EXTERNALITIES OF FDI: EVIDENCE FROM CHINA S EASTERN COASTAL AND CENTRAL PROVINCES CHAN YUEN TUNG STUDENT NO

Transcription

1 EXTERNALITIES OF FDI: EVIDENCE FROM CHINA S EASTERN COASTAL AND CENTRAL PROVINCES BY CHAN YUEN TUNG STUDENT NO A PROJECT SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF BACHELOR OF SOCIAL SCIENCES (HONOURS) DEGREE IN CHINA STUDIES ECONOMICS CONCENTRATION HONG KONG BAPTIST UNIVERSITY APRIL 2015

2 Page of Acceptance April 2015 We hereby recommend that the Project by Mr. CHAN Yuen Tung entitled Externalities of FDI: Evidence from China s Eastern Coastal and Central Provinces. be accepted in partial fulfillment of the requirements for the Bachelor of Social Sciences (Honours) Degree in China Studies in Economics. Dr. Erin SO Pik Ki Project Supervisor Second Examiner 1

3 Acknowledgement I would like to thank my supervisor Dr. Erin SO Pik Ki for guiding and enlightening me through out the entire study. Without her generous care and support, this paper is hardly finished. Thanks are also due to Dr. CHAN Hing Lin for his teaching of the econometric theories and applications and to Dr. LUK Sheung Kan for his pragmatic comments on the regression models. China Studies Degree Course Economics Concentration Hong Kong Baptist University

4 Table of Content PAGE OF ACCEPTANCE 1 ACKNOWLEDGEMENT 2 ABSTRACT 4 1. INTRODUCTION 5 2. LITERATURE REVIEW 9 3. METHODOLOGY AND REGRESSION MODEL DATA REGRESSION RESULT AND INTERPRETATION CONCLUSION AND POLICY IMPLICATION 37 REFERENCES 40 3

5 Abstract Using the panel data across 14 provinces in China s Eastern coastal and Central regions from 2002 to 2011, this paper finds that there are different levels of positive and significant externalities spilled out from the labors hired by FDI and HKMT firms in various sectors as well as the capital stocks invested by FDI firms. Moreover, export-led growth does exist in China s industry sector, but is limited to the domestic firms only. In addition, the export shares of FDI and HKMT firms do not affect the domestic economic growth. Lastly, an interesting finding is that increasing the capital inputs in construction sector does not necessarily generate efficient GDP output growth. 4

6 1. Introduction To take the advantages of the foreign direct investment (FDI), it is not uncommon to see that the less developed countries governments are competing with each other to offer different preferential policies, such as rental discounts, tax holidays and some special subsides, to attract the overseas investors. Doubtlessly, China is a case in point. For instance, in some sectors, the FDI firms can enjoy a 2-year tax holiday starting from the first year that they can make profit and after these 2 years, they can still have a 50% discount of the tax for the following 3 years. The major reason for doing so is that Chinese government realizes that there will be externalities brought from the FDI inflows, which will finally benefit the domestic economy. Thus, in the 1990s, with the government s efforts, China has become the largest recipient of the FDI among all other developing countries. Back to the late 1970s, China government has already set attracting the overseas capital as one of the economic reform strategies. Since the Law on Sino-Foreign Equity Joint Ventures 1 published in 1979, the annual inflow of FDI has stepped up steadily. In early 1992, China s top leader Deng promised to further open up the country and to accelerate the economic reform in his tour to the Southern provinces. Right after his speech, the annual FDI inflows of 1992 and 1993 have increased for more than the double and reached a peak of U.S billion in After China entered the World Trade Organization (WTO) in 2001, according to Figure I below, the inflows of FDI kept expending explosively until 2008, which was the year of global financial tsunami, to a level 1 It is a legal framework for FDI, which allows foreign investors to have equity joint ventures together with partners from China. 5

7 of U.S billion. Starting from 2009, the figure recovered and rebounded rapidly from U.S billion to another peak of U.S billion in FDI inflows in China (1990 to 2011) Billion (U.S.) Figure I China s FDI inflows ( ). Source: The World Bank. Although there are many FDI inflows in China, not every province can get the same amount of benefits. As Figure II below illustrates, the regional distribution of FDI is not even. The majority part, up to 85%, went to the Eastern coastal region; this is because in the beginning of the open door policies, the Eastern area acted as a white mouse, especially Guangdong province, as it is near Hong Kong and close to the coastal line, it has a better linkage with the overseas investors. As a result, Guangdong alone shared 25.3% of the total FDI inflows from 1990 to While the central region accounted for 10% during this period. Although the West regions shared only 5% of the total FDI from 1990 to 2011, this percentage indeed has already been increasing slowly from 3% in 1979 to

8 Regional Distribution of FDI in China (2002 to 2009) Central Region 10% West Region 5% Eastern Coastal Region 85% Figure II Regional Distribution of FDI Inflows in China ( ). Source: China Trade and External Economic Statistical Yearbook. Eastern coastal region: Hebei, Liaoning, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong and Hainan. Central Region: Shanxi, Anhui, Jiangxi, Henan, Hubei and Hunan. West Region: Shaanxi, Sichuan, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Ningxia and Xinjiang. Besides the uneven geographic distribution, the sectorial distribution of FDI inflows is also uneven. As the production costs in China are relatively cheap; therefore, according to the Figure III below, the secondary sector, especially the industry, benefited the most and accumulated for more than half of the FDI inflows; only industry alone got 56% of the total FDI inflows from 2002 to As above data shows, the FDI inflows after China entered WTO in 2001 have been increasing rapidly and majorly concentrates on the 2 nd sector, especially the industry sector, and in the Eastern coastal and the central regions. Therefore, this paper collects a panel data crossing 14 provinces, namely: Hebei, Liaoning, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan, Shanxi, Anhui, Jiangxi, Henan, Hubei and Hunan, from 2002 to 2011 to analyze the externalities brought by these investments to different levels of the domestic economic growths and will put more attentions on the industry sector. The current study 7

