WPS6056 Policy Research Working Paper 6056 Recent Labor Market Performance in Vietnam through a Gender Lens Gaëlle Pierre The World Bank East Asia and Pacific Region Poverty Reduction and Economic Management Unit & Human Development Unit May 2012 Policy Research Working Paper 6056 Abstract This paper provides an overview of the recent women especially, while educated women benefit from a performance of the labor market in Vietnam during skills wage premium. Reassuringly given the large youth the Great Recession. The analysis uses data from the share of the total workforce, the youth labor market Labor Force Survey and the Vietnam Household Living is dynamic and outcomes for youths have improved. Standard Survey. The author finds that, notwithstanding Meanwhile, participation in poverty alleviation programs the global crisis and domestic volatility, job creation and labor market programs has not changed, and few has been sustained in Vietnam, especially in the formal workers use the newly created employment services and sector, but that the overall quality of employment has unemployment benefits. suffered. Gender differentials are found to affect older This paper is a product of the Poverty Reduction and Economic Management Unit and the Human Development Unit, East Asia and Pacific Region. It is part of a larger effort by the World Bank to provide open access to its research and make a contribution to development policy discussions around the world. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at gpierre@worldbank.org. The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. 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. Produced by the Research Support Team Recent labor market performance in Vietnam through a gender lens Gaëlle Pierre1 Keywords: Vietnam, labor market outcomes, gender wage differentials JEL codes: J21, J31 Sector Boards are SOCPT and GEN 1 World Bank; gpierre@worldbank.org. This paper was done under the guidance of Daniel Mont and Christian Bodewig and with the excellent research assistance of Hieu Nguyen. The paper benefited from comments received during the workshop on gender and labor markets held in Hanoi on July 12, 2011. All remaining errors are my own. Introduction Vietnam has known significant changes over the past 25 years. The country has become gradually more integrated in the global economy, culminating in joining the WTO in 2006, shortly before a time of financial and economic turmoil in the world. The impact of the crisis translated in a slowdown in economic growth: the average annual growth rate was reduced to 6.5% in 2008 and 5.3% in 2009 from rates of 8.5% in 2004-2007. This impact was lower than in other Asian countries that went into recession. At the same time, the recent period 2007-2009 saw a renewed increase in the labor force participation rate; the slowing economy has hence faced the challenge of creating jobs. Some evidence has suggested that most of these newly created jobs occurred in the informal economy and that the quality of employment has decreased (Nguyen et al. 2010; Cling et al, 2010). In addition to achieving economic growth and promoting job creation, a challenge for any government is to make economic growth inclusive. This means in particular ensuring that the poorest sections of the population are living in conditions that are acceptable in terms of nutrition, access to services, and access to housing. Moreover, this also means making sure groups that may be at a particular disadvantage, such as in terms of access to gainful employment (for example, the youth, women, the unskilled, and ethnic minorities), also benefit from improved economic conditions and are not disproportionately affected in times of downturns. Having a sound set of macroeconomic policies that foster economic growth and job creation is a necessary component for achieving these aims. At the same time, having a comprehensive social protection system that helps individuals and families cope during times of hardship is a timely and worthwhile endeavor for countries that have reached a sufficient level of development. In 2007, Vietnam introduced important reforms in this area, introducing unemployment insurance, and changing the laws that regulate pensions, sickness and maternity benefits, and work injury. The motivation for this paper is threefold. First, given the availability of recent labor market data from newly improved datasets, the paper gives a picture of the evolution of the supply side of the labor market in Vietnam in the recent period (2007-2009), putting it 2 in perspective with the evolution that took place in the period 1997-2007,2 and with the current performance of labor markets in comparator countries. Second, given the challenges described above, the paper then focuses on the quality of employment (wages, benefits, working conditions), migration patterns, and households’ coping strategies and the impact of poverty alleviation programs. Third, the paper focuses on the relative situation of women, looking at all the above issues through a gender lens. The gender analysis that is carried out in this paper should be placed in the context of recent evidence (World Economic Forum, 2010) that suggests that although Vietnam is ranked in the middle of the range in terms of broad gender equality (72 out of 132 countries), it is doing well in terms of women’s achievements in the labor market. The main results of the paper can be summarized as follows: ï‚· Job creation has been sustained, especially in the formal sector, but the quality of employment has suffered. ï‚· Gender differentials affect older women especially. Women with higher levels of education are benefiting particularly from skills wage premiums, and from better labor market outcomes than similar men. ï‚· The youth labor market was dynamic and improved over the period. Youth, being among the most educated workers, were able to find relatively good jobs more easily than other workers. ï‚· Regional patterns show evidence of a divide between rural areas and urban areas that is common in many lower-middle-income countries. ï‚· Participation in poverty alleviation programs and labor market programs did not change. Newly introduced unemployment insurance and employment services are little used. 2 Due to a lack of comparability of the data, it is not possible to look at the evolution over the period 1997- 2010. The paper therefore compares the evolution between 1997 and 2007 to the more recent evolution. 3 1. Labor Force Survey and Vietnam Household Living Standard Survey This paper uses two main sources of data: the Labor Force Surveys (LFS) of 2007 and 2009, and the Vietnam Household Living Standard Surveys (VHLSS) of 2006 and 2008. The Labor Force Surveys of 2007 and 2009 were collected by the GSO. The 2007 LFS was conducted in the month of August among about 170,000 households using a fairly short questionnaire. The GSO implemented a newly improved questionnaire in 2009. The sample size of the 2009 survey was about 18,000 households. Households were selected randomly in two stages from the list of 15% sample enumeration areas of the 2009 Population and Housing census. All usual residents of the selected households were interviewed and enumerated. The idea is to have such LFS on a more frequent (potentially quarterly) basis. These LFS data are not directly comparable with the data from the previous period, since the methodology has changed significantly (i.e. levels cannot be compared between the two surveys; see ILO, 2010a). The VHLSS is a household survey that contains information on demographics, education, health, labor market status, consumption, assets, dwelling and non-labor income. The surveys collected information through face-to-face interviews with household heads and key commune officials. These surveys have been conducted every two years by the GSO since 2002. The sample size consists in 2006 and 2008 of about 45,000 households (about 36,000 households in the income survey and about 9,000 households surveyed on both income and expenditure) in about 3,000 communes/wards which were representative at national, regional, urban, rural and provincial levels. A subset of households that were interviewed in 2006 were re-interviewed in 2008. These are clearly identified in the 2008 dataset, and we are able to use information on 4086 such households. The labor market information collected in LFS and VHLSS is not directly comparable. Each dataset has its own reference period and questions. For example, while it is possible to construct indicators such as unemployment and informal employment in the LFS, this is not possible in the VHLSS. On the other hand the VHLSS have valuable information regarding educational attainment. We therefore make use of both types of data. 4 2. Labor market performance 1997-2010 Changing patterns: recent data show an increase in labor force participation rates The evidence shows that in recent years there was a reversal of the decline in the labor force participation found in studies of the Labor Force Surveys that covered the period 1997-2007 (ILO, 2009; Phan, 2009). The steady decline in the labor force participation for both men and women, which these studies found, was broken. That decline was largely explained by the youth staying on longer in formal education and by older workers exiting the labor force at earlier ages. In contracts, labor force participation increased for both sexes in the latter period 2007-2009. Importantly, two age groups recently increased their participation significantly: the 15-19 year olds (from 37.3 percent in 2007 to 43.8 percent in 2009) and the over 50 (from 55.6 percent in 2007 to 58.9 percent in 2009). This evolution was true for both men and women, but 15-19 years old women increased their participation more than their male counterparts (from 36.4 percent to 43.6 percent for women, against 38.1 percent to 43.9 percent for men). Although the contemporaneous local inflationary crisis of 2008 and the global financial crisis may have contributed to this evolution, such a descriptive analysis does not prove a causal relationship between the two. One possible reason for the fact that the youth and older workers changed most their participation may be that these age categories were the most flexible in terms of their labor force participation choice. It is slightly worrying that some youth decided to abandon their studies early, presumably to help their family, therefore potentially diminishing their lifetime income. It will be interesting to see whether labor force participation rates return to previous levels, and whether individuals who left school return to formal education. A high participation rate and employment-to-population ratio As it has been noted before (e.g. Phan, 2009; ILO, 2010a), labor force participation is high in Vietnam; it is toward the high-end of the range of the participation rates of other lower-middle-income countries (Table 1). The participation rate of women is lower than for men in all countries, but the participation of women in the labor force in Vietnam has 5 remained high compared to the other countries. Moreover, the employment-to-population ratio for both sexes, but for women especially, is high compared to comparator countries. Throughout its transition to a market economy, its experience of financial and other crises, Vietnam has successfully maintained a high proportion of women within the labor force and in employment. This is consistent with previous findings (Pham et al., 2007). Table 1: Main labor market indicators - % - (2008) Labor for participation (15+) Employment-to-Population ratio Unemployment rate (15+) All Men Women All Men Women All Men Women Cambodia* 80.4 85.9 75.5 64.8 65.1 64.5 - - - Indonesia 67.2 83.5 51.1 61.5 77.1 46.1 8.4 7.6 9.7 Philippines 63.7 78.9 48.6 58.6 72.3 44.1 7.4 7.6 7.1 Thailand 73.4 81.5 65.8 72.4 80.4 64.9 1.2 1.3 1.0 Vietnam* 76.5 81.0 72.3 74.5 79.0 70.4 2.6 2.5 2.7 Sources: Laborsta (ILO) – Labor Force Survey data. *Cambodia (2004) 10+; Vietnam (2009, VET2) Low unemployment rates The evolution described above translates into low unemployment rates, and little difference in unemployment rates between men and women (Table 1). A low unemployment rate in a lower-middle-income economy cannot be interpreted in the same way as in developed economies. In the latter, comprehensive social security systems, and in particular unemployment benefits, mean that workers who are out of a job have the opportunity and time to look for another suitable job. The poorest people are therefore found among the unemployed. In middle-income countries that offer no or limited financial help to those looking for work, workers who are newly unemployed often cannot afford to stay out of a job very long and therefore take up new employment more quickly, even if the job is not suitable in terms of pay, required qualification, etc. Often, such jobs are found in the informal sector, which has much more flexible labor practices than the formal sector, but which also afford lower employment security. The unemployed are not therefore especially the poorest people; they are on the contrary those who can afford to remain unemployed. The structure and quality of employment therefore matter even more than unemployment in these countries. There was a modest increase in the unemployment rate between 2007 and 2009. The introduction of unemployment insurance just before the financial crisis could have been associated with a large increase in the proportion of people registering as unemployed, 6 both because of greater job destruction in the economy