Horizontal and vertical spillover effects of foreign direct investment in Chinese manufacturing

The Authors

Chengqi Wang, Nottingham University Business School, University of Nottingham, Nottingham, UK

Zhongxiu Zhao, School of International Economics and Trade, University of International Business and Economics, Beijing, People's Republic of China

Abstract

Purpose – The purpose of this paper is to examine foreign direct investment (FDI) spillovers accruing to Chinese local industry at both intra- and inter-industry levels and how such effects vary with the nationality of the investor.

Design/methodology/approach – A panel dataset for Chinese industry over the period 2000-2002 is employed for an augmented production function with external factors being defined as the influences of externalities in production.

Findings – The paper provides evidence of positive spillovers working both within industries and between industries, and evidence of vertical effects being more important than horizontal effects. The results also show that Western multinational enterprises (MNEs) generate more vertical spillovers than their overseas Chinese counterparts. This latter finding confirms that it is necessary to break down inward FDI by foreign ownership, as this makes a significant improvement over previous findings.

Research limitations/implications – Owing to data constraints, the design of the empirical framework does not allow a tightly defined source of potential vertical spillovers, i.e. a distinction of vertical spillovers through backward or forward linkages.

Practical implications – Given the strong evidence of vertical spillovers, the role of foreign MNEs in bringing local suppliers up-to-date with best practices should be counted amongst the policy benefits when judging whether it is appropriate to provide policy incentives to FDI. Policy effectiveness could be maximized at targeting specific types of foreign investor.

Originality/value – The paper assesses the relative importance of horizontal and vertical spillovers in a unified framework and the role of nationality of the investor.

Article Type:

Research paper

Keyword(s):

International investments; China; Inward investment; Multinational companies; Manufacturing industries.

Journal:

Journal of Chinese Economic and Foreign Trade Studies

Volume:

Number:

Year:

2008

pp:

8-20

Copyright ©

Emerald Group Publishing Limited

ISSN:

1754-4408

1. Introduction

It has long been thought that foreign direct investment (FDI) generates spillovers which are regarded as the most significant channel for the dissemination of modern technology (Liu et al., 2000). Spillovers have attracted much attention not only in the academic literature on FDI and but also in government policy agencies because of their relevance for indigenous industrial development. Indeed, attracting inward investment now gains prominence in the policy agenda of many governments. Yet, there is an ongoing debate about the existence and extent of such effects. This debate assumes special importance in view of the emergence of a few developing countries as major recipients of FDI and the particularly important role of FDI in diffusing modern technology to these countries.

Since Caves' (1974) pioneering work on spillovers in Canadian and Australian manufacturing, an extensive empirical literature has emerged. The bulk of the work however focuses only on horizontal spillovers of inward FDI. While this strand of research has provided important insights into the nature, direction, and magnitude of spillovers, it fails to capture vertical spillovers which, it has been argued, may be greater than horizontal spillovers (Kugler, 2001). The 2001 Issue of the UN World Investment Report (UNCTAD, 2001) focused entirely on the effects of FDI on backward linkages. Since then research on vertical spillovers has started to grow, but much of the evidence presented is based on case study approach. There remains a pertinent need to gain a better understanding on the issue by presenting substantive empirical evidence.

The purpose of this paper is twofold. First, it examines FDI spillovers accruing to Chinese local industry at both intra- and inter-industry levels. Second, it investigates the extent to which both horizontal and vertical spillovers vary with nationality of the investor. In doing so, this article complements and deepens existing research and promotes the research stream's capability to effectively inform policy.

A panel dataset for Chinese industry over the period of 2000-2002 is employed in this study. China is an appropriate context for this research because the last two decades have witnessed a striking transformation in the Chinese economy from a centrally planned to an essentially market-oriented system, and it is now the world's second largest recipient of FDI. China offers an unusually diverse, complex, and dynamic setting that warrants an in-depth examination of the issue of FDI spillovers.

The remainder of the paper is structured as follows. In section 2, we present a theoretical framework by outlining the literature. Then a description of the model and data follows. The penultimate section discusses the econometric results. The final section contains conclusions.

2. Literature review and hypothesis development

The theory of spillovers associated with inward FDI is well established, and little controversy remains. Multinational enterprises' (MNEs) ownership advantages consist of technological know-how, marketing, and managerial skills, which motivate FDI to take place. However, due to the public good nature of these advantages, MNEs are often unable to fully internalize them in the host country, causing the possibility of the uncompensated benefits to leak from the MNE into local industry. Horizontal spillovers arise where domestic firms benefit from foreign firms operating within the same industry through the channels such as demonstration effects and movement of labour or direct competition.

