0
Home Economy Effects
of China‟s Outward
Foreign Direct
Investment
Rujie Wang
Abstract
The rapid rise of China‟s outward foreign direct investment (OFDI) is an emerging phenomenon worldwide, and its major motives are global market expansion, natural resource procurement and strategic asset seeking. This paper explores the home economy effects of China‟s OFDI. More specifically, the causal linkages from market-seeking OFDI to China‟s export, from natural resource-seeking OFDI to China‟s import of natural resource, and from strategic asset-seeking OFDI to domestic R&D activities and regional gross production are theoretically analyzed and empirically examined. Supportive evidences are found for the first two home economy effects as the trade effects, but not for the last one. However, Granger Causality tests and Arellano-Bover estimators confirm that domestic R&D activities and regional GDP greatly benefit from the local aggregate OFDI stock. Therefore, in general the positive home economy effects are verified in the current research.
1
Contents
I. Introduction_____________________________________________________________________2
II. Home Economy Effect____________________________________________________________4 2.1Home Economy Effects of developed countries‟ OFDI ____________________________4 2.1.1 Trade Effect_____________________________________________________4 2.1.2 Technology Effect________________________________________________5 2.2 Home Economy Effects of China‟s OFDI______________________________________ 6
2.2.1 Distinctive Characteristics of China‟s OFDI____________________________6 2.2.2 Home Economy Effects of China‟s OFDI______________________________9 2.2.2.1 Trade Effects___________________________________________ 9 2.2.2.2 R&D Spillover Effect ___________________________________13
III. Data and Model specification ____________________________________________________15 3.1 Data___________________________________________________________________15 3.2 Model Specification ______________________________________________________16 3.2.1 Trade Effects___________________________________________________ 16 3.2.2 R&D Spillover Effect ____________________________________________18 IV. Empirical Results ______________________________________________________________20 4.1 Trade Effect____________________________________________________________ 20 4.2 R&D Spillover Effect_____________________________________________________24 4.2.1 Reverse Causality _______________________________________________24 4.2.2 Arellano-BoverEstimators_________________________________________25 V. Conclusion ____________________________________________________________________ 27 Appendix ________________________________________________________________________29 Reference ________________________________________________________________________30
LIST OF FIGURES AND TABLES
2
Introduction
Home economy effects of outward foreign direct investment (OFDI) has been an interesting issue for economists and policy makers in developed economies over decades, and developed countries have served as a major source of FDI as tradition. Recently, upward trends in relative size and importance of OFDI in emerging and developing economies have been observed, indicating an important change. The current research implores the home economy effect of China‟s OFDI, since the rapid emergence of Chinese multinational entrepreneurs (MNEs) is a quite representative case, which is expected to accelerate in the coming years. (Zhang, 2009) In 2009, China‟s OFDI flow by value was the largest among the developing countries and the sixth all over the world with its share over 4 percent1.
A substantial amount of literatures has emerged to evaluate the effects in developed countries. In the survey of Navaretti et al. (2006), overseas investment of multinational firms in general is influential in home plant output and sales, employment, input composition and R&D activities at firm level, and in trade and domestic economic growth at country level. However, the home economy effects of OFDI is inconclusive, varying from case to case and greatly affected by different types of OFDI, such as by whether home and foreign operations are horizontally or vertically integrated, and by whether the destination is a developing or developed country. (Also see Kokko, 2006; Denzer, 2011) Empirics have revealed that developing-country multinationals deny the conventional wisdom that engaging in large scale capital export rather than capital import would be harmful for the economy and disadvantaged for the MNEs. Samsung from Korea and Acer from Taiwan are the examples of successful MNEs from developing world. (Gammeltoft, 2008) But it remains unclear and puzzling as to whether OFDI home economy effects in developing economies and specifically in China could mirror those in developed economies.
Without theoretical analyses and empirical evidence, no one could guarantee OFDI has been always influential in upgrading aggregate economic performance in the case of China. The most straightforward and preliminary approach is by viewing the rent trends of the selected macroeconomic aggregates. In Figure 1, it is shown that aggregate output, total export and total import volumes of China experienced markedly rapid growth, accompanied with an accelerating increase in OFDI flow and stock during the same period. Positive correlations among China‟s OFDI, GDP and trade seem to be valid. In relevant literatures, Kokko (2006) contends that the effects will be similar but the magnitudes of effects on either trade or on economic structure are relatively small for developing countries. Cai (1999) argues that Chinese OFDI can directly promote the export of Chinese capital goods and
3
related materials to developing countries, and Chinese foreign affiliates play an important role in collecting up-to-date information on the international market and effectively circumvent trade barriers, which is crucial for expanding export share in many developed countries. Yeung and Liu (2008) predict that China‟s OFDI has potentially major impacts on the country‟s economy, since foreign technology development helps to boost China‟s efforts at industrial upgrading and increasing the competiveness of its higher-value-added manufacturing activities.
China‟s trade, domestic technology growth, and consequently the whole economy might benefit from China‟s OFDI, according the theoretical arguments in literatures. However, few literatures provide relevant and solid empirical evidences for China, and the effects are always mixed with opposite signs theoretically and are too complicated to jump into any conclusion by merely eyeballing those macroeconomic indicators.
The main purpose of this paper is to look for empirical evidences of the home economy effects of China‟s OFDI and this study contributes to the literatures profoundly. Firstly, two distinctive features are identified: the three motives that are specifically classify China‟s OFDI projects, and the special defining features of China‟s vertical OFDI. Then basing on the features, causal linkages from OFDI of different motives to trade at national level and to R&D spillovers at regional level are hypothesized and tested. The most important data resource is Ministry of Commerce People‟s Republic of China (MOFCOM) online registration system for Chinese MNEs that engage in OFDI2, where basic information including corporate name, original province, destination country, business scope and registration date, is listed for over 15000 OFDI projects. Using the information provided by MOFCOM and the MNEs‟ websites, vertical and horizontal OFDI, as well as its motive, for each project is broadly and roughly decided. Positive trade effects are validated by solid and 2 http://wszw.hzs.mofcom.gov.cn/fecp/fem/corp/fem_cert_stat_view_list.jsp GDP (left) Export (left) Import (left) OFDI flow (right) OFDI stock (right) 0 50 100 150 200 250 300 350 400 450 500 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 2002 2003 2004 2005 2006 2007 2008 2009 Year
Figure 1 Trends in selected macroeconomic aggregates in China, 2002-2009: GDP, PPC, billion constant 2005 international USD$; OFDI flow and stock, export and Import, billion USD$ at current prices.
4
robust empirical results, while R&D spillover effects are not. In addition, horizontal OFDI is revealed to have a greater promotive impact on China‟s trade than vertical OFDI does. Finally, specific and convincing policy implications are made for China‟s long march of internationalization and regional economic growth.
The structure of the paper is as follows. Section 2 firstly reviews standard literatures about the home economy effects in developed countries and then specifically discusses the two distinctive features of China‟s OFDI, basing on which relevant home economy effects are consolidated with the three motives of China‟s OFDI and formalized as main hypotheses for further examination; Data and model specification are described in section 3; section 4 discusses empirical results while section 5 concludes with relevant policy implication.
