Jobs Series Issue No. 4 Labor Market Outcomes Kazakhstan Victoria Strokova, Angela Elzir and David Margolis Achievements and Remaining Challenges © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immuni- ties of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creative commons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, dis- tribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Victoria Strokova, Angela Elzir and David Margolis. 2016. “Kazakhstan Labor Market Outcomes: Achievements and Remaining Challenges.” A note prepared for Kazakh- stan’s “Jobs: Sector Specific Analysis” (a product of the Joint Economic Research Program). World Bank, Wash- ington, DC. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attri- bution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Contents 1. Introduction 1 2. Jobs and Productivity 3 3. Demographics and Labor Market Outcomes 7 Employment and Unemployment..................................................................................................................................... 10 Determinants of Activity and Employment Type................................................................................................................ 14 Determinants of Wages................................................................................................................................................... 17 4. Access to Jobs 21 Impact of Labor Markets on Poverty................................................................................................................................. 23 Labor Market Outcomes of the Bottom 40 Percent.......................................................................................................... 25 5. Challenges Ahead and Policy Implications 30 Annex A: Additional Tables and Graphs................................................................................................................. 32 Annex B: Methodology and Additional Results on Informality....................................................................... 42 Annex C: Methodology of the Bottom 40 percent Analysis and Results........................................................ 44 Annex D: Methodology and Results of the Earnings Analysis.......................................................................... 50 References.................................................................................................................................................................... 56 iii Acknowledgments This note was prepared by Victoria Strokova (Economist, Jobs Group, co-Task Team Leader), Angela Elzir (Junior Professional Associate, Jobs Group), and David Margolis (Consultant, Jobs Group), with contributions from Aida Imanbekova (Consul- tant, Jobs Group), Aizhan Shamurzaeva (Consultant, Jobs Group), Aizhan Imasheva (Consultant, Jobs Group), Chengyan Gu (Consultant, Jobs Group), Shafique Jamal (Consultant, Jobs Group), and Judy Yang (Economist, Poverty and Equity). Valu- able comments and guidance were received from Namita Datta (Senior Private Sector Development Specialist, Jobs Group, co-Task Team Leader) and Thomas Farole (Led Economist, Jobs Group). The note is a product of the Joint Economic Research Program. The team worked with close guidance from David Robalino (Practice Manager, Jobs Group) and Naveed Naqvi (Program Leader, Human Development and Jobs, Central Asia). Peer reviewers were Roberta Gatti (Lead Economist, Social Protection and Labor), Christos Kostopoulos (Lead Economist, Macroeconomics and Fiscal Management), and Steven R. Dimitriyev (Lead Private Sector Development Specialist, Trade and Competitiveness). The work was carried out under strategic guidance from Saroj Kumar (Country Director), Francis Ato Brown (Country Man- ager, Kazakhstan), and Ludmilla Butenko (Former Country Manager, Kazakhstan). iv 1. Introduction This note presents a detailed analysis of jobs in Kazakhstan at the macro and individual levels, including regional and socio-economic disparities. At the macro level,1 it includes a diagnostic of the links between economic growth, jobs, and productivity across different economic sectors. At the individual level,2 the analysis focuses on labor market outcomes of women and men, young and adult workers, residents of urban and rural areas, and people in the bottom 40 percent of the consumption distribution.3 It also presents a detailed analysis of determinants of employment and wages. The results show that economic growth over the past decade has led to sustained job creation and rapid poverty reduction. The country has benefited from the global commodities boom to become one of the top 10 fastest growing economies in the world, achieving annual real per capita income growth of close to 7 percent. Kazakhstan is also characterized by strong labor market performance, even during periods of economic slowdown, with high labor force participation rates, low inactivity, and low unemployment. Increases in real wages have been the main contributors to pov- erty reduction—from 80 percent in 2001 to 15 percent in 2013 (measured by the PPP-­ corrected US$5 per capita per day). Prosperity was also shared over this period since the growth in consumption of the bottom 40 percent of the population outpaced that of the top 60 percent, resulting in upward mobility and a growing middle class.4 However, the economy remains highly natural resource-dependent and the concentration in capital intensive sectors means that the impact of growth on jobs remains relatively weak. Despite impressive strides, diversification remains a challenge for the country, with minerals, oil, and natural gas accounting for 73 percent of exports and 39 percent of GDP, which has important implications for the number and types of jobs being created. In particular, the dominance of capital-intensive extractive sectors in economic production has implications for the scale of job creation. As a result, the impact of growth on jobs remains relatively weak. Between 2003 and 2013, real GDP grew on average by 7 percent per year while employment expanded only by about 2 percent per year. Furthermore, Kazakhstan faces the same competitive- ness challenges as other resource rich economies. Some of the labor market outcomes, such as the expansion of employ- ment in low-productivity non-tradable sectors (construction, services, etc.), can be attributed to the fact that domestic production of tradable goods remains uncompetitive. Consequently, even though Kazakhstan experienced productivity-enhancing structural changes, jobs continue to be concentrated in low productivity activities. The concentration of investments in capital-intensive sectors means that the impact of growth on jobs remains relatively weak. In fact, while the sector with the largest relative decrease in employment was agriculture (below average productivity), there was little increase in the share of employment in high labor productivity sectors (mining and real estate activities). The main contributors to employment growth were low and below average productivity sectors, such as construction, education, wholesale and retail trade. Despite some reallocation 1 This analysis uses national accounts data and official government statistics on employment and utilizes a recently developed JobStructure tool. 2 This analysis utilizes most recent data for Kazakhstan: the Labor Force Surveys (LFS) for 2010–13 and the Household Budget Surveys (HBS) for 2011–2013. 3 The definition of “bottom 40 percent” (B40) or “top 60 percent” (T60) refers to the distribution of per capita household consumption, in which individuals are ranked by the expenditure of their household, deflated by regional price indices, based on data available in the 2011, 2012 and 2013 Household Budget Survey. Consumption aggregate used is the one developed by the ECAPOV team. 4 Azevedo, Joao Pedro, Sarosh Sattar, and Judy Yang. (2015). “Kazakhstan Economic Mobility and the Middle Class (2006–2013).” Manuscript in progress. 1 of labor in the last decade, a large share of the workforce remains in low productivity employment, particularly in agricul- ture. Self-employment remains relatively high at almost 30 percent of all employed.5 There are important variations in labor market outcomes and wages across regions and between different population groups, potentially exacerbated by limited geographic mobility. While there are relatively small differ- ences in labor force participation rates across regions and different population groups, there are major gaps in access to different types of employment. Individuals in the bottom 40 percent of the consumption distribution are more likely to live in lagging (mostly agricultural) regions, are more likely to be unemployed, or work in low productivity jobs, particularly as self-employed or personal farmstead workers.6 There are also important differentials in wage earnings across sector and regions, and by educational attainment, age and gender. These large wage differentials, especially across regions and sec- tors, indicate some constraints to labor mobility. Removing some of these constraints or lowering individual costs of mobility could help improve access to better jobs. One of the important factors to ensure access to better jobs among the bottom 40 percent of the population is education. The education gap between individuals in the bottom 40 and top 60 percent of the consumption distribution may grow in the future. Although tertiary education has become more prevalent in the population, the gap in educational attainment between the bottom 40 and top 60 percent has widened in more recent generations. As the older generations leave the workforce and the younger ones enter, the skills gap is expected to widen. Therefore, improving the educational attainment and skills of those in the bottom 40 percent could allow them to access higher productivity, better paid jobs. Improving the relevance and quality of education overall is also needed so new entrants into the labor market can convert educational attainment into better skills and higher earnings. Economy-wide institutional reforms continue to be needed to enable private sector job creation and diversifica- tion.7 While Kazakhstan has made impressive improvements in several areas, it remains critical to continue to improve the business environment. This includes reforming laws and regulations to address the specific obstacles faced by enterprises in different sectors; further strengthening the rule of law and improving the quality and effectiveness of service delivery; mak- ing room for the private sector and encouraging competition; improving the performance of the financial sector; and insti- tutionalizing a professional and merit-based civil service.8 It is important to note that a comprehensive government reform program called “100 Steps” was announced in 2015 with the aim to tackle some of these issues. In addition, Kazakhstan could consider additional regional and sectoral policies targeted to the bottom 40 per- cent of the population. Broad-based policies that promote investments into the private sector are fundamental, but the relationship between overall investment and jobs is complex, not always resulting in the types of jobs that may benefit the bottom 40 percent or create jobs for the older workers or the highly skilled. While the government has been promot- ing investments in particular sectors, they are primarily targeted at relatively low labor intensity sectors (manufacturing) or the construction sector which mostly creates temporary jobs. Going forward, in addition to the broad-based reforms, Kazakhstan could consider more targeted policies that aim to promote jobs for specific population groups in given regions. These policies would seek to address barriers to job creation in particular sectors and regions and complement economy-wide policies. The rest of the note is organized as follows: Section 2 discusses the relationship between economic growth, jobs, and productivity across different economic sectors. Section 3 discusses demographic trends and overall labor market outcomes. Section 4 focuses on assessing spatial and sectoral differences in access to jobs, including for those in the bottom 40 per- cent. Section 5 concludes with a discussion of challenges and broad policy implications. 5 While this share is actually not too high for low and middle income countries, it is almost twice as high as the average of the countries in the Organization of Economic Development and Cooperation (OECD) (16.8 percent) and is higher than the Europe and Central Asia (ECA) average of 20 percent. 6 Personal farmstead is defined as work on personal plots of land, either for self-consumption or trade/barter, or both, for at least one hour during the reference week. 7 World Bank. 2013. Beyond Oil : Kazakhstan’s Path to Greater Prosperity through Diversifying, Volume 2. Main report. Washington, DC. 8 Ibid. 2 2. Jobs and Productivity Strong economic growth in the past decade has contributed to robust employment gains, although the employment-growth elasticity has been low by international standards. Between 2003 and 2013, the economy has added 1.5 million jobs. During this period, real GDP grew on average by 7 percent per year, while employment expanded by about 2 percent per year (Figure 1). This implies an elasticity of around 0.28, which is lower than the aver- age for Europe and Central Asia (0.48) and OECD countries (0.5) (Figure 2). In fact, in 2013, employment grew by just 0.7 percent compared to 2012, while GDP grew 6 percent in real terms. Figure 1 Figure 2 GDP and employment growth, 2003–2013 Employment-growth elasticities, 2003–2013 Employment growth (annual %), 2003–2013 Chile 8% +" # Canada 7% *" # Australia 66=# 6% Brazil ! $&# )" # ! ! &<% 5% (" # Kyrgyz Republic % " :;# Tajikistan 6- 7- 8- 9# 4% '"# Malaysia 5 0123/- . /4# 3% &" # Uzbekistan 2% , - . /# %"# KAZ Armenia 1% $" # Korea, Rep. 0% !" # USA !" # %"# '"# )" # +" # $! " # $%"# $' " # $) " # 0% 2% 4% 6% 8% " :;#% 10% 12% 14% 16% , - . /# >>? # 5 6- 7- 8- 9# ! ! &<%! $&# 66=# Kazakhstan GDP Growth (annual %), 2003–2013 Azerbaijan Russia Source: Authors’ calculations using World Development Indicators. Turkmenistan Ukraine China Georgia OECD members ECA 0.0 0.2 0.4 0.6 0.8 1.0 Source: Authors’ calculations using World Development Indicators. Growth has been sufficient to create jobs, as the growth of employment was growing at a faster rate than the growth of the labor force. In the majority of countries, employment has been growing as fast as the labor force (Fig- ure 3). However, in Kazakhstan, employment growth has been higher (2.1 percent) than the labor force growth (1.7 percent) between 2003 and 2013. As a result, unemployment has decreased during this period, as discussed in detail later. Employment gains were driven by the services, construction, trade, and the education sectors. During 2003–2013, employment expanded in construction (contributing 21 percent of the total increase in employment), wholesale and retail trade (15 percent), education (18 percent): transportation and warehousing (10 percent), and other services (35 percent). Employment in manufacturing increased by 15 percent but contributed only 4 percent to employment gains. Employment 3 Figure 3 Employment and labor force growth, 2003–2013 Labor force growth (annual %), 2003–2013 4.0% )"!#$ 3.5% ("%#$ 3.0% *+,-.$/-.01$2.-345$6+778+9$#:;$'!!(<'!&($ ("!#$ 2.5% '"%#$ 2.0% '"!#$ 1.5% &"%#$ KAZ 1.0% &"!#$ 0.5% !"%#$ 0.0% !"!#$ !"!#$ !"%#$ &"!#$ &"%#$ '"!#$ '"%#$ ("!#$ ("%#$ )"!#$ 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% =-4+9$1>?9-@1A$2.-345$6+778+9$#:;$'!!(<'!