Segmentation and Discrimination in China’s Emerging Industrial Labour Market∗ ∗

Xiao-yuan Dong Associate Professor Department of Economics University of Winnipeg Winnipeg, Manitoba, Canada, R3B 2E9 Phone: 204-786 9307 Fax: 204-774 4134 and Paul Bowles Professor Economics Program University of Northern British Columbia Prince George B.C.

Abstract This paper analyses wage-setting behaviour in four types of enterprise: state-owned enterprises, township and village enterprises, joint ventures and foreign-invested firms in China’s light consumer goods industry in 1998. We find that there is no significant difference among the four types of firms in the returns to human capital and that wage discrimination against women and migrant workers exists across ownership types. Journal of Economic Literature Classification Numbers: J42, J71, O53, P23 Key words: Market Segmentation and Discrimination

∗ The corresponding author: Xiao-yuan Dong, Department of Economics, University of Winnipeg,

Telephone: 1-204-786-9307, fax: 1-204-774-4134, and E-mail:[email protected]. 1

Segmentation and Discrimination in China’s Emerging Industrial Labour Market∗ ∗ 1. Introduction After more than two decades of market-oriented reforms, China’s economy has changed in remarkable ways. Economic growth averaged 8.9 percent per year over the 1978-1997 period, China has become increasingly integrated with the global market and, by the mid-1990s, had become a major player in the world market, dominating global exports for products such as toys, electrical tools, garments and footwear. Accompanying this shift to a more market oriented economic system, the dominance of stateowned enterprises in industrial output has been continuously and substantially reduced as township and village enterprises, joint ventures, foreign-invested firms and other private firms have increasingly entered production.1 And yet while this transformation of the economic system has occurred, significant elements of the pre-reform administrative system continue to influence the allocation of resources in China’s economy with the result that China has been described as having, in Lardy’s (1998) words, an “unfinished economic revolution”. One area where this has been particularly evident is in factor markets where land continues to be owned by local governments, where the banking system remains influenced by credit allocation guidelines, and where inter-firm labour mobility remains low and labour markets are argued to be segmented. In this paper, we focus on labour market issues and examine the characteristics of China’s industrial labour market. In western industrial economies, it has been argued that labour markets may be segmented between

∗ We are grateful to Fiona MacPhail

for her comments on this paper. In 1978 state-owned enterprises produced 77.6 percent of the Gross Value of Industrial Output. By 1997, the composition of GVIO was 25.5 percent by state-owned enterprises, 38.1 percent by urban and rural township and village enterprises, 17.9 percent by private firms, and 18.5 percent by joint ventures, foreign-invested firms, and other types of firms. 1

2

>primary’ and >secondary’ sectors with workers with identical levels of human capital receiving higher returns to education and experience in the primary sector. (See Rosenberg, 1989 for review). The segmentation of the labour market in this way is based on firm-specific characteristics, such as, the capital intensity of the technology used and the market and institutional structures within which firms operate; capital intensive, oligopolistic firms with Trade Unions are typical representatives of firms in the primary sector. In transitional economies, economic >dualism’ takes a different form.2 Within the industrial sector, dualism is typically characterised as resulting from the co-existence of firms which still operate, at least partially, according to the norms established under the centrally planned economy and new firms which have emerged during as part of the transition to a market-oriented economy. The difference between these types of firms can be captured by ownership; the state owned enterprises representing the >old-style’, partially reformed, firms and private enterprises, the >new style’, market oriented, firms. The existence of both types of firms during the transition is likely to lead to segmented labour markets with wage setting behaviour varying between the two sectors as a result of differences in the market orientation of firms, differences in the strength of the profit objective, differences in market structure and differences in the degree to which labour recruitment and allocation are subject to administrative control. In China, after more than two decades of economic reform, a central question in analysing industrial labour markets, is the extent to which the process of market-oriented reform has eroded or eliminated this form of segmentation. Market oriented reform is also expected to have implications to another aspect of wage determination, namely, wage discrimination. Some theorists posit that the shift towards a more market oriented economic system will lead to reductions in various forms of wage discrimination; others argue, 2

See Putterman (1992) for a discussion of China as a >dual economy=. 3

however, that no such outcome should be expected. In this paper, we analyse the industrial labour market in China and examine the extent to which labour markets continue to be segmented by ownership type and the impact of marketisation on patterns of wage discrimination, using survey data drawn from four types of enterprises, namely, state-owned enterprises (SOEs), township and village enterprises (TVEs), joint ventures between urban collectives and foreign investors (JVs), and foreign-invested firms (FIFs) in 1998. The paper is organized as follows. In section II, we provide a review of the labour market issues introduced above highlighting the debates over segmentation and discrimination. In section III, we describe the data used in the empirical analysis and provide some descriptive statistics of our sample. In section IV, the results of the regression analysis are discussed. Our conclusions, presented in section V, indicate that market pressures appear to have forced firms of all ownership types to reward the human capital characteristics of individual workers in a similar manner. However, there were significant variations in the wage response to other institutional and economic factors between the foreign-invested firms and the three types of administered firms. Wage variation across firms, instead of ownership types, has become an important source of earnings inequality among Chinese industrial workers. We also find that wage discrimination, by gender and by residential status, exists in all types of firms and that the restrictions of labor mobility depressed wages more in foreign-invested firms than in administered firms. II. Overview of Labour Market Issues It has been widely argued the pace of reforms in China has been uneven with labour reforms lagging behind reforms in other areas. Indeed, while the reforms were initially focussed on the agricultural and trade sectors, before expanding to include the management of the industrial sector in the mid-1980s, Knight and Song (1994: p.1) could still write in the early 1990s that “in a sense, China does not yet have

