Developing Inland China: How Coastal FDI and Export Activity Matters

Transcription

1 Developing Inland China: How Coastal FDI and Export Activity Matters Puman Ouyang Department of Economics Syracuse University 110 Eggers Hall Syracuse University Syracuse, New York, USA, November 2009 Abstract This paper investigates whether coastal FDI and export activity have regional spillovers into inland regions in China and how these spillovers depend on different types of FDI and export activity. Applying 2SLS fixed effect models to two provincial level datasets, one for and the other for , I find that on average, coastal FDI has positive spillovers into inland regions while coastal export activity has negative spillovers into inland regions. Estimated inter-regional spillovers vary by location of activity. Eastern coastal FDI generates the largest positive spillovers into inland regions, possibly due to the type of firms concentrated there. Joint venture FDI has a negative impact on inland GDP while contract cooperative FDI has a positive impact on inland GDP. FDI from Hongkong, Macao and Taiwan has a large negative impact on inland GDP. Export activity by domestic firms in coastal provinces appears to have a positive impact on inland economies while foreign firms export activity does not. JEL classification: F19, F21, R11, R12, O18, O40 Keywords: Foreign direct investment, Exports, Spillovers, China address: pouyang@maxwell.syr.edu Tel.: Fax: Address: 110 Eggers Hall, Syracuse University, Syracuse, New York, USA, I thank my advisors, Mary E. Lovely (Chair), Jeffery D. Kubik, and Devashish Mitra, for their advice and encouragement.

2 1. Introduction Since 1978, China has experienced rapid economic growth rates and has gradually become one of the worlds leading emerging economies for export activity and attracting foreign direct investment (FDI). For example, during the period , China experienced FDI inflows of 0.53 billion dollars, and totaled exports of billion dollars. These figures increased to and 1, billion dollars respectively in FDI and exports are viewed as big driving forces for China s strong growth. Meanwhile, 90% of FDI and export activity in China tends to cluster in coastal provinces because they have geographical advantages in exports, receive preferential treatment from the central government, and better infrastructure. These provinces, however, account for only 40% of the population and 30% of the land area of China (Fu (2004); Ouyang (2009)). Thus, there are clearly questions of economic distribution that need to be explored. This paper seeks to understand the role played by spatially concentrated FDI and exports in inland economies. Previous literature in this area has focused on linking regional economic growth to local FDI inflows and exports. There is strong empirical evidence of widening regional level income inequality resulting from FDI and exports concentration. Not surprisingly, studies have found that coastal regions grow faster than inland regions in terms of GDP per capita (Ng and Tuan (2006); Fu (2004); Fujita and Hu (2001); Zhang and Zhang (2003)). The logical extension to this literature is to explore whether FDI and exports in coastal provinces have positive spillovers on the GDP growth of inland regions in China. Coastal FDI or export activity may influence inland economies in several ways. They may foster internal migration from inland regions to coastal regions, which creates remittances to inland regions and 1 Available at and 2

3 cause technology spillovers via labor turnover, and may improve inland domestic firms productivity through the inter-industrial linkages. Ouyang (2009) uses city level data in China from 1996 through 2004 and finds that FDI concentrated in coastal regions has a positive impact on inland cities GDP per capita growth rates. In other words, coastal FDI generates positive spillovers into inland China. In addition, the paper finds that inter-industrial linkages are important in inter-regional spillovers from FDI and hence more industrialized inland cities absorb more spillovers from coastal FDI than less industrialized ones (Ouyang (2009)). Fu (2004) finds no evidence of positive inter-regional spillovers from exports in China s provincial level data. She concludes her argument by stating that this is because export activity in China is heavily involved in processing exports, which do not generate inter-industrial linkages with domestic firms in inland regions. There are many ways in which this literature can be extended. Fu (2004) and Ouyang (2009) only focus on spillovers from either coastal FDI or exports. An obvious extension is to test spillovers from both. FDI and exports are intertwined since nearly half of FDI in China is export-oriented; however, not all foreign firms are export-oriented. In addition, domestic firms account for about 17% of exports. Ignoring either FDI or exports in estimation has the potential to lead to biased results. Further, Fu (2004) and Ouyang (2009) do not provide an assessment of how inter-regional spillovers vary across different types of FDI and export activity. The answer to this question has important policy implications for regional development. In this paper, I employ a two-stage least square (2SLS) fixed effect model to estimate whether FDI and exports in coastal regions affect inland regions via spillovers. I first use provincial statistics from 1993 to 2008 to examine the average spillovers. This data unfortunately does not include information on characteristics of FDI and exports, so I construct another dataset 3

