You are on page 1of 20

UIS RIP-9th Regional Innovation Policies Conference

Paper ID 1379
The conceptual support of the academy to the implementation of Regional Innovations Policies in
Brazil: Learning, Cooperation and Innovation in Local Productive Arrangements (LPAs)
Jorge Britto - Fluminense Federal University (UFF) and RedeSist
Marco Vargas - Fluminense Federal University (UFF) and RedeSist
Fabio Stallivieri - Fluminense Federal University (UFF) and RedeSist
Abstract: Based on the Brazilian experience, the paper discusses how concepts originally formulated in
the academic sphere can be used as instrumental and methodological tools for the implementation of
regional innovation policies. A horizontal-territorial focus of the regional innovation policy is discussed,
comprising the use of the concept of Local Productive Arrangements (LPAs), originally developed in the
academic sphere and widely adopted as a tool for the implementation of public policies in Brazil. The
analysis tries to identify general patterns regarding the innovative dynamics of enterprises inserted in
LPAs, evaluating the influence of learning processes and cooperative networks at the local level to the
strengthening of the innovative performance of those companies.
Key words: Conceptual Support to Regional Innovation Policies; Local Productive Arrangements;
Learning and cooperation; Horizontal-territorial policies

Introduction
The evolutionary approach of Industrial Economics has pointed the importance of connecting the
characteristics of the knowledge generation and the identification of critical dimensions of industrial
agglomerations. The discussion about how knowledge is generated, appropriated, distributed and
enhanced might contribute to understand how those agglomerations work, allowing not only to
differentiate them according to a greater or lesser degree of complexity but also to evaluate their potential
to evolve along a virtuous path of competence growth. In an evolutionary perspective, a major feature of
those agglomerations refers precisely to their ability to operate as a mediator between the firm and the
external environment, which increases the capacity of absorbing knowledge potentially useful for the
strengthening of efficiency, innovativeness and competitiveness. These agglomerations might redefine the
dichotomy between "internal" and "external" sources of knowledge, acting as an intermediate instance
which allows to "format" the knowledge according to the requirements of the competitive process,
providing relevant externalities, stimulating the integration of competences and generating multiple spillover effects. However, despite the recognition of the learning process as a critical aspect of this dynamics
- empirically illustrated by a growing number of case studies there still a gap regarding cross-sector
analyzes that enable the identification and quantification of those gains at the firm-level.
Two main objectives orient the analysis developed in the paper. First, it describes and analyzes a
relevant experience related to the transfer of analytical concepts developed in the academic sphere to the
design and implementation of a regional innovation policy in Brazil. Specifically, this experience refers to
the contribution of the concept of "Local Production Arrangements" (LPAs), elaborated as a variant of the
broader concept of "industrial agglomerations to the strengthening of a regional focus of the innovation
policy. This concept refers to a specific set of economic activities spatially located and sectorally
specialized, oriented to learning practices and to the generation and diffusion of new products and
processes. Second, the article tries to expand the understanding of the relationship between territorial
proximity, cooperation and innovation, based on an analytical framework that seeks to articulate the
intensity of learning and innovative processes to elements that emerge from territorial specificities.
The article comprises three sections. First, the analysis discusses the functionality of the concept
of Local Production Arrangements (LPAs) developed from a theoretical evolutionary perspective for the
1

definition an implementation of regional innovation policies. Second, the analysis discusses the support of
the academic sphere to provide an instrumental and methodological basis to the implementation of
regional innovation policies in Brazil, comprising the use of the concept of Local Productive
Arrangements (LPAs) as a tool for the implementation of public policies. Third, the analysis tries to
identify general patterns regarding the innovative dynamics of enterprises inserted in Local Productive
Arrangements (LPAs), evaluating the influence of learning processes and cooperative networks at the
local level to the strengthening of the innovative performance, of those companies. The analysis is based
on data collected from 1,187 companies inserted in 29 Local Productive Arrangements (LPA), trying to
identify clusters of firms with similar patterns regarding the characteristics of innovative efforts, learning
practices and insertion in local cooperative networks.
1. The concept of Local Production Arrangements (LPAs)
The concept of Local Production Arrangements (LPA) was developed as a variant of the broader
concept of industrial agglomerations, recurrently used as an analytical approach to discuss aspects related
to the territorial competitiveness by the modern literature of Industrial Economics and Regional
Economics. The basic assumption of those analyses is that industrial agglomerations might provide
positive externalities at the territorial level, increasing productive efficiency and creating a suitable
environment to the raise of innovativeness and competitiveness of the firms located in the territory
(Maskell and Malmberg, 1999; Maskell and Kelbir, 2005).. The use of this analytical category to discuss
structural conditions that affect firms competitiveness goes back to classical theoretical approaches,
starting from the works of Marshall (1890). These approaches have generated important analytical
developments in the field of the New Economic Geography (Krugman, 1991 and 1995), Structuralist
Regional Economics (Storper, 1996 and 1997, Scott and Storper, 1986, Piore and Sabel, 1984),
Innovation Economics (Audretsch, 1995; Audretsch and Feldmam, 2004; Maillat, 1995 and 1998; Maillat
and Kelbir, 1999) and in the literature about modern Industrial Districts (Schmitz, 1997; Nadvi and
Schmitz, 1994; Musyck and Schmitz, 1995; Pyke, Becattini and Sengenberger, 1990). At the same time,
the concept is becoming increasingly present in the policy guidelines of international development
agencies (OECD, 2001 and 2007, World Bank, 2009)
A proliferation of empirical studies developed from an evolutionary theoretical perspective
contributes to the refinement of the concept of industrial agglomerations. From these studies, some
relevant attributes of those agglomerations may be stressed: 1) Geographical proximity; 2) Sectoral
specialization and intra-sectoral division of work; 3) Close inter-firm collaboration; 4) Inter-firm
competition essentially based on innovation rather than on lower wages; 5) Social embeddedness that
facilitates trust, reciprocity and social sanction; 6) Different forms of state support. Bell and Albu (1999)
develop an analysis of the elements that strengthen the integration of capabilities in the knowledge
systems associated to those agglomerations, stressing the differences between elements that increase
knowledge-using capabilities and elements that increase knowledge-changing capabilities. Concerning
the first aspect, they mention the passive experience of production (learning by doing in production'') at
the firm level, the active efforts to adopt and improve specific technologies and the improved practices
derived from trial and experimentation on specific tasks. At the level of the agglomeration, they mention
the mobility of skilled labor, the improvement of operational skills and the knowledge diffusion of
specialized machinery or production-related services. Concerning the knowledge-changing capabilities at
the firm level, they mention the technological understanding gained from investment efforts (learning by
doing investment'') and the generic technological insights gained from adapting and improving existing
technologies (learning by changing''). This dimension involves collective practices in planning and
technology management, as well as collaboration to adapt machinery, to improve processes or to develop
product designs. The creative collaboration between firms and local technology-based institutions seems
also to be very important.
The evolutionary approach also argues that geographical proximity is not enough for the
achievement of collective learning processes and innovative dynamism. To amplify these effects, this
proximity has to be articulated with other elements, such as the institutional, cultural and technological
2

context, in order to foster the existence of an innovative system. The Regional Systems of Innovation
(RSI) concept rests on the relationship between technology, innovation and industrial location (Mothe and
Paquet, 1998; DAllura, Galvagno and Destri, 2012), highlighting the regional dimension of the
production and the exploitation of new knowledge, thereby helping to explain regional differences in
innovation capacity and economic strength. RSIs usually consist of a set of interacting private, semiprivate and public organizations, interacting within an institutional framework which stress the generation,
exploitation and dissemination of knowledge and thus supports innovative activities on a regional level
(Cooke, Uranga and Etxebarria, 1998; Doloreux, 2002).
According to this perspective, the presence of multiple ties among local actors performs a critical
role to strengthen competence-building processes in industrial agglomerations that conform RSIs. The
establishment of those ties may provide the necessary conditions to promote localized learning processes
and to consolidate innovative paths based on incremental innovations. On the other hand, in order to
avoid the danger of a geographical lock-in related to the exhaustion of learning processes, the
agglomerations might also retain capabilities to break productive practices and to change technological
paths (Iammarino and Mccann, 2006; Paniccia, 2005; Cooke and Morgan, 1998; Cooke, 2001; Menzel
and Fornahl, 2009). According to Christopherson, Michiel and Tyler (2010), these processes generate a
kind of regional resilience, defined as the capacity of a territory to overcome short-term or long-term
economic adversity (Hudson, 2010; Martin, and Sunley, 2006; Pike, Dawley and Tomaney, 2010). This
resilience would be provided by a strong regional system of innovation (Clark et al., 2010; Howells, 1999)
and by the effective creation of a learning region (Archibugi and Lundvall, 2001).
The territorial proximity between agents inserted in a similar social, cultural and institutional
context enhances cooperative practices that reinforce learning gains (Johnson and Lundvall, 1994). Noneconomic factors, socially defined rules and local institutional conditions affect the interactions between
economic agents, generating incentives for cooperation and learning. The evolutionary approach tries to
articulate the static competitive advantages generated by the spatial agglomeration with dynamic
competitive advantages obtained through the strengthening of learning practices and multiple forms of
cooperation. The systematic interchange of information and knowledge generates a process of collective
learning, which accelerates the diffusion of technological and organizational innovations. These flows
involve intangible assets and the circulation of tacit knowledge. Although innovations intentionally
developed in co-operation tends to occur only in more structured systems, there are many possibilities to
improve the competitiveness of local productive systems due to informal mechanisms of learning. The
evidence also shows that the circulation of information and skilled workers would improve the
competences of the firms inserted in those agglomerations. Another aspect refers to the impacts of the
interchange of information to the definition of industrial standards, normalization procedures and quality
control techniques.Given the tacit character of knowledge, innovation usually requires several forms of
interaction among economic agents, who in turn interact with technology-based and knowledge-based
institutions. In this sense, learning-by-interaction becomes a critical aspect of industrial agglomerations.
Typically, interactions develop in the form of cooperative efforts, formal or informal. Then, cooperation
constitutes the main instrument to improve learning-by-interacting practices. While cooperation is an
effective tool for information processing, it is also an important alternative to enable the binding of
complementary skills, to increase productive efficiency and to improve the innovative potential of interindustry arrangements.
In this context, the technological development of a firm becomes increasingly dependent on the
capabilities of other firms, competitors, clients and suppliers, being possible to differentiate horizontal
cooperative links among firms inserted in similar stages of the value chain and vertical cooperative links
involving firms, suppliers, customers and other organizations. Among those organizations, it can be
mentioned research centers, technical schools, public institutions and private representative associations.
All these agents conform the complex institutional context in which cooperative links emerge. The
external learning can complement but not replace the internal, increasing its effectiveness or changing its
direction. Particularly in knowledge intensive sectors, the viability of the innovation process requires a
direct and permanent interaction between firms and different sources of information, through which
capabilities could be calibrated, adjusted and incremented over time.
3