4

a >labour market’. By comparison with other market reforms, in China labour market reform is tardy and limited.” While this statement may be in need of some reassessment in the light of the increasing ruralurban migration of the 1990s and the increasing prevalence of lay-offs (xiagang) in the state sector (Dong and Putterman, 2000), it is still argued that the industrial labour market is still segmented by ownership structures (Howell, 1997). That is, segmentation occurs because firms from different ownership categories may have different wage determination mechanisms, different objectives, are subject to varying degrees of market pressure, and draw labour from different pools as a result of the continued influence of administrative policies in the allocation of labour. To illustrate the differences between firms in different ownership categories, the allocation and reward of labour can be summarised by the following >stylised facts= (see Bowles and White 1998). SOEs tend to be more capital intensive, are still largely under the control and regulation of the state administrative agencies, and operate partly according to the principles of the previous system of central planning although they have been increasingly subject to market pressures. They are penetrated by political institutions such as the Communist Party and the official trade union (the All China Federation of Trade Unions (ACFTU)) and are subject to various systems of administrative labour regulation in terms of wage levels, welfare provisions and job allocation. SOEs are still largely populated by workers who are recruited from the immediate urban environment in line with the employment demarcation established for urban and rural residents during the Maoist period, although the reforms have seen an increasing penumbra of casual or contract workers who are recruited from further afield. SOEs contain a relatively high proportion of middle-aged and older workers, have limited new recruitment, and mobility between enterprises is low. To a certain degree, urban collectives resemble SOEs in all these aspects. The TVE sector is more diverse and contains various forms of collective, shareholding, partnership

5

and private enterprises although all are typically classified in the collective sector in official statistics. TVEs have been, and in important respects continue to be, linked with township and village governments who have influenced the hiring levels and practices of the enterprises, often pursuing local employment maximisation objectives (Dong, 1998). TVEs are staffed predominantly by local residents from rural areas although here too, especially in the southeastern coastal provinces, migrant workers from outside of the locality have also been an important source of workers. TVEs, although linked to local governments, are typically subject to harder budget constraints than SOEs and are more market oriented and are not subject to any of the formal planning mechanisms which structured much of the SOE sector. The joint venture sector is found across the country but is concentrated in the special economic zones and open areas in coastal provinces. They tend to have greater flexibility in their ability to hire and fire labour and employ a contract labour from both urban and rural sectors. However, joint ventures also share some of the same labour market characteristics as the Chinese partner whether it is an SOE, an urban collective or a TVE. Some studies, for example, have indicated that the larger, more capital-intensive joint ventures often take on more of the employment characteristics of the state-owned enterprises with which they are partnered than of foreign private firms (See Goodall and Warner, 1997). JVs therefore represent hybrid forms of firms which share some of the characteristics of the Chinese partner, who themselves are subject to various degrees of administrative control, and some of the characteristics of private firms. FIFs are similar to JVs in geographical concentration and in their reliance on contract labour drawn from both urban and rural sectors especially young and low-skilled (especially female) labour. While workers in this sector are sometimes highly paid in monetary terms, they often lack the welfare benefits and organisational protection afforded to their counterparts in the state sector and are subject to forms of work organisation and discipline more characteristic of foreign (mainly overseas Chinese) capitalist

6

enterprises. The FIFs typically operate under a much looser regulatory regime and exercise more enterprise autonomy than other enterprises. Thus, the four types of firms, SOEs, TVEs, JVs, and FIFs, examined by this paper can be located on a continuum with SOEs representing the most >administered= type of firm and FIFs the most independent and market-oriented. These differences between firms in different ownership categories provide the underlying rationale for viewing China=s industrial labour markets as segmented. Most of the literature has highlighted the differences between wage setting behaviour in SOEs and other enterprises. For example, Gordon and Li (1999) argue that the effect of the regulation of wages in the state sector has resulted in SOEs having much greater degrees of wage compression, and less reward for skill, than in other sectors. The effect of this has been, in their analysis, that the SOE sector has lost skilled workers to the other sectors during the reform period as skilled workers have moved to sectors which reward their skills more highly.3 While this may seem compelling, other commentators have been equally at pains to point out that significant changes have taken place in China’s industrial labour market. These changes can be seen both in the increases in labour mobility, as evidenced by labour migration, and by changes in the practices of SOEs. With respect to migration, Rawksi (1998: p.3), for example, argues that Astatements about China’s retarded development of labour markets and the >failure’ of reform overlook the smooth transfer of enormous numbers of farm workers into new occupations, a phenomenon that indeed amounts to the largest migration in human history, and one mediated almost exclusively through spontaneous market 3

Lardy (1998: p. 53) also points to differences between the wage setting behaviour of SOEs and other enterprises and argues that, with respect to SOEs, Aenterprise managers too often are able to ignore the interests of the state in favour of the interests of managers and workers. In particular, they are likely to maximise wages, bonuses, benefits in kind for workers, managerial emoluments, and retained funds, including depreciation and social welfare funds, rather than maximising profits. Since wages are not determined by a competitive labour market but set by firms, managers can overpay workers.@ 7

mechanisms.” With respect to the behaviour of SOEs, while some authors have seen joint ventures and firms in high-tech zones reproduce the institutional framework of >old-style SOEs= (see Corinna-Francis 1996), others have seen the effective ending of the >iron rice bowl= and the adoption of capitalist work organisation in the SOEs (see, for example, Zhao and Nichols 1998, and Chan 2000). Despite the debate on the extent to which market forces now influence firms in different ownership categories, the issue of segmentation by ownership types in China’s industrial labor market has not been examined by rigorous econometric analysis.4 Hence, the first question that we address in this paper is the degree to which labour markets continue to be segmented by ownership type. We test for the existence of market segmentation by examining wage-setting behaviour and focus on the returns to human capital in enterprises which differ by ownership types. In a perfectly competitive labour market, firms pay a wage according to the productivity of a worker; hence, individual workers who have the same productive characteristics are expected to receive the same wage. The labour market is considered segmented if workers with the same productive attributes are rewarded differently when employed in different sectors. The variation of rewards to human capital across sectors has been regarded as one of the central pieces of evidence that reject the hypothesis of a competitive labour market in favour of a segmented labour market hypothesis (Dickens and Lang, 1985). We also examine the wage effects of other economic characteristics of enterprises, such as the enterprise’s involvement in the global market, its financial situation, the degree of market competition, the presence of Trade Unions5, and enterprise size to analyse whether other forms of segmentation have

