JOBS SERIES 10 Issue No. 11 DIAGNOSTIC SIERRA LEONE A l v ar o S . G on z ale z and Ver onic a Mic hel G u t ier r e z DIAGNOSTIC SIERRA LEONE A l v ar o S . G on z ale z and Ver onic a Mic hel G u t ier r e z The publication of this report has been made possible through a grant from the Jobs Umbrella Trust Fund, which is supported by the Department for International Development/UK AID, and the Governments of Norway, Germany, Austria, the Austrian Development Agency, and the Swedish International Development Cooperation Agency. © 2017 International Bank for Reconstruction and Development / The World Bank. 1818 H Street NW, Washington, DC 20433, USA. 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 govern- ments 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 immunities 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://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: Alvaro S. Gonzalez and Veronica Michel Gutierrez. 2017. “Sierra Leone Jobs Diagnostic” World Bank, Washington, 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 attribution: 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. Images: © World Bank Sierra Leone. Further permission required for reuse. Project Number: P156896 ABSTRACT Sierra Leone is a relatively small economy with potential for jobs-rich growth. It is an extremely poor nation with substantial mineral, agricultural, and fishery resources. Nearly half of the working-age population engages in subsistence agriculture. Alluvial diamond mining remains the major source of hard currency earnings, accounting for nearly half of exports. However, the anticipated job-rich recovery is unlikely to be realized simply from the continued exploitation of the minerals underneath Sierra Leonean soil. This Jobs Diagnostic of Sierra Leone is a first, modest, but necessary, step to formulate that focused effort to create good jobs. It provides a better understanding of the job creation performance of the economy. It identifies the major problems that hold back the economy from creating good, formal-sector jobs that will lift people out of poverty and get them on the road to upward mobility. The Jobs Diagnostic begins with a look at the jobs creation performance of the economy. It then provides an examination of the labor supply—the state of the present and future workers and entrepreneurs. Finally, this diagnostic looks at job performance of firms—the potential employers and job creators. The Jobs Diagnostic is the beginning of a dialogue among the government, the private sector, nongovernmental institutions, donors, and others interested in an economy in Sierra Leone that creates goods for all, but especially for the most disadvantaged groups in the country. ACKNOWLEDGMENTS This report was prepared by the World Bank Group’s (WBG) Jobs Group. The principal authors are Alvaro Gonzalez and Veronica Michel Gutierrez. This diagnostic could not have been written without the irreplaceable assistance of Adrian Scutaru and David Keith de Padua. The report was prepared under the general direction and ongoing support of Dino Merotto and David Robalino. Colleagues in the World Bank, including the Sierra Leone Country Office, provided excellent input, comments, and guidance. The authors are particularly grateful to World Bank country manager Parminder P. S. Brar for his support. The authors acknowledge the rich comments provided by the peer reviewers of this documents: Kristen Himelein Kastelic (World Bank Group–Global Practice on Poverty) and Siddharth Sharma (World Bank Group–Global Practice on Trade and Competitiveness). This report has been made possible through a grant from the World Bank’s Jobs Umbrella Trust Fund, which is supported by the Department for International Development/UK AID and the governments of Norway, Germany, and Austria; the Austrian Development Agency; and the Swedish International Development Cooperation Agency. CONTENTS ABBREVIATIONS VI EXECUTIVE SUMMARY 1 Macroeconomic Environment............................................................................................................................... 1 Workers................................................................................................................................................................ 1 Firms.................................................................................................................................................................... 2 MACROECONOMIC PERFORMANCE, GROWTH, AND JOBS IN A VOLATILE ENVIRONMENT 3 Background.......................................................................................................................................................... 3 Economic growth has yet to translate into employment growth........................................................................... 4 Summary and conclusions.................................................................................................................................. 14 LABOR SUPPLY IN A CONTEXT OF LIMITED OPPORTUNITIES AND SEGMENTATION 16 Introduction........................................................................................................................................................ 16 Labor force participation..................................................................................................................................... 19 Employment opportunities................................................................................................................................. 22 Human capital investments................................................................................................................................. 28 Summary and conclusions.................................................................................................................................. 30 DEMAND FOR LABOR AND FIRM PERFORMANCE 33 Introduction........................................................................................................................................................ 33 Cross-country comparisons................................................................................................................................. 33 Firms in Sierra Leone........................................................................................................................................... 37 Employment, labor productivity, and worker compensation............................................................................... 40 Conclusions........................................................................................................................................................ 46 BIBLIOGRAPHY 48 ANNEX A: FIRM-LEVEL DATA SUMMARY STATISTICS......................................................................................... 49 ANNEX B: MISSING DATA AND REPORTED ZEROS.............................................................................................. 56 ANNEX C: MULTIVARIATE REGRESSIONS............................................................................................................ 60 ANNEX D: ADMINISTRATIVE MAP OF SIERRA LEONE.......................................................................................... 67 ABBREVIATIONS DMT............... Dry Metric Ton GDP................ Gross Domestic Product ILO.................. International Labour Organization ISIC................. International Standard Industrial Classification LFS.................. Labor Force Survey LCU................ Local Currency Unit LMC............... Lower middle income country n.a.................. not applicable [in tables and figures] NGO .............. Nongovernmental Organization PPP ................ Purchasing Power Parity SCD................ Strategic Country Diagnostic TFP................. Total Factor Productivity US$................ United States dollars WAP............... Working-age population WDR............... World Development Report EXECUTIVE SUMMARY The economy of Sierra Leone has yet to generate the much-desired gainful employment required to lift many of its people out of poverty. The country’s demography implies that substantial job creation will be needed in coming years. According to the most recent United Nations population estimates, at current rates of population growth, new jobs will have to be created for approximately 100,000 labor market entrants per year. The economy struggles to create jobs because it is undiversified, trapped in subsistence agriculture, and prone to shocks. The economy remains dependent largely on natural resource exploitation for revenue and is dominated by subsistence agriculture for employment. Lacking options apart from natural resource exploita- tion, the economy fails to provide job opportunities for all. Subsistence agriculture employs many but does not provide income to lift people out of poverty. Together, this dependence and dominance leave the economy exposed to shocks. Between 2014 and 2015, the Ebola epidemic and the sharp decline in iron ore prices revealed the country’s high vulnerability to shocks. Continuous shocks make sustaining livelihoods difficult. The question is whether Sierra Leone is poised for a job-rich economic expansion that will take the most vulnerable and disadvantaged out of poverty. Creating more jobs, improving the quality of jobs, and providing for the most disadvantaged people are vital to poverty reduction. This report looks to understand if, and to what extent, Sierra Leone is achieving these labor outcomes. This is done by examining the question from three angles: macroeconomic and sectorial growth, labor supply (the workforce), and labor demand (firms). MACROECONOMIC ENVIRONMENT Volatile and undiversified economic growth has limited job expansion in the postconflict period. Chapter 1 finds that postwar economic growth has not translated into the desired employment growth. Hit by two macro shocks in the past decade and a half, the country has gone through a recovery period for which annual economic growth rates fluctuated from 26 percent to -21 percent. This volatility limits the potential to sustain and spread the gains of any economic upturn. Moreover, economic growth was uneven across sectors. In 2014, 78 percent of the gains in aggregate value added came from the mining sector (exports), whereas 14 percent came from the service sector and 7 percent from agriculture. If anything, employment shares were the opposite. The agriculture sector employed 61 percent of the workforce in that year, whereas only 5.5 percent worked for industry. This weak association between economic output and employment explains the remarkable differences in labor productivity across sectors, with productivity dwindling in the sectors that employ the most workers— agriculture and services. WORKERS Formal-sector job opportunities for the average Sierra Leonean are particularly few. Chapter 2 looks at the characteristics of the workforce and the labor outcomes: participation, employment by type of job, and earnings. Although employment is relatively high, job prospects are limited for many potential workers. One-third of the working-age population remains out of the labor force, although some are capable of working. Under- employment is high, especially among youth and women, and some people work without pay. Informal employ- ment remains predominant and is likely increasing because the working-age population is growing faster than jobs in the formal sector are. Finally, self-employment is the norm since wage jobs are scarce. Sierra Leone’s labor market is segmented, which means that access to “good” jobs is more restricted for individ- uals from certain segments of the population, conditional on human capital endowments (health, education). 1 Low levels of education indeed explain some barriers to good jobs. Much of the labor force has little education or training, especially women and people who live in rural areas. Men who live in urban areas are relatively better able to gain the skills to prepare them to work than are their female and rural counterparts. This implies that even where jobs are created, not everyone has the same chance to acquire the skills and traits needed to get jobs. In addition, the scarcity of good jobs further restricts opportunities. Wage jobs and formal jobs are mostly clustered in mining and construction and in the Western province, around Freetown. For this reason, many indi- viduals have few options other than to create their own jobs through self-employment in small, informal, and largely inefficient household enterprises. One of the fundamental socioeconomic challenges is to ensure that the country’s growth is inclusive and spreads across different segments of society, irrespective of age, gender, or region. Widespread job creation, for all skill levels, in all regions, and in growing sectors, is key to that kind of inclusive growth. FIRMS Firms also exhibit some of the scars of a volatile and fragile economic environment, and this situation affects jobs outcomes. Chapter 3 highlights the firm-level consequences of a constrained investment climate and of the uncertainty caused by a volatile macroeconomic environment. As in many other African econo- mies, the commercial sector outside of agriculture is large in terms of the number of firms in the economy, but it is relatively small in the number of jobs. The disproportionate size of commerce can be read as a sign that the business environment is so harsh that all that may thrive are enterprises that can enter and exit markets quickly in case of rapid changes in economic fortunes. The manufacturing sector carries on, but it is relatively small and endures in the low-cost, low-skill, and low-pay niche of the market. If volatility were reduced and the investment climate were to improve, this sector might be able to employ more of the underemployed, low-skilled workers that are common in Sierra Leone. 2 MACROECONOMIC PERFORMANCE, GROWTH, AND JOBS IN A VOLATILE ENVIRONMENT BACKGROUND Following the end of the 10-year-long civil war in 2002, Sierra Leone made progress in transitioning from the postconflict period to one of sustained economic growth. The economy consistently registered positive growth. Real growth domestic product (GDP) grew at a rate of 7.6 percent per annum between 2003 and 2014, which was 2 points higher than Sub-Saharan Africa’s growth rate over that period. The average annual per capita growth was 4.8 percent. The period was initially marked by reconstruction and slow resumption of economic activity as displaced populations returned to their homes and responded to the establishment of peace. Population growth also picked up its prewar pace and the proportion of the working-age population (WAP) rose at an increasing rate. The mining sector, driven by iron ore exports, represented 78 percent of the growth in GDP in 2014, whereas agriculture, which accounted for most of the country’s labor, represented only 7 percent. Apart from agriculture, which is mainly subsistence farming, the economy is undiversified and largely reliant on diamond production and iron ore exports. Unfortunately, postconflict growth was interrupted, and even reversed, by two shocks that plunged the country into a period of social and economic turmoil. The first was the onset of the Ebola epidemic in 2014. The second was the precipitous drop in the price for iron ore. The rapid spread of the Ebola virus created a public health emergency throughout the country and threatened macro stability, growth prospects, and gains made on poverty and human development. As Ebola spread, private sector activity came to a halt when businesses closed and movement of goods and people was restricted. There were sharp job losses in the private sector. Agricultural exports and manufacturing output declined 30 and 60 percent, respectively. As the government was trying to generate additional resources to fight Ebola, iron ore production came to a halt in early 2015, largely because of the sharp drop in iron ore prices. GDP fell more than 21 percent in that year. The banking sector vulnerabilities also increased, partly because of the pressures created by Ebola and the iron ore crisis and because the banking sector had yet to develop sufficiently to withstand these kinds of shocks. The structure of the economy remains undiversified. In 2015, agriculture made up 61.3 percent of GDP, while services accounted for 33.9 percent and industry contributed 4.8 percent of output. In comparison, the service sector was a greater contributor to output in economies such as Burkina Faso and Guinea, where this sector accounted for 45.6 and 42.9 percent of GDP, respectively. The gap between output of agriculture and out- put of industry was around 15 percentage points in those countries, whereas it was over 56 percentage points in Sierra Leone (see figure 1). Concentration also has persisted across the components of the GDP. In Sierra Leone, private consumption led aggregate demand and exceeded production by the entire economy. Further, relatively little contribution came from government consumption and gross fixed investment (figure 2). The composition of Guinea’s economy resembled the structure of Sierra Leone, whereas Burkina Faso was more diversified. More similar across all three countries is their dependence on the external sector. In Sierra Leone, the current account balance was negative and accounted for more than 25 percent of aggregate demand. Moreover, 70 percent of exports relied on three commercial partners: China (31 percent), Belgium (28 percent), and the United States (11 percent). Other macroeconomic indicators point to the magnitude of risks that stem from the comparatively nascent diversification in the economy. 3 Figure 1 Sierra Leone’s economy has been relatively unidiversified and more concentrated in agriculture, 2015 100 90 33.9 80 42.9 45.6 SHARE OF ECONOMY (%) 70 4.8 60 50 20.2 40 37 30 61.3 20 34.2 10 20.2 0 BURKINA FASO GUINEA SIERRA LEONE AGRICULTURE INDUSTRY SERVICES Source: World Bank staff depiction of information from The Economist Intelligence Unit, Country Report—Main Report, 1st quarter 2017 for each country. Figure 2 Aggregate demand has come mostly from private consumption, 2015 –47.4 IMPORTS OF GOODS & SERVICES –51.3 –40.2 19.4 EXPORTS OF GOODS & SERVICES 26.8 25.2 15.5 GROSS FIXED INVESTMENT 13.2 31.6 10.2 GOVERNMENT CONSUMPTION 8.6 22.7 102.4 PRIVATE CONSUMPTION 102.7 60.6 –60 –40 –20 0 20 40 60 80 100 120 SHARE OF GDP (%) SIERRA LEONE GUINEA BURKINA FASO Source: World Bank staff analysis of information from The Economist Intelligence Unit, Country Report—Main Report, 1st quarter 2017 for each country. Note: GDP = gross domestic product. ECONOMIC GROWTH HAS YET TO TRANSLATE INTO EMPLOYMENT GROWTH Despite a confluence of a rising aggregate demand and a growing labor force, recent growth has not translated into more jobs. As in the past 25 years, no correlation has been observed between economic growth and employment growth (Figure 3).1 Although real value added increased by 127 percent in 2003–2013, employ- ment increased by 45 percent. This implies that for every 1 percent increase in aggregate output, employment 1 In this chapter, data on employment are drawn from the World Development Indicators database, which uses the International Labour Organization ILOSTAT database as its main source. The original data were collected through the Sierra Leone integrated household surveys, which entailed some limitations but provided the time series needed for this analysis. The next chapter uses data collected through the 2014 labor force survey, which is of better quality but had not been implemented in almost 30 years. 