Based on this perspective, the Research Network on Local Production Arrangements and
Innovation Systems (RedeSist) - established as a Brazilian academic network with international links
since 1997 - developed the concept of Local Production Arrangements (LPAs) in the late 1990s, with a
focus on a set of economic activities spatially located and sectorally specialized. Some critical aspects of
these arrangements involve learning practices and the generation and diffusion of new products and
processes, combining elements of the Evolutionary Neo-Schumpeterian theoretical approach about
innovation systems with contributions about industrial development and structural change elaborated by
the Latin American structuralist school (Lastres, 2007; Cassiolato and Lastres, 2008; Cassiolato, Lastres
and Maciel, 2003). The focus on LPAs comprises a systemic view of the productive and innovative
activities, considering a multiplicity of economic, political and social actors that define the contours of
those activities in the territory.
The methodological framework that sustains the analysis of LPAs argue that geographical
proximity is not enough for the achievement of collective learning processes and innovative dynamism.
The concept tries to articulate the static competitive advantages generated by the spatial agglomeration
with dynamic competitive advantages obtained through the strengthening of learning practices and
multiple forms of cooperation. The complexity of knowledge flows, the multiplicity of the relations, the
intensity of interactive learning mechanisms and the degree of cooperation among agents are factors that
interfere in how learning processes take place, and therefore, in the generation, use and diffusion of
knowledge. The mapping of this diversity seems to be an important analytical tool to understand how
those processes occur and change over time.
2- Policies to Local Production Arrangements (LPAs) in Brazil
The concept of Local Production Arrangements (LPAs) has quickly disseminated not only in the
academic sphere as well at the policy level, covering a broad phenomena referring to the concentration of
similar or interdependent activities in economic space, without restrictions about the size of the
companies, nor about the nature of economic activity, which can be primary, secondary or tertiary. The
incorporation of this approach in the sphere of public policies occurred so early and fast, going to replace
other similar frameworks on political agendas. Given the breadth of the concept to characterize
productive agglomerations, the Federal Government, under the Permanent Working Group for Local
Productive Arrangements (GTP-APL), integrated to the Ministry of Development, Industry and Foreign
Trade (MDIC), opted to incorporate the general terminology of LPA. The efforts to internalize the LPA
framework to the support the federal programs involve a compromise between the original framework
elaborated by the academia and the experience accumulated by policy institutions over many years. In
this sense, the adoption of the concept seems to be very flexible, becoming a general guide to the
operational strategies of different institutions, whose strategies effectively have remained focused on the
traditional institutional subjects and on the general mission of each agency.
The institutional set-up of agencies engaged with the implementation of the innovation policy in
Brazil comprised some key actors as well as some specific mechanisms of coordination. The
improvement of the effectiveness to the Innovation Policy has involved the integration of the Science and
Technology Policy with the Industrial Policy. Each of these policies have been implemented by a specific
Ministry respectively, the Ministry of Science, Technology and Innovation (MCTI) and the Ministry of
Development, Industry and Foreign Trade (MDIC). MDIC has also under its purview BNDES the (the
National Bank for Economic and Social Development) which constitutes the main source of longterm
finance for Brazilian companies.
The incorporation of the approach of LPAs as an instrument of policy occurred since 1999 under
the Ministry of Science, Technology and Innovation (MCTI). In partnership with federative states, LPAs
were identified in projects that aimed to support the cooperation between research institutes and industrial
firms. It also included for the first time the mention of LPAs in the general framework of the
governments Multiyear Plan (PPA 2000-2003), in which the responsibility of the MCTI to support those
arrangements was explicitly mentioned. In this period, there was also an increase of government support
to academic research about the theme, both theoretical and empirical. The actions of MCTI were
4

implemented through the support provided by its agencies, the Council of Scientific and Technological
Development (CNPq) and the Financier of Studies and Projects (FINEP).
In 1999 began the collaboration between the MCTI and the Forum of Federal States Secretaries of
S&T for the identification and support of LPAs. This work adopted a methodology based on the broader
concept of Technology Platforms, mobilizing local actors to identify bottlenecks and to propose solutions
for specific problems of LPAs. In this effort, each federative state was tasked to identify three
arrangements to be supported. Platforms supported by MCTI were 54 in 2000, 53 in 2001 and 42 in 2002.
In this period, CNPq handled the operation of the program to support innovation in LPAs, supported by
grants from FINEP, which had also created in 2001 a specific action to run this program, called
Structured Action for LPAs, implemented by the Area of Innovation for Regional Development - ADRE.
Since 2003, with the beginning of the first term of the Lula government, the MCTI had undergone an
internal restructuring and the coordination of initiatives related to the support of LPAs was transferred to
the Minister of Development, Industry and Foreign Trade (MDIC), with MCTI no longer having a
specific line to support LPAs in the period 2004-2007.
Since 2003, while the policy for LPAs gained greater prestige and political support, not only at the
federal but also at state and municipal levels, the MCTI pioneer in the adoption and implementation of
this approach - lowers the priority given to it. This trend might be connected to a recurring issue
concerning the main focus of the Brazilian policy for LPAs that can be identify not only in MCTI actions,
but also in other institutional bodies that have taken this approach in their actions: an apparent
dissociation between promoting innovation and local industrial development. This attitude, in turn,
reflects a supposed decoupling between technological development and social development. Thus, it
appears that while some bodies associated with the promotion of LPAs have tended to focus on the
support for innovation (such as FINEP in the period 1999-2003), others have seemed to be primarily
concerned about the promotion of local development (such as MDIC in Lula's first term).
Under the FINEP, the Program for Support Research and Innovation in Local Productive
Arrangements (PPI - APL) was reformatted to support activities undertaken by institutions of science and
technology (ICTs) in cooperation with companies inserted in LPAs, focused on R&D, technological
assistance, technological services and technological troubleshooting. Simultaneously to FINEP actions, it
was created in the first half of 2003 an inter-ministerial group to support LPAs, aiming to integrate
actions implemented by different bodies, coordinated by the Ministry of Development, Industry and
Foreign Trade (MDIC), with the participation of 21 agencies working at the federal level, as well as other
governmental and non-governmental bodies. This group was formalized in August 2004, with the name
of the Permanent Working Group for LPAs (GTP-APLs). The Multi-Year Plan (PPA) 2004-2007, and
later the PPA 2008-2011, have also incorporated the theme LPA in their structure. Coordinated by the
Ministry of Development, Industry and Foreign Trade (MDIC), the GTP - APL was expanded in 2005,
with the inclusion of over 10 institutions (totaling 33). The first goal of GTP - APL was to coordinate,
articulate and integrate the different actors, policies and actions to promote LPAs at the federal level,
carried out by public and private bodies. The main actions developed initially were directed to conceptual
issues and to the establishment of a consensus criteria for the classification of LPAs, in order to permit a
general identification of LPAs around the country and to construct a database containing the mapping of
all actions performed by existing bodies involved with LPAs or similar approaches.
Reflecting this effort, 27 Support Nuclei to LPAs were installed in federative states in the period
2006-2008. The institutional design of these Nuclei was very heterogeneous, following the historicalinstitutional trajectory of each federative state concerning this issue, with different institutional settings
and different legal frameworks. The federal banks including not only BNDES, but also Banco do Brasil,
Caixa Econmica Federal, Banco do Nordeste and Banco da Amaznia - also began to use or expand the
use of a LPA approach in their operations. There is also an increasing interest not only of these public
banks, but also from private banks that have integrated later the GTP-APL. This interest reflects the aim
to extend financial services to smaller companies inserted in LPAs and the recognition that such a
strategy facilitates the proximity of actors, their formalization and their access to the financial system.
The creation of the GTP-APL occurred within the framework of the first experience of industrial
policy in Lulas government, the Industrial, Technological and Foreign Trade Policy (PITCE), elaborated
5

