Paper submitted to 19th Chinese Economic Association (CEA) (UK) Annual Conference
DOES LOCAL COMPETITION THREATEN FOREIGN AFFILIATES:
EVIDENCE FROM CHINA
Yi Wang
Doctoral researcher
Centre for International Business, University of Leeds (CIBUL), Leeds University Business School,
The University of Leeds, Maurice Keyworth Building, Leeds LS2 9JT, UK
Feb 2008
Abstract
Using data for 153 Chinese manufacturing industries for 1995 and 2000-2003, this paper
examines mutual productivity spillovers between foreign-owned (FOEs) and locally-owned
(LOEs) enterprises. It is hypothesised that there is a curvilinear relationship with either
negative or positive primary and secondary effects in each of the directions of mutual
spillovers. Mutual spillovers are expected to be stronger where the technological gaps
between FOEs and LOEs are smaller. The empirical model is applied to full sample (all
industries) and sub-samples (science based, scale intensive, specialised suppliers, supplier
dominated industries classified according to Pavitt's taxonomy). Results for 1995 are
compared against those for 2000-2003 in order to examine the changes of mutual spillovers
over time. This paper found evidence supporting the hypotheses. The main findings are (1)
labour-productivity-measured mutual spillovers are significant and positive, (2) mutual
spillovers principally follow a curvilinear relationship in each of their directions but the
evidence is limited for specialised suppliers; the finding of positive primary effects confirm
the widely observed simultaneous growth of FOEs and LOEs in China, (3) the scale and
magnitude of mutual spillovers have been unbalanced in each of their directions, and (4) the
hypothesis that mutual spillovers are subject to threshold effects of technological gaps is not
fully supported in the estimations for the four types of industries, and the interactions may be
conditional on the innovative features of industries.
Keywords: Foreign direct investment, productivity spillovers, mutual spillovers
Correspondence: e-mail address: bus3y6w@leeds.ac.uk
1
DOES LOCAL COMPETITION THREATEN FOREIGN AFFILIATES:
EVIDENCE FROM CHINA
Introduction
Recent years have witnessed the magnification of the intensity and propensity of
foreign direct investment (FDI) in China. Numerous studies have reached consensus
that FDI has been a key engine driving the remarkable expansion of the Chinese
economy. Indeed, as FDI represents “the transmission to the host country of a package
of capital managerial skills, and technical skills” (Johnson 1972), the entry of
multinational enterprises (MNEs) may bring significant benefits to local Chinese firms
in terms of productivity. Among the issues on the impacts of FDI on host country local
firms, one particular research topic that has been standing out in recent years is
so-called “spillover effects”. That is, how far the presence of foreign affiliates
stimulates the local-owned sector of industries, through ‘contagion effects’,
‘demonstration effects’, and competition between FOEs and LOEs. Empirical evidence
shows that the positive spillovers have been most compelling, with a predominance of
work finding in favour of the enhancement of local firms’ productivity (Caves 1974;
Globerman 1979; Liu et al. 2000; and Zhu and Tan 2000).
The literature agrees that there could be theoretically two way linkages between
inward FDI and the performance of domestic firms. Inward FDI is expected to have a
positive effect on the performance of domestic firms (Buckley, Clegg, and Wang 2002;
Caves 1974; Globerman 1979; Li, Liu, and Parker 2001), and foreign MNEs are more
likely to be attracted to industries where domestic productivity is higher and above
2
average profits realized. The recent FDI literature suggests the existence of reverse
spillovers (e.g., Cantwell, 1995; Driffield and Love, 2003, Wei, Liu and Wang 2008).
Driffield and Love (2003) observe that after initial transfer of firm-specific assets
abroad, “firms operating in the foreign country then have to undertake the process of
adapting this technology to a new environment, to take account of local working
practices, available human capital and customers’ tastes for example” (p. 662) . Wei,
Liu and Wang (2008) argue that mutual spillovers arise in a developing country
through the successful combination of the firm-specific advantages of MNEs’ and the
indigenous technology, and positive effects from LOEs can spill over to FOEs through
the diffusion of local knowledge.
Very few attempts have been made to empirically examine whether or not the
presence of LOEs is linked to the improvement of the performance of foreign firms.
Using the data for China, Buckley, Clegg, and Wang (2002) find that the direction of
the relationships run from ‘presence of foreign affiliates’ to ‘ the productivity of LOEs’,
not quite the reverse. This paper investigates whether the presence of local Chinese
firms impacts on the productivity of their foreign counterparts, positively or negatively.
This question severely warrants a close examination.
The rest of the paper is organized as follows. Section 2 reviews the literature and
proposes hypotheses. Section 3 presents data and methodology. Section 4 presents
some of the empirical results and conclusions are summarized in section 5.
Literature review
Mutual spillovers arise when foreign investments bring technological and
3
productivity benefits to indigenous firms and the performance of indigenous firms also
benefit the growth of foreign firms in the host country. The concept of mutual
spillovers originates from the analysis of motives for FDI (Dunning 1993). Ownership
advantages and advanced technological assets possessed by foreign firms enable MNEs
to obtain a higher rate of return from their FDI operations (e.g., Hymer 1976). By
attracting foreign investment, host countries may benefit from inward FDI through
direct international technology transfer, productivity gains, and technological
spillovers. Survey studies identified that such spillovers do not happen automatically,
but many channels catalyse spillovers. Blomstrom and Kokko (1998) suggest that
externalities from MNEs’ activities occur through demonstration effects, forward and
backward linkages, competition, market access externalities, employee turnover, and
imitation.
The literature agrees that technology sourcing is an important motive for FDI
despite of little empirical investigation has been done about this issue. Wei, Liu, and
Wang (2008) note that the hypothesis of reverse spillovers implies that firms tend to
invest in industries or places closer to technology leaders in order to benefit form
technological spillovers. Driffield and Love (2005) found supporting evidence that
technology generated by domestic firms in the manufacturing sector spilled over to
MNEs in the UK, and this effect was significant only in research and development
(R&D) intensive sectors. Studies on technology sourcing FDI stress that the concept of
technological compatibility of MNEs is ambiguous and the significance of
technological capabilities of domestic firms generates country and industry specific
4
advantages which attract inward FDI (e.g., Cantwell 1989; Neven and Siotis 1996).
Mutinelli
and
Piscitello
(1998)
argue
that,
despite
of
possessing
ownership-location-internalization (OLI) advantages (Dunning 1998), MNEs may still
need to access local technology. This is because that firms’ inability to internalise all
the needed knowledge and competencies forces them to acquire it outside.
Empirical studies suggest that MNEs are more likely to explore superior local
technologies in developed countries. Given the technological distance between foreign
and local firms is not too long, channels of spillovers may be reversed. The
establishment of joint ventures may create valuable learning opportunities for both
foreign and domestic venture partners (Inkpen 2000). Mutual spillovers arise from the
movement of knowledge-bearing workforce. Using a game theory approach, Gersbach
and Schmutzler (2003) examine endogenised technological spillovers arising from
employee turnover. They suggest that mutual spillovers occur when R&D personnel
working for foreign firms changing to work for domestic firms and “take all their
knowledge with them” (p. 180). Firms could obtain industry specific knowledge and
establish a strong position in the market through the employment of a large number of
experienced employees from competitors.
With regard to developing countries, only a few studies have been carried out on
mutual spillovers. Two conventional thoughts explain the reasons for the lack of this
type of research. First, for MNEs that seek to access superior technologies,
developing countries are not likely to be their desired location because competitive
technologies are usually unavailable there. This view is opposed by Wei, Liu, and
5
Wang (2008), who argue that foreign investors may learn indigenous knowledge from
domestic counterpart and, for foreign firms to be competitive in a developing country,
the indigenous knowledge, such as the labour intensive technologies in manufacturing ,
local language, local customs, and local work ethic, is essential. Second, empirical
studies present mixed results for spillovers from FOEs to LOEs in developing
countries (e.g., Haddad and Harrison 1993; Hu and Jefferson 2002; Young and Lan
1997), which is usually reasoned that LOEs have inadequate absorbability to benefit
from FOEs and therefore reverse spillovers may not occur since foreign investors can
learn nothing from uncompetitive local firms. Opposing to this view, some literature
argues that negative spillovers from competition effects may be only one part of the
story and competitions in the long term will stimulate technological catch up of local
firms (e.g., Cantwell 1995). Liu et al. (2000) stress that “the improvement in local
technology in turn reduces the technological gap and forces foreign firms to import
new technology to remain competitive and profitable in the host market”
(pp.409-410).
