Frm Performance and Foregn Drect Investment: Evdence from Transton Economes Mahmut Yasar Department of Economc, Unversty of Texas at Arlngton Catherne J. Morrson Paul Department of Agrcultural and Resource Economcs, Unversty of Calforna, Davs Abstract Frm Performance and Foregn Drect Investment: Evdence from Transton Economes Mahmut Yasar, Unversty of Texas at Arlngton, and Catherne J. Morrson Paul, Unversty of Calforna, Davs* Abstract We evaluate the performance of foregn-owned versus domestc frms, and the spllover effects of ndustry foregn share for fve transton economes, namely Poland, Moldova, Tajkstan, Uzbekstan, and the Kyrgyz Republc. We fnd hgher productvty, captal ntensty, export and mport shares, employment, and wages for frms wth foregn ownershp. Further, we fnd that ndustry presence of foregn afflates of multnatonal frms leads to performance mprovements for domestc frms; that s, spllovers from foregn frms beneft domestc frms n these transton economes. We would lke to thank John Nasr and Fabano Bastos for ther helpful data suggestons. Ctaton: Yasar, Mahmut and Catherne J. Morrson Paul, (2007) "Frm Performance and Foregn Drect Investment: Evdence from Transton Economes." Economcs Bulletn, Vol. 15, No. 21 pp. 1-11 Submtted: August 27, 2007. Accepted: October 10, 2007. URL: http://economcsbulletn.vanderblt.edu/2007/volume15/eb-07o50003a.pdf
Frm Performance and Foregn Drect Investment: Evdence from Transton Economes Mahmut Yasar, Unversty of Texas at Arlngton, and Catherne J. Morrson Paul, Unversty of Calforna, Davs* Abstract We evaluate the performance of foregn-owned versus domestc frms, and the spllover effects of ndustry foregn share for fve transton economes, namely Poland, Moldova, Tajkstan, Uzbekstan, and the Kyrgyz Republc. We fnd hgher productvty, captal ntensty, export and mport shares, employment, and wages for frms wth foregn ownershp. Further, we fnd that ndustry presence of foregn afflates of multnatonal frms leads to performance mprovements for domestc frms; that s, spllovers from foregn frms beneft domestc frms n these transton economes. Keywords: transton economes, Eastern Europe and Central Asa, frm performance, foregn drect nvestment, spllovers JEL Classfcatons: F10, F14, D21, L60 * We would lke to thank John Nasr and Fabano Bastos for ther helpful data suggestons. 1
1. Introducton Early studes such as Vernon (1966) and Caves (1974) suggested that multnatonal frms are more productve than ther domestc counterparts and that the presence of multnatonal frms has a postve mpact on domestc frms performance. Foregn drect nvestment (FDI) by multnatonal frms has snce been advocated as a prmary source of technology dffuson and economc growth for developng countres. In addton to havng postve mpacts on trade, employment and captal, FDI has therefore been vewed as a key channel for transferrng knowledge, sklls, and technology, especally from ndustralzed to developng countres (Blomström and Kokko, 1998). Such postve externaltes or spllovers dffuse to the domestc economy through nteractons between domestc and foregn frms, whch facltate the adopton of and nvestment n advanced technologes and the accumulaton of knowledge and sklls to effectvely utlze the technology by domestc frms. These spllovers thus may have mportant mpacts on the effcency, productvty and compettveness of frms, ndustres, and countres partcularly for mddle- and low-ncome countres. Emprcal evdence of ths phenomenon has, however been lmted and conflctng. Most emprcal studes (e.g., Blomström and Wolff, 1989, for Mexco, Haddad and Harrson, 1994, for Morocco, Atken and Harrson, 1999, for Venezuela, and Grma, 2002, for the U.K.), have not found evdence of ncreases n domestc frm productvty from the ndustry presence of foregn frms. However, some ndustry level studes (e.g., Caves, 1974, for Canada and Australa, Globerman, 1979, for Canada, and Blomström, 1986, for Mexco), as well as frm level studes (e.g., Kokko et al., 1996, for Uruguay, Dmels and Lour, 2002, for Greece, and Karpaty and Lundberg, 2004, for Sweden), do fnd evdence of spllovers to the domestc