Front. Econ. China 2016, 11(2): 265–301 DOI 10.3868/s060-005-016-0016-7

RESEARCH ARTICLE

 

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

Labour Markets in China: A Study of the Structure and Evolution of Wages Abstract This paper examines the development of labour markets and the evolution of a structure of wages in China, using household surveys for 1988, 1995, 2002 and 2007. It finds evidence of both provincial and sectoral segmentation in labour markets, with eastern regions and the state-controlled sector enjoying high wage premiums in the early reform period. During the reform, China has progressed slowly towards an integrated labour market with convergence in incomes between the rural non-agriculture sector and the urban market-based sector by 1995, when industry flourished in the rural areas. The wage gap between the rural non-agriculture sector and other sectors increased and the urban state-controlled sector remained segmented with respect to all other sectors up to 2002. However, the data from 2007 show there has been increasing sectoral and spatial integration. Keywords China, labour markets, wage structure, migration JEL Classification J31, J42, J61, R23

1

Introduction

During the past 35 years, China has witnessed significant economic growth with Received August 18, 2014 Xiaobing Wang ( ) Department of Economics, The University of Manchester, Manchester, M13 9PL, UK E-mail: [email protected] Jenifer Piesse Department of Management, King’s College London, London, SE1 9NH, UK; University of stellenbosch, South Africa E-mail: [email protected] Zhengmao Ye Key Laboratory of Mathematical Economics, Ministry of Education; and School of Economics, Shanghai University of Finance and Economics, Shanghai 200433, China E-mail: [email protected]

266

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

a programme of structural transformation from agriculture to industry. As part of this process, the country has experienced the largest labour migration in history. In 2005, 120 million people migrated, with 100 million moving from the rural to the urban areas, and 60 million moving to seek work away from their home province.1 Furthermore, another 80 million rural agricultural labourers work for township and village manufacturing enterprises close to their homes (State Council of China, 2006). China’s economic growth is supported by these labour-intensive manufacturing industries and labour is a crucial factor in this growth. Thus, it is important to achieve a better understanding of the efficient use of labour in China, which will in turn enrich our understanding of the development of the economy and the progress of the economic transition. China is moving from a planned economy to a market oriented one but previous research has shown that labour markets in China have been highly segmented for many years (Knight and Song, 1999; Appleton et al., 2004; Knight and Song, 2005). However, there is a move towards a more competitive labour market, which by necessity, includes a move away from segmentation towards integration (Dong and Bowles, 2002; Appleton et al., 2005; Knight and Song, 2005). However, most of these studies have concentrated on geographic segmentation in labour markets across provinces in China, and paid little attention to sector segmentation. Two exceptions are Meng and Zhang (2001) and Zhao (1999), where in both cases small samples were used to examine the rural and urban sectors separately. Appleton et al. (2005, 2014) studied labour market segmentation but excluded rural areas. Thus far, no study has used sub-sectors within urban and rural populations and none have used a comprehensive dataset that is representative of China as a whole. This paper focuses on the structure and evolution of wage and investigates the evolution of labour markets across regions and employment classification sectors2 to establish the degree of market segmentation and integration in four time periods that correspond to national survey dates. Labour market segmentation and wage differentials across industries that are a function of specific industry characteristics is not surprising, but the institutional barriers that 1

Migrants here are defined as individuals who worked away from their rural villages in non-agriculture work for at least a month. 2 It should be noted that the sectoral differences discussed here are not between industrial sectors but classes of labour. It relates to the wage levels of labourers thus can be used to study the labour market structure. The next section explains this in detail.

Labour Markets in China: A Study of the Structure and Evolution of Wages

267

restrict labour mobility is a legacy of state planning in China and these must be eliminated in order to achieve an efficient and fair labour market. In the transition from a planned to a market economy, income will inevitably become more unequal and regional differences will continue but it is essential that sectoral convergence takes place.3 Under the period of central planning, wages were set by the government and the labour status of individuals was strictly controlled. However, as the transition progresses, the expectation is that the system of segmented labour will diminish and wages should reflect individual skills and productivity. Clearly, this process will take some time and this paper estimates the progress of labour market convergence.4 Four representative household surveys are used for 1988, 1995, 2002 and 2007; the first follows the early reforms, the second and the third when the reforms have escalated considerably, the fourth after China joined the WTO and becoming more integrated into the world economy. Three major aspects of labour markets are addressed. First, intertemporal differences across provinces will be examined as the transition has been uneven nationally and geographical differences in wages are expected to increase. Second, under central planning, legal barriers to labour mobility not only applied to regional movements but importantly, also to movement between labour sectors. This created separate labour markets for people in the same province but with a different status in society. These differences were determined by their Hukou classification and subsequently affected employment sector to which they belonged. Thus, it is important to establish whether the wage premium exists for different sectors and the changes as the reforms progressed. Finally, as labour mobility becomes easier, some employment sectors become more integrated, although cross-sector migration is occurring at an unequal rate as institutional constraints are more rigid between some sectors than others. Thus, in the earlier period, segmentation is a constraint to the development of a competitive labour market, but by the third and fourth periods, integration between the rural non-agriculture sector and the urban market-based sector is expected. This may differ between provinces as the barriers to labour movement 3

In a market economy, regional differences such as geography and local policies are reflected in wage income. 4 See chapter two of Knight and Song (2005) titled “labour policy and progress: overview” for a detailed discussion of the timing and content of the main reforms that took place in China that have affected wage differentials.

268

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

are generally weakened but some provinces offer more opportunities than others. This paper makes three contributions to the literature on labour markets in China: (1) it identifies the extent to which segmentation exists between labour sectors; (2) it confirms the presence of regional labour market segmentation; and (3) it establishes that there is continued unequal economic growth in China in terms of wage incomes across labour market sectors and across provinces. The rest of the paper is structured as follows. The next section discusses sectoral segmentation in detail and provides a general overview of the development of labour markets in China. Section 3 introduces the data sets used and provides some descriptive data analysis. Section 4 discusses the model, the estimation technique and regression results. The final section concludes.

2

The Legacy of Segmented Labour Markets in China

China was partitioned into two major divisions, rural and urban, by the Household registration system or Hukou, whereby every Chinese citizen was registered as resident in a specific place and with either a rural or urban Hukou. 5 This functioned as a de facto internal passport mechanism (Knight and Song, 1999, pp. 258–259).6 In the pre-reform era, residents in the rural sector were assigned rights to land for farming, while those in the urban sector were provided with various benefits, including state-subsidised food and housing and, for many, access to permanent jobs. No unauthorised movement from the sector defined by registration status was permitted. This unique system of household registration confined the population to their place of birth and created two separate entities, a rural and an urban China, where migration between the two only occurred on an extremely small scale and under strict state control. Under this system, employment restrictions and state distribution policies created a highly segmented system, which still remains in many areas of the economy. As part of the reform, controls have been loosened overall. Restrictions on labour mobility to non-agricultural activities in rural areas and to employment in cities have been relaxed. However, it still remains difficult for an individual or household with rural Hukou to establish a household in an urban area and obtain 5

The Hukou Registration Act of the People’s Republic of China came into force in 1958, which strictly controlled rural to urban migration, and led to a dualistic rural-urban divide. 6 It should be noted that an under-discussed feature of the Hukou system is that it ties people not to just either rural or urban areas but to particular geographical locations.

Labour Markets in China: A Study of the Structure and Evolution of Wages

269

urban Hukou. Although temporary migration is now permitted, city governments impose restrictions on the employment of migrants in enterprises under their jurisdiction in order to minimise unemployment and maintain social order (Knight and Song, 2005, p. 184). As of 2011, it is estimated that roughly 250 million people with rural Hukou status were resident in urban areas7 (NBS, 2012). However, the urban and rural division in China is made more complicated still by sub-divisions within these sectors. For rural households, those employed in agriculture have very low marginal productivity and there is a large amount of surplus labour, whereas those in the non-agricultural sector have relatively high productivity and higher wages. Thus, there is an income gap between agricultural and non-agricultural activities resulting in a segmented rural labour market (Zhao, 1999). The urban sector is further divided and is comprised of two groups, state- and market-based. Here, the state sector is made up of individuals in employment that is part of the state plan and this functions as a quota system (zhibiao) up to 2002. These employees work for the government at some level, either national, provincial or city, or in state-owned enterprises (SOEs). These individuals are known as zhengshigong and are protected by both their urban Hukou and zhibiao status. Their labour mobility is relatively low, with limited employment opportunities available.8 These employees are a highly privileged group and even those who have urban Hukou are not automatically able to move freely into the state-controlled employment sector. The urban market-based employees, non-zhengshigong, include the self-employed and those working for non state-owned enterprises (non-SOEs) and while protected by their Hukou urban status are excluded from the state-controlled sector with respect to employment. These two urban sectors are quite distinct in terms of the rights and privileges available to employees. For example, the zhengshigong can claim many state welfare supports while non-zhengshigong cannot. As noted by Knight and Song (1999, p. 255) “a distinction is drawn in the literature between the high-wage formal sector and the low-income informal sector, which provides either 7 Residency here in this data is defined by the Chinese State Council as “at least three months” in the particular urban area. 8 Zhengshigong can be directly translated as formal workers while non-zhengshigong implies informal or temporary workers.

