The Improvement and Application of Econometric Model for Reverse Spillover Effect of Technology Sourcing FDI

[a]Department of Finance and Economics， Shandong University of Science and Technology Jinan， China.

*Corresponding Author.

Address: Department of Finance and Economics， Shandong University of Science and Technology Jinan， China.

Supported by Shandong province higher school of humanities and social science research project (number:J11WF63).

Received 11 June 2012; Accepted 23 July 2012

Abstract

Domestic enterprises expect to obtain foreign advanced technology through foreign direct investment， but its reverse spillover effect has been the lack of effective measurement. This paper use foreign research results for reference， and tries to improve and apply the econometric model for reverse spillover effect of technology sourcing FDI.

Key words: Technology sourcing FDI; Econometric model; Test

FU Yuanzhai， ZHAO Jiaying， LI Hongyu (2012). The Improvement and Application of Econometric Model for Reverse Spillover Effect of Technology Sourcing FDI. International Business and Management， 5(1)， -0. Available from: URL: http://www.cscanada.net/index.php/ibm/article/view/j.ibm.1923842820120501.Z0789

DOI: http://dx.doi.org/10.3968/j.ibm.1923842820120501.Z0789

INTRODUCTION

Under the unsatisfactory circumstances of technology spillover effect of FDI， more and more competitive China’s enterprises began to conduct foreign direct investment to acquire foreign advanced technology， and the research proves that the reverse spillover effect of technology sourcing FDI could improve total factor productivity of the national industries or enterprises of home country. However， the actual effect of this kind of FDI has been the lack of corresponding metrological analysis， so it has important practical significance to analyze the reverse spillover effect of technology sourcing FDI of China’s enterprises at the national level.

1. THE IMPROVEMENT OF REVERSE SPILLOVER EFFECT MODEL

Coe and Helpman (1995) proposed international RD spillover effect model:

(1)

Wherein， i means different countries; t indicates time; LogF shows the natural logarithm form of total factor productivity; SD stands for domestic RD capital stock; SF expresses foreign RD capital stock; is cross-sectional data items of country ; signifies the output elasticity of domestic RD capital stock; represents the output elasticity of foreign RD capital stock; is error items. This model reflects that the growth rate of a country’s total factor productivity is not only influenced by domestic RD capital stock， but also affected by foreign RD capital stock.

Van Pottelsberghe de la Potterie (2001) took FDI outflows as spillover channel and introduced it into this model for the first time to test the spillover effect of technology sourcing FDI. can be shown as: ， of which， tij means the foreign direct investment flows of investor i to jcountry; shows the gross fixed capital formation of investee country j; expresses domestic RD capital stock of investee country. stands for adjusting weight of the investee country’s RD capital stock， which reflects the contrast relation between FDI from investor country and domestic investment of investee country.

Jürgen Bitzer (2005) believes that the above model ignored the third-country effect of RD spillovers. So he improved foreign RD stock based on FDI:， of which， indicates the FDI stock of the investor country in period t， and kct signifies the physical capital stock of the investor country in period t. represents the sum of RD capital stock within countries except for investor country， including the third-country effect.

Considering the advantages and disadvantages of these two models， and combined with the actual situation of China’s statistical data， the author improved the model as follows:

(2)

Of which， TFPt is total factor productivity of China in period t; SDt means the domestic RD capital stock in China in period t; SFt refers to the reverse spillover of foreign RD capital stock through FDI in China in period t. ， of which，OFDIt indicates China’s FDI stock in period t;Kt stands for China’s physical capital stock in period t. represents the sum of RD capital stock of foreign countries in period t. The meaning of this model is that the growth rate of China’s total factor productivity is not only influenced by domestic RD capital stock but also affected by foreign RD capital stock through the way of China’s FDI.

2. THE APPLICATION OF IMPROVED ECONOMETRIC MODEL

Select the 23 years’ data from 1988 to 2010 to conduct analysis.

2.1 The Calculation of Total Factor Productivity TFPt

Take Solow Residual Method to estimate the LnTFPt. Assuming that the Cobb-Douglas production functionto satisfy the Hicks neutral， which means technical progress does not affect the marginal rate of substitution between K and L; additionally， the returns to scale of K and L are invariable， which means ， and the total factor productivity:

(3)

In order to obtain LnTFPt， it needs to estimate and firstly. selects 23 the value of real GDP; means physical capital stock of each year;stands for the quantity of employment at each end of year. Taking use of regression equation with the constraints of to conduct OLS estimation for， and adds variables of AR(1) and AR(2) to eliminate the autocorrelation， so it is: =-0.856018+0.762697+ [AR (1) =1.200582， AR (2) =-0.480634]， so =0.763，=0.237， takes them into equation(3)to calculate the value of .

2.2 The Calculation of Domestic RD Capital

Stock SDt

， of which，is depreciation rate of RD capital stock， as 5%; means RD expenditure in period t; the RD stock in period 0 refers to the RD stock in 1987， g stands for the average growth rate in form of logarithm of RD expenditure in each period.

2.3 The Calculation of Foreign RD Capital Stock

Spillover SFt

Using the calculation method of Jürgen Bitzer (2005) for reference，， calculates the reverse spillover to China of foreign RD capital stock through FDI channels of China’s enterprises. The calculation method of SDmt is the same as that of domestic RD capital stock， and is also 5%. For the selection of foreign country， combined with the mainly destinations of technology sourcing FDI of China’s enterprises， the author chooses 14 countries， such as the United States， to represent the RD overall stock of the rest countries in the world except for China， and the ever year RD expenditure of different countries is obtained by the proportion of yearly RD expenditure to GDP multiplies by the yearly real GDP.

