There is keen awareness across developed and middle-income countries of the developing world that increased longevity and aging populations will place significant and growing burdens on working age adults in the relatively near future. In the United States and other economies with pay-as-you-go social security systems, these burdens will be transmitted through fiscal systems. Even where public transfer mechanisms are not as important, working-age adults may nonetheless face increasing burdens associated with supporting the elderly through both financial and in-kind transfers.2 Increasing the retirement age is frequently viewed as one feasible means of easing burdens on working-
1 This chapter has benefited from conversations with Fang Cai, Xiaoyan Lei, Philip O’Keefe, Albert Park, James Smith, John Strauss, Firman Witoelar, Kyeongwon Yoo, Xiaoqing Yu, and Yaohui Zhao, and also from comments of David Wise, Yaohui Zhao, and other participants in the Conference on Aging in Asia, sponsored by the U.S. National Academy of Sciences and Chinese Academy of Social Sciences (Beijing, December 9-10, 2010). We are grateful for financial support for this work from two sources at the World Bank: the Gender Action Program of the PREM Network and the Knowledge for Change Trust Fund managed by the Development Economics Vice Presidency (DEC) of the Bank. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the executive directors of the World Bank or the governments they represent.
2 Lee and Mason (2011) highlight the implications of population aging for sustainability of public and private transfer systems across Africa, Asia, Europe, Latin America, and the United States.
age populations, and yet it is likely that exits from productive activity are shaped by household wealth and individual preferences, as well as by institutions and policy. Alternatively, continued participation in the workforce may reflect the ability of workers to learn new skills and to remain productive into older age. With an eye toward providing insight into the retirement decision in East Asia, this chapter presents descriptive evidence on retirement and labor supply patterns in China, Indonesia, and Korea.
While China’s rapid demographic transition is frequently highlighted in news accounts because of the sheer size of its aging population, Korea and Indonesia are also confronting rapidly aging populations.3 In contrast to most developed countries, however, rural and urban populations face significantly different retirement systems. Differences across rural and urban areas in both retirement patterns and access to financial support are most extreme in China, where most long-term residents in urban areas have had formal wage employment, retire at a relatively young age, and receive substantial support from pensions. Rural residents, by contrast, have lacked pension support and may expect to work in farming or other agriculture-related activities until relatively late in their lives.4 In this sense, urban residents of China with formal sector employment face retirement decisions that are more similar to those of residents in developed countries. Residents of China’s rural areas, by contrast, share more in common with residents of other developing countries, and make labor supply decisions in the absence of both pension availability and the constraint imposed by a mandatory age of retirement from the formal sector.
This chapter brings together information from several data sources to highlight differences in labor supply of older workers across urban and rural China, Indonesia, and Korea, and to review patterns and trends in the context of institutional differences across these three countries and between urban and rural areas. For perspective on retirement patterns in East Asian economies, we then place the retirement decision in China, Indonesia, and Korea in the context of employment patterns of older
3 Recent research by demographers at the U.S. Census Bureau suggests that the old age dependency rates in 2020 will reach 22, 19, and 13%, for Korea, China, and Indonesia, respectively, and by 2040 these rates will rise to 53, 40, and 25% (Kinsella and He, 2009). A preliminary release from China’s 2010 census informs us that 13.3% of China’s population is now over 60 as opposed to 10.3% in 2000, while the size of the future workforce has dwindled, with individuals under 14 accounting for 16.6% of the population, down from 23% in 2000 (National Bureau of Statistics, 2011).
4 New initiatives are currently under way in rural China. A government-subsidized contributory rural pension piloted in 2009 will be rolled out to cover all rural counties over the next three years. In cities, a new pension scheme, modeled on the rural pension program, was first introduced in July 2011 with the aim of providing financial protection in old age to nonworking urban residents and informal sector workers.
workers in the United States and the United Kingdom. In common with findings from the retirement literature focusing on developed economies, the descriptive evidence presented in this chapter is suggestive of the role that ability to collect a pension (or social security benefit) plays in the retirement decision.5
Mandatory retirement provisions in each of these East Asian economies, however, condition decisions of when and how to exit from productive activity. While significant numbers of retirees return to work in self-employed activities or informal work after reaching mandatory retirement age, the types of work that “retirees” are able to find may be unattractive for some older workers. Differences in the mandatory retirement age for men and women in China likely contributes to differences across genders in participation in work later in life, with important consequences for relative pension wealth and relative financial security of older men and women.
After reviewing descriptive trends, we lay out an empirical model to examine correlates of labor supply with own and spouse eligibility to receive a pension, own and spouse health status, and proxies for household wealth. We next review data sources and correlates of employment separately for China, Indonesia, and Korea. The chapter presents comparative descriptive evidence from East Asia and highlights important questions on retirement behavior in developing countries that may be addressed from new panel data initiatives currently under way.
EMPLOYMENT PATTERNS AND THE RETIREMENT OF OLDER WORKERS
In China, long-term urban residents with formal sector employment can expect to receive a pension upon retirement, but face mandatory retirement at a relatively young age.6 Where urban employed men confront mandatory retirement at age 60, women in blue collar occupations are frequently required to retire at age 50, those in white collar occupations at
5 Blau (1994) suggests that social security eligibility contributes to relatively high exit from the labor force at age 65; Krueger and Pischke (1992) exploit design features of the U.S. social security system to demonstrate the effects of benefits on labor force participation. Gruber and Wise (1999, 2004) present evidence from Organisation for Economic Co-operation and Development economies of the effects of social security and public pension systems on labor supply decisions of older workers.
6 We define a long-term urban resident as an urban dweller with an urban (nonagricultural) residential registration (hukou) status. While considerable efforts have been made recently to extend social insurance benefits to migrants living in the city, migrants are much less likely to have employment contracts and to have employers who are making mandated contributions to pension, health, and disability insurance programs (Giles, Wang, and Park, 2012).
age 55, with women in some categories (e.g., university professors) able to work until age 60. Among current retirement-age residents of urban areas, a large share is receiving relatively generous pension support. By contrast, rural elderly, who had lower incomes during their working lives and less accumulated wealth than their urban counterparts (Kanbur and Zhang, 1999; Ravallion and Chen, 2007), do not typically have pension income. According to the 2005 1% population sample, 45.4% of urban residents over age 60 report pension income as their most important source of financial support, but only 4.6% of rural residents note an important role for pension income. Instead, 38% of rural respondents over age 60 report that income from their own labor is their most important source of support.7
The stark difference in employment rates of rural and urban residents reflects differences in both pension wealth and mandatory retirement provisions across urban and rural areas.8 As evident in Panel A of Figure 6-1, which presents locally weighted regression (LOWESS) estimates of employment rates by age, China’s rural residents are far more likely to be employed well after the mandatory retirement ages faced by urban residents. From the China Health and Retirement Longitudinal Study (CHARLS) pilot conducted in 2008, we note 45% of urban men aged 60-64 were still employed at least one hour per week (Panel A), but in rural areas nearly 86% of men in this age range were still working. The difference in employment of urban and rural women aged 60-64 was even wider, with only 16% working in urban areas and nearly 57% still employed in rural areas.