&($ 3.0% 3.5% 4.0% Total employed growth (annual %), 2003–2013 Source: Authors’ calculations using World Development Indicators. in the public administration and social sectors (such as health and education), both of which have a high share of public employment, contributed almost a third of employment increases during this period. Agriculture was the only sector that contracted, declining in absolute terms by 14 percent. Overall, reallocation of employment away from agriculture toward services progressed at a relatively slow pace, while job creation in industry was very low (Figure 4). The sectors with the highest elasticity of employment to value added during the same period were construction and education; while the lowest was wholesale and retail trade9 (Figure 5). Figure 4 Figure 5 Trend in employment by sector, 2003–2013 Employment-growth elasticities by sector, 2003–2013 3.0 Construction Education 2.5 Transportation, storage and communications 2.0 Human health and social work activities Millions Manufacturing 1.5 Public administration and defence 1.0 Mining and quarrying Wholesale and retail trade 0.5 Agriculture, forestry and fishing 0.0 (1.5) (1.0) (0.5) 0.0 0.5 1.0 1.5 03 04 05 06 07 08 09 10 11 12 13 20 20 20 20 20 20 20 20 20 20 20 Source: WEO; Authors’ calculations using data from the Statistical Committee of RK. Agriculture Mining Construction Public and social Other services Manufacturing Trade Source: WEO; Authors’ calculations using data from the Statistical Committee of RK. 9 Given increasing output, low employment-growth elasticity could indicate increasing labor productivity, which is discussed next. 4 Growth in labor productivity has been the main contributor to GDP per capita growth. In Kazakhstan, GDP per capita grew by 9.1 percent annually between 2003 and 2013. Growing employment and participation contributed posi- tively, albeit marginally (0.46 and 0.37 percent, respectively), to the growth of per capita GDP. The bulk of the per capita value added growth came from increases in labor productivity (8.33 percent). The slightly declining share of the working age population contributed negatively to GDP per capita growth, but by a very small amount (0.06 percent) (Table 1). Table 1 Decomposition of growth in per capita value added, 2003–2013 Period: 2003 to 2013 % of % Yearly Total Contribution Change Change to Growth Change in per capita value added 390.74 100.00 9.10   due to changes in productivity 357.80 91.57 8.33   due to changes in employment rate 19.70 5.04 0.46   due to changes in participation rate 15.73 4.03 0.37   due to changes in share of working age population –2.49 –0.64 –0.06 Source: Authors’ calculations using JobStructure tool and data from Statistical Committee of RK. Labor productivity growth has been driven mainly by the wholesale and retail trade as well as the mining sec- tor. The decomposition of value added per worker changes shows that the majority of labor productivity growth came from within sectors, such as mining and wholesale and retail trade, which contributed approximately 22 and 23.4 percent, respec- tively (Table 2). Other activities10 and manufacturing contributed an additional 17.5 and 7.6 percent, respectively. Agriculture contributed only 4 percent to labor productivity growth. Intersectoral shifts (i.e. changes in productivity due to reallocation of workers from less to more productive sectors) contributed about one fifth to overall labor productivity growth. 10 Other activities include the following sectors: public administration; education; health, public utilities; finance and insurance; real estate activities; accommodation and food services; professional, scientific and technical activities; arts, entertainment and recreation; and administrative and support service activities. 5 Table 2 Decomposition of value added per worker into sector changes and inter-sectoral shifts Period: 2003 to 2013 Change % Contribution Agriculture 19.35 4.01 Mining & Utilities 105.88 21.96 Manufacturing 36.72 7.62 Construction –1.09 –0.23 Wholesale & Retail 112.73 23.38 Transport & Communications 23.32 4.84 Other Activities 84.43 17.51 Not Defined 3.98 0.83 Intersectoral Shift 96.91 20.10 Total change in productivity (value added per worker) 482.23 100.00 Source: Authors’ calculations using JobStructure tool and data from Statistical Committee of RK. While Kazakhstan has experienced some ­ productivity-enhancing structural changes, a large share of the popu- lation continues to be employed in low productivity sectors. Despite a significant reduction in the share of employ- ment in agriculture (10 percentage points between 2003 and 2013), there are still about 2 million people, or a quarter of all employed, who remain engaged in this sector. Furthermore, many of the other sectors that increased their share of employment, such as construction and education, also have below average productivity (Figure 6). As a result, overall labor productivity remains low, especially for the non-oil sectors, and compared to countries with similar GDP per capita levels.11 Figure 6 Sectors and structural change in Kazakhstan, 2003–13 1.0 Log (sectorial productivity/total productivity), 2013 Real estate activities 0.8 Mining 0.6 0.4 Financial & insurance Manufacturing 0.2 Transportation & communication Wholesale & retail trade Other 0.0 –12% –10% –8% –6% –4% –2% 0% 2% 4% 6% –0.2 Construction Accommodation & food service Public administration & defence –0.4 Health & social work –0.6 Agriculture Education –0.8 –1.0 Change in employment share, 2003–2013 Note: The size of the bubble represents the sectoral employment shares in 2003. Source: Authors’ calculations using data from Statistical Committee of RK. 11 World Bank. 2013. Beyond Oil : Kazakhstan’s Path to Greater Prosperity through Diversifying, Volume 2. Main report. Washington, DC. 6 3. Demographics and Labor Market Outcomes Demographic trends have important implications on the labor market. In Kazakhstan, fertility rates dropped sig- nificantly in the 1990s and early 2000s. Fertility increased sharply starting in 2005 and peaked in 2010 at 2.5 children per woman, constituting a short “baby boom” period. Fertility subsequently started to decline and is projected to continue decreasing to the levels of the early 2000s (2 children per woman). These trends affect the size of the working age popula- tion and, therefore, of the labor force (Figure 7). Figure 7 Trend in fertility, population pyramid (2010 and 2050) in Kazakhstan 3.5 2010 100 Total fertility (children per woman) 3 Males Females 90 2.5 80 2 70 60 1.5 50 1 40 30 0.5 20 0 10 1990 2000 2005 2010 2015 2020 2030 2050 2075 2100 0        1000 800 600 400 200 0 200 400 600 800 1000 2050 100 Males Females 90 80 70 60 Age 50 40 30 20 10 0 1000 800 600 400 200 0 200 400 600 800 1000 Source: UN projections. 7 At the same time, labor force participation rates are already high in Kazakhstan and unlikely to contribute much more to labor force growth. Almost 78 percent of men and 68 percent of women aged 15+ participate in the labor mar- ket.12 While women participate less than men, the share of women participating in the labor force is very high by international standards (Figure 8 and Figure 9). The share of individuals who are not in employment, education or training (NEET) is low throughout age cohorts, including among youth, with the exception of people aged 60–64 as they retire (Figure 10). The number of youth entering the labor market increases sharply between the ages of 18–25, after which the activity rate flattens (Figure 11). Around two-thirds of workers exit the labor market between the ages of 60 and 64. Women tend to exit the labor force earlier due to lower retirement ages.13 Figure 8 Figure 9 Male labor force participation rates, 2013 Female labor force participation rates, 2013 Moldova Moldova Belarus Malaysia Ukraine Mexico Lithuania Turkmenistan Latvia Uzbekistan United States Chile Estonia Belarus Azerbaijan Ukraine Canada Armenia Russian Federation Latvia Australia Lithuania Armenia Kyrgyz Republic Chile Estonia Georgia United States Malaysia Georgia Uzbekistan Russian Federation Turkmenistan Australia Tajikistan Tajikistan Kazakhstan Canada China Azerbaijan Kyrgyz Republic China Mexico Kazakhstan 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% Source: WDI (labor force participation rate, 15+); Statistical Committee of RK. Source: WDI (labor force participation rate, 15+); Statistical Committee of RK. 12 In 2013, 82.5 percent of men and 75.5 percent of women of working age, defined as those 15–64 years old, were in the labor force. 13 Until recently, retirement ages were 63 years old for men and only 58 for women. However, in 2013 a law was signed gradually increasing the retirement age for women from the current 58 to 63 years old within a decade. 8 Figure 10 Figure 11 Status of working age population by age, 2013 School to work transition, 2013 100% 90% $!!"# 100% 80% 90%,!"# 70% 80%+!"# 60% 70%*!"# 50% 60%)!"# 40% 50%(!"# 30% 40%'!"# 20% 30%&!"# 10% 20%%!"# 0% 10%$!"# 0% !"# l 9 4 9 4 9 4 9 4 9 4 al –1 –2 –2 –3 –3 –4 –4 –5 –5 –6 r $(# 16 15 $)# 17 $+# 19 $*# 18 $,# 20 %$# 22 %%# %&# %!# 21 %(# 26 %'# 25 23 24 %)# %*# %,# 30 %+# 29 27 28 &!# ve 15 20 25 30 35 40 45 50 55 60 O Age - Age . /0123# Employment Unemployment Inactivity NEET Employment Unemployment Inactivity 4156378# 98/01:.3578# ;81<3=>578# Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. Due to the fall in fertility in the 1990s and constant participation rates, the growth of the labor force in the next few years is expected to be much slower. Kazakhstan will experience a relatively slow growth in the labor force14 in the next few years (Figure 12). In fact, while the labor force grew by 1.5 percent annually in 2003–2010, it is projected to increase only by 0.5 percent annually in 2015–2020, which means that it will increase by less than 60,000 persons per year. As a result, the pressure on the labor market may be reduced in the short run.15 The growth rate of the labor force will peak around 2030 when the “baby boom” generation enters the labor market (Figure 12). This will mean that annually, the labor force will grow by as many as 135,000 people, requiring a much faster job creation pace during that period. Figure 12 Projected labor force growth, 2010–2050 160,000 4,500 140,000 4,000 3,500 Thousands 120,000 3,000 100,000 2,500 80,000 2,000 60,000 1,500 40,000 1,000 20,000 500 – 0 2010 2012 2014 2016 2018 2020 2022 2024 2026 2028 2030 2032 2034 2036 2038 2040 2042 2044 2046 2048 2050 y/y LF increase, left axis Cumulative LF increase, right axis Note: For working age population of 16+, assuming 72 percent labor force participation rate. Source: Demographics projections and trends in working age population and labor force are based on the Statistical Committee of RK. 14 Assuming constant labor force participation rate of 72 percent. 15 Assuming the current deterioration of the economic situation does not lead to a significant drop in job creation and a spike in unemployment. So far the impact of the crisis on the labor market has been limited (World Bank Biannual Economic Update, 2015), but simulations presented later in the note indicate that unemployment may increase somewhat by 2020 should current economic trends persist. 9 External migration flows remain relatively low, especially in comparison to the 1990s. Compared to the 1990s, migration flows have decreased dramatically in Kazakhstan. Estimates for official inflows and outflows are less than 30,000 persons per year, even though unofficial migration could be higher. It should be noted that during this period Kazakhstan has been actively pursuing a policy of encouraging repatriation of ethnic Kazakhs living abroad. Almost a million (944,500) ethnic Kazakhs immigrated to Kazakhstan during 1991–2013, which helped to smooth the decrease in the demographic growth observed in the country in the 1990s.16 While migration trends may change in the future,17 the impact of external migration on the labor market remains relatively low (Figure 13). Figure 13 Official migration flows in Kazakhstan, 1991–2013 500 400 477.1 Thousands of persons 300 299.5 200 100 70.4 38.1 24.4 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Emigrants Immigrants Source: Economic Research Institute. Employment and Unemployment Employment rates have been traditionally high in Kazakhstan, especially for women, and unemployment has been declining steadily. The employment rate18 was 75 percent for men and 63 percent for women in 2013, which are very high levels by international standards.19 Employment rates are slightly higher in rural areas (Figure 14). As a result of the rapid growth in the number of jobs, the unemployment rate was halved between 2001 and 2013, from 10.4 to 5.2 percent (Figure 15). Long-term unemployment remains very low. The unemployment rate is slightly higher among women (5.9 per- cent compared to 4.6 percent among men) and urban areas (5.4 percent vis-a-vis 4.9 percent in rural areas) (Figure 14). The recent economic slowdown has not yet affected employment in any significant way, with the unemployment rate holding steady at 5 percent in the first quarter of 2015.20 16 Source: Economic Research Institute. 17 For instance, a slowdown in neighboring countries may increase the number of labor migrants coming to Kazakhstan. 18 Calculated as the number of employed divided by the number of working age population (15+). 19 Figure 44 in Annex A. 20 Data from the Statistical Committee of RK for Q1 of 2015. 10 Figure 14 Figure 15 Labor force status in working age population, 2013 Unemployment rates, 2001–2013 100% 25% 90% 20% 80% 15% 70% 10% 60% 5% 50% 0% 40% 01 02 03 04 05 06 07 08 09 10 11 12 13 30% 20 20 20 20 20 20 20 20 20 20 20 20 20 20% 10% Unemployment rate, % 0% Youth unemployment rate, % (15–24 years) Overall Male Female Urban Rural Long- term unemployment, % Employment Unemployment Inactive Source: Statistical Committee of RK. Note: Working age = 15–64 years of age. Source: Authors’ calculations using LFS 2013. While unemployment among 15–24 year olds decreased dramatically, younger people (aged 25–34) do have somewhat higher unemployment rates. Youth unemployment fell drastically from almost 20 percent in 2001 to just under 4 percent in 2013, which can be attributed in large part to increasing enrollment in tertiary education.21 However, younger people aged 25–29 and 30–34 do have somewhat higher unemployment rates (approximately 7 percent compared to a national average of 5 percent) (Figure 16). Even controlling for other characteristics, this age group is more likely to be unemployed. Similarly, unemployment is slightly less likely among males and the more educated (those with upper second- ary and tertiary education) (Table 5 in Annex A). Those living in urban areas are also slightly less likely to be unemployed. Figure 16 Unemployment rates by age, 2013 8% 7% 6% 5% 4% 3% 2% 1% 0% 9 4 9 4 9 4 9 4 9 4 –1 –2 –2 –3 –3 –4 –4 –5 –5 –6 15 20 25 30 35 40 45 50 55 60 Source: Authors’ calculations using LFS 2013. 21 Tertiary enrollment (gross) increased from 29 percent in 2000 to 45 percent in 2012 (WDI). 11 While the majority of workers are wage employees, self-employment still remains relatively high. Although there has been a gradual shift from self-­ employment toward wage employment, the share of self-employment remains high at 29 percent of the employed (Figure 17) and more than half of them (52.5 percent) are working in agriculture. In contrast, the share of self-employed is about 20 percent in ECA countries and 16.8 percent among OECD countries.22 Wage employment has increased from about 61 percent of total employment in 2003 to about 66 percent in 2013, mirrored by a decrease in the number of self-employed. Wage employment is significantly lower in rural areas (55 percent). Most of the difference in employ- ment between rural and urban areas is due to personal farmstead employment,23 although own-account workers and employ- ers are also relatively more frequent in rural areas. The share of public employment24 among wage employees is also high—at 42.3 percent—reflecting the big role, albeit decreasing,25 that the state continues to play as an employer. Figure 17 Overview of the working age population in Kazakhstan, 2013 Public 2.5m; 42.3% Wage Employee Informal: 5.9m; 65.8% 14.4% Active Private 9.0m; 72.1% 3.4m; 57.7% Unemployed Formal: 0.5m; 5.2% 85.6% Working Age Agriculture Population (15+) 1.4m; 52.6% 12.5m; 74.4% Self-Employed 2.6m; 29.0% Inactive Non-Agriculture 3.5m; 27.9% 1.2m; 47.4% Note: Self-employed include own-account workers, employers, farmstead workers, members of cooperatives and unpaid family workers. Source: Authors’ calculations using LFS 2013. Wage employment is high in almost all sectors, except agriculture and trade. Wage employment is nearly universal in mining, manufacturing, public administration, public utilities, and other services. Self-employment is more prevalent in retail and wholesale trade (commerce), construction, transport and communications, and agriculture (Figure 18). 