LABOR MARKETS IN SWAZILAND: THE CHALLENGE OF YOUTH UNEMPLOYMENT 1 Zuzana Brixiova 2, Robert Fakudze 3, Kumiko Imai 4, and Thierry Kangoye 5 Very preliminary draft - March 2012 Abstract Utilizing the 2007 and 2010 recent labor force surveys, this paper provides the characteristics of the Swaziland labor markets, with focus on youth, and analyzes policy options to address key challenges. First, features of the labor market and its youth segment are documented with descriptive statistics. Second, the paper identifies determinants of the labor market position of different types of workers aged 20 29 years through a multinomial logit model. Third, drawing on a search model, the paper analyzes impacts of active labor market policies at different segments of the population. While active labor market policies such as training or job search provided to people leaving the public sector may speed up transition to the private sector, they would exacerbate inequalities. Effective employment policies to integrate the vulnerable groups, especially the youth, into the labor market still need to be developed. 1 The authors thank Neil Rankin and participants of the workshop on Labor Markets Dynamics in Africa, organized by the Wits University (November 2011) for helpful comments. This research started when Kumiko Imai was with UNICEF Swaziland. The views expressed are those of the authors and do not necessarily reflects views of the institutions of their affiliation. 2 UNDP Swaziland 3 Swaziland Ministry of Labor and Social Security 4 UNICEF, East Jerusalem 5 United Nations Economic Commission for Africa and University of Auvergne (CERDI) 1
I. Introduction In contrast to Africa s strong economic performance before the global financial crisis (GFC), at about 2.5% a year in 2001-07, Swaziland s growth was subdued before the global financial crisis, in contrast to high growth in the sub-saharan Africa. The 2007 and 2010 labor force surveys for the Swaziland also revealed high unemployment rate, which at almost 30% of the labor force is one of the highest levels among Africa s middle income countries. Moreover, the aggregate numbers hide substantial differences across subgroups, as unemployment is disproportionally high among youth and the less educated. The country s socio-economic challenges are exacerbated by the highest prevalence of HIV and AIDS rates in the world (26.1% of population aged 15-49), which is also reflected in its very young population (40% of population is younger than 15 years). Even though Swaziland was not impacted by the global financial crisis directly due to its limited integration into the global economy, it was impacted through trade with South Africa and revenues. SACU revenues fell by 57% between 2008/09 and 2010/11, reducing space for pro-growth public outlays and payments to private sector contractors. The weak business environment and decline in FDI flows have also constrained growth. The growth thus declined to 2% or less in 2009-2011, adding to already significant labor market challenges that the country has been facing. The objective of this paper is to characterize Swaziland s labor market, with focus on its youth, and point out its main challenges and policy options. The paper first outlines the key features of the Swaziland labor market in general and its youth segment with descriptive statistics. Second, it identifies determinants of the labor market position of different types of workers (by age, gender, etc.) through a multinomial logit model. Third, drawing on a search model, the paper analyzes impacts of targeting active labor market policies at different segments of the population on labor market outcomes. The paper is organized as follows. After this Introduction, Section 2 outlines main characteristics of Swaziland s labor market. Multivariate analysis based on the multinomial logit model is in Section 3, while the trade-offs involved in applying active labor market policies to different segments of population are in Section 4. II. Labor Markets in Swaziland Key Stylized Facts Utilizing the first two integrated labor force surveys (2007 and 2010) and other sources, this section illustrates the differences in unemployment and other labor market outcomes across various groups of population, so as to highlight areas for policymakers attention and actions. Over the past several years, Swaziland labor market has exhibited the following features: 6 a. High unemployment, especially among youth and less educated In 2007 and 2010, unemployment rate reached 28.2% and 28.5% of the labor force, respectively (Government of Swaziland, 2007 and 2010), one of the highest rates in sub- 6 The labor market characteristics listed in this section are not exhaustive, but instead focused on those that underpin the empirical and policy analysis in Sections 3 and 4. 2