4 that covers characteristics of FDI and exports from 2002 to I find that coastal FDI has positive spillovers among inland regions in general, while coastal export activity has negative spillovers into inland regions. Spillovers vary substantially, depending on the characteristics of FDI or export activity. FDI concentrated on the eastern coast generates the largest positive spillovers. FDI from Hong Kong, Macao and Taiwan (hereafter HMT) has a negative impact on inland GDP mainly due to the processing export oriented nature of these firms. Contract cooperative FDI has a positive impact on inland economies. Similarly, spillovers from coastal export activity are not uniform. While coastal exports on average have a negative impact on inland economies, northern coastal exports generate large positive spillovers into inland regions. Exports from domestic firms in coastal regions appear to have a positive impact on inland economies, while exports from foreign firms do not. The next section of this paper discusses previously published literature relevant to the topic. Section 3 discusses data. Section 4 estimates the average inter-regional spillovers from coastal FDI and exports. Section 5 examines inter-regional spillovers from FDI and exports in different coastal regions. Section 6 breaks down inter-regional spillovers based on characteristics of coastal FDI and exports. The paper is concluded in Section Literature Review There is little consensus in the literature on inter-regional spillovers from FDI. In a study using firm level UK data, Girma and Wakelin (2000) find that domestic firms are not positively influenced by foreign firms in other regions in the UK. In a subsequent study on Chinese firms, Girma and Gong (2008), using firm level data, do not find that foreign firms affect state-owned enterprises in other regions. Madariaga and Poncet (2007) use spatial econometric techniques to find that a city s income level is positively affected by FDI in surrounding cities in 4

5 Chinese data in Another use of city level data in China during indicates that FDI concentrated in coastal regions has a positive impact on inland cities GDP per capita growth rate (Ouyang (2009)). Furthermore, her results indicate that the level of FDI regional spillovers in an inland city depends on the development of the city s mining and manufacturing industries. In addition, because city level datasets do not contain information on FDI characteristics, her paper is unable to investigate how FDI regional spillovers vary due to different characteristics or structures of FDI. Unlike the positive spillovers accrued due to FDI, exports generate no such inter-regional benefits. Fu (2004) finds that coastal exports benefit coastal regions but not inland regions because a large part of export activity in coastal provinces is processing trade, which does not provide inter-industry linkages to inland regions. However, although she argues that the concentration of processing trade is the main reason for no spillovers from coastal exports, she does not use econometric techniques to test the hypothesis. In addition, her paper has no information on characteristics of export activity and hence does not analyze how spillovers vary across different types of coastal exports. In addition to the studies mentioned above, there are studies on inter-regional spillovers from other factors. For example, Brun et al. (2002) and Zhang and Felmingham (2002) find that economic growth in coastal regions has a positive impact on economic growth in inland regions in China. They mention coastal FDI as a potential reason but do not use econometric tools to test causality. Bronzini and Piselli (2009), Funke and Niebuhr (2005), Keller (2002), and Kuo and Yang (2008) find evidence of inter-regional spillovers from R&D activities in a variety of datasets. 5

6 3. Data This paper uses two provincial level datasets authorized by the National Bureau of Statistics of China. 2 Average inter-regional spillovers from FDI and exports are estimated over a long period with the use of the dataset. A short term analysis of regional spillovers with FDI and exports heterogeneity is constructed with the use of the dataset. The primary dataset covers official statistics of 31 administrative districts that include provinces, autonomous districts and municipalities during For simplicity, I use province to refer to all provincial level administrative districts. 3 This dataset contains variables listed in China Provincial Yearbooks, such as GDP, population, employment, domestic investment, total FDI flow, and total export value and so on. There are 491 observations, of which 191 are for coastal provinces and the rest for inland provinces. I employ this dataset rather than the city level dataset in Ouyang (2009) because the latter has no statistics on export value. This provincial dataset is used to estimate the overall level of inter-regional spillovers from coastal FDI and exports in this time period. The dataset has a shortcoming in that there is no information about different types of FDI or export activity occurring in a province. In addition to examining the overall level of regional spillovers from coastal FDI and exports, this paper analyzes the impacts of different types of coastal FDI and exports on inland economies. In fact, many provinces yearbooks do not just contain variables such as total amounts of FDI and exports. Detailed information on characteristics of FDI and exports can be 2 Available at 3 Coastal provinces: Beijing, Fujian, Guangdong, Guangxi, Hainan, Hebei, Jinagsu, Liaoning, Shandong, Shanghai, Tianjing, and Zhejiang. Inland provinces: Anhui, Chongqing,Gansu, Guizhou, Heilongjiang, Henan, Hubei, Hunan, Inner Mongolia, Jiangxi, Jilin, Ningxia, Qinghai, Shaanxi, Shanxi, Sichuan, Tibet, Xinjiang, and Yunan. 6

7 drawn from China Provincial Yearbooks, though over a shorter period of time. 4 There are seven years, from 2002 to 2007, for which all coastal provinces yearbooks are available. Therefore, I construct the second dataset by including variables such as FDI of different investment types and from different origins, exports by different types of firms and so on. Table 1 compares main variables in inland and coastal provinces, based on the provincial statistics data during Inland provinces, on average, have smaller economies than coastal provinces. During the sample years, the average GDP in coastal provinces was around twice that in inland provinces. China s FDI and export activity are highly concentrated in coastal regions. The average FDI in coastal provinces was at least 5 times higher than that in inland provinces. The average value of exports concentrated in coastal provinces was at least 8 times higher than that in inland provinces. 4. Average Inter-Regional Spillovers from Coastal FDI and Exports This section estimates the overall extent to which coastal FDI and exports affect inland economy growth by using the provincial data in I divide the entire sample into two groups: 19 inland provinces and 12 coastal provinces. Based on geographic proximity, I construct effective coastal FDI and coastal exports variables to which an inland province may be exposed and incorporate these variables into an inland province s production function Model I define an augmented Cobb-Douglas production function of an inland province as the following: =, 1 4 Although my first dataset for the provincial statistics in contains major variables from China Provincial Yearbooks, it does not include all information in those yearbooks. China Provincial Yearbooks can be found at 7

