R&D-FDI-TG-Productivity Link 1 Running head: R&D-FDI-TG-PRODUCTIVITY LINK R&D-FDI-TG-Productivity Link in Chinese Industries Qian Lu The University of Hong Kong Renmin University of China
R&D-FDI-TG-Productivity Link 2 Abstract In this paper, I try to analyze relationships between innovation, ITT (international technology transfer) and productivity. After incorporating the technological gap (TG) into the model, I build a R&D-FDI-TG-Productivity Link. An empirical model that contains a system of 2 equations (. the production function, and the R&D equation), is estimated using data from 26 manufacturing industries over the period 1995 to 1997 in Shanghai. My results show that when we are considering the effects of FDI on R&D and productivity, we have to take into account technological gaps as well as the crowding-out effect across different industries. Hence, we should provide more incentives to industries which have larger TG.
R&D-FDI-TG-Productivity Link 3 R&D-FDI-TG-Productivity Link in Chinese Industries Many reports have shown that FDI in China has reached a new record $ billion, which accounts for 10% of the global total FDI. This large influx of FDI may be attributed to the belief that FDI can narrow the technological gap (TG), lead to a spillover effect, and technological transfer (Findlay, 1978; Walz, 1997; Saggi, 1998). Holding this belief, the Chinese government has adopted various methods to attract FDI. For example, depending on sectors, foreign-invested firms are exempt from paying income tax for two years from the first profit-making year and allowed 50% tax reduction thereafter for three years. 1However, empirical results are mixed and a consensus cannot be established. Some economists have concerned about the negative impacts of FDI on indigenous technological creation in developing countries (Stewart & James, 1982; Fransman, 1986; Kim, 1991; Lall, 1993, 2001). Recently, this concern which comes from the concept Latin-Americanization is also becoming a hot topic in China. Latin-Americanization refers to the phenomenon that in some developing countries FDI will weaken domestic R&D and crowd out the market for domestic firms. China’s automobile industry acts as a case in point. After many years of cooperation with foreign firms, domestic firms still cannot control the core technology and the R&D level is very low compared with other countries. While China has been relatively open to FDI, Korea has tended to limit FDI but relied on R&D. As time goes on, she has established her own R&D 2operation and many developed industries, such as the automobile industry. This paper tries to explore the relationships between FDI, R&D, and Productivity. What is the effect of the FDI on R&D and Productivity? Will it complement R&D or substitute it?
R&D-FDI-TG-Productivity Link 4 What’s the determinate of R&D? Is the driving force of R&D one of market pull or supply push? Do firms rely on internal funds to do R&D? This paper seeks to answer these questions. As the core of our analysis is the TG, as to build a R&D-FDI-TG-Productivity Link. With a relatively rich panel dataset from Shanghai in twenty-six two-digit industries, I can construct empirical models to test our hypothesis. From the test results, I conclude some policy implications. The remainder of this paper is organized as follows: in Section 2, an in-depth theoretical analysis takes place and the hypothesis is stated. Section 3 consists of the data and variables. In Section 4, I build a two-equation empirical model. The results will be shown in Section 5, and section 6 will conclude with policy implications. 2. R&D-FDI-TG-Productivity Link &D, Technological Transfer and Productivity Growth Innovation (R&D) and international technology transfer (ITT) are sources of the indigenous technology growth. FDI is a very important source of ITT, especially when the market fails to offer the channel to purchase the technology directly from Northern firms due to the inability to absorb the advanced technology and legal barriers such as patents. Many economists and policy makers believe that FDI will generate externalities which 3will lead to technology transfer for domestic firms, and also enhance the returns for domestic R&D. Externalities can arise from the accelerated diffusion of new technology, observing from 4the nearby foreign firms and the experienced labor turnover. Kim and Nelson(2000) suggested that imitation through the adoption of existing technologies would serve as an effective learning experience which will enhance the indigenous technological growth. Furthermore, Cohen and
R&D-FDI-TG-Productivity Link 5 Levinthal(1989) argue that R&D involves both innovation and learning, which will enhance the ability to absorb the technology transferred from FDI, thus in turn boosts the transfer efficiency. This relationship can be identified as complementary effect. However, besides the complementary relationship, FDI and R&D can also exhibit substitution relationship if R&D and technological transfer have independent and similar effects on the productivity. Aitken and Harrison(1997) suggest that the foreign presence can reduce the productivity of domestic firms, particular in short run, since foreign firms have lower marginal costs which will enhance their competitive ability compared with the domestic firms. When the foreign presence enter the domestic market, it will draw demand from domestic firms, causing them to reduce production. When