9 will also analyze and compare the effects brought from 2 different origins of the FDI, which are the investments from the foreign and HKMT investors. Sectorial Distribution of FDI in China (2002 to 2009) 1st Sector 1% 3rd Sector 42% 2nd Sector - Industry 56% 2nd Sector - Construction 1% Figure III Sectorial Distribution of FDI Inflows in China (2002 to 2009). Source: China Trade and External Economic Statistical Yearbook. The rest of this paper is organized as follows. The previous literatures of the related topics are summarized in section 2. The regression models are shown in section 3. Section 4 describes the data and the processing procedures. The econometric results and the related interpretations are contained in section 5. The concluding remarks and some policy implications are in the final section 6. 8

10 2. Literature Review According to Qi and Li (2008), there are 3 main ways to increase the technology, innovation and creativity level, which are: I. Creating knowledge independently by research and development (R&D); II. III. Purchasing advanced technology and know-how from international trade; Spilling over the knowledge from FDI enterprises to the host countries; and hence, they will increase the productivity of the society and the economic growth as well. Unlike the former two ways, the externalities in form of knowledge spillover generated by FDI are relatively indirect and there are 5 major channels for it to carry out. First and foremost, the local partners can learn the production processes and technology directly from the investors. Labor mobility effect can be the second channel. The turnover of the trained and skilled labor from the FDI firms to the domestic related counterparts will also bring the technological know-how to their new domestic firms. Demonstration effect, in which the products and the inventions of the FDI companies can stimulate and enlighten the local R&D activities, can be another channel (Jianhong Qi & Hong Li, 2008). The fourth channel is the competitive effect. Last but not least, there will be the vertically knowledge spillover through the forward and backward linages of supply chains. For example, according to Smarzynska (2002), the local suppliers in Lithuania will have higher productivities when there is a greater presence of the FDI companies. 9

11 Economists generally agree that FDI does insert an externality - the positive knowledge spillover effect and as a result, the production technology, efficiency and thus the economies will grow in the host countries, which are usually the Lesser-Developed Countries (LDCs), can be enhanced. For instance, Findlay (1978) suggests that the speed of technical progress in the host country can be increased infectiously by the more advanced know-how and management methods used in the FDI corporations. Walz (1997) claims that the multinational companies in LDCs will spill their knowledge over to the domestic R&D sectors and causing the economic growths in the host countries happen eventually. Although most of the related theoretical literatures show the positive relations between the levels of presence of the FDI firms, the technical progresses and the productivities of the host countries, interestingly, the results of empirical studies are somehow mixed and diverse. Major views from the empirical literatures are summarized as follow: FDI has a positive effect on the technology progresses and productivities in the host countries. Many scholars find out that the domestic firms are likely to absorb the knowledge spillover by the FDI and learn from them to increase their own productivities and competitiveness in the market. Rhee and Belot (1989) discover that the creations as well as the growth of the local-owned textile firms in Bangladesh and Mauritius are stimulated by the entry of a few foreign-owned firms. Hanel (2000) assumes that there are relations between the shares of sales of the foreign subsidiaries in 19 industries of Canada 10

12 and those subsidiaries related knowledge stocks; then calculates how R&D in those subsidiaries affect the growth of the domestic factor productivity; and draws the conclusion that the knowledge creation and the productivity growth can be contributed by those non-local knowledge stocks. Branstetter (2000) figures out that FDI is an important way for knowledge spillovers between both the indigenous companies in US and the investing companies from Japan using the data of patent citations. More recently, Lee (2006) summarizes that the global knowledge spillover via FDI is obvious after analyzing the data of 16 Organization for Economic Cooperation and Development (OECD) countries from 1981 to Wang et al. (2006) find that the inflows of FDI are one big incentive for the domestic companies in China to lift their R&D inputs in order to maintain their competitiveness under the international and globalization pressures. FDI does not have a positive effect on the technology progresses and productivities in the host countries. Some scholars think that the spillover effect from the FDI is not strong and robust enough, whereas some may even deem that the effect does not exist at all. As a result, the domestic parties do not learn or adopt the more advanced technology, know-how or managements used by the FDI firms to increase their productivities. A case in point is that the orders about assembly of electronic devices received by the local firms from the foreign investors do not help the former to learn any new or advanced technology besides the simple and low-value-added assembly work. Another case in point is that, according to Xian and Yan (2005), in order for 11

13 the spillover effects from FDI to occur, the local firms have to achieve certain basic technology levels; yet, the provinces in the middle and west part of China have not pass through the doorsill and thus, no robust spillover effects can be observed. Chen (2007) argues that the influences from FDI to China s regional innovation capabilities are feeble and furthermore, the FDI inflows may even crowd out the domestic R&D activities. The spillover effect of FDI on the technology progresses and productivities in the host countries is unclear. Besides the R&D stocks, economists generally understand and accept that the capital stocks of the FDI firms can also spill over and enhance the productivities of the local companies. However, Todo (2006) refutes that, basing on the data of Japanese manufacturing industries from 1995 to 2002, there is a positive effect of R&D stocks, but not capital stocks, on the knowledge and productivity growths of the host countries, and that result can be interpreted as the daily production practices of the FDI firms do not generate spillover effects, but only the R&D activities do. This uncertain spillover effect is somewhat similar in China in the eyes of Jiang and Xia (2005); using the data of China s hi-tech industries, they claim that FDI does provide some positive effects towards the domestic counterparts, yet, their own R&D inputs and the numbers of the scientific and technical researchers, staff employed are the more important factors contributing the knowledge creations of the domestic firms; in addition, they find that the competitive effect does not only bring goods to the firms, while the local firms may also lose their markets as well as their own strategic R&D resources due to the intensive competition pressures. 12