8 where Y it, K it, L it and F it represent output, domestic capital stock, labor input and foreign capital stock respectively. 5 Subscripts i and t refer to inland province i and year t. As in Solow (1957), A is the technology level, which includes any kind of shift in the production function, such as macroeconomic situations and improvements in the education of the labor force. Coastal FDI and exports do not directly produce outputs but they may affect a province s technology level and economic environment in many ways. Here, I posit that A is a function of regional spillovers of coastal FDI and exports such that, = f(, ). (2) I substitute (2) into (1) and take the log-linear form of (1). Since the functional form of A it is unknown, I replace ln A it by and for simplicity. This gives us, = θ 3 (3) yields the following specification, = ε 4 The key parameters are β 4 and β 5. If β 4 is positive, coastal FDI has a positive impact on an inland province s economy and hence generates positive spillovers into inland regions. If β 4 is negative, coastal FDI has negative spillovers in inland regions. If β 4 is zero, there are no spillovers from coastal FDI. The estimated value of β 5 can be interpreted in a similar manner. 5 Ideally, K or F should be the volume of domestic or foreign capital services, assumed to be a proportion to the productive domestic or foreign capital stock. See OECD manual Measuring Productivity: Measurement of Aggregate and Industry-Level Productivity Growth (Available at However, calculating the volume of capital service requires data on the asset combination and the user cost price per unit of capital services, which is not available with most regional level statistics. Therefore, in practice, people usually use capital stock in the estimation of a production function. 8

9 4.2. Specification and Econometric Issues Based on (4), we can rewrite the specification such that, ln = , 5 where subscripts i and t refer to an inland province and year respectively. J refers to a coastal city j and the subscript c emphasizes that city j belongs to the coastal province group c. Dependent and independent variables are logged so the parameters refer to the elasticities with respect to independent variables. The dependent variable is real GDP of an inland province. To alleviate the influence of inflation, I convert nominal GDP to real terms by the province level consumer price index (CPI) from China Statistical Yearbooks, with the base year being I construct effective coastal FDI and exports with cross time and cross sectional variation based on Keller (2002). is effective coastal FDI flow which may affect the inland city i. is FDI flow in a coastal province j. 6 D ij is the distance between capital cities of provinces i and j, with the unit being 1000 kilometers. 7 δ is a positive parameter and implies that coastal FDI becomes less effective at influencing province GDP as distance between the two provinces grows. is effective FDI flow in the coastal province j. stands for total effective FDI flow in all coastal provinces which may affect 6 The FDI statistics include foreign investors asset acquisition in China which does not establish new equipments. However, this should not be a problem in my study of regional spillovers from FDI because asset acquisition is highly concentrated in capital intensive industries in China but China s FDI inflows is distributed across industries (Dean et al.(2009); Huang (2003)). In addition, if the spillovers are from foreign technology or management practices, the broader measure of foreign involvement may be preferred. 7 The reason to use 1000 kilometers instead of 1 kilometer as the unit of distance is that the former will make close to zero. 9

10 the inland city i at year t. Since there are 19 inland provinces and 12 coastal provinces, the distance combinations ensure a large variation in effective coastal FDI. As noted in Ouyang (2009), I assume δ equal to 1. 8 Note that, FDI in US dollars is converted to Chinese currency Yuan by the annual average exchange rates published by the Federal Reserve Bank. Similarly,. refers to effective coastal exports which may affect an inland province. To isolate the spillover effects of FDI and exports, I control for the following variables: Domestic Capital it refers to K it in (4), domestic capital stock in an inland province at year t, while Foreign Capital it refers to F it in (4), an inland province s own foreign capital stock at year t. I calculate domestic and foreign capital stock by the perpetual inventory method. 9 Employment it is the total employment in province i at year t, a proxy for labor input L it in (4). Year Dummies control for macro shocks that affect an inland province s GDP and the independent variables simultaneously. Province Fixed Effects control for province-specific unobservable factors, particularly policy variation. Chinese provinces make and implement policies independent of the central government to a large extent. Without controlling for these policies, my estimation has the potential to yield biased results. Since it is difficult to parameterize cross-provincial policy variation, I control for policy variation by province fixed effects with the assumption that policies are time-invariant. An important point to note is that a region s own FDI may be inversely affected by its own GDP, which may cause bias in estimates ((Li and Liu (2005); Alfaro et al. (2004); Lee et al. (2003); Cheng and Kwan (2000); Borensztein et al. (1998); Wheeler and Mody (1998)). 8 I also try different values for δ and the results do not change significantly. 9 I first regress on to estimate the investment growth rate to be 17%, noted as. Then the initial capital stock =, where is the capital depreciation rate, assumed as 7.5% in China (Kuo and Yang (2008)). Using, I calculate by the formula = 1 +.When calculate foreign capital stock, is replaced by the growth rate of and is replaced by. 10