270

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

wage-employment or self-employment in the urban areas. The former is difficult to enter whereas the latter contains a free-entry segment.”9 As noted, some restrictions have lessened since the reforms that began in 1978, and increased throughout the 1980s and 1990s, but many barriers in the form of government restrictions on labour remain in rural China. Individuals with rural Hukou are aware that higher returns to their labour exist outside agriculture but face limited opportunities for non-agricultural activities. Thus, migration from rural to urban areas remains difficult for many people. Prior to China joining the WTO in 2001, many migrants needed a permit before being allowed to leave their village to work elsewhere, with permits issued on a quota basis. On arrival in a new city, they are still required to register with the police, and pay a fee for both the permit and registration.10 These fees may be very high and pose an insurmountable barrier for some, although many seasonal and even long-term migrants remain as illegal workers and manage to avoid police registration (Knight and Song, 1999). When China reached the first Lewis turning point around 2004 that some parts of China started to face labour shortages and increasing wages, some previous policies on labour restriction were relaxed (Wang and Weaver, 2013). However, the economic and social discrimination remain very strong (Wang et al., 2013). Despite the movement from the rural to the urban sector, there is still insufficient across-province migration to ensure a fully-integrated labour market at the national level (Johnson, 2003). Furthermore, any progress towards an open labour market in the urban market-based sector is not replicated in the state sector. The centrally administered labour system that operates the state sector remains largely intact and seems likely to continue, firstly because labour mobility between different work units is very low, and secondly, this sector is protected from competition from labour in the urban market-based sector, while the reverse is not the case. Thus, there is an inequality between levels of competition in these sectors, with the urban private 9

The formal and informal sector in Knight and Song (1999) is not the conventional use in labour economics in the West, but more of the sense of Zhengshigong and non-zhengshigong, with the former implying a tenured post with the administration, whether state or city and the latter excluded. 10 China’s extra-legal detention (Shourong) of people residing in a city for which they lack a residential permit was for many years strictly implemented and was only abolished in 2003, following an incident where Sun Zhigang, a university graduate, died while in police custody.

Labour Markets in China: A Study of the Structure and Evolution of Wages

271

labour market facing increasing pressure from employees that have been made redundant from the state sector, all of whom have urban Hukou, and from the migrants from the rural sector. This latter group is willing to take any available employment opportunity and is increasingly dominated by young and better educated workers, particularly from the more economically strong provinces (De Brauw et al., 2002). Therefore, the greatest scope for integration and the development of a competitive market lies with the convergence of the rural non-agricultural and urban market-based sectors.

3

Data

3.1

Sample and Data

The data used in the empirical estimation are four cross section national household surveys that form the Chinese Household Income Project (CHIP) for 1988, 1995, 2002 and 2007. The 1988 survey covers several provinces, although the spatial distribution is uneven. Ten provinces are well represented with both rural and urban data: Anhui, Beijing, Gansu, Guangdong, Henan, Hubei, Jiangsu, Liaoning, Shanxi, and Yunnan. In some other provinces only rural data are available. In the 1995 survey, Sichuan was added in both the urban and rural datasets. In the 2002 survey, Chongqing was treated as a provincial- level city separate from Sichuan. The 2002 and 2007 surveys collected information on migrants as well as household in the rural and urban areas. This is an adequate representative sample of China although it might not be strictly balanced. The samples of all these four surveys were drawn from a large-scale sample selected by the National Bureau of Statistics (NBS) for the annual household survey (approximately 65,000 rural households and 35,000 urban households), using a multistage stratified probability sampling method. The sample size is large. The original CHIP 1988 dataset has 31,827 individuals in urban areas and 51,352 individuals in rural areas. The original CHIP 1995 dataset has 21,694 individuals in urban areas and 34,739 in rural areas. The CHIP 2002 dataset has 20,632 individuals in urban areas and 37,968 in rural areas plus an additional 5,318 migrants. The CHIP 2007 dataset has 14,703 individuals in urban areas and 31,795 in rural areas plus an additional 8,446 migrants. The CHIP 2007 surveys

272

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

fewer provinces than previous rounds.11 The CHIP study is arguably the best publicly available data source on Chinese household income and expenditures (Riskin, Zhao and Li, 2001). These data remain the only source of household level information on income and other individual and household characteristics that are representative of China as a whole and cover a time span from 1988 to 2007 (Gustafsson, Li and Sicular, 2008). These four datasets enable comparisons of 1988, 1995, 2002 and 2007 and the changes and restructuring that occurred over this twenty-year period.12 They have been widely used in the academic literature and the main research results published in volumes edited by Griffin and Zhao (1993), Riskin et al. (2001), Gustafsson et al. (2008) and Li et al. (2013).13 Part of the classification in the surveys is the definition of individuals as urban and rural, distinguished according to their Hukou status. For those defined as urban, this paper identifies whether they are part of the state or market-based sector. Individuals in the urban state sector are characterised as employed in state-owned enterprises and various levels of government organisations. To be included, they must be permanent employees of an enterprise or institution (including state cadres and civil servants) or long-term contract workers. Individuals in the urban market-based sector are included if their primary workplace is either an urban collective or private enterprise, they are self-employed,14 work in a privately owned or township or village enterprise, or if they are in a state-owned, or local publicly owned enterprise but not permanently employed or in long-term contract.15 All other entries are deleted. For those defined as having rural Hukou, only those in the non-agricultural sector are used in this paper as the focus is on wages and the survey of rural 11

For CHIP 2007, we only retain the five provinces that appeared in our analysis in the first three rounds. 12 Given the structure of the census data it is not possible to construct a panel. 13 Details of the survey methods and the scope of the data are in the appendixes of the books of each survey. 14 Although employee wages and the income of the self-employed exhibit quite distinct characteristics in western developed economies, the difference in transitional China is small, especially in the presence of surplus labour. 15 Our definition of sector is different from many other studies that examine labour markets and wages but common to others that specifically study wages in China. The sector definitions here are a function of the way that wages are determined, i.e., state or market based. Thus, temporary workers in SOEs are not included in the state- sector because their wages are paid at market rates and they do not have the privileges enjoyed by permanent workers.

Labour Markets in China: A Study of the Structure and Evolution of Wages

273

agricultural income is at the household level and thus not applicable.16 In this case, the workplace is defined as being a private enterprise, a non-farming individual enterprise, a township or village enterprise, or other rural collective enterprise. For both the rural and urban samples, those still in full time education were deleted in this study and only individuals between 16 and 60 years old retained in order to concentrate on adult labour.17 Finally, in CHIP 2002 and 2007 data, a survey of rural-urban migrants is also included. This enables us to test the characteristics of those who have rural Hukou but migrate to work in an urban area. Their place of residence, rather their place of Hukou registration, is used to identify their location. All sector definitions are consistent with those used in the literature such as Meng and Zhang (2001), Knight and Song (1999, 2005) and Zhao (1999). To reconcile the four data periods, inflation rates were captured. China’s National Bureau of Statistics (NBS) publishes a number of official price indices, including national and provincial consumer price indices (CPIs), as well as separate CPIs for rural and urban areas at both national and provincial level. These price indices allow a comparison of changes in the level of consumer prices over time across different localities, but do not permit a comparison of absolute price levels between different localities. However, Brandt and Holz (2006) constructed a set of rural, urban and total provincial level spatial price deflators for the years 1984–2002, which are based on prices of a common basket of 64 goods from 10 broad categories of consumer goods and services. These were used here as they provide province level income measures adjusted for purchasing power and allow the adjustment of cross-sectional differences in prices across provinces, and between rural and urban areas within provinces.18

16

It is preferable to have both rural non-agricultural and agricultural sector included although this is not possible in these surveys. However, excluding this sector will not result in selection bias as the rural population have only very limited land. These households may make investments that increase their labour productivity on the farm but even then, they are near to subsistence. People always tend to work off the farm for a wage or migrate when the opportunity arises. 17 It is noted that for the state sector, the state retirement age for women is 55 and for men is 60. However, because of cross gender differences and variations of working age across sectors, 60 years is adopted as our data attrition benchmark. We have excluded those who are not working, so retired persons do not affect our results. 18 In few cases, when third party deflators are not available, authors’ own calculation was used.

