Technology Gap, Human Capital and Multinational Enterprises: Evidence from Country-Level

6th Global Conference on Business & Economics ISBN : 0-9742114-6-X

Technology Gap, Human Capital and Multinational Enterprises: Evidence from Country-Level Panel Data

Khaled Elmawazini*, Economics Department, University of Ottawa and KIMEP Samir Saadi, School of Management, University of Ottawa, Ontario, Canada

ABSTRACT

This paper investigates human capital and technology gap as measures of host country absorptive capacity in 38 developed and developing countries. Park/SUR method are used to estimate panel data regression equations. In many previous studies, human capital is measured by schooling years and technology gap is measured by total factor productivity relative to US total factor productivity. The results of these empirical studies are sensitive to the measure of absorptive capacity used. Using data on US multinational enterprises (MNEs), this paper shows that total factor productivity level is more significant than average schooling years to capture the absorptive capacity of host countries. This conclusion may indicate that the results of previous studies that used average schooling years should be interpreted with caution. In addition, it may support the hypothesis that MNEs technology diffusion effect in developing countries is weak and less significant than in developed countries.

INTRODUCTION

Recent empirical studies show mixed support for the hypothesis that the magnitude of technology transfer from multinatinal enterprises (MNEs) depends on host country absorptive capacity. The results of these empirical studies are sensitive to the measure of absorptive capacity used. Many recent studies argued that a country must have a certain level of human capital in order to benefit from multinational enterprises. Borensztein and et al. (1998) measured the human capital level by male secondary school attainment in the population over age 25. They examined the impact of FDI outflows from OECD countries to 69 developing countries during 1970 to 1989. They divided the sample into two periods: 1970-79 and 1980-89. Their results indicated that the threshold value is 0.52 years. Xu (2000) used the same measure of absorptive capacity as Borensztein et al. (1998) but he obtained different threshold value (1.4 years). He used MNEs data from the Bureau of Economic Analysis (BEA) of the US Department of Commerce in 1966, 1977, 1982, 1989, and 1994 for 40 developed and developing countries.

Xu (2000) argued that the difference between the threshold value in his study and Borensztein and et al. (1998) is due to the difference in the measures of technology diffusion being used . More specifically, Borenstein et al. (1998) used FDI to GDP ratio to capture the technology diffusion from FDI. On the other hand, Xu (2000) used technology transfer spending of foreign affiliates to GDP in the host country; this can measure the technology diffusion effect of foreign affiliates in the host country. Xu’s measure can extract the technology diffusion effect from other productivity-enhancing effects of US MNCs affiliates.

The panel data regression equation that is used in Xu’s is: 0 0 GTFPit  a i  a t  a1 GAPit  a 2 MNEit  a 3H it  eit (1.1) Where: GTFPit is the growth rate of total factor productivity (TFP) . GAP variable captures the technology level and it is measured by TFP for the host country relative to TFP * Corresponding author. Address: 23 Impasse Du Sillon, Gatineau, Quebec, J8Z 2Y7 Canada. Tel. 1(819)777-5068. E-mail: [email protected]

OCTOBER 15-17, 2006 1 GUTMAN CONFERENCE CENTER, USA 6th Global Conference on Business & Economics ISBN : 0-9742114-6-X

for the US. The MNE variable has three measures to determine the technology diffusion from US MNEs’ affiliates. The first measure, the value-added of MNEs’ affiliates / GDP in the host country, reflects the effect of the presence of MNEs on total factor productivity growth in the host country. The second measure, technology transfer spending of MNEs’ affiliates (affiliates’ spending on royalties and license fees) / (the value-added of MNEs’ affiliates), can reflect the effect of technology transfer intensity by MNEs affiliates on total productivity growth in the host countries. The third is technology transfer spending of MNEs’ affiliates / GDP in the host country; this can measure the technology diffusion effect of MNEs in the host country. H is the human capital level of the country; Human capital is measured by average years of male secondary school attainment in the population over age 25. This variable is calculated by Barro and Lee 1996, and finally eit is the error term.