11 Therefore, I use to be an instrumental variable for. There may be another concern about potential endogeneity because an inland province s economy may affect coastal FDI and exports, since the economies of all provinces are interrelated. However, given that the economies of inland provinces are much smaller than those of coastal provinces (see Table 1), an individual inland province is unlikely to influence total FDI and exports in coastal regions. Further, in later robustness checks, I will find instrumental variables for effective coastal FDI and exports variables. Table 2 reports summary statistics of the inland province group. There is large variation across inland provinces. The real GDP ranges from 2.65 to billion Yuan. Domestic capital stock is on average billion Yuan, from 6.01 to billion Yuan. Foreign capital stock is on average billion Yuan, from 0 to billion Yuan. Coastal FDI varies from 0.21 to 136 billion Yuan Results and Robustness Checks Since coastal FDI and exports are intertwined, it is hard to distinguish between the effect of coastal FDI and the highly correlated influence of coastal exports. Therefore, I run regressions which exclude either coastal FDI or coastal exports, and also run regressions with both coastal FDI and coastal exports. By this method, although I cannot completely parcel out the effect from FDI and the effect from exports, the estimates will give some hints of the direction of these effects. Although my analysis is primarily based on 2SLS fixed effect estimation, I first run benchmark OLS regressions and then show how different the results from 2SLS fixed effect estimation will be. The OLS results are shown in Table 3. Columns (1) - (3) include effective coastal FDI and other independent variables but exclude effective coastal exports. The estimated 11

12 elasticities of GDP with respect to effective coastal FDI are positive and significant. The estimates of coastal FDI are 0.21 and in columns (1) and (2) respectively. They are significant at the 1% level. When including all independent variables in column (3), the estimated elasticity declines to and is still significant at the 10% level. Columns (4) - (6) include effective coastal exports and other independent variables, excluding effective coastal FDI. The estimated elasticities of GDP with respect to effective coastal exports are positive, from to 0.19, and are significant at the 1% level as shown in columns (4) and (5) and at the 5% level as shown in column (6). Columns (7) - (9) present results from the regression with both coastal FDI and exports variables. The coefficients of coastal FDI are still positive and significant but coastal exports coefficients become negative and significant. OLS estimation shows that coastal FDI has a positive impact on inland economies while coastal export activity has a mixed influence. Labor input, domestic capital and foreign capital are found positive and significant. Domestic capital stock has the largest impact on an inland province s GDP. The estimated elasticities of GDP with respect to domestic capital stock vary from 0.92 to 1.04 and are significant at the 1% level. The 2SLS fixed effect estimation results are reported in Table 4. 2SLS fixed effect estimation alleviates potential endogeneity in the model by the use of province fixed effects and an instrumental variable for foreign capital stock, yielding results different from OLS. Coastal FDI elasticity is still positive, from to Except for column (1), the coefficients of coastal FDI are significant. For example, in column (9), the estimated elasticity of GDP with respect to coastal FDI is 0.19 and significant at the 1% level. This means when effective coastal FDI increases by 10%, GDP of an inland province increases by 1.9%. The results are consistent with Ouyang s (2009) argument that coastal FDI has a positive impact on inland regions 12

13 economic growth. The sign of coastal exports coefficients is negative, shown in columns (4)-(9). The estimated elasticities of GDP with respect to coastal exports vary from to The results imply that coastal exports have a negative impact on an inland province s GDP. To summarize, both OLS and 2SLS fixed effect estimations yield positive coefficients for coastal FDI, while coastal exports appear to have a negative impact on inland economies. These results confirm conclusions reached in Ouyang (2009) and Fu (2004). Coastal FDI, although part of it is involved in processing exports which may not create enough inter-industrial linkages to affect inland economies positively, there are some foreign firms which do business with inland domestic firms and have positive spillovers into inland provinces. By contrast, export activity in coastal regions is heavily concentrated in processing exports and hardly generates any positive spillovers into inland regions. It is very likely that coastal export activity even has negative spillovers into inland regions. Estimates for other variables including employment are also reported in Table 4. Employment coefficients are found to be positive and significant. The coefficients of domestic capital stock are around 0.50 and significant at the 1% level, meaning that domestic capital stock has a large impact on an inland province s GDP. By contrast, foreign capital stock is not significant. The reason why foreign capital stock is not significant in 2SLS fixed effect estimation is that the use of province fixed effects and instrumental variables helps minimize the possibility of reverse causality between GDP and foreign capital stock. The results illustrate that in inland provinces, domestic investment plays a big role while the amount of foreign capital is too small to boost the economy. The conclusions on the impacts of domestic and foreign capital are consistent with Ouyang (2009). Various methods are used to test result sensitivity. I first use different definitions for 13

14 effective coastal FDI. One alternative definition is, a simple way to discount effective coastal FDI. Another definition is effective coastal FDI stock, defined as 14 where is the foreign capital stock in coastal province j at year t, calculated by the perpetual inventory method. Using effective coastal FDI stock can help estimate how previous FDI affects inland economy. All other independent variables are included and defined as before. The results are presented in columns (1) and (2) in Table 5 and do not change significantly from Table 4. Coastal FDI has a positive impact on inland GDP. The estimated elasticity with respect to coastal FDI is 0.33 and significant at the 1% level in column (1). In column (2), the coefficient is 0.67 and significant at the 1% level, larger than coefficients in column (9) in Table 4. Coastal exports have a negative impact on inland GDP. The estimated elasticities with respect to coastal exports are in column (1) and in column (2) and significant at the 1% level. The results indicate that coastal FDI stock has a larger impact than coastal FDI flow and imply that inland economy is also affected by previous FDI. This is intuitively plausible because if a foreign firm makes an investment in coastal regions in a particular year, the investment may carry over and affect inland cities in years following the original expenditure. There is a potential question about endogeneity since foreign firms investing in coastal regions, may decide whether or not to focus on products for the domestic market based on the economic situation in inland provinces. I argue that the total amount of coastal FDI and exports are unlikely to be affected by an individual inland province s GDP. However, I still find instrumental variables for effective coastal FDI and exports to see if the results change. Ideally, an instrumental variable (IV) should be completely exogenous and highly correlated with coastal FDI and exports variables and uncorrelated with the error term. However, such an IV is hard to find in practice. Using what has become standard practice in the trade literature, I use the logs of