Empirical studies on China have been centred on productivity spillovers from
FOEs to LOEs. Studies on mutual spillovers in China are very rare, and the only
exception is the research conducted by Wei, Liu, and Wang (2008). While some
empirical evidence supports the view that the productivity of Chinese LOEs is
positively related to the presence of foreign investment in manufacturing industries
(e.g., Buckley, Clegg, and Wang 2002), others found a mixture of both positive and
negative spillovers (e.g., Hu and Jefferson 2002). For those mixed results, Buckley,
6
Clegg, and Wang (2007) argue that productivity spillovers from FOEs to LOEs actually
follow a curvilinear relationship, which means that local labour productivity tend to be
increasing (or decreasing) at an increasing (or decreasing) rate when the presence of
foreign investment exceed (or below) a certain level. According to Buckley, Clegg, and
Wang (2007), this curvilinear relationship implies the intricacy of spillovers, reflected
by the many mixed results that traditional linear models can not explain. A significant
failure of the traditional linear models is that it can not distinguish between the primary
and secondary effects of spillovers. Primary effects measure the level of spillovers in
relation with the level of foreign presence, which is what linear models intend to
examine. The secondary effects measure the incremental effects of spillovers which
relates to the changes of foreign presence.
Assuming there are bidirectional externalities between LOEs and FOEs, it is
expected that, after a certain period of interaction, some level of foreign productivity is
linked with a certain level of local presence. In industries where local presence is low,
foreign labour productivity may be higher than those with a high level of local
presence. The available LOEs that survived through the competitions due to the early
entry of FOEs may become very competitive. The improvement of local technological
ability has pressed MNEs to transfer more advanced technologies to its affiliates in the
host country. The reverse spillovers may be composed of primary and secondary
effects. The primary effects of reverse spillovers are that an additional increase of local
presence is (positively or negatively) related to the performance of FOEs. In a
firm-level study, Wei, Liu, and Wang (2008) found that indigenous knowledge
7
spillovers have an impact on the productivity of OECD firms in China. Assuming the
presence of LOEs in an industry proxies a pool of indigenous industry-specific
knowledge, a greater primary reverse effect is expected to be positively related to a
better resource of indigenous knowledge from which FOEs can draw during their
operation in China.
The secondary effects of reverse spillovers are that the performance of FOEs’ may
increase (or decrease) beyond a certain level of local presence. First, an additional
decrease of local presence may be related to an increase of FOEs’ performance. This is
straightforward because efficient FOEs are able to force unproductive LOEs out of the
market, and enjoy the growth of their market share as a result of reduced number of
domestic competitors. Second, an additional decrease of local presence may be also
related to a decrease of FOEs’ performance if the decrease of local presence is related
to a general deficiency of an industry. Third, an additional increase of local presence,
in the form of improved LOEs’ efficiency, may result in a decrease of FOEs’
performance if the LOEs are able to eventually overtake FOEs in the competition. In
China, there are examples that LOEs that used to be downstream suppliers for MNEs
are now able to exploit their comparative advantages of being a local and surpass FOEs
after many years of development (e.g., the Lenovo’s acquisition of IBM’s PC
manufacturing and services division in 2004, see Liu, 2007 for details).
In summary, this study proposes a brief illustration of mutual spillovers through a
set of curves as shown in figure 1. The scale of mutual spillovers (the dot curve) varies
with the size of the technological gap between FOEs and LOEs. In a position where
8
technological gap is near zero, local labour productivity is too low to attract FDI and
there are little spillovers from LOEs to FOEs, but spillovers from FOEs to LOEs can be
maximally negative. In a position where the gap value is near one, LOEs are highly
competitive and there are likely to be no overall benefit from LOEs to FOEs because
there may be simultaneous positive and negative effects. In contrast, impacts from
FOEs towards LOEs in this point are likely to be the maximum because of LOEs’
strong technology absorbability. In a position where the gap is greater than one, LOEs
are significantly more productive than FOEs. Negative impacts from LOEs towards
FOEs are most likely to dominant this stage. It is expected that the mutual spillovers
are the greatest when the size of technological gap is moderate. In the case of figure 1,
a moderate technological gap will be closer but not greater than one.
Insert figure 1 here
Following the framework discussed above, this study examines both directions of
the mutual spillovers respectively, assuming that spillovers in each of the directions
follow a curvilinear relationship. Three main research questions are put forward: (1)
Are the presence of FOEs or LOEs positively and/or negatively related to the
performance of their counterparts? (2) Do the magnitude of the mutual effects change
beyond some level of the presence of FOEs or LOEs? (3) Are mutual spillovers
stronger when there are smaller technological gaps between LOEs and FOEs?
Data and methodology
Data
This paper uses two sets of pooled data for China’s 153 manufacturing industries
from 2000 to 2003 and 121 out of 153 industries in 1995. Data are collected from
9
various issues of China’s Annual Industrial Statistical Report and the Third Industrial
Census of the People’s Republic of China in 1995 published by the National Bureau of
Statistics of China (NBS, 1997). The former is compiled by NBS but not publicly
published. It follows the statistical standards as consistent as the Third Industrial
Census. Both datasets cover all state-owned firms and those domestic and foreign firms
whose annual turn over is above five million RMB. The industries refer to a three-digit
classification according to Chinese convention. There are 196 sectors in total. To
obtain balanced pooled data set, the total observations are 628 for 2000-2003 after
deducting those industries without a complete data set of the four years examined or
failing to pass other basic error checks, such as excluding those sectors that have not
been largely liberalized and that do not enjoy free entry and exit (Wang and Yu , 2007).
Figure 2 shows a matching trend of productivity growth for FOEs and LOEs during
the two periods examined. Local labour productivity increased by around 1.6 times,
while foreign labour productivity remained averagely 2.5 times higher than local
labour productivity. The value of technological gaps between FOEs and LOEs has
increased from 0.36 to 0.64, which indicates that LOEs have taken a faster pace than
FOEs on the improvement of efficiency. These facts suggest the existence of
technological catch-up of LOEs which is stimulated by the presence of foreign
investment. The descriptive analysis initiates the empirical investigation to further
inquire whether there are different levels of benefits on each of the directions of mutual
spillovers and whether LOEs have benefited more from FOEs through mutual
spillovers than do FOEs.
10
Insert figure 2 here
Methodology
Two expanded production functions are developed as follows:
log( Y it( f ,d ) )   0  1 log( MGTit( f ,d ) )   2 log( SIZE it( f ,d ) )   3GAPit
  4 PRESENCE it( d , f )   5 ( PRESENCEit( d , f ) ) 2   6 log( K it( d ) )   7 log( L(`itd )` )   it
( f ,d )
log( Y it
)   0  1 log( MGTit
( f ,d )
  4 PRESENCE it
(d , f )
)   2 log( SIZE it
( f ,d )
  5 PRESENCE
(d , f )
it
(1.1, 1.2, 1.3, 1.4)
)   3GAPit
 GAPit   6 log( K it( f ,d ) )   7 log( L(itf ,d ) )   it
(1.5, 1.6)
where subscripts i and t denote the industry and the year; superscripts d and f denote
LOEs and FOEs, respectively. νit is the error term. For comparison purpose, the first
model
is
initially
estimated
as
a
linear
function,
namely
without
the
term ( PRESENCEit( d , f ) ) 2 . The linear regressions are denoted as model (1.1) for FOEs and
model (1.2) for LOEs. Similarly, the curvilinear regressions are denoted as model (1.3)
for FOEs and model (1.4) for LOEs. Models (1.5) and (1.6) examine the interactions
between mutual spillovers and technological gaps.
Y it is the output, measured by firms’ sales, SALESit , or value added, VADit . K it and
L it are the capital and employment, respectively. All three quantities are conventional
variables for a production function. According to Buckley, Clegg and Wang (2002), the
control variables have been chosen to be the management input of firms, MGTit( f ,d ) , and
the size of firms, SIZE it( f ,d ) . MGTit( f ,d ) is measured by sales fares, MGT1(it f ,d ) , or
management fees, MGT 2(it f ,d ) . SIZE it( f ,d ) is measured by the value of fixed assets per
firm, SIZE1(it f ,d ) , net fixed assets per firm, SIZE 2 (it f ,d ) , or total assets per firm, SIZE3(it f ,d ) .
GAPit is the ratio of labour productivity of local firms to that of foreign firms.
Following Liu et al. (2000), it is assumed that a high labour productivity reflects
11
advanced technology. Thus, a larger ratio means a smaller technological gap between
LOEs and FOEs. A negative sign for the coefficient of GAP it is expected for the
regression of FOEs’ performance, and, vice versa, a positive sign is expected to be
linked with LOEs’ performance.