2. Home Economy Effects
In this section, standard literatures of the home economy effects of OFDI in developed economies are reviewed firstly; then attentions are paid to the distinctive features of OFDI of China and other developing economies. Due to the fact that developing countries are usually scarce on capital and abundant on labour as the traditional recipients of FDI, and the fact that the special conditions of the home market do not seem to play an important role in generating advantages that the advanced country MNEs exploit, the features and home economy effects for developing economies would greatly differ from those from developed economies. (Euh and Min, 1986; Gammeltoft, 2008)
2.1 Home Economy Effects of developed countries’ OFDI
By and large, two dimensions of effects on home economy in developed countries from OFDI have been intensively discussed. The first is the impact on domestic production, depending on the interaction between OFDI and trade, if more specifically, between OFDI and domestic trade-oriented production of intermediate and final goods. The second is the impact on domestic technological growth and R&D activities. (Kokko, 2006; Navaretti et al., 2006) For the current research, trade effects and technology effect are also the basic two dimensions under investigation for home economy effects of China‟s OFDI.
2.1.1 Trade Effects
5
On the one hand, horizontal OFDI, according to Navaretti et al. (2006), is identified when a firm split geographically and duplicates just a subset of its activities in foreign market. Usually horizontal OFDI suffers from returns to scale foregone at plant level and increased transportation costs if intermediate goods are needed to export from home plant to foreign plants for further processing, and benefits from convenient access to foreign market. The main motive of horizontal OFDI is consequently argued to be market expansion. Theoretically, under horizontal OFDI, products in home and foreign plants substitutes each other thus export will be replaced; however, if foreign plants need input or other complementary goods produced at home, domestic output of intermediate goods for export will increase. (Rob and Vettas, 2003; Blonigen, 2005) Empirically no clear-cutting evidences have been found supportive for the complementary or substitutive relation between horizontal OFDI and export.
On the other hand, vertical OFDI is defined when a firm split its activities by function; for example, a particular part, especially the downstream production part, in the case of MNEs from developed economies, is moved to a separate foreign plant. The main benefits of vertical OFDI are the reduced production costs by exploiting local factors, while the costs involve reduced economies of integration of production process. Hence, the motive of vertical OFDI in standard literatures is largely efficiency seeking for relatively low costs of production. As to the trade effect of vertical OFDI, a complimentary relation is expected due to production in foreign plants and export of intermediate goods from home plants is unambiguously predicted and commonly observed in empirical studies; on the other hand, import of final goods from the host country back to home will increase, but import of intermediate goods might decline correspondingly.
The relationship between OFDI and trade largely remain an empirical question and is widely tested. For example, Lundan (2007) confirms the positive relation between OFDI and export in most empirical studies of developed countries. For instance, cases studies of US have lent great support to the complementarity relationship. Lipsey and Weiss (1981, 1984) have found that US OFDI had a positive effect on export, of the parent company, of the industry and of the whole nation; also, it is verified that OFDI is a method for oligopolistic US-owned firms to advance its position in host country markets. Significant complementarity is also found in Singapore, but the trade balance with regard to manufactured goods is hardly affected by OFDI. (Ellingsen et al., 2006) The trade effect remains an empirical issue, and seldom empirical studies distinguish horizontal and vertical OFDI and examine their respective impacts on export of intermediate and final goods.
6
The technology effect addressed in the literatures normally explores whether OFDI from developed economies has impacts on technological upgrading and R&D activities at host economies; however, theoretically the reverse technology enhancement effect on home operation is also worth noticing. Such an effect is commonly discussed along with the impact of changing composition of home employment between skilled and unskilled labour, and the impact of knowledge flows through MNEs to the home country. (Kokko, 2006; Navaretti et al., 2006)
Firstly, it is argued that “White-collar” employees are favored at the expense of “blue-collar” workers, because the relocation of activities may be lead to concentration of skilled labour intensive activities and increase the relative demand for skilled labour and also stimulate R&D activities at home base. In this way, industry structure of the home economy is significantly impacted. In addition, most of vertical investments are found to be responsible for the effect, since in general terms headquarters activities are more skilled intensive than production.
Secondly, by investing overseas the MNEs are believed to simulate knowledge flows from host countries, which are supposed to be beneficial to the home countries. Strategic asset-seeking FDI is the most relevant type for North-North investment. (Dunning and Narula, 1995) For example, according to Falzoni and Grasseni (2005), investing in developed countries is found to be quite beneficial to the more advanced MNEs from Italy in term of productivity; and it might be due to their relatively high abilities in learning and absorbing technologies and skills used in local host country firms.
2.2 Home Economy Effects of China’s OFDI 2.2.1 Distinctive Characteristics of China’s OFDI
Before further discussion, a better understanding towards the specific characteristics of FDI from China would be of great necessity to unravel the home economy effects. In general, two characteristics of China‟s OFDI distinctive from that of developed countries‟ OFDI can be observed and summarized from the previous literature, i.e., the three motives of China‟s OFDI and the defining features of China‟s vertical OFDI, which are consequently reflected in the discussion of home economy effects of China‟s OFDI.
Three Motives of China’s OFDI
7
for the three motives will be explained in association with the home economy effects in detail afterwards; because more importantly, such a classification deviates from that in Dunning (1977) which comprises market-seeking, efficiency-seeking, and resource-seeking, which is stressed as follows.
First, the efficiency (cost reduction) seeking OFDI, is not particularly considered here. The reason is that efficiency-seeking OFDI will occur when MNEs seek low-cost production factors; however, Chinese MNEs have already enjoyed a comparative advantage in low-cost labour and labour-intensive labour production at home operation. (Buckley et al., 2007) Of course for a quantity of MNEs from other developing economies (such as South Korea and Taiwan), efficiency seeking is a dominating motive of foreign investment in smaller and poorer countries, because they have acquired technology from industrial countries and adapted even better in those countries than their developed-countries rivals. (Wells, 1981; Euh and Min, 1986) In China, this type of OFDI only takes up a trivial share of the total value, most of which flow into small Asian countries, such as Thailand, Vietnam, Laos and Cambodia, mainly in the industries of clothing and textile. It is reasonable to predict that this type of OFDI will noticeably increase in ASEAN countries with the uprising production prices in China. (Wu and Sia, 2002; UNCTAD, 2003)
Second, resource seeking motive, as defined by Dunning (1977), is further divided into natural resource-seeking and strategic asset-seeking for China‟s OFDI. Although both of natural resource and strategic asset could be regarded as foreign resource, seeking different kinds of resource will impact different aspects of China‟s economy; the former is supposed to have a major impact on China‟s import of natural resource, while the latter might exert influences on domestic R&D and productivity. Since this paper focuses on the home economy effects, such a division is important and even necessary for discussion.
China’s Vertical OFDI
The second specific aspect involves the defining features of vertical OFDI that is driven by either foreign market or strategic assets, since horizontal and vertical OFDI need to be further divided under each of the above-mentioned three types. Natural resource-seeking vertical OFDI won‟t be discussed here but will be specified in the next section.
8
in China is thoroughly different from that in developed countries, revealed by the information provided by MOFCOM. Firstly, due to the comparative advantages in labour-intensive production, efficiency-seeking OFDI that aims at reducing production costs is still not frequently observed. Secondly and more importantly, a majority of Chinese MNEs choose to maintain the production section at home and shift upstream activities to foreign plants, which serve as export-platforms or R&D centers, or both, depending on the underlying motives.
