Foreign Ownership and Firm Productivity in Bangladesh Garment Sector

Hiau Looi Kee* May 2005

Abstract This paper studies the productivity advantage and spillover of FDI firms in Bangladesh garment sector. This is based on a newly collected exclusive firm level data, supported by a unique custom firm level export data. Firm productivity is first estimated from a firm production function, controlling for input endogeneity, selectivity, as well as firm and year fixed effects. Results show that FDI firms are on average 20 percent more productive than domestic firms. Moreover, there are statistical evidence suggesting that productivity spillover occurs such that domestic firms may benefit from the productivity increase in FDI firms. These findings support a more open FDI policy for the Bangladesh garment sector.

__________________________________________ * Development Research Group Trade, the World Bank, 1818 H ST NW (MSN MC3303), Washington, DC 20433, USA; Tel: (202) 473 4155; Fax: (202) 522 1159; E-mail: hlkee@worldbank.org; I thank the World Bank, CIDA and DFID for providing research funding. The findings, interpretations, and conclusions expressed in this paper are entirely those of the author, and do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent.

Introduction Conventional wisdoms have it that firms with foreign equity tend to be more productive. This could be due to the firm specific tangible assets such as exclusive technology and product designs, or the intangible know-how embodied in foreign equity such as marketing, networking and sourcing. Such assets may be more readily available in big multinational corporations (MNC). As such, being part of MNCs allow the local subsidiaries with foreign equity to gain access to these assets, which in turn make them to produce more output given the same level of inputs, and thus a higher level of total factor productivity (TFP) than the solely domestic owned firms. Such hypothesis has some empirical support based on samples of Venezuela manufacturing firms studied in Aiken and Harrison (AER, 1999) and Malaysia service sector firms in Kee (forthcoming). Unlike many developing countries such as Cambodia, Mauritius and Mongolia, where most of the garment firms are part of some larger multinational corporations in the form of foreign direction investment, less than 15 percent of Bangladesh garment firms have foreign equity. This is partly due to the industrial policies of Bangladesh in order to safe guard quota allocations of garment export to US to the domestic firms. Furthermore, foreign firms are allowed to invest in Bangladesh garment sector only if they locate the plants in the export processing zones, and are not competing with the subcontracting domestic firms supplying to the exporting firms who have quota access. Thus, almost all FDI firms export all of their products from Bangladesh. The objective of this paper is to study the potential productivity advantage of FDI firms operating in Bangladesh. In addition, this paper aims to identify the possible channels by which local firms may benefit from the FDI firms. We focus on the productivity spillover effects, beyond the physical presence of FDI firms.

The paper first presents an overview of the garment sector, in terms of industry structure and export performance. The paper proceeds to study the firm export performance according to a unique custom export data. By dissecting the firms in terms of the markets they participated, this paper is able to assess the productivity distribution among Bangladesh garment export. The main part of the paper focuses on estimating firm productivity by modifying the state of the art technique due to Olley and Pakes (Econometrica, 1996), to control for firm and year specific biases. We relate the estimated productivity, which is the level of output not explained by the level of inputs, to the ownership structure of the firms using between firm panel regression, controlling for industry, year, location fixed effects. It is shown that firms with foreign equity are on average 20 percent more productive than otherwise identical domestic firms. The productivity advantage of FDI firms is robust to age and export destinations. In addition, we relate the productivity performance of domestic firms to that of FDI firms and show that there are indeed positive and significant productivity spillovers. For every 10 percent increase in the productivity level of FDI in the industry, productivity of domestic firms increases by 1.4 percent. An Overview of Bangladesh Garment Sector According to data obtained from the Bangladesh Garment Manufacturers and Exporters Association (BGMEA) Members Directory 2004-2005, there are more than 4,000 firms operating in Bangladesh garment sector, of which 2,800 are in Dhaka area. Almost 65% of the firms are in the woven industry, 20% in the knitting industry with the sweater industry rounds up the remaining 15%. About 13% of the woven firms also engage in the knitting industry. These are usually the larger and more productive woven firms. Most of the garment firms in Bangladesh are locally own, with about 1% of them
3