4 Figure 3 Employment growth has not seemed to follow real value added growth in Sierra Leone, 1992–2015 0.3 0.2 REAL VALUE ADDED GROWTH 0.1 0 –0.1 –0.2 –0.3 92 93 94 95 96 97 98 99 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 REAL VALUE ADDED GROWTH TOTAL EMPLOYMENT GROWTH Source: World Bank staff calculations using data on gross value added at factor cost, labor force size, and unemployment rate from the World Development Indicators database. increased by 0.36 percent. This value is below most comparator countries in West Africa, except Burkina Faso (Figure 4). The relatively low elasticity means that in Sierra Leone improved economic performance is not the result of a larger labor force. It has resulted from skyrocketing labor productivity, which brought about a large increase in real income but had little effect on employment. In other words, economic growth derived from more productive labor and less labor overall—particularly in the mining sector. Between 2003 and 2014, every 1 percent of output growth in the mining sector was concomitant to 0.16 percent of sectoral employment growth. By contrast, this ratio was 0.39 in the agriculture sector and 0.99 in services. Sierra Leone’s economic growth is comparatively volatile The volatility of Sierra Leone’s economic growth could be impeding the economy’s better job creation performance. Although it is a mystery why some growth is job rich and other growth spells are not, growth and job creation are associated with macroeconomic stability (World Bank 2012). This chapter identifies the pos- sible reasons that employment growth has not followed economic growth in the postwar period in Sierra Leone. The analysis focuses on the lack of economic stability to explain the relatively poor results in overall job creation. First, the analysis provides a characterization of the growth that the country has experienced—including volatility, growth differentials across economic sectors, and drivers of growth. Second, the analysis identifies and describes the changes in aggregate employment, as well as in employment by sector. Third, it examines the decomposition of the changes in GDP per capita and the changes in labor productivity—the juncture between growth and employment. Sierra Leone continues to be classified as a “fragile state.” As signatory to the New Deal for Engage- ment in Fragile States and a pilot country for the New Deal, in 2012 Sierra Leone was one of the first coun- tries to conduct a fragility assessment. This assessment confirmed that the country had made considerable progress in moving out of fragility. However, evidence shows the persistence of some of the underlying drivers of fragility and suggests the need to carefully assess and manage Sierra Leone’s transition. For example, one of the fundamental socioeconomic challenges is to ensure that the country’s growth dividend is inclusive and spreads across different segments of society, irrespective of age, sex, or region. The Gini coefficient is still high, with 50 percent of national income attributable to the top 20 percent of the popu- lation. Rapid economic growth has had only a limited effect on poverty reduction and employment gener- ation, a condition that creates resentment and limits trust in the government. Jobs are key to both poverty reduction and stability. 5 Figure 4 The elasticity of employment growth to economic growth has been relatively low, 2003–2014 GNB SEN CIV NIC UGA SSA BGD SDN LBR LIC ZMB SLE ETH LMC BFA GIN –0.7 –0.2 0.3 0.8 1.3 1.8 PERCENTAGE INCREASE IN EMPLOYMENT FOR EVERY PERCENTAGE OF GDP GROWTH Source: World Bank staff calculations using data on gross value added at factor cost, labor force, and unemployment rate from World Development Indicators database. Note: GDP = gross domestic product; GNB = Guinea-Bissau; SEN = Senegal; CIV = Côte d’Ivoire; NIC = Nicaragua; UGA = Uganda; SSA = Sub-Saharan Africa; BGD = Bangladesh; SDN = Sudan; LBR = Liberia; LIC = low-income countries; ZMB = Zambia; SLE = Sierra Leone; ETH = Ethiopia; LMC = lower middle income countries; BFA = Burkina Faso; GIN = Guinea. Volatility creates uncertainty and reduces investment, suppressing job creation. Investors must be able to predict outcomes reasonably well, especially if they plan to make long-term investments. Volatility makes prediction difficult and therefore generates uncertainty. Until that uncertainty is resolved, investors remain, with their money, on the sidelines. A relative lack of investment, in turn, reduces productivity growth, and this affects both the quantity and quality of jobs an economy produces. For this reason emphasis is placed on the volatility of economic growth in Sierra Leone. The economy has suffered many recent shocks and it was not yet on solid footing, in any case, before those shocks hit. The forces and shocks have caused volatility in Sierra Leone’s GDP growth. Between 1991 and 2015, the economy grew as much as 26 percent (in 2002, at the end of the civil war), but it also plummeted by 21 percent (in 2015, when iron ore prices slumped and the Ebola crisis persisted). Dispersion measures help to encapsulate the experience of economic fluctuations in the context of fragile economies and characterize this type of growth. When there is high variation across years, average values are likely to tell a limited story about the performance of the economy. Consider, for example, the growth rates in Sierra Leone between 2003 and 2014. Real GDP grew at an average annual rate of 7.7 percent, outperforming other similar economies. But when 2015 is included in the period, the average performance drops to 4.9 percent, below most comparators (figure 5). GDP growth in Sierra Leone has been markedly more volatile than in similar economies. Except for Liberia, which had larger fluctuations in GDP growth, Sierra Leone’s experience of growth has been more unstable than that of comparable economies. Figure 6 illustrates this observation using a box-whisker plot. This plot provides a graphic description of the dispersion of 24 years of growth rates observed between 1992 and 2015 for each economy. The end points of the vertical lines (the whiskers) indicate the lowest and the highest rate values, respectively. The widest range is observed in Liberia, with a difference of 141 percentage points between the minimum growth rate (-35 percent) and the maximum (106 percent—not displayed). Sierra Leone 6 Figure 5 Sierra Leone’s economic growth has not been as sustained as that of comparator economies 12 COMPOUND ANNUAL GROWTH RATE OF REAL GDP (%) 10 8 6 4 2 0 –2 LIC LMC SSA BFA BGD CAF CIV ETH GNB LBR NIC SEN SLE SDN 2003–14 2003–15 Source: World Bank staff calculations using data on gross value added at factor cost from the World Development Indicators database. Note: GDP = gross domestic product; LIC = Low-income countries; LMC = lower middle income countries; SSA = Sub-Saharan Africa; BFA = Burkina Faso; CAF = Central African Republic; CIV = Côte d’Ivoire; ETH = Ethiopia; GNB = Guinea-Bissau; LBR = Liberia; NIC = Nicaragua; SEN = Senegal; SLE = Sierra Leone; SDN = Sudan. Figure 6 Sierra Leone has experienced comparably large slumps and surges of the economic growth rate, 1992–2015 0.40 0.35 0.30 YEAR-ON-YEAR REAL GDP GROWTH RATE 0.25 0.20 0.15 0.10 0.05 0.00 –0.05 –0.10 –0.15 –0.20 –0.25 –0.30 –0.35 –0.40 LIC LMC SSA BGD BFA CAF CIV ETH GIN LBR NIC SEN SLE SDN UGA ZMB Source: World Bank staff calculations using data on gross value added at factor cost from the World Development Indicators database. Note: The upper limit for LBR (Liberia) is not shown. LIC = Low-income countries; LMC = lower middle income countries; SSA = Sub-Saharan Africa; BGD = Bangladesh; BFA = Burkina Faso; CAF = Central African Republic; CIV = Côte d’Ivoire; ETH = Ethiopia; GIN = Guinea; LBR = Liberia; NIC = Nicaragua; SEN = Senegal; SLE = Sierra Leone; SDN = Sudan; UGA = Uganda; ZMB = Zambia; GDP = gross domestic product. comes second with a range of 48 percentage points. The horizontal lines outlining each box indicate the quartile values of the data distribution. The bottom line depicts the first quartile of the growth distribution, the middle line indicates the second quartile, and the top line represents the third quartile. For Sierra Leone, this means that the bottom 25 percent of the yearly growth rates in the sample (quartile 1) have a value below 0 percent, while the top 25 percent (quartile 3) are above 6.4 percent. The median is at 4.7 percent, which means that 12 of the 24 years covered had a growth rate above this level. The size of the box gives the interquartile range, or how spread out the middle half of the data is. Again, Liberia and Sierra Leone have the highest interquartile ranges at 8.4 and 7.0 percentage points, respectively—that is, the difference between the lowest and the highest of the growth rates in the middle half of the data. Other economies, such as the region of Sub-Saharan Africa (excluding high-income economies), Bangladesh, and Nicaragua, have a similar median growth rate but with an interquartile spread of less than 3 percentage points. 7 Without steady growth, it is hard to sustain the type of investments that strengthen economic sectors and foster job creation. Even before the war ended, gross capital formation reflected the prospect of future sta- bility in Sierra Leone. It rose from 1 percent of GDP in 2000 to fluctuating between 9.8 and 11.8 percent through- out the decade and finally peaked at 42 percent in 2011. But this investment was ephemeral. In 2015, capital investment was down at 15 percent of aggregate output. Comparatively, Bangladesh consistently increased capi- tal formation to output year by year since 1990, rising from 16 percent then to 29 percent in 2015. Ghana is in a similar situation, although with somewhat more variability. Sustained periods of growth usher in long-term invest- ments in key sectors, which in turn may generate spillovers for the rest of the economy and promote employment. Volatility might be attributable to Sierra Leone’s concentrated economic structure. As will be evidenced in the following section, output in the country is rather concentrated. The literature recognizes a link between income level, volatility, and lack of diversification. First, Pritchett (2000) finds that poorer economies are characterized by unsustained growth and volatility: “peaks and valleys.” In comparison, better economic performers are less volatile and are characterized by “peaks and plateaus.” Koren and Tenreyro (2007) show that less developed countries experience greater growth volatility because of increased concentration in volatile sectors. Moore and Walkes (2010) show that less diversified economies have higher rates of output, investment, and consumption growth volatility. High economic concentration makes an economy vulnerable to the fate of fewer economic events, such as changes in the price of the most prevalent commodity sold, looming as a stumbling block for economic improvement. Growth in Sierra Leone has been uneven across sectors and largely driven by the export-led mining sector The postwar economic recovery was first driven by agriculture and then propelled by industry. In the past 25 years (except for 2013–14), more than half of the economic activity in the country emanated from the agriculture fields of Sierra Leone (figure 7). After the war was over, agriculture picked up quickly, growing at an average rate of 8.3 percent between 2003 and 2008. During this period, 70 percent of economic growth came from agriculture, on average (figure 8). Industry and the service sector rose more moderately (at an annual aver- age rate of 1.7 and 3.8 percent, respectively). However, the spell of vertiginous growth shifted toward industry in the following years. While the growth rate of agricultural output dropped to an average of 3.5 percent between 2009 and 2014, industry grew at a pace of 44.6 percent. The industrial sector’s value added went from composing 9 percent of the economy in 2011 to 30 percent in 2014. By that year, 78 percent of economic growth was driven by the expansion of industry, whereas 14 percent was attributable to the service sector and 7 percent to agriculture. Figure 7 Postwar growth was steady across sectors until the upsurge and slump in industry, 1990–2015 4.5 4.0 VALUE ADDED (BILLIONS CONSTANT 2010 US$) 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0 1990 1992 1993 1994 1995 1997 1998 2000 2001 2003 2004 2005 2006 2008 2011 2012 2013 2014 2015 1991 1996 1999 2002 2007 2010 2009 AGRICULTURE SERVICES INDUSTRY Source: World Bank staff calculations using data on gross value added at factor cost from the World Development Indicators database. Note: GDP = gross domestic product. 8 Figure 8 The industry sector contributed less than 2 percent of GDP growth before driving almost 80 percent of GDP growth and then reversing, 2003–15 125 100 PERCENTAGE OF THE GDP GROWTH RATE 75 50 25 0 –25 –50 –75 –100 –125 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 AGRICULTURE SERVICES INDUSTRY Source: World Bank staff calculations using data on gross value added at factor cost from the World Development Indicators database. Note: GDP = gross domestic product. The growing export-led mining sector drove years of robust economic growth. The performance of industry was led by a growing mining sector. Iron ore, which was not produced before 2011, represented nearly 15 percent of GDP alone in 2013. Rising commodity prices fueled this expansion. Iron ore exports played an important role, growing at an average annual rate of 15.2 percent between 2009 and 2014 and making up 65 percent of total value of exports by the end of the period. However, iron ore’s time as a driver of the economy proved brief. The following year, when international prices for iron fell sharply (figure 9), the production of iron ore came to a halt. The two com- panies operating in the sector (Shandong Iron and Steel Group and Timis Mining Corporation) suspended activities because of the low world prices. Average annual prices in 2015 stood at just a third of their 2011 highs ($55.8 per dry metric ton (dmt) in 2015 compared with $167.8/dmt in 2011) due to new low-cost supplies, primarily from Australia but also from Brazil. The new supplies led high-cost production in different countries to close, dwarfing demand for Sierra Leone’s exports. Crucial in this crisis, was the weakened demand from the Chinese steel industry, which consumes nearly all iron ore output by Sierra Leone. The share of iron ore in total exports fell to 21 percent and the total value of exports shrunk by 73 percent, going from $2.14 billion to $582 million (figure 10). GDP contracted by 21 percent from 2014 to 2015, just what the economy had gained two years earlier. Figure 9 After a period of rapid increase, international iron ore prices plummeted in 2015 180 160 140 120 ANNUAL PRICE 100 80 60 40 20 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 NOMINAL $/DMTU REAL $/DMTU, 2010 US$ Source: World Bank 2017. Note: DMTU = dry metric ton unit. 9 Figure 10 Iron ore exports represented 65 percent of total exports in 2014; after prices plummeted, this share dropped to 21 percent A. 2014—TOTAL VALUE OF EXPORTS: US$2.14 BILLION B. 2015—TOTAL VALUE OF EXPORTS: US$5.82 MILLION Total: $2.14B Total: $582M Iron Ore Tin Ores Diamonds Iron Ore Aluminium Ore Diamonds Cocoa Beans 3.8% 8.2% Cocoa 21% Shells 22% 0.47% 11% Scrap Vessels Cocoa... 1.1% Titanium Ore Titanium Aluminium Ore Cars 1.4% Cocoa Shells 9.3% Excavation Machinery Non-Fillet Frozen Fish Coffee 1.2% Cars 0.46% Ore Rough Plastic Niobium, Tantalum, Vehicle... 1.7% 1.6% Wood Lids Vanadium and Zirconium Ore 0.85% 0.62% Rubber Tires 3.6% 2.3% 3.4% Other Sulfites Scrap Iron Scrap... 0.55% 18% Coloring... 65% Scrap Iron Models... Non... Niobium, Tantalum, Vanadium and... 0.68% 1.5% Iron Chains 1.2% –0.67 0 15.2 –0.73 0 6.49 Note: The number indicates share of exports by product and the color scale represents 5-year annual growth rate. Source: The Observatory of Economic Complexity, http://atlas.media.mit.edu/en/. Sector-level growth has varied with respect to job creation The employment rate was already high in Sierra Leone and economic growth in the postwar period pushed it up slightly. The employment rate budged slightly from 96.6 percent of the labor force in 2003 to 96.7 percent in 2014.2 Each sector performed differently in contributing to this nearly imperceptible improve- ment in employment numbers. Of the total 745,000 jobs added to the economy, 48 percent were jobs in agri- culture, 8 percent in industry, and 44 percent in the service sector (figure 11). Agriculture is the largest provider of jobs, but its contribution to overall employment has dwindled. In 2003, 67.3 percent of laborers were employed in agriculture. Ten years later, the sector remained the larg- est provider of jobs, with a share of total employment reduced to 61.1 percent, compared with 33.4 percent Figure 11 Agriculture and Services drove net job growth, and the industrial sector contributed little, 2003–14 400 350 CHANGE IN NUMBER OF PEOPLE IN A JOB 300 250 (1,000 PEOPLE) 200 150 100 50 - AGRICULTURE INDUSTRY SERVICES Source: World Bank staff calculations using employment data from the World Development Indicators database. 2 In this chapter, the employment rate is presented as the number of people employed as a percentage of the labor force. Chapter 2, on the other hand, uses a definition that calculates the number of people employed as a percentage of the working-age population. To avoid confusion, each definition is used to bring attention to the main point of the analysis in each chapter. Chapter 1 focuses on the relationship between economic growth and macro-level employment. Therefore, employment rates are needed that describe the ultimate equilibrium in the labor market. Because chapter 2 examines labor supply issues, the employment ratios need to capture the proportion of the working-age population that is inactive. 10 employed in services and 5.5 in the industrial sector (figure 12).3 The reduced participation in employment might be a reflection of the sector’s slowdown of its value added growth. Figure 12 Most of Sierra Leone’s jobs are in agriculture, but workers have shifted toward services, 2014 SHARE OF EMPLOYMENT BY SECTOR A. 2003 B. 2014 28.5% 33.4% 61.1% 4.2% 5.5% 67.3% AGRICULTURE INDUSTRY SERVICES Source: World Bank staff calculations using employment data from the World Development Indicators database. The service sector has begun to employ a higher proportion of the labor force, ushering in incipi- ent structural transformation of the economy. The postwar period witnessed the shifting of employment from agriculture to services. While agricultural employment shrunk, reducing the overall employment rate by 5.9 points, the service sector added 4.8 percentage points (figure 13). The shift of employment from agri- culture to services is projected to continue toward 2020 but with a slightly more optimistic increase of the aggregate employment rate of 0.8. The out-sized contribution of the industry to overall growth has not been mirrored in terms of employment. Throughout the 11 years in which industry grew at an annual average rate of 18.4 percent, the sector created 62,000 jobs. This contribution of 1.26 percentage points to the employment rate is relatively low when benchmarked against the service sector, which grew at a 5 percent annual rate and created 328,000 jobs. The service sector’s share of employment increased to a small extent (figures 11–13). Figure 13 Services have led a slight increase in the employment rate, with some industrial job growth, while the agricultural workforce shrinks SERVICES INDUSTRY AGRICULTURE –6 –4 –2 0 2 4 6 PERCENTAGE POINTS OF THE EMPLOYMENT RATE PERIOD: 2003 TO 2014 (CHANGE = 0.1%) PROJECTIONS: 2014 TO 2020 (CHANGE = 0.78%) Source: World Bank Group, Jobs Structure Tool, version February 2017. 3 A caveat on comparability of employment indicators over time is in order. For example, the International Labour Organization changed the definition of employment to no longer include unpaid work. This could explain, in part, the declining share of employment coming from agriculture. 