A strong research tradition exists that tests the extent of such spillovers by identifying whether foreign presence has induced higher level of productivity to local firms in the same sector through these channels. Although positive effects are often expected, the empirical literature has, however, produced rather mixed results, ranging from a positive, indeterminate, to even a negative effect (Buckley et al., 2002, 2005, 2007a, b; Wang et al., 2005; Wei and Liu, 2006). A possible explanation for this puzzle, according to Aitken and Harrison (1999), is that the observed productivity spillovers of FDI are the result of the interactions between positive and negative effects. This means that whether the overall impact is positive or negative would depend to a large extent on which of the effects prevails.

Despite the prevailing arguments for positive horizontal spillovers, recent studies have suggested that negative effects may emerge from the competition in final goods markets where incoming foreign firms are able to monopolise markets and draw demand away from local firms or confine competing local firms to less profitable segments of industry (Aitken and Harrison, 1999), thereby suffocating local unproductive competitors. The negative effects may also arise in factor market where foreign investors increase demand for scarce resources such as skilled labour and domestic credit, and hence raise production costs for local firms (Wang and Yu, 2006). These possibilities lead to the reasoning that in sectors where foreign capital is large, there may be limited scope for positive spillovers, while negative effects will intensify with rising foreign presence. In fact, large foreign presence is a sign that domestic firms lack the capabilities to defend their market share against foreign firms. It is clear that further evidence on horizontal effects is still needed to unveil insights to advance our current understanding of FDI spillovers.

Although the importance of linkage effects of inward FDI has entered mainstream economics for a long time (Lall, 1980), the potential for inter-industry linkages has only recently been considered as a channel through which spillovers might impact on the domestic economy (Harris and Robinson, 2004). For example, Giroud and Scott-Kennel (2006) suggest that linkages as spillover mechanisms are differentiated from other spillover mechanisms due to the direct influence of the foreign subsidiary on the local firms' capabilities. The importance of linkages leads Dunning (1993) to define spillovers as a direct consequence of the linkages forged between foreign investors and other economic agents in the countries in which they operate. Vertical spillovers are effects carried by FDI across industries through, for example, contacts between domestic suppliers of intermediate inputs and their multinational clients. Vertical spillovers are effects carried by FDI across industries through, for example, contacts between MNEs and their local suppliers of intermediate inputs. Many studies discuss the channels through which such spillovers would manifest themselves. The major channels, according to Javorcik (2004) include: direct knowledge transfer from foreign affiliates to local suppliers in the complementary sectors; imposing higher requirements for product quality and their management or technology; and MNEs' entry increasing the demand for intermediate products, which allows local suppliers to reap the benefits of scale economies. Linkages thus can be regarded as a propagation mechanism for technological externalities above and beyond the pecuniary externalities highlighted by Hirschman (1977), leading to a nexus between FDI penetration and productivity improvements of local non-competing upstream and downstream manufacturers (Rodriguez-Clare, 1996).

H1. There are both horizontal and vertical spillovers of FDI in Chinese industry.

Some authors further suggest that vertical spillovers are more likely to emerge compared with horizontal effects either by their prevalence or by the extent of resource transfer that occurs (e.g. Harris and Robinson, 2004). The literature on MNEs' optimal market penetration strategy emphasizes the minimization of the probability of leakage of knowledge to local competitors operating in the same industry (Ethier, 1986; Kugler, 2006). This can be achieved through formal protection of their critical inventions, trade secrets, and trade marks, and by locating in industries where intellectual property protection enforcement is effective or indigenous firms have limited capacities to imitate their foreign competitors. Thus horizontal effects may be less probable especially where intellectual right protection regime is weak, as is often seen in developing countries. This contrasts markedly with the situations where MNEs may voluntarily diffuse technology to upstream sectors for the benefit of more efficient supply chains for their operations. Furthermore, MNEs tend to operate in industries where the market structure yields less direct competition with local firms within its industry but in which upstream sectors are competitive, resulting in limited horizontal spillovers but stronger impact in non-competing and complementary sectors (Kugler, 2006). Despite the keen interest of policy makers in the subject, empirically little is known about the relative importance of these two types of spillovers in the form of productivity effects, for example. The above discussion leads to the following hypothesis:

H2. The magnitude of vertical spillovers is greater than that of horizontal spillovers.