If anything, the CHARLS pilot, with relatively small sample sizes in urban Zhejiang and Gansu and representative of only two provinces, may overstate the employment rates of older men and women. Also presented in Panel A of Figure 6-1 are the 1991 and 2009 estimates from the China Health Nutrition Survey (CHNS), which show a substantially lower employment rate of 31% for urban men in the 60-64 age range. When comparing employment rates across the age distribution over time using the CHNS, one observes declines for both men and women in urban
7 Additional descriptive statistics on sources of support from the 2005 1% population subsample are reported in Cai et al. (2012) and Giles, Wang, and Zhao (2010).
8 In defining “employment” in this chapter, we include wage employment in the informal sector, casual work, self-employed activities, and unpaid work in family-run enterprises, all of which may be important for older workers in these economies. We focus on employment as opposed to labor force participation, per se, for two reasons. Job search is often not well documented, and where it is (e.g., the CHARLS data for China), there are a vanishingly small number of respondents (five in CHARLS) aged 45 and older who are not employed but report active searches for work. We have no doubt that a search process exists for older workers who wish to work, but it is difficult to capture, and this is particularly true when large shares of older workers are self-employed.
FIGURE 6-1 Employment rates by age cohort of older workers and elderly in urban and rural China.
NOTE: Employment rates by age cohort are calculated using nonparametric locally weighted regression (LOWESS) with a bandwidth of 0.3.
SOURCES: Panel A and B use data from the common provinces surveyed across waves of the China Health and Nutrition Survey (CHNS) conducted from 1991 to 2009. Panel A also includes data from the 2008 China Health and Retirement Longitudinal Study (CHARLS) pilot.
areas, but less pronounced declines in employment rates in rural areas.9 Rural women between age 45 and 65 were somewhat less likely to be working in 2009 than in 1991, but this does not appear to be true for older women or for rural men.
One of the sharper changes from 1991 to 2009, as viewed from the CHNS, lies in the decline in employment rates of women over age 45 in urban China.10 The decline in older women’s labor force participation raises an important question for labor research in China: Does the decline in women’s employment reflect a resurgence of gender discrimination in post-reform China or the effects of increases in household wealth and the ability of women to exit the labor force at a younger age? Differences across genders in mandatory retirement ages likely create an institutional bias against women’s employment. Even in the absence of discrimination, the employment decision of older women in urban China reflects a constrained choice. Women may return to work as consultants or in self-employed activities after reaching mandatory retirement age, but retired women are frequently receiving pensions, which raises reservation wages for new employment. Those urban women uninterested in working in typical self-employed activities held by blue collar workers (e.g., nannies or housekeepers) may choose to stay out of work if their pension incomes are sufficient.
Earlier research on labor force participation in China has noted the drop in employment rates of urban residents, and urban women in particular (Cai, Park, and Zhao, 2008; Maurer-Fazio et al., 2011) and attributed the drop to the effects of state sector restructuring after 1997. After losing work during state sector restructuring, men, the young, and the well educated generally faced shorter durations out of work (Appleton et al., 2002; Giles, Park, and Cai, 2006a; Maurer-Fazio, 2007). Moreover, some researchers found that a woman’s decision to reenter the workforce was affected not only by permanent and relatively generous pensions, but also by family circumstances (Giles, Park, and Cai, 2006b).
Given support available through pensions to relatively young workers
9 We review evidence from the CHNS as it is the publicly available data source that researchers used (pre-CHARLS) to study labor supply, health status, and retirement of older workers in China (e.g., Benjamin, Brandt, and Fan, 2003; Dong, 2010). As it is based on a panel of households that does not enumerate complete information on family members who have split off of households, one should be concerned that later waves of the CHNS over-represent those who remain in the households. As we are interested in the over-45 population, this is less of a problem than if we were considering the employment decisions of individuals who were children in 1991 and likely to have moved out of households.
10 Additional corroborating evidence on the decline in women’s labor force participation from the census and the China Urban Labor Survey (CULS) can be found in the expanded working paper version of this paper (Giles, Wang, and Cai, 2011).
during economic restructuring and potential biases against hiring displaced workers relatively close to retirement age, one would expect to see sharp drops in employment of urban women from 1991 to 1997 and 2000, as is shown in Panel B of Figure 6-1. If reduced employment rates of urban women over 45 were simply the effect of restructuring in the late 1990s, however, we would expect that employment of women in the 45-55 age range would return to higher pre-1997 levels by 2009. The fact that older working-age women, who have not yet reached mandatory retirement age, continue to have lower employment rates, even after labor markets tightened during the 2000s, raises the possibility that exits from the labor force may be a choice facilitated by higher wealth or by increasing demands on time for nonmarket activities, such as caring for children, elderly, or other ill family members.
Exits from employment among older workers as they approach pension eligibility are not unusual in more developed economies such as the United Kingdom and the United States, but exit rates in the years before retirement age are not typically as high as one observes in China. Figure 6-2 highlights differences in employment rates across five countries and shows that 68 and 70% of women aged 50 to 54 are still employed in the United States and United Kingdom, respectively. Employment rates of women in this age range are somewhat lower in urban Indonesia at 63%, but above the 38 and 30% employment rates witnessed in urban Korea and China, respectively.11 Relative employment patterns of men in the 55 to 59 age range follow a similar pattern: 68 and 67% of urban men in this age range in China and Korea, respectively, are employed, while 82% of urban Indonesia men of this age are still working.
Mandatory Retirement Provisions and Retirement Patterns
Incentives created by gender differences in the mandatory retirement age may encourage early exits from the labor force by women, particularly for women who face the prospect of a job search in their 40s. Career changes and job changes later in working life can be difficult in developed and developing economies alike. As beginning a new job requires learning processes, technology, and culture of the new workplace, a new employee will not reach peak productivity in a position immediately upon being hired. An employer may be less likely to consider hiring a worker who is close to mandatory retirement age simply because there is not sufficient time to earn a return on initial training and start-up costs relative to a younger worker. In the United States, where many workers
11 Note that this is the urban employment rate for women using the CHARLS sample; in the CHNS sample, the rate is higher at 42%.
