Journal of Asian Economics

Transcription

1 Journal of Asian Economics 20 (2009) Contents lists available at ScienceDirect Journal of Asian Economics China s post-economic reform growth: The role of FDI and productivity progress Chyau Tuan a, Linda F.Y. Ng a, *, Bo Zhao b a Department of Decision Sciences and Managerial Economics, Faculty of Business Administration, The Chinese University of Hong Kong, Hong Kong, China b Stephen M. Ross School of Business, University of Michigan, USA ARTICLE INFO ABSTRACT JEL classification: F21 O16 Keywords: FDI TFP Growth Productivity China China has become the top FDI destination among all developing countries and remained host to the world s largest share of foreign direct investment receipts since its accession to the WTO in Given the impressive growth performance and FDI influx into China, the two globalized delta economies (GDEs), Pearl River Delta (PRD) and Yangtze River Delta (YRD), have continued to out-perform all other regions in China in terms of FDI absorption and growth. The role that inward FDI plays in the process of regional development and the exact channels through which economic growth would be affected are investigated by panel data estimations of the GDEs at the city level since China s economic opening and reform. This research shows consistent results with some recent findings in other developing countries in that FDI exerted spillover effects and affected productivity (TFP) growth of the recipients. While TFP was found to be increasing overtime in the GDEs cities and facilitated economic growth in both PRD/YRD regions, major technology- and knowledge-related factors including R&D and human capital other than FDI also played critical roles in TFP enhancement and regional growth. The endogeneity of TFP and the simultaneous relations of FDI in affecting TFP and output growth are also addressed in this regard. ß 2009 Elsevier Inc. All rights reserved. 1. Introduction China has become the top foreign direct investment (FDI) destination among all developing countries and remained host to the world s largest share of FDI receipts since its accession to the WTO in Together with the extremely impressive record of FDI influx since China s economic reform in 1979, China has also experienced remarkable economic growth and development by achieving a high growth rate of almost 10% per annum, on average, and over 10% during the period after Deng s speech in 1992 reaffirming China s continuous economic opening. Given such a notable economic growth performance, the quest of China s sustainable growth continues to stimulate much discussion and vigorous debates among academics during recent years. However, in studying China s growth, special attention has to be paid to the two major globalized delta economies (GDEs), Pearl River Delta (PRD) and Yangtze River Delta (YRD), located respectively in the southand east-coastal China. 1 * Corresponding author. Tel.: ; fax: address: lindang@cuhk.edu.hk (Linda F.Y. Ng). 1 According to the official definition, the PRD economic region, which comprises of 9 cities/counties in Guangdong Province, has an area of 41.5 thousand square kilometers. The YRD economic region, which comprises of 16 cities (and officially know as ), includes 8 cities in Jiangsu Province, 7 cities in Zhejiang Province, and a metropolitan city (municipality), Shanghai, has a total area of thousand square kilometers. For a visual inspection of the economic indicators and investment environment of PRD and YRD, please see Tuan, Ng, and Lin (2006, /$ see front matter ß 2009 Elsevier Inc. All rights reserved. doi: /j.asieco

2 C. Tuan et al. / Journal of Asian Economics 20 (2009) Table 1 Realized FDI and real GDP growth in the globalized delta economies ( ). Year China Greater Yangtze River Delta (GYRD) a Greater Pearl River Delta (GPRD) a Greater YRD Shanghai Jiangsu Province Zhejiang Province Guangdong Province Real GDP (100 million RMB) (100) (18.2) (7.5) (7.1) (3.6) (5.1) (100) (20.6) (6.2) (9.3) (5.1) (8.8) (100) (24.7) (7.0) (11.1) (6.5) (10.9) (100) (28.2) (7.9) (12.7) (7.5) (12.8) (100) (29.2) (8.2) (13.2) (7.8) (13.3) (100) (29.8) (8.2) (13.6) (7.9) (13.7) average (100) (27.4) (7.8) (12.4) (7.2) (12.3) Realized FDI (US$ 100 million) (100) 29.6 (26.9) 12.6 (11.4) 14.0 (12.7) 2.9 (2.6) 35.5 (32.3) (100) (26.7) 48.1 (10.6) 57.9 (12.8) 15.0 (3.3) (25.9) (100) (50.7) 58.5 (10.9) (29.5) 54.5 (10.2) (29.1) (100) (41.8) 65.4 (10.8) (20.0) 66.8 (11.0) (16.5) (100) (46.0) 68.5 (11.4) (21.8) 77.2 (12.8) (20.5) average (100) 86.7 (36.1) 25.9 (9.6) 44.8 (16.6) 16.0 (5.9) 70.6 (26.2) a Notes: Provincial level data is used as approximation for the measurements of the two globalized delta economies (GDEs), Yangtze River Delta (YRD) and Pearl River Delta (PRD) and denoted as the Greater regions; percentage share of the total in parentheses. Source: Compiled from the Statistical Yearbooks of China, Guangdong, Jiangsu, Zhejiang and Shanghai, China s economic reform and opening policy launched in 1979 successively made PRD and YRD the two most dynamic regions in China. 2 These two GDEs represented the two most dominating, fastest growing regions and continued to outperform all other regions in China in terms of GDP performance and FDI absorption. 3 The PRD and YRD, which together account for only 1.4% of China s total land area and 7.7% of the total population, 4 have contributed (on average) 12.3% and 27.4%, respectively to China s national GDP in recent years; they absorbed 26.2% and 36.1%, respectively, of the total national inward FDI during the post-reform period (Table 1). In addition, the GDP growth rates of PRD and YRD reached an average of 13.8% and 11.9%, respectively, during the period from 1979 to 2005, a much higher pace than the national average growth rate of 9.7% during the same period. Detailed statistics of the GDP performance and realized FDI of the two delta regions during the post-reform period are presented in Table 1, and relevant basic economic indicators comparing the two regions are further presented in Table 2. Among the studies of economic growth, Krugman s (1994) input driven framework has been used in the literature to examine whether or not China will face the same problems of other Asian countries in terms of excessive capital and labor input accumulation without rising production efficiency (Collins & Boseworth, 1996). High rates of factor input accumulation have been posited as a possible explanation in the case of China s remarkable growth. Until recently, inward FDI has been considered a critical factor among the many factors contributing to sustained economic growth in China (Wei, 1995; Borensztein, de Gregorio, & Lee, 1998; Wu, 1999; Wei & Liu, 2001; Graham & Wada, 2001; Whalley & Xin, 2006; Tuan & Ng, 2004, 2007; Ng & Tuan, 2006; Yao & Wei, 2007). 