Saharan Africa s middle income countries. Moreover, during 2000s, the gap between female and male unemployment has widened. Unemployment, which has been rising steadily since the mid-1990s (Figure 1), has likely increased further due to arrears that the public sector accumulated with private contractors in the context of the liquidity squeeze of 2011. Figure 1. Unemployment rates, by gender, 1995 2010 (% of labor force) 35 30 25 20 15 10 5 Total Males Females 21,8 23,7 29,1 28,2 28,5 0 1995 1997 2001 2007 2010 Source: Swaziland Integrated Labor Force Surveys (SLFIS), 2007 and 2010. Unemployment in Swaziland is a youth phenomenon. At 53.3% of the relevant labor force, it impacted especially the youth (aged 15-24 years). These young people account for 43.3% of all the unemployed; people aged 25-34 years constitute another 34% of all unemployed. 7 Youth in Swaziland also seem to be the most affected by job mobility (Figure 2). Analysis of 2007 LFS data reveals that the age group 10-17 is clearly the most mobile and is dedicating more often some time in other work activities as they face the need to supplement their revenues. While their rates are not as high as those for youth, women are also more affected by unemployment than men. For example, while unemployment among men decreased from 29.7% in 2001 to 25.7% in 2010, it has increased from 29.7% in 2001 to 31.3% in 2010 for women. According to the SLFS 2007 and 2010, Swaziland, similarly to other developing countries has experienced feminization of unemployment, where women are more likely to be unemployed or discouraged from searching for work than men. Figure 2. Job mobility distribution by age groups, 2007 Source: Swaziland Integrated Labor Force Surveys (SLFIS), 2007. 7 The youth (ages 15-24) accounts for 40% and those aged 25-34 years for 22% of discouraged workers. 3
Unemployment has been also disproportionally concentrated among less educated segments of the population, namely people with primary or no education. Specifically, while unemployment rate for people with tertiary education was only 7.9%, it reached 34% for those with less than or primary education. Moreover, the 29% unemployment rate for high school graduates points to high return to tertiary education. The high rate of unemployment has been even more concentrated among people with less education for youth (32% of the young unemployed) than for the overall population (52% of all unemployed) (Figure 2). Figure 3. Distribution of unemployment by education, 2007 (% of total) 100% 80% 60% 40% 20% 0% 19,5 29,8 22,4 Source: 2007 Swaziland Labor Force Survey. In addition to 81,334 unemployed in 2007, another 49,023 people (more than 50% of the unemployed) above age 15 were discouraged workers. Put differently, if broader definition of unemployment were used with discouraged workers included, 2007 unemployment rate would have amounted to 38% of labor force. The causes of discouragement stated in the 2007 SLFS point to the need to expand access to higher levels of education and training programs. Concretely, about one third of discouraged workers stated not enough work experience, not enough skills or not enough education as the main reason for not securing employment. Another 13% of discouraged workers highlighted the lack of information about available jobs, underscoring the importance of developing active labor market policies, so far mostly missing in Swaziland. b. Long unemployment duration 3,3 3,3 25,0 21,7 Unemployed, all Another notable feature of the Swaziland labor market is the long duration of unemployment and/or underemployment periods, reflecting low utilization of available human resources. For example, about 80% of respondents in 2007 Labor Force Survey indicated that they have been available for work for over a year, and more than half of respondents have been available for more than two years (Table 1). Figure 2 also highlights the fact that this deteriorated a bit from 2007 to 2010, especially for longer availability for work. Indeed, 58.18% of respondents indicated in 2010 that they have been available for work for over two years, while 57.33% of the respondents provided the same answer in 2007. Unlike the level of unemployment, its duration is slightly lower for the youth (ages 15-24 years) than for the older cohorts. While more than 70% of people in this age category were unemployed or underemployed for more than a year, for ages 45 and above more than 90% were unemployed or underemployed for over one year. Nevertheless the still very long unemployment and underemployment duration for young people is of concern, as the 4 12,1 30,2 32,7 Unemployed youth Ter@ary High secondary Secondary Primary Less than primary