domestic firms are facing a smaller market, they have to move back up their average cost curves. Then, even if FDI may have a spillover effect, the net demand productivity can also be declined. This effect will offset the positive spillover effect from FDI. We can analyze this situation in Figure 1: Figure 1 is about here In Figure 1, the positive spillover will cause the average cost curve fall from ACto AC. 0 1However, the competition from foreign presence will draw market share from the domestic firms and cause output to move back up to point B. Thus, FDI can reduce domestic firms’ productivity by increasing the average fixed cost. Overall, it is ambiguous how FDI will affect the productivity since we are not sure where B will end up on AC. 1Since R&D is also a kind of fixed costs, after the entrance of FDI, the average cost of R&D will also increase. When R&D is increased, the magnitude of the increased average cost
R&D-FDI-TG-Productivity Link 6 will be larger than before. Thus although R&D can help reduce production cost increase productivity, the average cost due to the increase of R&D will also increase. Overall it is ambiguous that under the continued injection of FDI, how the increase of R&D will affect productivity. This can be identified as a substitution relationship between FDI and R&D since FDI may reduce the productivity gain from R&D. A model to test the relationship by using an interactive term of FDI and R&D is given in Section 5. &D, FDI and the technological gap ---- a simple analysis From the analysis in Section , we can derive two effects of FDI on R&D: one is a 5complementary effect, the other is a substitution effect. Since FDI can enhance the effectiveness of R&D, firms may have the incentive to conduct R&D, then FDI will have a complementary effect on R&D. However, because FDI can also reduce the productivity from R&D due to the shrink in the market share, domestic firms may have less incentive to do R&D and rely on the cheaper imitation and spillover from FDI. Thus, the overall effect of FDI is inconclusive. Because R&D has an external effect and can enhance the ability to absorb existing knowledge, if 6FDI reduce the R&D effort, the level of R&D will be further below the social optimum. However, when we take into account the effects of TGs, the analysis will be different. The spillover effect of FDI in an industry with a larger gap tends to be smaller than the one with a smaller gap, since it is more difficult to absorb and learn the foreign advanced technology, and also the legal barrier will be stronger. Thus the productivity returns from the spillover of FDI tends to be smaller for larger gap industries. When FDI enters the host country, larger gap
R&D-FDI-TG-Productivity Link 7 industries will be hurt by both smaller gains from spillover but as well as the market share. Thus an industry with a larger gap will have smaller output and higher average cost than an industry with a smaller gap. We interpret it more clearly using the following figure 2: Figure 2 is about here In figure2, ACis the common original average cost of two industries without the entrance 0 or increase of FDI. After the entrance of FDI, due to spillover effect, AC will go down. AC is 01the average cost curve of the larger TG industry, and ACis the average cost of the smaller TG 2 industry. Because the spillover effect of the larger TG industry is smaller, AC is above ACAlso 12. because of the different competitive disadvantages, the output reduction of the larger TG industry will have larger market shrinkage than the industry with a smaller TG. Thus the output decisions of different industries will be different, B represents the output decision of the industry with a larger TG, and C represents the output decision of the industry with a smaller TG. From the figure, we can see that the industry with a smaller TG has a smaller average cost and larger market share since the spillover effect outweigh the crowd-out effect. Since the average cost of R&D is smaller in the industry with a smaller TG, it will have more incentive to invest in R&D. If we analyze the direct effect of TG on the decision on R&D effort, there will be a positive relationship between R&D and TG. Since the spillover from FDI is smaller, the industry with a larger TG has to rely on its own R&D effort. And also because R&D can enhance the effectiveness of spillover, which is more important for the industry with a larger TG, the 7incentive to do R&D for an industry with a larger gap may be larger. &D, FDI and the technological gap ---- a more complex analysis
R&D-FDI-TG-Productivity Link 8 In section , we assume FDI is the sole channel of ITT. Then FDI will have a more positive effect on R&D in industries which have relative smaller TG. This analysis can be applied to countries which are very undeveloped or industries which do not have the ability to absorb the technology from the North firms. However, it cannot hold for countries like China where TG in some industries is not very large. For these countries, imitations both from FDI and from North firms are sources of ITT. Thus, for the least developed industries which have the largest TG, FDI is the sole channel of ITT because it is much easier to learn and apply the technology from multinationals than from the North firms directly; and for the less developed industries for which TG is not too large, they can imitate both from the FDI and from North firms. When TG