in 2003. The issue was articulated to the strategic theme "Regionalization" included in the Productive
Development Policy (PDP), the general framework of the industrial policy implemented in 2008, as well
as to the cross-cutting theme Special Actions in Regional Development, mentioned the in the
subsequent Brasil Maior Plan launched in 2011. It is also important to highlight the spread of the concept
of LPAs as a guiding policy to other instances of the government. In the case of BNDES, the main agency
for the promotion of industrial development, the incorporation of that concept in operational practices
resulted from the need to articulate long-term national policies with regional and local priorities. In this
context, challenges were associated with the need to increase the presence of BNDES in the regions and
federative states less attended, reducing intra-regional imbalances. This process was reflected in the
establishment in 2007 of the Committee of Productive Arrangements, Regional Development, Innovation
and the Environment, as well as in the creation of a Special Office for the Development and Local
Productive Arrangements, directly linked to the BNDES President. These instances try to articulate
various operational areas of BNDES, contributing to incorporate regional development and a territorial
systemic approach to their lines of action.
The GTP - APL had contributed to the spread of the concept of LPA, allowing the exchange of
experiences and cooperation among public and private stakeholders at the national and federative state
levels. It also contributed decisively to the improvement of information systems and to the use of
indicators for the assessment of policies to support LPAs (Lastres, 2011). Three key mechanisms of
learning emerged from this experience. First, it indicates that it is possible to overcome the limits of
occasional, strictly sectoral and one-dimensional policies, advancing in the understanding that the
productive development depends on the interaction between multiple actors and institutions, including
those responsible for the generation of knowledge, funding and representation. Second, it indicates that it
is possible to overcome policy models driven by an administrative logic that reduce the political
management to unique and de-contextualized frameworks, based on general parameters and
methodologies, which, at the end, tend to reinforce sectoral, social and territorial inequalities. Thirdly, it
indicates that the territorial dimension constitutes a fundamental issue to be considered in infrastructure
projects, not only at the macro-regional, but also at the sub-regional and local levels.
However, despite the progress achieved, there were some limitations intrinsic to this pattern of
policymaking. Some of these limitations involves the development of general methodologies to select and
classify LPAs, many of which start from a recognition of the existence of arrangements in different
"stages" incipient, potential, stagnated, dynamic, mature, world-class etc. Additionally, there were also
limits to the application of traditional rules and quantitative methods to characterize territorial
agglomerations (and consequently LPAs) and to define their formats, which are often based on models
and typologies overly schematic. There were also situations in which policy programs seek to "build"
cooperation and governance, treating firms and other actors as patients who hypothetically need to learn
how to interact, cooperate and innovate. Thus, it is common to impose policy prescriptions based on
models that ignore local conditions in terms of their historical and socio-political context. Another
problem comprises the implementation of policies to support LPAs as part of a strategy that lacks
convergence with other policies of the Federal Government. In this sense, Brazilian experience shows an
emphasis on traditional compensatory actions in which the promotion of innovation in LPAs is often
placed in a disconnected manner, or even in opposition to the promotion of local development and social
inclusion. The need to train a staff of policy-makers and local agents qualified to deals with the
complexity of the development of LPAs also constitutes an important challenge for the advancement of
the policies.
To overcome these threats, a relevant issue refers to the effective contribution of the academia to
the improvement of regional innovation policies based on a LPA approach. In this sense, three types of
contribution might be stressed. The first involves academic professionals specialized in discussing and
formulating normative conceptual guidelines for the definition and implementation of those policies. The
second involves a more direct contribution, through which qualified personnel originated from academia
assume a relevant executive role in the implementation of those policies. The third comprises the
contribution of the academic sphere to the training of the staff responsible for the implementation of those
policies at different institutional instances. Regarding these contributions, it is important to highlight the
6

role of research networks with a "dual" character, dedicated to both academic research as well as to
methodological and operational support of policy implementation through consulting activities. The
continuity of the policies implemented favors a process of learning and the accumulation of a "critical
mass", which enables their adjustment and calibration in order to achieve a greater efficiency in the
implementation of actions.
The Brazilian experience concerning the implementation of a regional innovation policies based
on a LPA approach reflects this trend. The establishment of channels of communication between the
academic sphere and policy-makers has allowed an exchange of information and experiences that
promote policy adjustments, increasing or reducing the scope of actions according to the results generated
and to the responses of the audience of the policies. These adjustments reflect a process of "institutional
interactive learning" that mobilizes and articulates the academic sphere and policy-makers instances,
which can improve the effectiveness of the innovation policies. Reflecting this trend, the identification of
a set of problems related to the implementation of regional innovation policies based on a LPA approach
resulted in a movement towards a "2nd generation" of policies. These adjustments reflect three dynamics:
1) a natural process of "institutional learning " on the part of policy makers; 2) a (self) diagnosis about the
critical limits of the current standard implementation of these policies and of the challenges for their
further development; 3) a return to the academic sphere in order to get a conceptual and methodological
basis suitable to the reformulation of the policies.
Two general principles guide this evolution. The first refers to the reinforcement of the cohesion
of the local development through an orientation to the economic potential of the territories and to their
institutional and productive environment, even when considering the surroundings of large projects. The
second principle incorporates the notion of a sustainable development trajectory, linking the different
dimensions of economic development with an emphasis on sustainable exploitation of socio-biodiversity,
exploiting new niches based on clean technologies and activities socially and environmentally responsible.
This evolution towards a 2nd generation of policies to support LPAs relies too heavily on a coordination
with the academic sphere, both in terms of conceptual and methodological foundations as well as in the
format of actions and programs .
Concerning these changes, it is possible to mention the demand of the GTP- APL towards the
format of a training program by the academy directed to managers responsible for formulating and
implementing policies for LPAs at different governmental levels. The creation of the Thematic
Committee "National Training Plan" by GTP-APL reflects this trend. To format this program three
different audiences were identified: 1) Formulators and Operators of LPAs Policies; 2) Managers and
local multipliers agents; 3) Entrepreneurs and local productive agents. Another contribution of the
academy refers to the structuring of an Integrated Knowledge Management System to LPA policies,
through the establishment of the Brazilian Observatory of LPAs. The structure of this national
observatory comprises the creation of a database, an Internet portal with various features and a social
network, operated by the GTP-APL with the support of the representatives of the Nuclei established in
federative states. Another important contribution of the academy refers to the identification and the use of
a set of indicators for the assessment of policies to support LPAs..
3 Learning and Cooperation in Brazilian Local Productive Arrangements: some characteristics
3.1 - Data and Indicators
The use of the concept of Local Productive and Innovative Arrangements requires the
development of analytical tools with the aim to capture dimensions not found in statistics based on
traditional sectoral and territorial dimensions. It also points the importance of conducting case studies
based on a common methodological framework that captures the relevance of learning practices and
cooperation to the generation and diffusion of new products and processes.
The analysis developed in this section attempts to fill up the gap from the lack of systematized
information about the structure, the internal processes and the innovative performance of LPAs in
Brazilian economy. The methodology comprises empirical surveys based on direct collection of data through
7

interviews with firms inserted in LPAs developed by academic researchers integrated to REDESIST.
Information was collected from a questionnaire designed with the aim of understanding learning and
interaction processes, evaluating externalities of the local environment and assessing different aspects that
could affect firms performance. The questionnaire was designed in such a way to make it compatible with
the Brazilian Innovation Survey (PINTEC) carried by IBGE (the Brazilian Institute of Geography and
Statistics), incorporating indicators to evaluate learning and cooperation processes in LPAs. Specific
questions were formulated to evaluate the origin of the information used for learning, internal or external
to the firm. Other questions verify the intensity of interactions and the strength of the relationships with
other agents in LPAs.
To identify the influence of technological efforts, learning processes, cooperation, local
externalities and networking to the innovative performance, the analysis recurs to the use of 30 indicators,
divided into four distinct groups: 1) indicators of innovative efforts; 2) indicators of external learning and
cooperative actions; 3) indicators of the density of externalities to local production; 4) indicators of
innovative performance. These indicators were estimated based on questionnaires applied to the firms,
transforming qualitative attributes, such as the importance attributed by the company to a particular event,
in quantitative ones, establishing a value between 0 and 1 to express the opinion of the company about
each event. These indicators were individually calculated for each one of the 1.187 firms in the sample.
Table A in the annex summarizes the indicators used in the analysis and the events captured by each one
of them.
The analysis comprises a self-evaluation of the surveyed firms about the main factors that
influenced their innovative efforts, learning, cooperation and performance. Although recognizing that
such procedures may distort results, once interviewee not always has the proper understanding about what
is questioned, they are fully recognized as adequate to evaluate innovative dynamics at the firm level,
being accepted as an important tool by OECDs Oslo Manual (2005) that establishes the methodological
principles that have guided national Innovation Surveys in several countries. Furthermore, the possibility
of obtaining empirical data from different sources based on common methodologies and concepts tends to
minimize problems related to the diversity of the interpretation of questions among the agents.
Based on the set of indicators, we sought to identify general patterns using statistical procedures
related to Multivariate Analysis 1 . Through the implementation of four Factor Analyses, one for each
group of indicators, we try to systematize and reduce the number of aspects related to each dimension,
starting from 30 indicators extracted from data collected by questionnaires. A second step applies
techniques of Cluster Analysis, based on the synthetic factors identified and on the scores attributed to
them. This procedure might be justified because of the heterogeneity of the sample. The common patterns
identified operate as an analytic tool to format this diversity, elucidating aspects of the innovative
dynamics in LPAs. The sample from which the analysis was developed has some specificities. On the one
hand, the companies are located in territorial agglomerations of different areas of economic activity,
which affects the intensity of the learning process and the orientation of technological efforts. On the
other hand, the sample consists mainly of micro and small enterprises. Table 1 shows the stratification of
the sample according to the economic activity performed and the size of the firms. There is a strong
predominance of micro enterprises, which together represent 59.2% of the sample; small enterprises reach
32% of the total and medium enterprises 8%. We also find a greater concentration on activities related to
traditional industries (furniture, clothing, footwear, plastics and fishing), which together represent 68% of
the sample. The most capital-intensive and knowledge-intensive activities (mechanical equipment and
components, oil and gas, software, computer equipment and telecommunications and biotechnology)
represent a smaller part of the firms in the sample (32%).