5 Although inward FDI is believed to have a positive significant effect on economic growth, the exact mechanism of how FDI has impacted the development process is far from being well understood (Yao & Wei, 2007; Tuan & Ng, 2007). The fascinating developments in China s globalization also provide us with a tempting opportunity to study the role of inward FDI in the country s economic growth. In this connection, the total factor productivity (TFP) framework may provide us with an effective tool to measure production efficiency and, in addition, provide some evidence regarding the sources of economic growth in the context of FDI absorption. The early international economics literature suggests that FDI is a significant source of innovation and technology transfer (Caves, 1974; Findlay, 1978; Mansfield & Romeo, 1980). It is believed that through either the multinationals competitive force or a demonstration effect, locally owned firms operating in imperfect markets may be induced to achieve higher levels of technical or X-efficiency (Leibenstein, 1966). FDI not only serves as a capital injection to the domestic market but also plays a central role for technological spillover and advancement of managerial skills. FDI is believed to be embedded with new technologies and know-how not available in the host countries and could also accelerate the speed of the adoption of technology as well as improve production efficiency in the host countries. 2 Following the implementations of preferential FDI policies in PRD, Guangdong was first designed as the showroom for FDI promotion in the early 1980s and in YRD, including Shanghai and 15 selected cities in Jiangsu and Zhejiang provinces, was further opened up to receive inward FDI in For the stepwise economic opening of China to receive FDI and the implementations of FDI preferential policies by stages, please see Ng and Tuan (2001). 3 The reconfirmation of China s commitment to economic opening in 1992 has stimulated a much larger volume of inward FDI of more diversified sources, including those from Korea, Europe, and U.S.A., other than the original dominating sources from Hong Kong and Taiwan. 4 When measured at the provincial level to be denoted as the Greater regions as presented in Tables 1 and 2, the two Greater regions together accounted for 4% of China s total land area and 16.6% of the total population. 5 For a review of literatures on FDI and regional growth theories, see Berthelemy and Demurger (2000), Lim (2001), Mullen and Williams (2005), Brock (2005), andwhalley and Xin (2006).

3 282 C. Tuan et al. / Journal of Asian Economics 20 (2009) Table 2 Brief profiles of the globalized delta economies: major economic indicators ( ). Economic indicators Greater Yangtze-River Delta (GYRD) a Greater Pearl-River Delta (GPRD) a Shanghai Zhejiang Province Jiangsu Province Guangdong Province Land area and population Land area (thousand km 2 ) Average population (year-end) (10 thousand person) Population density (person/km 2 ) Output performance Real GDP in 1978 price (100 million RMB) Real GDP per capita in 1978 price (RMB per person) Investments Investment in fixed assets (100 million RMB) Realized FDI (US$ million) R&D expenses R&D expenditures (100 million RMB) R&D expenditures as percentage of GDP (%) Employment Average employment (10 thousand person) Education Number of student enrollment in higher education (10 thousand person) Student enrollment in higher education as percentage of population (%) Number of student enrollment in secondary education (10 thousand person) Student enrollment in secondary education percentage of population (%) Source: Compiled from the Statistical Yearbooks of China, Guangdong, Jiangsu, Zhejiang and Shanghai, a Notes: AsinTable 1. From a technical perspective, technological aspects have been considered to relate more to innovation, other than R&D, where firms would draw improvements from technology advancements in sciences and technical progress as sources of innovation. FDI is believed to transfer technology and technological know-how to the host countries via channels such as spillovers, demonstration, transfer of management know-how, and competitive effects (Teece, 1977; Aitken & Harrison, 1999; Blomstrom & Kokko, 2001; Keller, 2004; Javorcik, 2004). The presence of multinational subsidiaries in an industry may speed up the process or lower the cost of the technology transfer. Moreover, the threat of competition may also stimulate domestic firms to innovate. Both imitation effects and the movement of personnel trained by multinational subsidiaries would enhance the transfer of technology to local firms (de Mello, 1997). Moreover, the location and spatial agglomerations of the foreign and local firms and their interactions and synergies would contribute to regional economic growth (Ng & Tuan, 2006). In this context, the impressive growth phenomena observed in PRD/YRD have given rise to a number of interesting research questions. What role does inward FDI play in the process of regional growth and development in these two delta regions and exactly how or through what channels would FDI affect growth, particularly in a huge country like China? By focusing on the regional development in the GDEs at the city level, the aims of this study include: (1) to measure production efficiency by TFP and make comparisons over time and across the cities in the GDEs; (2) to study the role and effects of inward FDI on TFP growth; (3) to measure the contributions of other technology-related factors including R&D and human capital on TFP; and (4) to examine the effects of FDI and TFP on growth. The endogeneity of TFP in facilitating economic growth is further addressed. The research findings of this study may provide important implications and direction in the future studies of regional development in China. The rest of this paper is organized as follows. Section 2 reviews the major literature on the relationship between FDI, productivity progress, and economic growth. Section 3 describes the research hypotheses, estimation methods, and the data. Section 4 presents the research findings, and Section 5 provides some concluding remarks. 2. FDI, productivity progress, and growth: literature review 2.1. Evidence on TFP and growth in China Following the neoclassical framework of economic growth (Solow, 1957) focusing on factor accumulation and productivity growth, TFP is well-recognized as a possible measurement of factor productivities in the process of economic growth. The relationships among technology change, factor productivities, and economic growth have been extensively debated and reported in a voluminous literature on growth. According to the well-known East Asian Miracle (World Bank,