15 1-year lagged effective coastal FDI and exports as IVs. The estimation result is shown in column (3) in Table 5. Coastal FDI has a large positive impact on an inland province s GDP. The estimated elasticity of GDP with respect to coastal FDI is 0.82 and significant at the 10% level. By contrast, coastal exports, once again, have a large negative impact on inland province s GDP. The estimated elasticity of GDP with respect to coastal exports is and significant at the 5% level. To dispel the influence of extreme observations on my results, I run regression (5) by excluding extreme observations in the inland province group. Column (4) excludes the observations with real GDP below the 1 st percentile or above the 99 th percentile. Column (5) excludes those observations with real GDP per capita below 1 st percentile or above the 99 th percentile. The results do not change significantly. All of the robustness checks indicate that coastal FDI generates positive spillovers while coastal exports has negative spillovers into inland provinces. 5. Inter-Regional Spillovers from FDI and Exports in Different Coastal Regions This section analyzes inter-regional spillovers from FDI and exports in different coastal regions. Officially, there are four coastal regions in China: northern coastal provinces, which include Beijing, Tianjin, and Hebei; eastern coastal provinces, which include Shanghai, Jiangsu, Zhejiang, Fujian, and Shandong; a northeastern coastal province called Liaoning; and the southern coastal provinces Guangdong, Guangxi, and Hainan. 10 To separate spillovers from coastal FDI and exports concentrated in different coastal 10 Available at This website combines southern China and central China into Central China. In this paper, I separate three coastal provinces located in southern China from Central China. 15

16 regions, I decompose effective coastal FDI into four effective coastal FDI variables, one for each coastal region. I also construct four effective coastal exports variables in a similar way. I incorporate these effective FDI and exports variables into the production function and run regressions. The results are represented in Table 6. Column (1) includes results for four effective coastal FDI variables, total effective coastal exports, employment, domestic capital stock and foreign capital stock. Column (2) includes results for four coastal exports variables, total effective coastal FDI, employment, domestic capital stock and foreign capital stock. Column (3) includes results for four coastal FDI variables and four coastal exports variables and all other independent variables. Eastern coastal FDI coefficients are approximately 0.16 and significant at the 1% level. This indicates that if eastern coastal FDI increases by 10%, inland GDP increases by 1.6%. The other three coastal regions separately do not have a significant impact on an inland GDP. I run an F-test to see if the other three coastal regions have a jointly significant impact on inland GDP. The p-values from the F-test are 0.73 and 0.17 in columns (1) and (3), which indicates that they jointly do not have spillovers into inland economy. All positive spillovers from coastal FDI stem from eastern FDI. This indicates that eastern coastal FDI has more inter-regional linkages with inland regions possibly due to its relatively lower concentration of processing trade; a result consistent with Ouyang (2009). Although coastal exports on average have a negative impact on inland economy, northern coastal exports positively affect inland economy. The estimated elasticities of GDP with respect to exports from northern coastal provinces are about 0.82 and significant at the 1% level, shown in columns (2) and (3). By contrast, unlike eastern coastal FDI, eastern coastal exports generate 16

17 negative spillovers into inland economy. The estimated elasticities of GDP with respect to eastern coastal exports are about and significant at the 1% level in columns (2) and (3). 6. Decomposition of Regional Spillovers Based on Characteristics of Coastal FDI and Exports In addition to examining the average inter-regional spillovers from coastal FDI and exports in China, this paper also analyzes these spillovers based on characteristics of coastal FDI and exports. The data used in this section comes from the provincial level dataset mentioned previously Coastal FDI from Different Types of Foreign Firms There are mainly three types of FDI in China: joint ventures, which refer to enterprises co-founded by foreign and Chinese investors and where both parties divide revenues and expenses according to their shares, contract cooperative enterprises, which are established by foreign and Chinese investors and where both parties divide obligations and rights according to the contract for their mutual benefit, and foreign solely funded enterprises which are founded and run solely by foreign investors. Based on the data, I decompose effective coastal FDI variable into three effective coastal FDI variables, each for a type of FDI. I incorporate these effective coastal FDI variables into the production function for the inland province group to estimate the production function. The results are presented in Table 7. Joint venture FDI coefficients have a negative sign and are significant at the 5% level as shown in columns (1) and (2). Contract cooperative FDI coefficients have a positive sign and are significant at the 5% level as shown in columns (1) - (3). Coefficients for FDI by foreign solely funded enterprises are positive. These coefficients are 17