becomes smaller, the imitation from North firms will substitute the imitation from FDI. FDI will have less and less spillover effect on domestic technological growth as TG continues to become smaller. And also because of the smaller TG, FDI will not grasp as much market demand as in the industries with large TG. Thus the analysis based on the spillover effect and the shrinkage of market demand in section is not complete. After some critical point of TG, the smaller the TG, the less effect the FDI will have on R&D. To make it clearer, we can refer to the following figure: Figure 3 the Relationship between FDI and R&D From Figure 3, when TG is very large, FDI is the sole channel of ITT, then FDI will have a positive relationship with R&D as TG is decreasing. After the critical point, when TG is not too large, domestic firms can also imitate from North firms, then the effect of FDI on R&D will 8decrease as TG becomes smaller and smaller. Thus, we should expect TG will have a quadratic
R&D-FDI-TG-Productivity Link 9 or non-monotonic effect on the relationship between R&D and FDI as shown below: Figure 4 is about here In section 4, we derive two equations to estimate relationships between R&D, FDI and TG. In section 5, the results will be discussed. 3. Data description The econometric analyses in this study are based on data of independently calculated 9enterprises on 26 2-digit manufacturing industries from 3 statistical yearbooks and a survey conducted by CEPII. The 3 statistical yearbook are Shanghai Statistical Yearbook，Shanghai Statistical Yearbook on Science and Technology, and Shanghai Statistical Yearbook on Industry, Energy and Transport, which are edited by the Shanghai Municipal Statistics Bureau; the adopted survey is FDI and The Opening Up of China’s Economy, from which I directly adopt the measurement of comparative advantage in 1997 as a proxy for TG. Table 1 reports the summary statistics of key variables used. The combined panel dataset contains information for twenty-six manufacturing industries on the value-added, the total capital assets, the net value of fixed assets, the inventory, the R&D 10expenditure from 1995-1997 . Labor input is measured by the total number of employees rather than working hours due to the lack of data. Also because of the same reason, I cannot follow 11Chow (1993) in constructing the real capital stock. However, I use an indirect way to calculate 12the capital stock by summing of the net value of fixed assets and the inventory. For FDI, I use the percentage of foreign funded capital in total capital assets as the proxy. The revenue from products sales in the previous year is used as a proxy for the expected market demand of an
R&D-FDI-TG-Productivity Link 1 0 industry and the total after-tax profit in the previous year is used as a proxy for the availability of internal funds. These two variables will be interpreted in next section. R&D is the sum of the research expenditures incurred by large and intermediate enterprises at the industry level under the assumption that the small firms do not have the ability to do R&D, or their R&D expenditures only accounts for a small share of the total amount. We use the revealed comparative advantage by industry as the proxy of TG. The comparative advantage is calculated as the difference between an industry’s share in total exports and its share in total imports. The comparative advantage is adopted directly from a survey data have various limitations. First, it is not a balanced panel data, so that there is a sample attrition problem. However, since we have a large N (cross-sectional variables) and small T (time variables), this sample selection bias in the fixed effect framework is not as critical an issue as it could be because the first-differencing time-demean estimation will naturally correct for attrition bias. Second, because the TG is not time-variant, it cannot be used in the regression when we are using the time-demeaning operation to do econometric work. Third, the sample is restricted to Shanghai only, which introduces a potential source of selection bias from ignoring domestic technology transfer. However, under the reality that most industries in Shanghai have the most advanced technology in China, the source of technological transfer from other areas should not be a big problem, we can not eliminate the potential and possible transfer from other areas. Also, since R&D is a path-dependent process, the inability to construct a measure of knowledge capital and how innovation contributes to the productivity at a dynamic
R&D-FDI-TG-Productivity Link 1 1 14level, is another shortcoming. Besides, I did not adjust the value-added, the FDI and other current values using the price index, due to limitations. However, it is not a big problem because we only have 3-year datasets and in these three years (1995-1997), China is developing smoothly 15as a consequence of the “soft-landed” policy. Thus, we could just use the nominal values. Having acknowledged the limitations of the data, we have to admit that they provide rich information for the investigation of various important issues. 4. The Model I provide three empirical equations in this paper. The first is the production function which represents the relationship between R&D, FDI and productivity. The second and the third models specify the relationships between FDI, R&D and TG. With these 3 models, we can construct the R&D, FDI, and productivity link. The Production Function In this part, I incorporate the contribution of R&D and technological transfer from FDI into the production model to investigate the following questions: (1) whether FDI has a significant positive or negative spillover effect on productivity；(2) whether R&D input has a significant effect on the growth of productivity. We first specify a value-added Cobb-Douglas production function: αβY= AKL (1) ijijijij Where Yis the value-added in industry i, and time j, which represents the productivity; α ij