22 Source: WDI (ECA) and OECD (https://data.oecd.org/emp/self-employment-rate.htm), respectively. 23 Defined as those who have worked on their personal farmstead for at least one hour during the reference week. 24 Defined as those wage employees reporting state ownership of the organization they work in. 25 Analysis of HBS data shows that the share of public employment in total wage employment decreased by almost 10 percentage points from 2006 to 2013. 12 Figure 18 Distribution of employment by type and sector in Kazakhstan, 2013 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% re es g s n ce . te n s m tie ice rin tio tio tu iv ta er m ili rv ct tu uc es ra ul m co ut ra se ric m ist ac tr l d a t ic ns Co er ex uf in Ag re n bl ta Co th m an d d Pu ad O an an or M sp ic e g nc an bl in Pu in ra Tr M su in e, nc na Fi Wage employment Own-account self-employed Employer Farmstead worker Note: Excluding unpaid family workers, which represent 1 percent of employed in agriculture and less than 1 percent in other sectors. Source: Authors’ calculations using LFS 2013. The majority of self-employed in agriculture work on personal farmsteads, while self-employed in non-agriculture are mostly own-account workers. Almost two-thirds (64 percent) of self-employed in agriculture are those who work on per- sonal farmsteads. Only about 30 percent are own-account workers, while the rest are split among employers (4 percent), unpaid family workers (1.5 percent), and cooperative members (0.5 percent). Among self-employed in non-agricultural activities, 90 percent are own-account workers and another 8.5 percent are employers. Only a very small share are unpaid family workers (1 percent) and members of cooperatives (0.5 percent). Informality is concentrated in self-employment, particularly in agriculture. A relatively small share (only 8.3 per- cent) of wage employees are informal, but over two-thirds of agricultural self-employment (which is largely personal farmstead work, as noted above) is informal26 (Table 3 and Annex 2). Very few employers are informal (only 3.4 per- agricultural self-employed are informal. However, it should be noted cent), while about one fifth (18 percent) of non-­ that measurement of informality for self-employment is somewhat less reliable.27 Due to this measurement issue and prominence of wage employment in Kazakhstan, this note primarily focuses on informal wage employment. 26 For wage employment, this note adopts a broad definition of informality based on two standard questions. First, a wage worker is considered informal if he or she does not have a written contract. Wage workers are also considered informal if their employer does not contribute to social insurance/the pension fund on their behalf. All unpaid family workers, cooperative members and personal farmstead workers are considered informal, and among the self-employed (employers or own-account workers), those whose enterprise is not registered are considered informal. 27 The question on registration changed in 2012 to make it applicable to all employment categories, whereas it previously applied only to the self- employed. As a result, the LFS shows a significant difference in the share of formal self-employed between 2011 and 2012. 13 Table 3 Distribution of informality by job type, 2013 Share of Share of Percent Type of Employment28 Employed Total Informal Public Wage Employment 29.35 42.30 0 Private Formal Wage Employment 34.27 49.40 0 Private Informal Wage Employment 5.76 8.30 100 Agricultural Self Employment 16.11 52.61 69.31 Personal Farmstead 10.46 34.15 100 Non-Agricultural Employer 1.23 4.03 3.36 Non-Agricultural Self Employment 13.28 43.36 18.17 Source: Authors’ calculations using LFS 2013. 28 Determinants of Activity and Employment Type Men and the more educated are more likely to participate in the labor market. Controlling for a set of characteristics,29 men are 10 percentage points more likely to be active than females. Besides gender, education plays a big role, with those with upper secondary and tertiary education being 17.7 and 15.6 percentage points, respectively, more likely to be active than those with lower secondary or less. Age is related to activity in a predictable manner: 15–19 year olds, who are over- whelmingly in school, are less likely to be active (by 15 percentage points), as well as those aged 55 and above, particularly those 60 years and older, who are 70 percentage points more likely to be inactive compared to 25–29 year olds. There are also some regional differences, with those in Astana particularly less likely to be active compared to the Akmola region, and those in urban areas are only slightly more likely to be inactive than those in rural areas (Table 5 in Annex A). Those who are more likely to secure public wage employment compared to private formal employment appear to be different in several respects. Keeping other characteristics constant, males are significantly more likely to be employed in formal private sector jobs compared to women, who are more likely to work in the public sector (Figure 19). Younger people (20–24 years old) are more likely to work in the private sector even compared to a relatively young cohort of 25–29 year olds. On the other hand, older people (40 and above) are much more likely to be in the public wage employment. While higher levels of education increase the probability of both public and private wage employment com- pared to lower secondary education, the effect is most dramatic for those with tertiary education, who are 31 percentage points more likely to be in the public sector. On the other hand, those with upper secondary education are much more likely to be in the private formal sector than those with lower secondary. There are also significant regional differences; however, while those in urban areas are more likely to be in the private formal sector compared to those in rural areas, this is not the case for the public sector, where there are no statistically significant differences between rural and urban areas. 28 Due to changes in the LFS questionnaire, the classification of public and private formal sector employment shifted between 2011 and 2012, so most of the analysis that distinguishes between these two types of formal employment focuses on 2013 only. Shares of employment by public/ private among formal workers: 2010 = 32.8 percent/58.3 percent; 2011 = 33.0 percent/58.8 percent; 2012 = 42.6 percent/48.4 percent; 2013 = 42.3 percent/49.4 percent. 29 Multinomial logits estimated on the following outcomes: NEET (Not in Employment, Education or Training), in school or training, unemployment, farmstead worker, non-agricultural self-employed, public wage employment and private formal wage employment. Reference categories include female, 25–29 years old, lower secondary or less, rural resident, in Akmola region, and single person household. Informal wage employment and its determinants are considered separately. 14 Figure 19 Marginal effects for multinomial logit model, 2013 Urban Almaty City East Kazakhstan Pavlodar South Kazakhstan Mangystau Kyzylorda Kostanay Karaganda Jambyl Atyrau Aktobe Tertiary education Upper secondary education 60–64 Years old 50–54 Years old 45–49 Years old 40–44 Years old 20–24 Years old Male –40.00 –30.00 –20.00 –10.00 0.00 10.00 20.00 30.00 40.00 Private formal wage employment Public wage employment Interpretation: A male has 11.4 percentage points higher chance of having private formal wage employment than a female when both have the average characteristics of the working age population (15–64 years old). Note: Only coefficients significant for both public wage employment and private formal wage employment are included, with the exception of urban which is only significant for private formal wage employment. Reference categories include female, 25–29 years old, lower secondary or less, rural resident, in Akmola region, and single person household. Source: Authors’ calculations using LFS 2013. Determinants of self-employment in agriculture and non-agriculture differ, but younger people tend to be more likely to engage in both. Not surprisingly, residents of urban areas are considerably less likely to be in agricultural self- employment while location (urban-rural) does not make a difference for non-­ agricultural self-employment (Figure 20). Younger workers (15–19 and 20–24 years old) are more likely to be self-employed in agriculture compared to 25–29 year olds, but the number of such workers is very small.30 Younger people are also more likely to engage in non-agricultural self- employment, but there are also relatively few of them.31 Older workers, on the other hand, especially those over 60 years of agriculture than prime-age workers (25–29 year olds). Workers age, are significantly less likely to be self-employed in non-­ with tertiary education are less likely to be self-employed compared to those with lower secondary or less, especially outside of agriculture. 30 Approximately 16 percent of all self-employed in agriculture or about 232,000 workers. 31 Only 13 percent of all self-employed outside of agriculture or about 172,500 workers. 15 Figure 20 Marginal effects for multinomial logit model, 2013 Urban Almaty City East Kazakhstan Pavlodar South Kazakhstan Mangystau Kyzylorda Kostanay Karaganda Atyrau Aktobe Tertiary education 60–64 Years old 45–49 Years old 40–44 Years old 20–24 Years old 15–19 Years old Male –15.00 –10.00 –5.00 0.00 5.00 10.00 Self-employment in agriculture Self-employment outside of agriculture Interpretation: A male has 3.75 percentage points higher chance of being self-employed outside of agriculture than a female when both have the average characteristics of the working age population (15–64 years old). Note: Only coefficients significant for both self-employed in agriculture and non-agriculture are included. Reference categories include female, 25–29 years old, lower secondary education or less, rural resident, in Akmola region, and single person household. Source: Authors’ calculations using LFS 2013. Informal wage workers tend to be younger, less educated and more likely to be found in rural areas than for- mal private wage workers. Informal wage workers start working younger, and nearly a third of all informal workers are under the age of 30 (Figure 21). In addition, only 13.5 percent of informal wage workers have a tertiary education, relative to 34.9 percent for formal wage workers (and 52.5 percent for public sector workers) (Table 6 in Annex B). Informal work- ers also drop out of education earlier. Less than half (46.4 percent) of informal wage workers live in urban areas, against 61.7 percent of public sector and 69.6 percent of formal private wage workers. On the other hand, very few people work in the public sector at young ages. 16 Figure 21 Age distribution by formality status 20% Share of workers 15% 10% 5% 0% 9 4 9 4 9 4 9 4 9 4 –1 –2 –2 –3 –3 –4 –4 –5 –5 –6 15 20 25 30 35 40 45 50 55 60 Age group Public wage workers Private formal wage workers Private informal wage workers Source: Authors’ calculations using LFS 2013. Sector and location appear to play a large role for informality. Detailed statistical analysis32 shows that while certain determinants of informal status are very strong (there is almost no informal wage employment in mining, health and social services, education and public administration services, for example), other effects are less clear. For example, informality in agriculture is significantly higher than in other sectors with the exception of commerce, while informality is significantly lower in Astana and Almaty cities. However, outside of agriculture and the main cities, a wage worker is no more likely to be informal in rural areas than in urban areas. Similarly, the apparent differences in formality related to education levels appear to be largely due to the fact that the more educated workers are more likely to work in low-informality sectors and low-informality regions. Thus, it seems that people with an upper secondary education degree are more likely to be informal than those with a lower educational attainment, but these workers are also more likely to be wage employees. The combination of these factors implies that the level of educational attainment actually does not play a significant role in explaining formality among those who have wage jobs (Table 7 in Annex B). Determinants of Wages Similar to many other countries, an analysis of wages33 in Kazakhstan reveals a large gender gap and significant wage differentials by age (Annex 4). Men tend to earn more than women with identical characteristics; wages for men were 27–31 percent higher than for women over the 2011–2013 period. These estimates control for sector of work and worker characteristics so these wage gaps do not result from the fact that women are less likely to work in high-paying sec- tors such as mining, manufacturing or transportation, and are more likely to work in low paying sectors like commerce and other services (especially the education and health and social services sectors). They also do not reflect education differentials 32 See Annex B for more information about the methodology. 33 The analysis in this section exploits data from the 2011–2013 Household Budget Surveys to estimate the determinants of wages, overall and on a sector-specific basis. It then uses the 2011–2013 Labor Force Surveys to estimate how much individuals with different skill levels could expect to earn in different regions, and compares this to the actual distribution of employment of these skill levels in the different regions. In the absence of constraints, one should see a higher concentration of skills in the regions where these skills are the most highly rewarded; any deviation from this distribution is indicative of constraints to labor reallocation (see Annex D for more details on the methodology). 17 or disproportionate work in rural areas.34 Holding these characteristics constant, women are paid less than men suggest- ing that there are other factors contributing to the gender wage gap. There are also significant wage differentials by age. Wages in Kazakhstan increase with age up to around 45 years of age, after which they tend to decline. This trend has been strengthening in recent years (Figure 22). This age profile of wages is also typical of most countries, as it reflects the process of skills accumulation, depreciation and selection into and out of the labor market. Figure 22 Wage differentials (relative to 15–19 year olds) by age group, 2011–2013 40% 30% 20% 10% 0% 9 4 9 4 9 4 9 4 9 4 –1 –2 –2 –3 –3 –4 –4 –5 –5 –6 15 20 25 30 35 40 45 50 55 60 2011 2012 2013 Source: Authors’ calculations using 2013 HBS. Additional skills—in the form of education—bring a wage premium, particularly for those with tertiary educa- tion. The estimated wage premium is 7–12 percent for an upper secondary education and 39–48 percent for a tertiary education (both relative to a lower secondary education or less).35 These estimates suggest that employers do value an upper secondary education more highly than simply a lower secondary education or less, but not by much. However, there is a significant premium in wages that comes with having a tertiary education. The lack of a significant premium for upper secondary education may be related to the low quality of education at secondary and upper secondary levels. At the inter- national level, despite high levels of enrollment and completion of secondary education, Kazakhstan fares poorly in educa- tion quality, as reflected in the poor, though markedly improved, performance on international student assessments,36 such as the OECD’s Programme for International Student Assessment (PISA). The 2012 PISA results suggested that Kazakhstani students underperformed compared to their peers in comparator countries in reading, mathematics and science (Figure 67 in Annex A). Not surprisingly, different sectors of the economy pay different wages, even after controlling for worker characteris- tics (Figure 23).37 The mining and extractive industries sector pays the most on average, with its workers earning wages between 78 and 84 percent above what similar wage workers earn in agriculture in the same year.38 Manufacturing, transportation and communication, and finance, insurance and real estate are also relatively high paying sectors, while wage workers in agriculture, other services, and commerce earn the least. Not only do different sectors pay similar workers different amounts, they reward the same characteristics differently as well (Table 11 in Annex 4). Returns to skills, age, residing in an urban area or gender can vary 34 In fact, women make up a larger share of tertiary educated wage workers (56 percent) and slightly more than half of urban wage workers (51 percent), while men occupy 53 percent of wage jobs in rural areas. 