and greater incentive to stay unemployed longer to look for a new job. The increase in the unemployment rate over the period was however small (from 2 percent to 2.6 percent) reflecting either a low take-up of unemployment benefits or simply a lack of net job loss, or both. The fact that both labor force participation and the employment-to-population ratio increased over the period suggests that there was a net increase of entrants in the labor force and that job creation nearly kept up with this net influx of workers. However, adjustments may have been done elsewhere, in particular in the type and quality of employment created. The unemployment rate among youth is much higher than for adults but has remained stable between 2007 and 2009 (Table 2). Youth unemployment is generally higher than the adult rate of unemployment because youth tend to have shorter spells of employment while gaining experience and searching for a more permanent job. In middle-income countries, this may also be due to the fact that they are more likely to work in the informal sector, which has greater turnover. However, as will be shown below, this is not the case in Vietnam. Interestingly, the youth unemployment rate has not changed since 2007, while the unemployment rate for adult increased (from a low level). Moreover, these results are found for both young men and women. Table 2: Youth labor market performance (%) Labor force participation Employment to population Unemployment rate ratio (15+) All Men Women All Men Women All Men Women 2009 Youth (15-24) 60.5 62.5 58.5 56.7 58.6 54.9 6.2 6.2 6.1 Adults (25+) 81.5 87.0 76.3 80.0 85.6 74.9 1.8 1.7 1.9 2007 Youth (15-24) 55.9 56.9 54.8 52.5 53.4 51.6 6.0 6.2 5.8 Adults (25+) 80.4 86.0 75.4 79.5 85.3 74.6 1.0 1.1 1.0 Source: LFS, 2007, 2009 Overall, many new entrants joined the labor force probably in order to complement income; the youth labor market is dynamic and youth have been relatively successful in securing employment. The labor market of youth saw an increase in labor force participation and employment-to-population ratio and a stable unemployment rate over the period 2007-2009. In addition, there is evidence that although a large proportion of 7 young workers are first-time job seekers, this share decreased between the two years (78 percent of 15-24 unemployed in 2007 were first-time job seekers (64 percent in 2009)). The overall share of first-time job seekers decreased from 60 percent to 40 percent of the unemployed between 2007 and 2009. Moreover, the share of youth in unemployment decreased over the period: they went for representing 56 percent in 2007 of the unemployed to 44 percent of the unemployed in 2009. This age group was the only one to decrease its share in unemployment. Regional patterns reveal some gender disparities A national picture does not give a full understanding of the labor market as strong regional disparities have important implications for policy. Moreover, regional patterns may reveal further disparities between groups of workers, as some may be less mobile than others. In particular, gender differentials become apparent when looking at the regional data as was noted before in Rodgers et al. (2010). The regional evidence suggests a picture where individuals in rural areas have little alternative but to work in order to contribute to the income of the household while those in urban areas are more likely to have the financial means to look for a well-suited job. Distinguishing between rural and urban areas (Table 3), it is apparent that labor force participation in rural areas is higher than in urban areas (79.1 percent versus 70.6 percent in urban areas). Similarly employment-to-population ratio is greater in rural areas (77.6 percent versus 67.5 percent in urban areas). Moreover, the unemployment rate is much higher in urban areas (4.4 percent) than in rural areas (1.9 percent). Table 3: Regional patterns of labor market performance - % - (2009) Labor force participation Employment to Unemployment rate population ratio (15+) All Men Women All Men Women All Men Women Area Urban 70.6 76.1 65.6 67.5 72.9 62.6 4.4 4.1 4.6 Rural 79.1 83.1 75.3 77.6 81.5 73.9 1.9 1.9 1.9 Region Northern Midlands and Mountain 83.5 84.0 83.1 82.5 83.1 81.9 1.3 1.1 1.4 Red River Delta 77.4 79.2 75.8 75.9 77.2 74.8 2.0 2.5 1.4 North Central Coast 76.8 79.4 74.3 74.7 77.1 72.4 2.7 2.9 2.4 Central Highlands 83.3 86.9 79.9 82.2 85.8 78.8 1.3 1.2 1.4 South East 76.5 83.9 69.6 74.8 82.1 67.9 2.3 2.2 2.5 Cuu Long River Delta 76.2 84.7 67.9 73.5 82.6 64.8 3.5 2.5 4.6 Ha Noi (new definition) 69.6 72.7 66.8 67.6 70.4 65.0 2.9 3.1 2.7 8 HoChiMinh city 67.3 76.0 59.4 63.9 72.3 56.4 5.0 4.8 5.1 Source: LFS, 2009 There are also strong differences across regions that are linked to the degree of urbanization and geographical circumstances. Higher unemployment rate are found in more dynamic provinces where workers may move in order to find work. The highest unemployment rate for men and women is found in HoChiMinh city (Table 3). Looking at the changes over the period 2007-2009 (Table 4) reveals important regional and gender differences in the evolution of labor market indicators. The labor force increased at a higher pace than the working population (thereby leading to the above- mentioned increase in labor force participation). However, Cuu Long River Delta saw a decrease in working age population and labor force. All regions saw important increases in unemployment (starting from low levels). Three regions (Northern Midlands and Mountain, North Central Coast, and Cuu Long River Delta) saw a decrease in working age population. Moreover, rural areas saw a slight decrease in working age population. Table 4: Average annual growth rate between 2007 and 2009 (%) Areas/ Regions Population 15+ Labor Force Employment Unemployment All Men Women All Men Women All Men Women All Men Women Country 0.9 1.4 0.4 2.3 3.0 1.5 1.9 2.7 1.1 18.4 15.0 22.2 Urban 4.8 4.9 4.7 8.3 8.9 7.7 7.8 8.3 7.3 20.3 23.9 16.8 Rural -0.7 -1.4 0.1 0.2 -1.0 1.3 -0.1 -1.3 1.1 16.8 20.6 13.5 Northern Midlands and -0.4 -1.5 0.7 0.8 0.02 1.6 0.7 -0.3 1.7 14.6 36.8 -2.5 Mountain Red River Delta 0.4 0.4 0.5 1.1 0.3 1.9 0.8 0.1 1.6 15.1 18.2 13.0 North Central Coast -0.7 -1.6 0.2 1.2 0.3 2.1 0.9 0.0 1.8 12.2 11.9 12.5 Central Highlands 2.1 2.9 1.3 5.2 5.9 4.6 5.3 6.1 4.5 2.7 -5.6 14.6 South East 7.3 7.1 7.5 9.6 9.4 9.8 9.2 8.9 9.4 24.3 23.0 25.5 Cuu Long River Delta -1.2 -2.0 -0.3 -0.1 -1.2 0.8 -0.8 -2.2 0.5 25.9 33.4 16.6 Source: LFS, 2007; LFS, 2009 There is evidence of a dichotomy between men and women: men left rural areas to go to urban areas; women increased participation everywhere, in particular in urban areas. The decrease in the working age population in rural areas was mostly due to a decrease in male working age population in rural areas, suggesting that men are leaving these areas. Except in Central highlands, the rate of annual growth of working age population was greater for women than for men. At the same time, the labor force of women increased in 9 rural areas while men’s labor force decreased in these areas. The growth of labor force and employment was greater for women in all regions, except in Central Highlands and South East. Educational attainment of the labor pool The skills pool of the labor supply is an important factor in determining the dynamism of the labor market of a country. A skilled labor force is more productive, better adaptable, and can contribute to the adoption of new technologies, and the development of new products. A skilled labor force can also attract foreign investors and companies. At the domestic level a skilled labor force can lead to a better functioning economy, to the creation of more successful business endeavors. Moreover, a skilled labor force is likely to lead to further skills development in future generations, as more educated parents tend to push their children to stay longer in school. Looking at this issue is important as some studies have expressed concerns about a shortage of skilled labor in the Vietnamese economy and have identified important increases in the return to higher education (e.g., di Gropello and Sakellariou, 2010). The evidence shows gender and cohort differences in skills levels. In the working age population (age 15+), 21 percent of individuals have no degree, and about 5 percent have college and above education (Table 5). Looking at 10-year cohorts shows an improvement in educational attainment for more recent cohorts, with a decrease in the share of individuals with no degree, and an increase in the share of individuals with higher degrees. Women are more likely than men to have no degree, similar probabilities to have College and University education. This suggests that apart from a small proportion of women who go on to tertiary education, women drop out of school earlier than men. The 15-24 years old have much better educational attainment than later cohorts, suggesting that better level of education may be a reason behind the dynamism of their labor market that was found above. Table 5: Educational attainment by gender and age group (2008) % No Primary Lower Upper College University Degree school secondary secondary Men 15.6 23.8 32.7 22.1 1.1 4.7 Women 26.3 22.9 28.7 17.5 1.5 3.1 10 15-24 6.7 16.5 40.9 33.5 1.0 1.4 25-34 17.4 31.0 21.9 19.6 2.4 7.7 35-44 17.6 27.7 33.8 16.1 1.1 3.7 45-54 19.8 23.7 34.8 15.8 1.5 4.3 55-64 30.6 23.8 26.9 12.4 1.0 5.4 65+ 64.1 19.4 8.8 4.8 0.5 2.4 All 21.1 23.4 30.6 19.7 1.3 3.9 Source: VHLSS 2008 But what do these numbers mean for labor market performance? Looking at this issue from the perspective of employers helps locate in which sectors of the economy the skills shortages are likely to be. The existing data provides limited information as to the mismatches that may exist between skills supply and demand in Vietnam. Looking at this issue from the employers’ side, the evidence shows that overall, about 9 percent of firms report suffering from a shortage of skilled labor.3 This number is somewhat higher than other lower-middle-income countries (Table 6), but it is not particularly high, especially when compared to low-income countries such as Cambodia or Lao PDR. Table 6: % of Firms Identifying Labor Skill Level as a Major Constraint All firms Apparel Exporter Non- Small Medium Large Domestic Foreign sector exporter firms firms firms firms firms Cambodia 15.5 - 10.3 15.6 12.8 22.0 18.1 13.4 22.3 Indonesia 4.5 4.2 2.4 4.5 4.5 3.7 6.3 4.3 12.9 Lao PDR 18.7 - 19.1 18.7 10.3 46.9 46.3 16.7 52.4 Philippines 7.8 28.7 2.4 8.2 10.0 5.3 5.5 8.4 4.1 Vietnam 8.9 11.9 5.5 9.3 5.6 11.1 12.3 7.9 18.3 Source: World Bank Enterprise Surveys (Indonesia, Lao PDR, Philippines, Vietnam 2009; Cambodia 2007) Looking beyond the national level, it becomes apparent that a significant proportion of foreign firms, as well as medium and large firms, report being constrained by shortage of skills in the labor force. It is important to look beyond the national level, as certain types of firms are more likely than others to be constrained by skills shortages. These results show that firms that tend to use more advanced technologies, and therefore are in need of more skilled workers, are facing greater constraints in terms of the availability of skills. This tends to be true in most Asian comparator countries, except the Philippines (Table 6). Compared with the other countries, the Philippines has an English-speaking workforce, and has had a policy of training its labor force to meet international demand 3 These numbers are obtained from World Bank Employer Surveys. An Enterprise Survey is a firm-level survey of a representative sample of an economy’s private sector. The surveys cover a broad range of business environment topics including access to finance, corruption, infrastructure, crime, competition, and performance measures. 11 for skills, in particular in the view to export its human capital. Although the Philippines’ strategy presents its own difficulties and specificities, these comparisons suggest that there is a potentially unmet demand for skills in Vietnam. The return to higher educational attainment There are important returns to higher education in terms of activity rates and type of employment. The patterns of labor force status by educational attainment (Table 7) suggest that college educated individuals are the most likely to participate in the labor force: they have the lowest inactivity rate, but also the highest unemployment rate. Workers with higher education are more likely to work in wage employment than workers with lower levels of educational attainment. University graduates and primary school graduates have the second highest labor force participation; only about 15 percent of these two groups are inactive. However, the similarities between the two groups end here: while more than 73 percent of primary school graduates are self-employed, more than 90 percent of University graduates are wage workers, including 66 percent of University graduate working in the state-owned sector (Table 8). This is remarkable since the latter sector includes only about 10 percent of employment. In fact, up to lower secondary schooling, over 70 percent of workers are self-employed; this share falls to about 48 percent for upper secondary education, about 14 percent for college education, and less than 9 percent for University education (Table 8). Table 7: Educational attainment and activity status Activity status No degree Primary Lower secondary Upper secondary College University Total school school school Work 69.3 84.8 75.6 64.2 88.3 83.5 74.7 Unemployed 0.2 0.2 0.2 0.5 1.2 0.3 0.3 Inactive 30.5 15.0 24.2 35.3 10.5 16.2 25.