Ejefas 10

1
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)
EUROPEAN JOURNAL OF ECONOMICS, FINANCE AND ADMINISTRATIVE SCIENCES

ISSN: 1450-2275 Issue 10 March, 2008
EUROPEAN JOURNAL OF ECONOMICS, FINANCE AND ADMINISTRATIVE SCIENCES http://www.eurojournals.com/EJEFAS.htm Editor-In-Chief Adrian M. Steinberg, Wissenschaftlicher Forscher Editorial Advisory Board Bansi Sawhney, University of Baltimore Jwyang Jiawen Yang, The George Washington University Zhihong Shi, State University of New York Zeljko Bogetic, The World Bank Jatin Pancholi, Middlesex University Christos Giannikos, Columbia University Hector Lozada, Seton Hall University Jan Dutta, Rutgers University Chiaku Chukwuogor-Ndu, Eastern Connecticut State University Neil Reid, University of Toledo John Mylonakis, Hellenic Open University (Tutor) M. Femi Ayadi, University of Houston-Clear Lake Emmanuel Anoruo, Coppin State University H. Young Baek, Nova Southeastern University Jean-Luc Grosso, University of South Carolina Sumter Richard Omotoye, Virginia State University Mahdi Hadi, Kuwait University Jean-Luc Grosso, University of South Carolina Ali Argun Karacabey, Ankara University Felix Ayadi, Texas Southern University Bansi Sawhney, University of Baltimore David Wang, Hsuan Chuang University Cornelis A. Los, Kazakh-British Technical University Leo V. Ryan, DePaul University Richard J. Hunter, Seton Hall University Said Elnashaie, Auburn University Panayiotis Tahinakis, University of Macedonia Mukhopadhyay Bappaditya, Management Development Institute M. Carmen Guisan, University of Santiago de Compostela Subrata Chowdhury, University of Rhode Island Teresa Smith, University of South Carolina Wassim Shahin, Lebanese American University Mete Feridun, Cyprus International University Teresa Smith, University of South Carolina Sumter Ranjit Biswas, Philadelphia University Katerina Lyroudi, University of Macedonia Maria Elena Garcia-Ruiz, University of Cantabria Zulkarnain Muhamad Sori, University Putra Malaysia Indexing / Abstracting European Journal of Economics, Finance and Administrative Sciences is indexed in Scopus, Elsevier Bibliographic Databases, EMBASE, Ulrich, DOAJ, Cabell, Compendex, GEOBASE, and Mosby.
Aims and Scope The European Journal of Scientific Research is a quarterly, peer-reviewed international research journal that addresses both applied and theoretical issues. The scope of the journal encompasses research articles, original research reports, reviews, short communications and scientific commentaries in the fields of economics, finance and administrative sciences. Editorial Policies 1) The journal realizes the meaning of fast publication to researchers, particularly to those working in competitive & dynamic fields. Hence, it offers an exceptionally fast publication schedule including prompt peer-review by the experts in the field and immediate publication upon acceptance. It is the major editorial policy to review the submitted articles as fast as possible and promptly include them in the forthcoming issues should they pass the evaluation process. 2) All research and reviews published in the journal have been fully peer-reviewed by two, and in some cases, three internal or external reviewers. Unless they are out of scope for the journal, or are of an unacceptably low standard of presentation, submitted articles will be sent to peer reviewers. They will generally be reviewed by two experts with the aim of reaching a first decision within a two-month period. Suggested reviewers will be considered alongside potential reviewers identified by their publication record or recommended by Editorial Board members. Reviewers are asked whether the manuscript is scientifically sound and coherent, how interesting it is and whether the quality of the writing is acceptable. Where possible, the final decision is made on the basis that the peer reviewers are in accordance with one another, or that at least there is no strong dissenting view. 3) In cases where there is strong disagreement either among peer reviewers or between the authors and peer reviewers, advice is sought from a member of the journal's Editorial Board. The journal allows a maximum of three revisions of any manuscripts. The ultimate responsibility for any decision lies with the Editor-in-Chief. Reviewers are also asked to indicate which articles they consider to be especially interesting or significant. These articles may be given greater prominence and greater external publicity. 4) Any manuscript submitted to the journals must not already have been published in another journal or be under consideration by any other journal. Manuscripts that are derived from papers presented at conferences can be submitted even if they have been published as part of the conference proceedings in a peer reviewed journal. Authors are required to ensure that no material submitted as part of a manuscript infringes existing copyrights, or the rights of a third party. Contributing authors retain copyright to their work. 5) The journal makes all published original research immediately accessible through www.EuroJournals.com without subscription charges or registration barriers. Through its open access policy, the journal is committed permanently to maintaining this policy. This process is streamlined thanks to a user-friendly, web-based system for submission and for referees to view manuscripts and return their reviews. The journal does not have page charges, color figure charges or submission fees. However, there is an article-processing and publication fee payable only if the article is accepted for publication. Submissions All papers are subjected to a blind peer review process. Manuscripts are invited from academicians, research students, and scientists for publication consideration. The journal welcomes submissions in all areas related to science. Each manuscript must include a 200 word abstract. Authors should list their contact information on a separate paper. Electronic submissions are acceptable. The journal publishes both applied and conceptual research. Articles for consideration are to be directed to the editor through ejefas@eurojournals.com. In the subject line of your e-mail please write "EJEFAS submission"
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Articles are accepted in MS-Word or pdf formats Contributors should adhere to the format of the journal. All correspondence should be directed to the editor There is no submission fee Publication fee for each accepted article is $150 USD
European Journal of Economics, Finance and Administrative Sciences is published in the United States of America at Lulu Press, Inc (Morrisville, North Carolina) by EuroJournals, Inc.
European Journal of Economics, Finance and Administrative Sciences

Issue 10 March, 2008
Contents
Rural Non-Farm Employment in India: The Trends and Determinants Rudra Prakash Pradhan A Substitute for Riskless Assets and the Market Portfolio Jose Rigoberto Parada Daza Effects of M&As on the Spanish Savings Banks Pedro L. Contreras, Victor Barrios, Romn Ferrer and C. Gonzlez The Effect of Model-Selection Uncertainty on Error Bands for Estimated Impulse Response Functions in Vector Autoregressive Models Islam Azzam Comparative Analysis of Socio-Economic Constraints in Niger-Delta Reuben Adeolu Alabi Market Efficiency and Company Size. Empirical Evidence from the Athens Stock Exchange Christos A. Alexakis Relationship between Strategic Human Resource Management and Organizational Performance: Evidence from Selected Malaysian Firms Raduan Che Rose, Naresh Kumar and Hazril Izwar Ibrahim The Reaction of Bank Lending to Macroeconomic Fluctuations of Monetary Policy Transmission in Greece Aikaterini Markidou and Eftychia Nikolaidou Private Cost of Teacher Education in a Nigerian University Roseline O. Olubor Management Training in Small and Medium Sized Enterprises: A Study of North Cyprus erife Zihni Eypolu Attitudes Toward Large-Scale Risks: A Discrete Estimation using Dutch Data Joseph G. Eisenhauer 7-18 19-28 29-45
46-62 63-74
75-85
86-97
98-115 116-128
129-137 138-150
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) 151-158 159-174 175-183 184-199 200-212
Cointegration and Priority Relationships between Stock Markets of Turkey, Brazil and Argentina Erman Erbaykal, H. Aydn Okuyan and zgr Kadolu Factors of Administrative Corruption in Nigeria and the Implication for Policy-Makers Abdul Raufu Ambali Job Satisfaction of Public Sector Managers in North Cyprus Tlen Saner and Berna Kocaman Management of Financial Margin Through Funds Transfer Pricing Jordi Carenys Fuster Methods of Researching Strategy Processes in Developing Countries Abdallah M Elamin The Enterprise Risk Management Model for Corporate Sustainability and Selection of the Best ERM Operator in the Turkish Automotive Distributor Company: ANP Based Approach Ayse Kucuk Yilmaz Effect of Capital Structure on Firms Performance: The Nigeria Experience Ishola Rufus Akintoye
213-232 233-243
European Journal of Economics, Finance and Administrative Sciences ISSN 1450-2887 Issue 10 (2008) EuroJournals, Inc. 2008 http://www.eurojournalsn.com
Rural Non-Farm Employment in India: The Trends and Determinants

Rudra Prakash Pradhan Assistant Professor, Vinod Gupta School of Management Indian Institute of Technology, Kharagpur, West Bengal- 721 302, India Fax: +91- 3222- 278027/255303 E-mail: rudrap@vgsom.iitkgp.ernet.in, pradhanrp@yahoo.com Abstract This article attempts to examine the trends and determinants of rural non-farm employment in the Indian economy. It has been analyzed under two phases. First phase focuses Indias state-wise and gender-wise trends of rural non-farm employment for the three cross sectional years viz. 1983, 1993-94 and 1999-2000. Second phase focuses the overall trends of rural non-farm employment and its determinants during the period 1970-71 to 2003-04. It finds that the trends of rural non-farm employment has been increasing in the Indian economy but varies across the states, gender and time periods. It further notices that Indias rural non-farm employment is substantially influenced by HYV coverage, rural literacy and rural road. While the impact of HYV is negative, rural literacy and rural road are positively determined the same. The paper finally suggests that boosting rural infrastructure has a substantial solution to expand rural non-farm employment in the Indian economy.
1. Introduction
The most conspicuous feature of the Indian economy is that India is steadily- if, not rapidly- urbanizing but millions of its population still live in rural areas (i. e. about 70%). Most of them are engaged in agriculture (i. e. about 70%) and millions are in the conditions of hunger, ill health, homelessness, illiteracy and subject to different forms of class, caste and gender oppression (Ramachandran, 1997). Alleviating such poverty and increasing the provision of basic civic amenities to its population, especially those living in rural areas, has remained one of the most emphatic goals of Indian planning since its inception in the beginning of 1950s. To put in other way, India always has some comprehensive rural development approach to alleviate its rural poverty and other social insecurities since its planning era. The term rural development represents a process of progressive improvement in economic securities of the people in the rural areas (i. e. mostly in the form of income and employment). It involves more than just expanding agricultural output and growth in real per capita agricultural income (Lanjouw et al., 2001). In many developing countries including India, agriculture is not the sole sectorsometimes not even the dominant sector- for household employment and income in the rural sector. Non-agricultural activities in those countries most often account for a large percentage of rural income (Narayanamoorthy et al., 2002) and rural employment, household income (World Bank, 1999) and overall economic growth (Lanjouw and Shariff, 2004). The percentage distribution, however, varies across the regions of the economy (Reardon et al., 1998) and even within these regions; there is also considerable variation across the states and districts. Over and above, rural development needs substantial improvement in both farm economy and non-farm economy. The farm economy involves agriculture and its allied and these include field crop
production, plantation, diary, livestock, agricultural services, forestry, logging, fishing and haunting. On the contrary, non-farm economy necessitates wide spectrum of activities in the rural areas, which are not directly associated with agriculture and its allied but plays an important in rural economy. These include employed and self-employed activities in manufacturing, commerce and other services. The development of farm economy is the first choice, as most of the rural people depend upon agriculture and also as per the low skills of rural poor. But the recent experience shows that the introduction of land-saving and labour-using technologies in agriculture is not in favour of generating adequate employment opportunities for rural poor. Also agriculture cannot bring forth passable employment in the long run and thus, it is alone not sufficient to meet the challenges of rural unemployment and poverty in the economy. If agriculture and its allied are failed to generate employment, rural poor has no other alternatives. They either migrate to urban areas or look for some supplementary opportunities in the rural areas in the form of non-farm activities. The first track is not good sign for the Indian economy so far as its balanced regional development is concerned. This is because it leads to capital drain from rural to urban areas. Also some people frequently migrate to urban areas without having necessary skills for their survive. Nevertheless they not only face the difficulty to find a job but also put pressure on urban economy and thus, disrupt urban development in the country. All these factors resulted a mismatch growth in the economy and the standard of living between rural and urban people. The second track on the way of planned and regulated development of non-farm sectors has a certainty to bring balanced regional development in the economy (Mishra, 2005). Keeping in above view, present paper seeks to examine the trends and determinants of rural non-farm employment in the Indian economy and suggests some policy implications for its boost in the rural economy. The rest of the paper is organized as follows. Section 2 provides a brief introduction about rural non-farm economy and its national importance. Section 3 describes the size and composition of rural non-farm sector in the Indian economy. Section 4 highlights the database and modeling. Section 5 discusses the results and section 6 concludes with policy implications.
2. Non-Farm Sectors and its Importance

As we have cited above, non-farm sector simply represents all economic activities in the rural areas except agriculture, livestock, fishing and haunting. Since it is defined negatively, as against agriculture, it is not a homogenous sector. For this reason, in the developing countries including India, it is a poorly understood component and we know relatively about its role in the broader development process only (Lanjouw and Lanjouw, 2001). The judgments about the viability and importance of rural non-farm sector hinges crucially on what is meant by rural (Lanjouw and Lanjouw, 2001). It is generally defined as a settlement of about 5000 population or a fewer. But the clear picture of rural locality based on the population size and/or functions of a rural locality based on the characteristics of settlement such as school, hospital seat of local government, etc. (Lanjouw and Feder, 2001). It is also true that as more and more people engaged in non-farm activities, a community may classify as urban, even if it is not changed in any other aspects (Haggblade et al., 1989; Acharya and Mitra, 2000). Rural economy is usually described as the combination of farm and non-farm sectors. Thus, rural development depends upon the performance of both farm sectors and non-farm sectors. Both are sometimes depends upon each other and the lack of one resulted the cost of other (Harriss, 1987) and leads to overall degradation of rural development. According to Mellor (1976), modern agriculture has strong forward and backward linkages with industry and other non-agricultural sectors, which are somewhat available in rural areas. Also the prospects of rural non-farm activities critically hinges on the performance of agriculture also (Hazell-Haggblade, 1990). These activities, which are hard to define with clarity, includes both off-farm and on-farm activities. Off-farm activities include straightforward agricultural activities like income earned by peasants and workers, as hired labour on farms owned by others. The empirical literature suggests that the level of rural poverty is lowest in
those households, whose income stems solely from off-farm self employment (Lanjouw, 2001; Van de Walle, 2000; Lanjouw, 1999). On-farm activities, on the contrary, comprise non-agricultural components and that includes mining, quarrying, manufacturing, utilities and construction under the secondary sector, trade-hotels-restaurants, transport-storage-communications, banking-insurance, real estate-business services and community-social-personal services under the tertiary sector (Lanjouw, and Lanjouw, 2001; Chadha, 2002). The diverging varieties of rural non-farm activities are substantially needed for a country like India because of its certain advantages in the object behind balanced regional development in the economy. These are as follows: First, the farm and non-farm linkage. It is mostly visualized in three forms: (a) an increase in farm income increases demand for consumer goods, which are more or less produced by local non-farm economy; (b) the growing demands for modern agricultural inputs, which are either produced or distributed by local non-farm enterprises (Collier and Lal, 1986; Evans and Ngau, 1991); (c) the rising agricultural productivity and wages will enhance the opportunity cost of labour in non-farm activities. This induces a shift in the composition of non-farm activities out of very labour-intensive, low-return activities into more skilled, higher investment, high-return activities. Rural non-farm activities are very useful for utilizing local talents and local resources, which cannot be easily transferred and utilized in the urban centres. According to Saith (1992), rural non-farm activities are usually labour- and local- resource intensive in nature, which would be in line with the perceived comparative advantage in most of the developing countries including India. The development of rural non-farm activities certainly prevents rural migration to urban areas (Islam, 1987) and encourages self-sufficiency among the rural people. The expansion of rural non-farm activities considerably bridges the gap between rural and urban economy and also among the peoples in the rural areas. The empirical studies suggest that rural income inequality is substantially less in areas, where a wide network of rural nonfarm employments are available. Rural industries are usually less capital intensive and more labour absorbing. This certainly ensures the social objectives of deriving higher employment and output gains for every unit of capital invested, which are readily fulfilled through a chain of rural non-farm activities. There is a substantial solution to rural poverty as well as urban poverty through a wide network of non-farm activities. This is because people having no land base of their own, can engage in these activities and their productivity and earning are much higher in non-farm rather than farm employment (Lanjouw and Shariff, 2004; Reardon, 1997; Bagachwa and Stewart, 1992). The growth of rural non-farm activities also establishes the linkage between rural and urban economy, which certainly reduces overall poverty in the economy. The empowerment of rural women is substantially visible in the above process, as it allows them to participate in non-farm activities (Rosegrant and Hazell, 2000).
3. Rural Non-Farm Sectors in India: Its Size and Composition

The rationale behind rural non-farm development in India is mostly due to population pressure only and can contribute to the economy with respect to employment, poverty, productivity, income distribution and efficient utilization of countrys economic resources (both physical and human). In this section, we highlight the size and composition of rural non-farm sectors particularly with regards to employment only. The need to diversifying the Indian rural economy was felt since Indian planning era. Some bold policy initiatives were mounted afterwards to expand the rural non-farm activities in the economy. India also accordingly received some diversification of employment from agriculture to non-agricultural activities. It is visible both in the gender-wise and state-wise. While male workers are
10
diversified from agriculture to non-farm activities, female workers are more or less stable in agriculture. The proportion of male in agriculture has been declined from 77.5% in 1983 to 74.1% in 1993-94 and 71.4% in 1999-00. On the other side, their proportion in non-farm activities has been increasing from 22.5% to 25.9% and 28.6% in the same successive periods. The proportion of female in agriculture has been varying from 86% to 87% since 1983. Though some kind of declining has been visible in 1993-94, it has increased considerably in the later period. Their employment in the non-farm activities has witnessed a steady increase from 13.5% in 1983 to 13.8% in 1993-94 and than declined to 13.7% in 1999-2000 (See Table 1).
Table 1:
Sector Description Agriculture Mining and Quarrying Manufacturing Utilities Construction Secondary Sector Trade and Hotels Transport and Communication Services Tertiary Sector All Non-agricultural
Source: Chadha, GK (2002)
Sectoral Distribution of Usual Status Rural Workers in Indian States by Workers Sex
1983 77.5 10.6 0.6 1.2 7.0 26.8 0.2 1.1 2.2 5.1 10.0 34.2 4.4 20.4 1.7 10.0 6.1 24.7 12.5 55.2 22.5 89.4 Male 1993-94 74.1 9.0 0.7 1.3 7.0 23.5 0.3 1.2 3.2 6.9 11.2 32.9 5.5 21.9 2.2 9.7 7.0 26.4 14.7 58.0 25.9 91.0 1999-00 71.4 6.5 0.6 0.9 7.3 22.4 0.2 0.8 4.5 8.7 12.6 32.8 6.8 29.4 3.2 10.4 6.2 19.0 16.2 58.8 28.6 93.5 1983 87.5 31.5 0.3 0.7 6.4 26.7 NA 0.2 0.7 3.2 8.7 30.8 1.9 9.5 0.1 0.6 2.8 26.7 4.8 37.7 13.5 68.5 Female 1993-94 86.2 24.7 0.4 0.6 7.0 24.1 NA 0.3 0.9 4.1 8.4 29.1 2.1 10.0 0.1 1.3 3.4 35.0 5.6 46.3 13.8 75.3 1999-00 86.3 17.6 0.3 0.4 7.6 24.0 NA 0.2 1.1 1.1 9.0 29.4 2.0 16.9 0.1 1.8 3.6 34.2 5.7 52.9 13.7 82.4
Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban Rural Urban
All in all, we have been observed two significant changes in the Indian economy with regards to diversification of rural employment. First, an inter-sectoral shifts, which was relatively sharper in the 1980s in contrast to 1990s. Second, the diversification of male workers, from agriculture to nonfarm activities, is considerably significant than female workers. In other words, females are badly affected especially during the reforms era of 1990s. This is because they not only compete with their male counter parts in the rural areas but also with their urban sisters (Pradhan, 2005; Chadha, 2002). In the state-wise comparison, it is of diverging in nature. In the states like Assam, Bihar, Himachal Pradesh, Kerala and Andhra Pradesh, it has been observed a significant diversification of rural workforce from agriculture to non-agriculture. The similar trend is also visible in the states like Haryana, Jammu and Kashmir, Madhya Pradesh, Orissa, Punjab, Rajasthan, Tamil Nadu and West Bengal, but at a very low rate. In some states, the diversification of employment has substantially increased in the 1980s but subsequently declined in the 1990s. They include Gujarat, Karnataka and Maharashtra. The picture is somewhat more interesting between male and female workers especially during the reforms era. For male, Assam, Himachal Pradesh, Kerala, Punjab and Uttar Pradesh exhibited a diversification from agriculture to non-agriculture and for female, Bihar, Jammu and Kashmir, Orissa, Uttar Pradesh and West Bengal showed the above tendency. In the states viz. Andhra Pradesh, Gujarat, Haryana, Himachal Pradesh, Karnataka, Maharashtra and Rajasthan, the proportion
11
of female engaged in agriculture has relatively increased (either by differing or constant proportions). By and large, there has been structural transformation in the rural employment (both for male and female workers) during the reforms era. In most of the cases, either it got reversed in some states or witnessed a halting pace in others. Only in few states, the diversification has been continuing over the years (See Table 2).
Table 2:
State AP Assam Bihar Gujarat Haryana HP JK Karn Kerala MP Mah Orissa Punjab Raj TN UP WB
Note:
Sectoral Distribution of Usual Status Rural Workers in Indian States by Workers Sex
Year 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 1983 1993-94 1999-00 RM 77.1 75.60 74.40 78.6 77.7 64.5 81.2 81.9 79.0 78.9 71.0 71.9 71.2 60.8 59.5 77.0 65.8 55.3 71.7 61.3 64.1 81.6 80.4 78.5 57.6 52.8 43.0 87.3 87.2 84.2 79.5 75.3 73.9 78.1 78.8 77.0 77.0 68.0 64.0 80.7 69.5 67.1 68.7 63.8 62.6 78.5 76.2 71.7 73.0 64.8 66.0 Agriculture RF 83.4 83.7 84.4 79.8 82.9 79.3 88.1 91.8 85.8 92.0 90.6 92.2 89.5 93.0 92.7 97.5 95.5 95.1 96.1 94.7 91.0 88.2 84.1 88.0 70.4 62.8 60.6 93.8 93.9 91.7 92.7 91.2 94.0 81.0 85.1 81.2 92.1 92.7 90.7 94.0 93.0 92.1 81.7 78.4 76.4 89.5 90.0 87.7 74.8 59.6 52.6 RT 80.1 79.3 78.8 80.0 78.7 67.7 83.5 84.2 80.7 84.4 78.7 80.4 77.7 71.9 69.8 87.6 80.2 74.8 79.5 75.5 73.0 84.2 81.9 82.2 63.1 56.1 48.8 90.0 89.9 87.2 85.6 82.6 82.8 79.1 81.0 78.6 82.5 74.6 72.9 86.7 79.8 77.9 74.6 70.4 68.3 82.0 80.0 76.4 73.6 63.6 63.0 RM 8.0 7.1 6.2 3.3 4.0 4.2 6.3 4.2 5.8 7.4 12.9 10.2 7.6 6.5 10.5 5.5 5.3 7.8 5.9 6.5 5.6 5.6 5.9 6.0 12.6 10.7 11.6 4.4 3.7 4.3 6.9 7.2 7.6 8.0 6.3 6.2 7.3 7.5 10.0 5.7 6.9 6.4 12.5 14.0 14.4 8.3 7.9 9.4 9.3 12.8 11.9 Manufacturing RF 7.6 7.4 6.1 9.2 10.5 10.0 6.1 4.1 8.5 3.3 4.2 2.1 3.8 1.5 2.1 1.1 1.7 1.1 2.1 0.9 5.7 6.6 7.8 5.8 17.7 19.4 20.2 3.3 3.3 4.0 2.7 3.1 2.2 10.0 7.6 12.6 4.2 1.3 3.0 2.5 1.5 2.9 9.9 13.1 14.4 5.4 4.8 6.5 16.6 30.3 38.0 RT 7.7 7.3 6.2 4.4 5.4 5.4 6.3 4.1 6.5 5.7 9.5 6.8 6.1 4.8 7.9 3.3 3.5 4.5 4.6 4.1 5.6 6.0 6.6 5.9 14.5 13.6 14.4 3.9 3.5 4.2 5.0 5.3 5.2 8.7 6.8 8.6 6.3 5.9 7.7 4.2 4.6 4.8 10.9 13.6 14.4 7.4 7.1 8.5 11.1 17.0 17.7 RM 22.6 24.4 25.6 21.2 22.3 35.5 18.6 18.1 21.0 20.3 28.8 28.1 28.5 39.2 40.5 22.1 34.2 44.7 27.9 38.6 35.9 18.2 19.6 21.5 42.3 47.1 57.0 12.5 12.8 15.8 20.2 24.7 26.1 21.8 21.2 23.0 22.3 31.9 36.0 19.0 30.4 32.9 31.2 36.2 37.4 21.0 23.8 28.3 26.8 35.1 34.0 Non-agriculture RF 16.3 16.3 15.6 18.0 17.1 20.7 11.8 18.2 14.2 7.1 9.4 7.8 9.9 6.8 7.3 2.4 4.5 4.9 3.6 5.2 9.0 11.6 15.9 12.0 29.5 37.1 39.4 5.8 6.1 8.3 7.0 8.8 6.0 19.0 14.9 18.8 7.2 7.3 9.3 6.0 7.0 7.9 18.1 21.6 23.6 11.1 10.0 12.3 24.8 40.4 47.4 RT 19.9 20.7 21.2 20.0 21.3 32.3 16.5 15.8 19.3 15.6 21.3 19.6 22.3 28.1 30.2 12.4 19.8 25.2 19.8 24.5 27.0 15.8 18.1 17.8 36.9 43.9 51.2 10.0 10.1 12.8 14.4 17.4 17.2 20.9 19.0 21.4 17.5 25.4 27.1 13.3 20.2 22.1 25.4 29.6 31.7 18.0 20.0 23.6 26.4 36.4 37.0
RM: Rural Male; RF: Rural Female; RT: Rural Total; AP: Andhra Pradesh; HP: Himachal Pradesh; JK: Jammu and Kashmir; Kar: Karnataka; MP: Madhya Pradesh; Mah: Maharashtra; Raj: Rajasthan; TN: Tamil Nadu; UP: Uttar Pradesh; WB: West Bengal Source: GOI, Economic Survey (1990), Sarvekshana, Vol. 19, Nos 1 & 2; GOI, NSSO.
12
In India, agriculture still continues to be the mainstay for rural female in number of states. There are around eight states, where the share of agricultural employment exceeds 90% and in 15 states, it is about 75%. It is only in West Bengal and Kerala, where rural females commanded a fairly respectable proportion in non-farm activities. In most of the states, about 15% of females are in nonfarm activities. But in the states like Gujarat, Haryana, Himachal Pradesh, Jammu and Kashmir, Madhya Pradesh, Maharashtra, Punjab and Rajasthan, their proportion in off-farms are extremely low (i.e. from 4.9% in Himachal Pradesh to 9.3% in Punjab). For sector-specific, rural employment in the manufacturing sector has been at a very low web and has not witnessed any improvement in the recent years. In 10 states, about 5-6% of the rural workers are engaged in manufacturing sector. In the states like Gujarat, Himachal Pradesh, Haryana, Maharashtra, Punjab and Rajasthan, the females presence in manufacturing sector is only 2-3%. But a noticeable presence of rural workers in the manufacturing sector has substantially visible in West Bengal, Kerala and Tamil Nadu and to a lesser extent in Assam and Orissa. In West Bengal and Kerala, the presence of female workers in the non-farm sectors has been substantially high than other states and their trend also improving in recent years. In short, it is of divergent experiences that we have been observed in the India economy with respect to employment diversification from agriculture to non-agriculture activities in the rural areas. In the sector-wise classification, some sectors progress well, while others are not. The trend is also very similar to interstate as well as male-female comparisons.
4. Database and Modeling

The preceding section clarifies that rural non-farm activities have a substantial contribution to employment in rural India. Though it is of diverging experiences across the states, activities, gender and different time periods, its importance has to be considered in a great extent in the interest of balanced regional development in the economy. In this section, we have been attempted to find out the overall trends and determinants of rural non-farm employment in the Indian economy. As per the availability of literature, we found numbers of studies that have been undertaken to examine the determinants of rural non-farm employment and its diversification both at the micro level and macro level. Most of them are based on cross sectional analysis and that related with different time periods and different regions/countries in the world. Some of the important studies, as per our knowledge, are: Lanjouw and Shariff (2004), Narayanmoorthy et al. (2002), Mecharla (2002), Lanjouw (2001), Lanjouw and Lanjouw, (2001), Lanjouw and Feder (2001), Fan et al. (1999), Verma and Verma (1995), Visaria (1995), Binswanger and Khandker (1995), Rao, (1995) Shylendra and Thomas (1995), Vaidyanathan (1994), Unni (1991), Ranis et al. (1990), and Dev (1990). The above investigation suggests the factors that affecting rural non-farm employment are: growth of crop output, size of operational holdings, per capita GDP, per capita development expenditure and the availability of socio-economic infrastructure in the rural areas like irrigation, transport, power, literacy, etc. The present paper picked up some relevant variables from these literature and applied time series analysis for examining the determinants of rural non-farm employment in the Indian economy. The data used under this study are secondary in nature and covered from the period 1970-71 to 200304. It has been mostly collected from Economic Survey, Government of India, Centre for Monitoring Indian Economy (CMIE), Mumbai, Census of India, Hand Books of Statistics, Reserve Bank of India, Mumbai and International Food Policy Research Institute (IFPRI), Washington. To attain our objective, we fit a multi-variate regression model, which is as follows: RNFEt = + 1PCDEVEXt + 2HYVt + 3IRRt + 4RLITt + 5RROADt + 6AGGRt + 7PCGDPt + Ut Where, RNFE = Rural non-farm employment (in %); PCDEVEX = Per capita development expenditure (in rupees); HYV = Area under high yielding varieties (in %); IRR = Area under irrigation (in %); RLIT = Rural literacy (in %); RROAD = Rural road (in thousand per square kilometers);
13
AGGR = Agricultural growth (in %); PCGDP = Per capita GDP (in rupees); U = Error term; t = Time periods; and and s are the parameters, which to be estimated. The variables, except agricultural growth and per capita GDP, used under the model are aggregated form of district-level data, which we obtained from IFPRI research report (See Fan et al., 1999). The report provided data from 1970-71 to 1995-96 and the rest of the data obtained from various other sources, which we have mentioned above. In some cases, we obtained the data by applying the method of interpolation. The data of agricultural growth and per capita GDP were directly obtained from Economic Survey, Government of India. And the variable rural non-farm employment, which is as a percentage of total rural employment, is based on NSS data and that obtained from IFPRI report and the Sources of Economic Growth in India: 1950-51 to 1999-2000 (Sivasubramonian, 2004).
5. Results and Discussion

The basic objective of this paper is to examine the trends of rural non-farm employment and its determinants in the Indian economy. We have already discussed the first part of our objectives in the preceding section for the state-wise, sector-wise and gender-wise picture of three cross sectional years viz. 1983, 1993-94 and 1999-2000. In this section, we examined the overall time series trends of rural non-farm employment and its determinants. While the trend of rural non-farm employment has been reported in Figure 1, the determinants are reported in Tables 3, 4 and 5. The rural non-farm employment, as a percentage of total rural employment, has been substantially increasing in the Indian economy since 1970-71. It has been around 19-20% in the 1970s, 20-25% in the 1980s and 25-30% in the 1990s and beyond. The turn is now to examine the factors that substantially influencing this trend of rural non-farm employment. To know the same, we initially assumed seven different factors, which we have mentioned in the earlier section. We first examined the interrelationship among these variables. For this, we calculated the correlation matrix and the results are reported in Table 3. The results ensured that agricultural growth has weak correlation, while all other factors have strong correlations with each other at a positive level. This reflects that there is presence of some multicolinearity in our model. This is somewhat natural, as some of the variables are considerably interdependent to each other. To avoid such problems and to get a better econometric model, we have applied step-wise regression in the subsequent stage.
Figure 1: The trends of Rural Non-farm Employment in the Indian Economy
Source: Sivasubramonian, S. (2004); and Fan et al. (1999).
14
Table 3:
X1 X2 X3 X4 X5 X6 X7 X8

Variance-covariance matrix of Rural Non-farm Employment Determinants
X1 1.000 X2 0.988 1.000 X3 0.977 0.991 1.000 X4 0.99 0.992 0.987 1.000 X5 0.982 0.982 0.989 0.981 1.000 X6 0.991 0.996 0.991 0.996 0.984 1.000 X7 0.033 0.020 0.033 0.058 0.033 0.046 1.000 X8 0.944 0.931 0.941 0.938 0.959 0.937 0.049 1.000
Note:
X1: Rural non-farm employment; X2: Per capita development expenditure; X3: HYV area; X4: Irrigation; X5: Rural literacy; X6: Rural road; X7: Agricultural growth; and X8: Per capita GDP.
The study at first hypothesized that per capita GDP (PCGDP), growth of crops output, agricultural modernization, availability of rural infrastructure and per capita development expenditure (PCDEVEX) have a substantial influence on Indias rural non-farm employment. In other words, besides host of other factors, increasing- agricultural growth, HYV coverage (as a proxy to agricultural modernization), PCGDP, PCDEVEX and the availability of rural infrastructure (such as irrigation, literacy and rural road) may certainly increase the level of rural non-farm employment. But our empirical investigation suggested that agricultural growth and HYV coverage have a negative impact on Indian rural non-farm employment. While HYV is statistically significant at 1% probability level, agricultural growth is not in any way significant. As a result, they go against our hypothesis. On the contrary, Indias rural non-farm employment is positively determined by PCGDP, PCDEVEX, irrigation, rural literacy and rural road. Among them, rural literacy and rural road have a larger impact than others and both are statistically significant at 5% probability level. The coefficient of determination (R2 = 0.988) reflects that about 99% of the systematic variations of rural non-farm employment is explained by these determinants. The F-test, which is asymptotically justified the class of models, indicate that R2 is statistically significant at 1% probability level (See Table 4). The DurbinWatson (DW) statistics clarifies that the estimated model has no autocorrelation. But the model is not free from multicolinearity, as some of the variables are not statistically significant at a high level of R2. This problem also earlier identified through correlation matrix. This has been subsequently wiped out by step-wise method and the estimated results under this method are reported in Table 5.
Table 4: Estimated Results of Rural Non-farm Employment Determinants
Estimated Coefficients 5.913 (2.72) 0.0196 (0.02) -0.153 (0.04) 0.126 (0.17) 0.236 (0.09) 0.0023 (0.001) -0.008 (0.013) 0.0064 (0.006) 0.991 (0.988) 400.15 1.5060 t- Statistics 2.169 1.097 -3.864 0.740 2.561 2.392 -0.622 1.046 Significance Level 0.039 0.283 0.001 0.466 0.017 0.001 0.534 0.305 0.000
Determinants Constant PCDEVEX HYV IRR RLIT RROAD AGGR PCGDP R2 F DW

Note:
PCDEVEX: Per capita development expenditure; HYV: Area under High Yielding Varieties; IRR: Area under irrigation; RLIT: Rural literacy; RROAD: Rural road; AGGR: Agricultural growth; PCGDP: Per capita GDP; R2: Coefficient of Determination; F: F-statistics; and the parentheses indicate the standard errors.
15
Table 5:

Step-wise Estimated Results of Rural Non-farm Employment Determinants
Estimated Coefficients 5.831 (1.143) 0.0033 (0.000) 0.295 (0.075) -0.148 (0.037) 0.990 (0.989) 952.82 1.5110 t- Statistics 5.103 8.244 3.950 -3.971 Significance Level 0.000 0.000 0.000 0.000 0.000
Determinants Constant RROAD RLIT HYV R2 F DW

Note: All notations area defined earlier.
The results indicate that rural literacy, rural road and HYV coverage are the most noteworthy factors that determining the level of rural non-farm employment in the Indian economy during 1970-71 to 2003-04. They are all statistically significant at 1% probability level. The nature of these variables has remained unchanged like our earlier estimated model. That means rural literacy and rural road have a positive influence on rural non-farm employment, while HYV has a negative impact on the same. The coefficient of determination (R2 = 0.99) reflects that 99% of the systematic variations of Indias rural non-farm employment is explained by rural road, rural literacy and HYV coverage. The model is asymptotically justified, as F-test ensured that R2 is highly statistically significant at 1% probability level. The Durbin-Watson (DW) statistics also clarifies the absence of autocorrelation. The present model is now free from multicolinearity problems, as all the estimated variables are highly significant with high value of R2.
6. Conclusions and Policy Implications

The present study analyzed the trends and determinants of rural non-farm employment in the Indian economy. It has been discussed under two heads: cross-sectional analysis and time series analysis. The first part focused Indias state-wise and gender-wise trends of rural non-farm employment for the three cross-sectional years viz., 1983, 1993-94 and 1999-2000. The second part focused the overall time series trends and determinants of Indian rural non-farm employment during the period 1970-71 to 2003-04. So far as trends are concerned, rural non-farm employment has been increasing in the Indian economy but varies across the states, between male and female groups and between different time periods. This represents that there is employment diversification in the rural India from agriculture to non-agriculture but at a varying rate with respect to states, gender and time periods. The study further finds that Indias rural non-farm employment is substantially influenced by per capita GDP, agricultural growth, HYV coverage, irrigation, rural road, rural literacy and per capita development expenditure. Except agricultural growth and HYV coverage, all others are positively related with rural non-farm employment. The impact of agricultural growth is not statistically significant, while HYV coverage is highly significant. This indicates that an increase of HYV coverage will lead to decline of rural non-farm employment. But in reality, the picture may reflect differently. That means HYV can influence rural non-farm employment indirectly. For instance, a substantial adoption of HYV will lead to increase the productivity of agriculture and thus, augments farm income. An increase of farm income further leads to increasing the demands of consumer goods, which is more or less supplied by rural non-farm sectors. As a result, it improves the production of rural non-farm sectors and also the rural non-farm employment. Similarly, agricultural growth can also influence rural non-farm employment. For instance, a sustainable agricultural growth can enhance overall economic growth (both in rural as well as urban) and thus, generates employment opportunities in both rural and urban economy. But in India, agricultural growth is lack of sustainability. That means there exists large-scale variation over time and across regions/states. As a result, it not only affects employment in the agricultural sector but also in the non-farm sectors. This is because, besides host of other factors,
16
growth of rural non-farm sectors is considerably guided by growth of agriculture. The failure of one leads to failure of others with respect output, employment and income. Surprisingly, Indian agriculture still depends upon gamble of monsoon. With the low availability of rural infrastructure (with regards to quantity, quality and as per the needs of rural economy), agricultural modernization always remains question mark. Though our empirical investigation suggests that rural infrastructure like road, literacy and irrigation have a substantial positive impact on rural non-farm employment, but it could be much higher (both directly and via growth of agriculture and other non-farm sectors), if India has high availability of these facilities in terms of both quantity and quality. Moreover, boosting of rural infrastructure in India is a matter of government intervention. It is, thus, the duty of government to heighten its rural infrastructure at any cost. This requires comprehensive fiscal reforms in the form of cutting debt service payments and other current spendings like wages and salaries and subsidies to public sector enterprises and makes availability of funds for infrastructure investment and its maintenance. Central government simultaneously asks the local governments to look after the development of rural infrastructure in their locality. Thats what Chinese government is doing now for attracting its foreign direct investment. Moreover, government has to go by public-private partnership and/or allow corporate sector independently to invest in rural infrastructure. To conclude, a substantial progress of rural India depends upon its growth of rural non-farm sectors. India needs a vigorous policy decision to boost its rural non-farm sectors particularly with regards to rural employment. The above process certainly reduces the burden of agriculture in particular and can improve the overall rural development in general.
17
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Acharya, S and Mitra, A (2000). The Potential of Rural Industries and Trade to Provide Decent Work Conditions: A Data Reconnaissance in India. SAAT Working Paper, South Asia Multidisciplinary Advisory Team. International Labour Organization, New Delhi. Bagachwa, MD and Stewart, F (1992). Rural Industries and Rural Linkages. In Stewart, F, Lal, S & Wangwe, S (eds) Alternative Development Strategies in Sub-Saharan Africa. Macmillan Press, London. Binswanger, HP & Khandker, S (1995). The Impact of Formal Finance on the Rural Economy of India. Journal of Development Studies, 32 (2): 234-262. Chadha, GK (2002). Rural Non-farm Employment in India: What Does Recent Experience Teach Us. Indian Journal of Labour Economics, 45 (4): 665-694. Collier, P and Lal, D (1986). Labour and Poverty in Kenya: 1900-1980. Clarendon Press, Oxford. Dev, SM (1990). Non-agricultural Employment in Rural India: Evidence at a Disaggregate Level. Economic and Political Weekly, 25 (28): 1526-1536. Evans, H and Nagu, P (1991). Rural-urban Relations, Household Income Diversification and Agricultural Productivity. Development and Change, 22 (3): 519-545. Fan, S, Hazell, P and Thorat, S (1999). Government Spending, Growth and Poverty in Rural India. International Food Policy Research Institute Research Report No. 110. International Food Policy Research Institute, Washington DC. Haggblade, S, Hazell, P & Brown, J (1989). Farm-Non-farm Linkages in Rural Sub-Saharan Africa. World Development, 17 (8): 1173-1201. Harriss, B (1987). Regional Growth Linkages from Agriculture. Journal of Development Studies, 23 (2): 275-289. Hazell, PBR and Haggblade, S (1990). Rural-urban Growth Linkages in India. PRE Working Paper No. 430. Washington DC: The World Bank. Islam, R (1987). Rural Industrialization in Asia. ILO-ARTEP, New Delhi. Lanjouw, JO and Lanjouw, P (2001). The Rural Non-farm Sector: Issues and Evidence from Developing Countries. Agricultural Economics, 26 (1): 1-23. Lanjouw, P (1999). Rural Non-agricultural Employment and Poverty in Ecuador. Economic Development and Cultural Change, 48 (1): 91-122. Lanjouw, P (2001). Non-farm Employment and Poverty in Rural El Salvador. World Development, 29 (3): 529-547. Lanjouw, P and Shariff, A, (2004). Rural Non-farm Employment in India: Access, Incomes and Poverty Impact. Economic and Political Weekly, 39 (40): 4429-4446. Lanjouw, P and Feder, G (2001). Rural Non-farm Activities and Rural Development: From Experience Towards Strategy. World Bank Rural Development Department, Rural Strategy Background Paper No. 4. Washington DC: The World Bank. Lanjouw, P, Quizon, J and Sparrow, R (2001). Non-agricultural Earnings in Peri-urban Areas of Tanzania: Evidence from Household Survey Data. Food Policy, 26 (3): 385-403. Mecharla, PR (2002). Determinants of Inter-District Variations in Rural Non-farm Employment in Andhra Pradesh. Indian Journal of Labour Economic, 45 (4): 807-819. Mellor, J (1976). The New Economics of Growth. Cornell Press, Ithaca. Mishra, AK (2005). SSVI and Rural Development. Yojana, 49 (11): 79-82. Narayanamoorthy, A, Rodrigues, Q and Phadins A (2002). Determinants of Rural Non-farm Employment: An Analysis of 256 Districts. Indian Journal of Labour Economics, 45 (4): 759769. Pradhan, R. P. (2005). Economic Reforms India and Its Impact on Rural Employment. Paper presented at South Asia Regional Conference, Hyderabad, 23-25 March.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
18 [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Ramachandran, VK (1997). On Keralas Development Achievements. In Dreze, J & Sen, A (eds) Indian Development: Selective Perspective. Oxford University Press, New Delhi. Ranis, G, Stewart, F and Angeles-Reyes, E (1990). Linkages in Developing Economies: A Philippines Study. ICS Press, San Francisco. Rao, GG (1995). Rural Non-farm Employment and Pattern of Rural Non-farm Employment by Geo-Agro Base: A Study of Western Godavari District. Indian Journal of Agricultural Economics, 50 (1): 86-92. Reardon, T (1997). Using Evidence of Household Income Diversification to Inform Study of the Rural Non-farm Labour Market in Africa. World Development, 25 (5): 735-747. Reardon, T, Stamoulis, K, Balisacan, A, Curz, ME, Berdegue, J and Banks, B (1998). Rural Non-farm Income in Developing Countries: Importance and Policy Implications. The state of Food and Agriculture. Food and Agricultural Organization, Rome. Rosergrant, MW & Hazell, P (2000). Transforming the Rural Asian Economy: The Unfinished Revolution. Asian Development Bank, Manila. Saith, A (1992), The Rural Non-farm Economy: Process. International Labour Organization, Geneva. Shylendra, HS and Thomas, P (1995). Non-farm Employment: Nature, Magnitude and Determinants in a Semi-Arid Village of Western India. Indian Journal of Agricultural Economics, 50 (3): 410-421. Sivasubramonian, S (2004). The Sources of Economic Growth in India. Oxford University Press, New Delhi. Unni, J (1991). Regional Variations in Rural Non-Agricultural Employment: An Explanatory Analysis. Economic and Political Weekly, 26 (3): 109-122. Vaidyanathan, A (1994). Employment Situation: Some Emerging Perspectives. Economic and Political Weekly, 29 (50): 3147-3156. Van De Walle, DV (2000). Is the Emerging Non-farm Market Economy the Route Out of Poverty in Vietnam. The World Bank, Washington DC. Verma, BN and Verma, N (1995). Distress Diversification from Farm to Non-farm Rural Employment Sector in the Eastern Region. Indian Journal Agricultural Economics, 56 (3): 422429. Visaria, P (1995). Rural Non-farm Employment in India: Trends and Issues for Research. Indian Journal of Agricultural Economics, 50 (3): 398-409. World Bank (1999). Tanzania: Peri Urban Development in the African Mirror. World Bank Research Report No. 19526-TA. The World Bank, Washington DC.
A Substitute for Riskless Assets and the Market Portfolio

Jose Rigoberto Parada Daza Professor of Finances at the Universidad de Concepcion- Chile Department of Administration, Economics And Administrative Sciences, Casilla 160-C Tel: (56 41) 20 4172; Fax: (56 41) 204172 E-mail: rparada@udec.cl Abstract A review of the traditional literature of the efficient frontier of mean-variance (MV) reveals the development of an alternative to CAPM and the Capital Market Line, the re-elaboration of a test to measure the efficient frontier without riskless assets, and the development of the efficient frontier with short sales. These perspectives let us re-elaborate the concept of a risky substitute portfolio that generates a profit equal to that of a riskless financial asset. This papers develops some propositions for the formation of a portfolio that can substitute for the riskless asset, but that is made up of risky assets and has set investment proportions. Likewise, the paper analyzes the creation of a substitute for the market portfolio. Keywords: Riskless asset, CAPM, mean-variance, efficient frontier
1. Introduction
The CAPM model is based on what is called the Security Market Line, and is represented by: E(Ri ) = RF + i [E(Rm ) RF ] i M where i = cov(Ri,Rm)/var(Rm), M = Security Market. i is the Beta of asset i, E(Rm) is the market portfolio return, and RF = the riskless interest rate. We develop the Capital Market Line by investing % of our resources in the market portfolio and (1-)% in a riskless asset; the return of this new asset is: Ri = (1 - )RF + E(Rm). Finally, in equilibirum, the Global Model is as follows: E(Ri) = RF + (i/m) (E(Rm) - RF) where i = standard deviation of an assets return, and m = standard deviation of the market portfolios return. In an early work, Black [1972] demonstrated that the model is verified even without a riskless asset, or one having = 0. The model defines market portfolios as the percentage of risky assets j, for j = 1, k and having = 1 so that it satisfies the following relationship according to Jarrow, [1988]: k 0 Pm = ij p(x j )/ xk p(xk ) k =1
where
P
j =1
mj
=1
20
Thus, CAPM requires a riskless asset ( = 0) as well as risky assets from the market portfolio ( = 1). With this article, we will show how a portfolio with a different definition of riskless and market portfolios, and which can be substituted by other equivalent portfolios, can exist.
2. Substitute Portfolio of Riskless Assets

Using Markowitzs [1958] definition of the efficient frontier of mean-variance (MV) Jarrow [1988] lays out an alternative to CAPM and the Capital Market Line. He then offers us an interesting perspective by using this same MV methodology to re-elaborate the Capital Market Line using nonlinear optimization. Kandel [1984] elaborates a test to measure the efficient frontier without riskless assets. Elton and Gruber [1995] develop the case of the efficient frontier based on short sales and in which classic methods of determination are described. We use these perspectives to re-elaborate the concept of a substitute portfolio whose return is equal to the return of a riskless portfolio. Since our methodology is based on the development of Sharpes initial Market Model [1964, 1970], we use the following assumptions and propositions. Assumptions 1. Suppose that CAPM and MV assumptions are met, and that there is a possibility for short sales. 2. There are two risky assets (1 and 2) that have the following conditions: 2.1) These assets have the following returns when in equilibrium, i.e. lie along the Security Market Line (SML), (Jarrow, pg. 223) E(R1) = RF + 1 [E(Rm) - RF] E(R2) = RF + 2 [E(Rm) - RF] In general: E(Ri) = RF + i [E(Rm) - RF] 2.2) When in equilibrium, i.e lie along SML, their total risks are given by (Sharpe,1970): 2 2 2 2 1 = 12 m and 2 = 22 m In general: i2 = (im)2 = systematic risk. 2.3) When assets 1 and 2 are in equilibrium, i.e. lie along SML, the covariances of these two assets in function of the Betas of the CAPM model are as follows: (See Appendix 1) (1,2) = (2/1)/21 = (12m2) where 2m = market portfolio variance. In general: (i,j) = (i/j)/j2 Proposition I. By mixing risky assets, a portfolio can be made that generates a return equivalent to the return of a riskless asset and which has zero risk. This portfolio includes a risky asset and is financed with personal equity and debt or short sales. Proof We can create a portfolio with these two assets by investing x1 and x2 so that x1 + x2 = 1; the risk, 2c, can be minimized following the MV model, in other words: 2 2 2 c = (1 m )2 x1 + (1 x1 )2 (2 m )2 + 2x1 (1 x1 )(1 2 ) m (1) 2 By resolving the first order conditions c/x1=0 and performing algebraic operations, the following proportions are obtained: (See Appendix 2) x*1 = 2/(2 - 1) (2) x*2 = -1/(2 - 1) (3) A portfolio having the proportions x*1 and x*2 has the following characteristics in expected return, E(Rc), and risk, 2c:
21
1 2 [RF + 2 (E(Rm ) RF )] E(Rc ) = [RF + 1 (E(Rm ) RF )] 1 1 2 2 By rearranging, it can be shown that: E(Rc) = RF and 2 2 2 2 c = (1 m ) 2 [ 2 /( 2 1 )] + ( 2 M ) 2 [ 1 /( 2 1 )] 2 (1 2 )/( 2 1 ) 2 (1 2 M ) By algebraic reduction, it can be shown that: 2c = 0 Therefore, a portfolio investing x*1 and x*2 in two assets has an expected return of RF and a risk equal to 2c = 0. In other words, this portfolio generates returns equivalent to a riskless rate and with zero risk, which is the same as investing all our personal equity in a riskless asset.
Investing and Financing Analysis of new portfolio From x*1 and x*2, it can be seen that, if x*1>1 x*2<0 or if x*1<0x*2>1. So for any x*1>1, a greater proportion of the resources available will be invested, being summarized as: x*1>1 x*2<0 or x2*<0x*2>1 We know that x1 + x2 = 1, and if x*1 > 1 and x*2 < 0, then: x*1 = x*2 + 1; this indicates investment in the risky asset x*1 and financing x*2 with debt (or short sales) and one unit of personal equity. In this case, x*2 indicates the times that we must go into debt (or make short sales) per unit of personal equity. Supposing that we have $C, then the equation becomes the following: x*1C = x*2C + C, or investment = financing Summarizing: Investment: x*1 Financing: Debt (or short sale): x*2 Personal equity: 1 The debt (or short sale) is made at a risky rate of: RF + 2[E(Rm) - RF]; the investment in asset 1 generates a return of: RF + 1[E(Rm) - RF]. Supposing that we can invest a total of $C of personal equity, then the profit to be invested in a riskless asset is $RFC. A riskless portfolio that is equivalent to investing in a riskless asset will be made up as follows:
Investment Financing Debt or short sale Personal equity Proportion x*1 x*2 1 Value x*1C x*2C C
Asset 1 Investment Earnings = x*1[RF + 1(E(Rm) - RF)] Financing Cost = x*2[RF + 2(E(Rm) - RF)] By replacing the values of x*1 and x*2 with x*1 = 1/(2 - 1) and x*2 = -2/(2 - 1), we arrive at: Net Utility = Investment Earning Financing Cost = RF A particular case of x*1 and x*2, that is deduced from equations (2) and (3) is that, if asset i has i = 0, then x*1 = 1, which is the very definition of a riskless asset and coincides with the minimum point of the Capital Market Line. This first proposition shows that an asset of i=0 is not vital. It can be replaced by an alternative portfolio that invests x*1 in risky assets, finances them with debt or short sales plus personal equity, and can generate the same riskless return with null risk.
22
3. A Substitute for the Market Portfolio

Proposition II. According to the Security Market Line, alternative market portfolios, offering the same risk and return as a market portfolio with =1, can be made with risky assets whose Betas are not equal to one. Proof a) Alternative market portfolios The same assumptions hold as in Proposition I. In this case, it is necessary that: x1E(R1) + (1 - x1)E(R2) = E(Rm), where E(Ri) = expected return of risky asset i when in equilibrium according to CAPM, E(Rm) = expected return of a market portfolio, and xi = proportion to invest in risky asset i, having a known i. Accordingly, a new portfolio is attempted that has the same expected market return risk, mathematically implying the following: 2 2 MIN L = x1 (1 m )2 + (1 x1 )2 (2 m )2 + 2x1(1 x1 )1 2 m + [E(Rm ) x1 E(R1 ) (1 x1 )E(R2 )] By solving for 2c/x1 = 0 and 2c/ = 0, and by rearranging algebraically, the following results are obtained for x1 and x2: (See Appendix No. 3) (4) x*1 = (2 - 1)/(2 - 1) x*2 = (1 - 1)/(2 - 1) (5) 2 We can arrive at a return of E(Rc) and a risk of c for a portfolio formed with x*1 and x*2: E(Rc ) = [R F + 1 (E(Rm ) R F )]( 2 1)/( 2 1 ) + [R F + 2 (E(Rm ) R F )](1 1 )/( 2 1 ) By arranging, we arrive at: E(Rc) = E(Rm) and
1 1 1 1 1 1 2 ( 2 m ) 2 + 2 2 (1 m ) 2 + = 2 (1 2 m ) 1 1 1 2 1 2 2 2 2 2 By arranging, we arrive at: c = m In other words, the new portfolio formed of x*1 and x*2 has an expected return equal to the return of a Market Portfolio, E(Rm), and the same risk, 2m, as this Portfolio. We can see from (5) and (6) that, if i = 1 in any case, all resources should be invested in this asset. The situation used for the Security Market Line is a specific case of (5) and (6), in which there is an exact coincidence of this line with the efficient frontier
2 c 2 2
b) Investment and Financing of the Alternative Portfolio b.1. Investment and personal equity If 0 < x*1 < 1 0 < x*2 < 1, when 2 > 1, 2 > 1, and 1< 1 < 2 In this case, we invest in x*1 and x*2, and finance everything with personal equity. If we have $C of personal equity, then the investment in asset 1 is: [(2 - 1)/(2 - 1)]C and in asset 2 is: [(1 1)/(2 - 1)]C with an expected return of E(Rc) = E(Rm) and a total risk of 2c = 2m. In other words, the return is equal to that of a market portfolio and the risk is equivalent to that of a market portfolio, 2m. b.2. Investment and financing with debt (or short sales) plus personal equity. We may see the following situations, depending on each assets Beta value: x*1 > 1 x*2 < 0 or x*1 < 0 x*2 > 1 In each of these two cases, we obtain a portfolio that has the following relationships: if x*1 >1 and x*2 < 0 x*1 = x*2 + 1 if x*1 <0 and x*2 > 1 or x*2 = x*1 + 1 In general, if: x*i = x*j + 1, then a proportion of x*i is invested in asset i, generating a return equal to RF + i [E(Rm) - RF]. To finance this investment, we use one unit of personal equity plus debt
23
or short sales in proportion x*j at a cost equal to RF + j [E(Rm) - RF]. This portfolio generates a return of E(Rm) with a risk equal to 2m In effect: by substituting the values of x*1 and x*2 in the return function of the new portfolio, we arrive at: 1 1 1 E(Rc ) = 2 [RF + 1 (E(Rm ) RF )] + [RF + 2 (E(Rm ) Ri )] 1 1 2 2 By arranging, we arrive at: E(Rc) = E(Rm). Regarding the risk, we arrive at: 2c = 2m Thus, we show how a portfolio including an investment in asset 1, financed with debt and personal equity, has a risk of 2m and a return of E(Rm), which is equivalent to investing personal equity alone in a market portfolio with =1.
4. The Efficient Frontier and Capital Market Line. A General Case

Both previously mentioned propositions are specific cases of a more general situation determining alternative portfolios having any Beta other than zero or one, the respective values initially established for riskless and market portfolios. In a more general formulation, Brennan [1971] focused on the formation of portfolios with debt and borrowing at different interest rates by using the concept of the efficient frontier. In an analogous approach, Jarrow [1988] focused on the efficient frontier and its relationship with CAPM. In this paper, we sought out alternative portfolios using Jarrows and Sharpes methodologies, since both authors demonstrate market portfolios to be in the efficient frontier, and, therefore, efficient portfolios. However, in general terms, the following question can be asked: Is there a tangency point with the Capital Market Line for any efficient portfolio (or any portfolio within the efficient frontier)? The answer, according to the methodological scheme of propositions I and II, is yes. The proof is as follows: Our idea is to use two assets, having 0 and 1, to form a portfolio subject to desired return, Rd. According to Lagranges conditions, the function to be minimized is: 2 2 L = x 1 ( 1 m ) 2 + (1 x 1 ) 2 ( 2 m ) 2 + 2x 1 (1 x 2 ) 1 2 m + [R d x 1 R 1 (1 x 1 ) R 2 ] where the same assumptions hold as in propositions I and II. Using the First Order minimization conditions, we arrive at: 2 2 2 L / x 1 = 2x 1 m [ 2 1 ] + [R 2 R 1 ] 2 m 2 ( 2 1 ) = 0 L/ = Rd x1 R1 + x1 R2 R2 = 0 By solving for the two equations simultaneously, the results are the following: (See Appendix No. 4) 2 (R RF ) 1 with 1 2 y Rm RF x1, = d 2 1 (Rm RF ) (2 1 ) (6) (7) The portfolios created by following an investment strategy, be it for asset 1 or 2 (each of these having 1 and 2), that generates a return equal to RF + k (E(Rm) - RF) and is financed with personal equity and short sales at a cost equal to RF + i(E(Rm) - RF), are as much in the efficient frontier as the Capital Market Line. Moreover, such portfolios have desired return of Rd and minimum risk. In order to know that solutions (6) and (7) are in the efficient frontier, we must show that the portfolios risk is at a minimum and unique. In order to show that this portfolio is on the Capital Market Line (CML), we must show that it offers the same desired return, Rd, and risk as a Portfolio in the Efficient Frontier.
, x2 =
1 ( R RF ) + d 2 1 ( Rm RF )
1 ( 2 1 )
24 Proof
It is known that the CML is: R j = RF +

2 2 2
(E (Rm ) RF ) m j = Standard Deviation of alternative portfolio made up of the proportions x,1 and x,2 or:
, , 2 2 = (x1 ) ( 1 m ) + (1 x1, ) ( 2 m ) + 2x1, (1 x1 )1 2 m j 2
By replacing the values of x,1 and x,2, and performing algebraic operations, we arrive at: R RF j = m d and R R m F by replacing in the CML, we arrive at: (R R F ) (R m R F ) R j = RF + m d m (R m R F ) In other words Rj = Rd, which in turn is equal to the return obtained in the Portfolio in the Efficient Frontier. Therefore, by using the Efficient Frontier or the CML, we finance the alternative portfolio by investing and financing with debt or short sales, and if CAPM is verified, then it has the same return and risk. The Efficient Frontier and the CML, then, lead to the same conclusion. We can also see that, in x,1 and x,2, if Rd = RF, we arrives at the same conclusion as Proposition I, in other words, the desired return of the portfolio is the riskless rate. Now, if Rd = Rm, then we arrive at Proposition II, the desired return of this formulation being the general case and Propositions I and II being specific cases. On the other hand, from the general expressions of x1, and x2, we saw that the definitions of the riskless Portfolio with = 0 is a very specific case, as is the Market Portfolio with = 1. Both portfolios can be created with assets having 0 and 1 and will achieve the same results, as can be seen in this paper.
5. Investing and Financing the New Portfolio. General case

Proposition II has already shown that the new portfolio can have an asset, be it financed with debt or short sales plus personal equity, based on the Investment = Financing identity, which breaks down into: Investment (in Asset 1 or 2) = Debt (or short sale) + Personal equity. If we have $C of personal equity, we may see the following three situations: a) 0 < x1, < 1 and 0 < x2, < 1 In this case: Investment (in $): x1C + x2C Financing (in $): C Or: x1, C + x2C = C b) x1, > 1 and x2, < 0 In this case: Investment (in $): x1C Financing (in $): x2C + C Or: x1,C = x2,C + C c) x1, < 0 and x2, > 0 In this case: Investment (in $): x2C Financing (in $): x1C+C Or: x2C = x1C + C The investments return and financing cost is as follows: i = 1 or 2 The return of investing in asset i is: RF + i(E(Rm) - RF) The financing cost for asset k is: RF + k(E(Rm) - RF) k = 1 or 2
25
Personal equity return: Rd The three situations mentioned above are directly derived from the relationship x1+x2=1, since if x1 > 1 and x2 < 0 x1 = x2+1; or x1 <0 and x2 > 1 x2 = x1 + 1 We can see that the alternative portfolio can be widened for i = 1,... k; so long as the Lagrange function is set out as follows:
k k k k 2 2 L = m x 2 i2 + 2 m x i x j i j + R a x i R i i i =1 i =1 j=1 i =1
We can clarify the above situations with the following example: there are two risky assets in which we may invest or make short sales.
Appendix N 1
Relationship between ij with im and jm For the portfolios function of risk, we must know ij = f(im, jm,), where: ij = Covariance between the returns of asset i and j. km = Coefficient between the returns of asset k and the market portfolio m, k = 1,...j Proof Assuming two assets, i and j, whose returns Rk have the following relationships: Ri = i + imRm Rj = j + jm Rm Where Rm = Return of a market portfolio. For (1) and (2) we arrive at: R j j R i Rm = i and R m = im jm By balancing, we arrive at: Rj j Ri i = im jm By clearing Ri we arrive at: j i im j im R Ri = m + j jm jm Likewise, it is known that a relationship can be established of the following type: R i = o + ,R j Where: i j , = 2 j Of (3) it is known that: , = im jm Or: ij im = 2 jm j By clearing:
(1) (2)
(3)
(4)
26
ij =
i m 2 j jm
Because asset j is in equilibrium, then: j2 = ( jm m )2 From which we can see that: 2 ij = im jm m
Appendix N 2
Calculation of proportions x1 and x2 that minimize the risk of a portfolio with two risky assets We know that: 2 2 c2 = x 1 12 + x 2 2 + 2x 1 x 2 12 2
2 2 Where 1 and 2 , the variance of the returns of assets 1 and 2, respectively, 12 = Covariance of
the returns of assets 1 and 2 We know that: (from x1 + x2 = 1) x2 = x1 -1 2 2 (When in equilibrium, according to CAPM) 1 = (1m) 22 = (2m)2 (When in equilibrium, according to CAPM) 2 12 = 12 m (from Appendix 1) Assets 1 and 2 are two risky assets. So, the function to be optimized is: 2 2 c2 = x 1 ( 1 m )2 + (1 x 1 ) 2 ( 2 m )2 + 2 x 1 (1 x 1 ) 1 2 m By using a minimization process, we arrive at: c2 2 2 = 2x 1 ( 1 m ) 2 2(1 x 1 )( 2 m ) 2 + 2(1 x 1 ) 1 2 m 2x 1 1 2 m = 0 x 1 By clearing for x1, the optimum value (x*1) for x1 is: 2 2 2 m x 1 ( 12 2 1 2 + 22 ) = 2 m ( 22 1 2 ) Since x1 + x2 = 1, then the optimum value for x2 is: 1 x* = with 2 1 2 2 1 We have shown that: 2 c2 2 = 2 m ( 1 2 ) 2 > 0 2 x 1 So 2c in (x*1,x*2) is at a minimum.
Appendix N 3
Calculation of an alternative market portfolio, investment proportions, and financing We know that: 2 2 L = x 1 (1 m ) 2 + (1 x1 ) 2 ( 2 m ) 2 + 2x(1 x)1 2 m + [Rm x1 R1 (1 x1 )R 2 ] Where: Rm = market portfolio return, with any Beta = Lagrange multiplier. The assumptions of Appendix No. 2 are maintained.
27
Our idea is to try to find the x*1 and x*2 proportions that conform a portfolio having minimum risk and a return equivalent to the return of a market portfolio Rm. To do so, we calculate the Lagrange multipliers, for which the First Order conditions are the following: L/x 1 = 2x 1 2 ( 2 1 ) 2 + [R 2 R 1 ] 2 2 2 ( 2 1 ) = 0 m m L/ = x1 (R2 R1 ) (R2 Rm ) = 0 To solve this system of equations, we must use the following matrix procedure: L/ = x1 (R2 R1 ) (R2 Rm ) = 0
x1
2
Constant
2 ( 2 1 )
2 m
R2 R1 0 R2 R1 2 2 m ( 2 1 ) 2
(R2 R1 ) 2 2 2 m ( 2 1 ) 2
2 2 ( 2 1 )
2 m
R2 R1
1 0
R2 Rm 2 2 1 2 (R2 R1 ) + (R2 Rm ) 2 1
Based on this last matrix, we can establish the following relationship between x1 and : R R 2 x1 + 2 2 1 2 = 2m(2 1 ) 2 1
(1)
(R R1 )2 (R R1 ) 2 2 = 2 2 + (R2 Rm ) (2) 2 2 1 2 m (2 1 ) by clearing from (2) and substituting it in (1), we arrive at the following x1 value: R Rm * x1 = 2 with R2 R1 R2 R1 By replacing the returns of assets 1 and 2 with those obtained according to CAPM, we arrive at: R F + 2 [Rm R F ] Rm * x1 = RF + 2 [Rm R F ] RF 1 [Rm R F ] By reducing, we are left with the following: 1 * x1 = 2 with 2 1 2 1 Since x*1 + x*2 = 1, then 1 1 x* = with 2 1 2 2 1 The second order conditions imply that: 2F 2 2 L(x1 ) = 2 = 2 M ( 2 1 ) 2 > 0, therefore,it is a conditioned minimum. x1
We took this formulation directly from Appendix No. 3, replacing the x*1 value with the Rm value for Rd in x*1, or: RF + 2 [Rm RF ] Rd , x1 = 2 [Rm R F ] 1 [Rm R F ]
Appendix N 4
28
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) By arranging, we arrive at: 2 (R R F ) 1 x' 1 = d with 1 2 y Rm R F 2 1 (Rm RF ) ( 2 1 ) Since x1 + x2 = 1, then x2 is: 1 (R R F ) 1 , x2 = + d . 2 1 (Rm R F ) ( 2 1 )
References
[1] [2] [3] [4] [5] [6] [7] Black, F. (1972) Capital Market Equilibrium with Restricted Borrowing, Journal of Business 45, July 1972, 444-455. Brennan, Michael J. (1971) Capital Asset Market Equilibrium with Divergent Borrowing and Lending Rates, Journal of Financial and Quantitative Analysis, No. 5, Dec. Pp. 1197-1205. Jarrow, R. (1988), Finance Theory, Prentice-Hall International Editions, NJ. Elton E. and Gruber, M. (2000). Modern Portfolio Theory and Investment Analysis, 5th. Edition, Ch. 9. Kandel, S. (1984) The Likelihood Ratio Test Statistics of Mean- Variance Efficiency without a Riskless Asset. Journal of Financial Economic, 13, pp 575-592. Sharpe, W. (1964) Capital Asset Price: A Theory of Market Equilibrium Under Conditions of Risk, Journal of Finance, 19, pp. 425-442. Sharpe, W. Portfolio Theory and Capital Markets (New York: McGraw-Hill, 1970).
Effects of M&As on the Spanish Savings Banks

Pedro L. Contreras University of Castilla-La Mancha (Spain) E-mail: pedroluiscontreras@yahoo.es Victor Barrios University of Valencia (Spain) E-mail: Victor.E.Barrios@uv.es Romn Ferrer University of Valencia (Spain) E-mail: Roman.Ferrer@uv.es C. Gonzlez University of Valencia (Spain) E-mail: Cristobal.Gonzalez@uv.es Abstract The Spanish banking industry has experienced an unprecedented wave of mergers and acquisitions (M&As) along the past two decades, particularly intense in the savings banks sector over the period 1989-1993. This paper analyzes the effects of M&As on the performance of the Spanish savings banks industry in terms of their market power, risk, and efficiency. With that aim, a statistical analysis is carried out comparing the evolution of a set of ratios widely used in the literature for both merged and non-merged savings banks before and after M&As. The results indicate that, with the only exception of the capital ratio, M&As have not had a significant impact on the financial performance of Spanish savings banks. On the contrary, the changes in the economic and market conditions, affecting to the whole banking industry, seem to be the main driving forces of savings banks performance. This evidence, broadly consistent with the previous literature, provides important implications for public policymakers, bank managers, and bank customers on the primary determinants of the performance of Spanish savings banks in the sense that M&As have less impact on performance than expected a priori. Keywords: M&As, Savings Banks, Market power, Risk, Efficiency. JEL Classification Codes: G21, G34, L10
I. Introduction
Over the past two decades, consolidation of the banking industry has had an extraordinary expansion in the majority of the developed countries, being mergers and acquisitions (M&As hereafter) the principal way in which this process has come about. Concerning the causes of financial consolidation, two main reasons have been generally argued in the literature: the value-maximizing purpose and the non-value maximizing motive. Company value maximization constitutes the primary motivation to undertake a
30
consolidation process. This increase of value has been traditionally justified for different reasons: economies of scale and scope, market power in setting prices, improvement of firm efficiency, and decrease of risk through geographic and product diversification. In contrast, non-maximizing value motives include managerial objectives and the role of government. The boost given to the M&A activity in the banking sector throughout the world, starting in the 1980s in the United States and reaching Europe in the 1990s, can be explained by a number of key factors: technological progress that generates greater economies of scale in producing financial services, improvement in the financial condition of banking firms, increased shareholder pressure for financial performance, globalization of financial and real markets, financial innovation, and deregulation of both geographical and product restrictions. Given the central role played by financial institutions in the economy in general, consolidation in banking industry has attracted substantial attention not only from managers and shareholders, but also from borrowers, depositors, and policy-makers, concerning to aspects such as the possible impact of M&As on the degree of competition and the stability of the banking sector, and the transmission mechanisms of monetary policy. In the case of Spain, and similarly to other European countries, the banking industry has experienced an unprecedented wave of M&As along last years, particularly intense in the savings banks sector over the period 1989-1993. The Spanish banking system is analogous to those in the main continental European countries: it is a bank-oriented financial system, characterized by the prevalence of a universal banking model. The study of bank M&As for the Spanish case is especially interesting for several reasons. First, a drastic deregulation and liberalization process affecting the operating environment faced by Spanish financial institutions has been occurring since the 1980s. As a result, the competitive viability of Spanish banking firms has been conspicuously stimulated in a framework where the barriers to entry in the European financial markets were falling down due to the European integration. One of the firms responses to this new scenario was to initiate a wave of M&As that changed the structure of the Spanish banking industry in few years. A second reason that may have favored the proliferation of M&As is the tax reduction benefit for the banking firms resulting from the revaluation of hidden reserves at a lower tax cost. A final reason indicates the need to extend geographically the academic literature about this topic. Since the majority of studies have mainly focused on the US market, it seems reasonable analyzing the impact of M&As on a market where the deregulation processes and the institutional framework of the banking system is different from the US one. This paper carries out a statistical analysis of the effects of the consolidation process among Spanish savings banks by using a panel data set with those firms. Specifically, it compares the evolution of a set of financial indicators that represent several aspects of the performance such as power market, risk, and efficiency of savings banks before and after M&As with the corresponding ones of other firms that have not been engaged in M&As. Hence, it permits evaluating whether M&As have given Spanish merged savings banks the opportunity to increase their market power in setting prices, to decrease their level of risk, and to obtain significant gains in efficiency. The remainder of the paper is structured as follows. Section 2 contains a review of the literature about this topic. Section 3 describes the methodology and data used. Section 4 discusses the major empirical results obtained. Finally, section 5 provides some concluding comments.
II. Review of the Literature