35 Estimates from 2011–2013. See Table 11 in Annex D for results for 2013. 36 Kazakhstan’s performance on PISA improved markedly since 2009, especially in math and science and also among the lowest achievers, but its overall achievement remains significantly behind other countries with similar income per capita levels. 37 Without controlling for other characteristics, the variation in wages within a sector can be quite large, although the ranking of sectors discussed here holds on average. 38 There are few consistent trends over the 2011–2013 period, although wage gaps appear to be shrinking somewhat in finance, insurance and real estate, transportation and communication, and public administration services. 18 Figure 23 Trends in relative wages across sectors (relative to agriculture), 2011–2013 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% g g s n ce n e s n s s tie ce ce ice at in rin io tio tio er vi vi in t st ili rv tu uc ica ca m er er e M ut se m ac u tr al s ls un Ed ns Co ic n er uf re cia m tio bl Co th an d m Pu so ra O an M co ist d an e d in nc an m th ra ad n al su tio He in ic ta bl e, or Pu nc sp na an Fi Tr 2011 2012 2013 Source: Authors’ calculations using 2013 HBS. dramatically across different industries. For example, gains to having a tertiary education degree range from 12 percent above a lower secondary education degree in the agriculture sector to 58 percent higher in manufacturing. There are also significant regional and urban-rural wage differentials, mainly explained by the economic sec- tors prevalent in different parts of the country. In fact, urban residents earn 16–18 percent more than rural residents. Such rural-urban wage differentials are common around the world and reflect a higher level of labor demand in urban areas (Table 11 in Annex D). In terms of regional wage variation, the same worker earns 51 percent more than the national aver- age in Mangystau and 9 percent less than the national average in North Kazakhstan (Figure 24), and these premia do not simply serve to cover differences in cost of living. Such large wage differences across sector and location of employment, controlling for worker characteristics, imply that there may be barriers to mobility which would allow workers to move where they can obtain higher paid jobs given the skills they have. Labor mobility is known to be low in Kazakhstan,39 while constraints to mobility could be cultural (desire to stay near the historic family residence), financial (not enough money to pay for a move or to acquire housing in the new location) or informational (not knowing where the high paying jobs are located). Finally, while the public sector commands only a slight wage premium, it attracts a very high share of the ter- tiary educated. As was shown earlier, those with tertiary education are much more likely to work in the public sector com- pared to those with lower secondary education. In fact, half of those working in public wage employment have completed tertiary education compared to only 30 percent for private formal wage employment and just 10 percent of private non- formal wage employment.40 However, on aggregate, controlling for employee characteristics the public sector enjoys only a 39 Arias, Omar S.; Sanchez-Paramo, Carolina; Davalos, Maria E.; Santos, Indhira; Tiongson, Erwin R.; Gruen, Carola; de Andrade Falcao, Natasha; Saiovici, Gady; Cancho, Cesar A. 2014. Back to work: growing with jobs in Europe and Central Asia. Europe and Central Asia Reports. Washington, DC: World Bank Group.  40 Based on LFS 2013. 19 Figure 24 Wage differentials and cost of living (relative to Akmola) by region, 2013 51% 43% 28% 25% 15% 13% 11% 10% 6% 7% 8% 6% 7% 5% 5% 4% 2% 0% 0% –1% –2% –2% –4% –5% –5% –6% –4% –7% –8% –9% be y au ty an l a ay a u an ar an an ty by at nd rd ta Ci Ci od an to yr st st st st m m lo s ga kh gy kh kh kh At na y Ak vl st Ja Al zy at ra Pa Ko an za za za za ta Ky m Ka Ka Ka Ka Ka As M Al t h h st es ut rt Ea W No So Wages Cost of living Note: The effects pictured are drawn from the overall regression, in which sectors are controlled for with a set of indicator variables, and not the sector specific regressions. Thus, this specification imposes the same differentials for all sectors. 2 percent wage premium compared to the private sector (Table 11 in Annex D). Returns to higher education are high in some sectors with a large share of public employment, such as education (Table 12 in Annex D), but wages in those sectors tend to be on the lower side overall. Hence, wage differentials do not fully explain the appeal of the public sector. Public sector employment is likely to be associated with other benefits (such as better social benefits or social prestige, etc.) that make it particularly attractive to the well-educated. 20 4. Access to Jobs There are relatively small differences in labor force participation rates across regions, even when taking into account the gender dimension. The differences in participation rates and unemployment rates are not large in Kazakhstan, even between men and women (Figure 25 and Figure 26). The regions with the lowest participation rates are Almaty City and East Kazakhstan, at around 65 percent each, while the regions with the highest rates are Akmola and Zhambyl, at almost 80 percent. Similarly, female labor force participation rates are highest and lowest in those four regions. Figure 25 Figure 26 Working age population by region, 2013 Labor force participation rates by gender and region, 2013 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Almaty Akmola Aktobe Atyrau East. Kaz. Mangystau North Kaz. Pavlodar Karagandy Kostanay Kyzylorda South Kaz. West Kaz. Zhambyl Astana City Almaty City Akmola Aktobe Almaty Atyrau West Kaz. Jambyl Karaganda Kostanay Kyzylorda Mangystau South Kaz. Pavlodar North Kaz. East Kaz. Astana City Almaty City Male Female Total population Employed Unemployed Not in a labor force Note: Working age population = 15–64 years old. Note: Working age population = 15–64 years old. Source: Authors’ calculations using 2013 LFS. Source: Authors’ calculations using 2013 LFS. However, the types of employment differ substantially across regions and are largely dependent on the eco- nomic sectors prevalent in different parts of the country (Figure 27 and Figure 28). Regions that are relatively special- ized in mining and extractive industries (Mangystau, Atyrau, Karaganda), as well as the cities of Almaty and Astana, have particularly high concentrations of wage employment, while agriculture-intensive areas like North Kazakhstan, Zhambyl and Kostanay had high shares of self-employed working in agriculture, including farmstead workers. 21 Figure 27 Figure 28 Employment type by region, 2013 Employment by sector and region, 2013 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% Ak t y Ak la At e u gy . u v z. ra ar Ko ndy zy y ut da tK . Zh az. na l at ty ty z W Kaz y Ak la Al e At y W yrau Ja z. ga l Ko nda zy y an da u . rt ar . . y ty Ky na b y Pa az Ea Kaz z a No s t a M t Ka Ka As m b at Ky ana t o b a i Ci So ysta Ka lod a Ka m b As t Ka o to yr r C i Ci No lod m m to M lor Al a C lo tK K a ga m m y st h h a Al y st h h Ak s es g v at rt n Pa es s Ea an ut ta ra m ta m So Al Wage formal public Wage formal private Services Agriculture Wage informal Self-employed agriculture Mining, manufacturing and construction Self-employed non-agriculture Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. Informal wage employment is much more prevalent in South Kazakhstan and in the agriculture, commerce and other services sectors. The South Kazakhstan region has the largest share of informal wage workers (almost a third), by a significant margin, while Astana and Almaty cities have a high share of formal workers (Figure 29). Informal wage workers are employed primarily in the agriculture, commerce and other services sectors. In contrast, public employment is largely concentrated in education, health and social services and public administration (Figure 30). Formal private wage work, on the other hand, is not dominated by any particular sector. Figure 29 Wage employment types within region, 2013 NORTH KAZAKHSTAN PAV L O D A R AKMOLA KOSTANAI WEST Astana KAZAKHSTAN EAST K A Z A K H S TA N AKTOBE ATYRAU KARAGHANDY ALMATY KYZYLORDA MANGYSTAU ZHAMBYL SOUTH KAZAKHSTAN Public Wage Employment Private Formal Wage Employment Private Informal Wage Employment Note: Green = public wage employment; dark blue = private formal wage employment; light blue = private informal wage employment. Cities of Astana and Almaty excluded. Their shares of public wage employment, private formal wage employment and private informal wage employment are as follows: Astana (42 percent, 55 percent, 3 percent) and Almaty (41 percent, 54 percent, 5 percent). Source: Authors’ calculations using LFS 2013. 22 Figure 30 Sectors by formality status, 2013 100% 80% 60% 40% 20% 0% Public wage Private formal Private informal workers wage workers wage workers Other services Commerce Finance, insurance & real estate Construction Transport & communication Manufacturing Public administration, utilities & social sectors Note: Social sectors include education, health and social services. Source: Authors’ calculations using LFS 2013. Impact of Labor Markets on Poverty Over the period 2006–2013, labor incomes contributed a large share to the decline in Kazakhstan’s poverty rate.41 Extreme poverty levels are now very low in Kazakhstan,42 and the share of the population living in poverty went down from 80 percent in 2001 to 15 percent in 2013, as measured by the PPP-corrected US$5 per capita per day.43 The main sources of observed poverty reduction were wage earnings and, to some extent, pensions (Figure 31). Consistent with the considerable contribution of earnings to poverty reduction in 2006–2007 (6.8 percentage points), real wages increased significantly during this period but dropped in 2008 during the economic crisis (Figure 32). Wage growth rebounded quickly after 2008, but at a significantly lower rate, thus contributing to a lesser extent to poverty reduction. Still, wages contributed 4.8 percentage points to the decrease in the poverty rate over the 2012–2013 period. Growth in consumption of the bottom 40 percent of the population outpaced that of the top 60 percent, resulting in increased upward mobility. As a result, the size of the middle class more than doubled between 2006 and 2013 (Azevedo, Sattar and Yang, 2015). Continued poverty reduction will depend on wage increases, but recent wage growth has been outpacing labor productivity. Overall productivity growth over the past decade has been lower than wage increases, even at the sectoral levels (Figure 33 and Figure 34). In the period 2004–08, wage growth (9 percent annually) outstripped labor productiv- ity growth (5 percent annually) to some extent. However, after 2008, productivity growth stagnated (1 percent annually), but robust wage growth continued (7 percent annually). The sectors with the highest increases in wages are mainly low- productivity sectors (health & social services, education, public administration). In high productivity sectors (mining, real estate transactions, financial and insurance activities), productivity has fallen in recent years while wages have continued to increase (Figure 33). Although growth in wages has contributed to poverty reduction and shared prosperity, the growing gap between labor productivity and wages raises concerns over competitiveness and sustainability. 41 Azevedo, Joao Pedro, Sarosh Sattar, and Judy Yang. (2015). “Kazakhstan Economic Mobility and the Middle Class (2006–2013).” Manuscript in progress. 42 The extreme poverty line is $1.25/day PPP per person. 43 World Bank’s cross-country poverty lines for ECA countries include PPP-corrected US$5 per capita per day (ECAPOV). 23 Figure 31 Figure 32 Contribution to poverty reduction Year to year change in real monthly wage index (2003 = 100) 4.0 18% 2.0 16% (difference, percentage points) 0.0 Poverty Rates ($5/day) –2.0 14% –4.0 12% –6.0 10% –8.0 –10.0 8% –12.0 4% –14.0 2% 7 8 9 0 1 2 3 00 00 00 01 01 01 01 –2 –2 –2 –2 –2 –2 –2 06 07 08 09 10 11 12 0% 20 20 20 20 20 20 20 Shares of Adults Share Employed Wages –2% 04 05 06 07 08 09 10 11 12 13 Social Assistance Pension Agriculture 20 20 20 20 20 20 20 20 20 20 Other Income Source: Azevedo, Joao Pedro, Sarosh Sattar, and Judy Yang. (2015). “Kazakhstan Source: Azevedo, Joao Pedro, Sarosh Sattar, and Judy Yang. (2015). “Kazakhstan Economic Mobility and the Middle Class (2006–2013).” Manuscript in progress. Economic Mobility and the Middle Class (2006–2013)”. Manuscript in progress. Figure 33 Figure 34 Trends in real wages and labor productivity index, 2004 = 100 Growth in real wages and labor productivity index, 2004–2013 280 20% 260 15% 240 10% 220 5% 200 0% 180 –5% 160 –10% 140 –15% 120 re g g n e es ns se n es ad in rin tio io tu iti en ic tio 100 ry at rv tr u uc tiv ul f ac ar uc t de se il ic ac tr ac s qu ta 80 Ed gr an ns al uf d re e A an ci Co d tr an nc so an d e 60 ra n M an at tio d g su an t in le es in ra 40 in sa lth ist al nd M le Re in ea ho la 20 m H W ia ad nc ic 0 na bl Fi Pu 04 05 06 07 08 09 10 11 12 13 20 20 20 20 20 20 20 20 20 20 Productivity growth, 2004–2008 Wage growth, 2004–2008 Productivity growth, 2008–2013 Wage growth, 2008–2013 Real monthly wages index, 2004 = 100 Real VA per worker index, 2004 = 100 Note: Labor productivity is measured by real value added per worker. Note: VA = value added. Source: Authors’ calculations using Statistical Committee of RK. Source: Authors’ calculations using Statistical Committee of RK. 24 Labor Market Outcomes of the Bottom 40 Percent 44 Two thirds of household incomes in Kazakhstan comes from wages, but the relative importance of other sources varies somewhat between the bottom 40 and top 60 percent of the population.45 Wages, on average, contribute two-thirds to household incomes, while self-employment and agricultural incomes contribute less than 10 percent com- bined. However, self-employment and incomes from agricultural activities, in particular, are more prominent for the bot- tom 40 percent, while pensions are a larger share of income for the top 60 percent. An additional 1.7 percent (top 60) to 2.5 percent (bottom 40) comes from social assistance programs. Other income sources, including financial income, make up less than 10 percent of total household income, even for the wealthiest households (Figure 35). Figure 35 Sources of household income, by B40/T60 status, 2013 100% 80% 60% 40% 20% 0% Overall Bottom 40% Top 60% Wage income Self-employment income Agricultural income Pension income Social assistance income Other income sources Note: The welfare aggregate is household consumption per capita. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Households in the bottom 40 percent are not less likely to get income from various sources, except pensions, but they receive smaller amounts for each type of income, except social assistance (Figure 36). The lack of income does not come from receiving each source of income less often, as bottom 40 households are more likely to receive at least some income from each different source except pensions (Figure 37). However, the amounts received from each source, including wage and especially pension income, are smaller. Wage income of those in the bottom 40 percent was 38 percent less than the national average, whereas top 60 households earned wages that were 16 percent higher than the national average. Pension income is even more divergent, with bottom 40 households receiving 52 percent less than the average, while top 60 households receive 22 percent more, on average. This result is partially driven by demographics, as is shown above. 