FIGURE 6-2 Employment rates by age cohort in China, Indonesia, Korea, United Kingdom, and the United States.
NOTE: Employment rates by age cohort are calculated using non-parametric locally weighted regression (LOWESS) with a bandwidth of 0.3.
SOURCES: Data from China: 2009 CHNS and the 2008 CHARLS pilot; Indonesia: 2007 Indonesia Family Life Survey (IFLS); Korea: 2006 Korean Longitudinal Study of Aging (KLoSA); United States: 2008 Health and Retirement Study (HRS); United Kingdom: 2008/9 English Longitudinal Study of Ageing (ELSA).
start to leave the workforce upon eligibility for Social Security benefits, difficulties finding new employment among older displaced workers are well documented. Research using the U.S. Health and Retirement Study (HRS), for example, has demonstrated that of workers who lose jobs after age 55, only 60% of men and 55% of women are employed again within two years, while 80% of nondisplaced workers are employed (Chan and Stevens, 2001).
In China, and to some extent Indonesia and Korea, a mandatory retirement age for some occupations and types of employers is even more binding than Social Security eligibility in the United States. Differences in mandatory retirement ages of men and women in China may have a significant impact on how employers view the relative returns to hiring male and female employees who are in their 40s and older.12
In Korea and Indonesia, retirement ages are not mandated by the law, but employment laws allow for firms to set mandatory ages, and government employees and civil servants face mandatory retirement. After the East Asian financial crisis in 1997/1998, firms that had not implemented mandatory retirement started to do so as a way of slowing wage increases associated with seniority-based wage systems (Cho and Kim, 2005).
The higher employment rates of older urban men and women in Indonesia and Korea (Figure 6-2) suggest that mandatory retirement from some occupations and types of employers does not mechanically lead to permanent exit from productive employment. Existing research on older workers in both countries suggests that an impending retirement creates incentives for forward-looking workers to leave employers preemptively, either to start their own businesses or to start second careers working for smaller private-sector employers.13
One significant difference across workers in urban and rural areas of China, Indonesia, and Korea lies in the share of the retirement-age workforce with access to pensions. In China, evidence from the CHARLS pilot suggests that, of urban residents aged 60 and older, 79% of men and 54% of women have access to pension support. In urban areas of Indo-
12 Of course, employers may already perceive older workers to be less productive (Chan and Stevens, 2001; Dalen, Henkens, and Joop, 2010), but this is a problem faced by both older men and women when looking for work.
13 McKee (2006) finds that in Indonesia, half of government workers move into either the private sector or into self-employment, and 61% of workers who leave their private-sector jobs move into self-employment. In Korea, self-employment was one response to lay-offs from larger employers in the wake of the 1997/1998 financial crisis (Sohn, 2007) and continues to be an important source of employment for older males (Lee, 2009). Lee and Lee (2011) note differences in the retirement ages of self-employed and wage-salary earners, but do not discuss the incentive to move into self-employment ahead of mandatory retirement among wage and salary earners who wish to continue working.
nesia and Korea, by contrast, fewer than 35% of men and 15% of women have access to pensions.14 Another unique feature of pension eligibility in Korea lies with an imperfect correspondence between mandatory retirement age and age at which pension-eligible retirees may start receiving pension benefits. The National Pension scheme, the largest of the three main sources of pensions in Korea, does not begin paying benefits until age 60, yet a significant share of employees face mandatory retirement at age 55 (Cho and Kim, 2005). This imperfect correspondence likely increases incentives for those employees facing mandatory retirement to look for new career opportunities, including in the self-employed sector.
Across rural areas of China, Indonesia, and Korea, both men and women remain actively employed until much later in their lives than urban residents. In rural areas of all three economies, agricultural production on the family farm continues to be a significant source of employment for older workers, and this is necessitated by the fact that rural residents tend to accumulate less wealth over their working lifetimes. In addition, older workers in rural China and Indonesia are far less likely to have access to pensions than their urban counterparts. Both the CHARLS and the 2005 Population Census suggest that, of rural residents over 60, roughly 5% of men and less than 1% of women have pension support. Similarly, in rural Indonesia, the Indonesia Family Life Survey (IFLS) shows that only 8% of men and 3% of women aged 60 and older have access to pensions.
Rural Korea is a somewhat different matter. Efforts to bring rural residents into the National Pension scheme established in 1988 have led to pension coverage rates in rural areas that do not differ significantly from urban areas; indeed, 34% of older rural men in the 2006 Korean Longitudinal Study of Aging (KLoSA) wave report receiving pensions. Nonetheless, the continued labor supply of rural men is viewed as an important factor contributing to high rates of economic activity among older Koreans (Lee, 2009). In spite of availability of pensions, levels of support are not sufficient to permit retirement of the rural elderly. Lee (2009) suggests that the out-migration of the young has left elderly who remain behind with both a lack of young labor and insufficient wealth to cease productive activity. Given that the history of rapid urbanization in Korea mirrors the process of rural-to-urban migration taking place in China, one might be concerned that incidence of delayed retirement of China’s rural farmers, who are less affluent and have lacked pension support, may follow a similar pattern to Korea.
14 More specifically, in Korea, 34% of urban men and 15% of urban women have access to pensions; and in Indonesia, 27% of urban men and 10% of urban women have access to pensions. Figures on pension coverage are drawn from the 2006 wave of KLoSA and the 2007 wave of IFLS.
RETIREMENT IN CHINA, INDONESIA, AND KOREA
The descriptive patterns shown above make use of data from early stages of longitudinal studies from China and Korea that are modeled on the Health and Retirement Study (HRS) from the United States and the English Longitudinal Study on Ageing (ELSA) in the United Kingdom. New survey efforts, like CHARLS in China, promise to facilitate the study of retirement and labor supply behavior in regions where pension and social security systems are not well established and population aging is occurring at a rapid pace. In the analytical models that we estimate, we make use of the cross-sections from China, Indonesia, and Korea used for basic descriptive statistics presented above.15 Below, we provide some additional discussion of the analysis sample used for each country.