The accelerated pace of bank consolidation over last decades has given rise to an extensive amount of academic literature about this topic. A comprehensive review of studies evaluating M&As in the financial industry can be found in Berger et al. (1999). It can be noted that the consequences on market power, risk, and efficiency of the financial institutions involved in bank consolidation processes represent three of the areas where the researchers have typically focused their attention on, due to their
31
importance for the overall performance of banking firms. The most relevant findings about these issues are detailed below, paying special attention to the ones corresponding to the Spanish case. Market power As Berger et al. (1999) point out, M&As among banking firms that have a significant share of local market may increase substantially their market power. A great market power is potentially harmful for the retail customers in deposits and small business loans, since consolidated firms will be able to raise profits by setting prices less favorable to customers (higher rates on small business loans and, conversely, lower rates on retail deposits). Empirical literature on bank mergers has generally found evidence that increases in market power during the 1980s affected adversely consumer prices in concentrated markets. However, the effect of M&As on the market power of banking firms seems to have substantially declined in the 1990s due to an increase in the degree of contestability of financial services markets and the advances in new technologies (see for example, Berger and Hannan, 1997; Berger et al., 1999; Shull and Hanweck, 2001). Concerning to Spain, there is not evidence of any collusion practices in the banking system since the mid-1980s (see for example, Gual and Vives, 1992; Fuentes and Sastre 1999; Humphrey and Carb, 2000). However, it must be pointed out that the main goal of the majority of these papers is not to determine the specific effects from M&As. More recently, Carb et al. (2003) did not find significant differences in terms of prices and profitability between merged savings banks and the rest of the industry either, concluding that branching deregulation and improved economic conditions are more determinant factors for the performance of banking firms than M&As. Risk Although the results in the theoretical literature are not totally conclusive about the final effect, bank concentration can affect banks risk taking and it can also have a significant influence on the risk of global banking system. For example, using a simple model of bank deposits, Allen and Gale (2000) showed that acquisitions reduce the number of competitors and, consequently, the degree of competition in the deposit market, which eventually would reduce bank risk-taking. Conversely, in a revised version of that paper incorporating the loan market, Boyd and De Nicolo (2005) concluded that concentration increases bank risk-taking. In this model, when competition declines financial intermediaries charge higher loan rates, what is counterbalanced by companies increasing their own risk of failure. The empirical literature concerning the effects of bank consolidation is not either conclusive, as it is remarked by the Group of Ten (2001). The area where consolidation seems most likely to diminish firm risk is the potential for geographic and product diversification gains, although even here the possibilities are quite complex, because merged firms with more diversified portfolios can encourage more risk taking (Boyd and Graham, 1998). Moreover, other risks, such as operating risks and managerial complexities, may also increase. The net impact of consolidation on systemic risk is also uncertain. However, the fact that M&As amplify the size of institutions and reduce their number together with the increasing interdependencies between banks observed over the last decade could lead to financial problems of one institution being transmitted to other firms. Furthermore, consolidation may also impose costs on the financial system by expanding the financial safety net, since it can provide additional protection to the firms considered too big to fail, which are created by this kind of processes. In this sense, De Nicolo and Kwast (2002) showed a significant increase of potential systemic risk among large complex banking organizations during the 1990s. In the case of Spain, the empirical evidence obtained by Fuentes and Sastre (1999), Humphrey and Carb (2000), and Fuentes (2003) suggest that banking firms improved their capital positions after M&As. This increase obeys to a large extent to the disclosure in books of hidden reserves at a low tax cost arising from the revaluation of assets during merger processes. Although this effect is a purely
32
accounting phenomenon, it enables banking firms to widen the possibilities to growth by increasing their available capital and also contributes to improve their financial ratios. Efficiency Bank consolidation may increase efficiency in a number of different ways. First, M&As may allow banking firms to achieve a more profitable scale, scope, or mix of output. Second, consolidation also may improve X-efficiency through a change in the organizational focus or the managerial behavior. Third, efficiency gains also may include improvements in the firms risk-expected return tradeoffs. The changes in efficiency associated with bank M&As have been extensively examined by a growing body of research focused on different markets and time periods, and using a wide range of methodologies. However, the empirical results show that the effect of M&As on bank efficiency is quite ambiguous. In particular, a number of studies have shown little or no evidence of cost efficiency gains from the exploitation of scale or scope economies (Peristiani, 1997; Boyd and Graham, 1998; Rhoades, 1998, and Berger et al., 1999). The difficulties in improving cost efficiency could be related to the obstacles often encountered in reducing a banks labor force after the M&A. In contrast, some papers suggest that bank consolidation could increase profit efficiency since the improvement on diversification permits merged firms to reduce their risk and facilitate a better use of their capital. Specifically, merged institutions tend to undertake a portfolio shift from security investments into activities with higher expected values, such as consumer and business loans (Berger, 1998; and Boyd and Graham, 1998). With respect to Spain, the seminal work of Fanjul and Maravall (1985) about the relationship amongst bank M&As and efficiency was continued by, among others, Raymond (1994), Fuentes and Sastre (1999), Humphrey and Carb (2000), Contreras (2001), and Carb et al. (2003). In general, their results do not show a clear linkage between M&As and significant improvements in the efficiency of banking firms. On the other hand, Cuesta and Orea (2002) conclude that M&As have a positive impact on technical efficiency of savings banks, but only in the long term. The majority of the studies for the Spanish banking industry mentioned above are based in data from the 1980s concerning to savings banks, and their main aim was not to evaluate the specific effects from M&As. For these reasons, it appears appealing to carry out further research in order to identify potential improvements in the performance of Spanish savings banks industry due to M&As.
III. Methodology and data used

Methodology The methodology used to assess the impact of M&As on the performance of Spanish savings banks is based on comparing the evolution over time of a set of financial ratios widely used in the literature and representative of various aspects of bank activity. These ratios are calculated for both the firms involved in a merger process and the non-merged firms. In particular, three groups of ratios have been selected: first, a group of indicators of market power; second, two ratios to assess the risk of the firms, and finally, an indicator of the level of efficiency. All of them are detailed below. Market Power For each savings bank, five profitability ratios tied to the market power have been computed for each period t. R1: Interest income/Loans This ratio provides a proxy of the average rate charged on loans granted for a banking firm. It has been frequently associated in the literature with the market power of merged banks, since loan rates
33
will tend to rise after M&As if banking firms exercise the increased market power associated with larger market concentration. R2: Interest expense/Deposits This ratio measures the average rate paid on deposits for a banking firm. It can also be used as a market power indicator because it is expected that depositors will be able to exert less influence on this ratio when the market power of banking firms increases. R3: R2 R1 This ratio represents the average loan to deposit rate spread or interest margin. It reflects the commission that the bank receives for its role as intermediary between borrowers and depositors, and can also be interpreted as an indicator of the degree of market power. The higher this ratio, the higher the market power of banking firms. R4: Fee income/Average total assets plus off balance sheet instruments It measures the ability of the banking firm to generate non-interest income not only from the balance sheet but also from the off balance sheet. A high level of fee income is indicative of both an extensive provision of banking services and an important income diversification towards nontraditional banking activities. Large universal banks are able to offer such a variety of financial services. In this sense, it seems to be necessary a minimum bank size to provide those services, which would imply banks providing these ones have a certain market power. R5: Market share (in terms of average total assets) This ratio measures the market share in terms of average total assets of each firm with respect to the total assets of savings banks. It attempts to capture the possible external effects of banking consolidation, defined as the reactions of other firms to M&As in their market that could affect the volume of total assets of each banking firm. Risk Two indicators of a banking firm risk exposure have been employed. R6: Security holdings/Total deposits R7: Capital plus Reserves/Total assets The first ratio illustrates the destination of the funds obtained through deposits. It is, therefore, indicative of the sheet balance structure from the asset side of a banking firm. Specifically, it shows whether the asset portfolio of a banking firm is more focused on secure and liquid assets, such as security holdings mainly government securities or on riskier assets, such as loans. The higher this ratio, the lower the risk. On the other hand, the capital ratio, defined as capital plus reserves (including revalorization reserves) over total assets, captures the capital strength of the banking firm and its ability to absorb losses. It can be considered as an indicator of capital adequacy and its evolution is indicative of the solvency of the firm and its capacity of risk-taking. The higher the capital ratio, the lower the probability of bankruptcy. Efficiency The operating efficiency ratio, defined as the non-interest operating expenses over the gross income, has been traditionally employed to measure the cost efficiency of firms. It accounts for all of the operating expenses such as personnel, back-office operations, and branches that should be directly affected by the cost savings frequently cited as resulting from bank M&As. The lower this ratio, the higher efficiency. The approach to gauge efficiency used in this paper is based on a frontier analysis method, by comparing for each period t how close each firm is to the best practice firm from the banking industry. R8: Operating efficiency ratio minus operating efficiency ratio of the best practice firm Thus, this indicator reflects the distance between the operating efficiency ratio of each banking firm and the ratio corresponding to the best practice firm in the sample, which adopts the lowest value. The higher this indicator, the greater distance to the most efficient firm and, therefore, the lower operating efficiency.
34
The comparative analysis of ratios has been tackled from a twofold perspective. First, by using a graphical approach that displays the evolution of the yearly average values of each ratio for the groups of merged and non-merged savings banks. This approach provides an intuitive way to identify the impact of M&As on the performance of firms. Second, by carrying out a statistical test of significance for the equality of group means. Specifically, a t-test based on the equality of means across independent populations is used to analyze whether the average of each ratio is significantly different for merged and non-merged savings banks along the period of study. Data All the indicators have been calculated annually from the information in the balance sheets and the profit and loss accounts of Spanish savings banks published by the Spanish Confederation of Savings Banks. It must be pointed out that the ratios have been computed after carrying out a homogenization process due to the changes introduced by Circular 4/1991 of the Bank of Spain with respect to the way in which banking firms disclose information. After obtaining all the ratios for each individual firm, a simple average has been calculated for each ratio (except R5) and for each period t, distinguishing between two groups of firms, the merged savings banks and the non-merged ones. In the case of the market share in terms of average total assets of the whole group, R5, it has been calculated simply as the sum of the market share of all the members of that group for each period t. Therefore, a time series of each indicator corresponding to each group of firms is available to be statistically compared. The correct interpretation of the empirical results requires a previous assessment about the distinctive features of savings banks and their role in the banking system. In our case, Spanish savings banks are primarily focused on retail banking, competing in this area with commercial banks. Nevertheless, whereas commercial banks are profit-maximizing firms, savings banks are non-profit institutions, with an unclear ownership structure which is shared among different stakeholders (public authorities, founding entities, depositors, and employees) with preferences different from shareholders value maximization. Furthermore, they cannot issue shares and are forced to invest a percentage of their retained earnings in social programs. The period of study extends from 1986 to 1998, with 20 M&As among savings banks taking place, some of them involving more than two firms. The wave of mergers occurred over 1989-1993 and it was mainly instigated by the process of financial deregulation of the Spanish banking system. In particular, the last major liberalization decision was taken in 1989 when savings banks were allowed to expand themselves beyond their traditional geographic markets. As a consequence of this merger wave, the number of Spanish savings banks was reduced from 78 in 1986 to 52 in 1993. In order to achieve a better understanding of the effect of M&As, three periods have been distinguished: pre-merger (1986-1989), merger wave or transition (1990-1993), and post-merger (1994-1998) periods. With regard to the latter, there is a general agreement that about half of any efficiency gains should be apparent one year after merger, and the bulk of merger-induced changes in financial and operational performance should be realized within three years (Rhoades, 1998). That is why five years have been taken as the duration of the post-merger period in this paper, since it is consistent with the above statement and is similar to the length of the other two periods considered. Therefore, all the indicators cited above have been calculated during the pre-merger, merger wave, and post-merger periods for the merged savings banks. The same ratios have been computed for a peer group composed by the non-merged savings banks in order to compare the performance of the merged firms to the non-merged ones that are similar in terms of size, type, and location. An unbiased evaluation requires a comparison among firms that are homogeneous in their productive specialization. In this sense, the sample considered in this paper has been taken from the classification in terms of productive specialization proposed by Prez et al. (1999), where, using cluster analysis to identify homogeneous groups of banking firms, they assemble practically all the Spanish savings banks into the same group.
35
IV. Empirical Results

Graphs 1 to 8 show the evolution of the average values of each ratio along the sample period for merged and non-merged savings banks. In turn, Table 1 contains the average values of all the ratios for the pre-merger, merger wave, and post-merger periods, and the outcomes of the statistical test of mean difference between merged and non-merged firms for each individual ratio and period. The discussion of all of the results has been structured into the three areas of bank performance already referred in the previous sections of the paper. Market Power Since Spanish savings banks have been traditionally focused on concentrated local markets, their M&As are expected to lead to less favorable prices for their clientele than the corresponding to other banking firms. However, as Graphs 1 to 4 reveal, the effects of bank M&As on the market power of savings banks do not seem to be very important. This finding is confirmed by the results of the mean difference test, since no statistically significant differences between the indicators of market power for merged and non-merged firms have been found. Graph 1 shows a very interesting pattern for the average rates on loans for merged and nonmerged savings banks. Both groups of firms reduced significantly the average rate of their loans, mainly during the post-merger period. Moreover, the differences between merged and non-merged firms in terms of the average of prices charged on loans are practically negligible, especially at the end of the sample period. This implies that the decline in loan rates cannot be primarily attributed to the effect of M&As. On the contrary, this reduction seems to be related to external factors, such as the dramatic decrease of inflation and interest rates in the Spanish economy since the mid 1990s due to the convergence with the rest of the euro zone, and the intensification of bank competition, as a result of the forenamed deregulation process. The results of the statistical test for equal means reported in Table 1 support this evidence since the null hypothesis of equal average retail loan rates between merged and non-merged firms cannot be rejected at the usual levels for any of the three periods considered.
Graph 1: Interest income/Loans of merged (R1M) VS non-merged savings banks (R1)
Graph 2 displays the evolution of the average rates on retail deposits for merged and nonmerged savings banks. Analogously to the previous indicator, the values of the average rates paid on retail deposits are very similar for both groups of firms along the sample period, with a smooth increasing trend up to 1993 and a dramatic downward trend from then, although the average deposit rates are slightly higher for non-merged firms. Again, this graph suggests that the behavior of the deposit rates of the Spanish savings banks is fundamentally driven by changes in economic conditions and not as a consequence of the M&As, result that is in line with Carb et al. (2003). This is corroborated by the statistical test, since the null hypothesis cannot be rejected for any period of the sample.
36

Graph 2: Interest expense/Deposits of merged (R2M) VS non-merged savings banks (R2)
Graph 3 shows the average loan to deposit rate spread for merged and non-merged savings banks. A twice-humped line can be appreciated in the framework of a clear decreasing trend along the period of study. The sharpest reduction of the net interest margin takes place during the merger wave period within an environment characterized by a fierce bank competition, mainly on the bank asset side. As expected from the previous results, the differences in the average loan to deposit rate spread between merged and non-merged firms are very small, although the average spread is generally higher for merged firms. In this line, the statistical test reveals that the means of this ratio for merged and nonmerged savings banks do not differ significantly for any of the periods considered. From these results, it can be stated that the important reduction of the average loan-deposit rate spread can be largely ascribed to changes in the overall economic and market conditions, and not to the impact of M&As. This finding agrees with prior studies (e.g. Humphrey and Carb, 2000; or Carb et al., 2003). Another implication is that borrowers were not harmed by the greater market concentration derived from the wave of mergers among the Spanish savings banks in terms of the interest rates applied.
Graph 3: Average loan to deposit rate spread (R2-R1) for merged (R3M) VS non-merged savings banks (R3)
With regard to the indicator of the fee income to the average total assets plus off-balance sheet instruments, a common feature observed in Graph 4 for both merged and non-merged savings banks is the remarkable increase experienced by the fee income along the period of study. The two steeped lines with a hump over the years 1991-1994 show that the differences between merged and non-merged firms are practically negligible during the first half of the sample period, while the fee income ratio is slightly higher for non-merged savings banks from the end of the merger wave period. As it could be predicted, the test for equal means does not permit to establish any significant differences between
37
merged and non-merged firms with respect to this indicator. According to this, it can be concluded that M&As are not the primary factor behind the increasing importance of non-interest income for Spanish savings banks.
Graph 4: Fee income/Average total assets plus off balance sheet instruments of merged (R4M) VS nonmerged savings banks (R4)
A possible interpretation of these findings is that, in an attempt to somehow compensate the lower net interest margins resulting from stronger competition in the traditional segments of deposits and loans, Spanish savings banks have tended to expand their business to off-balance sheet activities so giving more weight to the non-interest income in their income structure. The variation over time of the market share in terms of average total assets of merged and nonmerged savings banks as a whole is exhibited in Graph 5. It can be noticed that the group of merged firms has been steadily losing market share relative to the non-merged firms along the whole period of study. In fact, while at the end of the pre-merger period the total market share of merged firms was over 55%, at the end of the post-merger period it was under 50%, so having been overcome by the nonmerged firms. This negative impact of M&As on the market share of merged savings banks may be attributed to the fact that in the short-term in a highly competitive environment the rival banking firms can react belligerently in their pricing policies, so achieving an increase in their market share at the expense of consolidated firms.1
Notice that in this case the statistical test has not been used since data do not reflect an average of the market share per company, but the market share of the whole group.
38
Graph 5: Market share (in terms of average total assets) of merged (R5M) VS non-merged savings banks (R5)
In summary, the empirical evidence obtained does not support the existence of a significantly different behavior in the setting of prices on loans and deposits between merged and non-merged Spanish savings banks. Furthermore, it is consistent with previous literature in the Spanish banking industry (e.g., Fuentes and Sastre, 1999; Fuentes, 2003; or Carb et al., 2003). Risk Graphs 6 and 7 illustrate the evolution of the two indicators of risk considered. As Graph 6 displays, the path followed by the ratio security holdings to total deposits is practically identical for both groups of savings banks along most of the sample period. Specifically, and as a consequence of the deregulated operating environment, from the beginning of the 1990s all firms did reduce their security holdings ratio in a similar way. In addition, the remarkable increase of bank competition in a context of general decline of the net interest margin may have also contributed to the progressive substitution of security holdings by loans with higher expected return and risk in their portfolios. In this sense, the real estate boom occurred in Spain along the last decade, and the subsequent spectacular expansion of mortgages in credit portfolios of banking firms clearly favored this substitution process. As predictable, the test of equal means does not detect statistically significant differences between merged and non-merged firms for any of the three periods examined. These results are consistent with Humphrey and Carb (2000).
Graph 6: Security holdings/Total deposits of merged (R6M) VS non-merged savings banks (R6)
39
Graph 7 shows the increasing general trend of the capital ratio for merged and non-merged savings banks, suggesting that the level of capitalization and, consequently, of solvency of Spanish savings banks did improve along the sample period, with the subsequent positive effect on the whole Spanish economy. In this case, however, a different pattern for merged and non-merged savings banks can be appreciated during the pre-merger and merger wave periods. In particular, whereas at the premerger period the non-merged firms held higher capital ratios than the merged ones, during the merger wave period the latter took advantage of the important tax benefit obtained in the consolidation process to raise up their capital ratios to such an extent that they did overtake the values of the ratios of nonmerged firms. Nevertheless, after the merger wave period a convergence process took place so at the end of the sample period the ratios of capital adequacy for both groups of firms were practically equal. This convergence process is quite logical within an environment marked by a recent capital adequacy regulation, in which all firms should hold the regulatory minimum capital ratio plus a possible capital cushion for contingencies.
Graph 7: Capital plus Reserves/Total assets of merged (R7M) VS non-merged savings banks (R7)
Undoubtedly, the most remarkable feature of the capital positions evolution is the substantial increase of the capital ratio of savings banks just after the merger.2 This improvement has its primary origin in the incorporation into reserves at a low tax cost of the capital gains arising from M&As, due to the revaluation of assets which were recorded on the books at historical cost. However, this situation was not consolidated, but rather this ratio did start a smooth decline after the merger wave period. This effect was particularly outstanding in the case of savings banks, a type of mutually owned banking firm which lacks of share capital due to their special legal status, since it did allow them to increase their available own resources and thereby have a larger margin for making new investments (Fuentes and Sastre, 1999). In this case the difference of means between merged and non-merged savings banks is statistically significant at usual levels in the pre-merger and the merger wave periods. From these results, it can be stated that the capital ratio is the indicator most significantly affected by M&As. More specifically, the evidence obtained indicates that the merger wave had a positive impact on the level of capitalization of merged savings banks. This finding coincides with the achieved by Fuentes and Sastre (1999), Humphrey and Carb (2000), and Fuentes (2003).
From 1985 to 1992 reserves of savings banks increased from EUR 3,570 million to EUR 11,289 million.
40 Efficiency
Graph 8 displays the evolution of the distance between the average operating efficiency ratio for the groups of merged and non-merged savings banks and the operating efficiency ratio of the best practice firm. It can be noted that merged savings banks show a greater distance to the best practice firm along most of the period of study, so implying a lower efficiency, although the differences are not, in general, very important. In particular, the greatest divergence between merged and non-merged firms takes place during the merger wave period, in coincidence with the increase of non-recurring costs that typically occur after M&As. Since then, the difference in efficiency between them has tended to be reduced gradually. The test of equal means confirms this finding, indicating that the differences in operating efficiency between merged and non-merged firms are statistically significant only during the merger wave period.
Graph 8: Operating efficiency ratio minus operating efficiency ratio of the best practice firm of merged (R8M) VS non-merged savings banks (R8)
Therefore, it can be concluded that M&As have not had a significant positive effect on the Spanish savings bank industry in terms of operating efficiency. This poor performance can be ascribed to several factors, such as the difficulty to reduce workforce expenses after M&As due to the prevailing limitations in plans for personnel reduction, with huge costs arising from severance payments or earlyretirement plans, and, to a lesser extent, the increase in overheads associated with the process of internal reorganization of merged firms. This finding is in line with the evidence presented in previous papers on the Spanish banking industry (Fuentes and Sastre, 1999; Humphrey and Carb, 2000; and Fuentes, 2003).
41
Table 1:

Average values of ratios and test of equal means
Pre-Merger (1) 0.1056 0.1078 -0.0022 (-0.93) 0.0557 0.0582 -0.0025 (-1.62) 0.0499 0.0496 0.0003 (0.09) 0.0018 0.0019 -0.0001 (-0.82) 0.2112 0.2065 0.0047 (0.56) 0.0462 5.44 -0.0082 (-3.46)** 0.2127 0.2054 0.0073 (0.52) PERIODS Merger wave (2) 0.1072 0.1053 0.0019 (0.22) 0.0679 0.0694 -0.0015 (-0.99) 0.0393 0.0359 0.0034 (0.41) 0.0031 0.0032 -0.0001 (-0.15) 0.2247 0.2184 0.0063(0.18) 0.0587 5.28 0.0059 (2.47)** 0.1908 0.1638 0.0270 (2.10)** Post-Merger (3) 0.0692 0.0685 0.0007 (0.08) 0.0439 0.0450 -0.0011 (-0.16) 0.0253 0.0234 0.0019 (0.38) 0.0044 0.0046 -0.0002 (-0.62) 0.1994 0.1911 0.0083 (0.59) 0.0567 5.42 0.0025 (1.50) 0.1841 0.1723 0.0118 (0.95) Dif. Pre- Post Merger (4) = (3)- (1) -0.0364 (-16.02)** -0.0393 (-24.61)** -0.0118 (-7.38)** -0.0132 (-10.76)** -0.0246 (-10.76)** -0.0262 (-21.89)** 0.0026(13.69)** 0.0027 (18.90)** -0.0118(-0.83) -0.0154 (-1.46) 0.0105 (4.22)** -0.02 (-0.11) -0.0286 (-1.98)* -0.0331 (-3.26)**
Ratios R1 R2 Market power R3 R4 R6 Risk R7 Operating efficiency

Note:
R8
Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b) Merged (a) Non-merged (b) Dif. (c) = (a) (b)
The three first columns of this table show the average value of every ratio in the three periods considered (pre-merger (1), merger wave (2), and post-merger periods (3), respectively) for (a) the merged and (b) non-merged savings banks, and (c) their differences. The fourth column displays the differences between the values of the pre-merger and post-merger periods for every ratio and group of firms. The results of a t-statistic which tests the equality of means across independent populations are shown in parentheses. * and ** denote whether the null hypothesis of equal means is rejected at the 10% and 5% levels, respectively.
Extensions In order to obtain a greater degree of understanding of the differential effect of M&As and changes in economic conditions on the performance of Spanish savings banks, last column of Table 1 displays the results of a test of equal means between the pre-merger and post-merger periods for each ratio and each group of firms. A statistically significant difference between both periods of time simultaneously for each group separately would be interpreted as a new evidence of the prevalence of market conditions in the explanation of the performance of the Spanish banking firms. With regard to the ratios of market power, the null hypothesis of equal means for the pre- and post-merger periods is clearly rejected in all cases. The absence of significant differences between merged and non-merged firms suggests that the evolution of these ratios along the sample period has been basically determined by the changes in the overall economic conditions and not by M&As. With reference to the risk indicators the most interesting finding refers to the capital ratio. In particular, the different outcomes achieved for merged firms (rejection of the null hypothesis) and non-merged firms (no rejection) support the previous conclusion that merged savings banks took advantage of the special tax treatment given to the capital gains derived from M&As, raising significantly their capital positions. Concerning the operating efficiency ratio, the rejection of the null hypothesis for both kinds of firms is indicative of their effort in increasing efficiency via reduction personnel costs and overhead expenses in order to compensate the dramatic downfall of intermediation margins. All of these results are in line with the ones discussed above, confirming the preponderance of the economic conditions over the M&As themselves. Given the relevance of the results obtained for the capital ratio, in order to check their robustness an additional analysis based on the Theil index has been carried out. The Theil index constitutes a widely used instrument in inequality studies, since it allows splitting a population into homogeneous, exhaustive and mutually exclusive groups, in order to estimate for each group which is
42
its contribution to total inequality within the population (Cowell, 2000). One of its main advantages is its capacity to decompose the total inequality into the sum of the differences or inequalities between groups and within groups. In this particular case, the Theil index has been computed for the three periods considered using merged versus non-merged savings banks as a groups. The results achieved (Graph 9) show a downturn of the total inequality of capital ratios for Spanish savings banks between the merger wave and post-merger periods. This process of convergence took place in a framework of an identical capital adequacy regulation for all banking firms. Thus, savings banks whose capital ratio lay below the minimum regulatory ratio woud have to reach this one in order to enforce the rule.
Graph 9: Theil Index for Spanish savings banks
Regarding the inequality between groups a decreasing trend along the three periods is observed. This decline is smooth between the two first periods, but the downturn gets sharper from the wave merger to the post-merger period, where this indicator reaches a value close to zero. A different pattern, however, can be observed in relation to the within group Theil index. In this case, the inequalities within groups increased during the wave merger period due to the differences among the firms immersed in the M&As processes but after that period this inequality decreases as could be expected. This fact leads (Graph 10) that the weight of inequality between groups with respect to total inequality has been steadily decreasing along three periods from 5% to 0.4%. Thus, in the post-merger period practically the whole inequality could be considered inequality within groups. These results are indicative of how differences between capital ratios for merged and non-merged Spanish savings banks have lost importance over time, supporting the previous conclusions about this ratio.
Graph 10: Distribution of the whole inequality
43
V. Conclusions
The Spanish banking industry has undergone profound transformations over the last two decades due to the confluence of a number of factors such as the financial deregulation, the globalization of the economy, the advances in information technology, and the creation of the European monetary and economic union. As a consequence, the consolidation of the banking sector via M&As experienced a strong development, leading to a substantial reduction in the number of firms operating in the banking market. However, this process of concentration was not accompanied with a parallel decline in the level of bank competition. On the contrary, the degree of competition in the Spanish banking system did increase notably, mainly as a result of the financial deregulation and liberalization process. Thus, the margins of traditional banking activities have been squeezed, leading banking firms to expand their business to off-balance sheet activities and to change their income structure giving more weight to the non-interest income side. In this context, the purpose of this paper has been to shed some light on the effects of M&As on several areas of performance (market power, risk, and efficiency) of the Spanish savings banks industry. With that aim, a statistical analysis has been carried out by employing a set of financial indicators widely used in the literature for merged and non-merged savings banks and comparing them before and after the wave of M&As occurred over 1990-1993. The main conclusion of the paper is that M&As have not had, in general lines, a significant impact on the financial performance of Spanish savings banks. On the contrary, the changes in the overall economic and market conditions, affecting to the whole banking industry, seem to be the main driving forces of the performance of savings banks. Concerning to market power, M&As do not seem to have propitiated a greater capacity for Spanish savings banks to set prices on retail loans and deposits out of line with the market. Moreover, the increasing relevance of non-interest income cannot be primarily ascribed to M&As either, but it rather can be seen as a reaction to the progressive narrowing of the net interest margin in traditional banking business within an environment of fierce competition. Additionally, M&As seem to have a negative impact on savings banks market share, indicating that mergers may help to improve the size of firms and to reorganize their productive structure, but they have a cost in terms of clientele. With regard to the level of risk, M&As do not play a prominent role in the reallocation of funds from securities to loans with higher expected return and risk. The clearest effect of M&As on the performance of Spanish savings banks is the significant increase of their capital ratio during the merger processes. This is due basically to the disclosure in books of reserves arising from the revaluation of assets recorded at historical cost which obtain an important tax benefit. Finally, operating efficiency does not improve significantly after M&As; instead merged savings banks tend to present a poorer efficiency than the non-merged ones. This finding is very interesting since cutting duplicative and wasteful costs is one of the primary stated motives argued in many of these M&As. The evidence presented in this paper is consistent with the previous literature concerning the impact of M&As on the Spanish banking industry, characterized by the general absence of significant differences in performance between banking firms resulting from mergers compared with the remaining non-merged firms. In this sense, concerning the economic policies to be implemented, these results would suggest that it is not clear that consolidation processes should be promoted since no distinguishable effects in terms of performance are appreciated between merged and non-merged savings banks. However, the fact that M&As per se do not constitute the primary determinant of the performance of saving banks does not imply that they did not have a certain positive impact in terms of reduction in costs, internal reorganization, or improvements in capital ratios, so contributing to place the merged savings banks in a better position to confront the growing bank competition.
44
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Allen, F. and Gale, D. (2000), Comparing financial systems, MIT Press, Cambridge, MA. Berger, A.N. (1998), The efficiency effects on bank mergers and acquisitions: A preliminary look at the 1990s data, in Amihud, Y. and Miller, G. (Eds.), Bank Mergers and Acquisitions, Kluwer Academic Publishers, Boston, MA, pp. 79-111. Berger, A.N., Demsetz, R. and Strahan, P. (1999), The consolidation of the financial services industry: causes, consequences, and implications for the future, Journal of Banking and Finance, Vol. 23 No. 2-4, pp.135-194. Berger, A.N. and Hannan, T.H. (1997), Using measures of firm efficiency to distinguish among alternative explanations of the structure-performance relationship Managerial Finance, Vol. 23 No. 1, pp. 6-31. Boyd, J.H. and Graham, S.L. (1998), Consolidation in U.S. banking: Implications for efficiency and risk, in Amihud, Y. and Miller, G. (Eds.), Bank Mergers and Acquisitions, Kluwer Academic Publishers, Boston, MA, pp. 113-135. Boyd, J.H. and De Nicolo, G. (2005), The theory of bank risk-taking and competition revisited, Journal of Finance, Vol. 60 No. 3, pp. 1329-1343. Carb, S., Humphrey, D.B. and Rodrguez, F. (2003), Bank deregulation is better than mergers, Journal of International Financial Markets, Institutions, and Money, Vol.13 No.5, pp. 429-449. Confederacin Espaola de Cajas de Ahorro (1986-2004), Anuario estadstico de las Cajas de Ahorro Confederadas (CECA), Confederacin Espaola de Cajas de Ahorro, Madrid. Contreras, P.L. (2001), La concentracin del sector bancario espaol (1988-1992): un estudio dinmico, UNED, Ph.D. Dissertation. Cowell, F.A. (2000), Measurement of inequality, in Atkinson, A. B. and Bourguignon, F. (Eds.), Handbook of Income Distribution, North-Holland, Amsterdam, pp. 87-166. Cuesta, R.A. and Orea, L. (2002), Mergers and technical efficiency in Spanish savings banks: A stochastic distance functions approach, Journal of Banking and Finance, Vol. 26 No. 12, pp. 2231-2247. De Nicolo, G. and Kwast, M.L. (2002), Systemic risk and financial consolidation: Are they related?, Journal of Banking and Finance, Vol.26 No. 5, pp. 861-880. Fanjul, O. and Maravall, A. (1985), La eficiencia del sistema bancario espaol, Alianza Editorial, Madrid. Fuentes, I. and Sastre, T. (1999), Mergers and acquisitions in the Spanish banking industry: Some empirical evidence, Bank of Spain, Working Paper Series, No. 9924. Fuentes, I. (2003), El crdito bancario en Espaa, Economic Bulletin Bank of Spain, Vol. 12, pp. 31-43. Group of Ten (2001), Consolidation in the Financial Sector, G10, Working Group Report to the Governors of the Group of Ten. Gual, J. and Vives, X. (1992), Ensayos sobre el sector bancario espaol, Fundacin de Estudios de Economa Aplicada (FEDEA), Madrid. Humphrey, D.B. and Carb, S. (2000), Las fusiones de las entidades financieras: costes, beneficios, servicio y precios, Papeles de Economa Espaola, Vol. 84-85, pp. 88-107. Prez, F., Maudos, J. and Pastor, J.M. (1999), Sector bancario espaol (1985-1997). Cambio estructural y competencia, Caja de Ahorros del Mediterrneo, Alicante. Peristiani, S. (1997), Do mergers improve the X-efficiency and scale efficiency of U.S. banks? Evidence from the 1980s, Journal of Money, Credit, and Banking, Vol. 29 No. 3, pp. 326-337. Raymond, J.L. (1994), Economas de escala y fusiones en el sector de cajas de ahorro, Papeles de Economa Espaola, Vol. 58, pp. 113-122. Rhoades, S.A. (1998), The efficiency effects of bank mergers: An overview of case studies of nine mergers, Journal of Banking and Finance, Vol. 22 No. 3, pp. 273-291.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]
45 [23]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Shull, B. and Hanweck, G. (2001), Bank mergers in a deregulated environment: Promise and peril, Quorum Books, Westport, CT.
The Effect of Model-Selection Uncertainty on Error Bands for Estimated Impulse Response Functions in Vector Autoregressive Models
Islam Azzam American University in Cairo, April 2007, Assistant professor of Finance Department of Management, American University in Cairo 113 Kasr El Aini Street, P.O. Box 2511, Cairo 11511, Egypt E-mail: iazzam@aucegypt.edu Abstract Model selection uncertainty adds to the variability in the coefficient estimates when small samples are used because model-selection criteria perform poorly in small samples. Previous literatures account for model-selection uncertainty to improve inference by endogenizing the lag order selection using bootstrap methods. This paper shows that all bootstrap methods fail in cases that are most common in macroeconomic applications. As the maximum eigenvalue of the vector autoregressive model gets closer to one, the bias of the impulse response estimates increases. As a result, the standard bootstrap resampling produces low interval coverage accuracy while bootstrap subsampling produces zero coverage. A proposed solution for this problem is using the first-order bias correction with bootstrap-t interval for impulse response estimates, which corrects for the first and secondorder bias of these estimators. This dramatically improves the interval coverage accuracy for impulse response estimates. Keywords: Model-selection uncertainty, endogenous lag order, bootstrap, and impulse response functions.
1. Introduction
Here model-selection uncertainty corresponds to the uncertainty about the true lag order of the autoregressive process that should be picked. None of the model-selection criteria performs well with small sample sizes, giving rise to the problem of model-selection uncertainty. This uncertainty increases the bias and variability of the estimated coefficients. Ltkepohl (1993) indicates that selecting a higher lag order than the true lag order causes an increase in the mean square forecast errors of the Vector Autoregressive (VAR), while selecting a lower lag order than the true lag order generates autocorrelated errors. Braun and Mittnik (1993) show that estimates from a VAR whose lag order differs from the true lag order are inconsistent. Using the traditional impulse responses that are sensitive to the order of the variables, they show that the impulse response estimates are sensitive to changes in the lag structures of their autoregressive model. Kilian (1998b) shows that ignoring model selection causes under-coverage for traditional confidence intervals for impulse response estimates of the VAR. Few analyses in the literature incorporate the lag-order uncertainty in calculating the impulse response estimates and their coverage accuracy in autoregressive models. Faraway (1992) suggests
47
using pairwise resampling of the original data without conditioning on the selected lag-order for regression model in inference. He shows that the resulting bootstrap distribution tends to be wider because of the additional uncertainty arising from model selection. Other possible approaches in the literature include sample splitting and jack-knife methods. Hjorth (1994) suggests combining crossvalidation techniques with the model-free bootstrap methods for inference in regression models. However, model-free resampling methods for time series are known to require very large samples for reasonable accuracy. Kilian (1998b) introduces a modified bias-correction bootstrap algorithm that reflects the true extent of sampling uncertainty in the regression estimates. This endogenous lag order bootstrap algorithm recognizes that lag order selection is an integral part of the sampling procedure by re-estimating the lag order in each bootstrap iteration. He suggests that the endogenous lag order bootstrap algorithm should replace the standard bootstrap algorithm in applications. Kilians (1998b) algorithm compares the coverage accuracy and average length of the endogenous lag order interval with the exogenous interval for impulse response estimates. The exogenous lag order interval is the standard interval from conditioning on the estimated lag order as if it were known to be the true lag order. Normally, confidence bands for impulse response estimates are based on Ltkepohls (1990) asymptotic normal approximation, Runkles (1987) nonparametric bootstrap method, or the parametric Monte Carlo integration procedure of Doan (1990). Kilian (1998a) shows that his exogenous biascorrected bootstrap intervals tend to be more accurate than delta method intervals, standard bootstrap intervals, and Monte Carlo integration intervals. In this paper we examine the effect of pushing the stationary VAR process closer to the nonstationary region which is the common case in macroeconomic applications on the interval coverage accuracy and average length for impulse response estimates. The analysis uses Monte Carlo simulation based on an empirical application of Obstfeld-Rogoffs (1999) current account VAR model using US quarterly data for 1960.1 2006.1. Different Monte Carlo designs based on different lag orders for this estimated current account VAR are used. This paper shows that, as the maximum eigenvalue of the VAR model gets closer to one, the bias of the impulse response estimates increases. The interval coverage accuracy is very sensitive to the bias of the impulse response estimates. As a result, nonparametric bootstrap resampling produces low interval coverage accuracy for impulse response estimates while bootstrap subsampling produces zero coverage accuracy. Bootstrap subsampling, which is consistent, produces zero coverage accuracy because subsampling increases the bias of the impulse response estimates more than resampling which is inconsistent. Bootstrap subsampling requires a big sample size, such as 20,000, to produce interval coverage accuracy for impulse response estimates close to the nominal coverage. This big sample size is not practically feasible. This paper proposes a solution to this problem: correcting the first and second-order bias through using the first-order bias and stationarity correction with bootstrap-t interval. This method dramatically improves the interval coverage accuracy for impulse response estimates. However, using bootstrap-t interval is computationally intensive. This paper starts by presenting the methodology, the model, the data, and then the empirical study. The empirical study uses Monte Carlo simulation based on an application of Obstfeld-Rogoffs (1999) current account VAR model. This empirical study consists of two parts. First, we estimate the bias in the impulse response estimates and its effect on the interval coverage accuracy for these estimates. Second, we examine the effect of moving closer to the non-stationary region on the coverage accuracy and average length for impulse response functions using different Monte Carlo designs. The reason we start by examining the effect of the bias in the impulse response estimates on the interval coverage accuracy for these estimates is to provide an explanation for the low coverage accuracy when bootstrap methods including Kilians (1998b) are used.
48
2. Methodology
This paper uses both the standard non-parametric bootstrap and Killians (1998b) endogenous lag order bootstrap algorithm Kilian (1998b) introduces the endogenous lag order bootstrap algorithm which reflects the true extent of sampling uncertainty in the regression estimates. He described his method as follows: Consider a stationary VAR process with a finite lag order 0 :
yt = B1 yt 1 + B2 yt 2 + L + B0 yt 0 + ut
(1)
where u t is an identically distributed disturbance with zero mean and covariance matrix u . The dynamic multipliers are obtained from the coefficient matrices by recursion
i = i j B j
j =1
(i = 1, 2,K) where 0 = I N
and B j = 0 for
j > 0 . Then, i = [ kl ,i ] and
i = [ kl ,i ] = i P where P satisfies PP' = u .
Let = vec( B1 , B2 ,K, B 0 ) and = vech( u ) where vec is the column stacking operator and vech is the column stacking operator that stack the elements on and below the diagonal. Pope (1990) 3 shows that, under some regularity conditions, = E ( ) = b( ) T + O(T 2 ) where denotes the ordinary least squares estimate. kl ,i ( , , ) denotes the estimated response of variable k to a shock in variable l , i periods
ago. has been added to account for the dependence of the impulse response on the estimate of the lag order. The steps for his endogenous lag order bootstrap algorithm are: a. Determine the lag order using a model-selection criterion then fit a VAR ( ) model to the data. Estimate and calculate its first-order bias = b( ) T using the closed form solution proposed by Pope (1990). b. Calculate the modulus of the largest root of the companion matrix associated with , ~ denoted by m( ) . If m( ) 1 , set = without any adjustments. If m( ) < 1 , construct ~ ~ the bias-corrected coefficient estimate = . If m( ) 1 , let 1 = and 1 = 1 , then ~ ~ ~ define i +1 = i i and i +1 = i 0.01 . Set = i after iterating on i = i , i = 1,2,K, ~ ~ until m( i ) < 1 . By changing the grid for , m( ) can be made arbitrary close to unity. This stationarity correction is to prevent the stationary impulse response estimates from going to the non-stationary region. c. Using standard nonparametric resampling techniques, recursively generate bootstrap replications {yt* } of {yt } as follows: ~ ~ ~ yt* = B1 yt*1 + B2 yt*2 + K + B yt* + ut* (2)
For each bootstrap replication {yt* }, re-estimate the lag order by the same model selection criterion used in step (a). Denote this selected lag order by * , then fit VAR( * ) * * * * * to {yt }. Estimate and u and construct the first-order bias estimate = b( ) T . ~ d. Calculate the bias-corrected estimate * from * and * following step (b). Then read off the and 1 percentile interval endpoints of the distribution of the bootstrap impulse ~ response estimates kl ,i ( * , * , ) . I calculate the effective coverage accuracy and average length for endogenous and exogenous intervals for impulse response estimates. It is expected that the endogenous interval will be wider than
49
the exogenous interval. The effective coverage accuracy is the relative frequency at which the confidence interval covers the true impulse response.
3. Model and Data

I use Obstfeld-Rogoffs (1999) current account model. In this model, the current account is a predicator of the future net output. CAt = Bt+1 - Bt is the current account where Bt+1 is the value of the economys net foreign assets at the end of period t. Z = Y G I is the net output which is output less government consumption and investment. Consumers base their forecasts on the current and lagged current account in addition to current and lagged net output. The empirical implementation is based on US quarterly data for 1960.1 2006.1 taken from International Financial Statistics (IFS). The VAR model considered here uses CAt and Zt. Based on US data, this VAR model is stationary. Obstfeld and Rogoff (1999) apply their current account model to Belgium data for the period 1954 1990. Our estimated VAR (1) using the US data is qualitatively similar to their estimated VAR (1) using Belgium data.
4. Empirical Study
4.1. The Effect of the Bias in The Impulse Response Estimates on the Interval Coverage Accuracy for these Estimates 4.1.1. An estimation of the bias in the impulse response estimates The bias in the impulse response estimates reduces the interval coverage accuracy for these estimates in the VAR models. This section estimates the bias in the impulse response estimates. In order to simplify the analysis, we switch off the model selection process by conditioning on the true lag order, which will be one. We use the following estimated current account VAR (1) as the true model, where standard errors are in parentheses and estimated coefficients with asterisk are significant at 90% level of confidence: Z t = 21 .74 * 0.04 Z t 1 0.57 * CAt 1
2.75
CAt = 0.19 0.03 Z t 1 + 0.96 * CAt 1

0 .44 0 .01 0.02
0.08 *
0 .13
(3)
4 .7 u = 0.006
0.006 0.12
Conditioning on the estimated VAR (1) shown in equation (3), we calculate the percentage bias and standard deviation of the impulse response estimates using 4000 Monte Carlo iterations. A histogram for the impulse response estimates of the first forecasting period (h = 1) is also plotted to examine the direction of the bias in these estimates. We use small samples of 30 and 60 and a big sample of 10,000. By searching in the International Financial Statistics (2006), you find that the annual data for many countries in the world starts from 1960 for variables like GDP, government consumption, export, investment and many other low frequency variables. In addition, the quarterly and monthly data for those variables are not available for most of the developing countries. This means macroeconomists face the problem of having small sample sizes when they deal with annual data for both developing and developed countries. They also face the same problem when they deal with quarterly and monthly data for developing countries. Figures 1 and 2 present the percentage bias and standard deviation for impulse response estimates at N = 30 and N = 10,000. At N = 30, Figure 1 shows that the percentage bias in the impulse response estimates is 17% in the first period then this bias increases as the forecasting horizon
50
increases until it exceeds 40%. Also, the standard deviation of the impulse response estimates increases as the forecasting horizon increases indicating a higher uncertainty about the longer forecasting horizons. The histogram in Figure 1 shows that the impulse response estimates are biased to the left as the true impulse response for h = 1 is 3.46. At N = 60, the percentage bias in the impulse response estimates is 10% in the first period then this bias increases as the forecasting horizon increases until it reaches 30%. In Figure 2, the percentage bias over all forecasting period is less than 1.5% with N = 10,000. Also, the standard deviation of the impulse response estimates is very small. From the histogram in Figure 2, the impulse response estimates for the first forecasting period are concentrated on the true impulse response.
Figure 1: The percentage bias and standard deviation of impulse response estimates for N = 30 with nine forecasting periods (h = 1,,9), and a histogram of the impulse response estimates for h =1. (the true impulse response for h = 1 is 3.46)
51
Figure 2: The percentage bias and standard deviation of impulse response estimates for N = 10,000 with nine forecasting periods (h = 1,,9), and a histogram of the impulse response estimates for h =1. (the true impulse response for h = 1 is 3.46)
The bias in the impulse response estimates results in a low interval coverage accuracy with small sample sizes. This interval coverage accuracy is very sensitive to the bias in these estimates. A small increase in the bias of the impulse response estimates has a large negative effect on the interval coverage accuracy for these estimates. 4.1.2. The effect of the bias in the impulse response estimates on the interval coverage accuracy for these estimates under no model selection The bias in the impulse response estimates reduces the interval coverage accuracy based on these estimates. Kilians (1998b) first-order bias and stationarity correction is not effective in eliminating the bias in the impulse response estimates in the case considered here. Using Kilians (1998b) first-order
52
bias and stationarity correction with percentile interval produces very low interval coverage accuracy with small sample sizes. Hall (1992) shows that if the bootstrap is applied to asymptotically pivotal statistics, such as bootstrapping the t-statistic, the rate of convergence of the bootstrap increases to T for one-sided distributions and to T 2 for symmetric distributions. A pivotal statistic has the property that its distribution does not depend on the unknown population distribution. This is the justification for the claim that the bootstrap provides better small sample performance than traditional asymptotic procedures. As mentioned in Horowitz (2001), theory and a fair amount of Monte Carlo evidence and examples show that bootstrapping these pivotal statistics provides improvements equivalent to using second-order Edgeworth expansions in standard asymptotic theory. Accordingly, I use Kilians (1998b) first-order bias and stationarity correction with bootstrap-t interval to correct for the first and second-order bias of the impulse response estimates. I calculate the coverage accuracy for impulse response estimates using percentile, such as in Kilians (1998b) method, and bootstrap-t intervals. The analysis conditions on the true lag order which is VAR (1). Therefore, no model selection is included in this section. In calculating the percentile interval, I use 4000 Monte Carlo iterations with 400 bootstrap iterations. The bootstrap-t interval uses 4000 Monte Carlo iterations with 1000 iterations for the first bootstrap loop and 100 iterations for the second bootstrap loop. Only 100 iterations are used in the second bootstrap loop because this loop only estimates the standard deviation of the impulse response estimates. The percentile and the first layer of bootstrap-t need large number of bootstrap iterations to construct the confidence interval of impulse response estimates. Over the 4000 Monte Carlo iterations, I record how many times the true impulse responses lay inside the bootstrap-t and percentile intervals using a nominal coverage of 90%. Table 1 reports the coverage accuracy for percentile and bootstrap-t intervals for impulse response estimates of the first forecasting period (h = 1) under different sample sizes.
Table 1: Coverage Accuracy for Percentile and Bootstrap-T Intervals for Impulse Response Estimates Conditioning on the True Lag Order (No Model Selection) for H = 1 Under Different Sample Sizes (N). (nominal coverage is 90%)
N 30 60 160 600 Percentile 0.54 0.68 0.84 0.89 Bootstrap-t 0.88 0.90 0.90 0.90
3
Percentile intervals with bias correction, as in Kilians (1998b) method, produce very low interval coverage accuracy for impulse response estimates with small sample sizes. In Table 1, with N = 30 and 90% nominal coverage, the coverage accuracy for percentile and bootstrap-t intervals are 54% and 88%, respectively. Bootstrap-t interval increases the coverage accuracy by 34%. This is because using Kilians (1998b) bias and stationarity correction with bootstrap-t interval corrects for the first and second-order bias of the impulse response estimates. With N = 60, the coverage accuracy for the percentile and bootstrap-t intervals are 68% and 90. With large sample sizes, both the percentile and bootstrap-t intervals give similar interval coverage accuracy as they are asymptotically equivalent. The percentile and bootstrap-t intervals produce interval coverage accuracy equals to the nominal coverage at N = 600. The interval coverage accuracy for h > 1 is similar to that for h = 1. 4.1.3. The effect of the bias in the impulse response estimates on the interval coverage accuracy for these estimates under model selection Now, I examine the effect of the bias in the impulse response estimates on the interval coverage accuracy for these estimates under model selection. I examine the effect of using Kilians (1998b) bias
53
and stationarity correction with bootstrap-t on the interval coverage accuracy for impulse response estimates under bootstrap subsampling and resampling. Over the Monte Carlo iterations, I record how many times the true impulse response lies inside the bootstrap exogenous and endogenous intervals. I have eight candidate VAR () models where = 1,,8. Two types of confidence intervals are analyzed: the exogenous and endogenous confidence intervals. The impulse response estimates from conditioning on an estimated lag order construct the exogenous interval. The impulse response estimates from the lag order estimated in each bootstrap iteration construct the endogenous interval. In constructing the percentile interval, I use 4000 Monte Carlo iterations with 400 bootstrap iterations. The bootstrap-t interval requires double bootstrap which is computationally intensive in this endogenous lag order algorithm. So, the bootstrap-t interval is based on 100 Monte Carlo iterations with 1000 iterations for the first loop of bootstrap and 100 for the second loop. Only 100 Monte Carlo iterations were used because each iteration needs 1.5 hour on a Pentium 4 machine. The analysis examines the effect of correcting the first and second-order bias in the impulse response estimates on the interval coverage accuracy for these estimates using bootstrap subsampling under model selection. With N = 300, I use two different bootstrap subsample sizes to examine the sensitivity of the analysis to changes in the subsample size. In Table 2, using the percentile interval, as = 0.5 and very low coverage with in Kilians (1998b) method, produces zero coverage with M N M = 0.9. Bootstrap subsampling increases the bias of the impulse response estimates. As the N subsample size decreases, the bias of the impulse response estimates increases and it becomes harder to eliminate this bias even if we correct for the first and second-order bias of these estimates. Using the bootstrap-t interval with first-order bias and stationarity correction improves the coverage accuracy. At M = 0.5, the coverage accuracy is above 53% while it was zero with the N percentile interval. At M = 0.9, the bootstrap-t interval performs much better than the percentile N interval. Under bootstrap subsampling, the bootstrap-t interval with first-order bias correction performs very well in correcting the bias of the impulse response estimates. Bootstrap subsampling with percentile interval needs a huge sample, such as 20,000, to reach the nominal coverage which is not feasible in practice. Accordingly, two possible solutions to the problem of producing zero coverage with bootstrap subsampling are (1) to use a huge sample size with the percentile interval which is not feasible, and (2) use bootstrap-t interval which is feasible but computationally difficult. Bootstrap-t interval is sensitive to changes in the subsample size.
Table 2: Coverage Accuracy for Percentile and Bootstrap-T Intervals for Impulse Response Estimates Under Model Selection using Bootstrap Subsampling with N = 300. (Nominal Coverage is 90%)
Subsample = S = 0.5 1 2 3 4 S = 0.9 1 2 3 4
M N
Endog. 0.00 0.00 0.00 0.00 Endog. 0.38 0.57 0.62 0.61
Percentile Exogen. 0.00 0.00 0.00 0.00 Exogen. 0.39 0.59 0.63 0.59 Endog. 0.53 0.55 0.55 0.58 Endog. 0.81 0.86 0.89 0.89
Bootstrap-t Exogen. 0.53 0.49 0.49 0.57 Exogen. 0.81 0.83 0.89 0.89
54
The analysis also examines the effect of correcting the first and second-order bias in the impulse response estimates on the interval coverage accuracy for these estimates using bootstrap resampling under model selection. Table 3 presents the coverage accuracy for the endogenous interval of the impulse response estimates using the percentile and bootstrap-t intervals under bootstrap resampling with N = 60. Table 3 shows that bootstrap-t interval produces over-coverage while percentile interval produces under-coverage. The coverage accuracy for bootstrap-t interval is much closer to the nominal coverage than percentile interval. The coverage accuracy for the percentile interval is lower than the nominal coverage by 21%, while this coverage for bootstrap-t interval is higher than the nominal coverage by 5%. This over-coverage in the bootstrap-t and under-coverage in the percentile intervals are expected to go away as the sample size increases. Bootstrap resampling produces lower bias in the impulse response estimates than bootstrap subsampling because you loss efficiency by using bootstrap subsampling. However, bootstrap resampling is inconsistent.
Table 3: Coverage Accuracy for Percentile and Bootstrap-T Endogenous Intervals for Impulse Response Estimates Under Model Selection Using Bootstrap Resampling With N = 60. (Nominal Coverage is 90%)
Period 1 2 3 4 Percentile 0.69 0.69 0.66 0.64 Bootstrap-t 0.95 0.95 0.95 0.90
4.2 An application of the endogenous lag order bootstrap algorithm on the current account var model under different cases The analysis applies Kilians (1998b) bootstrap bias correction method using Monte Carlo simulation based on empirical applications to the two-dimensional current account VAR model. I have both the AIC and SC as model-selection criteria. The analysis starts with the following current account VAR (4) as the true model: Z t = 1 + B11 Z t 1 + B12 Z t 2 + B13 Z t 3 + B14 Z t 4 + B15 CAt 1 + B16 CAt 2 + B17 CAt 3 + B18 CAt 4 + u1t CAt = 2 + B21 Z t 1 + B22 Z t 2 + B23 Z t 3 + B24 Z t 4 + B25 CAt 1 (4)
+ B26 CAt 2 + B27 CAt 3 + B28 CAt 4 + u 2t CAt = Bt+1 - Bt is the current account where Bt+1 is the value of the economys net foreign assets at the end of period t. Z = Y G I is the net output which is output less government consumption and investment. The estimated current account VAR (4) is as follows, where standard errors are in parentheses and estimated coefficients with asterisk are significant at 90% level of confidence: * Z t = 19.2* 0.06 Z t 1 + 0.14* Z t 2 + 0.16 Z t 3 0.12 Z t 4 + 0.15 CAt 1
3.87 0.08 0.08 0.08 0.08 0.49
1.10 CAt 2 0.21 CAt 3 + 0.73 CAt 4

0.70 0.69 0.51
CAt = 0.15 0.03 Z t 1 0.009 Z t 2 + 0.006 Z t 3 + 0.008 Z t 4 + 0.95 * CAt 1

0.65 0.01 0.01 0.01 0.01 0.08
0.01 CAt 2 + 0.008 CAt 3 + 0.01 CAt 4

0.11 0.11 0.08
(5)
4.3 0.008 u = 0.008 0.12
55
4.2.1. What is the effect of moving closer to the non-stationary region on the interval coverage accuracy and average length for impulse response estimates? In this section, I estimate the coverage accuracy and average length for exogenous and endogenous confidence intervals for impulse response estimates using different lag orders for the true VAR process. Using Monte Carlo simulation based on empirical applications, the analysis examines the effect of having a stationary VAR process close to the non-stationary region on the coverage accuracy and average length for exogenous and endogenous intervals. VAR models with a maximum eigenvalue of 0.9 are common in practical macroeconomics and finance. The Obstfeld-Rogoffs (1999) current account VAR model with US quarterly data for 1960.1 2006.1 has a maximum eigenvalue equals 0.9. This indicates that this two-dimensional VAR model is close to the non-stationary region but it is still stationary. The analysis considers VAR (1), VAR (4), and VAR (8) as true models. Figure 3 plots the coverage accuracy and average interval for the exogenous and endogenous percentile intervals for impulse response estimates using N = 60 where VAR (4) is the true model. The difference between the average length of the endogenous and exogenous intervals represents the model-selection uncertainty. Figure 3 shows that uncertainty about the true impulse response increases as the forecasting horizon increases, while model-selection uncertainty is approximately the same over periods. While the coverage accuracy for the endogenous interval declines over periods, the coverage for the exogenous interval slightly changes. The coverage accuracy for the endogenous interval in almost all periods is less than 65%, while the nominal coverage is 90%. The endogenous interval has higher coverage accuracy than the exogenous one on the expense of having higher average length. Based on different lag orders for the true model, I apply Kilians (1998b) bootstrap bias correction method on the current account VAR example. The coverage accuracy and average length for exogenous and endogenous intervals for shock 22 under different lag orders for the true VAR model using the AIC and SC as model-selection criteria is given in Table 4. It shows the effect of having a near to non-stationary VAR model on the coverage accuracy and average length for the endogenous and exogenous intervals. The results for the first four forecasting periods with N = 60 are reported. In Table 4, the coverage accuracy for the endogenous interval is less than 70% in most cases for the AIC and SC with N = 60. The coverage accuracy for the exogenous interval is less than 60% under different lag orders for the true model in most cases. The lag order of the true model matters in determining the coverage accuracy. Having a higher lag order for the true model resulted in a decline in the interval coverage accuracy for the first forecasting period. For the second and third forecasting periods, higher lag order for the true model resulted in an increase in the coverage accuracy for both criteria on the expense of increasing the average length. For the fourth period, a dramatic decline in the coverage accuracy happened when I use VAR (8) as the true model. The coverage accuracy dropped to 40% using the AIC and 30% using the SC with VAR (8) as the true model. The difference between the average length of the exogenous and endogenous intervals reflects the model-selection uncertainty. Model-selection uncertainty is less with the SC than the AIC because the SC tends to pick lower lag order than the AIC. Changing the lag order for the true model has little effect on the average length of the two intervals. The average length for the endogenous interval increases as we increase the forecasting horizon reflecting the increase in the uncertainty level at higher forecasting periods. The average length for the endogenous interval is higher than that for the exogenous interval by 20% to 40% under different lag orders for the true models. This means the model-selection uncertainty increases the interval length by 20% to 40%. The endogenous and exogenous intervals with the SC have higher coverage accuracy with lower average length than with the AIC. In this example, the SC produces lower model-selection uncertainty under different lag orders for the true model than the AIC. Kilians (1998b) method still generates low coverage accuracy where the coverage accuracy for the endogenous interval is lower by 20% to 30% than the nominal coverage level of 90%.
56
Figure 3: The average length and coverage accuracy for the exogenous and endogenous intervals where var (4) is the true model and n = 60 with 90% nominal coverage(solid line is the endogenous interval, dotted line is the exogenous interval)
57
Table 4:

The Coverage Accuracy and Average Length for a Near Non-Stationary VAR Process Under Different Lag Orders for the True Model using Stationarity and Bias Correction (Nominal Coverage is 90%)
AIC with N = 60 Endog interval Exogen interval Cover. Leng. Cover. Leng. 0.69 2.13 0.59 1.55 0.69 2.61 0.56 1.96 0.66 2.92 0.55 2.29 0.64 3.12 0.56 2.54 Cover. Leng. Cover. Leng. 0.65 2.15 0.55 1.59 0.71 2.62 0.62 1.99 0.67 2.92 0.59 2.31 0.60 3.11 0.53 2.55 Cover. Leng. Cover. Leng. 0.59 2.17 0.50 1.65 0.74 2.65 0.67 2.04 0.72 2.94 0.66 2.35 0.42 3.17 0.30 2.61 SC with N = 60 Endog interval Exogen interval Cover. Leng. Cover. Leng. 0.70 1.91 0.63 1.43 0.66 2.32 0.59 1.86 0.61 2.58 0.59 2.21 0.59 2.76 0.60 2.48 Cover. Leng. Cover. Leng. 0.67 1.93 0.59 1.45 0.76 2.34 0.70 1.87 0.70 2.61 0.68 2.23 0.59 2.79 0.59 2.50 Cover. Leng. Cover. Leng. 0.62 1.97 0.53 1.49 0.83 2.38 0.78 1.91 0.80 2.66 0.79 2.27 0.30 2.86 0.28 2.54
TRUE VAR(1) 1 2 3 4 VAR(4) 1 2 3 4 VAR(8) 1 2 3 4
Kilian (1998a) shows that his exogenous bias-corrected bootstrap intervals tend to be more accurate than asymptotic method intervals, standard bootstrap intervals and Monte Carlo integration intervals. Do exogenous intervals with bias-correction outperform endogenous intervals without biascorrection using the current account example? In other words, is it better to correct for the bias or endogenize the lag order? To answer this question, I endogenize the lag order using the standard bootstrap, which does not include bias-correction, to compare it with Kilians (1998b) exogenous bootstrap bias-corrected interval. Table 5 presents the coverage accuracy and average length for endogenous and exogenous standard bootstrap intervals which does not include bias-correction. It shows that the lag endogenized bootstrap bias-corrected interval outperforms the lag endogenized standard bootstrap interval in coverage accuracy. This means bias-correction improves the interval coverage accuracy. The standard bootstrap produces lower coverage accuracy than the bias-corrected bootstrap interval. Using the standard bootstrap, the coverage accuracy for the endogenous interval is from 24% to 60% with the AIC, and 11% to 60% with the SC under different lag orders for the true model. With the standard bootstrap, the coverage accuracy for the exogenous interval does not exceed 45% for most cases. The coverage accuracy for the exogenous interval using the standard nonparametric bootstrap resampling is very low in this example. This example shows that bootstrap methods perform poorly when the stationary VAR process is near the non-stationary region because of the increase in the bias of the impulse response estimates. For the considered current account example, endogenzing the lag order and correcting for the bias of the impulse response estimates improve the interval coverage accuracy for these estimates.
58
Table 5:

The Coverage Accuracy and Average Length for a Near Non-Stationary VAR Process Under Different Lag Orders for the True Model using Standard Bootstrap with N = 60 (Nominal Coverage is 90%)
AIC SC Exogen interval Cover. Leng. 0.44 1.52 0.32 1.87 0.25 2.09 0.21 2.22 Cover. Leng. 0.42 1.57 0.39 1.91 0.30 2.12 0.20 2.24 Cover. Leng. 0.39 1.63 0.47 1.96 0.41 2.18 0.14 2.33 Endog interval Cover. Leng. 0.58 1.88 0.44 2.22 0.33 2.41 0.27 2.51 Cover. Leng. 0.55 1.90 0.57 2.25 0.43 2.44 0.27 2.55 Cover. Leng. 0.51 1.94 0.67 2.30 0.58 2.51 0.11 2.63 Exogen interval Cover. Leng. 0.43 1.41 0.27 1.76 0.20 2.00 0.17 2.12 Cover. Leng. 0.40 1.43 0.39 1.78 0.29 2.02 0.16 2.15 Cover. Leng. 0.37 1.46 0.51 1.82 0.43 2.07 0.06 2.23 Endog interval Cover. Leng. 0.62 2.10 0.55 2.53 0.45 2.76 0.39 2.89 Cover. Leng. 0.58 2.12 0.60 2.54 0.49 2.77 0.38 2.90 Cover. Leng. 0.53 2.15 0.64 2.58 0.59 2.81 0.24 2.97
TRUE VAR(1) 1 2 3 4 VAR(4) 1 2 3 4 VAR(8) 1 2 3 4
However, the exogenous interval under bias correction has higher coverage accuracy than the endogenous interval under no bias correction, although the first interval has a smaller average length than the second one. In this example, it is better to use bias correction with exogenous interval than endogenizing the lag order without bias correction. Bias correction is more effective than endogenzing the lag order in improving the coverage accuracy in this example. Using both the bias correction and endogenizing the lag order improves the coverage accuracy, but at the expense of larger increase in the average length. On top of that, this procedure has a much greater computational expense. 4.2.2. What is the effect of using bootstrap subsampling with the endogenous lag order bootstrap algorithm when the var process is near the non-stationary region? Kilians (1998b) method involves a nonparametric resampling technique which is not consistent. Under resampling, the bootstrap fails due to violations of the continuity conditions. Reducing the bootstrap sample size (subsampling) is a general technique for remedying this bootstrap failure. A key requirement for subsampling to work is that the ratio of the bootstrap subsample size to the original sample size goes to zero as the original sample size goes to infinity. The analysis also examines the performance of Kilians (1998b) method under bootstrap subsampling instead of resampling when the stationary VAR model has a maximum eigenvalue close to one. Table 6 shows that using subsampling with Kilians (1998b) method when the VAR process is near the non-stationary region is a bad idea. At N = 1000, subsampling produces a zero coverage accuracy while resampling produces coverage accuracy equals to the nominal coverage for both the exogenous and endogenous intervals using the AIC. With resampling at N = 1000, the average length for exogenous interval equals the average length for endogenous interval. This indicates that the model-selection uncertainty disappears for large sample sizes and we do not need to endogenize the lag order.
59
Table 6:

The Effect of Subsampling with a Near Non-Stationary VAR Process using the Stationarity and Bias Correction Method Where Var(4) is the True Model with N = 1000 (Nominal Coverage is 90%)
AIC Endog. interval Cover. Leng. 0.00 0.51 0.00 0.65 0.00 0.75 0.00 0.82 Cover. Leng. 0.00 0.47 0.00 0.60 0.00 0.70 0.00 0.73 Cover. Leng. 0.89 0.46 0.90 0.59 0.90 0.69 0.91 0.72 Exogen. interval Cover. Leng. 0.00 0.51 0.00 0.65 0.00 0.76 0.00 0.79 Cover. Leng. 0.00 0.47 0.00 0.60 0.00 0.70 0.00 0.71 Cover. Leng. 0.89 0.46 0.90 0.59 0.90 0.69 0.90 0.70 Endog. interval Cover. Leng. 0.00 0.45 0.00 0.56 0.00 0.63 0.00 0.68 Cover. Leng. 0.00 0.40 0.00 0.53 0.00 0.57 0.00 0.58 Cover. Leng. 0.90 0.44 0.85 0.57 0.87 0.63 0.90 0.64 SC Exogen. interval Cover. Leng. 0.00 0.49 0.00 0.62 0.00 0.71 0.00 0.74 Cover. Leng. 0.00 0.45 0.00 0.56 0.00 0.64 0.00 0.64 Cover. Leng. 0.88 0.44 0.83 0.55 0.85 0.63 0.87 0.63
Subsamp MN s = 0.1 1 2 3 4 s = 0.5 1 2 3 4 s=1 1 2 3 4
This zero coverage accuracy with bootstrap subsampling is due to the increase in the bias of the impulse response estimates under subsampling. The current account VAR example is stationary but its maximum eigenvalue equals 0.9. Using three different Monte Carlo designs, the analysis shows that as the stationary process moves closer to the non-stationary region, the bootstrap procedures become less reliable and we should be very careful when we make inferences based on these VAR models. Even with Kilians (1998b) procedure where we correct for the bias in the impulse response estimates and make sure the VAR process is in the stationary region, the bootstrap is still not reliable. This result is important for macroeconomists as the VAR models with a maximum eigenvalue not too far from unity are very common in applied macroeconomics and finance. In this case, we should not try to use bootstrap subsampling as it produces zero coverage accuracy. In addition, when we use bootstrap resampling, which is inconsistent, we still get low coverage accuracy. My comparison involves changing the value of B25 in equation 2.4 from 0.95 to 0.1 which results in a maximum eigenvalue of 0.6. This means I pushed the VAR process far away from the nonstationary region. The results of changing the value of B25 are reported in Table 2.7 where I use a sample of size 160 and the estimated VAR (4) in equation 5 as my true model after changing the value of B25 to 0.1. Table 7 indicates that, with bootstrap resampling, pushing the process away from the nonstationary region results in over-coverage for the endogenous interval under both model-selection criteria. The SC has lower interval average length than the AIC. This means the SC also outperforms the AIC when the process is far away from the non-stationary region in this example. By comparing Table 4 and Table 7, we find that when the VAR process moves away from the non-stationary region, the bootstrap resampling has much smaller average length with much higher coverage accuracy.
60
Table 7:

The Effect of Pushing the VAR Process Away from the Non-Stationary Region (B25 = 0.1) using Bootstrap Resampling with Bias-Correction where VAR(4) is the True Model (Nominal Coverage is 90%)
AIC with N = 60 Endog interval Exogen interval Cover. Leng. Cover. Leng. 0.94 1.79 0.90 1.57 0.97 1.74 0.81 1.10 0.97 1.65 0.73 0.79 0.98 1.57 0.60 0.62 SC with N = 60 Endog interval Exogen interval Cover. Leng. Cover. Leng. 0.94 1.70 0.92 1.60 0.99 1.50 0.76 0.70 1.00 1.20 0.55 0.30 1.00 1.00 0.30 0.20
Period 1 2 3 4
Table 6 shows that subsampling produces zero coverage when B25 equals to 0.95. In Table 8, when B25 equals to 0.1, subsampling produces over-coverage using both the AIC and SC. This implies that with a VAR model far away from the non-stationary region, bootstrap subsampling and resampling produce over-coverage. As can be seen in Table 7 and 8, bootstrap resampling produces smaller interval average length than bootstrap subsampling. In addition, as mentioned before, bootstrap subsampling produces a higher bias in the impulse response estimates than the bootstrap resampling. In other words, bootstrap resampling, which is inconsistent, outperforms bootstrap subsampling.
Table 8: The Effect of Pushing the VAR Process Away from the Non-Stationary Region (B25 = 0.1) using Bootstrap Subsampling with Bias-Correction where VAR (4) is the True Model (Nominal Coverage is 90%)
AIC with N = 160 22 Endog interval Exogen interval Cover. Leng. Cover. Leng. 0.99 2.47 0.96 1.70 1.00 2.45 0.96 1.22 1.00 2.51 0.82 0.90 1.00 2.63 0.70 0.74 SC with N = 160 22 Endog interval Exogen interval Cover. Leng. Cover. Leng. 1.00 2.46 0.97 1.64 1.00 2.44 0.95 0.82 1.00 2.48 0.69 0.43 1.00 2.61 0.44 0.25
Subsamp. M/N s = 0.60 1 2 3 4
These results indicates that, as the maximum eigenvalue of the VAR model gets closer to one, it is not a good idea to use bootstrap to construct confidence intervals for impulse response estimates even with corrections for the first-order bias and stationarity. We need to correct for the first and second-order bias of the estimated impulse responses by using bootstrap-t interval with first-order bias correction. However, as mentioned before, this bootstrap-t interval is computationally intensive. Unfortunately, it is a common case in macroeconomics to have VAR process with a maximum root not far from one. Besides, bootstrap is the most commonly used alternative to construct confidence interval for impulse response estimates from the classical point of view. Overall, the results based on the current account VAR example indicate the superiority of bootstrap resampling over bootstrap subsampling because bootstrap subsampling increases the bias of the impulse response estimates. However, bootstrap resampling is inconsistent. I showed in the first chapter of this dissertation that bootstrap subsampling outperforms bootstrap resampling when the Shaos (1996) model-selection criterion was used. This is because the sensitivity of the Shaos (1996) criterion to the bias in the estimated coefficients is much less than the sensitivity of Kilians (1998b) interval coverage accuracy to the bias in the impulse response estimates.
Conclusions
Modifying Kilians (1998b) method, I apply nonparametric bootstrap subsampling using the SC as model selection criterion with stationary VAR processes having high maximum eigenvalue. I found
61
that the interval coverage accuracy for impulse response estimates for such VAR processes is low when one uses bootstrap resampling and goes to zero when one uses bootstrap subsampling. Thus, when the stationary VAR process has a maximum root that is close to unity which is not uncommon in most macroeconomic applications, Kilians (1998b) method performs poorly. The problem here is that as the maximum eigenvalue for the VAR process increases, the bias in the impulse response estimates increases. My results suggest using a first-order bias correction with bootstrap-t interval to correct for the first and second-order bias in the impulse response estimates. This dramatically improves the interval coverage accuracy for impulse response estimates under bootstrap subsampling and resampling. On the basis of different Monte Carlo designs, I am able to establish the robustness of my results.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] Andrews, D. W. K. (1997). A simple counterexample to the bootstrap. Cowles foundation Discussion Chapter No. 1157, Yale University, August, 1997. Braun, P. and Mittnik, S. (1993). Misspecifications in vector autoregressions and their effects on impulse responses and variance decompositions. Journal of Econometrics, 59, 319-341. Burnham, K. and Anderson, D. (2002). Model-selection and Multimodel Inference: A Practical Information-Theoretical Approach. Springer-Verlag, New York. Faraway, J. F. (1992). On the cost of data analysis. J. Comput. Graph. Stat. 1 (3), 213-29. Freedman, D. A., Navidi, W., and Peters, S. C. (1988). On the impact of variable selection in fitting regression equations. In T. K. Dijkstra, ed., On Model Uncertainty and its Statistical Implications, Springer-Verlag, New York. Goodfriend, M. and King, R. (1997). The new neoclassical synthesis and the role of monetary policy. In: Bernanke, B., Rotemberg, J. (Eds.), Macroeconomics Annual 1997. MIT Press, Cambridge, pp. 231-282. Hall, P. (1992). The Bootstrap and Edgeworth Expansion. Springer Verlag, New York. Hastings, W. K. (1970). Monte Carlo sampling using Markov chains and their applications. Biometrika 57, 97-109. Hjorth, J. S. U. (1994). Computer Intensive Statistical Methods: Valuation, Model-selection, and Bootstrap. London: Chapman and Hall. Horowitz, J. (2001). "The Bootstrap." Handbook of Econometrics, Vol. 5, J.J. Heckman and E.E. Leamer, eds., Elsevier Science B.V., Ch. 52, pp. 3159-3228. International Financial Statistics (2006). Data CD, IMF. Kilian, L. (1998a). Small-sample confidence intervals for impulse response functions. The Review of Economics and Statistics. 80 (2), 218-229. Kilian, L. (1998b). Accounting for lag order uncertainty in autoregressions: The endogenous lag order bootstrap algorithm. Journal of Time Series Analysis. 19 (5), 531-547. Ltkepohl, H. (1990). Asymptotic distribution of impulse response functions and forecast error variance decompositions of vector autoregressive models. The review of Economics and Statistics. 72, 53-78. Ltkepohl, H. (1993). Introduction to Multiple Time Series Analysis. Berlin, Heidelberg: Springer-Verlag. McFadden, D. (1997). A note on selecting variables to maximize their significance. Department of Economics, University of California, Berkeley, February 1997. Obstfeld, M. and Rogoff, K. (1999). Foundation of International Macroeconomics. Cambridge, Massachusetts: MIT press. Paulsen, J. and Tjstheim, D. (1985). On the estimation of residual variance and order in autoregressive time series. Journal of the Royal Statistical Society. B 47, 216-228.
62 [19] [20] [21] [22] [23] [24] [25]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Pesaran, H. and Shin, Y. (1998). Generalized impulse response analysis in linear multivariate models. Economics Letters, 58, 17-29. Politis, D. and Joseph P. R. (1994). The stationary bootstrap. Journal of the American Statistical Association 89(428): 1303 1313. Pope, A. L. (1990). Biases of estimators in multivariate non-Guassian autoregressions. Journal of Time Series Analysis. 11, 249-258. Runkle, D. (1987). Vector autoregressions and reality. Journal of Business and Economic Statistics. 5, 437-442. Shao, J. (1996). Bootstrap model-selection. Journal of the American Statistical Association, June 1996, 91 (434), 655-665. Sims, C. and Zha, T. (1999). Error bands for impulse responses. Econometrica. 67 (5), 11131155. White, H. (2000). A reality check for data snooping. Econometrica, September 2000, 68 (5), 1097-1126.
Comparative Analysis of Socio-Economic Constraints in Niger-Delta