2.4 The Stability ADF Test

Because the macroeconomic variables are usually non-stationary， therefore， it is necessary to conduct the stability ADF test for time series before analysis. Take the ADF unit root test method to conduct the stability test for，and . Shown as Table 1:

Table 1

Adf Unit Root Test Results

VariableTest formADF test StatisticCritical valueStabilityOrder of integration

Lntfp(1，c)-2.355613-3.012363unstableI(1)

D(Lntfp)(0，n)-2.973797-1.958088stable

LnSD(1，n)2.056422-3.012363unstableI(2)

D (LnSD)(0，n)3.889785-1.958088unstable

D(LnSD，2)(0，t)-3.935869-3.658446stable

LnSF(0，c)-2.795166-3.004861unstableI(1)

D(LnSf)(0，c)-4.426329-3.012363stable

It is clear that ，andare all unstable time series， so it needs further co-integration analysis to and.

2.5 Johanson Co-Integration Test

According to the Akaike Information Criterion (AIC) and Schwarz Criterion (SC) minimum standards， it confirms the lag order is 3 orders. The test form is to choose a sequence of a linear trend and the co-integration equation only with cross-sectional data items. The test results are shown as Table 2:

Table 2

Johansen Co-Integration Test

Null HypothesisEigenvalueTrace statisticTrace statistic 5% critical valueP value

No Co-integration Relationship 0.866004 38.19548 15.49471 0.0000

At Least Existence of a Co-integration Vector 0.000340 0.006466 3.841466 0.9354

It is clear that: the trace statistic of the first 1 hypothesis is larger than its critical value (38.08230﹥15.49471)， so the first 1 hypothesis is rejected; however， the trace statistic of the second 1 hypothesis is smaller than its critical value (0.006108 ﹤3.841466)， which means the second 1 hypothesis is accepted. Therefore， it shows that there is a long-term co-integration relationship between and. Additionally， Eviews5.0 provides co-integration regression equation between and as follows:

=+ 0.037927(4)

Of which， indicates non-equilibrium error， and it reflects that there is a negative correlation relationship between the growth rate of China’s total factor productivity and foreign RD stock， which means the growth of China’s FDI not only failed to bring positive reverse spillover effect to promote China’s total factor productivity; on the contrary， it produced negative effect， which hinders the improvement of China’s total factor productivity.

2.6 The Error Correct Model

Granger theory points that if several non-stationary variables exist co-integration relationship， these variables must have error correction model expression. Established error correction model of and as follows:

Of which， is a lag of non-equilibrium error， and it means the control and correction of long-term co-integration relationship of andfor short waves of ()， and its coefficient is a correction factor， which means the correction speed of non-equilibrium error for the short-term fluctuations of， and is called error correction item. is the optimal lag order number which takes the residual as white noise， and selects the lag order 3. The short waves of are influenced not only by error correction term but also by the short waves of hysteretic. Take advantage of Eviews5.0 to estimate error correct model as follows:

=-0.429779(

+ .037927

+ 0.673961) +0.701634

-0.323169+0.274358

–0.023356-0.036205

-0.033089+ 0.02101

It can be perceived that the correction factor is -0.429779， and its absolute value is smaller than 1， which means it is not very obvious for long-term equilibrium system of andto the short waves of. The coefficient of is significantly negative， which illustrates the negative correlation between short-term total factor productivity in China and foreign RD spillover， and the growth or decline of the short-term foreign RD spillover would lead to reverse change of China’s total factor productivity.

CONCLUSION

Based on above analysis， it is clear that， no matter long-term or short-term， the technology sourcing FDI of China’s enterprises not only failed to form positive reserves spillover effect， but it hinders the improvement of China’s total factor productivity. Although it may be derived from the model and data problems， from the model point of view， it is mainly caused by the scale limitation of the technology sourcing FDI of China’s enterprises. In the model， as the adjustment coefficient of foreign RD stock， which means the larger the scale of FDI， the more foreign RD stock would spill， and it would contribute to the improvement of total factor productivity in the home country. According to the survey made by Roland Berger Company， only 16% China’s leading enterprises operate oversea business in order to obtain advanced technology and brand assets; in the selection of oversea operational aspects， RD only accounts for 16%. Compared with the large scale economic capacity， such small scale foreign investment is not enough to become the effective carrier and channel of reserve spillover to transfer foreign RD resources. In addition， many reasons， for example， the insufficiency of absorptive and embedded ability of China’s enterprises in subsidiaries or RD institutions in host countries， the dislocation between obtained technology and the development demand of domestic enterprise and market， the potential crowding-out effect for domestic RD investment and the insufficiency of spillover mechanism at the industry and country level， lead to the weak spillover effect， which should be taken actions based on above reasons.

REFERENCEs

Bruno van Pottelsberghe de la Potterie and Frank Lichtenberg (2001). Does Foreign Direct Investment Transfer Technology Across Borders? The Review of Economics and Statistics， 83(3)， 490-497.

Coe， D.T.， Helpman， E. (1995). International RD Spillovers. European Economic Review， 39， 859-887.

J?urgen Bitzer Monika Kerekes (2005). Does Foreign Direct Investment Transfer Technology Across Borders? A Reexamination http://www.wiwiss.fu-berlin.de/files/K6UAD7B/discpaper07_05.pdf

ZHANG， X.T. (2005). Foundation of Econometrics. Tianjin: Nankai University Press.

ZHAO， J.Y. (2007). Analysis of Technology Sourcing FDI of China’s Enterprise and Its Reverse Spillover Effect. Shandong: Shandong University.