The China Health and Retirement Longitudinal Study (CHARLS), directed by Yaohui Zhao at the National School of Development at Peking University, was developed and implemented by an accomplished team of collaborators from China and abroad. The sample used in this analysis is from the pilot survey conducted in 2008 in Gansu and Zhejiang provinces and is representative of adults over 45 years of age in these provinces. Gansu, in the northwest, is a relatively poor province, and Zhejiang, on the coast south of Shanghai, is relatively affluent. CHARLS follows the model of the HRS, ELSA, and KLoSA in the United States, United Kingdom, and Korea, and enumerates information on respondent and spouse socioeconomic status, health conditions, work and employment, and incomes and pensions, and it also includes household and community-level information. In the pilot sample used in the analysis below, there are 1,570 households and 2,685 eligible respondents.16
The Indonesia Family Life Survey (IFLS) is an ongoing project conducted by the RAND Corporation and universities in both Indonesia and the United States. To date, four waves have been conducted in 1993, 1997/1998, 2000, and 2007/2008. The survey covers individuals, households, communities, and the community-level health and educational
15 Additional summary statistics for important variables used in the analysis are shown in Table 1 of the longer working paper version of this paper (Giles, Wang, and Cai, 2011).
16 The CHARLS project has funding for nationally representative surveys in 2011 and 2013, and fieldwork on the first national wave began in July 2011. Public-use data should be available for the research community by June 2012.
facilities that respondent households access in their daily life. While it is not a formal member of the “HRS family” of aging surveys, the data are comprehensive and include information on respondent health status, employment and labor allocation, pension participation, and household and family characteristics, all of which are crucial for study of retirement and labor supply decisions of the elderly. The original survey was representative of 83% of the Indonesian population and 13 of 27 provinces. As it tracks the movers and split-offs from the original sample, IFLS was a pioneer among household surveys in the developing world. Tracking is important when working with the 2007/2008 round of IFLS, because it allows us to avoid using a selected sample of individuals who have not moved.17 The most recent round has 50,583 observations and 13,507 households. The analytical work below uses detailed information collected from 7,283 direct respondents aged 45 and over for whom information on pension eligibility was enumerated.
The Korean Longitudinal Study of Ageing (KLoSA) is a nationally representative survey conducted by the Korea Labor Institute to investigate the overall conditions of people aged 45 and older in South Korea (excluding Jeju island). KLoSA belongs to the same family of aging surveys as CHARLS, HRS, and ELSA. The first baseline survey was conducted in 2006, and it was followed by a job history survey in 2007. While KLoSA is a longitudinal survey, only the baseline survey is publically available. In common with CHARLS and ELSA, but differing from the HRS, the first baseline survey does not cover the institutionalized elderly. Overall, 6,171 households and 10,254 individuals participated in the baseline survey, and there are 10,030 observations with complete information.
A MODEL OF THE EMPLOYMENT OF OLDER ADULTS
Below, we examine determinants of the labor supply (employment) of older workers, recognizing that important correlates, such as access to pensions, may be systematically related to both the ability to retire and unobservable characteristics (e.g., ability). While these models should be viewed as providing descriptive evidence, we choose measures of health status and proxies for wealth and family characteristics with the aim of minimizing endogeneity biases.
17 Even though older residents of both this survey, as well as the CHNS in China, do not move or split off with the same frequency as household members who were under 25 during the initial wave, one might nonetheless be concerned about this potential source of bias.
The three data sources used to examine employment outcomes in China (CHARLS 2008), Indonesia (IFLS, 2007), and Korea (KLoSA, 2006) provide roughly comparable information on labor supply, health status, and education of both the respondent and spouse, and information on parents and children regardless of current residence in the household.18 We estimate the reduced form labor supply model in each country:
where labor supply, LSi, is a binary indicator of whether individual i worked for one hour or more during the previous week. We expect that higher values of educational attainment of elderly, Ei, will be associated with higher wealth and savings and, as leisure is a normal good, may be negatively related to elderly labor supply.19 Similarly, we expect that access to own pension income, Peni, and pension income of a spouse, Pen_i, will be negatively related to employment.
The health of older workers and the elderly are measured using two z-scores calculated from responses to a set of activities of daily living questions, ADL.20 For each of the three data sources, we make use of questions related to activities of daily living to construct two indices: one is based on the number of activities that the respondent may perform with difficulty (Difficulty), and the second is based on the number of activities that a respondent cannot perform (Unable). Because each survey has some differences in activities and asks a different number of activities, we use these responses to construct within country z-scores
18 An important benefit of knowing the numbers of grandchildren and living parents independent of whether they are members of the household is that we are able to control for potential demand for a family care-provider without introducing biases related to endogenous household composition, as would occur if relying on information about household members alone.
19 Of course, an individual with more education may also be able to earn significantly higher returns, and so the coefficient on education will reflect the net effect of returns and accumulated wealth on employment.
20 Bound (1991) cautions that general health status questions are likely to be correlated with unobservable individual characteristics and, further, that they may suffer from justification bias. Several studies (e.g., Bound, 1999; Dwyer and Mitchell, 1999) have suggested that proxies constructed from ADLs do not suffer from such serious bias. Bound, Stinebrickner, and Waidmann (2010) cautions that financial wealth may affect ADL outcomes, and that even proxies developed from ADLs may lead to underestimating the negative effects of poor health on labor supply. As we do not yet have appropriate panel data for the three countries in this study, we are not able to control for dynamic relationships between health and wealth.
(respondent count - average count)/(standard deviation of count).21 Increases in each of these two z-scores reflect declining physical ability and worsening general health status. A particular value of the Unable-ADL z-score will be associated with more severe limitations than the same value of the Difficulty-ADL z-score.
We expect that declining health will have a negative impact on work activity, particularly for those workers in occupations in which physical strength and mobility are important (Bound, 1999). As the education of a spouse or adult children will be related to additional sources of income (from spouse income or transfers from children), we include average education of these family members,_i. Finally, we control for a vector of individual and household characteristics, Xi, which include age and age-squared and are associated with own productivity, numbers of grandchildren, and living parents of the head and spouse, which are associated with preferences for employment, and the ln (1 + per capita value of the household dwelling), as a proxy for household wealth.