operating in the export processing zones (EPZs) in Dhaka and Chittagong. Finally, more than 63% of EPZ firms have some foreign ownership, from countries such as South Korea and Hong Kong. The sector as a whole employ 2.1 million workers, with 53,000 workers in the firms with foreign ownership. Overall, firms in Dhaka are larger and more productive, relative to firms in Chittagong. In addition, firms in EPZs are the better firms than those out side of the EPZs. Finally, firms with foreign capital are the most productive of all firms: On average, FDI firms are larger, they hire more workers given the same number of machine. FDI firms are more capital intensive, they use less workers per machine given the same number of plant capacity. EPZ firms are also more capital intensive relative to non-EPZ firms. Given after taking into account the numbers of product variety, FDI firms are still on average larger in capacity than domestic firms. 76% of the FDI firms are in the woven industry.

Export Performance of Garment Sector The past few years have witnessed an expansion of Bangladesh garment export to the world market. In 1998, the total value of garment export from Bangladesh was about US$3.8 billion, it increased to US$4.2 billion in 2001 and settled at US$3.6 billion in 2003. This Information is obtained from the United Nations Comtrade Database according to the reporting of the Bangladesh government. Figure 1 presents the breakdown of the aggregate export of the Bangladesh garment sector by destinations, in 1998, 2001 and 2003. In both 1998 and 2001, the share of EU in Bangladesh garment

export was about 50 percent, closely followed by the US at 45 percent, while other countries, noticeably Canada, made up the remaining 5 percent of aggregate garment export. In 2003, the importance of EU further increased to 58 percent, while the share of the US dropped to 37 percent.

Figure 1: Breakdown of Garment Export

Total Garment Exports (HS Catagories 61 & 62)
2500.0 2117.1 2000.0 1874.0 1689.2 Million of US$ 1500.0 1339.5 1897.6 2081.7

EU USA Others

1000.0

500.0 220.8 0.0 1998 2001 2003 201.8 175.1

The surprising fall in the garment export to the US could be due to transshipment or misclassification of goods. Based on US custom data from the US International Trade Commission (USITC), garment export to US from Bangladesh in fact has been steadily climbing from US$1.5 billion in 1998 to US$1.8 billion in 2003. In 2004, the value of garment export from Bangladesh further increased to US$1.9 billion, which makes Bangladesh the 10th largest garment supplier for the US market. Figure 2 presents the values of garment imports of US from 1998 to 2004 by major exporting countries. In

2004, the top ten garment exporting countries to the US market and their market shares are China (16%), Mexico (10%), Hong Kong (5.8%), Honduras (4.1%), Vietnam (3.7%), Indonesia (3.6%), India (3.4%), Dominican Republic (3.1%), Guatemala (2.9%) and Bangladesh (2.8%). Figure 2: Breakdown of Major Garment Exporters in the US Market
US Garment Imports (HS Catagories 61 & 62)
12000

10000

8000 Million of US$

BDG CHN MEX HKG HON IND

6000

4000

2000

0 1998 1999 2000 2001 2002 2003 2004

We further use a firm level export data set obtained from the Textile Unit of the Export Promotion Board (EPB) of Bangladesh to analyze the export performance of the Bangladesh garment sector. This information is compiled from those firms that applied for Country of Origin Certificates in 2004. This certificate is often requested by the importing countries to verify the origins of the imported goods in order to grant trade preferences. In this firm level data set there are 2387 garment firms exporting in 2004. The total value of garment export is US$5.7 billion, with more than 400 million dozens of garment

exported. Overall 57 percent of garment export headed to the EU, 20 percent for the US and the remaining 23 percent went to the other countries such as Canada and Australia. Table 1 presents the breakdown of garment export volume by destinations.
Table 1: Garment Export by Destination, 2004 Description EU under GSP Others USA with quota USA without quota Total Quantity (dozen) 319,718,411 32,044,542 42,196,576 19,785,482 413,745,011 Value (US$) 3,244,562,889 1,306,109,811 976,267,029 159,150,271 5,686,090,000