negative impact on their skills at early stage may impact their employment opportunities for the rest of their lives and also hamper Swaziland s human capital accumulation and growth. Table 1. Average duration of unemployment or underemployment, 2007 1/ 15 to 24 25 to 34 35 to 44 45 to 64 65 and above Total Less than 3 months 3.1% 1.8% 0.7% 0.3% 0.0% 5.9% 3 months - 1 year 7.3% 4.1% 1.2% 0.7% 0.2% 13.4% 1-2 years 13.9% 5.5% 2.5% 1.1% 0.1% 23.0% over 2 years 17.8% 17.9% 9.9% 11.0% 1.2% 57.7% Total 42.0% 29.3% 14.3% 13.1% 1.4% 100.0% Source: Authors calculations based on the 2007 Swaziland Labor Force Integrated Survey. 1/ The question underpinning this table was For how long has the person been available for work?. Figure 4. Average duration of unemployment of underemployment, 2007 versus 2010 2007 2010 Source: Authors calculations based on the 2007 and 2010 Swaziland labor force survey. Unemployment and underemployment duration rises with the educational attainment, but remains high at all educational levels. For example, in 2007 43% of the unemployed or underemployed with tertiary education were in that state for more than two years (Figure 3). c. Large share of employment in low value-added activities The Swaziland economy contains a large share of low value added and low paid activities in subsistence agriculture and low value added services. Moreover, relative to the overall population, the youth is even more engaged in low value-added and low paying activities such as community services or agriculture. In contrast, the youth is underrepresented in highpaying sectors such as the public service, the financial sector and business activities (Table 2). The public sector (consisting of the public service and public enterprises) accounts for a large share of employment, and the formal private sector remains underdeveloped. Development of the private sector that would drive structural transformation from low to high value added industries and services is thus needed to stimulate job-rich growth. 5
Figure 4. Duration of unemployment of underemployment, by education (2007) Source: Authors calculations based on the 2007 Swaziland Labor Force Survey. Table 2. Sectoral distribution of employment, 2007 (% of total) All Youth Agriculture, forestry, fishing 9 11 Mining, manufacturing, electricity 23 22 Construction 6 6 Trade, hotels, transport 24 23 Financial intermediation 6 3 Real estate, renting, and business activities 8 5 Public sector services 13 5 Other community services 12 25 Source: Authors calculations based on the 2007 Swaziland Labor Force Integrated Survey. d. Existence of a gender gap in the labor market opportunities Analyses of the Labor Force Survey data also reveal some gender imbalances in the labor market participation and opportunities in Swaziland, with women being in less advantageous position than men. For example, at 52.7%, work activity is indeed much higher for men against 39.4% for women (Figure 5). The Labor Force Survey also confirms the gender imbalance regarding the duration of work. Figure 6, which gives a picture of the work duration concentration regarding the gender clustering, interestingly show that men work longer than women (median work duration is 41 hours for women and 45 hours for men in 2007). Also when considering a given duration of work, more women have been found working less than this time as compared with men. As a 6
matter of fact, no man has been found working less than 18 hours, while about 15% of women are working less than 18 hours. Figure 5. Labor market participation by gender, taking into account the working age population (2007) Source: Authors calculations based on the 2007 Swaziland Labor Force Survey. Regarding age groups clusters consideration on top of gender clusters, interesting evidence is also found. Indeed, one can see from Figure 7 that the evidence provided by Figure 6 (that is women are more affected by underemployment than men) is mostly valid for people who are older than 18. Indeed, a close look at the Graph 7 reveals that for the age group 10-17, females work longer than males (median work duration is about 50 hours for females and 35 for males). This suggests that the labor market in Swaziland is not very protective of young females, who are more exposed to child labor than their male counterparts. This evidence could also reflect the fact that for this age group, boys have more chances to keep enrolled in school (and thus tend to work less longer) as compared with girls. LFS data analyses also unveil that women were most hit by the transition from 2007 to 2010, as their over-two years duration of unemployment of underemployment raised from 57.25% to 58.9% (Table 3). The table also shows that short durations of employment or underemployment (less than 6 months) deteriorated the most from 2007 to 2010. Figure 6. Work duration (Hours worked for main activity) concentration by gender (2007) Source: Authors calculations based on the 2007 Swaziland Labor Force Survey. 7