Within the retirement literature in the United States, recent research has focused on the important roles of spouse employment and spouse health status in labor supply and retirement decisions. First, the retirement decisions of husbands and wives may be interdependent. Structural models (Blau, 1994; Gustman and Steinmeier, 2004) suggest that labor supply decisions of older couples reflect preferences for shared retirement. Second, work decisions may be affected by the health status of a spouse. One may plausibly observe an added worker effect, in which a spouse’s health shock leads to increased labor supply so as to ensure income against the earnings-loss associated with the health shock (e.g., Coile, 2004a), or alternatively find that spouse care requirements will require exit from the labor force (e.g., McGeary, 2009). To gauge the potential importance of these factors in China, Indonesia, and Korea, we next estimate model (2) below:
where LS_i is an indicator of whether or not a spouse is employed, and ADL_i are measures of spouse health status. As the labor supply decisions of husbands and wives are likely to be determined jointly and have a
21 CHARLS asks 19, IFLS asks 18, and KLoSa asks 17. Each of the surveys asks questions related to functional abilities such as the ability to walk for a kilometer, go up and down stairs, lift 10 or 20 lbs., draw water from a well, and perform household chores, bathe oneself, or take medication.
dynamic relationship with health and shocks to health and employment, we view these models as purely descriptive but informative of the extent to which joint labor supply decisions may affect the timing of retirement. As the HRS-type surveys mature into full panel studies, it will be feasible to control for unobservables and to unlock directions of causality among these variables.22
DETERMINANTS OF “RETIREMENT” IN CHINA, INDONESIA, AND KOREA
Employment and Pension Availability
Descriptive patterns of employment in productive activity across urban and rural areas of China, Indonesia, and Korea (see Figure 6-3) are suggestive of the role that eligibility for a pension may play in retirement of older workers and elderly. In these figures, the line shows the share of population in two-year age cohorts that is employed in some productive activity (salaried work, informal sector wage labor, casual work, self-employed activities, or unpaid employment in family-run enterprises) and the bars represent the pension-eligible share. In urban China, the employed share decreases sharply with increases in the pension-eligible share. In rural areas of China, one does not observe a similar decline in employment, and shares of workers with access to pensions are quite low. In Indonesia and Korea, where pension coverage rates at young ages are lower than for urban China, one observes a much more gradual decline in the employment of men in urban areas, and a far less pronounced decline in rural areas of both countries.
In Table 6-1, we show results from estimating the labor supply model presented in model (2) above.23 The results are suggestive of the role that pension income plays in decisions to exit from employment and the labor
22 In estimating models (2) and (3), it should be noted that for each country, there are subsamples of the population for which there is no spouse or for which spouse information is not available, and this may lead to concerns about biases introduced by selection into marriage. In the models presented in this paper, we have handled this problem by including indicator variables for marital status and for absence of spouse information, and then set spouse information to zero for those observations for which no data on a spouse are available (for ADL z-scores, of course, this corresponds to the mean). We have also estimated these models on the subset of the data for which we have information on both spouses. We take comfort in the fact that there is no appreciable difference in the coefficients of interest across models using the full sample and the one estimated on married couples with spouse information.
23 We present linear probability models by gender and residence location (urban and rural). While magnitudes differ somewhat, marginal effects using probit models do not lead to significant qualitative differences in results.
FIGURE 6-3 Employment patterns and access to pensions across rural and urban areas of China, Indonesia, and Korea.
SOURCES: Calculated using the 2008 CHARLS pilot (China), 2007 IFLS (Indonesia), and the 2006 KLoSA (Korea).
force. After controlling for age, education, health status of respondent and spouse, proxies for wealth, and family demographic characteristics, men and women in urban China are 15.2% and 18.3% less likely to be working if eligible for a pension. Among the rural population, we observe a marginally significant negative relationship between pension eligibility and employment for men, but no effect on employment of women. In Indonesia, where coverage by pensions is most common among retired civil servants, access to pensions is also negatively associated with work activity for both genders in rural and urban areas. In Korea, by contrast, access to a pension only affects the employment status of men in urban areas. This may not be surprising if pension coverage or income is particularly low in Korea, or is earned by wealthier households whose labor force participation is affected by other factors. Lee (2009) has noted that although coverage expansion to rural areas was an objective of the National Pension Scheme after 1988, wealth distribution in rural areas remains skewed and that the majority of rural households lacked sufficient pension income or wealth to retire.
Ability to retire, of course, not only depends on pension access, but also on whether a household has sufficient accumulated wealth to allow for retirement. A respondent’s years of schooling will be associated both with lifetime wealth, and with potential current returns available in the labor market. In the models reported in Table 6-1, we observe a nonlinear relationship between years of education for urban and rural women and their probability of working. For urban Chinese women, probability of working declines with each additional year of schooling through completion of middle school, and then starts to increase with higher levels of education. We also observe that increases in housing wealth per capita in urban China are associated with reduced probability of employment. In rural China, the probability of male and female employment declines with education through 11 years of schooling. As educational attainment is not very high in rural China, increasing education (and lifetime accumulated wealth) is associated with reduced likelihood of employment for all but those individuals who have completed high school. The more highly educated rural women are likely to be working in village administration, run small enterprises, or be involved in other activities that have higher returns and do not require the same physical demands as work in agriculture. Turning to Indonesia and Korea, we only observe a similarly strong negative relationship between years of schooling and participation in work among women in urban Indonesia.
The pension income of a spouse, as well as other incentives affecting spouse retirement decisions, may also influence the retirement decision (Coile, 2004b). In the results shown in Table 6-1, we find that spouse pension eligibility has a significant impact on the employment status of rural
Table 6-1 Pension Eligibility and Health Status in the Labor Supply of Adults Aged 45 and Older
|Linear Probability Models, Dependent Variable: Worked at Least One Hour per Week|
|Years of Education||0.003||-0.032*||-0.009||-0.040**|
|Years of Education-Squared||0.000||0.003***||0.001||0.004**|
|Spouse Pension Eligible||0.048||0.029||0.060||-0.182***|
|Average Education of Spouse and||-0.006||0.003||-0.008||0.007|
|Ln(Housing Wealth P.C.+1)||-0.061**||-0.013||-0.010||-0.028*|
|Number of Living Parents||0.011||0.065**||0.018||-0.003|
|Number of Grandchildren||0.005||-0.009||-0.009**||-0.006|
|ADL Z-Score (w/difficulty)||-0.049||-0.027||-0.075***||-0.072***|
|ADL Z-Score (unable)||-0.003||-0.010||-0.080***||-0.050***|
|Indonesia (IFLS)||Korea (KLoSA)|
women. A rural woman married to someone with a pension is likely to be in a much wealthier household. Pension recipients in rural areas tend to be retired cadres with much greater lifetime savings. Lack of an effect of spouse pension eligibility on employment of women in urban China likely reflects the fact that urban women, who are 5 to 10 years younger than their spouses, are frequently pension-eligible and already out of the labor force. A husband’s pension eligibility does not have an independent effect on labor supply.