PRESENCEit( f ,d ) is the presence of FOEs or LOEs. PRESENCE it(d ) is local
presence measured by employment share of LOEs, LP1it , local capital share, LP2 it ,
local sales share, LP3it , local value added share, LP4 it , or LOEs’ labour
productivity, LP5 it . Accordingly, PRESENCE it( f ) is foreign presence measured by
employment share of FOEs, FP1it , foreign capital share, FP2 it , foreign sales share, FP3it ,
foreign domestic value added share, FP4 it , or foreign labour productivity, FP5it . The
usage of different proxies for presence can help to justify the robustness of the
estimation results. It is argued that labour productivity essentially reflects the
technological ability of firms and it is a better proxy of the presence (Kokko 1996).
( PRESENCEit( d , f ) ) 2 is the squared form of (local or foreign) presence. The
coefficient of PRESENCEit( d , f ) ,  4 , measures the primary effects of the spillovers, and
the coefficient of ( PRESENCEit( d , f ) ) 2 ,  5 , measures the secondary effects of the
spillovers. According to the discussion in previous section, the primary and secondary
spillovers could be either positive, negative, or zero.
PRESENCEit( d , f )  GAPit is the terms of interaction between the presence with
technological gap. The coefficient of the interaction term,  5 , measures whether the
influence of (local or foreign) presence on the performance of FOEs or LOEs become
different if those effects are bounded with a technological gap multiplier.
12
Data are further divided into four sub-samples which are classified using Pavitt
Taxonomy. According to Pavitt (1984), there are four types of industries with
distinguished characteristics. They are the science based, scale intensive, supplier
dominated, and specialised suppliers industries. For details of the industrial
characteristics, see a summary by Malerba and Orsenigo (1996). Owing to the nature of
the datasets used in this paper, it is not possible to distinguish the 153 (or 121) Chinese
manufacturing industries by the input and output of R&D innovations within each
industry, a methodology used by Pavitt (1984). However, this study employed several
descriptive checks (using indicators show in table 1) to justify that a similar
classification to Pavitt’s is appropriate for Chinese industries. More specifically, this
study defines that science based industries at the two-digit level include chemicals,
electrical and electronic engineering, petroleum, coking, and pharmaceuticals. The
scale intensive industries include food and beverages, metals, rubber and plastics,
transport equipment, and non-metallic mineral products. The supplier dominated
industries include textiles, leather and footwear, lumber, wood and paper mill products,
printing and publishing. The specialised suppliers include manufacturing of general
and special purpose machinery, and instrument engineering.
Table 1 shows that the science based industries have the largest value of scale
indicators, and the specialised suppliers have the smallest value for most of the
indicators. This is consistent with the characteristics predict ed by Pavitt taxonomy. It is
worth noting that the supplier dominated industries have the largest total employment,
number of firms and foreign presence (by capital and employment shares), and lowest
13
foreign labour productivity (during 2000-2003). This implies that this group of
industries, particularly textiles, have been largely developed into labour intensive
sectors where inward FDI are least productive among four types of industries because
of its low technological content. It seems that LOEs have become most competitive in
this group, as the technological gaps have been shortened dramatically since 1995 and
became the smallest among all types during 2000-2003.
Insert table 1 here
Results
All estimations are pooled ordinary least squares (POLS) with r andom effects
models. The descriptive statistics and correlation matrices are shown in tables 2 and 3.
Because all proxies of management input are highly correlated with most of other
variables, only results without MGTit( f ,d ) are shown here. There is a potential serial
correlation problem due to the introduction of interaction term into equations (1.5) and
(1.6). However, the exclusion of variable GAP it may be unjustified if it causes the
omission of important explanatory variable. (Note that most of the coefficients for
GAP it are highly significant in the results). Therefore, for comparison purpose, all
regressions have been re-estimated without GAP it , for which the results are highlighted
by a * sign over the shoulder of the model numbers. Table 4 presents full sample
results for 2000-2003. Tables 5 and 6 present results of sub-samples for 2000-2003.
Table 7 presents full sample results for 1995. Because of limited number of
observations in 1995, it is not possible to run regressions again for each sub-sample
using the data of 1995. Due to space constrains, only the results using labour
14
productivity as proxy of PRESENCEit( f ,d ) are presented here. Results using capital
share as a proxy of PRESENCEit( f ,d ) are similar to those of labour productivity.
Regressions using employment share, sales share and value added share as proxies for
PRESENCEit( f ,d ) did not produce reasonable results. The results that are not presented
here are available from the author upon request.
Insert table 2-7 here
Full samples results
In table 4, the full sample results for models (1.1) – (1.4) show significant positive mutual
spillovers during 2000-2003. The results of linear estimations (models 1.1 and 1.3) are that
mutual spillovers are positive and significant and the coefficients for FP5it are greater than
those for LP5 it . This demonstrates that there is a general mutual beneficial relationship
between FOEs and LOEs, and LOEs have produced greater benefits to FOEs than do FOEs.
The results of curvilinear regressions are that all coefficients for primary and secondary
effects are significant, although the secondary effects are very weak (mostly with a near-zero
value for coefficient β5). While primary effects are all positive, secondary effects are
unanimously negative, indicating that mutual spillovers on each of their directions are
increasing at a decreasing rate. For spillovers from FOEs to LOEs, this means that the LOEs
benefit from a low or moderate level of foreign productivity, but the scale of this benefit tends
to fall when the efficiency of FOEs exceeds some level. This result is consistent with the
findings by Buckley, Clegg, and Wang (2007), who argue that the industrial concentration of
(overseas Chinese firms’) FDI in standardised goods market segments within industries
means that the scope for technological spillovers is limited. Following their argument, this
15
paper suggests that the enhanced efficiency of foreign production as a whole within an
industry indicates that spillovers from FOEs to LOEs are not sustainable. For spillovers from
LOEs to FOEs, the results demonstrate that improved productivity of LOEs leads to an
increase of reverse spillovers, but reverse spillovers may start to fall if LOEs’ efficiency is
continually improved and goes beyond some level.
The results for spillovers interacting with technological gaps (models 1.5 and 1.6 in table
4) are that the coefficient for FP5it becomes insignificant when the interaction term is
introduced into the regression, while the coefficient for LP5it  GAPit is only significant at 10%
level. The results demonstrate that spillovers from FOEs to LOEs are significant and positive
when the technological gaps are not too wide, while reverse spillovers are not strongly
interacted with technological gaps, indicating that only one direction of the mutual spillovers
are subject to the threshold effect of technological gaps during 2000-2003.
In table 7, the results for 1995 are that mutual spillovers are significant and positive in
linear regressions and the coefficient for FP5it are greater than those for LP5 it , which is
consistent with the results for 2000-2003. This demonstrates that the scale and magnitude of
mutual spillovers have been unbalanced in each of the directions for both periods examined.
The greater benefits enjoyed by FOEs may be explained by MNEs’ comparative advantages,
especially the ownership and internalization capability in the host country. Comparing the two
periods examined, there were a larger portion of joint ventures (JVs) in 1995 than recent years.
For most of foreign investors of JVs, they initially acquired local knowledge (Wei, Liu, and
Wang 2008) and established vertical linkages with downstream and upstream suppliers
through their Chinese partners who played an important role during the process where reverse
16
spillovers arose.
Results of the curvilinear regressions for 1995 are that all primary effects are positive and
significant and the secondary effects of spillovers from LOEs to FOEs have a particularly
large negative coefficient. This demonstrates that, in industries where LOEs are very
productive, there is a considerable declining trend of the benefits enjoyed by FOEs because
FOEs are faced with competition pressure on the host factor and resource markets. During
current years, more and more wholly owned foreign enterprises (WFOEs) are established,
particularly by investors of previous JVs, or through the acquisition of the remainder of
previous JVs’ ownership, such as the case of Electolux China (see Yang, 2004 for details).
Compared with JVs, WFOEs have more experience of doing business in China and are better
in coping with market and policy changes, and competition forces arising from improving
local competence.
The results for the regressions with interaction terms for 1995 contrast those for
2000-2003. It is found that interaction terms have only significant coefficients in models of
reverse spillovers, but not for spillovers from FOEs to LOEs. This may be explained by the
fact that FOEs have chosen to invest in industries where LOEs were most productive during
the early period, while, during recent years, foreign investors have started to operate in
industries where local production are not necessarily most efficient. This effect may have
been enhanced with China becoming a member of WTO in 2001. After the WTO entry,
foreign investors in China benefited from a series of reforms by the Chinese government, in
particular the improvement of transparency and predictability of doing business in China. The
reduced political and environmental risks have helped ensure the confidence of foreign
17
investors and stimulate FDI flows into all encouraged sectors. Also surprisingly, the results
show that the coefficient for LP5it  GAPit is positive and highly significant at the 1% level.