A further key issue is whether certain types of foreign investors are more likely to generate a particular type of spillovers than others. Dunning's eclectic paradigm (Dunning, 1993) posits that ownership advantages and the motivations for FDI vary with the nationality of the investor. In a similar vein, Schroath et al. (1992) argue that MNEs in certain industries of a particular country possess specific advantage that accrued to them because of the way their industries developed in their home country. According to received theory, the differences in ownership advantage are capable of leading to different patterns of FDI and, by logical extension, to different patterns of spillovers. Most of the studies to date on FDI spillovers, however, tend to ignore the importance of heterogeneity of spillovers associated with the nationality of the MNE.

MNEs that conduct FDI in China can be broadly divided into overseas Chinese group consisting of investors from Hong Kong, Macao, and Taiwan, and western group including all other countries, notably the USA, EU, and Japan. It is well recognized that MNEs from western countries have state-of-the-art technology from heavy investment in R&D, while overseas Chinese MNEs' ownership advantages are thought to lie in labour-intensive production technology (Buckley et al., 2002, 2007a, b). Firms from overseas Chinese sources are large in number but small in scale. They are concentrated in light industries and textile projects using labour-intensive technology (Fung et al., 2002).

Because of their technology-intensive nature, western investors are more likely to fall into market-seeking type of investment motivated by their ability to provide differentiated products to Chinese market, while overseas Chinese investors are more probable to be export-oriented operating in low-technology sectors to seek efficiency, reflecting their particular type of ownership advantages. The technological profile of western MNEs means that their investment is the catalyst of technology transfer and therefore more technology spillovers can be expected through the channels such as demonstration effects, labour turnover, and competition. On the other hand, the industrial concentration of overseas Chinese firms within low-technology industries means that the generation of technological spillovers should be less pro rata than for western capital, or slower in its realization.

H3. Horizontal spillovers arising from western FDI are greater than those arising from overseas Chinese FDI.

It can be further inferred that the two groups of investors, because of their different motivations for undertaking FDI, may have different sourcing strategy and therefore vary in their interactions with the local complementary industry and the spillovers they create. Local market-oriented affiliates are likely to source more locally than do export-oriented firms (Belderbos et al., 2001). Domestic suppliers are often capable of serving foreign affiliates aimed at local markets, because quality and technical requirements are lower. On the other hand, export-oriented foreign firms may be more integrated in their parents' global networks of sales and supply and source more in the international market since local suppliers may not meet their global standards. With regards to China, it is observed that much of the exports in Chinese manufacturing are in processing trade (e.g. electronics industry) with both material sourcing and product market being outside China (Fu, 2004). This observation is consistent with Lardy (1995) who categorizes the export industries as enclaves. A larger share of local sourcing implies greater integration between MNEs and local firms in upstream sectors and a greater potential for knowledge transfer. This reasoning suggests that market-oriented western FDI generates more spillovers at inter-industry levels than export-oriented overseas Chinese FDI.

Contrary to the above market orientation thesis is a line of argument which relates the extent of local sourcing to technology profile of the MNE. According to UNCTAD (2001), foreign affiliates producing standardized products with mature, non-proprietary technologies (e.g. overseas Chinese affiliates) tend to prefer externalized, arm's length procurement. On the other hand, foreign firms which produce specialized and technologically advanced products (e.g. western affiliates) are more likely to prefer in-house production or to retain relationships with a few selected suppliers. An example in case is Japanese MNEs which are technology-intensive and have relatively low local content in developing host countries (Belderbos et al., 2001). This line of arguments leads to the possibility that the presence of overseas Chinese firms produces more vertical spillovers than that of western investors.

H4. Vertical spillovers arising from western MNEs are greater than those from overseas Chinese FDI.

3. Methodology

The approach adopted in this study generally follows the set of studies initiated by Aitken and Harrison (1999), but goes further in that it allows for not only horizontal but also vertical effects. Essentially, our model stems from Griliches (1992), who postulates an augmented production function including both internal and external factors of production. The external factors in this case are defined as the influences of externalities in production. The model reads as follows: (Equation 1) where i and t are the index industry and year, respectively. Q _i
t is value-added, S K _i
t is state-owned capital, N S K _i
t is domestic non-state capital, L _i
t is total number of employees, S I Z E _i
t is the total assets per firm, H O R I Z _i
t is the share of capital accounted for by foreign firms, V E R T _i
t is calculated as V E R T _i
t=∑_{k=1,k ≠ i} ⁿ H O R I Z _k
t /n − 1. η_i is the unobserved industry effect, λ _t is the unobserved time effect, ε is the remainder stochastic disturbance term, and ε ∼ IID(0, σ _u ²). All the variables except spillover variables in the equation are defined for domestic firms only in each industry. This should reduce the estimation bias caused by the possibility that foreign investors tend to invest in above-average productivity industries.