18 positive and significant in columns (1) and (2). To conclude that joint venture FDI has negative regional spillovers into inland regions while contract cooperative FDI positively affects inland economy since there is a potential for inter-firm economies as foreign firms might communicate with and affect each other, I need to rule out several other possibilities. I run an F-test to see if the elasticity of joint venture FDI is equal to that of contract cooperative FDI. The p-values from the F-test are from to This rejects the hypothesis that regional spillovers from joint venture FDI and contract cooperative FDI are the same. The results indicate that joint venture FDI has negative regional spillovers into inland regions while contract cooperative FDI generates positive spillovers. The reason behind this effect needs further exploration, currently beyond the scope of this paper due to a lack of relevant firm level data Coastal FDI from Different Origins For political and historical reasons, China categorizes investment from Hong Kong, Macao and Taiwan (HMT for abbreviation) as FDI as well. Usually, FDI from HMT is more involved in processing exports, which may generate relatively lower inter-regional spillovers due to limited transactions with inland regions. However, those firms use Chinese as their primary language of communication, which may make it easier for them to interact with inland regions and hence generate positive regional spillovers into inland regions. Therefore, FDI from HMT and FDI from regions, mostly OECD countries, may have different impacts on inland economy. I construct effective HMT FDI variable and effective other origins FDI variable in a manner similar to the construction of effective coastal FDI in (5). The results are reported in Table 8. Coastal FDI from other origins does not have a significant impact on inland GDP, which seems strange because FDI from OECD countries usually brings advanced technology and 18

19 should have positive spillovers into inland regions. But note that the dataset only contains six years. There is insufficient evidence to draw the conclusion that FDI from OECD countries does not have a positive influence on inland economy. However, an interesting result shown in Table 8 is that FDI from HMT has a large negative impact on inland GDP: the estimated elasticities of GDP with respect to coastal FDI from HMT are from to -0.39, significant at the 1% level in columns (1) and (2) and at the 5% level in column (3). These may be because HMT foreign firms are heavily concentrated in processing exports, which have negative spillovers into inland regions, as shown in Section 4. I also try to estimate if coastal FDI from Asian origins and from other origins have different spillovers into inland economy. The unreported results show neither has separate significant regional spillovers. In addition, coastal FDI from European countries and from other origins do not have separate significant regional spillovers into inland economy Spillovers from Exports from Different Types of Firms The Chinese economy is heavily driven by growth in its export sector. While many studies show that exports boost local growth in coastal regions, the estimates in this paper show that benefits from coastal exports do not spillover into inland regions. This subsection investigates the variation in inter-regional spillovers from exports by different types of firms. The dataset has information on exports from domestic firms and foreign firms. As with FDI, I construct effective exports from coastal domestic firms and effective exports from coastal foreign firms and estimate the inland provinces production function, including these two variables. The results are reported in Table 9. Domestic firms exports have a positive impact on inland provinces GDP, with coefficients ranging from to The coefficients are 19

20 significant at columns (2) and (3). Foreign firms exports, on the other hand, show a negative impact on inland GDP, but are significant only in column (1). Therefore, exports from domestic firms in coastal provinces appear to have a positive impact on inland economy while foreign firms exports do not. 7. Conclusion Since 90% of FDI and exports in China are concentrated in coastal regions, which account for only 40% of the population and 30% of the land area, there are distributional questions of net benefits to be explored. This paper investigates whether coastal FDI and exports have regional spillovers into inland regions, and how the spillovers differ by type of FDI and exports. This paper applies a 2SLS fixed effect model to two provincial level datasets, one for and the other for , and finds that in general, coastal FDI has positive spillovers into inland economies while coastal exports have negative spillovers. Furthermore, this paper finds that different coastal regions generate different FDI or exports spillovers, with eastern coastal FDI generating the largest spillovers into inland regions. Northern coastal exports have positive spillovers into inland regions while FDI from HMT has a large negative impact on inland GDP. Exports from domestic firms in coastal provinces appear to have a positive impact on inland economy, while foreign firms exports seem to have no impact. 20

21 Table 1: Average GDP, FDI and Exports: Coastal Provinces vs. Inland Provinces Unit: 1 billion Yuan Average GDP Average FDI Average Exports Year Coastal Inland Coastal Inland Coastal Inland Data source: China Provincial Yearbooks, The National Bureau of Statistics of China. Available at 21

22 Table 2: Summary Statistics Obs Mean Std. Dev. Min Max Real GDP Employment Domestic capital stock Foreign capital stock Effective coastal FDI Effective coastal exports Notes: The unit of real GDP, domestic capital stock, foreign capital stock, effective coastal FDI and effective coastal exports is 1 billion Yuan. The unit of employment is people. 22

23 Table 3: OLS Estimation for Regional Spillovers of Coastal FDI and Exports Dependent variable: the log of real GDP (1) (2) (3) (4) (5) (6) (7) (8) (9) Coastal FDI 0.21*** 0.030*** 0.016* *** 0.20*** 0.12** (0.04) (0.0063) (0.0062) (0.5) (0.064) (0.06) Coastal exports *** 0.027*** 0.014** -1.16** -0.16*** * (0.04 ) (0.0060) (0.0057) (0.46) (0.060) (0.056) Employment *** 0.14*** *** 0.14*** *** 0.14*** (0.021) (0.019) (0.021 ) (0.020) (0.021 ) (0.020) Domestic capital stock *** 0.93*** *** 0.92*** *** 0.92*** (0.024) (0.027) (0.024) (0.027) (0.024) (0.027) Foreign capital stock *** *** *** (0.012) (0.012) (0.012) Province fixed effects No No No No No No No No No Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes R Obs Notes: 1) Independent variables are logged. 2) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance.