274

3.2

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

Descriptive Data Analysis

Variables used are as follows. Income (net) is total annual gross income minus property income and transfers (gifts, government subsidies, etc.). 19 This comprises wages, bonuses and various kinds of benefits from the employer that make up total income from labour services. Incomes in 1988 are deflated by rural and urban spatial deflators across provinces. Incomes in 1995 are first deflated with the adjusted official provincial CPI’s to convert all nominal values into 1988 yuan, so that comparisons across the period are valid, and then rural and urban spatial deflators across province are used in order to make rural and urban and cross province wages comparable. This was repeated for the 2002 and 2007 data. As is common practice in the empirical literature, the top and bottom 2.5% of the ranked income series in rural and urban (and for migrants in 2002 and 2007) were deleted to exclude extreme values. Following previous studies, (Knight and Song, 2003; Appleton et al., 2005; Yang, 2005) other control variables are included, specifically, dummy variables for gender, Communist Party membership and non-Han ethnic minority status20. These are included because they may have separate effects on earnings independent of schooling and experience. For example, on average, males earn more than females and Communist Party members earn more than non-members21. Finally, non-Han ethnic Chinese tend to live in poorer provinces where wages are typically low. The CHIP 1995, 2002 and 2007 surveys report years of schooling directly while the 1988 survey only reports educational attainment in seven categories based on 19

Ideally hourly wages should be used rather than annual income because working hours per day and per month can be different across sectors. For example, migrant workers and the self-employed tend to work longer hours (or less if no work is available), and rural non-farming workers report their cash earnings from different sources for the reasons mentioned above. However, this does not create any serious problem as we are examining sectoral differentials in earning ability where people maximise their income rather than consider actual hours worked. 20 According to the China 2000 census, 91.39% of the population are Han Chinese and the rest from one of the 55 minority races. 21 Appleton et al. (2005) show that for a worker with average characteristics, earnings were 10% more in 1988 for Communist Party members, rising to 14% in 1995 and 1999. Knight and Song (2003) and Appleton et al. (2004) find trends in the rise of both gender wage differentials and a premium for Communist Party membership. These authors did not control for endogeneity of party membership. Rather, individuals with higher ability tend to be selected for Communist party membership, although no causal link is claimed in this paper.

Labour Markets in China: A Study of the Structure and Evolution of Wages

275

completion levels. To estimate returns to years of education, each completion category was matched with years of schooling.22 In the 1995 survey, years of work experience is included, but this was not available in the other surveys. This item is more problematic than the school categories, as employment and the benefits of experience are more variable, so the usual convention was followed and experience is approximated by subtracting years of schooling completed from the age of the individual, plus a further six years (the age of primary school admission in China). The number of observations in each sector, by province, plus the mean and standard deviation of yearly income from all four samples is in Table 1. In the Table, rural non-agricultural income is lower than in the urban market-based sector, which is in turn lower than the urban state sector for almost every province in all four samples. There is also an increase in the real wage for every sector and every province between 1988 and 2007. The dispersion of rural non-agricultural income across regions reflects the different levels of development of rural industry (township and village enterprises) across provinces, with the eastern provinces in general having higher levels of development in all sectors than the western regions.23 This sector also has more dispersion overall. For the total sample, the mean values show that those in the urban state sector are paid a premium over those in the market-based sector of 22% in 1988, 29% in 1995, 45% in 2002 and 24% in 2007. Across the provinces, this premium varies from 11% to 38% in 1988, from 17% to 43% in 1995, from 36% to 64% in 2002 and from 4% in Guangdong to 41% in Anhui in 2007, suggesting that central government still protects this privileged sector to a considerable extent and that income variation across provinces increases over time, although the eastern province Guangdong are less so. The 2007 data also confirms the increased bias toward the public sector and the retreat of the private sector (guojin mintui). This marks a beginning of a period in which an increased number of graduates wanted to become civil servants or join state-owned enterprises. 22

Based on previous studies on the educational system in China, for example, Yang (2005), the following assignments are made: college and above (16 years); community and vocational college (14 years); high school (12 years); lower middle school (9 years); primary school (6 years); three years or more of primary school (4 years); and less than three years of primary school (2 years). 23 In our sample, Beijing, Guangdong, Liaoning, Jiangsu are eastern, Anhui, Hubei, Shanxi are Middle, Gansu, Henan, Yunnan and Western.

Liaoning

Jiangsu

Beijing

Guangdong

1,241 574

US UM

1,727

2,067

1,171 537

639

676

203

972

240

354

842

2,053

2,596

2,401

2,194

2,916

972

1,091

960

960

1,182

194

445

301

504

329

119

586

619

917

113

384

RN

1,756

2,108

1,937

956

974

305

711

58

2,085

2,760

M

1,464

US UM

352

65

705

958

139

1,449

605

738

2,317

221

RN

60

2,119 1,630

141

1,093

337

170

107

UM

744

2,752

1,179

M

694

US

1,243

214

3,235

1,033

314

51

728

527

2,019

1,502

RN

1,650

646

632

970

4,667

Obs.

328

473

UM

1,837

711

2,072

1,085

Std. Dev.

M

1,379

US

1,317

1,661

2,678

Mean

2,074

382

631

7,831

Obs.

RN

1,639

693

Std. Dev.

2,626

3,545

UM

2,002

Mean

1995

M

13,040

US

Total

Obs.

Sector

1988

Mean Income, by Sector and Province

Province

Table 1

3,418

5,184

2,104

3,341

3,253

5,534

1,494

2,271

2,979

3,984

1,702

3,672

4,273

5,825

1,458

3,392

3,468

5,017

Mean

2002

2,281

2,353

1,419

1,746

1,791

2,581

778

1,222

1,505

1,494

1,364

1,770

2,368

2,560

1,267

1,905

2,138

2,278

Std. Dev.

2,123

823

459

152

2,830

1,278

485

257

11,883

4,186

2,036

1,033

Obs.

3,400

2,860

4,628

5,472

3,186

2,484

5,450

5,369

3,212

3,416

4,991

5,405

Std. Dev.

(To be continued)

3,762

6,701

7,766

10,963

2,779

5,781

10,329

10,739

2,792

6,084

8,285

10,248

Mean

2007

Henan

Shanxi

Anhui

Hubei

Province

968

75

1,527

2,017

783

1,010

945

1,179

167

579 167

1,546

144

1,289 2,671

267

42

508

769

27

85

RN

1,342

627

722

568

546

M

312

UM

1,731

1,415

1,469

580 1,037

373

1,619

US

38

115

1,140 2,323

RN

663

930

36

120

301

1,401

657

580

859

360

M

246

UM

1,793

1,307

1,830

530 1,038

132

1,519

US

17

208

2,627

917

110

574

RN

632

576

18

884

1,013

325

1,623

747

694

1,907

2,570

M

456

UM

2,141

1,155

149

1,038

937

161

1,178

US

19

585

649

1,594

RN

1,724

2,034

25

273

317

UM

843

Obs.

M

1,541

US

1,554

Std. Dev. 161

25

Obs.

RN

Std. Dev. 328

Obs.

1995 Mean

M

Sector

1988 Mean

1,227

3,127

3,806

5,209

1,528

3,971

2,992

4,217

849

3,325

3,467

5,229

1,024

3,363

3,136

4,664

1,892

3,463

2002 Mean

1,022

1,705

2,464

2,444

1,234

2,349

1,846

1,878

811

1,711

2,126

2,377

790

1,922

1,582

2,052

1,664

1,942

Std. Dev.

2,339

751

401

246

1,977

792

430

221

2,614

542

261

157

Obs.

3,365

3,993

4,770

5,036

3,087

4,492

4,782

5,576

2,829

3,099

4,027

5,222

Std. Dev.

(Continued)

(To be continued)

2,581

6,656

8,378

10,116

2,855

5,994

7,672

10,854

2,159

5,199

6,263

8,108

2007 Mean

573

101

10

68 1,004

1,524

2,326

428

598

915 102

381

100

5

837

530

1,067

RN

1,650

774

2,127

89

202

147

UM

2,169

2

819

557

Obs.

M

899

US

865

2,092

946

Std. Dev.

619

3,057

3,182

4,805

488

3,650

3,289

5,631

Mean

2002

657

1,723

2,436

2,213

640

2,109

1,679

2,026

Std. Dev.

Obs.

Mean

2007 Std. Dev.

(Continued)

Note: (1) Income data are annual income in 1988 real terms (yuan). They were first deflated with 1988–2007 rural and urban CPI by provinces to make rural and urban income comparable and then deflated using rural and urban spatial deflators to make provinces comparable (This applies to all the tables below). (2) US, UM, M and RN are short for Urban state, Urban market-based, Migrant and Rural non-agricultural sectors respectively.

Gansu

1,487

141

2,889

Mean

98

6

616

942

Obs.

RN

1,592

700

Std. Dev.

227

202

UM

2,203

Mean

1995

M

1,506

US

Yunnan

Obs.