R&D-FDI-TG-Productivity Link 1 2 and β are output elasticities of capital and labor. A is the total factor productivity parameter, which is driven by R&D and technological transfer from FDI. We characterize the evolution of 16productivity by: γf(.)A= eR (2) ijijwhere f(.)= aF+bI+cT (3) ijj This evolution function characterizes the contribution from both R&D and FDI, which are two different channels of technological progress. R is the technical factor associated with R&D, ijmeasured by the R&D expenditures and γ is the elasticity of R which captures the return of R&D on the productivity of industry i; F is the technical factor associated with FDI and a captures the ijspillover effect of FDI in the manufacturing industry i; Iand Tare industry dummies and time j s dummies which are used for controlling for differences across industries and years. We obtain the production function by replacing A in (1) by (2) and (3) and taking log: lnY=a+ γlnR+aF+αlnK+βlnL+bI+cT+ε(4) ijijijijijji We have different ways to estimate the equation (4), whereas the simplest one is the OLS method. However, the simple OLS estimates do not consider the potential omitted variable bias which violates the zero conditional mean assumption. Thus, the OLS estimates will be subject to omitted variable misspecification and bias. Under this consideration, I use two ways to correct for the potential bias. With panel data,
R&D-FDI-TG-Productivity Link 1 3 17we can control for the bias using “demean” or “first-differencing” methods. This method would difference out the time invariant characteristics and allow for unbiased estimates. However, it also has a problem of eliminating the inter-industry as well as the unobservable characteristics since for a short panel, most variations are from the cross-section dimension. Another problem is the within bias that may exacerbate the bias by reducing the amount of useful information in 18variable. The results will be shown in section 4. R&D, Technological Transfer and Technological Gap Based on the analysis in section , we construct an equation to estimate relationships between FDI, TG and R&D. I use the interaction between TG and FDI to estimate the effect of TG on d(R&D)/d(FDI). Economists have long debated on what drives the R&D efforts. The demand-pull hypothesis argues that firms are pulled by market demand to do R&D whereas the supply-push hypothesis emphasizes the opportunity to carry out innovation. Both have empirical results to support their ideas. We can use sales revenue as a proxy for the market demand and TG as a 19proxy for market opportunities. Besides the sales revenue, I also incorporate the total after-tax profit as a proxy for internal funds since economists widely agree that firms may have to rely on 20internal funds to conduct R&D. The econometric model is constructed as following: lnR= a+bF+eTG+ f F*TG +cln(S)+dln(P)+Y+I＋ε(5) ijijijijiji,j-1i,j-1ij Where Ris the R&D expenditure in industry i and year j; F is FDI in industry i and ij ijyear j; TG is the technological gap which is represented by the comparative advantage, and here ij
R&D-FDI-TG-Productivity Link 1 4 I change the sign to make it more intuitively understandable. Thus a higher value of TG means a ijlarger technological gap. F*TG is the interaction between TG and FDI. S is the revenue of ijiji,j-1the product sales in industry i and year j-1. P is the total after-tax profit in industry i and year i,j-1j-1. I and Y are industry dummies and year dummies respectively. The results are shown in section 5. If we consider the imitation from North firms as a channel of ITT which as analyzed in section , we will construct the following econometric equation: 2lnR= a+bF+eTG+f F*TG+gF*(TG)+cln(S)+dln(P)+Y+I＋εijijijijijijiji,j-1i,j-1ij (6) and 23lnR= a+bF+eTG+f F*TG+gF*(TG)+hF*(TG)+cln(S)+dln(P)+Y+I＋εijijijijijijijijiji,j-1i,j-1ij (7) We just add more interactive terms between F and TG to equation (5), which allow us to test the non-monotonic relationship. 5. Results and Discussions &D, FDI and Productivity Link We report the results of the estimation of equation (4) in table 2. The industry dummies eliminate a potential source of omitted-variable bias. First, we do a simple test which omits the interaction between FDI and R&D. To correct the potential serial correlation problem, we use the random effect analysis. Then we use the Hausman Test to test the hypothesis that the Fix Effects and Random Effects are equal, which