24 Table 4: 2SLS Fixed Effect Estimation for Regional Spillovers of Coastal FDI and Exports Dependent variable: the log of real GDP (1) (2) (3) (4) (5) (6) (7) (8) (9) Coastal FDI *** 0.074** * 0.21*** 0.19*** (0.038) (0.033) (0.036) (0.054) (0.049) (0.053) Coastal exports *** -0.45* *** -0.49*** (0.12 ) (0.11) (0.12) (0.17) (0.15) (0.17) Employment * 0.17* * 0.17* ** 0.22** (0.087) (0.096) (0.088) (0.097) (0.086) (0.096) Domestic capital stock *** 0.49*** *** 0.49*** *** 0.53*** (0.052) (0.056) (0.053 ) (0.057) (0.052) (0.057) Foreign capital stock (0.018) (0.017) (0.018) Province fixed effects Yes Yes Yes Yes Yes Yes Yes Yes Yes Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Obs Notes: 1) Independent variables are logged. 2) The instrumental variable for foreign capital is the 1-year lagged foreign capital stock. 3) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance.

25 Table 5: Robustness Checks for Regional Spillovers of Coastal FDI and Exports Dependent variable: the log of real GDP Alternative variable definitions Alternative specification Exclude extreme observations Another distance discounted effective coastal FDI flow Coastal foreign capital stock IV for coastal FDI and exports Without the 1 th and 99 th percentile of real GDP Without the 1 th and 99 th percentile of real GDP per capita (1) (2) (3) (4) (5) Coastal FDI 0.33*** 0.67 *** 0.82* 0.18*** 0.18*** (0.083) (0.13) (0.32) (0.053) (0.053) Coastal exports -0.95*** -0.69*** -2.40** -0.53*** -0.44** (0.20) (0.16) (0.98) (0.17) (0.18) Employment 0.28*** 0.23** 0.42*** 0.17* 0.26*** (0.093) (0.093) (0.16) (0.096) (0.09) Domestic capital stock 0.54*** 0.54*** 0.67*** 0.50*** 0.62*** (0.054) (0.055) (0.10) (0.056) (0.058) Foreign capital stock * (0.016) (0.017) (0.028) (0.018) (0.018) Province fixed effects Yes Yes Yes Yes Yes Year dummies Yes Yes Yes Yes Yes Obs Notes: 1) Independent variables are logged. 2) The instrumental variable for foreign capital is the 1-year lagged foreign capital stock. 3) Standard errors are in parentheses. * at 10% significance; **at 5% significance; *** at 1% significance. 25

26 Table 6: Allow Spillovers from Coastal Regions to Vary Dependent variable: the log of real GDP (1) (2) (3) Northern coastal FDI (0.57) (0.61) Eastern coastal FDI 0.16*** *** (0.041) (0.037) Northeastern coastal FDI (0.30 ) (0.84 ) Southern coastal FDI 0.38** (0.18 ) (0.35) Northern coastal exports *** 0.85*** (0.27) (0.29) Eastern coastal exports *** -0.30*** (0.11) (0.11) Northeastern coastal exports * dropped (0.77) Southern coastal exports (0.24) (0.46) Coastal FDI *** - (0.048) Coastal exports -0.39* - - (0.15) Province fixed effects Yes Yes Yes Year dummies Yes Yes Yes Obs Notes: 1) Northern coastal provinces are Beijing, Tianjin, and Hebei. Eastern coastal provinces are Shanghai, Jiangsu, Zhejiang, Fujian, and Shandong. Northeastern coastal province is Liaoning. Southern coastal provinces are Guangdong, Guangxi, and Hainan. 2) The unreported independent variables include employment, domestic capital stock and foreign capital stock. 3) All independent variables are logged. 4) The instrumental variable for foreign capital is the 1-year lagged foreign capital stock. 5) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance. 26

27 Table 7: Spillovers Vary across Investment Types Dependent variable: the log of real GDP (1) (2) (3) Joint venture FDI ** ** (0.037) (0.029) (0.029) Cooperative FDI 0.062** 0.047** 0.038** (0.025) (0.020) (0.019) Foreign solely funded FDI 0.17* 0.12* (0.095) (0.075) (0.074) Coastal exports (0.18) (0.14 ) (0.14) Employment (0.10) (0.10) Domestic capital stock *** 0.28 (0.050) (0.057) Foreign capital stock *** (0.024) Province fixed effects Yes Yes Yes Year dummies Yes Yes Yes Obs Notes: 1) Independent variables are logged. 2) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance. 27

28 Table 8: Spillovers Vary across FDI from Different Origins Dependent variable: the log of real GDP (1) (2) (3) FDI from HMT -0.39*** -0.35*** -0.26** (0.10) (0.10) (0.10) FDI from other origins (0.0063) (0.0061) (0.0060) Coastal exports -0.22** -0.28** -0.25** (0.11) (0.10) (0.099) Employment (0.093) (0.091) Domestic capital stock ** (0.066) (0.067) Foreign capital stock *** (0.029) Province fixed effects Yes Yes Yes Year dummies Yes Yes Yes Obs Notes: 1) Independent variables are logged. 2) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance. 28

