Labour Market Informality and Economic Transition: Employment Regulation and Adjustment to Economic Crisis in China

Transcription

1 Labour Market Informality and Economic Transition: Employment Regulation and Adjustment to Economic Crisis in China A Proposal on Understanding Labor Market Informality in Developing Countries to the World Bank s Social Protection & Labor Unit Fang Cai Director, Institute of Population and Labor Economics (IPLE), CASS Albert Park Reader in the Economy of China, University of Oxford April 6,

2 Labour Market Informality and Economic Transition: Employment Regulation and Adjustment to Economic Crisis in China 1. Motivation and background This proposal outlines a research plan to increase understanding of labour market informality in China, with particular emphasis on the role of informality in adjustments to large economic shocks and the impact of labor regulation on informal employment. The case for studying the informal labor market in China is compelling. By nearly any measure, China is the world s largest transition economy and the world s largest developing country. Its remarkable development record over the past three decades of reform has increasingly made it a model for other developing and transition economies, especially in Africa. China stands at a critical moment in her economic transition, with the role of the informal sector featuring prominently in debates over the future direction of China s labour market development. The forces at work in China are similar to those in many other transition and developing countries but are operating with magnified speed and momentum, so that improved understanding of informality in China may provide key lessons and insights for other countries. These forces include rapid informalization of the urban labor market due to the decline of the state sector and limited success in establishing effective new social insurance programs; rapid structural change and urbanization accompanied by large-scale internal migration of rural labor to urban areas; and passage of ambitious new labor regulations designed to protect workers from unfair treatment and provide them with greater economic security. On top of all this, the ongoing global economic crisis has put tremendous pressure on the system to respond and adjust to an unforeseen sharp drop in labor demand driven mainly by significant reductions in exports. The crisis provides an unusual opportunity to study how firms and government respond to adversity and what it means for workers and their families, and for structural adjustment, economic growth, and poverty. Despite the great importance of research on these issues, there is a glaring lack of research and data sources for studying China s informal labor market. The few studies of informal employment in China of which we are aware, mostly completed by members of this project s research team, are primarily descriptive due to data limitations, and focus on defining and documenting the extent of informal employment (e.g., Cai and Park, 2008; Wu and Cai, 2006). The household and firm surveys conducted by China s National Bureau of Statistics have broad coverage, but are not publicly available, do not survey migrant households who comprise a large share of the informally employed, and have limited coverage of labor regulations, contracts, and employment conditions. Household surveys organized by academic researchers share one or more of these weaknesses. 1 No firm surveys of which we are aware contain sufficient detail on employment and labor contracting to allow for systematic analysis of the impact of labor regulations, nor are we aware of any completed or planned firm surveys to be conducted in the post-crisis period. As a result of these limitations there is very little existing knowledge about any of the research questions that will be addressed by the proposed research project. The data produced by the project will eventually be publicly released, providing a valuable resource for researchers. This proposal outlines a plan for data collection and analysis that addresses the shortcomings describe above in order to answer a series of research questions related to two broad topic areas: the role of 1 For example, the China Health and Nutrition Survey has a very small sample of migrants and little information on employment relations; the China Income Project Survey (CHIPS) data was collected in collaboration with NBS, focuses on income and expenditures with little detail on labor contracts and employment conditions, surveyed migrants in 2002 in a way that is not comparable to the NBS rural and urban household survey data, and has no completed waves after the crisis; and the recently begun Rural-Urban Migration in China and India survey shares the weaknesses of the CHIPS data. 2

3 informality in China s response to the economic crisis, and the impact of labor regulation on informal employment. Before presenting the specific research questions to be examined, in the remainder of this section we briefly describe in greater detail the key forces influencing labor market informality. Increasing Informalization. From 1995 to 2005, the percentage of China s urban labour force that was undocumented (not officially registered as employees of formal sector work units or as selfemployed workers) increased from 9 percent to 36 percent (Park and Cai, 2008). This appears as the other category in Figure 1, which also illustrates the rapid decline of the state and collective sectors and the rapid rise of the private sector. Adding the 10 percent of the urban workforce registered as self-employed to the undocumented category in 2005, all told nearly half of the urban labour force was employed informally. This dramatic informalization of the Chinese labour market was the result of several major changes: the rapid increase in rural-urban migration, massive layoffs of state-sector workers in the late 1990s, and fast growth of unregulated private sector activity, especially in services (Cook, 2008; Du, Cai, and Wang, 2007). Part of the rise in informal sector employment was likely due to the exclusion of rural migrants from formal jobs in many sectors, as China has maintained a residence permit system that formally excludes rural residents from job access and benefits available to urban residents. Informal employment thus has undoubtedly played an important role in expanding employment opportunities (Cai and Wang, 2007). However, evidence also suggests that a significant minority of those working in state- and collectively owned firms are employed informally (not registered and without a formal labour contract). Non-registered employment is the rule rather than the exception for private firms. Firms of all ownership types are motivated to keep workers off the Figure 1: Urban employment shares by ownership, State-owned Collective Joint, share Foreign Private, selfemployed Other books to avoid high required social insurance contributions and to allow for flexibility in adjusting their labour force (Park and Cai, 2008). Enforcement is often lax, partly because local governments fear scaring away potential investors. Labor Regulation and the 2008 Employment Contract Law. China passed a landmark Labor Law in 1994, formalizing for the first time the system of labor contracting and (Cai, Du, and Wang, 2009). Since then, other incremental reforms have been passed. But none of these was as important as China s new Labour Contract Law which came into effect on January 1, It increases protections for workers, strengthening rights to collective bargaining, requiring severance payments be paid when contracts are not renewed, and, most controversially, requiring that workers with 10 consecutive years of service, or having signed two consecutive fixed-term contracts with a company, be given a contract without a fixed end date. Many employers have complained that the new law will significantly raise labour costs and limit employment flexibility. At the same time, the Labour Contract Law contains considerable ambiguity. Thus, many questions remain about implementation and enforcement of the new law, and about the Law s actual impacts on the welfare of workers and the productivity of enterprises. In recent years, China also has implemented other reforms affecting the treatment of labor, including a comprehensive document to protect the rights of migrant workers (2006), new rules 3