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: VHLSS 2008 Table 8: Composition of employment by educational status (%) No Primary Lower Upper College University Total degree school secondary secondary school school SELF-EMPLOYED/Private sector 0.2 0.3 0.5 1.2 0.3 0.6 0.5 Self-employed excluding private sector 77.0 73.1 70.8 47.9 13.8 8.6 65.3 OTHER HOUSEHOLDS 20.2 18.9 16.1 11.8 4.0 1.8 16.1 12 STATE-OWNED 1.0 2.1 5.2 23.2 69.3 66.2 10.1 COLLECTIVE SECTOR 0.1 0.5 1.3 1.8 3.6 3.2 1.0 PRIVATE SECTOR 1.1 3.6 4.4 8.5 7.0 12.2 4.6 FOREIGN-INVESTED 0.4 1.6 1.7 5.5 2.0 7.3 2.3 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Source: VHLSS 2008 Some gender differentials are apparent in the returns to education in terms of labor force participation and employment; these tend to be worse at lower educational levels. Women with no degree are more likely than their male counterparts to be inactive. Women with primary or secondary education are slightly less likely than their male counterparts to be working, while those with college or university education are more likely than their male counterparts to be working. The analysis so far has shown a relatively positive evolution of labor market performance with an increase in labor force participation and employment-to-population ratio for both men and women. Some regional and gender differences have appeared, illustrating that women in certain regions may be more at risk of being disadvantaged in the labor market. This analysis therefore raises several questions. In particular, has the increase in employment been in productive jobs or not? To what extent has migration influenced and/or counter-balanced regional disparities? Although it seems that most women enjoy equal access to employment, has this employment been of the same quality as that of men’s? Employment in the informal sector A challenge in most low and middle-income countries is the size of the informal sector. This sector is important because it operates outside legal boundaries, not contributing to taxation and providing working conditions that are not supervised. Moreover, employment in the informal sector is often of low productivity and has high turnover, which means workers experience repeated spells of unemployment. Of course, this sector is far from homogenous, including street vendors trying to survive on meager earnings as well as successful small entrepreneurs deliberately choosing not to burden themselves with sometimes overwhelming rules and regulations (for recent evidence on this see Bosch and Maloney, 2010). 13 The Labor Force Surveys (2007, 2009) contain the necessary information to estimate the size of the informal sector. Following the ILO definition (Hussmanns, 2004), employment is divided into three sectors: (i) agricultural sector (wage or non-wage employment); (ii) informal sector; (iii) formal sector. The informal sector is defined as businesses that are owned by a household or individual, and all other businesses (private sector, state sector, foreign-owned) that do not have a business registration to operate. Although this definition treats the agricultural sector separately, most of the employment in this sector is likely to be of an informal nature. The LFS surveys show that the share of employment in the informal sector (and agricultural sector) has decreased in the period 2007-2009 for both men and women (Table 9). Women are less likely to be in the informal sector than men, but they are more likely to be in the agricultural sector. The increase in the share of formal sector employment is found for all age groups. Table 9: Informal sector employment - %. 2007 2009 All Men Women All Men Women Agricultural employment 50.1 47.9 52.4 47.6 45.4 50.0 Informal sector employment 27.7 28.2 27.1 24.4 25.1 23.7 Formal sector employment 22.2 23.9 20.5 27.9 29.5 26.3 Young workers, especially women, are more likely to belong to the formal sector (Table 10). The share of formal sector employment decreases with age especially after the age group 35-44. Young female workers aged 15-24 are more likely than their male counterparts to belong to the formal sector. For all other age groups, men are more likely to belong to the formal sector. This result suggests that as women grow older, get married and have children, they may lose their job in the formal sector or choose to work in the informal sector that provides greater flexibility in working hours. This shows how traditional gender roles affect women’s labor market outcomes. The increase in formal sector employment has mostly taken place for the younger cohorts (15-24 and 25-34). Combined with the relatively high youth (15-24) unemployment rate, this suggests that youth who are unemployed may be waiting for wage employment in the formal sector. 14 Table 10: Informal sector employment by gender and age - % . 2007 2009 Men Women Men Women AGR INF FOR AGR INF FOR AGR INF FOR AGR INF FOR 15-24 52.5 24.1 23.5 49.5 21.2 29.4 48.6 21.6 29.8 47.4 16.0 36.6 25-34 40.1 30.4 29.5 45.4 27.7 26.9 35.7 25.2 39.2 41.5 24.6 34.0 35-44 43.8 32.2 24.0 50.2 30.8 19.0 41.4 31.0 27.6 45.9 28.9 25.3 45-54 48.9 27.8 23.3 56.3 27.7 16.1 48.0 25.3 26.7 56.8 23.5 19.8 55-64 61.7 22.7 15.6 68.5 24.8 6.7 61.9 19.2 19.0 67.4 23.2 9.4 65+ 73.9 19.0 7.1 72.9 25.0 2.1 79.2 14.3 6.6 73.5 22.9 3.6 AGR = Agricultural employment; INF = Informal sector employment; FOR = Formal sector employment 3. The quality of employment Defining and measuring the quality of employment is not straightforward, as the former depends on the perspective that one has and the latter depends on data availability. In particular, there are trade-offs between the good of greater quality of employment (in terms of benefits and wages for example) for workers and the downside for firms of greater labor costs and labor regulations that governments may choose to use to improve the quality of employment. We look in this section into four main dimensions of job quality: type of employment, benefits, hours worked, and earnings. Small share of “secureâ€? employment The structure of employment in Vietnam is such that a small share of workers is in jobs that are generally considered secure. Wage employment is typically the type of employment that offers the highest job security. It is also the most common type of employment in developed countries (for example, Western European countries have rates of wage employment of over 80 percent). Although the share of wage employment in total employment has significantly increased in Vietnam over the recent period (2007- 2009) from 30.5 percent to 33.4 percent, it remains relatively low by comparison with similar countries (Table 11). Table 11: Structure of employment in Vietnam and comparator countries in 2008 (%) Indonesia Philippines Thailand Vietnam* Wage employment 32.6 52.4 43.2 33.4 Employers 2.9 4.2 2.6 4.8 Own-account workers 47.5 31.3 31.8 44.7 Unpaid family members 16.9 12.2 22.3 16.8 Other - - - 0.3 Sources: Laborsta (ILO); Vietnam (LFS, 2009) 15 Moreover, precarious types of employment, such as unpaid work, are significant. Unpaid family workers represent a sixth of employment. The great majority of these workers are in the agricultural sector (Table 12). This means that farm households in the rural sector are a significant source of employment, especially for women. Agricultural employment is notoriously of low productivity, and offers low work hours. Unpaid family workers and workers in the agricultural sector have the lowest number of work hours (Table 13). In addition, there is evidence that unpaid work has been a growing source of employment in the period 2007-2009 (going from 12.7 percent of employment in 2007 to 16.8 percent in 2009), despite an overall decrease in the share of agriculture in total employment (from 49.3 percent to 47.6 percent). Table 12: Employment by sector in Vietnam in 2009 (%) Employment Status Agriculture Industry Service Total Employer 33.4 25.7 41.0 100 Self-account worker 63.3 8.3 28.4 100 Unpaid family worker 80.4 4.9 14.7 100 Wage worker 12.4 47.7 39.8 100 Member of cooperative 35.0 50.0 15.0 100 Others 0.0 48.4 51.6 100 Total 47.6 21.8 30.6 100 Source: (LFS, 2009) Table 13: Average weekly work hours and standard deviation (2009) Agricultural Industry Service All Employer 44.1 53.9 55.1 51.1 13.9 12.5 14.7 14.7 Self-account worker 36.6 44.4 48.4 40.6 13.8 13.9 16.7 15.6 Unpaid family worker 36.2 42.6 44.9 37.8 13.5 14.2 18.4 14.7 Wage work 46.5 51.2 46.2 48.6 14.9 9.4 11.9 11.5 Member of cooperative 34.2 45.8 50.0 42.3 12.1 15.1 20.9 16.0 Others -- 51.4 49.6 50.5 -- 7.7 11.3 9.7 Total 37.6 49.8 47.6 43.3 14.2 11.1 15.0 14.9 Source: (LFS, 2009) 16 The evolution of unpaid family work has been strongly gender biased. The increase in the share of unpaid family work in employment was due to the increase of women’s employment in this type of work. The share of working women in this type of employment went from 13.8 percent to 22.2 percent of employment between 2007 and 2009, while the share of men’s employment remained around 11.8 percent. Women shifted from own-account work into unpaid family work and to some extent into wage employment and to a small extent into being employers (Table 14).4 Table 14: Structure of employment for working age population (15+) in 2007 and 2009 (%) 2007 2009 All Male Female All Male Female Employer 3.2 3.8 2.7 4.8 6.3 3.2 Self-account worker 53.5 48.55 58.6 44.7 42.6 46.9 Unpaid family worker 12.9 11.8 13.95 16.8 11.8 22.2 Wage worker 30.0 35.4 25.45 33.4 38.9 27.5 Member of cooperative 0.1 0.2 0.1 0.1 0.2 0.1 Others 0.2 0.25 0.2 0.2 0.3 0.1 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: (LFS, 2007, 2009) This gender bias is also translated into the finding that a smaller proportion of women than men are employers (Table 14). Analysis using the VHLSS (Rodgers et al, 2010) confirms and throws light on these results. They suggest that although the non- agricultural household enterprises sector is buoyant and mostly headed by women, female headed businesses tend to be smaller in scale, with fewer employees, smaller revenues streams, lower likelihood of being licensed and higher likelihood of operating within marketplace rather than established shops. Moreover, a large share of working youth is in unpaid employment. The great majority of young people who are working are mostly working as unpaid family workers or wage workers. Between 2007 and 2009, the proportion of youth working as wage workers has increased while the proportion of unpaid family workers has decreased (Table 15). The same gender differences exist as in the 15+ population, but they are less marked; the structures of employment of young men and women are similar. 4 Nguyá»…n et al (2010) suggest that the shift between own-account work and unpaid family work may be due to misallocation of women between the two, very similar, categories. 17 Table 15: Structure of employment for the youth in 2007 and 2009 (%) 2007 2009 All Male Female All Male Female Employer 0.8 0.8 0.7 0.9 0.8 1.0 Self-account worker 18.4 14.7 22.5 17.5 15.9 19.2 Unpaid family worker 40.6 42.2 38.9 36.3 35.0 37.6 Wage worker 39.4 41.2 37.3 44.4 47.0 41.6 Member of cooperative 0.04 0.1 0.02 0.1 0.1 0.1 Others 0.8 0.9 0.6 0.8 1.2 0.4 Total 100.0 100.0 100.0 100.0 100.0 100.0 Source: (LFS, 2007, 2009) Interestingly, the share of employers is relatively high compared to other countries (Table 11). Because they have grown beyond a one-person business, employers can be considered as successful entrepreneurs. It shows workers are willing to take risks and to launch their own venture; it also suggests an environment favorable to such business expansion. It would be interesting to see how Vietnamese employers fare compared with employers in other countries in terms of their business success and job creation capacity. Unskilled employment is significant A large proportion of workers have unskilled occupations (Table 16). The differences between men and women are consistent with usual occupational gender patterns: women tend to be more present in sales and personal services, men tend to be in manual occupations. Although women are largely under-represented in high level management positions, it is a good sign that there does not seem to be much gender differential in professional occupations/officers. However, they are more likely than men to be unskilled workers. Table 16: Occupational structure of employment (%) Occupation Men Women All Leaders in all fields 1.5 0.4 1.0 Top-level professionals 4.6 4.7 4.6 Mid-level professionals 3.2 4.4 3.8 Officers 1.7 1.6 1.6 Skilled workers in personal services, security protection and sales 10.8 20.5 15.6 Skilled workers in agriculture, sylviculture, and aquaculture 16.1 13.4 14.8 Skilled handicraftsmen and other relating skilled manual workers 17.1 7.7 12.5 Assemblers and machine operators 8.8 4.4 6.7 Unskilled workers 36.2 42.9 39.4 Total 100.0 100.0 100.0 Source: (LFS, 2009) 18 Few benefits in wage employment Digging deeper into the status of wage workers, it becomes apparent that wage employment does not often afford the benefits that would be expected from this type of employment. The great majority of wage workers are employed on fixed-term contracts or without a formal contract (Table 17). Women are slightly better off than men, a third of employed women report having a permanent contract compared to a quarter of employed men. About 10 percent of women and 12 percent of men were working without having any type of contract or agreement in 2009. Although the share of workers with no contract decreased between 2007 and 2009, there