4 C. Tuan et al. / Journal of Asian Economics 20 (2009) ) study, the rapid economic growth experience of the eight East Asian economies over the previous decade was mainly attributable to TFP growth. Krugman (1994), however, argued that the secret of the rapid growth in the four Asian NIEs was input-driven, which means that economic growth depended heavily on factor inputs but had nothing to do with TFP or production efficiency improvements. The issue of accumulation versus assimilation in the growth of East Asian countries has also been delineated (Collins & Boseworth, 1996). Although some scholars have pointed out that there may still exist some problems under the notion and measurement of TFP (Felipe, 1997) 6 and a call for a theory of TFP is considered warranted (Prescott, 1998), studying the growth process in both developed and developing countries is still a major focus. In China, the study of TFP growth is quite limited and the current available studies are mainly based on some highly aggregate data at country or provincial level. Regional TFP studies using more disaggregated data at the city/county level are particularly scarce due to either lack of information or of data. Among the few studies, Zhang, Wu, & Zhang (2004) observed that China s TFP using country-level data has begun to improve since her economic reform but the TFP gain in 1979 is rather small when compared with that in Arayama and Miyoshi (2004) computed provincial TFP and concluded that the regional diversity and growth in China could be attributable to the sources of factor inputs, human capital, and TFP growth Evidence on TFP enhancement: FDI and technology-related factors FDI is believed to transfer technology and technological know-how via channels such as technology spillovers, demonstration effect, transfers of management know-how, and competitive effects. Recent voluminous research papers have discussed how FDI, especially that from developed countries, would facilitate technology flows into the recipients via technology advancements, R&D spillover activities, and innovations (Teece, 1977; Coe & Helpman, 1995; Coe, Helpman, & Hoffmaister, 1997; Aitken & Harrison, 1999; de Mello, 1997, 1999; Hejazi & Safarian, 1999; Liu, Siler, Wang, & Wei, 2000; de la Potterie & Lichtenberg, 2001; Campos & Kinoshita, 2002). The impact of FDI on the technological catching up, especially made by developing countries, can become more important than its role as a source of capital (Molero, 1995; Mytelka & Barclay, 2004). Both direct and indirect (spillover) effects of inward FDI on productivity enhancement should be in evidence. As noted by Balasubramanyam, Salisu, and Sapsford (1996), it is the ability of FDI to transfer not only production knowledge but also managerial skills that distinguishes it from all other forms of investment, including portfolio investment and foreign aid. FDI is further considered an effective way to introduce advanced technology to the host countries (Haskel, Pereira, & Slaughter 2002). However, mixed results have been cited. It is argued that a country has to reach a minimum human capital threshold in order to benefit from FDI (Xu, 2000). Moreover, the results of some other studies demonstrate an inverse relationship between FDI and industrial productivity and an absence of evidence for positive spillover in the host countries (Haddad & Harrison, 1993; Aitken & Harrison, 1999). In studying the contributions of FDI to developing countries, country evidence shows that FDI is critical in facilitating technology upgrade in the host country via spillovers and interactions of foreign and local firms. In Venezuela, a positive long-run relation between TFP and FDI is found (de Mello, 1999). The cases of Trinidad/Tobago and Costa Rica further show that the strategic use of FDI for innovation has proven the significant goals of FDI in the long-term development process. Its impacts would benefit more than a single enterprise and further strengthen local innovation capabilities (Mytelka & Barclay, 2004). In Argentina, MNCs subsidiaries are found to play an active role in the technology progress (Marin & Bell, 2006). In the case of Africa, FDI is claimed to be an effective medium for technology transfer through industry technology spillovers such as in Zambia (Bwalya, 2006) and Morocco (Haddad & Harrison, 1993; Gorg & Strobl, 2001). Based on micro (firm)-level data from Lithuania, evidence further suggests the existence of positive productivity spillovers from FDI to the host country through the contacts between foreign affiliates and their local suppliers in the upstream sectors and between shared domestic and foreign ownership (Javorcik, 2004). In the case of China, Fang and Parker (2004) computed the TFP of the Chinese electronics industry and confirmed that while the performance under all ownership forms has improved, productivity in state-owned enterprises continues to lag behind productivity in collectives and joint ventures, including wholly foreign-owned firms. Evidence from the electronics and textile industries in China further reported that inward FDI enhances the productivity of FDI-receiving firms but depresses that of domestic firms (Hu & Jefferson, 2002), which is consistent with the market-stealing explanation offered by Aitken and Harrison (1999). Moreover, FDI by ownership and industry type affects both the productivity and the growth of the FDI recipient. Micro (firm)-level evidence from the manufacturing joint ventures in China further show that foreign investments equipped with higher technology and innovation capabilities have enabled those joint ventures to achieve much higher industry technology performance than their local Chinese counterparts (Ng & Tuan, 2005). Some further exploration of the determinants of TFP is meaningful. Among many possible factors being suggested to contribute to production efficiency, evidence indicates that technology-related factors do matter in TFP enhancement. Direct links between human capital and TFP may exist (Islam, 1995; Benhabib & Spiegel, 1994; Liu & Wang, 2003). Related factors being further examined include technology and entrepreneurial background (Wu, 1999); outward orientation and human capital (Miller & Upadhyay, 2000); firms internal environment, such as human capital, and external environment, such as 6 A survey of TFP growth in East Asia can be found in Felipe (1997). An overview of the TFP measurements can be found in Carlaw and Lipsey (2003) where productivity has been argued as not a measure of technology change.