4 to improve labor dispute resolution, and new social insurance programs to make migrants benefits more portable and to extend health insurance and pension coverage to the self-employed and nonemployed. Because the changes are so recent, little research has assessed their impacts on how firms treat workers and on the welfare of different types of workers. Migration. The last decade has witnessed an explosion of rural-urban migration in China. According to one monitoring survey, the number of rural migrants increased from 39 million in 1997 to 103 million in 2004 (World Bank, 2009). Data from the most recent NBS national rural household survey finds that rural migrants numbered 137 million in 2007, with up to ninety percent migrating to urban areas, where they form a large constituency within the growing informal sector. The large inflow migrants present challenges to urban governments, who on the one hand want to attract migrants to work in local factories and provide needed services, but on the other hand lack financial resources to extend equal benefits to migrants. Economic crisis. Like the rest of the world, China has been negatively impacted by the current global economic crisis. Because China never liberalized its capital account, Chinese financial institutions did not own any of the toxic assets that brought down banks in other countries and remained healthy. The main negative effects of the crisis have been through a sharp reduction in the demand for Chinese exports as global GDP may have fallen worldwide by 5% in the last quarter of 2008 and another 5% in the first quarter of 2009 (World Bank Beijing Office, 2009). Chinese exports fell sharply beginning in November 2008, falling by 2.2% year-on-year in comparison to a 19.1% increase in October. By February 2009, exports had declined by 25.7% year-on-year and imports had fallen by a similar amount (World Bank Beijing Office, 2009). In China, production of export goods is highly concentrated in the rich provinces of coastal China, where millions of migrants from rural areas throughout the country flock to work in labor-intensive manufacturing firms. There has been much public concern about the consequences of the crisis for labor demand and employment, especially of migrant workers but also of recent college graduates and local workers in urban areas. Recently, a top government economic official estimated that 15.3 percent of China s 136 million migrant workers (20.8 million persons) had lost their jobs due to the crisis (Chen, 2009). However, there is as yet little hard data to measure the employment impacts of the crisis on the urban workforce. One recent (undocumented) firm survey in China conducted in March 2009 by a consulting firm found that 21% of firms planned to cut employment in the second quarter of 2009 while only 8% planned to hire new workers; nearly two thirds of firms had reduced human resource-related costs in the previous six months by reducing workers, bonuses, etc. (Li, 2009). 2. Research questions The economic crisis raises a number of important questions regarding the role of the informal sector. First, do informal workers bear the brunt of adjustment costs as firms downsize in response to falling labor demand? Second, does the informal sector, especially through self-employment opportunities, provide a cushion, or alternative employment option, for dislocated workers from both the formal and informal sectors? Third, what is the extent of vulnerability of workers, both formal and informal, who are adversely affected by negative economic shocks? How well do social insurance and social assistance programs as well as private transfers support the consumption of families suffering from temporary hardship? We are particularly interested in the impact of the crisis on migrant workers, who comprise a large part of informal employment and who are less protected by institutions and policies in China. Another set of key questions are related to the impact of labor regulation on informal employment and the welfare of workers. On the one hand, labor regulations are meant to protect the interests of workers by providing greater economic security through access to social insurance programs, by stipulating rules for compensation when workers are dismissed, and providing dispute resolution mechanisms so that workers are able to air grievances. On the other hand, excessive labor regulation can increase the costs of employers, which can lead them to reduce employment or evade regulations 4