The two main parameters of interest are γ ₁ and γ ₂, whose sign and magnitude will inform the direction and intensity of intra- and inter-industry spillovers, respectively, controlling for the use of inputs and other sources of heterogeneity. As noted by Aitken and Harrison (1999), the two parameters can also be interpreted to account for total factor productivity as inputs are controlled for.

Note that λ _t is industry-invariant and takes account of any time-specific effect that is not captured in the equation. This makes sense for our analysis because spillover effects involve a process of learning or imitation by a local firm from a foreign firm and this takes time. η _i is also important in this analysis since China has adopted an industry-specific FDI policy, which leads to considerable variation of foreign ownership across Chinese industry. Failure to take account of such time and industry-specific effects may result in a biased assessment of FDI spillovers.

To allow for the two types of spillovers to vary with the nationality of the investor, equation (1) changes into: (Equation 2) where hmt and west denote FDI from overseas Chinese group and western group as defined above, respectively.

The study is based on data obtained from the unpublished Industrial Annual Reports for 2000-2002, which are compiled by the State Statistical Bureau (1988, 1994) of the People's Republic of China. There are 196 three-digit industries in total for each year but our sample only includes 162 industries. This arises from our decision on theoretical grounds to include only those industries which are liberalized to FDI over our sample period, in order to avoid biased results. FDI in the selected industries has been encouraged, completely without restriction. Industries for which data are imperfect are also excluded[1]. This gives us a panel of 486 observations in total.

Table I documents the top 20 industries with the largest foreign presence. Key inferences can be drawn from this table. First, the data on foreign capital share conform to the relatively well-understood pattern that over the time period some 30 per cent of capital investment in Chinese manufacturing was carried out by foreign MNEs, with overseas Chinese investors taking a share of 16 per cent and other investors (mainly from western countries) accounting for the remaining 13 per cent. The greater proportion of western capital contrasts remarkably with the long-lasting dominant role of overseas Chinese capital in the past and has important implications for spillovers[2]. Second, foreign capital is concentrated in traditional low-technology industries, e.g. textile and toys, but large foreign presence also emerges in electronics related industries, such as computers and office machinery. Furthermore, overseas Chinese capital is concentrated in industries characterized as low technology, such as textile products and plastic products, while western capital has a larger presence in high-tech industries, especially electronics related industries.

Despite a growing number of studies that have tended to employ firm-level data, following the study by Aitken and Harrison (1999), industry-level data remain attractive for its unique advantages for studying spillovers. First, many of the spillover determinants are industry-wide such as the degree of competition and FDI policy, which are well-known important factors influencing the size of FDI spillovers but cannot be captured in firm-level research. Second, there is more variation in the foreign presence variables in industry-level data (Buckley et al., 2005, 2007b).

The employment of panel data is well in accordance with some of the most promising lines of research on FDI spillovers. Panel data can minimize the possible simultaneity bias problem arising from the “performing industries” effect by controlling industry-specific effects. Earlier contributions using cross-sector data were typically unable to control for time invariant differences in productivity across industries, which might be correlated with foreign presence without being caused by it. The use of panel data also serves to overcome the issue of the omitted variables that are correlated with explanatory variables. Since both equations (1) and (2) accommodate fixed effects (FEs) in the cross-section, intercepts are allowed not to be identical across different industries. As such, those unique but missing factors affecting FDI spillovers of individual industries would be captured in the respective intercepts in the equations.

4. Results

Results of pooled ordinary least squares (POLS) with standard errors robust to heteroschedasticity, FEs, and random effects (REs) models for equation (1) are displayed in Table II. The POLS estimation shows that both H O R I Z _i
t and V E R T _i
t is statistically significant. This estimation, however, is biased due to the fundamental flaw of POLS estimator, that is, it fails to take account of industry-specific effects. The consequence of ignoring these effects may be particularly magnified in the Chinese setting. The Chinese industry has in recent years been undergoing fundamental structural transformation, along with the overall transition of the economy from state-controlled central planning to privatization and decentralization (Luo and Tan, 1997). This transformation has resulted in considerable inter-industry variation in terms of, for example, the level of government policy support, technological opportunity, and the degree of competition. Without appropriate controlling of these factors, the estimated FDI effects would be inaccurate.