14 As mentioned above, the empirical studies towards this issue come with different and diverse opinions. And the related studies towards China s regional situations after 2001, which is the entry year of WTO, comparing the direct investments from different origins are relatively rare. Hence, this current paper tries to analyze the externality, which can drive the economic growth, brought by the FDI from the foreign investors and HKMT investors in the Eastern coastal and the central regions of China. 13

15 3. Methodology and Regression Model The Gross Domestic Product (GDP), measures the total gross values added by all the residents participated in productions, which also is the aggregate production level of a country. It is commonly used to measure to the economic performances of the countries as well as various sectors contributions. In this current study, instead of Nominal GDP (NGDP), Real GDP (RGDP) is used as it separates the effects of inflation or deflation from NGDP and therefore, it can better indicate the economic performances. In order to observe the real economic growth caused by the externalities, taking the natural logarithm of the RGDP is an essential step to change it into the growth rate. Thus, lngdp, which is the natural logarithm of the deflated NGDP, will act as the dependent variables in following 9 regression models. The basic components and regression model to measure the externalities, according to Chen (2011) in his book 2, are as follow: lngdp = Constant + αlnl + βlnk + γ (Other independent variable) + ε where α, β and γ are the coefficients that capture the impacts from the growth of L (labor), the growth of K (fixed capital stock) and the other independent variable related to the externalities. ε represents the random error term of the model. 3 And lnl and lnk are the growth rates of labor number and capital stock respectively. While other independent 2 Indeed, the model originally comes from the production function: Y = AL α K β. 3 The mean of this term should be zero. 14

16 variable can be any variable that can insert an impact on the dependent variable, such as the FDI inflows growth. This current paper mainly focuses on the labor and the export side; since, comparing with the spillover effect brought by the capital stock, these two factors effects are relatively less studied and indirect in the sense that the knowledge or know-how basically has to spill through learning and adapting but not directly to acquire through purchasing. Basing on the above production function, 9 gradually established regression models are shown below (see Table I for definitions of the subscripts); to avoid the multi-collinear problem, expect model (7) and (8), independent variables of the domestic firms are not placed into the models. InGDP p,t = C + β 1 InL p,t + β 2 InK p,t + β 3 In(L F,p,t / L p,t ) + β 4 In(L HKMT,p,t / L p,t ) + ε (1) Subscript Definition p Pinvince t Time Period (Year) t-1 The Last Time Period (Lagged 1 Year) F Foreign Direct Investment Firms HKMT Hong Kong, Macau and Taiwan Investment Firms D Domestic Firms 1st Sector The First Economic Sector 2nd Sector The Second Economic Sector 3rd Sector The Third Economic Sector NON 2nd Sector The First and the Third Economic Sectors Industry Industry Sector Construction Construction Sector TEV Total Export Value TPV Total Production Value Table I Definitions of the Subscripts. 15

17 where In(L F,p,t / L p,t ) and In(L HKMT,p,t / L p,t ) are the growth rates of the ratios of labors employed by FDI firms as well as HKMT investment firms among the total employed labors respectively; while the coefficient β 3 and β 4 measure the magnitudes of the impacts from the growths of the FDI and HKMT investment firms labors to the total employment ratio severally. This model is set to study whether changes of the portions of labors hired by the FDI and HKMT investment firms will affect the GDP growth. Model (2) is somehow similar with model (1), except it is set to study whether changes of the portions of capital stocks, instead of the employed labors, invested by the FDI and HKMT investment firms will affect the GDP growth. InGDP p,t = C + β 1 InL p,t + β 2 InK p,t + β 3 In(K F,p,t / K p,t ) + β 4 In(K HKMT,p,t / K p,t ) + ε (2) where In(K F,p,t / K p,t ) and In(K HKMT,p,t / K p,t ) are the changes of the ratios of fixed capital stocks invested by FDI firms as well as HKMT investment firms among the total fixed capital stock respectively; while the coefficient β 3 and β 4 indicate the degrees of the impacts from the growths of the FDI and HKMT investment firms fixed capital stocks to the total fixed capital stock ratio severally. In order to further study the spillover effects through the employed labors in different economic sectors, model (3) and (4) are gradually developed from model (1) above. InGDP 2nd Sector,p,t = C + β 1 InL 2nd Sector,p,t + β 2 InK 2nd Sector,p,t + β 3 In(L 2nd Sector,F,p,t / L 2nd Sector,p,t ) + β 4 In(L 2nd Sector,HKMT,p,t / L 2nd Sector,p,t ) + ε (3) 16