5 by hiring workers informally. Conceptualizing informality. In China s urban areas, most employment had been formal in nature until recently when labour mobility (and migration) increased and aggressive economic restructuring dismantled the iron rice bowl enjoyed by workers in state-owned and collective enterprises. Thus, informality is a relatively new phenomenon in China. In the proposed research, we define informal workers to include the self-employed, workers in informal, unregulated businesses, and workers employed informally (casually, temporarily, off the books, with no access to social insurance programs) in formal work units that in principle should be subject to labour regulations. Accurately measuring informal employment in survey research is challenging, because simple indicators such as whether workers have signed a labor contract may not fully capture whether work is formal. In a recent paper, two of the project researchers identified nine criteria for identifying informal employment analyzing a rich survey dataset collected by China s Ministry of Labor in 66 large cities (Wu and Cai, 2006). In addition, research team members (Sarah Cook, Cai Fang and other IPLE researchers) have successfully applied for a project supported by IDRC (Canada) to collaborate with Indian researchers with lengthy experience studying the informal sector in India to apply relevant lessons for measuring informality in household surveys in China. These innovations will be incorporated into the proposed data collection effort and should enable us to make progress refining measurement concepts for studying informality in developing countries. Keeping these broader motivations in mind, the proposed project will focus on the following six specific research questions: 1. What were the nature and magnitude of negative economic shocks associated with the economic crisis? 2. Which types of workers were most adversely affected by the crisis? How did informal sector workers fare in comparison to formal sector workers? 3.What were the responses to employment shocks, and how did they differ for informal and formal sector workers? Which workers found new jobs or additional jobs, which migrant workers returned home, which workers were able to access government social assistance or social insurance programs, which workers received private assistance from relatives and friends, and which workers cut back on their expenditures? 4. What was the impact of the economic crisis on the welfare of return migrants and their families? 5. What is the extent of firm compliance with labor regulations and what are the determinants of firm compliance? Are firms suffering from negative economic shocks less likely to comply with labor regulations? 6. How has enforcement of labor regulations affected employment, wages and benefits, and working conditions of formal and informal sector workers? 3. Preliminary Studies The $30,000 award for proposal preparation was used to support the following activities by research team members: field research in 5 cities, focused mainly on open-ended interviews with government labor bureau officials, firms, community leaders, households, and survey collaborators in 4 cities; a pilot survey of 200 local resident and migrant households in 2 cities and a tracking exercise in a third city; international travel of co-pi to participate in these activities as well as to take part in intensive project planning meetings in Beijing; and analysis of existing datasets and questionnaire development by CASS researchers. In this section, we present preliminary findings from these activities and other previous research by project researchers to provide evidence of the importance of the proposed 5

6 research project. In the data section below, we also describe how the pilot study helped us refine our survey design and demonstrated the feasibility of our proposed data collection efforts. Previous studies As described earlier, aggregate employment statistics suggest that nearly half of workers in urban China are employed informally. What about the evidence from micro-data? Using survey data from 66 cities collected by the Ministry of Labor in 2002, project researchers Wu and Cai (2006) estimate that 45.3% of local resident workers (migrants not sampled) were employed informally using a multifaceted definition of informality (Table 1). 2 As seen in Table 1, they also reveal that informality is substantial even in state and collective enterprise (22.9%) and is greater for women, the young, the old, and the less educated. Thus informality appears to be associated with more vulnerable population groups. Table 1: Informal Employment Shares of Local Residents in 66 Chinese Cities in 2002 Informal Category Informal Category Informal Category (%) (%) (%) All 45.3 By age: By education: By sector: Primary and below 78.4 State and collective Middle school 62.6 Other High school 48.6 By sex: Junior technical school 35.9 Male Specialized college 24.6 Female 49.5 > College and above 14.8 Source: Wu and Cai (2006), analysis of data from Ministry of Labor survey in 66 large cities. The dataset with the most comprehensive population coverage is the 2005 mini-census data, which like the census has relatively good coverage of migrants. Defining informality simply as the selfemployed plus those without a labor contract (which could overestimate informality if some permanent workers do not sign contracts), project researchers calculate that 52.6 percent of urban workers were employed informally (Table 2). The percentage is much higher among rural migrants (69%) than for local residents (50%) or urban migrants (38%), and is higher in townships (62%) than in larger cities (48%). Table 2: Urban Informal Employment Shares in 2005 (%) Local residents Rural migrants Urban All migrants Cities Townships All urban Source: IPLE Project Group (2007), analyzing % mini-census data. Informal employment includes self-employed, those without formal contracts. Interestingly, a nonparametric analysis of the relationship between informal employment share and city GDP per capita conducted by one of the project researchers, again using the 2005 mini-census data combined with aggregate city-level data, reveals a striking inverse relationship between the level of economic development and informalization (Figure 2). It could be that resources available to 2 The following types of workers were categorized as informal: 1) hired workers without formal contract not listed as formal employees; 2) domestic workers, temporary agency workers, and casual laborers; 3) community service workers without formal contract; 4) workers hired on the basis of hourly pay, daily pay, weekly pay, and uncertain pay (in terms of time and/or account); 5) paid helpers in family and self-employed businesses; 6) workers hired by individual entrepreneurs (getihu); 7) individual business owners (getihu), note that the ILO recommends that small business of less than 10 workers be considered informal, in China individual businesses are those with 7 or fewer workers. Workers in the agricultural sector are excluded in the analysis. 6