Sector

Province

1988

Labour Markets in China: A Study of the Structure and Evolution of Wages

279

For the total sample, the mean value of the market-based sector premium over the rural non-agriculture sector fell from 24% in 1988 to close to parity in 1995. In some provinces, the mean income in the rural non-agriculture sector is even higher than that of the urban market-based sector, reflecting the development of rural industries and the increasing employment pressures in the urban market-based sector. However, in 2002 and 2007, the difference between the urban market-based sector and the rural non-agriculture sector increased sharply, while the wage difference is trivial between the rural-urban migrants and those in the urban market-based sector. The decrease in the rural industries (the number of township and village enterprises peaked in the mid 1990s) resulted in lower incomes and pushes wage levels in this sector towards those in the rural agriculture sector. This partly explains the increase in rural-urban migration and as discussed above, urban residents employed in the market-based sector now face huge competition from rural migrants. In particular, two patterns are worth noting with respect to the mean income of the sectors in all years for the sample provinces. First, the income ranking is the same for the three sectors with rural at the bottom for almost every province; second, the regional differences increased between 1988 and 2007. The summary statistics for all variables are in Table 2. In the Table, the mean real annual income for the whole sample increases from 1,910 yuan in 1988 to 2,538 yuan in 1995, to 3,731 yuan in 2002 and 4,227 yuan in 2007, reflecting the growth of the Chinese economy and the increase in incomes overall. However, as shown in Table 1, in some provinces the real wage in the rural non-agriculture sector actually falls. This is consistent with the increasing urban rural gap during this period. There is a more stable picture between the four samples for the other variables, with average years of schooling and experience changing very little and evidence of a slightly higher share of Han Chinese over minority communities. Communist Party membership nationally has expanded from 3.8% of the population in 1978 to 5.2% in 2002 and this sample partly reflects this trend, with a slight increase from 1988 to 1995, but a decrease in 2002. The 2007 survey did not include Communist Party membership and as our analysis only include five provinces, the share of ethnic minority population should not be interpreted literarily.24 24 It should be noted that this is only summary statistics of our sample and not all of them necessarily be representative of whole China.

280

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

Table 2

Summary Statistics

Variable

1988 Mean

1995

2002

2007

Standard Mean Standard Mean Standard Mean Standard Dev Dev Dev Dev 702 2,538 1,095 3,731 2,403 4,227 4,353

Annual Income (yuan) Years of Schooling

1,910 10.5

2.91

Years of Experience

20.5

Gender (% male)

52.4

Communist Party Member (%) Ethnic Minority (% population)

10.5

3.17

10.0

3.35

8.79 3.10

10.7

21.6

50

53.9

10.2

22.8

10.3

23.0

13.1

50

60.5

49

54

50

23.5

42

23.9

43

21.6

41

3.7

19

4.57

21

6.3

24

0.88

9.3

4

Modelling Labour Markets in China

4.1

Wage Differentials and Market Segmentation

China is undergoing a transformation from a system of centrally planned allocation of labour into a competitive labour market. To assess the extent to which China has moved in this direction Appleton et al. (2005) argues that it is necessary to have information on worker productivity to see if this is reflected in wages; evidence on barriers to job mobility; and information on whether employers are allowed to determine the optimal size and composition of their workforce. However, where such information is lacking or difficult to quantify, we can use income data to investigate changes in productivity over time. China experiences large quantity of surplus labour in both the rural and urban sectors (Knight and Song, 2005, p. 14). However, the wage behaviour in these two sectors is different. In the rural areas, wages in the non-agriculture sector were held at subsistence levels by the existence of large numbers of underemployed agricultural labourers. However, in the urban state sector, the state has prevented large scale open unemployment by controlling rural-urban migration, creating surplus jobs (Knight and Song 2005, p. 14) and providing additional subsidies to those who are employed (Wang and Piesse, 2010). These institutional factors have lead to an overall higher wage level in the urban state sector, where wages are not determined by productivity. In a well-functioning labour market, wage is determined by the productivity of

Labour Markets in China: A Study of the Structure and Evolution of Wages

281

labour, that is, the wage rate equals the value of the marginal product of labour rather than other factors. Thus, in the presence of surplus labour, the market price of labour would tend toward the subsistence level (Wang and Piesse, 2013). If the market is efficient, the wage differential between the urban market-based sector and the rural non-agricultural sector will not reflect the urban unemployment rate, higher cost of living or migration costs. In this scenario, since the surplus labour in the rural agricultural sector will work at subsistence level but are still unemployed, wage income will also reflect employment opportunities. 4.2

Estimating Equation

The work on earnings models introduced by Mincer (1974) is still regarded as the standard for both the determination of wages and the returns to education. The analysis begins with the estimation of Mincer (1974) equation for the three labour market sectors in China. It is specified 3

ln Yij = β 0 j + β1 j Sij + β 2 j Eij + β 3 j Eij2 + ∑ γ mj X mij + ε ij ,

(1)

m =1

where Yij is annual earnings for worker i in sector j, and Sij and Eij are years of education and experience, respectively. Experience is also included in a non-linear form, Eij2 , where the expected coefficient is negative, indicating diminishing return to experience. ε ij is an iid error term with E( ε ij ) = 0. The error term reflects other omitted factors in the determination of personal income, such as government regulations and personal preferences regarding labour/leisure choice. X mij is a matrix of independent variables representing characteristics that may affect personal earnings, including gender, Communist Party membership, and minority ethnic status. Later specifications also include sector and province dummies. The Mincer earnings specification has been widely used over the last five decades to estimate returns to schooling, to quality adjusted education and to measure the impact of work experience on male-female wage gaps. More recent research captures many important empirical aspects of the data, such as concavity of the logarithm of earnings, age and experience profiles, steeper profiles for persons with more years of education, and a U-shaped interpersonal variance of earnings over the life-cycle (Heckman et al., 2006). The Mincer earnings model is a hedonic wage function revealing expectations

282

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

about how the labour market rewards productive attributes, such as schooling and work experience. These two variables act as proxies for human capital and labour productivity and in a competitive market, human capital is fully reflected in earnings. However, this model was developed assuming a well-functioning labour market, where income reflects labour productivity. This paper uses the Mincer model to test the returns to education and experience in a labour market that is segmented within and between the sectors discussed above. If different groups of people are treated differently, it is expected that returns to their human capital will not be equal. Thus, in the present state of labour markets in China, the expectation is that wages, as they reflect productivity and returns to education and experience, will be different across the sectors. Further, if there are barriers to mobility between the sectors, this inequality will continue to hold, and the markets will function poorly. Only when these barriers are lifted and they are fully integrated will equilibrium be achieved. Thus, the coefficients of interest are those on schooling and experience in each of the three sectors separately and the dummy variables for these sectors and provinces. This estimation first measures the extent of segmentation between these markets and second, the degree to which integration has taken place.25 4.3

Empirical Results: Wage Structure and Market Segmentation

Table 3 reports the results of estimating equation (2), using a pooled sample of all three sectors and the sample for each sector for 1988 separately. In column 1, when only core variables are included, the findings from the Mincer model are confirmed, with positive and significant coefficients on schooling and education, and diminishing returns to experience as predicted. The coefficients are fairly small, that is around 3.3% for education and 4% for experience.26 One might expect that a greater scarcity of highly educated labour would be associated with higher returns in China. However, if there is not a competitive wage, the returns 25

While it should be noted that there is a disagreement on whether Mincer’s equation can be used to study wage gaps and market segmentation, Heckman and Hotz (2006) have used it for this purpose. Alternative methodologies have been used for testing labour market segmentation and segregation; for example Meng and Zhang (2001). 26 This is low by international standards. Mincer (1974) found the returns to an additional year of education to be between 7% and 10% in the US, although this was tested on a competitive market with free mobility of labour.