18 InGDP NON 2nd Sector,p,t = C + β 1 InL NON 2nd Sector,p,t + β 2 InK NON 2nd Sector,p,t + β 3 In(L NON 2nd Sector,F,p,t / L NON 2nd Sector,p,t ) + β 4 In(L NON 2nd Sector,HKMT,p,t / L NON 2nd Sector,p,t ) + ε (4) Unlike model (1), model (3) and (4) above are set to test whether the growths of the ratios of L employed by the FDI firms, HKMT investments firms in the secondary and nonsecondary sector to the total employment in these 2 sectors will affect the growths of GDP in these sectors respectively. If model (3) shows significant and robust coefficient β 3 or β 4, then by using model (5) and (6), it is possible to predict whether the spillover effects mainly occur in the industry or the construction sector. InGDP Industry,p,t = C + β 1 InL Industry,p,t + β 2 InK Industry,p,t + β 3 In(L Industry,F,p,t / L Industry,p,t ) + β 4 In(L Industry,HKMT,p,t / L Industry,p,t ) + ε (5) InGDP Construction,p,t = C + β 1 InL Construction,p,t + β 2 InK Construction,p,t + β 3 In(L Construction,F,p,t / L Construction,p,t ) + β 4 In(L Construction,HKMT,p,t / L Construction,p,t ) + ε (6) The coefficient β 3 and β 4 in model (5) are set to capture the impacts from the growths of the labors working in industry sector, which are hired by FDI firms as well as HKMT investment firms respectively, over the total industrial employment to the industry s GDP growth; while β 3 and β 4 in model (6) are not far-off but targeted at the construction sector. Besides studying L and K, the export of industry sector is also worthwhile to pay attention as the export sector, which is mainly composed by the industrial export sector in 17

19 China, can cause the export-led growth in many developing countries. 4 Therefore, model (7) below is established to test the relation between the GDP growth of the industry sector and the changes of the ratio of total export value to total production value. InGDP Industry,p,t = C + β 1 InL Industry,p,t + β 2 InK Industry,p,t + β 3 In(TEV Industry,p,t / TPV Industry,p,t ) + ε (7) If β 3 in model (7) is significant and robust, then, by using model (8), whether the GDP change of the industry sector, which is driven by the export, is mainly from the growths of the export portions of the FDI firms, HKMT investment firms or domestic firms to their own production values can possibly be identified. InGDP Industry,p,t = C + β 1 InL Industry,p,t + β 2 InK Industry,p,t + β 3 In(TEV Industry,F,p,t / TPV Industry,F,p,t ) + β 4 In(TEV Industry,HKMT,p,t / TPV Industry,HKMT,p,t ) + β 5 In(TEV Industry,D,p,t / TPV Industry,D,p,t ) + ε (8) Apart from the export to production ratio, the growths of the shares of exports from FDI firms as well as from HKMT investment firms among the total export value may also drive the GDP growth in industry sector, and thus, model (9) below is set to test the above statement. InGDP Industry,p,t = C + β 1 InL Industry,p,t + β 2 InK Industry,p,t + β 3 In(TEV Industry,F,p,t / TEV Industry,p,t ) + β 4 In(TEV Industry,HKMT,p,t / TEV Industry,p,t ) + ε (9) 4 See Tiwari and Mutascu (2011). 18

20 In order to prevent the problem of reverse causality, another 9 regression models are also formulated with the subscript t of all the independent variables in above 9 models replaced by t-1 (lagged for 1 year) for robustness checking as follows: InGDP p,t = C + β 1 InL p,t-1 + β 2 InK p,t-1 + β 3 In(L F,p,t-1 / L p,t-1 ) + β 4 In(L HKMT,p,t-1 / L p,t-1 ) + ε 1(t-1) InGDP p,t = C + β 1 InL p,t-1 + β 2 InK p,t-1 + β 3 In(K F,p,t-1 / K p,t-1 ) + β 4 In(K HKMT,p,t-1 / K p,t-1 ) + ε 2(t-1) InGDP 2nd Sector,p,t = C + β 1 InL 2nd Sector,p,t-1 + β 2 InK 2nd Sector,p,t-1 + β 3 In(L 2nd Sector,F,p,t-1 / L 2nd Sector,p,t-1 ) + β 4 In(L 2nd Sector,HKMT,p,t-1 / L 2nd Sector,p,t-1 ) + ε 3(t-1) InGDP NON 2nd Sector,p,t = C + β 1 InL NON 2nd Sector,p,t-1 + β 2 InK NON 2nd Sector,p,t-1 + β 3 In(L NON 2nd Sector,F,p,t-1 / L NON 2nd Sector,p,t-1 ) + β 4 In(L NON 2nd Sector,HKMT,p,t-1 / L NON 2nd Sector,p,t-1 ) + ε 4(t-1) InGDP Industry,p,t = C + β 1 InL Industry,p,t-1 + β 2 InK Industry,p,t-1 + β 3 In(L Industry,F,p,t-1 / L Industry,p,t-1 ) + β 4 In(L Industry,HKMT,p,t-1 / L Industry,p,t-1 ) + ε 5(t-1) 19

21 InGDP Construction,p,t = C + β 1 InL Construction,p,t-1 + β 2 InK Construction,p,t-1 + β 3 In(L Construction,F,p,t-1 / L Construction,p,t-1 ) + β 4 In(L Construction,HKMT,p,t-1 / L Construction,p,t-1 ) + ε 6(t-1) InGDP Industry,p,t = C + β 1 InL Industry,p,t-1 + β 2 InK Industry,p,t-1 + β 3 In(TEV Industry,p,t-1 / TPV Industry,p,t-1 ) + ε 7(t-1) InGDP Industry,p,t = C + β 1 InL Industry,p,t-1 + β 2 InK Industry,p,t-1 + β 3 In(TEV Industry,F,p,t-1 / TPV Industry,F,p,t-1 ) + β 4 In(TEV Industry,HKMT,p,t-1 / TPV Industry,HKMT,p,t-1 ) + β 5 In(TEV Industry,D,p,t-1 / TPV Industry,D,p,t-1 ) + ε 8(t-1) InGDP Industry,p,t = C + β 1 InL Industry,p,t-1 + β 2 InK Industry,p,t-1 + β 3 In(TEV Industry,F,p,t-1 / TEV Industry,p,t-1 ) + β 4 In(TEV Industry,HKMT,p,t-1 / TEV Industry,p,t-1 ) + ε 9(t-1) 20