economy. Further, Sjöholm (1999) fnds for Indonesa that productvty spllovers are greater when the degree of competton n an ndustry s hgher and the technology gap among frms s larger. Smlarly, Kokko (1994), who fnds no evdence of spllovers from foregn-owned frms for Mexco n hgh technology ndustres, suggests that spllovers depend on the complexty of the technology beng transferred by foregn-owned frms and the (labor) productvty gap between domestc and multnatonal frms. Although the mpact of FDI on the performance of the domestc frms has not been examned for most of the countres n our dataset, some studes have consdered ths ssue for Poland. For example, Konngs (2001) examnes the mpact of foregn drect nvestment on the performance of domestc frms n Bulgara, Romana and Poland, and fnds that n Poland the most advanced of the three countres foregn frms perform better than ther domestc counterparts. However, he fnds no evdence of spllovers to domestc frms, and concludes that t may take tme for ownershp effects to nfluence domestc frms performance due to restructurng lags. Damjan et al. (2001) obtan smlar results to those obtaned by Konngs (2001). 1 Barell and Holland (2000) examne 1 See Navarett and Venables for the lterature revew of foregn drect nvestment. 2
the effect of FDI for 11 manufacturng ndustres n Hungary, Poland and the Czech Republc, and fnd that FDI ncreases the labor productvty n most ndustres. The objectve of ths paper s to provde further emprcal evdence on the performance effects of foregn ownershp, from the perspectve of frms n fve Eastern Europe and Central Asan countres. We use cross secton survey data collected by the World Bank for garment and food processng frms n Poland, Moldova, Tajkstan, Uzbekstan, and the Kyrgyz Republc ndustres n whch the complexty of the technology s relatvely low although ther captal ntensty s ncreasng. We frst estmate proportonal dfferences n performance ndcators for foregn versus domestc frms n the data. We then examne whether the presence of foregn-owned frms n an ndustry ncreases the productve performance of domestc frms. We fnd that frms wth a foregn ownershp share are more productve than ther domestc counterparts, and that a greater foregn share mples hgher productvty. Such frms are also larger, pay more, hre more, and have a greater export share of sales and mport share of materals. Further, we fnd that ndustry presence of foregn afflates of multnatonal frms leads to performance mprovements for domestc frms; that s, spllovers from foregn frms beneft domestc frms. 2. Plant Performance of Frms wth Foregn Drect Investment We frst examne the relatonshps between foregn ownershp and frm performance by measurng proportonal dfferences between performance characterstcs (P ) of frms that have and do not have or have a greater foregn share, by estmatng the equaton: C = o + FDI FDI + EMP EMP + IND IND + 1 β β β c= 1 ln P α β D + u. (1) c c Our P varables nclude a varety of producton characterstcs related to economc performance: levels of total factor productvty (TFP), labor productvty (LP), 2 output (Y), captal nput (K), captal nput per worker (KI), average wage per worker (WI), average sklled worker wage (WSI), average unsklled worker wage (WUSI), employment (EMP), manageral workers (MW), professonal workers (PW), sklled producton workers (SP), and unsklled producton workers (USP). Addtonal P varables n our data are the share of sales exported (EXPS), the share of materals mported (IMPS), whether the frm developed/upgraded a new major product lne or ntroduced a new technology (INNOV), and whether the frm regularly used the nternet or emal n ts nteractons wth clents and supplers (INT). 2 Labor productvty (LP) s defned as ln LP = ln Y ln L. Total factor productvty, ln TFP = ln Y-ln X, where X stands for captal, labor, energy and materals nputs, s calculated usng output elastctes estmated from a Cobb-Douglas producton functon, ncludng ndustry and country dummes and sklled labor share. We pooled the data but nteracted the dummy varables wth the nputs to test whether the coeffcents vared by ndustry and country, whch they dd not. 3