This indicates that LOEs produced negative spillovers to FOEs when the technological gaps
between them were small. Relating to the technological gap indicator for 1995 in table 1, this
may be explained by the fact that the most competitive LOEs in 1995 were in science based
industries. This reverse threshold effect of gaps in 1995 may be mainly sourced from this group
of industries, of which, according to the results for 2000-2003 in table 5, the oligopolistic,
high-tech and process innovation nature leads to the development of vigorous local rivalries for
FOEs. The results for science based industries will be discussed further in the next section.
Regressions for models without GAPit variable have produced unsatisfactory results, and
have therefore been excluded in tables 5, 6 and 7.
Sub-samples results
In tables 5 and 6, the sub-samples results are that mutual spillovers are present and
positive in all four types of industries, but there are different patterns of the curvilinear form
of spillovers for each of the sub-samples.
With regard to the linear models (1.1) and (1.2), all signs of spillovers are significant and
positive except for science based industries, for which none of the coefficients of FP5it and
LP5 it is significant at the 10% level. According to Pavitt (1984), science based industries are
characterized by large oligopolistic firms, diversified competences, and process innovations.
Firms of this type use innovations produced by firms within this group and also produce
innovations that mainly serve users in the same group. The insignificance of linear form
18
mutual spillovers suggests that there are limited and possibly indirect connections between
FOEs and LOEs within a science based industry, which may be due to a lack of incentives for
intra-industry technological spillovers.
With regard to the curvilinear models (1.3) and (1.4), the results are that mutual spillovers
follow a curvilinear relationship in all sub-samples except for specialised suppliers. In table 6
(the eighth column from the left), the coefficient for LP5 2it is not significant at the 10% level.
Specialised suppliers are characterized by technology-intensive activities by small and medium
sized firms. Table 1 shows that technological gaps in this group are the widest among four types
of industries, and the level of foreign presence and scale in this group are very low. The lack of
secondary effects for reverse spillovers demonstrate that an increase of LOEs’ presence beyond
some level can hardly produce either positive or negative influence on FOEs’ performance if
the technological ability of LOEs is very low. The results also demonstrate that mutual
spillovers occur despite of the highly idiosyncratic and tacit nature of knowledge specifically
used in this type of industries (Malerba and Orsenigo 1996). The finding suggests that the
possibility that FOEs and LOEs as specialised suppliers in the same industry are able to benefit
from each other lie on some channels of inter-industry spillovers, such as improving their
ability of adapting and tailoring products to the needs of users who are in a different type of
industry.
For models (1.5) and (1.6), the results show that the patterns of interactions are distinctive
for each sub-sample. Results for science based industries are that mutual spillovers arise when
there is a certain level of technological gaps, but, opposing the hypothesis, reverse spillovers
become negative when there are smaller technological gaps (see the significantly positive
19
coefficient for LP5it  GAPit ). The results demonstrate that mutual spillovers do not tend to be
strong when both of LOEs and FOEs are very productivity in an industry because negative
effects arise on one of the directions of mutual spillovers.
Results for scale intensive industries show that the coefficient for FP5it  GAPit are
significant and positive while coefficients for the variables FP5it and GAPit become
insignificant with the introduction of interaction term. This demonstrates that there is clearly a
threshold effect for the ability LOEs are able to benefit from spillovers from FOEs in this type
of industries. Another finding for scale intensive industries is that the coefficient
for LP5it  GAPit is not significant at the 10% level, implying the reverse spillovers are not
subject to the threshold effect of gaps. This may be because that firms in these industries
achieve competence mainly through scale economies. FOEs of this type naturally benefit
from MNEs’ advantages of internalization, which enables them to integrate upstream or
downstream suppliers efficiently. In China, LOEs of the type are largely state-owned
enterprises (SOEs) for which the conventional proxy of presence may be biased because they
do not have free entry and exit from the market.
Results for supplier dominated industries are that the coefficient for FP5it  GAPit (the
sixth column from the left in table 6) is not significant at 10% level, implying the scale and
magnitude of spillovers from FOEs to LOEs in these industries do not change with a certain
level of technological gaps. Relating to the fact that LOEs in this group are most competitive
relative to FOEs among four types of industries (see the largest value of technological gap
indicator in table 1), this result is broadly consistent with the argument made by Wang and
Blomstrom (1992) that spillovers are basically endogenous outcomes of the interactions
20
between local and foreign firms. More specifically, this result demonstrates that spillovers
from FOEs to LOEs are not conditional on technological gaps when LOEs are highly
competitive in a none-technology-intensive industry where industrial innovations are
produced and applied in industries of the similar type.
Results for specialised suppliers are that the coefficient for LP5it  GAPit are not
significant at the 10% level, indicating that reverse spillovers do not change if the relative
technological ability of LOEs rises above a certain level. This result is consistent with the
results for model (1.3) (the seventh column in table 6). It suggests that the deterministic factor
that FOEs benefit from LOEs are not conditional on the size of technological gaps or the
productivity of LOEs within the same industry, but the interaction with the end users in
different industries that both FOEs and LOEs are serving.
Coefficients for variables Kit and Lit are positive and significant in all estimations.
Coefficients for the control variable log( SIZE it( f ,d ) ) are significant in full sample estimations,
and vary in estimations for sub-samples. The varied results for log( SIZE it( f ,d ) ) justify the
application of Pavitt taxonomy for Chinese industries. For example, all coefficients of
log( SIZE it( f ,d ) ) are significant for scale intensive industries, while they are insignificant for
supplier dominated industries which is composed of mostly only small firms. In addition, the
coefficients for log( SIZE it(d ) ) in models for LOEs in science based industries (the third to sixth
column from the left in table 5) are all insignificant, implying the lack of economies of scale for
science based Chinese firms.
Conclusions
The results support the existence of mutual spillovers between FOEs and LOEs
21
within Chinese manufacturing industries. It is found that spillovers from FOEs to LOEs
during 2000-2003 arose when there were smaller technological gaps, while spillovers
from LOEs to FOEs in 1995 became negative when the technological gaps were small.
More specifically, four findings of this paper follow. First, this paper finds that, among
many proxies of (local or foreign) presence, labour-productivity-measured mutual
spillovers are significant and positive. This result supports the view that China’s
developmental strategies which have been to promote the growth of indigenous firms
by encouraging inward FDI into the manufacturing industries have initiated a mutual
beneficial relationship between FOEs and LOEs within the same industry. This mutual
beneficial relationship has been essentially the improvement of labour productivit ies of
both FOEs and LOEs.
Second, it is found that mutual spillovers principally follow a curvilinear
relationship in each of their directions, but this evidence is limited for the specialised
suppliers. The unanimous results of positive and significant primary effects are broadly
consistent with the findings of many previous studies which observed simultaneous
growth of FOEs and LOEs in China (e.g., Buckley, Jeremy, and Wang 2002; Li, Liu,
and Parker 2001). The unexpected results for specialised suppliers argue that the
curvilinear relationship of spillovers from LOEs to FOEs may be conditional on a
moderate level of technological ability of LOEs, and on the nature of an industry in
terms of whether its innovations largely involve knowledge spillovers arising from
interactions with upstream customers who are in a industry different from the one that
innovation inventors belong to.
22
Third, the scale and magnitude of mutual spillovers have been unbalanced in each
of their directions. It is found that FOEs generally benefit more from LOEs than do
LOEs. The inequality of mutual spillovers suggests that China’s policy of encouraging
inward FDI in order to benefit from technological spillovers from foreign investment
have produced more benefits to FOEs than to LOEs. This, on one hand, seems to
provide supporting evidence for the long lasting attraction of China as one of the most
important host countries for MNEs. On the other hand, it indicates that there are still a
lot more to do for Chinese government in terms of leveraging FDI to promote economic
growth, such as creating more incentives for LOEs to learn, accumulate, and
substantialise the benefit from FOEs through this mutual spillovers relationship.
Last but not least, the evidence that mutual spillovers are subject to the threshold
effects of technological gaps are limited for four types of industries classified using
Pavitt taxonomy. This finding suggests that the hypothesis that mutual spillovers tend
to be stronger when the technological gaps are smaller may be conditional on the
innovative features of industries. The policy implication is that host country
government need to carry out different approaches in the monitoring or promotion of
mutual spillovers and need to pay attention to the industry-specific innovative factors
when scheming these approaches.