4

Yao (1999) examines labor market segmentation between the TVE sector and self-employment in agriculture using a household surveys from in a county in eastern China. He finds that the two sectors are segmented by two types of rationing in labor allocation. 5 Trade unions, represented by the ACFTU, were initially only present in SOEs. However, a national Labour Law, passed in 1994, sought to extend the role of the ACFTU to all enterprises. Given 8

accompanied the reform process. As noted, according to standard competitive theories, wages depend only on workers’ abilities and not on the characteristics of their employers that do not affect non-pecuniary aspects of employment (Krueger and Summers, 1988). After controlling for human capital variables and ownership effects, the presence of significant inter-firm wage differentials therefore constitutes a deviation from the norms of a competitive market and indicates the presence of labour market segmentation caused by factors other than ownership category. The second issue which we address is that of wage discrimination and the possible impacts of the increasing market orientation of the economy on the degree of wage discrimination. Three types of wage discrimination have been discussed in the literature. The first is discrimination against rural migrants. While this type of discrimination is common in many countries, it has particular roots in China in the unique hukou (or household registration) system which conditions (and limits) the terms on which rural migrants enter the industrial labour market. The second type is wage discrimination against women, again a phenomenon observed in many countries and based upon women=s dual roles in the productive and household sectors. A priori, it is difficult to predict how ownership structure affects the status of rural migrants and women. Studies of wage discrimination among different types of firms in the post-reform economy have focussed on two opposing forces. On one hand, the >egalitarianism’ embodied in public ownership, and the gender equality espoused in official ideology, might be expected to lessen wage discrimination against disadvantaged workers in the SOE sector, whereas the influence of more traditional patriarchal and Confucian values might hold more sway in the less regulated non-state sector. On the other hand, the soft

the role of the ACFTU as a part of the state apparatus and as operating on the traditional Leninist model of a Aconveyor belt@ between Party and workers, there has been considerable scepticism about whether the ACFTU can effectively represent the interests of workers. (For discussion of Trade Unions 9

budget constraint that protects state firms from competition in both product and labour markets would make it less costly for their managers to discriminate against rural migrants and female workers, according to Becker’s theory of discrimination (Becker, 1966). This >ideology’ versus >market forces’ approach has been applied to analysing gender wage gaps; for example, Meng (1996) and Liu, Meng and Zhang (1997) find that wage discrimination was more prevalent in the state-owned sector while the opposite pattern of gender discrimination is reported by Maurer-Fazio, Rawski, and Zhang (1999) and Maurer-Fazio and Hughes (1999). Rozelle, Dong, Zhang and Mason (2000) report that ownership type had no effect on the gender wage gap. The third type of wage discrimination is that against workers who are subject to restrictions on their ability to move from one firm to another. It has been observed that workers (especially migrant workers) are often required to submit their temporary resident permits or to give a lump sum employment deposit to their employers (Chan, 2000). Workers lose their residential permits or employment deposits if they quit without management permission before their contracts expire. This type of practice has been observed in firms from all ownership categories. Standard theory predicts that restrictions on labour mobility give monopsony power to managers which can be used to discriminate (in wages) against workers who cannot quit their jobs freely. The degree of wage discrimination of this type is expected to be positively correlated with the extent to which the firm seeks to maximise profits. III. Data The data used in this paper are derived from a survey undertaken in Dalian and Xiamen in 1998. Dalian, located in northeast China, is one of China's 14 opening-up coastal cities with a population of 5.4 million, and Xiamen, situated on the southeast coast, is one of China's four special economic zones with

in China see Ng and Warner (1998)). 10

a population of 683,000.6 Given the considerable differences in the patterns of regional development in China, the sample therefore also enables us to examine regional differences in wage setting behaviour. The data were gathered from questionnaires from 36 enterprise managers and over 700 workers. The sample of enterprises consisted of 12 SOEs, 10 TVEs, 7 JVs, and 7 FIFs (see Part 1 of Table 1). The SOE and TVE samples were equally divided between the two cities, whereas the JV sample was taken entirely from Dalian, and the FIF sample entirely from Xiamen. All of the sample TVEs were located in the suburbs of Dalian and Xiamen. In all of the JVs the Chinese partners, which were urban collectives, held the majority stake. The seven FIFs in Xiamen were owned entirely by overseas Chinese businessmen.7 The sample enterprises all produced light consumer goods. Enterprises in the sample were all responsible for marketing their own products. Most sample enterprises sold a part or all of their output in international markets with the FIFs holding the highest export share of 65% among the four ownership categories and the SOEs holding the lowest of 22.1%. The majority of the managers interviewed characterised competition in their product markets as very intense. According to the managers' assessments, based on profits in the preceding three years, four of the sample enterprises were in an >excellent’ financial situation, 17 were in a >good’ situation, 12 were >breaking-even’, and 3 were >losing money’. Two of the three money-losing enterprises were SOEs, and the remaining one was a JV. The average size was 1,104 6

1996 per capita GDP in Dalian and Xiamen at 13,641 yuan and 23,739 yuan respectively, compared favourably with the national average of 5,605 yuan. International trade and foreign investment have played an important role in the economic development of both cities. In 1996, foreign joint ventures and wholly foreign-owned enterprises produced 29.3% and 73.7% of industrial output in Dalian and Xiamen, respectively. The dominance of SOEs in industry has been on the decline in both locations although more rapidly in Xiamen than in Dalian. In 1996, the share of state-owned and urban collectively-owned enterprises in the gross value of industrial output was 62.6% in Dalian but only 14.7% in Xiamen. 7