7 municipal governments matter a lot for supporting formal employment; it will be of great interest to explore this relationship further in the proposed project. Figure 2: Informal Employment Share and City GDP Per Capita in 2005 Informal Employment Share (%) in Ln(GDP per capita) 2005 Source: Wu (2009) Why does informality matter? For one, informality is strongly associated with lack of participation in social insurance programs. Again analyzing the 2005 mini-census data, we find that among those hired in the formal sector, 74% have participate in pension programs, 78% have health insurance, and 55% have unemployment insurance, while among those hired informally, the participation rates are only 20%, 27%, and 8% respectively (Table 3). Coverage rates are even lower for the self-employed or those working as household labor. Table 3: China Social Insurance Coverage Rates in 2005 (%) Pensions Health Insurance Unemployment All By location type: Cities Towns Rural By employment formality (cities and towns): Formal employment Hired Employer Informal employment Hired Self-employed Household labor Source: IPLE Project Group (2007), analyzing % mini-census data. The lack of social insurance coverage is a particular problem for migrant workers. Analysis of the China Urban Labor Survey data for 2005 shows clearly that in both the formal and informal sectors, migrants have lower coverage by any of the main social insurance programs compared to local residents (Table 4). In fact, among migrants in the informal sector social insurance coverage was very close to zero! But even most migrants with labor contracts do not enjoy social insurance coverage, suggesting systematic discrimination against migrants by employers and/or local governments. 7

8 Table 4: Comparison of Local and Migrant Workers in Urban China, 2005 Formal Informal Local Workers Migrant Workers Local Workers Migrant Workers Working days per week Working hours per day Monthly earning (yuan) Pension 82.1% 29.0% 54.8% 2.1% Unemployment insurance 39.7% 17.8% 12.6% 0.4% Working Injury Insurance 29.1% 31.7% 6.0% 1.2% Health Insurance 71.4% 29.7% 32.6% 1.3% Source: CULS2. While these previous studies provide interesting facts to motivate more in-depth study of informal employment, they provide no information on the recent economic crisis or the current implementation status of the new Labor Law. For that we turn to evidence from the qualitative and quantitative field research completed for the preparation of this proposal. Qualitative research findings As part of the preliminary fieldwork in preparation for this proposal, members of the research team visited Guangzhou and Dongguang in Guangdong Province, Suzhou in Jiangsu Province, Shenyang, and Shanghai to conduct unstructured interviews with local government officials, especially Labor Bureau officials, local survey counterparts, especially statistical bureau s, firm managers, and households, in order to better understand the employment impacts of the economic crisis and enforcement mechanisms and impacts of the newly enacted Labor Law. Regarding the economic crisis, the interviews painted a consistent picture that the crisis had much more pronounced impact on export-oriented firms than those dependent more on the domestic market. For instance, we found that in Guangzhou, enterprises producing goods such as furniture and equipment that were primarily produced for export experienced drops in orders (by 15% and 33%). But a company in Shanghai that was one of the country s largest bottle cap producer had experience little impact on his business which was tied mainly to domestic beer consumption. The central and local governments have adopted measures and policies to tackle the financial crisis. They have sought to support medium- and small-sized enterprises, which employ a lot of labor but who are more vulnerable to financial crises, in obtaining loans. The Shanghai government provided subsidies to encourage firms not to fire local workers during the slowdown. Troubled enterprises throughout the country have been allowed to delay payment of social security contributions in 2009 for up to 6 months. Insurance premium rates for medical, work injury, unemployment and maternity have been temporarily cut back in some regions with the authorization of provincial governments. Regarding the The Labor Contract Law implemented since Jan.1, 2008, most enterprises seemed to think that the implementation of the Law was appropriate, but that it did not begin at a good time. In 2008, many enterprises suffered due to RMB appreciation, the financial crisis, and labor cost increases, which created a heavy burden. Implementation of the Employment Contract Law has made things more difficult. One enterprise manager in Suzhou complained that his company would not be able to afford following regulations of the Labor Contract Law because it would become much harder to adjust employment levels in response to changing conditions. An enterprise leader in Dongguan estimated that one-third or one-half of enterprises in Dongguan didn t follow the regulations of the Law strictly. In fact, they wouldn t be able to do so even if they wanted to. Another private manager in Shenyang said that he had workers sign contracts that he kept hidden away in case he was inspected, but that these were not actually followed. Nonetheless, a number of interviewed enterprise managers said they will be able to follow the regulations of the law if given more time. Many firms had spent considerable time studying the new Law preparing for its implementation. Shanghai labor 8