Work and Health Status
For those workers in developing countries involved in manual tasks, such as work in agriculture, we would expect to observe a strong relationship between health status and labor force participation. From the results shown in Table 6-2, health status has a far more pronounced effect on work activities of China’s rural residents than urban residents. A one standard deviation increase in the Difficulty-ADL z-score is associated with 7.5% and 7.2% declines in probabilities of working for rural men and women, respectively, and a one standard deviation increase in the ADL-Unable z-score is associated with an 8.0% reduction for men and 5.0% reduction for women. In contrast to China, decline in physical functioning has a negative effect on work status of men and women in both urban and rural areas of Indonesia. Given that urban residents of Indonesia tend to work until much later in their lives, frequently in self-employed activities, the difference in importance of health status between urban residents of China and Indonesia makes sense. China’s urban residents retire before marked declines in physical functioning. Finally, in Korea, poor health status has a statistically significant negative effect on employment of men in both urban and rural areas, but the effect is more pronounced in rural areas where work is likely to be more strenuous.
Family Care Provision and Employment?
One frequently raised hypothesis concerning the exit of older women from the labor force lies with growing demands for provision of elder care and care of grandchildren. In order to reduce potential bias, the models estimated in Table 6-2 use numbers of living parents and grandchildren, respectively, rather than presence of parents or grandchildren as household members as covariates. This distinction is important as a significant negative correlation when an elder is present in the household may lead to misleading causal interpretations if arrangements for care-provision
and older worker labor supply are jointly determined.24 The lack of a systematic negative relationship between employment of China’s older urban workers and the number of family members for whom they might provide care, suggests that the explanation for decline in women’s labor force participation lies elsewhere. In Indonesia and Korea as well, the composition of the extended family does not seem to be strongly and systematically associated with labor supply and retirement decisions.
Interdependence of Spouse Retirement Decisions
In Table 6-2, we present results from estimates of model (2), which provide insight into the joint retirement decisions of spouses and the role of spouse health status in retirement. Across rural and urban areas of China, Indonesia, and Korea, men are more likely to work if their spouses are working as well. In urban China, where the effect of a spouse working is associated with a 19.6% and 20.2% higher probability of employment for men and women, respectively, one policy implication of joint retirement decisions is that efforts to encourage retirement at older ages for both men and women will likely have a positive impact on employment of spouses (Falkinger, Winter-Ebmer, and Zweimuller, 1996). Moreover, in rural China and both urban and rural areas of Indonesia, we find that a woman’s employment is more strongly correlated with spouse work status than men’s. In part, this reflects the youth of wives relative to their husbands and the likelihood that women will choose to retire at roughly the same time as their older husbands when there are (effectively) no gender differences in mandatory retirement and pension eligibility ages. In Korea, we observe a positive correlation between employment of men and the work status of wives in urban areas, but a strong and roughly equal positive association between the employment of husbands and wives in rural Korea, which likely reflects joint decisions to retire from farming among older rural Koreans.
Employment and Spouse Health Status
Inclusion of Spouse Difficulty-ADL and Spouse Unable-ADL z-scores among covariates in model (2), and presented in Table 6-2, allow us to shed light on the relationship between spouse health status and respon-
24 Research using census data, which is unable to control for endogeneity of household composition, provides weak evidence that eldercare and childcare may explain a small share of women’s exits from the labor force (Maurer-Fazio et al., 2011). Giles et al. (2006b) find that urban women are more likely to work if they have a college-age adult child, which may reflect a labor supply response to the sharp increase in postsecondary tuitions after 1996.
TABLE 6-2 Spouse Health and Work Status in Labor Supply Decisions of Adults Aged 45 and Older
|Linear Probability Models, Dependent Variable: Worked at Least One Hour per Week|
|Spouse Pension Eligible||0.106||0.087||0.084||-0.136**|
|Ln(Housing Wealth P.C.+1)||-0.062**||-0.001||-0.007||-0.026|
|ADL Z-Score (w/difficulty)||-0.061||-0.027||-0.083***||-0.083***|
|ADL Z-Score (unable)||0.005||-0.005||-0.079***||-0.048***|
|Spouse ADL Z-Score (w/difficulty)||0.068||0.031||0.031**||0.052***|
|Spouse ADL Z-Score (unable)||-0.016||-0.014||0.003||0.003|
|Indonesia (IFLS)||Korea (KLoSA)|
dent work decisions. Similar to the relationship with own health status, we find no correlation between spouse health status and employment in urban China. In rural China, by contrast, increases in the number of functions that a spouse has difficulty performing (Spouse Difficulty-ADL z-score) are positively associated with labor supply, suggesting an added worker effect dominates. A one standard deviation increase in the Spouse Difficulty-ADL z-score is associated with 3.1% and 5.2% increases that men and women, respectively, will be employed. The gender difference in the added worker effect, evident in rural areas of China and Indonesia, stands in contrast to the results using data from the U.S. Health and Retirement Study, which suggest that shocks to health of a spouse have a small positive effect on labor force participation of men and no effect on women (Coile, 2004a). While results for the United States can be interpreted as suggesting that there is little opportunity to smooth income loss associated with health shocks through the labor market (as Coile concludes), the employment response for both genders in rural China and men in rural Indonesia likely reflects both a stronger need to smooth income and fewer constraints to returning to work on the family farm than when looking for formal wage work.
As in other regions of the developing and developed world, population aging in China raises the prospect that both formal and informal mechanisms for supporting the elderly will come under strain over the next 20 years. In common with Indonesia and other developing countries, however, China is experiencing population aging at lower income levels and prior to the extension of pensions to rural residents and to urban residents in the informal sector. In a sense, China has two retirement systems: a formal system, under which urban employees receive generous pensions and face mandatory retirement by 60, and an informal system, under which rural residents and individuals in the urban informal sector rely on family support in old age and have much longer working lives. The retirement patterns presented in this chapter illuminate the employment context for the decisions that China’s policymakers are currently facing as they work to extend new pension programs in rural areas and to the urban informal sector. Several issues warrant consideration when thinking about employment-related policy for an aging population.