More specifically, if market seeking is the major concern, foreign affiliates are established to facilitate export from home plants of final goods that would be sold locally if foreign plants don‟t engage in any production activities, or export of intermediate goods for further processing in foreign plants. For this type, the MNEs provide business consulting and after-sale services, advertising and developing local customer networks. It has been surveyed and estimated that 47 percent of Chinese executives have the ambition to expand overseas markets, revealed by the questionnaire by the Ministry of Foreign Trade and Economic Cooperation. (Li, 2000) A good example is provided by Henley et al. (2008) about OFDI from China, India and South Africa in sub-Saharan Africa (SSA); and it is found that the investments in SSA are mainly driven by market seeking. Figure A in Appendix compares the sectoral distribution of China‟s OFDI by percent of total value between year 2004 and 2009. The share of Sector of Wholesale and Retail Trades drops mildly but that of Manufacturing plummets, but Sector of Leasing and Business Services expands from 14 percent to 36 of the total value. Such a noticeable boost of OFDI in service sector generally involves an increasing number of small scale investments in trade-supporting activates by Chinese trading companies.
9
by the recently emerging empirical studies, such as Chinese MNES investing in Germany (Schüler-Zhou and Schüller, 2009), the United Kingdom (Liu and Tian, 2008) and Italy (Pietrobelli, et al. 2010)
Thus, the distinctiveness lies in that it is vertical OFDI, but it is not efficiency seeking, but either market seeking or strategic asset seeking. Such a feature can be also illustrated in term of costs and benefits of horizontal and vertical OFDI of developed countries and China, as summarized in Table 1 where the first two columns are taken from Navaretti et al. (Table 2.2, 2006) and the final column is basing on the previous discussion. For China‟s vertical OFDI, main costs are related to disintegration of business at firm level, such costs of transportation and tariff barriers, while benefits are from foreign market expansion and increased economies of scales of production at home base for market-seeking OFDI, and from obtained foreign strategic assets. The distinction is important because the home economy effect from market-seeking vertical OFDI needs to focus on export of final and intermediate goods.
Table 1 Benefits and costs to firm of horizontal and vertical OFDI Horizontal (developed
countries and China: market-seeking)
Vertical (developed countries:
efficiency-seeking)
Vertical (China: market-seeking or strategic
asset-seeking) Costs Returns to scale
foregone
Disintegration costs
Disintegration costs Disintegration costs Benefits
Market access Factor cost saving
Market access
Increased returns to scale
Foreign strategic assets 2.2.2 Home Economy Effects of China’s OFDI
According to literatures reviewed next, the impacts on trade and on domestic technology development are the essential two aspects of home economy effects of China‟s OFDI, namely trade effects and R&D spillover effects. (Cai, 1999; Yeung and Liu, 2008)
2.2.2.1 Trade Effects
Trade effects include the effects on export of intermediate goods and final goods, which could be from either market-seeking horizontal and vertical OFDI, and the effect on import of natural resource from natural resource-seeking OFDI.
10
significant correlation coefficient, 0.242, between OFDI and export for developing countries, both of which are measured as percent of GDP. Lecraw (1993) finds that the performance of export-enhancing Indonesian MNEs had improved substantially in terms of many aspects, including export, and that those MNEs became export-intensive, low-cost producers of high quality products. But in this paper trade effects on export will be distinguished between export of intermediate and finals goods, and between from horizontal and vertical OFDI.
Market-seeking Horizontal OFDI
In China, the market-seeking horizontal OFDI is defined when the foreign business scopes include production, marketing and sale functions in manufacturing sectors, or cover the entire business scopes in service industries, utilizing local factors and serving local customers.
Theoretically, that horizontal investment is positively associated with export of intermediate goods to target markets; and that‟s because intermediate goods are needed as input to produce the final goods. In addition, the relevant effect on export of final goods might be negative due to the substitutionary relation in between, and the empirical results are found mixed with negative and positive signs. (Kokko, 2006; Navaretti et al., 2006) It might due to the aggregate export measure applied in estimation. Therefore, it is expected market-seeking horizontal OFDI is positively correlated with export of intermediate goods and negatively correlated with export of final goods. Hereby the hypotheses are derived as:
H1(a): China’s export of intermediate goods increases with the market-seeking horizontal OFDI;
H1(b): China’s export of final good decreases with the market-seeking horizontal OFDI.
Market-seeking Vertical OFDI
For the market-seeking vertical OFDI of China, the defining features have been explained in detail above. In developed countries vertical OFDI is traditionally predicted to be highly correlated with export of intermediate goods or import of final goods.
11
especially when investing in developed countries. For instance, Hong Kong is the primary hosting market of China‟ OFDI in 2009; and regarding its sectoral distribution, Wholesale and Retailing is the third largest that investment went to in 20093, accounting for 13.7 percent of total flows; and it is obvious that most of MENs of this industry sector establish affiliate in Hong Kong to facilitate export of final products produced in China. At the same year, Hong Kong is also the second largest export market (13.8 percent of the total export) of China. Therefore, the relevant economy effect is predicted from market-seeking OFDI on export of both intermediate and final goods; and the hypotheses take the form as:
H2(a): China’s export of intermediate goods increases with the market-seeking vertical OFDI;
H2(b): China’s export of final goods increases with the market-seeking vertical OFDI. Natural Resource-seeking OFDI
Natural resource seeking is another important motive of China‟s OFDI, driven by an almost insatiable demand for raw material and other primary goods, and stimulating the MNEs to invest in natural-resource oriented projects in resource-rich countries, especially in East and Central Asia and Africa. In 2009, mining industry takes up 23.6 percent of total values of China‟s OFDI, amounting to over 13.3 billion USD, and Australia with its abundant natural resource is a major target. In 2009, over 2.44 billion USD of China‟s OFDI goes to Australia as the second largest destination by value of China‟s OFDI, 85.9 percent of which is attributed to mining industry. Meanwhile, Australia is topped in ore import markets of China, the second largest country where China imported aluminum, and the sixth in the copper markets; respectively, the shares are 35 percent, 27 percent and 5.3 percent of the total value in 20074.
Theoretically, natural resource-seeking OFDI would have a predominant effect on home economy, and the most direct result is the increase in import of natural resource from host countries. According to Ozawa (1992) oversea investments by MENs that mainly engages in natural resource extraction (or labour-intensive, low-skill manufacturing) happen at the initial stage of the evolutionary path of industrial upgrading and development. Similarly, Porter (1990) observes an interesting pattern of competitive advantage in a nation‟s firms with four distinct stages. Most developing countries are in the first two stages: factor driven and investment driven, whilst the other two advanced stages are innovation-driven and wealth-driven. In general, the changing pattern of comparative advantages is contended to play an
3 Data source: MOFCOM (2009)
12
important role in economic growth and transformation. Specifically, the transition from stage one to stage two would generate outward investments either in resource extraction abroad if the economy is natural resource-scarce, or towards lower-wage countries in labour-intensive manufacturing.
