Shang-Jin Wei (1996), in an analysis of Chinese city-level data covering 1988-90, finds that holding the growth of inputs constant, a one percent increase in the share of foreign-invested firms in city output in 1988 is associated with a 0. 32 percent higher growth rate in output, although he does not control for policy. In a different approach that focuses on total factor productivity, Dees (1998) adds a policy variable by controlling for openness, and finds that FDI contributed to Chinese growth only in the 1990s. In a similar analysis, Wei (1995) finds that FDI becomes significant for growth beginning in the late 1980s.
Like the standard growth regression studies, these studies also fail to test for the direction of causality, so it could be that multinationals are attracted to high productivity localities. An exception is the work of Gangti Zhu and Ding Lu (1998), who, applying a Granger causality test to a panel data set of 50 Chinese cities during 1985-95, find a causal linkage between the presence of FDI and productivity growth (as measured by value-added per employee), though they do not discuss the opposite causation, from productivity growth to FDI.
They also find that spillover efficiencies from FDI are stronger at promoting labor productivity than in boosting total factor productivity, suggesting that the benefits of FDI in China have come from improvements in human resource allocation efficiency versus overall technological progress. Gordon Hanson (2001), in an UNCTAD study of the benefits of FDI for developing countries, is not very confident about the findings of these types of studies.
He argues that although the early empirical literature was optimistic about the impact of MNCs on host-country productivity, its findings are open to the sorts of criticisms described above about the direction of causality as well as omitted variables, such as policy. He describes more recent and promising work done on the micro-level, where time series data of manufacturing plants provides solutions to these empirical problems by showing how the productivity of domestic plants changes over time in response to the presence of MNCs.
Haddad and Harrison (1993), using data for Moroccan manufacturing plants in 1985-89, find a weak negative correlation between plant total factor productivity growth and the presence of MNCs in that sector. Aitken and Harrison (1999), using data on Venezuelan manufacturing plants for 1976-89, find productivity growth in domestic plants is negatively correlated with foreign presence in that sector. Hanson concludes that micro-level data undermines empirical support for productivity spillovers from FDI, perhaps indicating that MNCs confine competing domestic firms to less profitable segments of industry. Clearly, for the case of China, more work needs to be done on the micro-level to assess the productivity spillover effects of MNCs.
Lacking data at the micro level, we choose not to revisit the already large literature on FDI, growth and productivity. Instead, using the insights gained from this work on growth and productivity, we decided to study more directly the impact that FDI has on workers and communities in China. We have refined our inquiry both in terms of questions asked and in accounting for policy in a more direct way. In terms of the former, in the next sections we take up the issue of FDI’s effects on wages, employment and domestic investment. In all the empirical work we undertake, we use panel data from the 29 Chinese provinces over the period 1986- 1999.
3 Before discussing the regression analysis, though, we develop a new policy variable that directly measures economic liberalization. One of the biggest flaws in the previous literature has been its confounding of two processes occurring in China: one is liberalization and the other is openness to trade and foreign investment. While these have often gone hand in hand, they have not always done so. Liberalization – the freeing up of prices and the liberalization of private ownership of business and control over profits, has also been an important part of the Chinese economic story.
It is important then to distinguish these two processes and their impact on China’s economic development. Liberalization need not be equivalent to openness. As any scholar of East Asian development will attest, high levels of trade integration can co-exist with strong industrial policies that actively guide development or protect particular economic sectors from the effects of international integration. Empirical studies of China that do control for policy tend to use some measure of openness as a proxy for liberalization and reform.
These measures most often reflect to what extent a geographical area has been incorporated into China’s Open Door Policy, ranging from SEZ status to whether a locality is situated in the coastal development zone (Dees 1998; Cheng and Kwan 1999, 2000; Wei 1996). But China has exercised a clear industrial policy where trade and FDI are concerned. Its increasing openness to trade and investment has co-existed with strong administrative controls, and so a distinct measure of liberalization is necessary to control for the effects of policy separate from openness.
For the PRC, the transition from a centrally-planned to a market economy – what we are calling its course of liberalization – has been a gradualist and ongoing one. In the early years of industrial reform, the state created a “dual track system” by maintaining clear delineations between production for the central plan and production in the new and growing sphere of marketbased industry. By actually freezing the scope of the traditional mandatory balance plan, the 3 We thank Robert Feenstra for supplying some of these data. state’s strategy meant the marketized sector would overwhelm the planned sector over time in terms of economic significance, a process that Barry Naughton has termed “growing out of the plan” (Naughton 1995).
Part of this process entailed breaking up the government monopoly in industry and easing entry by nonstate firms. Although some new firms established at this time were state-owned in the sense of being started up by local governments, most of these types of additions came in the 1980s. The vast majority of new firms were “collectively” owned (an ownership category that in practice exhibits a wide variety of attributes and is thus treated as separate from the traditional state sector), but have become increasingly private or joint-venture in ownership structure as reforms have proceeded.
In light of this history, a good way to capture liberalization is to use the ratio of state sector output to all industrial output. The data we use for this measure and the regressions below are annual provincial-level data between the years 1986 and 1999; 1986 was chosen as a starting point both because that is when policy towards FDI undergoes a marked shift, and because of data availability. Where possible, the data we use comes from the China Statistical Yearbook published by the PRC’s State Statistical Bureau, as it is regarded as more reliable than the alternative provincial statistical yearbooks that we used as a backup. There is a total of 29 provinces, listed in Table 3 below.