9 bureau officials said that there remained aspects of the law that were unclear, which had led to a number of disputes that were hard to resolve. Pilot survey findings We conducted pilot household surveys in 2 of the 6 cities where we plan to conduct the third wave of the China Urban Labor Survey. The two cities were Shenyang in Liaoning Province in the northeast of China and Shenzhen in Guangdong Province on China s southern coast. Shenyang had a large state-owned enterprise sector before economic restructuring, and has struggled to keep pace with the rest of China. Shenzhen is a stone s throw away from Hong Kong and is a center of export-oriented manufacturing. In Shenzhen, the private sector, including many foreign firms, is dominant and most of the city s residents are migrants. The other cities to be survey as part of CULS3 are Shanghai and Fuzhou, two well-developed coastal cities, Wuhan in central China, and Xian in western China. Thus three of the six cities are in coastal China where the impact of the crisis was hardest felt, and three are in other regions that are less export-oriented and so have been less affected. In each city, 100 households were surveyed, 10 households in each of 10 neighborhoods which were selected to be regionally representative of each city. In half of the neighborhoods local residents were randomly sampled and in the other half migrants were randomly sampled, yielding a sample of 50 local resident households and 50 migrant households in each city. Information was collected for all adults living each of the sample households, yielding a sample of about 400 individuals, with slightly more local residents than migrants (220 versus 180). The goals of the pilot surveys were to test newly designed questionnaires and the appropriateness and feasibility of asking specific questions, to provide preliminary results to motivate the research project, and to demonstrate the overall feasibility of the proposed data collection effort, including sampling, especially of migrants, tracking previously surveyed households, and collaborating effectively with local counterparts to successfully complete the surveys. English versions of the questionnaires used in the pilot are included as attachments to the proposal, and will be the starting point for future households surveys for the project. Next, we present some initial findings from the collected pilot survey data. Table 5 describes the impact of the economic crisis on employment outcomes. Comparing reported monthly incomes for August 2008 and March 2009, we find that in both cities, migrants reported mean income declines of 7-8%, whereas local residents in Shenzhen reported a similar decline (8.5%) but local residents in Shenyang reported virtually no change. Note that shocks to migrants are likely to be understated since the worst hit will return to their homes and not be surveyed. The result is consistent with locals and migrants being treated similarly in Shenzhen but not in Shenyang, consistent with the findings of Gao and Wang, 2008) who found when analyzing firm data from a survey commissioned by the NPC that migrants were less discriminated against in richer areas.. We also report in Table 5 the percentage of respondents who report experiencing different adverse shocks during the crisis, including reduced work days per week, reduced regular and overtime work hours per day, reduced incomes, and work unit layoffs and output reductions. Some interesting patterns emerge. First, workers in Shenzhen were hit much harder by the crisis than workers in Shenyang. Over half of migrant workers in Shenzhen report that their employers reduced output and 40% said their employers laid off workers, with similar results for local workers in the city. In comparison, virtually no local workers in Shenyang reported output reductions or layoffs by their employers, although 20 percent of migrants reported output reductions (only 6% reported layoffs). A high share of migrants reported income reductions in both cities, 33% in Shenyang and 38% in Shenzhen, while only local residents in Shenzhen reported a large incidence of falling incomes (47%). Work time reductions were less common than income reductions, but in Shenzhen up to 30 percent of respondents (both migrants and local residents) reported reductions in working days per week, and reduced regular or overtime hours. 9

10 Table 5: Impact of the Economic Crisis on Chinese Workers Local workers Migrant workers All SY SZ All SY SZ Monthly income 8/2008 (RMB) Monthly income 3/2009 (RMB) Change in monthly income (%) % reporting following changes since August 2008: Reduced working days per week Reduced regular working hours Reduced overtime hours Reduced income Work unit laid off workers? Work unit received subsidy? Work unit reduced output? Source: Pilot household surveys of 199 households conducted in March 2009 in Shenyang (SY) and Shenzhen (SZ). Notes: mean monthly incomes restricted to sample working in both periods. These results, although admittedly from small samples, suggest that employment shocks to workers were widespread in major coastal cities, but thus far much less pronounced in other regions. This accords well with current understandings of the crisis, and lends credence to the proposed identification strategy to exploit variations in export demand shocks in different sectors and regions. One last interesting finding from Table 5 is that virtually no workers reported that their employers received government financial assistance to deal with the crisis as was mentioned in field interviews with labor officials in Shanghai. Thus, there appears to be considerable variation in the extent of the shocks associated with the crisis, and the responses of local governments. Finally, let s turn to evidence on people s views of the labor law and its enforcement, and their awareness of the specific provisions of the law (Tables 6 and 7). A surprisingly low percentage of local workers signed labor contracts in Shenyang (34%), and even fewer migrants (19%). In Shenzhen, a much higher percentage of both local workers and migrants signed contracts (61% and 67%, respectively). Again the evidence suggests that migrants are treated similarly as local residents in Shenzhen, but not in Shenyang. Migrant workers in both cities who do sign contracts are more likely to sign fixed-term contracts than local residents. Respondents are mostly satisfied with how well employers are abiding by the new Labor Law, but with different levels of enthusiasm. Very few respondents are highly critical (rank employers as very poor ). Migrants have a somewhat more negative assessment than local residents. The survey also finds no evidence that adherence to the Labor Law has been undermined by the onset of the economic crisis, with assessments of compliance changing little between August 2008 and March 2009, the time of the survey. Overall, the results yield some optimism that the new Law is being taken seriously despite the economic downturn. Table 6: Compliance with the Labor Contract Law Local workers Migrant workers Total SY SZ Total SY SZ Have you signed labor contract with your working unit or employer for your current job? Yes No Self employment What kind of labor contract have you signed? Fixed period Indefinite period For specific work