Our sample cannot be said to be random since the enterprises included in the sample are restricted to those who were willing to co-operate with the Chinese institutes which conducted the surveys. 11

employees per enterprise for SOEs, 217 for TVEs, 318 for JVs, and 541 for FIFs. The ACFTU had representation in all of the 12 SOEs, 8 out of 10 TVEs, 5 out of 7 JVs, and 4 out of 7 FIFs. About 20 workers, most of them shop- floor workers, were randomly selected from each of the enterprises in order to obtain data on the characteristics and wages of individual workers. Descriptive statistics of the individual characteristics of workers are presented in Part 2 of Table 1. As can be seen, firms from different ownership categories drew labour from different pools, consistent with the >stylised facts’ discussed in Section II above. The sample SOEs and JVs had a similar composition of labour force in terms of age, gender and residential status. These two types of urban enterprises, most of them founded in the 1960s, employed a relatively older labour force with a typical worker being six or seven years older than his/her counterpart in TVEs and eleven or twelve years older than the typical worker in FIFs. The gender composition of the SOEs and JVs was more or less balanced and 77 to 81 percent workers were married. The labour force of the sample SOEs and JVs consisted primarily of urban residents, although the JVs employed a noticeably larger proportion of rural migrants than did the SOEs. The sample TVEs and FIFs were all established in the post-1978 reform era. The workforces of these enterprises were dominated by rural residents, although TVEs appeared to employ more local residents, whereas FIFs hired more migrant workers from other rural areas. There was also a gender gap between these two types of firms with TVEs employing more male workers and FIFs employing more female workers. In terms of human capital characteristics, a typical worker in each type of firms had about 10 to 11 years of schooling. However, the average years of work experience varied considerably across ownership types, ranging from 5.6 in FIFs to 18.2 in SOEs, reflecting in part the older workforce of SOEs. Resembling the wage structure of the country, the average monthly wage in SOEs and TVEs was lower than that in the JVs and FIFs. The practice of bonded labour was found in all four types of firms with 8.6%

12

of the workers in the sample reporting that their managers would forfeit their temporary residential permits or job deposit if they quit their jobs without permission. IV. Results We estimate a set of standard human capital-based wage equations using OLS8 and present the results in Table 2. Equation (1) is the unrestricted version of the wage equation in which the wage-setting behaviour of firms from the four ownership categories is allowed to differ from each other in terms of returns to human capital, wage discrimination, the effects of the practice of bonded labour, trade union, and enterprise=s involvement in the global market. Following Mincer (1974), we use years of schooling, years of work experience, and squared years of work experience as the proxy variables for human capital characteristics. Dummy variables are introduced to estimate the effects of residential status, gender, restriction to labour mobility, and trade unions; the benchmark in the respective set of dummies is an urban resident, female worker, who can quit her present job freely, and who is employed in an enterprise with no union representation. The share of exports in total sales is used to measure the degree of the enterprise’s involvement in the global market. To measure the ownership effect, we include three ownership dummy variables which interact with the variables on relevant characteristics of workers and firms; SOEs are the base category. We examine the effects of enterprise size, financial health, the degree of market competition and enterprise location but assume, for the sake of simplicity, that such effects do not vary across ownership types.9 The base category

8 Here we assume that a worker’s choice of type of ownership is exogenous. In Dong and Bowles

(2000), we relax this assumption and estimate the wage regressions by Heckman’s two-step method and find that the OLS results reported here are robust to self-selection bias. 9

We assume that all types of firms respond to market competition and financial situation in the same manner also due to data limitations. Because each ownership category in the sample does not cover all types of situations in these two aspects, a relaxation of this assumption creates the problem of perfect multi-collinearity. 13

for the dummy variables on financial health10, market situation and location are, respectively, an enterprise in an excellent financial situation, facing a market that is regarded as competitive versus that is regarded as very competitive and in Xiamen. Equations (2) to (6) are the restricted versions of the wage equation, respectively, under the null hypothesis that there is no variation in returns to human capital, wage discrimination, the wage effect of bonded labour, trade unions, and exports. As discussed in section II, firms from the four ownership categories lie on a continuum from the more administered SOEs through to the relatively unregulated FIFs. Equation (7) is specified under the null hypothesis that there is no difference among three types of firms - SOEs, TVEs and JVs - which could be characterised as >administered firms=, in the aforementioned areas. We first examine the ownership effect on returns to human capital. As can be seen from the estimates of equation (1), the rate of return to education is higher in TVEs and FIFs than in SOEs and JVs, and the rate of return to experience in SOEs is lower than that in the other three types of firms. However, none of the estimates of the interactive ownership dummies on human capital variables, which measure the difference between the enterprises in respective ownership categories and SOEs, is statistically significant. The F-statistic based on the R2 s of equations (1) and (2) fails to reject the null hypothesis that the returns to human capital characteristics are the same in all four types of enterprises at any conventional level of significance. These results suggest that firms from the four ownership categories are equally competitive in attracting and retaining productive workers. However, the estimated rate of returns to education, 2.3 percent for one additional year of schooling, is low relative to the rates for other developing countries.11 10

The dummy variable for financial health is derived based on the manager=s assessment of profitability in the preceding three years, and hence is a predetermined variable in the wage equation. 11

For instance, the average rate of return to education in Asia is 9 to 10 percent per additional 14

Low rates of returns to education have been reported in other studies of the post-reform Chinese economy (Parish, et. al., 1995, Xie and Hannum, 1996, and Meng and Kidd, 1997), and have often been used as an indicator of the lack of progress in China=s market reforms. While there is certainly some truth to this interpretation, interestingly, the rate of return to education in FIFs is found to be just as low as that in SOEs, TVEs, and JVs in our sample. It appears that the jobs provided by all types of firms to shop-floor workers in the light consumer goods industry are the type of jobs for which the returns to human capital are low12. Low returns to human capital are typical characteristics of workers in >secondary= labour markets in western industrial economies; the results from our study indicate that workers in light consumer goods industry might constitute such a >secondary= market in China. However, since we have data from only this one industrial sector we are unable to test for this further although this would certainly be an avenue worth exploring in future research. We next test for behavioural differences in wage discrimination among different ownership forms. According to the estimates of dummy variables on gender, residential status and the interactive ownership dummies in equation (1), the gender wage gap in SOEs is larger than that in all of the other ownership forms, whereas the wage differential between urban residents and rural migrants in SOEs is larger than that in FIFs but smaller than that in TVEs and JVs.