Reuben Adeolu Alabi Research Fellow, Institute of World Economics and International Management Faculty of Economics, University of Bremen, Germany E-mail: bayobimb@yahoo.com Abstract This study empirically investigates the socio-economic conditions of Niger-Delta and makes policy recommendations. This study relied on the most recent 2005 and 2006 data from National Bureau of Statistics in Nigeria. Mean, percentage and probability distributions were employed to analyse the relevant data. The results of analysis show that poverty level, unemployment, cost of health, cost of education, inaccessibility to social infrastructure are higher in Niger Delta than the national average, coupled with increased cost of living. The crises and conflicts in Niger Delta are predicated on these socio economic constraints that are rampart in the region. The study also indicates that the enormous financial resource that got to the region has not significantly improved health, education and infrastructural facilities in the region. Therefore, all the stakeholders in Niger-Delta have a role to play to improve socio-economic situation in the region so that the imminent danger can be averted. The role of these stakeholders were specified in the paper. This paper concludes that the challenge of socio-economic development of Niger Delta is complex but not insurmountable. Government owes a moral obligation to the Niger Delta people to create an enabling environment for peace and provide sustainable development in the region. Keywords: Socio-economic, Niger-Delta, Policy, Implication, Intervention
1. Introduction
Nigeria is the most oil-dependent economy in the world. According to the 2008 budget, oil sector is expected to contribute N3.63 trillion or 80% of total government revenue for the year (Arowolo, 2008). The non-oil sectors, comprising manufacturing, agriculture, solid minerals, etc, will contribute N910 billion or 20 %. The Niger Delta region is the bedrock of Nigerias crude oil production. The area accounts for over 90% of the nations export earning and up to 80% of revenue accruing to the Federation Account mainly from oil exploration and production in Niger Delta (ECA, 2005). The oil and gas reserves in the region are estimated at 25 billion barrels and 130 billion cubic feet respectively. Crude oil production stands at 2.45 million barrels per day. The region contains 5284 oil wells, 10 gas plants, 275 flow stations, 10 export terminals (Aigbokhan et al, 2007). Niger-Delta is crisis prone. It is inconceivable that any lasting peace on this planet can be achieved without resolving the complex problems of social and economic development that afflict contemporary societies (Afolabi et al, 2003). By the same token, it is a general knowledge that Nigeria cannot enjoy tranquillity and peace if the socio-economic problems of Niger Delta are not solved. These crises have socio-economic and diplomatic consequences, not only in Niger-Delta, but also in
64
Nigeria in general. For example, when Ateke Tom-led group, Niger-Delta Vigilante, invaded Porthartcourt on the 1st of January, 2008, 14 peole were killed, including policemen. This singular event reverberated across globe, pushing crude oil price at world market at a new record of $100 per barrel (Arowolo, 2008). In 2006, Shell Petroleum Development Company (SPDC) disclosed that it lost about $ 800 million from frequent shut down of its operations. Chevron, Elf and other oil firms have equally been licking their wounds, following disruptions in their production operations (Aigbokhan, 2007). It is estimated that attacks on oil facilities and the kidnapping of oil workers have slowed down oil production in Nigeria, costing multinational companies and the government billions of dollars in revenue. While an estimated 30 billion US dollars (over N 300billion) was lost to communal unrest and theft of oil in the region between 1999 and 2003, a fifth of the 2.45 million barrels exported per day is now being stolen daily. It is estimated that the country loses about $6.8m on every violence-induced production deferment, and another $6.8bn spent annually to replace vandalised equipment in the oil communities (Punch, 2007). The kidnapping of oil workers and expatriate is a popular knowledge in this region (Ifidon, 2006; Oil Revenue Watch, 2006). These crises and conflicts has potential to cripple Nigerian economy in no distant time, if urgent steps are not taken to remedy the situation in Niger-Delta. According to CSAE (2005), one year of conflict reduces a countrys economic growth rate by 2.2%. Conflict also has severe effect on human health. An average conflict causes an estimated 0.5 million DALYs each year (Disability Affected Life Years) a measure of the total number of people affected and the period for which their disability lasts). If each DALY is valued at $1000 (roughly the per capita income in many at risk countries), the economic cost of harm to human health in a typical conflict is around $5 billion (CSAE, 2005). Therefore, efforts need to be made to improve socio-economic conditions in Niger-Delta and thereby reduce crises and conflicts in the region. However, since past efforts have failed, there is need for accurate and analytical information on socio-economic conditions of Niger-Delta so that the government can formulate policies based on these analyses. This will help the government to make informed decision so that the recent plans to develop the region will not be an expensive failure as the past projects and plan. This study therefore, analyses information on socio-economic conditions in Niger-Delta and makes recommendations based these analyses on how to influence these socio-economic conditions in order to develop Niger-Delta. The rest of the paper is divided into six sections. Section 2 reviews past developmental efforts in Niger-Delta, section 3 deals with unemployment situation in Niger-Delta with average for Nigeria. Section 4 assesses health conditions in Niger-Delta, section 5 examines the education and schooling in Niger-Delta, section six presents poverty and cost of living in Niger-Delta, section 7 focuses on acessibility to infrastructure in Niger-Delta, while section 8 makes policy recommendation and concludes the paper
2. Review of Developmental Efforts in Niger-Delta

Previous efforts have been made to develop Niger-Delta, though, they failed. The first constitutional step taken to develop the Niger Delta Region was the appointment of Henry Willink-led Minority Rights Commission in September 1957, Henry Willink Commission then recommended a Federal Board to consider the problems of the Niger Delta. Based on the recommendations of Henry Willink Commission, subsequently the Niger Delta Development Board was created by the 1960 Constitution. Section 14 of the 1960 Constitution established the Niger Delta Development Board to provide physical development for the Niger Delta Region. The section enjoins that the Niger Delta Development Board shall be responsible for advising the government of the Federation of Nigeria and the government of Western Nigeria and Eastern Nigeria with respect to the physical development of the Niger Delta. In 1961, Nigerias federal parliament enacted the Niger Delta Development Board Act of 1961 to comply with Section 14 of the 1960 Constitution. The Board, however, could not provide any
65
meaningful development for the Niger Delta Region because of lack of political will and commitment. The Federal Military Government then nationalized every aspect of the national economy and life in Nigeria. The autonomy of the region was consequently taken away by military decrees. For example, Decree No 9 of 1971 repealed Section 140(6) of the 1963 Constitution which ceded oil revenue from the continental shelf to the region.The Commission existed for seven years, but by the time it went to oblivion in 1967, it had insignificant achievement to show. Following growing agitations for a special focus on the development of the region, the President Shehu Shagaris administration set up a task force in 1980, known as 1.5 percent committee to see to the developmental peculiarities of the region; 1.5 percent of federation account was allocated to the committee to carry out its mandate (Olukorede, 2007). The Federal Military Government of General Ibrahim Babangida promulgated Oil Mineral Producing Area Development Commission (OMPADEC) Decree No 23 of 1992 to address the years of neglect of the Niger Delta Region, Section 2 of the decree states that OMPADEC is to receive and administer the monthly sums from the allocation of the Federation Account in accordance with confirmed ratio of oil production in each state for the rehabilitation and socio-economic development of Niger-Delta areas. Section 4a(2) of the Allocation of Revenue (Federation Account) (Amendment Act No 106 of 1992) provides that 3 percent of the federation account derived from mineral revenue be paid into a fund to be administered by OMPADEC. Once again, OMPADEC also failed to ensure the development of the Niger Delta Region. Abandoned, uncompleted white elephant projects of OMPADEC are now common features in the region (Aigbokhan, 2007). Government officials and their cronies appointed into OMPADEC looted the funds set aside for the regions development. For instance, the implementation of the objectives of OMPADEC was stalled and defeated by large-scale fraud, corruption and fund diversions by some of the chairman and staff of the commission. Failure to implement the law destroyed its genuine intention while the commission became a conduit pipe through which large sums of money belonging to the Niger Delta and its people were misappropriated (Anayochukwu, 2007). The Nsiakak Eduok committee found that out of 2239 project by OMPADEC, 76 had N3.91billion ($39 million) paid for without job done (Joab- Peterson, 2004). At the time the commission was dissolved, it was owned over N100 billion ($1.2 billion) in short fall and miscellaneous unpaid entitlements (Aigbokhan, 2007). The failure of this intervention agency again fuelled more agitation and violence, which reached its peak in 1998 when youth disrupt oil production activities more frequently (Joab-Peterside, 2004). Given these scenarios, OMPADEC was scrapped by President Olusegun Obasanjo, and was replaced with Niger- Delta Development commission (NDDC). In response to local demands for greater resource ownership and benefits, and in an attempt to defuse the demands for resource control, President Obasanjo established Niger Delta Development Commission (NDDC). The commission is charged with a wide range of tasks, in particular, to: conceive, plan and implement, in accordance with set rules and regulations, projects and programmes for the sustainable development of the Niger Delta area in the areas of transportation, roads, jetties and waterways, health, education, employment, industrialisation, agriculture and fisheries, housing, urban development, water supply, electricity and telecommunications. The commission, according to the Act establishing it, is funded by a combination of contributions from the federal and state governments and oil companies as follows: 15 percent of the allocations due to the member states of the commission under the derivation principles of revenue allocation; 3 percent of the total annual budget of any oil company operating in the Niger Delta; 50 percent of funds due to the member states from the ecological fund. NDDC has recorded some progress in infrastructural development. Three hundred and fiftyeight projects have been completed across the region. These include forty major roads, forty-one landing jetties and ninety three water schemes. Also, eighty-one electricity projects and one hundred and ninety building construction works for schools and health centers have been completed. The breakdown of completed projects by state as at 2003 was as follows: Abia 40, Akwa Ibom 53, Bayelsa 47, Cross River 4, Delta 97, Edo 38, Imo 31, Ondo 31, and Rivers 17(ANEEJ, 2006).
66
This notwithstanding, youth restiveness persists, so some critics feel NDDC should do more. The commission is accused of not carrying the communities along in planning developmental projects, and thus there is lack of project ownership. It is also accused of operating independently ignoring inputs from state governments by way of ideas. One finds that the NDDC, thus, faces criticism from both the civil society for its lack of transparency and collaboration with stakeholders in its operations. The NDDC maintains that it does not have enough resources to do much. NDDC chairman, Chief Onyema Ugochukwu said that out of the expected N 105 billion, the commission has only received about N 60 billion. This leaves a shortfall of about N 45 billion (Gas and Oil, 2003). Oil companies are accused of not remitting the right amount of money to NDDC. In the year 2001, the approved budget for Shell Petroleum Development Company (SPDC) was $2.004billion. It remitted only $380.074 million (1.30%). This was contrary to the stipulation in the NDDC act requesting it to remit three percent of its annual budget. Mobils approved budget was $1.029billion, but it released only 2.55 percent being $26.631million. Chevrons approved budget was $1.100billion and it gave $29million (2.69 percent). Agips budget was $461.551 million but only remitted $11.173 million (2.41 percent). ELF/Texacos budget was $516.390 million but remitted $8194. While Elfs contribution amounted to 2.44 percent, that of Texaco was 1.19 percent (Guardian, 2006). ANEEJ (2006) noted that oil companies are not complying with the provision of the NDDC Act. Aigbokhan (2007) has also established that even the Federal Government is not fully complying with the provision of the Act. As in June 2003 what the oil companies have contributed to NDDC is N25 billion, Federal Governments contribution is N20 billion and is paying 10% instead of fifteen percent. The most recent efforts at developing Niger-Delta is the launching of Niger-Delta Development Master Plan. At the launching of the Plan, Managing Director of Niger Delta Development Commission stated that The master plan is a 15- year development roadmap for the region, from 2006 to 2020, to be implemented in three phases of five years each. The broad goals of the master plan with mutually reinforcing components include poverty reduction, industrialization and socio-economic transformation to prosperity. The aim of this development strategy is to raise the standard of living of the people above the poverty line up to the standard of the global development initiatives of Millennium Development Goals and New Partnership for Africas Development. However, Alabi et al (2008a) have noted that the master plan envisaged that there would be collaboration between the federal Government, NDDC (Niger- Delta Development Commission), state governments in NigerDelta, oil companies, oil producing communities, NGO and other stakeholders. It is interesting to know that most of these past developmental efforts and policies in NigerDelta are crisis-driven. They are not based on analytical information and the needs in Niger-Delta. For example, Isaac Adaka Boro led a short-lived rebellion in 1967. The April 1990 Gideon Orkar led attempted coup detat in Nigeria in which Niger-Delta coup soldiers participated significantly (Isumonah, 2004). The Ogonis agitations soon followed coup detat. In response, government established the Oil Mineral Producing Areas Development Commission (OMPADEC). In response to local demands for greater resource ownership and benefits, and in an attempt to defuse the demands for resource control, President Obasanjo established Niger Delta Development Commission (NDDC). This fire brigade approach to policy and progemme formulation can not be sustainable in Niger-Delta. This study intends to correct the anomaly by providing analytical information that will guide intervention in for socio- economic development in Niger-Delta.
3. Unemployment Situation in Niger-Delta

If the people are engaged, they will be less prone to criminal activities. However, analysis of registered unemployment rate in Niger Delta is higher than in most other regions in Nigeria. The unemployment in the region was 22% which is higher than 18% national average (Aigbokhan et al, 2007). The results of analysis show that unemployment was higher in core Niger Delta (Delta, Bayelsa and River States)
67
than the than the rest of the states in Niger-Delta. Unemployment in Akwa Ibom, Bayelsa, Delta, Imo and Rivers States were 37%, 24%, 23%, 22% and 34% respectively, compared with 18% national average. The fact remains that these employment rates were even under estimated, because there are many unregistered unemployed people in Nigeria (Alabi and Osasogie, 2006). The higher unemployment rate in Niger-Delta can be blamed on oil exploitation and exploration in the region. Due to oil exploitation activities, farmers, fishermen, herbalist and rural women are out of job, leading to starvation, malnutrition and poor health. This is because of oil exploration and exploitation, the water is polluted, the land is degraded, the herb and crops are destroyed, and aquatic resources are depleted. Alabi and Oviasogie (2005) have indicated that river and streams in Niger-Delta contain oil that is above Maximum Permissible Concentration (MPC) for human for human safety.
Table 1:
State Abia Akwa Ibom Bayelsa Cross River Delta Edo Imo Ondo Rivers Niger-Delta All Nigeria
Unemployment Rates (%) in Nigeria and in Niger Delta

Composite 10.6 36.9 23.6 16.6 23.3 14.3 22.3 17.0 34.2 22.09 18.1 Urban 8.7 29.8 20.7 7.3 23.5 24.0 23.8 14.0 27.5 19.92 14.2 Rural 10.8 37.1 24.1 18.3 19.0 11.8 32.8 19.8 35.2 23.21 19.8
Source: Computed from National Bureau of Statistics, 2005.
The nature of unemployment in Niger-Delta is of interest to crisis generation in Niger-Delta. Table 2 shows a particular nature of unemployment problem in Niger-Delta, namely, that of youth unemployment. Unemployment rate was highest among the 15 24 age group and secondary school leavers. This explains why youth restiveness is quite pronounced in the Niger Delta region. Alabi and Ailemen(2005) explained that the high unemployment in Niger-Delta is induced by the fact that the people in the region did not have required skills and education that would have enable them to be employed in oil companies and oil subsidiaries. Since, these youth are unemployed, they can be induced to cause conflict, can be involved in oil bunkering and hostage taking, to keep themselves busy and get money to keep their mind and body together.
Table 2: Unemployment rates by Educational level and age group in Niger-Delta
Composite 16.5 17.8 21.9 15.8 39.7 15.6 10.9 13.6 17.6 Urban 12.9 13.8 17.6 10.1 42.1 10.4 8.1 9.0 21.3 Rural 18.0 19.5 23.8 18.3 38.7 17.8 12.1 15.6 16.0
Parameter No schooling Primary Secondary Above secondary Age Group 15-24 25-44 45-59 60-64 65-70
Source: Computed from National Bureau of Statistics,2005
68
4. Health Conditions in Niger-Delta

Generally, the average population growth in Niger-Delta is 3% as against 2.8 % for the rest of the country, with life expectancy of 45 years compared to Nigerias national life expectancy rate of 57 years (Aigbokhan,2003a). The lower life expectancy in Niger-Delta can be attributed to poor health conditions in Niger-Delta. According to Alabi and Ailemen(2005), in Port Harcourt, the regions biggest city, there is no citywide sewage system. Water related diseases are widespread and probably the central health problem in the Niger Delta. The health situations in Niger Delta is worst in all the health parameters examined in Table 3. Average of about 6 days were the numbers of days that people fell sick within two weeks in Niger Delta, while the national average was 5 days. This indicates that people of Niger Delta had 0.43 probability of falling sick within two weeks, while the national average was 0.36. The economic consequence of illness is its ability to stop economic activities. The number of days activities stopped due to ill health and injuries were about 7 days within two weeks; the national average was 6 days. This suggests that there was about 50% chance that people of Niger Delta will stop activities due to illness within two weeks. The numbers of people who were injured within two weeks were correspondingly higher in Niger Delta (5.08) than the national average (4.86).
Table 3:
State Abia Akwa Ibom Bayelsa Cross river Delta Edo Imo Ondo Rivers Niger Delta Nigeria
Summary of Health Information in Niger Delta and Nigeria

Days ill in the last 2 weeks. 5.35 4.38 3.86 6.05 5.49 4.52 6.31 4.00 5.81 5.09 5.04 Days activities stopped due to illness in the last 2 weeks 6.61 5.04 5.21 6.25 6.56 5.63 7.24 6.08 6.72 6.14 5.68 Days injured in the last 2 weeks 4.42 4.02 5.84 6.28 3.87 6.18 4.46 5.75 4.88 5.08 4.86
Source: Computed from National Bureau of Statistics Survey, 2005
This poorer health condition in Niger Delta could have being minimized if people have access to good health care delivery system. The accessibility to health care delivery was hampered by high cost of health services. Table 4 shows that the average hospital fee in Niger Delta was N 3821.54 ($27) which was also higher than national average of N 3017.04 ($22). Hospital fee in Imo and Edo state (N 6929) ($50) were more than double of the national average hospital fee. When the poor health condition in Niger Delta is coupled with the exorbitant hospital fee, the poor human capital status of people in Niger Delta can be best imagined. Table 4 also implies that health facilities available to the people in Niger-Delta were lower. The average total health facilities in Niger-Delta were 296, while the average for Nigeria were 396. The lower health facilities situation become pathetic when the health facilities in Niger-Delta is compared with states like Bauchi, Kaduna, Kano, Katsina and Niger, which had 666, 827, 666, 721and 737 health facilities respectively. The Primary Health Centers (PHC) in Niger-Delta was still lower than the national average. The average PHC per state in Nigeria was 370, while the average in Niger-Delta was 271. This lower PHC in Niger-Delta is unacceptable when you compare with average for states like Bauchi, Kaduna, Kano, Katsina and Niger, which had 648, 813, 637, 719 and 718 PHC respectively. On population basis, the ratio of population to Health Facilities (HF) also shows that many people in Niger-Delta had to be sharing the few health facilities available. While 5521 people on the average in Nigeria will be sharing one health facility, 7674 will be sharing the same one health facility in Niger-
69
Delta. Alabi et al (2008b) have also supported the fact that accessibility to health facilities in Edo state was lower than the average for Nigeria. The low health facilities will compound the poverty level in Niger-Delta. Many indicators show that poorer individuals are generally less healthy (Gwatkin et al, 2003), and one may presume that they are in greater need of health facilities than non-poor. Therefore, the low health facilities in Niger-Delta may explain prevalence of sick people in the region.
Table 4:
State Abia Akwa Ibom Bayelsa Cross River Delta Edo Imo Ondo Rivers Niger Delta Nigeria
Summary of Information on Health in Nigeria and Niger Delta

Average Hospital fees 3250.00 1911.94 2616.67 5247.39 1135.00 6928.51 6928.52 2117.50 4258.33 3021.54 3017.04 *PST Health Facilities 235 390 153 429 285 292 235 356 292 296 395 Primary Health Centers (PHC) 221 344 142 406 259 254 220 331 260 271 370 **Ratio of Population to Health Facilities 3678 6419 25577 5231 6846 4726 4180 5278 7128 7674 5521
Source: Computed from National Bureau of Statistics, 2006 *PST is the combination of primary, secondary and tertiary health institutions
5. Education an Schooling in Niger-Delta

Another important determinant of human capital is education. Table 5 reveals low accessibility to education in Niger-Delta. The average school fee in Niger Delta was N 4788.84 ($34) which was also higher than the national average of N 4231.59 ($30). This may explain low level of formal education and high unemployment rate as earlier noted by Aigbokhan et al (2007). The average number of Public Primary School (PPS) in Niger-Delta was 956, while the average for Nigeria was 1371. The public primary schools in Bauchi, Kaduna, Kano, Katsina and Niger were 1783, 3034, 3066 and 2450 respectively (these are the Northern states that are classified as educationally disadvantaged areas by the federal government). The primary school is the place to acquire the basic education, with which one can make progress in academic leader, so where this is lacking, the chance of academic progress of the place can be low. The primary and secondary school enrolment in Niger-Delta were 535453 and 139349 respectively, which were lower than the national average primary and secondary school enrolment of 566265 and 146557 respectively. The fact that there is lack of education personnel in Niger-Delta reflects in higher primary and secondary school pupil/teacher ratio in the region. The primary and secondary pupil/teacher ratios in Niger-Delta were 42.00 and 32.39 respectively, while the averages for Nigeria were 40.00 and 32.10 respectively. In fact, the table shows that pupil/teacher ratio in primary school in Bayelsa was about 96. Edo and River states had pupil/teacher ratios of about 54 and 43 in primary and secondary school respectively. These figures are disheartening when you compare it with the same figures for Niger and Kaduna states. Niger state had primary and secondary school pupil/teacher ratios of 25 and 23 respectively, while Kaduna had primary and secondary school pupil/teacher ratios of 31 and 34 respectively. In fact, the low accessibility to education in Niger-Delta may suggests that the region is more educationally disadvantaged than the so called educationally disadvantaged areas in the Northern Nigeria. This is because Bauchi, Kaduna, Kano, Katsina and Niger have more access to educational facilities and institution than the average for Niger-Delta.
70
Table 5:
State Abia Akwa Ibom Bayelsa Cross River Delta Edo Imo Ondo Rivers Niger Delta Nigeria

Summary of Information Education in Nigeria and Niger Delta
ASF 5017.34 5454.95 7304.49 3935.94 5021.70 2754.15 4686.15 2334.46 6590.39 4788.84 4231.59 No. PPS 759 1093 502 869 1047 970 1230 1150 982 956 1371 EPS 356254 1443848 470217 434607 315604 244414 632022 593375 328735 535453 566265 ESS 109022 165764 49487 84703 195934 129254 139596 182354 198032 139349 146557 PSPTR 24.86 70.96 95.87 31.37 17.53 27.36 38.00 36.46 32.97 42.00 40.00 SSPTR 29.38 34.59 25.59 26.75 22.29 54.20 32.71 23.22 42.80 32.39 32.10
Source: Computed from National Bureau of Statistics, 2006
The situation of Niger Delta in terms of school and hospital fees is pathetic considering the huge amount of money devoted to the place. The authoritative report said that more than N2 trillion ($14) were given to Niger Delta within 7 years of democratic governance in Nigeria (Oil Revenue Watch, 2006). The revenue in Table 6 include the monthly statutory allocation from Federation Account, Excess Crude Oil proceeds, 13% Oil Derivation fund, Value Added Tax and Ecological Funds for States and Local Government within the states. If the volume of monies sent to Niger Delta is used judiciously, the people in Niger Delta could have been enjoying free education and health provision.
Table 6:
State Abia Akwa Ibom Bayelsa Cross river Delta Edo Imo Ondo Rivers Niger Delta
Source: Oil Revenue Watch, 2006
Revenue flow to Niger Delta (June 1999 to June 2005)

Amount (N) 102,853,702,423.15 264,231,991,046.22 230,935,505,788.08 108,001,279,844.88 332,495,013,802.40 112,319,975,389.28 133,407,024,267.19 139,051,991,269.82 285,320,125,530.50 1,708,616,609,361.54
6. Poverty and Cost of Living in Niger-Delta

Despite this huge financial resources that flowed into Niger Delta, the people of the region are very poor. The self rated poverty assessment conducted in 2005 shows that 31% of the people in Niger Delta were very poor compared with 24% which was the national average. In fact, Table 7 shows that 61% and 41% of people of people in Bayelsa and Edo state were very poor respectively. Aborribo (2002) also supported the fact that more people are in poverty bracket in Niger-Delta than the average for the country. This is indeed a provocative situation of lack in the midst of plenty. This implies that the people of Niger-Delta may not only have poverty of knowledge as a result of low accessibility to education, the poverty of material things is also predominant in the region. The poverty level is aggravated by high cost of living in Niger Delta. Due to influx of oil workers in the region, the cost of living is higher than the national average. Table 7 shows that composite per capita expenditure in Niger-Delta was N43827 ($313), while the national composite average per capita expenditure composite was N39155 ($280).
71
Table 7:
State Abia Akwa ibom Bayelsa Cross river Delta Edo Imo Ondo Rivers Niger Delta Nigeria

Self Rated Poverty Incidence and Composite Per Capita Expenditure in Niger Delta and Nigeria.
Core poor 36 15 61 25 26 41 28 33 17 31 24 Per capita expenditure (Composite) (N) 48770 40612 42516 42526 36050 40402 63341 43416 36809 43827 39155
Source: Computed from National Bureau of Statistics Survey, 2005
7. Accessibilitiy to Infrastructure in Niger-Delta

Table 8 indicates lower accessibility to Personal Computer (PC), Credit Facility (CF), Improved Waste Disposal (IWD) and treated Pipe-Borne Water (PBW) in Niger-Delta, compared with Average for Nigeria. The average accessibility to PC, CF, IWD and PBW in Niger-Delta were 1.1%, 10.5%, 11.3% and 6.4% respectively, while the national averages for these facilities were 1.3%, 10.7%, 16% and 13.6% respectively. Availability to Personal Computer will enable the people to have access to productive information that can transform their economic bases to posperity. Credit facility is important in improving the production and productivity of the people (Kifle, 2007). The importance of finance in poverty alleviation and income generation in Nigeria context has been emphasized by Anyawu (2004). Where credit facilities are lacking, people will not be able to overcome capital constraints that are limiting their productive activity. In fact, low access to credit facilities can limit access to other infrastructure. Lack of improved waste disposal system may explain the reason why many of the people in Niger-Delta are disposing their wastes in the streams they drink. Poor accessibility to pipe borne water in Niger-Delta is difficult to explain, because people are surrounded by water. However, Alabi and Ovasogie(2005) have demonstrated that this water is polluted by oil spills which render it undrinkable unless treated. The general consequence of the low accessibility to water and poor waste disposal system is increase in water borne-diseases and illhealth as noted previously.
Table 8: Accessibility to Personal Computer (PC), Credit Facilities (CF), Improved Waste Disposal (IWD) and Pipe-Borne Water (PBW) in Niger-Delta (%).
PC 1.6 1.3 1.4 0.5 1.3 1.1 0.9 0.5 1.7 1.1 1.3 CF 4.4 13.0 8.7 13.8 11.9 8.3 11.7 17.6 4.7 10.5 10.7 IWD 17.2 4.0 2.0 18.0 13.5 23.1 3.4 9.6 11.1 11.3 16.1 PBW 3.4 7.4 7.5 2.4 2.9 9.7 6.2 6.0 12.4 6.4 13.6
State Abia Akwa ibom Bayelsa Cross river Delta Edo Imo Ondo Rivers Niger Delta Nigeria
Source: Computed from Core Welfare Indicator Questionnaire (CWIQ) Survey, 2006
72
8. Conclusion and Policy Recommendations

The paper established that poverty level, unemployment, cost of health, cost of education, inaccessibility to social infrastructure were higher in Niger Delta than the national average, coupled with increased cost of living. The crises and conflicts in Niger Delta are predicated on these socio economic constraints that are rampart in the region. The study also indicates that the enormous financial resource that got to the region has not significantly improved health, education and infrastructural facilities in the region. Therefore, all the stakeholders in Niger-Delta have a role to play to improve socio-economic situation in the region so that the imminent danger can be averted. The stakeholders are Federal government, state government in Niger-Delta area, oil companies, donor agencies, NDDC, Non governmental organisation, civil society, Niger-Delta militants and host communities. Since improved financial situation in Niger-Delta has not improved its socio-economic condition, there is need for transparency in managing the financial resource in the region. There should be strict implementation of Nigeria Extractive Industry Transparency Initiative (NEITI) and Publish What You Pay campaigns for transparency in management of oil revenues by the Federal and State governments. Nigeria has been known to have good policies, but implementation is always abysmal failure. The states governments in Niger-Delta should take human capital development in the region more seriously. The states in Niger-Delta should be able to implement free education and free health programme, even if other state can not implement it. State of emergency should be declared in Education and health sectors in Niger-Delta by the federal government. This should continue until accessibility to education and health care in the region becomes equal with the average for the country. The Federal government should include Niger-Delta among the educationally disadvantaged states. This is because, this study shows that the so called educationally disadvantaged states in Nigeria have more access to education than the average in Niger-Delta. Since the study shows that oil companies are not complying in paying the allocation due to NDDC, oil companies should release funds as at when due to NDDC and check that the fund is judiciously used by NDDC. The 3.0 percent allocated to NDDC from the Federation Account must be paid up by the federal goverment. Oil companies and donor agencies should lay emphasis more on human capital development in Niger-Delta in terms of health and education facilities provision. They should have and implement developmental programmes that related to health and education provision in the communities where they operate. Donor agencies, Niger Delta Regions developments institutions and the state governments in the Niger-Delta should always co-ordinate their activities. There a lot of duplication of projects and plan in Niger-Delta, so it is difuclt to know, the contribution of each intervention agency in improving the socio-economic conditions of the region. The new Niger- Delta Master Plan must also recognize the felt needs of the people in the region(education, health and employment) and make them the priorities in their intervention activities. In implementing the master plan, there should be collaboration between all the stakeholders in NigerDelta. This integrative and collaborative implementation would lead to successful intervention. The Plan should be implemented with unwavering commitment, unity of purpose, harmony and of course transparency in order to make optimal contributions to socio-economic development of Niger- Delta. The oil producing communities should give peace a chance. The development of the region is only possible in the midst of peace and tranqulity. Militants should be encouraged to dialogue with the present government for lasting peace in the region. Finally, this paper concludes that the challenge of socio-economic development of Niger-Delta is complex but not insurmountable. Government owes a moral obligation to the Niger-Delta people to create an enabling environment for peace and provide sustainable development in the region.
73
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Aboribo, R. (2000). The political economy of environmental pollution and compensation: The case of victims and the law. Proceedings of the Conference on Management of Petroleum Resources for Sustainable Development in the Niger Delta in 21st Century, held at Delta State University, Abraka, 20-23th September,2000. Afolabi, A.O, A.A.Adesina and S.M.Agbonkhese (2003). Peace and conflict assessment methodologies: An analysis of the establishment of Niger Delta Development Commission (NDDC) as response to youth restiveness in the Niger Delta. Women Youth Restiveness and Development in the Niger Delta. Iyoha, F.E., S.O. J. Ojo and P.O. Oviasuji(eds). The institute for Governance and Development, Ambrose Alli University, Ekpoma. Pp 141 150. Aigbokhan, B., Alabi, R.A and M.I Ailemen (2007). Oil resources and incidence of conflicts in Nigeria. In K.Wohlmuth, C, Eboue, A Gutowski, J.Afeikhena, T.Knedlik, M,Meyn and T. Mama (eds). Africa Commodity Dependence, Resource Cuse and Export Diversification. LIT Verlag, Berlin, Germany. Pp45-65. Aigbokhan, B(2007). Reconstruction of economic governance in the Niger-Delta region in Nigeria: The case of the Niger-Delta Commission. In K. Wohlmuth and T.Urban (eds). Reconstructing Economic Goernance After Conflict in Resource-rich African Countries. LIT Verlag, Berlin, Germany. Pp193-198. Aigbokhan, B.E. (2003a).Challenges and options for social welfare development in the Niger Delta. Report Prepared for the West African Institute for Financial and Economic Management, Lagos. P. 16. Alabi, R.A., M. Ailemen and Osasogie D.I (2008a). Perception, causes of poverty and poverty alleviations programmes in Niger-Delta, Nigeria. African Development Perspective, Volume 14. Institute of World Economics and International Management, Bremen, Germany (in Press). Alabi, R.A.,Osasogie D.I and A.A. Alabi(2008b). Estimating the cost of health in urban and rural areas of Edo State, Nigeria. Journal of Human Ecology(in Press). Alabi, R.A and M.I Ailemen (2005). Human capital development in Niger Delta: Pathway to crisis management. Paper presented at International Conference on Human Development, held on 4 - 6th June, 2005. Covenant University, Ota, Ogun State, Nigeria. Alabi, R. A and D. I Osasogie(2006). Income generation by participants in National Directorate of Employment (NDE) in Edo state. Paper presented at 48th Annual Conference of Nigeria Economic Society, 13th -15th August, 2006, Calabar, Nigeria Alabi, R.A and Oviasogie, D.I (2005). A preliminary assessment of effect of oil exploration on fish farming in Niger Delta: A case study of Delta State, Nigeria. Journal of Agriculture and Environmental Research Studies Vol. 1 (1): 29 36. Alabi, R.A and Oviasogie, D.I. (2003). Perception and dimension of poverty among farmers in sselected LGAs of Edo State, Nigeria. The Nigerian Journal of Research and Production 3(3): 90 96. ANEEJ (2006). Oil of Poverty in Niger Delta. African Network for Environment and Economic Justice. Benin, Nigeria Pp 1 88. Anayochukwu, A (2007). A new dawn in the Delta. Tell Magazine. Number 23. A Publication of Tell Communications Limited, Ikeja, Lagos, Nigeria. Pp 30 -31 Anyawu, C.M (2004). Micro- finance institutions in Nigeria: Policy, practice and potentials. Paper presented at G24 Workshop on Constraints to Growth in Sub- Sahara Africa. 29- 30th November, 2004. Pretoria, South Africa. Arowolo.A (2008). Niger-Delta: Threat to 2008 budget. Available on the internet via www.punchng.com/Artic.aspx?theartic=Art 2008010715312210. Accessed on 07/01/08 CSAE (2005). Reducing the global incidence of civil wars: A discussion of available policy instruments. Centre for Study Africa Economies Research Summary.Oxford University, London. Pp 43- 60
References
[1]
[2]
[3]
[4]
[5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16]
74 [17] [18] [19] [20] [21] [22] [23] [24]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) ECA (2005). Economic governance and public financial management. Economic Commission for Africa Annual Report. Addis Ababa, Ethiopia Pp 55 80. Gas and Oil (2003): Alexander Gas and Oil Connection: News and Trends in Africa. Vol 8, Issue 21 Pp 30-31. Guardian (2006): Supplement on Ondo State Oil Producing Areas Development Commission (OSOPADEC). The Guardian, June 23, 2006, page 20 Gwatkin, D.R., Rutstein, S., Johnson, K., Suliman, E.A., Wagstaff, A. (2003). Initial countrylevel information about socio-economic differences in health, nutrition, and population. Washington, DC, The World Bank. Ifidon, S.E (2006).Planning without information: The bane of national development. The 26th Inaugural Lecture in Ambrose Alli University. Published by Ambrose Alli University, Ekpoma, Nigeria. Pp 32 34. Isuomonah, V.I (2004). Making difference with Niger-Delta Development Commission. Centre for Advanced Social Sciences Newsletter, 11:14 -16 Joab-Peterson (2004). State and the Niger Delta crisis in historical perspective. Centre for Advanced Social Science Newsletter 11:8- 13. Kifle, T (2007). Challenges facing Eriteria in growth and Poverty reduction policies: Does microfinance Help?. Paper presented at International Conference on Pro-poor Growth and Poverty alleviation. Institute for Wold Economics and International Management. Held on 810th Novemeber, 2007. Faculty of Economics,University of Bremen, Bremen, Germany. Oil Revenue Watch (2006). The Newsletter of African Network for Environment and Economic Justice (ANEEJ) 1(1): 3-5. Benin City, Nigeria. Olukorede, Y (2007). Just before the intervention. Tell Magazine, Number 23. A Publication of Tell Communications Limited, Ikeja, Lagos, Nigeria. Pp28-29
[25] [26]
Market Efficiency and Company Size. Empirical Evidence from the Athens Stock Exchange
Christos A. Alexakis Assistant Professor, Department of Economics, University of Piraeus 80 Karaoli and Dimitriou str., 185 34 Piraeus, Greece Tel: +30 210 4142300; Fax: +30 210 4142301 E-mail: calexakis@tutors.eap.gr Abstract According to financial theory, in an efficient market investors reflect fully and instantaneously all relevant information on security prices under rational expectations. Empirical research on market efficiency investigates if there is past available information which can help to predict future returns profitably, as well as if factors not related to rational economic behaviour, influence stock prices. This study investigates whether efficiency is an issue influenced by factors like the size of the listed companies. Company size is a characteristic which may be taken into account by investors when they make their investment decisions. Thus, different company size may attract different types of investors i.e. big capitalization companies may attract different kind of investors as compared to small size companies. The behavioural characteristics of these investors may affect security prices and consequently we may obtain evidence of different degree of efficiency among different classes of companies based on size. Keywords: Informational efficiency, causality, stock returns, trading volume. JEL Classification Codes: G14.
1. Introduction
According to the Efficient Market Hypothesis (EMH), Fama (1970), (1976), asset prices reflect fully and instantaneously all relevant available information in a rational, i.e. in accordance with economic theory manner, and factors not linked with economic theory, like investors psychology, should not affect asset prices. In an efficient market, past information is of no use in predicting profitably future asset returns, since it has been already fully reflected on asset prices by a number of competing, profit maximizing investors. An efficient market should react only to new information, but since this is unpredictable by definition, asset price changes or asset returns cannot be predicted. Thus, the empirical research for market efficiency investigates if there is past available information which can help to predict future returns profitably, but also investigates if non economic factors like investors psychology influence asset prices. In this study we will test the Efficient Market Hypothesis for different classes of stocks based on company size. Analytically, we will examine the possibility of a predictive relationship between stock returns and stock returns and trading volume for small, medium and large capitalization companies in order to test if company size matters in market efficiency. We suspect that size may matter, since the size of a company may attract, for a number of reasons, different types of investors which in turn, because of their behavioural characteristics, may affect stock returns differently.
76
Section two (2), presents a review of the relevant theory, section three (3), presents the data sets and the methodology used in this study, section four (4) presents the empirical findings and finally, section five (5) summarizes the results and provides some possible explanations.
Under the Efficient Market Hypothesis the Fair Game1 model holds for stock price changes and consequently for stock returns: (1) E[Pt-(P*t/It-1)]=0 or E(rt/It-1)=0 where It-1 is the information set available at time t-1, Pt is the actual price at time t, P*t is the expected price which is based on the information set It-1, and Pt-P*t is the forecast error which is uncorrelated with variables in the information set It-1. Similarly, rt is the stock return which is uncorrelated with variables in the information set It-1, Le Roy (1989, 1990). According to Samuelson (1965), under the assumption of a non zero equilibrium return and assuming that agents have constant and common time preferences, common probabilities and are risk neutral, then if all assets are to be held willingly, as must be the case for equilibrium, all should therefore earn the same expected rate of return, equal to the equilibrium return. Fama (1970), rejected the hypothesis that returns themselves are a Fair Game and proposed the following definition of market efficiency, which makes the EMH a joint hypothesis: Zt= r t -E(rt/It-1 (2) with: E(zt)=E[r t-E(rt/It-1)]=0 (3) In economic terms zt is the return at time t, in excess of the equilibrium expected return projected at time t, on the basis of the information set It-1. With the additional assumption that the equilibrium return is constant through time 2, then returns themselves are uncorrelated with variables in past information sets. Research conducted in 60s and 70s generally supported the market efficiency but recent evidence however does not support the same conclusion. The evidence now suggests, contrary to the prediction of the efficient market model, that most fluctuations in stock prices can not be traced to changes in rational forecasts of future dividends. The recent evidence arises from two areas of research. First, analysts came to realise that stock returns display a variety of systematic patterns, some kind of anomalies, which are difficult to be explained by the Efficient Market Hypothesis. Second, analysts realised that the same models which imply that returns should be unforecastable also imply that asset prices should have a volatility which is low relative to the volatility of dividends. The excess price volatility can be explained from the fact that investors could be reacting to information which is irrelevant to stock prices and that forces other than rational forecasts of future dividends may influence stock prices. Roll (1988) found that irrelevant information appeared to be of dominant importance since economic factors were able to explain only a small fraction of the variance in stock prices. Almost at the same time, Cutler et al (1989) provided evidence that stock returns are unrelated to news. Prior to Roll and Cutler, Black (1986), in his presidential address in the American Finance Association used the term noise as a large number of small events which is often a causal factor much more powerful than a small number of large events. In this context, psychological factors may be considered as noise and psychology driven investment decisions as noise trading.
2. Theoretical Framework
The Fair Game model is derived from the Martingale model: E(Pt/It-1)=Pt-1. According to the Martingale model, if the price of a stock, is a Martingale the best forecast of price Pt that could be constructed based on the available information set It-1, would just equal Pt-1, assuming that Pt-1 is in It-1. The assumption that the equilibrium return is constant through time is crucial for empirical tests because as Leroy (1989) noted, "On Fama's definition any capital market is efficient and no empirical evidence can possibly bear the question of market efficiency."
77
Since noise may influence investors decisions, it is realistic to assume that there may be a segmented asset market. Smart money i.e. investors who act under rational expectations may be one group; and noise traders i.e. investors who are influenced by factors not related to economic theory, like psychology, may be another group. It is also plausible to assume that the characteristics and the interaction of these two groups may affect asset price behaviour, De Long et al (1990a,b). In this study, apart from stock returns, we will use as a variable in order to test the Efficient Market Hypothesis, the trading volume, since it has been recognized by economists as an important factor indicating investors interest. In some empirical tests for market efficiency, price changes are interpreted as the market evaluation of new information, while the corresponding trading volume is considered as an indication of the extent to which investors disagree about the meaning of the information, Karpoff (1987), Hiemstra and Jones (1994). Also, some researchers argue that trading volume may provide insights regarding the quality of trader information that cannot be obtained from price statistics, and the joint analysis of past price and volume data can prove useful in providing information about future price movements, Blume et al (1994). In an early empirical examination of the price - volume relationship, Granger and Morgerstern (1963), discerned no relation between movements in a Securities and Exchange Commission Composite Index and the aggregate level of volume for the New York Stock Exchange. It was then argued that in the stock market the classical theory of demand and supply does not apply and the reason offered was that market participants can not be neatly divided into the groups of buyers and sellers and so it is not likely to exist a clear cut relationship between trading volume and price or price change. Furthermore, Godfrey, Granger and Morgerstern (1964) did not find any relationship between price changes or the absolute value of price differences and volume. Subsequent empirical evidence by Ying (1966), indicated that a small volume is accompanied by a fall in price, a large volume by a price rise, while a large increase in volume is accompanied by either a large rise or fall in price. Further empirical research by Crouch (1970), indicated a price change - volume correlation but with no evidence for causality i.e. predictability. The discovered relation was almost entirely contemporaneous, as most leading and lagged variables were statistically insignificant, contradicting the old Wall Street proverb it takes Volume to make prices move. Following Yings results, empirical studies have shown a positive correlation between volume and price change per se but again no lagged relationship has been found, implying a contemporaneous relationship between price change per se and volume, Rogalski (1978), Harris and Garel (1986). Nevertheless, recent international evidence, and especially evidence from emerging markets, gives support for causality relationships between stock returns and trading volume, but the evidence is not clear in terms of the involved dynamics. Moosa and Al-Loughani (1995) examine Asian stock markets and they find strong causality evidence running from volume to absolute price changes and from price changes to volume in almost all markets under examination. Silvapulle and Choi (1999) tested the dynamics between daily returns and trading volume in the Korean market and they report strong linear and non linear bidirectional causality. Lee and Rui (2000) analysed daily data on stock returns and trading volume of four Chinese stock market indices and they report a strong causality relationship running from returns to volume in all cases. Chordia and Swaminathan (2000) examine the interrelationship between trading volume and the predictability of short-term stock returns. They find that daily and weekly returns of high volume portfolios lead returns of low volume portfolios. The authors attribute these findings to the differences in the speed of price adjustment to information between the two types of stocks; stocks in low volume portfolios respond slowly to market-wide information while their high volume counterparts respond promptly to such information. Chen et al. (2001), conduct a comprehensive study examining causal relation between stock returns, trading volume and volatility using daily data for nine major markets. They document strong evidence for the argument that return causes volume; however, they were able to uncover only limited evidence to suggest that volume causes returns since only four markets displayed such causality.
78
3. The methodology employed

A very popular way to test the existence of any temporal statistical relationship, in terms of prediction, between two variables is the Granger causality test, Granger (1969). Grangers tests for causality, in the sense of precedence, are based on the following statistical reasoning: if we consider two time series as Yt and Xt, the series Xt fails to Granger cause Yt, if in a regression of Yt on lagged Ys and lagged Xs the coefficients of the latter are zero. The presence of causality obviously implies market inefficiency: Under the Efficient Market Hypothesis (EMH), it is true that past information is of no use in predicting future stock price movements, that is stock price changes and consequently stock returns should be unpredictable in an efficient market. Thus, for the return of a stock index, say j, it must be true that: E(rjt/It-1) = 0 (3) where It-1=[Pj,t-1, Pj,t-2, Pj,t-3,Pj,t-n] and Pjt-1,Pjt-n is the price history of the stock index j. If it is also true that: E(rjt/Ht-1)=0 (4) where H t-1=[Pj,t-1, Pj,t-2, Pj,t-3,Pj,t-n, Yk,t-1, Yk,t-2, Yk,t-3,Yk,t-n] and Yk,t-1,Yk,t-n is the history, of a variable k different than j, then no Granger causality exists and the market is still efficient with respect to the information set, Ht-1. The opposite case implies that past values of variable k can help to predict the return of stock j, and the market is inefficient with respect to the information set, Ht-1. The standard Granger causality tests are usually performed on stationary data. Nevertheless, the first difference transformation, which is often used to attain stationarity filters out low frequency (long run) information. Cointegration reintroduces in a statistically acceptable way, the low frequency information. The basic idea of cointegration is that when two or more series move closely together in the long run, even though the series themselves are trended, the difference between them is constant. We may regard the cointegrating series as defining a long run equilibrium relationship and the difference between them to be stationary. The term equilibrium in this case suggests a relationship which, on average, has been maintained by a set of variables for a long period, Engle and Granger (1987), Johansen and Juselius (1990). Cointegrated variables in the bivariate case must possess temporal causality in the Granger sense, in at least one direction, since for a pair of series to have an attainable equilibrium, there must be some causation between them to provide the necessary dynamics3.
4. Data sets and results

In this analysis we used daily observations i.e. closing prices for the indices of the Athens Stock Exchange FTSE20 (big capitalization stocks), FTSE40 (medium capitalization stocks) and FTSE80 (small capitalization stocks) and the corresponding trading volume, expressed by the number of shares traded. Data cover the time span 01/02/2003 to 03/31/2007; a total of 1062 observations (Graphs I and II). The indices are adjusted for dividends, stock splits and reverse stock splits. Finally, in all cases the logarithmic transformation of the original series was used.
MacDonald and Kearney (1987) point to the fact that the standard Granger causality test, are misspecified in the case of cointegrated variables.
79

Graph I: Stock Prices FTSE20-40-80
Graph II: Stock Returns and Trading Volumes, FTSE20-40-80
80
The basic statistics of the series under examination are presented in Table I. s expected by the theory, risk as it is measured by the standard deviation, is in a negative relation with the size of the companies which are included in the index. Nevertheless, the mean return for the small companies index is less then the mean return for the other indices. Finally, the distribution of the series under examination is found to be leptokurtic.
Table I: Basic Statistics
Mean 0.0010 0.0010 0.0006 0.0017 0.0010 -0.0002 St. Dev. 0.0114 0.0119 0.0149 0.502 0.451 0.356 Skewness -0.1266 -0.5511 -0.4820 -0.0150 0.1810 0.0400 Kurtosis 4.8028 6.7339 7.7337 7.3780 6.8480 4.4390
Variable Price FTSE20 Price FTSE40 Price FTSE80 Volume FTSE20 Volume FTSE40 Volume FTSE80
Table II presents the unit root test results. Analytically, we performed the Phillips- Perron test allowing for an intercept term and a time trend when this was necessary. For the case of the stock prices, it is clear from the table that the null hypothesis that any of the series have unit roots cannot be rejected. This is confirmed by the statistics which test for unit roots in the first differenced series. In each case the null hypothesis is easily rejected. Together with the results in the level series, it strongly implies that each of the stock price series are integrated of order one, I~(1). The unit root test statistics for the case of the trading volume series indicated that these are stationary at their levels i.e. they are integrated of order zero, I~(0)4. Based on the above results, the Granger causality tests can be performed on the first logarithmic difference of the original series.
Table II: Unit Root Tests
Levels -0.82 0.65 -0.62 -19.20** -14.44** -13.63** Transformation -23.75** -16.33** -13.84.** -
Variable Price FTSE20 Price FTSE40 Price FTSE80 Volume FTSE20 Volume FTSE40 Volume FTSE80
Double star(**) indicates significance at 99 % confidence interval.
The results obtained from the standard Granger causality tests are presented in Tables IIIa, b and c. The lag selection in the above tests ensured white noise residuals but also the Akaike information criterion was taken in to account. The relevant F statistics indicate that the lagged returns of the FTSE20 and FTSE40 indices can help to predict the return of the FTSE80 index. On the other hand, there is no causality evidence between the FTSE20 and the FTSE 40 indices. Nevertheless, in light of the possibility of cointegration between the price series of the examined indices, the above results may not be valid. Thus, we proceeded to cointegration tests. In table IVa, b and c we present the Johansen cointegration results As we can observe in Tables IVa, b and c, no co-integrating vector was traced (i.e. r = 0) among stock prices of the indices under examination. The above results lead us to conclude that there is no long run statistical equilibrium between the examined series and that the standard Granger causality tests were correctly specified.
The stationarity results of the unit root tests were confirmed be a visual inspection of the series and the behaviour of their sample autocorrelation function, (Mills 1993).
81
Table IIIa,b,c: FTSE20-40-80 - Returns Granger causality Tests Table IIIa: FTSE20-FTSE40 Granger causality Results
Depended Variable: FTSE20 Causality statistic:0.38 Depended Variable: FTSE40 Causality statistic:0.87 causality direction No causality
Table IIIb: FTSE20-FTS80 Granger causality Results

Depended Variable: FTSE20 Causality statistic:0.05 Depended Variable FTSE80 Causality statistic:3.46* causality direction No causality
Table IIIc: FTSE40-FTSE80 - Granger causality Results

Depended Variable: FTSE40 Causality statistic:0.37 Depended Variable: FTSE80 Causality statistic:3.65* causality direction FTSE40 cause FTSE80
Single star(*) denotes significance at 95% confidence interval
Table IVa,b,c: FTSE20-40-80 - Stock Prices Cointegration Tests Table IVa: Johansen cointegration statistics FTSE20 FTSE40
Ho r=1 Trace results - Selected lag length, p =4 Eigenvalue Likelihood Ratio 0.0066 7.22 5% Critical value 15.41
Table IVb: Johansen cointegration statistics FTSE20 FTSE80

Table IVc: Johansen cointegration statistics FTSE40 FTSE80

Finally, we performed the standard causality tests between returns and trading volume of the examined indices. The results of the above tests are presented in Table V a, b and c. According to the statistics the only causality relation was found between the stock returns and the trading volume of the index FTSE80. Analytically, the relevant F statistic indicates that lagged returns of the index FTSE80 precedes changes in the trading volume of the index FTSE80, i.e. returns cause trading volume.
82
Table Va,b,c: FTSE20-40-80 - Return and Volume Granger causality tests Table Va: FTSE20 - Volume of FTSE20 Granger causality Results
Depended Variable: FTSE20 Causality statistic:1.42 Depended Variable: Trading Volume FTSE20 Causality statistic:0.07 causality direction No causality
Table Vb: FTSE40 - Volume of FTSE40 Granger causality Results

Depended Variable: FTSE40 Causality statistic:1.99 Depended Variable: Trading Volume FTSE40 Causality statistic:2.20 causality direction No causality
Table Vc: FTSE80 Volume of FTSE80 Granger causality Results

Depended Variable: FTSE80 Causality statistic:0.38 Depended Variable: Trading Volume FTSE80 Causality statistic:3.63* causality direction return cause trading volume
Single star(*) denotes significance at 95% confidence interval
5. Conclusions
In this study we examined the dynamics between stock returns and stock returns and trading volume of three different indices of the Athens Stock Exchange. The examined indices represent big capitalization stocks (FTSE20), medium capitalization stock (FTSE40) and small capitalization stocks (FTSE80). The results obtained from the standard Granger causality tests indicated that the FTSE20 and FTSE40 stock returns cause the returns of the FTSE80. In addition, we obtained statistical evidence that there is a causality relationship between stock returns and trading volume only for the small cap index FTSE80, and that the causality runs from returns to trading volume. The above results suggest that the small capitalization market may not be as efficient compared to the medium and large capitalization markets. First, there is a slow reaction to news, as the returns of the small capitalization companies are led by the returns of the medium and large capitalization companies. Second, in the case where lagged stock returns influence trading volume, for the small capitalization companies, we can argue that market participants take into account past information in their trading actions or that their trading actions (investment interest as expressed by trading volume) are influenced by some psychological factors which are generated by the lagged returns e.g. a positive lagged return may create optimism for future returns and a negative return may cause pessimism, and that investors trades are based on that feelings i.e. investors of the small capitalization companies are influenced by psychological factors. In order to investigate the reason why the above statistical evidence is prominent for the small capitalization companies we analysed the investors profile of the examined indices as this is provided by the Central Depository of the Athens Stock Exchange. The analysis of the investors profile is presented in Table VI. It is clear from the table that the participation of the individual investors in the small capitalization index is much higher that their participation in the big and medium capitalization indices, which are dominated by institutional investors. Small investors like individuals, are more likely to act as noise traders and create the observed evidence against market efficiency, something of which evidence is reported in other studies5. The reason why small investors prefer to invest in small capitalization companies is a question which we hope to examine in the future.
Table VI: 2003-2007, Investor Profile Participation FTES20 FTSE40 FTESE80
5
Weiss (1989) examined the Closed End Found Anomaly and he reported that individual investors own a larger proportion of Closed End Funds shares compared to Institutional Investors.
83

Private Maximum 21.96 37.45 62.24 Institutional Maximum 82.57 74.97 46.74
FTSE20 FTSE40 FTSE80
Average 19.91 32.01 57.96
Minimum 17.43 25.03 53.26
Average 80.09 67.99 42.04
Minimum 78.04 62.55 35.76
Chart I: 2003 -2007, Investor Profile Participation FTES20 FTSE40 FTESE80
84
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Black F.,(1986), Noise, Journal of Finance, 41, 529-534. Blume L., Easley D. and OHara M. (1994), Market Statistics and Technical Analysis: The Role of Volume, Journal of Finance, 49, 153-181. Chen, G.; Firth, M; and Rui, O.M., (2001), The dynamic relation between stock returns, trading volume and volatility, Financial Review, 36, 153-173. Chordia, T., and Swaminathan, B., (2000), Trading volume and cross-autocorrelations in stock returns, Journal of Finance 55, 913-935. Crouch R. L. (1970), The volume of Transactions and Price Changes on the New York Stock Exchange, Financial Analyst Journal, 26, 104-109. Cutler D. M. Poterba J. M. and Summers L.H., (1989), What Moves Stock Prices Journal of Portfolio Management, 15, 4-12. DeLong B. Shleifer A. Summers L.H. and Waldmann R. J., (1990), Noise Trader Risk in Financial Markets, Journal of Political Economy, 98, 703-738 DeLong B. Shleifer A. Summers L.H. and Waldmann R. J., (1990), Positive Feedback Investment Strategies and Destibilising Rational Speculation, Journal of Finance, XLV, 379395. Engle R. and Granger C. W. S. (1987), Cointegration and Error Correction: Representation Estimation and Testing, Econometrica, 55,251-276. Fama E. (1970), Efficient Capital Markets: A Review of the Theory and Empirical Work, Journal of Finance, 25, 383-416. Fama E. (1976), Efficient Capital Markets: Reply, Journal of Finance, 3, 143-145. Godfrey M. D., Granger C.W. and Morgerstern O., (1964), The Random Walk Hypothesis of the Stock Market Behaviour, Kyklos, 17, 1-30. Granger C. W. (1969), Investigating Causal Relations by Econometric Methods and Cross Spectral, Econometrica, 37, 24-36. Granger C. W and Morgerstern O., (1963), "Spectral Analysis of New York Stock Market Prices", Kyklos, 16, 1-27. Harris L. and Garel E. (1986), Price and Volume Effects Associated with Changes in the S&P 500 List. New Evidence for the Existence of Price Pressures, Journal of Finance, 41, 815 829. Hiemstra C. and Jones D. D. (1994) Testing for Linear and Nonlinear Granger Causality in the Stock-Volume Relation Journal of Finance, 49, 1639-1664. Johansen S. and Juselius K. (1990), Maximum Likelihood Estimation and Inference on Cointegration - With Applications to the Demand of Money, Oxford Bulletin of Economics and Statistics, 52, 169-210. Karpoff J. M. (1987), "The Relation Between Price Changes and Trading Volume: A Survey", Journal of Financial and Quantitative Analysis, 22, 109-126. Lee, C.L., and Rui, O.M., (2000), Does trading volume contain information to predict stock returns? Evidence from Chinas stock markets, Review of Quantitative Finance and Accounting, 14, 341-360. LeRoy S. F. (1989), Efficient Capital Markets and Martingales, Journal of Economic Literature, XXVII, 1583-1621. LeRoy S. F. (1990), Capital Market Efficiency: An Update, Federal Reserve Bank of San Francisco Economic Review, Spring 1990, 29-40. McDonald R. and Kearney C (1987), On the Specification of Granger causality tests using the cointegration methodology, Economic Letters, 25, 149-153. Mills T. C. (1993), The Economic Modelling of Financial Time Series, Cambridge University Press.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
85 [24] [25] [26] [27] [28] [29] [30]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Moosa, I.A., and Al-Loughani, N.E. (1995), Testing the price-volume relation in emerging Asian stock markets, Journal of Asian Economics, 6, 407-422. Rogalski R. J. (1978), The Dependence of Price and Volume,The Review of Economics and Statistics, 36, 268 - 274. Roll R. (1988), R, Journal of Finance, 43, 541-556. Samuelson P.A.(1965), Proof the Properly Anticipated Prices Fluctuate Randomly, Industrial Management Review, 6, 41-49. Silvapulle, P. and Choi, J-S., (1999), Testing for linear and nonlinear Granger causality in the stock price-volume relation: Korean evidence, Quarterly Review of Economics and Finance, 39, 59-76. Weiss K. (1989), The Post-Offering Price Performance of Closed End Funds, Financial Management, 18(3), 57-67. Ying C. C. (1966), Stock Market Prices and Volume of Sales, Econometrica, 34, 676 - 686.
Relationship between Strategic Human Resource Management and Organizational Performance: Evidence from Selected Malaysian Firms
Raduan Che Rose Graduate School of Management, Universiti Putra Malaysia E-mail: rcr@putra.upm.edu.my Naresh Kumar Graduate School of Management, Universiti Putra Malaysia E-mail: naresh@putra.upm.edu.my Hazril Izwar Ibrahim Graduate School of Management, Universiti Putra Malaysia Abstract Based on the universal and contingency approach, the relationship between strategic human resource management (HRM) practices and organizational performance have been examined in the Malaysian electrical and electronic sector. The finding provides support for the universal perspective and the resource-based theory that a firms human capital pool that is embedded within a synergistic HRM practices will lead to a better organizational performance. It has also proven the idea propagated by institutional theorist that firms will adopt best practices to survive external pressures like economic uncertainty and the market environment. Keywords: Strategy, Human resource management, organizational performance, Malaysia
Introduction
The move towards market liberalization, preceded or forced by globalization, in many countries has been reflected in deregulatory policies by governments, including reduction of tariff barriers, facilitating the flows of capital and investment, and privatization of State owned enterprises. The phenomenon of globalization has been facilitated by the significant growth in world trade and foreign direct investment and by information technology which has facilitated rapid financial transactions and changes in production and service locations around the world (MacDonald, 1997). Firms of varying sizes, both domestic and multinational, are part of the dynamic, global environment, where business strategies, organizational structures and HRM policies have to be continually re- appraised to provide flexibility and innovation required for ongoing competitiveness. The process of globalization has brought with it the need for increased competitiveness at the level of the individual enterprise. A number of consequences resulted from the restructuring of the economy including relocation of factories to cheaper locations, unemployment and retrenchment in the manufacturing sector, re-training and disturbances in traditional industrial relations arrangements (MacDonald, 1997; Warner, 2001). According to De Cieri (2003) the impact of such volatile business
87
environment has led to varying outcomes across different countries and industries in Asia-Pacific, among them are reengineering, restructuring, retrenchment, recruitment and retention of employees. All the above factors requires for a different breed of organizational form and employee type in order to exploit available opportunities and succeed against fierce global competition (Perkins, 2003). One of the key challenges for organizations is to attract, retain and develop the right type of employees (Ashraff, 2002). According to Mendenhall, Black, Jensen and Gregersen (2003) people are the key to effective globalization and failure to identify the people problems results in faltering globalization efforts. The belief that human resources as an important source of competitive advantage has been gaining wide acceptance (Black & Lynch, 2000; Delaney & Huselid, 1996; Capelli & Newmark, 2001; Fey & Bjorkman, 2000; Harel & Tzafrir, 1999). Effective management of human resource is critical in the manufacturing sector, as the human effort constitutes the operational process. This has lead to increased interest in the impact of human resource management practices on organizational performance, and a number of studies have found positive relationship between High-performances HR practices or strategic HRM and different measures of organizational performance (Guest, Michie, Sheehan & Conway, 2000; Guest, Michie, Sheehan, Conway & Metochi, 2000).
Relationship between Strategic HRM and Organizational Performance

Universalistic Approach The relationship between HRM practices and organizational performance has been the subject of significant empirical examination (Arthur, 1994; Huselid, 1995; Huselid & Becker, 1996; Ichniowski, 1990; Pferrer, 1998). According to Delery and Doty (1996) the three following ways has been used in previous research to examine the effectiveness of HRM practices on firm performance: universalistic, contingent and configurational. Researchers in the universalistic perspective are micro-analytical in nature and posit that some HRM practices are always better than others and that all organizations should adopt these practices (Khatri, 2000). Guest and Peccei (1994) termed universalistic, contingency and configurational perspectives as internal fit (HRM as an ideal set of practices), external fit (HRM as strategic integration) and configurational fit (HRM as bundles). Khatri (2000) noted that there is empirical support for each of the three perspectives but consistently stronger support for the internal fit model. The connection between HRM practices and firms performance is supported by theoretical arguments from a number of disciplines, even though the underlying assumptions or arguments for the universalistic model may seem somewhat simplistic: that there is a linear relationship between HR practices or systems and organizational performance; that best practices are universally applicable and successful; and that organizational success is best measured in terms of the shareholders perspective such as financial performance indicators like, profits or by market share (Boselie, Paauwe & Jansen,2000). From micro-economics, human capital theory posits that employee have skills, knowledge and abilities that provide economic value to the organization (Becker, 1964). From strategic management and organizational economics, the resource- based theory of the firm suggests that human resources can constitute a source of competitive advantage for the firm (Rodriguez & Ventura, 2003; Wright, Smart & McMahan, 1995). Assuming heterogeneity among firms with respect to their human capital, competitive advantage is possible if a firm ensures that its people add value to its production processes and that its pool of human capital is a unique resource, both difficult to replicate and difficult to substitute for (Huselid, Jackson & Schuler;1997). There are considerable empirical supports for the universalistic predictions, Leonard (1990) found that organizations having long term incentives plans for their executives had larger increases in return on equity over a four year period compared to other organizations. Abowd (1990) proved that the degree to which managerial compensation was tied to an organizations performance was significantly correlated with future financial performance. Terpstra and Rozelli (1993) posited five best practices and concluded that the use of these practices had
88
positive relationship with organization- level outcomes. Arthur (1994) found that HR practices that focused on enhancing employee, such as decentralized decision- making, comprehensive training, salaried compensation and employee participation are related to higher firm performance. Huselid (1995) found that high performance work practices increased organizational performance through their impact on the employee skill development and motivation. At this juncture one might question how exactly the implementation of strategic HRM practices lead to better performance? These high- involvement practices are anticipated to be effective because they act as a synergy and have multiplicative effect. The commitment and flexibility provided by highly involving actions lead to behaviour changes among employees, because employees show high levels of motivation, co-operation and citizenship, they adopt better performing behaviours, leading to lower absenteeism and turnover rates, to better productivity and quality. Guest and Peccei (1994) associated this behaviour with social results (turnover, conflicts, absenteeism) and organizational results (productivity, quality of goods and services, customer complaints). Consequently, these types of performance were supposed to affect financial performance (Guerrero & Barraud Didier, 2004). A survey of 97 manufacturing plants in the US by Youndt, et al. (1996) found significant relationship between enhanced HR systems and manufacturing performance of the plants and the findings provide support for HR system that focused on enhancing human capital is a valuable approach for strengthening operational performance in manufacturing. Arthur (1994) conducted a study of 30 US steel minimills where the result of the showed that the mills with commitment systems had higher productivity, lower scrap rates and lower employee turnover than those with control system. In another study, Delaney and Huselid (1996) found that practices consistent with a high- involvement HRM strategy, such as highly selective staffing, incentive compensation and training were positively linked to organizational performance, however their result did not support that complementarities among HR practices enhanced firm performance. They contended that the lack of support could be due to the lack of development of reliable and valid measures of progressive HRM practices. Huselid, et al. (1997) in their survey of 293 US firms made a distinction between technical HRM and strategic HRM perspective by technical HRM; they referred to the alignment of HR practices to the interest of the companys stakeholders. These stakeholders include the government and various professional organizations that regulate a wide range of employment practices. Strategic HRM practices refer to a set of internally consistent policies and practices that ensure a firms human capital contributes to the achievement of its business objective. They found that large firms in the U.S. are more proficient in their technical HRM than their strategic HRM practices. In addition to this finding, they noted significant correlation between strategic HRM effectiveness and employee productivity, cash flow and market value. Bae and Lawler (2000) in their study of 138 firms in Korea found that firms that place high value on their HRM and consider their employees as source of competitive advantage were more likely to have high involvement HRM practices and in addition, these high- involvement practices result in better firm performance. Bayo-Moriones and de Cerio (2002) in their survey of 965 factories in Spain provided evidence of the presence of a positive, statistically significant correlation between the adoption of high commitment HR practices and operational performance and the study further confirmed that the result is universal and not dependent on the strategy used by firms. Richard and Johnson (2001) discovered that SHRM effectiveness is unrelated to productivity but is correlated to lower turnover. Numerous other studies have similarly shown positive relationships between strategic HRM practices and firm performance (e.g. Delery & Doty, 1996; Ichniowski, 1990). Michie and Sheehan (2003) in their survey of 362 firms from both manufacturing and service sector in the UK found that Strategic HR HR practices are significantly correlated with all the firms performance measures. Rodriguez and Ventura (2003) in their study of 120 manufacturing firms in Spain found that the implementation of a make HR system has a significant and positive effect on the organizations overall performance and this leads to the enhancement of the employees commitment to the firm.
89
Khatri (2000) in a survey of 222 firms in Singapore found that the hypothesis where high performance HR practices are positively associated with firm performance to receive only modest support as it was found that it only has marginal direct effect on sales and non- financial performance. Wan, Ong and Kok (2002) in a study consisting of 191 firms in Singapore found that high performance work practices to be positively related to firm performance. Datta, Guthrie and Wright (2003) found in their study of 132 manufacturing firms in United States that the use of high performance work practices indicated is positively associated with firm productivity. Guerrero and Barraud-Didier (2004) in their study on 180 firms in France, found that high-involvement practices as a bundle or system is significantly and positively related to social and organizational performance, lending further support for the universal approach. As observed by Armstrong (2001) in contrast to the US- oriented researches, UK researches (e.g. Guest & Peccei, 1994) apply a stakeholder perspective or pluralist framework and include such outcomes as absenteeism, employee turnover, commitment, motivation, satisfaction, trust, conflicts and social climate. Most of the UK academics seems to be skeptical of the typical American Dream view on the existence of best practices emerging from universal modeling and Guest et al. (2000) is of the opinion that the use of productivity or financial performance indicators is attractive if only to remind them of past neglect of a potentially untapped source. This approach has also been open to criticism; it was criticized as a manipulative management exercise to increase control, stress and efforts for the good of the firm performance but not for the employees (Pil & MacDuffie, 1996). Chang and Huang (2005) in their survey of 380 Taiwanese manufacturing firms found no support for the universal approach, contributing this result to the very different cultural and institutional environment. Contingency Approach In accordance with the contingency approach, the link between HRM practices and organizational performance is contingent on other aspects of the organization. The organizations strategy is considered to be the primary contingency factor in the strategic human resource management literature (Delery & Doty, 1996). The presupposition implicitly implied is that the alignment of strategy and HRM practices allows organizations to achieve superior performance (Youndt, et al., 1996). Contingency predictions are more complex than universalistic approach because contingency predictions imply interactions rather than the simple linear relationships incorporated in universalistic theories, where the organizations strategy is considered to be the prime contingency factor (Delery & Doty, 1996). According to Youndt, et al. (1996) each strategy implies something different about the potential role of human resources in improving firm performance, suggesting that the best HRM practices is contingent on the manufacturing strategy of a firm. Miles and Snow (1984) are of the opinion that a tighter fit between HR practices and strategy leads to superior performance. The concept of fit refers mainly to the close linkage of HRM and the firms strategies in ways that will help retain and motivate employees, moreover, application of the strategic fit concept help firms to manage their resources more efficiently, so they can reduce operational costs and respond effectively to environmental restraints and new opportunities (Tung, 2001). Wright (1998) point out that the notion of fit in SHRM implies a static, stationary contextual template to which practices and people can be fitted. Datta, Guthrie and Wright (2003) argued that in reality, the task environment faced by firms vary according to the competitive environment, where firm facing more dynamic environments may find greater value in investing in high performance work practices promoting acquisition, development and motivation of individuals who are able and willing to adapt to the needs of the environment. Thus, the contingency approach based on the behavioural theory requires that a firm select a strategy and then align the HR practices with the firms strategy to improve organizational performance (Delery & Doty, 1996). Similarly control theory, which bears similarities with behavioural theory, contends that effective performance depends on matching appropriate HRM
90
practices with the administrative established by a given strategy. Although the behavioural perspective and control theory tend to focus attention on managing a firms current employees behaviour in an effort to maximize performance, firms also focus on competency management through acquiring, developing and utilizing employees with particular knowledge, skills and abilities (Wright, 1998; Wright, et al., 1995). According to Bayo- Moriones and de Cerio (2002) the literature does not make it quite clear in what direction this relationship actually works. Miles and Snow (1984) argue that defender firms may, in their effort to achieve stability through efficiency, find it better to establish along term relationship with their workers, by offering them job security, establishing internal labour markets and investing in their training. This would reduce the turnover of employees, thereby making reactions on the part of the workforce easier to predict. In contrast to this, prospector firms have difficulty in investing in their workforce, since they cannot be sure of the type of demands they will have to make on them in the future. Arthur (1994) and Youndt et al. (1996) using competitive priorities established by Porter (1980) suggested the opposite relationship compared to those of Miles and Snow (1984). Arthur (1994) claims that a cost leadership strategy is hardly likely to benefit from the introduction of innovative HRM practices. Firms attempting to compete on the market on cost leadership will try to keep their costs to a minimum, a feat that is achievable via a traditional style of management. This enables standardized goods to produce through strict division of labour, thereby keeping costs low, since by not requiring highly- qualified personnel, it reduces the need for worker training. This makes employees easily replaceable, thereby eliminating the need for the firm to increase wages in order to retain them and the cost of setting up and running a high- performance programs can be avoided (Bayo- Moriones & de Cerio, 2002). However, firm that chooses to adopt a differentiation (flexible) strategy must be flexible enough to adapt in order to meet the various demands of their clients (Arthur, 1992). In this scenario, the employees will be required to carry out a variety of activities and they are likely to be able to make their own decisions (Youndt et al., 1996). Therefore extensive training is essential and they must also be sufficiently motivated to make the choices that will serve the best interest of the firm. This can only be achieved by implementing a high- commitment, high- performance HRM practices rather than through traditional practices (Bayo- Moriones & de Cerio, 2002). Youndt et al. (1996) and Hoque (1999) found that human- capital enhancing systems or high- commitment management practices have stronger effects on performance when coupled with a quality strategy. Wright, et al. (1995) found that organizations exhibited higher performance when they recruited and acquired employees possessing competencies consistent with the organizations current strategies. Reversing the causal arrow, they also found that organizations exhibited higher performance when they sought out a strategy that matched their current employees competencies. Bird and Beechler (1994) examined the linkages between business strategy and human resource management strategy in Japanese subsidiaries in the U.S., they found that subsidiaries with matched combinations did outperform their unmatched counterparts, particularly with regard to HRM- related outcomes like employee morale, tenure, promotion and turnover (Wan, et al., 2000). Bird and Beechler (1995) further established that employee performance in firms that successfully adopted the strategic fit concept was significantly better than in firms that did not do so, however the difference was not significant. This approach implies an interaction effect involving organizational and HRM practices, but to date the evidence in favour of such a contingency perspective has been mixed at best (Bae & Lawler, 2000). Delery and Doty (1996) also found modest support for fit with the Miles and Snow typology. Huselid (1995) concluded that the adoption of high performance work practices is more important that any efforts to ensure that HRM practices are aligned with the firms competitive strategy. Similarly MacDuffie (1995) found no evidence to suggest that a fit between appropriate HRM practices and mass production was able to compete with specific production methods (Tung, 2001). Fey and Bjorkman (2001) in a survey of 101 multinational corporations in Russia found that HRM strategy alignment to be significantly related to organizational performance. Richard and Johnson
91
(2001) in their study on the American banking industry found that SHRM alignment with capital intensity as the firms strategy increase both firm productivity and ROE. Khatri (2000) in a survey on 222 firms in Singapore found strong support for the hypothesis that the relationship between HR practices and firms performance is moderated by organizational strategy. The interactions of HR practices and strategy showed highly significant effects across all the performance measures. Michie and Sheehan (2003) in their survey of 362 manufacturing and service firms in the UK found that in quality enhancing firms, there is a strong positive and significant relationship between the HR practices and all the performance measures. However, in cost reducer firms it was found that the statistical significance of this relationship disappears for labour productivity and financial performance and becomes negative for sales growth. These results gives support to the fit hypothesis that the effectiveness of HR is strongly dependent upon the business strategy pursued (Michie & Sheehan, 2003). However, there is considerable evidence from studies in United States, which stated that, the integration or linking of HRM with the firms strategy is in practice rare, even among the larger and more established corporations (Springer & Springer, 1990; Kochan & Osterman, 1994). BayoMoriones and de Cerio (2002) in their survey of 965 factories in Spain found no support for the contingency approach as strategic HRM practices on its own were capable in influencing the organizational outcomes. The relatively weak support for the effects of HRM practices strategy alignment on organizational outcomes should come as no surprise as it is difficult to specify what constitutes good alignment across firm and industries (Becker et al., 1997). Barney (1991) suggests that the appropriate configuration and strategic alignment of HRM practices may be idiosyncratic and complex. Furthermore, Hiltrop (1996) contends that the whole idea of fit within a certain strategy seems inappropriate for a world in which there are high levels of dynamic and unpredictable change. Taking into account the debates regarding the strategy construct in the contingency perspective of strategic HRM, Wright and Sherman (1999) noted that very little empirical research has supported the efficacy of the concept of fit between strategy and HR practices. With Pfeffer (1998) contending that it is HRs best practices that impact performance and not their fit with the firms strategy (Wright & Gardner, 2000). Wright (1998) reasoned that while some practices (e.g., performance- based pay, rigorous selection, etc.) might be universally effective, the efficacy of fit comes more at the product level (i.e., pay promoting the kind of performance or selection system selecting the right kinds of people, given a particular strategy). Becker et al. (1997) support his idea, arguing that fit between HR and strategy might be best achieved at the level of competencies required for a strategy. Despite, the debates regarding the contingency approach, Wright and Gardner (2000) stated that infusing the strategy construct into a research can aid in understanding HR practices and also how an alignment of these practices with the strategy can provide a significant incremental effect on performance. The issue here is whether HRM practices implemented and practiced by MNCs and local firms in the electrical and electronic sector can be deemed to be strategic in nature, to face the challenges posed by globalize economies as the current business environment requires a workforce that are versatile and adaptable to changing technological and industrial environment. Hence, a strategic human resource management practices will be essential to create a more knowledgeable and skillful human capital to sustain a firms competitive advantage. Therefore this study was initiated to identify the prevailing HRM practices implemented by firms in the electrical and electronic sector. The relationship between strategic HRM practices and organizational performance also has been studies. Based on the universal and contingency approach, the variable organizational performance was used to explain and predict the effectiveness of such relationship.
92
Research Method
A structured questionnaire was employed to gather data for the present study. The questionnaire consisted of two sections. This first section contains 6 items adapted from Youndt, et al. (1990) to measure organizational performance, in this instance the perception of operational performance. Respondents were required to assess the firms current performance compared to other firms in the industry. Assessment was made on: product quality, inventory management, employee productivity, equipment utilization, production lead time and scrap minimization. The five point Likert scale indicated 1= worst in the industry and 5= best in the industry. The second section contains 40 items to measure strategic human resource management practices which were adapted from Delery and Doty (1996); Huselid et al. (1997); and Bae and Lawler (2000). This measure requires the HR manager to state the extent these HR practices were being implemented in the particular firms. Responses will be measured on a five point Likert scale ranging from 1= never to 5= always. In order to meet the representativeness criteria of a good sample collection, the questionnaires for the study were posted to all 700 companies in the electrical and electronics manufacturing sector. The database for the population was derived from the Federation of Malaysian Manufacturers MATRADE Electrical and Electronics Industry Directory. Apparently, this study is conducted in a single industry. The database was based on the directory, specifically because it has the most comprehensive industry coverage and an updated list. Out of the 700 firms that had been posted the surveys, only 121 responses were received. Thus, the response rate of survey was 17.2 %. This response rate is not different from other surveys in Malaysia, which tend to obtain a response rate of 15-25 per cent (Sarachek & Aziz, 1983; Rozhan, 1998; Hazman, 1998; Kanapathy & Jabnoun, 1998; Rozhan, Rohayu & Rasidah, 2001).
Findings and Discussion