11 Other Before August of 2008, how well did your employer abide by the labor contract law? Very well Satisfactory Just so so Very poorly Currently, how well does your enterprise abide by the labor contract law? Very well Satisfactory Just so so Very poorly But are people even aware of the specific content of the Labor Contract Law provisions? Lack of awareness among workers could lead to lack of enforcement by employers. To test this, we asked a sequence of factual questions about the substance of different aspects of the Labor Contract Law. Table 7 presents tabulations for the percentage of respondents who answered these questions correctly. Table 7: Knowledge of Labor Law Provisions (% answered correctly) Local workers Migrant workers Total SY SZ Total SY SZ 1. Within how long should a labor contract be signed after a person is employed? (one month ) For a one-year labor contract, what is the maximum probationary period? (one month) ,0 3. Must the wage during the probationary period be higher than the local minimum wage? (yes) If a worker breaks enterprise rules, can the enterprise terminate the labor contract? (yes) If a worker is ill, can an enterprise terminate the labor contract? (no) Overall, the respondents performed well. The only question that most people got wrong was the first one concerning how long after being hired a labor contract needs to be signed. For the last 3 questions, the percent correct is very high. Migrants perform slightly worse across the board (except the first question) but not by much. Methodology In this section we describe how we will use the survey data to answer research questions of interest. The empirical analysis will primarily analyze three datasets: a household survey of local workers and migrant workers in 6 cities, a linked firm-worker survey in the same cities, and a panel rural household survey dataset collected by the Research Center for Rural Economy under the Ministry of Agriculture. The first two datasets will be collected as part of the project, and several of our research team members have previously worked extensively with the RCRE data. 1. What were the nature and magnitude of negative economic shocks associated with the economic crisis? In China the negative impacts of the economic crisis have propagated primarily through large reductions in export demand. Define S i,t generally as an exogenous shock affecting worker i between periods t-1 and t. One relatively exogenous specific measure of negative demand shocks is the change in exports (E) in narrowly defined sectors (s) specific to each city (c), or S j,t = Δlog(E c,s ),which can be calculated from Chinese customs data. These can be matched to the sectors of workers and firms 11

12 prior to the crisis to construct plausibly exogenous negative shocks to demand faced by firms and in turn, labor demand for workers. Of course, there may be other macro-level impacts of the crisis, including government policy responses, which can affect credit access and interest rates, exchange rates, prices (for outputs, inputs, wages, housing, etc.), or access to government social assistance or social insurance programs. To the extent possible, measures for these outcomes will be collected for China as a whole and for each surveyed city. They can be used to describe the impacts of the crisis, to analyze the responses to the crisis, and as control variables in some regressions. To the extent that such impacts are national in scope, they will not vary in the cross section and so they will not affect before-after comparisons of how different firms or individuals respond to different export demand shocks. We can also construct individual- and firm-level shock variables from the survey data that we will collect. For example, we can record whether firms reduced exports or faced lower prices as a result of the crisis, or whether individuals lost jobs involuntarily or saw reductions in hours worked, wages paid, or benefits received. 2. Which types of workers were most adversely affected by the crisis? How did informal sector workers fare in comparison to formal sector workers? Who was hit hardest by the crisis? Possible impacts on workers include involuntary job loss, reductions in work hours as well as in wages and/or benefits. Household and firm survey data can first enable rich description of the nature of employment shocks and the characteristics of those affected to a greater or less extent by the crisis. Defining W i,t to be a vector of employment outcomes of worker i in sector j at period t, we will first estimate the following reduced form equation for the determinants of changes in W ij,t : ΔW ij,t = α 1 I ij,t-1 + α 2 X ij,t-1 + Δε ij,t (1) Here, I ij,t-1 is an indicator for whether the individual was informally employed prior to the economic crisis, X ij,t-1 includes other initial period characteristics of the worker or the worker s employer, family, neighborhood, or city, and Δε ij,t includes measurement error and omitted factors affecting changes in employment outcomes. It is possible in this specification also to add interactions between I ij,t-1 and variables in X ij,t-1, By examining changes rather than levels, this specification controls for the impact of individual unobserved factors (or fixed effects) that do not change over time. The impacts of city /sector factors on labor outcomes are likely to vary by time given different vulnerability to external shocks in different city specific sectors. Therefore, in the first-differenced model shown in (1), we will also add city-sector fixed effects. Similarly family, neighborhood fixed effects may be used as additional controls when examining impacts of the crisis on changes in employment outcomes. There may be some difficulty distinguishing whether the reduced form relationships just described reflect the impacts of the economic crisis or unobserved worker characteristics correlated with both changes in employment outcomes and observed worker or job characteristics (including informality). For example, less educated workers might be less stable by nature and thus more likely to change (lose) jobs. To examine how employment impacts are associated directly with exogenous economic shocks associated with the economic crisis, we will estimate the following model: ΔW ij,t = α 1 I ij,t-1 + α 2 X ij,t-1 + α 3 S j,t + α 4 S j,t I ij,t-1 + α 5 S j,t X ij,t-1 + Δε ij,t (2) The coefficients on the interaction terms with the shock variable for sector j S j,t tell us whether being informally employed or having other characteristics (e.g., gender or age) influence how vulnerable one is to negative economic shocks. As before, when using the household data it is possible to add family, neighborhood, and city fixed effects as well as sector fixed effects. We will also estimate versions of equation (2) using firm retrospective panel data to examine the impact of economic shocks on firm employment, including employment of different types of workers. This will be especially 12