In terms of the distribution of firms across different markets in 2004, there are 1967 firms exporting under GSP, mainly to the European market, 1039 firms exporting to the US, of which 709 export under quota allocations, and 1231 firms exporting to other countries. Figure 3 presents the distribution of firms by export destinations. Among these firms, 46 percent only supply to one market, 34 percent supply to two markets, 14 percent to three markets, and 5 percent to all four markets. This is clearly presented in Figure 4.

Figure 3: Number of Firms in Different Markets

Distribution of Firms by Markets

2500

2000 number of firms

1500

1000

500

0 US-quota US-no quota export destinations EU Other

Figure 4: Number of Firms vs. Number of Markets

Distribution of Firms by Number of Markets

1200 1000 800 600 400 200 0 1 2 3 4 number of export distinations

according to the number of export market the firms supply. It is very clear that EU is the most popular destination, especially among firms that have only one export market. Among the 1109 firms that only supply one market, nearly 850 firms concentrate on EU which is about 76 percent. The US market appears to be toughest to break in among this group of firms, less than 8 percent only export to the US with and without quota. For firms that supply two markets, both the EU and the others are the favorites. Together, they account for 80 percent of the markets among the 1640 firms that export to two markets. The US in quota market is popular for firms that export to more than 2 markets.

number of firms

Figure 5 presents the choice of export markets of Bangladesh garment exporters

Figure 5: Market Choice by Firms with Different Markets

Choice of Export Market

900 800 700 number of firms 600 500 400 300 200 100 0 US-quota US-no quota EU Others export destinations one market firms two market firms three market firms four market firms

In addition, according to Eaton, Kortum and Kramarz (AER, 2004) who study the export performance of French firms, the number of markets a firm supplies reflects the productivity and competitiveness of the firm in the world market. The above distribution of firms implies that more than 35 percent of Bangladesh garment exporters participate in world markets widely with at least 3 export destinations, and are thus very competitive. This is quite evidence in Figure 6, when we plot the unit value of garment export (left axis) and total export value (right axis) against the number of export destinations. Firms that export to more destinations tend to have higher average unit values and larger in size, with the former reflects better quality and the latter indicates greater scale economies, both signal higher productivity of the firms. The differences in unit values and total size among firms with different number of markets are statistically significant. Figure 6: Exporting and Productivity

10

Unit Value, Total Export by Number of Markets

45 40 35 unit value in US$ 30 25 20 15 10 5 0 1 2 3 4 number of export destinations unit value per dozen total value of export 9 8 total export in million of US$ 7 6 5 4 3 2 1 0

Preliminary Findings Based on Firm Survey Firm level survey was conducted from the period of November 2004 to April 2005, which covers a stratified random sample of 350 firms, which is about 10% of the total population of the garment firms currently operating in Bangladesh. After cleaning up the data to exclude outliers and firms with incomplete information, there are a total of 231 firms in the unbalanced final panel data set of 1026, from 1999 to 2003. In this unbalanced panel data set, the composition of sub-industries of knitwear, sweaters and woven is 24%, 8% and 68% respectively. Among the sampled firms, 13% have positive foreign equity, while the remaining 87% are purely domestic owned. Moreover, 15% of the sampled firms are in the Dhaka and Chittagong EPZs, 63% in Dhaka and 15% in Chittagong. Tables 2-4 and Figures 7-9 present the sample means of the key variables of the sub-industries of knitwear, sweaters and woven, by foreign versus domestic firms. It is

11

clear that FDI firms are in general larger in sales, in exports, they purchase more material inputs, including imported materials, they hire more employees, including production workers. FDI firms also have larger capital stock and investment. All these suggest that FDI firms are larger in scale and presumably more profitable and productive. To formally study the productivity superiority of FDI firms, and the possible productivity spillover to domestic firms, we will need to first estimate firm level productivity for the firm sample. The estimated firm productivity is then relate to the ownership of the firms, and the relationship between productivity of domestic and FDI firms in the same sub-industries will be statistically examined.