Hubei

Liaoning

Jiangsu

Beijing

Guangdong

Ethnic Minority (% pop)

CCP Member (% pop)

Male (% pop)

Experience2

Experience

Education

0.033 (33.86)*** 0.040 (49.11)*** –0.000546 (29.87)*** 0.091 (18.05)*** 0.066 (10.22)*** 0.022 (1.71)*

Base

+ Provinces 0.034 (36.30)*** 0.039 (48.98)*** –0.000517 (29.08)*** 0.091 (18.71)*** 0.063 (10.05)*** –0.012 (0.95) –0.134 (11.54)*** –0.013 (0.96) –0.014 (1.21) –0.031 (2.67)*** –0.041 (3.51)***

Pooled Sample

Estimated Earnings Equations (1988)

Regressor

Table 3 + Sectors 0.027 (28.25)*** 0.037 (48.30)*** –0.000509 (29.25)*** 0.084 (17.59)*** 0.054 (8.89)*** –0.009 (0.75) –0.118 (10.33)*** 0.005 (0.38) 0.017 (1.58) –0.008 (0.66) –0.035 (3.06)***

Urban State 0.026 (26.33)*** 0.038 (45.70)*** –0.000515 (27.32)*** 0.071 (13.95)*** 0.050 (8.10)*** –0.006 (0.43) –0.152 (12.84)*** –0.013 (0.95) 0.009 (0.80) –0.045 (3.70)*** –0.057 (4.93)***

Urban Market 0.029 (10.94)*** 0.036 (20.10)*** –0.000534 (12.97)*** 0.115 (9.83)*** 0.131 (6.40)*** 0.011 (0.36) –0.015 (0.47) 0.019 (0.46) 0.092 (3.09)*** 0.087 (2.87)*** 0.060 (1.84)*

Sector Samples

(To be continued)

Rural Non-ag 0.032 (3.41)*** 0.019 (2.59)*** –0.000110 (0.71) 0.180 (3.42)*** –0.156 (1.54) –0.393 (1.63) 0.988 (4.42)*** 1.126 (5.16)*** 0.842 (3.88)*** 1.177 (5.12)*** 0.847 (3.58)***

6.561 (478.87)*** 16772 0.3195

Base

6.620 (410.14)*** 16772 0.3624

+ Provinces –0.007 (0.61) –0.171 (14.52)*** –0.201 (17.43)*** 0.038 (3.24)*** Reference

Pooled Sample + Sectors 0.007 (0.60) –0.168 (14.60)*** –0.194 (17.21)*** 0.031 (2.64)*** Reference 0.304 (19.05)*** 0.173 (10.64)*** Reference 6.440 (310.22)*** 16772 0.3895 6.760 (396.75)*** 12889 0.3833

Urban State 0.014 (1.12) –0.183 (15.70)*** –0.208 (18.13)*** 0.025 (2.14)** Reference

Sector Samples

6.552 (156.46)*** 3509 0.2644

Urban Market 0.015 (0.48) –0.126 (3.67)*** –0.177 (5.42)*** –0.004 (0.11) Reference

Note: (1) The estimation uses Ordinary Least Squares. (2) The dependent variable = ln(income). (3) Absolute value of t statistics in parentheses, * significant at 10%; ** significant at 5%; *** significant at 1%.

Observations Adjusted R2

Rural Non-ag Constant

Urban Market

Gansu Urban State

Yunnan

Henan

Shanxi

Anhui

Regressor

5.499 (21.62)*** 374 0.1890

Rural Non-ag 0.853 (3.56)*** 0.963 (4.31)*** 1.031 (4.64)*** 1.011 (3.53)*** Reference

(Continued)

Labour Markets in China: A Study of the Structure and Evolution of Wages

285

to any economic activity will not be reflected in the wage. If the overall education level is high, there may be less variance across the different classes of education. The results also confirm the previous literature that male earnings are higher than female, and party membership has a positive impact on earnings.27 Finally, ethnic minorities are predicted to be lower down the earnings scale, but this regression shows a very small positive impact, with a low level of significance. This may be caused by the sample size that may not be representative of the population. Most of these results are supported by recent studies of wage inequality in China (Knight and Song, 2003; Appleton et al., 2005; Yang, 2005). In column 2, ten province dummies are included with Gansu as a reference group. Some of these coefficients are not significantly different from zero and when they are significant they have small coefficients, indicating that even though the reforms have been progressing, there was little to differentiate between the provinces by 1988. However, they do increase the explanatory power of the model from 0.32 to 0.36. Results from the final model estimated on the pooled data are in column 3. In addition to the province effects, the sector effects, which reflect the wage premium of the sectors, are now included, with rural non-agriculture as the reference sector. The urban state and market-based sectors have a positive and significant impact on income. Wages are higher in both these sectors compared to the rural non-agriculture sector. More precisely, as the coefficients for urban state (0.304) and market-based sectors (0.173) show, individuals in the urban state sector earn 14% more than those in the urban market-based sector in 1988 and 35.5% more than those in the rural non-agriculture sector.28 Individuals in the urban market-based sector earn 18.9% more than those in the rural non-agriculture sector. Furthermore, the explanatory power of the model has risen to 0.39. The results indicate these markets are clearly segmented.29 27 It should be noted that a positive coefficient here does not necessarily mean that communist party members have been given special treatment. Their higher income may be due to some unobserved quality that also contributed to their being party member in the first place. However, this endogeneity problem does not affect our overall result. 28 The coefficient of a continuous variable in a semi logarithmic equation can be interpreted as the percent effect of that variable on the dependent variable, while for the coefficient of a dummy variable β, the percentage effect is equal to exp(β)−1. 29 It has been argued that wage difference might also be due to sector selection, in addition to the market segmentation explanation. For example, the state-owned sector has the ability to attract more productive workers, thus even if there is no market segmentation, workers in the sector will get paid more. However, when both education and age are controlled as in our estimation, the wage difference shows exactly the sector bias and segmentation.

286

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

The remaining half of the table reports the results from estimating the same models, but at the sector level with all the province dummy variables. Here, the impact of years of schooling on wages is found to be different across sectors, with the lowest in the urban state sector, reflecting the lower effects of education on income in this sector.30 However, the higher return to experience in the urban state sector, followed by the urban market-based and rural non-agriculture sectors, represents the importance of seniority in determination of wage in this sector. Again, the returns to experience are subject to diminishing returns in all sectors. The importance of the province level effects varies. As mentioned above, wages are fixed by the state in the urban state sector and there is little variation across provinces, and the urban market-based sector seems to have little variance nationally. But there is considerable variation across provinces in the rural non-agriculture sector. The development of industry in the rural areas was extensive during the 1980s and this was very uneven across China. The result is that several of the coefficients in column 6 are statistically significant. Table 4 reports results from similar estimation using the 1995 sample. The relative values of the coefficients are unchanged, with the urban state exceeding the urban market-based and rural non-agriculture sectors with respect to education, experience and experience squared. This shows an improving contribution of education to income in almost every sector. The controls for gender and Communist Party membership are unchanged and the ethnic minority dummy is now negative but still insignificant. The last of these can be explained by the fact that ethnic minorities tend to live in poorer communities and the wages in these areas are lower than average, not just for ethnic minorities but for Han Chinese too. In column 2 of this table, it is clear that wages in Guangdong, Beijing, Jiangsu and Yunnan are significantly higher than in the other provinces, which reflects the pattern of economic development across China. From column 3, the segmentation remains in 1995, with the urban state sector dummy variables significant, although the value of the coefficient is small. The premium on wages in the urban state sector is 26.4% higher than those in the rural non-agriculture sector. The final three columns show the sector level results, with more province 30

It should be noted that coefficients on education among workers in these sectors cannot be directly interpreted as evidence of labour market segmentation, given the convexity in returns to education in China, where the urban state sector is likely to have higher levels of educated workers.

Hubei

Liaoning

Jiangsu

Beijing

Guangdong

EthnicMinority (% pop)

CCP Member (% pop)

Male (% pop)

Experience2

Experience

Education

0.038 (25.50)*** 0.036 (22.92)*** –0.000495 (14.09)*** 0.093 (10.66)*** 0.117 (10.79)*** –0.022 (1.07)

Base

+ Provinces 0.041 (27.21)*** 0.036 (23.58)*** –0.000506 (14.72)*** 0.097 (11.38)*** 0.115 (10.77)*** –0.074 (3.60)*** 0.322 (14.74)*** 0.101 (4.53)*** 0.193 (9.63)*** 0.074 (3.60)*** 0.068 (3.29) ***

Pooled Sample

Estimated Earnings Equations (1995)

Regressor

Table 4 + Sectors 0.032 (20.66)*** 0.034 (22.62)*** –0.000485 (14.39)*** 0.093 (11.10)*** 0.096 (9.17)*** –0.073 (3.63)*** 0.377 (17.48)*** 0.119 (5.46)*** 0.252 (12.68)*** 0.090 (4.45)*** 0.072 (3.59) ***

Urban State 0.033 (20.57)*** 0.036 (21.14)*** –0.000467 (12.55)*** 0.078 (8.54)*** 0.088 (8.26)*** –0.067 (3.12)*** 0.298 (12.51)*** 0.096 (4.28)*** 0.197 (9.17)*** 0.065 (3.10)*** 0.066 (3.19) ***

Urban Market 0.031 (7.24)*** 0.039 (10.81)*** –0.000679 (8.40)*** 0.158 (7.52)*** 0.114 (3.23)*** –0.104 (1.98)** 0.543 (9.39)*** 0.201 (2.77)*** 0.270 (4.90)*** 0.175 (2.99)*** 0.113 (1.85)*

Sector Samples

(To be continued)

Rural Non-ag –0.004 (0.46) 0.014 (2.01)** –0.000348 (2.13)** 0.181 (4.29)*** 0.145 (1.37) –0.153 (0.92) 0.888 (5.48)*** 0.651 (3.82)*** 0.943 (5.93)*** 0.499 (2.42)** –0.115 (0.59)

6.759 (273.92) *** 10124 0.1925

Base

6.721 (228.20) *** 10124 0.2325

+ Provinces 0.039 (1.78)* 0.001 (0.02) 0.091 (4.25) *** 0.233 (10.95) *** Reference

Pooled Sample + Sectors 0.077 (3.56) *** –0.001 (0.04) 0.111 (5.30) *** 0.236 (11.36) *** Reference 0.234 (12.71) *** 0.011 (0.56) Reference 6.623 (222.56) *** 10124 0.2656 6.751 (213.49) *** 7831 0.2301

Urban State 0.092 (3.96) *** 0.004 (0.20) 0.113 (5.18) *** 0.216 (10.17) *** Reference

6.685 (80.39)*** 1661 0.2276

Urban Market 0.106 (1.82)* –0.043 (0.68) 0.197 (3.22) *** 0.306 (4.85)*** Reference

Sector Samples

Note: (1) The estimation uses Ordinary Least Squares. (2) The dependent variable = ln(income). (3) Absolute value of t statistics in parentheses, * significant at 10%; ** significant at 5%; *** significant at 1%.