44 Unless explicitly stated otherwise, the statistics presented in this section are based on the working age (15–64 years old) population from the most recent (2013) Household Budget Survey. 45 As noted earlier, the definition of “bottom 40 percent” (B40) or “top 60 percent” (T60) refers to the distribution of per capita household consumption, in which individuals are ranked by the expenditure of their household, deflated by regional price indices, based on data available in the 2011, 2012 and 2013 Household Budget Survey. See Annex C for methodology. 25 Figure 36 Figure 37 Relative average per capita income by source and B40/T60 status, 2013 Percentage of households with any income by source and B40/T60 status, 2013 Social assistance Social assistance Share of households with any income by source income income Pension income Pension income Agricultural income Agricultural income Self-employment Self-employment income income Wage income Wage income –60% –50% –40% –30% –20% –10% 0% 10% 20% 30% 0% 20% 40% 60% 80% 100% Top 60% Bottom 40% Top 60% Bottom 40% Overall Note: The welfare aggregate is household consumption per capita. Note: The welfare aggregate is household consumption per capita. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Household incomes have been slowly becoming more equal, and this has been driven by a convergence in wage income (Figure 38). Per capita total income for bottom 40 households increased, in relative terms, from 41 percent to 38 percent below the national average, while household income for the top 60 percent fell, in relative terms, from 18 per- cent to 16 percent above the national average. This is consistent with impacts on poverty presented earlier. Figure 38 Trends in relative average total per capita income and per capita wage income by B40/T60 status, 2013 40% 20% 0% 2011 2012 2013 –20% –40% –60% Total per capita income (bottom 40%) Total per capita income (top 60%) Note: The welfare aggregate is household consumption per capita. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Part of the reason behind the lower per capita income levels of bottom 40 households is due to larger household sizes relative to richer households (Table 8 in Annex 3). Average household size is 4.8 people among bottom 40 house- holds, but only 3.1 people amongst the top 60. This implies that whatever income is generated by earners in the household is spread over more people. Moreover, 31 percent of household members are under 15 in bottom 40 households versus only 15 percent in top 60 households. Although bottom 40 households also have more people in their prime wage-earning years, they have fewer retirement-age people. Neither children nor retirees generate labor income for the household directly, but retirees can contribute to household resources through pension benefits. This effect is clearly inequality-increasing in Kazakhstan, as income from pensions is far higher in top 60 than in bottom 40 households. Bottom 40 households also reside in areas where earnings potential is lower. More than half (55 percent) of bot- tom 40 households reside in rural areas, whereas only 32 percent of top 60 households are found in rural areas. Moreover, 26 bottom 40 households are concentrated in South Kazakhstan and East Kazakhstan, and make up the largest shares of the population in South Kazakhstan and North Kazakhstan (Figure 39). However, these are the regions where wage employment provides the least income to households (Figure 40). Conversely, the regions where wage employment provides the most income (Almaty City, Astana City and Mangystau) are also the regions with the fewest bottom 40 households. Figure 39 Share of bottom 40 households in each region, 2013 NORTH KAZAKHSTAN PAV L O D A R KOSTANAI AKMOLA Astana WEST KAZAKHSTAN EAST K A Z A K H S TA N AKTOBE ATYRAU KARAGHANDY ALMATY KYZYLORDA MANGYSTAU ZHAMBYL Almaty SOUTH KAZAKHSTAN Astana City: 20.82 49.70 – 61.23 35.53 – 43.19 21.08 – 30.33 Almaty City: 12.32 43.85 – 49.70 32.41 – 35.53 20.31 – 21.08 43.19 – 43.85 30.33 – 32.41 Note: The welfare aggregate is household consumption per capita. Cities of Astana and Almaty excluded. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Figure 40 Relative average income from wage employment by region, 2013 NORTH KAZAKHSTAN PAV L O D A R KOSTANAI AKMOLA Astana WEST KAZAKHSTAN EAST K A Z A K H S TA N AKTOBE ATYRAU KARAGHANDY ALMATY KYZYLORDA MANGYSTAU ZHAMBYL Almaty SOUTH KAZAKHSTAN Astana City: 144.15 116.83 – 149.70 83.12 – 88.31 59.22 – 73.30 Almaty City: 147.68 97.76 – 116.83 77.66 – 83.12 57.59 – 59.22 88.31 – 97.76 73.30 – 77.66 Note: Cities of Astana and Almaty excluded. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. 27 Employed members of bottom 40 households disproportionately work in low paying sectors (Figure 41). Members of households in the bottom 40 are much more likely to work in the low-paying agriculture sector, and slightly more in the somewhat better paying construction sector. However, members of top 60 households are more likely to be found in high paying sectors such as mining, manufacturing, and finance, insurance and real estate. Part of this effect is related to the geo- graphic localization of bottom 40 households: as there are relatively few of them in Almaty City, Astana City and Mangystau, where the high paying jobs are found, it is normal that they are less likely to work in the high paying sectors. Figure 41 Sector of activity by B40/T60 status, 2013 Other services Health and social services Education Public administration services Finance, insurance and real estate Transportation and communication Commerce Construction Public utilities Manufacturing Mining Agriculture 0% 2% 4% 6% 8% 10% 12% 14% 16% Top 60% Bottom 40% Note: The welfare aggregate is household consumption per capita. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. This difference in sectors can be related to a skills gap between top 60 and bottom 40 households, which is likely to grow in the future. Whereas 33 percent of household members in top 60 households have a tertiary educa- tion, the figure for bottom 40 households is only 20 percent. Conversely, bottom 40 households have a higher share of low skilled individuals, at the upper secondary (63 versus 56 percent), lower secondary (14 versus 9 percent) and primary or below levels (4 percent versus 2 percent). As higher skills bring more earnings capacity and better wages, the lack of skills may be contributing to keeping bottom 40 households poor. Although tertiary education has been more prevalent in top 60 households than in bottom 40 households across generations, the gap has widened in more recent generations (Figure 42). Among 60–64 year olds, the difference was 11 percentage points, while among the most recent school-leaving cohort (25–29  year olds), the difference is 19 percentage points.46 If the skills gap continues to widen, recent trends toward reduced inequality could reverse, with the gap in income and expenditure between bottom 40 and top 60 house- holds potentially growing. The combination of skills, household size and geography pushes individuals from bottom 40 households into different types of jobs than those occupied by individuals from top 60 households (Table 8 and Table 9 in Annex C). People from bottom 40 households are more likely to be out of the labor market or unemployed, and less likely to be found in wage employment. The difference in activity rates is driven primarily by the larger share of individuals in schooling and training among the bottom 40 percent. However, among active individuals, those from top 60 households are much more likely to be wage employees, while those in bottom 40 households are more often found in farmstead work and self- employment, or not employed at all (Figure 43). 46 Conversely, upper secondary education accounted for 71 percent of people in the oldest generation of bottom 40 households (relative to 65 percent in top 60 households); the gap has widened in recent generations with 61 percent of poor 25–29 year olds having an upper secondary education versus only 45 percent of top 60 25–29 year olds. 28 Figure 42 Figure 43 Educational attainment by age and B40/T60 status, 2013 Labor force status by B40/T60 status, 2013 100% Private wage employee 90% Public wage employee 80% Non-agricultural self-employed 70% 60% Non-agricultural employer 50% Farmstead worker 40% Unemployed 30% 20% In school or training 10% NEET 0% 0% 5% 10% 15% 20% 25% 30% 35% 40% 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 Age group Top 60% Bottom 40% Note: The welfare aggregate is household consumption per capita. Upper secondary (top 60) Tertiary (top 60) Upper secondary (bottom 40) Tertiary (bottom 40) Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Note: The welfare aggregate is household consumption per capita. Source: Authors’ calculations based on the Kazakhstan Household Budget Survey. Not only do individuals in bottom 40 and top 60 households differ in terms of their characteristics, they also likely face a different set of constraints affecting their labor market outcomes (Table 10 in Annex C). A detailed statistical analysis47 of what makes people more likely to be wage employees relative to self-employment, farmstead workers or out of the labor force (for example) shows that most variables work in the same direction (e.g. women are more likely to be out of the labor force for reasons other than education or training and less likely to be in private wage employment for both bottom 40 and top 60 individuals).48 However, some factors play a larger role in the bottom 40 population than in the top 60 population (for instance, location of residence). This suggests that there may be differences in unobservable characteristics and constraints facing individuals from the bottom 40 and top 60 households, that cannot be captured with the existing data. Individuals from bottom 40 households typically earn less than those from top 60 households from wage employ- ment and non-agricultural self-employment (Figure 68 in Annex C). For both upper secondary and tertiary educated people, a larger share of members of bottom 40 households earn low wages and low income from self-employment than among members of top 60 households, and conversely a larger share of members of top 60 households earn high wages and high levels of income from self-employment. Besides education levels, other observable characteristics cannot explain the differences in wages between those in the bottom 40 and top 60 percent (Figure 69 in Annex C). After controlling for a large set of observable characteristics49 and the fact that individuals in wage employment differ from others along observable and unobserved dimensions, much of the difference between bottom 40 and top 60 individuals disappears, especially for those with upper secondary education. However, it appears that bottom 40 wage workers with tertiary education are more likely to be found at high incomes than would be expected given their education, gender, age, region, urban/rural residency and sec- tor of activity, while top 60 individuals earn slightly less than would have been expected, in particular at the top end of the wage distribution. The fact that such large unexplained variation remains further suggests a need for better data.50 47 The results of this analysis are presented in Table 10 in Annex C. The analysis of the determinants of job type was undertaken using a series of multinomial logit models. For more details, please refer to Annex C. 48 Note that the magnitude of coefficients presented based on HBS differs from LFS. LFS-based estimates presented earlier are likely to be more representative of the labor force. Hence, HBS-based estimates should be taken as indicative. 49 Including gender, 5-year age groups, education, region, urban residency and public/private sector employer. 50 Household Budget Survey is the only household survey which has information on both labor market status and wages and other incomes, as well as consumption of households. However, the set of labor market and demographic variables is a lot more limited compared to LFS. 29 5. Challenges Ahead and Policy Implications In summary, despite a good labor market performance in the last decade, Kazakhstan continues to face several jobs and skills challenges. During the past decade, Kazakhstan benefited from the global commodities boom to become one of the top 10 fastest growing economies in the world. This sustained growth has enabled Kazakhstan to achieve rapid reductions in poverty rates, resulting from strong labor market performance: labor force participation and employment rates are high in Kazakhstan by international standards, especially for women; unemployment halved between 2001 and 2013, from 10.4 to 5.2 percent; and real wages doubled between 2003 and 2013. Despite these gains, Kazakhstan faces a num- ber of jobs challenges, which are related to some of the challenges resource rich economies face: • Many people continue to be employed in low productivity jobs, particularly, in agriculture. Despite a significant reduction in the share of employment in agriculture, there are still about 2 million people, or a quarter of all employed, who remain engaged in this sector. Furthermore, many of the other sectors that increased their share of employment, such as construction and education, also have below average productivity. As a result, overall labor productivity remains low, especially in the non-oil sectors, and compared to countries with similar GDP per capita levels.51 Despite efforts to increase import substitution in higher value added sectors, Kazakhstan’s tradable sectors have remained less competitive, in part due to the exchange rate which was boosted by the resource exports during the period of high oil prices.52 • There are significant inequalities when it comes to the types of jobs and earnings workers have access to across population groups. Men and more educated people tend to have better labor market outcomes, including more access to formal wage employment and higher wages. There are large differences in labor market outcomes and earnings between those in the bottom 40 percent compared to those in the top 60 percent, which cannot always be explained by observable characteristics, suggesting that there may be unobservable barriers that people in the bottom 40 percent face on the labor market. Significant wage differentials across regions and sectors, even after controlling for worker attributes, point to constraints to labor mobility, which appear to play a role in inhibiting some individuals from moving to access higher paid jobs. • Education is a key determinant of labor market outcomes, but the gap in attainment between those in the bottom 40 percent and those in the top 60 percent is increasing, and low education quality may constrain labor productivity growth. Educational attainment is strongly linked with labor market outcomes and largely deter- mines access to wage employment and higher earnings. Although tertiary education has become more prevalent across generations, the gap between the top 60 households and bottom 40 households has widened in more recent genera- tions. Continued widening of the skills gap could result in a reversal of the recent trend toward reduced inequality. In addition, the low quality of education may hamper individuals’ prospects on the labor market as well constrain labor productivity growth. Despite high levels of enrollment and completion of secondary education, Kazakhstan does not provide a quality education, as reflected in the poor, though markedly improved, performance on international student assessments, such as the OECD’s PISA. 51 World Bank. 2013. Beyond Oil : Kazakhstan’s Path to Greater Prosperity through Diversifying, Volume 2. Main report. Washington, DC. 52 Under the pressures of low oil prices, the tenge has depreciated significantly in 2015, which could aid the price competitiveness of the economy. 