Exploitation of natural resource all over the world so as to guarantee a stable supply would be regarded as an important strategy to support China‟s rapid growth. That‟s because one of the characteristics of China‟s industrialization is known as “wage, tech, low-productivity manufacturing at the low end of global supply chain on the one hand, and high investment, high resource consumption and pollution, high exploitation, and high degree of foreign capital/trade dependency on the other” (Tarmidi and Gammeltoft, 2008). Besides, although China is well-endowed with natural resources, its per capita level is relatively low, especially for iron ore, aluminum, copper, timber and fish. (Zhan, 1995) Also, China‟s economic growth has exceeded its domestic natural resource base, thus import of fuel and minerals plays a critical role in domestic production. (Butts and Bankus, 2009) Imports of natural resource have been observed to surge with China‟s resource-seeking OFDI. Friedberg (2006) notices that at an explosive pace, China‟s dependence on imported natural resources is growing; and this trend is most noticeable in raw oil. Zweig et al. (2005) also find that in 2004 China is the world‟s second largest importer and it alone accounted for 31 percent of global growth in oil demand. Consequently, it is predicted with no complication that China‟s OFDI driven by natural resource extraction would increase the relevant imports.
13
H3(a): China’s imports of natural resource increases with the natural resource-seeking OFDI in the target markets;
H3(b): The import effect of natural resource-seeking horizontal OFDI is greater than that of vertical OFDI.
2.2.2.2 R&D Spillover Effect
The R&D spillover effect refers to the effect on domestic R&D activeness and regional gross production; and strategic asset-seeking OFDI are largely responsible for the spillover effects, because OFDI in this context serves as a channel for MNEs to exploit market intelligence, technological know-how, management expertise, and reputation for upgrading their competitiveness and being established in a prestigious market. In China, the strategic asset-seeking vertical OFDI has been defined above, while the horizontal type largely refers to those acquisition projects for brands and complementary assets, or to those investment holding companies that own and control outstanding stock. Lenovo‟s acquisition of IBM PC business in 2005 is the very example, which affectively helps expand its key brands and trade names into foreign markets.
By operating in foreign countries, the MNEs learn from local counterparts, and tap the advanced knowledge of technology-intensive production. In this way, OFDI, as a channel for transferring and diffusing foreign technology to other firms in home country, has been contended to be one of the key determinants of domestic R&D development and home economic growth. (Globerman, Kokko, and Sjöholm, 2000; Herzer, 2009) It is believed that although outward orientation alone is not a sufficient condition for rapid growth, it does create the financial benefits generated will largely belong to home operation; hence MNEs will be able to invest more in marketing, R&D and other fixed costs that are concentrated to the home country (Kokko, 2006) Therefore the primer effect should arise from the increased productivity of the MNC itself. Also, the linkages between MNEs and other firms in home country yield similar effects through spillovers, and the linkages are stronger the more MNCs are vertically integrated in home operations. Consequently, domestic R&D activeness and then gross production may benefit from the foreign activities of MNEs. (Blomstrom and Kokko, 1998)
14
highest productivity level and largest R&D stock among others, the causality in between is likely to be two-way. On the one hand, local R&D and production might benefit greatly from technology spillovers through MNEs‟ foreign activities, and it is one of the home economy effects of main interest in this section; on the other hand, a high productivity level on average in the region could grant local firms with large R&D capabilities and high productivity levels when competing in foreign markets. Consequently, the R&D spillover effect of OFDI on domestic economy should be estimated with special care, in term of the causality issue.
Most empirical studies have paid attention to this issue in developed economies. Herzer (2008) finds for 14 developed countries that outward FDI has positive long-run effects on domestic output. Desai et al. (2009) claim that in the case of American manufacturing firms 10% greater foreign investment is associated with 2.6% greater domestic level and 3.7%. In the case of China, Huang and Wang (2009) have found supportive empirical evidences for the existence of inverse technological spillover effect in term of the number of patent application in China from OFDI. Wei and Ling (2008) have the similar findings for the OFDI‟s inverse technology effect on domestic economy in China, but in term of total factor productivity.
Accordingly, it is hypothesized that strategic-seeking OFDI from China positively affect domestic R&D activeness and GDP through technology spillovers. Additionally, the discussion above largely points to an unambiguous opinion that strategic-seeking vertical OFDI is supposed to have a greater influence than the horizontal type does. In addition, since China is the only home economy under investigation, it would be more interesting and reasonable to examine Chinese provincial economies and the R&D spillover effect at regional level. The final main hypotheses and one other hypothesis are set as:
H4(a):China’s regional R&D activities are stimulated by the strategic asset-seeking OFDI;
15 Figure 2 Main hypotheses
Figure 2 provides an overview of the main hypotheses stated above, which also constitutes the essential empirical questions to be tested next. It is worth noticing that there are two other relevant hypotheses, H3(b) and H4(c) about the relative size of effects of horizontal versus vertical OFDI, driven by natural resource and foreign strategic assets, separately.
Dada and Model Specification
3.1 Data
The data sample for the empirical analysis consists of two balanced data sets; the first is for trade effects including 160 host countries from the year 2003 to 2009, and the second for R&D spillover effect, covering 30 Chinese provinces during the same period.
The vital step is the measurement of China‟s OFDI for each type. Due to the lack of micro firm level data, China‟s OFDI with different motives can only be indicated by the number of active firms in each target market, using the information provided by MOFCOM online registration system. Market-seeking, natural resource-seeking and strategic asset-seeking OFDI are identified broadly and roughly. Furthermore, under each type, whether the project is horizontal or vertical OFDI can be decided, according to the above-mentioned defining features. In addition, the system also lists each project‟s original province. Consequently, both at national and provincial level, the six types of OFDI, i.e. market-seeking horizontal and vertical OFDI (MA_H, MA_V), natural resource-seeking total, horizontal and vertical OFDI (NR, NR_H, NR_V), and strategic asset-seeking total, horizontal and vertical
China' OFDI
Market-seeking horizontal OFDI
H1(a): positive Export of Intermediate Goods H 1(b):
negative Export of Final Goods
Market-seeking vertical OFDI
H2(a): positve Export of Intermeidate Goods
H2(b): positive Export of Final Goods Natural
Resource-seeking OFDI H3(a): positive
Import of Natural Resource
Strategic Asset-seeking OFDI
H4(a): positve Regional R&D activities
16
OFDI (SA, SA_H, SA_V), are counted. In this paper only MA_H, MA_V, NR, NR_H and NR_V at national level, and SA, SA_H and SA_V at provincial level are utilized in estimation.
Importantly, projects with multiple motives are commonly observed. For instance, if the MNE aims at not only selling the firm products in the host country but also moving its R&D center to the foreign plant for developing new products, which is most frequently observed, it is both market seeking and strategic asset seeking. Therefore, double counting takes place and the summed number exceeds the total number. The distribution of China‟s OFDI by number for each type, summing from the year 2003 to 2009, is demonstrated in Figure 2. It is observed that market-seeking OFDI (including MA_H and MA_V) account for a lion‟s share of total number of China‟s OFDI project, and strategic asset-seeking vertical OFDI is the third largest type in term of project number.