23
Figure 1
An illustration of mutual spillovers
Note: This is not a simulation from the empirical models, but this figure serves as an
abstractive illustration. The breakdown point is assumed to be 1. The strength of spillovers
from FOEs to LOEs is assumed to follow y=cos(x) and spillovers from LOEs to follow
y=-sin(x), where y denotes spillovers, x denotes gap, and E(x)=3.2.
Figure 2
Labour productivity and technological gap index (1995-2003)
Firm performance
GAP
12
0.95
10
0.85
8
0.75
6
0.65
4
0.55
2
0.45
0.35
0
1995
2000
2001
2002
1995
2003
Labour productivity of LOEs
Labour productivity of FOEs
2000
2001
2002
2003
Ratio of LOEs labour productivity
LOEs to FOEs
24
Table 1 Descriptive indicators for four types of industries classified according to Pavitt taxonomy
Mean
No. of industries
Number of firms
Averaged total assets of a firm
Averaged total assets of a FOE
Averaged total assets of a LOE
Averaged sales of a firm
Averaged value added of a firm
Total capital
Averaged capital of a firm
Total employment
Science
based
29
(27)
1101.25
(2337.11)
1.33
(0.33)
1.45
(0.34)
1.22
(0.32)
1.26
(0.25)
0.30
(0.07)
296.72
(139.88)
0.33
(0.08)
Scale
intensive
78
(55)
878.68
(3250.13)
0.87
(0.25)
1.07
(0.55)
0.80
(0.23)
0.59
(0.17)
0.17
(0.04)
160.60
(124.18)
0.22
(0.06)
Supplier
dominated
28
(22)
1359.53
(4585.27)
0.86
(0.31)
0.75
(0.27)
0.81
(0.30)
0.68
(0.24)
0.34
(0.12)
155.09
(139.55)
0.18
(0.06)
Specialised
suppliers
18
(17)
1039.19
(2806.77)
0.55
(0.15)
0.80
(0.24)
0.48
(0.15)
0.46
(0.09)
0.12
(0.03)
143.52
(99.22)
0.15
(0.04)
33.66
(44.61)
24.36
(47.32)
40.52
(77.88)
28.65
(52.60)
0.04
0.04
0.03
0.03
(0.02)
(0.02)
(0.02)
(0.02)
9.07
5.68
5.41
5.19
Averaged capital labour ratio
(3.36)
(2.50)
(2.03)
(1.92)
0.36
0.34
0.44
0.34
Foreign presence by capital share
(0.23)
(0.21)
(0.22)
(0.13)
Foreign presence by employment 0.24
0.19
0.30
0.19
share
(0.13)
(0.14)
(0.21)
(0.08)
14.88
8.55
6.95
8.24
Labour productivity of FOEs
(5.82)
(3.88)
(4.27)
(4.48)
6.88
4.32
5.69
3.40
Labour productivity of LOEs
(2.19)
(1.39)
(1.84)
(1.12)
0.71
0.77
0.83
0.47
Technological gap
(0.48)
(0.45)
(0.44)
(0.29)
Notes: Figures within parentheses are calculated using data of 121 industries in 1995.
Averaged employment of a firm
25
Table 2 Descriptive statistics of all variables and correlation matrix (for models of FOEs 1.1, 1.3, and 1.5)
(f)
1 log( MGT1it )
(f)
2 log( MGT 2 it )
(f)
3 log( SIZE1 it )
(f)
4 log( SIZE 2 it )
(f)
5 log( SIZE3it )
6 GAPit
7 LP1it
8 LP2it
9 LP3it
10 LP4it
11 LP5it
2
12 LP1it
2
13 LP 2it
2
14 LP3 it
2
15 LP4 it
2
16 LP5 it
(f)
17 log( K it )
(f)
18 log( Lit )
(f)
19 log( SALES it
(f)
20 log( VADit
)
)
Mean*
8.69
(NA)
9.19
(NA)
0.44
(0.10)
0.41
(0.14)
1.05
(0.41)
0.73
(0.43)
0.78
(0.89)
0.64
(0.79)
0.70
(0.76)
0.71
(0.77)
4.95
(1.62)
0.65
(0.80)
0.45
(0.65)
0.53
(0.62)
0.54
(0.62)
54.3
(6.31)
63.33
(21.46)
6.10
(4.74)
187.31
(73.34)
48.27
(17.93)
Std. Dev.*
17.44
(NA)
17.64
(NA)
0.67
(0.21)
0.61
(0.27)
1.30
(0.74)
0.70
(0.32)
0.18
(0.10)
0.22
(0.14)
0.20
(0.18)
0.20
(0.17)
5.46
(1.93)
0.26
(0.16)
0.28
(0.20)
0.27
(0.24)
0.26
(0.23)
371.45
(37.28)
94.19
(23.31)
12.17
(8.92)
421.05
(102.87)
98.38
(25.82)
2
3
0.95 0.22
(NA) (NA)
0.24
(NA)
4
NA
(NA)
NA
(NA)
NA
(NA)
5
0.32
(NA)
0.33
(NA)
0.90
(0.91)
NA
(0.93)
6
-0.45
(NA)
-0.48
(NA)
0.00
(-0.07)
NA
(-0.19)
-0.12
(-0.19)
7
-0.38
(NA)
-0.40
(NA)
0.18
(0.08)
NA
(0.14)
0.09
(0.11)
-0.03
(0.04)
8
-0.53
(NA)
-0.54
(NA)
0.05
(-0.09)
NA
(0.02)
-0.02
(0.03)
0.14
(0.13)
0.86
(0.84)
9
-0.49
(NA)
-0.50
(NA)
0.07
(-0.01)
NA
(0.04)
-0.05
(-0.02)
0.13
(0.19)
0.92
(0.92)
0.91
(0.85)
10
-0.50
(NA)
-0.51
(NA)
0.06
(0.00)
NA
(0.05)
-0.06
(-0.02)
0.16
(0.27)
0.91
(0.89)
0.91
(0.83)
0.98
(0.96)
11
0.10
(NA)
0.10
(NA)
0.28
(0.29)
NA
(0.30)
0.35
(0.33)
0.07
(0.27)
0.03
(0.12)
0.09
(0.18)
0.05
(0.14)
0.06
(0.16)
12
-0.42
(NA)
-0.43
(NA)
0.18
(0.08)
NA
(0.14)
0.09
(0.11)
-0.01
(0.05)
0.99
(1.00)
0.89
(0.86)
0.93
(0.93)
0.92
(0.90)
0.05
(0.14)
13
-0.59
(NA)
-0.60
(NA)
0.04
(-0.10)
NA
(0.00)
-0.02
(0.01)
0.18
(0.15)
0.82
(0.82)
0.98
(0.99)
0.88
(0.85)
0.89
(0.83)
0.10
(0.20)
0.86
(0.84)
14
-0.53
(NA)
-0.55
(NA)
0.09
(-0.00)
NA
(0.04)
-0.02
(-0.02)
0.17
(0.22)
0.89
(0.90)
0.93
(0.87)
0.98
(0.99)
0.97
(0.96)
0.08
(0.18)
0.92
(0.92)
0.93
(0.87)
15
-0.54
(NA)
-0.56
(NA)
0.07
(0.01)
NA
(0.05)
-0.04
(-0.02)
0.19
(0.31)
0.88
(0.87)
0.93
(0.84)
0.97
(0.95)
0.98
(0.99)
0.08
(0.20)
0.91
(0.89)
0.93
(0.85)
0.98
(0.97)
16
-0.02
(NA)
-0.01
(NA)
0.15
(0.20)
NA
(0.23)
0.18
(0.23)
0.05
(0.14)
0.07
(0.11)
0.11
(0.15)
0.09
(0.13)
0.09
(0.14)
0.91
(0.95)
0.08
(0.12)
0.13
(0.17)
0.12
(0.16)
0.11
(0.17)
17
0.93
(NA)
0.98
(NA)
0.26
(0.46)
NA
(0.41)
0.33
(0.41)
-0.52
(-0.26)
-0.35
(-0.34)
-0.52
(-0.45)
-0.44
(-0.44)
-0.45
(-0.44)
0.09
(-0.07)
-0.39
(-0.35)
-0.58
(-0.48)
-0.49
(-0.46)
-0.50
(-0.45)
-0.01
(-0.13)
18
19 20
0.91
(NA)
0.95
(NA)
0.11
(0.23)
NA
(0.22)
0.17
(0.24)
-0.39
(-0.26)
-0.52
(-0.49)
-0.61
(-0.48)
-0.55
(-0.53)
-0.56
(-0.54)
-0.02
(-0.25)
-0.56
(-0.50)
-0.66
(-0.51)
-0.60
(-0.56)
-0.60
(-0.56)
-0.09
(-0.26)
0.94
(0.91)
0.99
(0.98)
Notes: * is the statistics of variables in original form without natural logarithm. NA denotes ‘not applicable’. The variable SIZE2(f)it has negative values, of
which the natural logarithm become invalid. This variable has therefore been excluded in the correlation matrix. Figures without parentheses are from data of
153 industries from 2000 to 2003. Figures within parentheses are from data of 121 industries in 1995. Data of MGT1(f)it and MGT1(f)it are not available for
1995.