However, none of these differences are statistically

significant. The F-test based on a comparison of equations (1) and (3) also fails to reject the null hypothesis that there is no difference in these two types of wage discrimination among four types of firms at any conventional level of significance. The estimates in equation (3) show that the wage gap between male

year of schooling (Psacharopoulos, 1994). 12

Parish (1996) pointed out that the findings of low returns to education in Chinese firms resemble the records of small firms in Taiwan ( approximately 4% for each year of education). The returns to education were low because these small firms operate in a highly competitive environment and with few internal markets that might allow high returns to high school and college education. 15

and female workers is 11.9 percent, and the wage of an urban worker is 6.8 percent higher than that of a local rural worker and 18 percent higher than that of a rural migrant worker. These results fit intuition based on the institutional structure of Chinese labour markets and are consistent with other studies. The wage effect of the practice of bonded labour (i.e. workers whose mobility is restricted by the payment of employment deposits) in different types of firms is also tested by both the t-statistics in equation (1) and the F-statistic derived from equations (1) and (4). According to equation (1), the bonded labour effect varies over ownership forms; it would reduce the monthly wage by 7.9 percent in SOEs, 24 percent in TVEs, have no effect in JVs, and reduce the wage by 27 percent in FIFs. This pattern of variation over SOEs, TVEs, and FIFs is consistent with the expectation that the degree of monopsonist discrimination is positively correlated with the degree of the firm=s pursuit of financial objectives. The difference between SOEs and FIFs is significant at the 5 % level, whereas the differences between SOEs and TVEs or JVs is insignificant. The F-test rejects the null hypothesis of no variation over ownership forms at the 10% level. We now look at the impact of trade unions in firms from different ownership categories. The estimates of trade union and the interactive ownership dummy variables in equation (1) show that trade unions had a negative effect on wages in the combined SOE and TVE category13, but this effect is not statistically significant. In contrast, trade unions have positive effect on wages in JVs and FIFs. The difference in the trade union effect between the base category of SOEs and TVEs, and JVs and FIFs is significant at the 10% and 1% level, respectively. The F-statistic derived from equations (1) and (5) rejects the null hypothesis that trade unions have the same wage effect in all four types of firms at the 1% level of significance. 13

A combined interactive dummy is used here to avoid the problem of multicollinearity, since 16

The estimates of exports and the interactive ownership variables in equation (1) indicate that the share of exports has no impact on wages in SOEs, TVEs, and JVs, but has a positive effect in FIFs. Specifically, if the share of exports increased by one percentage point, the wage in FIFs would increase by 0.28 percent. Thus, fluctuations in the global market have more impact on the earnings of workers in FIFs than in SOEs, TVEs, and JVs. The F-test based on equations (1) and (6) rejects the null hypothesis that the degree of an enterprise’s involvement in the global market has the same effect on wages in different types of firms at the 1% level. The results from equations (1) to (6) show that the wage-setting behaviour of the three types of >administered firms= are similar in all aforementioned aspects (with the one exception of the trade union effect in JVs). The F-statistic based on equations (1) and (7) further confirms this conclusion. The FIFs resembled the administered firms regarding to returns to human capital. They appear to have been less discriminatory in their wage structure than do the three types of administered firms, although the differences are statistically insignificant (see the t- ratios of the interactive FIF dummies with rural migrant and female dummies in equation (7)). However, the wage structure of the FIFs is significantly different from the administered firms with respect to the effect of bonded labour, trade unions, and exports. The impact of other economic characteristics of enterprises is examined under the assumption of no variation in these effects among ownership forms. As has been found in market economies, wages are positively correlated with firm size and the estimate of firm size is significant at the 10% level or higher in five out of seven regressions. The intensity of market competition has a negative effect on wages, but this effect is significant at the 10% level in only one run. Wages appear to be highly correlated with the financial situation of the enterprise with all the relevant dummy variables being significant at the 1% level

all of the SOEs had the representation of trade unions. 17

in all runs. The average monthly wage in an enterprise in an >excellent’ financial situation was about 30 to 34 percent higher than that in an enterprise that was either in a >good’ situation or breaking-even, and 63 to 65 percent higher than that in a money-losing enterprise. These results suggest that while a competitive industrial labour market began to emerge in the light consumer goods industry in the two coastal cities, there remained considerable barriers to labour mobility across individual enterprises. Labour market segmentation by firm, rather than by ownership, appears to have been one of the most important factors contributing to earnings inequality among industrial workers in the late 1990s. V. Conclusions In this paper we have examined the nature of China’s emerging industrial labour market and sought to assess the extent to which it is characterised by segmentation based on the ownership category of the enterprise and the extent to which wage discrimination exists and differs between firms of different ownership categories. As indicated in section II of the paper, both of these issues are contentious points within the existing literature. We tested for the existence of market segmentation by examining wage structures one of our main findings is that there was no significant difference among firms in the four ownership types in their response to the human capital characteristics of individual workers. The return to education was low, resembling the experience of small firms in a highly competitive environment in other developing countries. Judging by the estimates of the returns to human capital, a standard indicator of labour market segmentation, we find that segmentation by ownership has begun to collapse and that firms from all ownership categories behave similarly in this aspect of wage determination. However, there were significant variations in the wage response to other institutional and economic factors between ownership types. Specifically, trade unions and enterprise’s involvement in the global economy had positive effects

18

on the wages in FIFs but had no effects in AFs. We also find that the wages of workers were sensitive to the characteristics of an enterprise, especially, the financial health of the enterprise, indicating that segmentation by firm (rather than by ownership) has become an important source of income inequality among Chinese industrial labourers in the 1990s. Regarding the patterns of wage discrimination, Our results show that wages were significantly higher for men than for women and higher for urban workers than for rural migrants across all ownership categories. These wage gaps are not significantly different among the four types of firms. We also find that firms in all categories imposed restrictions on labour mobility, which depressed wages, but the wage differential between the workers who can quit their job freely and those who cannot was larger in FIFs than in AFs.