Observations Adjusted R2

Rural Non-ag Constant

Urban Market

Gansu Urban State

Yunnan

Henan

Shanxi

Anhui

Regressor

7.491 (20.28)*** 632 0.2979

Rural Non-ag 0.131 (0.75) 0.173 (0.96) 0.349 (2.11)** 0.905 (2.32)** Reference

(Continued)

Labour Markets in China: A Study of the Structure and Evolution of Wages

289

effects (which shows that wages are more dispersed regionally when reform deepens), with the exception of the rural non-agriculture sector where education, experience and Communist Party membership are also unimportant. Table 5 reports results from similar estimation using the 2002 sample. As in the previous tables, columns 1 to 3 report the regression results on the pooled sample, with column 2 including the province dummy variables and column three the province and sector dummies. The addition of sectoral dummies increases the explanatory power of the model significantly to 0.497. In column 3, there are coefficients on three sectors with rural non-agriculture as the reference group. The wage level in the urban state sector is still higher than the other sectors. However, the coefficient on the urban market-based sector is less different from that with migrants. Since rural migrants in the urban sector also enter the market-based sector, the two coefficients are not measuring two different sectors but two groups of people in the same sector. It is clear that, on average, migrants earn more than urban residents in the market-based sector.31 This might be because the migrants work longer hours and are employed in less favourable activities so these two groups are complementary in many cases, that is, migrants do the jobs that local non-migrants will not or cannot do. The striking fact is that now there is a big gap between the rural sector and other sectors. Migrants earn as much as 160% more than people in the rural non-agricultural sector. This is probably due to factors that not included in the model, as discussed in the theory section. For example, higher cost of living and migration have to be compensated and people in the rural non-agricultural sector may only work part time, or only part of the year, since the rural industries were reduced significantly after the peak of the mid 1990s in most provinces. The lack of jobs and the low wages in the rural sector also contributes to the large migration flows to the urban areas. Table 6 reports results from the 2007 data in a similar way to the previous tables. As shown in column 3, the addition of sectoral dummies increases the explanatory power of the model. It may be true that many urban state sector workers have higher education levels and thus their higher wage may be the result of their education and skills. Our regression results show that there is still a 31

It should be noted that two of the principal scholars involved in the CHIPs survey (Li Shi and Meng Xin) have been critical of the migrant data that they collected. They acknowledge that the CHIPs sample provides a poor representation of migrants.

Hubei

Liaoning

Jiangsu

Beijing

CCP Member (% pop) Eth-Minority (% pop) Guangdong

Male (% pop)

Experience2

Experience

Education

0.110 (40.54)*** 0.049 (16.85) *** –0.000872 (14.28) *** 0.024 (1.48) 0.132 (6.39) *** 0.055 (1.74)*

Base 0.113 (41.50) *** 0.050 (17.09) *** –0.000873 (14.38) *** 0.032 (2.00)** 0.125 (6.09) *** 0.015 (0.45) 0.163 (4.51) *** –0.126 (3.17) *** 0.193 (5.39) *** 0.176 (4.92) *** 0.041

Pooled Sample + Provinces

Estimated Earnings Equations (2002)

Regressor

Table 5

0.054 (20.47) *** 0.028 (11.91) *** –0.000468 (9.61) *** 0.158 (12.40) *** 0.133 (8.02) *** 0.040 (1.53) 0.348 (11.99) *** –0.009 (0.27) 0.356 (12.43) *** 0.199 (6.95) *** 0.079

+ Sectors 0.063 (23.04) *** 0.024 (8.89) *** –0.000258 (4.47) *** 0.100 (7.54) *** 0.085 (6.09) *** 0.017 (0.57) 0.221 (6.92) *** –0.185 (5.98) *** 0.137 (4.62) *** 0.072 (2.50)** –0.012

Urban State

Sector Samples Urban Rural Market Migrant 0.032 0.028 (5.44) *** (6.66) *** 0.007 0.026 (1.37) (7.21) *** –0.000071 –0.000571 (0.65) (7.55) *** 0.277 0.212 (10.48) *** (11.12) *** 0.060 –0.067 (1.31) (1.22) –0.055 –0.038 (0.87) (1.20) 0.322 0.198 (5.27) *** (4.67) *** –0.041 –0.329 (0.58) (6.69) *** 0.086 0.116 (1.37) (2.71) *** 0.084 0.134 (1.44) (3.17) *** 0.022 0.071

(To be continued)

Rural Non-ag 0.068 (6.45) *** 0.033 (4.29) *** –0.000675 (4.55) *** 0.221 (4.18) *** 0.349 (5.63) *** 0.574 (4.81) *** 1.085 (9.56) *** 1.076 (8.21) *** 1.503 (13.31) *** 0.734 (5.25) *** 0.567

6.221 (133.07) *** 10862 0.1933

Base

6.090 (112.90) *** 10862 0.2082

(1.12) 0.135 (3.55) *** –0.089 (2.44)** 0.153 (4.32) *** 0.209 (5.47) *** Reference

Pooled Sample + Provinces (2.69) *** 0.134 (4.40) *** 0.122 (4.19) *** 0.243 (8.57) *** 0.159 (5.21) *** Reference 1.407 (71.02) *** 1.124 (49.99) *** 1.258 (65.41) *** Reference 5.691 (122.50) *** 10862 0.4972

+ Sectors

7.120 (132.61) *** 4667 0.2100

(0.43) 0.073 (2.32)** –0.098 (3.44) *** 0.126 (4.46) *** 0.209 (7.19) *** Reference

Urban State

7.303 (69.21) *** 1502 0.1196

7.304 (108.81) *** 2626 0.1489

Sector Samples Urban Rural Market Migrant (0.32) (1.63) 0.113 0.146 (1.64) (3.44) *** 0.001 0.027 (0.02) (0.54) 0.230 0.275 (3.59) *** (6.42) *** 0.147 0.194 (1.95)* (4.27) *** Reference Reference

Note: (1) The estimation uses Ordinary Least Squares. (2) The dependent variable = ln(income). (3) Absolute value of t statistics in parentheses, * significant at 10%; ** significant at 5%; *** significant at 1%.

Observations Adjusted R2

Rural Non-ag Constant

Migrant

Urban Market

Gansu Urban State

Yunnan

Henan

Shanxi

Anhui

Regressor

4.851 (29.89) *** 2067 0.2349

Rural Non-ag (4.51) *** 0.366 (2.80) *** 1.013 (9.12) *** 0.789 (6.83) *** –0.529 (3.57) *** Reference

(Continued)

Henan

Hubei

Anhui

Jiangsu

Eth-Minority (% pop) Guangdong

Male (% pop)

Experience2

Experience

Education

(0.16) 0.068

–0.013 (0.94)

–0.025 (1.72)*

Reference

Reference

(11.44) ***

(12.81) ***

(0.47) –0.183

(0.21) –0.211

–0.007

–0.003

Reference

(5.04) ***

–0.290

(0.66)

0.034

0.058 (1.00)

0.040 (2.72) ***

0.004 (0.23)

(1.35)

–0.025

(0.25)

–0.009

(0.93)

–0.000133

(1.26)

0.009

(7.65)

***

0.046

Urban State

–0.090

(0.72)

(0.47)

(15.01) ***

0.144

(15.46) ***

–0.000513

(13.07)

***

0.020

(12.23)

***

0.024

+ Sectors

(1.99)**

–0.034

–0.022

0.111 (11.35) ***

0.113

(16.25) ***

(11.45) ***

–0.000554

(16.54) ***

(16.45)

***

0.025

(30.04)

***

0.051

+ Provinces

–0.000568

(17.02)

***

0.026

(29.89)

***

0.051

Base

Pooled Sample

Estimated Earnings Equations (2007)