5 284 C. Tuan et al. / Journal of Asian Economics 20 (2009) FDI (Yao, 2006); R&D intensity (Murakami, 2007); and FDI and licensing (Yasar, Paul, & Morrison, 2007). Based on more detailed Japanese firm-level data, Todo (2006) confirmed the positive spillover effects of R&D incurred by foreign firms on improving the productivity of local firms. Cheung and Lin (2004) chose provincial patent applications data in China as a measure of innovation and found positive spillover effects of FDI on the number of domestic patent applications, which highlighted a demonstration effect of FDI. In sum, the above evidence suggests that the critical role of FDI in technological transfer and diffusion and innovative activities in the host country should in no way be underestimated. Transnational corporations not only set the pace for technological change but also influence the opportunities for learning and innovations. Multinationals constitute a salient actor in providing global generation of innovations in the form of R&D and innovative activities, both at home and abroad, as well as in the acquisitions of existing R&D laboratories or green-field R&D investment of the recipients (Archibugi & Iammarino, 2002). 3. Research methodology 3.1. Research hypotheses Five research hypotheses are postulated to examine China s productivity growth since its economic opening and the factors that may have produced significant impact on regional productivity and economic growth with specific reference to the two GDEs, that is, PRD/YRD, in China. TFP and its changes are adopted as a convenient measurement of productivity progress as well as in the understanding of the sources of productivity changes. Hypothesis 1. TFP is hypothesized to increase after China s post-economic reform such that productivity improves along with economic opening over time. Hypothesis 2. TFP is positively affected by factors that might bring along and/or facilitate innovative ideas and technology progress that contribute to productivity enhancement. The effects of such factors, which include inward FDI, research and development activities (R&D), and human capital endowments, are examined in such a way as to test whether: (a) inward FDI plays a positive key role in enhancing regional productivities; (b) R&D activities have significant positive impacts on productivity enhancement; and (c) human capital, which refers to the stock of productive skills and technical knowledge embodied in labor, has a positive impact on productivity growth. Hypothesis 3. TFP is endogenous to economic growth such that FDI, R&D, and human capital affect growth positively through TFP. Hypothesis 4. FDI directly and positively affects economic growth. Hypothesis 5. The human capital endowment directly affects economic growth, such that the quality of human capital and/ or types of human capital and their interactions or synergies are significant and positively affect growth The model Following the traditional Solow residual framework of a production function with neutral technology change and competitive factors (Solow, 1957), the Cobb Douglas production function for each of the two GDEs (r) (where r = 1 2, the two regions, are omitted for simplicity) is specified as below: Y it ¼ A it K a it Lbe it it (1) where Y it, K it, and L it are the production output, capital input, and labor input of city i at time t, respectively; a and b are the capital and labor elasticities, respectively, where 0 < a < 1 and 0 < b < 1; and e it is the stochastic shocks to production of cross section i at time t in region r. Hence, A it measures TFP it, which can be used to pick up the changes of production efficiency of i cross sections over time t. We allow for the possibility of non-constant return to scales by not restricting (a + b) to equal one. Rewriting Eq. (1) in natural logarithms form yields: ln Y it ¼ ln A it þ a ln K it þ b ln L it þ ln e it (2) Eq. (2) is estimated by the methods of pooled OLS and two-way fixed effects panel estimations. The OLS estimation that pools together all the time series and cross-section data provides us with the baseline estimation results. In the two-way fixed effects panel data estimation, we define city-specific effects as cs i, and time-specific effects as ts t, where i and t are the cross-section number and time point, respectively, to capture the fixed effects over different cities and various time periods. These two variables together with the constant term c form the logarithmic index of TFP as presented by Eq. (3) such that ln Y it ¼ðcþcs i þ ts t Þþaln K it þ b ln L it þ g it (3) Using the estimated elasticities of capital (a) and labor (b) and by rewriting Eq. (2), we can construct the series of citylevel logarithmic index of the TFP measurement by i cross-sectional units in t years as: T ˆFP it ¼ ln  it ¼ ln Y it â ln K it ˆb ln L it (4)

6 C. Tuan et al. / Journal of Asian Economics 20 (2009) In order to examine the sources of growth in the two GDEs regions, the following set of simultaneous Eqs. (5) and (6) are constructed for r. Eq. (5) explains the effects of FDI together with other technology-, knowledge-related explanatory variables on TFP, that is: TFP it ¼ f ðfdi it ; R&D it ; HC it Þe it (5) where FDI, R&D, and HC are assumed to be the three major sources affecting technological productivity growth (TFP). In addition, TFP (in Eq. (5)) is endogenous to GDP growth (Y) which is also assumed to be affected by both FDI and HC such that: Y it ¼ f ðtfp it ; FDI it ; HC it Þe 0 it (6) Furthermore, when the quality of HC is reflected by distinct types, such as HC1 and HC2, the joint (interaction) effects measuring the synergies between HC types (HC1 HC2) are used for alternate estimations. Eqs. (5) and (6), therefore, become: TFP it ¼ f ðfdi it ; R&D it ; HC1 it ; HC2 it ; HC1 it HC2 it Þe it Y it ¼ f ðtfp it ; FDI it ; HC1 it ; HC2 it ; HC1 it HC2 it Þe 0 it (5a) (6a) Eqs. (5) and (6) (Version 1) and Eqs. (5a) and (6a) (Version 2) are the two sets of simultaneous equations to be estimated by 2SLS in log-linear form. All of the estimated parameters of the simultaneous equations are expected to carry positive ( + ) signs representing the positive effects of the predictors on TFP and Y it Data and measurement The panel data used for the estimation of the production function covers the city level data to include 23 GDEs cities including 8 cities in PRD and 15 cities in YRD over the period from 1978 to The distinction between PRD and YRD cities for statistical estimation purposes is important because the impacts of FDI and other technology- and knowledgerelated variables may not necessarily show homogeneous effects across these two regions. The data are compiled from various issues of the official statistical yearbooks of relevant cities from 1979 to 2005, Statistical Data and Materials on 55 Years of New China, Statistical Data and Materials on 50 Years of Zhejiang,andScience and Technology Statistical Year Books of China from 1996 to The measurements of the variables used to perform the statistical estimations are summarized below. (1) Output Production The values of output production (Y) is measured by real gross domestic product (GDP) at the city level in the unit of ten thousand RMB (Yuan) and are all converted to 1978 constant price using the GDP deflator. Both real GDP and GDP per capita are used as measurements. (2) Capital stock There is no published data on physical capital