19

References Bowles, Paul., and White, Gordon., 1998, ALabour Systems in Transitional Economies: An Analysis of China=s Township and Village Enterprises@, International Review of Comparative Public Policy, 10: 243270. Chan, Anita., 2000, AGlobalisation, China=s Free (Read Bonded) labour Market, and the Chinese Trade Unions@, Asia Pacific Business Review, 6, 3, (forthcoming). Corinna-Francis, B., 1996, AReproduction of Danwei Institutional Features in the Context of China=s Market Economy: The Case of Haidian District=s High-Tech Sector@, The China Quarterly, 147, pp. 839859. Davison, J. and MacKinnon, J, 1993, Estimation and Inference in Econometrics, Oxford, Oxford University Press. Dickens, William. T. and Lang, Kevin., 1985, AA Test of Dual Labour Market Theory@, American Economic Review, Vol. 75:792-805, September. Dong, Xiao-yuan, 1998, “Employment and Wage Determination in China’s Rural Industry: Investigation Using 1984-1990 Panel Data.” Journal of Comparative Economics, 26: 485-501. Dong, Xiao-yuan and Paul Bowles, 2000, “Segmentation and Discrimination in China’s Emerging Industrial Labour Market: Results from the Workers and Firm Matched Data.” Working paper, Department of Economics, University of Winnipeg. Dong, Xiao-yuan and Louis Putterman, 2000, “Investigating the Rise of Labour Redundancy in China’s State Industry”, Working Paper, 2000-09, Department of Economics, Brown University. Goodall, K., and Warner, M., (1997), AHuman Resources in Sino-Foreign Joint Ventures: Selected Casestudies in Shanghai, compared with Beijing@, International Journal of Human Resource Management, 8,

20

5: 569-594. Gordon, R. H., and Li, D., 1999, AThe Effects of Wage Distortions on the Transition: Theory and Evidence from China”, European Economic Review, 43: 163-183. Howell, Jude, 1997, AThe Chinese Economic Miracle and Urban Workers”, The European Journal of Development Research, Vol.9, No.2: 148-175, December. Khan, A., and Riskin, Carl, 1998, AIncome Inequality in China: Composition, Distribution and Growth of Household Income, 1988-1995.” The China Quarterly, 154:221-253. Knight, John. and Song., Lina., 1995, AToward a Labour Market in China”, Oxford Review of Economic Policy, Vol.11, 97-117. Krueger, Alan B. and Summers, Lawrence H., 1988, AEfficiency Wages and The Inter-Industry Wage Structure”, Econometrica, Vol. 56, No. 2:259-293. Lardy, Nicholas R., 1998, China’s Unfinished Economic Revolution, Brookings Institution: Washington, D.C. Maurer-Fazio, M. and Hughes, J., 2000, AThe Effect of Institutional Change on the Relative Earnings of Chinese Women: Traditional Values vs Market Forces.@ Working Paper, Department of Economics, Bates College, Maine. Meng, Xin, 1998, AMale-Female Wage Determination and Gender Wage Discrimination in China=s Rural Industrial Sector.@ Labour Economics, Vol.5:67-89. Ng, S.K., and Warner, M., 1998, China=s Trade Unions and Management, New York: St. Martin=s Press. Parish, William L. and Michelson, Ethan, 1996, APolitics and Markets: Dual Transformations.@ American Journal of Sociology, Vol. 101, No. 4:1042-59, January. Putterman, L., 1992, ADualism and Economic Reform in China”, Economic Development and Cultural

21

Change, 40, 3, pp. 467-93. Psacharopoulos, G., 1994 AReturns to Investment in Education: A Global Up-date.” World Development, Vol.22:1325-1343. Rawski, Thomas, 1998, AChina: Prospects for Full Employment”, mimeo., University of Pittsburgh, October. Rosenberg, S., 1989, AFrom Segmentation to Flexibility”, Labour and Society, October, pp. 363-407. Rozelle, Scott, Dong, Xiao-Yuan, Zhang, Linxin, and Mason, Andrew, 2000, AOpportunities and Barriers in Reform China: Gender, Work, and Wages in the Rural Economy”, Paper presented at the 2000 ASSA Meetings, Boston, January. Xie, Yu. and Hannum, Emily, 1996, ARegional Variation in Earnings Inequality in Reform-Era Urban China@, American Journal of Sociology, Vol.101, No.4:950-992, January. Yao, Yang, 1999, “Rural Industry and Labour Market Integration in Eastern China” Journal of Development Economics, Vol. 59: 463-496. Zhao, Minghua and Nichols, Theo, 1998, AManagement Control of Labour in State-Owned Enterprises: Cases from the Textile Industry@, in G. O=Leary, (ed.) Adjusting to Capitalism: Chinese Workers and the State, Armonk: M.E.Sharpe: 75-100.

22

Table 1: Descriptive Statistics -------------------------------------------------------------------------------------------------------------------------------All Firms SOEs TVEs JVs FIFs -------------------------------------------------------------------------------------------------------------------------------Part 1: Enterprise information No. of enterprises

36

12

10

7

7

No. of employees

611.4 (727.40 39.7 (39.9)

1103.5 (949.2) 22.1 (30.6)

216.6 (98.4) 37.5 (48.5)

318.3 (482.6) 47.9 (37.2)

541.3 (582.1) 65.0 (35.2)

Financial situation Excellent Good and Break-even Money-losing

4 29 3

0 10 2

1 9 0

2 4 1

1 6 0

Market situation Very competitive Competitive

33 3

12 0

8 2

6 1

7 0

No. of enterprises having trade union

29

12

8

5

4

260 35.1

204 27.5

133 17.9

144 19.5

745.1 (285.5)

699.1 (244.5)

691.2 (267.9)

727.9 (284.1)

917.18 (312.1)

32.6 (9.8)