was a small decrease in the share of workers under permanent contracts. Table 17: Contract types for wage workers (2009, %) 2007 2009 Men Women Total Men Women Total Permanent 28.2 38.3 32.25 26.1 34.2 29.4 Fixed term (3 years or less) 22.4 27.5 24.5 22.8 30.0 25.8 Verbal agreement 34.3 22.3 29.45 38.9 25.9 33.7 No contract 14.7 11.4 13.4 12.1 9.5 11.0 Others 0.5 0.5 0.5 0.1 0.04 0.06 Total 100 100 100 100 100 100 Source: (LFS, 2007 and 2009) Informal employment Vietnam is not the only country where such results are found. In fact, the recognition that formal sector firms may employ workers informally, or provide limited or none of the benefits prescribed by labor laws has prompted statisticians across the world to define formally the concept of informal employment. This has the advantage of providing a measure that is comparable across countries. Moreover, it has the further advantage of being relevant for any country, including developed countries, which have seen an increase in “precariousâ€? employment (i.e. employment that does not offer the usual benefits associated with permanent wage employment contracts). For measurement purposes, informal employment has been defined as unpaid family workers and wage and 19 salaried workers who do not receive social security in non-agricultural sectors (ILO, 2010a).5 This means that some workers may be informally employed in the formal sector. The results from the LFS suggest that about 47 percent of workers working in the formal sector were informally employed (i.e. are not covered by social insurance) in 2009. Being informally employed in the formal sector concerns all age groups, and men and women alike. About 15 percent of men were informally employed in the formal sector against 11 percent of women. The likelihood of being informally employed in the formal sector decreases by age group for both men and women. At all age groups, women are less likely than men to be informally employed in the formal sector. The results also suggest that women are less likely than men to be informally employed in the non-agricultural sector. While about 58 percent of non-agricultural female workers were formally employed in 2009, only 49 percent of their male counterparts were formally employed. Informal employment in Vietnam represented about 37 percent of total employment in 2009 (up from 36 percent in 2007). Men were at a disadvantage compared with women: 40 percent of male workers, 34.5 percent of female workers were informally employed (compared with 38 percent and 34 percent respectively in 2007). The small increase in informal employment was mostly due to the increase of informal employment among men. Younger cohorts of workers are at somewhat of an advantage. The share of informal employment in total employment increases with age up to the 35-44 age group and then decreases with age (Table 18). The increase in informal employment that is observed over the period 2007-2009 concerns mostly workers aged 35 to 64. Although the youth labor force participation and employment-to-population ratio increased (Table 2) and the share of youth working in wage employment increased (Table 15), the share of youth who are informally employed did not increase. Given that in the meantime the share of youth working in the agricultural sector decreased, there was an improvement in the quality of employment of youth (measured as working informally or not). The relative 5 With the LFS dataset, this means that informal employment is constructed in the following way: are considered informally employed in the non-agricultural sector all unpaid family workers and all others (employers, self-account workers, members of cooperatives) whose establishment does not provide social insurance. 20 situation of young women improved as their share in youth informal employment decreased from 45 percent to 42 percent between 2007 and 2009; at the same time, the share of young women among young agricultural workers remained around 47 percent. Table 18: Informal and agricultural employment (%) 2007 2009 Agricultural Informal Formal Agricultural Informal Formal employment Employment employment employment Employment employment Women 52.4 34.2 13.5 50.0 34.6 15.4 Men 47.9 38.1 14.1 45.4 39.9 14.7 Age 15-24 51.0 36.1 12.9 47.9 36.4 15.7 25-34 42.7 38.4 18.9 38.4 39.0 22.5 35-44 47.0 39.7 13.3 43.6 42.2 14.2 45-54 52.6 33.6 13.8 52.2 35.3 12.5 55-64 65.1 28.6 6.3 64.6 30.45 4.9 65+ 73.4 24.7 1.9 76.3 23.9 0.5 Total 50.1 36.1 13.8 47.6 37.3 15.1 Source: (LFS, 2007, 2009) Underemployment Underemployment complements the unemployment indicator in a country like Vietnam. It is a particularly interesting indicator: workers who lose their job and are unable to find similar employment may be forced to accept jobs that are below their expectations. In particular, they may be forced to work fewer hours than they would like. Persons in under-employment formally consist of “all employed persons who during a specified period work less than 35 hours a week and who wish to work additional hoursâ€? (ILO). For measurement purposes, using the LFS surveys, we use the following definition of underemployment: workers who work less than 35 hours in all jobs and want and are available to work additional hours. We also look at the hours worked in the main job, the number of jobs held, and the reasons for working less than usual. There is evidence that a significant proportion of workers are underemployed and that underemployment is a greater issue than unemployment. The proportion of underemployed6 among active workers was greater than the proportion of unemployed: while the unemployment rate was 2.6 percent, the underemployment rate was 5.4 percent. 6 Under-employment is defined as workers who work less than 35 hours a week in total (i.e. in all jobs) and want to work more and are ready to work more. 21 Men were slightly7 more affected by underemployment than women (5.0 percent of female workers were underemployed, versus 5.8 percent of male workers). Underemployment is more prevalent in rural areas where 6.2 percent of workers are in underemployment against 3.3 percent of workers in urban areas. The difference between men and women is slightly more marked in urban areas: 2.8 percent of working women are in underemployment versus 3.8 percent of working men, while in rural areas the differential between men and women is smaller: 5.8 percent of working women are in underemployment against 6.6 percent of working men. Comparing 2007 and 2009 shows that all the previous indicators worsened over the recent period: underemployment has become more of an issue in the context of the crisis. The underemployment rate was up by 2 percentage points from 3.5 percent in 2007 to 5.5 percent in 2009 (restricting sample to workers who hold 1 or 2 jobs8). Significant proportions of workers do not have sufficient hours in their main job. About 25 percent of workers held more than one job in 2009.9 The results show that 33 percent of workers spent less than 35 hours on their job in the previous week. Moreover, 85 percent worked fewer hours than “usualâ€?. Of course, those who work fewer than 35 hours in their main job are much more likely to hold several jobs than those who work full- time. While 19 percent of those who worked full-time in their main job had a second job or more, about 39 percent of those who worked less than 35 hours a week in their main job had a second job or more in 2009. The proportion of workers who held more than one job was up from 18.2 percent in 2007 to about 25 percent in 2009; this share increased significantly for both full-time workers and persons working less than 35 hours.10 What are the reasons underlying short hours? Looking at the reasons stated by workers who report working fewer hours than their usual number of hours in their main job in the previous week, we see that most (about 61 percent) do so because of reasons that are 7 This result is not statistically significant. 8 The share of persons who are in underemployment is not readily comparable since the 2007 LFS requested information about the first two jobs only. We therefore restrict the 2009 sample to those with 1 or 2 jobs. This leaves out less than 2 percent of workers, and the share of workers who are in underemployment is virtually unchanged. 9 This proportion was slightly higher for women (27.1) than for men (23.6). 10 About 17 percent of those who worked full-time had a second or more, while 37 percent of those who worked less than 35 hours a week in their main job had a second job or more in 2007 (LFS, 2007). 22 beyond their control (especially job loss or reduced workload, and bad weather). The three main reasons were: bad weather (48.4 percent); family responsibilities (22 percent); and job related - shortage of work, lack of work hours, or factory closure (12.6 percent). Wage differentials by type of contract, sector, and type of employer The above results show that significant proportions of workers are involuntarily working low hours, and compensating low hours by taking up additional employment. But they also show that a significant proportion of full-time workers complement their labor income with a secondary job. The evidence confirms that wage workers who are in underemployment are paid significantly less on average than others: in 2009 they were paid on average 62 percent of the average hourly wage of workers who are not in underemployment (LFS). Other wage differentials are apparent. Wage workers from vulnerable groups such as unskilled youth, women, and ethnic minorities are found to be paid less than others. In 2009, non-Kinh wage workers were paid 90 percent of the average hourly wage of Kinh wage workers (Table A 1). Gender wage differentials existed within these ethnic groups in particular the Kinh group: Kinh female wage workers were paid 89 percent of Kinh male wage workers, while non-Kinh female wage workers were paid 95 percent of non- Kinh male wage workers (a 5 percentage point insignificant wage differential). Youth wage workers were paid 75 percent of the wage of adults. On average, there is a premium for wage workers in large firms and in the services sector. Wage workers in larger firms (21 and more employees) are paid more than those in smaller firms, (mean hourly wage are 13.2 thousand VND in large firms versus 9.2 thousand VND in small firms in 2009). Wage workers in the services sector are paid significantly more (13.7 thousand VND per hour) than those in the agriculture (7.6 thousand VND per hour) and manufacturing sectors (10.1 thousand VND per hour). (See Table A 2 for disaggregated sectors.) 23 Low-wage earners represent a significant proportion of wage workers, especially affecting women 11 Low-wage earners represented 8.6 percent of wage earners in 2008; women were more affected than men by low-wage (11.8 percent versus 6.4 percent). Women are overrepresented in low-wage employment; this result is consistent with worldwide results (ILO, 2010b). Compared with 2006, there was an increase in the share of low wage earners (from 6.5 percent in 2006). This increase in low-wage earners occurred for both men and especially women (from 5.2 and 8.5 percent for men and women respectively). Low wage workers are found mostly in the agriculture sector. In the latter sector, 22.4 percent of wage workers were in low-wage employment, while only 7.1 percent of manufacturing wage workers, and 5.8 percent of service workers were in low-wage employment. Interestingly, the difference in proportion of low-wage workers in the manufacturing and service sectors is not statistically significant. Low-wage female workers are about as likely as male low-wage workers to be in the agricultural sector, but they are slightly more likely to be in the industrial sector and less likely to be in the services sector (Table 19). Table 19: Low-wage workers by sector and gender (2008) Women Men All Agriculture 35.9 32.1 34.2 Industry 41.1 35.3 38.5 Service 23.0 32.6 27.3 Total 100.0 100.0 100.0 Source: (VHLSS, 2008) When looking at disaggregated sectors, three sectors (electricity, gas; real estate, renting and business activities; and extraterritorial) have no low-wage workers, for the rest, there is too much variability to find statistically significant differences (Table A 3). The large majority (about 91.5 percent in 2008 and 94 percent in 2006) of low-wage earners are underemployed. Low hours seem therefore to be the main reason for earning wages below the poverty line, although it is less the case in 2008 than in 2006. In this 11 Low wage earners are defined as earning less than the per capita income poverty line in their main job. We use the national income poverty line: (2006: 200 thousand VND for rural areas and 260 thousand VND for urban areas; 2008: 290 thousand VND for rural areas and 370 thousand VND for urban areas ) (see GSO, 2009). 