29 Table 9: Spillovers Vary across Exports from Different Firms Dependent variable: the log of real GDP (1) (2) (3) Coastal FDI (0.066) (0.052) (0.052) Exports from foreign firms -0.19* (0.11) (0.092) (0.090) Exports from domestic firms *** 0.017** (0.012) (0.0091) (0.0090) Employment (0.11) (0.11) Domestic capital stock *** 0.31*** (0.050) (0.057) Foreign capital stock (0.025 ) Province fixed effects Yes Yes Yes Year dummies Yes Yes Yes Obs Notes: 1) Independent variables are logged. 2) Standard errors are in parentheses.* at 10% significance; **at 5% significance; *** at 1% significance. 29

30 References [1] Alfaro, Laura, Areendam Chanda, Sebnem Kalemli-Ozcan and Selin Sayek, FDI and Economic Growth: the Role of Local Financial Markets, Journal of International Economics, 2004; 64(1); [2] Borensztein, Eduardo, Jose De Gregorio, and Jong-Wha Lee, Does Foreign Direct Investment Affect Economic Growth? Journal of International Economics, 1998; 45(1); [3] Bronzini, Raffaello and Paolo Piselli, Determinants of Long-Run Regional Productivity with Geographic Spillovers: the Role of R&D, Human Capital and Public Infrastructure, Regional Science and Urban Economics, 2009; 39(2); [4] Brun, Jean-Frana, Jean-Louis Combes and Mary-Frana Renard, Are there Spillover Effects between Coastal and Noncoastal Regions in China? China Economic Review, 2002; 13(2-3); [5] Cheng, Leonard K. and Yum K. Kwan, What are the Determinants of the Location of Foreign Direct Investment? The Chinese Experience, Journal of International Economics, 2000; 51(2); [6] China Provincial Yearbooks, the National Bureau of Statistics of China. [7] Dean, Judith M., Mary E. Lovely and Hua Wang, Are Foreign Investors Attracted to Weak Environmental Regulations? Evaluating the Evidence from China, Journal of Development Economics, 2009; 90(1); [8] Fu, Xiaolan, Limited Linkages from Growth Engines and Regional Disparities in China, Journal of Comparative Economics, 2004; 32(1); [9] Fujita, Masahisa and Dapeng Hu, The Effects of Globalization and Economic Liberalization, Annals of Regional Science, 2001; 35 (1); [10] Funke, Michael and Annekatrin Niebuhr, Regional Geographic Research and Development Spillovers and Economic Growth: Evidence from West Germany, Regional Studies, 2005; 39(1); [11] Girma, Sourafel and Katharine Wakelin, Are there Regional Spillovers from FDI in the UK? University of Nottingham GEP Research Paper No. 16, [12] Girma, Sourafel and Yundan Gong, FDI, Linkages and the Efficiency of State-Owned Enterprises in China, Journal of Development Studies, 2008; 44(5); [13] Huang, Yasheng, Selling China: Foreign Direct Investment during the Reform Era, 2003, Cambridge University Press. 30

31 [14] Keller, Wolfgang, Geographic Localization of International Technology Diffusion, American Economic Review, 2002; 92 (1); [15] Kuo, Chunchien and Chihhai Yang, Knowledge Capital and Spillover on Regional Economic Growth: Evidence from China, China Economic Review, 2008; 19(4); [16] Lee, Hayan, Luca A. Ricci and Roberto Rigobon, Once Again, is Openness Good for Growth? Journal of Development Economics, 2004; 75(2); [17] Li, Xiaoying and Xiaming Liu, Foreign Direct Investment and Economic Growth: An Increasingly Endogenous Relationship, World Development, 2005; 33(3); [18] Madariaga, Nicole and Sandra Poncet, FDI in Chinese Cities: Spillovers and Impact on Growth, World Economy, 2007; 30(5); [19] Ng, Fungyee Linda and Chyau Tuan, Spatial Agglomeration, FDI, and Regional Growth in China: Locality of Local and Foreign Manufacturing Investments, Journal of Asian Economics, 2006; 17 (4); [20] OECD Manual-Measuring Productivity: Measurement of Aggregate and Industry-Level Productivity Growth, OECD. [21] Ouyang, Puman, Economic Growth, Industrial Development and the Inter-Regional Spillovers from Foreign Direct Investment, Evidence from China, 2009; Ph.D. dissertation. [22] Solow, Robert M., Technical Change and the Aggregate Production Function, Review of Economics and Statistics, 1957; 39 (3); [23] Wheeler, David and Ashoka Mody, International Investment Location Decisions: the Case of US Firms, Journal of International Economics, 1992; 33(1-2); [24] Zhang, Qing and Bruce Felmingham, The Role of FDI, Exports and Spillover Effects in the Regional Development of China, Journal of Development Studies, 2002; 38(4); [25] Zhang, Xiaobo and Kevin Zhang, How does Globalisation Affect Regional Inequality within a Developing Country? Evidence from China, Journal of Development Studies, 2003; 39(4);