Table 2: Summary for Knitwear

12

Knitwear (Thousands US$) Domestic Firm FDI Firm sales 3050.894 5044.482 export 2951.962 5044.482 cost 2917.288 4195.379 material 2037.68 2888.019 imp material 1560.666 2569.798 employee 582.9438 996.2333 prod worker 501.5181 943.7 capital 2033.417 1510.171 investment 817.5825 79.3077

Figure 7: Summary for Knitwear

US$ Thousands 5044 5000 5044

Knitwear
Domestic Firm
4195

FDI Firm

4000

3051 3000

2952

2917

2888 2570

2038 2000 1561

2033

1510

1000 583

996

944 502

818

79 0 sales export cost material imp material employee prod worker capital investment

13

Table 3: Summary for Sweater Industry

Sweater (Thousands US$) Domestic Firm FDI Firm sales 2363.506 3603.465 export 2362.946 3603.465 cost 2141.488 3350.958 material 1435.532 2389.08 imp material 564.879 1811.852 employee 906.8947 1305.85 prod worker 859.6316 1214.75 capital 1002.338 4231.342 investment 215.6552 344.9167

Figure 8: Summary for Sweater Industry

US$ Thousands 4500

Sweaters
4231

4000 3603 3500 3603 3351

Domestic Firm FDI Firm

3000

2500

2364

2363 2141

2389

2000

1812

1500

1436 1306 1215 1002 860

1000 565 500

907

345 216

0 sales export cost material imp material employee prod worker capital investment

14

Table 4: Summary for Woven Industry

Woven (Thousands US$) Domestic Firm FDI Firm sales 2926.72 13900 export 2919.786 13900 cost 2587.394 12700 material 2015.82 9665.94 imp material 1590.774 8393.138 employee 600.5773 1893.183 prod worker 560.29 1790.3 capital 639.5217 5076.089 investment 57.77929 315.9224

Figure 9: Summary for Woven Industry

US$ Thousands

Woven
13900 Domestic Firm FDI Firm

14000

13900

12700 12000

10000

9666

8393 8000

6000 5076

4000 2927 2920 2587 2016 1591 601 0 sales export cost material imp material employee prod worker capital investment 1893 1790 640 58

2000

560

316

15

Estimating firm productivity To formally study the overall productivity of firms, we need to estimate firm production function, taking into account total factor usage per unit of output. In the firm survey we asked firms to provide the annual increase in the main product price and the main material input price. The firm level price information allows us to construct firm level price indexes of output and material, which we use to deflate sales and material costs to obtain real output and material level. We estimate the following production function, Yit = Ait Lit L M it
M

K it

ln Yit = ln Ait + L ln Lit + M ln M it + K ln K it , where i and t are the indexes for firm and year, respectively. In log, output, Y, is linearly related to labor, L, materials, M, and capital stock, K. Any part of Y that are not explained by the three factors of production are attributed to productivity, A, which varies by firms and years. In other words, if we regress lnY on lnL, lnM and lnK using ordinary least squares (OLS) estimation, the regression errors are the firms productivity, lnA. However, firms input choices are likely to be endogenous. How many workers to hire, how many unit of fabrics to purchase, and how many new machines to set up each year depends on the productivity of the firms, which is known to the firms, but not the researchers or economists. Such input endogeneity will bias OLS estimates of labor and materials upward. In addition, if larger and older firms tend to stay in business despite low productivity, will younger and smaller firms tend to quit easier, such entry/exit decision of the firm will bias OLS estimates of capital downward. To address input endogeneity bias and selectivity bias, we follow a 3-step nonlinear estimation methodology developed by Olley and Pakes (Econometrica, 1996).