24 respect, the relative situation of women worsened slightly as female low-earners were slightly more likely than their male counterparts to be underemployed in 2008 (93 percent of female low-wage earners are underemployed, versus 89.6 percent of male low- wage earners), while in 2006 these shares were similar for men and women (95 percent for women and 93 percent for men). The share of low-wage earners who work more than 48 hours per week in their main job decreased slightly from 2.1 to 1.6 percent between 2006 and 2008. The latter is such a small percentage that such changes are difficult to interpret. In any case, underpaid long hours do not appear to be an issue in Vietnam. However, it is interesting to note that the situation seems to have worsened for men as in 2008 it affected men more than women (2.9 percent of male low-wage earners work over 48 hours per week versus 0.5 percent of female low-wage earners), while in 2006 the situations of men and women were similar (around 2 percent). Not surprisingly, low-wage earners are more likely to belong to a poor household: about 24 percent of them live in a poor household (slightly up from 22.9 in 2006), against about 7.5 percent of non low-wage earners in 2008 (8.7 percent in 2006). Although most low-wage earners were underemployed, only about 47.5 percent of low wage earners had more than one job in 2008 (up from 42 percent in 2006), compared to about 32 percent for the non-low-wage earners in 2008 (down from 34.2 percent in 2006). Comparing men and women, the biggest difference between low-wage earners and the others is for women: in 2008 only about 25 percent of female non low-wage earners worked more than one job, while about 47 percent of the low-wage earners did.12 There is also a strong link between low-wage and education. In 2008, about 51.5 percent of low-wage earners have primary education or no degree, while nearly a third of the non low-wage earners have such levels of education. It is noticeable that a small proportion of college and university educated wage workers are low-wage earners. At most levels of education, women are more likely to be low-wage earners. But women with college and university education are less likely than their male counterparts to be low-wage earners. 12 The corresponding percentages for men are 36 percent and 48 percent respectively. 25 Self-employment earnings are lower than wages, while more risky entrepreneurship potentially brings more reward. It is more complex to measure earnings from self- employment as such information relies mostly on what people report (unless a full analysis of their business is done). The LFS however attempts to measure such earnings. Own-account workers earn on average about 80 percent of what wage workers earn (Table 20). An even greater share of these workers is therefore likely to be low-wage earners. Employers on the other hand earn on average much more than wage workers, but their earnings are much more volatile. This is consistent with this sort of employment being more risky than wage employment, but potentially more rewarding. Table 20: Hourly earnings by employment status (2009) Mean Std. Dev. Median Frequency Thousand VND Employer 22.2 57.6 13.9 2273176 Self-account worker 9.1 35.8 6.1 21156565 Wage worker 11.2 10.1 8.8 15956925 Member of cooperative 6.7 4.8 6.9 51772.2 Others 5.9 7.7 5.2 35969.1 Total 10.7 30.4 7.4 39474407 Source: (LFS, 2009) Average hourly wages are not significantly different for underemployed in self- employment compared with other self-employed (Table A 1). But ethnic wage differentials observed for wage workers hold for the self-employed: non-Kinh self- employed earn on average 64 percent of the earnings of Kinh self-employed workers. The gender earnings differential is slightly more marked for the Kinh ethnic group: women are paid 76.8 percent of men’ hourly earnings, while non-Kinh women earn 79.4 percent of what non-Kinh men earn on average (LFS, 2009). Young self-employed are paid 78 percent of the earnings of adults. Employers’ earnings are very variable and it is difficult to find any significant differentials by sector or other firms characteristics. However, on average they are better paid than wage workers and self-account workers in all sectors. Self-account workers do well in all sectors compared to wage workers, and they are paid significantly less only in the manufacturing sector. 26 The working poor Looking at the poverty status of households and the labor force status of their members confirms the idea that the most disadvantaged individuals are not especially found among the unemployed and the inactive. While 14 percent of workers lived in households that were below the poverty line in 2008, only 10 percent of unemployed individuals and 9 percent of inactive did so (VHLSS). This is consistent with the idea that those who are unemployed or inactive can afford to do so because of their household situation. There are two underlying issues here: (i) the extent to which work pays. In other words, what is the type of employment that workers are able to secure, does it differ across gender and compared to other lower-middle-income countries? (ii) The extent to which the labor supply of household members differs between poor and non-poor households. 4. Mobility and migration The results suggest that movers and migrants tend to live in urban areas, to be employed as wage workers, and to have moved for job related reasons. Women living in urban areas are more likely than urban men to have moved, while rural men and women are similarly likely to have moved. Men are more likely to have moved/migrated for job- related reasons than women; women are more likely to have moved/migrated in order to get married. Only a static view of mobility and migration is available. Among respondents, few report having moved or migrated in recent years A small minority of individuals (6.6 percent) report having moved in the 5 years before 2009 (LFS, 2009).13 Among those who have moved, 70 percent are working and 57.1 percent are women. Individuals living in urban areas are much more likely to have moved than those living in rural areas: while 12.3 of urban individuals have moved or migrated within the previous 5 years, only 4.1 of individuals in rural areas have moved or migrated in the previous five years. This is consistent with a move towards urban areas. Young people (up to the age of 30 years old) are more likely to have moved than older individuals. 13 Comparison with Census data for 2009 suggests that this proportion underestimates the share of movers in the population. 27 Among working individuals, wage workers are the most likely to have moved or migrated within the previous 5 years The Labor Force Survey shows that among working individuals who moved within the previous 5 years, 64 percent are wage workers. Among the latter, over half of them report having moved for job related reasons (either to find work or to start a new job). Moreover, nearly half moved within Vietnam, while the other half reports having moved locally (i.e. within their province), a small minority having moved internationally. International in-migration is a marginal phenomenon The findings suggest that international in-migration is a marginal phenomenon. The Labor Force Survey gives a pretty good picture of within-country migration and moves (see footnote 13 for caveat), but it is not adapted to look at international migration since those who have moved abroad are not interviewed. In fact most moves reported in the LFS are at the local or national level; very few individuals (0.5 percent of workers who have moved in the previous five years) report having migrated from another country. Among the employed, own-account workers are the most likely to have moved from another country, they represent 51.7 percent of all international in-migration. This suggests that international migrants may have saved enough money while working abroad to set-up their own business upon return to Vietnam. But it is noticeable that no employers report having moved within the 5 years preceding their interview. This may reflect a slow growth of firms that are set-up by international migrants, but it also suggests that few international migrants come back with large investment capacity. This is consistent with previous evidence that suggests that remittances are mainly used to purchase land and housing, and are received by relatively better-off households (Nguyen and Mont, 2011).14 Gender differentials are apparent Family tops the list of reasons for women’s moves, while labor market reasons dominate men’ moves. It is not surprising to find that men are more likely than women to have 14 Of course these interpretations have to be taken cautiously given the important caveat of a very small number of observations. Specific migration studies that contain labor market modules are best to understand these issues. 28 moved for work related reasons (46.8 percent of all moves, versus 34.3 percent for women). It is also not surprising to find that a large share of women’s moves (36.4 percent) occurred upon getting married, while it is the case for only 18.2 percent of men. Interestingly, moves due to schooling represent about the same share of men’s and women’s moves. These gender patterns vary by age group. Own-work related reasons are most common for men between the age of 25 and 54 years old (about half of men movers of this age group report having moved to find or start a job) and for women between 35 and 54 years old (again about half of women movers of this age group report having moved to find or start a job). Own-work related reasons are the main reason to move for young men age 15-24 just before schooling (38.5 percent of men movers report having moved to find or start a job, 36.7 percent of them report having moved for schooling reasons). For young women, the main reason for having moved is getting married, own- work and schooling reasons coming second and third respectively. Similar gender differentials are apparent when focusing on internal migration Internal migration is limited. The share of urban dwellers who have moved from another province (internal migrant) stands at 5.8 percent. Among those urban internal migrants, 46 percent have moved for job-related reasons. Urban women are slightly more likely than urban men to report having migrated from another province. For both urban men and women, the main reason for internal migration is job-related (looking for a job or taking up a new job). The second main reason is schooling, but a close third for women is getting married.15 The share of rural dwellers who are internal migrants is significantly lower at 2 percent of rural dwellers. Their main reason for having migrated is still however job-related (especially for men). The second most important reason is getting married. This is true for both men and especially women. 15 Urban men’s reasons are: 50 percent job related; 34 percent schooling; and 10 percent getting married. Urban women’s reasons are: 43 percent job related; 27 percent schooling; and 25.5 percent getting married. 29 5. Gender differentials: Evidence from the VHLSS As mentioned in the introduction, cross-country research by the World Economic Forum (2010) suggests that, relative to its rank in terms of overall gender equality, Vietnam is doing well in terms of gender equality in the labor market. Vietnam ranks 12th out of 134 countries in terms of women’s labor force participation, and 39th in terms of pay equality for similar work (versus a rank of 72 for overall gender equality). The descriptive analysis above gives us a useful picture of the performance of the Vietnamese labor market, and confirms that Vietnamese women have good access to the labor market. The extent of the gender differentials are better analyzed using econometric modeling (see Annex 2 for methodology). 6.1 Women’s participation The labor force participation decision of women tends to be of a different nature compared with that of men. Although single women tend to have similar labor force participation to men, married women, and in particular women with children, have to make a choice between staying at home and participation in work outside the home. This is because they retain the main role in doing housework and caring for children/sick or elderly relatives. While it is generally assumed that men participate in the labor market unless they are incapacitated, the labor force choice of women is very much related to intra-household labor supply decisions. Since Vietnamese women have a high labor force participation, which is similar to men’s, it is interesting to see whether there are gender differences in the determinants of participation, and how these compare with the usual patterns found in countries with very dissimilar participation rates for men and women. Moreover, gender differences are likely to exist in other dimensions, such as the type of employment. The findings suggest that statistically significant gender differentials exist in participation with respect to age, educational attainment, regions, children ratio and marital status.16 Participation in various economic activities and inactivity varies over the lifecycle and in a different manner for men and women (controlling for other characteristics). Women are 16 The multinomial logit model results are not shown. 