stock (K) at the city level and only statistics on annual total investment in fixed assets is available. Therefore, the perpetual inventory method (PIM) is used to estimate city-level capital stock for the years PIM is defined as K it = K it 1 (1 d it )+I it, where K denotes capital stock, I denotes investment, d denotes depreciation rate, and i and t denote city and time, respectively. The amount of physical capital stock of the initial year in 1978 (for the estimation of the production function) is approached as I 0 /a, where a is obtained by comparing the provincial investment in fixed assets in 1978 and the estimated physical capital stock in The values of the estimated physical capital stock in Shanghai, Guangdong, Jiangsu, and Zhejiang in 1978 are obtained from Zhang et al. (2004). A depreciation rate of 9.6%, as suggested by Zhang et al. (2004),is used for the computation. All investment figures are converted to the 1978 constant values using the price indices of investment in fixed assets by region. With the estimated capital stock in 1978, a depreciation rate of 9%, and fixed assets investment obtained from the secondary data published by the China government, a panel of city-level physical capital stock in the PRD/YRD cities for the period from 1978 to 2004 is constructed and measured in units of ten thousand RMB. (3) Labor inputlabor input (L) is measured by the number of workers employed in each of the 23 cities. As reported in the statistical yearbooks, this measure does not adjust for labor hours or quality of employment. (4) Foreign direct investmentrealized FDI (FDI) is measured by the total amount of realized inward FDI received by each city in the unit of 100 million US dollars. (5) Research and developmentprovincial data for R&D expenditures in units of 100 million RMB are available only since 1995 and are used as a proxy because city-level data are not available for most cities in the two regions. City-level R&D expenditures are measured by the total R&D expenses and R&D share in GDP (that is, R&D/GDP) by cities. 7 The 9 cities/counties (Shi) of PRD are Guangzhou City, Shenzhen City, Zhuhai City, Foshan City, Jiangmen City, Dongguan City, Zhongshan City, Huizhou Urban District, and Zhaoqing City. The 16 cities of YRD are Shanghai (municipality) City; Nanjing City, Suzhou City, Wuxi City, Changzhou City, Zhenjiang City, Nantong City, Yangzhou City, and Taizhou City in Zhejiang Province; and Hangzhou City, Ningbo City, Jiaxing City, Huzhou City, Shaoxing City, Zhoushan City, and Taizhou City in Jiangsu Province. 8 See Arayama and Miyoshi (2004) and Zhang et al. (2004).

7 286 C. Tuan et al. / Journal of Asian Economics 20 (2009) (6) Human capitalthree types of human capital (HC) are examined, namely, higher or tertiary education, secondary education, and the sum of secondary and higher education. Correspondingly, three measurements of HC in terms of the proportion of the three types of enrollments in the total population are constructed at the city level for the purpose of statistical estimation. These include: (a) the number of students enrolled in higher education (HC1) to total population (P), that is, HC1/P; (b) the number of students enrolled in secondary education (HC2) to total population, where HC2 includes students in both regular secondary schools and specialized secondary schools, that is, HC2/P; and (c) total student enrollment in both secondary schools and higher education (HC where HC = HC1 + HC2) to total population, that is, HC/P. The three measurements represent three different quality levels of HC such that: (a) HC with a relatively high quality at the tertiary level; (b) HC with basic quality at the secondary level; and (c) HC of basic (secondary) quality level and above. Further, the interactions or synergies between the two types of HC, HC1/P and HC2/P (that is, HC1/P HC2/P), are also assumed to create a join (interaction) effect contributing to TFP and GDP growth. 4. Statistical results 4.1. TFP and TFP growth in YRD/PRD: panel data estimations ( ) Estimations of Cobb Douglas production function: panel estimations with fixed effects The Cobb Douglas production function as specified by Eq. (1) for the two regions, YRD/PRD, during the post-economic opening period of is estimated by OLS and panel data estimation assuming a two-way fixed effect model. The statistical estimation results presented in Table 3 show that the panel estimations yield higher goodness-of-fit (Adj-R 2 > 0.98), as well as statistically significant two-way fixed effects in both YRD/PRD. In terms of the production elasticities, YRD has a lower capital elasticity than PRD (that is, a K = versus a K = 0.41, respectively) but much higher labor elasticity (that is, b L = versus b L = 0.558, respectively). YRD s output production is subject to increasing returns to scale since the sum of elasticities is significantly larger than unity (a K + b L =1.246> 1; p < 0.05), whereas PRD is subject to decreasing returns to scale (that is, a K + b L = < 1; p < 0.05). Moreover, the significant two-way fixed effects estimated by the constant terms for time (ts t )and cross-section (cs i ) effects are used to compute TFP growth (ln A it ) through time and by cities (below) TFP growth through time: estimated results of panel estimations Significant time effects can be found in both YRD and PRD (r) as observed from the panel data estimations of the Cobb Douglas production functions for r (=1 2). TFP over the time period (ts t )inr is computed from the estimated time-effects and the corresponding result is presented in Table 4. TFP growth since the economic opening, indicated by major economic events in four sub-periods, are computed and illustrated in Table 4. The periods include: (a) sub-period 1 ( ): economic opening of Special Economic Zones (SEZs); (b) sub-period 2 ( ): economic opening of PRD; (c) sub-period 3 ( ): economic opening of YRD; and (d) sub-period 4 ( ): Post-Asian Financial Crisis. The resulting TFP growth behavior through time in both GDEs can be observed from the statistical results presented in Table 4, which generally confirms the positive impact of economic opening in enhancing TFP and TFP growth in both GDEs. Hence, Hypothesis 1 is well supported by the statistical results from the panel data estimations. (1) Overall TFP growth ( ) (a) Continuous TFP growth: Both YRD/PRD have achieved high TFP growth overtime. PRD exhibits a slightly higher average TFP of 4.94 and lower annual growth of 1.21%, versus YRD s 4.89% and 1.54%, respectively, for the whole period of the study. Table 3 Estimations of Cobb Douglas production function in GDEs: results from OLS and panel data estimations ( ). Dependent = Y Yangtze River Delta (YRD) Pearl River Delta (PRD) OLS Panel OLS Panel b (0.262) *** (1.300) *** (0.162) *** (0.612) Fixed effects TS(T) Yes a Yes a CS(N) Yes a Yes a K *** (0.020) *** (0.050) *** (0.018) *** (0.049) L *** (0.050) *** (0.170) *** (0.037) *** (0.045) Length (T, N) 26, 14 26, 7 Adj-R F-Stat b *** *** *** *** DF N Notes: All variables in log-linear form; standard errors in parentheses. a Estimated parameters omitted for simplicity; both time and cross-section effects for various time periods and cities/counties were computed from the estimated coefficients and presented in Tables 4 and 5. b F-Stat for panel estimation represents F-test for absence of fixed effects. *** Represents statistical significance of p < 0.01.