First, as researchers have found in the United States, the United Kingdom, and Organisation for Economic Co-operation and Development (OECD) economies, we observe a strong association between pension eligibility and exit from productive employment in China, Indonesia, and Korea. Moreover, in rural areas of China and Indonesia, where
work is physically demanding, we observe a strong correlation between employment status and physical functioning abilities. In China, this raises concerns that the “ceaseless toil” characterization of rural elderly lives (Davis-Friedman, 1991) may remain accurate for a significant share of the rural elderly population. China’s government has taken steps recently to improve pension support for the elderly. In 2009, a Rural Pension Pilot scheme was rolled out, and current plans are to extend it to all counties before the end of the 12th Five-Year Plan in 2012. The New Rural Pension Plan is a contributory plan with matched public contributions. Participants are eligible to receive the pension at age 60 with 15 years of contributions (or an equivalent buy-in).25 Receiving pensions from these sources would not contain a mandate that the recipient stopped working, but also the income would facilitate reduced work, whether complete or gradual.
Second, the gender disparity in mandatory retirement and pension eligibility ages for formal sector workers creates strong incentives for women to exit productive work at younger ages. While labor force participation rates of women are similar to those of men at younger ages, they fall precipitously after age 40. The evidence presented and reviewed in this chapter suggests that the probability an urban woman is employed is strongly related to pension eligibility, which also corresponds to working for an employer enforcing mandatory retirement. Changing the age of pension eligibility rapidly may cause hardship for those women who are close to the current retirement age and want to retire, but allowing women to retire at the same age as their male counterparts would remove an obstacle that women face relative to men when looking for work later in life. In Indonesia and Korea, where there is no difference in mandatory retirement ages for men and women in the civil service or formal sector, women’s labor force participation also declines after 45, but these declines are not nearly as steep as those observed for urban Chinese women.
Apart from the disparity between mandatory retirement age for men and women from government and formal sector employers, retirement ages are quite low for those with formal sector employment in urban China. Given the rate of population aging, the argument that mandatory retirement is important for providing opportunities for younger workers makes less sense.26 Indeed both macroeconomic and fiscal considerations warrant encouraging workers to remain employed until older ages. As noted above, eliminating (or raising) mandatory retirement ages will be
25 To cover the urban informal sector (the self-employed and workers whose employers are not participating in employer-based programs), a New Urban Residents Pension Scheme was announced in June 2011, with roll-out of pilots beginning in July 2011.
26 Gruber and Wise (2010) raise questions as to whether older and younger workers are substitutes, and the extent to which raising retirement ages could limit opportunities for new entrants into the workforce.
less problematic than increases in the age of pension eligibility, but both are likely to be unpopular because they force changes in expectations and long-term planning. One politically palatable approach may be to gradually raise the retirement age in three-month increments. Were such a reform started during the 12th Five Year Plan period, the retirement age for men would reach 65 by 2030, though it would take longer at this pace for full equalization for women. Research conducted in the United States and Europe, however, suggest that one might provide incentives within the pension system to encourage retirement later in life.27 Moreover, correlations in retirement of spouses, reflecting coordination of retirement planning, raises the prospect that eliminating disincentives for women to remain in the labor force after 50 may encourage delayed retirement of their husbands as well.28
Among older residents who work, the paper has shown that reductions in work hours are quite gradual. A key area of employment experimentation in OECD countries is through introduction of flexible work arrangements, accompanied by removal of mandatory retirement ages and promotion of “job-sharing,” which has received positive reviews as a component of labor market reforms in Germany. China could benefit from assessment of international lessons and expansion of pilots domestically in these areas.29
While this study has focused primarily on the relationship between pension eligibility, mandatory retirement, and work activity, the positive relationship in urban China between educational attainment and continued employment at the high end of the education distribution (for those with more than high school education) reflects the possibility that workers with more skills, or with the ability to learn new skills, may find it easier to work at older ages. In a review of policies followed in Europe, the OECD has recognized this phenomenon and noted that support for skills-upgrading at mid-career can be attractive for employers and employees alike, and may help to enhance the skills and employability of workers later in their careers.
27 Coile and Gruber (2007) find that changes in expected Social Security benefits in the United States have an impact on retirement planning well ahead of retirement. Gustman and Steinmeier (2009) and Vere (2011) find that changes in social security rules or benefits help to increase the labor force participation of older workers, and may even lead to increases in hours worked “after retirement” in one’s 70s. Robalino et al. (2009) suggest that changes to social insurance policies in Brazil could have an important impact on the labor supply and retirement decisions of older workers.
28 Falkinger, Winter-Ebmer, and Zweimuller (1996) find that increasing the retirement age of women through social security reforms may lead to longer working lives for men as well.
29 OECD (2006) provides a useful review of policies and approaches to reducing barriers and disincentives to continue working.
Appleton, S., J. Knight, L. Song, and Q. Xia. (2002). Labor retrenchment in China: Determinants and consequences. China Economic Review 13:252-275.
Benjamin, D., L. Brandt, and J. Fan. (2003). Health and labor supply of the elderly in rural China. Unpublished manuscript, Department of Economics, University of Toronto.
Blau, D.M. (1994). Labor force dynamics of older men. Econometrica 62(1):117-156.
Bound, J. (1991). Self-reported versus objective measures of health in retirement models. Journal of Human Resources 26(1):106-138.
Bound, J. (1999). The dynamic effect of health on labor force transition of older workers. Labor Economics 6:179-202.
Bound, J., T. Stinebrickner, and T. Waidmann. (2010). Health, economic resources and the work decisions of older men. Journal of Econometrics 156(1):106-129.
Cai, F., J. Giles, P. O’Keefe, and D. Wang. (2012). Old-Age Support in Rural China: Challenges and Prospects. Directions in Development Monograph. Washington, DC: World Bank.
Cai, F., A. Park, and Y. Zhao. (2008). The Chinese labor market in the reform era in China’s great economic transformation. Pp. 167-214 in The Chinese Labor Market in the Reform Era, L. Brandt and T. Rawski (Eds.). Cambridge: Cambridge University Press.
Chan, S., and A.H. Stevens. (2001). Job loss and employment patterns of older workers. Journal of Labor Economics 19:484-521.
China Health and Retirement Longitudinal Survey. (2008). China Center for Economic Research, Peking University. Available: http://charls.ccer.edu.cn/charls/data.asp.
China Health and Nutrition Survey. (1991, 2009). Available: http://www.cpc.unc.edu/projects/china.
Cho, J.M., and S.W. Kim. (2005). On using mandatory retirement to reduce workforce in Korea. International Economic Journal 19:283-303.