13 informative in quantifying employment impacts on migrant workers, who are difficult to capture in urban household survey data because many of them will have returned to rural areas. With matched firm-worker data, one may define S j,t to be firm-level shocks, and control for firm fixed effects to examine whether workers within the same firm are affected differently by negative shocks to the firm. One limitation of using matched data is that we would not be able to examine how shocks affect firing of workers, because only currently employed workers will be interviewed. 3.What were the responses to employment shocks, and how did they differ for informal and formal sector workers? Which workers found new jobs or additional jobs, which migrant workers returned home, which workers were able to access government social assistance or social insurance programs, which workers received private assistance from relatives and friends, and which workers cut back on their expenditures? Define ΔR ij,t as the responses to negative employment shocks ΔW ij,t. A general specification for estimating the determinants of responses to negative employment shocks is the following: ΔR ij,t = β 1 I ij,t-1 + β 2 X ij,t-1 + β 3 ΔW ij,t + β 4 ΔW ij,t I ij,t-1 + β 5 ΔW ij,t X ij,t-1 + Δe i,t (3) Note that ΔW ij,t may be a vector employment shocks, for example it could include reductions in both wages and benefits. In this example, we will examine whether lost wages and benefits are compensated by government social assistance, and whether the responsiveness of social assistance is a function of prior informal employment status or other individual, job, or community characteristics. To address the potential endogeneity of shocks to unobserved individual attributes that also influence responses, it is natural to choose the exogenous economic shock S j,t and its interactions to serve as instruments for ΔW ij,t, treating equation (2) as the first stage equation. Note that interaction terms in equation (2) act as additional instruments, enabling the identification of the impact of more than one endogenous employment-related shock in equation (3). To construct the instruments for the interaction terms with ΔW ij,t in (3), we simply interact the instruments in (2) with the variables that are being interacted with ΔW ij,t in (3). For studying responses that only occur after discrete events such as involuntary job loss, it is common to restrict attention to the sample of individuals experiencing the event. For instance, we will examine which factors influence the probability of reemployment conditional on job loss. We will use a discrete duration model to examine which workers dislocated by the economic crisis were more likely to find new jobs as well as whether reemployment probabilities were affected by whether one had previously been employed in the informal sector. We will also examine which factors influence whether new jobs are in the informal or formal sector What was the impact of the economic crisis on the welfare of return migrants and their families? Much has been made of the large impact of the crisis on rural-urban migrants. It was widely reported that 20 million migrants returned to the countryside early before the 2009 spring festival, with concern that return migrants would be unable to find work to do at home. On the other hand, many city officials were unconcerned with the plight of migrants, feeling that their rural homes would provide adequate support to act as an effective safety net. Studying return migration is thus an important aspect of the impact of the crisis on Chinese workers. To do so requires rural household survey data that has broad enough coverage over and time and space to reflect the overall impact of the crisis on return migrants. We will utilize the best available dataset for this purpose the RCRE panel survey of rural households in 20 provinces. 4 Since 2003, the RCRE panel survey has also 3 Giles, Park, and Cai (2006) develop a useful approach to estimating a discrete duration model for reemployment of urban workers in China who lost jobs during economic restructuring of the state sector during the late 1990s. 4 Evidence from a survey in RCRE villages carried out at Chinese New Year in 2009 provides one of the sources 13