R&D-FDI-TG-Productivity Link 1 5 does not reject the random effects at any level. The random effect results show that capital input and labor input are strongly significant at 1% level, whereas R&D and FDI are very insignificant. Then we do the same test adding the interaction between R&D and FDI. Hausman Test does not reject the random effects at any level either. The random effects results show that, R&D, FDI, and the interactive term are not significant. Thus, the results in column (1) and (2) show that FDI does not have obvious spillover effect or crowd-out effect on productivity, or the crowd-out effect offset the spillover effect; R&D does not have the positive relationship with productivity because R&D is a kind of fixed effects. The insignificant interaction shows we cannot observe the complementary relationship or substitution relationship between FDI and R&D, since they may offset each other. The test results confirm the analysis in section 2. R&D, FDI and Technological Gap I now explore the relationship between TG, FDI, and R&D, and find whether FDI will substitute or complement R&D. The results are reported in Table 3. First we do a test based on equation (4) without the interaction between TG and FDI. The Hausman Test cannot reject the random effects at any level. The results show that TG is significant and positive at 10% level, whereas FDI is very insignificant. The coefficient on TG means that the larger the TG, the more incentive the firms have to do R&D, which is in line with my analysis in section 2. Then I proceed to add the interaction between R&D and FDI. The Hausman Test shows that we cannot reject the random effects at any level. The results show that TG, FDI, and the interactive term are not significant. This result does not support the analysis in section which suggests that the interaction term should be significant and negative. It may be because in China, some industries can imitate both FDI and North firms, thus the relationship between TG and d(R&D)/d(FDI) may be non-monotonic. Thus we test the equation (6) and equation (7) using random-effects models given Hausman test does not reject the random effects at any level in both equations. The interactive 23term FDI*(TG) is not significant, whereas the interactive term FDI*(TG) is significant at 10% level. Although we cannot get the ideal quadratic relationship shown in section , we can still
R&D-FDI-TG-Productivity Link 1 6 draw the conclusion that the relationship between TG and d(R&D)/d(FDI) is non-monotonic, and 21with a better dataset, we more easily see what the precise relationship is. The results in all tests show that FDI does not have complementary effect for R&D, which confirms our analysis in section 2. Also, the test results show that R&D is pulled by the market demand since the coefficient of the sales revenue is positive and significant. The generally significant and positive coefficients of TG confirm the hypothesis that it will have more incentive to conduct R&D in industries with larger gap. The interaction between FDI and TG shows the monotonic relationship between TG and R&D, which is also in line with the analysis in section 2. 6. Conclusions and Policy Implications I first analyze the relationship between R&D, FDI and productivity and point out that since FDI can also draw market demand from domestic firms, it is not conclusive that FDI will have the positive spillover effect on productivity. And also FDI has a crowding-out effect, as a kind of fixed cost, R&D may increase the average cost. Thus FDI and R&D may exhibit a substitution relationship as well as a complementary relationship on productivity which is suggested by many studies. And in section and , I tried to do a deeper analysis on how FDI will affect R&D if we take into consideration the TG in different industries. Our analysis shows that if FDI is the sole channel of ITT, the larger the TG, the smaller the effect of FDI will have on R&D. And if the industries can imitate both from FDI and North firms, the relationship will be quadratic or non-monotonic. In this section, I also analyze other relationships such as TG and R&D. In section 3 I construct empirical models consisting of a two-equation system (. a Cobb-Douglas production function and a R&D expenditure equation with different interactive terms). I estimate the models using the panel data from Shanghai. And the results basically confirm the analysis in section 2. From the analysis and empirical tests, we do not find FDI has a positive relationship with productivity, or FDI can strongly promote or substitute for R&D as is suggested by other papers.
R&D-FDI-TG-Productivity Link 1 7 Thus, two important implications follow our study. First, although people widely admit that FDI may have spillover effects on productivity, in the short run, and particularly in industries with larger TG, the spillover effect is not obvious and FDI may even draw demand from domestic firms. Thus, when we are encouraging the influx of FDI, we may have to put more emphasis on R&D since the positive effect of FDI may be overestimated. Second, since the effect of FDI on R&D may depend on the technological gaps, we may need to have different policies on different industries. For industries with higher TG, FDI may have a negative effect on R&D, thus we should put more emphasis on the innovation in these industries whereas for industries with smaller TG, they will learn from both FDI and North firms and have more incentive to do R&D, thus we do not need to specially back these industries. The current paper is only one step towards a better understanding of the R&D-FDI-TG-Productivity Link. Due to limitations of data and time, our empirical analysis has many shortcomings. If we can get a better data with a longer time span, it may be more interesting to do analysis in the dynamic dimension. And also, the same results and analysis can be conducted in other developing countries.