The common remedy for the flaw of POLS could be re-estimation of equation (1) employing both FE and RE models. While both models allow us to purge any unobserved time invariant object-specific effects from our equation, it is possible that they may produce significantly different results because of their different assumptions on the essence of these effects. This possibility would be especially greater where the number of cross-sections is large and the number of years is small like our case. A Hausman specification test is thus implemented to compare the two models. The test indicates a clear rejection of REs, implying a correlation between specific industry effects and the explanatory variables. FE model is thus considered.

Compared with POLS, the FE model does provide additional explanatory power, as reflected in higher adjusted R-square value. This suggests that industry heterogeneity is indeed important and accounting for it leads to more accurate estimates of FDI effects. Both H O R I Z _i
t and V E R T _i
t are positive and statistically significant, revealing the presence of both horizontal and vertical spillover effects in Chinese industry[3]. This result corroborates H1. Furthermore, our test of joint significance of the two coefficients shows that vertical spillovers are larger than horizontal spillovers. This suggests that FDI produces a larger effect on firm productivity through spillovers across sectors than through within sectors. This finding lends support to H2. It is also in line with the argument that vertical spillovers are more important than horizontal spillovers (Harris and Robinson, 2004).

As discussed previously, there are reasons to expect differences between spillovers stemming from overseas Chinese capital and those from western capital. In pursuit of this line of enquiry, equation (2) is estimated and the results are reported in Table III. Again, FE model compares favourably to RE model in terms of Hausman specification test.

We first notice from the FE estimation that HORIZ_it ^htm and HORIZ_it ^west are significant, providing evidence for the occurrence of positive spillovers arising from both groups of foreign firms at intra-industry level. However, our test on equality of the coefficients shows no difference in the magnitude of the effects. The results thus do not provide support for H3. This finding is somewhat surprising given the supposition of superiority of western MNEs in terms of proprietary assets. A possible explanation for the lower-than-expected importance of western MNEs is that they are effective in protecting their intangible assets facing weak intellectual property regime in China. Such interpretation is well supported by some studies on China which find that transfer of advanced technology by MNEs is limited (e.g. Lan and Yong, 1996; Hu et al., 2005). Our second explanation is that the technology gap between western and Chinese local firms may remain large and thereby acting as an obstacle to technology transfer through FDI. This latter reasoning is based on the so-called “technological accumulation hypothesis” (Cantwell, 1989), which argues that the higher the technology gap between foreign and local firms, the lower the absorptive capacity of the latter, and thus the lower the expected benefits in terms of technology transfer to local firms. This perspective suggests that a minimum level of technological capability for local firms is required in order to absorb technology spillovers. Lastly, in interpreting this unexpected finding, caution needs to be exercised taking into account the fact that a significant share of inward FDI in China is due to “round-tripping”, whereby local Chinese firms direct capital to affiliates in Hong Kong which in turn re-originate this capital as FDI by Hong Kong firms in China in order to obtain favourable policies[4]. While the existence of such “round-tripping” capital should tend to reduce spillover effects arising from overseas Chinese investors, more precise evaluation of how it affects the relative magnitude of spillovers generated by the two groups of investors will depend on the size and nature of the “round-tripping” domestic capital in which we have limited knowledge owing to data unavailability.

The results show that both sources of inward FDI generate significant vertical spillovers. It is possible that both local market- and export-oriented foreign firms involve some level of local sourcing. The two groups of investors fall broadly into “local market-oriented” and “export-oriented” categories, respectively, but some firms in one group would nevertheless enter the other in terms of market orientation. Our test of joint significance of VERT_it ^hmt and VERT_it ^west however indicates much stronger vertical spillover effects are in the group of western investors. H4 is supported. This demonstrates that local market-oriented western MNEs source more locally than export-oriented overseas Chinese firms, and thereby generating more spillovers at the inter-sector level. The finding appears to support the “market orientation argument” rather than “the technology argument”, as discussed in section 2. Overall, the results in Table III confirm that the nature and strength of FDI spillovers vary within the foreign sector, so that affiliates from different country groups generate different spillover effects within Chinese manufacturing.