For a mathematical and statistical formalization of Multivariate Analysis, see Hair et al (2005), Johnson and Wichern (1998).

Table 1 Stratification of the sample according to the activity performed by the companies and
their size (N=1.187)
Size

Activity
Metal Mechanic,
Equipment and
Components, Oil and Gas
Furniture and Wood

Textile, Clothing and


Footwear

Hardware, software and


telecommunication

Micro
Location of LPAs
(Municipality and
federative state)
Joinville - SC; Camaari
BA; Ribeiro Preto
SP; Maca - RJ
Ub - MG; Linhares - ES;
Vitria - ES; Chapec
SC; Unio da Vitria - SC
Colatina - ES; Apucarana
- PR; Terra Roxa - PR;
Petrpolis - RJ; Cabo Frio
- RJ; Ibitinga - SP;
Campina Grande - PB;
Jaragu - GO; Natal
RN; Tobias Barreto SE;
Birigui SP
Petrpolis - RJ; Ilhus BA; Curitiba - PR; Recife
PE; Braslia DF; Santa
Rita do Sapuca-MG
Belo Horizonte - MG
Cricima - SC
Itaja - SC

Small

Medium

Total

%.

%.

54

7,68%

69

17,92%

32

32,32%

155

13,06%

164

23,33%

91

23,64%

20

20,20%

275

23,17%

284

40,40%

140

36,36%

23

23,23%

447

37,66%

136

19,35%

51

13,25%

14

14,14%

201

16,93%

Biotechnology
15
2,13%
4
1,04%
0
0,00%
19
Plastics
12
1,71%
14
3,64%
8
8,08%
34
Fishery
38
5,41%
16
4,16%
2
2,02%
56
703
59,22%
385
32,43%
99
8,34%
1187
Total
Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs).

1,60%
2,86%
4,72%
100%

The descriptive statistics presented in Table 2 refer to the average value of the indicators in the
entire sample of 1.187 companies. Innovative efforts seem to be mainly associated with information
obtained internally, since the indicators related to internal learning show high values - 0.59 for internal
learning derived from different organizational sources2 (LEARN ORGFONT) and 0.28 for the use of
R&D department (LEARNR&D) as a source of information for innovation. Despite the high relative
importance of the R&D department, these activities are conducted in a small scale by the firms in the
sample, since the indicator relating to the constancy of R&D activities (CONSTR&D) presents a reduced
value (0.21). The firms develop similar efforts related to pre-innovative activities (CONSTPREINOV)
and to organizational upgrade (CONSTORGUPDATE), since these indicators show similar values0.26 and 0.28 respectively. Efforts related to acquisition of incorporated technologies are significant,
since this indicator (CONSTINCTECH) assumes relatively high value (0.33). For all the companies, the
training effort of the workforce (TRAINEEWORK) is also significant, with an indicator of 0.34, but the
effort to absorb skilled workers (ABSSKILLWOR) is more limited (0.15). These data indicate that the
firms give more importance to the collection, systematization and dissemination of information obtained
from various departments of the company. In parallel, there is a reasonable effort oriented to a
technological upgrade, mainly based on the purchase of machinery and equipment. In contrast,
conducting pre-innovative activities and organizational update tend to be less intense. Additionally, the
importance attributed to actions related to the training of the workforce is higher than that attributed to the
hire of more qualified workers.
Indicators of external learning and cooperative actions show that the main form of interaction
observed in the sample refers to vertical learning. From the value obtained by the indicator of external
vertical learning (LEARNVERT = 0.63), it appears that the information obtained from customers and
suppliers are very important to innovative firms. There is also a relatively high value of the indicator
related to horizontal learning based on information derived from competitors in the industry
(LEARNHOR = 0.41). It should be noted that the indicators that capture the importance attached to
information obtained from S&T institutions (LEARNS&T) and specialized services
(LEARNTECHSERV)) have the lowest values among the indicators related to external learning (0.13 and
0.21, respectively), demonstrating the limited priority attributed to these sources of information.
2

Production, marketing and sales, customers services

Regarding cooperation, the data reinforce the perception that it occurs in a limited scalefor the
sample as a whole, since all indicators relatedto this dimension assume valuesrelatively reduced. Two
forms of cooperation might be highlighted: vertical cooperation (COOPVER) related to cooperation with
suppliers and clients, with an average value of 0.21; and horizontal cooperation with competitors and
other agents in the industry (COOPHOR), with an average value of 0.16. Indicators of cooperation with
other agents (COOPOTHERS), with S&T institutions (COOPS&T) and with specialized services
(COOPTECHSERV) showed lower values (0.09, 0.07 and 0.06, respectively).
Considering these indicators as whole, general evidences about the pattern of external learning and
cooperative actions might be captured from the sample. Among the different forms of interaction, the
firms attach considerable importance to vertical relationships, giving priority to the exchange of
information with customers and suppliers and, to a lesser extent, to horizontal interactions with
competitors and others agents in the industry. However, the interactions developed with S&T institutions
and specialized services tend to belimited, suggesting a great difficulty to capture information and to
access complementary competences from these sources.
Table2 Descriptive statistics of the indicators d (N = 1.187).
Indicators

Mean

Min

Max

Indicators
of
innovative
performan
ce

Indicators of
local
externalities

Indicators of
external learning
and cooperation

Indicators of
innovative
efforts

Constancy in conducting R&D (CONSTR&D)


0,2193
0
1
Constancy in the acquisition of incorporated technologies (CONSTINCTECH)
0,3385
0
1
Constancy in pre-innovative efforts (CONSTPREINOV)
0,2677
0
1
Constancy in organizational update (CONSTORGUPDATE)
0,2888
0
1
Training of the workforce (TRAINEEWORK)
0,3437
0
1
Absorption of skilled workers (ABSSKILLWORK)
0,1538
0
1
Internal learning from R&D department (LEARNR&D)
0,2867
0
1
Internal Learning from other sources (LEARNORGFONT)
0,5943
0
1
Vertical Learning (LEARNVERT)
0,6331
0
1
Horizontal Learning (LEARNHOR)
0,4136
0
1
Learning through S&T institutions (LEARNS&T)
0,1359
0
1
Learning through technical services (LEARNTECHSERV)
0,2133
0
1
Learning through other agents (LEARNOTHERS)
0,2159
0
1
Vertical cooperation (COOPVER)
0,1625
0
1
Horizontal cooperation (COOPHOR)
0,0611
0
1
Cooperation with S&T institutions (COOPS&T)
0,0714
0
1
Cooperation with technical services (COOPTECHSERV)
0,0909
0
1
Local Subcontracting Networks (EXTERSUBCONTNET)
0,2700
0
1
Externalities of Skilled Workers (EXTERSKILLWORK)
0,6076
0
1
Externalities of Inputs to Production (EXTERINPUTS)
0,5665
0
1
Externalities of Equipments to Processes (EXTERNEQUIP)
0,4262
0
1
Sales Externalities (EXTERSALES)
0,5594
0
1
Externalities related to Technical Services (EXTERNTECHSERV)
0,6281
0
1
Local S&T Externalities (EXTERNS&T)
0,2938
0
1
Radical innovation in products (INRDPRD)
0,1449
0
1
Radical innovation in processes (INRDPRC)
0,1601
0
1
Incremental innovation in products (ININCPRD)
0,5687
0
1
Incremental innovation in (ININCPRC)
0,5366
0
1
Organizational innovations Type 1 (INORG1)
0,3095
0
1
Organizational innovations Type 2 (INORG2)
0,3787
0
1
Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs).