Figure 7. Work duration (Hours worked for main activity) concentration by gender and age groups (2007) Source: Authors calculations based on the 2007 Swaziland Labor Force Survey. Table 3: Duration of unemployment of underemployment, by sex (2007 versus 2010) Male Female 2007 2010 2007 2010 < 1 month 2.11% 2.16% 2.08% 3.23% 1 month - 3 months 4.32% 4.46% 3.54% 5.66% 3 months - 6 months 4.6% 5.61% 4.28% 5.98% 6 months - 1 year 9.3% 8.67% 8.87% 7.77% 1 year - 2 years 22.19% 21.81% 23.98% 18.45% > 2 years 57.46% 57.27% 57.25% 58.9% Total 100% 100% 100% 100% Source: Authors calculations based on the 2007 and 2010 Swaziland labor force survey. e. Labor market policies to integrate the youth As discussed above, employment prospects are especially bleak for the young people, with negative implications for the country s human capital and ultimately growth with equity. The persistently high unemployment stems from both demand and supply side factors. On the demand side, job creation is insufficient to absorb young job seekers; barriers to competition are particularly severe in the ICT sector. While unemployment rate for people with tertiary education is relatively low (less than 8% in 2010), skill shortages in selected areas have been emerging, (e.g., managerial skills, ICT). Labor exchange offices, that would facilitate flow of information between searching employers and job seekers, have so far been mostly missing. Recognizing the youth unemployment as a key challenge, the government has undertaken some steps to tackle it. For example, in 2009 the Youth Enterprise Fund was established, aiming to provide start up business capital for qualifying young people (ages 18 35) and companies. In the first phase during 2010, the fund distributed E 5.8 million (580,000 euros) to about 800 young entrepreneurs. In the ongoing phase II in 2011, the Fund has so far 8
distributed about E2 million (200,000 euros) to 200 entrepreneurs. 8 This initiative would need to be markedly scaled up though to achieve meaningful reduction in youth unemployment. Currently, with the public sector offering the best paid and most secure jobs, the public sector employs the majority of workers with tertiary education (Table 3). Many recent university graduates also 'que' for jobs in this sector, since experience in the public sector is viewed as entry into better paid private industries, such as banking. Hence a vibrant private sector that would provide high paying and productive employment opportunities (e.g., banking, ICT sub-sectors) is key to inclusive growth and lasting reduction in youth unemployment. Table 4. Sectoral distribution of employment by education (% of the education category) Public Formal private Informal private Domestic workers Total Primary or less 8.7 60.5 27.0 3.7 100.0 Secondary (incl. High) 21.4 63.0 14.4 1.2 100.0 Tertiary 58.2 37.7 4.1 0.0 100.0 Source: Authors calculations based on the 2007 Swaziland labor force survey. III. Multivariate Analysis This section presents the results of multinomial logit regression that assesses the determinants of labor market status among Swazi youth aged 20-29 years in 2007. Among five labor force status groups, unemployment was the reference status. The independent variables were: age in years, gender, regions, educational level and the interaction term between gender and educational level. The results are weighted to account for the complex survey sampling. The descriptive statistics for the variables included in the regression are in Table 1, Annex ( means ). The results show that the most frequently reported labor force status for the youth aged 20-29 years is unemployment (47%), followed by wage employment in the formal private sector (25%) and idle/inactive (17%). In contrast, the least frequently reported status categories were: wage employment in the informal private sector (1%), wage employment (public sector) (4%), and self-employed (6%). There are more females than males that fall in this age group (55% vs. 45%). Among four regions, youth are more likely to come from Manzini and Hhohho (38% and 31%, respectively), compared to Lubombo and Shiselweni (18% and 13%, respectively). Among six education levels, the most frequently reported levels are primary school (28%), high school (27%) and junior high school (21%). The regression results are shown in Table 5. Columns (1) to (6) indicate the labor market status, i.e. whether the interviewee is inactive, is employed in the formal private sector, the informal private sector, the public sector or is self-employed. Beyond age, gender, education and geographic location (i.e. whether the interviewee is located in one of the four districts of Swaziland -- Hhohho, Manzini, Shiselweni and Lubombo), the regression model also test interactive effects between sex and education. First, age has a positive relationship with the likelihood of being in the private sector, public sector employment or self-employed, rather than being unemployed. Second, compared with men, females are more likely to be 8 No collateral is required. Young entrepreneurs have up to 3 months to start their business upon receiving the funds; they have to repay loans within 24 months. Interest rate is about 10%, well below the commercial rates. 9