26
Table 3 Descriptive statistics of all variables and correlation matrix (for models of LOEs 1.2, 1.4, 1.6)
(d )
1 log( MGT1it )
(d )
2 log( MGT 2 it )
(d )
3 log( SIZE1it )
(d )
4 log( SIZE 2 it )
(d )
5 log( SIZE 3 it )
6 GAPit
7 FP1it
8 FP2it
9 FP3it
10 FP4it
11 FP5it
2
12 FP1it
2
13 FP2 it
2
14 FP3 it
2
15 FP4 it
2
16 FP5 it
(d )
17 log( K it )
(d )
18 log( Lit )
(d )
19 log( SALES it )
(d )
20 log( VADit )
Mean*
14.99
(NA)
25.50
(NA)
0.33
(0.06)
0.33
(0.09)
0.85
(0.25)
0.73
(0.43)
0.22
(0.11)
0.36
(0.21)
0.29
(0.24)
0.30
(0.23)
9.42
(4.47)
0.08
(0.02)
0.18
(0.06)
0.12
(0.09)
0.13
(0.08)
220.96
(39.46)
120.05
(105.52)
23.49
(48.27)
406.56
(280.31)
116.46
(75.36)
Std. Dev.*
26.32
(NA)
38.61
(NA)
0.85
(0.10)
1.23
(0.19)
1.88
(0.56)
0.70
(0.32)
0.18
(0.10)
0.22
(0.14)
0.20
(0.18)
0.20
(0.17)
11.51
(4.43)
0.12
(0.04)
0.17
(0.07)
0.15
(0.12)
0.15
(0.11)
1444.29
(170.05)
187.57
(131.04)
31.35
(57.05)
655.04
(362.72)
185.32
(104.99)
2
3
0.93 0.45
(NA) (NA)
0.59
(NA)
4
0.44
(NA)
0.58
(NA)
0.97
(0.95)
5
0.43
(NA)
0.56
(NA)
0.92
(0.97)
0.91
(0.98)
6
-0.25
(NA)
-0.26
(NA)
-0.13
(0.14)
-0.12
(0.13)
-0.09
(0.11)
7
-0.15
(NA)
-0.26
(NA)
-0.35
(-0.13)
-0.35
(-0.31)
-0.24
(-0.24)
-0.03
(-0.04)
8
-0.07
(NA)
-0.19
(NA)
-0.42
(-0.25)
-0.42
(-0.35)
-0.34
(-0.30)
-0.14
(-0.13)
0.86
(0.84)
9
-0.09
(NA)
-0.19
(NA)
-0.33
(-0.10)
-0.33
(-0.27)
-0.22
(-0.20)
-0.16
(-0.19)
0.91
(0.92)
0.91
(0.85)
10
-0.11
(NA)
-0.21
(NA)
-0.33
(-0.11)
-0.34
(-0.27)
-0.22
(-0.20)
-0.13
(-0.27)
0.92
(0.89)
0.91
(0.83)
0.98
(0.96)
11
0.24
(NA)
0.27
(NA)
0.33
(0.46)
0.32
(0.42)
0.33
(0.48)
-0.24
(-0.22)
-0.17
(-0.09)
-0.07
(-0.09)
0.00
(0.03)
-0.03
(0.07)
12
-0.15
(NA)
-0.23
(NA)
-0.26
(-0.05)
-0.26
(-0.23)
-0.15
(-0.17)
0.08
(0.02)
0.95
(0.94)
0.74
(0.70)
0.83
(0.81)
0.83
(0.77)
-0.13
(-0.06)
13
-0.14
(NA)
-0.24
(NA)
-0.36
(-0.19)
-0.36
(-0.28)
-0.28
(-0.24)
-0.05
(-0.05)
0.88
(0.84)
0.95
(0.95)
0.89
(0.81)
0.88
(0.79)
-0.06
(-0.06)
0.81
(0.78)
14
-0.12
(NA)
-0.20
(NA)
-0.23
(-0.00)
-0.23
(-0.17)
-0.11
(-0.11)
-0.08
(-0.11)
0.89
(0.90)
0.81
(0.77)
0.94
(0.96)
0.93
(0.91)
0.02
(0.08)
0.89
(0.87)
0.86
(0.80)
15
-0.13
(NA)
-0.20
(NA)
-0.22
(0.00)
-0.23
(-0.17)
-0.10
(-0.09)
-0.06
(-0.18)
0.89
(0.87)
0.79
(0.75)
0.92
(0.92)
0.94
(0.95)
0.00
(0.15)
0.89
(0.82)
0.84
(0.77)
0.98
(0.95)
16
0.09
(NA)
0.09
(NA)
0.13
(0.36)
0.12
(0.33)
0.12
(0.37)
-0.08
(-0.06)
-0.10
(-0.07)
-0.02
(-0.10)
0.01
(-0.05)
-0.01
(-0.04)
0.85
(0.91)
-0.06
(-0.03)
-0.01
(-0.05)
0.02
(0.01)
0.01
(0.04)
17
0.92
(NA)
0.98
(NA)
0.61
(0.47)
0.61
(0.46)
0.55
(0.42)
-0.29
(-0.02)
-0.27
(-0.30)
-0.22
(-0.35)
-0.21
(-0.24)
-0.23
(-0.23)
0.24
(0.23)
-0.25
(-0.22)
-0.27
(-0.34)
-0.22
(-0.18)
-0.22
(-0.16)
0.08
(0.13)
18
19 20
0.90
(NA)
0.94
(NA)
0.43
(0.26)
0.43
(0.30)
0.37
(0.25)
-0.28
(-0.10)
-0.25
(-0.35)
-0.15
(-0.34)
-0.21
(-0.30)
-0.22
(-0.28)
0.14
(0.08)
-0.23
(-0.27)
-0.21
(-0.34)
-0.23
(-0.27)
-0.24
(-0.25)
0.05
(0.02)
0.95
(0.93)
0.99
(0.98)
Notes: * is the statistics of variables in original form without natural logarithm. NA denotes ‘not applicable’. Figures without parentheses are from data of 153
industries from 2000 to 2003. Figures within parentheses are from data of 121 industries in 1995. Data of MGT1(d)it and MGT1(d)it are not available for 1995.