30 Furthermore, the current economic slowdown is likely to increase unemployment and reduce real earnings in the short to medium term, with some groups being more affected than others. Kazakhstan’s GDP growth has slowed significantly due to a terms-of-trade (oil price) shock as well as the slowdown in Russia and the rebalancing of the China’s economy.53 Simulations show that with lower projected GDP growth in the next few years, unemployment may increase, particularly among older workers and the high skilled. This is despite the expected slowing down of labor force growth in the next few years.54 Removing constraints at the macroeconomic level might not be enough to address Kazakhstan’s jobs challenge, particularly for those in the bottom 40 percent of the population. Part of the solution to the jobs challenge may be in economy-wide policies that promote investments, innovation and economic diversification. However, the relationship between overall investment and employment is complex, not always resulting in the types of jobs that may benefit the bottom 40 percent. Reallocating a given amount of capital to a given sector can have very different consequences in terms of the number and composition of jobs created. Existing government programs such as the State Program for Industrial and Innovative Development (SPIID) for 2015–2019 and economic stimulus program “Nurly Zhol” are aimed at economic diversification and supporting the economy, respectively. However, these programs tend to favor capital intensive sectors or generate temporary jobs.55 Economy-wide reforms need to be complemented with sectoral and regional approaches. In addition to the reforms that target improvements in macro-economic management, investment climate, as well as business environment, it is impor- tant to focus on targeted sectoral policies that aim to promote jobs for specific population groups in given regions. This bottom-up approach implies that the government, in close consultation with social partners, also identifies sectoral policies for employment creation, such as targeted interventions to create jobs in specific regions/sectors. This entails mapping key sub-sectors and value chains within the economy to understand the potential for job creation and the types of bottlenecks and regulatory failures that would need to be removed to achieve it. This mapping would provide information about the types and level of investments that are necessary, the quantity of jobs that can be created, their composition in terms of skills, and their regional distribution. Going forward, addressing the jobs challenge would involve consideration of policies and programs at three levels. These are further elaborated in the Jobs Strategy for Kazakhstan,56 which will help to enhance the impact of the Government’s policies, programs, and projects on the availability, diversity, quality, and sustainability of jobs. 1. Facilitating the creation of new jobs through private sector investments, taking into account regional and population disparities in terms of labor market outcomes. This involves interventions to remove constraints to the creation and expansion of businesses—the main sources of jobs. 2. Upgrading production technologies and increasing the productivity of jobs in economic activities that are already underway, with a focus on lagging regions. To improve living standards, programs and policies to improve the pro- ductivity of existing jobs would be needed, with a focus on the poor. These could include programs that support self- employment and small-scale entrepreneurship, aiming to improve earnings, or policies that help upgrade workers’ skills and enhance their productivity. 3. Connecting individuals to jobs by facilitating labor market transitions; from inactivity or unemployment into jobs, or from low to high productivity jobs. This includes addressing constraints to mobility and reforming and expanding active labor market programs (counseling, intermediation, job search assistance, support to self-employment, and training). These programs are important to address information problems in the labor market and address skills mismatches. 53 Kazakhstan’s GDP growth slowed to 4.1 percent in 2014 and to 1.2 percent in 2015, while prospects for 2016 growth are 0.1 percent. Source: World Bank (2016). Kazakhstan First Fiscal Management and Resilience Programmatic Development Policy Financing. Program information document. 54 See Mohamed Ali Marouani, Bjorn Nilsson, Angela Elzir, Victoria Strokova and Namita Datta (2015) “Kazakhstan Jobs Impacts of Sectoral Investments: Simulations using Dynamic Computable General Equilibrium Model Applied to Kazakhstan”. A note prepared for Kazakhstan: Jobs: Sector Specific Analysis JERP (P153621). 55 These effects operate through two main channels: i) a substitution between capital and labor (as more capital reduces its rate of return, capital becomes cheaper relative to labor); and ii) the reduction of investments in other sectors, which reduces intermediate consumption from the sector in question. 56 World Bank (2016) Kazakhstan: Towards Development of a Jobs Strategy. 31 Annex A: Additional Tables and Graphs Table A1 Key Labor market indicators, Kazakhstan, 2013 Overall Male Female Urban Rural All Adults (15+)   Total Population 12,534,701 5,917,084 6,617,617 7,079,761 5,454,940   Labor Force Participation Rate 71.76% 77.32% 66.79% 69.39% 74.85%  Employment/Population 68.04% 73.81% 62.88% 65.62% 71.18% Working Age (15–64)   Total Population 11,387,615 5,494,509 5,893,106 6,376,155 5,011,460  Employment 8,480,801 4,340,221 4,140,580 4,626,300 3,854,501  Unemployment 466,939 207,858 259,081 266,814 200,125  Inactive 2,439,875 946,430 1,493,445 1,483,041 956,834  NEET 1,075,294 302,224 773,070 669,513 405,781   Employment Rate 74.47% 78.99% 70.26% 72.56% 76.91%   Unemployment Rate 5.22% 4.57% 5.89% 5.45% 4.94%   Percentage NEET 9.44% 5.50% 13.12% 10.50% 8.10%   Labor Force Participation Rate 78.57% 82.77% 74.66% 76.74% 80.91% Working Age Employed   Wage Employment 69.60% 68.85% 70.38% 82.09% 54.60%  Employers 1.87% 2.57% 1.13% 1.25% 2.62%   Own Account Workers 17.95% 19.43% 16.40% 15.16% 21.30%   Farmstead Workers 10.19% 8.77% 11.68% 14.08% 20.72% Source: 2013 LFS. 32 Figure A1 Employment rates (15+), male and female, 2013 Male Female Mexico Kazakhstan 60% Kazakhstan 70% China China Azerbaijan Kyrgyzstan Canada Malaysia Australia Chile Russian Federation Turkmenistan Tajikistan Tajikistan United States Australia Estonia Uzbekistan Kyrgyzstan Russian Federation Lithuania Azerbaijan Ukraine Canada Latvia Georgia Georgia United States Belarus Estonia Chile Armenia Armenia Ukraine Malaysia Latvia Uzbekistan Belarus Mexico Lithuania Turkmenistan Moldova Moldova 0% 10% 20% 30% 40% 50% 60% 70% 80%       0% 10% 20% 30% 40% 50% 60% 70% 80% Source: ILO. 33 Table A2 Marginal effects for multinomial logit model, 2013 Private Public Wage Formal Wage Non-Ag Self- Ag Self- NEET Unemployed Employment Employment Employment Employment Male –10.35 *** –0.75 ** –4.81 *** 11.44 *** 3.75 *** –0.09 *** 15–19 Years Old 15.22 *** –4.06 *** –14.86 *** –3.22 4.70 ** 1.47 *** 20–24 Years Old 1.68 * –2.70 *** –4.42 *** 2.82 ** 1.78 * 0.18 *** 30–34 Years Old –1.11 –0.13 1.14 –0.78 1.73 * –0.09 35–39 Years Old –1.05 –1.67 *** 5.52 *** –1.24 –1.02 –0.04 40–44 Years Old –2.89 *** –2.20 *** 8.78 *** –3.21 *** 0.41 –0.08 45–49 Years Old –0.16 –2.64 *** 8.33 *** –4.47 *** 0.59 0.00 50–54 Years Old 1.83 ** –2.40 *** 6.69 *** –3.69 *** –0.81 –0.08 55–59 Years Old 18.77 *** –2.73 *** 1.50 –11.56 *** –3.65 *** –0.04 60–64 Years Old 69.99 *** –5.46 *** –20.39 *** –27.89 *** –11.01 *** –0.35 *** Upper Secondary Education –17.70 *** –4.29 *** 9.93 *** 14.44 *** –2.25 –0.23 ** Tertiary Education –15.66 *** –5.71 *** 31.01 *** 4.62 –7.60 *** –0.78 *** Aktobe 3.13 * –1.48 ** –14.03 *** 20.03 *** –3.19 ** –0.52 *** Almaty 13.63 *** –1.01 –1.96 –7.46 *** –0.27 –0.36 *** Atyrau 3.60 ** –1.92 *** –15.23 *** 28.08 *** –9.40 *** –0.82 *** West Kazakhstan 10.67 *** –0.68 –3.97 ** –0.25 –0.84 –0.21 *** Jambyl 10.95 *** –0.35 6.12 *** –19.67 *** 6.43 *** –0.03 Karaganda 6.83 *** –1.31 ** –9.39 *** 16.43 *** –9.90 *** –0.59 *** Kostanay –1.42 –0.05 –11.50 *** 19.99 *** –3.75 *** 0.17 * Kyzylorda 15.54 *** –1.85 *** 7.12 *** –18.86 *** 3.08 * –0.73 *** Mangystau 15.09 *** –2.00 *** 16.55 *** –11.68 *** –11.77 *** –0.94 *** South Kazakhstan 15.66 *** –0.05 –4.82 *** –18.90 *** 7.08 *** –0.25 *** Pavlodar 3.24 ** –1.11 * –11.27 *** 24.03 *** –12.43 *** –0.19 *** North Kazakhstan 3.74 ** –0.41 1.11 8.46 *** –10.44 *** –0.08 East Kazakhstan 11.62 *** –1.21 * –17.08 *** 15.57 *** –5.35 *** –0.20 *** Astana City 25.03 *** –1.63 ** –10.71 *** 4.72 * –12.35 *** –0.94 *** Almaty City 9.06 *** 0.40 –6.98 *** 15.67 *** –8.71 *** –8.66 *** Urban 0.89 ** –0.96 *** –0.76 4.88 *** 0.60 –2.26 *** 2 Person Household –2.29 *** –1.27 *** 2.08 ** 2.82 *** –1.17 * 0.03 3 Person Household –2.04 *** –1.54 *** 2.41 *** 2.63 *** –0.07 –0.08 4 Person Household –2.62 *** –1.47 *** 3.26 *** 1.86 –0.29 –0.03 5 Person Household –0.86 –1.95 *** 3.13 ** –1.32 1.93 * 0.07 6 Person Household 0.91 –0.23 0.65 –2.35 1.22 0.16 7 Person Household –0.76 –3.79 *** –2.65 6.50 –0.17 0.25 8 Person Household –2.72 –2.38 –5.52 7.61 1.47 0.65 9 Person Household –6.85 *** –5.99 *** 7.95 11.33 –2.57 0.21 10+ Person Household –0.48 2.08 –15.94 ** –19.06 *** 9.53 1.50 Interpretation: A male has a 10.35 percentage point lower chance of being NEET (not in employment, education or training) than a female when both have the average characteristics of the working age population. Note: Multinomial logits estimated separately on the following outcomes: NEET, in school or training, unemployment, agricultural self-employed, non- agricultural employer, non-agricultural self-employed, public wage employment and private wage employment. Reference categories were female, 25–29 years old, lower secondary or less, rural resident and one-person household. The models also include controls for region of residence. *** signifies a marginal effect that is significant at the 1 percent level, ** signifies a marginal effect that is significant at the 5 percent level and * signifies a marginal effect that is significant at the 10 percent level. 34 Figure A2 Figure A3 Age by labor force status, shares, 2013 Age by type of employment, shares, 2013 100% 100% 90% 90% 80% 80% 70% 70% 60% 60% 50% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% p. ed ed ol ET p. t ed r r en ke ke po ho po NE oy oy oy or or m sc pl pl pl e e oy tw w ag ag Em em In Em pl d un ea g g em Un in in co st k k rm e ac or or ag W Fa W n W w O 15–24 25–34 35–44 45–54 55–64 15–24 25–34 35–44 45–54 55–64 Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. Figure A4 Figure A5 Age by labor force status, number of people, 2013 Age by type of employment, number of people, 2013 9,000 7,000 8,000 6,000 7,000 5,000 6,000 Thousands Thousands 5,000 4,000 4,000 3,000 3,000 2,000 2,000 1,000 1,000 0 0 ed ed ol ET t er r r en ke ke ho oy NE oy oy or or m pl sc pl pl oy tw w Em Em em In pl d un ea em Un co st rm e ac ag Fa n W w 15–24 25–34 35–44 45–54 55–64 O Source: Authors’ calculations using LFS 2013. 15–24 25–34 35–44 45–54 55–64 Source: Authors’ calculations using LFS 2013. 35 Figure A6 Figure A7 Job type by age for employed population, in %, 2013 Job type by age for employed population, number of employed, 2013 100% 3,000 2,500 80% 2,000 Thousands 60% 1,500 40% 1,000 20% 500 0% – 15–24 25–34 35–44 45–54 55–64 15–24 25–34 35–44 45–54 55–64 Wage employment Employer Wage employment Employer Own account worker Farmstead worker Own account worker Farmstead worker Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. Figure A8 Figure A9 Distribution of self-employed by age group, 2013 Distribution of self-employed by education, 2013 250 100% 600 100% 90% 90% 500 200 80% 80% 70% 400 70% Thousands Thousands 150 60% 60% 50% 300 50% 100 40% 40% 200 30% 30% 50 20% 20% 100 10% 10% – 0% – 0% 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 s ry l l) ry ry na ar ia da tia ia ye tio ec rt on r sp te te 12 ca ec l( vo e ed n ls et na ha et al ra pl io t pl Total (left axis) iti Percentage (right axis) m ne at ss m In co Le Ge c Co Vo In Source: Authors’ calculations using LFS 2013. Total (left axis) Percentage (right axis) Source: Authors’ calculations using LFS 2013. 36 Figure A10 Figure A11 Distribution of self-employed by household size and gender, 2013 Distribution of self-employed by relation to head of household and gender, 2013 60% 600 500 50% 400 40% 300 30% 200 20% 100 10% 0 Head Spouse Children Parents Other or not relate 0% n ns ns ns ns 5+ e e Male Female al al so so so so so M m en r pe r r r r Fe pe pe pe pe th 1 Source: Authors’ calculations using LFS 2013. 2 3 4 5 e or M Source: Authors’ calculations using LFS 2013. Figure A12 Figure A13 Distribution of farmstead workers by age group, 2013 Distribution of farmstead workers by education, 2013 140 100% 500 100% 90% 90% 120 400 80% 80% 70% Thousands 100 70% 300 60% Thousands 60% 80 50% 50% 200 40% 60 40% 30% 100 20% 40 30% 10% 20% 20 0 0% 10% rs ry l l) ry ry na cia a da ia ia ye tio rt rt e 0% on – sp te te 12 ca ec l( vo e ed 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 n ls et na a et al th ra pl io pl iti m ne at ss m In co Le Ge c Co Vo In Total (left axis) Percentage (right axis) Source: Authors’ calculations using LFS 2013. Total (left axis) Percentage (right axis) Source: Authors’ calculations using LFS 2013. 37 Figure A14 Figure A15 Distribution of farmstead workers by household size and gender, 2013 Distribution of farmstead workers by relation to head of household and gender, 2013 60% 300 50% 250 40% 200 Thousands 150 30% 100 20% 50 10% 0 Head Spouse Children Parents Other or 0% not related n ns ns ns ns 5+ e e al al so so so so so M m en r Male Female pe r r r r Fe pe pe pe pe th 1 2 3 4 5 e or Source: Authors’ calculations using LFS 2013. M Source: Authors’ calculations using LFS 2013. Figure A16 Figure A17 Distribution of wage employees by age group, 2013 Distribution of wage employees by education, 2013 1,200 100% 2,500 100% 90% 90% 1,000 2,000 80% 80% 70% Thousands 70% 800 1,500 60% 60% 50% Thousands 600 50% 1,000 40% 40% 30% 400 500 20% 30% 10% 20% 200 – 0% 10% s y l l) ry ry na ar ar cia ia tia ye nd tio rt e r – 0% sp te te 12 co ca l( se vo e ed 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 55–59 60–64 n et na ha et l al ra pl io st pl iti m ne at m s In co Le Ge c Co Vo In Total (left axis) Percentage (right axis) Total (left axis) Percentage (right axis) Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. 38 Figure A18 Figure A19 Distribution of wage employees by household size and gender, 2013 Distribution of wage employees by relation to head of household and gender, 2013 60% 3,000 50% 2,500 40% 2,000 Thousands 30% 1,500 1,000 20% 500 10% 0 Head Spouse Children Parents Other or 0% not related n ns ns ns ns 5+ e e al al so so so so so M m en r pe r r r r Male Female Fe pe pe pe pe th 1 2 3 4 5 e or M Source: Authors’ calculations using LFS 2013. Source: Authors’ calculations using LFS 2013. Figure A20 Figure A21 Labor force status by education level, percentage, 2013 Labor force status by education level, total number, 2013 100% 3,500 3,000 80% 2,500 Thousands 2,000 60% 1,500 40% 1,000 500 20% – s y l l) ry ry na ar ar cia tia ia ye nd io rt 0% e r at sp te te 12 co c l( se vo e ed n et na s y l l) ry ry na ha ar ar et l ia al ra pl ia ia io ye nd tio ec t pl iti rt rt m ne at ss m sp te te In co 12 co ca Le Ge c Co Vo l( In se vo e ed n et na a et al al th pl tio er pl iti m ss n m ca In co Le Ge Co Vo In Employed Unemployed In school NEET Source: Authors’ calculations using LFS 2013. Employed Unemployed In school NEET Source: Authors’ calculations using LFS 2013. 