3.2 Model Specification
3.2.1 Trade Effects
With respect to the trade effects, a gravity model for trade can fit well and be adopted to examine the relevant impacts from China‟s OFDI. The gravity equation has been recognized as a popular instrument for its consistent empirical success in elucidating foreign trade flows. According to this model, exports from country i to country j can be explained by their national incomes, and geographical distances and a set of control variables. The following specification is commonly used as in Tinbergen (1962):
𝑃𝑋𝑖𝑗 = 𝛽0(𝑌𝑖)𝛽1(𝑌
𝑗)𝛽2(𝐷𝑖𝑗)𝛽3(𝐴𝑖𝑗)𝛽4𝜇𝑖𝑗 (1) where 𝑃𝑋𝑖𝑗 is the value of bilateral trade flow between country 𝑖 and 𝑗, 𝑌𝑖 and 𝑌𝑗 refer to the nominal GDP in the two countries, 𝐷𝑖𝑗 denotes distance between the economic centers in 𝑖
0 2000 4000 6000 8000 10000 12000 14000 16000
MA_H MA_V NR_H NR_V SA_H SA_V OFDI by project number
17
and j, and 𝐴𝑖𝑗 controls for other factors that either encouraging or resisting the trade flows in between.
The Baseline Gravity Model
In this paper country i is limited to China, 𝑌𝑖 is then captured in a constant term. Furthermore, since GDP is the product of GDP per capita and population, 𝑌𝑗 can be divided into the two explanatory variables: GDP per capita (HGDPPC) that implies consumption ability of customers in host country j, and population (HPopulation) that indicates the size of consumer market in j. 𝐷𝑖𝑗 then becomes the geographical distance (𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑗) between capital city of country j and Beijing. The baseline model follows that in Ellingsen et al. (2006) and is further modified. The number of China‟s OFDI for each type (indicated by a general term,
OFDI_Project) is added as the variable of major interest. Total values of OFDI (OFDI) in
country j and of inward FDI (IFDI) from country j are also included to estimate the effects from capital movement between two countries. In addition, whether the destination is a developed or developing country is controlled to account for the host country‟s economy characteristic; therefore, the dummy variable Developed is added to the model, which takes the value 1 if the destination is developed and 0 otherwise, according to the World Bank standard. Typically, a long-linear equation is utilized in estimation. Finally, the baseline gravity model takes the form as follows:
𝐿𝑛𝑇𝑟𝑎𝑑𝑒𝑗 ,𝑡 =
𝛽0+ 𝛽1𝑂𝐹𝐷𝐼_𝑃𝑟𝑜𝑗𝑒𝑐𝑡𝑗 ,𝑡−1+ 𝛽2𝑙𝑛𝐻𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛𝑗 ,𝑡−1+ 𝛽3𝑙𝑛 𝐻𝐺𝐷𝑃𝑃𝐶𝑗 ,𝑡−1+ 𝛽4𝑙𝑛𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑗 + +𝛽5ln𝑂𝐹𝐷𝐼𝑗 ,𝑡−1+
𝛽6𝑙𝑛𝐼𝐹𝐷𝐼𝑗 ,𝑡−1+ 𝛽7𝐷𝑒𝑣𝑒𝑙𝑜𝑝𝑒𝑑 + 𝜀𝑗 ,𝑡 (2)
18
The Fixed Effect Model
According to Baldwin and Taglioni (2006), there are several common errors in the literatures when estimating a gravity equation. In particular, the omitted variables bias will be caused due to the ignorance of the “gravitational un-constant” which is in regression residual and correlated with other independent variables in the model, especially for panel data set. It is suggested that the inclusion of pair country dummies might be more helpful than country dummies in correcting the bias, which, however, is not an option in this paper. Therefore, the fixed effect model is applied to rule out the country-specific and period-specific terms that might be correlated with existing explanatory variables. Meanwhile, the time-invariant variable, distance, is omitted due to the colinearity problem.
Furthermore, in the current version of gravity models the determinants of OFDI might also be the economic mass and geographical distance, which might greatly overlap with those of trade flows. The inclusion of OFDI proxies in a trade gravity model might seriously bias the estimates. So as to avoid the endogeneity issue in estimation, an alternative method is utilized to estimate the three gravity models, following the approach in Brenton et al. (1999) and in Ellingsen et al. (2006).
First, an OFDI gravity model of the basic type is estimated:
𝑙𝑛 𝑂𝐹𝐷𝐼𝑗 ,𝑡 = 𝛽0+ 𝛽1𝑙𝑛𝐻𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛𝑗 ,𝑡+ 𝛽2𝑙𝑛 𝐻𝐺𝐷𝑃𝑃𝐶𝑗 ,𝑡 + 𝛽3𝑙𝑛𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒𝑗 + 𝜀𝑗 ,𝑡 (3) Then the residuals can be estimated from Equation (3) and replace the variable 𝑙𝑛𝑂𝐹𝐷𝐼𝑛,𝑡−1 in Equations (2). In this way, the determinants of China‟s OFDI, other than host country‟s population, GDP per capita and its distance to Beijing, are instrumented and controlled, which is supposed to reduce the estimation bias.
3.2.2 R&D Spillover Effect
Another two models are specified to explore the R&D spillover effect of China‟s OFDI on regional R&D activities and GDP. The former adopts the econometric model in Huang and Wang (2009), using the number of patent application (PA) as the proxy of domestic R&D trend. The latter follows the cross-country model in Barro (2001), Bleaney and Castilleja-Vargas (2007) and Herzer (2009), and it is specially adapted for the current research.
19
regional R&D activities and regional GDP; to this end, OFDI_Project, included as a proxy for project number, measures either total, or horizontal or vertical OFDI, depending on the hypothesis under investigation. Furthermore, the control variables include the lagged values of each province‟s OFDI (OFDI), inward foreign investment (FI), GDP per capita (GDPPC), and the ratio of government consumption to GDP (Gi), according to Levine and Renelt (1992) and Herzer (2009). Likewise, one year lag between the dependent and independent variables is needed. The models take the form as in equation (4) and (5), where n denotes the province where the MNE originally locates:
𝑙𝑛𝑃𝐴𝑛,𝑡 = 𝛽0+ 𝛽1𝑂𝐹𝐷𝐼_𝑃𝑟𝑜𝑗𝑒𝑐𝑡𝑛,𝑡−1+𝛽2𝑙𝑛𝑂𝐹𝐷𝐼𝑛,𝑡−1+ 𝛽3𝑙𝑛𝐺𝑛,𝑡−1+
𝛽4𝑙𝑛𝐹𝐼𝑛,𝑡−1+ 𝛽5𝑙𝑛𝐺𝐷𝑃𝑃𝐶𝑛,𝑡−1+ 𝜀𝑛,𝑡 (4) 𝑙𝑛𝐺𝐷𝑃𝑛,𝑡 = 𝛽0+ 𝛽1𝑂𝐹𝐷𝐼_𝑃𝑟𝑜𝑗𝑒𝑐𝑡𝑛,𝑡−1+𝛽2𝑙𝑛𝑂𝐹𝐷𝐼𝑛,𝑡−1+ 𝛽3𝑙𝑛𝐺𝑛,𝑡−1+
𝛽4𝑙𝑛𝐹𝐼𝑛,𝑡−1+ 𝛽5𝑙𝑛𝐺𝐷𝑃𝑃𝐶𝑛,𝑡−1+ 𝜀𝑛,𝑡 (5)
Table A in Appendix provides descriptive statistics of OFDI, (a) for the market seeking and natural resource-seeking OFDI of different types, and (b) for the strategic asset-seeking OFDI of different types. Finally, Table 2 lists all the variables that are mentioned above, and summarizes their respective linkages to hypotheses in this paper, as well as the data sources.