Regressor

Table 6

0.074

Reference

(6.62) ***

–0.313

Reference

(6.54) ***

–0.162

(1.89)

*

(2.49)

0.043 **

(3.40) *** –0.103

(1.63)

–0.066

–0.116 (5.85) ***

0.223

(0.03)

0.001

(10.74) ***

0.151

(13.48) ***

–0.000667

(14.43)

***

0.030

(7.83)

***

0.024

(5.55) ***

(1.69)*

–0.179

(2.28)**

0.060

(0.19)

0.000020

(1.26)

–0.006

(7.38)

***

0.032

Sector Samples Urban Rural Market Migrant

(To be continued)

Reference

(6.55) ***

–0.156

(0.81)

–0.020

(2.88) ***

0.066

(0.88)

–0.020

(1.42)

–0.215

(13.75) ***

0.207

(10.17) ***

–0.000511

(6.98) ***

0.016

(4.16) ***

0.015

Rural Non-ag

12882 0.1137

12882 0.0973

7.971 (311.67) ***

7.916

+ Provinces

(337.85) ***

Base

Pooled Sample

0.1654

12882

(299.40) ***

8.150

Reference

(11.26) ***

0.125

(16.36) ***

0.251

(27.13) ***

0.545

+ Sectors

0.0909

1030

(70.53) ***

8.495

Urban State

0.1210

2022

(97.29) ***

8.611

0.1174

3914

(207.07) ***

8.183

Sector Samples Urban Rural Market Migrant

Note: (1) The estimation uses Ordinary Least Squares. (2) The dependent variable = ln(income). (3) Absolute value of t statistics in parentheses, * significant at 10%; ** significant at 5%; *** significant at 1%.

Observations Adjusted R2

Rural Non-ag Constant

Migrant

Urban Market

Urban State

Regressor

0.0797

5909

(188.66) ***

8.248

Rural Non-ag

(Continued)

294

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

large proportion of the gap that could not be explained by education and other listed variables. As can be seen from the sector samples, education continued to have a small role in explaining wage earnings, with the lowest wages being in the rural non-agriculture sector. Being a male is important in the rural non-agriculture sector but makes no difference in urban state sector. Sectoral differences still exist but are to a lesser extent compared with the previous rounds. It should be noted that, on average, and this is different from 2002, migrants now earn less than those in the urban market-based sector. The spatial difference at the provincial level also decreased. 4.4

Empirical Results: Convergence and Market Integration

If migrants optimise their personal welfare, the decision to migrate is a function of the difference in expected wage rates between sectors, weighted by the appropriate regional costs of living, plus the social costs. If there is a well-functioning labour market, for workers with similar education level and experience, wages would be equal across sectors and regions and in equilibrium there are zero net benefits from migration. Thus, the wage level of workers with similar education levels and experience in different sectors in the same region can be used to test for the degree of labour market integration across sectors. Equally, the wage level in different regions can be used to test the degree of labour market integration across regions and sectors.32 The results for the 1988, 1995, 2002 and 2007 samples indicate market segmentation at different levels. However, from these results, it is reasonable to suggest that at least two of these sectors are converging, given some integration has taken place. Examining the differences between the samples in 1988 and 1995 shows that the urban market-based and the rural non-agriculture sectors are converging as a result of labour mobility and that in the latter period wages are determined on a more 32

If different wages are paid to workers with the same characteristics except for the places where they are working, markets segmentation must exist. Without such segmentation, workers would migrate to the higher paid sector. See Knight and Song (2005) for further discussions on this. Of course, certain industrial sectors may benefit unevenly from technological change and thus able to enjoy higher wages as a result of biased technological change. However, it should be noted that the “sectors” in this paper are not technologically defined “industrial sectors” but are rather the “artificial” sectors created by government policy.

Labour Markets in China: A Study of the Structure and Evolution of Wages

295

competitive basis than previously. From 1988 to 1995, an analysis of current policy shows that the restrictive control of labour migration from the rural to the urban market-based sector has been loosened, but there are still very strict controls over migration towards urban state sector employment. If the control of migration is lessened, labour productivity should rise, as wages begin to match labour effort and there is a competitive market for labour that takes account of specific skills and experience. This will be achieved as convergence between the two market-based sectors continues. Two main conclusions arise from a comparative analysis of the sectors. Firstly, the coefficients on the sector dummies have fallen from 0.304 in 1988 to 0.234 in 1995 in the urban state sector and from 0.173 in 1988 to 0.011 in 1995 in the urban market-based sector. The gap between the urban market-based sector and the reference group has diminished significantly and had become almost identical by 1995. This implies that the premium paid to urban market-based sector employees has been decreasing. Secondly, the difference between the urban state and the urban market-based sectors was increasing. This is because the urban state sector is protected from competition, and wages had risen as labour mobility into that sector remained restricted.33 However, labour in the urban market-based sector is competing with migrants from the rural non-agriculture sector, where reform has loosened the constraints on movement both between the rural and urban sectors, and between provinces. This is supported by the returns to education, which are lower in both the urban private and rural non-agriculture sectors. The oversupply in the former has resulted in a fall in wages, and in the latter, the individuals with less education and experience are the group that remain in the rural areas. Thus, there is evidence of integration between some labour markets but others remain segregated. These empirical results confirm those of Johnson (2003) that cross-province migration in China is well below the level consistent with fully-integrated labour markets. However, when the sub-divisions in the urban sector are taken into account, and the impact of migration from the rural to the urban market-based but not to the urban state sector is clear, the real determinants of wages in these labour markets becomes 33 It is noted that lower education levels may also contribute to the explanation of why most migrants cannot be employed by the urban state sector (see Chen and Hamori, 2009). However, strict control such as Hukou exacerbates both job and wage discrimination against migrants (Knight and Song, 2005).

296

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

apparent. This is confirmed in the 2002 sample when migrant data are available. It is shown that the wage income for urban market-based and rural migrants is almost the same. Migrants may earn higher wages because of longer working hours and harsh working conditions. However, while the urban market-based sector is integrating rural migrants, the urban state sector remains segmented. As shown in table 6 and as discussed above, in 2007, both spatial and sectoral segmentation were reduced. A couple of reasons may be behind this. First, after China joining the WTO at the end of 2001, there has been a significant increase in demand of workers in eastern and urban areas. To response to this demand, the implementation of the Hukou system has been relaxed. Labour mobility has increased (Meng, 2012; Appleton et al., 2014; Xia et al., 2014). Second, China has reached the first Lewis turning point and the workers’ wages have been increasing steadily since around 2005 (Wang and Weaver, 2013). Third, from 2004 to 2006, the agricultural tax has been abolished and some basic healthcare and pension schemes have been introduced in rural areas. These reforms have greatly increased the welfare of rural residents and positively affected the wages of those in the rural sector (Cai, 2011; Wang et al., 2013). Fourth, there may have some unreported grey income for the residents in the urban state sector. If the trillions of Renminbi of grey income is included (Wang, 2010), the gap would significantly be higher than what is shown here (see Wang, 2010). 4.5

Discussion

OLS estimates used in this paper do not account for population heterogeneity, quality of education and variation in levels of school attendance that can lead to a bias downwards in the estimation of rates of returns to education (Heckman and Li, 2004; Yang, 2005). The average return to education in China is low by international standards, although with considerable variance and despite significant increases since the reforms, as long as labour markets remain segmented, there is little incentive to invest in education (Yang, 2005). A number of factors can be important in explaining this. Labour mobility within the urban state sector is very low and the wage is pre-determined. This high level of compensation is a function of political affiliation and family connections, rather than education and experience. In the urban market-based and rural non-agriculture sectors there is a large proportion of low skilled jobs that do not

Labour Markets in China: A Study of the Structure and Evolution of Wages

297

require highly educated employees. One interpretation relies implicitly on human capital theory, so there is a potential problem in that education and experience are incomplete measures of labour quality. It is possible that unmeasured aspects of labour quality, for example, overall ability and motivation are more important influences on earnings, and are positively associated with high wages. Effects, such as those predicted by efficiency wage and agency models, suggest that higher pay in a sector delivers a gain in productivity that exceeds the incremental cost of wages, partly because where monitoring is more difficult, higher pay gives workers more incentives. Other research shows a positive relation between the size of the firm and the wage rate and the state sector enterprises are large, which may also account for the higher salaries. Another explanation of the higher wage in the urban state sector may reflect the reality that the Chinese state still owns a large proportion of the previous State-owned assets, and has still monopolised the most profitable economic sectors. Thus, the high wage premium can result from such a monopoly. However, for whatever reason, the segmentation is obvious. With respect to the overall low returns to human capital in China, it could be argued that this reflects the results of studies from other developing countries, which find that schools are less effective in imparting cognitive skills to their students than those in developed countries, resulting in lower labour productivity and hence a lower correlation between years of education and income (Glewwe, 2002). One unresolved question in the context of China is that while workers in all sectors attend similar schools, wages are different because of their inclusion in a particular sector. Thus, education and experience is less important than the status of the workers and the Hukou they have obtained. In the analysis in this paper, the restrictions on mobility into the urban state sector were attributed to the increase in the wage premium relative to the other sectors. However, there may be other causes. For example, the official wage policy in state owned enterprises is to raise the pay of employees who have not been retrenched in order to gain their support for the layoffs of their peers. This will not show up in the data used in this paper as the retrenched workers will no longer be in the urban state sector so the mean value does not fall when the unemployed leave the sector. However, this does not contradict the basic premise that sectoral segmentation remains very strong in China. Thus, while it cannot be claimed with certainty that restrictions on inward mobility cause wage

298

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

differentials between the urban state and other sectors to continue, it is the case that restrictions prevent market forces acting to eliminate these differences. Indeed, loosening the restrictions and allowing migrants to move to the urban market-based sector has enhanced convergence and integration of the urban private and rural non-agriculture sectors.