R&D-FDI-TG-Productivity Link 1 8 References Aitken, Brian J. and Ann E. Harrison (1997), Do Domestic Firms Benefit from Foreign Direct Investment? Evidence from Panel Data, Mimeo, Columbia University. Aitken, Brian J. and Ann E. Harrison (1999), Do Domestic Firms Benefit from Direct Foreign Investment? Evidence from Venezuela, American Economic Review, 89(3)：605－18. Balassa, B. (1965) Trade Liberalization and Revealed Comparative Advantage, Manchester School, 33, 2, 99-123. Balassa, B. (1977) Revealed Comparative Advantage Revisited: An Analysis of Relative Export Shares of the Industrial Countries, 1953-71, Manchester School, 45, 4, 327-344. Caves, Richard E. (1971) International Corporations: The industrial Economics of Foreign Investment. Economica, 38(149), pp. 1-27. Cohen, Wesley M. and Richard C. Levin (1989), Empirical Studies of Innovation and Market Structure, Handbook of Industrial Organization, Volume 2, pp. 1059-1107, North-Holland: Amsterdam and New York. Cohen, Wesley M. and Daniel A. Levinthal (1989), Innovation and Learning: The Two Faces of R&D, Economic Journal, 99(397): 569-96 Chow and Gregory (1993), Capital Formation and Economic Growth in China. Quarterly Journal of Economics. 108(3): 809-842. Edfelt, Ralph B. (1975) Direct Investment in a Developing Economy: Towards Evaluating the Human Resource Development Impact in Brazil. . dissertation, University of California, Los Angeles. Findlay and Ronald (1978), Relative Backwardness, Direct Foreign Investment, and the Transfer of Technology: A simple Dynamic Model, Quarterly Journal of Economics, 92(1):1-16. Francoise Lemoine (.) FDI and Opening Up of China’s Economy, CEPII, N 2000-N11 Fransman, Martin (1986), Technology and Economic Development, Boulder, CO: Westview Press. Germidis, Dimitri, (1977). Transfer of Technology by Multinational Corporations. Paris:
R&D-FDI-TG-Productivity Link 1 9 Development Center of the OECD. Glass, Amy. J. and Kamal. Saggi (1998), International Technology Transfer and the Technology Gap. Journal of Development Economics. 55(2): 369-298. Goncalves, Reinaldo. (1979) Foreign Direct Investment and ‘Spillover’ Efficiency Benefits in Canadian Manufacturing Industries. Canadian Journal of Economics, 12(1): 42-56. Griliches, Zvi (1984) Patent, R&D, Productivity, Chicago: University of Chicago Press. Griliches, Zvi (1979) Issues in Assessing the Contribution of Research and Development to Productivity Growth Bell Journal of Economics, 10(1):92-116. Haddad, Mona and Ann Harrison (1993), Are There Positive Spillovers from Direct Foreign Investment? Evidence from Panel Data for Morocco, Journal of Development Economics, 42(1): 51-74. Hall, Bronwyn H. (1992) Investment and Research and Development at the Firm Level: Does the Source of Financing Matter? NBER Working Paper no. 4096. Cambridge, MA: National Bureau of Economic Research. Helleiner, Gerald K. (1989) Transnational Corporations and Direct Foreign Investment, in Hollis Chenery and T. N. Srinivasan, eds., Handbook of Development Economics. Amsterdam: North-Holland. Hu, Albert G. Z. and Gary H. Jefferson (2002). FDI Impact and Spillover: Evidence from China’s Electronic and Textile Industries, World Economy, 25(8): 1063-76. Hu, Albert G. Z. and Gary H. Jefferson (2004). Returns to Research and Development in Chinese Industry. Evidence from State-Owned Enterprises in Beijing, China Economic Review, 15(1): 86-107. Hu, Albert G. Z. and Gary H. Jefferson (2004). R&D and Technology Transfer: Firm-level Evidence from Chinese Industry, Forthcoming, Review of Economics and Statistics. Javorcik, Beata Smarzynska (2004), Does Foreign Direct Investment Increase the Productivity of Domestic Firms? Is Search of Spillovers Through Backward Linkages, American Economic Review, 94(3): 605-627.
R&D-FDI-TG-Productivity Link 2 0 Jaffe, Adam B. (1986) Technological Opportunity and Spillover of R&D: Evidence from Firm’s Patents, Profits and Market Value. American Economic Review. 76(5): 984-1001. Jefferson, Gary H. and Zhong Kaifeng (2004). An Investigation of Firm-level R&D Capabilities in East Asia, in Yusuf, M. Anjum Altaf and Kaoru Nabeshima, eds., Global Production Networks and Technical Change in East Asia, World Bank and Oxford University Press, Washington, . Jones, Charles I. and John C. Williams (1998), Measuring the Social Return to R&D, Quarterly Journal of Economics, 113(4): 1119-35. Kim, Linsu (1991). Pros and Cons of International Technological Transfer: A Developing Country’s View, in T. Agmon and . von Glinow (eds,), Technology Transfer in International Business, -39, Oxford: Oxford University Press. Kim, Linsu and Richard R. Nelson, (2000) Technology, Learning and Innovation: Experiences of Newly Industrializing Economics, Cambridge: Cambridge University Press. Lall, Sanjaya (1993), Promoting Technology Development: The Role of Technology Transfer and Indigenous Effort, Third World Quarterly, 14(1): 95-108. Lall, Sanjava (2001), Competitiveness, Technology and Skills, Cheltenham, UK: Edward Elgar. Liu, Zhiqiang (2002), Foreign Direct Investment and Technology Spillover: Evidence from China, Journal of Comparative Economics, 30(3): 579-602. Mansfied, Edwin and Romeo, Anthony. (1980). Technology Transfer to Overseas Subsidiaries by -Based Firms, Quarterly Journal of Economics, 95(4): 737-50. Pakes, Ariel, and Schankerman, Mark, (1984). An Exploration into the Determinants of Research Intensity. In Zvi Griliches, Ed., R&D, Patents, and Productivity, -232. Chicago: University of Chicago Press. Rhee, Jong Wong, and Belot, Therese (1989). Export Catalysts in Low-Income Countries. Working Paper, World Bank. Rosenberg, Nathan (1974). Science, Invention and Economic Growth. Economic Journal. 84(333): 90-108.