Summary statstcs for these data and other data reled on are reported n Table 1. The BEEPS II Busness Envronment 2002 data was collected by the World Bank through the 2003 Investment Clmate Survey for fve transton countres: the Kyrgyz Republc, Moldova, Poland, Tajkstan, and Uzbekstan. A sample of one hundred frms was drawn randomly from all sze categores n each country. The questons used n the survey were dentcal n each country and ntervewers conducted face-to-face ntervews wth frms managers and bookkeepers or accountants. 3 Although over 500 frms were surveyed, because of the mssng varables we ended up wth 437 frms. The numbers of frms are 76, 96, 88, 99 and 78 for Poland, Moldova, Tajkstan, Uzbekstan, and Kyrgyz Republc, respectvely. For the productvty measure, due to mssng captal data, the numbers of observatons are 66, 55, 57, 85 and 66 for Poland, Moldova, Tajkstan, Uzbekstan, and Kyrgyz Republc, respectvely. TABLE 1 HERE The ndependent varables nclude a sze measure (EMP, the natural log of employment, representng dfferences n the producton technologes of dfferent sze frm) 4, an ndustry dummy (IND=0 for food and IND =1 for garment manufacturers), and country dummes (D C, where Moldova s the base). 5 The varable of prmary nterest, FDI, s alternatvely measured as a dummy varable ndcatng whether the frm has any foregn share, and as the foregn ownershp share. Accordng to the OECD, foregn drect nvestment s defned as the ownershp or control of 10 percent, or more, of assets by a foregn company. In our sample, the lowest (postve) share s 18 percent, whch thus becomes the ownershp threshold of drect nvestment n our sample. The frms are therefore defned as at least partally foregn owned f foregn nterests control 18 percent or more of the assets. The β FDI coeffcent on the FDI dummy thus ndcates the average percentage dfference n P (the premum n terms of the performance characterstc P ) between the frms wth and wthout a foregn share, condtonal on sze, ndustry and country dummes. The β FDI estmate for the foregn share s smlarly nterpreted as the percentage dfference n P f the foregn share ncreased. Such estmates avod the potental endogenety problems often thought to plague producton functon estmaton (Bernard et al., 2003), and have been shown to be consstent wth other methods that drectly address such econometrc ssues by Yasar and Paul (2007a,b). The resultng estmates are presented n Table 2. 3 The BEEPS II Busness Envronment 2002 data was collected by the World Bank through the 2003 Investment Clmate Survey. See Bastos and Nasr (2004) for a detaled explanaton of the data. 4 Ths s omtted when the dependent varable s based on overall employment or on a per employee bass. 5 We tred other control varables, such as EXPS, IMPS, INT, and INNOV, but the results are very smlar to those reported here. We also nteracted the ndustry and country dummy varables wth the FDI varable to test whether the coeffcents vared across the ndustres and countres. The regresson coeffcents dd not, however, dffer sgnfcantly by ndustry and country. 4
TABLE 2 HERE All of the estmated coeffcents are sgnfcant and postve, ndcatng that frms wth a foregn share have greater productvty levels, captal ntensty, sze, payscales, and labor force. Ths s consstent wth Yasar and Paul (2007a), who fnd smlar performance effects for Turksh manufacturng plants usng both producton functon and prema estmaton, and Yasar and Paul (2007b) who use matchng methods. 6 These frms also export, mport, ntroduce new products or technologes, and take advantage of the nternet more than the domestc frms. The estmated effects are consstent, but even stronger, when estmated n terms of the share rather than presence of foregn ownershp. 