30 more likely to be self-employed or inactive over the life-cycle; they are more likely to remain inactive and to leave economic activity at an earlier age (Figure A. 1 in Annex 1). Keeping all other characteristics at their mean value, gender differentials are apparent. The findings suggest that in 2008 married women are 12 percentage points less likely than married men to be wage workers; this differential drops to 7 percentage points between single women and single men (Figure A. 2 in Annex 1). Single women are on average 4 percentage points more likely than married women to be wage workers and 35 percentage point more likely to be inactive. Married women are 39 percentage points more likely than single women to be self-employed and 8 percentage points more likely to be self-employed than married men. Women with low education fare worse than other women and than their male counterparts. Women with no degree are 17 percentage points more likely (twice as likely) as men with no degree to be inactive (Figure A.3 in Annex 1). Women seem to make better use of the college and university education: they are more likely than men with the same education to be wage workers and less likely to be inactive or self- employed. For example, college educated women are 12 percentage points more likely to be wage workers than college educated men. Having children in the household affects women’s labor force participation as well as their type of employment. Women who live in households with a ratio of children aged under 6 years hold to total number of household members that is above 0.1 are about 11 percentage points more likely than their male counterparts to be inactive. Women are on average less likely to be self-employed, and having children in the household worsens that difference. For example, women who live in a household that has a ratio of children aged under 6 years old to total number of household members that is above 0 and less than 0.1 are 19 percentage point less likely than similar men to be wage workers. The decomposition results show a slight gender differential in the probability of working in 2006 and 2008 (Table 21). The estimated predicted probability of working is slightly reduced for women between 2006 and 2008, while the men’s predicted probability of working remains the same, thereby increasing the gender differential in the probability of working. Despite its small size, the gender differential in the probability of working 31 remains mostly unexplained. Circumstances and “endowmentsâ€? explain only about 34 percent of the gender differential in the probability of working. Among these, age, children ratio, working adult ratio, education, and marital status are the most powerful in explaining the difference. Overall endowment in age and difference in working adult ratio increase the gap, while overall differences in circumstances with respect to child ratio and marital status, and endowments in educational attainment reduce the gap. Within the unexplained part, the returns (in terms of the probability of working) to age and children ratio reduce the participation gap, while returns to circumstances linked to region of residence and marital status increase this gap. In particular, being married explains a significant part of both the explained and unexplained gap, while the fact that many women are single in the pool of workers contributes to decreasing the gap. This shows that, despite very similar participation rates between men and women, the distinction between married women and single women is still valid in Vietnam; everything else held constant, married women are less likely than their male counterparts to be working. This result holds for 2006 and especially for 2008. Women are less likely than men to be working in wage employment (Table 21). The gap in the probability of being in wage employment is similar in 2006 and 2008. It is largely unexplained (around 88 percent of the gap is unexplained). Age and endowments in education are the most important factors in the unexplained part of the gap, bringing lower returns for women than for men. Table 21: Gender gaps in labor market status (percentage points) 2006 2008 2006 2008 Probability of working (Women) 72.8% 71.5% Probability of being in wage employment 19.4% 19.8% (Women) Probability of working (Men) 78.0% 78.1% Probability of being in wage employment 31.3% 32.2% (Men) Gap in probability of working -5.2*** -6.6*** Gap in probability of being in wage -11.9*** -12.4*** employment Of which: Of which: Endowments -1.4*** -2.3*** Endowments -1.4*** -1.4*** Of which: Of which: Age -1.7*** -2.0*** Age -1.0*** -1.2*** Ratio of children to household 0.3*** 0.2*** Ratio of children to household members 0.1*** 0.1*** members Ratio of working adults to -2.0*** -2.1*** Ratio of working adults to household -0.3*** -0.3*** household members members Household non-labor income 0.0 0.0 Household non-labor income -0.003 -0.004 Educational attainment 0.7** 0.8*** Educational attainment -0.2** -0.2** 32 Marital status 1.3** 0.9* Marital status 0.1 0.1 Region 0.004 0.009 Region 0.03 0.1 Ethnicity -0.006 -0.023 Ethnicity 0.004 0.02 Coefficients -3.4*** -4.4*** Coefficients -10.2*** -10.7*** Of which: Of which: Age 0.6 5.8** Age -23.6*** -13.2** Ratio of children to household -0.5*** -0.5*** Ratio of children to household members -0.6** -0.4 members Ratio of working adults to -0.8 -1.1 Ratio of working adults to household 2.4 2.8** household members members Household non-labor income -0.1 -0.05 Household non-labor income 0.1 0.1 Educational attainment -0.4 0.1 Educational attainment -3.3*** -4.0*** Marital status -1.3* -3.6*** Marital status -2.2 -0.9 Region -0.4** -0.6*** Region -0.5 -0.2 Ethnicity 0.4 0.4 Ethnicity -0.9 0.2 Constant -1.0 -4.9 Constant 18.4** 5.0 Interaction -0.5* 0.1 Interaction -0.3 -0.3 Note: the gap is the difference between the men’s probabilities and the women’s probabilities, taking the men distribution as non-discriminatory. The probabilities are calculated within the total working age population (aged 15+). *** means statistically significant at the 1% level, ** at 5% and, * at 10%. Data source: VHLSS 2006/2008 6.2 Gender wage differentials Gender wage differentials in Vietnam are largely unexplained and linked to the different returns to endowments, and to the differentiated treatments given specific circumstances that exist between men and women in Vietnam. Raw wage differentials vary by individuals’ characteristics and circumstances. On average, women were paid 75 percent of men’s wage in 2009 (LFS). This average hides differences across occupations (Table 22): women in higher level occupations had smaller wage differentials with men, while women in manual or unskilled occupations had greater wage differentials. The raw gender wage differential is explained by a number of factors that are best analyzed through multivariate regression. This average also hides differences in the type of employment; the gender wage differential is much smaller among wage workers than among self-employed workers. Using LFS data, we find that whilst the raw gender gap in monthly earnings of wage workers was 16 percent in 2009, up from 12 percent in 2007, it was 22 percent for self-employed workers in 2009 (down from 25 percent in 2006). This is confirmed in the VHLSS data, which show that at the mean hourly wage, female wage earners earn 90 percent of their male counterparts’ wages and at the median hourly wage they earn 83 percent of men’s wage (Figure 1). 33 Table 22: Raw gender gaps % (2007 and 2009) Occupation Gender gap in weekly hours Gender gap in monthly pay 2007 2009 2007 2009 Leaders in all fields 101.0 96.9 120.6 84.3 Top-level professionals 96.7 96.8 83.7 82.6 Mid-level professionals 98.8 96.6 94.4 94.2 Officers 102.4 108.4 103.8 116.1 Skilled workers in personal services, security protection and sales 98.5 98.1 77.6 72.5 Skilled workers in agriculture, sylviculture, and aquaculture 90.0 90.2 67.7 64.7 Skilled handicraftsmen and other relating skilled manual workers 99.5 92.5 69.4 60.4 Assemblers and machine operators 102.7 101.7 65.3 65.0 Unskilled workers 96.7 93.8 80.0 75.2 Total 95.8 94.3 73.7 75.1 Source: (LFS, 2007 and 2009) Note: ratio of women’s number of hours by men’s; ratio of women’s monthly wage by men’s. Figure 1: Kernel density of wages in main occupation by gender (2008) Natural log of hourly wage in main occupation by sex (2008) .8 .6 Density .4 .2 0 -2 0 2 4 6 Natural log of hourly wage Male Female kernel = epanechnikov, bandwidth = 0.0890 Source: VHLSS 2008 The gender wage gap is estimated to be similar in 2006 and in 2008 (Table 23). Compared with the results for previous years (Liu, 2004), the gender wage differential is lower than in the nineties (it is now about 65 percent of its value in 1992-93), but it has stopped its decrease in the last two years of VHLSS data (2006 and 2008). While in the estimations for 2006, correcting for selection into wage employment decreased 34 significantly the wage gap, it is not the case in the 2008 estimations. The decomposition of the gender wage differential (Table 23) shows that endowments of women actually contribute to reducing the gender wage gap: given their endowments, women should actually be paid more than men; if men had women’s endowments they would receive higher wages. The gender wage gap remains therefore “unexplainedâ€?. Given their endowments, if men were treated as women are, they would receive lower wages. In particular, there seems to be important differences in the returns to age in 2008, and to the sector of activity and to belonging to private/public sector. In particular, while the returns to being in the public sector are higher for women than men, it is the opposite in the private sector. Moreover, the returns to being in industry are lower for women than for men. The ethnicity of men and women does not significantly account for the gender wage differentials. Table 23: Oaxaca decomposition of gender wage differential 2006 2008 Natural log of real OLS Heckman OLS Heckman hourly wage Women 1.466 1.340 1.749 1.547 Men 1.642 1.434 1.926 1.721 Wage differential -0.175 -0.094 -0.177 -0.174 Of which: Endowments 0.019** 0.032*** 0.041*** 0.057*** Of which: Age -0.002 -0.002 0.005* 0.009** Job tenure -0.006 -0.005 -0.003 -0.003 Job tenure squared 0.000 0.000 0.0003 0.000 Educational 0.028*** 0.036*** 0.037*** 0.045*** attainment Marital status 0.000 0.002 -0.002 0.000 Region 0.007*** 0.010*** 0.008*** 0.010*** Ethnicity 0.001 0.001 0.0003 0.001 Sector -0.003 -0.003 -0.003 -0.003 Private/Public -0.006*** -0.006*** -0.002 -0.002 Coefficients (returns) -0.213*** -0.144 -0.232*** -0.252 Of which: Age -0.080 -0.285 -0.542*** -0.663** Job tenure 0.006 0.006 0.012 0.007 Job tenure squared 0.014 0.013 0.016 0.018 Educational -0.004 0.006 0.002 -0.008 attainment Marital status 0.006 0.008 0.016 0.016 Region -0.001 -0.010 -0.029* -0.032 35 Ethnicity -0.023 -0.038 0.011 0.008 Sector -0.038*** -0.037*** -0.026*** -0.025*** Private/Public -0.026*** -0.027*** -0.032*** -0.032*** Constant -0.067 0.219 0.340* 0.460 Interaction 0.018** 0.019 0.014* 0.021 6. Coping with economic fluctuations In the absence of state help, households have developed strategies over the years to cope with periods of financial difficulties, sickness, and retirement. Although some of these strategies may favor intergenerational help and community ties, many, such as sending children to work instead of school or selling off productive assets, are suboptimal and perpetuate the cycle of poverty across generations. Although the role of governments in mitigating risks for households and individuals has been recognized, the design of comprehensive social protection systems remains in its infancy in many countries. Asian countries have not had a tradition of state intervention in social protection, and have only recently started to get interested in these issues, in particular in the wakes of the financial crises.17 Vietnam recently introduced a series of reforms in social protection. However, substantial difficulties exist when setting up such programs. The following issues, which are common in countries with similar level of development as Vietnam, have been raised (MoLISA, 2010): (i) coverage is low, both because of non-compliance with compulsory schemes, and because of low voluntary take-up; (ii) the system is biased against the poor and in favor of the rich; (iii) the level of benefits tends to be low and to represent a small share of household income; (iv) monitoring and management of the scheme implementation need to be improved. In addition, the new system has to be improved keeping in mind the need to maintain its long-run financial sustainability and to avoid inducing the wrong incentives to beneficiaries (the aim should be for beneficiaries to “graduateâ€? from social protection programs and to stay in them for finite periods of time). 17 There are three main domains of intervention: (i) Insurance (pension, health, unemployment); (ii) Active Labor Market Policies (ALMPs include employment service, workfare programs such as training, public employment programs, subsidized employment); and (iii) Social Assistance. 36 The rest of this section looks at the potential gender differentials in private coping mechanism and in access to existing programs. 7.1 Private coping mechanism This section focuses on the impact on women of households’ private coping mechanisms. Its aim is not to provide a comprehensive review of all strategies by households, but rather to see whether women are at a disadvantage in the intra-household labor supply decisions that take place when households have to face challenges. In particular, in the context of the recent financial and economic crisis, this section analyses how individuals and households have coped in the recent past (2006-2008) and whether it produced disadvantages for women. A majority of households did not change their poverty status between 2006 and 2008 The large majority of the households in the merged sample18 were non-poor in both 2006 and 2008, while about 8 percent were poor in both years. The rest changed their status between the two years: 4 percent became poor while 6 percent became non-poor.19 Female-headed households were more likely to be non-poor (and less likely to be poor) than male-headed households. Both female- and male-headed households were equally likely to become poor or non-poor.20 A significant proportion of households have changed the type of employment21 that their members do In order to deal with poverty, households have a number of strategies: they can increase the number of household members who are in the labor force; they can change the type of employment that current working household members hold; they can change the sector of economic activity, shifting to the higher-paying non-agricultural sector. 