27
Table 4 Regression results (foreign presence proxied by labour productivity)
Full Sample (N=612)
Log(SALES(f)it)
Log(SALES(d)it)
1.1
1.1*
1.3
1.3*
1.5
1.5*
1.2
1.2*
1.4
1.4*
1.582
1.219
1.561
1.077
1.483
1.336
1.460
1.605
1.393
1.586
Constant
(15.57)*** (15.11)*** (17.07)*** (13.22)*** (12.59)*** (16.80) *** (15.32)*** (16.58)*** (15.13)*** (16.66)***
-0.027
-0.065
-0.045
-0.094
-0.040
-0.057
-0.033
-0.034
-0.040
-0.039
log( SIZE 3 ( f , d ) )
(-0.83)
(-1.92)* (-1.57)
(-2.87)*** (-1.18)
(-1.77)*
(-1.32)
(-1.30)
(-1.69)* (-1.55)
-0.151
-0.232
-0.083
0.139
0.174
GAP
(-5.36)***
(-8.46)***
(-1.67)*
(7.79)***
(9.86)***
0.028
0.022
0.085
0.059
0.040
0.049
LP5 it
(6.44)*** (5.02)*** (10.69)*** (7.17)*** (4.77)*** (7.49)***
-0.001
-0.001
LP5 2
(-8.15)*** (-5.20)***
-0.010
-0.018
LP5 it  GAPit
(-1.69)* (-5.37)***
0.004
0.002
0.025
0.017
FP5 it
(3.23)*** (1.98)** (8.63)*** (5.71)***
-0.000
-0.000
FP5 2
(-7.96)*** (-5.35)***
it
it
1.6
1.6*
1.464
1.562
(17.00)*** (18.17)***
-0.065
-0.070
(-2.90)*** (-3.03)***
0.098
(5.93)***
it
0.001
(1.15)
-0.000
(-0.02)
it
0.044
0.048
(12.17)*** (13.32)***
0.645
0.754
0.582
0.749
0.656
0.685
0.670
0.672
0.573
0.604
0.517
0.504
log( K ( f , d ) )
(18.35)*** (25.28)*** (18.22)*** (26.26)*** (18.39)*** (21.75)*** (15.98)*** (15.34)*** (13.58)*** (13.43)*** (12.93)*** (12.32)***
0.466
0.387
0.498
0.378
0.453
0.427
0.354
0.339
0.454
0.406
0.518
0.523
log( L( f , d ) )
(12.54)*** (11.08)*** (14.98)*** (11.40)*** (11.97)*** (12.35)*** (7.78)*** (7.17)*** (9.97)*** (8.44)*** (11.98)*** (11.84)***
Adjusted R2
0.946
0.940
0.957
0.946
0.946
0.946
0.857
0.852
0.870
0.859
0.887
0.885
F value
2137.642 2414.373 2290.121 2138.289 1800.068 2140.067 736.319 878.415 684.670 747.769
797.499 944.807
Note: All estimations use data for 153 Chinese manufacturing industries from 2000 to 2003. *, **, and *** denote significance at 10%, 5%, and 1% levels
respectively. Figures in parentheses are t statistics. The results remain qualitatively unchanged when (1) VAD ( f , d ) replaces SALES ( f , d ) as dependent variable,
FP5 it  GAPit
it
it
it
it
2) GAP is alternatively measured by the ratio of Total Factor Productivity (TFP), and (3) SIZE 3 ( f , d ) is replaced by SIZE 2 ( f , d ) or SIZE1( f ,d ) .
it
it
28
it
it
Table 5 Regression results for science based and scale intensive industries (foreign presence proxied by labour productivity)
Science based industries (N=116)
Scale intensive industries (N=312)
(f)
(d)
Log(SALES it)
Log(SALES it)
Log(SALES(f)it)
Log(SALES(d)it)
1.1
1.3
1.5
1.2
1.4
1.6
1.1
1.3
1.5
1.2
1.4
1.6
1.324
1.181
1.786
1.168
1.152
1.547
1.417
1.213
1.336
1.023
0.954
1.302
Constant
(7.79)*** (6.91)*** (8.70)*** (3.96)*** (4.09)*** (7.31)*** (10.69)*** (8.08)*** (8.46)*** (6.69)*** (7.22)*** (13.04) ***
0.268
0.247
0.300
0.129
0.076
0.076
-0.079
-0.084
-0.087
-0.166
-0.133
-0.096
log( SIZE 3 ( f , d ) )
(3.30)*** (3.06)*** (3.95)*** (1.46)
(0.90)
(1.23)
(-2.15)** (-2.28)** (-2.30)** (-3.66)*** (-3.42)*** (-3.47) ***
-0.149
-0.198
-0.547
0.058
0.088
0.026
-0.248
-0.259
-0.192
0.194
0.258
0.029
GAP
(-3.27)*** (-4.16)*** (-4.35)*** (2.28)*** (3.41)*** (1.32)
(-6.08)*** (-6.33)*** (-2.66)*** (7.33)*** (10.47)*** (1.31)
0.015
0.079
-0.010
0.124
0.220
0.134
LP5 it
(1.50)
(3.58)*** (-0.84)
(9.09)*** (6.16)*** (7.83)***
-0.002
-0.007
LP5 2
(-3.32)***
(-2.92)***
0.043
-0.009
LP5 it  GAPit
(3.31)***
(-0.94)
-0.000
0.025
-0.000
0.007
0.065
-0.002
FP5 it
(-0.22)
(3.64)*** (-0.13)
(2.79)*** (9.30)*** (-0.93)
-0.000
-0.001
2
FP5
(-3.74)***
(-8.75)***
0.076
0.157
FP5 it  GAPit
(9.86)***
(15.91) ***
0.836
0.803
0.756
0.776
0.650
0.367
0.562
0.542
0.572
0.774
0.550
0.312
log( K ( f , d ) )
(15.61)*** (14.84)*** (14.06)*** (8.32)*** (6.75)*** (4.54)*** (12.13)*** (11.56)*** (12.05)*** (11.18)*** (8.36)*** (5.50) ***
0.231
0.236
0.294
0.350
0.454
0.721
0.509
0.523
0.498
0.294
0.508
0.732
log( L( f , d ) )
(4.15)*** (4.31)*** (5.45)*** (3.45)*** (4.47)*** (8.64)*** (10.13)*** (10.36)*** (9.70)*** (3.93)*** (7.27)*** (12.28) ***
Adjusted R2
0.978
0.980
0.980
0.780
0.808
0.892
0.959
0.959
0.959
0.856
0.889
0.945
F value
1040.033 934.927
927.009
82.693
81.604
159.323 1439.919 1209.779 1200.605 370.588
416.490
899.022
Note: All estimations use data for 153 Chinese manufacturing industries from 2000 to 2003. *, **, and *** denote significance at 10%, 5%, and 1% levels
respectively. Figures in parentheses are t statistics. The results remain qualitatively unchanged when (1) VAD ( f , d ) replaces SALES ( f , d ) as dependent variable,
it
it
it
it
it
it
it
it
2) GAP is alternatively measured by the ratio of Total Factor Productivity (TFP), and (3) SIZE 3 ( f , d ) is replaced by SIZE 2 ( f , d ) or SIZE1( f ,d ) .
it
it
29
it
it
Table 6 Regression results for supplier dominated and specialised suppliers industries (foreign presence proxied by labour productivity)
Supplier dominated industries (N=112)
Specialised suppliers (N=72)
Log(SALES(f)it)
Log(SALES(d)it)
Log(SALES(f)it)
Log(SALES(d)it)
1.1
1.3
1.5
1.2
1.4
1.6
1.1
1.3
1.5
1.2
1.4
1.6
2.279
2.485
2.280
1.726
1.578
1.748
2.512
2.474
2.542
0.308
-0.203
1.101
Constant
(11.01)*** (14.40)*** (13.01)*** (9.29)*** (8.60)*** (9.24)*** (7.75)*** (7.36) *** (7.88) *** (2.01) ** (-1.16)
(8.13) ***
-0.049
-0.063
-0.056
-0.092
-0.072
-0.098
0.216
0.219
0.220
-0.031
-0.025
-0.013
( f ,d )
log( SIZE 3
)
(-0.76)
(-1.21)
(-1.04)
(-1.48)
(-1.20)
(-1.55)
(2.76)*** (2.68) *** (2.83) *** (-1.68)*
(-1.48)
(-1.01)
-0.506
-0.633
-0.450
0.176
0.268
0.144
-0.984
-0.985
-0.893
1.054
1.518
0.124
GAP
(-7.52)*** (-10.41)*** (-7.67)*** (3.78)*** (5.10)*** (2.61)** (-8.81)*** (-8.46) *** (-3.42) *** (10.17) *** (12.20) *** (1.09)
0.017
0.057
0.067
0.196
0.221
0.211
LP5 it
(5.30)*** (6.32)*** (4.96)***
(11.34)*** (2.21) ** (5.40) ***
-0.000
-0.003
LP5 2
(-4.56)***
(-0.25)
-0.032
-0.031
LP5 it  GAPit
(-3.78)***
(-0.42)
0.039
0.074
0.031
0.077
0.224
0.010
FP5 it
(7.87)*** (6.14)*** (3.38)***
(11.28) *** (7.63) *** (1.30)
-0.001
-0.006
FP5 2
(-3.11)***
(-5.03) ***
0.007
0.233
FP5 it  GAPit
(1.10)
(10.12) ***
0.439
0.303
0.351
0.263
0.204
0.273
0.275
0.274
0.245
0.499
0.224
0.155
log( K ( f , d ) )
(6.37)*** (4.93)*** (5.58)*** (3.98)*** (3.12)*** (4.03)*** (2.51)*** (2.33) ** (2.11) ** (6.06) *** (2.66) *** (2.20) **
0.596
0.722
0.689
0.737
0.798
0.727
0.691
0.691
0.725
0.584
0.845
0.888
log( L( f , d ) )
(8.63)*** (11.72)*** (10.84)*** (11.54)*** (12.45)*** (11.10)*** (6.64)*** (6.25) *** (6.26) *** (7.33) *** (10.49) *** (13.28) ***
AdjustedR2
0.968
0.972
0.970
0.958
0.957
0.957
0.939
0.938
0.939
0.979
0.988
0.988
F value
667.078 654.351
600.002 504.912 414.602 411.153 219.415 179.569 181.675
657.245
942.256
987.383
Note: All estimations use data for 153 Chinese manufacturing industries from 2000 to 2003. *, **, and *** denote significance at 10%, 5%, and 1% levels
respectively. Figures in parentheses are t statistics. The results remain qualitatively unchanged when (1) VAD ( f , d ) replaces SALES ( f , d ) as dependent variable,
it
it
it
it
it
it
it
it
2) GAP is alternatively measured by the ratio of Total Factor Productivity (TFP), and (3) SIZE 3 ( f , d ) is replaced by SIZE 2 ( f , d ) or SIZE1( f ,d ) .