8 C. Tuan et al. / Journal of Asian Economics 20 (2009) Table 4 Estimated TFP and TFP growth in GDEs: results from panel data estimation ( ). Time period Yangtze River Delta (YRD) (16 cities) Pearl River Delta (PRD) (9 Cities) Est-TFP Growth (%) a Est-TFP Est-TFP growth (%) a 1978 (Opening of SEZs) (Opening of PRD) (Opening of YRD) b b b (Asian Financial Crisis) Range: min max ( ) Period average Overall: Sub-period 1: Sub-period 2: Sub-period 3: Sub-period 4: Notes: Computed from the estimated time-effects of the Panel data estimations of Cobb Douglas production functions for YRD and PRD (Table 3); four periods of averages were computed: 1. Sub-period 1 ( ) represents the period of economic opening of the Special Economic Zones (SEZs) in Guangdong in 1978 to the opening of PRD in Sub-period 2 ( ) represents the period of economic opening of PRD till the reconfirmation of China s economic reform by Deng and the economic opening of YRD in Sub-period 3 ( ) represents the confirmation of continuous commitment of China to economic opening in receiving FDI (for details of China s economic opening and its liberalization policies by stages, see Ng & Tuan, 2001). 4. Sub-period 4 ( ) represents the period of post-asian Financial Crisis. a Indicates annual growth rate. b Indicates the year YRD surpassed PRD in its TFP. (b) Range of TFP growth: YRD, however, has a wider range of TFP growth of , representing a TFP gain of 1.92% or 48.3% over the study period, while PRD has a range of representing a gain of 1.73% or 42.9%. Higher fluctuations in TFP growth in YRD can also be observed by the range of annual average TFP growth rates of 1.43% to 5.89%, compared with that of PRD s %. (c) Declining TFP growth rate: TFP average annual growth is further observed to slow down in YRD after its peak in 1993 at 4.08% and drop to less than 1% in 2004, and in PRD in 1991 at 3.42% to less than 1% since (2) Sub-period TFP growth (a) Sub-period 1 ( ): At the beginning of the economic opening of SEZs, PRD exhibited slightly higher average TFP but a slower growth rate (=4.41% and 0.95%, respectively), compared with YRD (=4.32% and 1.56%, respectively); (b) Sub-period 2 ( ): After economic opening, PRD continued to achieve higher average TFP and a much higher annual growth rate (=4.77 and 1.71%, respectively) relative to YRD (=4.61 and 0.92%, respectively). (c) Sub-period 3 ( ): Since YRD s economic opening, however, YRD surpassed PRD in 1994 in terms of average TFP and annual growth, that is, 5.49% and 1.78%, respectively, versus PRD s 5.44% and 1.21%. (d) Sub-period 4 ( ): During the post-asian Financial Crisis period, similar TFP growth behavior as in sub-period 3 can be observed. YRD continued to surpass PRD in terms of TFP growth to attain an average TFP of 5.7 and annual growth of 1.35% while PRD recorded both lower TFP and annual growth of 5.6% and 1%, respectively TFP growth by cross-section: estimated results from panel estimations Cross-section (city) effects are significant in both YRD and PRD (r) following the results of the panel data estimations of the Cobb Douglas production functions for r (=1 2). TFP over cross-sections (cs i )inr is computed from the estimated cross-section effects. The results are presented in Table 5. For easy comparison, the cross-section TFP is converted to a TFP index with the highest TFP in each r set as unity. From the estimated TFP by cross-section, the following TFP growth behavior in the cities in r can be depicted following the statistical results presented in Table 5. Suchobservations generally suggest that PRD cities have higher TFP, and the impact of FDI in enhancing TFP growth is significant in both recipient regions. (1) Regional TFP performance: During the study period, PRD attained higher TFP (=5.95), on the average, than YRD (=5.87). When considering provincial TFP, Jiangsu had the highest average TFP (=6.26) and followed by Guangdong (PRD) (=5.95)

9 288 C. Tuan et al. / Journal of Asian Economics 20 (2009) Table 5 Estimated TFP in the globalized delta economies by YRD/PRD Cities: results from panel data estimation ( ). Yangtze River Delta (YRD) (16 Cities) Pearl River Delta (PRD) (9 Cities) Est-TFP TFP-Index a Est-TFP TFP-Index a Shanghai (5) * Jiangsu Province Guangdong Province Suzhou (1) Zhuhai (1) Wuxi (2) Jiangmen (2) Zhenjiang (3) Shenzhen (3) Nanjing (4) Guangzhou (4) Changzhou (5) * Dongguan (5) Yangzhou (12) Zhongshan (6) Nantong (13) Zhaoqing (7) ** Jiangsu Avg Huizhou (7) ** Zhejiang Province Zhoushan (5) * Ningbo (8) Shaoxing (9) Huzhou (10) Jiaxing (10) Taizhou (14) Hangzhou (15) Zhejiang Avg YRD Avg PRD Avg Notes: Computed from the estimated cross-section effects of the Panel data estimations of Cobb Douglas production functions for YRD and PRD (Table 3); ranking in parentheses. a TFP-Index is constructed by converting the highest estimated-tfp (that is, Suzhou = 6.925) to 1. * The estimated parameters of the respective sets of cities in YRD are statistically indifferent. ** The estimated parameters of the respective sets of cities in PRD are statistically indifferent. and Zhejiang (=5.87). By sub-regions, both Jiangsu and Guangdong were observed to receive relatively higher FDI flows than Zhejiang as illustrated in Table 1. (2) TFP performance by Cities (a) Inter-regional TFP (YRD versus PRD): When comparing TFP by cities between the two GDEs regions, four cities in Jiangsu (YRD), namely, Suzhou, Wuxi, Zhenjiang, and Nanjing, had accomplished