11 Increases in labor productivity, mostly in the industry sector, have driven growth in real GDP per capita A more productive labor force contributed largely to the increase in real GDP per capita. Annual real GDP per capita grew at a rate of 4.8 percent between 2003 and 2014. Of this growth, 86 percent (4.09 percent- age points) was due to increases in labor productivity. The enlargement of the WAP and higher participation in the labor force also pushed GDP per capita growth, though its effect was small in magnitude (figure 14). More productive jobs are better jobs because they lift the economy and generate higher wages. The productivity boost of the postwar period created better jobs, albeit unequally distributed across economic sectors. Labor productivity gains were concentrated in the industry sector. Slightly more than half (51 percent) of the gains in productivity came from the industry sector, in particular from the spike in commodity prices from 2011 to 2014. The sector responded to market demand by increasing output, investing in machinery, and modestly increasing its workforce (as observed in figures 11 and 12). Investments likely improved the productivity of this sector’s workers, but it created relatively few jobs. The opposite occurred in agriculture and services, in which the majority of jobs in the economy are found. Labor productivity in agriculture moved up by 1.4 percent points each year and remained stagnant in the service sector (figure 15). Figure 14 Figure 15 The increase in per capita value added has come mostly from an increase in Industry contributed largely to the 4.09 percentage points of annual labor labor productivity productivity growth, 2003–14 DECOMPOSITION OF REAL GDP PER CAPITA DECOMPOSITION OF THE ANNUAL LABOR PRODUCTIVITY GROWTH RATE (%), 2003–14 GROWTH RATE (%), 2003–14 0.007 0.25 0.01 0.60 0.43 2.10 1.38 4.09 INDUSTRY PRODUCTIVITY AGRICULTURE PARTICIPATION RATE INTERSECTORAL SHIFT DEMOGRAPHIC CHANGE SERVICES EMPLOYMENT RATE Source: World Bank Group, Jobs Structure Tool, version February 2017. Source: World Bank Group, Jobs Structure Tool, version February 2017. Intersectoral shifts of labor also contributed to the rise in productivity. The movement of workers across sectors explains 15 percent of the increase in value added per worker in the economy. Labor moved from agriculture—the least productive sector of the economy—to services, in which productivity was 1.8 times higher. However, when the receiving sector lacks capacity to increase output as more workers enter its workforce, employment reallocation can hurt productivity. In Sierra Leone, the service sector increased its share of employ- ment by 5 percentage points, and by the end of 2014 its productivity differential dropped to 1.3 times higher than in agriculture. Output per worker (the measure for labor productivity) increased by 38.6 percent from 2003 to 2014 in agriculture, whereas it grew by 0.2 percent in the service sector. Continued increases in GDP per capita will come from better jobs. Changes in labor productivity will continue to set the course for a growing GDP per capita. Increases in the employment rate will be difficult to improve on because the level of employment is close to 100 percent of the labor force. Demographic change will bring a youth bulge, which, if fully exploited, could boost growth. However, given that 33 percent of the WAP remained inactive in 2014, the economy will have to create an environment in which more people are more willing to participate in the labor force. Future increases in GDP per capita will still primarily stem from improvements in the value added per worker, possibly from the industrial sector but also from improving 12 Figure 16 Value added per capita rose significantly, then fell to only slightly above the prewar level, 1990–2015 700 VALUE ADDED PER CAPITA (CONSTANT 2010 US$) 590 604 600 497 500 464 428 439 422 416 376 395 407 400 368 315 327 300 200 100 - 1990 1995 2000 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Source: World Bank staff calculations using data from the World Development Indicators database. agriculture productivity. By the end of this decade, it is possible that the continuing shift of workers to the service sector will render the sector’s net contribution (within the sector plus the intersectoral shift) negative in productivity. Despite overall growth, GDP per capita has remained low and improvements in poverty have been modest GDP per capita levels remained virtually the same over the past 30 years. After almost doubling during the postwar period, GDP per capita levels came down again to $464 (in constant 2010 US$) in 2015 (figure 16) as a consequence of plunging export prices. This amount is below the average GDP per capita value during the first half of the 1980s. Again, this stagnation is an indication that growth is far from being widely supported by a productive WAP. Compared with other countries with similar GDP per capita at the beginning of 1990, Sierra Leone experienced limited growth in its value added per capita. Bangladesh, Burkina Faso, and Uganda showed sustained improvements and outperformed Sierra Leone over the period (figure 17). By the same token, economic growth has hardly translated into improvements in living standards. Still more than half of the population lives in extreme poverty. When measured by the international standard of $1.90 a day, the poverty head count ratio fell from 58.5 percent in 2003 to 52.3 in 2011 (an improve- ment of 6 percentage points in nine years). Other countries such as Bangladesh, Burkina Faso, and Uganda Figure 17 In GDP per capita, other countries outperformed Sierra Leone, whose economy was hurt by war and volatility, 1990–2015 1,600 1,400 VALUE ADDED PER CAPITA (CONSTANT 2010 US$) 1,200 1,000 800 600 400 200 - 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 LIBERIA SIERRA LEONE BURKINA FASO BANGLADESH CÔTE D'IVOIRE UGANDA Source: World Bank staff calculations using data from the World Development Indicators database. 13 experienced sharp declines in their poverty rates during the same period (figure 18). Moreover, when con- sidering a threshold of $3.10 a day, the poverty headcount showed even less improvement, falling from 80.8 percent to 79.9 percent in the same period. In 2014, the country was poorer than it was in 1989 (78.3 percent in poverty). Figure 18 The poverty headcount ratio at $1.90 a day (2011 PPP) has dropped moderately in Sierra Leone 90 80 PERCENTAGE OF TOTAL POPULATION (%) 70 60 50 40 30 20 10 0 92 93 91 94 95 96 97 98 00 99 01 02 03 04 05 07 06 08 09 10 11 12 14 13 19 19 19 19 19 19 19 20 20 19 19 20 20 20 20 20 20 20 20 20 20 20 20 20 LIBERIA BURKINA FASO CÔTE D'IVOIRE SIERRA LEONE BANGLADESH UGANDA Source: World Bank staff calculations using data from the World Development Indicators database. Note: PPP = purchasing power parity. Data for Liberia are available for 2007 only. Data for Sierra Leone are available after 2003. SUMMARY AND CONCLUSIONS Postwar economic growth created jobs and improved productivity, but only to a small extent. The econ- omy indeed is creating more jobs than it did in the past. However, the increase in the employment rate was limited to 0.1 percentage point over 11 years (2003–14) and mostly came from the service sector. The productivity of jobs also increased, adding 4.09 points to GDP per capita growth each year. Nevertheless, this rise came largely from the industry sector, whereas services were almost excluded from these gains. This disproportionate dis- tribution of employment and productivity across sectors is worrisome. Projections show an ongoing (and intensified) situation in which productivity growth is driven by capital-intensive industries that have limited capacity to generate jobs. Moreover, productivity is dwindling in the sectors that concentrate the most workers: agriculture and services (figures 19 and 20). Figure 19 Figure 20 Labor productivity by sector, 2014 and 2020 (projected) Employment by sector, 2014 and 2020 (projected) 1,800 18 NUMBER OF PEOPLE IN A JOB 1,600 16 MILLION CONSTANT 2010 US$ 1,400 (1,000 PEOPLE) 14 1,200 12 1,000 10 800 600 8 400 6 200 4 0 AGRICULTURE INDUSTRY SERVICES ETC. 2 2014 2020 PROJECTED 0 AGRICULTURE INDUSTRY SERVICES ETC. Source: World Bank Group, Jobs Structure Tool, version February 2017. 2014 2020 PROJECTED Source: World Bank Group, Jobs Structure Tool, version February 2017. 14 Structural transformation has yet to happen. The structure of the current economy (see box 1)—with employment and productivity highly concentrated in a few sectors—hinders the development of broad-based, sustainable growth. Following current trends, the elasticity of employment to growth might drop from 0.37 to 0.30, with 565,000 jobs added by 2020. At the sector level, reallocation of labor is not yet fully efficient. The recent shift of employment from agriculture to services resulted in declining relative productivity in agricul- ture. Continuing this path without finding drivers for productivity growth within the service sector will result in further reduction of overall productivity. Furthermore, productivity in the service sector is not much higher than that of agriculture, suggesting that while workers moved to relatively better jobs, they did not find highly productive jobs. BOX 1: EMPLOYMENT AND PRODUCTIVITY ACROSS SECTORS, 2003 – 14 On one side, the agriculture sector made up around 50 percent of real GDP. It also comprised more than ¬  60 percent of the country’s labor force. Productivity in this sector was lower than that of the other two groups of economic activity. On the other side, industry contributed to 51 percent of the overall increase in productivity in the economy from ¬  2003 to 2014, but it only added 8 percent of the jobs created in the same period. The service sector was in between. It held 33 percent of the labor force and it generated 44 of the jobs created ¬  in the period. Its labor productivity did not increase, although it was still higher than in the agriculture sector. Understanding where to focus to make growth more robust is imperative to take advantage of increases in the working-age population. Recently, the spike in commodity prices supercharged the econ- omy, but it was only temporary and did not provide enough time and boost to make growth robust within and across sectors. That type of growth creates a limited number of jobs and causes the inequality among workers of different sectors to intensify. The demographic bonus can help the country move from fragility and toward stability and diversification of the economy. The challenge is to create more productive jobs for the burgeoning youth population. 15 LABOR SUPPLY IN A CONTEXT OF LIMITED OPPORTUNITIES AND SEGMENTATION INTRODUCTION In the aftermath of war and the Ebola outbreak, population growth is recovering steadily, especially in the working-age population. The 10-year civil war and the Ebola virus took thousands of lives and cut birth rates. Life expectancy is among the lowest in the world (50.9 years). A postwar, declining fertility rate also contributed to the slow recovery in the population growth rate. The average number of children per woman dropped from 6.6 in 1990 to 4.6 in 2014. Nevertheless, in the period between 2002 and 2014, the population grew at an annual average rate of 3.0 percent, surpassing its prewar levels (2.3 percent between 1979 and 1991, and 0.7 percent during the years of the conflict). What is more, the working-age population (WAP), those between 15 and 64 years old, grew at a slightly more accelerated pace of 3.2 percent (figure 21). Figure 21 Population by age groups, 1985–2014 7 6 POPULATION SIZE (MILLION PEOPLE) 5 4 3 2 1 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 0–14 15–64 65+ An increasing working-age population offers a demographic dividend for Sierra Leone, but it also brings challenges. A combination of declining fertility rates, relative low life expectancy, and a youth bulge entering the workforce has resulted in an increase of the WAP’s share in the population. In 2014, of 6.4 million people living in the country, 54.7 percent (3.45 million) were among the WAP. Sierra Leone’s economic recovery should be taking better advantage of this demographic bonus. The dependency ratio—the ratio of those out- side the WAP to those in the WAP—already dropped from 89.8 in 1990 to 81.9 in 2015, and it may continue to decrease because of an expected youth bulge (figure 22). This trend represents a window of opportunity if it results in more producers than consumers in the economy. However, for such a demographic dividend to be secured, labor outcomes must be improved. Although declining, the fertility rate remains high and the dependency ratio is the 26th highest in the world. Also, the 16 Figure 22 The demographic trend in Sierra Leone is bringing dependency ratios down RATIO OF THOSE NOT IN THE WORKING-AGE 100 POPULATION TO THOSE IN WORKING-AGE 80 60 40 20 0 1990 2015 2020 2025 2030 2035 2040 2045 2050 OLD-AGE DEPENDENCY YOUTH DEPENDENCY Source: World Bank staff calculations using data on population by age from the World Development Indicators database. population will continue to increase, adding more people from the WAP to the economy. For this reason, it will be central that a high portion of the WAP enters the labor market, that new entrants find productive employ- ment, and that current workers improve the quality of their present jobs. A variety of challenges to attaining this scenario is already foreseeable. As a preamble, Figure 23 shows that more than one-third of the WAP remains outside the labor force, 62.2 percent work with some remuneration,4 and of this group, only 9.5 percent receive a secure salary. Continuing in this path will mean that consumers outnumber the productive population and that the demographic boom plays against economic prosperity. This chapter provides more detail about these labor outcomes and shows specifically that they are associated with the characteristics of the population. The volatile macroeconomic environment, discussed in the previous chapter, is the backdrop against which individuals make decisions about their participation in the labor force. Volatility introduces uncertainty into labor supply choices. Although uncertainty affects any supply decisions, the analysis in this chapter indicates that other factors are even more important to workers in Sierra Leone. Individuals make life Figure 23 Structure of the working-age population in Sierra Leone, 2014 WAGE WORKER 9.5% AGRICULTURAL EMPLOYED SELF-EMPLOYMENT 62.2% 59.2% NONAGRICULTURAL UNEMPLOYED WORKING AGE SELF-EMPLOYMENT 2.8% POPULATION 31.3% 54.7% UNPAID WORKER 11.6% OUT OF THE LABOR FORCE 35.0% OTHER 88.4% Source: World Bank staff calculations using information from Margolis et al. 2016. 4 See footnote 2 in chapter 1 for an explanation of the use of different definitions of the employment rate across this study. 17 choices that will provide them their livelihood. Individuals decide how many years of schooling to complete, when to get married and build a family, whether to start a household business or search for a job, and whether to undergo training or migrate, among other choices and options. These factors are important because in the aggregate, labor choices made by individuals affect not only the productivity of the country, but also its stability and its ability to reduce poverty through economic growth and employment. Yet, in Sierra Leone few people can choose from various options. This chapter finds that labor market segmentation is a strong determinant of labor outcomes. Seg- mentation means that, after accounting for skills, an individual’s characteristics are correlated with certain types of job outcomes. If there were less segmentation and more fluidity in the labor market in Sierra Leone, the rela- tionship between characteristics and occupations would be less apparent or more easily explained by preferences or the geographic restrictions of some economic activities—such as mining or agriculture. However, the data indicate that the labor market is segmented into “good” and “bad” jobs, with many in the latter category and few in the former. Many workers hold jobs that are informal, probably temporary or casual, and certainly low paid. Women, rural residents, and youth especially find their choices are rather limited to bad jobs. Good jobs can develop from the growth of the formal and higher value added sector of the economy, but such jobs are so few, especially in rural areas, that even better educated workers are often forced to take unskilled jobs and work as low-paid laborers. Understanding segmentation is important to understanding labor and livelihood outcomes in Sierra Leone. Segmentation helps explain the choices that participants in the labor force make and the opportunities that they have. Individuals are very limited by segmentation because it is associated with the person’s back- ground characteristics that cannot be changed at a given time. In addition to segmentation, two other root causes limit choices in the country: the scarcity of productive job opportunities and low education. This chapter addresses both. Labor outcomes affect the well-being of the population because limited access to regular and productive jobs results in low earning capacity and pervasive poverty. Also, widespread informality means that the vast majority of workers are unlikely to benefit from wage growth. An important part of this analysis is dedicated to identifying the populations most disadvantaged by the lack of opportunities to improve their skills and to get “good” jobs. The WAP in Sierra Leone is predominantly young and female, has never gone to school, and lives in rural areas (Figure 24). To illuminate Figure 24 Distribution of the working-age population by sociodemographic characteristics, 2014 TERTIARY DEGREE 100 TECH UPPER CERTIFICATE WESTERN SECONDARY PERCENTAGE OF THE WORKING-AGE POPULATION (%) LOWER 80 SECONDARY WOMEN SOUTHERN COMPLETED YOUNG PRIMARY RURAL 60 INCOMPLETE PRIMARY NORTHERN 40 NO SCHOOL MEN OTHER 20 ADULT URBAN AND ELDER EASTERN FREETOWN 0 AGE GENDER EDUCATIONAL ATTAINMENT LOCATION REGION Source: World Bank staff calculations using table 1.1 “Key Aggregate Labor Market Statistics” in Margolis et al. 2016. Note: Young = ages 15–24; Adult = age over 25. 18 the effects of segmentation on job outcomes, this report identifies characteristics—age, gender, and location— most correlated with gaining good job outcomes (a definition and measure for this process come later in the chapter). Determining these characteristics helps pinpoint the individuals least likely to have opportunities. The structure of the chapter is as follows. The first part provides indications that labor force participation is affected by a perception of exclusion from job opportunities. The second part describes segmentation in employment opportunities in three ways: it (a) acknowledges the predominant characteristics of jobs in Sierra Leone, (b) shows that good jobs are scarce and concentrated, and (c) examines what type of individual has the best chances to enter such jobs. Finally, the third part explains the barriers to a better-educated work- force. The information in this chapter is drawn from the World Bank publication Findings from the 2014 Labor Force Survey in Sierra Leone (Margolis et al. 2016). Many of the graphs, figures, and numbers are taken directly from that study and annotated to let the reader know where to find more information on the subject, table, or figure. This chapter leverages the data analysis and synthesis provided in that document and derives a complementary analysis that aligns with the discussion in this Job Diagnostic. See box 2 for more information. BOX 2. WORLD BANK STUDY: FINDINGS FROM THE 2014 LABOR FORCE SURVEY IN SIERRA LEONE The report seeks to provide a picture of the jobs landscape using the country’s first labor force survey since 1984. The Sierra Leone Labor Force Survey was collected between July and August 2014. The World Bank study analyzes labor market indicators at the national level and across various dimensions, includ- ing breakdowns by district and province, the urban–rural divide, and gender. As needed, further analysis of relevant subpopulations is included. The report considers three main types of current economic activity: wage and salary employment, self-employment (divided into nonagricultural self-employment and agricultural self-employment), and unpaid work. The analysis covers all key, traditional labor market indicators: employment, unemployment, time-related under- employment, broader measures of unemployment, labor force participation, inactivity, formality of employment and nonfarm household enterprises. It also includes indicators of household agricultural activities, indicators on nonfarm household enterprises, indicators on the extractive sector, indicators on migration and civil conflict, indicators of youth employment, and skills. Consult the appendix of report for definitions of those concepts. LABOR FORCE PARTICIPATION Together with scarcity of jobs and shortage of skills, labor market segmentation may discourage labor force participation. When jobs are localized, when they are highly skill specific, or when they are accessible for only a particular subpopulation—that is, when there is segmentation—potential workers who are excluded from the opportunities for good jobs might choose not to participate in the labor force. In Sierra Leone, although 2.8 percent of the WAP are unemployed according to the International Labour Organization (ILO) definition, 9.1 percent are included among the broad unemployed. The broad unemployment rate is a labor market indicator that takes discouraged job seekers into account and is defined as the ratio of the sum of the total unemployed and all discouraged workers to the sum of the labor force and all discouraged workers. In other words, among the WAP who are able to work, 6.3 percent do not have jobs and are not actively searching for work. The discouraged vary in proportion across types of people, educational level, and location. Among youth, women, and rural inhabitants, there are more unemployed people not seeking work than those who are seeking work (Figure 25). Of women who can work, 7.3 percent are not searching for a job, compared with the 2.4 percent who are searching. Also, 6.8 percent of the working-age rural dwellers are not searching for a job, whereas 2.1 percent are unemployed and searching. Similarly, among those who never went to school, those who have incomplete primary education, and those living in the Northern and Southern areas, the ratio of the unemployed and not actively searching to the unemployed searching is higher than two (Figure 26). 