Absorptive capacity is measured by technology gap in many empirical studies. Kokko (1994) is the first author who empirically uses the technology gap as a measure of absorptive capacity. He measured the technology gap by the labor productivity gap. His results indicate that domestic firms can not benefit from the technology diffusion from foreign firms if the technology gap between them is significant. Similar results can be found in Kokko et al. (1996), and Girma et al. (2001).

The above recent studies showed that the impact of foreign affiliate on productivity and economic growth mainly depends on host countries absorptive capacity. This may explain why foreign affiliate technology diffusion effect in developing countries is less significant than in developed country. However, De Mello (1999) argued that the impact of FDI on growth in OECD is less than that in non-OECD countries. In addition, Lensink and Morrisey (2001) argued that FDI has positive impact on economic growth regardless of the level of human capital. Similar result is obtained by McMillan (1999). This means that the results of these empirical studies are sensitive to the measure of absorptive capacity used. With this in mind, this article investigates human capital and technology gap as measures of host country absorptive capacity in 38 developed and developing countries. The reminder of this article is structured as follows. Section 2 is the empirical specification, Section 3 discusses the data, Section 4 is the empirical findings, and Section 5 is the conclusion.

EMPIRICAL SPECIFICATION

In this section, we will examine the significance of technology gap and human capital on the technology diffusion from multinational enterprises (MNEs). The panel data regression equation is: N MNE_Tech it   0j D jt   1 H 1 , it   2GAP 2 , it  e it (2.1) J1 Where: MNE_Tech is the technology transfer spending of US MNEs’ affiliates / GDP in the host country; this can measure the technology diffusion effect of MNEs in the host country. The calculation of MNE_Tech, H, and GAP is based on Xu (2000), see equation (1.1). The Djt are cross-section dummy variables. When j=1, Djt=1 and otherwise Djt = 0. Hence, equation (2.1) gives the panel data dummy variable model (i.e. fixed effect model). However, this model does not allow for the heteroskedasticity and autocorrelation. For this reason, we apply Park (1967) method to the dummy variable model. The assumptions of this method yield cross-sectionally heteroskedastic and timewise autoregressive model. Park (1967) model is discussed in Greene (2000).

The source of TFP data is Heston et al. (2002). We follow Xu(2000) method to 0.35 0.65 calculate total factor productivity for country i at time t: TFPit = Yit / (Kit Lit ). Y is real GDP in 1996 international prices; K is capital stock in 1996 international prices ; L is labor force in thousands. GAP is the technology gap which is measured by TFP relative to US

TFP in the initial year. H is the average years of male secondary school attainment in the population over age 25. The data of H variable is collected from Barro and Lee 1996.

MNE data source is the Bureau of Economic Analysis (BEA) of The US Department of Commerce in 1966, 1977, 1982, 1989, and 1994. We follow Xu (2000) who used MNEs data from manufacturing sector since this sector is most relevant for studying technology diffusion (Xu, 2000, p.481). The BEA data on technology transfer is used to extract the technology diffusion effect from other productivity-enhancing effects of US MNCs affiliates in 38 countries. By using BEA data, a panel of four periods and 38 countries is constructed. The following table indicates the countries that are included in our panel data set.

Table 3.1 Developed and developing countries by regions Developed countries Developing countries Canada Latin America: Argentina Europe: Brazil Austria Chile Belgium Colombia Denmark Costa Rica Finland Ecuador France Mexico Greece Peru Ireland Venezuela Italy Netherlands Asia: Norway Thailand Portugal Turkey Spain India Sweden Malaysia Switzerland Singapore UK Korea Philippines Other-DC Israel Africa: Japan Egypt New Zealand South Africa Australia

EMPIRICAL RESULTS

The empirical findings based on equation 2.1 can be summarized in the following table:

Table (4.1) presents panel data regressions results for the MNE_Tech variable from the full sample. Values in parentheses are t-statistics. Regression (Regression 4.1) (Regression 4.2) (Regression 4.3) Method:

H variable 0.24907 (1.324) Not included 0.89796E-02 (0.05160)