Standard
Deviation
0,3085
0,3392
0,3166
0,3446
0,2766
0,2360
0,4211
0,3553
0,3434
0,3528
0,2692
0,2591
0,3392
0,2844
0,1876
0,1748
0,1822
0,3223
0,2894
0,3191
0,3354
0,3658
0,2678
0,3518
0,2854
0,3668
0,4093
0,4989
0,3470
0,4351

Indicators of local externalities show that the firms attach considerable importance to externalities
of skilled workers (EXTERSKILLWORK = 0.60), as well as to the presence of local technical services
(EXTERNTECHSERV = 0.62).Local externalities related to the supply of raw materials, parts and
components (EXTERINPUTS = 0.56) are more relevant than those related to the supply of machinery and
equipment (EXTERNEQUIP = 0.42). Local externalities can also be associated with the increase of the
sales, since the average of this indicator is also significant (EXTERSALES= 0.55). On the other hand,
local infrastructure related to science and technology is considered deficient by the average of the firms in
the sample (EXTERNS&T = 0.29). The firms of the sample (on average) have a limited insertion in
subcontracting networks that operate at the local level, as indicated by the lowest value attributed to the
index related to this event (EXTERSUBCONTNET = 0.27).
Indicators related to innovative performance reflect the ability of the companies to introduce
different types of innovations. With regard to innovations in products and processes, it is clear that most
companies in the sample has a high capacity of imitation, with the indicators related to incremental
10

innovations in products (ININCPRD) and incremental innovations in processes (ININCPRC) reaching the
highest values (0.56 and 0.53, respectively). There is also a reasonable ability to introduce organizational
innovations, which tends to be higher for the introduction of organizational innovations related to
marketing practices (INORG2 = 0.37), vis--vis the introduction of changes in organizational structure
and the adoption of new management practices (INORG1 = 0.30).
However, indicators assume low values when they attempt to measure capabilities oriented to the
introduction of more "radical innovations. The indicator related to the introduction of new products for
domestic or international markets has a low value (INRDPRD = 0.14), reflecting the low ability of firms
to innovate in this field. The introduction of process innovations that are completely new to the sector has
also low values (INRDPRC = 0.16). The analysis of Table 3 indicates a general emphasis in imitative
innovations for the firms in the sample. For both types of innovation analyzed, most companies in the
sample are not innovative - 39.17% for product innovations and 55.52% for process innovation. When
companies innovate, the introduction of new products for the companies, but already existent on the
market, tends to be predominant (37.8%). The same occurs in terms of process innovations that are new
for the companies, but already existent in the sector (28.4%).
Table 3 - Distribution of the firms in the sample according to company size and type of innovation
(N=1.187)
Micro
Small
Medium
Total
N
%
N
%
N
%
N
%
Non innovative
314
44,67%
130
33,77%
21
21,21%
465
39,17%
Product Innovation
New to the firm
265
37,70%
151
39,22%
33
33,33%
449
37,83%
New to the market
124
17,64%
104
27,01%
45
45,45%
273
23,00%
Non innovative
443
63,02%
179
46,49%
37
37,37%
659
55,52%
Process Innovation
New to the firm
172
24,47%
139
36,10%
27
27,27%
338
28,48%
New to the sector
88
12,52%
67
17,40%
35
35,35%
190
16,01%
Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs).
Size / Type of innovation

The broader picture provided by the analysis of all indicators indicate that the firms in the sample
concentrate their technological efforts in the systematization of internal learning. The most relevant
interactive actions occur with other productive agents, especially with customers and suppliers and, to a
lesser extent, with competitors and other companies. These companies have a huge capacity to imitate
products and processes and to introduce innovations in their organizational structures, but a reduced
capacity to implement intensive innovations - new products for the market or new processes to the
industry. The combination of efforts and interactions based on this pattern reinforces the imitative
character assumed by the innovative performance. However, to develop dynamic capabilities required to
implement intensive innovations, efforts to acquire new technologies, to develop pre-innovative activities,
to absorb qualified personnel and to interact on a larger scale with S&T infrastructure might be necessary.
Another aspect concerns the heterogeneity observed not only at the sectoral level, but also at the
firm level of the sample. This trend reflects the fact that a large number of the indicators have a standard
deviation expressively higher or very close to its mean, indicating that the behaviors of the firms tend to
be significantly distinct in relation to the dimensions captured by those indicators. This feature suggests
the existence of distinct patterns concerning the links between innovative efforts, learning mechanisms,
cooperation practices and innovative performance. To capture this diversity, the application of
Multivariate Analysis seems particularly useful.

11

3.2. Learning and cooperation in LPAs: empirical evidences


The analysis developed in the previous section identified the general characteristics of the sample
in relation to the indicators used, which could be capture by its means. We have also verified the high
heterogeneity of the sample, with the standard deviation being higher than the average value of the
indicators, indicating that companies differ significantly in relation to the dimensions analyzed. Due to
these characteristics, statistical techniques are applied to identify general patterns between companies and
the processes. First, a factor analysis is applied, seeking to systematize and reduce the dimensions to be
investigated. Subsequently, a cluster analysis tries to clarify general patterns in the sample, resulting in
the identification of groups of firms with similar behavior in terms of the dimensions analyzed.
Starting from the calculated indicators, we sought to develop a factor analysis based on the
principal component method, using the criteria of normalized varimax3for each subset of indicators. The
main purpose of factor analysis is to capture and describe the covariance relationships among many
variables in terms of a few factors not directly observable. Therefore, factor analysis could identify the
main factors and the weight of the variables for each factor, characterizing the behavior of the sample in
relation to these factors. In this sense, Table 4 shows the eigenvalues related to each factor and the
percentage of data variation explained by each subset of indicators.
We opted to apply four factor analyzes separately, one for each subset of indicators. Concerning
"innovative efforts" four (4) factors were selected, which together explain 78.5% of the variance of the
data. Concerning "external learning and cooperative actions, four (4) factors were selected which
represent 75.8% of the data variation. Regarding indicators of local externalities three (3) factors were
selected, which together express 74.5% of the variance of the data. With respect to "innovative
performance", three (3) factors were selected which together explain 74% of the variations of the data.
The definition of the number of factors for each analysis considers the total explained variance of the data,
which was located in the range between, 74% and 78%.
Table 4 - Eigen values and variance related to selected factors (N = 1.187)
% of total
% of cumulative
Eigenvalue
variance
variance
Accumulated
explained
explained
Factor 1
3,8005
47,5064
3,8005
47,5064
Factor 2
0,9445
11,8063
4,7450
59,3127
Innovative efforts
Factor 3
0,7910
9,8869
5,5360
69,1996
Factor 4
0,7445
9,3067
6,2805
78,5063
Factor 1
3,7185
41,3162
3,7185
41,3162
Factor 2
1,2737
14,1520
4,9921
55,4682
External learning and
cooperative actions
Factor 3
1,1008
12,2313
6,0930
67,6995
Factor 4
0,7314
8,1265
6,8243
75,8261
Factor 1
2,4466
40,7759
2,4466
40,7759
Local Externalities
Factor 2
1,0800
17,9997
3,5265
58,7756
Factor 3
0,9442
15,7368
4,4707
74,5124
Factor 1
2,6121
43,5351
2,6121
43,5351
Factor 2
0,9842
16,4039
3,5963
59,9391
Innovative Performance o
Factor 3
0,8456
14,0930
4,4419
74,0321
Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Systems
Subsets of Indicators

Factor

Eigenvalue

Table 5 summarizes the information generated by the analysis of the factor loadings matrix. The
table presents the value of the factor load associated with the most relevant indicators for each factor.
Concerning the indicators of innovative effort, the Factor 1 may be called "innovative activities",
comprising indicators related to the constancy in the acquisition of incorporated technologies
(CONSTINCTECH), constancy in organizational update (CONSTORGUPDATE) and constancy in
pre-innovative efforts (CONSTPREINOV). The Factor 2 may be referred to the "R&D Factor
comprising the internal learning from R&D department (LEARNINTR&D) and the constancy in
conducting R&D (CONSTR&D). The Factor 3 is influenced only by the high intensity of one indicator
3

This method is more usual and more accurate since it promotes the rotation of the orthogonal axis related to the factors and
variables (indicators) in order to reach the best result concerning the frame of the indicators in the respective factors.

12

the "internal learning from other sources" (LEARNORGFONT) and can be entitled "Internal Learning
Factor. Finally, the Factor 4 groups the subset of indicators related to the absorption of skilled workers
(ABSSKILLWORK) and to the training of the workforce (TRAINEEWORK), so this factor is defined
as "Development and Training of the Workforce Factor".
For the subset of indicators of external learning and cooperative actions, four factors were selected.
The first factor is influenced by larger scale indicators that represent cooperation vertical cooperation
(COOPVER), horizontal cooperation (COOPHOR), cooperation with technical services
(COOPTECHSERV) and cooperation with other agents (COOPOTHERS) and can be called
"Cooperation Factor". The second factor comprises vertical learning (LEARNVER) and horizontal
learning (LEARNHOR), so this factor may be titled Learning with Productive Agents. The third factor
may be titled "Interaction with S&T institutions, including indicators related to learning with S&T
institutions (LEARNS&T) and cooperation with S&T institutions (COOPS&T). The fourth factor is
strongly influenced by the indicator of learning through technical services (LEARNTECHSERV) and
can be called Learning through t specialized services.
Concerning the local externalities, the indicators more directly related to the sphere of the
production - namely, the externalities of inputs to production (EXTERINPUTS), the externalities of
equipment (EXTERNEQUIP) and externalities related to technical services (EXTERNTECHSERV) are grouped in the first factor, which can be named as "Production Externalities. The second factor
groups the indicators related to two dimensions externalities of skilled workers
(EXTERSKILLWORK) and S&T Externalities (EXTERNS&T) - and can be called Workforce and
S&T Externalities ". The third factor is influenced only by one indicator sales externalities
(EXTERSALE) - and can be called "Sales Externalities.
Table 5 - Summary of extracted factors for the subsets of indicators (N = 1187)

Innovative efforts
Factor 1 - Innovative activities
Constancy in the acquisition of incorporated technologies

(CONSTINCTECH) 0,80

Constancy in organizational update


(CONSTORGUPDATE) 0,76

Constancy in pre-innovative efforts


(CONSTPREINOV) 0,71
Factor 3 - Internal Learning
Factor 4 Absorption and Training of the Workforce
Internal Learning from other sources (LEARNORGFONT)