Table 2 Variables
Main Variable Hypohesis Source
Trade (nominal value, current USD)
UNCTAD Export_I Export of intermediate goods H1(a), H2(a)
Export_F Export of final goods H1(b), H2(b)
Import_NR Import of natural resource H3(a), H3(b)
OFDI_Project (number of exisiting project) MOFCOM
Online Registration System (calculated by the author) MA_H Market-seeking horizontal OFDI H1(a), H1(b)
MA_V Market-seeking vertical OFDI H2(a), H2(b)
NR Natural resource-seeking OFDI H3(a)
NR_H, NR_V Natural resource-seeking horizontal and vertical FDI H3(b)
SA Strategic asset-seeking OFDI H4(a), H4(b)
SA_H, SA_V Strategic asset-seeking horizontal and vertical OFDI H4(c)
Other Variables Equation Source
OFDI China's national outward foreign direct investment
stock to the host country (current USD) (2)
Statistical Bulletin of China's OFDI (2009) OFDI China's provincial outward foreign direct investment
stock to the host country (current USD) (4), (5)
Hpopulation Host country popoulation (2)
World Bank
HGDPPC Host country GDP per capita (current USD) (2)
Distance Geographical distance between the capitals (2) CEPII
IFDI Inward FDI from host country (current USD) (2) Chinese
Statistic Yearbooks PA NO. Of patent application in each province (4), (5)
20 USD)
G Ratio of government consumption to GDP in each
province (4), (5)
FI Inward FDI from host country in each province
(current USD) (4), (5)
GDPPC GDP per capita in each province (current USD) (4), (5)
Develped 1, if developed country; 0, otherwise (2)
Empirical Results
In this section, the empirical results are demonstrated and discussed. Following the same structure as above, estimates for the hypotheses of trade effects and R&D spillover effect are interpreted respectively.
4.1 Trade Effects
The Baseline Gravity Model
Table 3 reports the OLS estimation results for the baseline gravity model for the first three sets of hypotheses for the trade effects, using the data ample of 160 countries that were hosting China‟s OFDI projects during 2003-2009. Each hypothesis and the corresponding dependent variable (D.V.) are listed in the first two rows for each column, according to Figure 2. Also, period fixed effect is controlled, and the country specific effect is captured by the time invariant variable, the geographical distance between the capital city of the host country and Beijing. Reading through the columns, there are three major points worthwhile to be emphasized.
21
In addition, the OFDI projects aiming at extracting natural resource are disclosed to effectively promote China‟s import of natural resource, since the lagged coefficient of NR is estimated to be 0.034, validating H3(a). Considering the relative size of the effect from horizontal and vertical OFDI and therefore comparing the relevant coefficients in Columns (6) and (7), I find that extracting foreign natural resource in the way of horizontal OFDI will be more influential than the vertical OFDI in importing natural resource back to China; in particular, the vertical OFDI type is revealed insignificantly related to the import.
Secondly, the coefficients of the three basic variables in the gravity model, i.e.
HPopulation, HGDPPC and Distance, are estimated as expected in general, and all of them
are at least statistically significantly at 5 percent level. Both population base and income level in the host country appear to be contributing factors to bilateral trade flows. If the coefficients of them in Columns (1) and (3) are compared with those in (2) and (4), it seems that with a larger consumer market (HPopulation), more export of final goods than export of intermediate goods will be stimulated; on the contrary, China‟s export of intermediate goods is mildly more responsive to the change in consumption ability (HGDPPC) in the host country than export of final goods. In addition, estimates of HPopulation in Columns (5)-(7) for China‟s import of natural resource are found extremely large in size, and that might be due to the high correlation between the host country‟s population and China‟s import of natural resource from it, 0.64. It is sensible because for instance, Russian and Indonesia are the two countries which are two of the most important partners for importing natural resource in China, and they are also known for their large population bases. Moreover, geographical distance is shown to be a more discouraging factor for exporting intermediate goods and importing natural resource than for exporting final goods.
22
Table 3 Empirical Results of The Baseline Gravity Model
(1) (2) (3) (4) (5) (6) (7)
. H1(a) H1(b) H2(a) H2(b) H3(a) H3(b) H3(b)
I.V. D.V. Export_I Export_F Export_I Export_F Import_NR Import_NR Import_NR 𝑙𝑛𝐻𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛(−1) 0.637*** 0.719*** 0.642*** 0.723*** 2.126*** 2.118*** 2.133*** (0.048) (0.037) (0.047) (0.037) (0.093) (0.093) (0.093) 𝑙𝑛 𝐻𝐺𝐷𝑃𝑃𝐶(−1) 0.509*** 0.481*** 0.514*** 0.485*** 0.768*** 0.782*** 0.766*** (0.053) (0.061) (0.078) (0.061) (0.152) (0.151) (0.153) 𝑙𝑛𝐷𝑖𝑠𝑡𝑎𝑛𝑐𝑒 -0.789*** -0.281*** -0.839*** -0.329*** -0.733** -0.683** -0.786** (0.158) (0.123) (0.155) (0.121) (0.307) (0.308) (0.305) 𝑙𝑛𝐼𝐹𝐷𝐼(−1) 0.083*** 0.066*** 0.083*** 0.067*** -0.023 -0.024 -0.022 (0.013) (0.010) (0.013) (0.010) (0.025) (0.025) (0.025) ln𝑂𝐹𝐷𝐼(−1) 0.097*** 0.073*** 0.098*** 0.075*** 0.190*** 0.190**** 0.192*** (0.018) (0.014) (0.018) (0.014) (0.035) (0.035) (0.035) 𝑀𝐴_𝐻(−1) 0.011*** 0.011*** (0.004) (0.003) 𝑀𝐴_𝑉(−1) 0.005** 0.005*** (0.002) (0.001) 𝑁𝑅(−1) 0.034* (0.018) 𝑁𝑅_𝐻(−1) 0.083** (0.035) 𝑁𝑅_𝑉(−1) 0.043 (0.020) Developed 0.126 0.484** 0.073 0.434** 1.392*** 1.381*** 1.366*** (0.266) (0.208) (0.265) (0.206) (0.522) (0.520) (0.523) Constant 7.276*** 5.211*** 7.624*** 5.549*** -21.349*** -21.730*** -20.953*** (1.688) (1.315) (1.683) (1.311) (3.326) (3.336) (3.317) Observation R-squared 960 960 960 960 960 960 960 0.487 0.592 0.486 0.591 0.521 0.522 0.520 F-statistic 74.992 114.524 74.668 113.987 85.783 86.073 85.464 Note: Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.01. Period-fixed effect is specified. Dependent variables (D.V.) are in log value.
The Fixed Effect Model
To avoid the endogeneity issue in the baseline gravity model, the fixed effect with an alternative measure for China‟s OFDI stock is estimated and the results are reported in Table 4. Estimation bias is supposed to be reduced and coefficients to be estimated greater precision. The supportive evidences are clearly found by comparing the results between Table 3 and 4, and the most impressive finding is that China‟s OFDI stock are revealed to have almost 9 times larger impact on export of intermediate goods, 6 times larger impact on export of final goods and 5 times larger impact on import of natural resource than those estimated in Table 3. After controlling for the time specific effects and regional specific effects and ruling out the overlapped determinants of OFDI stock and trade, the evidences are strongly in favor of a complementarity between OFDI and trade.