Coile, C.C. (2004a). Health Shocks and Couples’ Labor Supply Decisions. NBER Working Paper 10810. Cambridge, MA: National Bureau of Economic Research.
Coile, C.C. (2004b). Retirement incentives and couples’ retirement decisions. Topics in Economic Analysis and Policy 4(1):1-28.
Coile, C.C., and J. Gruber. (2007). Future social security entitlements and the retirement decision. The Review of Economics and Statistics 89(2):234-246.
Dalen, H.P.V., K. Henkens, and S. Joop. (2010). Productivity of older workers: Perceptions of employers and employees. Population and Development Review 36:309-330.
Davis-Friedman, D. (1991). Long Lives: Chinese Elderly and the Communist Revolution, Expanded edition. Stanford, CA: Stanford University Press.
Dong, X. (2010). Parental care and married women’s labor supply in urban China. Feminist Economics 16(3):169-192.
Dwyer, D.S., and O.S. Mitchell. (1999). Health problems as determinants of retirement. Journal of Health Economics 18:173-193.
English Longitudinal Study of Ageing. (2008/2009). Available: http://www.esds.ac.uk/longitudinal/access/elsa/l5050.asp.
Falkinger, J., R. Winter-Ebmer, and J. Zweimuller. (1996). Retirement of spouses and social security reform. European Economic Review 40:449-472.
Giles, J., A. Park, and F. Cai. (2006a). How has economic restructuring affected China’s urban workers? China Quarterly 185 (March 2006):61-95.
Giles, J., A. Park, and F. Cai. (2006b). Reemployment of dislocated workers in urban China: The roles of information and incentives. Journal of Comparative Economics 34(3) (September 2006):582-607.
Giles, J., A. Park, and D. Wang. (2011). Expanding social insurance coverage in urban China: The evololution of policy and participation. Mimeo, World Bank.
Giles, J., D. Wang, and W. Cai. (2011). The Labor Supply and Retirement Behavior of China’s Older Workers and Elderly in Comparative Perspective. Policy Research Working Paper 5853. Washington, DC: World Bank.
Giles, J., D. Wang, and A. Park. (2012). Expanding Social Insurance Coverage in Urban China. Policy Research Working Paper. Washington, DC: World Bank.
Giles, J., D. Wang, and C. Zhao. (2010). Can China’s rural elderly count on support from adult children? Implications of rural to urban migration. Journal of Population Ageing 3(3):183-204.
Gruber, J., and D. Wise. (1999). Social Security and Retirement around the World. Chicago: University of Chicago Press.
Gruber, J., and D. Wise. (2004). Social Security Programs and Retirement around the World: Micro-estimation. Chicago: University of Chicago Press.
Gruber, J., and D. Wise. (2010). Social Security Programs and Retirement around the World: The Relation to Youth Employment. Chicago: University of Chicago Press.
Gustman, A.L., and T.L. Steinmeier. (2004). Social security, pensions and retirement behavior within the family. Journal of Applied Econometrics 19:723-737.
Gustman, A.L., and T.L. Steinmeier. (2009). How do changes in social security affect retirement trends? Research on Aging 31:261-290.
Health and Retirement Study. (2008). Available: http://hrsonline.isr.umich.edu/.
Indonesia Family Life Survey, Wave 4. (2008). Available: http://www.rand.org/labor/FLS/IFLS/ifls4.html.
Kanbur, R., and X. Zhang. (1999). Which regional inequality? The evolution of rural-urban and inland-coastal inequality in China from 1983 to 1995. Journal of Comparative Economics 27(4):686-701.
Kinsella, K., and W. He. (2009). An Aging World: 2008. U.S. Census Bureau, International Population Reports, PS95/09-1. Washington, DC: U.S. Government Printing Office.
Korean Longitudinal Study of Ageing. (2006). Available: http://www.kli.re.kr/klosa/en/about/introduce.jsp.
Krueger, A.B., and J. Pischke. (1992). The effect of social security on labor supply: A cohort analysis of the notch generation. Journal of Labor Economics 10:412-437.
Lee, C. (2009). Labor Force Participation of Older Males in Korea: 1955-2005. NBER Working Paper Series 14800. Cambridge, MA: National Bureau of Economics Research.
Lee, C., and J. Lee. (2011). Employment Status, Quality of Matching and Retirement in Korea: Evidence from the Korean Longitudinal Study of Aging. RAND Working Paper Series WR-834. Santa Monica, CA: RAND.
Lee, R., and A. Mason. (2011). Population Aging and the Generational Economy: A Global Perspective. Abingdon, Oxon, UK: Edward Elgar.
Maurer-Fazio, M. (2007). The role of education in determining labor market outcomes in urban China. Pp. 260-275 in Education and Reform in China, E. Hannum and A. Park (Eds.). London and New York: Routledge.
Maurer-Fazio, M., R. Connelly, L. Chen, and L. Tang. (2011). Childcare, eldercare, and labor force participation of married women in urban China, 1982-2000. Journal of Human Resources 46(2):261-294.
McGeary, K. (2009). How do health shocks influence retirement decisions? Review of Economics of the Household 7:307-321.
McKee, D. (2006). Forward Thinking and Family Support: Explaining Retirement and Old Age Labor Supply in Indonesia. CCPR Working Paper No. 005-06. Los Angeles: California Center for Population Research, University of California.
National Bureau of Statistics. (2011). Communiqué of the National Bureau of Statistics of People’s Republic of China on Major Figures of the 2010 Population Census (No. 1).
National Bureau of Statistics of China, Beijing. Available: http://www.stats.gov.cn/english/newsandcomingevents/t20110428_402722244.htm.
Organisation for Economic Co-operation and Development. (2006). Live Longer, Work Longer. Paris: OECD Publishing.
Ravallion, M., and S. Chen. (2007). China’s (uneven) progress against poverty. Journal of Development Economics 82(1) (January 2007):1-42.
Robalino, D.A., E. Zylberstajn, H. Zylberstajn, and L.E. Afonso. (2009). Ex Ante Methods to Assess the Impact of Social Insurance Policies on Labor Supply with an Application to Brazil. World Bank Social Protection Discussion Paper No 0929. Washington, DC: World Bank.
Sohn, M.J. (2007). The characteristics of the recent trend of self-employment. SERI Economic Focus 146:39-55.
Vere, J.P. (2011). Social security and elderly labor supply: Evidence from the Health and Retirement Study. Labor Economics 18(5):676-686.