19 Figure 25 Discouraged unemployed people represent a large share among women, youth and rural inhabitants 14 SHARE OF THE WORKING-AGE POPULATION 12 10 6.0 BY SUBGROUP (%) 3.3 2.4 8 2.1 2.8 3.3 6 3.6 4 6.8 7.3 6.8 6.3 6.1 5.2 2 4.3 0 OVERALL YOUTH MEN WOMEN URBAN OTHER URBAN RURAL FREETOWN UNEMPLOYED AND NOT SEARCHING UNEMPLOYED AND SEARCHING Source: World Bank staff analysis using figures 1.15 and 1.16 of Margolis et al. 2016. Figure 26 Those with low educational attainment and those who live in North and South have twice the share of discouraged unemployed of other groups 18 SHARE OF THE WORKING-AGE POPULATION BY SUBGROUP (%) 16 14 12 9.8 10 5.1 5.8 3.4 1.9 8 2.1 8.2 2.8 6 2.0 2.6 2.9 4 7.6 7.0 7.1 6.8 5.7 6.1 6.2 5.3 2 4.6 4.3 3.4 0 NEVER WENT INCOMPLETE COMPLETED COMPLETED COMPLETED TECH TERTIARY EASTERN NORTHERN SOUTHERN WESTERN TO SCHOOL PRIMARY PRIMARY LOWER UPPER DEGREES + DEGREE AREA SECONDARY SECONDARY CERTIFICATE UNEMPLOYED AND NOT SEARCHING UNEMPLOYED AND SEARCHING Source: World Bank staff elaboration using figures 1.15 and 1.16 in Margolis et al. 2016. The relatively low shares of the unemployed engaging in an active work search may reflect the existence of extenuating circumstances. A lack of skills is mentioned by 10.1 percent of respondents and is the most common answer from men as one reason for not searching actively for work. Also of relevance, 7.7 percent of respondents reported to be discouraged by a perceived lack of available jobs. Migrants, urban residents, and men cited this noticeably more often than their counterparts (not migrants, rural residents, and women). Markedly, 31.7 percent of individuals with tertiary degrees and more than 60 percent of those with technical degrees plus certificates are discouraged by the low availability of jobs, an answer given disproportionately by those groups. Despite a sense of exclusion from job opportunities, a relatively high proportion of Sierra Leone’s working-age population remains in the labor force. Even when segmentation results in some exiting the 20 labor force, participation remains relatively high. This is because in low-income countries like Sierra Leone few can afford not to work, so they employ themselves. Over 65 percent of Sierra Leone’s WAP, which represents nearly 2 million people, participates in the labor market (Figure 27). Labor force participation varies widely by subgroups, however. Indeed, it has a strong correlation with education. Over 80.0 percent of the most educated participate in the labor market, whereas the share is 44.2 percent among those who have only completed primary school. Fur- thermore, labor force participation varies by background characteristics of the individual. Although slightly, it varies by gender. Women participate at almost the same proportion as men: 64.7 percent for women and 65.7 percent for men. Labor force participation also varies by age; it averages 85.1 percent among prime-age workers (ages 36–55), but is 75.3 percent among the oldest of working age (ages 55–64). Finally, participation varies across geographical areas and is lower in urban areas. The participation rate is 53.9 percent in urban Freetown and 54.1 percent in other urban areas, compared with the considerably higher rate of 69.4 percent in rural areas. The participation rate in the Northern Province is the highest: 72.3 percent of the WAP is either employed or unemployed (ILO data), whereas the Southern Province and the Western Area have the lowest participation rates and consequently the highest inactivity rates (41.4 and 45.4 percent are inactive, respectively). So, while there may be a general lack of good job opportunities, the labor force participation rate varies by individual characteristics and location. Figure 27 Labor force participation rates vary widely by subgroups 90 PERCENTAGE OF THE WORKING-AGE POPULATION IN THE 80 70 60 LABOR FORCE (%) 50 40 30 20 10 0 LL R H EN EN ED ED NT NT L Y Y Y Y TE E N N L N RN RN EA O RA RE DE AR AR AR AR UT W BA ER RA CA M M BL BL RA RA HO AR HE HE G TO RU EL ST YO IM IM ND ND UR O VE SA SA DE IFI IG IG RT UT SC T- RN EA W EE PR PR CO CO O RT DI DI M M ER NO UL SO Y FR TE TO E ED CE AR TH T T SE SE AD ET ES NO NO N ET T I O + PL RT BA ER R W EN PE PL ES M TE W UR W M UP RE CO LO CO R G ED VE IN ED DE ET NE ET CH PL PL M TE M CO CO Source: World Bank staff elaboration using table 1.1, “Key Aggregate Labor Market Statistics,” in Margolis et al. 2016. A significant number of people participate in the labor market but are underemployed. The under- employment rate is the percentage of individuals who desire to work more among all those who are working an average of less than 8 hours a day. In Sierra Leone, although the official unemployment rate is only 4.3 per- cent, the underemployment rate is 30.9 percent. Almost one-third of all workers would like to work more hours. The rate is higher among men (35.0 percent) than among women (28.1 percent), and higher in Freetown (47.0 percent) compared with rural areas (32.0 percent). While wage workers have less control over their hours and are thus more often underemployed (44.6 percent rate), those self-employed in nonagricultural activities also work less than they desire (36.0 percent rate), likely reflecting weak demand or other constraints to the expansion of business activities. Among the self-employed in agriculture, underemployment is lower (25.6 percent) and it may be due to the seasonality of agriculture. Overall, underemployment is an indication of the low capacity of the labor market to provide more full-time jobs to those who desire them, especially highly skilled people. 21 EMPLOYMENT OPPORTUNITIES Segmentation in the labor market is also seen in the type of employment a worker gets. Not everyone in the workforce has access to the same type of jobs—opportunities vary between wage jobs and self-employment, formal and informal, high-productive and low-productive. In other words, some types of jobs are segmented or correlated with demographic characteristics of the individual. This section identifies an important degree of segmentation in good quality jobs in Sierra Leone, whereas bad jobs are less likely to be segmented. Most important, because of this segmentation, bad jobs predominate in the country. Good jobs are productive employment, mostly in the form of formal wage jobs, but they also include productive and stable self-employment. For sustainable and inclusive growth, individuals must be able to gain productive employment.5 Employment growth generates new jobs for the individual, from wage jobs in all types of firms or from self-employment, usually in micro firms. In turn, productivity growth has the poten- tial to lift the wages of those employed and the returns to the self-employed. After all, in many low-income countries the problem is not unemployment but rather underemployment. Hence, inclusive growth is not only about employment growth but also about productivity growth.6 Moreover, inclusive growth is not only about wage-employment but also about self-employment that provides returns to capital, where land and other assets are needed to generate income. Conversely, bad jobs are unproductive, informal, and associated with fluctuating payments. In Sierra Leone, self-employment provides an example of little to no labor segmentation, whereas wage employment is a case of a segmented labor market. Virtually any individual in any location in the country can be self-employed or employed in the family enterprise. Statistical analysis shows that the probabil- ities of being self-employed, either in agriculture or in a nonagricultural activity, vary relatively little across char- acteristics of individuals, compared with the case of wage job employment. This suggests that self-employment is not segmented. On the contrary, wage employment is associated with a narrow set of location, worker, and firm characteristics, supporting the case of a segmented labor market. To create inclusive job growth, the factors that support that segmentation must be addressed. Characteristics of employment This analysis finds that the majority of jobs in Sierra Leone fall into the second category: the bad quality jobs. The following paragraphs describe the predominant features of jobs in the country. Sierra Leone’s labor force is informal, and its informality is related to low levels of education and skills. Informality is pervasive in Sierra Leone; more than 35 percent of wage jobs and more than 88 percent of nonagricultural self-employment are informal. Approximately 90 percent of laborers work in the informal sector, predominantly in subsistence or other small-scale agriculture. This informality is related to relatively low educational attainment and equally low literacy rates. More than half the WAP (56.7 percent) cannot read or write. A similar proportion have never attended school and among these people, almost all are illiterate. Most workers are self-employed or employed in farming or nonfarming household enterprises. The prevalence of these types of enterprises in rural areas translates into employment rates that are higher in rural areas than in urban areas. In comparison, the employment rates in Sierra Leone are roughly the same as rates across Sub-Saharan Africa, but the share in wage labor is slightly lower than in these other economies. In addi- tion, the share in nonagricultural self-employment in Sierra Leone is slightly higher than in other Sub-Saharan African economies (Margolis et al. 2016). The vast majority of these enterprises have adverse characteristics. They are likely informal and likely provide few workers with regular and steady wages or salaries in exchange for their labor. Also, a nonnegligible proportion of households (26.1 percent) diversify labor across farm and nonfarm self-employment. This suggests that having a job is not enough to achieve adequate income streams. 5 According to the Commission on Growth and Development report (2008), sustained high growth requires rapid incremental productive employment. 6 There is no preconception or bias in favor of labor-intensive industry policies. Indeed, the self-employed poor need improvements in productivity and leveling of the business environment in order to raise their incomes. 22 People engage in multiple jobs to increase revenue when the main job does not yield a desirable level of earnings. Also, diversifying across sectors may result from the need to smooth revenues and mitigate risk when facing demand shortfalls, price volatility, or scarcity of inputs. Although many workers are self-employed, the inadequate remuneration (Figure 28) likely means that they are underemployed—a theme noted in the previous section. Figure 28 Only 1 in 10 people in the workforce holds a wage job DISTRIBUTION OF TOTAL EMPLOYMENT BY JOB TYPE 9.5% 31.3% 59.2% AGRICULTURAL SELF-EMPLOYMENT NONAGRICULTURAL SELF-EMPLOYMENT WAGE EMPLOYMENT Source: World Bank staff elaboration using table 1.2 in Margolis et al. 2016. Farming household enterprises are the largest part of the employment picture in Sierra Leone and the least productive. The agricultural sector provides 61.1 percent of all jobs in Sierra Leone. Nearly all people working in this sector (90.7 percent) are self-employed. Most households (72.8 percent) include at least one household member involved in agricultural activities. In farm self-employment, women represent a larger share of the employed (53.5 percent). Lack of productivity in agricultural household enterprises is exacerbated by low-skill workers. Specifically, the agricultural sector has the least-skilled labor. Educational attainment is lower than among the overall population and the nonfarm self-employed. Most of the agricultural self-employed (80 percent) never attended school, compared with 67.5 percent of the overall WAP and 59.9 percent of the nonfarm self-employed. Nonfarm household enterprises constitute the second-largest source of jobs in the economy. About a third (31.4 percent) of employed individuals have a job in nonagricultural household enterprises. About half of all households (49.6 percent) have at least one member working in nonagricultural self-employment and 37.2 percent of the households report nonfarm activities as their main occupation. Approximately 84.0 per- cent of household enterprises are trader or shopkeeper enterprises; 9.9 percent provide other services; and 6.2 percent are producers. As in the agricultural sector, women make up a majority of the employed in nonfarm household enterprises (63.8 percent). Better jobs A few workers have more opportunities for employment in productive and better-quality jobs. This may be the result of one or a combination of (a) strong sectorial or geographical concentration of opportunities, (b) labor market segmentation, and (c) variation in education and skills. The following analysis draws on descrip- tive and statistical analysis to provide evidence of how each of these factors affects wage and productive employ- ment opportunities. The analysis identifies the workers’ sociodemographic characteristics that are related to the probability of entering such jobs. The few wage employment opportunities available are concentrated across sectors and geographies. About 9.5 percent of the employed population earns a wage, and few subgroups of workers have relatively more access to this type of job (Figure 28). Within industries, the largest shares of wage workers are found in 23 the construction sector (53 percent) and in the mining and extractive industries (45 percent) (Figure 29). Yet these sectors employ fairly small proportions of the employed in Sierra Leone—1.5 percent work in mining and extractives and 1.2 percent in construction. By the same token, wage employment is concentrated in a few regions, to some extent mirroring the location of each industry. In the Western Area, 38 percent of jobs pay a wage (Figure 30). Also, the majority of wage jobs (71 percent) are located in Freetown and other urban areas. In sum, the sectors providing higher opportunities for wage jobs, also associated with higher earnings, make up a small share of total employment in most districts. Also, they seem to target specific groups of workers and employ relatively few of them, thus explaining the limited penetration of wage employment in the labor market. Figure 29 Distribution of the employed in each sector of economic activity by job type, 2014 SERVICES CONSTRUCTION MANUFACTURING AND UTILITIES MINING AND EXTRACTIVE INDUSTRIES AGRICULTURE, FISHING, AND FORESTRY 0 10 20 30 40 50 60 70 80 90 100 PERCENTAGE SHARE OF EACH TYPE OF JOB WITHIN THE ECONOMIC SECTOR AGRICULTURAL SELF-EMPLOYMENT WAGE EMPLOYMENT NONAGRICULTURAL SELF-EMPLOYMENT UNPAID LABOR Source: World Bank staff elaboration using table 1.3, “Job Type, by Sector of Activity,” in Margolis et al. 2016. Some types of workers are, indeed, more likely to be able to obtain formal wage jobs. Formal jobs are considered better than informal jobs primarily because they offer higher earnings on average and, in some cases, come with additional benefits, such as paid leave and medical benefits. The likelihood of working in a formal job is greater among men than among women and increases with educational attainment. Formality also varies by sectors. Formal wage jobs are significantly more common in the service sector. Wage jobs in agri- culture are almost never formal, whereas the self-employed are most likely to be working in the mining sector and least likely to be working in the construction sector. Findings on the determinants of wage employment and earnings levels help identify workers with better chances of entering quality jobs. Although good jobs also include productive and stable self-em- ployment activities, the following quantitative analysis examines only wage jobs because an econometrical model for the determinants of good self-employment was not available. The underlying conclusions and anal- ysis can be found in the report on the 2014 Labor Force Survey (Margolis et al. 2016). That analysis was done in stages. First, an econometric model estimated the probability of job types—wage employee, non­ agricultural household enterprise, agricultural household enterprise, and unpaid contributing family member, controlling for gender, school, disability, age, location, and sector of activity (agriculture; mining and extractive industries; manufacturing and utilities; construction; and services). Second, the Labor Force Study analysis identifies which factors are correlated with current earnings for wage employees (including paid apprentices). The lat- ter analysis includes the following possible determinants of wage: gender, educational attainment, disability status, type of employer (private sector, public sector, or nongovernmental or international organization), sector of activity, age, and geographical location. Wage employment At least some schooling provides better chances to get a wage job. A person who attended primary school without completing it has a higher probability, by 32 percentage points, of getting a wage job than some- 24 Figure 30 Job types by district, 2014 WAGEEMP NONAGSELFEMP AGSELFEMP UNPAID Source: Map ES.1, “Types of Jobs by District,” in Margolis et al. 2016. Note: WageEmp = wage employment; NonAgSelfEmp = nonagricultural self-employment; AgSelfEmp = agricultural self-employment. one who never went to school. Someone who completed postsecondary education has a more than 54 percent- age points better chance of finding a wage job. However, after some years of primary education, the probability of wage employment does not go up by much (Figure 31). That means that, for this sample, the extra levels of education, past a few years of primary level, do not markedly improve the chances of wage employment. Opportunities to get a wage job are more likely in the construction sector. A person who is employed in construction is more likely to be in a wage job than if occupied in any other economic activity. For example, the probability is 21 percentage points higher than in the service sector and 34 percentage points higher than in manufacturing. To a lesser extent, the mining and extractive industries also provide comparatively good pros- pects for wage jobs, followed by the service sector. Moreover, the manufacturing and utilities sector offers the lowest chances to become a wage employee (Figure 32). Men, urban dwellers, and prime-age workers have better chances to find wage employment. On average, men have better opportunities (by 29 percentage points) to become wage employees than do women with otherwise the same demographic characteristics. This – coupled with the issue of teenage pregnancy— might explain why girls drop out school earlier than boys. Also, people living in Freetown and other urban areas have a higher probability of gaining wage employment than the average rural dweller. Individuals in the group ages 30–34 have on average the best prospects of finding a wage job. The youngest workers (ages 15–19) and 25 Figure 31 Change in the probability of being in a wage job, according to education level, compared with an individual who completed post-secondary education 0 RELATIVE PROBABILITY OF FINDING WAGE WORK (%) –10 –20 –30 –40 –50 –60 NEVER WENT TO SCHOOL COMPLETED PRIMARY COMPLETED UPPER SECONDARY INCOMPLETE PRIMARY COMPLETED LOWER SECONDARY Source: World Bank staff calculations using table B.2, “Marginal Effects for Multinomial Logit Model of Job Types,” in Margolis et al. 2016. the oldest workers (ages 60–64) have the worst prospects for this type of job. Possibly, these prospects may explain why the young enroll at a higher rate in school but at the same time stay fewer years than their older cohorts have done in the past. Of youth, 44.7 percent have never attended school, whereas the corresponding share is 75.5 percent among older people. Earnings Women, rural residents, and the young have lower prospects of gaining wage employment. Gender gaps in earnings are stark: holding other characteristics constant, statistics show that men earn nearly three times as much as women in wage employment. The wage job market is also highly segmented by location. Salaries in Freetown are 12 times higher than salaries in the rural areas and 3 times higher than in other urban areas. The wage rate also favors people older than the youngest workers. The younger cohorts earn the least. Moreover, it seems that experience and seniority are compensated accordingly until the age of 59— then the salary rate is reduced for the oldest age group (60–64), who earn less than younger cohorts (ages 25–39). Schooling is relevant to obtaining higher earnings in a wage job, but only for the tails of the distribu- tion across educational attainment. The earning gap between individuals with postsecondary education— Figure 32 Change in the probability of being in a wage job, compared with an individual working in the service sector CONSTRUCTION MANUFACTURING AND UTILITIES MINING AND EXTRACTIVE INDUSTRIES AGRICULTURE –20 –15 –10 –5 0 5 10 15 20 25 RELATIVE PROBABILITY OF FINDING WAGE WORK (%) Source: World Bank staff calculations using table B.2, “Marginal Effects for Multinomial Logit Model of Job Types,” in Margolis et al. 2016. 