inactive/idle than unemployed. Third, living outside Lubombo makes it more likely to be inactive or in the private sector employment or self-employed than unemployed. Controls Table 5. Multinomial logit regression of the determinants of the labor market state, 2007 (1) (3) (4) (5) (6) Inactive Wage employment (formal private sector) Wage employment (informal private sector) Wage employment (public sector) Self-employed Age -0.04 (-1.60) 0.17*** (9.52) 0.01 (0.15) 0.27*** (5.19) 0.23*** (6.29) Sex (female) 0.66** (2.05) -0.64*** (-2.30) 0.43 (0.64) -1.1 (-0.99) 0.99 (1.53) Primary 0.16 (0.53) 0.29 (1.23) -0.24 (-0.35) 0.19 (0.35) 1.16* (1.92) Low secondary 0.67** (2.12) 0.43* (1.67) -0.33 (-0.29) -0.54 (-0.67) 1.71*** (2.78) High secondary 0.30 (0.98) 0.15 (0.68) -2.67*** (-2.38) 1.42*** (2.60) 0.99* (1.66) Tertiary 1.10*** (2.28) 0.18 (0.46) -48.23 (--) 2.62*** (4.28) 1.66*** (2.29) District (Hhohho) 0.49*** (2.67) 0.23 (1.44) -0.52 (-0.89) 0.51* (1.79) 1.20*** (3.72) District (Manzini) 0.58*** (3.35) 0.62*** (4.30) -0.85 (-1.60) -0.07 (-0.24) 1.11*** (3.50) District (Shiselweni) 0.92*** (4.97) 0.41*** (2.31) -0.38 (-0.72) 0.70*** (2.58) 0.8*** (2.23) Female primary -0.35 (-0.91) -0.01 (-0.02) -0.03 (-0.03) 0.1 (0.07) -1.34* (-1.81) Female low secondary -0.97*** (-2.28) 0.07 (0.20) -0.29 (-0.22) 0.92 (0.67) -1.39* (-1.86) Female high secondary -0.58 (-1.50) 0.18 (0.56) 0.78 (0.58) 0.48 (0.41) -1.54** (-2.02) Female tertiary -0.76 (-1.19) 1.24*** (2.37) -7.54 (--) 1.53 (1.28) -1.42 (-1.42) Intercept -1.02* (-1.64) -5.05*** (-10.76) -3.32* (-1.82) -10.08*** (-7.06) -9.65*** (-9.14) Obs 12004 12004 12004 12004 12004 Source: Authors computations using the 2007 Labor Force Survey data Note: * denotes significance at 10%; ** denotes significance at 5%; *** denotes significance at 1% Regarding the level of education, having completed lower secondary education raises the likelihood of being inactive or self-employed rather than unemployed. Having completed tertiary education raises the likelihood of being inactive/idle or in the public sector employment or self-employed. Generally having more education increases the likelihood of being in the public sector rather than unemployed; these patterns do not systematically differ between males and females. Among females having completed tertiary education increases the likelihood of being in the public sector employment rather than unemployed. IV. Impact of Labor Market Policies In this section, the likely medium term outcomes of introducing active labor market policies (ALMPs) targeted at youth and raising minimum wages are examined in a dynamic search model is utilized. Specifically, the model outlined below is a dynamic version of Van Ours (2007), which incorporates participation of the unemployed workers in job search and/or training programs in the framework of Mortensen and Pissaridies (1999). The focus is on the effect of labor market reforms on incentives for the unemployed to participate in training programs or search for jobs and for firms to create jobs. 10
In the model, workers can be either employed in the private sector or unemployed (that is working in the informal sector). Unemployed workers receive income from the informal sector b, and search for jobs or put effort in training with intensity while incurring cost, where. Employed workers receive wage w. Firms post vacancies to fill jobs at cost c. Each filled job results in output y, with. The key component of the model is a matching function, where denotes the efficiency of the matching and is the elasticity of matches with respect to vacancies. The workers search/training effort results in job offers, which arrive at rate, where denotes the ratio of vacancy rate, v, to effective unemployment rate, xu, i.e. it describes the tightness of the labour market from firms perspective. Conversely, firms fill their vacancies at rate. With normalizing the labour force to 1, that is, an dall job matches dissolving at rate, the employment rate,, and unemployment rate,, change according to: The steady state equilibrium unemployment thus becomes: (2) (1) (3) A scenario where all unemployed workers participate in the job search assistance or retraining programme is considered to illustrate the impact of participation in ALMPs on workers search/training effort. Participation in such programmes lowers workers income from the informal sector by a fraction z and reduces the rate of search or training cost by a fraction. Workers enter formal employment only when the value of formal employment,, exceeds the value of unemployment/informal employment, : (4) where is the discount rate, and denote change in the present discounted value of being unemployed and employed in the formal sector, respectively. The equilibrium search/training intensity x can be derived from (4) as: (5) (6) 11