5

Conclusion

Increasing inequality in both the level and the growth of incomes has been driven by two major factors: the urban-rural income gap and inter-province disparities. Many studies have confirmed the uneven development of regions. However, few studies have addressed inter-province rural and urban and intra rural and urban labour market segmentation. The paper uses Mincer’s earnings model to study the structure and evolution of wages which could inform us about the extent of labour market segmentation and integration, using Chinese Household Income Project data for 1988, 1995, 2002 and 2007. The results show that labour markets in China are characterised as being not only regionally different but also segmented by sector. People in different sectors are paid not according to their productivity but the sector that they are in. This is clearly established in the 1988 sample but by 1995 some evidence of integration is emerging. However, in the 2002 sample, although the integration of the urban market-based sector and migrants from the rural private sector is clear, those that remain in the rural sector are worse off. In 2002, the gap between the urban and rural sectors deepens significantly as the rural population have not benefited from the reforms. In 2007, due to various factors such as the improvement of social welfare in the rural areas and the loosening of controls on labour mobility associated with the Hukou system, there has been a slow trend of both sectoral and spatial integration. Despite major improvements in the functioning of labour markets following the reforms in China, serious obstacles to a fully integrated national market remain. Due to the combination of the unique Hukou system and the quota system, the labour market in the state sector is shielded from competition by others.

Labour Markets in China: A Study of the Structure and Evolution of Wages

299

Acknowledgements The authors would like to thank Shi Li and Chuliang Luo for the provision of CHIPs data sets and Jeremy Clegg, John Knight, Xiaming Liu, Paul Ryan, Roger Strange and Nick Weaver for their comments on the early drafts of this paper, and Gary Fields for discussions on the use of the Mincer equation. We thank Rong Huang and Mengbing Zhu for their assistance on processing the CHIP 2007 data. We also thank four anonymous reviewers and seminar participants at King’s College London and the University of Manchester for valuable comments and suggestions. The research was supported by Program for Innovative Research Team of Shanghai University of Finance and Economics “IRTSHUFE.” An early version of this paper was circulated as “labour market segregation and integration in China: A spatial and sectoral analysis, 1988–2002”.

References Appleton S, Knight J, Song L, Xia Q (2004). Contrasting paradigms: Segmentation and competitiveness in the formation of the Chinese labour market. Journal of Chinese Economic and Business Studies, 2(3): 185–205 Appleton S, Song L, Xia Q (2005). Has China crossed the river? The evolution of wage structure in urban China during reform and retrenchment. Journal of Comparative Economics, 33(4): 644–663 Appleton S, Song L, Xia Q (2014). Understanding urban wage inequality in China 1988–2008: Evidence from quantile analysis. World Development, 62(1): 1–13 Brandt L, Holz C A (2006). Spatial price differences in China: Estimates and implications. Economic Development and Cultural Change, 55 (1): 43–86 Cai F (2011). Hukou system reform and unification of rural-urban social welfare. China & World Economy, 19(3): 33–48 Chen G, Hamori S (2009). Solution to the dilemma of the migrant labor shortage and the rural labor surplus in China, China & World Economy, 17(4): 53–71 De Brauw A, Huang J, Rozelle S, Zhang L, Zhang Y (2002). The evolution of China’s rural labour markets during the reforms. Journal of Comparative Economics, 30(2): 329–353 Dong X, Bowles P (2002). Segmentation and discrimination in China’s emerging industrial labour market. China Economic Review, 13(2–3): 170–196 Glewwe P (2002). Schools and skills in developing countries: Education policies and socioeconomic outcomes. Journal of Economic Literature, 40(2): 436–482 Griffin K, Zhao R (1993). The Distribution of Income in China. New York, NY: St. Martin’s Press, Inc. Gustafsson B, Li S, Sicular T (2008). Income Inequality and Public Policy in China Inequality and Public Policy in China, New York, NY: Cambridge University Press Heckman J J, Hotz V J (1986). An investigation of the labor market earnings of Panamanian males evaluating the sources of inequality. The Journal of Human Resources, 21(4): 507–542 Heckman J J, Li X (2004). Selection bias, comparative advantage and heterogeneous returns to

300

Xiaobing Wang, Jenifer Piesse, Zhengmao Ye

education: Evidence from China in 2000. Pacific Economic Review. 9(3): 155–171 Heckman J J, Lochner L, Todd P T (2006). Fifty years of Mincer earnings regressions. In: Hanushek, E. and F. Welch (eds.), Handbook of Education Economics, Vol. 1. Amsterdam, The Netherlands: Elsevier Johnson D G (2003). Provincial migration in China in the 1990s. China Economic Review, 14(1): 22–31 Knight J, Song L (1999). The Rural-Urban Divide: Economic Disparities and Interactions in China. Oxford, UK: Oxford University Press Knight J, Song L (2003). Increasing urban wage inequality in China. Economics of Transition 11(4): 597–619 Knight J, Song L (2005). Towards a Labour Market in China. New York, NY: Oxford University Press Li S, Sato H, Sicular T (2013). Rising inequality in China: Challenges to a harmonious society. New York, NY: Cambridge University Press Meng X, Zhang J (2001). The two-tier labour market in urban China: Occupational segregation and wage differentials between urban residents and rural migrants in Shanghai. Journal of Comparative Economics, 29(3): 485–504 Meng X (2012). Labor market outcomes and reforms in China. Journal of Economic Perspectives, 26(4): 75–102 Mincer J (1974). Schooling, Experience, and Earnings. New York, NY: NBER Press National Bureau of Statistics of China (NBS, 中 国 国 家 统 计 局 ) (2012). China Statistic Yearbook (various years) (中国统计年鉴). Beijing, China (中国北京): China Statistics Press (中国统计出版社) Riskin C, Zhao R, Li S (2001). China’s Retreat from Equality: Income Distribution and Economic Transition. Armonk, NY: M. E. Sharpe, Inc. State Council of China (中国国务院) (2006). The Research Report on Chinese Rural-Urban Migrants (中国农民工调研报告). Beijing, China (中国北京): China Yan Shi Press (中国 言实出版社) Wang X, Piesse J (2008). Labor market segregation and integration in China: A spatial and sectoral analysis, 1988–2002 (Whither Chinese Labour Market: A Study of Wage Structure Evolution in China). KCL Department of Management Research Paper Series, No. 51 Wang X, Piesse J (2010). Inequality and the urban-rural divide in China: Effects of regressive taxation, China and World Economy, 18(6): 36–55 Wang X, Piesse J (2013). The micro-foundations of dual economy models, The Manchester School, 81(1): 80–101 Wang X, Piesse J, Weaver N (2013). Mind the gaps: A political economy of the multiple dimensions of China’s rural-urban divide. Asian Pacific Economic Literature, 27(2): 52–67 Wang X, Weaver N (2013). Surplus labour and Lewis turning points in China. Journal of Chinese Economic and Business Studies, 11(1): 1–12 Wang X (王小鲁) (2010). Off-the-books income and national income distribution (灰色收入 与国民收入分配). Comparison (比较), Volume 48, Beijing, China (中国北京): Zhongxin

Labour Markets in China: A Study of the Structure and Evolution of Wages

301

Publishing house (中信出版社) Xia Q, Song L, Li S, Appleton S (2014). The effect of the state sector on wage inequality in urban China: 1988−2007. Journal of Chinese Economic and Business Studies, 12(1): 29–45 Yang D T (2005). Determinants of schooling returns during transition: Evidence from Chinese cities. Journal of Comparative Economics, 33(2): 244–264 Zhang J, Zhao Y, Park A (2005). Economic returns to schooling in urban China, 1988 to 2001. Journal of Comparative Economics, 33(4): 730–752 Zhao Y (1999). Leaving the countryside: Rural-to-urban migration decisions in China. American Economic Review, 89(2): 281–286