26 such as technical degrees, certificates, or tertiary degrees—and individuals with less education is prominent in wage employment. Moving from no schooling to some years of primary education nearly doubles the median earnings. This jump is much larger (by almost 7 times) among people with postsecondary education relative to people with complete upper secondary. However, in the middle of the education spectrum, the returns to education do not vary much. There is relatively little variation in earnings between incomplete primary school and completed upper-secondary school. This particular shape of the distribution of earnings seems to create disincentives for the population to engage in higher levels of education. Differences in earnings also depend on the sector in which the job is found. Construction provides the highest remuneration to labor among wage jobs, with an average of 54 percent more than a typical wage job in services. Mining offers the second highest with an average of 42 percent more than services (Figure 33). A wage employee in agriculture receives an average payment that equals 80 percent of the wage in the service sector. The average worker in manufacturing can earn as much as 26 percent more than the average worker in services if the service job is a nonwage job in a household enterprise. However, the manufacturing and util- ities sector pays the lowest among wage jobs. Public sector jobs on average pay less than private sector jobs, controlling for the relatively high levels of schooling that are required. A job in the government pays 18 percent lower than employees in services, and nongovernmental organizations and international organizations pay an average of 7 percent less for a comparable worker. Figure 33 Ratio of earnings relative to an individual working in the service sector 1.8 EARNINGS BY SECTOR AS A RATIO OF 1.6 EARNINGS IN A SERVICE JOB 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 WAGE EMPLOYEE AGRICULTURE MANUFACTURING AND UTILITIES PUBLIC SECTOR MINING AND EXTRACTIVE INDUSTRIES CONSTRUCTION NGO OR INTERNATIONAL ORGANIZATION Source: World Bank staff calculations using table B.3, “Heckman Selection-Corrected Regressions of Log Earnings in Main Job,” in Margolis et al. 2016 Note: NGO = nongovernmental organization. The observed locational variation in earnings does not appear to motivate migration to locations with better job opportunities, such as Freetown. As previously stated, Freetown offers 12 times the average wage in a rural area, while other urban centers triple the rural benchmark. Nevertheless, data from the 2014 Labor Force Survey show that marriage is the main reason for migration across all group ages, varying between 40 and 60 percent of migrants in each age group. The share of those who migrated for work reasons ranges from 5 percent to 27 percent, becoming more relevant as people age. Moreover, a multinomial logit model of the determinants of migration shows that, after controlling for characteristics related to gender, education, the civil conflict, landownership, disability, and age, geographical location does not seem to change much the motivation to migrate for work reasons. A person who resides in Freetown has a 0.26 percentage point higher chance to have migrated for work reasons than a person living in a rural area, while this chance is 0.42 percentage point higher than a rural dweller for people living in urban areas other than Freetown. However, the model finds too that the motivation for migration varies with the level of education. Better schooling and work opportunities are indeed drivers for young people and people with postsecondary education who live in urban areas, especially Freetown. Given the geography of Sierra Leone and the predominance of Freetown as a suitable location for firms, internal 27 migration has potential for improving the jobs situation. So far, labor opportunities at the urban centers are mainly attracting the better educated. HUMAN CAPITAL INVESTMENTS High returns to education in Sierra Leone can be gained, but not at all education levels. Getting an education pays off both in terms of being employed in a wage job and securing relatively high returns. The previous analysis indicates that marginal returns to education increase across levels of educational attainment. Those increases are slim going from a complete primary education to a complete upper-secondary education, but the improvement is significant for people who moved from no schooling to some years of primary school and for those who attained postsecondary education. Even so, educational attainment remains relatively low and is not improving in Sierra Leone. If higher earnings are an indication of higher productivity (and they usually are), why are firms not training their workers more? Why individuals may not invest more in their human capital has two parts: (a) the scarcity of jobs that exist in the formal, wage sector, a condition evidenced in the previous section; and, most important, (b) con- straints to opportunities to fruitfully invest in human capital. To highlight the latter, consider education provision in Sierra Leone. (For a discussion on training and apprenticeships, see the second chapter of Margolis et al. 2016.) Besides overall system flaws that affect educational outcomes, this section identifies particular lagging conditions for women and rural individuals. This attention is consistent with the findings on groups affected by labor market segmentation that hint that closing the gaps in education across gender and location groups may do something for more equal job opportunities. More generally, the overall level of education of the population needs to be signifi- cantly increased to create conditions for rapid skills development to respond to the growing private sector demand. A low level of education characterizes the country’s workforce, and it is considerably lower for women and for those in rural areas. More than half the WAP (56.7 percent) cannot read or write. Almost as many (55 percent) have never attended school, and almost 8 in 10 have attained, at most, a primary education (Figure 34). The average years of education of working-age individuals is 8.7. The proportion of those who have never attended school is greater among women than men (63.7 percent versus 44.9 percent). The urban-rural divide starts with disparities in educational attainment. The gap in average years of education Figure 34 Educational attainment by characteristics of working-age individuals, 2014 0.6 0.4 0.9 0.3 0.4 0.6 0.3 1.0 0.0 100 1.3 1.1 3.5 0.5 1.7 1.2 2.2 1.0 1.7 2.1 3.3 3.8 5.7 3.1 4.8 7.1 8.3 10.0 8.5 6.1 11.5 7.8 9.0 90 9.8 15.7 15.4 12.4 PERCENTAGE OF WORKING-AGE POPULATION (%) 12.7 9.5 12.8 80 16.3 12.5 16.1 12.0 25.3 7.2 14.2 70 14.0 14.1 14.9 21.1 7.7 7.2 18.1 60 16.4 7.6 7.5 7.6 13.3 23.0 50 9.2 8.0 18.4 7.5 40 6.9 70.9 17.1 66.4 63.7 30 57.4 55.2 54.7 44.7 44.9 42.0 7.4 20 33.6 10 17.5 0 OVERALL YOUTH MEN WOMEN DISABLED NOT MIGRANT NOT URBAN OTHER RURAL DISABLED MIGRANT FREETOWN URBAN NEVER WENT TO SCHOOL COMPLETED PRIMARY COMPLETED UPPER SECONDARY TERTIARY DEGREE INCOMPLETE PRIMARY COMPLETED LOWER SECONDARY TECH DEGREES + CERTIFICATES Source: Derived from figure 2.2 in Margolis et al. 2016. 28 between urban and rural residents is 2.4 years (10.2 years versus 7.8 years). In urban Freetown, individuals who have completed upper-secondary school make up the largest share of the WAP (25.3 percent), compared with 15.7 percent in other urban areas and only 3.8 percent in rural areas. Also, the share of those who have never attended school is 17.5 percent in Freetown, compared with 66.4 percent in rural areas. Opportunities to invest and improve human capital are hard to find. Getting a good and complete edu- cation in Sierra Leone is a challenge. The devastating Sierra Leone Civil War that lasted from 1991 to 2002 took the nation’s education system as an early casualty, wiping out 1,270 primary schools and forcing 67 percent of all school-aged children out of school in 2001. More than 15 years later, education in Sierra Leone is still recovering from the destruction caused by the conflict. The literacy rate among 15- to 24-year-olds is below 60 percent, and the total adult literacy rate is even lower, about 43 percent. Secondary school participation is low, with a net attendance ratio from 2008 to 2012 of 39.9 percent for boys and 33.2 percent for girls. Pupils’ learning outcomes are generally very poor at all levels. The Early Grade Reading Assessment (EGRA) results show 87 percent of pupils in second grade were unable to read a short text compared with 40 percent in Gambia, 30 percent in Liberia, and 53 percent in Uganda. Exam performance at the end of secondary school also attests to poor learning. More than a decade of research finds unequivocal the connection between teacher quality and student learning. Only half of teachers are qualified for their level and position. Teacher absenteeism is high (30 percent) and their allocation is inefficient. Moreover, opportunities to invest in education face household and institutional barriers. Among the WAP, the main reason for never attending school is financial constraints (42.2 percent). Other commonly cited reasons include the decision of families not to allow schooling (32.0 percent) and lack of trust in the value of education (16.5 percent), which represent a greater barrier for women than for men. A large and more educated young cohort raises opportunities for a more productive workforce but still bears the hardship of low educational attainment. Youth (ages 15–35) represent the largest share of the overall WAP (66 percent). They are also more educated than their older cohorts. Among young people, 51.8 percent report they can read and write, compared with 22 percent among older people (ages 36–64 age). However, their overall educational attainment could improve markedly. Although the young attain more schooling than older cohorts, still nearly half have no education. Finally, with respect to those who start education, a major wave of school exits occurs between ages 17 and 18 and another takes place between ages 19 and 20, when the share of this age group still in school drops from 34 percent to 18 percent (Figure 35). The vast majority of youth who leave school begin to work, including women. However, even with higher education, a smaller share of the young is employed, relative to older people (52.4 percent versus 81.3 percent). Figure 35 In the transition from school to work, a major wave of school exits occurs between ages 17 and 18, and between ages 19 and 20 100 90 PERCENTAGE OF TOTAL PEOPLE IN THE 80 70 AGE COHORT (%) 60 50 40 30 20 10 0 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 AGE IN SCHOOL EMPLOYED UNEMPLOYED NEET Source: Copy of figure 5.2 in Margolis et al. 2016. Note: NEET = Not in Employment, Education, or Training. 29 The education problems of the past may not be the problems of the future, but change brings oppor- tunities and pitfalls. As just discussed, the younger population has a very different educational profile than the current overall population. The gross primary enrollment ratio and the primary completion rate are quickly picking up during the postwar period. Although Sierra Leone is far from becoming a country with a surplus of educational attainment, the days when more than half of the labor force is illiterate are likely numbered as well. Like the demographic dividend story, this situation brings both opportunities and challenges. The main oppor- tunity is the potential to increase productivity and to move into higher value added production. This opportunity is balanced by the increasing expectations of what types of jobs are appropriate for a more educated person. It is an open question whether agriculture and informal service sector jobs will be able to absorb the labor market entrants at the same rate in a climate of shifted expectations. An especially serious problem that continues to plague education in Sierra Leone is the challenge of girls’ education. The gender gap in access to primary education has been closed, but access to secondary and tertiary education for girls is significantly lagging that of boys. Young women have an average of about 7 months less education than young men (8.3 versus 8.9 years). Relatively high rates of teenage pregnancy are largely reinforcing the high dropout rate among girls. Among young women, 66.5 percent had their first child between the ages of 15 and 19. Working-age women who were teen mothers have 7.8 years of education, compared with the 8.9 years among non-teen mothers. The disparity in human capital between men and women, urban and rural, explains the relatively poor jobs outcomes for women and rural residents. Relatively few workers are trained. Approximately 5.5 percent of the WAP took part in vocational training, seemingly to compensate for a lack of formal education. Most training is undertaken by people who either have never attended school or have started secondary school (Figure 36). This may be consistent to the observation that most of the variation in earnings is at the tails of the distribution. The average spell of training is 2.2 years, but people who had never attained education undergo training for 3.4 years. Not all areas of vocational training are accessible to the least educated workers, and the choice of field is closely determined by the level of formal education obtained before undertaking the training. None of the people who had never attended formal schooling enrolled in training in teaching or business services, whereas 47 percent of these people undertook training in construction and manufacturing, and 41 percent took personal services training (Figure 37). Importantly, the data show that these sectors are addressing the gap between demand and supply of specific labor skills by investing in training the labor force. This is relevant given that construction provides the highest pay in wage jobs and manufacturing also pays relatively well, even in the self-employment segment. Figure 36 Educational attainment before the start of training. NEVER WENT SECONDARY TO SCHOOL COMPLETED 25% 30% PRIMARY INCOMPLETE 10% PRIMARY COMPLETED SECONDARY 4% INCOMPLETE 31% Source: Figure 5.12 in Margolis et al. 2016. SUMMARY AND CONCLUSIONS Inclusive job growth would address disparities across the subgroups that are consistently disadvantaged— namely, the least educated, women, youth, and rural dwellers. Underemployment is high, especially among youth and women. Informal employment remains predominant and is likely growing. Much of the labor force 30 Figure 37 Fields of study by formal educational attainment before vocational training 6% 6% 7% SHARE OF PEOPLE TRAINED IN EACH AREA, BY LEVEL OF EDUCATIONAL ATTAINMENT (%) 11% 0% 11% 0% 1% 0% 4% 1% 1% 2% 1% 0% 7% 19% 4% 41% 47% 14% 42% 48% 22% 17% 47% 45% 45% 27% 19% 1% 0% 0% 1% 2% NEVER WENT TO SCHOOL INCOMPLETE PRIMARY COMPLETED PRIMARY INCOMPLETE SECONDARY COMPLETED SECONDARY AGRICULTURE CONSTRUCTION AND MANUFACTURING PERSONAL SERVICES BUSINESS SERVICES NURSING TEACHING OTHER Source: Figure 2.12 in Margolis et al. 2016. has little training or education. A few workers have productive and adequately remunerative employment oppor- tunities. These opportunities for “good” jobs are few and do not yet reach the poor and other disadvantaged groups sufficiently to qualify as inclusive job growth. Market segmentation limits choices; in turn, the limits lower the productive capacity of the work- force. Despite the aftermath of war and loss of property, the workforce of Sierra Leone is abundant and the majority of those participating are employed. However, the market still presents frictions against enhancing the productivity of workers. Most people find employment by creating household enterprises rather than by taking a wage job. As reviewed in this chapter, they make that choice because of the shortage of wage jobs, their own low educational attainment, and segmentation that marginalizes many from access to the best jobs. Moreover, although self-employment delivers relatively high earnings in specific sectors, productivity is still lagging. In this case, financial constraints coupled with uncertain economic prospects, lack of skills, and segmentation induce families to diversify across different activities rather than invest and specialize in one productive activity. Education is the key missing ingredient for an overall more productive workforce and for better liv- ing standards for everyone. With more than half the WAP having never attended school and almost 8 in 10 individuals in this group having attained, at most, primary education, Sierra Leone’s current labor force can only do so much to find productive, well-compensated employment. Better prospects are needed for the upcoming cohorts entering the workforce. Almost half of the population is below age 15. This sizable youth bulge rep- resents “workers in waiting.” Thus, the present-day challenge is to provide adequate educational opportunities for that group. Subsequently, the future challenge is to answer the forthcoming labor demands that those better-educated children will make upon entering the labor market. At present, graduates of tertiary education are mainly employed in the public sector. An expanded and more diverse set of job prospects for them and for the rest of the qualified youth will be essential. 