R&D-FDI-TG-Productivity Link 2 1 Scherer, Frederic M. (1992) Demand-Pull and Technological Innovation: Schmookler Revisited. Journal of Industrial Economics. 30(3): 225-238 Schmookler, Jacob, (1966). Invention and Economic Growth. Cambridge, MA: Harvard University Press. Shanghai Municipal Statistics Bureau (1995-1998), Shanghai statistical yearbook，Shanghai statistical yearbook on science and technology, and Shanghai statistical yearbook on industry, energy and transport. Beijing, China: Statistical Publishing House. Stewart, Frances and Jeffrey James (1982), The Economics of New Technology in Developing Countries, Boulder, CO: Westview Press. Teece, David J. (1977), Technology Transfer by Multinational Firms: The Resource Cost of Transferring Technological Know-how, Economic Journal, 87(346): 242-61 Vernon, Raymond (1966), International Investment and International Trade in the Product Cycle, Quarterly Journal of Economics 80, 190-207. 22Walz, Uwe, (1997). Innovation, Foreign Direct Investment and Growth. Economica 64: 63-79, Westphal, Larry E. (1990) Industrial Policy in an Export Propelled Economy: Lessons From South Korea’s Experience, Journal of Economic Perspectives, 4(3):41-59. Yusuf, Shahid and Simon J. Evenett (2002) Can East Asia Compete?, The World Bank and Oxford University Press, Washington, .
R&D-FDI-TG-Productivity Link 2 2 Footnotes 1 See Haddad and Harrison (1993), Aitken and Harrison(1999), Javorcik(2004), Germidis (1977), Rhee and Belot(1989), Mansfield(1980) 2 See Yusuf and Evenett(2002), Jefferson and Zhong(2004), and Westphal(1990) 3 See Caves (1982), and Helleiner (1989) for surveys of technology transfer and FDI. 4 See Teece(1977), Edifelt(1975), and Gonclaves(1986). 5 See Stewart and James (1982), Fransman (1986), Kim(1991), and Lall (1993, 2001) 6 See Jones and Williams (1998), they show that actual investment in R&D is always below the social optimum. 7 Please do not confuse this with the above analysis which considers the effect of entrance of FDI on the R&D effort. These two analyses are different since one analyzes the situation with the increased FDI, whereas the other only analyzes the relationship without considering the FDI. 8 I am trying to extend the analysis in a dynamic direction and construct models that incorporate R&D growth. However, due to the lack of data, an empirical study is not carried out. 9 The 26 manufacturing industries are food processing and manufacturing, beverage manufacturing, tobacco processing, textile, garments and other fiber products, leather( including furs and related products), timber processing, furniture, paper products, printing and record medium, educational and sports goods, petroleum processing, chemical material and products, medical and pharmaceutical products, chemical fiber, rubber products, plastic products, nonmetal mineral products, processing of ferrous metals, processing of nonferrous metals, metal products, ordinary machinery, transport equipment, electronic and telecommunications equipment, instruments, and other manufacturing. 10 The data sets from the 3 statistical yearbooks in different years do not match each other
R&D-FDI-TG-Productivity Link 2 3 perfectly; thus after careful consideration, only 3 years’ data information is adopted. 11 The method by Chow, please refer to his work with Gregory “Capital Formation and Economic Growth in China” in 1993, which offers a better way to deal with this issue. 12 I use the inventory as a proxy for working capital. Although it is not perfect, it is not a bad proxy since the inventory accounts for much part of the working capital. 13 There are many ways to calculate the revealed comparative advantage. See the basic works by Balassa, B. (1965/1977). Due to the complexity of the China Custom Statistics Yearbook, I directly use the results of the survey in 1997 as the proxy, on the assumption that the comparative advantage in a short period (3 years in this case) will not have obvious change dramatically. 14 Actually there are some papers which deal with this issue very well, see Hu and Jefferson (2003), (2004). In their papers, they construct a method to calculate the R&D capital. Due to the time constraint, I did not derive the R&D capital stock. 15 Many papers do not make the adjustment due to the complex of the comparison. However, there are also some exceptions in which the authors succeeded in making price adjustment, see Liu (2002), Aitken and Harrison (1999). 16 Griliches(1979) was the first to model the contribution of R&D to productivity in a micro production function framework. 17 Besides this method, I also used the instrumental variable (IV) approach. However, it is hard to find instruments to apply IV method since it is not easy to find a variable which correlates with the independent variables but not the unobservable. I used the industry and year dummies as the instruments to have a try, but I am not confident about the industry-specific bias, thus in this version of this working paper, I only show one approach to do econometric analysis. 18 See Griliches (1984) for a stylized finding that significant returns to R&D are usually in