5. Summary and conclusions

Although the importance of vertical spillovers is undisputable, prior studies focus primarily on horizontal spillovers of FDI which occur at intra-industry level. Empirical evidence about the size of FDI spillovers across industries within a host country is scant. This study helps respond to a general lack of substantial empirical work on the wider, dynamic effects of foreign investors on local complementary industries and going beyond the existing studies by shedding some light on the existence and extent of such spillovers relative to horizontal spillovers.

The study has found evidence of positive spillovers working both within industries and between industries in Chinese manufacturing favouring the performance of domestic firms. Furthermore, it is evidenced that vertical effects are more important than horizontal effects, consistent with theoretical predictions. Clearly, vertical spillover effects are central to understanding how FDI affects local firms. This finding implies that quite extensive upgrading of Chinese local suppliers have resulted from inward manufacturing FDI to China.

When the data are further broken down by ownership of foreign investors, the study presents results suggesting contrasting patterns of spillovers between overseas Chinese and western MNEs. While we find no observable differences in the size of horizontal spillovers between the two groups of investors, the results show that western MNEs generate more vertical spillovers than their overseas Chinese counterparts. This latter finding confirms that it is necessary to break down inward FDI by foreign ownership, as this makes a significant improvement over previous findings. The two categories of foreign capital that we employ are clearly vindicated.

These findings need to be considered in light of the limitations of this study. First, because of data unavailability, the design of the empirical framework does not allow a tightly defined source of potential vertical spillovers, i.e. a distinction of vertical spillovers through backward or forward linkages. Second, our measure of vertical spillovers puts equal weight on each of other industries and this limits our ability to identify the different spillovers from upstream, downstream, and irrelevant industries. Therefore, caution is needed when interpreting our findings. Third, account also needs to be taken of the short time frame on which the study is focused. Spillovers may require more time to grow and to be observable and it is also perfectly possible that spillovers vary with the passage of time. Lastly, the use of industry-level data is a limitation. Compared with industry-level data, firm-level data will enable us to better evaluate, in general, the effects of spillovers and compare the magnitude of within and across industry spillovers. Despite these limitations, this research represents a starting point which will stimulate subsequent research along the lines of this study.

The findings shed light on policy relevance in terms of advocating public support for attracting inward FDI. First, the wider benefits above conventional horizontal spillovers documented in this study should clearly be taken into account in the design and justification for inward FDI policies. Given the strong evidence of vertical spillovers, the role of foreign MNEs in bringing local suppliers up-to-date with best practices and supporting their competitiveness in other ways should be counted amongst the policy benefits to be set off against policy costs when judging whether it is appropriate to provide policy incentives to FDI. The policy on fostering foreign-local linkages are at least as important as those conventional measures such as encouraging labour movement and competition between foreign and local firms within the same industry for the benefits of horizontal spillovers. Second, policy effectiveness could be maximized at targeting specific types of foreign investor. For instance, a high level of encouragement should be given to inward investment by western MNEs in downstream industries and to spurring growth in the matching local supplier industries to maximize vertical spillovers.

Notes

These industries include crude oil mining, gas mining, petrol-shale mining, other black metal mining, precious metal mining, salt production, other non-metal mining, other mining, timbering, bambooing, salt processing, man-made crude oil, general mechanical repairing, tramcar, other transportation equipment repairing, electronic machine repairing, instrument and office machine repairing, household grocery, other production and life products, electricity production, electricity supply, hot water production and supply, coal gas production, coal gas supply, water production, water supply.
As references, the western to overseas Chinese capital share in Chinese industry is about 20-23 per cent in 1995 (Buckley et al., 2002), while in 1987 and 1993 investments from Hong Kong and Macao accounted for about 69 and 71 per cent of total FDI in China, respectively (State Statistical Bureau, 1988, 1994).
Our use of the type of Cobb-Douglas production function assumes constant return to scale, which renders the inclusion of ln(SIZE) in the model inappropriate. However, we find dropping this variable does not make material difference to our key results.
The authors are grateful for one of the Reviewers for this point.

Table ITop 20 industries in terms of foreign presence in Chinese manufacturing

Table IIRegression results

Table IIIRegression results: the nationality effects of spillovers

Equation 1

Equation 2

References

Aitken, B., Harrison, A.E. (1999), "Do domestic firms benefit from direct foreign investment? Evidence from Venezuela", The American Economic Review, Vol. 89 No.3, pp.605-18.