16

Moreover, to control for any factors that are specific to the firms, such as fraudulent accounting practice, or years, such as economic downturns, that may bias our estimates that are beyond the Olley-Pakes correction, we also include firm and year fixed effects in our regressions. We modified the three stage nonlinear estimation of the above production function due to Olley and Pakes to include firm and year fixed effects. Furthermore, even that older firms are more likely to stay in business despite temporary down turn in business, we also control for firm age in the estimation. ln Yit = i + t + L ln Lit + M ln M it + K ln K it To control for input endogeneity, we first regress lnY on lnL, lnM, a full set of firm and year fixed effects and a 3rd order polynomial function of real investment and capital, which is used to control for the unobserved productivity. The estimated coefficients on labor and materials are consistent. Firms real investment, I, is obtained by deflating nominal investment from the firm survey by the GDP deflator of domestic fixed capital formation of Bangladesh in the respective years. Capital is constructed by summing real investment over the years using perpetual inventory method with an annual depreciation rate, , of 10 percent: K it = K it 1 (1 ) + I it I 1 K i 0 = Fi1 + i1 , 2 with initial capital stock being constructed using average between firms first year fixed asset, F, and the infinite sum series of investment prior to the first year, assuming that the growth rate of investment of 0 and depreciation rate of 10 percent. To obtain consistent coefficient estimate of capital, we first estimate the entry/exit decision of the firms using a Probit regression on a 3rd order polynomial function of

17

investment, capital and age, controlling for year, region and industry fixed effects. This regression yields the propensity for a firm to stay in business. We then regress ln Yit L ln Lit M ln M it , constructed using the consistent estimates of L and M from the 1st step, on age, capital, and a 3rd order polynomial function of propensity of survival and E ( ln Yit ) L ln Lit M ln M it . This last-stage nonlinear regression gives us consistent estimated coefficient on capital, K . Results of the regressions are reported in Table 5. Column (1) of Table 5 shows the OLS estimation with no correction on endogeneity, selectivity, firm or year fixed effects. These estimates are likely to be biased. Column (2) shows the within estimates with firm and year fixed effects. While these estimates are robust to factor such as location which is specific to a firm and macro economic climates which is specific to a year, year to year variation of productivity within firm will still bias our estimates. Column (3) reports the first stage Olley-Pakes procedure, where a 3rd order polynomial function of investment, capital and age is included, in addition to firm and year fixed effects, to control for within firm year to year changes in the unobserved firm productivity. This procedure corrects for input endogeneity, which reduces the upward bias relative to the OLS estimates. The consistent estimated coefficients for labor and materials are 0.25 and 0.72, respectively. Without correcting for selectivity, the estimated coefficient on capital is too low.

Table 5: Dependent Variable log of firm output

18

Materials

(1) OLS 0.688*** (0.037) 0.283*** (0.036) 0.025*** (0.008)

(2) Within 0.718*** (0.065) 0.240*** (0.086) 0.017 (0.022)

(3) Olley-Pakes 0.718*** (0.065) 0.250*** (0.088) 0.013 (0.248) -0.173 (0.316) 0.137 (0.111)

(4) Olley-Pakes 0.718*** (0.065) 0.250*** (0.088) 0.021* (0.011) 0.032* (0.019)

Labor

Capital

Age

Investment

Endogeneity correction 1
2

No

No

Yes

Yes

Selectivity correction No No No Yes Firm fixed effects No Yes Yes Yes Year fixed effects No Yes Yes Yes Observations 1027 1027 1027 795 Notes: Heteroskadasticity corrected white robust standard errors in parentheses.
1

A 3rd order polynomial function of age, capital and investment are included.

A 3rd order polynomial function of propensity to stay in business and the fitted output net of labor and capital are included.