37.3 (8.6)

29.8 (9.3)

36.4 (9.7)

25.0 (5.7)

Male workers (%)

51.9 (50.0)

51.1 (50.1)

61.1 (48.9)

53.0 (50.1)

38.9 (48.9)

Married workers (%)

63.3 (48.2)

81.2 (39.1)

55.3 (49.8)

76.9 (42.3)

32.8 (47.1)

Urban residents (%)

65.6 (47.5)

95.1 (21.7)

42.1 (49.5)

77.8 (41.7)

38.9 (48.9)

% of exports in sales

Part 2: Worker information Sample distribution 741 % 100 Wage (yuan/month) Age (years)

Local rural 15.6 1.8 33.2 1.7 25.9 residents (%) (36.3) (13.3) (47.2) (13.0) (44.0) ----------------------------------------------------------------------------------------------------------------------------------

23

Table 1: Descriptive Statistics (cont=d) ---------------------------------------------------------------------------------------------------------------------------------All Firms SOEs TVEs JVs FIFs ---------------------------------------------------------------------------------------------------------------------------------Rural Migrants 18.7 3.1 24.7 20.5 35.1 (%) (39.1) (17.4) (43.3) (40.5) (47.9) Education (year)

11.3 (3.0)

11.9 (2.8)

10.4 (2.9)

10.8 (3.2)

11.8 (2.8)

Working experience (year)

12.9 (9.9)

18.2 (9.1)

9.7 (8.7)

16.8 (10.2)

5.6 (5.1)

47.8 (10.9)

42.5 (4.9)

47.9 (6.2)

49.4 (16.2)

55.0 (13.0)

Weekly working hours

Workers who would either lose job deposit or residential permit if quitting their job (%) 8.6 13.4 3.7 5.1 10.7 (28.1) (34.1) (18.9) (22.2) (31.0) ---------------------------------------------------------------------------------------------------------------------------------Note: The table presents the sample means of variables with standard deviations reported in parentheses.

24

Table 2: OLS Estimates of the Wage Equations1 Dependent variable = log wage ---------------------------------------------------------------------------------------------------------------------------------------

(1) (2) (3) (4) (5) (6) (7) ----------------------------------------------------------------------------------------------------------------------------Constant

6.714 6.686 6.737 6.731 6.636 6.650 6.678 (31.372)* (34.716)* (32.322)* (31.586)* (31.082)* (31.469)* (50.558)* TVEs -0.124 -0.021 -0.156 -0.148 -0.082 -0.005 0.082 (-0.351) (-0.071) (-0.454) (-0.419) (-0.227) (-0.013) (1.805)*** JVs 0.149 0.052 0.098 0.144 0.300 0.131 0.145 (0.402) (0.203) (0.262) (0.393) (0.869) (0.397) (3.810)* FIFs -0.532 -0.358 -0.534 -0.577 -0.260 -0.346 -0.481 (-1.879)***(-1.594) (-1.925)* (-2.075)**(-0.957) (-1.212) (-2.073)** Education 0.021 0.023 0.021 0.021 0.021 0.021 0.022 (3.534)* (5.486)* (3.856)* (3.572)* (3.529)* (3.452)* (4.861)* Edu.x TVEs 0.007 ------0.007 0.006 0.003 0.006 -----(0.660) (0.656) (0.598) (0.298) (0.527) Edu.x JVs -0.007 -------0.006 -0.008 -0.007 -0.007 -----(-0.624) (-0.513) (-0.700) (-0.587) (-0.611) Edu.x FIFS 0.006 ------0.002 0.006 0.006 0.005 0.005 (0.556) (0.182) (0.648) (0.584) (0.510) (0.512) Experience 0.008 0.016 0.007 0.007 0.009 0.009 0.014 (1.301) (3.621)* (1.108) (1.072) (1.342) (1.347) (2.908)* Exp.x TVEs 0.009 ------0.010 0.011 0.007 0.010 -----(0.778) (0.969) (0.961) (0.616) (0.924) Exp.x JVs 0.006 ------0.013 0.006 0.005 0.007 -----(0.426) (0.993) (0.408) (0.338) (0.543) Exp.x FIFS 0.019 ------0.022 0.022 0.016 0.024 0.014 (1.194) (1.327) (1.363) (0.992) (1.531) (0.930) Experience2 0.000 -0.0002 0.0001 0.0001 0.00002 0.00004 -0.0001 (0.211) (-1.519) (0.494) (0.296) (0.122) (0.271) (-1.103) Exp.2 x TVEs -0.0003 --------0.0003 -0.0003 -0.0003 -0.0003 ------(-0.883) (-1.052) (-0.978) (-0.801) (-1.108) Exp.2 x JVs -0.0003 --------0.0004 -0.0003 -0.0002 -0.0003 ------(0.769) (-1.226) (-0.743) (-0.686) (-0.921) Exp.2 x FIFS -0.0004 -------0.0005 -0.0005 -0.0003 -0.0006 -0.0003 (-0.661) (-0.789) (-0.730) (-0.491) (-1.049) (-0.414) Working hours2 -0.001 -0.002 -0.001 -0.001 -0.002 -0.002 -0.002 (-0.343) (-0.508) (-0.238) (-0.294) (-0.391) (-0.382) (-1.226) Working hours 0.002 0.003 0.002 0.002 0.003 0.0005 -----x TVEs (0.313) (0.437) (0.301) (0.289) (0.425) (0.087) Working hours -0.0003 0.0001 -0.003 -0.0002 0.0001 -0.0003 -----x JVs (-0.054) (0.028) (-0.529) (-0.043) (0.013) (-0.067) Working hours 0.006 0.006 0.006 0.005 0.005 0.006 0.007 x FIFS (1.283) (1.291) (1.298) (1.236) (1.103) (1.268) (2.446)** -------------------------------------------------------------------------------------------------------------------------------------