31 Productive employment is virtually unavailable for most of the workforce, therefore reducing the potential to reap high returns for providing labor in specific sectors. Wage employment is unlikely for workers who have not acquired postsecondary education. Further, wage work is offered mostly in urban areas. Only 3.6 percent of rural dwellers who are employed have a wage job. Thus, people living in rural areas have relatively low incentives to invest in higher levels of education. Furthermore, the construction and the mining industries offer the highest earnings in wage employment, but they offer slim probabilities for finding a job (about 0.65 percent of the employed have a wage job in each sector). The opportunities are also restricted by location, with mining activities mostly available for those living in Bo, Kono, and Kenema (Eastern and Southern provinces), and construction activities for those in the Western Area. Strengthening the productive capacity of household enterprises is the segue into more evenly distributed good job outcomes. A big part of the workforce in Sierra Leone employs itself, not because it is the best work option, but because they lack opportunities for wage jobs and employment in the formal sector. That situation is likely to continue. Often, household enterprises operate with low productivity, financial constraints, and limited human capital, and they are subject to volatile markets. The service sector seems to offer a less-segmented, higher-productive option among household enterprises. Some 26 percent of jobs in the country are already provided by the household enterprises in services. This sector is relatively well paid, and it employs far more of the most skilled workers than any other sector. This illustrates an opportunity to set out the transition toward a workforce that is more evenly spread across activities and toward a more diversified economy overall that reduces the concentration of employment in agriculture and concentration of productivity in mining and construction. 32 DEMAND FOR LABOR AND FIRM PERFORMANCE INTRODUCTION Businesses that operate in Sierra Leone’s difficult environment may have difficulties creating jobs. In the chapter that provided the macroeconomic overview of Sierra Leone, we noted the high levels of macro- economic volatility. Volatility introduces uncertainty. From the perspective of an owner or manager of a firm in Sierra Leone, uncertainty presents sizeable disincentives to invest and hire. Disincentives are most binding for investments in projects that take a relatively long time to gestate and provide returns—such as hiring new staff and training them. Under such volatility, most projects take a short-term outlook. Likely adding to macro- economic uncertainty is the arbitrary nature of law and regulation enforcement, which because of low capacity and governance increases uncertainty and makes it hard for business to grow and hire. The investment climate is also particularly challenging for firms. The World Bank Group ranked Sierra Leone 147th among 189 countries in 2016 for the ease of doing business; the World Bank identified particular challenges in getting electricity, registering property, and trading across borders. The difficulty in moving goods back and forth across the border is likely a missed opportunity given that Sierra Leone is a relatively small market that could get relatively easy access to larger markets. In addition, firms have significant difficulty obtaining credit and must pay comparatively high interest rates. Indications also suggest that corruption is endemic throughout the economy. Firms that are tiny, older, and risk averse may adapt to survive but likely will not grow in this envi- ronment. Currently, job creation and employment in Sierra Leone is generated by a population of firms that is tiny in size, older than one would expect given the average size, and predominantly in the commerce sector. The shocks have been difficult, and although firm entry seems healthy, survival does not appear likely unless the firm is well-established. The poor prospects of survival likely explain the advancing average age of firms and their small size. In addition, the adverse business environment reinforces incentives for firms to stay small and to operate in sectors in which they can exit or retreat quickly. CROSS-COUNTRY COMPARISONS We begin examining the characteristics and performance of firms in Sierra Leone by comparing them with firms in other parts of the world. Cross-country comparisons are attractive, and with good reason. In the absence of any practical criteria for judging optimal performance, cross-country comparisons offer one of the few ways to assess performance outcomes of policy measures and economic processes. However, heterogeneity needs to be taken into account, and care must be used to distinguish between real differences and variations in statis- tics or measurement. Here, comparisons are used to highlight firm-level performance topics that may be most important for Sierra Leone. In this section, formal firms are the unit of analysis across countries. We included firm-level data from other African economies against which to compare Sierra Leone (SLE 2005 and 2016): Burkina Faso (BFA 2008), Cabo Verde (CPV 2014), Lesotho (LSO 2013), Uganda (UGA 2010) and Zambia (ZMB 2010). We also included Afghanistan (AFG 2009) to compare another economy that suffered (and is suffering) from fragility, conflict, and violence. Finally, Bangladesh (BGD 2013) and Peru (PER 2012) are two economies that recently have had relatively high growth rates, are generating many new jobs, and have begun this positive trajectory from a 33 relatively low income base. The firm-level data come from censuses carried out in each country in the year that is shown next to the abbreviated name. The data from Afghanistan, however, come from a representa- tive survey of firms in the country. All statistics and figures, to the extent possible, are for firms in the formal sector only. Firms in Sierra Leone compared In Sierra Leone, employment is predominantly found in small firms. This is a pattern typical of developing countries in general, and in the less-developed countries in Africa in particular. However, even when compared with other African economies, the predominance of small firms in formal sector employment in Sierra Leone is a distinguishing feature. This group provides 60 percent of the formal jobs (Figure 38 (a)). In Bangladesh, also a stand- out when compared with other economies, the same subpopulation of firms comprises just below this employment share. In all other economies studied, the share is well below 50 percent. Large firms (with 100 employees or more) make up barely 20 percent of all employment in Sierra Leone. In stark contrast, in Peru large firms make up close to 90 percent of all employment. Figure 38 Cross-country comparisons of employment shares by size (a) and age (b) A. EMPLOYMENT SHARE BY SIZE 100 80 PERCENTAGE OF FIRMS (%) 60 40 20 0 16 08 3 14 9 12 10 13 10 01 00 20 20 20 20 20 20 20 –2 –2 E– A– V– R– B– D– A– O G SL PE BF ZM CP LS AF BG UG NUMBER OF EMPLOYEES 1–9 10–19 20–99 100+ B. EMPLOYMENT SHARE BY AGE OF FIRM 100 80 PERCENTAGE OF FIRMS (%) 60 40 20 0 16 08 3 9 14 12 10 0 13 01 00 01 20 20 20 20 20 20 –2 –2 2 E– A– V– R– B– A– D– O G SL PE BF ZM CP LS AF UG BG AGE OF FIRM (YEARS) 1 2–5 6–10 10+ Source: World Bank staff calculations. Note: SLE = Sierra Leone; BGD = Bangladesh; CPV = Cabo Verde; BFA = Burkina Faso; ZMB = Zambia; LSO = Lesotho; UGA = Uganda; AFG = Afghanistan; and PER = Peru. 34 In addition, firms in Sierra Leone seem relatively older. There are moderately few new firms while older firms survive (Figure 38, (b)). These two graphs (figure 38) indicate that firms do not grow in Sierra Leone com- pared to firms in other economies. The macro chapter highlighted the volatility to growth in Sierra Leone. The kind of uncertainty that volatility brings, coupled with the access to finance issues listed above, may be the rea- son why firms are not growing in size as they age. The formal sector is mainly composed of firms in commerce. The relatively high share of firms and employ- ment shares in the commerce sector (figure 39), a sector in which entry and exit are comparatively inexpensive, may be yet another indication that growth has been volatile and that firms are concentrated in sectors that allow them to exit easily in case business decline. Figure 39 Cross-country comparisons of share of firms (a) and employment shares (b) by sector A. SHARE FOR FIRMS BY SECTOR 100 PERCENTAGE OF FIRMS (%) 80 60 40 20 0 16 3 10 13 10 14 08 12 9 01 00 20 20 20 20 20 20 20 −2 −2 E− B− D− A− V− A− R− O G SL ZM PE CP BG UG BF LS AF AGRICULTURE UTILITIES TRANSPORT, STORAGE, COMM BUSINESS, FINANCE MINING, QUARRYING CONSTRUCTION HOTELS, RESTAURANTS OTHER SERVICES MANUFACTURING COMMERCE B. SHARE OF EMPLOYMENT BY SECTOR 100 80 PERCENTAGE OF FIRMS (%) 60 40 20 0 6 13 10 08 14 3 10 12 9 01 01 00 20 20 20 20 20 20 2 −2 −2 E− D− B− A− V− A− R− O G SL ZM PE CP BG BF UG LS AF AGRICULTURE UTILITIES TRANSPORT, STORAGE, COMM BUSINESS, FINANCE MINING, QUARRYING CONSTRUCTION HOTELS, RESTAURANTS OTHER SERVICES MANUFACTURING COMMERCE Source: World Bank staff calculations. Note: SLE = Sierra Leone; LSO = Lesotho; ZMB = Zambia; BGD = Bangladesh; UGA = Uganda; CPV = Cabo Verde; BFA = Burkina Faso; PER = Peru; and AFG = Afghanistan. Transport, storage, comm = transport, storage, and communications. In Sierra Leone, although sales appear to be concentrated in fewer firms, employment is not. Bench- marked against comparator countries, Sierra Leone’s firm revenue appears to be highly concentrated at the top 1 percent of enterprises in the sales revenue distribution. This percentile comprises 74.4 percent of all sales among formal firms in the 2016 census. At the same time, the top 1 percent makes up a relatively modest proportion of employment at 15.3 percent. This 5:1 ratio in sales to employment shares for the top 1 percent of firms is much higher than the approximately 2:1 ratio for Afghanistan, Burkina Faso, Peru, and Zambia (Figure 40 (a)). 35 Figure 40 Sales and employment shares and the ratio of output per worker between young and older firms A. SHARE OF TOTAL EMPLOYMENT AND SALES BELONGING TO THE TOP 1% OF FIRMS, BY COUNTRY 100 88% 80 74.4% 69.6% 67.3% 66.6% 64.4% PERCENTAGE (%) 60 58.8% 44.2% 39.3% 38.6% 40 29.9% 29.2% 16.9% 20 15.3% 0 3 10 14 9 08 12 16 01 00 20 20 20 20 20 −2 −2 B− V− A− R− E− O G SL ZM PE CP BF LS AF % EMPLOYMENT % SALES B. RATIO OF THE AVERAGE OUTPUT PER WORKER OF YOUNG FIRMS TO MATURE FIRMS, BY COUNTRY 1.5 1 RATIO .5 0 6 10 14 08 10 9 01 00 20 20 20 20 2 −2 E− A− V− A− B− G SL ZM CP UG BF AF Source: World Bank staff calculations. Note: LSO = Lesotho; ZMB = Zambia; CPV = Cabo Verde; AFG = Afghanistan; BFA = Burkina Faso; PER = Peru; SLE = Sierra Leone; and UGA = Uganda. Young firms = firms ages 1 to 5 years; mature firms = firms ages 6 and over. The concentration of sales may not be the result of better market performance and beating the com- petition. Preliminarily, larger firms in Sierra Leone do not seem to be as efficient as in other economies. Using high-level, aggregate data poses a challenge to providing sufficient evidence to support or reject these hypotheses, but there is some support for the poor competition hypothesis. See figure 40 (b), which shows a ratio of the average output per worker between young (1 to 5 years) and older firms (6+ years). A ratio greater than one means that the average young firm has more per-worker output than the average older firm. If age is a proxy for size (in terms of revenue generated) and output per worker is a proxy for productivity, this may indicate that smaller firms are more productive than larger firms are in Sierra Leone because the ratio is above 1.5. In 36 comparison, for all other economies, this ratio is below 1. Thus, although revenue is concentrated in a few large firms, these firms may not have gained their market share because they were the most efficient and were beating the competition. FIRMS IN SIERRA LEONE Productive firms create good jobs. Firms create jobs, but jobs are a cost for firms. They will not assume these costs if they do not obtain the benefits of increased and more efficient production. To produce productive jobs, firm productivity must also be relatively high and improving as well. In return, productivity grows as jobs become more productive, as new high-productivity jobs are created, and as low-productivity jobs disappear. Productive jobs are associated with better pay, benefits, and more secure and full-time employment. In Sierra Leone, businesses with limited productivity are the main job creators. In Sierra Leone, many people work in very small and not so dynamic economic units. Family farms predominate in agriculture. Outside agriculture, microenterprises and household businesses account for a large share of employment in Sierra Leone. Such businesses make a significant contribution to gross job creation and destruction, but they do not necessarily add to net job creation and productivity growth. Firm performance is often associated with certain firm characteristics, locations, and activities. Factors such as size, age, export orientation, ownership, and location, to name a few, are often correlated with perfor- mance of firms in creating jobs and improving productivity. Unfortunately, these correlations differ by economies; in some economies, big firms create the majority of jobs over a long period of time and in others young, small firms do so. In a subset of economies, a single sector is the job creation machine, whereas in other economies, the same sector is relatively dormant. For this reason, it is important to understand which firms create jobs. This section first provides a description of which firms employ the most workers. Summary statistics and firm characteristics A large population of microenterprises dominates the landscape of firms in Sierra Leone, and this has barely changed over time. This report categorizes firms according to Statistics Sierra Leone’s size classification (See box 3). Micro-sized firms in Sierra Leone (less than 10 employees) comprised 94.0 percent of all firms in 2005 and 91.4 percent of all firms in 2016. Small firms are a distant second and accounted for 4.1 percent in BOX 3: FIRM-LEVEL DATA This report uses data from the two business censuses carried out in 2005 and 2016. The two datasets were col- lected by Sierra Leone’s national statistics body—Statistics Sierra Leone. The two censuses supposedly covered the universe of registered firms that year with a sample size of 10,840 firms in 2005. The data collected in 2016 updated the industry classification according to the International Standard Industrial Classification (ISIC) version 4 and contains 16,603 businesses, according to Statistics Sierra Leone. The data were collected by the four admin- istrative districts: Northern, Southern, Eastern, and Western. The last comprises the Western Urban region (where Freetown is located) and the rural Western Rural region. See the map in annex D. The key limitation with the data is the fact that the information is restricted to formal firms. Firms in the informal sector—a sizeable set of firms in the economy—are excluded in the analysis. Also, these censuses take snapshots at given points in time, hence firms that exit the market are not captured. Finally, these are not panel data; firms are not followed from one dataset to the other. The 2005 and 2016 rounds of the surveys have data on employment, size, firm age, revenue, sector, regional location, and worker compensation (labor costs). Firm size categorization is based on the number of employees: micro (1–9), small (10–19), medium (20–99) and large (100 and above). The main sectors are agriculture/fishing/mining; food production; retail-wholesale; hotels and restaurants; manufacturing; construction; and other services. Firm employment is measured by total employment because that number is available across all the rounds. This is not ideal because it may include some part-time or non-numerated employees. However, a correlation of almost 100 percent between paid workers and total employees shows that these groups are almost identical. 37 2005 and 6.9 percent in 2016. The remaining 2 percent of firms in years 2005 and 2016 were medium and large firms (Figure 41). International comparisons (such as in the previous three figures) also indicate that Sierra Leone has far more of these types of enterprises than in other low-income economies. Figure 41 Sierra Leone’s formal sector remains disproportionately composed of micro-sized firms 15,000 NUMBER OF FIRMS 10,000 5,000 0 2005 2016 NUMBER OF WORKERS 1–9 10–19 20–99 100+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Note: Firm size is determined by the number of employees, with 1–9 being micro, 10–19 being small, 20–99 being medium and more than 100 being large. At the same time, the age distribution of firms shifted to more mature firms between 2005 and 2016. In 2005, approximately 21.9 percent of firms were 1 year of age or less. In 2016, that share dropped to 6.9 percent. The oldest firms, of 30 and more years, comprised 1.2 percent of all firms in 2005 but had increased to 6.3 percent of firms in 2016. The most dramatic differences were for middle-aged firms (between 10 and 19 years in business), which made up 8.2 percent of all firms in 2005 but 26 percent in 2016 (Figure 42). Figure 42 The share of middle-aged firms in Sierra Leone in 2016 increased considerably from 2005 6,000 NUMBER OF FIRMS 4,000 2,000 0 2005 2016 AGE OF FIRM (YEARS) 1 2−5 6−9 10−19 20−29 30+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Note: Firms are considered middle-aged if they have been in business 10 to 19 years. This weak relationship between the age and size of firms in Sierra Leone may indicate inefficiencies in the allocation of resources. In economies that perform well, a firm’s age and size are positively correlated. This correlation indicates that firms grow as they age, holding all other factors constant (location, sector, and economic activity, for example). The underlying explanation is that markets work in ways that efficiently allocate resources to 38 productive firms—firms that survive and thrive for every year in the market. When this relationship breaks down, it could signal allocative inefficiencies that block access to resources (labor and finance) for existing firms to grow. It also may signal allocative inefficiencies that block new firms from emerging. For this reason, firm size is looked at in relation to firm age. The aging of firms overall may indicate a change in patterns of entry and exit. Assuming that the censuses capture well and in the same way the two comparable time periods in the population of formal firms in Sierra Leone, the aging population of firms may indicate a drop off in entry in relation to firm exits. Given difficulties that Sierra Leone went through between these two periods (between 2005 and 2016), it would make sense that firm entry dropped off dramatically. More difficult to understand, however, is why and how older firms survived. Older firms, however, are usually larger, and larger firms may be better able to handle the volatility that shook the economy during those years. As a result, firms already in the market stayed, younger firms were more likely to exit, and new firms were discouraged from entry. This would all lead to a composition of firms that are aging. Another explanation may be that the war halted entry and forced some firms out so these changes in the age distribution of firms could be a return to normalcy. The interpretation that the aging of firms between 2005 and 2016 was caused by limited entry and exit due to shocks in the intervening years may not be the only explanation. It also could be that 2005 was an odd year, in that many firms may have shut down during the war years. If so, young firms were overrepresented in 2005. Thus, the aging of the firms between 2005 and 2016 could represent a return to the norm. As shown in figure 42, many firms fall into the 2–5 and 6–9 years old categories in 2016. Their appearance could suggest that some entry occurred in the intervening years. Entry and exit are important dimensions of a healthy economy. Pinning down the right explanation for this changing pattern in the age distribution of firms is worth understanding. Firms in commerce are many, and they endure in Sierra Leone. The distribution of firms by sector in 2005 and 2016 (Figure 43) looks nearly the same. By far, commerce remains the largest sector in terms of the number of firms. In 2016, agricultural firms are slightly more noticeable than in 2005. Mining, utility, and construction firms are slightly fewer in the more recent period. The shares of services and manufacturing firms remain unchanged. Figure 43 The predominance of firms in the commerce sector did not change between 2005 and 2016 8,000 6,000 NUMBER OF FIRMS 4,000 2,000 0 2005 2016 AGRICULTURE MANUFACTURE SERVICES MINUTILCONSTR COMMERCE Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Note: MinUtilConstr = mining, utilities, and construction. Firms in commerce likely dominate and endure because entry and exit costs are low, making it com- paratively easier to adapt to economic turmoil. Commerce has relatively low fixed costs that are consider- ably easier to recover in case of exit. This makes it a safe haven of activity when economic downturn is looming. On the other hand, businesses in the commerce sector are also relatively easy to start when an opportunity pres- ents itself. In a volatile economy, the commerce sector may be a safe choice to obtain an income and to provide an economic service that consumers in a stressed economy would appreciate. The resilience of this sector is not a surprise, and it may reflect the recent difficulties in the economy in Sierra Leone. 