To determine the relationship between manufacturing strategy and organizational performance, the SPSS procedure for ordinary least square regression, ANOVA (analysis of variance) was used with the maximum R2 option. The model summary (Table 1) showed an R = 0.090, meaning SHRM explained 9% of the variance in organizational performance. The ANOVA table indicated statistical significance (p value= 0.001, p< 0.05). This is further supported by Beta= 0.299, t value= 3.424, sig= 0.001, p< 0.05. The regression indicated that SHRM practices have significantly impact on organizational performance.
Table 1: Relationship between SHRM practices and Organizational Performance
B 0.418 Standard Error 0.122 Beta 0.299 Standard Error= 0.52280 t 3.424 Sig 0.001
SHRM R= 0.090 F value= 11.724 Sig. F= 0.001
While the finding on the direct influence of SHRM practices on organizational performance (B= 0.299, P < 0.01) showed significant result but the R= 0.09, meaning only 9 % of the variance was explained was quite low. However the significant relationship between SHRM practices and organizational performance, viewed in the Malaysian context, is encouraging as it is in agreement with the findings from prior studies. The finding provides support for the universal perspective and the resource- based theory that a firms human capital pool that is embedded within a synergistic HR practices will lead to a better organizational performance. It has also proven the idea propagated by institutional theorist that firms will adopt practices to survive external pressures like economic uncertainty and the market environment. Furthermore, the significance of the relationship also provided support for the measurement of a set or bundle of SHRM practices instead of measuring HR
93
practices individually as it was done in several previous studies for e.g. Tepstra and Rozell (1993). It also promotes that the notion of a bundle or set of SHRM practices are more effective than individual HR practices, because of the overlapping and mutually reinforcing effect of multiple practices.
Conclusion
The significant relationship between SHRM practices and organizational performance is encouraging and is in agreement with findings from prior studies (Huselid, 1995, Wan et al, 2000; Khatri, 2000). This finding lend a strong support to the universal perspective that a set of HR practices on its own are able to influence firm performance. It also provided support to the resource- based theory and institutional theory. The evidence provided by this study may justify the need for firms to procure or develop the resources needed to implement the strategic HR practices. In order to obtain the resources needed to enhance their efficiency, firms must understand how these activities affect their workforce. Implementing these SHRM practices without the support of the employees may render them ineffective or these practices may be outsourced to external parties. The result underlines the value of creating challenging and enriched activities such as open communication and power sharing to change the firms management style to improve the effectiveness of their HR practices and consequently performance (Guerrero & Barraud- Didier, 2004). While the results indicated that many of the respondents may actually have strategic HR practices but these practices might not have been effectively implemented. Richard and Johnson (2001) contributed this to half- hearted implementation by the managers simply because it represents the latest management fad than because it is sincerely believed to work within the organizational context to improve workforce effectiveness. However, it also important for firms to evaluate which practices may significantly contribute towards the firms performance because in the end what actually matters is the quality or effectiveness of the practices and not the quantity.
Limitation of the Study and Suggestions for Future Research

The study is conducted within a single sector of the manufacturing industry; while prior study by Michie and Sheehan (2003) contended that a study in a single sector would be add more value to the findings, there is also the question of generalizability and applicability to other sectors in the industry. Therefore it would be better for future study to obtain a cross- industry sample this even more relevant in the case of firms competing in globalized environment. SHRM measurement was measured based on the perspectives of a single respondent that is the HR manager, the problem with this is that it does not take into account the viewpoint of the employees. Therefore the result is open to potential bias; in the future it would be more objective to measure the concept from the viewpoint of multiple respondents, both the employees and the management. Two perspectives; universal and contingency perspectives, in the study of HRM were identified and studied. Future research will need to explore empirically whether firms integrate HRM issues in their strategic process, how they do it and whether the more proactive approach improves organizational approach. Future studies should also analyze configurational perspective of HRM perspectives and its effect on firm performance. This study utilized subjective manufacturing performance measures to measure organizational performance as the study is related to the manufacturing environment. Future studies in this area can look beyond the measures used in the present research in order to enhance the understanding of the impact of HR practices on different measures of performance. Another issue related to performance as the dependent variable in HRM research is the focus on who stands to gain from improved firm performance. In most cases, as in prior studies and the present study, the benefit of improved firm performance focused on shareholders or owner of the businesses. A
94
significant extension would be to consider the potential benefit of SHRM practices to other parties other than the shareholders (Huseselid, 1995). The present study failed to discover the moderating effects of the organizational context variables in their interaction with SHRM practices to influence organizational performance. Perhaps data collected across industry with a wider range of firm sizes, age, more varied firm nationality and larger coverage of union, will be able to capture some moderating effects. Finally, since this is a cross- sectional study that sought to capture the character of the practices and perception at a point of time; the questionnaire failed to capture the trends and changes the firms faced due to the highly volatile business environment. The design of future research would be significantly improved with the inclusion of interviews as a follow up to the questionnaire as this would provide more insightful views on the relevant trends and changes that firms undergo to remain competitive.
95
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Abowd, J.M. (1990) Does performance-based compensation affect corporate performance? Industrial and Labour Relations Review, 43: 52-73. Arthur, J.B. (1994) Effects of human resource systems on manufacturing performance and turnover. Academy of Management Journal, 37, 670-687. Ashraf,A.R.(2002) Effects of economic reform on human resources management in Egypt. Paper presented at Institutional and Policy Challenges Facing the Egyptian Economy, Cairo, Egypt. Armstrong, M. (2001) Human Resource Management Practices, Kogan- Page. Bae, J. & Lawler, J.L. (2000) Organizational and HRM strategies in Korea: Impact on firm performance in an emerging economy, Academy of Management Journal, 43(3): 502- 517. Barney, J.B. (1991) Firm resources and sustained competitive advantage. Journal of Management, 17: 99- 120. Bayo- Moriones, A. & de Cerio, J.M.D. (2002) Human resource management, strategy and operational performance in the Spanish manufacturing industry, Management, 5: 175-199. Becker,G.S. (1964) Human Capital, Natinal Bureau for Economic Research, New York. Becker, B.E., Huselid,M.A., Pickus, P.S. & Spratt,M.F. (1997) HR as a source of shareholder value: research and recommendations. Human Resource Management, 36: 39- 47. Black.S.E. & Lynch, L.M. (2000) Whats driving the new economy: the benefits of workplace innovation, NBER Working paper series no. 7479, Cambridge, National Bureau of Economic Research. Boselie,P., Pauuwe, J. & Jansen, P. (2000) Human resource management and performance: Lessons from the Netherlands, Research in Management, 46. Cappelli, P. & Neumark, D. (2001) Do high performance work practices improve establishment level outcomes? Industrial and Labor Realtions Review, 54: 737- 775. Chang, W.J. & Huang, T.C. (2005) Relationship between strategic human resource management and firm performance. International Journal of Manpower, 26: 434- 449. Datta,D.K., Guthrie,J.P. & Wright, P.M. (2003) HRM and productivity: does industry matter? Working paper for Center for Advanced Human resource Studies, Cornell University. De Cieri, H. (2003) International human resource management: Asia- Pacific Challenges. Working paper series, Monash University. Delaney, J.T. & Huselid M.A. (1996) The impact of human resource management on perceptions of organizational performance. Academy of Management Journal, 39 (4), 949- 969. Delery, J.E. & Doty, D.H. (1996) Modes of theorizing in strategic human resource management: test of universalistic, contingency and configurational perspectives. Academy of Management Journal, 39, 802- 835. Fey, C.F. & Bjorkman, I. (2000) The effect of human resource management practices on MNC subsidiary performance in Russia, SSE/EFI working paper series in Business Administration, Sweden. Guerrero, S. & Barraud- Didier, V. (2004) High involvement practices and performance of French firms. International Journal of Human Resource Management, 15: 1408- 1423. Guest, D.E. & Peccei, R. (1994) The nature and causes of effective human resource management. British Journal of Industrial Relations, June, 219- 242. Guest, D.E., Michie,J, Sheehan, M. & Conway, N. (2000) Employment relations, human resource management and business performance. In Armstrong (2001) Human Resource Management Practices, Kogan -Page. Guest, D.E., Michie, J, Sheehan, S. & Metochi, M. (2000) Effective people management: Initial findings of the future work survey. In Armstrong, M. (2001) Human Resource Management Practices, Kogan- Page.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22]
96 [23] [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Harel,G.H. & Tzafrir S.S. (1999) The effect of human resource management practices on the perceptions of organizational and market performance, Human Resource Management, 38(3): 185-200. Hazman,S. (1998) The level of participation and influence of HR managers in the strategic management process in Malaysian corporations. Malaysian Management Review, 33:47- 60. Hiltrop,J.M. (1996) The impact of human resource management on organizational performance: theory and research. European Management Journal, 14: 628- 637. Hoque, K. (1999) Human resource management and performance in the UK hotel industry. British Journal of Industrial Relations, 37:419- 443. Huselid,M.A. (1995) The impact of human resource management practices on turnover, productivity and corporate financial performance. Academy of Management Journal, 38, 635672. Huselid, M.A. & Becker, B.E. (1996) Methodological issues in cross- sectional and panel estimates of the human resource- firm performance link. Industrial Relations, 35, 400-422. Huselid,M.A., Jackson,S.E. & Schuler, R.S. (1997) Technical and strategic human resource effectiveness as determinants of firm performance. Academy of Management Journal, 40: 171188. Ichniowski, C. (1990) Human resource management systems and the performance of U.S. manufacturing businesses. NBER Working paper series no. 7479, Cambridge, National Bereau of Economic Research. Kanapathy,K. & Jabnoun,N. (1998) Are ISO 9000 and TQM programmes paying off for Malaysian manufacturing companies. Malaysian Management Review, 33: 40- 46. Khatri, N. (2000) Managing human resource for competitive advantage: a study of companies in Singapore. International Journal of Human Resource Management, 11: 336- 365. Kochan,T. & Osterman,P. (1994) The Mutual Gains Enterprise, Harvard Business School Press, Cambridge. Leonard, J.S. (1990) Executive pay and firm performance. Industrial and Labour Relations Review, 43: 13- 29. Macdonald, D. (1997) Industrial relations and globalization: Challenges for employers and their organizations. Paper presented at the ILO Workshop on Employers Organizations in Asia Pacific in the 21st century, Turin, Italy. Macduffie, J.P. (1995) Human resource bundles and manufacturing performance: organizational logic and flexible production systems in the world auto industry, Industrial and Labour Relations Review, 48: 197- 221. Mendenhall, M.E., Black,J.S., Jensen,R.J. & Gregersen H.B. (2003) Seeing the elephant: human resource management challenges in the age of globalization. Organizational Dynamics, 32: 261- 274. Michie, J. & Sheehan, M. (2003) Business strategy, human resources, labour market flexibility and competitive advantage. Paper for The Leverhume Trust, University of London Central Research and the University of Dallas. Miles,R.E. & Snow, C.C. (1984) Designing strategic human resource systems. Organizational Dynamics, 131: 36-52. Perkins, S. (2003) Globalization and IHRM: partners in comparative perspective. Journal of European Industrial Training, 27: 461- 472. Pil,F.K. & MacDuffie,J.P. (1996) The adoption of high- involvement work practices. Industrial Relations, 3: 423- 455. Pfeffer, J. (1998) Seven practices of successful organizations. California Management Review, 40: 96- 124. Richard, O.C. & Johnson, N.B. (2001) Strategic Human resource management effectiveness and firm performance, International Journal of Human Resource Management, 12: 299- 310.
97 [44] [45] [46] [47] [48] [49] [50] [51] [52] [53] [54] [55] [56]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Rodriguez, J.M. & Ventura, J. (2003) Human resource management systems and organizational performance: an analysis of the Spanish manufacturing industry, International Journal of Human Resource Management, 14(7): 1206- 1226. Rozhan,O. (1998) Human resource management practice of service organizations: Evidence from selected Malaysian firms. Journal of Asia- Pacific Business, 2: 65- 81. Rozhan,O., Rohayu,A.G. & Rasidah,A. (2001) Great expectations- CEOs perception of the performance gap of the HR function in the Malaysian manufacturing sector. Personnel Review, 61- 80. Sarachek, B. & Aziz, A.H. (1983) A survey of Malaysian personnel practices and problems. Jurnal Pengurusan, 2: 61-79. Springer, B. & Springer, S. (1990) Human resource management in the UK celebration of its centenary. In Peiper,R. (ed) Human resource management: an international comparison. Walter de Gruyter, Berlin. Terpstra, D.E. & Rozell, E.J. (1993) The relationship of staffing practices to organizational level of measures of performance. Personnel Psychology, 46, 27-48. Tung, C.H. (2001) the effects of linkage between business and human resource management strategies, Personnel review, 30: 132- 151. Wan, T.W.D., Ong, C.H. & Kok, C.F.V. (2002) Strategic HRM and organizational performance in Singapore. Compensation and Benefits Review, 34: 33- 42. Warner, M. (2001) Globalization, labour markets and human resources in Asia- Pacific economies: an overview. Working paper for the Judge Institute of Management Studies, University of Cambridge, www.jims.cam.ac.uk. Accessed on 30/9/2003. Wright, P.M. (1998) Introduction: Strategic human resource management in the 21st century, Human Resource Management Review, 8:187- 191 Wright, P.M. & Gardner, T.M. (2000) Theoretical and empirical challenges in studying the HR practice- firm performance relationship, Working paper for Center for Advanced Human Resource Studies, Cornell University. Wright, P.M., Smart, D.L. & McMahan, G.C. (1995) Matches between human resources and strategy among NCAA basketball teams. Academy of Management Journal, 38: 1052- 1074. Youndt, M.A., Snell, S.A., Dean, J.W. & Lepak, D.P. (1996) Human resource management, manufacturing strategy and firm performance. Academy of Management Journal, 39 (4), 836866.
The Reaction of Bank Lending to Macroeconomic Fluctuations of Monetary Policy Transmission in Greece
Aikaterini Markidou South East European Research Center (SEERC) A Research Center of the University of Sheffield and City Liberal Studies 17 Mitropoleos Str, 54624 Thessaloniki, Greece E-mail: amarkidou@seerc.org Eftychia Nikolaidou City College, Affiliated College of the University of Sheffield Business Administration and Economics Department, 2 Kalapothaki Str 54624 Thessaloniki, Greece E-mail: enikolaidou@city.academic.gr Abstract This paper investigates the relevance of the credit channel of monetary policy transmission in Greece by employing a SVAR model on both aggregated and disaggregated data and estimating the response of bank loans to different macroeconomic shocks. By distinguishing between households and firms instead of focusing on the response of total bank credit to a tightening monetary policy shock and by employing a SVAR methodology following the work by Safaei and Cameron (2003), this paper identifies structural models to study bank credit in Greece as a source of macroeconomic variation for the period 19802005. The findings suggest that the credit channel in Greece for the period 1980-2005 is inoperative when monetary base is considered as the main monetary policy variable. On the other hand, when interest rate is used to capture the role of monetary policy variable, there is promising evidence that the credit channel might be present and bank credit to individuals seems to be more vulnerable compared to bank credit to firms. Keywords: Monetary policy transmission; bank lending; SVAR model; Greece JEL Classification Codes: E44; E52; G21
1. Introduction
Monetary policy transmission mechanisms have been a subject of theoretical and empirical research over the last two decades in an effort to better comprehend how monetary policy affects the real economy1. A central point of this research has been the role played by banks in the transmission of monetary policy, inclusive of the effect of monetary policy on bank lending. Despite the general agreement on the active role of banks in the transmission of monetary policy, there is an extensive dispute over the exact role that banks play. The basic idea behind the concept of the credit channel is that central bank impulses affect output, as an upshot caused by shifts in the supply of loans. This comes in contrast with the traditional money view where there is no reference to loan supply shocks.
1
See Bernanke and Gertler (1995) and Kashyap and Stein (1993,1997) for excellent surveys.
99
According to the money channel theory the focus is placed on the effects of monetary policy on the attraction of investment and saving that predominate when financial markets are complete. The major shortcoming of money view is the considerable difficulty in the identification of a quantitatively meaningful effect on aggregate spending and investment that the theory indicates it should influence. The credit channel has emerged to fill the gap. Even though the credit channel and the money channel share a common starting point in highlighting the relevance of financial considerations, they are reckoned to be complementary, which implies the simultaneous coexistence of the two transmission channels. Two subchannels of the credit channel have been presented in the literature, the balance sheet channel (BSC) and the bank lending channel (BLC). The BSC states that shifts in monetary policy can severely affect the creditworthiness of borrowers due to information asymmetries that consequently affect the bank loan supply and thus borrowers investment and spending decisions stemming from the inability of borrowers to raise funds from any provider. The BLC, in turn, rests on the assumption that banks have a dual nature as holders of reserved-backed deposits and as originators of loans and this can affect the availability of bank loans. It is generally accepted that commercial banks are the major source for investment expenditure financing and consumption financing of small firms and households, particularly valid for most of the euro area countries that are prevailed by rather small-sized banks and strong customer relationships. The structure of a banking system is of great importance since it can create the required conditions for a BLC to be operative. The credit channel theory has considerable appeal in Greece due to the dominance of small size bank-dependent firms with no alternatives to nonbank finance and the specific features of the Greek financial system. The Greek capital market was rather restricted compared to other EU countries and the Stock Exchange was significantly immature until mid 1990s. Moreover, banks were to a great extent government regulated and characterised by financial interventionism, restricting particular economic sectors from accessing bank credit, imposing restrictions on bank capital flows and reserve requirements as well. Mutual funds experienced considerable growth only towards the end of the 1990s after the systematic liberalisation of the financial markets2. As a result, it is easily anticipated that the Greek banking sector is of primary importance for the transmission of monetary policy and that the bank lending channel (BLC) exerts a dominant role particularly during the period prior to the liberalisation and transformation of the banking system in 1996. Early literature has investigated the possible repercussions of monetary policy by studying the relationship between money and output as well as bank loans and output mainly by using correlations or through Granger-causality tests. It has approximately deduced in favor of the money channel. During the 1990s numerous studies have attempted to assess the existence of the BLC. Although the magnitude of financial market imperfections the credit view in general- has been recognized by a substantial number of surveys, the empirical evidence for the existence of a BLC is much less definite. A very influential study that made use of aggregated US data and supports the BLC is that of Bernanke and Blinder (1992). Other SVAR studies supported the presence of the credit channel are that of Kashyap et al. (1993), Gibson (1997) for the U.S. where the composition of bank balance sheets has an impact on the strength of the transmission mechanism, whereas Carlino and DeFina (1999) following the theory proposed by Peek and Rosengren (1995), offered contrasting outcomes revealing that the credit channel is not present at the state level and greater concentration of small banks minimise responsiveness to monetary policy shocks at the level state. This is in contrast to the theory advocated by Kashyap and Stein (1994)3. Ramaswamy and Sloek (1998) and Clements et al. (2001) reported that a credit channel is present in most EU countries. Output responses of core countries react slower but deeper than in periphery countries. It is interesting to notice for instance that Gerlach and Smets
2 3
For more details regarding the deregulation of the Greek banking sector see Kashyap and Stein (1997) and Brissimis et al. (2001). The hypothesis of Kashyap and Stein (1994) states that small banks can easier confront higher costs in attracting CDs and other non-deposit sources to counteract a loss in reserves resulting from monetary policy tightening. Consequently, small banks are forced to contract bank loans by relatively more than large banking groups.
100
(1995), Ehrmann and Worms (2001) revealed similar results against the efficacy of the credit channel in Germany while Hortemoller (2003) offers a contradictory outcome. Agung (1998) observed that the BLC is operative in Indonesia and consumer loans of all bank categories drop following monetary contraction. On the other hand, findings of the empirical part of Bacchetta and Ballabriga (2000) and Dedola and Lippi (2000) lie between the two opposite findings providing fairly supportive evidence for the BSC and inconclusive or partially consistent for the BLC. This study attempts to analyse a macro-dynamic system utilising the SVAR approach that examines different bank credit measures (heterogeneity of the loan types) in Greece. Even if there are some drawbacks in the use of macro aggregated data, empirical research based on such data might still yield valuable insights4. So far, tests of the credit channel based on the importance of bank and other macroeconomic characteristics in determining the response of loans to a shift in monetary policy have been limited to U.S. and other developed EU countries data. To our knowledge, studies involving macroeconomic time-series for the case of Greece have not attracted much attention by researchers due to data deficiencies. The only Greek study on the BLC is that of Brissimis et al. (2001) however it makes use of bank level panel data and analysed the effects of bank characteristics in terms of size and liquidity on the supply of bank loans. The remainder of the paper is organised as follows. Section 2 describes the data and presents the SVAR model. Subsequently, the estimated structural relationships are set out in section 3 along with the impulse response functions to the structural shocks and the variance decompositions. Section 4 provides concluding remarks.
2. The Model
This section summons up the specification of the benchmark models that have been employed by Safaei and Cameron (2003). The small-scale models include 5 variables to capture the economy and the domestic sector of Greece. Although a larger SVAR that makes use of more variables may allow for more affluent interactions, a more parsimonious model is likely to be easier to estimate and more constant and more reliable to grasp key macroeconomic interactions. Impulse responses of macroeconomic variables to monetary shocks are estimated so as to investigate the credit view of monetary policy for the case of Greece over the period 1980-2005, A0t = ut (1) where A is the matrix of short-run coefficients, t is the reduced form residual resulting from the first step estimation of the VAR model and ut is the independent structural disturbance term of the structural model. The A matrix contains n2 parameters. Hence, in a contemporaneous version of a five variable model implying five equations, n2 (n+1) n/2 = n (n-1)/2 unique elements or to put it differently ((52 5)/2 = 10) restrictions are required to identify the structural parameters (orthogonal shocks) on the covariance matrix. Therefore, just-identification of the parameters in matrix A calls for exactly 10 restrictions (9 zero restrictions and an equality restriction) on the off diagonal elements of the matrix. In order to set the short-run restrictions that are necessary for the identification of the structural model, economic theory is applied for this purpose. In this way, the aggregate demand relationship is included in the contemporaneous restrictions. It is generally accepted that IS shocks (uy) have a straightforward negative impact to real output which is greater comparing to that of price level. Taking into consideration shocks in output growth resulting from shocks in the growth rate of real credit (ci p) is central to the theoretical underpinning of credit constraints that arise from the ability and willingness of banks in the bank lending procedure along with the cost of borrowing money which severely affects both the level of investment, real economic activity and the level of output5.
4
The limitations inherent in studies based on aggregated data are associated with the identification problem, whether or not the cutting off of credit following a monetary contraction is a result of deterioration in bank supply or decline in borrowers demand for bank credit. Vermeulen (2002), Hubbard (1998), Boyd and Smith (1997), Oliner and Rudenbusch (1996) and Greenwald and Stiglitz (1993) are indicative examples in the literature providing empirical evidence of asymmetric effects resulting
101
Furthermore, the aggregate supply function is embraced, describing the shifts in price level and the corresponding impact in inflation rate stemming from shocks in aggregate demand. The thought behind shocks in aggregate demand is compatible with the short-run Philips curve, thus it is logical to recognize that inflation shocks (up) entail economic changes that drastically affect the cost of inputs derived from adverse shocks associated with factors of production such as oil prices, energy costs, unemployment rate, tax regime or technological advances (i.e. information processing and communications technology)6. In addition, a policy reaction function caused by the monetary policy decisions set by the Central Bank should be considered as well. In particular, two policy variables are taken into account, namely monetary base and interest rate that make distinction between two classes of models (monetary base - B models and interest rate R models). As far as Class B models are concerned, nominal reserve money responds in shifts of real output, price level as well as M1 balances (narrow money) reflecting a monetary base reaction function. It is reasonable to regard as it is also influenced by shocks in monetary base emerging from foreign exchange rate distractions7. Referring to Class R models, the policy variable selected is the interest rate. More precisely, interest rate is considered and is taken for granted to react to price and money balances but not to output at least in short-term. Responsiveness to output is added in the reaction function due to the just-identification of the model. Nonetheless, it seems rational to presume that in the bounds of the quarter, no data for real GDP is on hand. On contrary, data referring to inflation and money supply are offered in monthly basis. Also, an illustration of the money stock function and specification of the demand for money balances are embodied. Many factors that determine the demand for money balances are found in the literature. Nevertheless, the level of real GDP, the level of prices, the level of interest rates accompanied by the pace of financial innovation turn out to exhibit greater influence to the money stock function. For this reason, structural disturbances um represent such financial innovation factors like financial system/institutional innovations, process and product innovations that are not unambiguously incorporated in the models. An effort is made to incorporate a credit supply function attempting to capture the credit view of monetary transmission mechanism originated by the influential study of Bernanke (1983)8 and empirically tested by King (1986). To be more specific, in the credit supply function, responses in price shocks, monetary base shocks and money demand shocks are taken into account. Monetary base tightening weakens the accessibility of funds to the financial system and to banks in particular. This results to credit constraints of borrowers as a consequence of draining bank reserves depressing the amount of loanable funds. In this way, an attempt is made so as to identify the credit view of monetary transmission mechanism. In addition, the inclusion of money demand can be rationalised based on the fact that unanticipated money shocks have an effect on monetary aggregates and bank deposits that represent the principal source of funds for bank lending. This can lead to the deterioration of investment funds imposed by the negative influence on loanable funds which force bank-dependent borrowers to contract investments. Structural credit shocks, uci, correspond to exogenous financial innovations or monetary policy regulatory innovations. Given the short-run restrictions analysed above based on economic theory, the matrices for the two different classes of models B and R are as follows:
6 7 8
from monetary shocks that actually affect the supply of loans which in turn determine investment spending and the level of economic activity (balance sheet channel). See Ball and Mankiw (1995). A considerable amount of papers take into account the exchange rate in investigating the transmission mechanism of monetary policy, i.e. Taylor (2001) and Leitemo and Soderstrom (2005). Bernanke (1983) examined the correlations between bank lending and economic activity and the consequences of output on bank lending. In this framework, the author investigated the impact of the Great Depression in the United States for the period 1930-1933. However, in his SVAR model, credit grants an unresponsive prospect, which cannot disentangle the supply-versus-demand puzzle.
102
1 a2 a3 a6

a1 1 a4 a7 0 0 1 0 0 0 a5 1 a1 0 0 0
u ci 0 a8 a9 a10 1 By multiplying the matrices we obtain 5 equations that are clearly displayed in the structural equations [2] to [6] for the Class B models. On purpose, the time subscripts are neglected to have a more comprehensible presentation. y = a1 (ci - p) + uy (2) (+) (3) p = a2y + up (+) b = a3y + a4p + a5m + ub (4) (+) (-) (?) (5) m = a6y + a7p + um (+) (+) ci = a8p + a9b + a10m + uci (i = 1,2,3) (6) (+) (-) (?) In equations [2] to [6] y, p, b, m and ci correspond to the reduced form errors, innovations in the related variables. Similarly, uy, up, ub, um and uci, specify the orthogonal structural shocks. Respectively, the coefficients a1, a2,.,a10 are the structural parameters that measure instantaneous reactions and identify the short-run relationships among the shocks in variables. Lastly, ci (i = 1,2,3) symbolize the three alternative measures of bank credit. The corresponding relationship of the structural parameter is specified in the parentheses underneath the coefficients and a question mark (?) under a parameter implies theoretical uncertainty on the appropriate sign related to the economic relationship of that parameter. To this point, class B models considering the monetary base have been clarified. Regarding class R models reckon with interest rate, approximately similar theoretical relationships are applicable. Following the same procedure, the contemporaneous restrictions referring to Class R models lead to the matrices depicted below. y uy 1 a1 0 0 a1 up = ur 0 a3 1 a4 0 um a5 a 6 a 7 1 0 u ci 0 a8 a9 a10 1 Similarly, by multiplying the matrices we obtain another 5 equations that can be viewed distinctly in the structural equations (7) to (11) for the Class R models. Again, the time subscripts are omitted so as to facilitate the understanding of the structural relationship between the variables. y = a1 (ci - p) + uy (7) (+) p = a2y + up (8) (+) (9) r = a3p + a4m + ur (+) (?) m = a5y + a6p + a7r + um (10) (+) (+) (-) ci = a8p + a9r + a10m + uci (i = 1,2,3) (11) (+) (-) (?)
1 0 0 0 a2
y p b m ci
uy up = ub um
p r m ci
103
Equations [7] and [8] are precisely the same with Equations [2] and [3]. In Equation [9] where the policy response function is represented, instead of the monetary base, the interest rate is considered. Turning to Equation [10], the money demand function is enriched by permitting impacts arising from interest rate shocks, leaving the other elements unchanged. Consequently, in Equation [11] interest rate shocks instead of monetary base are set to influence credit supply function, and to put it differently interest rates shocks are behind the credit slowdown. An alternative method employed by Sims (1986) named Maximum Likelihood method (ML) and further developed by Amisano and Giannini (1997) is employed to estimate the parameters in Equations [2] to [11] for both classes and to calculate standard errors. The log likelihood is maximized by the method of scoring (with a Marquardt-type diagonal correction). 2.1. Data and variable selection Five endogenous variables are encompassed in the SVAR models to detect the macroeconomic responses of the Greek economy to monetary innovations in the spirit of other studies of the monetary policy transmission process: output (y), price level (p), money stock (m), credit (c) and a monetary policy variable. There are two classes of models under consideration namely Class B and Class R. For Class B model the monetary policy variable that has been selected is monetary base (b), whereas in Class R model is bank rate (r). In order to disentangle the impact of monetary policy, three different measures of bank credit (c) are considered that is to say credit to households (c1), credit to firms (c2) and total bank credit (c3). Hence, there are three models in class B and class R respectively. Seasonally adjusted quarterly time series for Greece over the period 1980:1 to 2005:4 are used. The time series in the SVAR analysis are: real gross domestic product (y), the consumer price index (p), the monetary base (b), overnight interbank rate (r), narrow money stock M3 (m), bank credit to households (c1), bank credit to firms (c2) and total bank credit (c3) used as three alternative measures of credit. The overnight interbank rate is the bank rate. For the analysis of macro data, economic aggregates such as money stock and monetary base are obtained from monthly statistical bulletins published by the Bank of Greece (BoG). Variables regarding bank credit are based on monthly data provided from the statistics department of Bank of Greece (BoG) as well as monthly data on the overnight interbank rate. Real GDP is taken from the National Statistical Service of Greece (NSSG).
3. Estimation Results
Having ensured stationarity of all variables, the estimation results based on the Maximum Likelihood method (ML) demonstrate the estimated structural relationships for both classes of models in Table 1 and 2. The significance of the estimated coefficients have been determined taken into account the pvalues along with the z-statistic that allow us to perform the test at the 5% significance level, a p-value lower than 0.05 is taken as evidence to reject the null hypothesis of a zero coefficient and a z-statistic greater than 2 at the same significance level respectively. To begin with the monetary-base models, it is apparent that the estimation results of B2 and B3 models are quite similar, however they are not significantly different than that of B1 model. More importantly, on average 7 out of 8 expected signs indicated in the previous section confirm the economic theory behind the equivalent relationship of the structural parameters for all B models. The positive sign and the significant coefficient of credit in the aggregate demand equation of all models of B class are in accordance with the economic theory. In particular, they verify the fact that credit constraints do have an effect on either the level of investment or real economic activity and the level of output. The positive and significant coefficient of output in the aggregate supply equation for all B models, also confirms the economic theory, implying that the short-run Philips curve holds to be true in Greece. Aggregate demand innovations are important determinants of the course of inflation, at least in the short run. Referring to monetary base reaction function, the response of monetary base to output in all models is positive and significant indicating that price level targeting has been undertaken
104
with the objective to achieve price stability. The positive but insignificant coefficient of price level to monetary base in B1 model can be interpreted as a lack of inflation targeting initiated by the monetary authority. In B2 and B3 models, the negative insignificant coefficient of price level implies that monetary policy can actually affect monetary variables such as price level. Furthermore, the positive but insignificant response of monetary base to money demand expresses the altering of interest rates by the monetary authority as being able to affect the money supply and keep it under its control by mandating specific types of interest rates that may lead to liquidity shocks. The negative coefficient of output in the money demand function is unexpected suggesting that the level of real output does not exhibit any influence to the money demand. On the other hand, the significant but positive coefficient of level of prices in B2 and B3 models proves to be a factor that determines the demand for money balances. All coefficients referring to credit supply function are insignificant but the majority of them assert the economic theory. More specifically, in all B models the credit supply is negatively affected by monetary base. It is notable that the negative sign in monetary base cannot affirm the existence of the credit view of monetary transmission mechanism for any form of credit. In other words, the respective negative coefficients in all B models point to the fact that the credit view seems not to be relevant in the individual, corporate or total credit implying a non-operative role of the credit view for any credit category. Also, is notable in household credit the decrease of credit supply resulting from inflation growth. Additionally, the theoretical uncertainty prevailing the effect of money demand in the credit supply function is evident due to the fact that leads to contradictory sings in the coefficients of B1 model on the one hand and that of B2 and B3 models on the other hand. In particular, model B1 points to a positive impact of money demand shocks on the credit supply function while models B2 and B3 underline a negative impact correspondingly. Having examined the B models it is reasonably to assume that the credit view cannot be supported when monetary base is taken into account as a monetary policy variable even if being insignificant. Turning to class R models, the results of the estimated coefficients for the 3 different models are fairly comparable as they are presented in Table 2. In the aggregate demand function there is a significant and positive impact of credit in R1 and R3 models supporting the economic theory in a way similar to monetary-base models. It is remarkable that all remaining coefficients in the aggregate supply function, interest rate function and money demand function are statistically insignificant but like in monetary base models the majority of the expected signs confirm the economic foundation imposed in the contemporaneous restrictions. More precisely, the aggregate supply function is characterized by the insignificant short-run impact of aggregate demand on price level. In the interest rate function which reflects the monetary policy variable for this class of models, it is evident that interest rate responds positively nonetheless insignificantly to price level in R1, R2 and R3 models, which indicates once again absence of inflation targeting by the monetary authority following the corresponding interest rate shock. In the same function, shocks in money demand lead to negative and insignificant effects on interest rate only in R2 model which is contrary to the responses of monetarybase models discussed above. In this sense, it is logically to assume that when interest rate is taken into account as a monetary-base variable, money demand shocks affect interest rates in a similar way than they affect monetary base. Looking at the money demand function, it doesnt verify the expected positive relationship between money demand and output in any R model, however there is validation for the expected positive relationship between money demand and price level in R1 and R3 models since nominal money demand is proportional to the price level in the economy (notable exception is the independent relationship between real money demand and price level). Additionally, there is validation for the expected reverse relationship between money demand and level of interest rates in R1 and R3 models. Lastly, the estimated coefficients of the credit supply function offer valuable insights for the credit view in the Greek economy.
Table 1: Estimated structural relationships for monetary-base models
105

Model B1
y = 1.313 (c1 - p) + uy (0.629) (0.036) p = 0.073 y + up (0.039) (0.059) b = 0.342 y + 0.601 p + 0.268 m + ub (0.148) 0.713) (0.229) (0.020) 0.399) (0.241) m = -0.158 y + 0.414 p + um (0.063) (0.312) (0.013) (0.184) c1 = -0.218 p 0.019 b + 0.037 m + uc1 (0.180) (0.020) (0.046) (0.225) (0.337) (0.413) Model B2 y = 0.717 (c2 - p) + uy (0.376) (0.056) p = 0.053 y + up (0.025) (0.035) b = 0.453 y - 0.230 p + 0.340 m + ub (0.141) (0.683) (0.230) (0.001) (0.736) (0.140) m = -0.107 y + 0.590 p + um (0.061) (0.296) (0.080) (0.046) c2 = 0.164 p - 0.004 b - 0.016 m + uc2 (0.185) (0.025) (0.058) (0.375) (0.872) (0.777) Model B3 y = 1.189 (c3 - p) + uy (0.531) (0.025) p = 0.068 y + up (0.031) (0.028) b = 0.446 y - 0.183 p + 0.326 m + ub (0.145) (0.698) (0.228) (0.002) (0.792) (0.153) m = -0.108 y + 0.626 p + um (0.063) (0.305) (0.087) (0.040) c3 = 0.070 p - 0.010 b - 0.029 m + uc3 (0.168) (0.020) (0.046) (0.676) (0.624) (0.519)
Note:
(uy) = 0.058 (up) = 0.011 (ub) = 0.076 (um) = 0.033 (uc1) = 0.014
(uy) = 0.053 (up) = 0.010 (ub) = 0.070 (um) = 0.031 (uc2) = 0.017
(uy) = 0.054 (up) = 0.010 (ub) = 0.071 (um) = 0.031 (uc3) = 0.013
The numbers in parentheses are asymptotic standard errors and P-values respectively. us are the estimated standard errors of the structural disturbances.
106
Table 2:

Estimated structural relationships for interest-rate models
Model R1 (uy) = 0.060 (up) = 0.011 (ub) = 0.012 (um) = 0.040 (uc1) = 0.012 Model R2
y = 1.622 (c1 - p) + uy (0.783) (0.038) p = 0.084 y + up (0.045) (0.061) r = 0.085 p + 0.180 m + ub (0.116) (0.238) (0.461) (0.449) m =-0.119 y + 0.131 p - 1.173 r + um (0.095) (0.443) (2.651) (0.208) (0.767) (0.658) c1 = -0.216 p + 0.377 r + 0.031 m + uc1 (0.176) (0.124) (0.040) (0.217) (0.002) (0.427) y = 0.535 (c2 - p) + uy (0.350) (0.127) p = 0.030 y + up (0.023) (0.194) r = 0.107 p - 0.098 m + ub (0.135) (0.669) (0.428) (0.883) m = -0.043 y - 0.034 p + 1.591 r + um (0.074) (0.633) (5.586) (0.562) (0.956) (0.775) c2 = 0.321 p + 0.444 r - 0.045 m + uc2 (0.155) (0.152) (0.048) (0.038) (0.003) (0.342) Model R3 y = 0.972 (c3 - p) + uy (0.470) (0.038) p = 0.044 y + up (0.028) (0.117) r = 0.028 p + 0.316 m + ub (0.159) (0.544) (0.855) (0.562) m = -0.106 y + 0.406 p - 3.049 r + um (0.187) (0.843) (8.359) (0.568) (0.629) (0.715) c3 = 0.206 p + 0.384 r - 0.045 m + uc3 (0.131) (0.121) (0.038) (0.118) (0.001) (0.237)
Note:
(uy) = 0.053 (up) = 0.011 (ub) = 0.012 (um) = 0.035 (uc2) = 0.016
(uy) = 0.052 (up) = 0.011 (ub) = 0.014 (um) = 0.055 (uc3) = 0.012
The numbers in parentheses are asymptotic standard errors and P-values respectively. us are the estimated standard errors of the structural disturbances.
R1 model responses in a way similar to B1 model where there is a negative relationship, yet not significant, between credit and price level innovations. Nevertheless, the given economic theory is justified in R2 and R3 models. Once again there is no definite impact of money demand shocks to credit supply, as it is evident in the interest rate models. In R1 model, is apparent the positive effect of money demand shocks to credit supply, while the opposite prevails R2 and R3 models, rationalising
107
thus the theoretical ambiguity on the issue. The unexpected positive response, even though significant, of credit supply to interest rate innovations in all models implies that banks in Greece induce bank loan supply in case of interest rate increase brought about by tighter monetary policy. This is in contrast with the economic underpinnings and an alternative explanation can be found in the parallel shifts of overnight interbank interest rates, deposit and lending interest rates and to put it differently similar interest spread across different types of interest rates. Admittedly, the impact of interest rate shocks is very determined and it is noticeable the comparable impact to all types of credit, however credit to individuals give the impression of being less affected by interest rate innovations compared to firm credit and total bank credit. To conclude, in spite of a considerable amount of insignificant estimated coefficients in R class models mainly, the general impression for the credit view in Greece is quite prompting. A large portion of the estimated coefficients conforms to the imposed restrictions of the economic structure and the outcomes are pretty stable for the 3 alternative credit measures. Having examined the estimated structural relationships, the next section deals with the structural impulse response functions to study the consequences of the different shocks and to provide evidence in support of the theoretical model. 4.1. Structural impulse response functions In the previous section, the structural parameters for the two classes of models have been estimated. The next step involves the study of the behaviour of the macro-dynamic system incorporated in the models. In other words, this section observes the econometric outcomes of the structural specification. For this reason, impulse response functions depict the dynamic impact following a one-time shock to one of the innovations on current and future values of the systems endogenous variables. The impulse responses facilitate to bring to light the interaction of credit with the rest of the economy. The impulse responses are estimated setting the impulse to one standard deviation of the residuals, thus ignoring the correlation in the VAR residuals. Impulse responses standard errors are estimated by using Runkles (1987) Monte Carlo simulation method and are plotted for a time period of 12 quarters (12 quarters/4=3 years) for all variables. The vertical axes depict changes in the variables from their baseline. The dashed lines are 90% confidence intervals obtained by employing Kilians (1998) technique. It is reasonable to assume that for a model with five variables, 25 impulse response functions are obtainable. Since the impulse responses for B2 are somewhat akin to B3 model, only the impulse responses for B1 and B2 are reviewed9.
The impulse responses for the interest-rate models also do not significantly deviate from that of monetary-base models.
108

Figure 1: Responses to a monetary policy shock B1 model
Note: The dashed lines enclose the one-standard deviation of the residuals that have been calculated from the asymptotic distributions of the responses.
Figure 2: Responses to a credit shock model B1
Note: The dashed lines enclose the one-standard deviation of the residuals that have been calculated from the asymptotic distributions of the responses
The adverse monetary shock of model B1 presented in Figure 1 has a rather weak negative effect initially on real output, price level, money demand and credit supply as indicated in the economic theory. Approximately 3 quarters after the monetary shock, real output, price level, credit and money demand reveal a minor almost insignificant rise, which then tend to approach the zero baseline. Especially, the indifferent long-run response of real output is well suited in the economic theory. The negative impact of credit to monetary base resulting from contractionary monetary policy is in favour of the credit view however this outcome is in contrast with estimated structural relationships at least in the short run. Responses to a credit shock are depicted in Figure 2 where it is noticeable the positive effect on real output that converges to zero in the following periods, the moderate short-term negative impact on money demand, the initial scarcely positive effect on monetary base is accompanied by a relatively unaffected period of no effect and the permanent adverse effect on price level for all the forecast
109
horizon. Hence, the impulse responses present an inconsistency with the estimated structural relationships in the credit supply function and the economic theory as well. Generally speaking, monetary base positively responds to credit shock according to the impulse responses, however, the estimated structural coefficient points to an insignificant but negative impact which fits the economic theory. In other words, a paradox is observed in this case. Monetary base should have the same effect with price level and real output, nevertheless it seems to have an independent response. Moreover, the corresponding response of money demand as implied by policy reaction function should exhibit a positive effect, though the impulse response of money demand is associated with a rather negative effect through the 12 quarters following the credit shock. It is reasonably to assume that the dynamics of the monetary policy shock and money demand may not be captured in the corresponding functions of the structural model.
Figure 3: Responses to a monetary policy shock B2 model
In Figure 3 there is roughly no difference in the impulse responses to a monetary policy shock in B1 and B2 models. Needless to repeat, the relevant impulse responses in B2 model exert similar characteristics like in B1 model. The only contrast can be found in credit response where right after the monetary shock credit seems to remain unaffected for about 2 quarters, then scarcely rises for another 2 quarters to decline and persists in a negative direction for the remaining period. The price response is the only one that is in line with the structural coefficients whereas real output and money demand responses pursue a reverse direction that then estimated ones.
110

Figure 4: Responses to a credit shock model B2
Surprisingly, there is a differentiation in the responses of B1 and B2 models and they can be found in the case of a credit shock. To be more specific, in Figure 4, it is clear that money demand in B2 model responds in an opposite way than that in B1 model and in particular it remains positive for almost 8 quarters after the credit shock, and then it becomes fairly neutral. The price response in B2 model is not that negative compared to the response in B1 model, which has a permanent negative effect. Nonetheless, the price response cannot be considered to have the expected response as indicated in the credit supply function. Likewise, the monetary base response is more dynamic in B2 model. Unexpectedly, the impulse responses referring to monetary base are in contrary with both the economic theory and the estimated coefficient in the credit supply function. More importantly, a credit shock has a positive effect in the real output with a tendency to reach the baseline following the fourth quarter. The two models present a difference in terms of the response of real output to a credit shock. The real output response in household credit is somewhat more definite than that of corporate credit. The respective effect on household credit model is steadily positive for the first 3 quarters and then there is another positive impact that lasts for another 3 quarters before becoming neutral. This can be taken as an indication of greater credit constrains for households in Greece. A possible explanation for this incident is that household credit in form of consumer and housing loans is a relatively new activity of the banking operations in Greece hence the openness to bank loans attracts the interest of the Greek banks basically following the transformation of the banking system after 1996. There are some implications for the credit view after presenting the impulse responses functions for the 2 classes of models that may contribute to the investigation on the bank lending channel of transmission mechanism and shed some light on the issue for the case of Greece. The feeling extracted from the responses is that there is no explicit evidence for the existence of the credit view just encouraging indication no matter if monetary base or interest rate is used as a monetary policy variable given the presence of certain anomalies. To sum up, the findings of the impulse responses support the view that households instead of firms give the impression of being more credit constrained. This argument is based on the findings referring to the impact of bank credit shock in real output. However, credit constrains to households are somewhat predicted by the structural economic theory, nonetheless it can be verified by the organization and structure of the Greek banking system. 4.2. Variance decompositions An alternative method to interpret system variables and the properties of the models is the variance decomposition. To assess the importance of the different shocks for the macroeconomic variables, the
111
forecast error variances of the variables are decomposed with reference to the shocks. Given that factorisation is based on structural orthogonalisation, the forecast standard errors are identical to those from the Cholesky factorisation since the SVAR is just identified. It is representative for a variable to elucidate all its forecast variance in short run horizons and smaller percentages in long run horizons. The results for different time horizons are presented in Tables 3 and 4 only for B1 and B2 models, since there is little variability between corporate and total bank credit for both classes of models. It is another paradox that in the short and long run, all shocks that correspond to each variables own shocks are important determinants for their forecast errors. For example in Table 3 it is apparent that aggregate supply, monetary policy, money demand and credit supply shocks contribute only a small part to the forecast error variance of real output during the whole forecast period. At the end of time horizon, IS shock accounts for 81,9 per cent of total variation in real output, AS shock for 3 per cent, monetary policy shock for 7 per cent respectively. In the study of Safaei and Cameron (2003) IS shock contributes approximately 45 per cent to the total real output variation after 40 quarters following the shock while aggregate supply shock gradually increases its influence reaching a peak up to 22 per cent.
Table 3:
Variable
Variance decompositions for the B1 model (Proportion of forecast error variable for variable)
Quarter(s) ahead 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 IS shock 100.0 93.7 92.0 83.0 82.4 81.9 0.6 0.7 1.5 2.4 2.1 2.5 3.9 4.1 9.5 9.9 9.9 9.8 5.1 4.5 5.0 4.8 4.7 4.7 5.6 7.1 10.4 9.7 9.4 9.4 Aggregate supply shock 0.0 1.3 1.3 2.3 2.7 3.0 99.4 93.6 89.9 84.3 82.3 81.6 0.9 3.1 3.5 4.0 4.4 4.8 1.9 1.7 8.3 12.2 12.5 14.3 0.7 2.9 6.2 11.1 12.5 13.8 Monetary policy shock 0.0 1.1 2.5 6.6 6.8 7.0 0.0 0.1 1.1 2.2 3.5 3.7 95.2 89.5 82.2 78.8 78.3 78.0 1.3 3.6 4.5 7.7 7.9 7.8 0.7 0.5 0.5 2.6 2.8 2.7 Money demand shock 0.0 3.6 3.9 5.3 5.3 5.3 0.0 4.3 5.1 6.7 7.5 7.2 0.0 2.1 2.7 2.6 2.5 2.5 91.7 89.2 81.3 71.3 70.5 68.9 0.7 2.2 3.1 4.0 4.6 4.8 Credit supply shock 0.000 0.3 0.3 2.8 2.8 2.8 0.0 1.3 2.4 4.4 4.6 5.0 0.0 1.2 2.1 4.7 4.9 4.9 0.0 1.0 0.9 4.0 4.4 4.3 92.3 87.3 79.8 72.6 70.7 69.3
Real output
Price
Monetary base
Money demand
Credit to persons
This is an indication that money supply shocks cannot be considered accurate indicator for monetary policy. As a result the negative impact of money supply shock on real output remains unclear for the case of Greece. Variability in the price level is ruled by aggregate supply shocks in the whole forecast period of 12 quarters leaving minor temporary effects to money demand shock (7,2 per cent). The aggregate supply shock has a moderate impact on money demand particularly in the long run,
112
whereas monetary policy shock seems to influence money demand 2 years following the shock. Monetary policy shocks have sound effects on monetary base which in turn is influenced in a lesser extent by aggregated demand shocks and even lesser by aggregate supply and credit supply shocks. Lastly, aggregate supply shock demonstrates slow but steady impact to credit to persons and money demand shock has a non-operative role in explaining credit variations.
Table 4:
Variable
Variance decompositions for the B2 model (Proportion of forecast error variable for variable)
Quarter(s) ahead 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 1 2 4 8 10 12 IS shock 100.0 91.4 84.0 71.7 71.2 70.2 2.2 2.2 3.3 4.8 5.2 5.4 8.7 9.8 16.8 15.8 16.1 15.9 2.5 2.4 4.3 3.5 3.4 3.4 5.5 9.6 11.6 11.8 11.5 11.9 Aggregate supply shock 0.0 1.2 2.7 4.5 4.7 5.4 97.8 89.7 83.1 76.2 73.0 73.1 0.0 3.4 4.6 6.3 6.5 6.9 3.8 3.3 8.5 14.3 14.9 16.1 1.5 1.4 2.6 3.3 3.3 3.6 Monetary policy shock 0.0 0.4 1.5 5.5 5.7 5.7 0.0 0.1 0.2 2.0 2.6 2.6 91.3 85.8 73.3 69.1 67.0 66.8 2.1 8.5 8.5 11.3 10.9 11.2 0.0 0.3 1.0 4.0 7.1 7.1 Money demand shock 0.0 6.1 6.6 8.7 8.7 9.2 0.0 7.9 8.5 12.0 13.5 13.0 0.0 0.9 1.8 4.2 4.6 4.7 91.6 82.7 74.8 66.7 65.6 64.1 0.1 0.1 1.8 2.7 2.7 2.7 Credit supply shock 0.0 0.9 5.2 9.6 9.7 9.5 0.0 0.1 4.9 5.0 5.7 5.9 0.0 0.1 3.5 4.6 5.8 5.7 0.0 3.1 3.9 4.2 5.2 5.2 92.9 88.6 83.0 78.2 75.4 74.7
Real output
Price
Monetary base
Money demand
Credit to firms
Overall, the variance decompositions for household bank credit are comparable to some of the impulse responses presented in the previous section. However, some of the variance decompositions are in contrast to the estimated results and the impulse response functions. The money supply shock has a very modest impact on real output and even smaller to the variance of price level and the contribution of aggregate supply shock is noticeable only to money demand and household credit. It is remarkable that credit shock is not an important factor in bringing about essential variations in other variables. The variance decompositions for B2 models are shown in Table 4. There are several differences but not very determined between household and corporate credit. For corporate credit, credit supply and money demand shocks are greater determinants of real output compared to that of household credit and this influence is apparent in the second quarter after the initial shock. Money demand shock plays a more active role in the variation of price level of corporate credit. There is also a considerable increase in the influence of IS shocks over the monetary base. In the same way, the contribution of monetary policy shock on money demand is greater for B2 model compared to B1 model. Finally, the role of
113
aggregate supply shock is deteriorated for B2 model and on the other hand, the significance of monetary policy shock is triple than that of B1 model. The outcomes from Table 3 and 4 support the belief that IS shocks along with money demand shocks play an important role in the variation of major macroeconomic variables and to a lesser extent monetary policy shocks. On the other hand, credit is assigned by a limited contribution in general macroeconomic fluctuations.
4. Concluding Remarks
This paper examined the importance of bank credit as a critical financial source of variation on output and other major macroeconomic variables in Greece for the period 1980-2005. Following a SVAR approach that imposes contemporaneous constraints in the structural models and distinguishing between household, corporate and total bank lending based on two different classes of models for two different monetary policy variables, the outcomes offer sensible rationale for the specifications embodied in the SVAR models. A reasonable number of precise estimates of the structural parameters is observed, which implies consistency with the theoretical grounds. The feeling derived from the impulse responses fairly fitted the choice of credit measure. However, lack of similar studies for the case of Greece does not permit comparison on the drawing conclusions although the results of this study mostly match those of Brissimis et al. (2001), the only Greek study that addresses the credit channel by using panel bank level data. The outcomes of this study clearly pointed to the significant impact of monetary policy on the supply of bank loans and on aggregate economic activity in general. However, in this study when monetary base is considered as a monetary policy variable, the structural estimates are not in favor of the existence of the bank lending channel in Greece. So inconclusive evidence calls for further investigation on the issue. A likely extension of this study would be to use a sample taking into consideration the years following the financial liberalisation in 1996 capturing this way the advances that took place in the banking sector. This may prove to be useful to uncover the credit channel in Greece.
114
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Agung, J. (1998) Financial deregulation and the bank lending channel in developing countries: The case of Indonesia. Asian Economic Journal, 12 (3), p. 273-294. Amisano, G., and Giannini, C. (1997) Topics in structural VAR econometrics. 2nd ed. Springer-Verlag: New York. Bacchetta, P., and Ballabriga, F. (2000) The impact of monetary policy and banks balance sheets: Some international evidence. Applied Financial Economics, 10 (1), p. 15-26. Ball, L., and Mankiw, N.G. (1995) Relative-price changes as aggregate supply shocks. The Quarterly Journal of Economics, 110 (1), p. 161-193. Bernanke, B.S., and Blinder, A.S. (1992) The federal funds rate and the channels of monetary transmission. American Economic Review, 82 (4), p. 901-921. Bernanke, B.S., and Gertler, M. (1995) Inside the black box: The credit channel of monetary policy transmission. Journal of Economic Perspectives, 9 (4), p. 27-48. Bernanke, B.S. (1983) Non monetary effects of the financial crisis in the propagation of the Great Depression. American Economic Review, 73 (3), p. 257-276. Boyd, J.H., and Smith, B.D. (1997) Capital market imperfections, international credit markets, and noncovergence. Journal of Economic Theory, 73 (2), p. 335-364. Brissimis, S.N., Kamberoglou, N.C., and Simigiannis, G.T. (2001) Is there a bank lending channel of monetary policy in Greece? Evidence from bank level data. ECB Working Paper No. 104. Carlino, G., and DeFina, R. (1999) The differential regional effects of monetary policy: Evidence from the U.S. States. Journal of Regional Science, 39 (2), p. 339-358. Cecchetti, S.G. (1995) Distinguishing theories of the monetary transmission mechanism. Federal Reserve Bank of St.Louis Review, 77 (3), p. 83-97. Clements, B., Kontolemis, Z.G., and Levy, J. (2001) Monetary policy under EMU: Differences in the transmission mechanism?. International Monetary Fund Working Paper 01102. Dedola, L., and Lippi, F. (2000) The monetary transmission mechanism: Evidence from the industries of five OECD countries. Banca dItalia Temi di Discussione 389. Ehrmann, M., and Worms, A. (2001) Interbank lending and monetary policy transmission: Evidence for Germany. ECB Working Paper Series 73. Gerlach, S., and Smets, F. (1995) The monetary transmission mechanism: Evidence from the G-7 countries. CEPR Discussion Paper 1219, July 1995. Gibson, M.S. (1997) The bank lending channel of monetary policy transmission: Evidence from a model of bank behavior that incorporates long-term customer relationships. Board of Governors of the Federal Reserve System, International Finance Discussion Paper 584. Greenwald, B.C., and Stiglitz, J.E. (1993) Financial market imperfections and business cycles. The Quarterly Journal of Economics, 108 (1), p. 77-114. Hortemoller, O. (2003) Further VAR evidence for the effectiveness of a credit channel in Germany. Applied Economics Quarterly, 49 (4), p. 359-381. Hubbard, R.G. (1998) Capital-market imperfections and investment. Journal of Economic Literature, 36 (1), p. 193-225. Hubbard, R.G. (1995) Is there a credit channel for monetary policy?. Federal Reserve Bank of St.Louis Review, 77(3), p. 63-77. Kashyap, A.K., and Stein, J.C. (1994) The impact of monetary policy on bank balance sheets. NBER Working Paper 4821. Kashyap, A.K., and Stein, J.C. (1997) The role of banks in monetary policy: A survey with implications for the European monetary union. Federal Reserve Bank of Chicago Economic Perspectives, 22 (5), 2-18.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
115 [24] [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Kashyap, A.K., and Stein, J.C., and Wilcox, D.W. (1993) Monetary policy and credit conditions: Evidence from the composition of external finance. The American Economic Review, 83 (1), p. 78-98. Kilian, L. (1998) Small-sample confidence intervals for impulse response functions. Review of Economics and Statistics, 80 (2), p. 218-230. King, S. (1986) Monetary transmission: Through bank loans or bank liabilities?. Journal of Money, Credit & Banking, 18 (3), p. 290-303. Leitemo, K., and Soderstrom, U. (2005) Simple monetary policy rules and exchange rate uncertainty. Journal of International Money and Finance, 24 (3), p. 481-507. Oliner, S.D., and Rudebusch, G.D. (1996) Is there a broad credit channel for monetary policy?. Federal Reserve Bank of San Francisco Economic Review, 1, p. 3-13. Peek, J., and Rosengren, E.S. (1995) Is bank lending important for the transmission of monetary policy?. Federal Reserve Bank of Boston Conference Series, 39, pp. 47-68. Ramaswamy, R., and Slok, T. (1998) The real effects of monetary policy in the European Union: What are the differences?. IMF Staff Papers, 45(2), p. 374-396. Runkle, D.E. (1987) Vector autoregressions and reality. Journal of Business and Economic Statistics, 5 (4), p. 437-432. Safaei, J., and Cameron, N.E. (2003) Credit channel and credit shocks in Canadian macrodynamics a structural VAR approach. Applied Financial Economics, 13 (4), p. 267277. Sims, C. (1986) Are forecasting models usable for policy analysis?. Federal Reserve Bank of Minneapolis Quarterly Review, 10 (1), p. 2-16. Taylor, J.B. (2001) The role of the exchange rate in monetary-policy rules. American Economic Review, 91 (2), p. 263-267. Vermeulen, P. (2002) Business fixed investment: Evidence of a financial accelerator in Europe. Oxford Bulletin of Economics and Statistics, 64 (3), p. 213-231.
Private Cost of Teacher Education in a Nigerian University

Roseline O. Olubor Senior Lecturer, Institute of Education, University of Benin P.M.B.1154, Ekewan Campus, Benin City, Edo State, Nigeria E-mail: olubor2004@yahoo.com Abstract The University of Benin at its inception in 1970 had the Faculty of Education and Institute of Education as part of its founding Faculties and Institutes. These two departments have therefore matured over the years and are well known for teacher education programmes. It is therefore appropriated to use the University of Benin to find out the private cost of teacher education programmes. Eight hundred and twenty students of the Faculty of Education and Institute of Education formed the sample of this study. Simple percentage, mean and rank orders were the statistical tools adopted for the analysis of data. The study inter alia revealed cost variations in the items of cost in the private cost of teacher education within and among programmes in University of Benin. The study also revealed that annual average private incidental costs were higher than the annual average private academic costs in all the programmes.. Thirdly cost of feeding ranked first among the items of cost of annual average private cost of Teacher Education in University of Benin. Among the recommendations is that multinational companies, banks and other NGOs committed to standard in education may have to also provide scholarship for education students as they are currently doing for some other courses like engineering, medicine etc. Also recommended is that students loans scheme may need to be introduced as an incentive to encourage people go into education programmes in Nigerian Universities. Keywords: Teacher Education, Nigeria, Incidental Cost, Academic Cost, Private Cost
Introduction
The mission of higher education in Nigeria is to provide access to quality post secondary education which will lead to the production of sound graduates. The Federal Ministry of Education (2002) in its National Summit on Higher Education resolved among others that all stakeholders in the higher education should be challenged to share in the cost of education by paying some fees in order to attain and sustain a reasonable level of funding of higher education in Nigeria. A strategy to be adopted was to set up a fact-finding committee on the actual cost of producing graduates of various disciplines at all levels of tertiary education. Until the late 1980s the management of teachers had hardly been a subject of public debate, research or innovative projects in the education sector (Durett, 2001). He also made it clear that over the recent years the professional status and motivation of teachers have been declining world wide while the expectations of parents, employers and others towards teachers are steadily on the increase. Teachers are leaders of the pupils community of work and they make a great difference not just in the functioning of a class but also in the lifes of their students hence the most important tools for
117
schools are teachers. Hernes (2001) pointed out that over the next half century the number of children to be educated in the world will increase by a half. They will demand not just a basic education but more years of higher quality education thus implying that the cost will also be enormous. Kitaev and Peano (2001) argued that families and communities financial contributions to education development cannot be ignored whether it is a direct fee imposed on parents, private tuition to pass a critical examination or payments hidden as voluntary contribution. This is to make for objective policy analysis and decision making. Private organization, NGOs and foreign nations are being called upon to participate in the financing of education in the developing world. What will form the basis of their contribution? Hence this study which is meant to find out what individuals or families are spending to keep their wards in the university in order to undergo teacher education programme. Solase (2000) in his project - Trends in Private Sector Development - pointed out that the private sector is playing an increasingly important role in financing and providing educational services in many countries. To him the term private sector encompasses household out of pocket expenses. The underlying principle of his project is that strengthening the private sectors role in non compulsory education over time will release public resources for the compulsory (primary) level. The demand for higher education is quite high all over the world. Globalization has further increased the thirst for higher education. Because of this growing demand all over the wold, tuition and other fees are rising faster than inflation rates and even family incomes. Under these circumstances, student loans perform a social role, facilitating access to education for those who are otherwise unable to pay expensive tuition fees and other related expenses. Altner and Kitaev (2001) quickly pointed out that the record of student loans in developing countries is notoriously bad. They identified the problems to be administration, targeting, accountability and tracking the defaulters. Since students loans are not very different from any other mortgage type loans, designed to provide credit to buy a house or a car, if the financial system functions well and the banks are motivated to get involved, students loans are no different from mortgages or other regular loans for the lending agencies. Globalization and new technologies have made it possible to track a defaulter anywhere in the world through bank account. Their names could also be posted in newspapers and on the internet. Earley (1998) in his paper titled Teacher Quality Enhancement Grants for States and Partnerships pointed out that the Higher Education Act (HEA) was enacted during the 1960s to help needy college students attend institution of higher learning via federal grants and loans in the United States. Ebmeiere and others (1991) in analyzing financial data on funding of teacher education from 6 research universities over 10 years revealed that schools of education do not hold favourable positions in research universities. They are generally the lowest funded unit on campus and have lost ground over time. Ever since man realized that human resources, not capital, nor income nor material resources constitute the ultimate basis for the wealth of nations, there has been massive expansion of educational facilities at all levels in most developing countries. Consequently, education industry has become the major consumer of public and private funds. Psacharopoulos asserted that: Human capital is created and the quality of human input in production is significantly improved by spending on education. This is why countries, particularly those with low per capita income, invest such a large proportion of their budgets on education and why, when the state does not, individuals do. Teachers are always in demand because of regular depletion. This is due to retirements, termination, death, resignation or abandonment of the job. Consequently, teachers recruitment is a continuous process. Orhibhabor (2004) reported that the katsina State government in Nigeria is soon to recruit about 600 secondary school teachers for subjects like science, mathematics, as well as English language. This is meant to replace those that have left the profession. And so it is happening in many other states of the federation. Egbobamwonyi (2002) pointed out that even the best professors at the most prestigious universities are indebted to teachers who took time, effort to prime and activate their desire for
118
knowledge and understanding. All top policy makers in the public and private sectors have been taught by a teacher at some point in their lives. It is worthy to note that teachers are the most important guarantors of quality in the educational system. Their apathy or resistance could be deadly to initiatives. No meaningful development can take place without adequate manpower and no adequate manpower training can take place without competent teachers who are products of good teacher education. Recently Edo State government had employed 1,300 graduates into its public schools. In spite of this, the recent study of Obanewa et al (2004) on Edo State Secondary Education Strategy revealed that in the fifty-one sample schools, virtually all the school subjects did not have adequate number of teachers. This study is based on the human capital theory. This theory rests fundamentally on the idea that people invest in themselves (private cost) in a variety of ways in the sense that they incur costs of schooling for the sake of future financial returns. This is evidenced in the phenomenon of self investment by choosing to stay on in school and perhaps also the act of choosing one kind of school rather than another. (Blaugh, 1974, p.1). According to Salim (1993) private cost is defined as that part of investment in education which is made either by the student or the parent. It is broadly divided into direct cost and indirect cost. Direct private cost is defined as the cost directly incurred by a household for the education of students. It indicated the amount of physical resources spent while receiving education. Direct cost is again subdivided into two groups: academic costs and incidental costs. Academic costs show the expenditure which is directly related to instruction whereas the costs which are not directly related to instruction are referred to as incidental costs. The major components of academic costs are pre-admission cost, fees given to the college, private tuition fee, cost of books, stationery, project work, excursions, group work and teaching practice. Incidental costs consist of feeding, subscriptions, travel costs, hostel expenses, cost of clothing, entertainment, and phone calls. The cost of students time which could have been devoted to different ends is part of the indirect cost of education. Alternative uses of students time could be either enjoying leisure or doing some work and being rewarded for that work. It is quite difficult to value the leisure opportunity that is forgone. Most often, indirect cost of education is assessed in terms of the foregone earnings (Kothari and Panchamukhi; 1980). In a country like Nigeria where there is so much unemployment, the alternative to schooling may not be work at all. It may be unemployment. In this type of situation the opportunity cost of students time is zero. Moreover non-graduates in Nigeria are rated poorly on the salary scale. Government and some companies sometimes owe workers salaries to the tune of four months. So it may not worth working rather it is better to acquire higher qualification and aim for better paid jobs. Oranefo (2003) in her study on the private cost of education in a private university in Nigeria found out that the cost ranged between N537,00 for Diploma students and N750,00 for degree students. In this study, tuition fee formed 18.64% of the academic cost for Diploma students and 20.0% of the cost of degree programme. Another item of academic cost that was substantial was cost of textbooks/learning materials. She attributed this marked variation to the fact that the quality of educational services rendered at the undergraduate level is more enhanced than at the Diploma level thereby demanding much financial support. Coombs and Hallaks (1972) assertion that educational cost varies greatly within the same institution and among different programmes has continued to be a truism. Salim (1993) in his study on private cost of Higher Education in Kerala revealed that the actual cost per student in technical education is only slightly higher at the degree level and substantially lower at the post graduate level than that of general education. Adekanmbi (2002) in analyzing the private cost of Post Graduate Diploma in Education showed that the private cost of Post Graduate Diploma in education for the 2000/2001 academic session was N136,666.86k while that of Post Graduate Diploma in technical education was
119
N122,510.30. Adekanmbis study therefore agreed with Salims study that cost per student in technical education is lower at the postgraduate level than that of general education. A worrisome trend in the education industry in Nigeria is the rising cost of education compounded by inflation. Unfortunately, the economies of developing countries in Sub-Saharan Africa are not buoyant and the available fiscal resources are being competed for by many sectors other than education more than ever before. Nwadiani (1993) and Salam (1993) have lamented on the nonavailability of data on the private cost of education. Hence this paper is an attempt to provide data in the private cost of teacher education program in a first generation university in Nigeria. Almost everything is being accounted for by the family or the individual because of governments inability to provide enough fund for the competing sectors of the economy, education inclusive. In fact allocation to education has continued to be on the decline. Table 1 shows the declining budgetary allocation to Education from 1999 to 2003.
Table 1: Budgetary Allocation to Education from 1999 to 2003.
Year 1999 2000 2001 2002 2003 Budgetary Allocation 11.2% 8.3% 7.0% 5.9% 1.8%
Source: Ndiribe, Obey (2003) Nigerias performance in WAEC exam poorest in West Africa Vanguard, Thursday, February, 13 2003, p.21
In modern society, the cognitive, affective and psychomotor abilities of individuals are developed in the formal setting of education. Teacher education therefore is that component of any educational system charged with the education and training of teachers to acquire the skills and competencies of teaching for the school system. The current trend is that scholarship and educational aids are made arbitrarily. Such gesture ought to be based on the actual amount that is being spent in training. Presently, the public does not know how much is spent in training teachers in Nigeria. To find out the actual private cost of teacher education in Nigeria, the following questions were raised 1. What is the annual average private cost of teacher education programme in Nigeria? 2. What is the annual average private academic cost of teacher education? 3. What is annual average private incidental cost of teacher education? 4. Is there any difference in cost among the programmes? Knowledge of the private cost of teacher education will provide benchmark data on planning norms such as unit cost and cost disparities among items of cost. This work is also a modest contribution to the Federal governments desire to find out the actual cost of producing graduates of various disciplines at all levels of tertiary education. The aspiration of any nation to transform into a great country can only be possible if there are competent and dedicated teachers. This is why it is acknowledged that the teacher holds the key to nation building. It is therefore a worthwhile exercise to examine the cost of training such a group of personnel in the school system. The teacher is the most crucial single element in the educative process. His education and training contribute to a more purposeful and better-planned education. A competent teacher directly or indirectly, is bound to influence the quality and quantity of service provided by all other jobs and profession. Teaching is not a lucrative and attractive profession in Nigeria. It is when all other options fail that people resort to teaching. It is therefore likely that the cost may discourage some individuals from investing into such a venture. Table II shows the number of candidates who applied to study education in the Institute of
120
Education and Faculty of Education and the number that was eventually cleared and matriculated, for the periods 2001/2002; 2002/2003 and 2003/2004 sessions.
Table II: Number of students who initially indicated interest in studying education courses and the number that were eventually cleared and matriculated.
2001/2002 No. that was No. that Cleared admitted & matriculated N.A N.A. 52 96 2002/2003 No. that was No. that cleared admitted & matriculated 56 229 65 262 2003/2004 No. that was No. that cleared & admitted matriculated 51 241 160 850
Session Program Diploma Degree