25

Table 2: OLS Estimates of the Wage Equations (cont’d) Dependent variable = log wage ------------------------------------------------------------------------------------------------------------------------------------(1) (2) (3) (4) (5) (6) (7) ------------------------------------------------------------------------------------------------------------------------------------Male 0.153 0.165 0.119 0.154 0.152 0.155 0.138 (4.713)* (5.113)* (5.561)* (4.662)* (4.673)* (4.744)* (5.994)* Male x TVEs -0.036 -0.052 ------0.037 -0.031 -0.036 -----(-0.674) (-0.986) (-0.691) (-0.571) (-0.662) Male x JVs -0.042 -0.062 ------0.031 -0.041 -0.033 -----(-0.795) (-1.261) (-0.593) (-0.788) (-0.593) Male x FIFS -0.077 -0.073 ------0.085 -0.104 -0.089 -0.064 (-1.226) (-1.187) (-1.332) (1.674)*** (-1.413) (-1.102) Rural resident

-0.072 -0.083 -0.068 -0.076 -0.083 -0.106 (-1.727)*** (-2.056)** (-1.664)*** (-1.829)*** (-1.967)** (-2.775)*

-0.091 (-2.351)**

Rural migrant

-0.156 (-1.402) -0.053 (-0.407) -0.192 (-1.091) 0.039 (0.305)

Rural migrant x TVEs Rural migrant x JVs Rural migrant x FIFS Bonded labour

-0.079 (-1.757)*** Bonded labour x TVEs-0.163 (-0.907) Bonded labour x JVs 0.095 (0.775) Bonded labour x FIFS -0.187 (-2.431)**

-0.153 (-1.473) -0.060 (-0.496) -0.127 (-0.794) 0.030 (0.253)

-0.180 (-4.191)* -------

-0.081 (-1.877)*** -0.153 (-0.852) 0.121 (1.056) -0.193 (-2.472)**

-0.082 (-1.855)*** -0.162 (-0.901) 0.099 (0.903) -0.198 (-2.577)*

-------------

Trade Union

-0.172 (-1.564) -0.041 (-0.317) -0.175 (-0.962) 0.065 (0.523)

-0.167 (-1.502) -0.081 (-0.621) -0.186 (-1.058) -0.014 (-0.115)

-0.156 (-1.372) -0.092 (-0.704) -0.216 (-1.212) 0.052 (0.398)

-0.248 (-5.148)* ------

-0.141 (-3.523)* -------

-0.086 (-1.890)*** -0.150 (-0.811) 0.104 (0.833) -0.163 (-2.036)**

-0.065 (-1.496) -0.171 (-0.930) 0.086 (0.717) -0.210 (-2.657)*

-0.103 (-2.233)** ------

-------------

-----0.126 (1.639)

------0.159 (-2.027)**

-0.084 -0.074 -0.090 -0.078 0.035 -0.078 -0.008 (-1.611) (-1.476) (-1.758)*** (-1.506) (1.018) (-1.528) (-0.212) Trade Union x JVs 0.153 0.132 0.179 0.158 ------0.158 -----(1.707)*** (1.459) (1.936)*** (1.735) (1.807)*** Trade Union x FIFs 0.251 0.238 0.242 0.237 ------0.258 0.176 (3.144)* (3.099)* (3.201)* (2.964)* (3.189)* (2.498)** -------------------------------------------------------------------------------------------------------------------------------------

26

Table 2: OLS Estimates of the Wage Equations (cont’d) Dependent variable = log wage -----------------------------------------------------------------------------------------------------------------------------------(1) (2) (3) (4) (5) (6) (7) ------------------------------------------------------------------------------------------------------------------------------------Exports -0.027 -0.033 -0.026 -0.033 -0.034 0.023 -0.032 (-0.527) (-0.663) (-0.506) (-0.621) (-0.656) (0.600) (-0.946) Exports x TVEs -0.003 0.003 -0.016 0.0007 0.016 ---------(-0.024) (0.041) (-0.204) (0.008) (0.190) Exports x JVs -0.024 -0.004 -0.031 -0.022 -0.048 ---------(-0.202) (-0.036) (-0.255) (-0.189) (-0.411) Exports x FIFs 0.277 0.288 0.293 0.289 0.286 ----0.276 (2.807)* (2.958)* (3.048)* (2.913)* (2.879)* (3.015)* Size

0.0001 (1.761)***

0.0004 0.0001 0.0001 (1.873)*** (1.801)*** (1.457)

Competition

-0.055 (-1.135)

-0.059 (-1.268)

Financial situation Break-even Money-losing

Dalian Adjusted R2 F-test p-value Observations

-0.057 (-1.187)

-0.051 (-1.043)

0.00004 0.0001 (1.754)*** (2.350)* -0.083 -0.036 (-1.792)*** (-0.868)

0.00004 (1.537) -0.060 (-1.594)

-0.324 -0.340 -0.334 -0.322 -0.315 -0.318 -0.317 (-6.926)* (-7.807)* (-7.342)* (-6.854)* (-6.743)* (-7.048)* (-7.464)* -0.641 -0.650 -0.634 -0.636 -0.630 -0.635 -0.631 (-11.096)* (-11.980)* (-11.279)* (-11.026)* (-10.932)* (-11.762)* (-12.266)* -0.322 -0.328 -0.330 -0.332 -0.338 -0.309 -0.346 (-10.468)* (-10.823)* (-10.775)* (-10.757)* (-10.915)* (-10.139)* (-11.551)* 0.529 -----

688

0.531

0.530

0.526

0.523

0.523

0.531

0.713 0.698

0.753 0.607

2.306 0.074

5.265 0.005

4.006 0.008

0.829 0.660

688

688

688

688

688

688

Notes: 1.The table presents the OLS estimates of the wage equation with t-statistics reported in parentheses. The t-statistics are derived from heteroscedasticity-consistent standard errors. *, **, and *** denotes, respectively the significance level of 1%, 5%, and 10% for a two-tailed test. 2. We control for the variation in weekly working hours between firms. This variable is treated as exogenous because the number of working hours per week is determined by the firm and hence it is mandatory for workers in Chinese factories.

27