Absorption of skilled workers (ABSSKILLWORK) 0,81


0,93

Training of the workforce (TRAINEEWORK) 0,77


External learning and cooperation
Factor1 - Cooperation
Factor 2 - Learning with productive agents
Horizontal Learning (LEARNHOR) 0,78

Horizontal Learning (LEARNHOR) 0,84


Cooperation with other agents (COOPOTHERS)

Vertical Learning (LEARNVERT) 0,79


0,78
Vertical cooperation (COOPVER) 0,70

Cooperation with technical services


(COOPTECHSERV) 0,70
Factor 3 - Interaction with S&T

Learning through S&T institutions (LEARNS&T)


0,85

Factor 2 - R&D
Internal learning from R&D department (LEARNR&D)
0,86
Constancy in conducting R&D (CONSTR&D) 0,80

Factor 4 - Learning through technical services

Learning through technical services


(LEARNTECHSERV) 0,81

Cooperation with S&T institutions (COOPS&T)


0,82

Factor 1 - Production externalities


Externalities of Inputs to Production
(EXTERINPUTS) 0,88
Externalities of Inputs to Production
(EXTERINPUTS) 0,84
Externalities related to Technical
Services (EXTERNTECHSERV) 0,63
Factor 1 - Organizational innovations

Organizational innovations
Type 2 (INORG2) 0,84

Organizational innovations
Type 1 (INORG1) 0,83

Local Externalities
Factor 2 - Workforce and S&T Externalities

Local S&T Externalities


(EXTERNS&T) 0,88

Externalities of Skilled

Factor 3 - Sales Externalities

Sales Externalities
(EXTERSALES) 0,93

Workers
(EXTERSKILLWORK) 0,58
Innovative Performance
Factor 2 - Radical Innovations

Factor 3 - Incremental Innovations

Radical innovation in processes

Incremental innovation in
(INRDPRC) 0,83
products (ININCPRD) 0,92

Radical innovation in products

Incremental innovation in
(INRDPRD) 0,81
process(ININCPRC) 0,61
Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs)

Regarding the subset of indicators related to innovative performance, Factor 1 represents the
two indicators related to the implementation of organizational innovations comprising new production
13

methods (INORG1) and new concepts and practices of marketing (INORG2) - being called as
"Organizational Innovations. The indicators related to radical innovation in products (INRDPRD) and
radical innovation in processes (INRDPRC), are grouped in Factor 2, which could be called "Radical
Innovations". The Factor 3 groups the indicators of incremental innovations in products (ININCPRD) and
processes (ININCPRC) and may be entitled as "Incremental Innovations".
Regarding the factor loadings, with the exception of the indicators proposed for each factor, the
others have less influence on their behavior. A similar trend is observed for indicators that have an
inverse relationship with the factors, which is generally very low (less than -0.05), slightly influencing the
final value of the factor. Considering the factor scores4 related to the sampled companies, a comparative
analysis was developed based on a reduction of the number of variables, which allows the formation of
clusters grouping the companies with similar characteristics in terms of the factors identified. It is
expected that theses clusters reflect similar patterns with respect to innovative efforts, external learning,
cooperative actions, local externalities and innovative performance.
The cluster analysis was applied to identify distinct groups of companies with similar
characteristics in terms of the factors identified. An additional indicator related to the participation in
local subcontracting networks was also included in the analysis 5 . Cluster analysis evaluates a set of
interdependent relationships between the data, without any distinction between dependent and
independent variables. It enables the classification of the objects in this case, the firms in the sample - in
relatively homogeneous groups, based on the set of variables or on the set of factors generated by
Multivariate Analysis. These procedures allow, at first, to reduce the dimensions of the analysis from
thirty (30) original indicators to fourteen (14) underlying factors. Secondly, based on the application of
these factors for the 1.187 companies in the sample and on the subsequent use of cluster analysis
techniques, four (4) groups of companies with similar characteristics were identified. The next step tries
to identify the specific characteristics of these groups in terms of the dimensions analyzed, stressing the
influence of these factors to the performance of the innovative companies with a higher degree of
accuracy.
3.3- Similar Firms inserted in LPAs: the characteristics of the clusters
The analysis developed identified four groups (clusters) of firms with similar characteristics from
the factors previously generated. Figure 3 shows the values of the factors identified for each cluster. For
the entire sample, the mean of a given factor is always zero (0) and its standard deviation is equal to one
(1) 6 . Table 6 summarizes the specific features identified for the four distinct clusters. The main
characteristics of each cluster are subsequently discussed.

The score comes from the factorial coefficients related to each indicator. The factor coefficients are multiplied by each
indicator of the companies, obtaining a final value equivalent to the individual factor score of each company.
5
For operational reasons, the indicator related to the insertion in local subcontracting networks was not included in the factor
analysis developed for the group of indicators related to local externalities. It was added in the second stage of analysis and
does not refer to any factor. To ensure uniformity of the analysis, this indicator was standardized for the sample companies,
6
In the Factor Analysis the value obtained by a given factor - associated to a group of companies with similar characteristics can only be analyzed in comparison with the entire sample mean or with other groups with distinct characteristics.

14

Figure 3 Mean of the factors for the firms inserted in the clusters identified

Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs)

Table 6 Characteristics of the clusters


Characteristics
/ Clusters

Cluster 1 (386 firms)


Medium / Low
Innovation-Intensive

Cluster 2 (352 firms)


Low InnovationIntensive

Cluster 3 (266 firms)


Medium / High
Innovation-Intensive

Innovative Efforts

Medium intensity and related


to internal learning

Occurs with very limited


intensity

Occurs with high intensity


being strongly associated with
informal innovation activities

External Learning
and Cooperation

Related solely to productive


agents

Almost inexistent

Development of cooperative
actions. Interaction with
specialized services and to a
lesser extent with other
productive agents

Local Externalities
and networking

Externalities in the orbit of


production activities and
presence of relatively
structured networks

Relevant only to improve the


companies sales. Limited in
the sphere of production.
Lack of structured networks

Externalities have medium


relevance. Structuring of
technical-productive networks
at the local level.

Innovative
Performance

Medium and based on the


introduction of innovative
products and processes

Low for all types of


innovations

Medium-high. Predominance
of organizational innovations
and, to a lesser extent,
incremental innovations

Cluster 4 (183 firms s)


High InnovationIntensive
Occurs with high intensity being
associated with R&D and with
the training and absorption of
skilled workers
Interaction with S & T
institutions and to a lesser extent
with specialized services.
Cooperation and interaction
with productive agents
Externalities and insertion in
local networks have low /
medium importance, except
those related to labor training
and to the presence of
universities and research
institutions
High, being associated with the
introduction of radical
innovations and, to a lesser
extent, organizational and
incremental innovations

Source: Micro data collected at the firm level. Information extracted from REDESIST Database on Local Productive Arrangements (LPAs)

The first group of companies identified (cluster 1) comprises 386 companies, from which 68.6%%
are micro enterprises, 29% are small enterprises and 2.3% are medium enterprises 7 . Concerning the
activities developed by these firms, there is a predominance of the textile and garment companies, with
46.37% of the firms in the cluster, and furniture and wood, with 34.71%. The remaining 19% would be
divided between others activities. These companies had an average revenue around R$ 900.000 in the
year 2003and had employed on average 25 employees. They present a low intensity concerning the
development of innovative efforts. These efforts comprise mainly internal learning, which occurs more
intensively than in the other clusters, and, to a lesser extent, R&D activities, which equals the intensity of
the sample mean. Concerning external learning and cooperative actions, these companies are intensively
involved with productive agents, but other forms of interaction are very restricted. Externalities directly
related to the orbit of production are very important, as well as other local sources of externalities, but
there is a low integration in technical-productive networks, reflecting a more passive attitude towards
those aspects. The innovative performance of these companies is strongly oriented to the introduction of
incremental innovations in products and processes, presenting a higher score attributed to these aspects
7

Comprising the size of the firms, four ranges was considered: 1) Micro enterprises (up to 19 employees), 2) Small enterprises
(between 20-99 employees) 3) Medium-sized enterprises (between 100-499 employees); 4) Large enterprises (more than 500
employees).