GAP variable -0.48219 (-3.169) -0.40526 (-2.883) Not included

BUSE RAW-MOMENT 0.9099 0.9096 0.9072 R2 DW 1.7596 1.7488 1.7519

RUNS (Geary) test 56 RUNS,75POS,77 NEG 53 RUNS, 76 POS, 76 NEG 51 RUNS, 75 POS,77 NEG

N 38 38 38 T 4 4 4 NT 152 152 152

Breusch-Pagan (1980) Lagrange multiplier test indicates that there is no evidence to reject the hypothesis of no cross-section correlation. Lagrange multiplier test for cross-section heteroskedasticity indicates that there is evidence to reject the the hypothesis of homoskedasticity. With these two results in mind, Park (1967) method are used to estimate regressions 4.1, 4.2, and 4.3. The assumptions of this method yield cross-sectionally heteroskedastic and timewise autoregressive model. Park (1967) model is discussed in Greene (2000). The results of regressions (4.1), (4.2), and (4.3) indicate that total factor productivity gap is better than average years of male secondary school attainment in capturing the host country absorptive capacity. These results may imply that the results of previous studies (e.g. Xu (2000) and Borensztein et al. (1998), which used average schooling years, should be interpreted with caution. In addition, it may support the hypothesis that MNEs technology diffusion effect in developing countries is weak and less significant than in developed countries.

CONCLUSION

The results of previous empirical studies, which examine the host countries benefits from multinational enterprises, are sensitive to the measure of absorptive capacity used. With this in mind, this article investigates human capital and technology gap as measures of host country absorptive capacity in 38 developed and developing countries. The LSDV model suffers from cross-section heteroskeastic. For this reason, we apply Park method to LSDV Model to estimate panel data regression equations. The assumptions of this method yield cross-sectionally heteroskedastic and timewise autoregressive model. MNE data source is the Bureau of Economic Analysis (BEA) of The US Department of Commerce in 1966, 1977, 1982, 1989, and 1994. This study follows Xu (2000) who used MNEs data from manufacturing sector since this sector is most relevant for studying technology diffusion (Xu, 2000, p.481). The BEA data on technology transfer is used to extract the technology diffusion effect from other productivity-enhancing effects of US MNCs affiliates in 38 countries. By using data on US multinational enterprises, this study shows that total factor productivity level can capture the absorptive capacity better than the average schooling years in host countries. These results may imply that the results of previous studies (e.g. Xu (2000) and Borensztein et al. (1998), which used average schooling years, should be interpreted with caution.

REFERENCES

Aitken, B. and A. Harrison (1999). Do domestic firms benefit from foreign investment? Evidence from Venezuela. American Economic Review, 89, 605-618.

Barro, R.J., and X. Sala-i-Martin (1997). Technological diffusion, Convergence, and growth. The Journal of economic growth, 2, 1-27.

Barro, R.J. and J.W. Lee (1996). International Measures of Schooling Years and Schooling Quality. American Economic Review, 86(2), 218-223.

Borensztein, E., J. Gregorio, J. Lee, “How does foreign direct investment affect economic growth,” Journal of International Economics 45, (1998): 115-135.

De Mello, L.R. (1997), ‘Foreign Direct Investment in Developing Countries and Growth: A Selective Survey’, Journal of Development Studies, 34, pp. 1-34.

De Mello, L.R. (1999). ‘Foreign Direct Investment-led Growth: Evidence from Time Series and Panel Data’, Oxford Economic Papers, 51, pp. 133-151.

Haddad, M. and A. Harrison. (1993). Are there Positive Spillovers from Direct Foreign Investment? Evidence from Panel Data for Morocco. Journal of Development Economics 42, 51-74.

Hermes, Niels and Robert Lensink (2003). Foreign Direct Investment, Financial Development and Economic Growth, Journal of Development Studies, 40(1), 142-63

Horstmann, I. and Markusen, J. (1996). Exploring new markets: direct investment, contractual relations and the multinational enterprise. International Economic Review, 37, 1-20.

Lesink, R. and Morrissey, O. (2001). Foreign Direct Investment: Flows, Volatility and Growth in Developing Countries. CREDIT Research Paper 01/06.

McMillan, M.(1999). Foreign Direct Investment: Leader or Follower? Discussion Paper 99-1, Department of Economics, Tufts University.

Xu, B. (2000). Multinational enterprises, technology diffusion, and host country productivity growth. Journal of Development Economics 62, 477-493.

OCTOBER 15-17, 2006 5 GUTMAN CONFERENCE CENTER, USA