22 4. Data The econometric analyses in this paper are based on data of 14 provinces across the Eastern coastal and central region of China from 2002 to The annual provincial data is collected from various statistical reports: CHINA STATISTICAL YEARBOOK, CHINA POPULATION & EMPLOYMENT STATISTICS YEARBOOK, CHINA LABOUR STATISTICAL YEARBOOK, STATISTICAL YEARBOOK OF THE CHINESE INVESTMENT IN FIXED ASSETS, CHINA INDUSTRY ECONOMY STATISTICAL YEARBOOK, CHINA REAL ESTATE STATISTICS YEARBOOK, CHINA EXTERNAL ECONOMIC STATISTICAL YEARBOOK and different provincial statistical yearbooks, such as JIANGSU STATISTICAL YEARBOOK. Since NGDP cannot reflect the actual economic performance, GDP used in this paper is the NGDP deflated by the GDP deflator, which takes 1978 as the base year. Labor is measured as the employed labor number at the end of the year. The export and production value are counted in the provinces which product the final goods. For example, if Guangxi produces a car and Guangdong acts as the exporter, the production value as well as the export value will both count into the former s account. As Liu (2002) mentions, the capital stock measurement is a well-recognized problem in the empirical studies; it does not only exist in the studies of China, but also other countries. There are several approaches to estimate the total capital stock in a country; unlike some fellows who will use the fixed capital investment figures directly as the capital stock, this paper uses a scientific method used by Kim and Lau (1994) and Lei and Yao (2009), which the initial capital stocks of China, Hong Kong and Macau, are assumed to be 21

23 5 times of the real gross fixed capital formation in the same year and the real gross fixed capital formation after the initial year can be used as a proxy for the change in the capital stock each year. And China s annual depreciation rate of the capital, as used by Perkins (1988), Woo (1998), Meng and Wang (2000) and Wang and Yao (2003), is assumed to be 5%. Therefore, in the current study, the nominal fixed capital formation of the initial year (2002) is firstly deflated by the investment in fixed assets price index, which takes 1978 as the base year; then by multiplying the value 5 times, the initial capital stocks can be found. And the capital stock of the year after 2002, i.e. 2003, is the sum of 95% of the capital stock of 2002 plus the newly real fixed capital formation of 2003 (deflating the nominal fixed capital formation by the investment in fixed assets price index of 2003). Furthermore, as Holz (2004) claims that China s official statistics are of questionable quality and inaccuracy, the inconsistency among the data used by this study is relatively apparent. To cope with this problem, if there is inconsistency of a data point, the data from a later time period will be given the first priority and the data from a higher authority will be given the second priority in this study. 5 5 This practice is in a belief that the data announced later may be amended and therefore they should be generally more consistent and competent; while the data published by a higher authority might be more accurate because the data may be processed more seriously. 22

24 5. Regression Result and Interpretation As there are various institutions and characteristics of each province after the decentralization, fixed-effects model should be used to process the panel data to separate the differences and let each province has a different constant term. However, probably due to the insufficient sample size as the time limitation, the results from 18 models are almost statistically insignificant. Hence, pooled ordinary least squares (pooled OLS) is used to carry out the regressions in the current paper. And the results are shown below. From the results of 9 models as well as 9 (t-1) models, generally speaking, the coefficients of lnl and lnk, which are β 1 and β 2, are positive and statistically significant at the 1% level. And the means of the all coefficients of them are and respectively. Therefore, it implies that, on average, every 1% of labor increase will lead to a 0.60% growth of GDP, while every 1% increase of capital stock will lead to 0.39% growth of GDP. And the results above are indeed within the expectation as L and K are two basic elements that have positive relations with the output in the production function. Moreover, it suggests that, from a macro view, putting 1% extra labor in economic activities will give a higher output growth then putting 1% more of the capital stock. And it denotes that in these regions, economic activities and growths are still relying more heavily on labor instead of using capital such as machines and computers and it can also be interpreted as the economy of China s Eastern coastal and central areas is relatively labor-intensive. However, one thing to be highlighted is that the coefficients of the InK Construction in model (6) and its (t-1) model, unlike all other coefficients of lnl and lnk in remaining models, are not statistically significant. This suggests that the increase in fixed capital stock 23

25 in the construction sector of those provinces does not push up the GDP growth of the related field and this can also be interpreted as the capital investments in China s constriction sector are inefficient to generate positive outputs. Table II Estimation of GDP growth effect by the changes of (L F /L), (L HKMT /L), (K F /K) and (K HKMT /K). a) Models with (t-1) mean that the independent variables are lagged for 1 year for robustness checking. b) The italic numbers in the table above are the p-values. c) *** stands for p-value < 0.01; ** stands for p-value between 0.01 & 0.05; * stands for p-value between 0.05 & 0.1. According to Table II, the results of model (1) show that if the ratio of labors hired by the FDI firms to the total employment of a province increase 1%, the GDP of that place will have a 0.12% increase; while the labor hired by HKMT investment firms to the total employment ratio of a province increase 1%, it will lead to a 0.06% rise in GDP of that 24