15

when compared with other clusters. On the other hand, the introduction of organizational innovations and
of radical innovation occurs on a very small scale. Considering these characteristics, this cluster might be
characterized as having a "medium- low innovative dynamism.
The second cluster (cluster 2) comprises 352 companies from which 256 (72%) consist of small
firms, 84 (23%) of small firms and 12 (3%) of medium-sized enterprises. These firms employed on
average 23 employees in 2003, with revenues of around R$ 1.2 million. Regarding the activities
performed, 44% of the firms work in the textile, clothing and footwear industry, 19% in furniture and
wood, 14% in fishing, 10% in software and computer and the remaining 14% distributed among a wide
spectrum of activities. Compared to the sample, these companies are not very involved with innovation
efforts, and even the information obtained from the various departments of the firms (characterizing
internal learning) are restricted used, which would be reflected in the scores attributed to this factor, the
lowest value compared with other clusters. The limited strategies related to external learning and
cooperative actions illustrate this aspect. Concerning factors related to local externalities and to the
insertion in subcontracting networks, the values assumed by the factors are somewhat higher.
Externalities directly related to the production have a value close to the sample mean and the externalities
related to local sales are high. This last feature shows that companies of this group concentrate its sales in
regions where they are located. On the other hand, participation in subcontracting networks is very low,
reflecting the fact that local sales do not involve inter-industry trade. Regarding the innovative
performance, these companies tend to generate different types of innovations in a very small scale. Due to
these characteristics - namely, the lower values in most of the factors related to innovative performance
when compared with other clusters and with the mean of the sample - the firms in cluster 2 can be
characterized as presenting "low innovative dynamism.
The third cluster (cluster 3) includes266 companies, from which 36% are micro enterprises, 46%
small enterprises and 16% medium-sized enterprises. These companies operate in activities related to
metal mechanics, equipment and components (35%), textiles and clothing (33%) and furniture and wood
(16%). Their sales revenue reached R$ 2.4 million on average in 2003 and they had employed 58
employees on average per company. Compared with other clusters, these companies carry intensive
innovative activities and, despite having R&D on a limited scale, they try very hard to exploit the
potential of internal learning. They also develop actions related to the absorption and training of skilled
workers at a high scale. The cooperative actions and learning related to specialized services are also very
significant in this cluster, as well as learning with productive agents. However, interactive actions with
S&T institutions are virtually nonexistent for this group of companies. Concerning the set of local
externalities analyzed in this study, companies in this cluster attach similar importance when compared
with the sample mean. However, these companies are the ones in the sample that participate more
intensively in subcontracting networks. They have high capacity to innovate and introduce organizational
innovations incrementally in products and processes on a scale considerably high. The introduction of
more radical innovations, however, is still restricted. The performing of innovative efforts, the
involvement with learning and cooperation and the integration to more complex networks are factors that
generate capabilities to innovate for these firms. However, the R&D activities are still performed on a
smaller scale, as well as the interaction with S&T institutions, limiting the possibilities of implementing
more virtuous innovation processes. Due to these characteristics, the firms in this cluster can be
characterized as presenting "medium-high technological dynamism.
The fourth cluster (cluster 4) involves 183 companies from which 46% are micro enterprises, 34%
are small enterprises and 18% are medium enterprises. In 2003, those firms earned on average R$ 3.1
million with 59 employees each. These companies are concentrated in computing and software-related
activities (33%), metal mechanics, equipment and components (33%), textiles and clothing (8.5%),
biotechnology (7.5%), furniture (6%) and other activities (10 %). Their innovative efforts are
concentrated mainly in conducting R&D and in training skilled workers. Innovative activities occur in
medium-high scale for these firms. They also establish deeper interactions with S&T institutions, when
compared with other clusters extracted from the sample. Factors related to learning processes and external
cooperation also have relatively high values, significantly above the average of the sample. The more
important local externalities identified by this group are related to availability of skilled workers and of
16

S&T infrastructure. Concerning the innovative performance, these companies are those that introduce
more intensive innovations in products and processes. They also innovate in their organizational
structures at medium-high intensity. Therefore, innovative efforts are higher, with an emphasis on high
interaction with S&T institutions and intensive R&D activities, leading them to develop strong
capabilities to implement intensive innovations, both in products and processes. Due to these
characteristics, the pattern of firms in this cluster reflects a situation of "high technological dynamism.
4- Concluding Remarks
The analysis indicates that concepts originally developed by the academic sphere can contribute
significantly to the improvement of regional innovation policies. Specifically, it deals with the
transference of concepts and competences from the academy to the policy sphere, identifying how those
concepts have been incorporated in the repertory of skills and practices of the policy makers. The analysis
points out some relevant trends that guide this transference. A relevant aspect refers to the provision of a
conceptual and methodological basis for these policies. This conceptual base can be important to enable
the implementation of actions in contexts and realities very heterogeneous in terms of their productive,
social and territorial environment, as in the case of policies to support LPAs. In this sense, the
mobilization of LPAs of different sizes and types seems to be a way to stimulate a better regional
distribution of economic activities, as well as to mitigate social and territorial inequalities.
A second important aspect highlighted by the analysis refers to the academic contribution to the
training of a qualified staff responsible for the implementation of innovation policies in different fields.
Situations in which personnel originated from academia assume an important executive role in the
implementation of these policies were identified in the case of policies to support LPAs, when researchers
originated by the academy assume an executive role in public organizations such as BNDES and the
Ministry of National Integration. Another important contribution refers to the role of communities of
experts originated from the academic sphere that come to play a systematic advisory role to the
formulation of innovation policies. Concerning policies to LPAs, we can mention the role of the Brazilian
Conference of Local Productive Arrangements held annually. In these Conferences, academic agents took
part with members of public bodies in various forums, thematic groups and advisory boards, much of
them articulated to the regional nuclei of the GTP - APL. A further contribution refers specifically to the
training of policy-makers, public managers and other staff responsible for operate concrete actions based
on that framework.
A third aspect that seems to be very important concerns the establishment of channels of
communication between the academic sphere and policy-makers, allowing an exchange of information
and experiences in order to promote policy adjustments, increasing or reducing the scope of actions
according to the results generated and to the responses of the target audience of the policies. Concerning
the support to LPAs, this adjustment might be articulate to the ability to evolve along different
"generations" of policies, through an adjustment of the focus according to the feed-back obtained from
the practical experience gained in policies operation. This adjustment reflects a process of "institutional
interactive learning" that mobilizes and articulates the academic sphere and policy-makers instances,
which can improve the effectiveness of the innovation policies. These processes seem to have a direct
impact on the consolidation of a more analytically grounded policy, better adapted to the reality on which
it intends to intervene and with a greater capacity to meet specific needs of the agents affected, becoming
more lasting and sustainable.
The analysis developed also tries to characterize the behavior of firms inserted in LPAs,
identifying the importance of interaction with other agents and how they are affected by externalities
associated with the local environment. Specifically, we sought to identify the dimensions that influence
the introduction of innovations by firms included in different LPAs. Four clusters of firms with similar
patterns were identified. In this sense, the innovative dynamics of a LPA would be influenced by the
relative participation of companies in each one of those clusters. This work is part of a broader research
program that seeks to identify and analyze indicators related to "innovative dynamics" of local productive
systems. To go beyond on the research agenda some additional steps are needed, such as a more detailed
17

analysis of the influence of different territorial specificities in the dynamics of innovative processes. It
also seems very important to advance in a better understanding of the learning processes and cooperative
practices at the level of structured industrial agglomerations, particularly through more detailed empirical
analysis. Another natural extension of the analysis consists in the use of econometric techniques to
evaluate possible relationships between the factors identified.
Annex - Table A Description of the Indicators.
Indicators

Events
1) Indicators of innovative efforts
Constancy in conducting R&D
Performing R&D within the firm; external acquisition of
(CONSTR&D)
R&D.
Acquisition of machinery and equipment that imply
Constancy in the acquisition of
significant technological improvement; acquisition of other
incorporated technologies
technologies (incorporated in software, licenses, patents,
(CONSTINCTECH)
trademarks and trade secrets).
Industrial project related to products / processes either
technologically new or significantly improved; and training
Constancy in pre-innovative efforts
program associated to the introduction of products /
(CONSTPREINOV)
processes either technologically new or significantly
improved.
Programs oriented to quality management or organizational
Constancy in organizational update
modernization; new forms of commercialization or
(CONSTORGUPDATE)
distribution of products either new or significantly
improved.

Indicators

Events

2) Indicators of external learning and cooperation


Vertical Learning (LEARNVERT)

Information absorbed from inputs suppliers (equipment, raw


materials); and customers.

Horizontal Learning (LEARNHOR)

Information absorbed from competitors and other firms in


the sector.

Learning through S&T institutions


(LEARNS&T)

Information absorbed from Universities; and Research


Institutions.

Learning through technical services


(LEARNTECHSERV)

Information absorbed from Centers of Professional training,


technical assistance and maintenance; Information absorbed
from Laboratories of tests and certification; Information
absorbed from Consulting Enterprises.

Training of the workforce


(TRAINEEWORK)

Training within the firm; training in local technical courses;


training in technical courses outside the cluster; training at
either supplier or customer firms.

Learning through other agents


(LEARNOTHERS)

Licenses patents and know-how; Conferences, seminars,


courses and technical publications; Fairs, Exhibitions and
Shops; leisure meetings; local business associations; and
information based on Internet or digital networks.

Absorption of skilled workers


(ABSSKILLWORK)

Hiring of technicians / engineers from other firms in the


cluster; hiring of technicians / engineers from firms outside
the cluster; absorption of graduates from universities.

Vertical cooperation (COOPVER)

Cooperation with Inputs suppliers (equipment, materials,


components and software); and Customers.

Internal learning from R&D


department (LEARNR&D)

Department of R&D as a relevant source of information for


innovation.

Horizontal cooperation (COOPHOR)

Cooperation with Competitors; and other firms of the


sector.

Cooperation with S&T institutions


(COOPS&T)

Cooperation with Universities; and Research institutes.

Internal Learning from other sources


(LEARNORGFONT)

Production area; Sales and marketing area and customer


service.