18 For this sub-section we restrict our analysis to the sample of households that were interviewed in both years (see dataset description in section 2). 19 In the following paragraphs, we refer to poor households (poor in both 2006 and 2008), non-poor households (in both years), households that became poor (between 2006 and 2008) and household that became non-poor (between 2006 and 2008). 20 85 percent (6 percent) of female-headed households were non-poor (poor) against 81 percent (8 percent) of male-headed households. 21 By type of employment we mean: wage work only, wage work and self-employment, self-employment only, no economic activities. 37 The results show that a significant proportion of households changed the structure of employment within the household. However, there is only weak evidence that these shifts were associated with better poverty outcomes. About a quarter of households modified the types of employment that their members did (Table 24). The households that became non-poor are the most likely to have changed the types of employment of household members. They are statistically significantly more likely than non-poor households to have done so. While one-third of the former households changed employment type, only a quarter of non-poor households changed employment type. Table 24: Type of employment by household poverty (2006-2008) Change in type of employment Non-poor in Poor in Non-poor in Poor in 2008 total both 2006 & both 2006 2008 and poor and non-poor 2008 & 2008 in 2006 in 2006 1. No change 74.3 72.0 64.8 69.4 73.4 2. FROM Wage+self-employment TO 8.2 10.6 12.1 13.0 8.8 self-employment only 3. FROM wage work only TO self- 0.2 0.7 1.1 0.0 0.3 employment only 4. FROM wage work only TO 1.7 0.9 0.8 0.7 1.5 Wage+self-employment 5. FROM self-employment only TO 0.6 0.3 0.0 0.0 0.6 wage work only 6. FROM self-employment only TO 8.4 12.1 11.7 11.1 9.0 Wage+self-employment 7. FROM Wage+self-employment TO 3.9 1.4 6.2 3.9 3.8 wage work only 8. FROM no economic activity TO work 0.9 0.0 0.3 0.0 0.8 9. FROM work TO no economic activity 1.7 1.9 2.8 2.0 1.8 Source: VHLSS 2006 and 2008. The most common changes that households do are to shift between having members working in wage work and in self-employment, to having workers working in self- employment only and vice versa. This is true for poor and non-poor households. Non- poor households are the least likely to go from a situation where members work both in wage work and self-employment to a situation where they only work in self-employment. Households that became poor are the most likely to have had this strategy, while households that were poor in both years are most likely to have had the opposite strategy (although these differences are not statistically significant). 38 Starting an economic activity and refocusing labor market participation can be strategies to exit poverty Interestingly, comparing the structure of employment within households that became non-poor and those that remained poor, we see that some of those that remained poor shifted from having only self-employed members to having only wage workers, while that did not happen to households that became non-poor. Inversely, some of the households that became non-poor went from no economic activity to some economic activity, while that did not happen in households that remained poor. Finally, households that became non-poor were significantly more likely than those that remained poor to have shifted from having members working in self-employment and in wage employment to having members in wage employment only.22 Households that exited poverty are more likely to have changed sector than others There is some evidence that getting into or out of non-agricultural work is associated with a change of poverty status: households that became non-poor are more likely to have started or stopped an activity in the non-agricultural sector than households that remained poor, or were non-poor in both 2006 and 2008. Households that became non-poor are also found more likely to have increased the share of adults that work, but this result is not statistically significant. Female-headed households are less likely to be poor Given the small number of observations that such analysis would entail, it is not possible to compare how female-headed and male-headed households changed over the period. We give, in this sub-section, cross-sectional views in 2006 and in 2008. Female-headed households are less likely than male-headed households to be poor. The share of household that are headed by female is about 22.5 percent (unchanged between 2006 and 2008). Female-headed households are under-represented among poor households (17 percent of poor households are female-headed); despite improving 22 There is some evidence that poor female-headed households behave differently than male-headed households, but it is not possible to make assertive conclusions regarding these differences as the number of observations become too small when looking at these sub-groups. 39 slightly their situation between 2006 and 2008, male-headed households were more likely to be poor than female-headed household both in 2006 and 2008. Female-headed households tend to have a lower ratio of adults working to the total number of members than similar male-headed households. They are more likely to have no members in any economic activity, and to have wage workers only. They are less likely than male-headed households to have members in both wage employment and self- employment (especially among households that are non-poor in 2008). 7.2 Participation in active and passive labor market programs As mentioned above, official statistics suggest that only about 12 percent of the labor force is insured against the risk of unemployment (MoLISA, 2010). It is still early days to assess the impact of the introduction of such programs in Vietnam, and in any case, the LFS and VHLSS datasets provide little information that can be used to understand such impact. The income module from the VHLSS 2008 reveals than less than 1 percent of households report having received one-time job loss allowance.23 The above analysis suggests that those who are unemployed are individuals who can afford to stay unemployed. It is interesting to find out the structure of unemployment has changed over the period 2007-2009 - a period with the introduction of unemployment insurance and the advent of several economic crises – and whether women and men are affected differently and behaving differently in their job search. Few use employment services Employment services (publicly or privately provided) can address potential inefficiencies in the process of matching workers with jobs. Such inefficiencies arise because information does not flow instantly, and some workers and firms are much better equipped to acquire information than others. Few unemployed individuals use employment services in Vietnam. The two main ways in which unemployed individuals look for jobs are (i) through their family and friends (about 49 percent of the unemployed in 2009) and (ii) by applying directly with 23 We would need to know the percentage of unemployed who received unemployment benefits, but it is not possible to calculate this. 40 companies (32 percent of the unemployed in 2009). Only about 5 percent of the unemployed use employment services. These numbers are not surprising; even in countries with higher incomes, the use of employment services tend to be low. For example, in Mexico, two-thirds of individuals report finding their job through networking and 20 percent by contacting their employer directly in 2001. In Brazil, only about 28 percent of the unemployed were registered with employment services in 2001. When looking at men and women separately, similar patterns are found. However, women have become even less likely to use employment services between 2006 and 2008; among other things, this may suggest that the types of jobs advertised through employment services have become even less attractive to women. For example, jobs advertised through the employment services may belong to male dominated sectors or sectors where employing women may be prohibited by the labor code. Registering with employment services is often a requirement to receive unemployment benefits, as the authorities want to make sure the beneficiaries are actually looking for a job. A low use of employment services tends therefore to go hand-in-hand with a low coverage of unemployment insurance. In countries with low coverage of unemployment benefits, or with little eligibility criteria/constraints on receiving unemployment benefits, the use of employment services is therefore likely to be low. Some countries prefer not to impose strict criteria on unemployment benefits recipients, because they lack the capacity to monitor these criteria, and their general aim may be to alleviate poverty, rather than focus on the employability of workers. However, if a country wants to target specifically the issue of unemployment and help the unemployed find better jobs, then employment services can play an important role. Studies have shown that supplementing income support with some form of targeted employment services has enabled job seekers to obtain more secure and better-paid employment (Robins et al, 2008). 7.3 Participation in poverty alleviation programs About 45 percent of poor households participated in one or more poverty alleviation programs in 2008. There is no gender differential in terms of access to poverty alleviation 41 programs. There is no significant difference between the probability that a poor female- headed household participate in at least one poverty alleviation program and that of a poor male-headed household -- 51 (44) percent of poor female-headed (male-headed) households participated in at least one program in 2008. At the same time, a small proportion of non-poor households also participated in such programs, among those, female-headed households were significantly more likely to do so (10 percent for female- headed households versus 7 percent for male-headed households). Male- and female- headed households participate in the same number of programs on average.24 There is a strong hysteresis in program participation: nearly 100 percent of households that participated in a poverty alleviation program in 2008 were participating in 2007. This may be due to the fact that the households that are recipients are the poorest. On average recipients of poverty alleviation programs had per capita expenditure equal to less than half of the expenditure of non-recipients (in 2006 and 2008). 7. Concluding remarks The main findings that are relevant for policymakers can be summarized as follows: ï‚· Policymakers should pay attention to the quality of jobs created. Recent developments include an increase in the labor force participation and employment-to-population ratio, especially among young workers. The recent period has seen an increase in underemployment, in adult unemployment, and in informal employment (although employment in the informal sector has not increased). At the same time, the share of wage employment has increased, together with a decrease in the benefits associated with this type of employment. ï‚· Current policies have done wonders to promote youths’ outcomes, but recent changes need to be followed up in the next few years. Youth are better equipped than older generations to compete in the labor market. The increase in the labor supply of youth has been well absorbed into employment. In the recent context of 24 These programs are: Preferential credit for poor people; Free healthcare for poor people; Tuition exemption and reduction for poor people; Vocational training for poor people; Provision of productive land for poor ethnic households; Agricultural - forestry - fishery extension; Housing and land support for poor households; Clean water for poor people. 42 economic crisis and post-crisis, the situation of working youth has slightly improved with respect to that of adults. ï‚· The increase in the labor supply of youth may be worrying because it means that youth are leaving the schooling system earlier. This raises a number of questions about how the labor market will evolve in the coming years. Whether the labor force participation of youth continues to increase or not may influence the educational attainment of future workers. In addition, the recent evolution may have long-term effects on the educational achievements of this “crisisâ€? generation of youth, and on their labor market achievements. Overall current young workers do have greater educational attainments than older workers. ï‚· Current policies and customs are not sufficient to lead to greater equality of outcomes between men and women. Gender differentials have overall remained stable over the period. Although women are at a disadvantage compared to men on the labor market, some of their outcomes have improved. For example, women have increased their share of employment in wage employment, but the gender wage differential in wage employment has remained the same over the period. ï‚· Gender differentials (in terms of the probability of working, the probability of being in age employment and wages) are mostly unexplained. In other words, they are not explained by the differences in characteristics between men and women. Instead, they are mostly explained by differences in the returns to these characteristics (be it for discrimination or other unobserved reasons). ï‚· Strong regional disparities exist, these may be exacerbated as the country further develops and certain areas become further left behind in the development process. ï‚· Finally, specific data collection efforts are needed in order to understand several aspects of the labor market. The data that are available currently are insufficient to carry out analysis to understand better the labor market of some vulnerable groups such as migrant workers. The analysis of the informal sector and informal employment would benefit from specific quantitative and qualitative data collection effort. 