14 included specific questions on migrants from the household, including their age, educational attainment and industry sector of employment. The estimating equation is still equation (3). In this context ΔW ij,t is whether the household had a migrant who returned home, and ΔR ij,t can include a large set of outcomes characterizing the return migrant and the household to which he or she returns, such as income per capita or poverty status; individual and household labor supply in self-employment, wage employment, and agriculture; and household agricultural productivity, consumption per capita, wealth per capita, and investment per capita. With information on migration destinations, which is available in the RCRE data, it is still possible to instrument ΔW ij,t with a measure of export shocks to migration destination cities or provinces. Because not all families have migrants prior to economic shocks, we can weight the regressions by the inverse probability of migration (estimated separately) to correct for potential selection bias based on observables (Wooldridge, 2002). 5. What is the extent of firm compliance with labor regulations and what are the determinants of firm compliance? Are firms suffering from negative economic shocks less likely to comply with labor regulations? The household and firm survey data will enable us to provide a rich description of the extent of actual compliance with the new Labor Law and other labor regulations. With the firm data, we will analyze the determinants of current compliance, and also investigate whether firms suffering large negative shocks were less willing, able, or obligated to enforce such regulations. We will estimate the following model: L ij,t = δ 1 L ij,t-1 + δ 2 X ij,t-1 + δ 3 S j,t + δ 4 S i,t L ij,t-1 + δ 5 S j,t X ij,t-1 + u i,t (4) The regressors include labor regulation enforcement prior to the Labor Law L ij,t-1, which captures firm heterogeneity in propensity toward compliance. We are most interested in which firm characteristics X ij,t-1 (lagged to reduce endogeneity) better predict enforcement of labor regulations, and the extent to which firms more exposed to economics shocks, S j,t, are more or less likely to comply with regulations. 6. How has enforcement of labor regulations affected employment, wages and benefits, and working conditions of formal and informal sector workers? We can also study additional firm outcomes such as firm labor costs, profitability, investment, etc. Methodologically, we take two approaches to this question, relying on different assumptions about enforcement of China s new Labor Law, which took effect in January If we first assume that implementation of the new Labor Law was strictly enforced and significantly affected how firms make employment decisions, we can exploit comparisons of firm behavior before and after implementation of the law. In addition to data on 2009 (depending on survey timing) the firm survey will collect retrospective firm-level data for the years 2006, 2007, 2008, thus providing two years of data prior to the Law and two years of data after the law went into force. Then one can apply standard panel data methods to estimate the treatment effect of the new Labor Law using the following equation: ΔW ij,t = α 1 X ij,0 + α 2 ΔT t + α 3i ΔT t X ij,0 + Δε ij,t (5) Here, we regress changes in employment outcomes ΔW ij,t on changes in treatment ΔT t, where T t =0 in 2006 and 2007 and T t =1 in 2008 and Differencing controls for firm fixed effects. We omit other firm changes because of endogeneity concerns. We do permit changes in W ij,t and treatment effects to be a function of firm predetermined or time-invariant characteristics X ij,0. A weakness of this specification is that there is no variation in treatment across firms, so it is not possible to control for the frequently quoted figure of 20 million return migrants. 14

15 separately for year effects, although we can still control for changes in aggregate variables such as sector output or sector employment as well as city-specific trends (by adding city fixed effects). One possible way to generate variation in the change in treatment is to exploit variation across firms in compliance with labor regulations prior to implementation of the Labor Law, in other words to replace ΔT t with ΔL ij,t. For instance, if some firms were already complying fully with labor regulations, then passage of the new Labor Law would be have minimal impact on their behavior, making such firms a potential control group to be compared with firms that had poor initial compliance and were targeted by the new law. In this case, treatment and control groups vary in initial compliance, selection bias associated with this difference is reduced by the removal of firm fixed effects through differencing, and can be further reduced by employing propensity matching methods. Variation in change in treatment across firms also makes it possible to control for year effects. How does the economic crisis impact estimation of (5)? Since the economic crisis did not have a large impact on employment until 2009, we could simply omit 2009 data and compare 2006 and 2007 with 2008 to get the estimates of treatment effect without confounding factors from the economic crisis. Alternatively, under the assumption that crisis shocks are exogenous, we can add S i,t and the interaction of S i,t with ΔT t to the equation to control for the impact of the crisis and then test whether the crisis influences the treatment effect. One might not believe that the Labor Law was well implemented during the period from 2008 to 2009 because it was new and then subverted by the economic crisis. For this reason, we will also focus on comparing the behavior of firms with high initial compliance and those with poor initial compliance throughout the period of study. Here we assume that initial compliance captures the firms obligation or commitment to complying with labor regulations, including the New Labor Law. Then we can estimate the following equation: ΔW ij,t = α 1 L ij,t-1 + α 2 X ij,t-1 + α 3 L ij,t-1 X ij,t-1 + Δε i,t (6) As before, W i,t is a vector of firm employment outcomes of interest. L i,t-1 is a measure of the extent to which firms followed labor regulations prior to implementation of the Labor Law (such as number of complaints filed against the firm, or share of casual or temporary workers). Since prior compliance is not randomly assigned, we still are left with a potential endogeneity problem. For this reason it also will be important to study what types of firms had poor compliance initially; it may be possible to employ matching methods to reduce remaining endogeneity bias. As with the previous specification, one can also add S i,t and S i,t L i,t-1 to equation (6) to examine how initial compliance with labor regulations influences how firms respond to the crisis with respect to hiring or firing workers or changing wages or benefits. With detailed employment and wage data on different types of workers (from retrospective records), one can also test whether certain types of workers suffer more from the crisis, and whether enforcement of labor regulations mediates this process. Originally, we had hoped in the preliminary fieldwork to identify additional exogenous sources of variation in enforcement of labor regulations to enable identification of the impact of enforcement on firm employment behavior, for example through random audits, or rules affecting the likelihood that different types of firms had different probabilities of being audited (for example based on ownership or size). Interviews in multiple cities, however, revealed that investigations of violations of labor regulations nearly always occurred in response to complaints from workers, which is highly endogenous to the firm s own compliance behavior. There still may be some scope for variation across cities in the enforcement of labor regulations due to differences in city budgetary strength (to allocate more staff to enforcement) or differences in the activity level of non-governmental organizations supporting workers rights. However, this variation is likely to be correlated with other factors influencing labor outcomes. 5. Data collection and other datasets 15