3) Indicators of local productive externalities


Externalities of Skilled Workers
(EXTERSKILLWORK)
Externalities of Inputs to Production
(EXTERINPUTS)
Externalities of Equipments to
Processes (EXTERNEQUIP)
Sales Externalities (EXTERSALES)

Externalities related to Technical


Services (EXTERNTECHSERV)
Local S&T Externalities
(EXTERNS&T)
Local Subcontracting Networks
(EXTERSUBCONTNET)

Cooperation with Centers of Professional training, technical


and maintenance assistance; Cooperation with laboratories
of tests and certifications and with consulting enterprises.
Cooperation with other agents
Cooperation with representation; trade union entities;
(COOPOTHERS)
bodies of support and promotion; and financing agents.
4) Indicators of innovative performance
Cooperation with technical services
(COOPTECHSERV)

Availability of skilled labor, low local labor costs


Proximity to suppliers of inputs and raw materials;
Acquisition of raw materials; Acquisition of components
and parts.
Proximity to equipment suppliers and acquisition of
equipment
Proximity to customers / consumers; Strong sales of
products to local markets
Physical infrastructure (energy, transport, communications),
availability of technical expertise and services, purchases of
local specialized services (maintenance, marketing, etc..).
Proximity to universities and research centers.

Multiple subcontracting links among local business:

Radical innovation in products


(INRDPRD)

New product to the international market; new product to the


national market.

Radical innovation in processes


(INRDPRC)

New process for the sector.

Incremental innovation in products


(ININCPRD)

New product for the firm, although existing in the market;


Innovation in design of products; creation of substantial
improvement of packaging

Incremental innovation in (ININCPRC)

New technological processes for the firm, although existing


in the sector.

Organizational innovations Type 1


(INORG1)
Organizational innovations Type 2
(INORG2)

Advanced management techniques; changes in the


organizational structure; implementation of new methods
related to ISSO 9000/ 14000.
Changes in the concepts and/or practices of marketing;
changes in the concepts and/ or practices of
commercialization.

5. References
ARCHIBUGI, D. and LUNDVALL, B.-(2001) eds.The Globalizing Learning Economy. Oxford:
Oxford University Press
AUDRETSCH, D. B.(1995) Innovation and Industry Evolution (MIT Press, Cambridge).
AUDRETSCH, D. B.; FELDMAM, M. P.(2004) Knowledge spillover and the geography of innovation.
In: HENDERSON, Vernon; THISSE, Jaques-Franois (Ed.). Handbook of urban and regional
economics. Amsterdam: Elsevier, v. 4, cap. 3.
BASANT, R. (2002) Knowledge Flows and Industrial Clusters, A background paper commissioned by
the International Development Research Centre (IDRC), 2002.
BELL, M. and ALBU, M. (1999) Knowledge Systems and Technological Dynamism in Industrial
Clusters in Developing Countries, World Development, Vol. 27, No. 9, pp. 1715-1734,
CASSIOLATO, J. E.; LASTRES, H. M. M. (2008). Discussing innovation and development: converging
points between the Latin American school and the innovation Systems perspective? GLOBELICS
2008
CASSIOLATO, J. E.; LASTRES, H. M. M. e MACIEL, M. L. (eds) (2003) Systems of Innovation and
Development. Cheltenham: Edward Elgar.
18

CHRISTOPHERSON, S. MICHIE, J.; TYLER, P. (2010)Regional resilience: theoretical and empirical


perspectives, Cambridge Journal of Regions, Economy and Society, 3, 310,
CLARK, J., HUANG, H.I. and WALSH, J.P. (2010) Typology of Innovation Districts: What it Means
for Regional Resilience", Cambridge Journal of Regions, Economy and Society, 3.1: 121-137. 2010
COOKE, P. and MORGAN, K. (1998)The associational economy: firms, regions, and innovation. Oxford
[England]; New York: Oxford University Press, 1998
COOKE, P. (2001) Regional innovation systems, clusters, and the knowledge economy, Industrial and
Corporate Change, 10(4): 945-974, 2001
COOKE, P.; URANGA, M. G.; ETXEBARRIA, G. (1998) Regional systems of innovation: an
evolutionary perspective. Environment and Planning, v. 30, n.9, p.1563-1584
DALLURA, G., GALVAGNO, M., and DESTRI , A.M. (2012). Regional Innovation Systems: A
Literature Review. Business Systems Review, 1(1), 139156.
MOTHE, J.; PAQUET, G. (1998) Local and regional systems of innovation. Economics of science,
technology, and innovation ; v. 14., Boston, Mass.: Kluwer Academic Publishers.
DOLOREUX, D. (2002) What we should know about regional systems of innovation, Technology in
Society, 24: 24363.
HOWELLS, J. (1999), Regional systems of innovation? In: Archibugi, D., et al. (Eds.), Innovation
Policy in a Global Economy. Cambridge University Press, pp. 6793
HUDSON, R. (2010) Resilient regions in an uncertain world: wishful thinking or a practical reality?
Cambridge Journal of Regions, Economy and Society, 3: 1125.
IAMMARINO, S.; MCCANN, P. (2006) The structure and evolution of industrial clusters: Transactions,
technology and knowledge spillovers, Research Policy 35 10181036,
JOHNSON, B. e LUNDVALL, B. (1994) The Learning Economy, Journal of Industry Studies, Vol.1,
no 2, 23-42. December, 1994
KRUGMAN, P. (1995) Development, geography and economic theory. Cambridge: MIT
KRUGMAN, P. (1991) Increasing returns and economic geography. Journal of Political Economy, v. 99,
n. 3, p. 483-499
LASTRES, H e CASSIOLATO, J. E. (2005) Innovation systems and local productive arrangements: New
strategies to promote the generation, acquisition and diffusion of knowledge. Innovation:
management, policy & practice. Vol 7 Issues 2-3 april-august
LASTRES, H. (2011) O Novo Ciclo de Desenvolvimento e a Poltica para APLs ,5 Conferncia
Brasileira de Arranjos Produtivos Locais, Braslia
LASTRES, H. (2007) Avaliao das polticas de promoo de Arranjos Produtivos Locais no Brasil e
proposio de aes. Nota Tcnica CGEE. Braslia, 2007.
LASTRES, H.M.M; CASSIOLATO, J.E. (2000) From Clusters to Innovation Systems: cases from
Brazil, Paper prepared for the GDN2000 Second Annual Global Development Network
Conference: "Beyond Economics: Multidisciplinary Approaches to Development." Tokyo,
December 2000
MAILLAT, D. (1998)'From the Industrial District to the Innovative Milieu: Contribution to an Analysis
of Territorialised Productive Organisations.' Recherches Economiques de Louvain 64: 111-129,
1998.
MAILLAT, D. (1995) Milieux innovateurs et dynamique Territoriale. In: Rallet, A.;Torre, A. conomie
industrielle et conomie spatiale. Paris: Economica,1995.
MAILLAT, D.; KELBIR, L (1999). Learning region et systemes territoriaux de production. Revue d'
Conomie et Urbaine, v.3, p. 429-448, 1999
MARSHALL, A. (1890) Princpios de Economia. So Paulo: Abril Cultural, 1982. (Os Economistas;
Primeira edio: 1890.
MARTIN, R. L. and SUNLEY, P. (2006) Path dependence and regional economic evolution. Journal of
Economic Geography, 6: 395437.
MASKELL, P. and KELBIR, L. (2005). What Qualifies as a Cluster Theory?, DRUID Working Paper
No 05-09, Department of Industrial Economics and Strategy, Copenhagen Business School.

19

MASKELL, P. and MALMBERG, A. (1999) Localised learning and industrial competitiveness.


Cambridge Journal of Economics 23 (2): 167-186., 1999
MCCORMICK, D. (1999)African Enterprise Clusters and Industrialization: Theory and Reality,
World Development, 27(9): 153152.1999
MENZEL, M.P. and FORNAHL, D. (2009) Cluster life cyclesdimensions and rationales of cluster
evolution, Industrial and Corporate Change, Volume 19, Number 1, pp. 205238 , July 22
NADVI, K. and SCHMITZ, H. , (1994) "Industrial Clusters in Less Developed Countries: Review of
Experiences and Research Agenda", Discussion Paper No. 339, Institute of Development Studies,
University of Sussex. 1994
OECD (2007), Competitive Regional Clusters: National Policy Approaches, OECD, Paris
OECD (2001) Innovative Clusters: Drivers of National Innovation Systems. OECD, Paris. 2001
PANICCIA, I. (2005)Cutting through the Chaos: Toward a New Typology of Industrial Districts and
Clusters, in: Chapter XX Clusters in Regional Development. Critical reflections and
explorations, edited by B. Asheim, P. Cooke, and R. Martin, Routledge, .
PIKE, A., DAWLEY, S.; TOMANEY, J. (2010) Resilience, adaptation and adaptability. Cambridge
Journal of Regions, Economy and Society, 3: 5970.
PIORE, M. J and SABEL, C.F. (1984) The Second Industrial Divide. Possibilities for Prosperity, Basic
Books, Nova York. 1984
PYKE, F., BECATTINI, G.. SENGENBERGER, W. (1990) Industrial districts and inter-firm cooperation in Italy. Geneva: International Institute for Labour Studies
SCHMITZ, H. and. MUSYCK, B., (1995) Industrial Districts: Policy Lessons for Developing
Countries?, World Development, 22(6), pp. 889-910. 1995
SCHMITZ, H. (1997) Collective efficiency and increasing return. Brighton, UK: Institute of
Development Studies,. (Working paper, n. 50)
SCOTT, A. J.; STORPER, M. (1986) Production, work, territory (the geographical of industrial
capitalism). Boston: Allen & Unwin, 1986.
STORPER, M. (1996) Innovation as collective action: conventions, products and technologies. Industrial
Corporate Change, v.5, n.3, p.761-789.
WORLD BANK (2009) Clusters For Competitiveness: A Practical Guide & Policy Implications for
Developing Cluster Initiatives, February

20

You might also like