43 Annex 1: Figures Figure A. 1: Labor force participation of men and women by age (2008) Men Women 1 1 .8 .8 .6 .6 .4 .4 .2 .2 0 0 20 40 60 80 100 20 40 60 80 100 age age Wage workers Self-employed Wage workers Self-employed Inactive Inactive Figure A. 2: Employment status by marital status 2008 (Predicted probabilities). 100% 90% 80% 70% 60% Self-employed 50% Wage workers Inactive 40% 30% 20% 10% 0% female male female male female male single married widowed Note: see note to Figure A.3 44 Figure A. 3: Employment status by educational attainment 2008 (Predicted probabilities). 100% 90% 80% 70% 60% Self-employed 50% Wage workers Inactive 40% 30% 20% 10% 0% female male female male female male female male female male female male No degree Primary school Lower secondary Upper secondary College University and school school higher Note: these are predicted probabilities obtained from the multinomial logit model. They are calculated holding all other characteristics at their average values. Table A 1a: Mean hourly wage and earnings – Thousand VND - 2009 Underemployed Not underemployed Kinh Non-Kinh All Men Women All Men Women All Men Women All Men Women Employer 17.5 - - 22.3 - - 22.2 23.5 19.6 22.2 - - Self- 8.5 9.7 7.1 9.1 10.2 8.1 9.7 11.0 8.4 6.2 6.8 5.4 account workers Wage 7.1 7.6 6.4 11.4 11.9 10.6 11.3 11.8 10.5 10.2 10.4 9.9 workers Source: LFS, 2009. Note: - indicates a low number of observations. Table A 1b: Mean hourly wage and earnings – Thousand VND - 2009 Youth (15-24) Adult All Men Women All Men Women Employer 13.8 - - 22.5 - - Self-account workers 7.2 8.7 5.9 9.2 10.3 8.2 Wage workers 8.9 9.6 8.0 11.9 12.3 11.3 45 Table A 2: Hourly earnings by sector and type of employment - Thousand VND - 2009 Employer Self-account worker Wage worker Mean Median Mean Median Mean Median 1. Agriculture, hunting, forestry and fishing 16.2 9.3 7.6 5.3 7.7 6.6 2. Mining and quarrying 29.0 25.7 7.1 5.9 14.1 12.4 3. Manufacturing 27.0 15.6 7.7 6.3 9.7 8.0 4. Electricity, gas 19.8 19.8 9.5 9.5 15.4 13.3 5. Water supply 38.2 41.2 7.8 6.1 13.3 11.0 6. Construction 19.3 14.7 11.4 9.3 10.0 8.7 7. Wholesale and retail trade, repair of motor vehicles 28.1 16.5 9.8 7.4 11.0 8.7 8. Transportation and storage 28.6 18.6 12.0 8.5 13.4 11.4 9. Accommodation and food service activities 18.5 14.1 9.8 7.4 8.4 6.4 10. Information and communication 27.7 13.0 12.5 10.3 18.9 16.5 11.Financial , banking and insurance activities - - 26.3 16.8 25.0 19.1 12. Real estate, renting and business activities 92.2 51.6 226.6 74.2 22.4 20.6 13. Professional, scientific and technical activities 38.0 23.6 13.6 10.8 17.8 15.5 14. Administrative and support service activities 30.5 26.5 56.7 9.9 13.0 9.5 15. Communist Party, socio-political organization - - 6.2 3.3 11.9 9.9 16. Education 52.7 27.2 18.4 14.1 16.4 14.8 17. Human health and social work activities 31.8 31.8 14.0 11.8 14.3 12.9 18. Arts, entertainment and recreation 22.2 17.7 8.0 6.2 11.3 9.7 19. Other services 14.5 11.9 9.3 7.1 8.8 7.4 20. Active households - - 11.8 5.9 6.4 4.9 21. Extraterritorial - - - - 24.5 16.1 Total 22.2 13.9 9.1 6.2 11.2 8.8 Source: LFS 2009. Table A 3: Proportion of low-wage earners by sector – Percent – 2008. Mean 95% confidence interval 1. Agriculture, hunting, forestry and fishing 22 20 25 2. Mining and quarrying 5 0 10 3. Manufacturing 10 8 11 4. Electricity, gas 0 . . 5. Water supply 5 -2 13 6. Construction 4 3 5 7. Wholesale and retail trade, repair of motor vehicles 7 5 10 8. Transportation and storage 2 1 4 9. Accommodation and food service activities 11 6 16 10. Information and communication 6 0 11 11.Financial , banking and insurance activities 3 -1 6 12. Real estate, renting and business activities 0 . . 13. Professional, scientific and technical activities 1 -1 3 14. Administrative and support service activities 5 -1 12 15. Communist Party, socio-political organization 8 6 11 16. Education 3 2 4 17. Human health and social work activities 4 1 7 18. Arts, entertainment and recreation 5 0 10 19. Other services 8 3 12 20. Active households 18 9 27 Source: VHLSS 2008 46 Annex 2: Econometric approach Participation into the various potential labor force statuses is modeled as a function of personal characteristics, household characteristics and regional dummies. We interact all these explanatory variables with a gender dummy. A set of three mutually exclusive outcomes is defined: wage employment, self-employment, and out of the labor force (+ unemployed).25 A multinomial logit model is used to estimate the quantitative effect of these determinants on these outcomes. Thanks to non-linear Oaxaca decomposition (Yun, 2004), we are able to decompose the gender differential in participation into work (versus staying inactive). We model participation through a probit model, using the same control variables as in the multinomial logit model. Gender wage differentials are found by comparing the mean wage for men and women. These means are calculated using three types of model: (i) OLS, which does not control for selection into wage employment; (ii) Heckman selection model; and (iii) an extension of the Durbin-McFadden model (as presented in Bourguignon et al., 2007). In the standard Heckman model, participation is modeled through a probit (Heckman, 1979). In the extension of the Durbin-McFadden model, each mutually exclusive outcome is taken into account and selectivity is modeled as a multinomial logit model (Bourguignon et al., 2007). In order to estimate such selection models, identifying variables, which are in the participation model but not in the wage equation, need to be selected. In other words, these variables need to capture factors that influence participation into wage employment but not wages themselves. The decision to participate in the labor force is likely to depend on the intra-household allocation of time. In order to take this into account we choose the following variables: the ratio of children under 6 years old over the total number of household members, the ratio of working adults over the total number of household members, and non-labor income of the household. 25 The unemployed are a very small category of individuals who are financially able to remain unemployed (i.e. to look for a job). They are not well defined in VHLSS (see data description) Because it is such a small category, it is not possible to leave them as a separate group. We add them to those who are out of the labor force as they seem the most similar category in the context of Vietnam. 47 Heckman (1979) estimates all of the parameters in the model: wage regression W  Xï?¢  ï?¥ 1 where W is the natural log of real hourly wage and X is the set of explanatory variables as described. and W is observed if Zï?¤  ï?¥ 2  0 where: ε1~ N(0, sigma); ε2 ~ N(0, 1); corr(ε1, ε2) = Ï? In the extension of Durbin-McFadden model (see Bourguignon et al. 2007), the wage regression is: W  Xï?¢  ï?¥ 1 where W is the natural log of real hourly wage and X is the set of explanatory variables as described and ε1~ N(0, sigma); W is observed if the individual chooses category one (wage employment) in a set of 4 alternatives. This happens when Y1*  max( Y j* ) , with Y j*  Zï?¡ j  v j for j=1 to 4 where Y* is the unobservable continuous variable representing the propensity to choose wage employment over other alternatives, and the residuals vj are assumed independent and identically Gumbel distributed. Decompositions of gender wage differential into “explainedâ€? and “unexplainedâ€? component are presented (Oaxaca). The share of the gender gap that is deemed “explainedâ€? relates to the fact that the different levels of endowments (different characteristics) of women can explain the wage differences. For example, women’s wages may be lower because they are on average less educated than men. In this case, the source of gender differentials does not lie in the labor market, but rather upstream with the education system. The unexplained part of the wage differential relates to the returns to these characteristics: for example the fact that women may have different returns to similar educational attainment can be due to discrimination, and the residual difference due to other unobservable factors such as the quality of education. 48 The Oaxaca decomposition consists in two steps: first estimating wage regressions for men and women separately (using OLS or Heckman models), then decomposing the mean outcome difference between the two groups. The two wage regressions for men (M) and women (F) can be written as: WM  X M ï?¢ M  ï?¥ M WF  X F ï?¢ F  ï?¥ F Where W is the wage and X is a vector of independent variables as described above. Using the vector of means of the regressors (xM and xF), the decomposition can be written as: D  ( xM  xF )' ï?¢ M  xF ' (ï?¢ M  ï?¢ F ) if the male wage distribution is considered to be non- discriminatory. D  ( xM  xF )' ï?¢ F  xM ' (ï?¢ M  ï?¢ F ) if the female wage distribution is considered to be non-discriminatory. The issue with such decomposition is therefore the choice of the non-discriminatory wage distribution. This choice can affect interpretation of the gender wage differential (index number problem). In the context of Vietnam, we take men’s wage distribution as being the non-discriminatory distribution, although this may overestimate the discrimination component of the wage differential.26 We present detailed decomposition results, i.e. we show the contribution of groups of explanatory variables. For categorical variables, the detailed decomposition results depend on the choice of the (omitted) base category. The solution that we use is to compute the decomposition based on "normalized" effects, i.e. effects that are expressed as deviation contrasts from the grand mean (Yun 2005, as applied by Jann 2008). 26 The idea is that discrimination may take two forms: men may be overpaid, while women may be underpaid. The reality is likely to be a bit of both. So when assuming that men receive the non- discriminatory wage, and comparing it to women’s wage, the differential that we find may be greater than if we were to compare with the “trueâ€? non-discriminatory wage. 49 Bibliography Bosch, Mariano, and Maloney, William. 2010. “Comparative analysis of labor market dynamics using Markov processes: An application to informalityâ€?. Labour Economics. Vol. 17, Issue 4, pp. 621-631. Bourguignon, François, Fournier, Martin and Gurgand, Marc. 2007. "Selection Bias Corrections Based On The Multinomial Logit Model: Monte Carlo Comparisons," Journal of Economic Surveys, Blackwell Publishing, vol. 21(1), pp. 174-205. Cling, Jean-Pierre, Nguyá»…n, Hữu Chí, Razafindrakoto, Mireille and Roubaud, François. 2010. “How deep was the impact of the economic crisis in Vietnam? A focus on the informal sector in Hanoi and Ho Chi Minh Cityâ€?. Policy brief. GSO-ISS/IRD-DIAL project. Ha Noi. di Gropello, Emanuela, and Sakellariou, Chris. 2010. “Industry and Skill Wage Premiums in East Asiaâ€?, World Bank Policy Research Working Paper no. 5379. Nguyá»…n, Hữu Chí, Nguyá»…n, Thị Thu Huyá»?n, Razafindrakoto, Mireille and Roubaud, François. 2010. “Vietnam labour market and informal economy in a time of crisis and recovery 2007-2009. Main findings of the Labour Force Surveys (LFS)â€?. GSO-ISS/IRD- DIAL project. Ha Noi. Nguyen, Viet Cuong and Mont, Daniel. 2011. “Economic Impacts of International Migration and Remittances on Household Welfare in Vietnamâ€?. Mimeo. Ha Noi. General Statistics Office. 2009. “Result of the survey on household living standards 2008â€?, Vietnam. Hussmanns, Ralf, 2004. “Measuring the informal economy: From employment in the informal sector to informal employmentâ€?. Working Paper no. 53, ILO, Geneva. International Labor Office. 2009. Vietnam Employment Trends 2009. Ha Noi. International Labor Office. 2010a. Vietnam Employment Trends (No. 2). Ha Noi. 50 International Labor Office. 2010b. Global Wage Report. Geneva. Jann, Ben (2008). The Blinder-Oaxaca decomposition for linear regression models. The Stata Journal 8(4), pp. 453-479. Liu, Amy. 2004. “Gender wage gap in Vietnam: 1993-1998â€?. Journal of Comparative Economics 32, pp. 586-596. Ministry of Labour, Invalids and Social Affairs. 2010. Social Protection Strategy of Vietnam, 2011-2020: New concept and approach. Pham, Thai-Hung, and Reilly, Barry. 2007. “The gender pay gap in Vietnam, 1993-2002: A quantile regression approachâ€?. Journal of Asian Economics 18, pp. 775-808. Phan, Diep. 2009. “A report on Vietnam’s Labor Marketâ€?. CIEM-DANIDA Project. Mimeo. Rodgers, Yana, and Menon, Nidhiya. 2010. “Gender differences in Socioeconomic Status and Health: Evidence from the 2008 Vietnam Household Living Standard Surveyâ€?. Mimeo. Yun, Myeong-Su. 2004. “Decomposing differences in the first momentâ€?. Economics Letters 82, pp. 275-280. Yun, Myeong-Su. 2005. “A Simple Solution to the Identification Problem in Detailed Wage Decompositionsâ€?. Economic Inquiry 43, pp. 766-772. 51