26 province. And both findings above are significant at the 1% level and the results still hold their significances and robustness in the (t-1) model. The empirical results reveal that the increases of ratios of labors from FDI firms and HKMT firms can both have positive effects on the GDP growth; but with the same 1% increase in the ratio, the labor hired by the FDI firms will give a nearly 100% higher return then the labors hired by HKMT firms to the GDP growth. It denotes that the spillover effects are stronger through the labors hired by the FDI firms to the domestic GDP growth. It may due to different reasons and for example, the worker training sections are better and/or the standards of production, such as the knowhows, technical requirements in the working environment, are higher in the FDI firms, comparing with HKMT ones; as there is a hypothesis that when the multinational firms decided to enter a country, i.e. China, they will have ensured that the revenues brought by their competitive advantages, such as high technologies, are big enough to cover the huge costs. And when the labors in the FDI firms go to domestic firms, the local firms can enjoy the relatively high-skilled labors. Hence, the general productivities and contributions to the local GDP growth of the labors in FDI firms are higher than those in HKMT firms. Model (2) s results in Table II above show that when the percentage of the capital stock owned by FDI firms among the total amount of the capital stock increases 1%, the GDP will have a rise of 0.22% and this finding is robust and significant at the 1% level both in Model (2) and its (t-1) model. Whereas the coefficients of the ratio of capital stock owned by HKMT firms to the total amount of the capital stock are statistically insignificant in Model (2) and also its (t-1) model, suggesting that there are no association between this ratio and the GDP growth. By assuming that the more advanced capital stocks can generate a higher output value and contribute more to the GDP, the above findings reveal that, 25

27 comparing to the fixed asset capital stocks owned by HKMT firms, the ones owned by FDI firms are more advanced and with higher productivities and this finding is also in line with the hypothesis mentioned above. Therefore, if the ratio of the capital owned by the FDI firms to the total capital stock amount in a province increases, there will be a positive spillover effect to domestic sector as well as a positive growth of the GDP. Table III Estimation of GDP growth of 2 nd sector and NON 2 nd sector effects by the changes of (L 2nd Sector,F / L 2nd Sector ), (L 2nd Sector,HKMT / L 2nd Sector ) as well as (L NON 2nd Sector,F / L NON 2nd Sector ) and (L NON 2nd Sector,HKMT / L NON 2nd Sector). a) Models with (t-1) mean that the independent variables are lagged for 1 year for robustness checking. b) The italic numbers in the table above are the p-values. c) *** stands for p-value < 0.01; ** stands for p-value between 0.01 & 0.05; * stands for p-value between 0.05 &

28 level to claim that when the ratio of labors employed by HKMT firms within the non-2 nd 27 Model (3) and (4) in Table III are further developed from model (1) to see how spillover effects occur through the labors employed by the FDI and HKMT firms in the secondary and the non-secondary sector. Basing on the regression results of model (3) as well as the related (t-1) model, the coefficients of the ratio of labors hired by the FDI firms within the secondary sector to the total employment number of the secondary sector are positive and statistically significant at the 1% level, showing that when there is a 1% increase in this ratio, the GDP of the secondary sector will increase by 0.04%. On the other hand, model (3) also shows that there are no statistical relation between the growth of GDP in secondary sector and the change of the ratio of labors hired by HKMT firms within the secondary sector to the total employment number of the secondary sector, since the coefficients of this ratio are insignificant in both model (3) and its (t-1) model. The empirical results above reveal that, within the secondary sector, the spillover effects to the domestic economic growth will transmit through the labors hired by the FDI firms but not through those who are hired by HKMT firms. And thus, it implies that the productivities of the labors employed by the FDI firms in the 2 nd sector are averagely higher; thence, when more and more labors work in the FDI firms, with the total labor number of the 2 nd sector unchanged, more and more labors may have a better knowledge and productivity and contribute more to the 2 nd sector s GDP growth. A reason behind may be the high standard and technology productions of the FDI firms, but not HKMT ones, in the industry sector that can spill the technical skills and the knowledge to the domestic labors and this reason is further confirmed in model (5), which will be discussed later. According to model (4) and its (t-1) model in Table III, there is a 99% confidence

29 sector to the total employment number of the non-secondary sector goes up 1%, it will lead to a 0.08% increase of the non-secondary GDP. However, in the non-2 nd sector, as the coefficients of the ratio of labors hired by the FDI firms to the total employment number are insignificant in model (4) as well as the (t-1) model, there are no statistical relations between this ratio and the GDP growth of the non-2 nd sector. It discloses that, in the nonsecondary sector, which is mainly composed by 3 rd sector 6, the workers hired by HKMT firms have a higher contribution to the GDP growth of the related sectors, comparing with the ones hired by the FDI firms; and there are some possible reasons to explain. Closer Economic Partnership Arrangement (CEPA) involving Hong Kong and Macau may be one of the reasons as it liberates various high value-added 3 rd sectors, such as banking sector, insurance service sector and security markets, to the companies from Hong Kong and Macau. The labors of HKMT firms can work in the higher economic output fields, which also require higher human capitals, than the ones work in FDI firms by assuming that higher value added sectors need higher human capitals, and consequently, the workers in HKMT firms of non-2 nd sector will conduct a stronger spillover effect to the domestic economic growth. Culture may also be another reason to explain. Since the 3 rd sector is majorly composed by the service sector, unlike the industry and construction sector in 2 nd sector, it is much more human and hence, with the same service quality, the culture of a company is relatively important. By assuming that HKMT firms will have a more similar cultural background with Chinese consumers; one can claim that HKMT firms will be more 6 The size and the economic output value from the 3 rd sector are much bigger than the 1 st sector. For instance, China 3 rd sector s GDP was around 4.6 times higher than the 1 st sector one in