39 Figure A22 Education attained by type of employment, 2013 100% 80% 60% 40% 20% 0% Wage Employer Own account Farmstead employment worker worker Less than 12 years General secondary Initial vocational Vocational (special) Incomplete tertiary Completed tertiary Source: Authors’ calculations using LFS 2013. Figure A23 Employment by sector and education, 2013 100% Less than 12 years 80% Incomplete tertiary General secondary 60% Completed tertiary 40% Vocational (special) 20% Initial vocational 0% re s g s n ce n te s n rie tie ice rin tio tio tio tu ta er st ili rv tu uc ica es ul m ra du ut se ric m ac tr al ist un in ns ic Co er uf Ag re in m bl Co th e an m d m tiv Pu O an ad M co ac ce ic d tr an bl an ex Pu ur n d io an ns at ,i g rt ce in o an sp in M an n Fi Tr Source: Authors’ calculations using LFS 2013. 40 Figure A24 Comparison of PISA 2012 scores for select countries and ECA/OECD averages, 2013 520 One year of schooling 500 480 PISA 2012 Score 460 432 425 440 420 393 400 380 360 Kazakhstan Malaysia Chile ECA Russia Turkey OECD Malaysia Chile Kazakhstan Turkey ECA Russia OECD Malaysia Kazakhstan Chile Turkey ECA Russia OECD Reading Math Science Source: OECD, PISA 2012 Results (2012); WB, Strengthening Kazakhstan’s Education System (2014). 41 Annex B: Methodology and Additional Results on Informality Methodology Most of the statistics presented are weighted sample statistics. The weights used were drawn directly from the vari- ous years’ labor force surveys, and the years 2010 to 2013 were used. The statistical analysis was based on a probit model with selection bias correction. The main equation explained the probability of formal wage employment (private or public) relative to informal wage employment. Since the set of individuals actually occupying a wage job (as opposed to self-employed or out of the labor force, for example) is not a random sample of the population, one has to correct the formal wage employment model estimates for the probability for being observed in wage employment. This is done through a (probit) selection equation for selection into wage employment. The disturbance terms between the selection equation and the formal wage employment equation can be correlated, and the system is esti- mated by maximum likelihood. Table B1 Characteristics of wage workers by formality status All Wage Public Sector Private Formal Private Informal Workers Wage Workers Wage Workers Wage Workers Number of Workers 5,899,592 2,495,590 2,916,665 487,337 Share of All Wage Workers 100.0% 42.3% 49.4% 8.3% Percentage Male 50.6% 43.6% 56.2% 53.7% Average Age 38.0 38.8 37.6 36.7 Percentage with Lower Secondary or Below Education 0.8% 0.5% 0.8% 1.8% Percentage with Upper Secondary Education 58.7% 47.0% 64.3% 84.7% Percentage with Tertiary Education 40.5% 52.5% 34.9% 13.5% Share Living in Urban Areas 64.4% 61.7% 69.6% 46.4% Average Household Size 2.00 2.01 1.98 2.05 Contract and Social Security 100% 98% 0% Contract Only 0% 2% 38% Social Security Only 0% 0% 6% No Contract or Social Security 0% 0% 56% Source: Authors’ calculations using LFS 2013. 42 Table B2 Marginal effects for probit model of informality with selection bias correction, 2013 Marginal Conditional Marginal Conditional Effect Marginal Effect Effect Marginal Effect Male 0.44 ** 0.49 *** Urban –0.51 –0.28 Age 15–19 –2.60 *** –2.86 *** Aktobe –2.56 *** –2.33 *** Age 20–24 –0.85 ** –1.02 *** Almaty –2.37 *** –2.16 *** Age 30–34 –0.28 –0.24 Atyrau –2.55 *** –2.30 *** Age 35–39 0.19 0.23 West Kazakhstan –2.78 *** –2.55 *** Age 40–44 –0.18 –0.10 Jambyl –1.97 *** –1.84 *** Age 45–49 –0.73 ** –0.63 ** Karaganda –1.30 *** –1.10 *** Age 50–54 –0.60 * –0.51 * Kostanay –2.71 *** –2.48 *** Age 55–59 –0.59 * –0.63 ** Kyzylorda –2.06 *** –1.88 *** Age 60–64 –2.21 *** –2.35 *** Mangystau –2.87 *** –2.62 *** Upper Secondary Education 1.25 1.45 ** South Kazakhstan 1.51 * 1.05 * Tertiary Education –1.62 –1.07 Pavlodar –1.78 *** –1.58 *** Mining and Extractive Industries –10.42 *** –9.29 *** North Kazakhstan –2.01 *** –1.81 *** Manufacturing –5.52 *** –4.92 *** East Kazakhstan –2.27 *** –2.07 *** Public Utilities –8.43 *** –7.52 *** Astana City –2.59 *** –2.35 *** Construction –1.16 ** –1.04 *** Almaty City –2.08 *** –1.87 *** Commerce 1.83 *** 1.63 *** Transportation and Communication –3.83 *** –3.41 *** Financial, Insurance and Real Estate –4.21 *** –3.75 *** Correlation Public Administration Services –8.69 *** –7.75 *** Coefficient Education –9.59 *** –8.55 *** (Selection Effect) = 0.554 *** Health and Social Services –10.66 *** –9.51 *** Other Services 0.01 0.01 Interpretation: Men have a 0.44 percentage point higher probability of being informal wage workers (as opposed to formal wage workers) than women when in wage employment. Accounting for the fact that men are also more likely to be in wage employment and that those with a higher probability of being in wage employment also have a higher probability of being informal (positive significant correlation coefficient), men have a 0.49 percentage point higher chance than women to be seen in informal wage employment overall. Notes: Reference categories include: female, age 25–29, lower secondary education and below, agriculture sector, rural and Akmola region. Exclusion restrictions for the selection equation included a set of 10 indicators for household size. *** signifies a marginal effect that is significant at the 1 percent level, ** signifies a marginal effect that is significant at the 5 percent level and * signifies a marginal effect that is significant at the 10 percent level. 43 Annex C: Methodology of the Bottom 40 percent Analysis and Results Methodology The descriptive analysis undertaken is based on weighted descriptive statistics. As the official sample weights were not provided, the World Bank poverty group designed a weighting scheme for the HBS that is representative of the number of households in each urban versus rural area in each region. There is, however, no calibration to a population census used in establishing sample weights, and thus the representativity of the data for particular population groups is not guaranteed. The analysis of the determinants of job type was undertaken using a series of multinomial logit models. This approach assumes that there is an unobservable (latent) value to each possible outcome, and the outcome with the highest value is the one observed in the data. This value need not reflect exclusively preference parameters of individuals; it can also incorporate matching processes on the labor market and firm objectives. This approach implicitly assumes “independence of irrelevant alternatives,” in that factors that change the value of one alternative do not affect the relative rankings of other alter- natives among themselves. The multinomial logit approach assumes that each latent value is a linear function of the observable characteristics in the model and a random variable that follows a standard type-1 extreme value distribution. The table presents marginal effects estimated from the multinomial logit models. Marginal effects describe how much the probability of a particular outcome changes when the variable in question changes by one unit (or goes from 0 to 1 in the case of indicator variables). As multinomial logit models are highly nonlinear, the estimated coefficients cannot be directly interpreted in terms of the probability of choosing a particular outcome. Moreover, the estimated marginal effects will typically depend on the reference values for the explanatory variables at which the calculations are made. This paper sets the covariates to the population average (overall, bottom 40 or top 60, depending on the model) and calculates the change in the probability of each outcome numerically. Kernel density estimates of different income sources are also presented. Kernel density estimators are similar to his- tograms, in that they count the number of observations with a given value of variable in question (for example, log income from wage employment). They differ, however, in that they use a weighted count that includes the number of observations with values near the one being evaluated, with the weighting function (kernel) and set of points over which the weighted count is taken (related to the bandwidth) being chosen by the analyst. The figures use an Epanechnikov kernel, a bandwidth of 0.04 for wage income and a bandwidth of 0.1 for self-employment and agricultural income. The kernel density estimates of residuals from a log wage regression were constructed using a Heckman (1979) selection-bias corrected wage regression. This model regressed, separately for 10 different sectors,57 log wage income on gender, 5-year age groups, education, region, urban residency and public/private sector employer. The selection equation 57 These effects operate through two main channels: i) a substitution between capital and labor (more easily available capital reduces its cost, making it cheaper relative to the price of labor); and ii) a reduction of investments in other sectors, which reduces intermediate consumption of goods and services produced by the sector receiving the injection of capital from the sectors from which the capital is taken.   These effects operate through two main channels: i) a substitution between capital and labor (more easily available capital reduces its cost, making it cheaper relative to the price of labor); and ii) a reduction of investments in other sectors, which reduces intermediate consumption of goods and services produced by the sector receiving the injection of capital from the sectors from which the capital is taken.   The sectors included agriculture, mining, manufacturing, public utilities, construction, commerce, transportation and communication, finance, insurance and real estate, public administration services and other services. 44 also included indicators for household size, but did not include the public/private indicator. Unconditional expected log wage income was subtracted from observed log wage income, and kernel density estimators of this residual were calculated with an Epanechnikov kernel using a bandwidth of 0.1. Figure C1 Earnings distributions by education, income source and bottom 40 classification, 2013 Wage Income—Upper Secondary Education Wage Income—Tertiary Education 100,000 200,000 300,000 400,000 500,000 100,000 200,000 300,000 400,000 500,000 0 0 .000028 .00003 .000032 .000034 .000028 .00003 .000032 .000034 Log (wage income (annual), relative to national average) Log (wage income (annual), relative to national average) Overall Bottom 40 percent Top 60 percent       Overall Bottom 40 percent Top 60 percent Self-Employment Income—Upper Secondary Education Self-Employment Income—Tertiary Education 40,000 40,000 30,000 30,000 20,000 20,000 10,000 10,000 0 0 .00018 .0002 .00022 .00024 .00026 .00028 .00018 .0002 .00022 .00024 .00026 .00028 Log (self-employment income (annual), relative to national average) Log (self-employment income (annual), relative to national average) Overall Bottom 40 percent Top 60 percent       Overall Bottom 40 percent Top 60 percent 45 Figure C2 Wage distributions after controlling for observables by education and bottom 40 classification, 2013 Upper Secondary Education Tertiary Education .01 .01 .008 .008 .006 .006 .004 .004 .002 .002 0 0 –100 –50 0 50 100 –100 –50 0 50 100 Log (unexplained component of log (wage income), Log (unexplained component of log (wage income), relative to national average) relative to national average) Overall Bottom 40 percent Top 60 percent           Overall Bottom 40 percent Top 60 percent 46 Table C1 Household level descriptive statistics 2011 2012 2013 Overall Bottom 40% Top 60% Overall Bottom 40% Top 60% Overall Bottom 40% Top 60% Average Household Head Age 47.7 46.1 48.4 47.9 46.0 48.8 48.0 46.3 48.7 Average Household Size 3.7 4.8 3.1 3.6 4.8 3.1 3.5 4.7 3.0 Children (Under 15) 20.4% 31.5% 15.6% 20.7% 32.6% 15.7% 19.6% 31.4% 14.7% Youth (15–24) 16.0% 17.0% 15.6% 14.5% 15.9% 14.0% 13.9% 15.0% 13.4% Prime Age (25–54) 42.7% 40.4% 43.8% 42.7% 40.0% 43.9% 43.5% 40.6% 44.7% Older (55–64) 12.4% 6.8% 14.8% 13.3% 6.8% 16.0% 13.5% 7.8% 15.9% Retirees (65 and Above) 8.5% 4.3% 10.3% 8.8% 4.7% 10.5% 9.5% 5.2% 11.3% Primary Education or Less 3.1% 4.2% 2.7% 3.0% 4.2% 2.5% 2.5% 3.6% 2.0% Lower Secondary 13.1% 17.0% 11.4% 11.6% 14.5% 10.4% 10.3% 13.8% 8.8% Upper Secondary 56.3% 62.6% 53.6% 57.1% 63.0% 54.7% 58.1% 62.6% 56.3% Tertiary 27.4% 16.2% 32.3% 28.2% 18.3% 32.4% 29.1% 20.0% 32.9% Urban 61.1% 45.2% 68.0% 61.3% 47.9% 66.9% 61.2% 47.8% 66.8% Akmola 4.9% 5.4% 4.7% 4.8% 5.3% 4.6% 4.8% 6.3% 4.2% Aktobe 4.6% 4.1% 4.8% 4.4% 3.9% 4.7% 4.2% 3.8% 4.4% Almaty 10.2% 5.7% 12.1% 10.4% 7.4% 11.7% 10.8% 6.4% 12.7% Atyrau 2.5% 2.6% 2.5% 2.4% 2.0% 2.6% 2.4% 2.2% 2.5% West Kazakhstan 3.8% 4.4% 3.5% 3.6% 4.6% 3.2% 3.5% 4.4% 3.1% Jambyl 5.7% 8.5% 4.5% 5.6% 9.5% 4.0% 5.8% 8.4% 4.8% Karaganda 9.3% 8.0% 9.8% 9.1% 7.7% 9.8% 8.9% 8.5% 9.1% Kostanay 6.4% 6.7% 6.3% 6.3% 6.6% 6.2% 6.2% 6.8% 5.9% Kyzylorda 3.1% 3.0% 3.1% 3.1% 3.3% 3.0% 3.0% 4.0% 2.6% Mangystau 2.6% 3.4% 2.3% 2.5% 3.1% 2.3% 2.6% 1.5% 3.1% South Kazakhstan 12.2% 20.5% 8.6% 12.2% 21.9% 8.1% 12.4% 22.5% 8.2% Pavlodar 5.3% 6.0% 5.0% 5.2% 5.2% 5.2% 5.4% 3.9% 6.0% North Kazakhstan 4.2% 5.0% 3.8% 4.2% 4.6% 4.0% 4.0% 5.2% 3.4% East Kazakhstan 10.0% 11.3% 9.5% 10.3% 9.1% 10.8% 10.3% 9.7% 10.5% Astana City 4.5% 2.7% 5.3% 4.9% 3.0% 5.8% 4.8% 2.8% 5.6% Almaty City 10.6% 2.6% 14.1% 10.7% 2.8% 14.0% 11.0% 3.7% 14.0% Relative Average per Capita Wage Income –41.0% 17.8% –38.0% 16.0% –38.1% 15.9% Self-Employment Income –17.5% 7.6% –17.0% 7.2% –19.2% 8.0% Agricultural Income –13.0% 5.6% –15.8% 6.6% –16.6% 6.9% Pension Income –56.2% 24.3% –54.4% 22.9% –51.9% 21.7% Social Assistance Income 18.0% –7.8% 8.7% –3.7% 16.7% –7.0% Notes: Per capita measures assign a weight of 1 to each household resident with a familial relation to the household head. Relative average compares the average income from a particular source in the sub-population to the overall mean. For example, average per capita wage income in households in the 40 percent poorest part of the population was 41 percent below the 47 national average, while average per capita wage income in households in the richest 60 percent was 17.8 percent above the national average. Table C2 48 Individual level descriptive statistics 2011 2012 2013 Bottom Bottom Bottom Overall 40% Top 60% Overall 40% Top 60% Overall 40% Top 60% Male 45.9% 47.5% 44.9% 45.7% 47.7% 44.6% 45.7% 47.7% 44.6% Youth (15–24) 25.7% 28.0% 24.4% 23.9% 26.8% 22.3% 22.7% 25.3% 21.2% Prime Age (25–54) 61.3% 62.4% 60.7% 62.2% 63.5% 61.5% 62.9% 64.1% 62.3% Older (55–64) 13.0% 9.6% 15.0% 13.9% 9.7% 16.2% 14.4% 10.6% 16.5% Primary Education or Less 3.1% 4.0% 2.6% 3.3% 4.4% 2.6% 2.8% 3.9% 2.2% Lower Secondary 13.8% 17.7% 11.6% 11.6% 14.2% 10.2% 10.2% 13.2% 8.4% Upper Secondary 57.3% 63.1% 54.0% 58.6% 64.1% 55.6% 59.7% 64.2% 57.1% Tertiary 25.8% 15.2% 31.9% 26.4% 17.3% 31.5% 27.4% 18.7% 32.2% Agriculture 9.3% 15.5% 6.2% 8.5% 12.9% 6.3% 8.4% 12.8% 6.2% Mining 3.6% 3.2% 3.7% 3.6% 3.4% 3.7% 3.6% 3.0% 3.8% Manufacturing 4.5% 3.9% 4.8% 4.2% 3.7% 4.5% 4.4% 3.9% 4.6% Public Utilities 3.9% 3.2% 4.2% 3.9% 3.6% 4.0% 3.6% 3.0% 3.9% Construction 9.0% 10.1% 8.4% 8.8% 10.0% 8.1% 8.8% 10.5% 7.9% Commerce 14.0% 14.2% 14.0% 13.9% 13.8% 13.9% 13.7% 13.7% 13.7% Transportation and Communication 10.8% 9.6% 11.4% 11.5% 10.8% 11.8% 12.2% 11.5% 12.5% Finance, Insurance and Real Estate 3.5% 1.6% 4.4% 3.8% 2.3% 4.5% 4.1% 2.0% 5.1% Public Administration Services 11.5% 9.0% 12.7% 11.1% 9.1% 12.1% 10.6% 9.0% 11.3% Education 13.8% 13.8% 13.8% 13.6% 13.2% 13.7% 14.0% 13.6% 14.2% Health and Social Services 5.9% 5.0% 6.4% 6.4% 5.5% 6.8% 6.1% 5.7% 6.3% Other Services 10.3% 11.0% 10.0% 10.8% 11.6% 10.4% 10.5% 11.2% 10.2% NEET 14.9% 18.1% 13.1% 15.0% 18.1% 13.3% 14.3% 17.0% 12.7% In School or Training 14.8% 16.9% 13.5% 13.9% 16.4% 12.5% 13.2% 15.6% 11.8% Unemployed 3.5% 4.8% 2.8% 2.6% 3.3% 2.1% 2.4% 3.1% 2.0% Farmstead Worker 2.3% 3.5% 1.6% 2.1% 2.6% 1.7% 2.4% 3.4% 1.9% Non-Agricultural Employer 0.1% 0.0% 0.2% 0.2% 0.1% 0.3% 0.2% 0.1% 0.2% Non-Agricultural Self-Employed 7.2% 8.5% 6.4% 6.6% 8.0% 5.8% 6.9% 8.4% 6.1% Public Wage Employee 21.9% 18.3% 24.0% 22.2% 19.0% 24.0% 21.2% 19.0% 22.4% Private Wage Employee 35.3% 29.9% 38.4% 37.5% 32.4% 40.3% 39.5% 33.4% 42.9% Notes: NEET = Not in Employment, unemployment, Education or Training. Table C3 Multinomial logit marginal effects by bottom 40 classification, 2013 Non-Agricultural Public Wage Private Wage NEET Unemployment Self-Employed