39 EMPLOYMENT, LABOR PRODUCTIVITY, AND WORKER COMPENSATION In the previous section, we examined the basic characteristics of firms in Sierra Leone—age, size, location, and sector—and how these changed between 2005 and 2016. In this section, we use these same dimensions and identify which firms employ more workers, compensate better, and have higher productivity. Then, we explain what those types of firms may indicate about the economy and prospects for better job creation performance. Employment In Sierra Leone, the most recent data indicate that micro firms remain the largest employers. For both 2005 and 2016, most of the wage employment generated came from micro firms, those having between 1 and 9 employees (Figure 44). Given the relatively large share of firms that are of micro scale—well over 90 percent of all firms for both years—that they continue to represent a large share of employment is unsurprising. The uptick in employment from larger firms likely represents the transient upturn in fortunes in the formal mining sector. These data may have recorded the employment response of mining companies to relatively high (but falling) commodity prices. Figure 44 Employment shares by firm size, 2005 and 2016 50,000 AGGREGATE NUMBER OF EMPLOYEES 40,000 30,000 20,000 10,000 0 2005 2016 NUMBER OF WORKERS 1−9 20−49 100−499 10−19 50−99 500+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Older firms employ more workers than before. Older firms (10 to 19 years) made up a larger share of employment in 2016 than in 2005 (Figure 45). The changing contribution to employment coming from older firms reflects changes to the age composition of firms that we pointed out in the previous section. This is poten- tially an important finding7 because young firms are associated with dynamism and job creation (Haltiwanger, Jarmin, and Miranda 2013). Not for this period in Sierra Leone. Incumbent firms may be contributing more to employment because they outlasted newer firms that could not weather difficult times. It is not uncommon for established, incumbent firms to better weather difficult times such as the episodes that the economy of Sierra Leone underwent between 2005 and 2016 (see the case of Russia in Gonzalez, Iacovone, and Subhash 2012). By comparing the age composition in figure 45 for 2005 (left) and 2016 (right), we may see the effect of surviving firms from 2005 in the 2016 data. The bulge of 2- to 5-year-old firms in 2005 may have become the bulge of 10- to 16-year-old firms in 2016 (10 years later). These graphs may suggest that firms are getting older, but they are not growing in terms of number of employees (not creating jobs). 7 This conclusion is based on the multivariate regression found in annex C. 40 Figure 45 Employment shares by firm age, 2005 and 2016 25,000 AGGREGATE NUMBER OF EMPLOYEES 20,000 15,000 10,000 5,000 0 2005 2016 AGE OF FIRM (YEARS) 1 6−9 20−29 2−5 10−19 30+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations Firms in the commerce sector continue to provide most of the employment in the country. Between 2005 and 2016, little changed in the distribution of employment across sectors. If anything, it became more concentrated in the commercial sector. Businesses engaged in commercial activities have employed more than half of the total number of workers in the country (figure 46). On the other hand, the distribution of jobs has become more even across regions (Figure 47). Notably, the share of jobs in the Northern and Eastern regions increased between 2005 and 2016, likely reflecting the emergence of activities different from agriculture in these areas. Figure 46 Employment shares by sector, 2005 and 2016 25,000 AGGREGATE NUMBER OF EMPLOYEES 20,000 15,000 10,000 5,000 0 2005 2016 AGRICULTURE 6−MANUFACTURE SERVICES MINUTILCONSTR COMMERCE Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Note: MinUtilConstr = mining, utilities, and construction. Public enterprises are on average larger than private enterprises, all other things being equal, in terms of number of employees. Using data for 2016, the results of the multivariate regression in annex C show that public enterprises hire a larger number of workers than private firms do. At the same time, addi- tional multivariate regressions indicate that, on average, public enterprises are not more productive than private ones (output per worker) nor do they compensate their workers more (labor costs). The public sector employs the most skilled people—those with postsecondary education (see chapter 2). This pull of the most educated 41 Figure 47 Employment shares by region, 2005 and 2016 20,000 AGGREGATE NUMBER OF EMPLOYEES 15,000 10,000 5,000 0 2005 2016 WESTERN SOUTHERN NORTHERN EASTERN Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. toward public sector employment could possibly create problems for private sector firms that may also need the most-skilled workers but cannot provide the security of government employment. Finally, as suggested in the earlier discussion of labor segmentation, firms with a workforce mainly of women employ fewer workers in general. Controlling for all other factors, average employment is lower among firms that have more women in their workforce. In addition, when the workforce is predominantly female—50 percent or more—within the average firm, employment is on average lower. This is a surprising factor and may indicate a segmentation of formal firms that is not captured by controlling for size, age, sector, or location. Output per worker Output per worker is significantly higher only for the very oldest and the very largest firms.8 In 2005, output per worker was similar among firms 2 years and older. In 2016, the oldest group of enterprises (10 years and more) shows an output per worker that is on average higher than the other age groups (Fig- ure 48). Similarly, output per worker is about the same for micro, small, medium, and even larger firms for 2005. However, in 2016 output per worker for small and medium-sized firms was on average less than for micro firms, whereas only very large firms indicated output per worker that was higher than that of micro firms (Figure 49). For both 2005 and 2016, output per worker for every other sector was higher than in manufac- turing (Figure 50). This result indicates that the manufacturing sector, mostly located in Freetown, is low skilled (see chapter 2), low productivity, and low wage (see next section). This is an important finding because manufacturing may be able to provide the kinds of jobs that the least educated and less skilled, and likely the poorest individuals may be able to fill. Therefore, addressing productivity in the sector will be essential. 8 Output per worker is the most reliable firm-level productivity measure that the business census data from Sierra Leone allow us to estimate. More specifically, we use firm-level annual revenue and divide this by total employment (paid, unpaid, family contributing workers, self- employed, and proprietors). Output per worker is a simple measure of productivity and must therefore be interpreted with some caution. It does help to use this measure while controlling for exogenous factors that can influence productivity, such as size of the firms, location, sector, and age. A multivariate regression is used to determine the correlates of employee productivity. The effort to look at productivity measures is worth it, even with all its warranted caveats. From a micro perspective, persistent differentials in productivity—meaning that some industries expand in terms of production, employment, or both while others shrink—lead to churning in the labor market and to structural changes in the economy. 42 Figure 48 Output per worker (log) by firm age, 2005 and 2016 A. 2005 B. 2016 0.2 0.4 EACH VALUE OF SALES PER WORKER) EACH VALUE OF SALES PER WORKER) DENSITY (PERCENTAGE OF FIRMS AT DENSITY (PERCENTAGE OF FIRMS AT 0.15 0.3 0.1 0.2 0.05 0.1 0 0 −5 0 5 10 15 4 6 8 10 12 14 SALES PER WORKER (LOG) SALES PER WORKER (LOG) AGE OF FIRM (YEARS) AGE OF FIRM (YEARS) 1 2−5 6−10 10+ 1 2−5 6−10 10+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Figure 49 Output per worker by firm size, 2005 and 2016 A. 2005 B. 2016 EACH VALUE OF SALES PER WORKER) 0.2 DENSITY (PERCENTAGE OF FIRMS AT EACH VALUE OF SALES PER WORKER) 0.3 DENSITY (PERCENTAGE OF FIRMS AT 0.15 0.2 0.1 0.1 0.05 0 0 −5 0 5 10 15 0 5 10 15 SALES PER WORKER (LOG) SALES PER WORKER (LOG) NUMBER OF EMPLOYEES NUMBER OF EMPLOYEES 1−9 10−19 20−99 100+ 1−9 10−19 20−99 100+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Figure 50 Output per worker (log) by sector, 2005 and 2006 A. 2005 B. 2016 EACH VALUE OF SALES PER WORKER) EACH VALUE OF SALES PER WORKER) DENSITY (PERCENTAGE OF FIRMS AT DENSITY (PERCENTAGE OF FIRMS AT 0.6 0.4 0.3 0.4 0.2 0.2 0.1 0 0 0 5 10 6 8 10 12 14 SALES PER WORKER (LOG) SALES PER WORKER (LOG) SECTOR SECTOR AGRICULTURE MANUFACTURE SERVICES AGRICULTURE MANUFACTURE SERVICES MINUTILCONSTR COMMERCE MINUTILCONSTR COMMERCE Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Note: MinUtilConstr = mining, utilities, and construction. 43 Firms in the Western region, especially Freetown, indicate significantly higher productivity. For both 2005 and 2016, output per worker was significantly lower for all other regions than in the Western region, except when controlling for the tradeable sector, which can be done only with data in 2016 (Figure 51). Once the data are controlled for the tradable sector, results on the relative productivity of firms in the Western region indicate that the tradable sector is largely located in Freetown. However, the coefficient for the tradable sector does not seem related to higher productivity. Figure 51 Output per worker (log) by region, 2005 and 2006 A. 2005 B. 2016 EACH VALUE OF SALES PER WORKER) EACH VALUE OF SALES PER WORKER) DENSITY (PERCENTAGE OF FIRMS AT DENSITY (PERCENTAGE OF FIRMS AT 0.2 0.4 0.15 0.3 0.1 0.2 0.05 0.1 0 0 −5 0 5 10 15 5 10 15 SALES PER WORKER (LOG) SALES PER WORKER (LOG) REGION REGION WESTERN SOUTHERN NORTHERN EASTERN WESTERN SOUTHERN NORTHERN EASTERN Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculations. Output per worker is lower for firms that comprise a large proportion of female workers. Using data on gender composition for the workforce in 2016, we found that as the workforce becomes increasingly female, the average output per worker decreases. When the workforce is mostly female (50 percent or more), average output per worker goes down compared with firms with a lower proportion of females in the firm-level workforce. This phenomenon may be related to women being, on average, less educated than men (see chapter 2) and having fewer opportunities to be employed in the formal wage sector. It is well worth noting that improving productivity is not a straight line to more jobs, but it is essential to producing good jobs. When productivity goes up, employment can, in fact, go down. Policymakers who seek to avoid job losses should understand the dynamic process of productivity changes and identify who, where, and how these changes will harm some workers, regions, and industries. Although improvements in productivity do not necessarily lead to more jobs, these changes do lead to more productive jobs. For some, increases in productivity represent job losses, but understanding this process will likely help mitigate those negative effects. Worker compensation More productive firms generally provide better worker compensation. Because wages received by workers cannot be directly observed, the measure of workers’ compensation is based on the firms’ reports of labor expenditures.9 In the micro and labor economics literature, worker compensation is said to be deter- mined by the marginal productivity of the worker. The correlation between average worker compensation (labor costs) and productivity (revenue per worker) is significantly positive, as expected (Figure 52). Further, the finding that workers in Sierra Leone benefit from higher firm productivity confirms a narrow and min- imal level of efficiency that exists in the labor market. Unfortunately, given the missing data on other firm inputs (such as capital), it is not possible to provide more conclusive evidence for the presence of allocative inefficiency in the labor market. 9 A significant number of firms reported zero labor costs—2,225 firms in 2005 and 28 firms in 2016. Also, the data on labor costs are missing for 6,995 firms in 2005 and 6,439 firms in 2016. 44 Figure 52 More productive firms tend to better compensate workers in Sierra Leone 12 PREDICTED LABOR COST PER WORKER (LCU) 10 8 6 4 7 8 9 10 PREDICTED OUTPUT PER WORKER (LCU) FITTED VALUES FITTED LINE Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculation. Note: Labor costs include total expenditure of wages and salaries. Output is measured as the revenue of the firm. These amounts are divided by the total number of workers (part time, full time, temporary, and permanent). LCU = local currency unit. As firms age, they do not offer a better compensation to workers. As Figure 53 shows, the density plots for all age groups of firms mostly overlap, except for the 1-year-old firms in 2005. Thus, in 2005 and 2016, age did not determine labor compensation in Sierra Leone. This is consistent with the finding that older firms did not increase output per worker, except for the very oldest firms. Finally, comparing panels a and b exposes a small shift in the distributions to the right, a change that indicates that overall, workers’ compensation increased between 2005 and 2016, but only slightly. Figure 53 Average worker compensation (expressed as cost per worker) by firm age, 2005 and 2016 A. 2005 B. 2016 EACH VALUE OF SALES PER WORKER) EACH VALUE OF SALES PER WORKER) DENSITY (PERCENTAGE OF FIRMS AT DENSITY (PERCENTAGE OF FIRMS AT 0.25 0.3 0.2 0.2 0.15 0.1 0.1 0.05 0 0 −5 0 5 10 15 0 5 10 15 LABOR COSTS PER WORKER, LCU (log) LABOR COSTS PER WORKER, LCU (log) AGE OF FIRM (YEARS) AGE OF FIRM (YEARS) 1 2.5 6.10 10+ 1 2.5 6.10 10+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculation. Note: LCU = local currency units. Excluding the very largest firms, larger firms do not pay significantly higher wages on average than smaller firms pay. Controlling for annual turnover, average worker compensation was significantly lower for larger firms relative to micro enterprises, for both 2005 and 2016. This means that the number of employees is not a likely determinant of worker compensation, but the size of a firm’s turnover is. Only the very largest firms (500+ employees) offered average compensation that was significantly higher (Figure 54).10 This finding 10 This result is only for 2016 data because the 500+ employee size category is not available for 2005. 45 Figure 54 Average worker compensation (expressed as labor costs per worker) by firm size, 2005 and 2016 A. 2005 B. 2016 EACH VALUE OF SALES PER WORKER) EACH VALUE OF SALES PER WORKER) DENSITY (PERCENTAGE OF FIRMS AT DENSITY (PERCENTAGE OF FIRMS AT 0.25 0.4 0.2 0.3 0.15 0.2 0.1 0.1 0.05 0 0 −10 −5 0 5 10 15 –10 0 10 20 LABOR COSTS PER WORKER, LCU (LOG) LABOR COSTS PER WORKER, LCU (LOG) NUMBER OF EMPLOYEES NUMBER OF EMPLOYEES 1−9 10−19 20−99 100+ 1−9 10−19 20−99 100+ Source: Sierra Leone Business Census (2005 and 2016); World Bank staff calculation. Note: LCU = local currency units. is inconsistent with a large body of empirical evidence in Africa (Bloom et al. 2010; Fafchamps and Söderbom 2006), where size is correlated with worker pay. One explanation is that larger firms are better able to fire or screen workers, and hence are more likely to have productive workers relative to smaller firms. Also likely is that productivity is dependent on other factors such as better access to finance. The very existence of large firms (which also tend to be older) suggests that financing constraints are less binding for these firms (Bloom et al. 2010). In Sierra Leone, however, the very oldest firms (10+ years) do indeed report on average higher labor costs and can be interpreted as providing higher worker compensation. Average worker compensation was significantly higher in the commerce sector, for both 2005 and 2016, when compared with the manufacturing sector. This finding is consistent with the relatively lower average productivity of the manufacturing sector highlighted in the previous section. For 2016, average worker compensation was significantly higher for all sectors (agriculture; mining, utilities, and construction; and services) than in manufacturing. Average worker compensation is significantly lower when females make up a larger portion of the firm’s workforce. In 2016, the data indicate that when 50 percent or more of a firm’s workforce is female, average com- pensation decreases. A negative interpretation of this finding would emphasize discrimination and disadvantageous (to females) labor market segmentation. A more positive interpretation may focus on the possibility that this finding results from some flexibility being provided to females in working hours or contracts that cannot be picked up by the data. Unfortunately, without better data, this remains inconclusive. It is worthy of more study and reflection. Average worker compensation varies significantly across regions. For both 2005 and 2016, compensa- tion is higher in the Western region than in all other regions, except for the Southern region in 2016, which displays similar levels as in the Western region. Regional differences in firms’ average worker compensation within Sierra Leone are also quite significant and show significant correlation between urbanization and wage levels. The Western province (where Freetown is located) is also the region where the highest skilled workers reside. In effect, workers sort spatially in Sierra Leone, as they do in other countries, to areas with better labor market opportunities and amenities. The sector with the highest average worker compensation is the service sector, which includes banking, insurance, transportation, and telecommunications. These industries are mostly located in urban areas and populated on average by relatively high-skilled workers. CONCLUSIONS The formal sector has a role in the economic prosperity of Sierra Leone. Chapter 2 highlighted that not everyone has the same access to formal jobs and wage jobs. Job opportunities in informal farm and nonfarm household enterprises are both more available and more inclusive than the opportunities in the formal sector. 