30 popular than the FDI firms in the service sector 7 ; in another words, the labors in these HKMT firms can adapt to the Chinese service market environment better and thus, there can be greater spillover effects regarding the knowledge and the skills, such as selling skills, which can generate higher outputs. And when these workers go to domestic firms, the skills they learnt could then spill to the domestic firms. 8 Table IV Estimation of GDP growth of industry sector and construction sector effects by changes of (L Industry,F / L Industry ), (L Industry,HKMT / L Industry ) as well as (L Construction,F / L Construction ) and (L Construction,HKMT / L Construction ). a) Models with (t-1) mean that the independent variables are lagged for 1 year for robustness checking. b) The italic numbers in the table above are the p-values. c) *** stands for p-value < 0.01; ** stands for p-value between 0.01 & 0.05; * stands for p-value between 0.05 & This assumes that the service qualities of the FDI firms and HKMT firms are the same. 8 This is because the service qualities of the FDI and HKMT firms are generally better than the local ones. 29

31 Model (5) and (6) are built up from model (3) to see and compare the spillover effects through the labors hired by the FDI firms and HKMT firms in 2 sectors of the 2 nd sector, namely the industry sector and the construction sector. From the regression results of model (5) in Table IV above, the estimated coefficient of the ratio of labors employed by FDI firms in the industry sector to the total labor employment number in the industry sector is positive and significant at the 10% level, which means that, in China s Eastern coastal and central areas, every 1% increase of this ratio, it will lead to a 0.08% growth of the industry s GDP. Yet, this finding, indeed, is relatively feeble comparing with other findings in this paper as the coefficient of above ratio in its (t-1) function is insignificant statistically; more works have to be done to solidify this finding. On the other hand, significant and robust relation does not exist in the labors employed by HKMT firms according to the results above. In both model (5) and its (t-1) model, the results shows that there are no relations found between the GDP growth of the industry sector and the change of the ratio of the industrial labors hired by HKMT firms to the total industrial employment number as both coefficients in these 2 models are insignificant. The aim of setting model (6) is to find out the relations between the GDP growth of the construction sector and the ratios of the workers in construction sector employed by FDI firms as well as HKMT firms to the total employment number of the construction sector. However, as the coefficients of both ratios are statistical insignificant in model (6) as well as in its (t-1) model. It denotes that no matter how the portion of workers in construction sector employed by the FDI firms or by HKMT firms among the total 30

32 construction sector employment changes, the GDP growth rate of the construction sector will not be affected. This may be explained by construction sector s situation. As in the construction sector, it is common to see that the large infrastructure projects are indeed launched by the state; therefore, the contracts are usually got by the domestic firms. As a result, the numbers of labors hired by the FDI firms as well as HKMT firms are small, i.e. the ratios of them to the total employment in construction sector are generally below 5% from 2002 to Thus, even if there are spillover effects through these labors, the effects might not be statistically significant. As illustrated previously in model (3) of Table II, if there is a 1% increase of the ratio of labors hired by the FDI firms in the 2 nd sector to the total employment of the 2 nd sector, it will lead to a 0.04% increase of the GDP of 2 nd sector. One possible reason behind is that the FDI firms in industry sector are using relatively higher technologies and/or having higher productivities and the labors working there should have absorbed the skills and technical know-hows. As a result, when this kind of labors portion becomes relatively bigger in the society, it will lead to a higher economic growth and their knowledge will also spill to the domestic firms someday later. And this reason is now solidified by the results of model (5) and (6). Increasing the number of labors employed by the FDI firms in the industry sector with total employment number of the industry sector unchanged does have a positive relation with the GDP growth of industry sector. As a major part of GDP growth of the 2 nd sector is in fact coming from the GDP growth of the industry sector 9 ; therefore, one 9 From 2012 to 2013, 2011 to 2012 and 2010 to 2011, the increases of the GDP of the industry sector accounted for 76%, 76% and 84% of the GDP growth of the 2 nd sector, while the rest of 24%, 24% and 16% of the GDP growths of the 2 nd sector are contributed by the construction sector respectively. 31

33 of the sources for the spillover effects to carry out through the labors hired by FDI firms in 2 nd sector to the domestic economic growth is probably from the spillover effects through labors hired by the FDI firms in the industry sector but not in the construction sector; yet, to further ensure this statement, new models as follow should be set and tested: InGDP 2nd Sector,p,t = C + β 1 InL 2nd Sector,p,t + β 2 InK 2nd Sector,p,t + β 3 In(L Industry,F,p,t / L Industry,p,t ) + β 4 In(L Industry,HKMT,p,t / L Industry,p,t ) + ε and InGDP 2nd Sector,p,t = C + β 1 InL 2nd Sector,p,t + β 2 InK 2nd Sector,p,t + β 3 In(L Construction,F,p,t / L Construction,p,t ) + β 4 In(L Construction,HKMT,p,t / L Construction,p,t ) + ε Table V Estimation of GDP growth of industry sector effect by changes of (TEV Industry / TPV Industry ), (TEV Industry,F / TPV Industry,F ), (TEV Industry,HKMT / TPV Industry,HKMT ) and (TEV Industry,D / TPV Industry,D ). a) Models with (t-1) mean that the independent variables are lagged for 1 year for robustness checking. b) The italic numbers in the table above are the p-values. c) *** stands for p-value < 0.01; ** stands for p-value between 0.01 & 0.05; * stands for p-value between 0.05 &