23
the same for market-seeking vertical OFDI, together suggesting that export of intermediate goods is more impacted by the foreign activities of Chinese MNEs than export of final goods. Moreover, as in Column (5)-(7), the coefficients of NR, NR_H and NR_V are larger than those in Table 3, and here natural resource-seeking vertical OFDI has a statistically significant coefficient. Apart from those, the findings for hypotheses are consistent, the positive relations between MA_H and Export_I, between MA_V and Export_I and Export_F, and between NR and Import_NR are further proved, declaring the strongly promoting effects on trade of China‟s OFDI. Again, H1(b) is rejected due to the positive parameter estimated, 0.01; and H3(b) is valid for the larger estimate of NR_H than that of NR_V.
Other prominent differences generated are estimates of the two basic variables in the gravity model. Host country‟s consumer market size still serves as an attractive factor for China‟s export of final goods, but not for export of intermediate goods evinced by the insignificant and negative estimated coefficients. Against to the previous result, export of final goods will rise more than export of intermediate goods in response to an increase in host country‟s GDP per capita. These new findings make more sense than the old ones. Export of final goods is largely motivated by market expansion of Chinese exporters, Chinese MNEs, or foreign investors, making it more closely related to consumer market size and local consumption capability than export of intermediate goods. The coefficients of HPopulation for import of natural resource still exceed 1, but are reduced in size relative to those in Table 3. Another conspicuous difference is that the news estimates show no evidence that choosing to invest in a developed economy over a developing economy will make statistically significant difference for export of both intermediate and final goods.
So far, H1(a), H2(a), H2(b), H3(a) and H3(b) are proved to be valid and H1(b) is rejected, manifesting solid home economy effects on trade of China‟s OFDI. Also importantly, China‟s OFDI has been found to have a complementary relation with trade, in accordance with what has been predicted in most previous literatures and contributing a better understanding towards the home economy effects of China‟s OFDI.
Table 4 Empirical Results of the Fixed Effect Model
(1) (2) (3) (4) (5) (6) (7)
. H1(a) H1(b) H2(a) H2(b) H3(a) H3(b) H3(b)
I.V. D.V. Export_I Export_F Export_I Export_F Import_NR Import_NR Import_NR 𝑙𝑛𝐻𝑃𝑜𝑝𝑢𝑙𝑎𝑡𝑖𝑜𝑛(−1) (0.105) -0.152 0.404*** (0.079) -0.196* (0.104) 0.362*** (0.078) 1.316*** (0.244) 1.345*** (0.243) 1.269*** (0.243)
𝑙𝑛 𝐻𝐺𝐷𝑃𝑃𝐶(−1) 0.311*** 0.463*** 0.304*** 0.457*** 0.636*** 0.666*** 0.638***
(0.087) (0.056) (0.088) (0.056) (0.177) (0.176) (0.178) 𝑙𝑛𝐼𝐹𝐷𝐼(−1) 0.084*** (0.013) 0.073*** (0.011) 0.084*** (0.014) 0.073*** (0.011) (0.028) 0.005 (0.028) 0.005 (0.028) 0.005
24 (0.098) (0.072) (0.095) (0.070) (0.237) (0.236) (0.235) 𝑀𝐴_𝐻(−1) 0.011*** 0.010*** (0.003) (0.002) 𝑀𝐴_𝑉(−1) 0.005** 0.004*** (0.001) (0.001) 𝑁𝑅(−1) 0.027** (0.535) 𝑁𝑅_𝐻(−1) 0.064** (0.018) 𝑁𝑅_𝑉(−1) 0.035** (0.017) Developed 0.117 0.279 0.053 0.218 1.187** 1.196** 1.128** (0.245) (0.185) (0.245) (0.187) (0.535) (0.526) (0.542) Constant 2.044*** 2.662*** 2.072** 2.692*** -26.441*** -26.560*** -26.466*** (0.973) (0.681) (0.984) (0.688) (1.931) (1.920) (1.952) Observation R-squared 960 960 960 960 960 960 960 0.471 0.577 0.469 0.575 0.499 0.501 0.501 F-statistic 166.23 285.23 106.09 286.10 207.63 210.52 207.87 Note: Robust errors in parentheses, *** p<0.01, ** p<0.05, * p<0.01. Period-fixed effect is specified. Dependent variables (D.V.) are in log value.
4.2 R&D Spillover Effect
4.2.1 Reverse Causality
In this part, the other panel data set comprising 30 Chinese provinces in the period 2003-2009 is utilized to estimate Equation (4) to examine the R&D spillover effect of China‟s OFDI on economic growth. But, according to the previous analyses, theoretically both R&D spillover effect from OFDI and the heterogeneous firm model can explain the correlation between regional domestic production and OFDI trends, so Granger causality tests for the main variables are conducted before further examination.
Results are reported in Table 5, according to which, the mutual causality is found between PA and China‟s regional OFDI stock, significant at 5 percent level. Similarly, GDP and OFDI stock is the Granger cause for each other. In addition, the hypotheses that SA does not Granger Cause PA is not able to be rejected statistically, while the reverse causality is valid. Finally, the causality between GDP and SA is also two-way. To sum up, mutual causality is in general confirmed; what‟s more, first two columns in Table 5 imply more significant causal effects from PA to SA and to OFDI than the other way around.
Table 5 Causality Test Results
Null Hypothesis: F-Statistic
(Prob.) Null Hypothesis:
F-Statistic (Prob.) PA does not Granger Cause OFDI 0.004 GDP does not Granger Cause
OFDI_stock 0.030
OFDI does not Granger Cause GDP 0.049 OFDI_stock does not Granger
Cause GDP 0.002
25
SA does not Granger Cause PA 0.465 SA does not Granger Cause GDP 0.046
4.2.2 Arellano-Bover Estimators
Given the indicated reverse causality of the main variables of interest, the relevant home economy effect is hard to be identified empirically and estimates would be biased if only OLS is utilized. To prevent the endogeneity issue, the Arellano-Bover GMM estimation technique is applied, which uses the lags of the level variables to be instruments in the first-differenced model and uses the lags of the first-differenced variables to be instruments in the level model. Arellano and Bover (1995) find that if the autoregressive process is too persistent, the Arellano-Bover estimator will generate more consistent parameters. Arellano-Bover estimation results of Equations (4) and (5) for the effects on domestic R&D activities and regional GDP are presented in Table 5.
Primarily, no positive effects of China‟s regional strategic asset-seeking OFDI stock on regional R&D and GDP are found, contradicting to the hypothesized positive causal effects. Thus, H4(a) and H4(b) are rejected. Even if the total number is divided into horizontal and vertical types, estimates are not improved in size or significance level, as in Columns (3)-(6). By virtue of the results, H4(c) is rejected as well.