Employment Employment Bottom 40% Top 60% Bottom 40% Top 60% Bottom 40% Top 60% Bottom 40% Top 60% Bottom 40% Top 60% Male –31.8 –22.0 –0.1 –0.2 4.7 6.4 –9.8 –15.5 37.0 31.3 15–19 Years Old 4.3 10.2 6.5 8.9 –0.2 –3.2 –7.0 –11.4 –3.6 –4.6 20–24 Years Old 0.3 –1.8 4.1 2.5 –0.2 0.2 –0.1 0.7 –4.1 –1.7 30–34 Years Old –2.6 –1.5 –1.9 –0.9 2.5 1.0 5.3 3.6 –3.2 –2.3 35–39 Years Old –7.9 –5.2 –3.1 –1.6 3.1 3.3 9.7 7.4 –1.8 –3.9 40–44 Years Old –10.3 –6.3 –3.3 –2.2 4.0 4.2 11.7 8.6 –2.1 –4.4 45–49 Years Old –8.2 –6.8 –3.2 –1.9 2.1 5.4 15.6 12.4 –6.2 –9.1 50–54 Years Old –5.9 –3.5 –2.9 –1.6 3.4 3.2 13.8 11.0 –8.5 –9.2 55–59 Years Old 8.8 12.0 –2.7 –1.7 0.2 1.1 10.2 4.4 –16.5 –15.7 60–64 Years Old 54.9 62.0 –3.3 –2.1 –3.7 –5.1 –9.0 –14.3 –39.0 –40.6 Upper Secondary Education –9.9 –6.6 –1.9 –0.6 –1.2 –1.4 9.8 9.9 3.2 –1.4 Tertiary Education –13.7 –11.1 –2.9 –1.8 –4.4 –4.5 40.4 35.0 –19.2 –17.5 Urban –4.2 –2.5 0.6 –0.1 –0.9 0.3 –7.8 –12.1 12.4 14.6 3 Person Household –0.2 3.4 –1.1 1.8 –2.5 0.8 0.7 0.2 3.2 –6.2 4 Person Household –0.9 5.6 –1.5 2.0 –1.8 2.1 4.0 0.7 0.2 –10.4 5 Person Household 0.3 8.0 –1.6 2.5 –0.9 2.5 3.2 –0.1 –1.0 –13.0 6 Person Household 1.3 9.7 –1.4 3.6 –0.6 3.7 5.9 0.0 –5.3 –17.0 7 Person Household 5.0 9.7 –1.3 3.3 –1.8 2.5 6.3 2.1 –8.1 –17.6 8 Person Household 4.5 8.3 0.2 7.1 –2.2 2.3 2.1 –2.8 –4.6 –14.9 9 Person Household 12.0 20.4 1.1 1.5 –0.6 0.0 –1.1 –2.2 –11.4 –19.8 10 or More Person Household 9.4 12.5 –0.5 4.0 –1.9 7.9 –0.4 –7.5 –6.6 –16.9 Interpretation: A male from a bottom 40 household had a 31.8 percentage point lower chance of being in neither employment, unemployment, education or training than a female when both have the average characteristics of individuals in bottom 40 households, and a male from a top 60 household had a 22.2 percentage point lower chance than a female with average characteristics of individuals in top 60 households. Note: Multinomial logits estimated separately on each Bottom 40 Classification on the following outcomes: NEET (Not in Employment, unemployment, Education or Training), in school or training, unemployment, farmstead worker, non-agricultural employer, non-agricultural self-employed, public wage employment and private wage employment. Reference categories were female, 25–29 years old, lower secondary or less, rural resident and one and two person households combined. The models also include controls for region of residence. Figures shown are rounded to 1 significant digit; marginal effects corresponding to coefficients that are not statistically significant at the 5% level are shown in light blue. There are no “true zeros” among the estimated marginal effects. 49 Annex D: Methodology and Results of the Earnings Analysis Methodology Estimating the determinants of wages is different from simply calculating average wages for different types of individuals, as some characteristics tend to go together (for example, urban areas tend to concentrate more educated workers). Thus when one sees higher wages in urban areas, it is not clear whether this is due to urban employers paying all workers more, or more educated workers earning more and employers in urban areas paying comparably skilled workers the same amount as rural employers. Regression analysis allows one to statistically control for differences in all of the characteristics simultaneously in order to draw conclusions about how each individual factor affects wages. The analysis here also attempts to address the problem of “selection bias”, using econometric techniques proposed by Heckman (1979). Since the objective is to study the determinants of wages, it is important to consider that some people are more likely to find wage jobs than others. Without special treatment, the estimates of the determinants of wages would be biased because only certain people (the “best”) among those unlikely to get wage jobs would actually be seen in wage employment, and we can only use the wages that we can measure in a regression. The value added of the analysis presented here stems from the focus on the sector-geography type of constraints. The analy- sis undertaken here exploits the labor force survey, which is designed to be representative of the working age population, for studying the structure of employment. However, estimates from the household budget survey are needed to quantify incentives for mobility, as the labor force survey contains no information on earnings. By combining both surveys, one gets a more complete picture of the type of skills that are the most highly rewarded in each region and where the people with those skills actually live. This is necessary to assess the importance of mobility constraints for people of different skill levels. A “spatial” analysis of employment and skills can help guide more nuanced policy making at a regional level. Technical Discussion of Results The wage equations were estimated in logarithmic form following the technique for correction of selection bias proposed by Heckman (1979). In this technique, participation in the estimation sample is modeled using a dichotomous probit model, and an additional term is then introduced into the wage equation that corresponds to the expected residual from this selec- tion equation. When the disturbance term from the wage equation and the disturbance term from the latent model of the selection equation are jointly normally distributed, the expected residual takes on a form known as the “inverse Mills ratio”. Moreover, if only a selected subset of observations are used to estimate the wage model, inclusion of the inverse mills ratio purges the disturbance term of its correlation with the variables in the wage equation, thereby eliminating bias due to selection. Once the coefficients are estimated for each sector, the average share of urban jobs and the average share of public sector jobs in region is calculated, as is the share of men and the share of the population in each age group for the whole country. Using these average values, expected log wages are calculated specific to each sector in each region for each education level by manipulating the set of indicator variables for region and education in each sector’s wage model. This provides expected log wages for all combinations of sector, region and education, and the exponential of the log wages is taken to give expected wages. 50 The available jobs in each region by sector are then estimated using the share of employment in the sector observed in each region. These shares are multiplied by the corresponding expected wages for the sector in the region, for each education level, and then summed across sectors to give a weighted average expected wage, where the weights reflect the observed share of jobs. This is how expected wages for each level of education in each region are calculated. Kernel density estimates of wages by sector are also presented. Kernel density estimators are similar to histograms, in that they count the number of observations with a given value of variable in question (for example, log income from wage employment). They differ, however, in that they use a weighted count that includes the number of observations with values near the one being evaluated, with the weighting function (kernel) and set of points over which the weighted count is taken (related to the bandwidth) being chosen by the analyst. The figures presented use an Epanechnikov kernel and a bandwidth of 0.5. 51 52 Table D1 Regression results from Heckman selection-corrected log wage regressions—2013 Finance, Transportation Insurance Public Health All Public and and Real Administration and Social Other Sectors Agriculture Mining Manufacturing Utilities Construction Commerce Communication Estate on Services Education Services Services Male 28.8% *** 111.9% *** 56.2% ** 54.3% *** 19.1% *** –1.3% 49.0% *** 19.0% *** 27.6% *** 41.1% *** –14.4% *** 9.8% 28.3% *** Age 20–24 8.5% ** 145.0% *** 60.2% 18.3% –23.4% * –1.7% –8.1% –10.3% 97.8% ** –15.0% 107.7% *** –35.1% * 0.8% Age 25–29 14.6% *** 230.7% *** 87.9% 37.4% ** –22.5% –3.3% –7.6% –1.7% 131.6% *** –9.7% 119.0% *** –35.4% 7.4% Age 30–34 26.2% *** 258.6% *** 123.2% 59.0% *** –13.0% 12.6% –2.9% 3.0% 182.4% *** 4.9% 149.4% *** –30.7% 16.8% Age 35–39 31.7% *** 252.2% *** 125.2% 59.5% *** –10.9% 16.8% 5.0% 14.0% 171.8% *** 12.2% 179.3% *** –29.6% 22.8% * Age 40–44 31.7% *** 310.8% *** 118.8% 62.1% *** –7.8% 21.0% 7.3% 9.6% 158.8% *** 8.1% 192.4% *** –29.2% 15.8% Age 45–49 31.8% *** 252.9% *** 127.5% 59.8% *** –4.7% 28.8% 3.6% 10.8% 171.6% *** 4.2% 201.0% *** –25.8% 16.4% Age 50–54 26.1% *** 278.5% *** 113.8% 57.9% *** –11.9% 13.9% 1.0% 9.0% 154.7% *** –4.1% 195.6% *** –29.2% 10.0% Age 55–59 22.8% *** 252.2% *** 118.4% 49.8% ** –13.5% 13.5% –3.1% 0.8% 179.8% *** –0.7% 174.0% *** –35.5% 8.3% Age 60–64 10.1% *** 146.5% *** 85.2% 29.7% * –22.9% 6.0% –9.9% * –1.6% 168.0% *** –14.0% 109.6% *** –30.0% –10.6% Upper Secondary 11.7% *** –1.6% 15.6% *** 32.3% *** 12.1% *** 13.8% *** 4.8% * 2.3% 33.6% *** 8.8% * 22.8% *** 18.6% *** 8.0% *** Tertiary 48.0% *** 12.3% 35.7% *** 57.8% *** 44.2% *** 41.2% *** 40.8% *** 23.7% *** 35.4% *** 40.1% *** 150.9% *** 53.7% *** 45.5% *** Urban 17.4% *** –48.6% *** 20.0% *** 31.4% *** 19.8% *** 14.0% *** 17.7% *** 13.8% *** –6.7% 22.3% *** –9.3% ** 13.2% *** 11.9% *** Private Sector –1.9% *** –11.6% *** 14.5% * 22.8% ** 1.6% –18.7% *** –3.8% –2.1% 14.6% *** –5.4% *** 0.4% –7.0% *** 3.5% ** Total Number of Observations 115,431 116,014 116,096 116,085 116,098 116,070 116,067 116,071 116,076 115,996 115,963 116,066 116,072 Observations in Earnings Model 69,589 5,056 3,584 3,212 3,097 5,381 7,189 7,499 2,628 8,689 11,803 4,612 6,839 Note: Differences from comparison group are calculated as exp(coefficient) minus 1. In addition to the variables shown, the wage model includes control variables for regions. The reference categories are female, age 15–19, lower secondary education and below, rural residency and public sector. The model for selection into wage employment in the given sector excludes the private sector indicator but includes a set of indicators for household size (1 person to 10 or more members). *** indicates statistical significance at the 1 percent level, ** indicates statistical significance at the 5 percent level, * indicates statistical significance at the 10 percent level. Table D2 Relative expected earnings by education level, sector and region (percent above/below national average for education level) Tertiary Finance, Health Transportation Insurance Public and Public and and Real Administration Social Other Overall Agriculture Mining Manufacturing Utilities Construction Commerce Communication Estate Services Education Services Services Akmola –30 –88 –35 –49 29 40 28 43 118 7 –58 –20 4 Aktobe 9 –93 –14 –37 52 68 83 73 88 22 –53 2 42 Almaty –23 –92 –56 –46 35 68 68 44 84 23 –57 –10 35 Atyrau 33 –89 24 6 83 88 42 72 124 25 –55 0 35 West Kazakhstan –16 –92 –2 –41 31 52 30 43 119 13 –55 –3 17 Jambyl –25 –90 –43 –49 33 50 39 41 74 22 –57 –6 18 Karaganda 3 –93 –6 –21 63 84 35 60 94 24 –59 –2 40 Kostanay –26 –84 –27 –52 31 55 18 53 143 22 –59 –2 35 Kyzylorda –10 –93 –34 –50 16 8 15 40 99 24 –58 –9 3 Mangystau 63 –94 50 2 101 143 117 132 145 37 –48 11 80 South Kazakhstan –24 –90 –27 –53 43 40 36 44 90 21 –55 –9 23 Pavlodar –9 –92 –28 –34 39 75 29 46 122 24 –53 –5 38 North Kazakhstan –38 –90 –29 –56 33 32 12 43 103 17 –56 –23 3 East Kazakhstan –24 –93 –27 –37 36 30 13 36 70 15 –63 –9 19 Astana City 70 –92 –35 –46 134 107 100 110 144 85 –53 23 56 Almaty City 53 –93 22 –13 81 98 89 76 94 52 –53 13 61 Upper Secondary Finance, Health Transportation Insurance Public and Public and and Real Administration Social Other Overall Agriculture Mining Manufacturing Utilities Construction Commerce Communication Estate Services Education Services Services Akmola –24 –86 –24 –41 40 53 32 62 187 16 –73 –14 5 Aktobe 17 –92 0 –28 64 84 59 97 148 32 –70 9 44 Almaty –18 –90 –49 –38 46 84 74 63 143 33 –73 –4 37 Atyrau 44 –87 44 23 98 107 47 95 196 35 –71 7 37 West Kazakhstan –10 –90 14 –32 42 67 35 62 189 22 –71 4 18 Jambyl –21 –88 –34 –41 44 65 44 60 129 31 –72 0 19 Karaganda 11 –91 10 –8 76 102 40 81 155 34 –73 56 42 Kostanay –20 –82 –15 –45 41 70 22 74 220 32 –74 5 36 Kyzylorda –5 –92 –24 –42 25 19 20 59 162 34 –73 –3 4 Mangystau 77 –93 75 18 117 166 125 163 223 48 –67 19 81 (continued) 53 54 Table D3 Relative expected earnings by education level, sector and region (percent above/below national average for education level)  (Continued) Upper Secondary Finance, Health Transportation Insurance Public and Public and and Real Administration Social Other Overall Agriculture Mining Manufacturing Utilities Construction Commerce Communication Estate Services Education Services Services South Kazakhstan –20 –88 –15 –45 54 54 41 64 151 31 –71 –3 24 Pavlodar –2 –90 –16 –24 50 92 34 66 192 34 –70 2 40 North Kazakhstan –35 –88 –18 –49 44 45 16 62 168 27 –72 –17 4 East Kazakhstan –18 –92 –15 –27 47 43 18 55 125 24 –76 –3 20 Astana City 86 –90 –24 –37 152 127 108 139 222 100 –70 32 58 Almaty City 65 –92 42 0 96 117 96 99 155 65 –70 21 63 Lower Secondary Finance, Health Transportation Insurance Public and Public and and Real Administration Social Other Overall Agriculture Mining Manufacturing Utilities Construction Commerce Communication Estate Services Education Services Services Akmola –29 –92 –52 –60 60 45 49 80 37 25 –86 7 5 Aktobe 14 –96 –37 –51 88 75 78 118 18 42 –85 36 44 Almaty –19 –95 –68 –58 67 75 95 81 16 43 –86 21 38 Atyrau 38 –93 –10 –16 127 96 65 116 41 46 –86 33 37 West Kazakhstan –14 –95 –28 –54 62 58 52 80 37 31 –85 30 19 Jambyl –21 –93 –58 –60 65 57 62 77 9 42 –86 26 20 Karaganda 7 –95 –31 –38 101 92 57 101 22 45 –87 31 43 Kostanay –24 –90 –47 –62 62 61 37 93 52 42 –87 31 37 Kyzylorda –5 –96 –52 –61 44 13 34 76 25 44 –86 22 5 Mangystau 65 –96 10 –20 148 153 152 192 53 60 –83 48 82 South Kazakhstan –21 –94 –47 –63 77 46 58 82 19 41 –85 21 25 Pavlodar –6 –95 –47 –48 72 82 51 84 39 45 –85 27 41 North Kazakhstan –37 –94 –48 –65 65 37 30 80 27 36 –86 3 4 East Kazakhstan –21 –96 –46 –50 69 36 32 71 7 34 –88 22 21 Astana City 80 –95 –52 –57 189 115 133 165 53 115 –85 65 58 Almaty City 59 –95 –11 –32 124 106 120 121 22 77 –85 52 64 Figure D1 Expected wages by region and education level, 2013 100 80 60 40 20 0 –20 –40 –60 l be y au an a ay a u an ar an an ty ty a by at nd rd ta ol Ci Ci od an to yr t st st st m m m hs lo s ga gy kh kh kh At na y Ak vl st Ja Al Ak zy ak at ra Pa Ko an za za za ta Ky m az Ka Ka Ka Ka As M Al tK h h st es ut rt Ea W No So Lower secondary and less Upper secondary Tertiary Figure D2 Wage distributions relative to the national average, overall and by sector, 2013 .15 .15 .1 .1 .05 .05 0 0 –10 –5 0 5 10 –10 –5 0 5 10 Log (wage income) relative to national average) Log (wage income) relative to national average) Overall Public utilities Manufacturing Mining Agriculture Construction Overall Commerce Transportation and communication Financial, insurance and real estate Public administration services Education Health and social services          Other services 55 References Arias, Omar S., Carolina Sanchez-Paramo, Maira E. Davalos, Indhira Santos, Erwin R. Tiongson, Carola Gruen, Natasha de Andrade Falcao, Gady Sajovici, and Cesar A. Cancho. 2014. Back to Work: Growing with Jobs in Europe and Central Asia. Europe and Central Asia Reports. Washington, DC: World Bank. Azevedo, Joao Pedro, Sarosh Sattar, and Judy Yang. 2015. Kazakhstan Economic Mobility and the Middle Class (2006–2013). Manuscript in progress. Washington, DC: World Bank. Marouani, Mohamed Ali, Bjorn Nilsson, Angela Elzir, Victoria Strokova, and Namita Datta. 2015. Kazakhstan Jobs Impacts of Sectoral Investments: Simulations using Dynamic Computable General Equilibrium Model Applied to Kazakhstan. A note prepared for Kazakhstan: Jobs: Sector Specific Analysis JERP (P153621), World Bank, Washington, DC. World Bank. 2013. Beyond Oil: Kazakhstan’s Path to Greater Prosperity through Diversifying, Volume 2. Main report. Wash- ington, DC: World Bank. ———. 2014. Strengthening Kazakhstan’s Education System. Washington, DC: World Bank. ———. 2015. World Bank Biannual Economic Update. Washington, DC: World Bank. ———. 2016. Kazakhstan: Towards Development of a Jobs Strategy. Washington, DC: World Bank. 56