Column (4) controls for selectivity bias by including a 3rd order polynomial function of the estimated survival probability and the net fitted output. The resulting estimated coefficient for capital is 0.02. All these coefficients are statistically significant, and are in line with the estimates in the literature. Finally, with the sum of the estimated coefficients of labor, capital and material equals to one, the production function in the garment sector is found be constant returns to scale. With these estimates, we constructed firm level productivity according to the following equations: ln Ait = ln Yit L ln Lit M ln M it K ln K it , Ait = exp( ln Yit L ln Lit M ln M it K ln K ).

19

Comparing firm productivity across all firms in all sub-industries and locations yields some interesting insights in terms of relative productivity of firms. When we compare different firms across different sub-industries, on average, knitwear firms are the most productive. An average knitwear firm has 10 percent higher productivity than a woven firm, and 17 percent more productive than a sweater firm. Figure 10 presents the distribution of the estimated firm productivity by the three sub-industries. In terms of locations, productivity of firms located in Dhaka-EPZ is the highest, follow by firms in Dhaka, Chittagong-EPZ and Chittagong. Figure 11 presents the distribution of firm productivity by location.

Figure 10: Distribution of firm productivity in different sub-industries

20

Firm Productivity by Sub-industries

2.75

2.7

2.65 log of TFP

2.6

2.55

2.5

2.45 Knitwear Sweater Woven

Figure 11: Distribution of firm productivity by location

Firm Productivity by Location
2.78 2.76 2.74 2.72 Log of TFP 2.7 2.68 2.66 2.64 2.62 2.6 2.58 Chittagong Chittagong-EPZ Dhaka Dhaka-EPZ Others

Comparing firm productivity from year to year within firms also sheds some

21

interesting new lights. On average, garment firms are 3 percent more productive in 2003 than in 1999. The improvement in productivity is especially clear for the sample of domestic firms -- on average, domestic firm productivity is 5.5 percent higher in 2003 than in 1999. Figure 12 presents the movement of firm productivity over time in the different sub-industries. It is clear that most of the improvements are driven by firms in the Sweater and Woven industries. These results purely reflect the growth in productivity within a given firm, and thus are not contaminated by the composition of firms in different industries. Such an increase in productivity within a firm suggests that there are some exogenous factors pushing firms to be more productive over time. We explore one such exogenous factors which is the productivity spillover effects of FDI firms. Figure 12: Productivity Growth of Domestic Firms by Sub-industries
Productivity Growth of Domestic Firms by sub-industries
2.75 2.7 2.65 2.6 2.55 2.5 2.45 2.4 1999 2000 2001 2002 2003 Knitwear Sweater Woven

Are FDI firms more productive? We relate the firm level productivity, Ait, to the ownership of the firms. As shown in

22

Figure 13, on average, productivity of firms with foreign equity are about 20 percent higher than purely domestically owned firms.

Figure 13: Productivity of Firms with Different Ownerships

Productivity vs Firm Ownership
20 total factor productivity (in unit of output) 18 16 14 12 10 8 6 4 2 0 Domestic firms FDI firms

What could have explained the 20 percent productivity advantage of FDI firms? Column (1) of Table (6) regress the estimated lnTFP of firms on a FDI indicator variable, controlling for industry, year and location fixed effects. This is to isolate the effect of foreign ownership from the influences of sub-industries, investment climate of the locations, and the macro economic shock in each year. Given that ownership seldom change within firms in our sample, between-firms variation in foreign ownership is used to identify the effect of FDI dummy on productivity. The result shows that a FDI firm is still about 20% more productive than a domestic firm in the same industry, location and