it
it
30
it
it
Table 7 Regression results using data of 1995 (foreign presence proxied by labour productivity)
Full Sample (N=121)
Log(SALES(f)it)
1.3
2.108
(11.06)***
0.103
(1.93)*
-1.078
(-6.62)***
0.638
(6.72)***
-0.026
(-6.35)***
1.1
2.293
(8.40)***
0.203
(2.76)***
-0.507
(-1.91)*
0.086
(1.71)*
Constant
log( SIZE 3 (it f , d ) )
GAPit
LP5 it
LP5 2it
1.5
2.728
(10.75)***
0.208
(2.86)***
-1.169
(-6.71)***
0.010
(0.23)
1.2
1.354
(5.63)***
0.105
(2.76)***
0.222
(1.18)
Log(SALES(d)it)
1.4
1.162
(4.86)***
0.090
(2.59)***
0.376
(1.88)*
1.6
1.409
(5.48)***
0.100
(2.62)***
0.078
(0.29)
0.137
(5.69)***
LP5 it  GAPit
0.032
(3.45)***
FP5 it
FP5 2it
0.085
(3.33)***
-0.001
(-2.58)***
FP5 it  GAPit
0.436
(5.00)***
0.600
(7.64)***
0.923
289.448
log( K it( f , d ) )
log( L(itf , d ) )
Adjusted R^2
F value
0.178
(2.42)**
0.836
(12.17)***
0.951
391.7178
0.364
(4.40)***
0.647
(8.50)***
0.932
277.181
0.505
(4.79)***
0.469
(4.83)***
0.935
347.776
0.414
(4.04)***
0.553
(5.67)***
0.939
309.639
0.013
(0.55)
0.052
(0.89)
0.501
(4.62)***
0.475
(4.77)***
0.936
292.285
Note: *, **, and *** denote significance at 10%, 5%, and 1% levels respectively. Figures in parentheses are t statistics. The results remain qualitatively
unchanged when (1) VAD ( f , d ) replaces SALES ( f , d ) as dependent variable, 2) GAP is alternatively measured by the ratio of Total Factor Productivity (TFP),
it
it
it
and (3) SIZE 3 ( f , d ) is replaced by SIZE 2 ( f , d ) or SIZE1( f , d ) .
it
it
it
31
References
Blomstrom, Magnus and Ari Kokko (1998), "Multinational Corporations and Spillovers", Journal
of Economic Surveys, 12 (3), 247-277.
Buckley, Peter J., Jeremy Clegg, and Chengqi Wang (2002), "The Impact of Inward FDI on the
Performance of Chinese Manufacturing Firms", Journal of International Business Studies, 33
(4), 637-655.
Buckley, Peter J., Jeremy Clegg, and Chengqi Wang (2007), "Is the Relationship between Inward
FDI and Spillover Effects Linear? An Empirical Examination of the Case of China", Journal of
International Business Studies, 38 (3), 447-459.
Cantwell, John (1995), "The Globalisation of Technology: What Remains of the Product Cycle
Model?", Cambridge Journal of Economics, 19(1), 155-174.
Caves, Richard E. (1974), "Multinational Firms, Competition, and Productivity in Host-Country
Markets", Economica, 41 (162), 176-193.
Driffield, Nigel and James H. Love (2003), "Foreign Direct Investment, Technology Sourcing and
Reverse Spillovers", The Manchester School, 71 (6), 659-672.
Driffield, Nigel and James H. Love (2005), "Who Gains From Whom?: Spillovers, Competition and
Technology Sourcing in the Foreign-Owned Sector of UK Manufacturing", Scottish Journal of
Political Economy, 52 (5), 663-686.
Dunning, John H. (1988), "The Eclectic Paradigm of International Production: A Restatement and
Some Possible Extensions", Journal of International Business Studies, 19 (1), 1-31.
Dunning, John H. (1993), Multinational Enterprises and the Global Economy, First edition, London:
Addison-Wesley.
Gersbach, Hans and Armin Schmutzler (2003), "Endogenous Technological Spillovers: Causes and
Consequences", Journal of Economics and Management Strategy, 12 (2), 179-205.
Globerman, Steven (1979), "Foreign Direct Investment and 'Spillover' Efficiency Benefits in
Canadian Manufacturing Industries", Canadian Journal of Economics, 12 (1), 42-56.
Haddad, Mona and Ann E. Harrison (1993), "Are There Positive Spillovers from Direct Foreign
Investment?: Evidence from Panel Data for Morocco", Journal of Development Economics, 42
(1), 51-74.
Hymer, Steven H. (1976), "The International Operations of National Firms: A Study of Direct
Foreign Investment", Cambridge, Massachusetts: MIT Press (Monographs in Economics).
Inkpen, Andrew C. (2000), "Learning Through Joint Ventures: A Framework of Knowledge
Acquisition", Journal of Management Studies, 37 (7), 1019-1044.
Johnson, Harry G. (1972), "Survey of the issues," in Direct Foreign Investment in Asia and Pacific,
Peter Drysdale (Ed.), Toronto: University of Toronto Press, 1-17.
Kokko, Ari (1996), "Productivity Spillovers from Competition between Local Firms and Foreign
Affiliates", Journal of International Development, 8, 517-530.
Li, Xiaoying, Xiaming Liu, and David Parker (2001), "Foreign direct investment and productivity
spillovers in the Chinese manufacturing sector", Economic Systems, 25 (4), 305-321.
Liu, Xiaming, Pamela Siler, Chengqi Wang, Yingqi Wei (2000), "Productivity Spillovers from
Foreign Direct Investment: Evidence from UK Industry Level Panel Data", Journal of
International Business Studies, 31 (3), 407-425.
Malerba, Franco and Luigi Orsenigo (1996), "The Dynamics and Evolution of Industries",
Industrial and Corporate Change, 5 (1), pp. 51-87.
Mutinelli, Marco and Lucia Piscitello (1998), "The Entry Mode Choice of MNEs: An Evolutionary
approach", Research Policy, 27 (5), 491-506.
Neven, Damien and George Siotis (1996), "Technology Sourcing and FDI in the EC: An Empirical
32
Evaluation", International Journal of Industrial Organization, 14 (5), 543-560.
National Bureau of Statistics (NBS), China. (1997), "The Data of the Third National Industrial
Census of The People's Republic of China in 1995", Beijing: China Statistics Press.
Pavitt, Keith (1984), "Sectoral Patterns of Technical Change: Towards a Taxonomy and a Theory",
Research Policy, 13 (6), 343-373.
Wang, JianYe and Magnus Blomstrom (1992), "Foreign Investment and Technology Transfer: A
Simple Model", European Economic Review, 36 (1), 137-155.
Wang, Chengqi and Li Yu (2007), "Do Spillover Benefits Grow with Rising Foreign Direct
Investment? An Empirical Examination of The Case of China", Applied Economics, 39 (3),
397-405.
Wei, Yingqi, Xiaming Liu, and Chengang Wang, (2008), "Mutual Productivity Spillovers between
Foreign and Local Firms in China", Cambridge Journal of Economics, 1-23.
Yang, Yan (2004), "Being a WFOE a Growing Trend", China Daily, page 11, 14 Aug 2004,
Beijing.
Young, Stephen, and Ping Lan (1997), "Technology Transfer to China through Foreign Direct
Investment", Regional Studies, 31 (7), 669-679.
Zhu, Gangti and Kong Yam Tan (2000), "Foreign Direct Investment and Labor Productivity: New
Evidence from China as the Host", Thunderbird International Business Review, 42 (5),
507-528.
33