higher TFP than the top-performed TFP PRD city (namely, Zhuhai); and together with Shanghai and the fifth Jiangsu (YRD) city (namely, Changzhou), a total of six YRD cities were shown to achieve higher TFP than the eight cities in PRD. Alternatively, the eight PRD cities out-performed the nine remaining YRD cities. (b) Intra-regional TFP by Cities: In YRD/PRD, the rankings of TFP by cities are further presented in Table 5. The topperformed TFP cities in YRD and PRD were Suzhou and Zhuhai, respectively, and the least-performed cities were Hangzhou and Huizhou, respectively. The better performed TFP cities were seemingly those receiving heavier flows of FDI as shown by the realized FDI observed at the city level. Impacts of TFP and FDI on Output Growth in YRD/PRD: 2SLS estimations ( ) The two sets of simultaneous equations (Eqs. (5) and (6) and Eqs (5a) and (6a)) constructed for each region, YRD/PRD, are both estimated by 2SLS in log-linear form and the estimation results are presented in Tables 6 and 7, respectively. The statistical results show that in general, the hypotheses (Hypotheses 2 and 3) that FDI affect TFP which in turn contributed to economic growth are strongly supported regardless of the measurements of growth by GDP or GDP per capita and human capital by total HC endowment or HC by type. Such research results may well illustrate the critical roles of TFP and particularly its endogeneity effect in channeling the impacts of FDI to output growth. Specific observations from the estimation results are as follows. (1) Effects on TFP (a) Impacts of FDI: TFP is positively affected by FDI in both GDEs (Tables 6 and 7, Eqs. (1.1) and (2.1); p < 0.05) with YRD carrying a slightly higher estimated partial FDI elasticities (est-b = ; p < 0.01) than that of PRD (estb = ; p < 0.01). (b) Impacts of R&D expenses: TFP is also positively affected by R&D in both GDEs with YRD exhibiting a slightly lower estimated partial R&D elasticities (est-b = , p < 0.05) than that of PRD (est-b = , p < 0.01). (c) Impacts of HC endowment: In YRD, total HC endowments (Table 6, Eq. (1.1); p < 0.05) and HC2 (but not HC1) as well as their synergy effects (HC1 HC2) have significantly affected TFP (Table 6, Eq. (2.1); p < 0.01). In PRD, however, HC endowments, regardless of total or by type, show no significant impact. The insignificance of the HC effect in PRD may suggest that human capital, whether higher education or secondary education, is not an important determinant of

10 C. Tuan et al. / Journal of Asian Economics 20 (2009) Table 6 Impacts of TFP and FDI on GDP growth in YRD: results of 2SLS regression ( ). YRD: Version 1 YRD: Version 2 2SLS Stage I Stage II Stage I Stage II Dependent TFP GDP GDP/P TFP GDP GDP/P X (1.1) (1.2.1) (1.2.2) (2.1) (2.2.1) (2.2.2) b *** (0.647) *** (0.743) (1.092) *** (1.682) (0.523) (0.767) FDI *** (0.035) *** (0.043) *** (0.033) *** (0.035) *** (0.035) *** (0.020) TFP-hat *** (0.142) *** (0.153) *** (0.100) *** (0.066) R&D a ** (0.071) *** (0.071) HC/P ** (0.179) *** (0.118) HC1/P (0.441) *** (0.109) HC2/P *** (0.606) *** (0.193) HC1/P HC2/P ** (0.153) *** (0.038) DF Adj-R F-Stat *** *** *** *** *** *** RMSE Notes: All variables in log-linear form; standard errors in parentheses. a The measure of R&D/GDP is statistically insignificant. ** Represent statistical significance of p < *** Represent statistical significance of p < Table 7 Impacts of TFP and FDI on GDP growth in PRD: results of 2SLS regression ( ). PRD: Version 1 PRD: Version 2 2SLS Stage I Stage II Stage I Stage II Dependent TFP GDP GDP/P TFP GDP GDP/P X (1.1) (1.2.1) (1.2.2) (2.1) (2.2.1) (2.2.2) b *** (0.361) (2.714) *** (1.656) ** (0.403) (1.907) *** (1.651) FDI *** (0.040) *** (0.118) *** (0.090) *** (0.038) *** (0.062) *** (0.090) TFP-hat * (0.363) *** (0.309) *** (0.218) *** (0.308) R&D/GDP *** (0.032) *** (0.034) HC/P (0.116) *** (0.337) HC1/P (0.020) *** (0.326) HC2/P (0.132) *** (0.635) HC1/P HC2/P *** (0.123) DF Adj-R F-Stat *** *** *** *** *** *** RMSE Notes: All variables in log-linear form; standard errors in parentheses. * Represent statistical significance of p < ** Represent statistical significance of p < *** Represent statistical significance of p < TFP. Since the production process in PRD cities were mainly labor-intensive, which means that a large pool of workers would be needed, rather than HC equipped with good or high quality education. Another possible explanation is that the proportion of student enrollments may not serve as a good measure of HC endowment because the large proportion of HC with good education were educated and migrated into PRD from other regions of China or from abroad. 9 (2) Effects on output growth (a) Impact of TFP: The results from Stage II of 2SLS show that the TFP-hat estimated from Stage I exerts a significant effect on growth in terms of the two output measurements (GDP and GDP per capita) in both YRD (Table 6, Eqs. ( ) and ( ) and PRD (Table 7, Eq. ( and ). When comparing the partial effects of TFP on growth between the two regions, YRD has much lower TFP elasticities, showing a range from 0.42 to 0.83 (that is, inelastic; p < 0.01), while that of PRD from (that is, inelastic; p < 0.1) to 1.67 (that is, highly elastic; p < 0.01), depending on HC or HC by type. 9 This is particularly true in some big PRD cities, such as in Shenzhen, Dongguan, and Guangzhou. Hence, local school enrollments may not serve as a good measurement of human capital in these cases.