23

year. This shows that the effect of foreign equity on firm productivity is independent on the location of the firms, the sub-industry of the firms and the macro economic fluctuations. Columns (2) and (3) further include age and export destinations of the firms in both the between and the OLS regressions. It is clear that FDI firms do have a higher level of productivity, even after we take into account the export destinations and thus the potential demand shocks of the firms, as well as the experience of the firms as proxied by age. Moreover, the OLS results show that firms export to US tend to be more productive, which concurs our previous finding using firm export data from EPB. Columns (4) to (6) repeat the exercise by using the actual foreign equity share in the regressions instead of a FDI dummy variable. The results are strikingly similar. This could be because most of the FDI firms in Bangladesh garment sector have 100 percent foreign equity, only 7 FDI firms are jointly venture firms with foreign equity no less than 25 percent. Thus overall there is convincing and statistical significant evidence suggesting that FDI firms are more productive than otherwise identical domestic firms operating in Bangladesh. This result is robust after taking into account the effects of locations, subindustries, macro fluctuation, export destinations and experience.

Table 6: Dependent Variable log of TFP

24

FDI dummy variable

(1) Between 0.194* (0.111)

(2) Between 0.181* (0.098)

(3) OLS 0.245*** (0.088)

(4) Between

(5) Between

(6) OLS

Foreign equity share

0.208* (0.114) 0.001 (0.003) 0.234 (0.166) 0.144 (0.160) 0.000 (0.001) 0.229*** (0.063) 0.130** (0.053)

0.194* (0.101) 0.001 (0.003) 0.237 (0.166) 0.144 (0.160)

0.256*** (0.089) 0.000 (0.001) 0.234*** (0.064) 0.132** (0.053)

Age

Export share of US

Export share of EU

Region fixed effects Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Industry fixed effects Yes Yes Yes Yes Yes Yes Observations 1027 1013 1013 1027 1013 1013 Notes: Asymptotic standard errors in parentheses in Columns (1), (2), (4) and (5). Heteroskadasticity corrected white robust standard errors in parentheses in Columns (3) and (6). Total number of firms in the unbalanced panel is 232 in Columns (1) and (4), and 227 for the rest . Dependent variable is constructed based on Column (4) of Table 5.

Productivity Spillover: Can Domestic Firms Benefit from FDI Firms? Many countries provide special incentives such as tax holidays or subsidies, and import duty exemptions to attract FDI, with the assumptions that the presence of FDI will benefit domestic economy through the some unmeasured spillover effects. To date, there is evidence of vertical spillover effects through the contact of domestic upstream suppliers to the downstream FDI firms (Javorcik, AER, 2004), evidence of horizontal spillover effects however have been quite elusive. To study whether such effects exists in Bangladeshs garment sector, we first relate the estimated TFP of the domestic firms to the presence of FDI firms in the subindustries. Presence of FDI firms in sub-industry j, FPjt, is captured by the share of employment of FDI firms collectively in the sub-industries in a given year, adjusted by the

25

percentage of foreign ownership of FDI firms, FSit, for all firm i in sub-industry j. This measure of the influence of FDI firms has been used in the literature (Aitken and Harrison, AER, 1999).

FPjt

L * FS = L
i j it i j it

it

In addition, we further relate the estimated TFP of domestic firms to the TFP of FDI firms in the same sub-industry and year. In order to capture the economic influence of the productivity of FDI firms, we weight the TFP of FDI firms with the share of foreign equity and the share of employment in the industry. Weighting by capital or output would not change the results.

ln A

FDI jt

L
i j

it

* FS it * ln Ait

L
i j

it

FDI jt

L
i j

it

* FS it * Ait

L
i j

it

Given that both the presence of FDI in the industry and the productivity of FDI firms in the industry do not vary within each firm observation, and are specific to each industry-year, we have aggregate variables in micro unit, which will artificially deflate the standard errors of the firm level panel regression (Moulton, RESTAT, 1990). We correct for such problem nonparametrically by clustering the standard errors of the regressions by industry-year. Table 7 presents the regression results. Column (1) shows that controlling for firm and year fixed effects, productivity of domestic firms increases with the presence of FDI firms in the sub-industry. However, while the effect is positive, it is not statistically