Sources:
(a) Admissions office University of Benin (2003) Diploma clearance list for the 2002/2003 session (b) Ibid.(2004) 2003/2004 session (c) Joint Admission and Matriculation Board (2001) list of Candidates that scored 200 and above for 2001/2002 session. (d) Ibid. (2002) for 2002/2003 session. (e) Ibid. (2003) for 2003/2004 session (f) Admissions Office University of Benin (2002) UME clearance list for the 2001/2002 session. (g) Ibid. (2003) for 2002/2003 session (h) Ibid. (2004) for 2003/2004 session.
Table II showed that many students having failed to gain admission in their areas of choice decided to study education. This is why the number of students that eventually cleared and matriculated in education courses after transferring from other courses where they were denied admission became by far higher than the number of students that initially applied to study education. This is not good enough. The education sector should have enough incentives to attract people in the first instance to the training.
Methodology
The study adopted the ex post facto design. The independent variables such as tuition fees, academic costs and incidental costs have already been incurred. The researcher observed the dependent variable (in this case the private cost to be calculated). The researcher has no direct control over variables because they have already occurred. Thus, only inferences about relations among variables were made. Stratified random sampling technique was adopted along the stratum of type of program. In all, 360 Diploma in education students of the Institute of Education; 400 undergraduates of the Faculty of Education and 60 Post Graduate Diploma of Education students of the Institute of Education of the 2002/2003 session were selected. This period was chosen because it is the last session that has been concluded and the private cost is the most current so far. University of Benin has been chosen for this study because of the fact that the university has matured over the years since its inception in 1970 in terms of provision of facilities and faculties. A checklist named Private Recurrent Unit Cost of Teacher Education was designed and 840 copies provided to elicit information on the private cost of the teacher education programs under study. The checklist was divided into two sections. Section A elicited background information of the students such as nature of program. Section B focused on the items of cost such as pre-admission fee, clearance fee, registration fee, private lesson fee, costs of textbooks/learning materials, stationery, project, excursion, photocopy, term papers and contribution towards group work and cost of teaching practice. Other items are costs of transportation for lectures, journeys home and back to school, clothing, feeding and telephone calls subscriptions to unions and hostel fees. The aforementioned checklists were distributed at the tail end of the session. Data generated were analyzed using simple percentages, mean and rank order.
121
Results
Research Question One What is the annual average private cost of teacher education programme in Nigeria? The research question was answered by making reference to the data in table III.
Table III: Annual average private cost of Teacher Education in Nigeria
S/N 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 ITEMS OF COST Pre-admission fee Clearance fee Registration fee Tuition fee Private lesson Cost of textbooks/learning materials Cost of stationery Cost of teaching practice Cost of project Cost of Excursion Cost of photocopying Cost of term papers Contributions towards group work Cost of transportation Subscription to unions Cost of journeys home & back Out of campus accommodation Cost of clothing Cost of feeding Cost of telephone calls Total Diploma (Naira) N 3186 4669 1486 17613 6917 6453 2000 8660 2482 2111 2303 9770 600 7646 25000 28137 32575 3061 164669 % 1.9 2.8 0.9 10.7 4.2 3.9 1.2 5.3 1.5 1.3 1.40 5.9 0.4 4.6 15.2 17.1 19.8 1.9 100 Degree (Naira) N 1851 4669 2175 4212 3550 10125 3294 3747 10000 1955 3293 4925 2490 5042 1000 9182 30000 34194 38377 5328 179409 % 1.0 2.6 1.2 2.4 2.0 5.6 1.8 2.1 5.6 1.1 1.8 2.8 1.4 2.8 0.6 5.1 16.7 19.1 21.4 2.9 100 PGDE (Naira) N 3000 4669 2300 37750 11100 3305 1600 20000 3380 6375 2140 11200 20450 25000 20200 48000 15300 235769 % 1.3 1.9 1.0 16.0 4.7 1.4 0.7 8.5 1.4 2.7 0.9 4.8 8.7 10.6 8.6 20.3 6.5 100
The data in Table 111 revealed that for the 2002/2003 session in University of Benin, Nigeria, the private cost of Diploma teacher education programme was N164,669; Bachelor teacher education programme was N179,409 and Post Graduate Diploma in teacher education programme was N235,769. In all the three programmes under study, cost of feeding contributed more than 19% of the total private cost of teacher education thus ranking the highest contributor. Research Question Two: What is the annual average private academic cost of teacher education? In order to answer this question recourse was made to the data in table III. Items 1 13 in the checklist were extracted and displayed in Table IV.
122
Table IV: Annual Average Private Academic Cost of Teacher Education. *$1= N137.50K (August 26th, 2005)
S/N 1 2 3 4 5 6 7 8 9 10 11 12 13 Items of cost Pre-admission fee Clearance fee Registration fee Tuition fee Private lesson Cost of textbooks learning material Cost of stationery Cost of teaching practice Cost of project Excursion Cost of photocopying Cost of term papers Contributions towards group work Total Diploma (Naira) N 3186 4669 1486 17613 6917 6453 2000 8660 2482 2111 2303 57880 % 5.5 8.0 2.6 30.4 00 12.0 11.1 3.5 15.0 00 4.3 3.6 4.0 100 Degree (Naira) N 1851 4669 2175 4212 3550 10125 3294 3747 10000 1955 3293 4925 2490 56286 % 3.3 8.3 3.8 7.5 6.3 18.0 5.8 6.7 17.8 3.5 5.9 8.7 4.4 100 PGDE (Naira) N 3000 4669 2300 37750 11100 3305 1600 20000 3380 6375 2140 95619 % 3.1 4.9 2.4 39.5 00 11.6 3.5 1.7 20.9 00 3.5 6.7 2.2 100.0
*Note: Exchange rate is not stable and can change after this date.
The data in Table IV showed that N57,880; N56,286 and N95,619 were the academic costs of teacher education program of Diploma, Bachelor and Post Graduate Diploma in Education respectively. Tuition fee constituted the highest contributor of private cost of teacher education to Diploma (30.4%). Post Graduate Diploma in Education programme. (39.5%) whereas tuition constituted only 7.5% of the academic cost of teacher education to Bachelor programme. The Diploma and Post Graduate Diploma in Education did not incur any cost in the areas of private lesson and excursion. However they constituted 6.3% and 3.5% respectively in the Bachelor programme. Cost of textbooks/learning materials constituted 18.0% of the private academic cost of Bachelor in Education programme thus ranking as the highest contributor to the private academic cost of teacher education. For the Diploma programme, four items of cost namely tuition fee, costs of textbooks/learning materials, stationary and project contributed more than 10% each to the private cost of that programme. Other items such as fees for pre-admission clearance, registration, costs of teaching practice, photocopying, term paper and contribution towards group work contributed less than 9% of each item to the private academic cost of Diploma Teacher Education Programme. As regards the Bachelor Teacher Education Programme, two items of cost namely cost of textbooks/learning materials and cost of project constituted more than 15% each to the private academic cost of the programme. Each of the other items contributed less than 9% each to the total academic cost of Bachelor teacher education. The Post Graduate Diploma in Education (PGDE) had three items of cost that contributed more than 10% each to the academic costs of teacher education. They are tuition fee (39.5%), cost of textbooks/learning materials (11.6%) and cost of project (20.9%). Other items of cost contributed less than 7% each. Research Question Three What is the annual average private incidental cost of teacher education? The research question was answered by extracting data of items 14 to 21 from Table III and displayed in Table V.
123
Table V:

Annual average private Incidental cost of teacher education programme in Nigeria. *$1= N137.50K (August 26th, 2005)
S/N 1 2 3 4 5 6 7
Items of cost Cost of transportation Subscription to unions Cost of journeys home and back Out of campus accomodation Cost of clothing Cost of feeding Cost of telephone calls Total
Diploma (Naira) N 9770 600 7646 25000 28137 32575 3061 106789
% 9.1 0.6 7.2 23.4 26.3 30.5 2.9 100
Degree (Naira) N 5042 1000 9182 30000 34194 38377 5328 123123
% 4.1 0.8 7.5 24.4 27.8 31.2 4.3 100
PGDE (Naira) N 11200 20450 25000 20200 48000 15300 140150
% 8.0 0.0 14.6 17.8 14.4 34.2 10.9 100
*Note: Exchange rate is not stable and can change after this date.
The data in Table 5 revealed that N106,789; N123,123 and N140,150 were the incidental costs of teacher education programme for Diploma, Bachelor and Post Graduate Diploma of education programmes respectively. The cost of feeding contributed more than 30% of the incidental cost of teacher education in the three programs under study thus ranking first among the contributors. Post Graduate Diploma in education did not incur any cost as regards subscription to unions. For the other two programmes, Diploma and Bachelor, the subscription to union was negligible contributing less than 1.0% to the total incidental cost of teacher education. In the area of incidental cost three items were towering, they are out of campus accommodation, cost of clothing and cost of feeding. These items contributed more than 14% each to the incidental cost of teacher education in the three programmes. Cost of journeys home and back during the duration of the programme contributed 14.6% to the incidental cost for Post Graduate Diploma in education. Cost of telephone calls contributed 2.9% and 4.3% to the incidental cost of Diploma and Bachelor programmes. It contributed up to 10.9% in the cost of Post Graduate Diploma in education. Research Question Four: Is there any difference in cost among the programmes enrolled in? In order to answer this question, recourse was made to the data in table III, IV and V. Relevant data were extracted and displayed in table VI.
Table VI: Variation In Cost Among and within Teacher Education Programmes in University of Benin, Nigeria. (In Naira (N))
Type of Cost Types of Programme and Costs Diploma Bachelor N57,880 (35.1) N56,286 (31.4) N106,789 (64.9) N123,123 (68.6) N164,669 (100) N179,409 (100)
Academic Incidental Total Private Cost

Note: Figures in parentheses show percentages.
PGDE N95,610 (40.6) N140,150 (59.4) N235,769 (100)
Variations exist among the programs enrolled in the said university as displayed in Table VI. Academic cost contributed 35.1% of the total cost in the case of Diploma programme, 31.4% of Bachelor programme and 40.6% of PGDE programme. On the other hand, the incidental cost accounted for 64.9% of the cost of Diploma programme, 68.6% of Bachelor programme and 59.4% of PGDE programme of the total private cost of teacher education. Post Graduate Diploma in Education (PGDE) incurred the highest cost among the three programmes N235,769. This is followed by N179,409 for the Bachelor programme and N164,669 for the Diploma programme.
124
Discussion of Results
The findings of research question one indicated that the private costs of teacher education in University of Benin were N164,669; N179,409 and N235,769 for Diploma, Bachelor and Post Graduate Diploma programmes respectively. Cost of feeding accounted for more than 19% of the total private cost in each of the programmes. It ranked first among other items of cost. In this study, expenditure on food is given its recognition. In Salims study on the private cost of higher Education in kerala, he excluded expenditure on food because he felt that had the student remained at home he would have eaten food also. However this study has decided to give cost of feeding its recognition because the absence of one person in a Nigerian home does not substantially reduce the cost of feeding in the home. Post Graduate Diploma in Education recorded the highest tuition cost which is about twice the Diploma and eight times the degree programme. Majority of students in the PGDE are adults and are already raising families. Consequently, they find it difficult to stay in school without visiting home. Moreover many of them are already employed and are on study leave. They therefore need to visit their offices especially at the end of the month to collect their salaries. In Nigerian universities, tuition fee of Post Graduate programme is higher than Diploma and Bachelor programmes. This study thus agrees with Coombs and Hallaks (1972) assertion that educational cost varies greatly within the same institution and among different programmes. In University of Benin where these three programmes exist, the study has shown that variations exist in the total private cost of Diploma, Bachelor and Post Graduate Diploma Programmes. Even within the same programme, variations exist between the different items of cost. The findings of research question two indicated that in University of Benin, the academic cost per student in the Post Graduate Diploma in Education is about 16.24% higher than in the Diploma and Bachelor programmes. As for the components, the Diploma and Post Graduate Diploma students incurred the highest amount on tuition fees. Tuition fee accounted for 30.4% of the academic cost of Diploma programme and 39.5% of the academic cost of Post Graduate Diploma in Education. In the two programmes, tuition fee ranked first among other items of cost. In the Bachelor programme, tuition fee accounted for only 7.5% of the academic cost of teacher education. Cost of private lesson and excursion were only peculiar to Bachelor students and the costs were low 6.3% and 3.5% respectively. Other items of cost that were quite substantial are cost of textbooks/learning materials and cost of project. In the three programmes, cost of textbooks/learning materials accounted for more than 11% of the total academic cost; cost of project accounted for more than 14.96% of the academic cost. The difference in school fees is due to the fact that University of Benin is a federal institution and so degree programmes are highly subsidized by the federal government. The Diploma and Postgraduate Diploma programmes are expected to be self-sustaining and not the responsibility of the federal government hence the relatively high tuition fee charged by the university in order to run the programmes. The Diploma programme provides opportunity for students to enter the university through direct entry. Despite its high charges, the Diploma programme has continued to enjoy a high patronage and so has become a veritable source of income for the University. Another reason for the high patronage is the fact that the Diploma certificate of University of Benin is recognized all over the country and beyond; so its acquisition is an asset for further studies. The Post Graduate Diploma programmes do not also enjoy federal government subsidy hence the high charges in tuition fee in order for the institution to sustain the programme. A lot of people in Nigeria, after acquiring non-teaching degrees from universities and Higher National Diploma from the polytechnics take up teaching appointments when they cannot obtain more competitive appointments. Thereafter, such people decide to pursue Post-Graduate Diploma in Education in order to be recognized as professional teachers. Cost of stationery was substantial for the Diploma programme. It accounted for 11.1% of the annual average private academic cost of Diploma programme. For the Degree and PGDE programmes, it accounted for only 5.9% and 3.5% of the programmes respectively. This sharp difference in cost can
125
be attributed to the fact that Diploma students are subjected more to series of assignments requiring the use of stationery. Due to the deficiencies in their entry certificates, they are milled through series of assignments to strengthen their academic capability. In all the three programmes, academic cost accounted for 35.1% in the case of Diploma programmes; 31.1% for Bachelor programme and 40.6% for PGDE. As earlier mentioned, the Bachelor program has the least academic cost because of its low school fees which is being subsidized by the federal government. The other two programs- Diploma and Post Graduate Diploma - are expected to be self- sustaining. An interesting observation from the data is the fact that academic cost of teacher education is higher in the Diploma program than in the degree program. This difference is accounted for by the high tuition fee paid by the Diploma students. Apart from cost of admission form, tuition fee and cost of stationary, the Bachelor programme incurred more cost in all the remaining items more than the Diploma programme. Another concern of the study was to find out the annual average private incidental cost. For the Diploma program it was N106,789; Bachelor programme N123,123 and PGDE N140,150. Incidental cost accounted for 64.9% for the Diploma programme; 68.6% for the Bachelor programme and 59.4% for the PGDE programme. Surprisingly, this aspect of cost- incidental cost which is not directly related to instruction, accounted for a higher percentage of the private cost of teacher education in all the three programmes under study. Among the items of incidental cost, cost of feeding ranked first in the three programmes. Salim (1993) in his study decided to exclude the cost of feeding. However it is being included here because its exclusion may lead to underestimation. It is true that if the student was at home he would still be fed. In Nigeria,it is economically cheaper to cook for a larger number of people than to cook for just one person. In that kind of situation, the exit of one child from the home may not necessarily result in a decline in the cost of feeding in that home. But to feed that single child differently in the university would necessitate noticeable extra expenditure on the purse of the home. Nwadiani (1993) complained that since the late 1970s the price of virtually everything has continued to rise in Nigeria. This therefore is demonstrated in Adekambis study on the analysis of private cost of Post Graduate Diploma in Education for 2000/2001 session. During this period, the cost of feeding was N20,738.82 per session while in the present study of 2002/2003 it was N48,000, thus showing a difference of N27,261.18. Also the academic cost showed the same scenario. In the year 2000/2001 academic session, tuition cost was N33,000 per session while in 2002/2003 session (the period under study) tuition cost was N37,750. Cost of textbooks and learning materials in 2000/2001 session was N7172. In 2002/2003 session it was N11100. These obvious differences in costs in such a short time is due largely to inflation. Although the incidental cost is not directly related to instruction yet it can make or mar the education pursuit of the students. For instance if a student is not well fed, he may not be healthy enough to go through the rigor of academic work. Shelter is also very important. These programmes are housed in different campuses. The Diploma and PGDE programmes are in Ekewan campus with low students population hence the rent is N25,000. Ugbowo campus is the main campus and has large student population. The demand for accommodation is therefore higher. The rent there ranges between N30,000 to N50,000 per session depending on the facilities in the buildings. In 1993, rent outside the university campus was N15,000 by 1999 it rose up to N20,000 per annum and today it is N30,000 and above. Cost of foodstuffs did not fare better. In 1993 a kilogram of beef was N35. By 1999 it rose up to N250 and today it is being sold for N450. Hitherto, the federal government took upon itself to feed Nigerian undergraduates. Unfortunately with the proliferation of universities and growing students population the federal government shelved that responsibility. The burden of feeding therefore shifted to the families. Cost of clothing took a recognizable percentage of the incidental cost of teacher education. For the Diploma students it accounted for 26.3%; Bachelor programme 27.8% and PGDE programme 14.4%. The contribution is higher for the Diploma and Bachelor programmes because the affected students are youths and are quite excited to find themselves in a federal university. Consequently, they
126
desire to look fine and well dressed. They are therefore quite inconsiderate while making demands for money to buy the clothes. However, the desire for more clothing is understandable because of the tropical weather in Nigeria, which leads to quick depreciation of clothes. For the PGDE, clothing only accounted for 14.4% of the cost of the programme. The students in this category are adults and have acquired reasonable number of clothes before the course. They have attended higher institutions before and so are not too excited over another round of higher education. Cost of telephone calls accounted for 10.9% of the cost of incidental cost of PGDE programme. This is higher than that incurred by the Diploma and Bachelor students. This is understandable because the students in the PGDE category are adults and need to keep contact with their immediate and extended family members and their places of work. The students of Diploma and Bachelor are not compelled to phone. Rather most of the time they receive phone calls. The findings of research question four revealed that there are differences in costs among and within the teacher education programmes as shown in table VI. Academic cost contributed less than 41% to the total private cost of teacher education programme whereas the incidental cost accounted for more than 59% of the total cost of teacher education programme. Despite the fact that incidental items are not directly related to instruction, they enhance the students ability to concentrate. These items such as food, clothing and shelter are primary needs that must be met in order to ensure good living that is required for good academic performance The incidental cost of the Bachelor programme is twice the academic cost. Also the incidental cost of the Diploma programme is almost twice its academic cost. In both academic and incidental costs, the PGDE programme incurred more costs than the other programmes per annum.
Implication for Educational Planning

The federal government resolved that all stakeholders in higher education should share in the cost of education by paying some fees in order to attain and sustain a reasonable level of funding of higher education in Nigeria. The government is interested in knowing the actual cost of producing graduates of various disciplines at all levels of tertiary education. This modest attempt in this work is in that direction. It has shown that the cost is substantial. It therefore implies that education as a course may not be able to attract enough people into its programmes because people will prefer to spend such enormous sum of money to pursue other more lucrative courses that will earn them more prestigious jobs.
Conclusion and Recommendations

Based on the findings of this study, it can be concluded that the annual financial burden on the families for teacher education is substantial for the Diploma, Bachelor and Post Graduate Diploma programmes. Out of the total private cost the PGDE students spent the largest amount in both incidental and academic costs. By way of total academic cost, school fees ranked first among the items of cost in the Diploma and Post Graduate Diploma programmes. Cost of textbooks/learning materials ranked first among the items of cost for Bachelor programme. Among the items of incidental cost, the largest amount is incurred on feeding in the three teacher education programmes. In fact it ranked first among the items of cost in the incidental categories. Therefore it is hereby recommended that multinational companies, banks and nongovernmental organizations may have to provide scholarship, loans and bursary for education students in Nigerian Universities. The amount should be substantial to take care of tuition fee, feeding and out of campus accommodation among others. Bursary and scholarship should not be for tuition fee alone as is currently being practiced. The record of students loans in developing countries is notoriously bad as opined by Altner and Kitaev(2001). However, they were quick to say that globalization and new
127
technologies have made it possible to track a defaulter anywhere in the world through the internet and other avenues. I also associate myself with that view. Further research should be focused on the cost of other courses in order to have a comprehensive cost of programmes in Nigeria
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] Adegoke, Bade (2002, June 7) Coping with the Challenges of distance learning in Nigeria: The Teachers Role A paper presented at the National Teachers Institute Graduation in Badagry. Reported in The Punch. Vol. 17 (18,578) p. 38. Adekanmbi, O. (2002) Analysis of Private Cost of Teacher Education (A Case Study of Post Graduate Diploma in Education, Ekewan Campus, University of Benin. An Unpublished Post Graduate Diploma in Technical Education Project, Institute of Education, University of Benin Admissions Office University of Benin (2002) UME clearance list for the 2001/2002 session. Ibid. (2002) for 2002/2003 session. Ibid. (2004) for 2003/2004 session Admissions Office University of Benin (2003) Diploma clearance list for the2002/2003 session. Altner, Dominique and Kitaev, Igor (2001) Why did student loans fail to work in the Philippines The Lessons Learned IIEP Newsletter Paris:International Institute for Educational Planning. Vol. XIX (4) Oct Dec. 2001 Blaug, M. (1974) An Economic Analysis of Personal Earnings in Thailand. Economic Development and Cultural Change, October. Coombs P.H. and Hallak J. (1971) Managing Educational Cost. London: Oxford University Press. Duret, Gabriele Gottelmann (2001) Teacher Management addressing the Challenges in IIEP Newsletter Paris: International Institute for Educational Planning. P.1. Vol. XIX, (2), April June 2001, P.1. Earley, Penelope M. (1998) Teacher Quality Enhancement Grants for States and partnerships American Assocation of Colleges for Teacher Education Publication, 1307 New York Avenue, N.W., Suite 3000, Washington, DC 20005 4701. Ebmeier, Howard and Others (1991) The Comparability and Adequacy of Financial Support for Schools of Education Journal of Teacher Education Vol. 42 (3) Pp. 226 39, May June 1991. Egbobamwonyi, Dorothy (2002, October 8th) Teaching profession in distress? The Nigerian Observer, Vol. 33,(9, 522) Federal Ministry of Education (2002) Plan of Action in the Resolutions of the National Summit on Higher Education. Abuja: Federal Ministry of Education. Hernes Gudmund (2001) Managing Learning by Managing Teachers in IIEP Newsletter, Paris: International Institute for Educational Planning.. Vol. XIX, No. 2, April June 2001, P. 2. Howard, Richard D., Hitz, Randy and Baker, larry (1998) A National Study Comparing the expenditures of Teacher Education Programs by Carnegie Classification and with other Disciplines Action in Teacher Education; Vol. 20, No. 3 Pp. 1-14. Fall 1998. Joint Admission and Matriculation Board (2001) List of UME Candidates that scored 200 and above for 2001/2002 session. Ibid. (2002) for 2002/2003 session. Ibid. (2003) for 2003/2004 session. Kitaev, Igor and Peano, Serge (2001) Come rain or shine, educational cost analysis wins through. IIEP Newsletter Paris: International Institute for Educational Planning Vol. XIX, (3) July September 2001. P.11. Koithari, V.N. and P.R. Panchamukhi (1980) Economies of Education: A Trend Report In ICSSR, A Survey of Research in Economics: Infrastructure, Allied, New Delhi.
128 [18] [19] [20] [21] [22] [23] [24] [25] [26] [27]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Ndiribe, Okey (2003) Nigerias performance in WAEC Exams Poorest in West Africa Vanguard Vol. 19 No. 51787, Thursday, February 13, p. 21. Nwadiani, Mon (1993): Cost in Education: Implication for Educational Management A Mimeograph for Post Secondary Education planning. Nwadiani, Mon (1993): Cost in Education: Implication for Educational Management. A Mimeograph. Benin City: Department of Educational Administration and foundation, University of Benin, Nigeria. P. 9. Obanewa, Olaitan, Afemikhe, O.A; Momoh, S.O. and Olubor, R.O. (2004) Edo State Secondary Education Strategy Study. Being a Project Commissioned by the Edo State Ministry of Education. Oranefo, P.C. (2003) Private Cost of Education in Nigerian University: A Case Study of Catholic Institute of West Africa Port Harcourt: An Unpublished PGDE Project, Institute of Education, University of Benin. Orhibhabor, Rose (2004) Teachers recruitment in Katsina: The Nigerian Observer, Tuesday, July 20, 2004, Vol. 33, No. 9979. Pandit, H.M. (1981) Cost of Education in Nigeria Memo. Ibadan: Department of Educational Management, University of Ibadan. Psacharopoulos, G. (1985) Education for Development. An Analysis of Investment Choices, New York: Oxford University Press for the World Bank. Salim, A. Abdul (1993) Private Cost of Higher Education in Kerala: Socio-Economic Analysis Journal of Higher Education Vol. 16. No. 3, Summer P. 443 451. Osole, Shobhana (2000) Trends in Private Sector Development in World bank Education Projects. Policy Research Working Paper Services. Research Report. World Bank Education Team and the Policy Research Dissemination Center.
Management Training in Small and Medium Sized Enterprises: A Study of North Cyprus
erife Zihni Eypolu Faculty of Economics and Administrative Sciences, Near East University, North Cyprus Tel: 0392 2236464 ext.271/278; Fax: 0392 2236461 E-mial: serifeeyupoglu@neu.edu.tr Abstract Not a great deal of research has been conducted on the topic of management training in small and medium sized enterprises (SMEs), but interest in this area is beginning to increase. It is important to have an understanding of the factors that have an influence on the take-up of management training in SMEs because SMEs play an important role in the economy of any country. This study investigates whether SMEs in North Cyprus conduct management training. A total of 78 SMEs took part in the study. Results indicate that management training was more widespread than had been anticipated. All of the sample SMEs considered that a link does exist between business success and management training and around 70% of them conducted management training. However the training conducted was related to the development of the technical abilities of the managers and not related to the development of their managerial competence. Also, it can be seen that industrial sector, the number of managers, and family management have a statistically significant influence on management training; however the number of years that the SME has been in service and the number of its employees do not. Keywords: Management training, small and medium sized enterprises, human resource management
Introduction
Since the late 1970s and early 1980s there has been a growing interest in the impact of small and medium sized enterprises (SMEs) on both local and national economies. Obviously, the role and importance of SMEs differ from one country to another, but the fact that they are key contributors to economic growth for any country cannot be denied. Storey (1994) has identified the desirability of a vibrant SME sector as a means of reducing unemployment, promoting flexibility and innovation, and improving the health of the economy. Of the 20 million or so firms in Europe, around 99% are small and medium sized that employ fewer than 250 people and account for two thirds of the European workforce (Gray, 2004). Upgrading the skills of all types of workers, including managers, is central to firm performance in knowledge-based economies. The quality of management is particularly important for SMEs, which must be able to adapt quickly to evolving markets and changing circumstances, but which often have limited resources. Studies indicate that there is a positive relationship between the degree of management training and the bottom-line performance of SMEs (Storey, 1994). This study aims to examine SMEs in Northern Cyprus in relation to management training. The main objectives are;
130
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Do SMEs in North Cyprus make investments in management training and development? If they do, what form of management training is conducted? Do SMEs in North Cyprus perceive a link between management training and business success?
A Review of the Literature

Small and Medium Sized Enterprises Defined Small businesses are the most common form of business worldwide, but researchers have always had difficulty in satisfactorily defining them and separating them from larger firms. Researchers do agree however that a small firm is not just one that is smaller than a larger organization; it is also a business that is often managed and run in ways fundamentally very different from a larger firm (Welsh and White, 1981). A variety of both qualitative and quantitative factors have been used to separate small firms from other businesses. From a qualitative perspective, the business has to be independently owned and operated, closely controlled and funded by the owner, and the principal decision-making functions must rest with the owner-manager (Schaper and Volley,2004). The small business can also be defined using quantitative indicators, such as number of employees, value of assets, turnover, and share of ownership retained by the owner-manager (Schaper and Volley,2004). Management Training in SMEs Environmental changes such as globalization and technological developments affect businesses all around the world. However, such changes pose a bigger challenge to SMEs. While SMEs have been a popular area for research, few studies appear to have specifically examined the role of managerial training within SMEs (Winch and McDonald, 1999). The studies that have been conducted show evidence of a postive impact of management and employee training on SMEs. Research conducted in countries like Canada, Finland, Germany, and the U.K. indicate that formal management training can reduce the failure rates of small firms, particularly in their early years, and failure rates could fall from one in three in the first three years to one in ten where training was undertaken (Storey, 1994) Also, Demich and OReilly (2000) indicate that provision of management training enhances long-term strategies and managerial competence of SMEs. Training has largely been associated with large firms, with SMEs lagging behind (Boocock et al., 1999). Research confirms that even with SMEs there is a direct relationship between size and training, where larger SMEs conduct more training than micro firms which report low levels of training (Hughes, 1997; Curran et al., 1997). In a study conducted in the U.K. by Loan-Clarke et al (1999) results indicated that the take-up of management training was more widespread among SMEs than has previously been thought, and that a number of characteristics, namely ownership, family management, size, and number of managers, amongst others, strongly influenced management training investment. Definitions of management training are not uniform when one examines the literature, however most definitions refer to management training is relation to the development of the managerial skills and abilities of owner-managers and/or managers of SMEs. In this study management training refers to any form of training given to the owner-managers and/or managers of SMEs, thus this may include the development of managerial skills, technical skills, and such.
SME in North Cyprus

The North Cyprus State Planning Organization (SPO) (1998) gives the following definitions for SMEs and is the definition used in this study; Micro enterprises: less than 5 employees
131
Small enterprises: 5-9 employees Medium enterprises: 10-49 employees Large enterprises: 50 and more employees. Small and medium sized enterprises make up almost all of the North Cyprus private sector. According to SPO (1998) of the private enterprises in North Cyprus 88.5% are micro enterprises, 6.9% are small, 3.9% are medium, and the remaining 0.7% large. Around 80% of these enterprises are sole proprietors, or family owned and operated (SPO, 1998). When it comes to sectorial distribution, micro and small enterprises are concentrated mainly in trade, and medium and large enterprises are concentrated mainly in manufacturing. SMEs contribution to employment in North Cyprus is significant. In 1998, micro organizations employed 40.8% of the total employment in the private sector, small and medium organizations employed 38%, and large organizations employed 21.2% (SPO, 1998). Due to many constraints businesses in North Cyprus have difficulties in growing, thus almost all of the businesses are SMEs. Limited resources and lack of demand due to the small population negatively influences the efficiency and diversification of production consequently increasing the extent of imports of final and intermediate products (afakl and Gryay, 2004). Similarly, North Cyprus businesses serve local markets and lacks competitiveness due to diseconomies of scale. (afakl and Gryay, 2004). Other important factors limiting the growth of local businesses are the economic and political embargoes, and the international non-recognition of the Turkish Republic of Northern Cyprus (TRNC) (afakl and Gryay, 2004). Therefore, most businesses in North Cyprus are SMEs, but are so not by choice.
Research Methodology
Based on the literature, and with special emphasis to the Loan-Clarke et al (1999) study, a range of independent variables considered to potentially influence management training were derived. The independent variables selected were those considered to be appropriate for SMEs in North Cyprus, namely; the number of employees (employee size), family management (family management), industrial sector (sector), age of the enterprise (years in service), and the number of managers (management size); The dependent variable for the study being the take-up of management training (management training). Research has shown that the bigger the enterprise (employee size) the more likely it is to conduct management training (Marshall et al., 1995; Loan-Clarke et al., 1999). Likewise, the greater the number of managers (management size) it is assumed that the firm is more likely to conduct some form of management training. Where the founding family is involved in the management of the business (family management) there tends to be less management training (Loan-Clarke et al., 1999). Different industrial sectors have different training requirements and thus place different emphasis on management training (Wong et al., 1997; Loan-Clarke et al., 1999). The benefits of training are generally considered to occur in the long run, therefore it is argued that older organizations (years in service) are more likely to invest in management training (Storey, 1994; Loan-Clarke et al., 1999). To ascertain whether SMEs in North Cyprus conduct any form of management training a questionnaire (adapted from the survey instrument used in the Loan-Clarke et al., 1999 study) was designed. A mix of closed and open ended questions were used. The original questionnaire was designed in English and then translated into Turkish, the local language of North Cyprus. In order to pre-test it, it was administered to 5 managers/owners from 5 different SMEs. Issues assessed during the pilot test were instructions and statement clarity, questionnaire layout and appearance, and the length
132
of the questionnaire. Improvements were made based on the comments of the owners/managers. The Cronbach alpha reliability coefficient of the survey items was 0.718 which is an acceptable value. Primary data was collected by applying a cross-sectional data collection method. Due to the reluctacy of Turkish Cypriot business owners to participate in such research, it was decided that an interviewer administered method be used. The interviewer administered method is time comsuming and expensive but was adopted to ensure as high a response rate as possible as well as for clarification of ambiguities related to the survey items (Churchill, 1995). A total of 150 owners/managers of SMEs were randomly selected and contacted, 78 of which agreed to take part in the study resulting in a 52% response rate. The data collection process was conducted over a period of several weeks (OctoberDecember 2006). The data collected was analyzed through the use of frequency distributions and regression analysis obtained from SPSS (Statistical Packages for Social Science). Regression analysis is the most widely used non-experimental data analysis technique (Mernard, 1995).
Results/Findings of the Study

The findings of the study have uncovered some interesting results which are summarized in the following section and are represented through the use of tables. Table 1 shows the demographic characteristics of the respondents and indicates that 32% of the respondent SMEs were from the manufacturing sector and the remaining 68% were from the service industry. Also, the table indicates that 5% of the respondent SMEs have been in operation for less than 1 year, 22% for 1 to 5 years, 36% 6 to 10 years, and 37% for more than 10 years. Additionally, 94% of the respondent SMEs have a member of the owner family in its management whereas 6% do not. Of the responding SMEs, 4% have less than 5 employees, 37% have 5 to 9 employees, 31% have 10 to 19 employees, 12% have 20 to 29 employees, 8% have 30 to 39 employees, and 9% have 40 to 49 employees. In other words, 4% are micro, 37% are small, and 59% are medium sized enterprises. Table 1 futher indicates that the greatest percentage of respondent SMEs have between 1 to 4 managers (97%) with only the remaining 3% having more than 4 managers in their organizations. Finally, almost 70% of the respondent SMEs conducted some form of management training with only just over 30% not taking up any management training.
133
Table 1:

Demographic Profile of Respondents
Frequency N=78 25 53 N=78 4 17 28 29 N=78 73 5 N=78 3 29 24 9 6 7 N=78 76 2 N=78 54 24 Percentage 32.1 67.9 5.1 21.8 35.9 37.2 93.6 6.4 3.8 37.2 30.8 11.5 7.7 9.0 97.4 2.6 69.2 30.8
Details Sector Manufacturing Service Years in Service Less than 1 Year 1-5 Years 6-10 Years More than 10 Years Family Management Family No Family Employee Size Less than 5 employees 5-9 employees 10-19 employees 20-29 employees 30-39 employees 40-49 employees Management Size 1-4 managers More than 4 managers Management Training Conduct Do Not Conduct
The model represented in Table 2 indicates that 26.1% of the variance (change) in the dependent variable (management training) can be explained by the independent variables (sector, age, family management, employee size, and management size). The model is statistically significant (F= 5.086, p<0.005). The independent variable sector, management size, and family management (ownership) are all statistically significantly related to the dependent variable management training. Sector has the greatest influence (=0.396, p<0.0001), followed by management size ( =0.319, p< 0.05), and then family management (ownership) (= -0.286, p<0.05). The direction of influence for sector and management size is positive but negative for family management. The variables years in service and employee size are not statistically significantly related to management training.
134
Table 2:

Regression Analysis Results
Model Summary
Model 1
R .511a
R Square .261
Adjusted R Square .210
Std. Error of the Estimate .413
a. Predictors: (Constant), Management size., Sector, Years in service, Family Management, Employee size.
ANOVAb Model 1 Sum of Squares 4.336 12.279 16.615 df 5 72 77 Mean Square .867 .171 F 5.086 Sig. .000a
Regression Residual Total
a. Predictors: (Constant), Management size., Sector, Years in service, Family Management, Employee size. b. Dependent Variable: Management training
Coefficientsa Unstandardized Coefficients B Std. Error .195 .281 .392 .105 -.009 .054 -.539 .219 .030 .041 .104 .039 Standardized Coefficients Beta .396 -.017 -.286 .087 .319
Model 1
t .696 3.735 -.161 -2.457 .741 2.651
(Constant) Sector Years in service Family Management Employee size. Management size.
Sig. .489 .000 .873 .016 .461 .010
a. Dependent Variable: Management training
The study also indicates that of the 70% that conduct management training, 20% are manufacturing firms and 80% are service firms. This supports Wong et al (1997) who states that service organizations may place grater emphasis on management training to maintain customer relations when compared to other firms. Additionally, of the SMEs that conduct management training 94% have members of the owner family in the management, and 6 % do not have a member of the owner family in the management. This is an interesting finding which does not fit the literature because according to Cromie et al (1995) family firms are less keen on management training than non-family firms. Furthermore, the training given to managers emphasizes technical skills development rather than managerial competence development. As Martin and Stains (1994) identifies, owners and partner place more emphasis on technical rather than managerial skills whereas non-owner managers reverse the relative importance of the two factors. Around 6% of the SMEs that take up management training have been in service for less than 1 years, 19% 1 to 5 years, 39% 6 to 10 years, and 37% have been in service for more than 10 years. This fits with Storeys (1994) suggestion that as the survival rates of small firm are significantly poorer than those of larger organizations, the small firm may not be in existence to see a return on management training therefore firms that have been in operation for longer are more likely to take-up management training. Of those that conduct management training, 4% of the SMEs have less than 5 employees, 37% 5 to 9 employees, 28% 10 to 19 employees, 13% 20 to 29 employees, 7% 30 to 39 employees,
135
and 11% have 40-49 employees. Perhaps the most consistent finding in SMEs research is that the larger the organization the more like it is to take-up management training (Marshall et al., 1995). This study indicates the same results by showing that the largest percentage of SMEs that conducted management training is medium sized. All of the sample SMEs considered there to be a link between business success and management training, and as previously mentioned, 70% of them conduct some form of management training. However, in the last 12 months only 11% of them have taken-up management training, with the majority of the training given being internal (93%). However, what is important to indicate is that the training given to managers in SMEs is technical, and not related to managerial competence. None of the respondent firms conduct any form of training for its managers related to the development of their managerial skills. According to Westhead and Storey (1997), one of the main reasons why SMEs do not take-up management training is ignorance. Ignorance refers to the lack of awareness by small business owners of the importance of training for skills development. So it seems that SMEs in North Cyprus are ignorant towards the importance of training for managerial skills development. Many writers consider that for businesses to be successful managers with adequate managerial knowledge and skills must support them (Drucker, 1954; Katz, 1974; MacMahon and Murphy, 1999). Also, various studies find that managerial problems are the root of the most frequently voiced reasons for the failure of the SMEs in developing countries (Burke and Collins, 2001). Therefore it is important that in order to develop managerial competencies and effectiveness it is necessary that managers become aware of the advantages of attending management training programs.
Conclusion
The main purpose of this study has been to shed some light on the take up of management trasining in SMEs in Northern Cyprus. It would seem that a significantly greater number of SMEs conduct some form of management training and that the take-up of management training in SMEs in North Cyprus is more widespread than had previously been thought. However, it would seem that SMEs are ignorant towards the importance of managerial competence in SMEs. Perhaps the most important finding of the study is the fact that the management training given to managers in SMEs is related to the development of their technical skills and not the development of their managerial competencies. This would seem to pose a disadvantage to SMEs due to managerial problems being the cause of many business failures, especially SMEs. After all, most SMEs in North Cyprus are small or medium sized not by choice but due to the political and economic constraints faced by the island, thus they have no choice but to be efficient in their operations. It is not possible for SMEs to overcome the political and economic constraints that they face however, enhancing their managerial competence and implementing contemporary management systems is possible.
136
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Boocock, J.G., Loan-Clarke, J., Smith, A.J., and Whittaker, J. (1999), Management training and development in small and medium-sized enterprises:an assessment of the effectiveness of training and enterprise councils in the east midlands, Journal of Small Business and Enterprise Development, Vol.6, No.2. Burke, S., and Collins, K. (2001), Gender differences in leadership styles and management skills, Journal of Women in Management Review, Vol.16, No.5. Cromie, S., Stephenson, B., and Montieth, D. (1995), The management of family firms:an empirical investigation, International Small Business Journal, Vol.13. Churchill, S. (1995), Projecting a career:industry and education work together in Bexley, Journal of Education and Training, Vol.37, No.5. Curran, J., Blackburn, R., Kitching, J., and North, J. (1997), Small firms and workforce training: Some results. Demich, D.H., and OReilly. (2000), Supporting SMEs internationisation:a collaborative project for accelerated export development, Irish Marketing Review, Vol. 13, No. 1. 7. Drucker, P.F. (1954), The Practice of Management, Harper and Row, New York, NY. Gray, G. (2004), Management development in european small and medium enterprises, Advances in Developing Human Resources, Vol. 6, No.4. Hughes, A. (1997), Small firms and unemployment, ESRC center for Business Research Working Paper:WP71, University of Cambridge. Katz, R.L. (1974), Skills of an effective administrator, Harvard Business Review, Vol. 52, September/October. Loan-Clarke, J., Boocock, G., Smith, A., and Whittaker, J. (1999), Investments in management training and development by small businesses, Employee Relations, Bradford, Vol. 21, No.3. MacMahon, J., and Murphy, E. (1999), Managerial effectiveness in small enterprises:implications for HRD, Journal of European Industrial Training, Vol. 123, No. 1. Marshall, J.N., Alderman, N., Wong, C., and Thwaites, A. (1995), The impact of management training and development on small and medium sized enterprises, International Small Business Journal, Vol.13. Martin, G., and Staines, H. (1994), Managerial competencies in small firms, Journal of Management Development, Vol.13, No.7. Mernard, S. (1995), Applied Logistic Regression Analysis, Sage, Thousand Oaks, CA, London, and New Delhi. North Cyprus State Planning Organisation, Industry and Business Census: 1998, Nicosia. afakl, O. & Gryay, E., The Possible Impacts Of EU Membership On The Competitiveness Of SMEs In The TRNC, Third International Symposium On Business Administration on The Future of the EU Competitive Power after the EU Enlargement, 27- 28 May, T.C. Canakkale Onsekiz Mart University Biga Faculty of Economics and Administrative Sciences & Silesian University in Opava School of Business Administration in Karvina, Biga/anakkale TURKEY, 2004. Schaper, M., and Volley, T. (2004), Entrepreneurship and Small Business:A Pasific Rim Perspective, John Wiley and Sons, Australia. Storey, D.J. (1994), Understanding the Small Business Sector, Routledge, London. Welsh, J.A., and White, J.F. (1981), A small business is not a big one, Harvard Business Review, Vol. 59, No. 4. Westhead, P., and Storey, D. (1997), Training Provisions and the Development of Small and Medium Sized Enterprises, WEE Publications. Westhead, P., and Storey, D. (1996), Management Training and small firm performance:why is the link so weak?, International Small Business Journal, Vol.17, No.4.
References
[1]
[2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
[18] [19] [20] [21] [22]
137 [23] [24]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Winch, G., and McDonald, J. (1999), SMEs in an environment of change:computer-based tools to aid learning and change management, Industrial and Commercial Training, Vol. 31, No. 2. Wong, C., Marshall, J.N., Alderman, N., and Thwaites, A. (1997), Management training in small and medium-sized enterprises:methodological and conceptual issues, The International Journal of Human Resource Management, Vol.8, No.1.
Attitudes Toward Large-Scale Risks: A Discrete Estimation using Dutch Data

Joseph G. Eisenhauer Professor and Chair of Economics, Wright State University Department of Economics, 248 Rike Hall, Raj Soin College of Business Wright State University, 3640 Colonel Glenn HighwayDayton Ohio 45435-0001 USA Tel: (937)775-3070 E-mail: joseph.eisenhauer@wright.edu Abstract The conventional measures of risk aversion developed by Pratt and Arrow, as well as Kimballs measures of prudence, assume differentiable utility functions and infinitesimal changes in wealth, and consequently yield strictly local assessments of preferences. Most of the meaningful risks encountered in practice, however, occur on a large-scale, making it necessary to measure preferences over broad intervals of wealth. To that end, the present paper uses discrete rather than continuous changes in utility to develop reduced-form index measures of risk aversion and prudence suitable for use with risks of any magnitude. Survey data from the Center for Economic Research (CentER) at Tilburg University in the Netherlands are then applied to the model to determine construct validity and illustrate mean and marginal demographic differences in attitudes toward large-scale risks. Keywords: Prudence, risk aversion, expected utility JEL Classification Codes: D81, criteria for decision-making under risk and uncertainty
I. Introduction
The fundamental indicators of aversion to risk are the concavity of utility and the convexity of marginal utility. A concave utility function implies that a fixed level of wealth, w, is preferred to stochastic wealth with an equivalent expected value; to gauge the concavity of the indirect utility function U(w), Pratt (1964) and Arrow (1971) introduced the now-familiar measures of absolute and relative risk aversion, a ( w) = U ( w) / U ( w) and r ( w) = wU ( w) / U ( w) , respectively. Convex marginal utility implies both a propensity for precautionary saving (Leland, 1968) and an aversion to downside risk in the sense that, for a given mean and variance, a symmetric or positively skewed wealth distribution is preferred to a negatively skewed distribution (Menezes, et al., 1980). To measure that convexity, Kimball (1990) defined the coefficients of absolute and relative prudence as ( w) = U ( w) / U ( w) and ( w) = wU ( w) / U ( w) , respectively. As applications of differential calculus, however, these traditional metrics require the utility function to be differentiable, and even then they are suitable only for assessing preferences at the point at which the derivatives are taken; they are therefore strictly local measures. More precisely, because a derivative evaluates the rate of change in a function as its argument changes by an increment that approaches zero in the limit, the Pratt-Arrow-Kimball instruments capture the curvature of utility only over an infinitesimally small interval of wealth, as each of the three authors duly emphasized. Pratt
139
(1964), for example, referred to a(w) and r(w) as measures of risk aversion in the small, noting that no comparable global measures were being introduced.1 Measuring attitudes toward risk at a single point is a rather curious convention, however, inasmuch as meaningful risk by its very nature involves potentially large fluctuations in wealth. Indeed, as Arrow (1971, p. 100) pointed out, for small amounts at risk, the utility function is approximately linear, and risk aversion disappears. That is, individuals are essentially risk neutral over small intervals of wealth, and display substantial degrees of risk aversion only when the stakes are large; Fullenkamp et al. (2003) have recently provided empirical evidence of this. Thus, important risksthose that induce precautionary saving, the purchase of insurance, or other behavioral responsesare generally of a magnitude far larger than the Pratt-Arrow and Kimball coefficients were designed to accommodate. Even where the risk is small, however, the process of weighting absolute risk aversion or prudence by wealth in order to construct the corresponding elasticity can result in enormous estimates of relative risk aversion or prudence, exaggerating an individuals preferences regarding trivial gambles. Moreover, the weighting process itself requires accurately measuring initial wealth, a task fraught with empirical difficulty. The present paper develops index measures of risk aversion and prudence that are not based on differential calculus and therefore are not restricted to miniscule risks. In contrast to the conventional measurement tools, the instruments proposed here gauge attitudes toward risks without explicit reference to initial wealth and without requiring an approximation of the utility function; they therefore facilitate empirical accuracy. Section 2 below outlines the theoretical model of aversion to large-scale risks, and section 3 does the same for prudence. Section 4 illustrates the application of the new measures using survey data from the Center for Economic Research (CentER) at Tilburg University in the Netherlands. A short conclusion is given in section 5.
2. Risk Aversion
Consider an individual with a concave von Neumann-Morgenstern utility function U defined over final wealth, who faces a gamble in which the probability of winning the wealth increment k is p, and the probability of losing z is 1 p. Then the expected value of the risk is given by = pk (1 p ) z and its variance is 2 = p (1 p )( z + k ) 2 . With an endowment of initial wealth w, the risk premium is implicitly defined by the expected utility equation U ( w + ) = pU ( w + k ) + (1 p )U ( w z ) . (1) For the purpose of estimating the individuals preferences regarding risk, it has become common to use surveys, experiments, or empirical observation of insurance purchases to elicit information on either the risk premium for a given gamble, or the probabilities that establish equation (1) for a particular gain, loss, and risk premium; see, for example, Hartog, et al. (2002).2 Because utility is unobservable, however, its functional form cannot be known a priori, and imposing a specific parametric form is highly restrictive, largely defeating the purpose of empirical investigation. Thus, in order to evaluate risk aversion on a non-parametric basis (i.e., when the functional form of utility is unknown), equation (1) is customarily approximated by way of a Taylor series expansion. However, doing so requires that utility be differentiable, and each derivative implicitly assumes that wealth increases by an increment that approaches zero in the limit; hence the procedure is only valid over an infinitesimally small range. If the intent is to measure the concavity of utility over the entire z + k interval and the increments to wealth are of sizable dimensions, differentiation is not a legitimate method. Additionally, other problems arise. Because the Taylor series
1 2
Pratt (1964) did observe that a utility function displaying local risk aversion at every level of wealth would exhibit global risk aversion, but no single measure of the latter was developed. For k > 0 and z > 0, equation (1) represents a speculative risk, in which either a gain or a loss is possible. In practice, individuals are often confronted by large-scale pure risks, in which the most favorable outcome is the absence of a loss rather than a gain. Such a situation simply implies a large value of z and a zero value of k.
140
expansion is made at a single point, one must decide whether to construct the approximation around w (initial wealth), w + (expected wealth), or some other level of wealth. There is also the question of how many terms to include in the Taylor series so that the residual is negligible. Pratts (1964) approximation expands the left-hand side of (1) to the first order ( U ) and the right-hand side to the second order ( U ); Szpiro (1991) suggests expanding the right-hand side to the third order, so as to include U terms. Eisenhauer (2003) shows that the approximation is biased if one side is expanded further than the other. Finally, the reduced-form coefficient of absolute risk aversion that results from the local approximation is not an elasticity, but rather is measured in the inverse of monetary units; it is therefore necessary to multiply absolute risk aversion by initial or expected wealth (depending upon where the approximation was taken) in order to obtain relative risk aversion. Thus, knowledge of the riski.e., the potential gain, the potential loss, and their respective probabilitiesis insufficient; the ability to accurately measure initial wealth is also required when using local measures. This raises numerous empirical questions, such as whether to consider only financial assets or include physical and intangible assets that may be more difficult to value, such as housing and real estate or human capital. Empirical estimates are highly sensitive to these measurement issues; see for example Bellante and Saba (1986). As an alternative to differential calculus, we follow Eisenhauer (2006) in employing a discrete approach that makes use of the entire wealth distribution for a given risk rather than a single point. For notational simplicity, let the increment to wealth that establishes certainty equivalence be c = ; equation (1) can then be rewritten as U ( w + c) = pU ( w + k ) + (1 p )U ( w z ) . (2) We now define marginal utility (MU) as the discrete change in utility per unit of wealth.3 Taking differences among the three levels of utility identified in equation (2) yields U ( w + c) U ( w z ) MU cz = , (3) c+z and U ( w + k ) U ( w + c) MU kc = . (4) k c The weighted average of the marginal utilities in (3) and (4) is given by U ( w + k ) U ( w z ) (k c) MU kc + (c + z ) MU cz MU kz = . (5) = k+z k+z Equation (2) can now be rearranged to give U ( w + c) U ( w z ) = p[U ( w + k ) U ( w z )] , or equivalently, (c + z ) MU cz = p(k + z ) MU kz , (6) which yields the ratio MU cz p(k + z ) = . (7) MU kz c+z Similarly, substituting between (5) and (6) results in MU kc (1 p)(k + z ) = . (8) MU kz k c From (7) and (8), the percentage change in MU over the entire interval from w z to w + k can be calculated as the difference in marginal utility divided by the average value of marginal utility, and
3
In a calculus-based model, the marginal utility of wealth is defined as the first derivative of utility, U ( w) = lim[U ( w + h) U ( w)] / h . Discrete marginal utility, MU = [U ( w + h) U ( w)] / h , is simply a more
h 0
general notion which does not require arbitrarily small values of the wealth increment h.
141
interpreted as a measure of risk aversion in the large: R = ( MU kc MU cz ) / MU kz . This can be expressed in reduced form as (k + z ) . (9) R= 2 + (2 + z k ) 2 Of course, a discrete measure of risk tolerance can be obtained by simply inverting (9). Notice that the (k + z) term in the numerator of (9) equals the range of wealth outcomes between the favorable and unfavorable states [(w + k) (w z)], and the (2 + z k) term in the denominator is the multiplicative product of that range and the difference in the probabilities of winning and losing the wager: (2 + z k ) = [ p (1 p )][( w + k ) ( w z )] . It should be emphasized that this measure is nonparametric in the sense that the functional form of utility need not be known, and it is exact in that it does not depend on a Taylor series or other approximation.4 Because R increases (decreases) with an increase (decrease) in the risk premium, it preserves the preference ordering generated by , and behaves with respect to wealth as absolute risk aversion in the small behavesincreasing, decreasing, or remaining constant as initial wealth rises when the underlying utility function exhibits increasing, decreasing, or constant local absolute risk aversion, respectively. Yet because both the numerator and denominator are measured in squared monetary units, R is a pure index numberi.e., it achieves the same invariance with respect to units of measurement as relative risk aversion in the small, but without resorting to the process of weighting a more basic measure by wealth. Indeed, unlike relative risk aversion, R can be measured empirically without specific information on initial wealth. In contrast to the conventional measurement devices, R is not a point estimate of the concavity of utility at a particular level of wealth. Rather, it is an interval measure that reflects an individuals attitude to the specific risk at hand. Nor does a positive value of R necessarily imply that the individual is globally risk averse in the sense of having a utility function that is concave at every conceivable level of wealth. Instead, R > 0 indicates that the utility function is broadly concave between the most and least favorable states of nature that could result from the prospective wager. Over shorter or longer spans of wealth, the utility function could display greater or lesser concavity, linearity, or even convexity. Before illustrating the calculation of R with survey data in section 4, we extend the model further to construct an empirically estimable measure of prudence in the large.
3. Prudence
The conventional approach to estimating prudence when the functional form of utility is unknown has been to expand an Euler equation by way of a Taylor series in order to obtain the necessary second and third derivatives of utility. That approach, however, is subject to precisely the same limitations as the estimation of local risk aversion: it is valid only at a single point (i.e., over an infinitesimal wealth interval) and initial wealth must be measured with precision. In practice, such an approximation is often highly inaccurate; Ludvigson and Paxson (2001), for example, demonstrate that linearized Euler equations provide extremely biased estimates of prudence. Indeed, the empirical challenges are so formidable that nearly two decades after Kimball (1990) formalized the concept, only a handful of studies have attempted to estimate prudence.5 Consider instead a second lottery, similar to that presented in equation (2):
4
Indeed, because U(w) is a differential approximation to the slope of a secant, the accuracy of which depends upon the magnitude of the wealth increment, the conventional measures of risk aversion based on differential calculus are essentially approximations, even when the functional form of utility is known. Dynan (1993) found implausibly low and statistically insignificant values. Merrigan and Normandin (1996) obtained a range for relative prudence from less than 1 to 2.4, Eisenhauer (2000) found a range from 1.5 to 5.1, Eisenhauer and Ventura (2003) reported estimates that were generally between 7 and 9, and Terazona-Gomez (2004) obtained results ranging from about 1.25 to 2. Those are, of course, all localized point estimates rather than interval estimates.
142
U ( w + v) = qU ( w + k ) + (1 q )U ( w z ) , (10) where v c and q p. Notice that the potential gain and loss are the same as those in (2) but the certainty-equivalent has been altered, along with the probabilities of gain and loss.6 Taken together, equations (2) and (10) identify four unique levels of utility. Without loss of generality, let v < c. Then U(w z) < U(w + v) < U(w + c) < U(w + k). (11) Taking discrete differences yields three measurements of marginal utility per unit of wealth, as follows. U ( w + v) U ( w z ) MU 1 = , (12) v+z U ( w + c) U ( w + v) MU 2 = , (13) cv and U ( w + k ) U ( w + c) MU 3 = . (14) k c Two discrete changes in marginal utility, denoted MU, can then be defined from equations (12) through (14) as U ( w + c ) U ( w + v ) U ( w + v ) U ( w z ) (15) MU 1 = MU 2 MU 1 = cv v+z and U ( w + k ) U ( w + c ) U ( w + c ) U ( w + v ) (16) MU 2 = MU 3 MU 2 = . k c cv Combining (2) and (10) to eliminate the U(w + k) terms yields (p q)U(w z) = pU(w + v) qU(w + c); (17) similarly, combining (2) and (10) to eliminate the U(w z) terms gives (p q)U(w + k) = (1 q)U(w + c) (1 p)U(w + v). (18) Equation (17) can now be used to replace the U(w z) term in (15) and equation (18) can be used to substitute for the U(w + k) term in (16). Then after some elementary algebraic manipulation, MU1 can be expressed in terms of MU2 as (v + z)(p q)MU1 = [(p q)(v + z) q(c v)]MU2 (19) and equation (16) likewise becomes (20) (p q)(k c)MU2 = [(1 p)(c v) (p q)(k c)]MU2 Together, (19) and (20) imply (21) xMU1 = yMU2 where x = (v + z)[(q p)k + (1 q)c (1 p)v] (22) and y = (k c)[p(v + z) q(c + z)]. (23) Because MU1 involves the wealth range (w + c) (w z) = c + z and MU2 covers the interval (w + k) (w + v) = k v, the weighted average of these two changes in marginal utility, MUA = m1MU1 + m2MU2, (24) is obtained by applying the following weights:
6
Another measurement of risk aversion can be obtained from equation (10) using the procedure outlined in Section 2 above. If the second lottery is less (more) favorable to the individual than the first, i.e., if v < (>) c, then aversion to the second risk will be greater (lower) than aversion to the first. Note also that using two measurements of risk aversion to obtain a measurement of prudence has a clear analogue among local measures. The derivative of absolute risk aversion with respect to wealth is a( w) = a ( w)[a ( w) ( w)] . Absolute prudence in the small can therefore be written as ( w) = a( w) [a ( w) / a( w)] , and interpreted as the average value of absolute risk aversion minus its percentage change over a tiny interval around w.
143 m1 = and
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) c+z ( w + c) ( w z ) = [( w + k ) ( w + v)] + [( w + c) ( w z )] k v + c + z (25)
k v ( w + k ) ( w + v) = . [( w + k ) ( w + v)] + [( w + c) ( w z )] k v + c + z Substituting between equations (21) and (24) gives the following ratios: MU 1 / MU A = y /(m1 y + m2 x) m2 =
(26)
(27)
and MU 2 / MU A = x /(m1 y + m2 x) . (28) Finally, the measure of large-scale prudence is defined as a discrete analogue of local prudence, P = (MU2 MU1)/MUA. (29) Substituting (27) and (28) into (29) allows us to express prudence as a function of the observable parameters defined by (22), (23), (25), and (26): ( y x) . (30) P= m1 y + m2 x Although initial wealth generally influences the c and v terms, as an empirical matter it is not necessary to measure w in order to estimate (30); an investigator only needs to know the parameters of the two risks (z, k, p, and q) and the magnitudes of their certainty-equivalents (c and v). This feature provides enormous empirical advantages over existing methods.7
4. Empirical Measurement
The application of the risk aversion measure in (9) and the prudence measure in (30) are now illustrated using the 1993 wave of the CentER Savings Survey (CSS) from Tilburg University in the Netherlands. The survey asked several variations of the following question. Imagine you have won Dfl 5,000 in a game. You can now choose between keeping that Dfl 5,000, or having a lottery ticket with a certain chance to win a prize of Dfl 20,000. How high would that chance to win Dfl 20,000 need to be such that you would prefer the lottery ticket to keeping the Dfl 5,000 that you had already won?8 The use of probability-equivalence questions of this type was anticipated by Arrow (1971), but it is only in recent years that such survey instruments have become popular vehicles for assessing risk preferences. Data from the CentER survey have been analyzed by Warneryd (1996), Donkers et al. (2001), and Eisenhauer (2005), while similar questions from other surveys have been used by Barsky, et al. (1997), Pennings and Smidts (2000), Hartog et al. (2002), Guiso and Paiella (2003), and Eisenhauer and Ventura (2003), among others. The present study is unique, however, in applying such data to models of risk aversion and prudence in the large.9
7
Eisenhauer (2006) suggests an alternative measure of prudence in the large that relies on the precautionary premium as defined by Kimball (1990). Such a measure can be implemented empirically only if precise information regarding precautionary saving can be obtained. The present measure avoids this difficulty, and instead makes use of information more commonly available in surveys. The CentER Saving Survey was recently renamed the DNB Household Survey. The survey is available by permission of Tilburg University, at http://www.uvt.nl/centerdata/dhs. In the 1993 wave, a guilder or one Dutch florin (Dfl 1) was worth roughly US$0.50. As with any such survey, one might questions the respondents truthfulnessor more generally, the consistency between their answers and their actions outside the interview setting. This issue is, however, neither more nor less applicable to estimating the models developed here than to estimating conventional models of risk aversion or prudence in the small. Because our primary purpose is to offer empirically estimable, large-scale alternatives to the local measures, the empirical implementation undertaken here is merely illustrative. Thus, we shall not be overly concerned with respondent veracity, though we do demonstrate in Tables 2 and 3 below that the respondents answers to several related questions were highly consistent throughout the survey. Those who prefer experimental methods may note that
144
The answer to this particular survey question corresponds to the probability-equivalence parameter p in the model above. The value of k is 20,000 guilders, and z = 0. The expected value (in guilders) of the hypothetical lottery ticket is therefore = 20,000 p and its variance is
2 = p(1 p)20,000 2 . With c = 5,000 guilders as given in the survey question, the risk premium is given by = 20,000 p 5,000 so that a response of p =.25 implies risk neutrality, p >.25 implies risk
aversion, and p <.25 implies a love of risk. After substituting the parameter values into (9), aversion to this risk is measured by R = (400p 100)/75. For example, an individual who is indifferent between a risk-free Dfl 5,000 and a 47.5 percent chance of winning the lottery is found to have R = 1.2, whereas another who refuses to forego the certain money unless the probability of winning is 62.5 percent registers R = 2 . Note that the nature of this survey question places clear limits on risk aversion; in particular, 1.33 R < 0 for all 0 p < .25 , and 0 R 4 for all .25 p 1 . In another iteration of the survey question, the certainty equivalent increment was 200 guilders rather than 5,000. That alternative version of the lottery implies v = 200 and each respondents answer produces a value for the second probability-equivalence parameter q, such that q >.01 among risk averters. As might be expected, the vast majority of respondents exhibited aversion to both lotteries, and it is on their answers that our analysis is focused.10 In addition, even infinitely risk averse individuals can insist on no more than a one hundred percent probability of winning either lottery, so that p = 1 and q = 1 reflect corner solutions. Thus, attention is restricted to respondents for whom.25 < p < 1 and.01 < q < 1. This provides 2,087 responses on which both risk aversion and prudence are calculated. Table 1 reports the minimum, maximum, mean, and quartile values for R and P, along with their standard errors. The standard errors for both variables are quite small, indicating that the mean values, R = 2.17 and P = 3.97 respectively, are both significantly greater than zero in this cross-section. On average, the level of prudence derived from these survey responses exceeds the level of risk aversion, and prudence displays somewhat greater variability, though both R and P have fairly narrow ranges and inter-quartile ranges. A small minority (approximately 3.3 percent) of risk averters exhibited imprudence, having values of P < 0.
Table 1: Summary Statistics
Risk Aversion 1.3333 2.4000 2.9333 2.1742 0.0232 0.0500 3.9500 2,087 Prudence 3.3387 4.5005 4.9175 3.9697 0.0318 -2.0926 5.9549 2,087
1st quartile: 2nd quartile: 3rd quartile: Mean: Standard error: Minimum: Maximum: Sample size:
Because these discrete indices of risk aversion and prudence are created through rather intricate manipulation of survey responses, it is useful to test whether the resulting values are reasonably consistent with other available information on risk-taking and precautionary saving, respectively; several tests of construct validity were therefore conducted. Elsewhere in the CSS, the respondents
confronting subjects with two wagers of the type given in (2) and (10) and eliciting their behavioral responses would also yield the information needed to estimate (9) and (30). 82.5 percent of respondents exhibited risk aversion when the certainty equivalent was Dfl 5,000; 94.8 percent displayed an aversion to the lottery whose certainty equivalent was only Dfl 200. A third version of the survey question offered a Dfl 1,000 certainty equivalent; to keep the analysis as simple as possible and to focus on the largest differences, the intermediate lottery is excluded.
10
145
were asked to indicate their level of agreement on a scale from 1 to 7 with the statement, I would never consider investments in [stock] shares because I find this too risky. As shown in Table 2, those who agreed most strongly (rating a 4 or higher) were found to have significantly greater risk aversion than others. Similarly, when asked what level of investment risk they had undertaken over the past few years, those who replied that they had taken no risk at all or small risks every now and then showed significantly greater risk aversion as measured by R than those who took some risks and those who either sometimes or often took great risks. On the other hand, respondents were asked to indicate their level of agreement (from 1 to 7) with the statements, I am prepared to take the risk to lose money, when there is also a chance to gain money, and If I think an investment will be profitable, I am prepared to borrow money to make this investment. Those who rated their level of agreement a 4 or higher constituted a minority of relatively risk-tolerant individuals; as the table shows, the index R correctly confirms such individuals as having significantly lower risk aversion than their counterparts.
Table 2: Construct Validation for Risk Aversion
Sample size 976 727 618 251 606 1,097 704 999 Mean value 2.2142 2.0601 2.0539 1.8556 1.9921 2.2348 2.0541 2.2149 Standard error .03269 .03983 .04284 .06765 .04293 .03114 .03953 .03292 Two-tailed significance .003 .014 .000 .002
Belief/Characteristic Shares are too risky Agree Disagree Risk taken recently Little or none Some or great Will risk money for chance to gain Agree Disagree Would borrow for investment Agree Disagree
Likewise, the CSS posed several questions related to precautionary saving. Respondents were asked to rate the importance (on a scale from 1 to 7) of each of the following: To have some savings to cover unforeseen expenses as a consequence of illness or accidents, and to have enough money in my bank accounts to be sure I will be able to meet my financial liabilities. They also rated their agreement with the statements, I always keep some reserve money because my income varies during the year, and Being careful with money is an important character trait. Those who replied to these statements with ratings of 4 or more might reasonably be inferred to have strong inclinations toward precautionary saving; if our measure of large-scale prudence is valid, such individuals should have higher values of P than others. Table 3 reveals that there is indeed a significant difference in measured prudence among respondents, such that greater prudence is correlated with greater proclivities for precautionary saving.
146
Table 3:

Construct Validation for Prudence
Sample size 1,400 452 1,451 401 426 1,426 1,282 570 Mean value 3.9757 3.7748 3.9724 3.7610 4.0474 3.8905 3.9677 3.8344 Standard error .03906 .07134 .03843 .07561 .06902 .03949 .04033 .06478 Two-tailed significance .014 .013 .049 .081
Belief/Characteristic Precautionary saving against illness Important Unimportant Saving to be sure liabilities are met Important Unimportant Reserve money because my income varies Agree Disagree Being careful with money is important Agree Disagree
Table 4 provides some simple comparisons of means across demographic groups. Most notably, males were found to be significantly less risk averse and less prudent on average than females. This gender difference is one of the few results that have consistently appeared in studies of local risk aversion; see for example, Powell and Ansic (1997) and Jianakoplos and Bernasek (1998).11 Compared to younger respondents, the elderly (those aged 60 or over) were slightly more risk averse but not significantly so; however, they did demonstrate significantly greater prudence on average than the nonelderly. Those living with spouses or non-marital partners were less risk averse but not less prudent than others. On the other hand, those with children in the home were significantly more prudent, but not more risk averse than those without children. On average, heads of larger households (those having 5 or more members) were neither significantly more risk averse, nor more prudent, than heads of smaller households. The survey also separated respondents into two income groups: a representative panel and a high-income panel. The high-income group displayed significantly lower average values than the representative panel, which suggests that both risk aversion and downside risk aversion (i.e., prudence) may decline with income. Similarly, both attitudes appeared to decline with human capital as represented by education, in that those with college degrees exhibited significantly lower average values on both measures than their counterparts without degrees.
Table 4: Demographic Differences in Risk Aversion (R) and Prudence (P)
Sample size 902 1,185 297 1,790 1,820 267 1,002 1,085 255 1,832 853 1,234 847 1,240 Mean R (std. error) 2.2830 (.03479) 2.0914 (.03084) 2.1761(.06067) 2.1739(.02507) 2.1585(.02493) 2.2815(.06228) 2.1679(.03373) 2.1801(.03189) 2.2707(.06698) 2.1608(.02468) 2.0367(.03708) 2.2693(.02934) 2.0262(.03685) 2.2754(.02945) Two-tailed significance .000 .974 .067 .793 .124 .000 .000 Mean P (std. error) 4.2270(.04124) 3.7739(.04563) 4.0977(.08307) 3.9485(.03443) 3.9721(.03365) 3.9531(.09637) 4.0515(.04396) 3.8942(.04571) 4.0541(.08718) 3.9579(.03416) 3.7823(.05384) 4.0993(.03847) 3.6562(.05567) 4.1839(.03651) Two-tailed significance .000 .098 .852 .013 .305 .000 .000
Characteristic Female Male Age >= 60 Age < 60 Spouse/Partner No partner Children No children Household >= 5 Household < 5 High-income Lower income College or University No college
11
Some possible explanations for this phenomenon include gender differences in biology, evolutionary development, and cultural expectations; for a succinct discussion, see Halek and Eisenhauer (2001).
147
A complementary perspective is obtained by considering the marginal effect of each factor in a multivariate context. Table 5 provides regression results for R and P, using the same independent variables as in Table 4, with the exception that age and household size are now continuous. The table reports outcomes both before and after the removal of the most insignificant variables through backward elimination. While the overall explanatory power of the independent variables is quite low in each case (as is common in cross-sections), a marginal increase in household size significantly increases both risk aversion and prudence, ceteris paribus, while an increase in either income or education has a significant negative effect.
Table 5: Regression Results for Risk Aversion and Prudence*
Risk Aversion 2.199 [.000] 0.16 [.001] -.001[.542] -.216[.100] -.104[.369] .134[.246] -.185[.000] -.174[.000] .024 1.803 2,087 Risk Aversion 2.256 [.000] .162[.001] -.125[.111] .036[.074] -.179[.000] -.176[.000] .025 1.804 2,087 Prudence 3.689[.000] .397[.000] .003[.261] -.098[.584] -.080[.584] .148[.341] -.229[.001] -.405[.000] .053 1.870 2,087 Prudence 3.890[.000] .391[.000]
Variable: Constant Female Age Spouse/Partner Children Household size High income College or University Adj. R-sq. Durbin-Watson Sample size
.059[.016] -.231[.001] -.401[.000] .053 1.868 2,087
* Two-tailed prob-values are given in brackets below coefficients.
5. Conclusion
While this appears to be the first paper to develop an empirically estimable measure of large-scale prudence, it is not the first to propose an alternative to the local measures of risk aversion. Szpiro (1991), for example, advocated employing a more extensive Taylor series approximation of (1), estimating local risk aversion at various wealth levels, and constructing a global risk aversion curve out of the resulting point estimates. Chander (2000) suggests that for a gamble with z = 0 and a positive probability of a positive gain (k), a parametric measure of large-scale risk aversion might be a ( w + k )[U ( w + k ) U ( w)] / U ( w + k ) , which must then be approximated by a Taylor series expansion in order to obtain a reduced-form expression. More recently, Caballe and Esteban (2007) have recommended using the infimum of absolute risk aversion as a measure of global risk aversion. The ongoing expansion of this literature indicates a continuing interest in finding suitable measures of attitudes toward large-scale risks. This is certainly a worthwhile pursuit; the most important pure risks encountered in practice, such as the possibility of automobile theft, identity theft, disability, or unemployment, as well as speculative risks such as entrepreneurial ventures and other investments, all involve potentially enormous quantities of wealth. However, any index that relies on taking derivatives of the utility function implicitly assumes that preferences are being measured over miniscule ranges of wealth. In that respect, the alternatives proposed to date generally share the principal limitation of the Pratt-Arrow-Kimball coefficients; employing such measurement instruments in the context of large-scale risk requires an extrapolation from a single point to a broad interval on the utility curve. Moreover, none of those alternatives have yet been empirically implemented. The present paper offers empirically estimable measures of risk aversion and prudence that are more suitable for use with large-scale risks. By soliciting information concerning the certaintyequivalent of a simple lottery and calculating discrete differences in the utility levels associated with the gain, loss, and certainty-equivalent positions, it is possible to obtain an exact measure of risk aversion without resorting to differential calculus. Combining that information with similar data from a second wager that is either more or less favorable to the individual than the first, it is possible to create
148
a similar measure of prudence. The implementation of these instruments has been demonstrated with survey data, and the results offer evidence that the coefficients constructed here generate valid measures of individual preferences. The empirical analysis, however, is merely illustrative. These models can and should be estimated more extensively with other surveys, as well as with experimental and observational dataa task deferred to future research.
Acknowledgement
The study uses data from the DNB Household Survey (formerly the CentER Savings Survey) at Tilburg University in the Netherlands. I thank Marcel Das for permission to use the data. Any errors are my own.
149
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Arrow, K. J., (1971). Essays in the Theory of Risk Bearing. Chicago: Markham. Barsky, R. B., Juster, F. T., Kimball, M. S., and Shapiro, M. D., (1997) Preference Parameters and Behavioral Heterogeneity: An Experimental Approach in the Health and Retirement Study Quarterly Journal of Economics, 112: 537-579. Bellante, D. and Saba, R. P., (1986) Human Capital and Life-Cycle Effects on Risk Aversion Journal of Financial Research, 9: 41-51. Caballe, J. and Esteban, J. M. (2007) Stochastic Dominance and Absolute Risk Aversion Social Choice and Welfare, 28: 89-110. Chander, P. (2000) A Simple Measure of Risk Aversion in the Large and an Application CORE discussion paper 41, August. Donkers, B., Melenberg, B., and van Soest, A. (2001), Estimating Risk Attitudes using Lotteries: A Large Sample Approach Journal of Risk and Uncertainty, 22: 165-195. Dynan, K. E. (1993), How Prudent are Consumers? Journal of Political Economy, 101: 11041113. Eisenhauer, J. G. (2000), Estimating Prudence, Eastern Economic Journal, 26: 379-392 Eisenhauer, J. G. (2003), Approximation Bias in Estimating Risk Aversion Economics Bulletin, 4: 1-10. Eisenhauer, J. G. (2005), How Prevalent are Friedman-Savage Utility Functions? Briefing Notes in Economics, 66: 1-7. Eisenhauer, J. G. (2006), Risk Aversion and Prudence in the Large Research in Economics, 60: 179-187. Eisenhauer, J. G. and Ventura, L. (2003), Survey Measures of Risk Aversion and Prudence Applied Economics, 35: 1477-1484. Fullenkamp, C., Tenorio, R., and Battalio, R., (2003) Assessing Individual Risk Attitudes using Field Data from Lottery Games Review of Economics and Statistics, 85: 218-225. Guiso, L. and Paiella, M., (2003), Risk Aversion, Wealth, and Background Risk Banca DItalia Discussion paper 483, September. Halek, M. and Eisenhauer, J. G., (2001), Demography of Risk Aversion Journal of Risk and Insurance, 68: 1-24. Hartog, J., Ferrer-I-Carbonell, A., and Jonker, N. (2002) Linking Measured Risk Aversion to Individual Characteristics Kyklos, 55: 3-26. Jianakoplos, N.A., and Bernasek, A. (1998), Are Women More Risk Averse? Economic Inquiry, 36: 620-630. Kimball, M. S. (1990) Precautionary Saving in the Small and in the Large Econometrica, 58: 53-73. Leland, H. E. (1968), Saving and Uncertainty: The Precautionary Demand for Saving Quarterly Journal of Economics, 82: 465-473. Ludvigson, S. and Paxson, C. H. (2001), Approximation Bias in Linearized Euler Equations Review of Economics and Statistics, 83: 242-256. Menezes, C., Geiss, C., and Tressler, J. (1980), Increasing Downside Risk American Economic Review, 70: 921-932. Merrigan, P. and Normandin, M. (1996), Precautionary Saving Motives: An Assessment from UK Time Series of Cross-Sections, Economic Journal, 106: 1193-1208. Pennings, J. M. E., and Smidts, A. (2000), Assessing the Construct Validity of Risk Attitude Management Science, 46: 1337-1348. Powell, M. and Ansic, D., (1997), Gender Differences in Risk Behaviour in Financial Decision-Making: An Experimental Analysis, Journal of Economic Psychology, 18: 605-628. Pratt, J. W. (1964), Risk Aversion in the Small and in the Large Econometrica 32: 122-136.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24] [25]
150 [26] [27] [28]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Szpiro, G. G. (1991), Risk Aversion as a Function of Variance and Skewness in: Progress in Decision, Utility, and Risk Theory, Chikan, A. (ed.), Dordrecht: Kluwer Academic Publishers: 355-363. Tarazona-Gomez, M. (2005), Are Individuals Prudent? An Experimental Approach using Lottery Choices University of Toulouse, LERNA working paper 05.13.177, September. Warneryd, K. (1996), Risk Attitudes and Risky Behavior Journal of Economic Psychology, 17: 749-770.
Cointegration and Priority Relationships between Stock Markets of Turkey, Brazil and Argentina
Erman Erbaykal Balikesir University Department of Economics and Administrative Sciences 10200 Bandirma, Balikesir, Turkey E-mail: eerbaykal@yahoo.com H. Aydn Okuyan Balikesir University Department of Economics and Administrative Sciences 10200 Bandirma, Balikesir, Turkey E-mail: aydinokuyan@hotmail.com zgr Kadolu Yildiz Technical University Department of Economics and Administrative Sciences Istanbu, Turkey E-mail: kadioglu_99@yahoo.com Abstract The aim of this study is to expose the relationships between Istanbul (Turkey), Merval (Argentina) and Bovespa (Brazil) stock exchange markets using monthly data covering the period 1997:03 2007:06. The long term co-movement of the stock exchanges has been tested by cointegration test developed by Johansen (1998) and Johansen Juselius (1990). Besides, the priority relationships and direction of the relation among the variables have been examined by impulse-response functions and variance decomposition. A long run relationship has been detected between the stock markets as a conclusion. According to the variance decomposition results, Bovespa has a major effect on ISE. In addition to this, even Bovespa has also an effect on Merval, it itself is not affected by the other two stock exchanges. Keywords: Stock Markets, Cointegration, Impulse-Response, Variance Decomposition JEL Classification Codes: C32, F41, F43
1. Introduction
The integration between the financial markets has fastened with the technological developments since the last quarter of 20th century. The restrictions on the capital flows have been lifted as a consequence of the changes during this period and interest in the foreign markets has risen. Especially the capital flow to developing countries is one of the most important reasons for market integrations. There are two main results of the increased integration between the markets: On the one side, legal barriers were broken, the transaction costs reduced and the efficiency of the markets increased. On the other side, the efficiency of the instruments used in portfolio diversification reduced (tak and Gzba, 2006:2). In this situation, the diversification possibility which can be done by exploiting the low correlation between international stock markets started to remove by and by.
152
With the fast integration process in the world, the number of studies in this field grew simultaneously. The primer of these studies was done by Grubel (1968). Grubel (1968) examined the stock market index portfolios of 11 countries and concluded that if portfolio had been diversified, a % 8 higher return could have been gained. Levy and Sarnat (1970), Panton et al.(1976) found evidence supporting Grubels (1968) results. Chan et al. (1997) investigated the integration degrees of financial markets in their study which contains the stock market data of 18 countries in the period 1961- 1992. According to the results obtained, they concluded that there is a very low level of integration between the markets. In the study investigating the relationships between the stock markets of USA, England, Japan, Germany, France and Canada, Yce (1997) stated a relationship between the stock markets of developed countries and added that stock markets of USA and Japan are the most dominant markets among the others. Flatzscher (2002) investigated the integration process of European stock markets and found out an increasing integration especially faster after the adoption of Euro as common currency. Adam et al. (2002), in their study of EU countries stock markets covering years 1994-2001, pointed out that integration reached to very high levels especially in 1997-1999 period and advanced in a stable level after this period. One of the last studies in this field done by Marashdeh (2006) investigated the integration of stock markets of Turkey, Jordan, Morocco and Egypt both among themselves and with the stock markets of USA, England and Germany in 1994- 2004 period and concluded that these countries are integrated among themselves but could not satisfy the integration process with the developed stock markets, except Egypt. Important studies have also been done in Turkey about the integration of stock markets. Mandac and Takn (2005) investigated whether returns of ISE (Istanbul Stock Exchange) and EU member countries stock markets move in the same direction in the period 1995-2004. They detected a high correlation among the stock markets of EU member countries but a low correlation with ISE contrarily. Bayri and Glolu (2005) investigated the integration level of ISE with the stock markets of EU and USA for both before and after 2001 crisis period. According to the evidence obtained, they claimed that there is a strong long run relationship between Turkey and both EU and USA markets. Besides, they figured out that Turkey is quiet sensible to the external shocks of German stock markets. tak and Gzba (2006) investigated the long run cointegration relationship between ISE and stock markets of USA, Germany, England, Japan, India, Malaysia in 1986-2006 period on the one hand and the effects of financial liberalization on integration in the mentioned period on the other hand. In the study, they concluded that there is an integration between ISE and England and Germany, and partially integration between ISE and USA and Indias basic indexes in 1986-2006 periods. In addition, in 2000- 2006 sub-period they detected integration between ISE and India, and partial integration between ISE and Japans basic indexes. In the light of results of studies in the literature, it is seen that integration among the stock markets in the world has risen since 1990s. However a concensus could not be reached about the relations of ISE with the other stock markets in the studies regarding Turkey. In this study, whether ISE moves in the same direction with Bovespa and Merval in the long run is investigated by cointegration test developed by Johansen (1998) and Johansen Juselius (1990). Also the interaction between the variables is examined by impulse-response functions and variance decomposition. The reason why Argentina and Brasil are chosen is their having similar economical structure to Turkeys. These three countries could not acquire stability and as a result, experienced important economic crisis. In addition to this, they emerged as an attraction center for portfolio investments flowing to developing countries. This paper consists of four sections. First section is the introduction. The model and data set is introduced in the second section. Section three presents the methodology and empirical results. Finally, section four concludes and evaluates the paper.
2. Model and the Data Set