11 290 C. Tuan et al. / Journal of Asian Economics 20 (2009) (b) Impact of FDI: The above results may well suggest the significant direct impact of FDI on TFP (at Stage I, 2SLS) and also output growth through TFP (Stage II, 2SLS). Such an indirect impact through the estimated-tfp in PRD is much stronger than that in YRD. Furthermore, FDI is also found to affect directly growth as illustrated by the significant statistical results of Stage II of 2SLS (Tables 6 and 7, Eqs. ( ) and Eqs. ( )). Hence, Hypothesis 4 is well supported by the data. Such direct partial impact of FDI on GDP growth is stronger in YRD when compared with PRD, and the reverse is true for GDP per capita. (c) Impact of HC endowment: HC endowment and HC by type are both found to affect growth and, therefore, Hypothesis 5 is well supported by the research findings. Both HC by type and HC synergies significantly affect directly per capita GDP growth in YRD and GDP growth in PRD. From Tables 6 and 7, HC endowments (in terms of total and by type) affect both TFP and growth (per capita GDP) in YRD, but only on GDP growth in PRD. Furthermore, HC by type and their synergies affect the two regions quite differently. In YRD, education and training of HC and especially by type and their synergies are important to both TFP and growth in terms of per capita GDP. In particular, the education and training of students at the secondary level shows a significant direct (highly elastic) impact that warrants special attention. In PRD, however, HC by type and their synergies have not affected TFP, but the direct effects of tertiary and secondary education on GDP growth are both critical (highly elastic). (d) Impact of R&D expenditures: Interestingly, R&D expenses have no direct effect on output growth but only through TFP enhancement (Tables 6 and 7) Associations of TFP with FDI and output growth by YRD/PRD Cities: simple correlation analyses Due to the limitations in terms of data availability and consistency, the estimation of the hypothesized relationships by city is impossible. With the panel of city-level TFP constructed from the above model, the positive associations of the estimated-tfp with FDI, GDP, R&D, HC and HC by type, and HC synergies by cities being described by the above hypothesized relationships can be examined instead by simple correlation analyses for further reference. The corresponding statistical results are presented in Table 8. The statistical findings show that with the exception of a few cases, majority of the city concurred with the hypothesized relations that TFP has close associations with FDI, GDP, R&D, HC and HC by type, and HC synergies. From the examinations of Table 8, the few cases are: (1) TFP with FDI: Absence of significant statistical association between TFP and FDI in two YRD cities, namely, Zhoushan and Zhenjiang. Table 8 Associations of TFP with FDI, GDP, R&D, and human capital in PRD and YRD: correlation analyses by cities. Yangtze River Delta (YRD) Cities Variables Shanghai Hangzhou Ningbo Jiaxing Huzhou Shaoxing Zhoushan Taizhou FDI *** *** *** *** *** ** *** GDP *** *** *** *** *** *** *** *** R&D *** *** *** *** *** *** *** *** R&D/GDP *** *** *** *** *** *** *** *** HC *** *** *** *** *** *** *** *** HC *** *** *** *** *** *** *** *** HC1 HC *** *** *** *** *** *** *** *** Variables Nanjing Wuxi Suzhou Nantung Yangzhou Zhenjiang Changzhou FDI *** *** *** ** *** *** GDP *** *** *** *** *** *** *** R&D *** *** *** *** *** *** *** R&D/GDP *** *** *** *** *** *** *** HC *** *** *** *** *** *** *** HC *** *** *** a *** *** HC1 HC ** *** *** a *** *** Pearl River Delta (PRD) Cities Variables Guangzhou Shenzhen Zhuhai Huizhou Dongguan Zhongshan Jiangmen Zhaoqing FDI *** *** ** *** *** *** *** ** GDP *** *** *** *** *** *** *** *** R&D *** *** *** *** *** *** *** *** R&D/GDP *** *** *** *** *** *** *** *** HC *** *** a *** *** *** *** *** HC *** a a *** *** *** *** *** HC1 HC *** a a *** *** *** *** *** Notes: Data periods for FDI, GDP, HC1, and HC2: ; and for R&D: HC1, HC2 denote the number of students enrolled in higher education and secondary education, respectively. a Inadequate or lack of data. ** Indicate statistical significance at p < *** Indicate statistical significance at p < 0.01.