Denoting as value of filled job and as value of vacancy, the Bellman equations are: (7) where is the output from the filled vacancy, is the social employers contribution tax paid, and and denote change in the value of filled job and vacancy over time, respectively. To complete the characterization of unemployment, a solution for the tightness of the labour market,, needs to be obtained through deriving wages. Regarding wage determination, the model assumes that private sector wages move with productivity changes:, where. From this assumption,. Moreover, with free entry into the job creation, value of posting a vacancy is. Hence (7) and (8) become: (8) (9) From (9), vacancy-to-(effective) unemployment, that is job-finding rate, rises with higher private sector profits (and lower wage and tax rate), improved efficiency of matching (A), and lower discount and destruction rates ( ), respectively. From (1), (2) and (6), these factors impact positively the equilibrium level on intensity of job search or participating in the training program and hence on job creation during transition as well as on the steady state level of formal private sector employment (Table 6 ). The impact of rising minimum wages is ambiguous since the positive impact on workers incentives to search for jobs or undertake training are hampered by disincentives of the private sector to create jobs. Table 6. Comparative statics of the model First round effect of an increase in Intensity of training (search) effort of workers in the informal sector Employment in the formal private sector Matching efficiency A + + Increase in binding minimum wage Income in the informal sector b +? - - Payroll tax none - Discount rate - - Separation rate - - 12
V. Conclusions Utilizing the 2007 and 2010 recent labor force surveys, this paper systematically characterized the main features of the Swaziland labor markets, with focus on the youth segment (ages 15 24 years). It also analyzed options for labor market policies with a view to bring down the stubbornly high unemployment, especially among the youth. First, features of the labor market and its youth segment were documented with descriptive statistics, showing that the youth and women have less favorable labor market outcomes than other segments of the population. Second, the paper identified determinants of the labor market position of different types of workers aged 20 29 years through a multinomial logit model, showing that age, gender (being a male) and education have negative relation with the likelihood of being unemployed. Third, drawing on search model, the paper analyzed impacts of labor market policies. While active labor market policies such as training or job search would speed up transition to the private sector, they would exacerbate existing inequalities if they are geared only at the public sector employee. Effective employment policies to integrate the vulnerable groups, especially the youth, into the labor market still need to be developed and implementing. 13
References Government of Swaziland, 2007. Labor Force Survey 2007. Government of Swaziland, 2010. Labor Force Survey 2010. Mortensen, D. and C. Pissaridis (1999), New Developments in Models of Search in the Labor Market, in Ashenfelter, O. and D. Card (ed.), Handbook of Labor Eco- nomics, Amsterdam: Elsevier Science, 2567-2627 Stampini, M. and A. Verdier-Chouchane, 2011. Labor Market Dynamics in Tunisia: The Issue of Youth Unemployment. AfDB Working Paper, N 123, February. Van Ours, J.C., 2007. Compulsion in active labor market programmes. National Institute Economic Review, 202(1), 67-78. 14
Annex 1 Table 1 A1. Frequency of the labor force status for the youth aged 20-29 years, 2007 Summary Statistics (% where otherwise indicated) Weighted proportions Linearized standard errors Idle/inactive 0.1666 0.0087 Unemployed 0.4718 0.0108 Wage employment (formal private sector) 0.2487 0.0081 Wage employment (informal private sector) 0.0118 0.0023 Wage employment (public sector) 0.0436 0.0044 Self-employed 0.0575 0.0048 Age 24.114 0.052 Female 0.547 0.009 Hhohho 0.31 0.013 Manzini 0.378 0.012 Shiselweni 0.133 0.008 Lubombo 0.179 0.01 Years of education (years) 9.647 0.069 No education 0 0 Less than primary 0.148 0.008 Primary 0.281 0.01 Secondary 0.209 0.008 High secondary 0.273 0.01 Tertiary 0.088 0.007 Source: Authors calculations based on the 2007 Labor Force Survey. 15