However, lagged regional OFDI stock is evidenced to have positive and statistically significant coefficients across columns. Regional R&D activeness which is measured by the number of patent application will increase by 1.7 percent if total value of regional OFDI stock increases by 1 percent in the previous year, in response to which, however, regional GDP will merely increase by 0.7 percent. The results are similar in each case across columns, so the model is considered robust in certain way. Basing on the estimates, it is clear that the general effects from China‟s total OFDI on regional R&D activities and GDP are substantiated, but the impacting channel could not be narrowed down to the strategic asset-seeking OFDI. However, with increased levels of regional R&D stock and GDP, strategic asset-seeking behaviors of the MNEs could be further stimulated. Such a linkage is quite likely, given the causality in between in Table 5. Intuitively, the firms with large R&D abilities and high productivity would like to keep their competing advantages by investing in global markets for foreign strategic asset seeking.
26
started their foreign adventures. The time span might be too short for the spillovers from strategic asset-seeking OFDI to actually take place, since most China‟s OFDI projects registered only since 2003, and it would be a long march for Chinese MNEs before the beneficial effects can be transferred and significantly impact China‟s macro economy. Such an argument is in line with the contention by Zhang (2009) that the OFDI boom in China is regarded as a natural result of China‟s economic development and its “going global” policy, and Chinese MNEs are still strongly motivated by the tradition factors (market-seeking and resource-seeking) and the Chinese factors (asset-seeking and overseas-opportunities-seeking); in addition, the first set of motivations dominates Chinese OFDI, while the other set just started to gain its importance in the recent years.
Chinese MNEs‟ failing to spillover foreign strategic assets to home economy might be attributed to the inherent insufficiency of learning or absorptive capability. Wells (1983) argues that FDI from emerging markets was largely incapable of exploiting the comparative advantages in developed countries. Although those MNEs are willing to expand business internationally but they usually adopt low-cost governance mechanisms such as network structures because of a poverty of resources. (Oviatt and McDougall, 2005) It might be an important and highly probably fact that due to the poor organizational system and institutional settings, even if Chinese MNEs have the ambition of procuring foreign strategic assets, the predicted effects are not realized. Another possible reason for the weak ability of the strategic asset-seeking OFDI projects in explaining regional GDP is that the total project number of this type is relatively small at national level; and it becomes even smaller when disaggregated into regional level. After all, only project number rather than the actual value of OFDI is accounted for estimation, so the estimates are supposed to be imprecise and could tell nothing more than a vague idea about the relation in between.
Table 6 Arellano -Bover GMM Results
27 (0.000) (0.000) SA_H(-1) -0.002 -0.000 (0.003) (0.001) SA_V(-1) -0.000 0.000 (0.001) (0.000) Constant -1.408 0.212 -1.363 -1.417 0.162 0.223 (1.052) (1.035) (0.060) (1.049) (1.058) (1.029) Observation Wald chi2(6) 210 210 210 210 210 210 4252.26 11298.62 4257.17 4249.55 11283.09 11309.68 Note: Standard errors in parentheses, *** p<0.01, ** p<0.05, * p<0.01. Dependent variables (D.V.) are in log value.
In brief, as demonstrated in Table 5, no convincing evidence is found to support the hypotheses H4(a)-H4(c). Notwithstanding, positive causal effects from OFDI stock regional R&D activities and on GDP have been clearly indicated, provoking some interesting speculation towards the source of positive effect. Although the strategic asset-seeking OFDI is reported to be irrelevant for the home economy effect, but it can by no means nullify the R&D spillover effect from other types of China‟s OFDI. Market-seeking and natural-resource seeking OFDI also have full potentials in obtaining market information, foreign high technology, operation know-how and other valuable intangible assets and transferring them to home markets and domestic firms, which remains as an issue for further discussion and examining.
Conclusion
The current research aims at providing theoretical analyses and empirical evidences to verify home economy effects of China‟s OFDI, and so far this paper has given a better understanding towards this issue and has made several contributions to the relevant literatures.
28
export and import. In addition, the distinction between horizontal and vertical OFD has yielded important results. Horizontal OFDI, motivated by either foreign market or foreign strategic asset, contributes more to China‟s export and import than vertical OFDI does. Finally, the Granger Causality tests in the last sector have provided sufficient evidences to support the mutual causality between China‟s OFDI and domestic economy performance, and the causal effect from the latter to the former might be even dominating at the current stage. Nonetheless, the Arellano–Bover estimators clearly verify the significant and positive technology spillover effects on domestic R&D activities and GDP from China‟s OFDI at regional level.
The rise of Chinese firms has been realized as an emerging political-economic force in the global economy, and it is commonly predicted that this phenomenon will be strengthened and intensified in the near future. (John and Sarah, 2003; Liu et al., 2005; Yeung and Liu, 2008) Globerman (2006) finds it difficult to conclude that OFDI should be encouraged by governments in emerging countries, but there is neither any plausible basis for them to restrict or discourage OFDI. Interestingly, Chinese central government has a clear-cutting positive attitude towards investment. Ever since implementing the „open-door‟ policy in the 1980s, cautiously but continuously, Chinese government commenced a new long march of internationalizing the domestic economic activities, or namely the “going global” strategy. Not only the government liberalized restrictive policies and regulation in late 80s, Beijing also formalized this encouragement in the 10th and 11th five-year plans, in which the “going global” policy were specifically highlighted. (Buckley et al., 2007) Given the important role that Chinese government is playing, it is the consequences and the home economy effects are the most important element to be evaluated. To this end, this paper has theoretically verifies and empirically confirms the positive home economy effects on trade and R&D activities.
29
This study has data limitations, which are commonly observed in other empirical literatures on China‟s OFDI. Firstly, the available OFDI measures, flow and stock, are provided by MOFCOM, which are in general criticized to be distorted and hide the real picture of the Chinese MNEs‟ foreign activities. Moreover, there is no official data identifying each OFDI project whether it is horizontal or vertical, and whether motivated by market, natural resource, or strategic asset. So the division and identification could only be conducted manually for over 15000 OFDI projects with other brief information provided by MOFCOM online system. Although the data set is able to illustrate the trends of China´s OFDI in a general and even vague way, it is reasonable to doubt the accuracy of the proxies. Besides, only the numbers of OFDI project for each type are obtained and utilized in estimation, it in no case implies the scale of investment for each project, which if used might tell a bit different and more precise story.
Appendix
0 5 10 15 20 25 30 35 40
Mining Transport, Storage and Post Wholesale and Retail Trades Manufacturing Leasing and Business Services Agriculture, Forestry, Animal Husbandry and Fishery Services to Households and Other Services Production and Supply of Electricity, Gas and Water Construction Information Transmission, Computer Services and Software Scientific Research, Technical Service
Real Estate 2004 2009 63% 9% 4% 3% 3% 2%1% 2% 2009 HongKong Cayman Islands Australia British virgin islands
Singapore U.S.
Canada Macao
Russia South Korea Algeria Indonesia U.K. Germany Nigeria Vietnam 0 20 40 60 80 100 2003 2004 2005 2006 2007 2008 2009 Asia Africa
Europe Latin America North America Oceania Figure A Sectoral distribution of China‟s OFDI, 2004 and 2009, percent of total value (non-financial OFDI) Source: calculated from data in Statistical Bulletin of China‟s OFDI (2009)
Note: only sectors with shares larger than 0.1 are included.
(b)
Figure B Geographical distribution of China‟s OFDI, (a) by country of China‟s OFDI, 2009, percent of total value: (b) by continent, 2003-2009, percent of total value
Source: calculated from data in Statistical Bulletin of China‟s OFDI (2009) Note: only host countries with shares larger than 0.1 are included in (a).