3. Spllover Effects Second, we evaluate productvty spllovers to the host economy as the effect of the extent of foregn presence n the ndustry on the performance of the domestc frms. We use two measures as ndcators of foregn presence, FDINS, for each country: the share of foregn frms n an ndustry s employment, and the share of foregn frms n an ndustry s sales. Our estmatng equaton s: ln PM = α + β + β o INT INT FDIS FDINS + β IND + β IND + EMP C 1 c= 1 EMP + β β D + u c c EXP EXP + β IMP IMP + β IN INNOV, (2) where the performance measures PM nclude total factor productvty, labor productvty, captal nput per worker, average wage per worker, and the percent of sales exported. A sgnfcant coeffcent on the FDIS varable, β FDIS, therefore ndcates spllovers from foregn-owned to domestc frms. EMP, EXPS (except when t s the dependent varable), IMPS, INNOV, INT, and ndustry and country dummy varables are the control varables. 7 The results presented n Table 3 show postve mpacts of relatve foregn presence on domestc frm performance, or ndustry spllover effects. For example, when foregn presence s measured by the foregn frm employment share (the second column), the estmates show that an ncrease of 0.1 n FDINS s assocated wth 0.14 percent hgher total factor productvty, 0.52 percent hgher labor productvty, 0.77 percent hgher captal ntensty, 0.48 percent hgher average wage, and 0.15 percent hgher export share. When the foregn sales share s nstead used as n ndcator of foregn presence, the estmates n the thrd column agan show a postve mpact of foregn presence on domestc frm performance for all ndcators. 6 The producton functon n ths case was estmable by nstrumental varables because t was based on tme seres data, and the prema estmaton s often used to avod such problems (see Bernard et al., 2003). The matchng methods used by Yasar and Paul (2007b) provde a drect counterfactual and thus mply causalty. 7 We also nteracted the ndustry and country dummy varables wth the varables for the degree of foregn presence and found that the coeffcents do not sgnfcantly dffer by ndustry and country; the results are avalable upon request. 5
TABLE 3 HERE 4. Conclusons For our data on garment and food manufacturers n transton economes we fnd that foregn owned frms perform better accordng to a number of ndcators than ther domestc counterparts. Controllng for sze, ndustry and country, they exhbt hgher productvty, captal ntensty, wages, employment, export share of sales, mport share of materals, and nnovaton. One would thus expect postve spllovers from such frms to beneft domestc frms n these developng countres, whch are also evdent from our data. As the ndustry foregn share ncreases the performance of domestc frms n terms of productvty, export share, and captal and average wage per worker s enhanced. These fndngs support the hypothess that foregn companes brng wth them technology and sklls, access to foregn markets, and new employment opportuntes that enhance the performance of host country frms. Ths n turn suggests that connectons wth foregn owned frms should be encouraged by polces regardng foregn presence to enhance frm and ndustry productvty and compettveness n transton economes. 6
References Atken, B., and A. Harrson (1999) Do Domestc Frms Beneft from Foregn Drect Investment? Evdence from Venezuela Amercan Economc Revew 89, 605-618. Barrell R., and D. Holland (2000) "Foregn Drect Investment and Enterprse Restructurng n Central Europe Economcs of Transton 8(2), 477-504. Bastos, F., and J. Nasr (2004) Productvty and the Investment Clmate: What matters Most? World Bank Polcy Research Workng Paper 3335. Bernard, A.B., Eaton, J., Jensen, J. B., and S. Kortum (2003) Plants and Productvty n Internatonal Trade Amercan Economc Revew 93,1268-1290. Blomström, M. (1986) Foregn Investment and Productve Effcency: The Case of Mexco Journal of Industral Economcs15, 97-110. Blomström, M., and A. Kokko (1998) Multnatonal Corporatons and Spllovers Journal of Economc Surveys12, 247-277. Blomström M., and E. N. Wolff (1989) Multnatonal Corporatons and Productvty Convergence n Mexco NBER Workng Paper 3141. Caves, R. E. (1974) Multnatonal Frms, Competton and Productvty n Host Country Markets Economca 41,176-93. Damjan, J. P., M. Knellb, B. Majcenc, and M. Rojecd (2003) "The role of FDI, R&D Accumulaton and Trade n Transferrng Technology to Transton Countres: Evdence from Frm Panel Data for Eght Transton Countres" Economc