The model employed in this study is as follows:
LTRt = 0 + 1 LBRt + 2 LARt + t
(1)
153
TR, BR and AR refer to the closing values of ISE 100 index, Bovespa index and Merval index respectively LTR, LBR and LAR imply that the natural logarithms of the variables are taken. 0 , 1 and 2 are the parameters to be estimated, and t is the error term. The variables in the study are composed of monthly observations in 1997:03-2007:06 period and obtained from EconStats database.
3. Methodology and the Empirical Results

3.1. Stationarity Test In time series analysis, stationarity of the series is examined by unit root tests. A time series is stationary if its average and variance do not change in time and if the common variance between two periods depends not on the calculated period but the distance between the periods (Gujarati, 1999:713). Variance and average of non-stationary series change in time and when time converges to infinity so variance does. Granger and Newbold (1974) showed that studying with non-stationary series; a spurious regression problem could be encountered. In this case, the results obtained by regression analysis do not reflect the accurate and real relationship. The most common used tests in literature determining the stationarity levels of series are ADF (1979) test develpoed by Dickey and Fuller and Phillips- Peron unit root test. These two tests are employed in this study for determining the stationarity level. Test results can be seen in Table 1.
Table 1:
Variables LTR LBR LAR Sig. level %1 %5 %10
Results of Unit Root Tests

Augmented Dickey Fuller (ADF) With trend No Trend Level First Diff. Level First Diff. -2.979 -8.316* -1.720 -8.306* -1.973 -11.727* -0.308 -11.685* -1.836 -10.360* -0.306 -10.237* Critical values -4.034 -3.484 -3.447 -2.885 -3.148 -2.579 Phillips Peron (PP) With Trend No Trend Level First Diff. Level First Diff. -2.634 -8.341* -1.595 -8.332* -1.947 -11.732* -0.230 -11.694* -1.684 -10.533* -0.152 -10.243* -4.034 -3.446 -3.148 -3.484 -2.885 -2.579
* %1 significance
According to the test results in Table 1, three of the variables are stationary in the first difference according to both ADF and PP unit root tests. Thus, the pre-condition to examine the cointegration relationship between the series by the cointegration test developed by Johansen (1988) and Johansen Juselius (1990) is satisfied. 3.2. Cointegration Test First, second, third, etc difference of series is taken in non-stationary series to satisfy stationarity. However, taking difference method does not only erase the effects of persistant shocks on the varible in the past periods but also may erase the long term relationships besides the shocks. Therefore, the regression between the series that are made stationary by this method would not explain the long term relatioship due to the terminated information belonging to long term. In this point, cointegration analysis says that even if economic variables are not stationary, there might exist a lineer combination of these series and if so, this might be modelled econometrically. In the cointegration approach developed by Engle and Granger (1987), time series which are not stationary in level but stationary in their first difference may be modelled in their level state and by this way, long term information loss can be prevented. However this approach is invalid if there are more than one cointegration vector. This situation may be overcomed by the help of approach of Johansen (1988) and Johansen and Juselius (1990) and VAR model which all the variables are taken granted as endogenous. By this method, number of cointegration vectors can be tested. Therefore, a more realistic examination can be
154
implemented without limiting the test in one cointegration vector expectation unlike the Engle-Granger method. Johansen cointegration test is composed of VAR estimation including the differences and levels of non-stationary time series. Lets consider two series named X and Y. In this case Z is a vector including X and Y series. According to Johansen cointegration approach, the VAR model would be as follows: t = 1 t 1 + .......... .......... + k 1 t k +1 + t k + t (2) Here i , (i=1,2,.,k-1) is the variables matrix of lags of first difference of t vector. refers to the levels of the variables and t represents the error-terms of VAR model. Johansen cointegration test is based on determining the rank of matrix. If the rank of matrix is zero, this implies that there is no cointegration relationship between the variables composing Z vector. If the rank of matrix is one, this means that there is a cointegration relationship between the series composing Z vector and finally if the rank is greater than one, there are more than one cointegration relationship between the series. In Johansen cointegration approach, cointegration relationship is examined by the help of trace and maximum eigenvalue test statistics. Trace test examines the null hypothesis saying rank of matrix is equal or less than r. Here, r represents the number of cointegration vectors. Maximum eigenvalue statistics tests the null hypothesis saying number of cointegration vectors is r against the alternative that says r+1. Null hypothesis is rejected if the calculated test statistics is greater than the critical valua at a given significance level. First of all, the lag number of the variables in VAR model should be identified in Johansen cointegration test. Akaike and Schwarz information criteria are used to determine the number of lags in this study and according to both criteria, number of lags in the model is 2. The results of cointegration test can be seen in Table 2.
Table 2: Results of Johansen Cointegration Test
Trace statistics 35.758* 8.860 1.685 %95 35.192 20.261 9.164 Max. Eigenvalue statistics 26.898* 7.174 1.685 %95 22.299 15.892 9.164
Rank r=0 r1 r2
*implies that related hypothesis is rejected at %5 significance level.
According to the results in Table 2, one cointegration vector is detected for both trace and maximum eigenvalue statistics at %5 significance level. In other words, null hypothesis, claiming there is no cointegration vector, is rejected according to two test statistics. In conclusion, there exists a long term relationship between the variables. 3.3. Vector Autoregressive Model (VAR) VAR models are basicly used in examining the dynamic effects of relations between variables and randam shocks on variables system (Greene, 1993:553). The variables in VAR model act in an integrated system and the model does not distinct the variables as indogenous or exogenous. Besides, these models do not require to depend on an economic base. Because there are only lagged values of endogenous variables in the right side, problem of being simultaneous is not confronted in VAR models. Impulse-response functions and variance decomposition can be obtained by the constructed VAR models. These two methods are used in this study to expose the interaction between the variables as well. 3.3.1. Impulse-Response Functions Impulse-response functions display the effect of a one unit standard deviation shock coming to one of the random error terms on the current and future values of endogenous variables. The most effective variable on a macroeconomic indicator is found out by variance decomposition and whether this
155
variable can be used as a policy instrument is determined by impulse-response functions (zgen and Glolu, 2004:97).
Figure 1: Response of Bovespa to ISE Figure 2: Response of Merval to ISE
Figure 3: Response of ISE to Merval
Figure 4: Response of Bovespa to Merval
Figure 5: Response of ISE to Bovespa
Figure 6: Response of Merval to Bovespa
Looking at figure 1, it is seen that Bovespa is unreactive to a shock on ISE. In figure 2, a shock in ISE results in a fall in Merval starting from the second month. In the end of the twelfth month, the fall sums up to nearly % 4. However, in figure 3, ISE stays unreactive to a shock on Merval. It is seen in figure 4 that a shock on Merval affects Bovespa positively in %2-%3 range. Figure 5 shows the response of ISE to a %4 shock on Bovespa. According to this, ISE started to rise since the first month and reached to % 9 levels in the second month. It remained in the same level in the third month and then even a fall is observed, it ended in % 5 levels rise in the twelfth month. Finally figure 6 shows the response of Merval to Bovespa. Merval started to rise in the beginning but then fell down since the second month. The effect of the shock completely died out in the end of twelfth month. 3.3.2. Variance Decomposition Variance decomposition decomposes the change in one of the endogenous variables as seperate shocks effecting all endogenous variables. Variance decomposition gives information about the dynamic structure of the system in this respect. The goal of the variance decomposition is to figure out the effect of a random shock on error variance of forecast for the future terms (zgen and Glolu, 2004:98). Variance decompositions of the three stock market indexes can be seen in table 3.
156
Table 3:

Variance Decompositions
LBR 100.0000 99.86347 99.58224 99.17924 98.67371 98.08317 97.42345 96.70881 95.95228 95.16578 94.36027 93.54579 15.57235 35.26151 41.21472 44.09914 45.88283 47.14548 48.11009 48.88015 49.51112 50.03627 50.47744 50.85005 26.69088 31.99529 31.74642 30.35690 28.65917 26.93127 25.28931 23.78557 22.44184 21.26355 20.24680 19.38229 LTR 0.000000 0.008882 0.011275 0.012125 0.013580 0.016828 0.023078 0.033645 0.049814 0.072696 0.103128 0.141618 84.42765 64.61023 58.58361 55.64735 53.82355 52.52937 51.54043 50.75226 50.10853 49.57512 49.12930 48.75483 0.003209 0.044000 0.211333 0.493865 0.868339 1.310946 1.800071 2.317036 2.846259 3.375155 3.893893 4.395070 LAR 0.000000 0.127645 0.406488 0.808630 1.312713 1.900000 2.553475 3.257543 3.997908 4.761527 5.536601 6.312588 0.000000 0.128260 0.201669 0.253511 0.293620 0.325146 0.349487 0.367595 0.380345 0.388608 0.393253 0.395116 73.30591 67.96071 68.04224 69.14923 70.47249 71.75778 72.91061 73.89740 74.71191 75.36129 75.85931 76.22264
Variance Decomposition of Bovespa Index Terms S.E. 1 0.103076 2 0.137168 3 0.160178 4 0.177441 5 0.190937 6 0.201724 7 0.210469 8 0.217629 9 0.223541 10 0.228455 11 0.232567 12 0.236030 Variance Decomposition of ISE 100 Index 1 0.103671 2 0.173854 3 0.217944 4 0.247258 5 0.267798 6 0.282718 7 0.293823 8 0.302233 9 0.308684 10 0.313681 11 0.317582 12 0.320646 Variance Decomposition of Merval Index 1 0.113666 2 0.160983 3 0.192260 4 0.215513 5 0.233840 6 0.248833 7 0.261426 8 0.272216 9 0.281601 10 0.289862 11 0.297197 12 0.303756
Examining Bovespa first, it is seen that Bovespa is affected by itself in % 95-%96 range in twelve months period. In other words, ISE and Merval do not have an important effect on Bovespa. Looking at ISE, even it seems like it is affected by itself at first sight, in the end of twelve month it is seen that Bovespa has a significant effect on ISE. Merval is affected by itself by %70 level and affected by Bovespa by % 20. In conclusion, according to the results of variance decomposition, even Bovespa is not affected by the other two indexes, it has an important effect on Merval and ISE.
4. Empirical Results and Evaluation

Cointegration and priority relationships among the stock markets ISE, Merval and Bovespa is investigated in this study. In conclusion, a long term relationship has been detected among these stock markets. The existence of this relationship implies that employing portfolio diversification among these stock markets would not be efficient. Also the interaction between the variables has been shown by the help of impulse- response functions and variance decomposition. According to the results of
157
variance decomposition, it has been figured out that Bovespa has a significant effect on both ISE and Merval. On the other hand, it is not affected by the mentioned two stock markets. Coming to Merval, while it affects ISE, it is not affected from ISE. Therefore, among these three markets, Bovespa appears to be the pivotal stock market. As a result, Bovespa and Merval can be exploited as an indicator for the investors of ISE. Besides, it can be said that portfolio diversification employed among these three markets might not provide the expected efficiency for the international portfolio investors.
References
[1] ADAM K., JAPPELLI T., MENICHINI A., PADULO M. and PAGANO M., (2002). Analyse, Compare and Apply Alternative Indicators and Monitoring Methodologies to Measure the Evolution of Capital Market Integration in the European Union, Centre for Studies in Economics and Finance (CSEF) Department of Economics and Statistics, University of Salerno, Report Prepared for EU Commission BAYR O. ve GLOLU B., (2005). Hisse Senedi ve Yabanc Para Piyasalarnn Entegrasyonu: Trkiye, AB ve ABD rnei, ktisat, letme ve Finans Dergisi, 234, 13 34 CHAN K.C., GUP B.E. and PAN M.S., (1997). International Stock Market Efficiency and Integration: A Study of Eighteen Nations, Journal of Business Finance and Accounting, 24 (6), 803 813 ITAK L. and GZBAI O., (2006). MKB ile Baz nde Gelen Gelimi ve Gelimekte Olan lke Borsalar Arasndaki Btnlemenin Temel Endeks ve Ana Sektr Endeksleri Temelinde Analizi, X. Ulusal Finans Sempozyumu, 1 4 Kasm 2006, zmir DICKEY, D. and FULLER,W. (1979) Distribution of the estimators for autoregressive time series with a unit root, Journal of American Statistical Association, Vol. 74, pp. 427-431. ENGEL, R.F. and GRANGER, C.W.J. (1987) Co-integration and Error Correction Representation, Estimation and Testing, Econometrica, Vol. 55, No.2, pp. 251-276. FLATZSCHER Marcel., (2002). Financial Market Integration in Europe: On Effects of EMU on Stock Markets, International Journal of Finance and Economics, 7, 165 193 GRANGER, C.W.J. and NEWBOLD, P. (1974) Spurious Regressions in Econometrics Journal of Econometrics, 2(2), pp. 111-120. GRENE, William H. (1993), Econometric Analysis, Second Edition, New York: Macmillan GRUBEL, Herbert G., (1968). Internationally Diversified Portfolios: Welfare Gains and Capital Flows, American Economic Review, 58, 1229 1314 GUJARATI, Damodar N., (1999) Temel Ekonometri, (ev. . enesen & G.G. enesen). stanbul: Literatr JOHANSEN, Soren. (1988): Statistical Analysis of Cointegration Vectors, Journal of Economic Dynamics and Control, Vol. 12, No. 2/3, pp. 231-254. JOHANSEN, S. and JUSELUS, K. (1990) Maximum Likelihood Estimation and Inference on Cointegration-With Applications to the Demand For Money, Oxford Bulletin of Economics and Statistics, Vol. 52, No. 2, pp.169-210. KAR M. ve KARA M.A., (2001). Finansal Entegrasyon ve Sermaye Akkanl Trkiye rnei, ktisat, letme ve Finans Dergisi, 180, 62 72 KORAJCZYK Robert, (1996). A Measure of Stock Market Integration for Developed and Emerging Markets, The World Bank Economic Review, 10 (2), 267 289 LEVY, H. and SARNAT M., (1970). International Diversification of Investment Portfolios, American Economic Review, 60, 668 675 MANDACI P.E. ve TAKIN D., (2005). ABye Uyum Srecinde MKBnin AB Piyasalar ile Karlatrlmas, MUFAD Muhasebe Finansman Dergisi, 26, 127 137
[2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
158 [18] [19] [20] [21] [22]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) MARASHDEH, Hazem, (2006). Stock Market Integration in the MENA Region: An Application of ARDL Bounds Testing Approach, Economic Working Paper Series, University of Wollongong ZGEN F. ve GLOLU B. (2004) Trkiyede Borlarn ktisadi Etkilerinin VAR Tekniiyle Analizi, ODT Gelime Dergisi 31, 93-114 PANTON D., LESSIG V. and JOY O., (1976). Comovements of International Equity Markets: A Taxonomic Approach, Journal of Financial and Quantitive Analysis, 11, 415 432 PHILLIPS, P.C.B. and PERRON, P.(1988) Testing for a unit root in time series regression. Biomtrika 75 (2) pp. 336-346. YCE Aye., (1997). Gelimi Dnya Borsalar Arasndaki Korelsyon, Gecikme ve ncelik likileri, Hazine Dergisi, 5, 1 8
Factors of Administrative Corruption in Nigeria and the Implication for Policy-Makers

Abdul Raufu Ambali Faculty of Administrative Science and Policy Studies University Technology MARA (UiTM), Shah Alam, Malaysia Abstract Given that the trend of corruption has spread over every aspect of life in Nigeria, no government public institutional and administrative agencies are free from the smear of corruption. Many Nigerian and non-Nigerian scholars have written voluminous articles and books on incidence of corruption throughout the world without focusing on factors of administrative corruption in public offices. A few of them who did so have attributed the cause mainly to materialistic factor. In this paper, the focus went beyond material factor. It is argued that corrupt attitudes of administrative practices by public officers should be traced back to other factors such as human nature itself, peoples tolerance, poverty and institutional problematic designation and weaknesses. The methodological approach of the study makes a difference from the usual theoretical based studies by going quantitatively to investigate causes of corruption in Nigerian administration. The paper develops a conceptual framework extracted from other views relating to administrative corruptions beyond material assumption as the sole cause of corruption in society and tested them in Nigerian context. The findings of the study have shown that corrupt attitudes of the public office holders in various Nigerian administrations were largely interrelated to personalistic nature of human greediness, low pay/poverty, institutional weakness, societal tolerance and an insatiable human appetite for material wealth. Finally, the implications behind all the identified factors were discussed to draw the attention of the policy-makers in the country to possible solution for solving corrupt practices. Keywords: Corruption, poverty, tolerance, greediness, human nature, institution, public office, weakness, material wealth and political will.
1. Introduction
It is sometimes unpleasant to write about frustrating and shameful issues such as corruption. At the same time in the academic circles, truth must be said and causes must be identified in order to arrive at concrete solutions to minimize a more damaging situation. A good number of scholarly works have been produced by local and international scholars- who diagnosed the problem and prescribed solutions to control the menace of corruption. Still, there is hardly any sign of improvement. Corruption with all its manifestations remains unabated in society and polity. Corruption is not merely a problem of law and administration; it also has to do with social and economic life of a community. The problem of corruption within or across nations is not a recent phenomenon, nor is it exclusively a Third-world concern. Corruption is a pervasive phenomenon that can be found in countries of widely varying ideology, economic condition and social development. Although, some societies may be more
160
vulnerable than others and may suffer more from its devastating effects, no country in the world today is immune from corruption's corrosive influence (Stapenhurst and Kpundeh, 1999: p.90). Corruption has devastating consequences on families, friends, morality, communities, public and private services, public institutions, democracy, the economy, and on all other facets of life. Corruption has created a lot of difficulties for innocent citizens. It has caused a psychological depression for the people either in their places of work or study. It is not uncommon to read newspapers or listen to news on the television and radio without coming across reports of corruption in any of its forms committed by some unscrupulous citizens of their respective nations. Corruption became the principal means of private accumulation today; in the absence of other means, it came to shape political activity and competition. In many countries of the world, all arms of government have been pervaded by corruption. It has resulted in a combination of scandalous wealth among the ruling class with growing poverty/low pay, misery and degradation among the people. Corruption has thus become a way of life in many countries, especially the Third world. Thus corruption is the most important problematic root leading to all other problems of inequality and underdevelopment. (Osoba, 1996: p.371). The aims of this paper are to examine and analyze the factors that help sustain corrupt administrative practices of any country in the world focusing on personalistic, institutional, materialistic, societal tolerance and poverty/low pay factors.
2. Literature Review
Corruption has been around in one form or another from the earliest days of social organization. What has only changed is that information about corrupt practices has become more available as governments have become increasingly unable to conceal evidence of wrongdoing rather than proper structure of the causative factors. The level of public tolerance for corruption has declined and the spread of democracy seems to afford less fertile ground in which corruption can flourish. Osoba argues that corruption is more visible because of new political and media conditions rather than because the government of some countries are more corrupt than the others. Since corruption is nothing new, it evoked condemnation of many in the past. It has also attracted the attention of social and political reformers- who advocated remedial measures. All these have led to the production of a huge amount of literature in the field. 2.1. Conceptualizing Corruption It is very easy to talk about corruption, but like many other complex phenomena, it is difficult to define corruption in a concise and concrete term. Seldom there is a consensus as to what exactly constitutes corruption. Consequently, there are always different opinions regarding the definition of the concept. Some scholars define corruption based on the Public Interest Approach while others apply the Public Opinion Approach. However, each of these approaches has its drawbacks. To avoid this confusion, this paper provides an operational definition of corruption as conceptualized in James T. Gires article. Before doing that however, it is pertinent to describe some of the commonly used definitions. Corruption is "the intentional misperformance or neglect of a recognized duty, or the unwarranted exercise of power, with the motive of gaining some advantage- more or less directly personal" (Brooks, 1974: p.18). Senturia sees it as the misuse of public power for private gains (Senturia, 1931: p118). Alatas (1990) characterizes corruption as the abuse of trust for the sake of private benefits. These definitions are fairly embrasive of the issues subsumed under corruption but do not sufficiently bring out the prioritization process of the individual, i.e. putting self above the collective interests. Therefore, in this study, corruption is defined as a betrayal of trust resulting directly or indirectly from the subordination of public goals over those of the individual. Thus a person who engages in nepotism has committed an act of corruption by putting his family interests over those of the larger society.
161
2.2. Forms of Coruption Corruption structurally entails different pictures in terms of what actually chracterizes it nature. In an elaborate analysis, Alatas (1990: p.38) has divided corruption into several distinct forms; they are: autogenic, defensive, extortive, investive, nepotistic, supportive, and transactive. Autogenic corruption is self-generating and typically involves only the perpetrator. A good example would be what happens in cases of insider trading. A person learns of some vital information that may influence stocks in a company and either quickly buys or gets rid of large amounts of stocks before the consequences arising from this information come to pass. Defensive corruption involves situations where a person needing a critical service is compelled to bribe in order to prevent unpleasant consequences being inflicted on his interests. For example, a person wanting to travel abroad within a certain time frame needs a passport in order to undertake the journey but is made to pay bribes or forfeit the trip. This persons corruption is in self-defense. Extortive corruption is the behavior of a person demanding personal compensation in exchange for services. Investive corruption entails the offer of goods or services without a direct link to any particular favor at the present, but in anticipation of future situations when the favor may be required. Nepotistic corruption refers to the preferential treatment of, or unjustified appointment of friends or relations to public office, in violation of the accepted guidelines. The supportive type usually does not involve money or immediate gains, but involves actions taken to protect or strengthen the existing corruption. For example, a corrupt regime or official may try to prevent the election or appointment of an honest person or government for fear that the individual or the regime might be probed by the successor(s). Finally, transactive corruption refers to situations where the two parties are mutual and willing participants in the corrupt practice to the advantage of both parties. For example, corrupt business person may willingly bribe a corrupt government official in order to win a tender for a certain contract. It must be declared that all forms of the above mentioned corruption charactrised most of the public administrative officers in Nigeria with no exception. It is a fact that cannot be denied that extortive, nepotistic and transactive corruption appear to be at the core of the corruption phenomenon in Nigeria context. All other forms appear to be the offshoot of these three fundamental types. For example, most complaints about corruption in Nigeria relate to the unfair advantages given to friends and relations over others in appointments, awards of contracts and so forth; the extortive practices of those in charge of vital services; and the collusion between corrupt private individuals, companies and public officials to deprive ordinary citizens of a fair chance at competing for goods and services. Moreover, other types such as supportive corruption arise as a protective shield to cover or defend the existing corruption. Also, there would be no defensive corruption in the absence of the extortive type. 2.3. Studies on Factors of Administrative Corruption Several authors have seen corruption, as a universal phenomenon of human life that can be found in every society -be it developed or underdeveloped, democratic or authoritarian. No matter what the volume of information about moral turpitude is in a particular society, the general assumption is that all societies irrespective of the nature of their ideologies are susceptible to corruption (Shaukat, 1985: p.67). Some of these authors apply an historical approach to their explanations about the universality of corruption. This is done by studying history and analyzing the incidence of corruption in different societies from the pre-enlightenment period up to modern times. Some of the authors in this group study corrupt practices in different societies and/or different periods of time, and draw case studies from both developed and underdeveloped countries. Shaukat (1985: p.82) argues that corruption is an unquestionable fact of history that no society is immunized from. David Gould (1991: p.4) is among those who compare corrupt practices in developed countries and underdeveloped countries. In his book, Administrative Corruption: Incidence, Causes and Remedial Strategies Gould analyzed the nature, incidence and effects of administrative corruption on economic, social and political development. He argues, to think that public morality in America and Europe is better and untainted is
162
to say the least, a false assumption. He has traced incidence of corruption to public morality. Gould concludes that corruption in developing countries is highlighted because of their under development. The inequality of incomes among various sections of the population is very wide. The independent earning capacity of a politician is very limited and political office is the only source by which they can become wealthy. Modernization has unhinged moral and social values, which has plunged the new states into a gruesome moral crisis. The weakness of the enforcement machinery and absence of national consciousness also are contributory factors in this matter. Huntington (1968: p.58) and Rick, Kpundeh (1999: p. 90) have attributed incidence of corruption in the Third world countries to traditional structure of those societies. Kpundeh (1995: p.93), combines extensive analytical knowledge of the nation with survey evidence gathered from interviews with Sierra Leonians to provide a rich and subtle account of how they view government, politics, corruption, and social life. Samuel Huntington argues that corruption in underdeveloped countries cannot be disconnected from their weak political institutions and structures. However, he concluded that corruption exists everywhere whether the country is developed or underdeveloped, and that it may be more prevalent in some cultures than others and in most cases it seems to be prevalent during the most intense phase of modernization. Corruption by those holding public office has to do with the refusal to separate the holding of public office from the private accumulation of wealth. A perusal of the literature has revealed a surprising dearth of psychological variables in the discussion of the mechanisms involved in corruption. James Gire (1986: p.6), has studied the phenomenon of corruption from a psychological perspective. He applies Maslows concept of hierarchy of needs and Banduras perspective of social learning theory to analyze the physical and psychological conditions in Nigeria that tend to have predictable effects on behavior-including corruption among the public office holders. In is his argument, overlooking the psychological viewpoint is a major reason why current attempts at tackling corruption have failed. Gire opines that Banduras observation learning theory and Maslows theory of hierarchy of needs explain why corruption took hold in Nigeria and its current proliferation. Furthermore, the increasingly worsening economic situation has led to a corresponding decrease in the standard of living of most Nigerians. This situation, coupled with the increased uncertainty about jobs and anxiety in many people about their ability to feed themselves and their families, may have contributed to the present escalation of corruption in the country. Brownsberger (1983: p.215) argues that the roots of corruption go to materialism and a political fragmentation that are products of development. Therefore, inappropriate development of a vital differentiated private economy with a broader, better and more patriotic elite leads to corruption. It remains skeptical about who will come to power to enforce the tough anti-corruption laws already available in the books. However, all the above sources of literature have their own limitations. They focus on the material aspect alone. In other words, material is the only yardstick used by many authors as factor for occurrence of administrative corruption in public offices. They also failed to see the contribution of what might be the personalistic, institutional, societal tolerance factors, materialism and poverty/low pay, which could be investigated through a pure survey research.
163
3. Theoretical Conceptual Framework/Model

Figure 1:
Personalistic factor Institutional factor Societal-tolerance factor Materialistic Factor Poverty/low pay Attitude towards Administrative corruption
3.1. Operationalizing the Research Model The incidence of corruption in public offices can be traced back using five basic framework factors, which are personalistic, institutional, societal-tolerance, materialistic, and poverty/low pay as plausible explanations relating to corrupt attitudes of the public office holders in society. In personalistic factor, corruption is seen as the work of the people by their nature. In other words, when individuals or small groups are found to have broken the rules governing public roles, it is tempting to search for causes in the personal qualities of those involved (Kpundeh, 1995). Politicians and bureaucrats are human as the rest of us; therefore, they are prone to act irrationally and contrary to the laid down rules and regulations. At a personal or individual level of analysis, it holds that corruption is simply a part of human nature, that is to say that human beings are all subject to greed and they engage in corruption in order to rationalize their gains in public offices. Therefore, the theorists of personalistic factor of corruption believe that corruption and any other vices are inherently embedded in human nature. If an individual plunders public funds, he does so not willingly but because he has been tempted to be corrupt. However, if this theory is considered to be a fundamental explanation for corruption, it would appear that policy makers must accept corruption as inevitable and, at best, make an attempt to limit manifestations of being greedy by reducing opportunities for corruption among public office holders and increasing the costs towards corruption of all forms. The institutional explanatory factor of corruption is a useful framework that helps describe corruption as a fundamental and deep-seated problem rather than merely as the working of a few bad individuals. Most institutional corruption can be attributed to structural problems- ordinary matters of administration such as inefficient auditing procedures or uncertain communication among the sections of large organizations. Institutional fraud occurs when a pattern of private gain in goods and funds operates in a particular public institution (Kpundeh, 1995). Therefore, in this study, I am of the view that even though, public administrative corruption in a country may be perpetuated by individuals, yet it takes place primarily within an institutional problematic context. This institutional explanation will assist us in pinpointing the weaknesses and suggest institutional remedies to policy-makers. It is fact that major scandal and commission of inquiry reports of many countries in the world have shown that there is a common system of channeling government funds for private use due to institutional factor (Robin. 1990: p.236). The societal tolerance is the most comprehensive factor because it regards corruption as a form of influence due to peoples tolerance, rather than the failure of institutions. Theorists of societal tolerance factor hold that corruption exists in a situation where wrong-doing has become the norm and an acceptable standard behavior necessary to accomplish organizational goals according to the notions of public responsibility. In this case, trust has become an exception, not a rule.
164
Thus, corruption has become so regularized that the people supports and backs wrongdoing and more or less actually penalize those who live up to the old norm. Hence, people in position of trust engage in economic sabotage, or orchestrate ways to channel government funds to personal accounts within and outside the countries and those in power kept quite or have no appropriate measure to take against them. It is argued that when wage scale fails to provide the average civil servant with the means to feed, house and clothe his/her family, it would leads the civil servants to turn to other means typically involving corruption for survival's sake. This often become a crucial factor where there is a heavy gap or gross inequalities between the top level officers and their subordinates pays. Finally, materialistic factors referred to insatiable appetite for material wealth. It is assumed that material wealth like power generates an effect of corrupt practice in administration, especially when it becomes an obsession; people generally tend to indulge in irregularities. Hence, the chances of corruption increase in manifold (Shaukat, 1985: p.139).
4. Research Methodology
A sample of three hundred and fifty subjects was chosen from public office holders in two different locations of Kwara State namely: Afon and Ilorin Local Governments of Nigeria. The designation of the study and sample technique was based on stratified sampling method to get the targeted respondents. The survey was carried out through a self administered questionnaire consists of 58 items that were expected to capture the fundamental dimensions of the five determinant factors and the administrative corruption issues. Statistically, at the inception of the study, factor analysis was used to reduce the data to smaller set of 48 underlying items that summarize the essential information that appropriately taped the intended constructs of the studys variables. Various conditions and assumptions such as normality of data distribution, linearity, multi-colinearity and singularity as well as factorability of the correlation matrix was satisfied before proceeded with result analyses. Principal component and rotated varimax method was used to identify the dimensions in scale of measurement for depended variable and independent predicted factors. Bartletts test of sphericity and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy were used to determine the factorability of the variable matrix as whole. Hence, factorability was assumed where Bartletts test of sphericity is large and significant, and the KMO greater than 0.6. To further test the internal consistency of scale used, a reliability analysis was carried out. Internal consistency of scale measure for this study was assumed where Cronbachs Alpha values were more than 0.7 (see details in the finding section). Finally, to realize the major objectives and the hypothesized links between the predicted factors and attitude towards administrative corruption, a regression analysis and correlation matrix was used as shown in the finding sections.
5. Findings
5.1. Profile of Respondents Out of three hundred and fifty initial self-administered questionnaires distributed, only three hundred and twenty were valid and accepted for analyzing the result of this study. The rests 30 were considered in valid due to too many deliberate omissions and incomplete items of the questionnaires. The respondents were drawn from different age groups below 30 years old, 30-40 and above 40 among the key public office holders. They were also from different ethnic compositions and different educational backgrounds as well as tenure of employment as indicated in the table 1 of the study.
165
Table 1:

Profile of Respondents
Frequency 107 120 93 Frequency 180 140 Frequency 145 40 55 80 Frequency 80 140 60 40 Frequency 165 85 70 Frequency 145 100 75 Percent 33.4 37.5 29.1 Percent 53.2 43.8 Percent 45.3 12.5 17.2 25.0 Percent 25.0 43.8 18.8 12.4 Percent
Age Group (n = 320) Below 30 years old 30-40 years old Above 40 years old Gender (n= 320) Male Female Ethnic (n= 320) Yoruba Fulani Baruba Tapa Education (n = 320) Diploma Bachelor Degree Masters PhD Public Offices (n = 320) Public servicing firms Public Manufacturing firms Higher Institutions Tenure of Office (n = 320) Les than 5 years 5-10 years More than 10 years
Percent 45.3 31.3 23.4
5.2. Pilot Test for Adequacy and Dimensions of Scale of Measurement Principal component and rotated varimax method was used to identify the dimensions in scale of measurement with 28 items for the five potential factors that have impact on corrupt habits of the administrative public office holders. Dimension with at least 4 items and Eigen value of more than 1 as shown in table 2 has been analyzed for further analyses in the study.
166
Table 2:

Result of Rotation Matrix and Dimensions (Loading value >0.3 in all)
1 0.882 0.859 0.839 0.826 0.809 0.802 0.752 0.728 Components 2 3 4 5
Items Personalistic 5 Personalistic 2 Personalistic 7 Personalistic 6 Personalistic 3 Personalistic 8 Personalistic 4 Personalistic 1 Institutional 5 Institutional 4 Institutional 1 Institutional 6 Institutional 3 Institutional 2 Societal tolerance 5 Societal tolerance 3 Societal tolerance 4 Societal tolerance 2 Societal tolerance 1 Materialism 3 Materialism 1 Materialism 4 Materialism 5 Materialism 2 Poverty/low pay 2 Poverty/low pay 1 Poverty/low pay 3 Poverty/low pay 4 Eigen value Cumulative variance %
0.814 0.805 0.777 0.768 0.760 0.698 0.831 0.792 0.770 0.752 0.741 0.799 0.742 0.739 0.679 0.641 0.825 0.790 0.749 0.448 1.31 73.10
12.19 42.63
3.67 54.78
2.80 63.82
1.81 69.18
As shown in table 3, for all the predicted factors of administrative corruption of the public office holders, the result of the analysis indicates that the underlying structure of the scale for the set of variables used to measure attitudes of public office holders towards administrative corruption are reflected in the five theoretically proposed subscales of the factors. As such, the Bartletts test of sphericity is significant and the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy is far than 0.6 with a value of 0.812 and cumulative variance value of 73.percent.(see table 2 & 3 respectively).
Table 3: KMO and Bartletts Test
0.918 2868.110 436 0.000
Kaiser-Meyer-Olkin Measure of Sampling Adequacy Approx. Chi-Square Bartletts Test of Sphericity df. Sig.
On the other hand, the scale measures of items for attitudes of public office holders towards administrative corruption with Eigen value more than 1 was also used for further analysis. The KMO of attitude is 0.918 as shown in table 4. All of the 20 items used produced a cumulative variance value of 63.75 percent (refer to table 5).
167
Table 4:

KMO and Bartletts Test
0.918 2868.110 436 0.000
Kaiser-Meyer-Olkin Measure of Sampling Adequacy Approx. Chi-Square Bartletts Test of Sphericity df. Sig.
Table 5:
Result of Factor Analysis for Attitudes towards Administrative Corruption (Loading value >0.3 in all)
Components 0.822 0.786 0.852 0.845 0.857 0.805 0.822 0.825 0.844 0.833 0.861 0.852 0.827 0.815 0.846 0.792 0.849 0.804 0.809 0.831 19.13 63.75
Items Attitude towards Administrative Corruption 1 Attitude towards Administrative Corruption 2 Attitude towards Administrative Corruption 3 Attitude towards Administrative Corruption 4 Attitude towards Administrative Corruption 5 Attitude towards Administrative Corruption 6 Attitude towards Administrative Corruption 7 Attitude towards Administrative Corruption 8 Attitude towards Administrative Corruption 9 Attitude towards Administrative Corruption 10 Attitude towards Administrative Corruption 11 Attitude towards Administrative Corruption12 Attitude towards Administrative Corruption 13 Attitude towards Administrative Corruption 14 Attitude towards Administrative Corruption 15 Attitude towards Administrative Corruption 16 Attitude towards Administrative Corruption 17 Attitude towards Administrative Corruption 18 Attitude towards Administrative Corruption 19 Attitude towards Administrative Corruption 20 Eigen value Cumulative variance %
5.3. Reliability Test The internal consistency of the scale of measurement for both factors and attitudes towards administrative corruption was examined using Cronbachs alpha for all items. The Cronbachs alpha value for individual items is ranged from 0.862 to 894, while the overall Cronbachs alpha is equal to 0.886. The results indicate a reliable internal consistency of the scale used as measurement (see table 6).
Table 6: Reliability Statistics
Cronbachs Alpha based on Standardized Items 0.891 No of Items 28
Cronbachs Alpha 0.886
5.4. Correlation Coefficient Analysis Factors and Administrative Corruption The results of correlation coefficient analysis carried out indicate a high significant relationship between the predicted factors and attitude towards administrative corruption of the public office holders as shown in table 7.
168
Table 7:

Correlation Coefficient Analysis of Factor and Administrative Corruption
1 1.000 0.485** 0.341* 0.682** 0.528** 0.295* 2 1.000 0.542** 0.367* 0.466** 0.386* 3 1.000 0.548** 0.311* 0.320* 4 5 6
Administrative Corruption Personalistic factor Institutional factor Societal tolerance factor Materialistic factor Poverty/low pay factor
Note: *p<0.05 **p<0.01
1.000 0.385* 0.310*
1.000 0.567**
1.000
5.5. The Strength of Relationship between Factors and Administrative Corruption. The general model summary of the regression analysis for all independent factors together explain 88 percent of the variance (R squared) in attitude of the public office holders toward the act of administrative corruption in Nigeria (see table 8). However, the results of coefficient analysis presented in table 9 further show the strength of relationships between individual factor and the attitudes of public office holders towards administrative corruption in Nigeria with respect to individual factor. As indicated in the table, personalistic, institutional, societal tolerance and materialistic factors were the major contributors of administrative corruption with Beta values of 0.354, 0.218, 0.169, and 0.179 respectively. As the results indicate, the highest contributive factor of all is personalistic factor, while the least contributive factor to administrative corruption of the public office holders is poverty/low pay with beta value of 0.146.
Table 8: Model Summaryb
R 0.945a R Square 0.880 Adjusted R Square 0.835 Std. Error of the Estimate 6.06905
Model 1
a b
Predictors: (Constant), personalistic, institutional, societal tolerance, materialistic, poverty/low pay Dependent variable: attitude towards administrative corruption
Table 9:
Coefficients Result
Standard Coefficients (Beta) 0.354 0.218 0.169 0.179 0.146 t 4.394 4.545 2.974 2.203 2.325 2.107 Sig. 0.000 0.000 0.003 0.029 0.021 0.032
Variables (Constant) Personalistic factor Institutional factor societal tolerance Materialistic factor Poverty/low pay factor
Dependent variable: Attitude towards Administrative Corruption
Table 10: ANOVAb

Model 1 Regression Sum of Square 2602.391 Mean Square 1352.194 F 35.091 Sig 0.000a
Note: a predictors (Constant), personalistic factor, institutional factor, societal tolerance, materialistic factor, poverty/low pay factor b dependent variable: attitude towards administrative corruption
Looking at the ANOVA in table 10, the result supports the argument and/or hypothesis that there are many other factors other than material aspects that dominate the literature as reason for administrative corruption by public office holders in Nigeria with an F-ratio of 35.081, p<0.001. Therefore, we can conclude that personalistic, institutional, societal tolerance, materialistic and poverty/low pay/low pay factors combined together to influence the attitude of public officers toward the habit of corruption in Nigerian public administrations.
169
6. Discussions on Findings
There was a belief in some literatures that corruption is cultural-laden, and therefore, the cause of corruption in a certain society could not be unconnected with the culture of that society. This assertion with the 'cultural' argument seems to make us believe that, in certain cultures, corruption is quite normal and morally acceptable. However, there is no single place on earth where getting rich through immoral means is considered lawful or morally acceptable! Universality of corruption and numerous philosophical aspects associated with it have not deterred scholars from searching out the possible causes for this malady which has been part of human life ever since the dawn of history. The research has been multidimensional and the investigators have tried to explore every aspect of man's social, political and economic activities, and knowing how difficult it is to find pertinent clues of corruption. However, the findings of this research pinpoint few factors that generate corruption in public administrative offices of all kinds. These factors include: personalistic, institutional, peoples tolerance, materialistic and poverty/low pay factors. 6.1. Personalistic factor As the finding reflects, the issues of corruption in Nigeria context can be traced back to personal greedy of human nature. To substantiate this finding, Rogow and Lasswell (1963) rejected the notion that corruption is an inherent aspect of power, but produced by a combination of factors relating to environments and personality of holder of power. Their research in child psychology concludes that sources of corruption of public office holders must be sought in their childhood experiences. It would depend on the kind of deprivations to which they were subjected as children. If a public officer had undergone an economic deprivation, most likely as he attains position of power in the office, he will use it to enrich himself and his family by virtue of his greedy nature (Rogow and Lasswell: p.58). In the Nigerian context, it is very true with most political leaders, administrators and various public office holders either in the past or present. All those who came to power purposely sought to use their office positions to enrich themselves, friends and families. It is uncommon to see any Nigerian administrative leader at the helm of power in any public organizations without using the power position at the expense of the subjects under his control. For instance, it was found that the former Nigerian Senate President, Chuba Okadigbo, who had 14 fleet of car of escort, and he unofficially increased the fleet of car into 25 with the public fund (Tell Magazine, 1999: p.28). 6.2. Institutional factor According to the findings of the study, one potential factor of corruption in Nigeria is due to institutional issues. These issues concern with too much of power, lack of transparency, accountability or absence of a civil service work ethics in Nigerian public institutions. Jabbra (1985: p.22) says, civil servants lack a sense of purpose and commitment to their responsibilities. They believe they serve no one but themselves and should exploit their positions for personal gains. Public office holders in various institutions in Nigeria believe they serve no one but themselves and should exploit their positions for personal gains. This feeling and belief, according to Jabbra, has furthered the institutionalization of bureaucratic behavior characterized by goals of self-aggrandizement, usurpation of power and the conception of public office as an avenue for wealth. The issue of an excessive and uncheckable power is one of the major causes of corruption from times immemorial in Nigerian public offices and institutions. For example, administrators in charge of many projects are given unlimited powers in granting loans, buying machinery, and selling land for industrial purposes. They also have authority in giving tax-holiday and tariff concessions to the industrialists. Due to this enormous power, it gives them the leverage to manipulate the privileges for personal gains (Stanislav, 1968; p.92).
170
6.3. Materialistic factor With regard to material factor as found in this study, it is widely acknowledged that man's insatiable appetite for wealth is one of the major causes of corruption. According to Ali, wealth like power generates an effect and when it becomes an obsession, people generally tend to indulge in irregularities and the chances of corruption increase manifold. When the distribution of wealth is unequal and unjust, it produces social and economic inequality, which leads to jealousy, envy and enmity, all of which pollute human character and conscience. Therefore, in many of Nigerian public offices and/or administrations, it is clear that corruption has become a phenomenon that is largely caused by many economy bribes and other material objects through, which could be kept secretly done. Thus, as public officers open their eyes to money particularly in Nigeria, possibilities of corruption through money bribes magnified (Shaukat: p.139). 6.4. Societal Tolerance Factor Though no one will dispute the fact that the lust for power and level of greediness among rulers are the fundamental sources of corruption, but any rational appraisal of this social ailment must also take into consideration the role of the peoples apathy, and their ignorance of national problems. According to Mr. Justice P. D. (Chairman of the Commission of Enquiry into Bribery and Corruption appointed by the Ghanaian Prime Minister in 1970):to study the area, prevalence, and methods of bribery and corruption in Ghanaian society, but also to determine whether there were factors in society which contributed to this. As people, do we frown upon and resist bribery and corruption or do we tend to regard them as natural and inevitable? (See: Herbert, 1972: p.247) In addition, Denis Joseph also in his examination of the causes of administrative corruption in India has concluded that masses in general are as much to be blamed for the continued existence of corruption, as are those who indulge in this nefarious transaction. In his opinion the extent of corruption in a particular country would depend on how much the people in that society are prepared to tolerate it. In Nigeria, rampant corruption in society is attributed to the permissive attitudes of the people towards corruption. When a case is brought to their notice they put it away by saying 'everybody does it' or 'it is as an accepted part of life'. In another argument on the societal tolerance as a factor for corruption is regarding the relationship between the leaders and the subjects. Some believe that if the leaders are corrupt the people are bound to be corrupt, because leaders after all, are the product of the same society. There are other views, however, whereby it is believed that masses more often than not tend to emulate their leaders and that it is from them that corruption spills over into the rest of the society. 6.5. Poverty/low pay According to Gould, corruption in many countries is mainly due to poverty/low pay and low, inequitable pay scales. Pay scales throughout Africa have remained poorly structured since colonial times and failed to keep pace with inflation. The buying power of a civil service salary, which might have been adequate 20 years ago, is in most countries surely inadequate. The prevailing wage scale fails to provide the average civil servant with the means to feed, house, and clothe his/her family. This, in turn, leads the civil servants to turn to other means typically involving corruption for survival's sake. There is a heavy gap between the top officers' pay and their subordinates pay; this creates gross inequalities. Following this in Nigeria the subordinates did not care much with their works instead, they remain preoccupied with surviving matters through corruption. In addition, a bi-polar type of income distribution has been created between the haves and have-nots. The consequence of this is that civil servants, who strive for high social status so that they can be counted among the haves resort to the instrument of demonstration effect or a conspicuous consumption of luxury goods which they cant afford given their low salaries.
171
7. Implications for Policy-Makers about Solution

Given that corruption is a universal phenomenon and that there will always be individuals and groups in a society who attempt to enrich themselves unjustly, the obvious question is: how should corruption be attacked? Solutions are difficult to design partly because corruption is difficult to detect or expose. It is widely believed by all scholars that no single line of attack is effective to curb corruption rather holistic approaches are required. Corruption is internationally recognized as a major problem that is endangering social stability, security, damaging economic and political development of a nation. With the present trends of globalization of people, markets and services, reducing corruption has become a priority nationally and internationally. In other words, various mechanisms and measures have been identified through out the world to fight corruption. An implication is that since corruption is not confined to a particular part of the world, the instruments to combat corruption vary in their scope, content and methods of their applications. Base on the findings relating to the underlying factors of corruption in this study, the step to battle administrative corruption may include an adherence to international conventions for tackling corruption throughout the world. The conventions are considered to be a joint effort by a number of countries to combat corrupt practices at international level. In 1997, OECD adopted convention on combating bribery of foreign public officials in international business transactions. The parties have been obliged to make bribery of foreign officials a crime. It also ensures that penalties are effective and proportionate and comparable to domestic ones. The convention urged all the OECD countries to adopt territorial jurisdiction against corruption. They should have mutual criminal investigations and other corrupt practices. The legislation adopted by countries to implement the convention is subject to a rigorous monitoring process to ensure that it meets the standards set by the convention. Other international instruments include the Criminal Law Convention of 1988 and the Civil Law Convention on Corruption of 1999 adopted by the council of Europe's legal instruments. The European Union (EU) has adopted some instruments against corruption. These include the treaty on the protection of financial interests of the community by focusing on criminalization of transnational bribery in Europe; the Supranational Law against corruption, which deals with bribes and rules on accounting and auditing. Furthermore, the 1988 United Nations Convention against Illicit Traffic in Narcotic Drugs has addressed the issue of money laundering through illicit proceeds. The convention contains provisions for signatory countries to criminalize money laundering from drug trafficking as considered to be the most serious form of corruption. Other measures against money laundering include the 1990 Council of Europe Convention on Laundering; the 1991 Council of European Communities Directive on Prevention of the Use of the Financial System for the Purpose of Money Laundering. In 1992, the International Organization of Securities Commissions (IOSCO) has encouraged members to take measures to curtail money laundering in securities, and further markets. In addition, the Caribbean countries have established the Financial Action Task Force (FATF) while the countries in Asia Pacific region established the Asia/Pacific Group to curb money-laundering corruption. With this mechanism, the participating countries were provided with information assistance in drafting workable laws and training investigators and prosecutors (Asian Development Bank, 1999: pp.27-33). Nigerian government and policy-makers should find a useful insight of tackling corruption or preventive action from the various international conventions against corruption discussed. The agency of anti-corruption in Nigeria is yet to be much better placed to ensure that effective preventive steps against corruption are identified and taken. However, to operate successfully an anticorruption agency must possess committed political backing at the highest levels of government. Many scholars have emphasized on strong political will as a crucial factor for eliminating corruption, from both administration and society. Political leaders must be sincerely committed to the lay down approaches for curbing corruption. According to Jon, if the "big fish" is protected from prosecution for corruption and the "small fish" is caught, the anti-corruption strategies would be doomed to failure (Jon, 1982: p.408). In order words, political will is a prerequisite for anti-corruption strategy. Jon cited the example of the Philippines, which relied on seven laws and 13 anti-graft agencies since its battle
172
against corruption began in the 1950s (Jon, 1999c). However, the Philippines had failed to eliminate corruption through anti-corruption mechanism due to lack of will of its leaders. One of the principal functions of a parliament is to watch the executive branch of government and government financing. By helping ensure accountability and transparency in public sector finances, the parliament can contribute to efforts for curbing corruption. However, mere existence of a parliament is not enough. Parliamentary activities should be as open to the public as possible. Select committees, particularly the Public Accounts Committee need to be empowered to hold the chief executive accountable. Parliamentarians are to be empowered to call civil servants to account and participate in the detailed scrutiny of budgetary estimates. Parliament can "question any action of an official in the House through question time/adjournment motion or a debate. The officials are bound to provide full information from their files and documents on the subject matter (Jon, 1999c: p.59). However, the legislature itself should be committed to building a public service system that is accountable if corruption is to be eliminated. The implication here is that certain factors or forces like inadequacy, inconsistency, and obsolescence of certain laws, rules, procedures, mechanisms and their application impede the practice of accountability at various levels of government. Hence, corruption can only be tackled when legislative body is committed to this goal and makes serious effort in this regard. The audit unit also should adequately check the auditing performance and 'value for money'. Audit unit is expected to monitor the efficiency and cost-effectiveness of administrative activities and to prevent corruption in financial procedures. Through a proper audit report, public accountability is partially ensured. This can be facilitated by regular submission of departmental manual accounts and reports of money, materials, manpower, equipment, facilities stores, assets etc. to the Auditor General (AG) who will report directly to the government. It has been emphasized that scope of duties undertaken by AG should not be limited to administrative departments alone; it should also include the accounts of the state and federal governments, local and statutory bodies. According to Salleh and Kar (1995: p.32), the duties should extend to the accounts of "companies in which majority shares are held by the government or the public authorities in general." The resources to be monitored include money, materials, manpower, equipment, facilities stores and assets. The main implication here to policy makers is that the audit mechanism could be effective only when law guarantees the independence of auditors from the executive control. In addition, AG must be allowed by the designated policy to carry his functions without the interference of legislature and that auditor and all his staff should have unimpeded access to all records, books, vouchers, documents cash, stamps, securities, stores and other properties, which are subject to proper discharge of duties. Hence, the existence of a sound and effective auditing system will force the administrators, officials and bureaucrats out of engagement in corrupt practices. It ensures that money is spent according to the purposes and programs approved in the government budget. Above all, a proper audit helps to maintain a high standard of public morality in all administrative and public trusts, especially, in all financial matters (see: UN, 1977: p.35; Anderson, A., 1995: p.252). Another implication for policy-makers in this study lies in the fact that the present level of corruption in Nigeria can largely be attributed to poor public sector pay. Mauro, Panol (1997: p.10), says "when civil service pay is too low, civil servants may be obliged to use their positions to collect bribes as a way of making ends meet, particularly when the expected cost of being caught is low." Many government budgets have been inflationary in nature. As such, it has eroded civil service salaries up to a critical point where the civil servants could not live on them. They simply resort to various kinds of petty corruption. It has been pointed out that the major causes of corruption in most of the contemporary Third world were the extremely low salaries of civil servants and politicians. Therefore, to eradicate corruption in government agencies, public sector wages must be reformed. According to Lee (1985: p.10), Pay political leaders the top salaries that they deserve and get honest, clean government or underpay them and risk the third world disease of corruption." It is clear from this statement that the corrupt practices of political leaders and the civil servants and/or public administrative office holders, particularly in Nigeria are the result of salaries, which do not
173
commensurate with the positions and responsibilities they discharge. However, a big implication here for policy-makers is that raising salaries alone will not solve the problem of corruption; if the government does not have the political will to minimize it from the body politic and administration. Finally, anti-corruption policy actions must be designed to win community support. Without community support and involvement in the efforts to combat corruption, the wars against corrupt practices are like zero-sum battle. Corruption cannot be detected unless members of community are willing to come forward and report it. Anderson (1995: p.252), believed that "civil society where it is free to organize and act can become a vital partner in developing and strengthening ethical practices in the public sector." This implies that a growing numbers of people awareness and action can fight out corruption in society. This mechanism, if adopted, gives average citizens vital tools for uncovering and stamping out corruption from all levels of society. Hence, efforts to raise public awareness of corruption should be directed primarily at grass-roots communities rather than just policymakers, elite or middle class levels. It is imperative for policy-makers to know that mass educational campaign on corruption is highly needed to overcome societal tolerance of corruption in public offices (ADB, 1999: p.272).
8. Conclusion
Administrative corruption in Nigeria cannot be disconnected from varying factors that include: highly centralized institutional stranglehold, weaknesses, widespread poverty/low pay coupled with domination of personalistic and self- aggrandizement attitudes of the public officers as shown in various findings of the study. On top of everything is the absence of a sense of mission of professionalism and monumental leading to monumental distrust among public servants. Public office holders are quick to use this to their advantages to perpetuate gross abnormities in the operation of governments public institutions and offices. Societal tolerance was also found to be a function of administrative corruption and helped nurture corrupt practices among office holders in Nigeria. It becomes imperative for policy-makers to draw their attentions to causes of administrative corruption and their implications before moving towards proposed solutions and/or preventive measures.
174
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Alatas, H.S. (1990), Corruption:Iits Nature, Causes and Functions. Kuala Lumpur: S. Abdul Majeed. Anderson, A. (ed.1995), 'Anti-corruption Initiatives of the United States Agency for International Developments in Combating Corruption'. Asian &Pacific Economies. Asian Development Bank.(1999), Combating Corruption in Asia and Pacific Economies. Manila: Philippine. Brooks, R.C. (1974), Corruption in American Politics and Life. New York: Arno Press. Brownsberger, William (1983), 'Development and Governmental Corruption Materialism and Political Fragmentation in Nigeria'. The Journal of Modern African Studies, vol. 21, No. 2 David, Gould. 1991. Administrative Corruption: Incidence, Causes and Remedial Strategies. New York: Marcel Dekker, Inc. Herbert, H. Werlin (1972), 'The Roots of Corruption-the Ghanaian Enquiry', The Journal of Modern African Studies, Vol.10, No 2, July. Jabbra, J.G. (1985), 'Bureaucratic Corruption in the Third World: Causes and Remedies' Indian Journal of Public Administration, Vol. 1 No. 2, June James, T. Gire. 1986. A Psychological Analysis of Corruption in Nigeria. Oxford: Heinemann. Jon S.T Quah (1982), 'Bureaucratic Corruption in the ASEAN Countries: A Comparative Analysis of Their Anti-Corruption Strategies'. Journal of Southeast Asian Studies, 13 (1) March. Jon S.T Quah. (1999c), 'Comparing Anti-corruption Measures in Asian countries: Lessons to be Learnt'. Paper presented at UNDP workshop on promoting to Integrity in Governance at World Conference on Governance in Manila 31 May- 4 June Kpundeh J., Sahr and Rick, Stapenhurst (1999), Curbing Corruption: Toward a Model for Building National Integrity. The World Bank, Washington, D C. Lee, Kaun Yew (1985), New Straight Times, 23 March Mauro, Panol. 1997. Why worry about corruption? Washington Dc: International Monetary Fund. Osoba, S.O. (1996), 'Corruption in Nigeria', Review of African Political Economy, Vol. 23, No. 69, Sept. Robin, Theobald (1990), Corruption, Development and Underdevelopment, Hong Kong: Macmillan Press Ltd. Rogow, A. A. & D.D. Lasswell. (1963), Power, Corruption and Rectitud,. Eaglewood Cliffs, N.J., Prentice Hall. Salleh, Sirajuddin, H. & Arabinda, Kar. (1995), Administrative and Financial Accountability: The ASEAN-SAARC Experience. Kuala Lumpur: Asian and Pacific development Center (APDC). Samuel, Huntington, P. 1968. Political Order in Changing Societies, New Haven Conn: Yale University Press. Senturia, J.J. 1931. 'Political Corruption'. Encyclopedia of the Social Sciences. Vols. 3-4. New York: Macmillan. Shaukat, Ali. 1985. Corruption: A Third World Perspective. Lahore: Aziz Publishers. Stanislav, Andreki (1968), The African Predicament. New York: Atherton Press. Tell Magazine (1999), 15th August. United Nations Hand Book. 1977. Government Auditing in Developing Countries. New York.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
Job Satisfaction of Public Sector Managers in North Cyprus

Tlen Saner Faculty of Economics and Administrative Sciences, Near East University, Cyprus Tel: 0392 2236464 ext.271/360; Fax: 0392 2236461 E-mail: saner@neu.edu.tr Berna Kocaman Faculty of Law, Ankara University, Turkey E-mail: kocaman@dialup.ankara.edu.tr Abstract This study examines the relationship between selected independent variables and job satisfaction of public sector managers in North Cyprus. Data was collected with a self administered questionnaire. For this study one of the most widely used questionnaires, Minnesota Satisfaction Questionnaire (MSQ). MSQ was administered on the full population to 111 directors and undersecretaries. Results indicated that overall directors and undersecretaries were satisfied with their jobs. Also, the study indicated interesting result in those male directors and undersecretaries were more satisfied with their jobs than female directors and undersecretaries. Keywords: Public sector, directors and undersecretaries, determinants of job satisfaction.
1. Introduction
Job satisfaction and psychology of motivation are the most important elements of organizations. Each organization acknowledges the importance of the two key components, job satisfaction and motivation. People are the central source of organizations; organizations can not exist without people. The main element of organizations, the human resources, should be treated differently from the rest of the factors of production. Satisfied and motivated employees will help the organizations to reach their goals and objectives in the best possible way. In organizations the managers main and the most important task is to motivate his or her team so that motivated staff will drive satisfaction from delivering goods and services which creates a better organization and customer satisfaction. Bruce and Blackburn (1992) write, Most of todays workers expect to derive much more satisfaction from their work then their grandparents ever dreamed was possible. Turn over rate of an organization and employee satisfaction is closely related. Unhappy employee is not likely to stay in their present job, sooner or later will leave for another.
2. Job Satisfaction
In reviewing the literature it becomes apparent that job satisfaction can be defined in a number of ways. Hoppock (1935) defined job satisfaction as any combination of psychological, physiological, and environmental circumstances that causes a person truthfully to say, I am satisfied with my job. Employees may be satisfied with some aspects of their jobs, while being dissatisfied with others. It is
176
assumed that employees are able to balance the specific satisfactions against the specific dissatisfactions and arrive at a composite satisfaction with the job as a whole (Hoppock, 1935). According to Poling (1990), the best predictor of job satisfaction is when the employees personal values match those of the organization (Scott, Swortzel, & Taylor, 2005). Ivancevich and Donnelly (1968) define job satisfaction as the favorable viewpoint of the worker toward the work role he presently occupies. According to Spector (1997) Job satisfaction is simply how people feel about their jobs and different aspects of their jobs. It is the extent to which people like (satisfaction) or dislike (dissatisfaction) their jobs.
3. A Reviw of the Litrature

In an extensive job satisfaction literature review it is apparent that the number of research on public sector job satisfaction is low. Many of the job satisfaction studies suggested that job satisfaction is determined by several factors. Maslow (1954), Herzberg (1959) and Alderfer (1952) are the studies on job satisfaction that form the basis for much of the modern studies. Focus was not only on the working conditions, pay and human relations aspect of the job, factors like recognition, responsibility and self esteem were emphasized. Maslows (1954) suggests that people are driven by unsatisfied needs. Five-stage needs hierarchy is one of the oldest to outline job satisfaction of employees. Maslow (1954) outlined the theory in five levels. He grouped hierarchy of needs as higher order and lower order needs. Lowest order needs includes basic physiological needs, safety and security needs. Higher order needs includes social needs, esteem needs, and self-actualization. Theory is a step by process. As lower order needs are met, higher order needs can be fulfilled. The implication of this theory to job satisfaction supposes that when an individuals lower order needs for things such as pay and security have been met, then higher order needs begin to be desired. Herzberg (1959) used Maslows (1954) needs hierarchy to formulate motivator-hygiene theory of employee motivation. Two-factor theory (motivator-hygiene theory) suggests that intrinsic factors produce job satisfaction where as presence of hygiene factors brings the employee to neutral job satisfaction level but absence of these factors will lead to dissatisfaction. Motivators or satisfiers are the factors which satisfy and motivate employees such as achievement, recognition, advancement, responsibility, and work itself. These higher order needs correspond to Maslows level of selfactualization. Hygiene or maintenance factors consist of variables such as pay, security, and physical working conditions. In the same manner, the presence of hygiene factors does not produce feelings of satisfaction, but in their absence they do lead to job dissatisfaction. Alderfer (1972) suggests that individual needs can be divided into three groups: Building upon earlier need models (primarily Maslows) Clayton Alderfer proposed a modified need hierarchy the ER-G model with just three levels as existence, relatedness and growth, focuses on understanding what contributed to workers job satisfaction in industrial organizations. E-R-G model suggests that any of the need can be active at any time. Another theory related to job satisfaction is work adjustment theory. Work adjustment theory is based on the concept correspondence between the individual and work environment (Davis & Lofquist, 1984). Weiss, Dawis, England, & Lofquist (1967) state that work adjustment depends on how well an individuals abilities correspond to the ability requirements in work, and how well his needs correspond to the reinforces available in the work environment Furthermore, Weiss et al.(1967) maintain that satisfaction and satisfactoriness are measurable indicators of work adjustment, and that they can be measured independently of each other. Gender is one of the most important demographic variables that receive huge attention in the literature. Gender differences also have been recognized as a factor in employees job satisfaction level. Findings on the impact of gender on job satisfaction are at present not consistent. While some of the findings suggest that there are no differences in the level of job satisfaction among men and women
177
(Mannheim, 1983), others do suggest that the expectations of working women in terms of job satisfaction are different from those of men (Martin & Hanson, 1985). Spector (1997) suggests that relationship with gender and job satisfaction is divergent. Mwange and McCaslin (1994), and Varca, Shaffer & McCauley (1983) found that male employees were more satisfied with their jobs than female employees. Conversely, Hodson (1989) and Kelly (1989) found that female employees have increased job satisfaction over males (Nestor, 2000). A number of empirical studies on job satisfaction have suggested that female workers have lower level of job satisfaction than their male counterparts because male officials dominate most of the public organizations (Kim, 1999). Another study by Hulin and Smith (1964) investigated 295 male workers and 163 female workers to determine the satisfaction level of employees. Results indicated male workers were more satisfied with their job than female workers. It has been suggested that reason for satisfaction or dissatisfaction is not the sex. It is determined by several other different factors which varies with gender such as pay, job level, or advancement opportunities. Hulin (1969) studied the effects of community characteristics on the job satisfaction of 470 male and female workers. Hulins (1969) results indicated that there were differences between males and females for the variables related to job and life satisfaction. Maynards (1986) surveyed 338 employees satisfaction level with work and related support networks, no significant differences were found between male and female employees. Ivancevich and Donnelly (1968) proposed that like other researchers that it is not gender differences that lead to job satisfaction but variations in societal treatment such as different compensation scales for males and females (DeMato, 2001). Results from Hulins (1969) study were not fully supported by the results of the Herzberg et al. (1957) study. There are no simple conclusions about the differences between males and females and their job satisfaction levels. Some studies reviewed by Herzberg et al. (1957) indicate that males are more satisfied with their jobs, while others indicate that females are more satisfied. Educational level is not clear either. Furthermore, these studies showed that workers with more education have a higher job satisfaction level, while other studies indicate that workers with more education have a lower job satisfaction level. Other studies showed no relationship between the two. Herzberg et al. (1957) suggested that a clear conclusion cannot be drawn concerning job satisfaction and its relationship to marital status, number of dependents, number of previous occupations, or ethnicity (Scott, Swortzel, & Taylor,2005). Clark (1997) put forward possible explanations for why women are happier at work than men. He found evidence for the expectations hypothesis, which says that women will be more satisfied because they expect less (cf. fit-hypothesis) from their job than men. These lower expectations may result from the poorer position of women in the labor market in the past. As the differences in the labor market between man and women diminish, so may disappear the gender difference in job satisfaction. Clarks results confirm this hypothesis because the gender difference disappears for young and highly educated employees and also for employees whose mother works outside. Those groups were likely to have higher expectations (i.e. the same expectations as their male counterparts) about their job. (Verhofstadt, 1998). One view of job satisfaction holds that women are satisfied with jobs in which they can interact with others in a supportive and cooperative way, even though the jobs may be only minimally demanding and challenging. The basis for this view is that women are socialized into values, attitudes, and behaviors that are communal in nature; whereas mens socialization reflects values and behaviors (Newby, J.E. 1999).
4. Data , Hypothesis, and Methodology

4.1. Data This study investigates the current level of job satisfaction of directors and undersecretaries in the North Cyprus public sector. Participants in this study are directors and undersecretaries of the North Cyprus public sector. Data were collected through face to face interviews with Minnesota Satisfaction Questionnaire short-form (Weiss, Dawis, English, and Lofquist,1967).
178
The general purpose of this study is twofold. One, the study is to document facet specific and general job satisfaction levels directors and undersecretaries of the North Cyprus public sector. Two, the study is to investigate whether a relationship existed between selected demographic variables and general job satisfaction. 4.2. Hypotheses Based on the work The Minnesota Satisfaction Questionnaire (MSQ) developed by Weiss, Dawis, English, and Lofquist (1967) hypotheses to be tested are; 1. There is a positive relationship (linear) between gender and general job satisfaction of the North Cyprus public sector managers; 2. There is a positive relationship between intrinsic factors and general job satisfaction of managers in North Cyprus public sector; a. Ability utilization b. Achievement c. Activity d. Advancement e. Compensation f. Co-workers g. Creativity h. Independence i. Moral values j. social service k. Social status l. Working conditions 3. There is a positive relationship between extrinsic factors and general job satisfaction of managers in North Cyprus public sector; a. Authority b. Government polices and practices c. Recognition d. Responsibility e. Security f. Variety 4.3. Sample and Data Collection The population for this study is comprised of directors and undersecretaries of Northern Cyprus. According to the Office of the Prime Ministry (2005) there is a total of 111 federal government directors and undersecretaries. 4.4. Measurement of Variables 4.4.1. The Job Satisfaction Variables (Independent Variables): Independent variables studied as intrinsic and extrinsic factors. Intrinsic Factors (Independent Variables): Ability utilization: The chance to do something that makes use of my abilities Achievement: The feeling of accomplishment I get from the job Activity: Being able to keep busy all the time Advancement: The chances for advancement on this job Compensation: My pay and the amount of work I do Coworkers: The way my coworkers get along with each other Creativity: The chance to try my own method of doing the job
179
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Independence: The chance to work alone on the job Moral values: Being able to do things that don't go against my conscience Social service: The chance to do things for other people Social status: The chance to be "somebody" in the community Working conditions: The working conditions Extrinsic Factors (Independent Variables): Authority: The chance to tell other people what to do Company policies and practices: The way company policies are put into practice Recognition: The praise I get for doing a good job. Responsibility: The freedom to use my own judgment Security: The way my job provides for steady employment Variety: The chance to do different things from time to time
4.4.2. Job Satisfaction Variable (Dependent Variable): The dependent variable is job satisfaction. In MSQ job satisfaction is composed of intrinsic satisfaction and extrinsic satisfaction and this study will be consistent with MSQ. 4.4.3. Demographic Factors Gender Age Education Length of Service 4.5. Data Analysis Data was collected with the Minnesota Satisfaction Questionnaire and the Statistical Package for the Social Sciences (SPSS) 13.0 used for the statistical analysis.
5. Results
5.1. General Findings There were more male respondents than female respondents. The number of male respondents was 74 (79.6%) and the number of female respondents was 19 (20.4%). The majority of the directors/undersecretaries, 61 (65.6%), obtained an undergraduate degree and 22 (23.7%) obtained masters degree. The number of high school graduate respondents was 6 (6.5.6). Very few chairs 4 (4.3%) earned doctorate degree. The largest 51 (54.8%) of respondents had been director/undersecretary for 1-3 years. Twenty one (22.6%) of respondents had been directors/undersecretaries less than 1 year, 8 (8.6%) of respondents had been directors/undersecretaries for 3-5 years, 8 (8.6%) of respondents had been directors/undersecretaries for 5-10 years and 5 (5.4%) of respondents had been directors/undersecretaries for 10 years or more. majority of the directors/undersecretaries, worked for ministry of interior 13 (14.0%) and ministry of national education and culture13 (14.0%). The number of respondents as directors was 82 (88.2%) and the number of respondents as undersecretaries was11 (11.8%). 5.2. Results for Directors and Undersecretaries General Job Satisfaction 5.2.1. Gender and Job Satisfaction As Table 1 indicates there is statistically significant relationship between male directors and undersecretaries and 20 dimensions of MSQ (at p<0.05) except for two dimensions. There were no statistically significant relationship between male directors and undersecretaries extrinsic satisfaction and status and also there were no statistically significant relationship between male directors and undersecretaries extrinsic satisfaction and compensation.
180
Table 1:

Male Correlation Coefficients
Intrinsic 567 606 369 395 .526 .582 .300 .210 .536 .271 .493 .343 .386 .444 .546 .543 .718 .412 .583 .642 Extrinsic 370 373 454 .161 .555 .553 .241 .510 .608 .441 .524 .567 .186 .326 .619 .532 .510 .277 .619 .613 General 528 556 429 .330 .574 .611 .297 .344 .602 .357 .540 .456 .333 .428 .613 .576 .685 .387 .638 .675
Activity Independence Variety Status Super HR Super tech Moral Security Social services Authority Ability Comp policies Compensation Advancement Responsibility Creativity Work conditions Co- workers Recognition Achievement
Table 2 also indicated that 6 dimensions of MSQ (variety, supervision human resources, supervision technical, security, government policies and achievement) were statistically significantly related with general satisfaction for female directors and undersecretaries. 14 dimensions of MSQ (activity, independence, status, moral values, social services, authority, ability, compensation, advancement, and responsibility, creativity, working conditions, co-workers and recognition) were not statistically significantly related with general satisfaction for female directors and undersecretaries. Male directors and undersecretaries were more satisfied than female secretaries and undersecretaries but there were satisfaction both for males and females.
Table 2: Female Correlation Coefficients
Intrinsic .008 .499 .566 .377 .386 .454 .344 .514 -.256 .132 .167 .375 .166 .415 .236 .342 .511 .492 .377 .480 Extrinsic -.177 .376 .788 .168 .404 .460 .378 .539 .135 .243 .564 .488 .057 .379 .574 .516* .077 .119 .464 .648 General -.076 .483 .714 .311 .425 .494 .387 .566 -.097 .194 .363 .457 .129 .431 .410 .449 .353 .360 .422 .596*
181
5.3. Results for intrinsic factors and general job satisfaction of directors and Undersecretaries Satisfaction scores for extrinsic dimensions of the MSQ were analyzed according to the extrinsic variables; authority, government policies and practices, recognition, responsibility and variety. Table 3 presents correlation relationship supporting the relationship between intrinsic satisfaction, extrinsic job satisfaction and general job satisfaction. Table 3 indicates positive statistically significant relationship between intrinsic scores and determinants of job satisfaction for ability utilization (0.000), achievement (0.000), activity (0.000), advancement (0.000), compensation (0.003), co-corkers (0.000), creativity (0.000), independence (0.000), moral values (0.005), social service (0.000), social status (0.007) and working conditions (0.000) at (p<0.05). 5.3.1. Results for extrinsic factors and general job satisfaction of directors and Undersecretaries Satisfaction scores for extrinsic dimensions of the MSQ were analyzed according to the independent variables, authority, government policies and practices, recognition, responsibility and variety. Table 3 also indicates positive statistically significant relationship between extrinsic scores and determinants of job satisfaction for authority (0.000), Government polices (0.000), recognition (0.000), responsibility (0.000), security (0.000), and variety (0.000) at (p<0.05). Positive statistically significant relationship between general job satisfaction and 20 dimensions of MSQ can also be seen in table 3.
Table 3: General Correlation Coefficients
Intrinsic .486 .576 .403 .389 .494 .546 .307 .234 .432 .252 .432 .347 .360 .435 .506 .512 .675 .426 .548 .601 Extrinsic 287 .372 .552 161 .521 .528 .268 .504 .532 .410 .532 .551 .169 .338 .613 .530 .422 .248 .594 .616 General .440 .534 .480 .324 .541 .578 .313 .361 .504 .334 .503 .455 .309 .427 .586 .556 .620 .384 .606 .650
6. Conclusion and Discussion