R&D-FDI-TG-Productivity Link 2 4 cross-section dimension. Also see Hu and Jefferson (2004). 19 Interested readers can refer to Schmookler(1966) and Rosenberg(1974) to find the basic literature for this argument. Also, Scherer (1982), Pakes and Schankerman(1984), and Jaffe(1986) have contributed empirical studies. 20 See studies by Hall (1992), who finds a positive impact of internal cash flow on R&D expenditure. 21 I believe that with a better dataset and adjustment, in dynamic direction, I can have much better results. Author Notes
R&D-FDI-TG-Productivity Link 2 5 LU Qian, School of Economics and Finance, The University of Hong Kong. I thank Dr. William Chan, Associate Professor of Economics in The University of Hong Kong for many important suggestions and kindly helps. I also thank Anil for his wonderful suggestions on writing. This paper is for the course Econ0606 Current Economic Affairs of The University of Hong Kong. Correspondence concerning this article should be addressed to LU Qian, 1406A, Starr Hall, Pokfulam Road, The University of Hong Kong, HKSAR, China. E-mail: luqian0807@ Table 1
R&D-FDI-TG-Productivity Link 2 6 Summary Statistics of Key Variables Used Variable | Obs Mean Std. Dev. Min Max -------------+----------------------------------------------------- date | 78 1996 .8217814 1995 1997 industry | 78 1 26 R&D | 67 11 529589 LnR&D | 67 .8869028 FDI | 77 .2745564 .1395298 .0063728 .571086 valueadded | 78 tg97 | 78 labor | 78 .7 capital | 78 sales | 78 profits | 78 Table 2
R&D-FDI-TG-Productivity Link 2 7 Production Function Regression (26 manufacturing Industries, 1995-1997) abVariables (1)(2)lnR () () fdi () () ******lnK () () ******lnL () () FDI_lnR - - () Industry dummies Yes Yes HN 2R Number of obs. 66 66 Number of groups. 24 24 Note. GLS random effects test; figures in parentheses are t statistics. The dependent variable is log( value-added). HN: Hausman statistics for testing the random-effects model against the fixed-effects model. Numbers are p-values.
R&D-FDI-TG-Productivity Link 2 8 2R is the within R-square a Random-effects specification without the interaction between FDI and lnR. b Random-effects specification with the interaction. * Significant at the 10% level. ** Significant at 5% level. *** Significant at 1% level. Table 3
R&D-FDI-TG-Productivity Link 2 9 Random Effects estimates of R&D abcdVariables (1)(2)(3)(4)fdi ***tg97 ************lns lnp () fdi_tg1 - - fdi_tg2 - - - - *fdi_tg3 - - - - - - Industry dummies Yes Yes Yes Yes HN 2R Number of obs. 65 65 65 65
R&D-FDI-TG-Productivity Link 3 0 Number of groups. 24 24 24 24 Note. GLS random effects test; figures in parentheses are t statistics. The dependent variable is log( R&D). HN: Hausman statistics for testing the random-effects model against the fixed-effects model. Numbers are p-values. 23fdi_tg1 is the interactive term fdi*tg, fdi_tg2 is fdi*(tg), fdi_tg3 is fdi*(tg). 2R is the within R-square a Random-effects specification without the interaction. b Random-effects specification with the interaction fdi_tg1. c Random-effects specification with the _tg2. d Random-effects specification with the _tg3. * Significant at the 10% level. ** Significant at 5% level. *** Significant at 1% level. Figure Captions
R&D-FDI-TG-Productivity Link 3 1 Figure 1. Output Response to FDI. Figure 2. Different Output Responses to FDI in different Industries. Figure3. Relationship between FDI and R&D as TG increases. Figure4. Relationship between TG and d(R&D)/d(FDI). Figure 1. Output Response to FDI.
R&D-FDI-TG-Productivity Link 3 2 Adopted from Aitken and Harrison (1997), drawn by author. Figure 2. Different Output Responses to FDI in different Industries.
R&D-FDI-TG-Productivity Link 3 3 Drawn by author Figure3. Relationship between FDI and R&D as TG increases.
R&D-FDI-TG-Productivity Link 3 4 Drawn by author Figure4. Relationship between TG and d(R&D)/d(FDI).
R&D-FDI-TG-Productivity Link 3 5 Drawn by author.