26

significant. This is quite in line with the finding of the previous literature, and is robust to the inclusion of other control variables such as age and export destinations in Column (2). The more interesting result is presented in Columns (3) where we find positive and significant effects of the productivity spillover of FDI firms on the domestic firms in the same sub-industry. For every 10 percent increase in the productivity of FDI firms, the productivity level of domestic firms in the same sub-industry improves by 1.4 percent. This result is robust to controlling for export shares and age of the firms as shown in Column (4). Columns (5) to (8) repeat the same exercise, but instead of using log of TFP as dependent variable, we use the level of TFP. This is closer to the usual notion of productivity (Olley and Pakes, Econometrica, 1996). In these specifications, both the presence of FDI firms and the productivity of FDI firms have positive and statistically significant effects on the productivity of domestic firms in the same sub-industry. Overall, there are sufficient statistically evidence suggesting that domestic firms may benefit from the productivity growth of FDI firms in their sub-industries. Thus, not only are FDI firms more productive than domestic firms, productivity growth of FDI firms may spillover to the domestic economy to benefit the domestic firms.

Table 7: Foreign Productivity Spillover

27

Dependent Variable

FDI share in industry

ltfp (1) Within 0.332 (0.224)

ltfp (2) Within 0.354 (0.223)

ltfp (3) Within

ltfp (4) Within

tfp (5) Within 8.790* (4.156)

tfp (6) Within 8.967** (4.167)

tfp (7) Within

tfp (8) Within

Productivity of FDI in industry Age -0.005*** (0.002) 0.000 (0.001) -0.001 (0.001)

0.142** (0.063)

0.150** (0.063) -0.006*** (0.001) 0.000 (0.001) -0.001 (0.001) -0.0678** (0.024) -0.010 (0.043) -0.016 (0.038)

0.251** (0.097)

0.253** (0.097) -0.117*** (0.024) -0.011 (0.043) -0.015 (0.038)

Export share of US

Export share of EU

Firm fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Year fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Observations 878 878 878 878 878 878 878 878 Notes: Both FDI presense and productivity are specific to industry and year. To correct for correlation of errors within industry-year, we cluster the standard errors in parentheses for each sub-industry-year. Sample consists of an unbalanced panel of 196 wholely domestic owned firms.

Conclusions This paper studies the relationship between foreign equity and firm productivity of Bangladesh garment sector. Firm productivity is measured by the total factor productivity (TFP), which is the level of output that is not explained by inputs, reflects efficiency in production of the firms. Using between firm variations, we show that FDI firms on average are 20 percent more productive than domestic firms in the same subindustry and location. Furthermore, there is statistically significant evidence suggesting that domestic firms may benefit from the productivity spillover from the FDI firms. For every 10 percent increase in FDI firm productivity, the productivity of domestic firms improve by 1.4 percent. The findings of this paper support a more open FDI policy in Bangladesh garment sector. References Aitken, Brian J. and Ann E. Harrison (1999). Do Domestic Firms Benefit from Direct

28

Foreign Investment? Evidence from Venezuela, American Economic Review 89, no. 3, 605-618. Eaton, Jonathan, Samuel Kortum, and Francis Kramarz (2004). Dissecting Trade: Firms, Industries, and Export Destinations, American Economic Review 94(2), 150-154. Javorcik, Beata (2004). Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages, American Economic Review 94, no. 3, 605-627. Kee, Hiau Looi (forthcoming). Firm Performance in the Services Sector, in Malaysia: Firm Competitiveness, Investment Climate, and Growth, the World Bank. Moulton, Brent R. (1990). An Illustration of a Pitfall in Estimating the Effects of Aggregate Variables on Micro Unit, Review of Economics and Statistics, 334-338. Olley, G. Steven and Ariel Pakes (1996). The Dynamics of Productivity in the Telecommunications Equipment Industry, Econometrica 64, no. 6, 1263-1297.

29