Directors and undersecretaries job satisfaction is measured with 20 different dimensions of the MSQ. Data collected provides intrinsic, extrinsic and general job satisfaction level of directors and undersecretaries. The full population is studied, and the response rate of the study is 84%. Of the respondents 79.6% are male and 20.4% of the respondents are female. There is a consistency with this study and previous studies that male employees are dominating the public sector.
182
The results of the study suggest that directors and undersecretaries of TRNC public sector are generally satisfied with their present job. The first five mean results of the study suggest that out of the twenty dimensions of directors and undersecretaries, social services, moral values and authority and achievement are important respectively. Directors and undersecretaries derive the most satisfaction from social services which indicate they are mostly satisfied with having chances to help other people. Helping other people is a part of directors and undersecretaries job but was not expected to be the most important one. It can be said that directors and undersecretaries derive satisfaction from helping people both inside and outside the organization which has a positive influence on other people. Authority assigned by the organization to the directors and undersecretaries provides more chance to help other people and satisfy their needs. Authority is used to delegate some work to employees and complete the process in the best way. Results of correlation relation statistics also indicates that job satisfaction levels have significant differences between male and female respondents. Studying gender differences was not the main aim of this study however it is crucial to mention the significant difference between male and female directors and undersecretaries. In general, male respondents are more satisfied then females. Correlation relation results between men and women indicates that there is a significant correlation relationship between the twenty dimensions of the questionnaire and general, intrinsic and extrinsic job satisfaction of male directors and undersecretaries. Opposite results are indicated for female directors and undersecretaries. Females are not statistically significantly related to the twenty dimensions of the questionnaire. There is a statistically significant relationship with variety, security and achievement and general job satisfaction for female directors and undersecretaries. Female directors and undersecretaries realize the importance of achievement to secure their jobs in the public sector. They strongly believe that they have to achieve more than their male colleagues to secure their present job. For the rest of the dimensions of questionnaire either there is negative relationship or no statistically significant relationship for female directors and undersecretaries. This finding is somewhat surprising; given the general feeling that women in TRNC are treaty equally both in society and organizations; however lower levels of satisfaction displayed by female respondents must be of serious concern of policy makers. Consistent with earlier empirical evidence in this study older respondents are more satisfied than younger respondents. This may be because older employees gets use to the environment and have lower expectations compared to younger employees.
183
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Alderfer, C. P. (1972). Existence Relatedness and Growth: Human Needs in Organizational Settings. New York: Free Press. Bengston, D.S. & Shields, D. (1985). A test of Marchants predictive formulas involving job satisfaction. Journal of Academic Librarianship 11(May), 88 - 92. Bruce, Willa M., and J. Walton Blackburn. Balancing Job Satisfaction & Performance: A Guide for Human Resource Professionals. Westport, Conn.: Quorum Books, 1992. Dawis, R. V., & Lofquist, L. H. (1984). A psychological theory of work adjustment: An individual-differences model and its applications. Minneapolis, MN: University of Minnesota Press. DeMato, S. D. (2001). Job satisfaction among elementary school counselors in Virginia. Unpublished doctoral dissertation, Virginia Polytechnic Institute and State University, Blacksburg, Virginia Gruneberg, M. M. (1979). Understanding job satisfaction. New York: Macmillan. Herzberg, F., Mausner, B., & Snyderman, B. (1959). The motivation to work. New York: John Wiley and Sons, Inc. Hoppock, R. (1935). Job Satisfaction. New York: Harper Brothers Ivancevich, J. M., & Donnelly, J. H. (1968). Job satisfaction research: A manageable guide for practitioners. Personnel Journal, 47, 172-177. Krejcie, R., & Morgan, D. (1970). Determining sample size for research activities. Educational and Psychological Measurement, 30, 607-610. Marchant, M (1970). The effects of the decision-making process and related organizational factors on alternative measures of performance in university libraries. Dissertation Abstracts International 31(12), 6639. (UMI No.7115228) Maslow, A. H. (1954). Motivation and personality. New York: Harper and Row. Mc Ginn, F. H. (2003). An investigation into factors that influence job Satisfaction of African American librarians in some urban public library systems in the United States. Unpublished doctoral dissertation, Emporia State University, Emporia, Kansas Newstorm, W.J. and Kaith, D. (2002) Human Behavior at Work. NY: McGraw-Hill. Poling, R. L. (1990). Factors associated with job satisfaction of faculty members at a land-grant university (Doctoral dissertation, The Ohio State University, 1990). Summary of Research in Extension, 5, 143. Scott, M., Swortzel, A.K., & Taylor, N.W. (2005). The Relationships between selected demographic factors and the level of job satisfaction of extension agents. Journal of Southern Agricultural Education Research, 55 (1), 102-115. Spector, P. E. (1985). Measurement of human service staff satisfaction: Development of the Job Satisfaction Survey. American Journal of Community Psychology, 13, 693-713. Weiss, D. J., Dawis, R. V., England, G. W., & Lofquist, L. H. (1967). Manual for the Minnesota Satisfaction Questionnaire. Minneapolis, MN: The University of Minnesota Press.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18]
Management of Financial Margin Through Funds Transfer Pricing

Jordi Carenys Fuster Professor, EADA (Escuela de Alta Direccin y Administracin) Arag 204 08011, Barcelona. Spain E-mail: jcarenys@eada.edu Abstract Funds transfer pricing is an accounting and management control tool that is widely used in the banking industry. Its usefulness for analysing and managing the financial margin of financial services companies is widely acknowledged and it also serves as a guide for decision making in banking commercial policy making. This document aims to look at the basic theoretical tenets behind the use of transfer pricing in banking institutions as well as different possible ways of calculating it. We will then go on to describe how it is used by Spanish banking institutions that operate under the legal form of savings banks. The study finishes off by drawing up conclusions on how to improve the use of this management planning and control tool in these types of companies.
1. Introduction
The banking sector has undergone deep transformations over the last two decades as a consequence of deregulation, disintermediation, innovation and European Monetary Union. Faced with these radical changes, banking companies have had to redesign their strategies and reorganise their structures in order to address the new conditions of competition. Income revenue derived from subsidiaries devoted to investment fund management, pension and insurance schemes is gaining increasingly more and more weight. At the same time, banks have increased their range of products and services which generate commissions for services other than those derived from financial mediation. Although revenue from commissions has gained substantially more weight in the global results of banking companies in recent years, about 80% of the total profits of these companies are still generated by profit margins from financial intermediation. Financial intermediation focuses on the one hand on borrowing funds from customers via deposit contracts in exchange for remuneration, and on the other hand, carrying out asset operations with depositor funds which are generally related to the issuing of loans. The deposit institutions take on the risks derived from these operations (loan risk, interest and liquidity risks, among others) and the transaction costs ensuing from these operations (deposit taking, solvency assessment of loan applicants, management of the receipt of loans or means of payment services). The difference between the interest rate that the banking institution charges its loan applicants and the interest rate it pays on deposits constitutes what is commonly known as the financial margin, which must cover the bank's operating expenses, remuneration of company capital and its selffinancing needs. Hence, the financial margin of the deposit institution is determined by the difference between the financial products produced by its assets and the financial costs generated by its deposits and other payables. However, when for planning and management control purposes, the institution is divided into different cost objects (i.e. products, customers or branches) the financial margins generated by these
185
cannot be calculated as the simple difference between financial turnover and costs, because the cost objects do not maintain a balanced financial position but may appear as users or generators of funds. The margin of these cost objects is calculated taking into account the corresponding financial revenue generated by the assets they maintain and the financial costs produced by their liabilities, which yields a margin derived from their commercial activity but that does not accurately reflect the financial margin they generate in their operations because it does not include yields from surplus funds from creditor units neither the cost of the additional funds necessary to finance the debtor units (Sovignet, 1989, p. 720). For example if we analyse the profitability of savings deposit products., these always show a loss given that the financial cost they generate is not accompanied by the yields obtained by the banking institution from investing these deposited funds. As Ormaecheverra points out (1994, p. 221) "in order to obtain a financial margin from cost objects, management accounting must balance out the alternation of debtor and creditor positions by means of an internal funds transfer system, by means of which the surplus resources from creditor units are remunerated at a pre-set price and the resources lent out to debtor units will entail a cost to the latter based on the transfer rates. This will enable the bank to balance the balance sheet of each unit and draw up a profit and loss account for cost objects that are fund users as well as for those that are fund generators, thus setting real and homogeneous financial margins for all the units." Given that the financial margin constitutes the principle component of a banking institution's final profit, its allocation to different cost objects via a funds transfer pricing system is a key element in the planning and control system of banking institutions (Rout and Kochvar, 1994, page 25). In order for this funds transfer system to work, we need to establish the debtor/creditor position of the cost objects whose financial margin we want to calculate. Furthermore, it requires an internal treasury unit (a funds pool) which acts as an intermediary between the units that manage surplus and deficit operations. Deficit positions are covered by funds received from the pool, whereas the surplus units pour their surpluses into the pool (Pedraja, 1993, p. 91-92, Warner and DAlessandro, 1993, p. 22; Kimball, 1997, p. 25-26). Before we can apply transfer prices to the funds transferred between the different business units and the treasury pool we first of all need to quantify the total sum of all the funds that are being mobilised. We will look at this aspect in the following section.
2. Allocation of funds to cost objects

In order to calculate the financial margin at the cost object level, the first issue we need to address is how we can match the cost objects with the funds they generate or use. There are various ways to measure and evaluate the funds that are exchanged between the management units and the internal treasury [(BAI (1972, p. 157-178), Fitz (1995, p.74-93), AECA (1994, p. 33-60), Rouach and Naulleau (1994, p.107-118), Cole (1995, p.176-184), De la Cuesta (1996, p. 64-68) and Kimball, 1997, p. 2529)]: Net cash flows Gross cash flows 2.1. Allocation of funds based on net cash flows In a net cash flows system there is supposed to be total homogeneity between deposited and invested funds, which allows us to compensate the resources and uses of different cost objects in order to estimate the deficit or surplus of each object (De la Cuesta, 1996, p. 66). Hence, from the difference between assets and liability, a balance is drawn up that reflects the funds that cost objects transfer to the internal treasury in order to invest their surplus balance or that they borrow in order to finance their deficit balance (Bonnier and Simon, 1994, p. 63). See Figure 1. for a graphic representation.
186
In Figure 1 we can see that branch A's investment volume is greater than the deposits it has captured. The ensuing imbalance is covered by resorting to the treasury pool which furnishes the necessary funds in exchange for which the branch must pay a transfer price. In the case of branch B, the situation is the other way round, given that its captured funds exceed its investment volume. The excess funds it generates are poured into the treasury pool which pays a transfer price to the branch in exchange for the received funds. When operating with net cash flows, the impact of transfer prices is applied solely on the balance of the resulting financial position and not on the total assets and liabilities generated by the cost object. Hence, a net cash flows system implies that the internal pool takes on a passive role, limited to covering treasury deficits or the accumulation of surpluses, whereas the primary role in the decision making process concerning deposit taking or issuing of loans falls on the people in charge of cost object management (Rouach and Naulleau, 1994, p. 112-113).
Figure 1: Net funds system
2.2. Allocation of funds based on gross cash flows In a gross cash flows system all customer operations are translated into opposed operations between the cost object and the funds pool. Hence there is no compensation between the resources and uses of each management unit. Instead, the total assets are refinanced by the pool at a transfer rate and all the liabilities are transferred to the pool at a transfer price (Sovignet, 1989, p. 496). The hypothesis of homogeneity of assets and liabilities is therefore not applied. Instead, the business units transfer on mass to the internal treasury all their captured resources and borrow from the treasury all the funds needed to cover their investments. It is therefore possible to separate the investment and savings components that make up the financial margin of cost objects (De la Cuesta, 1996, p. 66). For a graphic representation see Figure 2.
187

Figure 2: Gross funds system
It is worth noting that although the gross cash flows method is more complex to implement and manage it does nevertheless have significant advantages over the net cash flows method1: When transfer prices are applied on the total assets and liabilities instead of only on the balance after compensating assets and liabilities, the impact of transfer pricing on the investment or deposit taking decision making process is greater. Hence, the gross cash flows method enhances the influence that the commercial and financial managements of deposit institutions exercise on the management decisions of decentralised business units, because it gives them more influence over the structure of the company's assets and liabilities. (Simon and Bonnier, 1994, p. 64), The gross cash flows method enables us to break down the financial margin of the cost object into its two fundamental components: the margin generated by asset operations and the margin generated by liability operations. On the other hand, the net cash flows method does not allow us to distinguish which part of the financial margin generated by a cost object comes from asset operations and which part comes from liability operations (Rouach and Naulleau, 1994, p. 112113). The choice of one or the other alternative depends to a great extent on the degree of decentralisation within the institution given that larger banking institutions with highly developed treasury departments may opt to measure financial cash flows in gross terms because the treasury department in this case acts as if it were a real internal money market. In contrast, in small or very decentralised institutions the best alternative may be to measure cash flow in net terms, because in compensating the applications with the resources of each cost object, it leaves the funds pool with a marginal reinvestment or coverage function for the balances of each unit. (De la Cuesta, 1996, p. 67).
3. Fund transfer pricing in banking

After having looked at the various alternatives for allocating funds to cost objects let us now go on to consider the price at which the treasury pool is going to remunerate the resources poured into the pool by the business units or the other way round, the price at which it is going to charge for the cost of funds it lends out and the different possible options for setting the transfer rate or price. We shall also look at the objectives of the transfer pricing system in banking institutions.
1
It should be pointed out that the financial margin of cost objects obtained by applying net and gross funds procedures only coincides if we use a single funds transfer rate for both asset and liability operations. Therefore, it only makes sense to use the gross funds method in conjunction with various transfer prices, because otherwise, we could obtain the same financial margin via a simpler procedure (AECA, 1994, p. 53).
188
The funds cost/yield assessment is one of the most crucial aspects in measuring the financial margin at a customer, product or centre of responsibility level, given that it is especially sensitive to changes in transfer pricing and it also directly influences the investment or deposit-taking decision making process (Tewes, 1976, p. 27; Lucien, 1979, p. 214; Graddy and Spencer, 1990, p. 385; Fitz, 1995, p.74). 3.1. The funds transfer pricing system's objectives We have already mentioned the importance of using funds transfer pricing in order to calculate the financial margin of the cost objects that a banking institution can be broken down into for management purposes. Listed below are the other objectives of implementing a funds transfer pricing system in deposit institutions: To motivate managers to achieve the organisation's objectives as regards asset and liabilities management and commercial policy (p. 97; D Alessandro and Warner, 1993, p. 22-23; Ernst and Young, 1995, p. 175). They act as a lever making certain products more or less attractive to the heads of business units because they allow the institution to come closer to its structural objective of an ideal balance. Management accounting therefore uses transfer price allocation as a tool that contributes to achieving the institution's general objectives. To contribute to the performance appraisal of the management units and the people in charge of them (DAlessandro and Warner, 1993, p. 22-23; Fitz, 1995, p. 74-75; Ormaecheverra, 1994, p. 221; Ernst and Young, 1995, p. 175 and Kimball, 1997, p. 25-28), To facilitate analysis of the components that make up the financial margin of deposit institutions (D Alessandro and Warner, 1993, p. 22-23; Kimball, 1997, p. 25-28). The bank's capacity to generate its financial margin depends on three factors: The first factor has to do with the bank's capacity to place assets on financial markets at a higher price than the marginal cost of the funds that finance them. The second factor is the possibility of capturing liabilities at a lower price than the marginal cost of funds of the same maturity. The third factor is the effect that exposure to interest rate risk has on financial margin, due to the different maturity structure of profit and loss operations of the company's profit and loss account. In figure 3, the straight line AB represents interest rates for assets the bank has placed on the market. The straight line CD represents the cost of deposits and other liabilities for the bank. Finally, the straight line EF represents market interest rates at which the bank can place funds or capture them if need be, which we could identify as the market price (or marginal revenue/cost). This represents the opportunity cost of funds for each date of maturity. Kimball (1997, p. 27-28) analyses the breakdown of the financial margin of a deposit institution taking into account three lines of business, the first originates loans at one year, the second deposits at one year and the third is a treasury department in charge of managing assets and liabilities. It is assumed that the volume of mobilised funds for deposit-taking and loans coincides, so it is not strictly necessary to access the interbank market. In line with the above line of thinking, if the deposit taking business unit sells these funds to the treasury department at price H, it will obtain profit margin HI. If the business unit that originates loans finances them by purchasing the necessary funds from the treasury department, it will obtain a margin equivalent to GH. However, the treasury department, instead of adopting a passive stance, if it considers that market interest rates (and their corresponding yield curve) for the next year will remain at present levels, may opt to finance the loans with funds at three months maturity, which costs less, and to renew them every three months. Hence it sells funds captured at one year on the external market at price H, and purchases an equivalent sum of funds at three months at price J (renewable every term). If interest rates are expected to remain stable, this will yield an additional financial margin equivalent to (HJ). If we assume that the spread on the threemonth funds (JK) is equal to the spread on the one-year funds (HI), then the additional net interest margin derived from assets and liabilities management is equal to the distance (IK). Hence the bank's
189
total net interest margin (GK) can be broken down into its components, the margin earned from deposit taking, HI, the margin derived from fund placement (GH) and the margins from assets/liabilities management (IK).
Figure 3: Breakdown of the financial margin through transfer pricing. Source: Kimball (1997, p. 27).
The following section looks at the different alternatives put forward in the literature for setting funds transfer prices in banking institutions. To begin with, banks have the option of using a single rate or a multiple rates system (Fitz, 1995, p. 75-77; De la Cuesta, 1996, p. 69; Kimball, 1997, p. 26). In both cases, there are two ways the price can be fixed: by resorting to external market prices or associating the price to the average liability cost, which involves linking it to the financial and productive structure of each specific institution (Tewes, 1976, p. 27-33; Graddy and Spencer, 1990, p. 385; Bonnier and Simon, 1994, p. 64; De la Cuesta, 1996, p.70; AECA, 1994, p. 39). The possible options are displayed in Figure 4. 3.2. Single transfer price In a single transfer price setup, all the funds are assessed at a single price regardless of whether they are funds lent to the pool or funds borrowed from the pool, of their maturity period or other characteristics of financial assets (Rouach and Naulleau, 1994, p.119). As we have mentioned previously, this price can be fixed internally by calculating the average costs (historical and forecasted) or set in line with observed prices on financial markets. We shall look at the peculiarities of each of these alternatives later on in the following sections.
Figure 4: Funds transfer pricing. Source: original material.
Price origin Costs Market Single price based on Single price based on the costs market Multiple prices based on Multiple prices based on costs the market
Single Price # of Prices Multiple prices
190
3.2.1. Single transfer price based on average cost Let's begin by looking at the option of estimating the transfer price from the savings bank's own average financing costs. In this case we need to define what we mean by average cost. We can point to the following alternative definitions (Sartoris, 1971, p. 34; Tewes, 1976, p. 27-28; Fitz, 1995, p. 7577): Average cost of liabilities: this approach takes into account all liabilities (including equity) when calculating the transfer price. In this case, we need to allocate the use of equity that corresponds to each cost object depending on the risk level it involves and to calculate what cost is assigned to the equity that is used up by each cost object. It should be pointed out that in this case, the information obtained from the allocation of funds transfer prices goes beyond the calculation of the financial margin generated by each business unit in its operations, because it includes the opportunity cost of equity. Average cost of funds with financial cost: in this case only the average balances of interest bearing liabilities are taken into account, excluding all liabilities that do not generate financial costs. During periods of rapid interest rate changes, the transfer price calculated in this way will fluctuate in a more synchronised way with market prices than it would if it were calculated as the average cost of all funds, thus leading to better use of resources. Average cost of interest sensitive funds: this includes only those funds sources whose cost is influenced by market interest rate fluctuations and does not take into account liabilities that have been contracted at a fixed interest whose cost does not vary with market fluctuations. In this way market interest rate fluctuations are transferred even more directly to the transfer rates. As market interest rates fluctuate, these changes are transferred to transfer prices based on average costs, although to varying degrees, given that, for example, the average cost of liability has a much more stable transfer cost than the average cost of interest sensitive liabilities which reflect market interest rate fluctuations much more rapidly. (Fitz, 1995, p.76) When using a single transfer price based on average cost we also need to consider which resources should be included as part of the cost of credit for funds provided. There are basically two possibilities (Sartoris, 1971, p. 33-39; Bonnier and Simon, 1994, p. 64): To include only financial costs generated from deposit taking, which requires the introduction of a methodology based on the direct cost of funds. To also include transformation costs associated with real resources (personnel, branchs, advertising...) needed for deposit-taking and in this way approximate the transfer price to the complete cost of funds. In this respect, given that transformation costs are to a great extent fixed and indirect in relation to cost objectives, their inclusion in transfer prices may lead to decision making errors concerning deposit taking or investments which may lead to an unwanted balance sheet structure (Sartoris, 1971, p. 33). The main problem of using transfer rates based on average cost, no matter which way they are calculated, is that this doesn't reflect the true opportunity cost of funds. In other words, a banking institution cannot purchase funds at an average cost (nor can it make investments at an average profitability) given that the profitability of each operation is influenced only by the marginal costs and revenue derived from each decision (Tewes, 1976, p. 29). Given that the average cost dilutes the effect of marginal decisions concerning investment or deposit taking, the real economic impact of each decision is hidden by the total number of operations. Hence, a non profitable decision may appear to be profitable and vice versa (Graddy and Spencer, 1990, p.388). The transfer price based on average cost will tend to approach the marginal cost only in those cases where the composition and structure of fund sources remains stable and there are no interest rate fluctuations (Graddy and Spencer, 1990, p.388). 3.2.2. Single transfer price based on marginal cost The use of marginal cost as a transfer rate fulfils the basic function of maximising the profitability of investment or deposit taking decisions, ensuring economically correct decision making in the absence
191
of other objectives other than profit maximisation (Tewes, 1976, p. 29-32; Lucien, 1979, p. 24). The use of marginal cost is based on the tenet that each operation must be analysed according to the additional costs and revenues it generates (Graddy and Spencer, 1990, p.389). In deposit operations, captured funds can always be invested on the interbank market, obtaining a marginal revenue which is at least equivalent to the market price, or it may be allocated to reducing interbank debt, thus decreasing its financial costs by an amount equivalent to the market price (or alternatively, if the aim is to capture funds, this can always be done at a cost equivalent to the marginal cost). Hence, a transfer rate based on the marginal cost of funds will basically reflect the market price of money, which for the savings bank coincides with the opportunity cost of these funds. (De la Cuesta, 1996, p.70). The option of considering the market price as an approximation to the marginal cost is perfectly feasible in deposit institutions because the interbank market is so extensive that the price it presents can be used as a transfer price. What's more, given that it is calculated externally from accounting criteria, this limits the arbitrariness derived from different accounting alternatives used when calculating transfer prices based on average costs (Waden Berghe, 1993, p. 31). Using marginal cost as the transfer rate has a great advantage over using average cost, because it approximates the transfer price to the opportunity cost of funds. However, the use of a single rate presents significant drawbacks because the same transfer price is applied to short, medium and long term financial operations whereas the interest rate yield curve reflects different opportunity costs for different maturity periods. This makes the funds pricing transfer system less coherent because market interest rates for long term operations do not coincide with those for short term operations (Rouach and Naulleau, 1994, p.119; Bonnier and Simon, 1990, p. 384). This therefore leads to distortions when calculating profitability because it generates crossed subsidies between asset and liability products with different maturity or repricing periods (Villalonga y Passera, 1985, p. 137). Figure 5 reflects this state of affairs in graph form. Figure 5 shows us that when a single transfer rate based on market prices is used, assets with maturities below "n" periods and liabilities with interest maturity above "n" periods are penalised. This favours assets and liabilities with maturities above and below "n" periods respectively. What's more, if the managers in charge of business units are evaluated based on the results of their business units, they will try to maximise their financial margin, which will give them a clear incentive to favour and/or rule out certain maturities for their asset and liability operations. The company may therefore find that its assets and liabilities structure has a greater liquidity risk and is more exposed to interest rate fluctuations (Ernst and Young, 1995, p.175).
Figure 5: Single marginal transfer price Source: Villalonga and Passera, 1985, p. 137.
3.3. Multiple transfer prices In order to avoid the abovementioned drawbacks it has been suggested that the transfer price should reflect the maturity or repricing period for each specific operation, which essentially involves using multiple transfer prices (Lucien, 1979, p. 23-25). However, if different prices are applied, we still need to determine whether these are to remain fixed throughout the financial product's life or whether they
192
should be reviewed periodically. The financial margin of a business unit includes both revenue and the costs of operations closed in previous years as well as transactions carried out in the current financial year. For example, if a profit centre has been operating for a number of years, its profit and loss account includes revenue as well as financial costs incurred from mortgages issued in the current financial year as well as those from previous years (Lucien, 1979, p. 36). Management accounting needs to address the issue of whether to apply current transfer rates to all the operations of a profit making centre even though they were produced during previous financial years. If this is the case, the profitability of these operations is assessed using different rates to those applicable at the time the operation took place and this can have a significant effect on the business unit's financial margin without any intervention on the manager's part (Bonnier and Simon, 1994, p. 64). On the other hand, if each operation carried out by a business unit is allocated the price corresponding to the financial year in which the operation took place, and this price is then maintained throughout its entire lifespan in the case of fixed rate operations, or the price is updated every time that benchmark rates are reviewed, in the case of variable rate operations, then we can effectively separate interest rate risk effects in the profit and loss accounts of business units (Tewes, 1976, p.32; Ernst and Young, 1995, p.179). If we opt to maintain the historical rates at the moment the operation took place, responsibility for interest rate risk associated with assets and liabilities that are sensitive to interest rate fluctuations is passed on to the treasury pool. The remaining business units will reflect in their financial margins only the part of profit and loss under their control, basically commercial activity and credit risk (Ernst and Young, 1995, p. 179). As in the case of the single rate, when implementing a multiple transfer price system one can set the rates based on average costs or marginal costs. We will now go on to look at the characteristics of these two different approaches. 3.3.1. Multiple transfer prices based on average costs This procedure assumes that it is possible to group assets and liabilities according to certain characteristics, such as the repricing period, their liquidity or their interest rate sensitivity [(BAI, 1972, p. 169, 173; Graddy and Spencer, 1990, p. 394-397; Rouach and Naulleau, 1994, p. 119-123; De la Cuesta, 1996, p.74)]. A widely accepted practice in the literature is to set up funds pools that group together asset and liability operations which have similar maturity or repricing dates and to then apply the average cost of each funds category as the transfer price. 3.3.2. Multiple transfer prices based on marginal costs With this approach, the observable interest rate on the market that has been established for each fund category is used as the transfer price for that category. (Fitz,1995, p. 77; Lucien, 1979, p. 24). In this way, the transfer price for short term operations will be based on the market price for short term funds. (De la Cuesta, 1996, p. 77-78). As DAlessandro and Warner (1993, p. 26) or Cole (1995, p. 181) point out, the multiple prices method, because it uses different prices, is a closer approximation to the market interest rates curve.
4. Empirical study
In recent years margins have narrowed and this means that in order to improve management, these companies increasingly need to calculate the financial margin more accurately for different cost objects and business units. Thus, funds allocation systems and funds transfer prices have become a crucial element of the planning and control systems of these types of companies. This section presents an empirical study that was carried out to assess the prevailing characteristics of funds transfer pricing systems in Spanish savings banks in order to find out how these companies use this mechanism to control their management.
193
4.1. Survey universe and methodology The universe for our research study is made up of 47 savings banks which were already registered on the Official Register of Banking Institutions of Spain in December 2004. We sent out 47 questionnaires and received 28 responses, which represents a 59.5% response rate. The total assets in December of 2004 of the 28 savings banks that responded to the survey represented 76% of the total assets of all savings banks in Spain. This response rate is comparable to those of similar research initiatives, both nationwide and international. For example, the BAI (Bank Administration Institute) of the United States carried out a survey in 1996 with the aim of finding out what accounting and management control systems were being used by North American banks. The BAI questionnaire was sent out by post to the 250 biggest banks in the United States, and 40 banks responded (16%) although respondents accounted for 43% of the sector's total assets. Also in the United States, Gadner and Lammers (1984, p. 34-39) carried out a survey on the application of management accounting between banks and saving and loans associations. They selected the 50 largest banks and the 20 largest saving and loans associations according to their asset volume and sent out 70 questionnaires through the post. They received responses from 49 institutions, with a response rate of 68%. In Europe, Innes and Mitchel (1997, p. 290-203) carried out a survey with postal questionnaires in order to find out to what extent the ABC costs system was being used by financial institutions in Great Britain. Innes and Mitchell sent out their questions to the 60 largest financial companies in Great Britain and the response rate was 51.6% (31 companies). Hence, the response rate to our questionnaire is comparable to (and in some cases greater than) those obtained in other international research studies on banking management accounting carried out using the same methodology. The responses were segmented according to company size, measured based on their total assets. They were divided into three categories: the larger savings banks with assets above six billion euros (segment A); medium sized savings banks with assets between six and three billion euros (segment B); and smaller savings banks with assets below three billion euros (segment C). The research method we used consisted of a questionnaire sent out by post to the heads of management control of all Spanish savings banks. In cases where the questions required the respondent to give their opinion as well as cases where the questions were not totally factual, the questionnaire made use of a Likert scale of 1 to 5. The questionnaire was sent out in December 2004 together with a research presentation letter kindly requesting the savings banks to collaborate in the survey. A reminder letter was sent out in February 2005 and the last responses were received in March 2005. Two variant analysis was used to tabulate this questionnaire, which is what is recommended for studying relationships between variables taken two by two (Pedret, 1997, p. 31). This type of analysis is useful for giving answers to questions such as for example, whether the size of the savings bank influences the costs system that is used or whether there is any relation between the size of the institution and the funds transfer system it uses. The data was treated using the programme SPSS version 6.12. In our case we used two variant analysis between two qualitative variables and contingency tables were also used, given that they are a most widely used tool yielding results that can be easily interpreted and understood. Contingency needs to be interpreted in the sense of dependence. Hence, a contingency table represents the way two or more characteristics depend on one another, in such a way that the body of the table is made up of the frequencies with which each combination of variables takes place (Wonnacott and Wonnacott, 1981, p. 445) 4.2. The survey findings The questionnaire was drawn up to find answers to the following questions: 1. What are the priority objectives of transfer pricing? What is it used for? 2. Do savings banks use a single transfer price regardless of funds characteristics or do they use a multiple pricing system? 3. Are the funds ceded/allocated to each cost object estimated as net or gross funds? 4. Is the transfer price set internally using analytical criteria or is it set by taking observed market prices as a reference?
194
From Table 1 we can see that savings banks use the internal funds transfer system basically in order to calculate the financial margins of their business units (4.58 points). All three savings bank segments coincide in this respect. This reason was the most important, followed by that of facilitating decentralisation of commercial decisions which received a 3.38 valuation mark on the 1 to 5 scale. In any case, this latter reason seems to be more relevant for savings banks in segments A and C (3.62 and 3.5 points respectively) than it was for those in segment B which gave it a much lower score. The objective of facilitating the breakdown of the financial margin into its components was the third most relevant criterion (3.31 points), especially for the large and medium sized savings banks (3.87 and 3.62 points respectively), whereas smaller savings banks considered it a less important reason (2.6 points). Finally, separating the financial margin of cost objects from interest rate risk was also referred to although as in the previous case this seemed to be more relevant for segments A and B.
Table 1: The usefulness of the internal funds transfer system.
Total Size of Savings Bank To find out the financial margin of centres, products and customers average To separate the financial margin of business units from interest rate risk average To facilitate decentralisation of commercial decisions average To breakdown the financial margin into its components average A 4.50 2.80 3.50 2.60 B 4.50 3.50 2.75 3.62 C 4.75 3.62 3.87 3.87 Total 4.58 3.27 3.38 3.31
In order to achieve the above objectives, savings banks can choose different techniques as regards the way funds are allocated to different objects as well as the prices they use to assess these funds transfers. Basically, to begin with, they can choose between adopting a single transfer price or a multiple price system. Table 2 displays our findings on this first aspect. According to our findings displayed in Table 2, 57.7% of savings banks use the multiple transfer pricing system whereas 42.3% use a single price for transferring funds. It is worth pointing out that the larger the savings bank the more likely it is to use a multiple pricing system. In fact, only one savings bank within Segment A uses a single transfer price system (this accounts for 9.1% of all banks that use this system). 4 banks from Segment B use the single price system making up 36.4% of the total and in Segment C with 6 banks, the percentage rises to 54.5%. We can arrive at the same conclusion by looking at the column percentages. 40% of Segment C uses various prices and in segment B the percentage rises to 50%. The multiple pricing system is much more widespread amongst the larger savings banks where 87.5% of the total use this approach.
Table 2: Funds transfer pricing in savings banks (I). Single or Multiple.
A Single Transfer Price Count Row Pct Col Pct Tot Pct Various Transfer Prices Count Row Pct Col Pct Tot Pct Total. Count Row Pct Col Pct Tot Pct 6 54.5% 60.0% 23.1% 4 26.7% 40.0% 15.4% 10 38.5% 100.0% 38.5% Total Size of Savings Bank B 4 36.4% 50.0% 15.4% 4 26.7% 50.0% 15.4% 8 30.8% 100.0% 30.8% C 1 9.1% 12.5% 3.8% 7 46.7% 87.5% 26.9% 8 30.8% 100.0% 30.8% Total 11 100.0% 42.3% 42.3% 15 100.0% 57.7% 57.7% 26 100.0% 100.0% 100.0%
195
When the bank opts to use a multiple pricing system it needs to decide whether these prices should be linked to the maturity dates of each operation they are applied to or whether they should be estimated using other variables. From Table 3 we can see that 14 out of the 15 savings banks that use various transfer prices (93.3%) calculate transfer prices according to the different maturity periods of operations. In a net cash flow system complete homogeneity should exist between deposit taking and invested funds, which enables us to compensate uses and applications of funds of different cost objects in order to go on and calculate the deficit or surplus of each object. In this way a balance is drawn up which reflects the funds that cost objects cede to the internal treasury to invest their balance surplus or that they borrow in order to finance their balance deficit. In a gross cash flows system, all operations with customers are regarded in terms of operations between the cost object and the funds pool. Hence the uses and applications of each business unit are not compensated. Instead, the total assets are refinanced by the pool at a transfer rate and all liabilities are ceded to the pool at a transfer price. Under this approach, the business units cede all captured resources on mass to the internal treasury and then borrow all the necessary funds they need to cover their investments. In this way it is possible to separate the investment and savings components when drawing up the financial margin of cost objects. Table 4 displays information concerning this aspect. In Table 4 we can see that the majority of savings banks (65.4%) use a gross cash flow system.
Table 3: Funds transfer pricing in savings banks (II). Related to maturity periods.
A Yes. Count Row Pct Col Pct Tot Pct No. Count Row Pct Col Pct Tot Pct Single Transfer Price. Count Row Pct Col Pct Tot Pct Total Count Row Pct Col Pct Tot Pct 4 28.6% 40.0% 15.4% Total Size of Savings Bank B 4 28.6% 50.0% 15.4% C 6 42.9% 75.0% 23.1% 1 100.0% 12.5% 3.8% 6 54.5% 60.0% 23.1% 10 38.5% 100.0% 38.5% 4 36.4% 50.0% 15.4% 8 30.8% 100.0% 30.8% 1 9.1% 12.5% 3.8% 8 30.8% 100.0% 30.8% Total 14 100.0% 53.8% 53.8% 1 100.0% 3.8% 3.8% 11 100.0% 42.3% 42.3% 26 100.0% 100.0% 100.0%
This procedure is most frequently used by larger savings banks. In fact, only one savings bank from Segment A allocates funds using the net cash flow system, which means that only 12.5% of the savings banks in this segment use the net cash flow procedure. A bigger proportion of savings banks in Segment B (25%) use the net cash flow procedure. The use of the net cash flow procedure is most widespread in smaller savings banks where 60% of the savings banks in this segment state that they use this procedure to allocate funds.
196
Table 4:

Allocation of funds to cost objects.
A Total Size Of Savings Bank B 2 22.2% 25.0% 7.7% 6 35.3% 75.0% 23.1% 8 30.8% 100.0% 30.8% C 1 11.1% 12.5% 3.8% 7 41.2% 87.5% 26.9% 8 30.8% 100.0% 30.8% Total 9 100.0% 34.6% 34.6% 17 100.0% 65.4% 65.4% 26 100.0% 100.0% 100.0%
Net Funds. Count Row Pct Col Pct Tot Pct Gross Funds. Count Row Pct Col Pct Tot Pct Total. Count Row Pct Col Pct Tot Pct
6 66.7% 60.0% 23.1% 4 23.5% 40.0% 15.4% 10 38.5% 100.0% 38.5%
Finally, savings banks need to decide on whether to base the prices they will apply to transferred funds on observed prices in the money market or to estimate transfer prices internally after analysing the bank's costs. The use of market prices is based on the strategy that each operation should solely be analysed according to the additional costs or revenue it generates. Hence transfer rates based on market prices primarily aim to maximise the profitability of investment decisions or deposit taking, thus ensuring correct economic decision making in the absence of objectives other than profit maximisation. The aim of transfer prices based on the savings bank's internal costs is to ensure that transfer prices do not only depend on money market conditions but that they also reflect the company's cost structure and its own particular competitive conditions. Practically all the savings banks (96.2%) stated that the transfer prices they use are based on market prices and there were no notable differences between segments regarding this point (see Table 5).
5. Conclusions
Savings banks make different choices among the various options described above depending on the use they wish to give to the information they obtain via the financial margin analysis they carry out. Presenting aggregate margins is acceptable for information used in strategic decision making, (Bonnier and Simon, 1994, p. 65; Cornet and Duplaa, 1997, p. 44). Hence, as far as customers are concerned, if we look at retail banking, in aggregate terms it is a net creditor, whereas corporate banking (for companies) is a net debtor. Similarly, liability products are generators of funds whereas asset products are fund users, which means that there is a de facto transfer of funds from one cost object to another and we need to economically evaluate this for strategic decision making. It is normally considered sufficient to identify these funds exchange relationships in aggregate terms with a single pool, operating with net funds and a single transfer rate, which corresponds to the market rate for the period in question (Kimball, 1988, p. 21, Bonnier and Simon, 1994, p. 65).
197
Table 5:

Funds transfer pricing in savings banks (III).
A Total Size of Savings Bank B 8 32.0% 100.0% 30.8% C 7 28.0% 87.5% 26.9% 1 100.0% 12.5% 3.8% 10 38.5% 100.0% 38.5% 8 30.8% 100.0% 30.8% 8 30.8% 100.0% 30.8% Total 25 100.0% 96.2% 96.2% 1 100.0% 3.8% 3.8% 26 100.0% 100.0% 100.0%
Market Prices. Count Row Pct Col Pct Tot Pct A Combination OF Both. Count Row Pct Col Pct Tot Pct Total. Count Row Pct Col Pct Tot Pct
10 40.0% 100.0% 38.5%
When it comes to measuring the performance of business units and the orientation of their decisions in matters of commercial management, the technical decisions that are best suited to very decentralised companies are listed below: a multiple pricing system which allows for better follow-up of funds cash flows based on maturity periods, that considers funds in net terms which leaves the business unit with a funds balance that it can manage. They should consider multiple transfer funds aligned to market rates, which rapidly transmit the conditions of the environment and which affect the business unit's deposit taking or investment placement as little as possible. Finally, they should take into account prevailing rates, encouraging business units to base their decisions on their expectations concerning future transfer rate and interest rate trends (Cornet and Duplaa, 1997, p. 45). This fund transfer and price transfer approach allocates a passive role to the internal funds market by granting a great degree of autonomy to the business units (Bonnier and Simon, 1994, p.65). From the point of view of more centralised management, the following technical options seem more appropriate (Bonnier and Simon, 1994, p. 65): a multiple pricing system that will enable the treasury to send out a much more directed deposit taking and investment policy to the business units; organisation based on gross cash flows given that the effects of transfer prices will be applied to all the asset and liability operations of the business units and not only on the balance between them. Their transfer prices should be based more on costs than on market rates, in order to guide long term business unit decision making and avoid transferring temporary fluctuations in market prices immediately. By applying historical rates they can avoid a situation whereby the business units take decisions based on their expectations concerning interest rate trends. In this way they can reserve interest rate risk management for the treasury pool (Cornet and Duplaa, 1997, p. 45.)
198
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) AECA [1994]: Principios de contabilidad de gestin: La contabilidad de gestin en las entidades de crdito. Document no. 9. Madrid. BAI [Bank Administration Institute] [1972]: Bank costs for planning and control, Park Ridge, Illinois. BONNIER, C., SIMON, C. [1994]: "Comment fixer les prix de cession interne de capitaux?", Banque, no. 40000029 PINMAR S.L. 63-65. BOS, J.J., BRUGGINK, B., IJSKES, E., [1994]: "Cost accounting in banking". XVII Congress of the European Accounting Association. COLE, L.P. [1995]: Management accounting for financial institutions, Probus Publishing, Chicago. CORNET, B., DUPLAA, C [1997]: "Mesurer la performance grce aux taux de cession internes", Revue Banque, issue no: 582, June, p. 44-46. DALESSANDRO, M.J., WARNER, J.L. [1993]: "Funds transfer pricing for retail deposits", Bank Accounting and Finance, August, p. 22-32. DE LA CUESTA, M. [1996]: El control de gestin en las entidades bancarias. Los modelos tradicionales y el ABM, UNED, Madrid. ERNST & YOUNG [1994]: Performance measurement for financial institutions: methods for managing business results, Irwin Professional Publishing, Chicago. FITZ, C [1995]: "Design a framework for a bank profitability management systems", Journal of Bank Cost and Management Accounting, vol.9, nm.2, pg.74-93. GADNER, M., LAMMERS, L.[1988]: "Cost accounting in large banks", Management Accounting [US], vol. 29, no.10, April, p. 34-39. GRADDY, D., SPENCER, H. [1990]: Managing commercial banks, Ed Prentice-Hall International Editions, Englewood Cliffs, New Jersey. INNES, J., MITCHELL, F. [1997]: "The application of activity-based costing in the United Kingdom's largest financial institutions", The service industries journal, vol.17, no.1, January, p.190-203. KIMBALL, R.: [1998]: "Economic profit and performance measurement in banking", New England Economic Review, July-August, Boston, p. 35-53. [1997]: "Innovations in performance measurement in banking", New England Economic Review, Mau-June, Boston, p. 23-38. LUCIEN, K [1979]: "Transfer pricing for the cost of funds in a commercial bank". Management Accounting [US], vol.60, no.7, January, p.23-36. ORMAECHEVARRIA, J [1994]: "La contabilidad de gestin en las entidades de crdito", Papeles de Economa Espaola, no.58, p.216-223. PEDRAJA, P. [1993]: Contabilidad y anlisis de balances en la banca, Centro de Formacin, Banco de Espaa, Madrid. PEDRET, R. [1997]: Direccin Comercial III, Universitat Oberta de Catalunya. ROUACH, M., NAULLEAU, G. [1994]: Le contrle de gestion bancaire et financier, La Revue Banque Editeur, Paris. ROUT, R., KOCHVAR, M. [1994]: "Transfer pricing: a poor man's approach". Journal of Bank Cost and Management Accounting, vol.7, no.3, p.25-30. SARTORIS, W. [1971]: "Transfer pricing in a commercial bank", Management Accounting [US], vol.52, no.8, February, p.33-39. SOVIGNET, E. [1989]: "Contrle de gestion: comment approcher la notion de refinanacement pour determiner le produit net bancaire", La Revue Banque, no.496, July-August, p. 718-724. TEWES, J.A. [1976]: "Valuing bank funds for allocation and pricing decisions", Management Accounting [US], November, vol.58, no.5, p.27-33.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14]
[15] [16] [17] [18] [19] [20] [21] [22] [23]
199 [24] [25] [26]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) VILLALONGA, G.; PASSERA, C. [1985]: "Medicin de rentabilidad por oficinas, por productos y clientes" en El reto de la banca ante el nuevo entorno competitivo, Instituto de Empresa, Madrid. p. 133-142. WADEN-BERGHE, L. [1990]: "Contabilidad de gestin de la entidad de crdito", Tcnica contable, no.504, p. 569-585. WONNACOTT, T.H., WONNACOTT, R.J. [1981]: Fundamentos de estadstica para administracin y economa. Ed. Limusa. Mexico.
Methods of Researching Strategy Processes in Developing Countries

Abdallah M Elamin Assistant Professor of Management and Strategy Ajman University of Science and Technology, Po Box 346, Ajman, UAE Abstract This paper attempts to shed light on the potency of intermarriage between grounded theory and case studies for researching strategy processes in developing countries. To research strategy processes in developing contexts, the author of this paper argued for using grounded theory and case study methods based on a phenomenological rather than a positivist approach. The ground for such a combination is fully justified and the ways for its practical implementation is elucidated. The paper suggests some practical solutions for the difficulties that are likely to surface during negotiation of access and data collection in developing countries organisations. Moreover, guidelines were suggested for performing analysis using such a potent combination of case study and grounded approach. Keywords: Grounded approach, case study, strategy processes, developing countries
Objectives of the Research

The objectives of this paper are twofold. The first is to justify the potency of the intermarriage between case study and grounded theory methods for researching strategy processes in developing countries. The second objective is to describe how such methods should be implemented empirically. The first two parts of the paper are devoted to achieving the first objective, whilst the rest of the paper is concerned with proposing outlines for empirical implementation of such methods. The paper ends with concluding remarks.
On Research Approaches in Social Sciences

Research approach refers to a philosophical stance, which underpins the methods followed in research. An ongoing debate within the sphere of social science about the most appropriate philosophical stances has identified two approaches: positivism and phenomenology. Positivism primarily claims that reality is external and objective, and knowledge is only of significance if it is based on observations of this external reality (Easterby-Smith et al., 1991). Similarly it holds that facts and values are distinct and scientific knowledge consists only of facts (Archer, 1988). Accordingly, scientific knowledge is utterly objective and only scientific knowledge is valid, certain, and accurate (Crotty, 1998). To achieve such scientific knowledge, this approach, according to Lee (1991), involves the manipulation of theoretical propositions using the rules of formal logic and the rules of hypothetico-deductive logic, so that the theoretical propositions satisfy the four requirements of falsifiability, logical consistency, relative explanatory power, and survival. EasterbySmith et al. (1991: 23) develop eight features associated with the positivist viewpoint, comprising:
201
1. Independence: the observer is independent of what is being observed. 2. Value-freedom: the choice of what to study and how to study it can be determined by objective criteria rather than by human beliefs and interests. 3. Causality: the aim of social sciences should be to identify causal explanations and fundamental laws that explain regularities in human social behaviour. 4. Hypothetico-deductive: science proceeds through a process of hypothesising fundamental laws and then deducing what kinds of observations will demonstrate the truth or falsify of these hypotheses. 5. Operationalisation: concepts need to be operationalised in a way which enables facts to be measured quantitatively. 6. Reductionism: problems as a whole are better understood if they are reduced into the simplest possible elements. 7. Generalisation: in order to be able to generalise about regularities in human and social behaviour it is necessary to select samples of sufficient size. 8. Cross-sectional analysis: such regularities can most easily be identified by making comparisons of variations across samples. In a similar vein, Robson (1993: 18-19) identified the following five sequential steps for a research based on the positivist tradition: 1. Deducing a hypothesis (a testable proposition about the relationship between two or more events or concepts) from the theory. 2. Expressing the hypothesis in operational terms (i.e. ones indicating exactly how the variables are to be measured) which propose a relationship between two specific variables. 3. Testing this operational hypothesis. This will involve an experiment or some other form of empirical enquiry. 4. Examining the specific outcome of the enquiry. It will either tend to confirm the theory or indicate the need for its modification. 5. If necessary, modifying the theory in the light of the findings. An attempt is then made to verify the revised theory by going back to the first step and repeating the whole cycle. Phenomenology or a social constructionism or interpretivism approach holds that reality is socially constructed and subjective in nature. Such a stance aims to understand phenomena from the point of view of participants directly involved with the phenomenon understudy (Cavaye, 1996) or from the actors own frame of reference (Bogdan and Taylor, 1975). The focus of such approach, therefore, is on appreciation of different constructions and meanings that participants attach to their experience (Easterby-Smith et al., 1991). The fact that such participants interpretations are subjective suggests that these interpretations will be shaped by the participants experience in their particular contexts. This stance, therefore, favours research forms that enable the distinctive character of such contexts to be studied (Nandhakumar and Jones, 1997). Researchers working with the phenomenology approach are more likely to work with qualitative data, investigating small samples in depth or over time, and using multiple methods to establish different views of phenomena (Easterby-Smith et al., 1991). Similarly, Locke has specified the tasks of researchers following this type of research tradition, arguing that: Researchers working in this paradigm focus on particular situated actors who they construe as composing meaning out of events and phenomena through prolonged processes of interaction that involve history, language and action (2001:9). Walsahm (1995) has identified two different roles that can be assumed by researchers working with this tradition, namely that of the outside observer and that of the involved researcher, through participant observation or action research. Each one of these roles has its merits and demerits, which need to be carefully assessed before a researcher makes any choice about which role to assume (Walsham, 1995: 77-78). The most important point from this perspective, however, is neither of these roles should be viewed as that of an objective reporter, since the collection and analysis of data involves the researchers own subjectivity.
202
Table 1 summarizes the distinguishing features of positivism and phenomenology, considering the basic beliefs underpinning each approach, the role played by the researcher, and preferred methods of analysis associated with each approach.
Table 1: Key Features of Positivist and Phenomenological Paradigms
Positivist Paradigm The world is external and objective Observer is independent Science is value free Focus on facts Look for causality and fundamental laws Reduce phenomena to simplest elements Formulate hypotheses and then test them Operationalising concepts so that they can be measured Taking large samples Phenomenological Paradigm The world is socially constructed and subjective Observer is part of what observed Science is driven by human interest Focus on meaning Try to understand what is happening Look at the totality of each situation Develop idea through induction from data Using multiple methods to establish different views of phenomena Small samples investigated in depth or over time
Basic beliefs:
Researcher should:
Preferred methods include:
Source: Adopted from Easterby-Smith et al., (1991:27).
It is noteworthy that selection of a particular approach depends on the research objectives, research questions and the interest of the researcher (Easterby-Smith et al., 1991; Morgan and Smircich, 1980). Due the scarcity of strategy processes research in developing world the author of the present paper judges that at the present stage the exploratory research, based on phenomenological as opposed to the positivist paradigm, would be the appropriate approach to follow in studying strategy processes in such particular contexts.
On Research Methods
Following a phenomenological approach, this author argues for intermingling case study and grounded research methods for researching strategy processes in developing countries. The grounds for this choice are elucidated below. Why Case study? Technically a case study can be defined as an empirical inquiry that: investigates a contemporary phenomenon within its real-life context; when the boundaries between phenomenon and context are not clearly evident; and in which multiple sources of evidence are used (Yin, 1989:23). Another relevant definition of a case study is as follows: Case study research consists of detailed investigation, often with data collected over a period of time, of one or more organizations, or groups within organizations with a view to providing an analysis of the context and processes in the phenomenon under study (Hartley, 1994: 208-209). Both definitions indicate that a case study method is an appropriate investigative framework for understanding a phenomenon alongside its context. There are three types of case study that have been identified in the literature on research methods (Locke, 2001). The first type, the intrinsic case study, focuses on the uniqueness of a particular case; researchers are interested simply in understanding the case itself for its particularity and uniqueness. The second type, the instrumental case, is where researchers are interested in a particular case because of the potential it has to offer providing insight into a substantive issue or to advance a theory. The third type is collective cases, when the instrumental case extends to more than one case. This typology is embedded in Yins (1989) approach, which indicates that case studies can involve either single or multiple cases, and numerous levels of analysis. If there are multiple cases, the logic is of replication rather than sampling. On the flexibility of case
203
study method, Eisenhardt (1989) opines that case studies can be used to accomplish various aims: to provide description, test theory or generate theory. A common criticism of the case study method, however, is that it provides little basis for scientific generalisation. This researcher has much sympathy with Yins (1989:21) position that: case studies, like experiments, are generalisable to theoretical propositions and not to populations or universes. Yin argues that the case study, because it does not represent a sample, should be regarded in the same way as the experiment. With the case study method, the investigator aims to expand and generalise theories, analytic generalisation, and not to enumerate frequencies, statistical generalisation (Yin, 1989). In that sense, the detailed knowledge about processes underlying the behaviour and its context can help to specify the conditions under which the behaviour can be expected to occur (Hartley, 1994). Case study is widely used in the field of management and organisational studies. Within strategic management the case study has been particularly relevant to understanding phenomena such as strategy formation (e.g. Chandler, 1962; Allison, 1971; Quinn, 1980; Eisenhardt and Bourgeois, 1992; Qi, 2000, Elamin, 2001). Aharoni (1993) cogently argued for the usefulness of a case study method for studying the complexity and sophistication of strategy processes compared to cross sectional and survey methods. There are three major reasons why the case study method is an appealing option for researching strategy processes in developing contexts. Firstly, the strategic management field is poorly researched in the developing contexts. In this respect, most of the studies to be done in developing countries are the first of its kind to explore strategy processes in those particular contexts. These studies, therefore, are basically exploratory attempts. The intention, thereof, is to describe how strategy gets formed by organizations in developing countries and to explain why it is developed in that particular way. Yin (1989) argues that when the research objectives revolve around how and why with regard to a contemporary set of events, over which investigator has little or no control, the case study will be the most appropriate investigative framework. In a similar vein, Hartley (1994) argues that case studies are tailor-made for exploring new processes or behaviours, or those, which are little understood. Secondly, to understand a phenomenon such as strategy which was not previously probed within the developing contexts, the researcher should identify the people who were involved in its making since they are the best who can tell about it. This is true, for as Rosen (1991) puts it clearly to understand social process one must get inside the world of those generating it. The simplest way to do this would seem to be to get access to the actors themselves and to elicit their interpretations directly (Johnson, 1987; Nandhakumar and Jones, 1997). This could satisfactorily be done only through indepth interviewing of these actors. Therefore, it has been judged that the case study would be the most appropriate strategy for understanding the complexities surrounding strategy processes through an indepth interviewing of those who have involved in its making. Thirdly, it is well documented in the strategy literature, especially within the processual and systemic approaches to strategy, that strategic issues are inherently contextual (Pettigrew, 1987, 1990; Whittington 1993). This argument increases the doubts about the appropriateness of the rational model of strategy for describing and explaining strategy development processes within the developing contexts organizations because of its inherent inadequacy resulting from excluding context. For a sound understanding of strategy processes in developing countries organizations, both inner (organizational) and outer context (economical, political and socio-cultural) should be considered. Hartley (1994) cogently comments that case studies are useful where it is important to understand social processes in their organisational and environmental contexts. In this respect, case study could be considered as a suitable framework that enables inclusion of contexts. Why Grounded theory? The grounded theory approach has been effectively used in strategic management field (see Burgelman, 1985; Sharivastava and Grant, 1985; Rayman-Bacchus, 1996; Nutt, 1998, Elamin, 2001).
204
The author of this article argues for using grounded approach jointly with the case study method. The reason for following the grounded theory method is twofold. Firstly, it is an inductive, theory discovery methodology that allows the researcher to develop a theoretical account of the general features of a topic while simultaneously grounding the account in empirical observations or data (Martin and Turner, 1986:141). Similarly, its value as described by Strauss and Corbin lies in its ability not only to generate theory but also to ground that theory in data (1998:8). Given the fact most of strategy research in developing countries is exploratory and in most instances represents the first of its kinds to explore strategy, the grounded approach would allow much flexibility and openness in data analysis. Secondly, grounded theory allows the inclusion and investigation of the context, process and consequences and their mutual relationships. Given the uniqueness of the developing contexts, both inner and outer, this author believes that the grounded theory together with the case study method would allow the investigation of the strategy processes within its own context and would help in developing some theoretical concepts (Eisenhardt, 1989). Eisenhardt, in her ever-fascinating paper Building Theories from Case Study Research published in Academy of Management Review (1989) could be considered as a pioneer in suggesting systematically how this combination would work. According to Locke (2001:100-101) such a combination could be considered as an adaptation of the grounded theory in which researchers selectively integrate the logic and practices of other qualitative research styles with those of grounded theory. Within the subject of strategy formation, Burgelmans (1985) study on internal corporate venturing of one high technology firm, Rayman-Bacchus (1996) study on strategy formation in three Scottish organizations, and Elamins (2001) study of strategy development processes in Sudanese enterprises are examples to the point.
An empirical implementation of Case study and Grounded approach

This section describes the empirical demonstration of utilising the case study and the grounded theory methods in data collection and analysis for studying strategy processes in developing countries organisations. The themes covered are negotiation of access, selection of sites, selection of issues, data collections and data analysis. Negotiation of access Negotiations for access should start as early as possible, and the researcher should contact as many organisations as possible in order to secure a reasonable room for a further selection. Hartley (1994) provided two useful devices for gaining access. The first, researcher has to be introduced through a third party and the second, researcher has to conduct a broad-ranging interview with the senior managers in organisations that he/she thinks may be suitable. Following formal means (using the supporting letter, introducing himself to the reception officer and asking for an appointment to see the general manager) is not a fruitful option to gain access in developing contexts. Following such means may waste a researchers limited time. Relying on friends, relatives, and other social and personal connections may help in securing reasonable access in developing contexts (Elamin, 2001). Selection of sites Three criteria should be used for selection of the sites for conducting the in-depth investigation of strategy processes in organisations within developing countries. These include the suitability and relevance of organisations for observing strategy processes, the quality of access achieved during preliminary visits by the researcher to research sites, and resources available to the researcher in terms of money and time. At this stage, the researcher may judge that some orgaisations are inappropriate for conducting any further investigations due to unpromising quality of access, lack of records and documentary evidence, unenthusiastic attitudes, and hesitancy on the part of research subjects. Considering the three criteria identified above and the difficulties the researcher may encounter in
205
selection of relevant sites for studying strategy in developing countries, the following points seem to be useful: First, as most of the strategy studies conducted in developing countries are exploratory and the first of their kind, the intention of researchers, therefore, is not to test hypotheses or make generalisations in a statistical sense; rather the intent is to understand strategy processes within its context and to make a theoretical generalisation. Considering this, the number of organisations is not of crucial importance. Moreover, if researchers want to understand the phenomenon in-depth, they need not study a large number of cases (Gummesson, 1991). Therefore, the selection is based on the theoretical sampling ground where cases are chosen for theoretical, not statistical reasons (Glaser and Strauss 1967; Eisenhardt 1989, Strauss and Corbin, 1998, Locke 2001). The important point of theoretical sampling is theoretical relevance. With respect to relevance, the selection process ensures that the substantive area addressed by the study here the strategy processes is kept similar or as Eisenhardt put it, is likely to replicate or extend the emergent theory. Locke 2001: 55) interestingly comments on the rationale for theoretical sampling as to direct all data gathering efforts towards gathering information that will best support development of the theoretical framework. Secondly, the selected organizations should be similar in some respects and different in others. The differences will help to maximise opportunities to discover variations among concepts and to densify categories in terms of their properties and dimensions (Strauss and Corbin, 1998: 201). The similarity in the line of inquiry across organisations and differences with respect to the content of issues, organisational activities and variables are sought to be useful in extending the emergent theoretical categories and enabling across-case analysis. Thirdly, deciding to focus on a small number of organizations sometimes is triggered by more pragmatic considerations, explicitly time and financial constraints imposed on the researcher. Selection of issues A wide spectrum of research on strategy has justified a selection of strategic instances for studying strategy processes. Cray et al. (1991) argue that one track for studying strategy processes is to analyse a decision or a series of decisions in a single organisation. The works of Allison, 1971; Pettigrew 1973, 1985a; and Mintzberg et al., 1976 contain some examples to the point. Such an approach was also used effectively by the London Business Schools SID (strategic investment decisions) research during the 1980s. Yamamoto interestingly comments on the LBSs researchers approach by stating that: They studied SIDs based on the recognition that in order to trace strategy process it is more effective to describe and analyze the process in which particular SIDs are made than to assume that strategy is rationally formed and implemented to achieve given objectives (1998: 147). The following are some practical suggestions for identification of the strategic issues: firstly, the researcher contacts the general manger/managing director of the selected organization and the present to him/her the purpose of the study. Secondly, the contacted persons will be asked to nominate issues that had strategic importance to the enterprise and that had recently completed. The importance is broadly defined in terms of novelty (non-routine), complexity, demand for a commitment of substantial amount of resources, precedent for subsequent decisions, and organisation-wide in its consequences (Mintzberg et al., 1976; Hickson et al., 1986; Lu, 1998). Recently completed issues are to be selected because the researcher has to reduce as much as possible the errors of distortion and memory failure of actors when he comes to the phase of data collection (Mintzberg et al., 1976; Papadakis et al., 1998). Although allowing the contacted persons to identify issues ensure both interest and first-hand knowledge (Nutt, 1998), is also not free from pitfalls. Then, the contacted person will be asked to identify the main actors associated with each selected issue (Hafsi and Hafsi, 1989; Nutt, 1998).
206
Data Collection
Yin (1989) argues that evidence for case studies may come from six sources: documents, archival records, interviews, direct observation, participants observation and physical artefacts. Due to the unique nature of developing contexts organizations, the researcher may find it appropriate to assume the role of the outside observer. For the outside observer the prime technique that is likely to be used in collecting primary data about strategy processes is in-depth interviewing of people involved in these processes. Nandhakumar and Jones (1997) argue that interviews can enable sensitive exploration of actors interpretations at a smaller number of sites. Moreover, interview is considered as one of the most effective techniques for studying strategy processes. Mintzberg et al. cogently argue for the usefulness of interviewing as a data collection technique for researching strategy processes: Observation is certainly a powerful and reliable method, but extremely demanding of research resources because strategic decision processes typically span periods of years; often forced to study the process after completion, therefore the researcher is obliged to rely heavily on interviewing. The best trace of the completed process remains in the minds of those people who carried it out (1976:248). Similarly, Schwenk (1995) outlines the difficulty of direct observation for studying decisionmaking, and argued for in-depth interviewing of decision-makers to reconstruct a retrospective account of the decisions they made. Schwenk argues that: Research on strategic decision making is difficult to do because of the difficulty of observing the decision process in action. Because of this fact researchers have used a variety of creative approaches to collecting data in the absence of direct observation. Mintzberg (1973) and Hickson et al. (1986) for example, collected decision-makers retrospective accounts of decisions they had made (1995:486). There are two commonly-used types of interviews in research that follow a phenomenological tradition: the unstructured and semi-structured interview. According to Bryman (2001), in an unstructured interview, the researcher uses an aide memoire as a brief set of prompts to help deal with a certain range of topics. There may be just a single question that the interviewer asks and the interviewee is then allowed to respond freely, with the interviewer simply responding to points that seem worthy of being followed up. On the other hand, within a semi-structured interview the researcher has a list of questions or fairly specific topics to be covered, often referred to as an interview guide, but the interviewee has a great deal of leeway in how to reply. Questions may not follow exactly as outlined on the schedule. The researcher should prepare interview guides before each interview. Each interviews questions should be based on feedback from previous interviews. The focus of the interviews should be on the conditions that stimulate the identified strategic issue, actions or interactions of the key players, and the consequences of these responses. The emphasis of the researchers intention should be directed toward discovering the processes through which the identified strategic issues were carried out. During the interview, the researcher should maintain a balance between excessive passivity and over-direction. Finally, the data obtained from interviews should be cross-checked with interviewees against each other, and against the documentary evidence, i.e., triangulation. Interviews should be conducted in a language easily spoken and understood by interviewees. Interviews can also be tape-recorded with prior consent of the interviewees. If an interviewee refuses tape-recording, the researcher must rely on note-taking. There are several advantages and disadvantages of these two types of reporting media (see Walsham 1995: 78). Alongside the primary data, considerable materials might be collected about the issues, organizations and the broader context of the study. The main source of these materials is the organizations files and official reports.
207
Difficulties on access and data collection in developing countries

On conducting research in developing countries organisations, the researcher may face three critical difficulties. First, difficulty of maintaining access to the selected organisations. Researchers might be forced to re-negotiate access every time they visit the study sites because those who initially granted access may unpredictably change their minds and disallow them. It is worth noting that supporting official letters and documents that explain the purpose of research are, alone, not enough for granting access to developing countries organizations. Two factors remain very crucial for securing full access to the research sites and interviewees minds in developing countries. These are personal connections, and the political influence of the gatekeeper or the mediator between the interviewer and the top people of an organization. Secondly, lack of proper documentation and records. Such a problem is very serious because of a lack of proper documentation and records make cross-checking, and the avoidance of post rationalisation a difficult task. Lack of proper documentation and records might be attributed to the predominance of verbal communication and personal contacts over formal communication in developing countries organizations. Since top person(s) deal with every matter (including making decisions) in a verbal way, documentation seems unnecessary. Thirdly, difficulties resulted from reluctance and negative attitudes towards time. Reluctance is a commonly-observed trend with interviewees in developing countries organizations specifically when the focus of the study, are strategic; their sensitivity is such that no one is willing to confide in research about them. The researcher may need a long time to build trust in order to access interviewees minds. Nandhakumar and Jones (1997) comment that relations between field researchers and their subjects rest on a base of trust which in turn rests on a base of liking. The researcher needs to have strong background about the socio-cultural set up of the contexts of the study. In his study of strategy making processes within the Sudanese context, Elamin states that: Besides use of friends as mediators, I have resorted to, based on my own knowledge about the Sudanese culture, several styles that proved very effective for building trust. For instance, Sudanese society is highly tribalistic in nature. Prior to conducting an interview with a particular person I often identify his tribe and its basic characteristics from my own sources (friends, relatives, etc). This information is very helpful for building trust with my respondents. For instance, mentioning good deeds and characteristics of the respondents tribe (e.g. generosity, boldness, etc), during my interviews was very effective in creating a conducive atmosphere for an interview. Another example is the identification of the political orientation of the respondent before the interview. I often use this dimension very carefully, because if the respondent discovered that I was against his orientation, the trust might be very difficult to build. (2001:95) The other common difficulty is the negative attitudes of managers and officials in developing countries about time. Time is not an important factor; researchers may face serious disturbances in their schedule of interviews. Missing or postponing appointments are common practice in developing countries organisations.
Data Analysis
This paper argues for using grounded theory analytical procedures (Strauss and Corbin, 1998) supported by the within case and across case analysis recommended by Eisenhardt (1989) and replication logic suggested by Yin (1989). Strauss and Corbin offer a comprehensive set of procedures and techniques for qualitative data analysis. In researching strategy processes in developing countries the researcher can employ some of Strauss and Corbins procedures and techniques to analyze the data collected with respect to each individual case. Using across case analysis (to detect similarities) is the next step that the researcher has to perform. Later replication logic could also be used to capitalise on similarities across cases.
208
According to Yin (1989), replication logic helps to treat a series of cases as a series of experiments with each case serving to confirm or disprove the emerging hypotheses and relationships. Eisenhardt (1989) argues that cases, which confirm emergent relationships, enhance confidence in the validity of such relationships. To implement the above-prescribed procedures the researcher is advised to divide the data analysis stage into three main phases. Phase one includes description and generation of propositions; phase two involves interpretation of the generated propositions; and phase three deals with elaboration and reflections. In the following each phase is illuminated. Phase one consists of three stages. In the first stage each of the selected cases will be written up from the interviewee perspective, using a narrative approach describing what is going on, what the setting looks like, and what the people involved are doing. This produces a story about strategy processes within each of the identified sites. Such a data reduction and ordering strategy is strongly recommended to help a researcher to cope with a deluge of data. In Eisenhardts (1989) terms, this stage is equivalent to the within case analysis. In so doing, the researcher will get familiarised with empirical data and particularly with each case as a stand-alone entity (Easterby-Smith et. al., 1991; Eisenhardt, 1989). The second stage is labelled as conceptualisation. Having produced these case narratives, the researcher will utilize the grounded theory analytical procedures to assign meaning to these accounts. A number of researchers found focusing on case narratives appealing rather than on the raw data (see Gersick, 1988; Hartley, 1994; and Brown and Eisenhardt, 1997). The conceptualisation is to be used by the researcher to convert data described within stage one into abstract concepts, categories, and subcategories, with the intention of relating them to each other. According to Strauss and Corbin, this stage involves organisation of data into discrete categories (and some time ratings) according to their properties and dimensions, and then using descriptions to elucidate those categories (1998:19). In so doing the researcher may use the open coding process. The open coding process entails breaking down data into discrete parts, then examining and comparing it for similarities and differences. It also includes constructing categories by grouping events, happenings, objects and actions or interactions that are found to be conceptually similar in nature and meaning. Grouping concepts into categories helps to reduce data, offers an explanation, and allows interpretation and prediction. Open coding can be performed through line by line analysis of the case narratives. Despite being the most time consuming, line by line analysis of data proved to be the most generative form of coding. Appendix A shows how these processes are conducted. Having identified categories, the researcher should now move to the third stage within this phase, which entails two consecutive activities. The first is writing-up a descriptive account of the cases based on the categories generated in the previous stage. The second activity is to use these descriptive accounts to generate propositions about each case as a stand-alone entity. Having completed this stage, the researcher will come up with propositions which could be understood in Van Mannens (1979) and Stiles (2001) terms as firstorder themes about strategy processes within the context studied. Phase two is an interpretation of the generated propositions, which entails two stages. The first one is an interpretation of the generated propositions and generation of general characteristics about strategy processes within each case as a stand-alone entity. The second stage is performing across case analysis for detecting common patterns from the cases studied (Eisenhardt, 1989). In this stage, the cases are treated by consideration of replication logic (Yin, 1989), as each case confirms the emergent findings within the others. The emphasis should be on across case similarities. Phase three, elaboration and reflection involves two sub-stages. The first one is elaboration within which each of the generated patterns is explained from the cases. At this stage the researcher may find that the generated patterns fit the evidence from each case very well. In Eisenhardts (1989: 541) terms, such a close fit is important to building good theory. The second stage is reflection, in which each of the generated patterns is compared with the existing theories of strategy processes. This stage reveals how the findings of the study are both similar and different to the findings of others and endeavours to find interpretation for the similarities and contradictions. Eisenhardt (1989) identified
209
three distinct uses of theory as an initial guide to design and data collection, as part of an iterative process of data collection and analysis, and as a final product of the research.
Conclusion
This research reviewed the philosophical stances which underpin research methods in the social sciences, namely positivism and phenomenology. It indicates that researchers working with the phenomenology approach are more likely to work with Qualitative data, whereas those who work with positivist use Quantitative techniques. The research argues that the case study and grounded approach are very compatible option with the phenomenology approach with respect to researching strategy processes in developing countries. Such compatibility is fully justified and the ways for its practical implementation are demonstrated. Moreover, the research pointed out the likely difficulties that may surface during the negotiation for access and data collections and suggested some practical solutions. Finally, the paper shows how the analysis could be performed following the case study and grounded approach.
210
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Aharoni, Y. (1993) In search for the unique: can firm-specific advantage be evaluated? Journal of Management Studies 30(1): 31-49. Allison, G. T. (1971) Essence of decision: explaining the Cuban Missile crisis, Boston, MA: Little, Brown and Company. Archer, S. (1988) Qualitative research and the epistemological problems of the management disciplines, in A. Pettigrew, eds., Competitiveness and the Management Process, pp.265-302, Basil Blackwell, Oxford. Bogdan, R., and Taylor, S. (1975) Introduction to qualitative research methods: a phenomenological approach to social science, New York: Wiley. Brown, S. L. and Eisenhardt, K. M. (1997) The art of continuous change: linking complexity theory and time-paced evolution in relentlessly shifting organisations, Administrative Science Quarterly 42: 1-34. Bryman, A. (2001) Social research methods, Oxford University Press. Burgelman, R. A. (1985) Applying the methodology of grounded theorising in strategic management: a summary of recent findings and their implications, Advances in Strategic Management 3: 83-99. Cavaye, A.L.M. (1996) Case study research: a multi-faceted research approach for IS, Information Systems Journal 6: 227-242. Chandler, A. D. (1962) Strategy and Structure: chapters in the history of the industrial enterprise, Cambridge, Mass.: MIT Press. Cray, G. R., G. R. Mallory, R. J. Butler, Hickson, D. J., and D. C. Wilson (1988) Sporadic, fluid and constricted processes: three types of strategic decision making in organisations Journal of Management Studies 25(1): 13-39. Crotty, M. (1998) The foundations of social research: meaning and perspective in research process, Sage Publications Ltd. Easterby-Smith, M., R. Thorpe and A. Lowe (1991) Management research: an introduction, London: Sage. Eisenhardt, K. M. (1989) Building theories from case study research, Academy of Management Review 14: 532-550. Eisenhardt, K. M. and Bourgeois, L. J. (1992) Conflict and strategic decision making: how top management team disagree, working paper, Stanford University. Elamin, A. M. (2001) Strategy development processes in selected Sudanese enterprises, Unpublished Ph.D. thesis, Lancaster University. Gersick,C. J. (1988) Time and transition in work teams: toward a new model of group development, Academy of Management Journal 31: 9-41. Glaser, B. G. and Strauss, A. L. (1967) The discovery of grounded theory, Chicago: Aldine. Gummesson, E. (1991) Qualitative methods in management research, 2nd edition, Newbury Park, CA: Sage. Hafsi, T. and Hafsi, C. (1989) Business behaviour of the Algerian state-controlled enterprises: a decision making process approach, in T. Hafsi, eds., Strategic issues in state-controlled enterprises, JAI Press Inc., London. Hartley, J. F. (1994) Case studies in organisational research in Catherine Cassel and Gillian Symon, eds., Qualitative methods in organisational research, Sage Publications Ltd., London. Hickson, D. J., R. J. Butler, D. Cray, G. R. Mallory and D. C. Wilson (1986) Top decisions: strategic decision-making in organisation, Oxford: Basil Blakwell. San Francisco: Jossey-Bass. Johnson, G. (1987) Strategic change and the management process, Basil Blackwell Ltd. Lee, A.S. (1991) Integrating positivist and interpretivist approaches to organisational research, Organisation Science 2: 342-365. Locke, K. (2001) Grounded theory in management research, Sage Publications, London.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23] [24]
211 [25] [26] [27] [28] [29] [30] [31] [32] [33] [34] [35] [36] [37] [38] [39] [40] [41] [42] [43] [44] [45] [46] [47] [48] [49] [50]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Lu, Y. (1998) Strategic investment decisions in China in V. Papadakis and P. Barwise, eds., Strategic Decisions, Kluwer Academic Publishers, the Netherlands. Martin, P. Y. and Turner, B. A. (1986) Grounded theory and organisational research, The Journal of Applied Behavioural Science 22: 141-157. Mintzberg, H. (1973) The nature of managerial work, New York: Harper and Row. Mintzberg, H., D. Raisinghani and A. Theoret (1976) The structure of the unstructured decision processes Administrative Science Quarterly 21: 246-275. Morgan, G. and Smircich, L. (1980) The case for qualitative research Academy of Management review 5: 491-500. Nandhakumar, J. and Jones, M. (1997) Too close for comfort? distance and engagement in interpretative information systems research, Information Systems Journal 7: 109-131. Nutt, P. C. (1998) Framing strategic decisions, Organisation Science 9(2): 195-216. Papadakis, V., S. Lioukas, and D. Chambers (1998) Strategic decision-making processes: the role of management and context, Strategic Management Journal 19: 115-147. Pettigrew, A.M. (1973) The politics of organisational decision making, Tavistock, London. Pettigrew, A.M. (1985a) The awakening giant: continuity and change in Imperial Chemical Industries, Oxford, Basil Blackwell. Pettigrew, A.M. (1987) Introduction: researching strategic change, in Andrew M. Pettigrew, eds., The Management of Strategic Change, Basil Blackwell, Oxford. Pettigrew, A.M. (1990) Studying strategic choice and strategic change, Organisation Studies 11(1): 6-11. Qi, H. (2000) Exploring the strategic development of Chinese township and village enterprises, Unpublished Ph.D. thesis, Lancaster University. Quinn, J. B. (1980) Strategies for Change: logical incrementalism, Homewood, IL: Irwin. Rayman-Bacchus, L. (1996) The practice of strategy, Unpublished Ph.D. thesis, Edinburgh University. Robson, C.(1993) Real world research, Blackwell, Oxford. Rosen, M. (1991) Coming to terms with the field: understanding and doing organisational ethnography, Journal of Management Studies 28(1): 1-24. Schwenk, C. R. (1995) Strategic decision making, Journal of Management 21(3): 471-493. Sharivatava, P. and Grant, J. H. (1985) Empirically Derived Models of strategic Decisionmaking Processes, Strategic Management Journal 6, 97-113. Stiles, P. (2001) The impact of the board on strategy: an empirical examination, Journal of Management Studies 38 (5): 627-650. Strauss, A. L. and Corbin, J. (1998) Basics of qualitative research: techniques and procedures for developing grounded theory, 2nd edition, Thousand Oaks, CA: Sage. Van Mannen, J. (1979) The fact of fiction in organisational ethnography, Administrative Science Quarterly 24: 539-550. Walsham, G. (1995) Interpretative case studies in IS research: nature and method, European Journal of Information Systems 4: 74- 81. Whittington, R. (1993) What is strategy and does it matter?, Routledge, London. Yamamoto, M. (1998) Strategic decisions and corporate governance in Japan, in V. Papadakis and P. Barwise, eds., Strategic Decisions, Kluwer Academic Publishers, the Netherlands. Yin, R. K. (1989) Case study research: design and methods, revised edition, Sage publications, Inc.
212
Appendix (A)
Demonstration of the steps to develop a category using the grounded theory procedures.
CONCEPTS
TEXT
SUBCATEGORIES Resource check (Gum buffer stock)
Change of roles Buffer stock Activation Slack
When the present general manager took over the management of the company in January 1998 (change of roles). He found that the company possessed huge buffer stocks inside the Sudan. The value of this stock was approximately equivalent to around $50 or $60 millions (buffer stock). To make use of these stocks (activation). The GAC is rich with its huge stocks of gum (slack). Internal contacts Refusal Gum trade monopoly Government hands Insecurity Lack of financing value
Search within an immediate environment (internal contacts)
CATEGORY
SEARCH Search within a distant environment
The GM contacted local banks for finance with gum as deposit (internal contact). Refusal of local banks for accepting gum (refusal) resulted from state of monopoly (monopoly) of international gum trade. Refusal resulted from government association with the gum trade (government hands). Dealing with GAC is insecure avenue for local banks (insecurity). Strategic stock of Gum has no financing value inside Sudan (lack of financing value).
The Enterprise Risk Management Model for Corporate Sustainability and Selection of the Best ERM Operator in the Turkish Automotive Distributor Company: ANP Based Approach
Ayse Kucuk Yilmaz Anadolu University, School of Civil Aviation 2 Eylul Campus, 26470, Eskisehir, Turkey Tel: +90 535 462 55 90 E-mail: akucukyilmaz@gmail.com Abstract The corporate sustainability (CS) subject is important challenge for managers in the todays risky, competitive and complex business environment. The main objective of this paper is present new Enterprise Risk Management (ERM) model for the CS. The new ERM model for CS is offered to be a suitable managerial approach for the Turkish automotive distributor companies. The new ERM model is introduced as a useable way for implementation of corporate sustainability management (CSM) in order to meet financial, strategic operational, ecological and social goals of organizations. In order to meet the requirements of CS specific tools are necessary. The new ERM model is offered as the central tool for the development and implementation of sustainable business strategies for the Turkish automotive distributor companies. In this study, new model has been developed according to my new ERM framework model in doctorate dissertation (2007). This study is contributed to ERM literature and supported for CS efforts in the Turkish Automotive Distributor Companies. In order to developing of a new ERM model, interview with distributor managers is experienced and best ERM practice, CS surveys and ERM guidelines are analyzed. The study is also offered an application of the analytic network process (ANP) for the selection of best ERM operator in the Turkish automotive distributor companies. In order to identification of the best practice criteria to the ERM is realized to interview with high level managers of leading companies, analyzed best ERM practice surveys and ERM guidelines. ANP is used solving the problem. H.O. Sabanci Holding and Dogus Holding are determined as alternatives in the proposed ANP model. Keywords: Corporate sustainability, enterprise risk management model.
1. Introduction
Rapid dynamics of business environment due to globalization, regulations, re-engineering, changing market and competition structures increase the uncertainties, major source of risks for future decisions (Kucuk Yilmaz, 2008). Automotive sector, catalyst of the entire economy, has two primary concerns: ensuring strategic advantages and competitive superiority and obtaining maximum shareholder value in this volatile environment like any other business. Their survival strongly depends on their ability of
214
managing whole corporate risks. ERM is offered an efficient business framework for managing these risks. The study is offered to new ERM model for corporate sustainability and presents of best ERM practices in automotive distributors companies in the Turkey, a typical example of developing countries. Allocation and use of limited resources through an effective and systematic risk management effort are critical in Turkey. Determination of risk approaches and perceptions can provide a better understanding in establishing necessary infrastructure for ERM. The aim of this paper is firstly to define corporate sustainability, developing an ERM model for corporate sustainability and then to highlight the best ERM practice of Turkish automotive distributor companies. Corporate Sustainability management (CSM) is also part of the overall concept of sustainable development. Corporate Sustainability (CS) is examined within an ERM model. The model has been developed by author and it is integrated with sustainability model. Aim of the new model support to effective CS efforts to the companies since sustainability is major risk for companies in the highly volatile and uncertain business environment. Automotive sector is important to Turkish economies. It has been sustainable for national developing. This paper is offered to support CS efforts by a new ERM model to managing sustainability risks for the automotive distributor companies. Sustainability risks will be a catalyst for innovation and opportunities. The automotive industry is playing an important role in supporting these changes and developments in the economics of Turkey. The paper is organized into seven main sections and begins with The Importance of Enterprise Risk Management for Corporate Sustainability. The Corporate Sustainability Issues in the Automotive Sector is given section 3. The New Enterprise Risk Management Model for Corporate Sustainability is explained in section 4. Section 5 introduces an importance of Automotive Sector for the Turkish Economy. The Enterprise Risk Management Practice in the Turkish Automotive Sector: Sabanci Holding and Dogus Group, is given Section 6. Section 7 introduces to ANP model and its application to the selection of best ERM operator among the Turkish Automotive distributor companies that they have ERM systems and improved ERM practices. The overall conclusion is given last section. The ERM model for corporate sustainability (CS) is a new generation management framework, aimed to meet increased corporate complexity and support corporate transformation towards more sustainable ways of doing business. The new ERM model for corporate sustainability (CS) is offered as the important mechanism for improving CS management and CS performance. The new model may generate business value through measurement and management of sustainability risks and opportunities. New ERM model is based on an authors previous study (Kucuk Yilmaz, 2007). The model is offered to Turkish distributor companies in the automotive sector. The results indicate that the model makes it possible to improvement of the Corporate Sustainability management of the researched companies. The conclusion validates the proposed model as a tool for advance corporate sustainability through ERM model.
2. The Importance of Enterprise Risk Management for Corporate Sustainability