34 Unlike previous models, model (7) aims at figuring out the relation between the GDP growth of the industry sector and the ratio of the total industrial export value to the total production value in order to see whether there will be export-led growth 10 in China s Eastern coastal and central areas industry sectors. From the results of model (7) in Table V, the estimated coefficient of the total industrial export value to the total industrial production value is positive and significant at the 1% level. It reveals that when this ratio goes up by 1%, the GDP growth of the industry sector will increase 0.09% and this finding is also robust in the (t-1) model. Besides L and K, the growth of export to production ratio of the industry sector can also drive the GDP growth of the industry sector positively; it denotes that with the total industrial production value unchanged, one can increase the economic growth of the industry sector by increasing the export value of the industrial goods. According to Grossman and Helpman (1991), trade can promote technology diffusion and knowledge spillover and hence lead to a faster productivity growth. Therefore, as China exports more with the total production level unchanged, there should be a bigger spillover effect from the trade to the domestic economic sector. 10 Indeed, to see the export-led growth of the whole country should use the model below: InGDP p,t = C + β 1 InL p,t + β 2 InK p,t + β 3 In(TEV p,t / TPV p,t ) + ε but not only limited to the industry sector. Yet, the export sector is only composed of primary goods and manufactured goods in China and the export from the industry sector accounts the majority part of the export sector; i.e. in 2013, 2012 and 2011, manufactured goods export values account for 95%, 95% and 94% of China s total export values. The relations between the export-led growth of the whole society and the change of the ratio of the total export value (which is mainly from the industry sector) to the total production value may not be obvious as the spillover effects have to be strong enough to spill to the non-industry sector in a short period of time. Since the models set in this paper are only the present year and the lagged one year (t-1), this paper focuses only on the industry sector s export-led growth effect. 33

35 Model (8) tries to find out whether the spillover effect found in model (7) is from the FDI firms, HKMT investment firms or from the domestic firms. According to the results in Table V, the coefficient of the ratio of industrial total export value of the domestic firms to the industrial total production value of these local firms, unlike the ratios of the FDI firms and HKMT firms, is the only robust finding and it is significant at the 1% level in both model (8) and its (t-1) model. In fact, the findings above are not difficult to understand. Comparing to the FDI and HKMT investment firms, domestic firms, without doubt, are relatively less productive. As trade can increase the chances of technology diffusions by letting the less advanced party to exposes to the more productive ones and to learn from the latter, therefore, the spillover effect from the exports of industrial goods only appears in the relatively backward Chinese domestic industrial firms according to the regression results above. 34

36 Table VI Estimation of GDP growth of industry sector effect by changes of (TEV Industry,F / TEV Industry ) and (TEV Industry,HKMT / TEV Industry ). a) Models with (t-1) mean that the independent variables are lagged for 1 year for robustness checking. b) The italic numbers in the table above are the p-values. c) *** stands for p-value < 0.01; ** stands for p-value between 0.01 & 0.05; * stands for p-value between 0.05 & 0.1. Unlike Model (7) and (8), model (9) is used to see whether the shares of the industrial total export value of the FDI firms as well as HKMT firms among the total industrial export value will affect the growth rate of GDP of the industry sector. As the estimated results indicate that the coefficients of both ratios of the FDI firms as well as HKMT investment firms are statistical insignificant in both model (9) and the (t-1) model, the changes of the export shares amount the FDI firms and HKMT investment firms do not affect the industrial GDP growth rate. One of the reasons to explain the above findings is 35

37 that there are basically no export quotas for many manufacturing goods in China after the entering of WTO and thus, the export amount of the FDI firms and the export amount of HKMT investment firms do not necessarily have relation and hence, the export amounts solely depend on the firms decisions. As a result, only the changes of ratio of the export value to the production value will matter and may affect the GDP growth, as model (7) and (8) show, but not the relative export shares. 36

38 6. Conclusion and Policy Implication Using the panel data across China s 14 provinces in the Eastern coastal and the central regions from 2002 to 2011, this study tries to find out the evidences of the relations between labors, capital stocks and the exports of the FDI firms as well as HKMT investment firms and the domestic GDP growths in different sectors. Basing on the pooled OLS regression results from above 9 models, it is obvious to observe that the growth of the labor force and the growth of the fixed asset capital stock are two important and essential factors to drive the positive growth of the GDP as they are two basic components of the production function; therefore, it is not hard to understand that basically all the coefficients of the lnl and lnk in above regression results are positive and significant at the 1% level, except one of the InK in the construction sector. It indicates that the growth of the fixed asset capital stock in that sector does not have statistical relation with the growth of the GDP and in another words; the input increase of the capital will not lead to output growth in the construction sector. As for the spillover effects through the labors, the results indicate that, with the total labor employment number unchanged, both increases of the labors hired by the FDI and HKMT firms will lead to a positive growth of the GDP. Within the 2 nd sector, only the growth of the labors hired by the FDI firms to the total employment level will lead to a positive rise of the 2 nd sector s GDP, but not the ones employed by HKMT firms. Whereas the situation is totally different in the non-2 nd sector, the results reveal that, instead of the FDI firms, the portion of the labors hired by HKMT investment firms to the total employment number in non-2 nd sector has a positive relation with the GDP of that sector. 37