Systems 27 (2), 189-204. Dmels, S., and H. Lour (2002) Foregn Ownershp and Producton Effcency: A Quantle Regresson Analyss Oxford Economc Papers 54, 449-469. Grma, S. (2002) Adaptve Capacty and Productvty Spllovers from FDI: a Threshold Regresson Analyss GEP Research Paper 2002/08. Globerman, S. (1979) Foregn Drect Investment and Spllover Effcency Benefts n Canadan Manufacturng Industres Canadan Journal of Economcs12, 42-56. Haddad, M., and A. Harrson (1993) Are there Postve Spllovers from Drect Foregn Investment? Evdence from Panel Data for Morocco Journal of Development Economcs 42, 51-74. Karpaty, P. and L. Lundberg (2004) Foregn Drect Investment and Productvty Spllovers n Swedsh Manufacturng Örebro Unversty Workng Paper Seres 2. 7
Kokko, A. (1994) Technology, Market Characterstcs and Spllovers Journal of Development Economcs 43, 279-93. Konngs J. (2001) The Effects of Foregn Drect Investment on Domestc Frms: Evdence from Frm-Level Panel Data n Emergng Economes Economcs of Transton 9(3), 619-33. Navarett, G. B., and A.J. Venables (2005) Multnatonal Frms n the World Economy, Prnceton Unversty Press. Sjöholm, F. (1999) Technology Gap, Competton and Spllovers from Drect Foregn Investment: Evdence from Establshment Data Journal of Development Studes 36, 53-73. Vernon, R. (1966) Internatonal Investment and Internatonal Trade n the Product Cycle The Quarterly Journal of Economcs 106, 190-207. Yasar, M., and C.J. Morrson Paul, C.J. (2007a). Internatonal Lnkages and Productvty at the Plant Level: Foregn Drect Investment, Exports, Imports and Lcensng. Journal of Internatonal Economcs 72(1), 373-388. Yasar, M., and C. J. Morrson Paul (2007b) Technology Transfer and Productvty: Evdence from a Matched Sample. Journal of Busness and Economc Statstcs, forthcomng. 8
Table 1: Mean Values of the Varables Natural log of Output, Y Output per worker, LP Captal, K Materals, M Energy, E Average wage, W Average sklled wage, SW Average unsklled wage, USW Employment, EMP Manageral workers, MW Professonal workers, PW Share of Sklled producton workers, SPW Unsklled producton workers, USPW Export share, EXPS Import share, IMPS Foregn share, FDIS Innovaton, INNOV Internet, INT Poland (76) Moldova (96) Tajkstan (88) Uzbekstan (99) Kyrgyz R. (78) 6.405 4.921 2.797 3.751 4.081 3.190 1.295 0.582 0.410 0.519 5.041 3.666 2.592 3.295 3.312 5.868 4.074 2.013 3.027 3.346 4.039 2.644 0.428 1.687 1.826 4.613 3.242 1.266 1.710 2.601 7.222 5.283 4.681 4.248 4.789 5.530 3.469 3.058 2.587 3.076 3.343 3.701 2.216 3.319 3.581 0.604 1.011 0.276 0.535 0.675 1.293 1.402 1.247 1.266 1.793 2.637 3.013 1.538 2.965 2.937 2.077 2.588 1.309 2.043 2.352 0.032 0.235 0.008 0.023 0.090 0.224 0.412 0.051 0.055 0.136 0.032 0.059 0.011 0.134 0.063 0.579 0.746 0.636 0.290 0.671 0.692 0.534 0.034 0.140 0.354 Notes: Numbers of observatons are n parentheses n the frst row. For the productvty measure, due to mssng captal data, the numbers of observatons are 66, 55, 57, 85 and 66. 9
Table 2: Dfferences between Foregn Owned and Domestc Frms FDI Dummy ln TFP 0.137 (0.039)*** ln LP 0.686 (0.133)*** ln Y 2.177 (0.272)*** ln K 1.737 (0.282)*** ln K/L 0.633 (0.200)*** ln W 0.470 (0.104)*** ln SW 0.227 (0.112)** ln USW 0.391 (0.108)*** ln EMP 1.695 (0.235)*** ln MW 0.753 (0.137)*** ln PW 0.739 (0.196)*** ln SPWs 1.574 (0.257)*** ln USPW 0.865 (0.267)*** EXPS 0.269 (0.036)*** IMPS 0.089 (0.046)** INNOV 0.184 (0.078)** INT 0.235 (0.068)*** FDI Share 0.190 (0.051)*** 0.793 (0.179)*** 2.427 (0.378)*** 2.017 (0.390)*** 0.871 (0.268)*** 0.543 (0.139)*** 0.300 (0.146)** 0.552 (0.143)*** 1.934 (0.317)*** 0.712 (0.187)*** 0.881 (0.260)*** 1.665 (0.348)*** 0.950 (0.363)*** 0.368 (0.048)*** 0.135 (0.061)** 0.288 (0.103)*** 0.339 (0.090)*** Note: * denotes sgnfcant at the 10% level, ** at the 5% level, and ***sgnfcant at the 1% level. 10
Table 3: Spllover Effects Dependent Varables Employment Share Sales Share ln TFP ln LP ln K/L ln W EXPS 0.014 (0.007)** 0.052 (0.025)** 0.077 (0.035)** 0.048 (0.020)*** 0.015 (0.004)*** 0.007 (0.004)* 0.042 (0.015)*** 0.049 (0.020)*** 0.028 (0.011)*** 0.025 (0.004)*** 11