Sustainability continues to grow as both a challenge and opportunity for businesses and their investors. The aim of the study is to contribute to awareness of ERM as a corporate sustainability tool for the best practice. Sustainability issues can be a driver for competitive advantage, not only by lowering production costs as a result of more material and energy-efficient production processes, but also through corporate and product differentiation in the market. Business risks may be related to legal risks such as environmental or social liabilities, technical risks, social risks resulting from stakeholder reactions, natural risks or to general economic risks. A sustainability perspective helps managers identify such risks. Engaging with sustainability may also provide an attractive platform to position in media and among peers as well as to signal responsibility and to create trust among employees. If a company is
215
able to show that it manages business risks well it will get better access to (cheap) capital. Companies which deal with sustainability early will benefit from the experience and know-how they built up when demand patterns or regulations change. Approaching business processes and products from all perspectives on sustainability provides managers with a framework to change their focus. This in turn may help to identify innovation potential. An effort to establish corporate sustainability management can constitute a vision-based framework to coordinate activities to re-position a company and to make it fit for future. (ETH Zurich, SusTecGroup, 2005). Companies should concentrate on their risks and on their opportunities in the business environment. These are important to managing a business in an entrepreneurial and sustainable way; also require to remain competitive in the long term, it is a necessity and to lead the way in quality. Increasingly, businesses are being pressured to address environmental and social responsibility performance in addition to the traditional financial performance. The driving force of this trend is the increasing environmental and social risk costs that adversely impact the financial bottom line. If the firm's sustainability risks are not properly managed, its reputation, directors and officers, and financial viability are threatened (Anderson, 2005). Anderson's paper, "Sustainability Risk Management as a Critical Component of Enterprise Risk Management: Global Warming-Climate Change Risks," focuses on one of the most important and discussed corporate sustainability risk issues today. Anderson's paper shows how mitigation strategies can reduce these risks, and more importantly, how the incorporation of ERM-based corporate sustainability management strategies can produce substantial business opportunities. ERM-based corporate sustainability management (CSM) and related strategies may decrease managing of corporate sustainability risk costs, augment competitive positions, protect reputations and improve financial, environment and social lines. Chief Risk Officers (CRO) will need to anticipate corporate risks and develop best options to risk handling and financing strategies for them, but since many corporate sustainability risks are new and emerging, the best strategies for dealing with corporate sustainability might not be apparent. But in a world where the rules of business are changing faster than ever, now is not the time for ignorance (SciTec Today, 2008). New ERM model for corporate sustainability (CS) is important from distributor company perspectives because it is significant in determining the intangible and future value of the business; and it impacts on material risks and opportunities of the business. Both of these factors are essential for making better informed business management decisions. The offered new ERM model for corporate sustainability is considered to managing following issues: Human resource management, sustainability and risk culture and awareness Compliance with sustainable management issues Community and consumer support Good corporate governance practice Stakeholder interest and support Consideration and managing of financial, social and environmental risks CSM is affected to business in the national and international markets and corporate value as directly. In the CSM process, effective role of shareholders and importance of integration to the whole sectors are implies by specialist. Think globally and act locally is identified of todays strategy in the business management for CS. Sustainability in business is about ensuring long-term business success while contributing toward economic and social development, a healthy environment and a stable society. As part of their core principles, companies that are committed to sustainable business adopt high standards in areas that can include environmental protection, gender equity, working conditions, employee benefits, capacity development, community development and a set of transparent relationships between a companys management, its board, its shareholders and other stakeholders that fall under the term corporate governance. CS can be viewed as a new and evolving corporate management paradigm. The term paradigm is used deliberately, in that CS is an alternative to the traditional growth and profit-
216
maximization model. While CS recognizes that corporate growth and profitability are important, it also requires the corporation to pursue societal goals, specifically those relating to sustainable development environmental protection, social justice and equity, and economic development. Sustainable business presents opportunities for an improved economy, environment, and society. The goal of this research is to contribute to the ongoing effort of understanding CS and developing clear indicators for companies to use in their strategy development. Business managers need to understand better the dynamics of the global operating environment in order to manage its related risks effectively. Global companies face a new reality that has changed the nature of risk and ERM: networked operations and global value chains, empowered stakeholders, and the dynamic tension among sectors. The emergence of the new forms of social risk cannot be mitigated through traditional means. The new environment requires innovation by companies in both sensing and understanding these risks, and in adapting ERM systems to include new tools and network-based models of information sharing (Kytle and Ruggie, 2005). Holdings and their group companies are the fundamental cells of modern economic life and they have an important role in bringing about the changes needed for sustainable development. Parallel to the efforts to integrate the three dimensions economic, environmental and social- of sustainable development, businesses are accepting their role in sustainability. Corporate sustainability is a business approach that creates long term shareholder value by embracing opportunities and managing risks deriving from economic, environmental and social developments (Wilson, 2003). CS is a business approach that creates long term shareholder value by embracing opportunities and managing risks deriving from economic, social and environmental developments. For those organizations pursuing sustainability, sustainability is central to their corporate strategies and a vital ingredient in how they assess their effectiveness (Nemli, 2006). Companies that change their viewpoint about environmental and social pressures and integrate sustainability issues to their business strategy will have a competitive advantage over their rivals and accrue the benefits sustainability offers for them (Nemli, 2006).
3. The Corporate Sustainability Issues in the Automotive Sector

The most important underlying influencing factors on CS in the automotive industry are increased complexity and competition. The number of elements (e.g. number of stakeholders, number of markets) and the number of possible interactions between these elements is increasing dramatically (e.g. mutual dependence of sustainability issues, mobility concepts, energy and emissions) (Brunner, 2003). The automotive industrys approach to CSM is reactive rather than proactive. Stakeholders generally raise sustainability issues. They identify issues and create direct or indirect external pressure on the companies. Companies react to stakeholders demands by internal economic assessment and eventually take appropriate action. When no external pressure exists on a certain issue, companies comply with regulations. Where no regulations exist, companies act exclusively according to their economic interest. Taking action beyond compliance is only found when public pressure exceeds regulations or when regulations are on the horizon (Brunner, 2003). It focuses on the economic reasoning behind the major sustainability issues for the industry as well as on the progress of CSM. The Figure-1 provides the summary of results on the status of sustainable development in the automotive industry (Brunner, 2003).
4. The New Enterprise Risk Management Model for Corporate Sustainability

The new ERM model has been developed according to the corporate sustainability management context (see figure-2). Aims of corporate sustainability management are considered to determination of the steps of ERM model. Basically, the new model is offered to do best CSM practice. The model is provided a reasonable assurance for achieving of corporate sustainability objectives. The model is
217
basically based on: Kucuk Yilmaz, Ayse., Doctoral Dissertation, Anadolu University, Social Science Institute, Civil Aviation Management Branch, Eskisehir, Turkey, 2007. New ERM model identified by 32 steps, which is designed as the 6 key components and their sub components.
Figure 1: Results on the Status of Corporate Sustainability in the Automotive Industry (Brunner, 2003).
The ERM process consists of a series of steps that, when undertaken in sequence, enable continual improvement in decision-making. Communication and consultation will be reflected in each step of the process. Monitor and review is an essential and integral step in the risk management process. i Analysis of Internal and External Business Environment; and Determination of Requirements to ERM ii Establishment of ERM Strategy; Establishment of infrastructure Requirements about ERM iii Establishment of ERM Function and Committee iv Establishment of ERM Information System and Determination of ERM Framework v Establishment of Data Flow and Feedback systems; Analysis of ERM implementation Performance vi Providing the continuity of the ERM system for corporate sustainability.
218
Figure 2: The New Enterprise Risk Management Model for Corporate Sustainability (Kucuk Yilmaz, Ayse. 2008)
Following part of the study illustrates the components of each key step of the ERM process and illustrates the cyclical nature of the process: 1. Analysis of Internal and External Business Environment; and Determination of Requirements to ERM a. Corporate goals and objectives b. Determination of external and internal pressures (together with their risks) c. Analysis of internal audit structure, management systems and information systems d. ERM infrastructure, ERM culture and a common language for both the corporate sustainability management and the ERM. Reviewing of corporate culture and level of risk awareness. e. Analyzing of Stakeholder Relations Stakeholder identification: Employees, shareholders, community, partners, regulators and others. Stakeholder outreach: Transparency, inclusiveness, conferences, website, training and governance. Marketing Media Relations f. Determination of organizational resource and their allocation g. Determination of organizational strategy and its relationship with ERM h. Determination of corporate tolerances and appetite
219
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) i. Reviewing of corporate policies and procedure according to the ERM based approach to corporate sustainability j. Establishment of triple Bottom Line for corporate sustainability: social, economic and environmental as the conceptual commitment to a core sustainability theory and model k. Supporting of ERM by the current managerial approaches, applications and functions: strategic management and planning, sustainable development, corporate governance, line management, portfolio management, value management, etc. Establishment of ERM Strategy and Establishment of infrastructure to ERM a. Definition of organizational ERM mission and strategy and integrating with the strategic planning b. Establishment of highly corporate level to ERM awareness and risk perceptions c. Determination of useful ERM guidelines and reports, best practices; Decision making of which guides are using to developing ERM model d. Develop a common risk language and ERM terminology e. Establishment of organizational capabilities and abilities according to the best ERM implementation. f. Research and development g. System design, funding and staffing to ERM. h. Training and education about fundamental ERM and corporate sustainability. i. Establishment of corporate risk profile. Establishment of ERM Function and Committee a. Establishment of ERM function b. Place of the ERM Committee (CEO, CRO, CFO, CIO, etc.) in the organization scheme. c. Assignment of roles and responsibilities. d. Establishment of ERM philosophy as in the company, everyone is risk manager for corporate sustainability! Establishment of ERM Information System and Determination of ERM Framework a. Establishment of suitable ERMIS (Enterprise Risk Management Information System) b. Establishment of reporting format and line (top-down, across, down to top, external and internal parties) c. Analysis and Reviewing of the best ERM frameworks, guidelines and surveys. d. Determination of the risk identification tools, methods and models e. Determination of the risk assessment tools, methods and models f. Identify, assess and prioritize to corporate sustainability risks g. Analysis of key corporate sustainability risks and current capabilities h. Determining of strategies and design capabilities i. Classification of corporate sustainability risks according to the triple bottom line: social, economic and environment. j. Corporate risk mapping and selection of agreeable risk responses k. Measurement and reporting: Social capital, economic capital, natural capital, material flow analysis, energy flow analysis, internal environmental cost accounting, external environmental cost accounting, ecological footprinting, social footprinting and stakeholder reporting l. Establish continuously monitoring and review function Establishment of Data Flow and Feedback systems; Analysis of ERM implementation Performance a. Update and improvement of the ERM system b. Providing of the effective and well-timed data flow and feedback line c. Assessment of the organizational ERM practice performance. Providing the continuity of the ERM system for corporate sustainability. a. Providing the continuity of both the ERM system and its development
2.
3.
4.
5.
6.
220
b. Restart to ERM process over loop to new and developing risk and environment ERM model consists of interrelated components. These are derived from the way management runs an enterprise and are integrated with the management process. The model is tailored according to the qualifications and various parameters of the corporate sustainability management. All efforts have been achieved with coordinated studies of managers. First, corporate sustainability management model is analyzed in respect of Authors ERM model. New model steps are determined. The model is tailored according to the determinations about corporate sustainability. The new ERM-based corporate sustainability model is offered for Dogus Automotive and Sabanci Holding Managers to integrate with their ERM systems. Dogus Automotive and Sabanci Holding have the best and fully ERM systems in the Turkeys automotive sector. The automotive distributor companies should consider that ERM is an integral part of good governance and management practice and is committed to its application at all management levels within a firm-wide framework. ERM should be placed in the core of the CSM. ERM is important for sustainable development and business growth. Companies is needed a framework to manage and control significant sustainability risks that can interfere with achieving desired results and objectives. ERM should apply on the automotive industry and the automotive distribution companies since ERM is the fundamental part of the decision-making process, marketing, sales management and corporate governance. ERM is an important element of the competition and value creation since ERM is the fundamental sustainability component. In the Turkey, Sabanci Holding and Dogus Automotive Co. have improved and fully ERM systems. They are successfully practiced their ERM systems. ERM has a critical importance to achievement of organizational objectives for executive board and managers of the Sabanci holding and Dogus Automotive. They are improved their ERM policy and implementation guides best ERM implementation. The new ERM model of corporate sustainability (CS) is offered to integration into their current ERM systems for Sabanci Holding and Dogus Holding managers. The length of time basically needed to integrated new ERM model varies, depending on the current state of the risk management of the organization, its desired future state and the extent to which it wishes to dedicate resources to improve risk management capabilities. While, the completion of the applications taken into consideration takes twelve months time in any organization, most organizations will basically need from three to five years to reach their objectives in fully carrying out their ERM framework and related systems (Kucuk Yilmaz, 2008). Risk is a fundamental and inseparable aspect of business. Providing maximum shareholder value is possible if enterprise-wide risks are effectively and truly managed. Confronting opportunities includes many risks. Therefore, determining, measuring, and managing corporate risks are the most important factors for the achievement of sustainable development objectives in business management. Today's automotive business is under pressure as a result of the development of the international markets, globalization, liberalization, commercialization, expectation of raised effectiveness and cost reduction, hardly legal regulations, rapid technological developments, and increasing competition. The cost of making mistakes is increasing day by day. The possibility of compensation to the faults is gradually reducing. The risk-profit balance should be established by managers. It has been shown that earning should be optimized by increasing output and decreasing operational costs and loss. Enterprise Risk Management is increasing corporate value by providing sustainable competitive advantages and cost optimization, increasing business performance, and focusing all value sources on the company (Kucuk Yilmaz, Ayse, 2008).
5. Importance of Automotive Sector for the Turkish Economy

The automotive industry is one of the locomotive, largest exporter and leading investor sectors of the Turkish economy. Considering its remarkable contribution to the national production and development, to direct and indirect employment, and to the level of technology, automotive industry in Turkey is
221
now a crucially strategic sector on economical basis. In the 2000s, the Turkish automotive industry exhibits characteristics of an industry focusing on exporting. Nowadays, the shape and structure of Turkish automotive industry is changing. Foreign partners have begun to see the facilities that Turkey offers as their production center for their global markets. Due to the high export potential and Turkeys regional advantages, foreign capital has been showing an increasing interest in the automotive suppliers industry in Turkey. For instance, after Toyotas new investment project, new Japanese part manufacturing companies decided to invest in Turkey (TOSB Area) as well as bringing new business opportunities for local manufacturers (Bayraktar, 2008). Turkey is the only country in its own region having already established a well-advanced automotive industry, and it has proven itself to be a successful one in exporting every kind of automotive products to EU countries. Therefore, the automotive industry is strategically important both for Turkish and foreign companies that will invest in Turkey, a country which provides (Bayraktar, 2008): More than 1.000.000 annual modern vehicles production capacity, Developed suppliers industry with the power to compete, Young, educated and motivated man - power, Entrepreneurs ambitious to succeed, Global multinational shareholders, Adopted legislation in line with the international rules of free trade within the context of customs union and WTO, Easy access to potential emerging markets, Competitive labour cost and culture of productivity. The fluctuation in the Turkish economy in May 2006 sent exchange rates and interest rates upward. This trend had an adverse impact on automotive demand in the second half of the year, resulting in a 15% contraction in the passenger car segment and a 9% contraction in the commercial vehicle segment. The share of imports in 2006 remained unchanged at 55%. During the year, total automotive exports increased steeply by 26% to reach 706,000. Strong growth in exports offset the contraction in domestic sales; total automotive production rose 12% to over one million units. This growth was largely driven by the export boom in passenger cars. Turkey is an international hub for automotive sales and production. Turkey possesses a dynamic domestic market with nearly 50 competing brands and a strong and well-established automotive industry. Many international automotive brands have production facilities in Turkey, either through joint ventures or wholly owned subsidiaries, making Turkey an important international production hub. In 2006, the automotive sector, including the component industry, had exports of USD15.5 billion or 18% of Turkey's total exports, making it the country's top exporting sector. The Turkish automotive sector is in a strong position to attract new large-scale export projects in the coming years due to the cost advantages of its workforce and other production inputs, engineering skills and advanced technological infrastructure, geographic location and zero customs access to the EU market (Koc Holding, Automotive Group, 2008). Turkey is between Europe and Asia. Turkey's geographic situation has a strategic importance. Turkish automotive industry has an important strategic advantage and automotive companies from Italy, Germany, France, Japan, USA and S. Korea have important investments. They have export capacity and especially in crisis period this capacity helped them to survive from the crisis with less financial loss. In Turkey production costs are lower, compared to developed EC countries. Turkish automotive industry had begun production in 1950. Today there are 19 companies in production of which 5 are automobile producers and 14 are commercial vehicle producers. In this industry most of the companies have foreign partners and they produce for both domestic and international markets. Since the beginning, production is based on commercial vehicles (Suder and Cakmak, 2003). Turkey is rapidly becoming an alternative supplier centre due to its reasonable costs and efficiency of production which, in turn, encourage further investment (http://www.wbr.co.uk/autoturkey/turkey.pdf).
222
Automotive sector is one of the leading sectors in all developed and developing countries, because the sector has very important relations with all the other sectors of the economy. For instance, automotive sector is the main buyer of iron-steel sector, non-heavy materials sector, petroleum and chemical products sector. As a result, the technological developments in automotive sector force other sectors to show parallel technological developments. In addition, some sectors such as tourism, infrastructure and construction, agriculture, and transportation sectors depend on automotive sector for their transportation related needs. Moreover, automotive sector also creates employment for even unrelated sectors such as marketing, services, finance and insurance companies, and raw material suppliers, because these supporting sectors play a major role in transferring the goods and services to the consumers. So, considering all these interconnectedness of the automotive sector and all the other sectors, it is very obvious that any change in the sector will automatically affect the whole economy (Bekmez and Komut, 2006). As it is the case for all the countries of the world, the automotive sector has very dynamic process in regards to changes in market and competition conditions in Turkey as well. The sector entered this dynamic process with customs union agreement by passing through several hardships and high technological developments. Overcoming these hardships and adopting newer technologies, the Turkish automotive sector found itself a place and became a part of the global world. Thus, Turkish automotive sector created a new destiny for discussed in the following chapters, the sector has a long way to go in order to become really a competitive in the EU and also in global world. If we look at the dynamic effects of customs union agreement, the Turkish industrial sector has been generally opened to the world competition. As a result of this liberalization process, Turkish automotive sector increased the productivity and quality in order to become competitive in the global world. Within the context briefly discussed above, this study will analyze the goods and the bads of the Turkish automotive sector, and its place in the competitive world in general; and the sectors comparative advantages in the Europe. In order to do such an analysis, we discussed the competitiveness of automotive sector of the EU countries first; and then competitiveness of Turkish automotive sector. We used two ways to measure the competitiveness of Turkish automotive sectors: Revealed Comparative Advantages Index and Basic Regression Model. As discussion, results produced by each analysis have been compared within the national and international bases (Bekmez and Komut, 2006).
6. The Enterprise Risk Management Practice in the Turkish Automotive Sector

Turkey possesses a dynamic domestic market with nearly 50 competing brands and a strong and wellestablished automotive industry. Many international automotive brands have production facilities in Turkey, either through joint ventures or wholly owned subsidiaries, making Turkey an important international production hub. In 2006, the automotive sector, including the component industry, had exports of USD15.5 billion or 18% of Turkey's total exports, making it the country's top exporting sector. The Turkish automotive sector is in a strong position to attract new large-scale export projects in the coming years due to the cost advantages of its workforce and other production inputs, engineering skills and advanced technological infrastructure, geographic location and zero customs access to the EU market. This chapter is gives best samples of the corporate social sustainability practices in the Turkish Automotive Sector. The application of the proposed model has counted with the participation of two companies of the automotive sector, with the involvement of a person responsible for the answer of the questionnaire, who also worked as a facilitator in the contact with several internal departments. The interaction with this person was made mainly through telephone and e-mail, and it was not the objective of this model to accomplish an auditing of the rendered information. ERM is becoming critical management practice in the automotive business. In this section, ERM practices of the Turkish Automotive sector reviewed based on Sabanci Group and Dogus Group
223
cases since they are only current best samples related to the ERM implementation to automotive sector in the Turkey. 6.1. The Enterprise Risk Management Practices in the Sabanci Holding Sabanci Holding is the one of Turkey's leading industrial and financial conglomerates. Holding is composed of 30.000+ employees in 66 companies, 13 of which are listed on the Istanbul Stock Exchange. Holding is operates in about a dozen foreign countries and export products throughout Europe and the Middle East, as well as to parts of Asia, North Africa, and North and South America. Holding is grown through the expansion of existing core businesses and the formation of joint ventures. In 2004, the consolidated revenue was ~US$ 8.5 billion with a net income of ~US$ 700 million. The Sabanci family is collectively Sabanci Holding's major shareholder with 80.9% of the share capital (Saka, 2005). In October 2000, Toyotasa was established as a joint venture between Sabanci Holding (65%), Toyota Motor Corporation (25%) and Mitsui& Co. (10%). The Company is the exclusive distributor for Toyota-branded automotive products in Turkey and conducts marketing, sales and after-sales services of locally produced Corolla models and such leading imported models as Avensis, Corolla H/B, Yaris, RAV4, Land Cruiser, Camry and spare parts. Toyota vehicles are marketed, distributed and serviced by exclusive dealerships operating as Toyota Plazas throughout the country. These facilities were founded on the philosophy of superior quality, advanced technology and reliability. They provide a full range of world-class sales and after-sales services for Toyota customers. Since its establishment, Toyotasa has regularly increased its market share thanks to proactive marketing and sales strategies. In 2006, Toyotasa achieved an 8% share in the passenger car market by increasing sales volume for the fifth consecutive year and setting a new sales record with US$ 527 million. Ranking fifth place in the passenger car market, Toyotasa aspires to be among the top three brands within the next three years. Toyotasa also achieved significant improvements in both brand image and customer satisfaction, the latter being the main focus of the Companys vision (Sabanci Holding, 2008). Corporate Governance Principles and Rule of Conduct were set and communicated throughout the Group. Group is decided to implement ERM for all subsidiaries. Established Risk Management Department and introduced CRO position, which is the first of its kind in the Turkish business environment. Sabanci Holding Risk Management Vision is determined as Effectively apply sound risk management methods across the spectrum of the Holdings companies, thus preserving the investments of its shareholders and partners. Sabanci Holding ERM objectives are (Saka, 2005): To develop a widespread, systematic Risk Management concept throughout the organization To manage risks within defined risk tolerance levels To design and implement an effective risk management reporting and information management system To integrate ERM, as a proactive process, within the corporate culture so it becomes an important part of the strategic business planning and operational management processes Effective ERM helps provide maximum value to Sabanci Holding shareholders. Since each opportunity contains different level of risk, it is important to evaluate the impact of each specific opportunity against the total risk profile of the corporation. Integrated Enterprise Risk Management approach of the Sabanci Holding serves this purpose. In line with their aim to generalize Enterprise Risk Management approach in their Group, the Holding Risk Management Department has conducted appraisal and analysis studies in 2006. Furthermore, to minimize their operational losses and develop preventative techniques, a detailed risk evaluation study has been implemented. Risks encountered by their companies are identified and assessed in line with the Sabanc Holding Risk Model (see Figure-3) and Management Standards under the responsibility of the related companies' senior management. The probability of occurrence and the possible outcomes are evaluated while assessing the risks. The
224
magnitude of impact is generally stated in terms of the size of potential effect on operating profit. The relevant executives and managers have the responsibility for developing and implementing risk management actions. ERM approach of the Sabanci Holding aims to identify the potential risks of our operations in an early stage, analyze them, and promptly select and implement appropriate solutions to reduce the risks to appropriate levels.
Figure 3: Sabanci Holding Enterprise Risk Management (ERM) Process (Saka, 2005).
Risks and risk management actions are systematically monitored. In addition to regular reporting, there is also timely internal reporting and monitoring systems employed within the Group for unexpected risks. Central Risk Management Department informs the Board of Directors and Supervisory Board of Sabanci Holding when necessary. Overall, their companies successfully managed both anticipated and unexpected risks that arose in 2006. Sabanci Holding main strategies were established and implemented in accordance with the pre-determined Risk/Return profiles of all industries wherein we operated. 2006 was a successful year in terms of risk management for their financial services companies which constitute a substantial part of our business portfolio. The risk management strategy while conservative is based on exploiting reasonable opportunities in a timely manner. Akbank has achieved considerable improvements in risk management and Basel II compliance. Akbank Risk Management Group's main objectives are to form a risk management system in accordance with Basel II regulations and to have the best risk management practices compared to all emerging countries. The goal of the Sabanci Holding is to establish and develop the most effective systems for managing credit, market, operational and asset liability risks. The basic characteristics of the risk management philosophy that differentiates Akbank from other financial institutions are: To detect at an early stage, manage and monitor market, credit, operational, asset liability risk and other banking-related risks in a proactive manner and to appropriately allocate capital among business units, To establish a risk management system as an independent structure separate from business units,
225
To report all financial risks, To be open to learning and change in order to cope with new market conditions, To provide the highest value to shareholders, clients and employees, To be financially sound and reliable, establishing long-term business relationships with our shareholders and clients with the help of our long-term commitment to delivering the best, and To comply with Basel II regulations and other international guiding principles. Sabanci Holdings conservative approach toward risk is also in line with their long-term banking strategies. A substantial part of the risk management process employs various sophisticated models that evaluate market, credit, operational and asset/liability risks. These models are Basel II compliant and have been developed and/or validated by leading consulting companies (Sabanci Holding, Risk Management, 2008). 6.2. The Enterprise Risk Management Practices in the DOGUS Group Founded in 1951 with initial activity rooted in construction, the Dou Group has today become one of Turkeys largest conglomerates of impeccable track record. Dou Group pursues a customer-focused and productivity-centered management style, while also embodying the ethics of corporate citizenship that benefit broader society at large. The business volume derived from our multi-sectoral business structure features strategic international partnerships that enable us to offer a comprehensive range of brand products and services. (Dogus Group, 2008). Dou Automotive (Dou Otomotiv) is both the leading automotive importer and one of the largest distributors in Turkey. With its 249 sales and 276 after sales service points, Dou Otomotiv has the widest sales and service network within the Turkish automotive sector. Dou Otomotiv is the flagship company of the Dou Group's automotive segment. Dou Otomotiv is both the leading automotive importer and one of the largest distributors in Turkey. The company is home to 13 brands (Volkswagen Passenger Cars, Volkswagen Light Commercial Vehicles, Audi, Porsche, Bentley, Lamborghini, SEAT, Skoda, Scania, Krone, Meiller, Volkswagen Marine and Scania Marine) and 70 different models in the segments of passenger, light commercial and heavy transport vehicles. Dou Otomotiv is the official representative of the Volkswagen Group brands in Turkey since 1994 (Dogus Group, Group Companies, 2008). The Dou Group operates in the core businesses as financial services, automotive, construction, media, tourism, real estate and energy. The Dou Group features 73 companies, each a leader in its respective field. A workforce of over 18 thousand employees enables the Group to offer high quality products and services to its 6 million customers. The Dou Group is a key strategic player in the domestic economy by virtue of the employment opportunities it creates, taxes paid, and total business volume it generates. The Group focuses its strategies on customer satisfaction and trust, and strives for professional excellence in all business activities. By creating, developing and representing world-class brands, the Dou Group remains at the forefront of Turkey's transformation and innovation. The Group seeks to maximize the value of its brands, to grow through strong global alliances and become a regional leader in the services sector with its most valued assets, namely human resources and technological infrastructure (Dogus Group, Corporate Risk Management, 2008). Since April 2007, Dou Group has ranked among those institutions to have signed the UN Global Compact, a United Nations agreement indicative of its unswerving dedication to sustainable development. Dou Groups deep sense of social responsibility as a corporate approach is in alignment with the 10 key principles of the United Nations that may be summed up under four main headings; human rights, working conditions, the protection of nature and prevention of deterioration. The UN Global Compact agreement, whose vision is a sustainable and comprehensive global economy based on a sincere will to comply, began its journey with an invitation from former United Nations Secretary-General Kofi Annan to work towards Millennium Development Targets determined during the 1999 World Economic Forum in Davos. Thereafter, the major principles of the agreement were set. Companies worldwide then began to sign the agreement as confirmation of intent
226
to adopt these global principles as part of their working strategies, operations and corporate cultures. The UN Global Compact to date has been signed by 71 institutions in Turkey and over 3,000 institutions worldwide. By joining this list of companies Dou Group has emphasized the importance it places on sustainable development. Dou Group has conveyed its Corporate Risk Management approach in support of sustainable profitability and regional risk management-oriented development and a management style conducive to maximizing value added for its investors and shareholders. (Dogus Group, 2008).
7. Application of Proposed ANP Model to the Best ERM Practice

The ANP model which is presented in this research has been evaluated in an airport business, which is interested in the implementation of the ERM best practice. The analysis and the implementation of the ANP model are presented in the following six steps. The offered ANP model in the best ERM operator is given in Figure-4.
Figure 4: The proposed ANP model is the selection of the best ERM operator in the automotive business.
Step 1. In applying ANP, the first step is to build the model to be evaluated. The overall objective of this ANP model is to evaluate the ultimate relative importance of different factors that impact the best mentation of ERM and assess the best operator to ERM in the automotive sector. The factors that will be used to evaluate the alternatives are developed and explained in step 2. In this step, its important components are included in the decision problem. The relevant criteria and alternatives are chosen to be based on the review of literature and discussion with someone both from industry and academia. Also, the first step of the algorithm is the analysis of the best ERM practice problem. The fundamental aim of the best ERM practice problem is selecting the best operator in the automotive sector that meets the demands or criteria of the best ERM practice. Best ERM practice criteria are determined by the analysis of ERM framework guidelines and best practice survey results. These researches are explained in the literature review section of this study. Step 2: For the proposed best ERM practice model, overall 16 criteria are determined within the three main criteria sets mentioned below: i) Strategic ii) Operational and iii) Financial best practice criteria. All of the criteria are given in the following table-1. They will be used in the super matrix. Step 3: In this step alternatives are determined as the Sabanci and Dogus Holding. Selecting the alternatives among the more effective and successful ones in their ERM field of practices by using the preliminary elimination will increase the quality of the decision.
227
Step 4: Pairwaise comparisons are performed in between the factors. The interactions between cluster sets and elements are determined in this step. The selection of the control hierarchy of the best Enterprise Risk Management Operator Model according to the determined criteria is given in figure-5.
Table 1: The Criteria determination for the best ERM operator
Strategic Development and Establishment of ERM mission and strategy Integration of ERM into other management practices and management functions Creating and Promoting an ERM culture and common language; Development of organizational ERM policy and procedures Achievement of Corporate Risk Optimization Operational Determining corporate risk appetite and tolerance line; creating corporate risk profile Establishing open communication and feedback systems Establishing ERM information system Setting up ERM function and committee Setting up an ERM framework for all aspects of corporate-based risks Overall enterprise risk assessment and analysis, enterprise risk mapping and prioritization Financial Use and supply of outsource about ERM Supply of modeling tools and techniques Resource allocation to ERM efforts, requirements and infrastructure Establishing ERM Framework Providing resource to sustainability of the ERM development.
THE CRITERIA TO THE BEST ERM PRACTICE (Kucuk Yilmaz, 2007.)
Step 5: Obtaining the overall outcome. Following weighting is obtained by the pairwise. Table2 shows the priorities of all the factors in the decision-making model.
Table 2:
Graphic
Final reprioritization of alternatives: It is obtained from ANP model.

Alternatives Dogus Sabanci Total 0.2749 0.2251 Normal 0.5498 0.4502 Ideal 1.0000 0.8188 Ranking 1 2
228
Figure 5: The ERM best practice network models control hierarchy (Ayse Kucuk Yilmaz, 2008)
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) The proposed decision model of best operator selection is implemented for 16 criteria under three main criteria sets and two main alternative operators. After this step, I have mine unweighted super matrix. Then priority weights of the sets are calculated by using experts' opinion. Multiplying this priority weight by the unweighted super matrix, I have the weighted supermatrix. The final step is the calculation of the limiting priorities of the weighted supermatrix. According to the final prioritization of alternatives given in Table-2, the DOGUS is the highest priority from the SABANCI. For this reason DOGUS is selected as the best operator about ERM implementations. To sum up, it is hard to decide which has the best practice for ERM certainly since ERM is not a standardized application. Also, ERM is shaped according to the organizational factors. It is mentioned before; ERM is not a one size fits all approach. It really is depends on the differences of purposes, the condition of resources and capabilities, and the existing organizational cultures. However, this study shows that; DOGUS is a best ERM operator as a distributor according to the current practice in the automotive distribution business. Dogus Automotive is identified their enterprise risk profile. They have successful risk portfolio. Sabanci Holding has improved holding-wide risk management system and practice. However, they should do many improvements in their ERM system for achievement of best practice
Table 3:
1. 2. 3.
229
The Criteria of The Best ERM Practice (Kucuk Yilmaz, 2008)

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 3 3 3 3 3 3 3 3 3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 5 5 5 5 5 5 5 5 5 5 5 5 5 5 5
Development and Establishment of ERM mission and strategy Integration of ERM into other management practices and management functions Creating and Promoting an ERM culture and common language; Development of organizational ERM policy and procedures 4. Achievement of Corporate Risk Optimization 5. Determining corporate risk appetite and tolerance line; creating corporate risk profile 6. Establishment of the open communication and feedback systems 7. Establishment of the ERM information system 8. Setting up ERM function and committee 9. Setting up an ERM framework for all aspects of corporate-based risks 10. Overall enterprise risk assessment and analysis, enterprise risk mapping and prioritization 11. Use and supply of outsource about ERM 12. Supplying of modeling tools and techniques 13. Resource allocation to ERM efforts, requirements and infrastructure 14. Establishing ERM Framework: Shaping for company 15. Providing resource to sustainability of the ERM development and continuity Reviews and Suggestions:
1. 2. 3. 4. 5. Certainly Not enough Not enough Basic Level good practice Best Practice
In this study, an assessment form (Table-3) is prepared for high level managers in the Sabanci Holding and Dogus Group. The form is offered criteria to the best ERM practice. The practice level of ERM is searched with this form. The Sabanci Holding and Dogus group has improved systems to the ERM framework and its implementations. Results are assessed for selecting of the best ERM operator between automotive distributor companies in the Turkey. According to the research results, Sabanci Holding and Dogus Group has improved and successful ERM system. Sabanci Holding Group Companies has been successfully managed both anticipated and unexpected risks that arose in 2006. ERM is practiced both non-financial and financial group companies in the Sabanci Holding. Sabanci Holding has improved holding-wide risk management system and practice. Sabanci Holding is achieved considerable improvements in risk management and Basel II compliance. Although Sabanci Holding has been pioneer about ERM in the Turkey, Dogus is best ERM operator about managing enterprise risks as automotive distributor company. Dogus automotive company has various automotive brands. For this reason, Dogus is
230
managed their risk distribution in their improved risk profile concept. Finally, Dou Automotive is both the leading ERM operator in the automotive distribution business and one of the largest distributors in Turkey. Dou Group has conveyed its Corporate Risk Management approach in support of sustainable profitability and regional risk management-oriented development and a management style conducive to maximizing value added for its investors and shareholders. Dou Group is geared at continuous growth, efficient performance in parallel to global developments, and efficient risk management in competitive environments. It generates brand equity through a combination of dynamic human resources and state of the art technological infrastructure. It is precisely in this environment that the Group fosters and sustains customer loyalty. The Executive Management of Dou Group is one of the leaders of change, given its ability to focus on changes both global, and specific to group companies. Dou Group considers corporate values such as shareholder relations an inseparable part of the service sector. Indeed, the Group defines itself with the phrase We solve, we execute and we develop with integrity." Dou Automotive is Turkeys largest automotive value chain provider with 249 authorized sales points distributed nationwide as of 2006 year-end. The Company provides comprehensive sales, service, spare part sales and after sales service for the Volkswagen, Audi, Porsche, Bentley, Lamborghini, SEAT, Skoda, Scania, Krone and Meiller brands in Turkey. Geared at increasing its competitiveness and diversifying its sources of income, Dou Automotive has added various automotive-related services to the companys core business over the past few years. Hence, Dou Automotive is now active in import and distribution, retail sales, after-sales services, consumer finance for vehicle purchasing, spare part and accessory trading, car insurance services, second-hand vehicle trading, and car fleet rentals. In accordance with its expansion strategy, the Company signed an agreement in 2006 with Krone, the second largest trailer producer in Europe, on joint production in Turkey. Recently, Dou Automotive has also become the exclusive Turkish distributor of the luxury car brands English Bentley and Italian Lamborghini, of VW Marine Engines, and of Meiller tippers. VW AG has recently granted a license to Dou Automotive to import and distribute VW light commercial vehicles in Egypt.
7. Conclusion of the Study

The study is presented the new ERM model to CSM and the method for applying ANP in determination of the best ERM operator in the automotive sector. To assist companies to evaluate and select ERM best practice successfully, I have proposed an effective method based on the ANP. Additionally, this study has contributed to extend practical applications of ANP in ERM field. ERM requires to significant managerial efforts and organizational recourses. Both Sabanci Holding and The Dogus Group have the best practice criteria for achievement of the best implementation of the ERM. In this study, I have developed both new ERM model and a framework based on ANP to identify the degree of impact of factors affecting best ERM practice. I used the ANP to decide the dependence and feedback. ANP is a new methodology that incorporates feedback and interdependent relationships among decision attributes and alternatives. It leads us to a fresh insight about significant matters. Based on my model, I find that Dogus Holding is the best operator from the Sabanci Group is given %54 rates about current ERM practice. Most of the existing models that investigate the factors affecting ERM implementation have used various survey methods. I have shown an alternative approach using expert judgment. The ANP approach, as a part of this methodology, not only leads to a logical result but also enables the decision-makers to visualize the impact of various criteria in the result. Since ANP is capable of dealing with all kinds of feedback and dependence when modeling a complex decision environment, I conclude that my results are more accurate. ANP deals with uncertainty and complexity and provides new insights that other, more traditional methods can miss.
231
The ANP approach is capable of taking into consideration both qualitative and quantitative criteria. The model of the ERM for corporate sustainability offers a possibility for companies to translate sustainability visions and strategies into action. This approach seems to be interesting for both researchers and practitioners because it shows how intangible assets may contribute to the corporate sustainability. Moreover the new ERM model provides high potential for the integration of environmental and social aspects and objectives into the core management of companies. This is a huge step forward for the corporate sustainability management.
References
[1] [2] [3] [4] [5] [6] Anderson, Dan R. Managing sustainability risks is critical. (Perspectives)(Corporate Survival: The Critical Importance of Sustainability Risk Management), Business Insurance, November 14, 2005. Bayraktar, Ahmet. Turkish automotive industry, Potential to do Business, Association of Automotive Parts and Components Manufacturers in Turkey (TAYSAD), www.taysad.org.tr, February 01, 2008. Bekmez, Selahattin and Murat Komut, Competitiveness of Turkish Automotive Industry: A Comperison with European Union Countries, Growth and Development, p. 181-182, International Conference on Human and Economic Resources, Izmir, 2006. Brunner, Marc. Tool Automotive, Draft, Version 20, CSM / WWF Project on The Business Case for Sustainability, Forum for Corporate Sustainability Management, International Institute for Management Development, 13 Haziran 2003, s.5. Dogus Group, Corporate Risk Management, http://www.dogusgrubu.com.tr/en/icerik/8508/398/web_kurumsal_yonetim_ilkeleri_yatirimci/c orporate_risk_management/, January 30, 2008. Dogus Group, Corporate Governance, Investor Relationships and Sustainable Development http://www.dogusgrubu.com.tr/en/icerik/9847/1/webkurumsal/dogus_group, http://www.dogusgrubu.com.tr/en/icerik/8507/398/web_kurumsal_yonetim_ilkeleri_yatirimci/s upport_for_sustainable_development/, January 30, 2008. Dogus Group, Group Companies, Automotive, http://www.dogusgrubu.com.tr/en/icerik/184/2/web_sektorler/automotive/, February 05, 2008). ETH Zurich, SusTecGroup for Sustainability and Technology, Schaltegger St. et.al., 2005, Corporate Sustainability, The international yearbook of environmental and resource economics. Koc Holding, Automotive Group, Market Overview, http://www.koc.com.tr/enUS/Business/FordOtosan/#basla, January 30, 2008). Kucuk Yilmaz, Ayse, ENTERPRISE RISK MANAGEMENT IN THE AIRPORTS: THE MODEL SUGGESTION FOR THE ATATURK AIRPORTS TERMINAL OPERATIONS COMPANY, Doctorate Dissertation, Civil Aviation Management, Anadolu University, Eskisehir, Turkey, 2007. Kucuk Yilmaz, Ayse. Enterprise Risk Management Perceptions in Airports of Turkey, 2nd International Conference on Research in Air Transportation ICRAT 2006, Serbia, Belgrade, June 24-28, 2006. Kucuk Yilmaz, Ayse. Airport Enterprise Risk Management Model, A study on Business Management and Airline Management, VDM Verlag Dr Mller, Germany, 2008. Kytle, Beth and John Gerard Ruggie, Corporate Social Responsibility as Risk Management A Model for Multinationals Workin Paper No. 10. March 2005, p.15. Mercedes Benz Turk A.. About The Turkish Automotive Industry, http://www.wbr.co.uk/autoturkey/turkey.pdf, February 06, 2008.
[7] [8] [9] [10]
[11] [12] [13] [14]
232 [15] [16] [17]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Sabanci Holding, Strategic Business Units, Tire, Tire Reinforcement Materials and Automotive, Toyotosa, http://www.sabanci.com/en/grup_baskanliklari_sirket.asp?ID=101, February 05, 2008. Sabanci Holding, Risk Management, http://www.sabanci.com/En/st_riskyonetimi.asp, January 30, 2008. Saka, Tamer. H.O. SABANCI HOLDING Enterprise Risk Management, The IRM Risk Forum 2005, Keele University, Staffordshire, United Kingdom. http://www.theirm.org/riskforum/documents/creating_ERM_non_financial_corporations_AON _enterprise_rm_2005.pdf, 2008. SciTec Today, The Critical Importance of Risk Management, January 2, 2008, http://www.planet2025news.net). Suder, Asl and Ahmet Cakmak, Crisis Management in the Turkish Automotive Automotive Industry, Endstri Mhendislii, July August- September 2003 - Number 3., http://www.mmo.org.tr/endustrimuhendisligi/2003_3/crisis_management.htm).
[18] [19]
233
Effect of Capital Structure on Firms Performance: The Nigeria Experience

Ishola Rufus Akintoye Room 116, Department of Economics, Faculty of the Social Sciences University of Ibadan, Ibadan Nigeria, West Africa Tel: 234-8035369293; 234-8082130269 E-mail: irakintoye @yahoo.com Abstract In this paper, we examined the effect of capital structure on firms performance. We address the following questions: Does higher leverage lead to better firm performance? Is the effect of performance on leverage similar across the distribution of different capital structures? Using a sample of 10 Nigerian quoted firms with consideration of their financial statements for three years, we discover that an evenly distributed capital structure has positive effect on firms performance, while the effect of performance on leverage varies across the distribution of different capital structure as seen from the companies understudied. Most of the equity financed firms in this study performed as much as those who employed debt in their structure in term returns on equity and assets. Although we cannot generalize this fact as few other firms with debt finance performed more efficiently as in the case of Nestle Nig. Plc, Northern Nig Flour Mills Plc, hence the effect of leverage on efficiency varies across the distribution of different capital structure lending credence to the agency cost theory of Jensen and Meckling(1976). We therefore recommend that investors should concentrate on engagement of efficient management team, motivation and other developmental programmes so as to achieve goal congruence in the long run.
Introduction
Capital structure is the mix of the long term sources of fund by the firm, Arthur (1985). It basically involves how a company finances its operations. The concept amplifies the importance of, and provides a stronger theoretical foundation for financial analysis in Business Policy and Strategic Management.(Akintoye, 2006) Arthur (1985) observed that a lot of enduring controversies have emerged within the arena of financial theory. These controversies have taken many elegant forms in finance literature, He noted that most of these controversies appeal to academics as opposed to practitioners of the financial management art. The argument centres on the effect of financial leverage on the overall cost of capital to the enterprise. The least of the argument may be stated in the form of a question; Can the firm affect its overall cost of funds and efficiency, either favourably or unfavourably by varying the mixture of financing sources used? Hence the need to evaluate the concept of capital structure as it affects the firms efficiency. We shall consider the effect of leverage i.e. capital structure on the firm efficiency. In particular, we shall address the following questions: Does efficiency exert a significant effect on leverage over and above that of traditional financial measures of capital structure? These we intend to achieve by using financial ratio and market valuation ratios to measure firm efficiency. As opined by Berger and Bonaccorsi di Patti(2006), efficiency measures are closer to the theoretical definition of agency costs and have the advantage over the more traditional measures of firm performance which is based on financial ratios in that they control firm specific factors outside the control of management, that are not part of agency cost. However, for the purpose of this paper, we
234
shall adopt the traditional financial ratios to measure firm efficiency. We shall also briefly revisit production technology which has been traditionally left out in modern finance representations of the firm using the Leibenstens framework on allocative efficiency Vs X-Efficiency, and more recent theoretical work on production frontiers as put up by Fare, Gross Kopf, and Lovell (1985, 1994) in their papers on measurement of efficiency of production, and The Production Frontiers respectively so as to establish the link between productive efficiency and leverage and we shall briefly discuss the theories of firm efficiency firm efficiency and performance are interrelated..
Empirical Review
In a recent study, Berger and Bonaciorsi (2006) examined the relationship between capital structure and firm performance for the US Banking Industry. They used a parametric measure of profit efficiency as an indicator of (inverse) agency costs in analyzing the bi-directional relationship between efficiency and capital structure. They found evidence in support of the agency cost hypothesis whereby the effect of the cost of outside equity dominates the effect of debt over, almost, the entire range of observed data. Similarly, in a different but related content, Zavgren (1985), Keasey and Watson (1987), Beccheti and Sierra (2003) have emphasized the importance of non-financial data as predictors of company failures. Zavgren (1985) argued that econometric models that solely rely on financial statement information will not predict accurately business failures. Using a measure of productive efficiency obtained from a stochastic frontier model, Becchetic and Sierra (2003) discover that productive inefficiency is a significant ex-ante indicator of business failure while Leasey and Watson (1987) report that better predictions for small company failures are obtained from models using nonfinancial data rather than conventional financial indicators.
Theories of Firm Efficiency: Agency Cost Theory

The agency cost theory is premised on the idea that the interest of the companys management and its shareholders do not perfectly align. The Model Di = a0+a1Li+a2L2i+a3Z1i+ui Where D is the firm efficiency measure, L is the total debt to total assets ratio; Z1 is a vector of control variables; and u is a zero mean error term. According to the Agency Cost Hypothesis, the effect of leverage(L) on efficiency should be positive, i.e a1 > 0, a1+2a2L > 0. However, the possibility exists that at sufficiently high leverage levels, the effect of leverage on efficiency may be negative(Margaritis and Psillaki, 2006). The expropriation of wealth and the me-first rules illustrate the need for debt-holders to monitor the actions of equityholders. Monitoring requires the expenditure of resources, and the costs involved are one form of agency costs. As discussed earlier, Jensen and Meckling (1976) have expounded the theory of agency costs. Among other things, they show that regardless of who makes the monitoring expenditures, the cost is borne by stockholders (Akintoye, 2006). Agency costs can also exist from conflicts between debt and equity investors. These conflicts arise when there is a risk of default, which may create under investment or debt overhang problem (Myers, 1977). Whereas the agency costs of outside equity support a positive relationship between firm performance and leverage, the agency costs of outside debt result in a negative effect of leverage firms are more likely to pass up profitable investment opportunities or shift top riskier operating strategies (Margaritis and Psillaki, 2006). In line with Jensen and Meckling (1976) we expect the effect of leverage on agency costs to be negative overall. Therefore, under the agency costs Hypothesis (Hi) higher leverage is expected to lower agency costs, reduce inefficiency and thereby lead to our improvement in firms performance,
235
with the provisor that this relationship may switch at a point where the disciplinary effects of further increases in leverage become untenable. Efficiency-Risk Hypothesis More efficient firms choose higher leverage ratios because higher efficiency is expected to lower the costs of bankruptcy and financial distress. The efficiency risk hypothesis is an accidental benefit of the trade off theory of capital structure whereby differences in efficiency, other thing else constant, enable firms to alter their optimal capital structure either up or down. However, it is also possible for firms that expect to sustain high efficiency rates into the future to choose lower debt to equity ratios in an attempt to guard the economic rents or franchise value due to these efficiencies from the threat of liquidation. Franchise Value Hypothesis According to Berger and Bonaccorsi (2006), more efficient firms tend to hold extra equity capital and therefore, all else equal choose lower leverage ratios to protect their future income or franchise value. Thus the efficiency risk hypothesis and the franchise value hypothesis yield opposite predictions regarding the likely effects of firm efficiency on its choice of capital structure (Margaritis et al, 2006). Although we cannot identify the separate substitution an income effects, our empirical analysis is able to determine which effect dominates the other across the spectrum of different capital structure choices.
Methodology
Here, we analyse the model that supports the relationship between efficiency and capital structure. We considered the financial statements of companies in study for three years using ratio analysis of Tangible Assets, other assets, Owners Equity and Long Term Liabilities as indicators of Capital structure and Return on Equity, Return on Assets Earnings and Net asset per share as indicators of efficiency. The Leverage Model The capital structure equation relates the debt assets ratio to our measure of efficiency as well as to a number of other factors that have commonly been identified in the literature to be correlated with leverage (Raviv, 1991; Myers, 2001). The leverage equation is given by: (6) L1 = 0 + 1 Di + 2 Z2i + Vi Where Z2 is a vector of factors other than efficiency that correlate with leverage and V is an error tern. Under the efficiency risk hypothesis, efficiency has a positive effect on leverage, i.e. B1 > O; whereas under the franchise value hypothesis the effect of efficiency on leverage is negative, i.e. B1 < O. This is in line with Myers (2001) who emphasized that there is no universal theory but several useful conditional theories describing the firms debt-equity choice. The variables included in Z2 control for firm characteristics such as size, asset structure, profitability, risk and growth and for industry characteristics such as market power that are likely to influence the choice of capital structure (Raviv, 1991). Firm Size (SIZE) is measured by the logarithm of the firms sales (Rajan and Zingales, 1995). As larger firms are more diversified and tend to fail less often than smaller ones, we would expect that size will be positively related to leverage. However, Rajan and Zingales (1995) discuss the possibility that size may also be negatively correlated with leverage. They argue that size may act as a proxy for the information outside investors have and that informational asymmetries are lower for large firms. Tangibility (TANGIBLES) is another variable which is measured by the ratio of fixed tangible to total assets (Frank and Goyal, 2003). The existence of asymmetric information and agency costs
236
may induce lenders to require guarantees materialized in collaterial. For instance it will be more advantageous for a firm which retains large investments in land, equipment and other tangible assets. It will have smaller costs of financial distress than a firm that relies on intangible assets. We therefore expect that tangibility should be positively related to debt. Profitability (PROFIT) is measured by pre-interest and pre-tax operating surplus divided by total assets (Fama and French, 2002). There are conflicting theoretical predictions on the effects of profitability on leverage (Barclay and Smith, 2001; Booth et al, 2001) Myers and Majufs (1984) predict a negative relationship because they argue firms will prefer to finance new investments with internal funds rather than debt, as demonstrated from their packing order theory. Thus they argue that we should expect a negative relationship between past profitability and leverage. On the other hand, using the arguments on trade off and contracting cost theories, we can predict a positive relationship between profitability and leverage. RISK (RISK) is measured by the standard deviation of annual earnings before taxes (Mackie Mason, 1990). Several authors have included a measure of risk as an explanatory variable of the leverage level. Leverage increases the volatility of the net profit. A negative relation between risk and leverage is expected from a pecking order theory perspective: firms with high volatility on earnings try to accumulate cash to avoid under investment problems in the future. Growth (GROWTH) is measured as the annual percentage change on earnings. Growth is likely to put a strain on retained earnings and push the firm into borrowing. However, there are conflicting theoretical predictions on the effects of growth on leverage. Based on Myers (1977), we expect that growth will be negatively related to leverage. On the otherhand, Michelas et al, (1999) argue that growth will push firms into seeking external funds and require additional capital. Consequently, growth may also be expected to have a positive relationship with leverage.
Empirical Results
In this section, we provide answers to the question raised in section 1. As we mentioned in the introduction, we are interested in examining how capital structure choices affect from performance. We use a sample of 10 Nigerian firms from different industries precisely. Conglomerates (UTC Nig Plc and UAC Nig Plc), Building and materials (Benue Cement Company, Cement company of Northern Nigerian Plc, WAPCO), Food and beverages (Nestle Nig Plc, Northern Nigerian Flourmills Plc), Chemical and Paints (Chemical & Allied Products Company, IPWA, African Paints Nig Plc). Table 1 presents balance sheet information for the selected industries. Table 2 gives the descriptive statistics of the firms in the sample featuring ROE, ROA, turnover to total Assets, turnover to total Debts.
Data Analysis & Findings

Conglomerate: UTC & UAC NIG PLC Tangible Assets: % of tangible assets in both company of same industry is high which indicate strength an continuity of the company has long standing effect. The high level % improve and reflect on the turnover of the companys (UAC & UTC) for the periods under consideration. Other Assets: The insignificant % of other assets. In UTC may be traced to the reduction in profit made in the current year. However, UAC, has more reasonable % in other assets i.e. investments which is traceable to the increase in profit during the current period.
237
Owners Equity: Both companies under consideration are lowly geared with the accompanying effect of reduction in liquidation risk, little cost of debt, etc. For UTC, the more equity employed the higher the ROE which is a good indicator of performance of the firm. On the other hand, UAC is going in opposite direction, there is more return on equity on the least combination of capital structure than of the higher combination which suggests that owners equity is better invested in more profit oriented ventures (the lower the equity the greater the ROE). Turnover to Total Assets: The companies record higher turnover throughout the period under consideration without a corresponding increase in profit for same period understudy this shows negative implication on the firms performance, which implies that irrespective of the low gearing of the companies, the overall effect on the firms performance does not deserve an applause. Capital Structure: An evenly distributed capital structure has a positive effect on ROE, as seen from UACs i e. a higher capital structure may not necessarily yield or result in better firm performance, but an evenly distributed one will do.
Building & Materials

Benue Cement, CCNN, WAPCO Tangible Assets: Both BC & WAPCO record a very high % through BC, does not results in a corresponding increase in turnover, perhaps due to the long term effect of the tangible assets so acquired, as seen from the constancy of the revaluation reserve. This also depicts an indication of the performance of the firm. WAPCOs high % on tangible assets shows a corresponding increase in turnover in the current period, however this increase in turnover does not result in an improved profit this shows that more of the companies fund had been invested in idle assets which do not improve on the performance of the companies. CCNN records a reasonable distribution of assets, as the companies has enough current assets to meet immediate obligations as well as tangle assets, the positive effect of this is seen on the shift from loss to profit by the firm thereby enhancing the firms performance. Owners Equity: For the industry both BC & CCVV are lowly geared, resulting in reduction in liquidation risk, cost of debt among other advantages, however, only CCNN has the positive effects on low gearing, as records profit after the losses incurred in previous period. LongTerm Liabilities: WAPCO is highly geared, having a higher proportion of its capital structure as debt. However this dos not result in a corresponding increase in profit as the company records losses throughout the period under consideration, perhaps as a result of improper debt management. ROE & ROA BC and WAPCO makes no return on its equity finance, while CCNN was return in the current able to make a higher return in the current period than previous probably.
238
Summary None of the capital structure of the companies considered in this industry is evenlky distributed hence the positive effect on ROE & ROA as indicators of performance.
Food & Beverages

Nestle Nig. Plc & Northern Nig Flour Mills Current Assets & Tangible Assets: Both are evenly distributed in Nestle Nig Plc with a resultant effect of increase in both turnover & profit throughout the period under consideration this is a positive indicator of performance by the firm. However Northern Nig Flourmills high % in current asset bring about an improvement on turnover & profit, in the current period. Owners Equity & LongTerm Liabilities: Both companies in this industry employ more of financial leverage (debt) than equity finance, which implies that the company will pay more on cost of debt, however it is interesting to note that despite the higher debt finance the companies were still able to make increase in their turnover and profit. ROE & ROA: For the equity finance, the companies here are still able to make reasonable returns which shows a better management of Equity finance. ROA reduced with the level of Equity in both companies, this we could trace to the increase in LongTerm liabilities of the companies understudy in the industry. Summary: Nestle Nig Plc & Northern Nig Flour mills are both levered companies, the better management of debt have resulted in improved profit, leading to a better performance in the current period understudy.
Conclusion
From our analysis and findings, companies with evenly distributed capital structure have their return on equity being positively affected resulting in better performance of the firms. We also discovered that higher financial leverage does not necessarily lead to a better firm performance as an increased percentage in tangible assets does not outrightly indicate better firm performance, as tangible assets with long standing effect may contribute little or nothing in the short term. Investors should also note that increase in turnover is not an indication of firm performance, without a corresponding increase in profit. Our findings in this study is therefore similar to that of Dimitri Margaritis and Maria Psillaki of 2006, where using a sample of 12,240 New Zealand Firms, they found evidence supporting the theoretical prediction of Jensen and Meckling(1976) Agency Cost Hypothesis which says that the interests of the companys manager and its shareholders do no perfectly align and that managers maximize their own utility rather than the Firms value leading to the development of the Free Cash Flow Theory(Jensen, 1986).
Recommendation
We opine that increased turnover and assets do not necessarily give a corresponding increase in profit and earnings, as a result, shareholders/investors should not just concentrate their efforts in increasing the quantum of fund available for managers use, rather they should be concerned about the quality of management and other workforce being put in place. They should ensure optimal recruitment decision
239
and sensitizing managers through motivation via profit sharing/share allocation, staff development programmes etc such that the interest of investors and companys managers may meet leading to goal congruence rather than suboptimization.
Appendixes
Table 1: (Financial Ratios as Indicators of Capital Structure) Conglomerates
UTC NIG PLC

Tangible Assets Other Assets Owners Equity Long Term Liabilities 2004 0.99 0.01 0.99 0.01 2003 0.99 0.01 0.82 0.18 2002 0.99 0.01 0.70 0.30
UAC NIG PLC

Tangible Assets Other Assets Owners Equity Long Term Liabilities 2004 0.72 0.28 0.82 0.18 2003 0.72 0.28 0.77 0.23 2002 0.74 0.26 0.82 0.18
Table 2:
(Financial Ratios as Indicators of Performance)

2004 0.075 0.092 0.97 1.15 16 51 2003 0.067 0.12 0.98 1.23 16 93 2002 0.055 0.08 0.72 1.64 40 23
UTC NIG PLC

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) Net assets/Share
UAC NIG PLC

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) Net assets/Share 2004 0.14 0.03 1.83 9.27 1.37 85 1048 2003 0.28 0.04 1.56 8.18 240 60 957 2002 018 0.04 1.43 11.07 128 35 708
Table 1:
Financial Ratios (As Indicators of Capital Structure)
Building & Materials Benue Cement Company

Tangible Assets Current Assets 2004 0.88 0.12 2003 0.74 0.26 2002 0.70 0.30
240
Equities & Liabilities % of Capital Employed

Owners Equity LongTerm Liabilities 2004 0.99 0.01 2003 0.82 0.18 2002 0.70 0.30
Cement Company of Northern Nig Plc

Equities & Liabilities % of Capital Employed

Owners Equity LongTerm Liabilities 2004 0.94 0.06 2003 1.00 2002 1.00 -
West African Portland Cement Company

Equities & Liabilities % of Capital Employed)

Table 2:
Financial Ratios (As Indicators Of Performance)

2004 0.11 0.06 (146) 534 2003 0.13 0.09 (4.33) (387) 2002 0.23 0.24 (216) 46
Benue Cement Company

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) Net assets/Share
Cement Company of Northern Nigeria Plc

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) Net assets/Share 2004 0.59 0.16 1.03 1.07 35k 10 154.75 2003 (.16) (.03) 0.91 0.91 72k 102.4 2002 (1.30) (.21) 0.59 0.84 84k 159
West African Portland Cement Company

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) Net assets/Share 2004 0.70 1.07 (198) 152 2003 0.44 1.08 (185) 352 2002 0.52 1.11 (81) 537
241
Table 1:

Food & Beverages
Nestle Nig Plc

Tangible Assets Other Assets Current Assets 2004 0.30 0.03 0.68 2003 0.18 0.03 0.79 2002 0.14 0.04 0.82
Equities & Liabilities

Northern Nig Flour Mills Plc

Tangible Assets Other Assets Current Assets 2004 0.11 0.02 0.87 2003 0.08 0.02 0.90 2002 0.08 0.92
Equities & Liabilities

Table 2:
Nestle Nig Plc

2004 2.27 0.46 2.12 2.44 7.26 7 2003 2.38 0.49 2.07 2.39 7.20 7 2002 2.13 0.53 0.22 2.67 6.01 6
ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k)
Northern Nig Flour Mills plc

ROE ROA Turnover to total assets Turnover to total debts Earning/Share (k) Dividend/Share (k) 2004 0.19 0.11 2.81 4.86 99k 50k 2003 0.20 1.13 2.37 4.33 93k 22k 2002 0.25 0.18 3.37 6.65 100k 25k
242
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Akintoye I.R. (2006), Investment Decisions Concepts, Analysis and Management, Lagos Nigeria. Barclay, M.J., C.W. Smith, Jr (2001), The Capital Structure Puzzle: Another Look at the Evidence in D.H. chew (ed.) The New Finance: Where Theory Meets practice, McGraw-Hill, New York. Barclay, M.J., C.W. Smith, Jr and R.L. Watts (1995), The Determinants of Corporate Leverage and Dividend Policies, Journal of Applied Corporate Finance, vol. 7 (4). Becchetti, L. and j. Sierra (2003), Bankruptcy Risk and Productive Efficiency in Manufacturing firms, Journal of Banking and Finance, vol. 27.2099-2120. Booth, L., V. Aivazian, A. Dermirguc-Kunt and V. Maksimovic (2001), Capital structure in developing countries, The Journal of Finance, vol. 39 pp. 857-878. Castanias R. (1983), Bankruptcy risk and Optimal Capital Structure, The Journal of Finance, vol.38, n05, December, pp. 1617-1635. DeAngelo, H. and R. masulis (1980), Optimal Capital Structure Under Corporate and Personal Taxation, Journal of Financial Economics, vol 8 (1), pp 3-29. Dimitri Margaritis and Maria Psillaki(2006) Capital Structure and Firms Efficiency in Newzealand Journal of Finance. Fare, R. and S. Grosskopf (2004), New Directions: Efficiency and Productivity, Kluwer Academic Publishers, Boston. Fare, R., Grosskopf, S. and C.A.K. Lovell (1994), Production Frontiers, Cambridge University Press, Cambridge. Harris, M. and A. Raviv (1988), Corporate Structure and the Informaiton role of Debt. The Journal of Finance, vol. 45. n02, june, pp. 321-349. Harris, M. and A. Raviv (1991), The theory of Capital Structure, the Journal of Finance, vol. 46, pp 297-355. Jensen, M. (1986), Agency Costs of free Cash flow, Corporate Finance, and Takeovers, American Economic Review, vol. 76, pp. 323-329. Jensen, M. and W. Meckling (1976), Theory of the Firm: Managerial Behavior, Agency Costs and Capital Structure, Journal of Financial Economics, vol. 3.pp 305-360. Keasey, K. and R. Watson (1987), Non-Financial Symptoms and the Prediction of Small company Failure. A Test of Agentis Hypotheses Journal of Business Finance and Accounting, vol. 17, No. 1, 119-135. Koenker, R. and G. Bassett (1978), Regression Quantiles, Econometrica, vol. 46, 107-112. Kremp, E., E. Stoss and D. Gerdesmeir (1999), Estimation of a debt function: evidence from French and German Panel Data in A. Sauve and M. Scheur (Eds). Corporate Finance in Germany and Franca (Frankfurt-am0Main-and Paris: Deutsche Bundesbank and Banque de France). Law commission (2001), Insolvency Law Reform: Promoting Trust and Confidence, An Advisory Report to the ministry of Economic Development, May, Wellington, New Zealand. Leibenstein, H. (1966), allocative Efficiency vs. X-Efficiency, American Economic Review, vol. 56, 392-415. Modigliani, F. and M.H Miller (1958), The Cost of Capital, Corporation Finance and the Theory of Investment, American Economic Review, vol. 48, pp. 261-297. Myers, S. (2001), Capital Structure, The Journal of Economic Perspectives, vol. 15, n02, spring, pp. 81-102. Rajan, R.G. and L. Zingales (1995), What do We Know about Capital Structure? Some Evidence from International Data, The Journal of Finance, vol. L. No. 5, pp. 1421-1460. Scott, J. (1977), Bankruptcy, Secured Debt, and Optimal Capital Structure, the Journal of Finance, vol. 32, March, pp 1-19.
References
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17]
[18] [19] [20] [21] [22] [23]
243 [24] [25]
European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008) Titman, Sh. (1984), The Effect of Capital Structure on a Firms Liquidation Decision, Journal of Financial Economics, vol. 13, pp. 137-151. Titman, Sh. and R. Wessels (1988), The Determinants of Capital Structure Choice. The Journal of Finance, vol. 43, pp. 1-19.

Ejefas 10

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ejefas 10

Uploaded by

Copyright:

Available Formats

1

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

EUROPEAN JOURNAL OF ECONOMICS, FINANCE AND ADMINISTRATIVE SCIENCES

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance and Administrative Sciences

Rural Non-Farm Employment in India: The Trends and Determinants

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

2. Non-Farm Sectors and its Importance

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

3. Rural Non-Farm Sectors in India: Its Size and Composition

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

4. Database and Modeling

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

5. Results and Discussion

Source: Sivasubramonian, S. (2004); and Fan et al. (1999).

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

Determinants Constant PCDEVEX HYV IRR RLIT RROAD AGGR PCGDP R2 F DW

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

Determinants Constant RROAD RLIT HYV R2 F DW

6. Conclusions and Policy Implications

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

A Substitute for Riskless Assets and the Market Portfolio

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

2. Substitute Portfolio of Riskless Assets

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

3. A Substitute for the Market Portfolio

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

4. The Efficient Frontier and Capital Market Line. A General Case

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

It is known that the CML is: R j = RF +

5. Investing and Financing the New Portfolio. General case

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

Because asset j is in equilibrium, then: j2 = ( jm m )2 From which we can see that: 2 ij = im jm m

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

Effects of M&As on the Spanish Savings Banks

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

II. Review of the Literature

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

III. Methodology and data used

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

IV. Empirical Results

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

Ratios R1 R2 Market power R3 R4 R6 Risk R7 Operating efficiency

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

European Journal of Economics, Finance And Administrative Sciences - Issue 10 (2008)

j > 0 . Then, i = [ kl ,i ] and

i = [ kl ,i ] = i P where P satisfies PP' = u .