46 Thus they are a natural focus track in the journey toward poverty reduction. Yet, the formal sector, too, has a role in the country’s development. Relatively more competent—with a workforce that is better educated and with better access to more resourceful financing and to international markets—formal enterprises have a lead role in sectorial value-added enhancement and in the creation and expansion of value chains. Small as this group is—it employs less than 10 percent of the workforce—the formal sector can spill these gains in economic value over to their informal counterparts. Importantly, given the rapid urbanization and the rapid catch-up in educational attainment, formal businesses have a straight part as future providers of more and better jobs for the well-educated in the urban areas. Firms in the formal sector, however, face a number of challenges. Small firms employ 60 percent of the workers in the formal sector. They also represent 94 percent of this type of enterprise. This chapter found that, in general, firms that age are not growing over time in terms of number of employees. That is, they are not creating jobs. Also, productivity and compensation are not that distinct across different firm sizes and ages. Only the very large firms (100+ employees) and the very old firms (10+ years) are on average more productive and consistently able to offer better compensation. This is not unexpected given the recent severe economic shocks that imposed additional constraints to business growth and employment generation. However, too many of these constraints are based on policies that could be usefully improved. The series of regulations that govern firms from entry to growth to exit are a good example. Firms point to these constraints, such as limited access to finance and poor infrastructure, as impediments to growth. The skill level of workers is low, which puts the onus on firms to train these workers and invest in them. Given the constraints and macroeconomic environment in which these firms operate, it is unlikely that they will be willing to make such investments in their workers any time soon. 47 BIBLIOGRAPHY Africa Development Bank, Organisation for Co-operation and Development, and United Nations Development Programme. 2015. “African Economic Outlook: Sierra Leone.” Abidjan, Côte d’Ivoire: AfDB. Aterido, Reyes, and Leonardo Iacovone. 2015. “Firms Dynamics in Peru: Analysis of Jobs and Productivity.” World Bank, Washington, DC. Biggs, Tyler, Mayank Raturi, and Pradeep Srivastava. 2002. “Ethnic Networks and Access to Credit: Evidence from the Manufacturing Sector in Kenya.” Journal of Economic Behavior and Organization 49 (4): 473–86. Bloom, Nicholas, Aprajit Majahan, David McKenzie, and John Roberts. 2010. “Why Do Firms in Developing Countries Have Low Productivity?” American Economic Review 100 (2): 619–23. Commission on Growth and Development. 2008. The Growth Report: Strategies for Sustained Growth and Inclusive Development. Washington, DC: World Bank Davis, Peter. 2015. “Ebola in Sierra Leone: Economic Impact and Recovery.” Adam Smith International, Nairobi, Kenya. Demographic and Health Survey 2013, Sierra Leone. 2014. Report on 2013 DHS with comparisons with 2008, ICF Inter- national, Rockville, MD. Fafchamps, Marcel, and Bart Minten. 2002. “Returns to Social Network among Traders.” Oxford Economic Papers 54 (2): 173–206. Fafchamps, Marcel, and Måns Söderbom. 2006. “Wages and Labor Management in African Manufacturing.” Journal of Human Resources 41 (2): 356–79. Fu, Ning, Rachel Glennerster, Kristen Himelein, Nina Rosas Raffo, and Suri Tavneet Kaur. 2015. “The Socio-economic Impacts of Ebola in Sierra Leone: Results from a High Frequency Cell Phone Survey—Round One,” World Bank, Washington, DC. Gonzalez, Alvaro S., Leonardo Iacovone, and Hari Subhash. 2012. “Russian Volatility: Obstacle to Firm Survival and Diver- sification.” Policy Research Working Paper 6605, World Bank, Washington, DC. https://openknowledge.worldbank.org/ handle/10986/15828. Hallward-Driemeier, Mary, and Reyes Aterido. 2010. “The Impact of the Investment Climate on Employment Growth: Does Sub-Saharan Africa Mirror Other Low-Income Regions?” Policy Research Working Paper 5218, World Bank, Washington, DC. Haltiwanger, John, Ron S. Jarmin, and Javier Miranda. 2013. “Who Creates Jobs? Small vs Large vs Young.” Review of Economics and Statistics 95 (2): 347–61. Hsieh, Chang-Tai, and Peter J. Klenow. 2009. “Misallocation and Manufacturing TFP in China and India.” Quarterly Journal of Economics 124 (4): 1403–48. Koren, Miklós, and Silvana Tenreyro. 2007. “Volatility and Development.” Quarterly Journal of Economics 122 (1): 243–87. Margolis, David, Nina Rosas, Abubakarr Turay, and Samuel Turay. 2016. Findings from the 2014 Labor Force Survey in Sierra Leone. A World Bank Study. Washington, DC: World Bank. http://dx.doi.org/10.1596/978-1-4648-0742-8. Moore, Winston, and Carlon Walkes. 2010. “Does Industrial Concentration Impact on the Relationship between Policies and Volatility?” International Review of Applied Economics 24 (2): 179–202. Pritchett, Lant. 2000. “Understanding Patterns of Economic Growth: Searching for Hills among Plateaus, Mountains, and Plains.” World Bank Economic Review: 221–50. Santarelli, Enrico, Luuk Klomp, and A. Roy Thurik. 2006. “Gibrat’s Law: An Overview of the Empirical Literature.” In Entre- preneurship, Growth, and Innovation: The Dynamics of Firms and Industries, edited by Enrico Santarelli,. International Studies in Entrepreneurship. Berlin: Springer Science. Strobl, Eric, and Robert J. Thornton. 2002. “Do Large Employers Pay More in Developing Countries? The Case of Five Afri- can Countries.” IZA Discussion Paper 660, IZA, Bonn. World Bank. 2012. World Development Report 2013: Jobs. Washington, DC: World Development Report. ———. 2014. Doing Business 2015: Going Beyond Efficiency. Washington, DC. ———. 2015. World Development Indicators. Washington, DC. ———. 2016. World Development Indicators. Washington, DC. ———. 2017. “Commodity Markets Outlook.” http://www.worldbank.org/en/research/commodity-markets World Health Organization. 2015. Ebola Situation Reports (http://apps.who.int/ebola/ebola-situation-reports). Accessed: 4 September 2015. 48 ANNEX A: FIRM-LEVEL DATA SUMMARY STATISTICS 2005 Region Number of firms Number of employees Employees per firm Eastern 2,042 6,941 3.40 Northern 1,845 7,279 3.95 Southern 1,948 6,686 3.43 Western 4,603 17,556 3.81 Missing 402 1,171 2.91 Total 10,840 39,633 2016 Region Number of firms Number of employees Employees per firm Eastern 2,436 16,940 6.95 Northern 4,090 19,926 4.87 Southern 3,032 12,958 4.27 Western 6,219 22,074 3.55 Missing Total 15,777 71,898 2005 Industry Share of labor (%) Share of revenue (%) Agriculture  2.02  1.26 Industry 20.41 10.99 Services 77.57 87.74 2016 Industry Share of labor (%) Share of revenue (%) Agriculture  3.49  1.11 Industry 21.40 35.96 Services 75.11 62.93  49 2005 Sector Number of firms Number of employees Mean employees Median employees Agriculture 55 719 13.31 12.50 Mining, utilities, construction 69 862 12.87 6.00 Manufacture 1,118 6,416 5.74 4.00 Commerce 7,718 19,044 2.47 2.00 Services 1,340 8,620 6.44 3.00 ≥ISIC 84 540 3,972 7.41 3.00 2005 ISIC4 ê84 industry description Frequency 84 Activities of membership organizations 20 85 Creative, arts and entertainment 4 86 Education 82 87 Human health activities 62 88 Other personal service activities 240 90 Public administration and defense; compulsory social security 1 93 Repair of computers and personal 97 94 Residential care activities 2 95 Social work activities without accommodations 22 96 Sports activities and amusement and recreation activities. 7 NO ISIC 3 2016 Sector Number of firms Number of employees Mean employees Median employees Agriculture 190 1,356 7.17 3.00 Mining, utilities, construction 61 2,212 36.26 3.00 Manufacture 1,096 6,109 5.58 4.00 Commerce 8,613 23,195 2.70 2.00 Services 1,486 6,006 4.05 2.00 ≥ISIC 84 4,331 33,020 7.63 5.00 50 2016 ISIC4 ê84 industry description Frequency 84 Activities of extraterritorial organization 2 85 Activities of membership organizations 26 86 Creative, arts and entertainment 33 87 Education 2,833 88 Gambling and betting activities 5 90 Human health activities 895 91 Libraries, archives, museums and other 6 92 Other personal service activities 371 93 Public administration and defense 13 94 Repair of computers and personal 93 95 Residential care activities 28 96 Social work activities without accommodation 13 99 Sports activities and amusement 13 2005 2016 Employment Mean Median Mean Median Region Western region 3.82 2.00 3.55 2.00 Eastern region 3.43 2.00 4.28 3.00 Northern region 3.95 3.00 4.88 3.00 Southern region 3.40 2.00 6.95 4.00 n.a. 2.91 2.00 Age 1 to 5 3.14 2.00 3.00 2.00 6 to 9 4.07 2.00 3.92 3.00 10 to 19 6.28 3.00 5.70 4.00 20 to 29 5.68 3.00 7.75 5.00 30 plus 8.78 3.00 9.38 7.00 n.a. 7.20 2.00 Size (by employees) 1 to 9 2.50 2.00 3.12 2.00 10 to 19 13.08 12.00 12.41 12.00 20 to 49 28.69 26.00 26.73 24.00 50 to 99 65.88 64.00 69.36 67.00 100 to 499 155.44 112.00 228.29 191.00 500 plus n.a. n.a. 1037.00 685.00 n.a. n.a. n.a. 0.00 0.00 (continued on next page) 51 2005 2016 Employment Mean Median Mean Median Sector Agriculture 13.31 12.50 7.17 3.00 Manufacturing 12.87 6.00 36.26 3.00 Mining, utility, construction 5.74 4.00 5.58 4.00 Commerce 2.47 2.00 2.70 2.00 Services 6.44 3.00 4.05 2.00 n.a. 7.41 3.00 7.63 5.00 Ownership Private 3.55 2.00 3.96 2.00 Public 28.71 8.50 8.67 6.00 n.a. Note: n.a. = not applicable. 2005 2016 Revenue per worker (productivity) Mean Median Mean Median Region Western region 6823.88 694.57 31192.43 7000.00 Eastern region 6550.59 555.65 11725.09 3125.00 Northern region 7965.47 277.83 12480.95 4000.00 Southern region 6231.41 555.65 17573.01 7500.00 n.a. 8142.06 209.90 Age 1 to 5 6286.06 444.52 25425.42 6000.00 6 to 9 9141.09 632.05 17087.87 5000.00 10 to 19 7233.25 555.65 15838.87 5000.00 20 to 29 5364.33 427.42 23116.60 3948.67 30 plus 8001.69 777.91 8982.26 843.75 n.a. 0.00 0.00 19591.16 5000.00 Size (by employees) 1 to 9 7113.34 500.09 21339.00 5666.67 10 to 19 4339.45 232.26 7819.81 1288.69 20 to 49 2366.18 353.22 5592.94 1700.00 50 to 99 4468.32 555.65 5957.50 2549.57 100 to 499 2552.81 1038.59 1808.90 1808.90 500 plus n.a. n.a. 91684.55 91684.55 n.a. n.a. n.a. (continued on next page) 52 2005 2016 Revenue per worker (productivity) Mean Median Mean Median Sector Agriculture 837.99 555.65 20050.52 8875.00 Manufacturing 6708.88 1065.00 29509.84 14125.00 Mining, utility, construction 2267.82 277.83 19169.20 2500.00 Commerce 8202.63 555.65 24970.33 8000.00 Services 4569.27 416.74 24075.63 5000.00 n.a. 5843.13 222.26 7960.15 1333.33 Ownership Private 6904.26 494.53 22411.05 6250.00 Public 4346.62 352.76 5446.66 800.00 n.a. Note: n.a. = not applicable. 2005 2016 Labor costs Mean Median Mean Median Region Western region 27,603.30 1,666.96 171,830.20 10,000.00 Eastern region 20,247.33 694.57 337,489.20 7,000.00 Northern region 13,109.21 722.35 148,567.30 6,000.00 Southern region 35,207.09 1,200.21 62,850.11 15,000.00 n.a. 1,188.36 309.78 Age 1 to 5 23,210.70 866.82 182,828.50 7,000.00 6 to 9 17,368.48 1,541.94 78,873.09 8,924.00 10 to 19 30,835.13 2,010.35 127,890.70 10,000.00 20 to 29 50,877.85 2,222.61 90,727.27 18,000.00 30 plus 22,951.84 4,889.74 383,966.00 10,000.00 n.a. 527,182.80 6,200.00 Size (by employees) 1 to 9 18,191.38 1,000.17 142,107.40 7,500.00 10 to 19 37,691.80 4,045.71 181,082.20 28,000.00 20 to 49 55,440.34 6,667.83 685,018.20 50,000.00 50 to 99 215,780.30 133,356.60 283,218.30 110,014.50 100 to 499 148,129.90 3,471.60 20,200,000.00 45,000.00 500 plus 16,300,000.00 16,300,000.00 n.a. 13,310.00 5,000.00 (continued on next page) 53 2005 2016 Labor costs Mean Median Mean Median Sector Agriculture 15,721.58 2,500.44 26,335.72 10,000.00 Manufacturing 57,011.49 6,001.05 1,435,775.00 14,100.00 Mining, utility, construction 14,718.94 889.04 232,004.50 6,000.00 Commerce 23,439.00 1,186.87 162,211.80 7,000.00 Services 32,939.64 1,183.54 113,920.50 8,000.00 n.a. 21,402.58 1,666.96 181,642.70 12,000.00 Ownership Private 24,862.48 1,200.21 137,464.80 8,000.00 Public 20,277.92 1,363.46 310,262.20 9,737.35 n.a. Note: n.a. = not applicable. 2005 2016 Annual revenue Mean Median Mean Median Region Western region 35,545.54 1,666.96 323,581.40 16,500.00 Eastern region 24,293.67 1,389.13 84,243.30 8,000.00 Northern region 29,783.09 722.35 52,879.36 10,000.00 Southern region 28,837.75 1,278.00 285,179.30 21,000.00 n.a. 22,578.19 555.65 Age 1 to 5 23,716.83 1,049.91 151,293.80 13,000.00 6 to 9 36,246.28 1,666.96 115,960.60 12,000.00 10 to 19 38,452.90 1,666.96 206,258.30 15,000.00 20 to 29 28,127.40 1,972.57 224,476.60 15,377.00 30 plus 108,069.10 2,222.61 399,743.70 5,575.00 n.a. 1,232,395.00 15,000.00 Size (by employees) 1 to 9 25,403.92 1,111.31 135,511.30 12,000.00 10 to 19 60,570.38 2,778.26 105,996.70 16,275.00 20 to 49 86,609.89 8,334.79 968,488.80 38,935.00 50 to 99 261,558.10 27,782.62 2,342,804.00 142,500.00 100 to 499 234,895.00 30,976.69 75,200,000.00 340,000.00 500 plus 173,000,000.00 173,000,000.00 n.a. 34,031.82 5,000.00 (continued on next page) 54 2005 2016 Annual revenue Mean Median Mean Median Sector Agriculture 39,258.68 6,667.83 136,702.60 36,000.00 Manufacturing 91,184.02 9,266.20 7,787,860.00 45,000.00 Mining, utility, construction 21,028.22 986.28 403,741.10 10,000.00 Commerce 28,409.33 1,111.31 140,149.00 15,000.00 Services 34,475.00 1,111.31 193,730.30 12,500.00 n.a. 33,655.46 833.48 167,529.60 6,256.50 Ownership Private 29,035.70 1,111.31 190,908.10 15,000.00 Public 44,502.55 1,600.28 246,270.20 4,257.63 n.a. Note: n.a. = not applicable. PPP conversion factor, GDP GDP deflator (LCU per international $) (PA.NUS.PPP) (base year varies by country) (NY.GDP.DEFL.ZS) Year pa_nus_ppp ny_gdp_defl_zs 2005 890.523 88.8999 2006 971.856 100 2007 1,011 106.796 2008 1,090.63 117.468 2009 1,167.26 126.676 2010 1,351.37 148.447 2011 1,553.14 174.135 2012 1,708.68 195.102 2013 1,797.97 208.613 2014 1,798.09 212.363 2015 2,114.55 252.424 2016 Note: PPP = purchasing power parity; GDP = gross domestic product; LCU = local currency units. 55 ANNEX B: MISSING DATA AND REPORTED ZEROS 2005 2016 Number of Number of Number of firms employees in Number of firms employees in with labor firms with labor with no labor firms with labor costs missing costs missing costs costs missing Region Western region 2,737 7,652 3,136 8,786 Eastern region 1,480 4,370 1,592 4,535 Northern region 1,337 4,727 776 2,677 Southern region 1,190 3,826 935 4,156 n.a. 251 670 0 0 Age 1 to 5 5,469 15,709 3,385 8,457 6 to 9 821 2,574 1,061 3,122 10 to 19 475 2,068 1,433 5,313 20 to 29 169 662 151 691 30 plus 61 232 116 855 n.a. 0 0 293 1,716 Size (by number of employees) 1 to 9 6,710 15,856 6,201 15,145 10 to 19 201 2594 170 2,082 20 to 49 55 1,543 42 1,070 50 to 99 20 1,252 12 808 100 to 499 0 0 2 364 500 plus n.a. n.a. 1 685 n.a. 9 0 11 0 Sector Agriculture 15 149 53 189 Manufacturing 36 329 35 183 Mining, utility, construction 761 3,981 557 2,705 Commerce 5,050 11,219 4,192 9,505 Services 803 3792 750 2,524 n.a. 330 1775 852 5048 Ownership Private 6,975 21,048 6,263 18,029 Public 20 197 176 2,125 n.a. 0 0 0 0 Total 6,995 2,1245 6,439 20,154 Note: n.a. = not applicable. 56 2005 2016 Number of Number of Number of employees in Number of employees in firm reporting firms reporting firms reporting firms reporting no labor costs no labor costs no labor costs no labor costs Region Western region 1,100 3,128 10 29 Eastern region 153 429 18 100 Northern region 265 932 0 0 Southern region 584 1,707 0 0 n.a. 123 347 0 0 Age 1 to 5 1,731 4,641 15 74 6 to 9 267 875 3 9 10 to 19 159 634 3 7 20 to 29 49 298 1 8 30 plus 19 95 2 9 n.a. 0 0 4 22 Size (by number of employees) 1 to 9 2,139 4,793 25 96 10 to 19 64 811 3 33 20 to 49 15 392 0 0 50 to 99 4 217 0 0 100 to 499 3 330 0 0 500 plus n.a. n.a. 0 0 n.a. 0 0 0 0 Sector Agriculture 10 53 2 12 Manufacturing 6 28 1 4 Mining, utility & construction 197 965 0 0 Commerce 1,732 3,865 11 31 Services 201 1,020 5 23 n.a. 79 612 9 59 Ownership Private 2,219 6,397 22 89 Public 6 146 6 40 n.a. 0 0 0 0 Total 2,225 6,543 28 129 Note: n.a. = not applicable. 57 2005 2016 Number of Number of Number of employees in Number of employees in firms with firms with firms with firms with revenue missing revenue missing revenue missing revenue missing Region Western region 2,470 8,100 913 3,075 Eastern region 544 1,590 338 1,940 Northern region 671 2,439 493 3,158 Southern region 631 2,155 663 6,807 n.a. 135 388 0 0 Age 1 to 5 3,640 10,841 848 2,607 6 to 9 383 1,470 301 1,555 10 to 19 285 1,580 703 5,999 20 to 29 104 522 161 1,703 30 plus 39 259 140 1,401 n.a. 0 0 254 1,715 Size (by number of employees) 1 to 9 4,214 9,674 1,990 7,378 10 to 19 154 2,017 348 4,092 20 to 49 51 1,520 45 1,218 50 to 99 23 1,461 15 1,090 100 to 499 0 0 3 517 500 plus n.a. n.a. 1 685 n.a. 9 0 5 0 Sector Agriculture 9 104 17 316 Manufacturing 29 268 15 110 Mining, utility, construction 441 2,254 105 553 Commerce 3,147 7,103 883 2,175 Services 572 3,458 225 1,238 n.a. 253 1,485 1,162 10,588 Ownership Private 4,432 14,428 2,005 11,097 Public 19 244 402 3,883 n.a. 0 0 0 0 Total 4,451 14,672 2,407 14,980 Note: n.a. = not applicable. 58 2005 2016 Number of Number of Number of employees in Number of employees in firms reporting firms reporting firms reporting firms reporting no revenue no revenue no revenue no revenue Region Western region 1,016 3,468 3 12 Eastern region 74 255 14 90 Northern region 221 753 0 0 Southern region 348 987 0 0 n.a. 95 270 0 0 Age 1 to 5 1,413 3,850 12 70 6 to 9 181 746 1 4 10 to 19 109 684 1 8 20 to 29 33 198 0 0 30 plus 18 255 0 0 n.a. 0 0 3 20 Size (by number of employees) 1 to 9 1,666 3,775 14 69 10 to 19 64 814 3 33 20 to 49 15 377 0 0 50 to 99 5 277 0 0 100 to 499 4 490 0 0 500 plus n.a. n.a. 0 0 n.a. 0 0 0 0 Sector Agriculture 5 19 2 12 Manufacturing 8 135 1 4 Mining, utility & construction 149 689 0 0 Commerce 1,342 3356 4 21 Services 183 956 3 15 n.a. 67 578 7 50 Ownership Private 1,748 5,587 11 62 Public 6 146 6 40 n.a. 0 0 0 0 Total 1,754 5,733 17 102 Note: n.a. = not applicable. 59 ANNEX C: MULTIVARIATE REGRESSIONS 60 2005 2016 Output per worker (log) (1) (2) (3) (1) (2) (3) (4) (5) (6) Missing age category=age 1 to 5 Age 6 to 9 0.386* 0.339 0.339 -0.0892 -0.0816 -0.0834 -0.0834 -0.0494 -0.0514 (0.203) (0.217) (0.217) (0.0840) (0.0706) (0.0683) (0.0683) (0.0548) (0.0557) Age 10 to 19 0.300 0.304 0.304 0.0407 0.0605 0.0614 0.0604 0.0205 0.0216 (0.313) (0.350) (0.350) (0.130) (0.0787) (0.0788) (0.0786) (0.0895) (0.0889) Age 20 to 29 0.110 0.245 0.245 0.288*** 0.353*** 0.351*** 0.349*** 0.317* 0.279* (0.340) (0.369) (0.369) (0.0936) (0.0856) (0.0857) (0.0856) (0.177) (0.152) Age 30+ 1.085*** 0.972*** 0.972*** 0.192 0.305*** 0.303*** 0.304*** 0.330** 0.324* (0.299) (0.264) (0.264) (0.120) (0.0937) (0.0956) (0.0956) (0.159) (0.164) Missing size category=1 to 9 employees 10 to 19 employees -0.551 -0.353 -0.357 -0.837*** -0.546*** -0.550*** -0.549*** -0.402* -0.370 (0.414) (0.340) (0.336) (0.195) (0.174) (0.173) (0.173) (0.237) (0.238) 20 to 49 employees -0.337 -0.358 -0.367 -1.062*** -0.867*** -0.883*** -0.882*** -0.605** -0.581** (0.515) (0.384) (0.393) (0.292) (0.238) (0.237) (0.237) (0.234) (0.229) 50 to 249 employees -0.295 -0.557 -0.554 -0.891** -0.827 -0.823 -0.817 -0.785 -0.755 (0.674) (0.616) (0.613) (0.377) (0.597) (0.596) (0.595) (0.575) (0.574) 250 to 499 employees 1.219*** 0.705** 0.679* -1.984*** -2.237*** -2.238*** -2.238*** -2.357*** -2.344*** (0.396) (0.282) (0.379) (0.159) (0.144) (0.143) (0.143) (0.118) (0.121) 500+ employees n.a. n.a. n.a. 3.187*** 2.442*** 2.440*** 2.388*** 2.411*** 2.413*** (0.111) (0.204) (0.205) (0.227) (0.240) (0.235) Missing sector category=manufacturing Agriculture 1.420** 1.423** 1.098*** 1.097*** 1.098*** 1.021*** 1.026*** (0.692) (0.691) (0.148) (0.148) (0.148) (0.0885) (0.0902) Mining, utility, construction 1.991*** 1.938*** 1.569*** 1.569*** 1.623*** 1.444*** 1.461*** (0.377) (0.437) (0.204) (0.205) (0.224) (0.237) (0.232) Commerce 0.949*** 0.948*** 1.016*** 1.013*** 1.013*** 0.908*** 0.895*** (0.230) (0.230) (0.0538) (0.0551) (0.0552) (0.0266) (0.0233) (continued on next page) 61 62 2005 2016 Output per worker (log) (1) (2) (3) (1) (2) (3) (4) (5) (6) Services 0.840** 0.806*** 0.654*** 0.647*** 0.700*** 0.756*** 0.753*** (0.346) (0.285) (0.134) (0.131) (0.0921) (0.0990) (0.0993) Missing location category=western Southern -0.555*** -0.553*** -0.303*** -0.301*** -0.302*** -0.353*** -0.363*** (0.117) (0.119) (0.0645) (0.0647) (0.0642) (0.0637) (0.0647) Northern -1.206*** -1.203*** -0.338*** -0.337*** -0.339*** -0.260*** -0.272*** (0.333) (0.339) (0.0812) (0.0810) (0.0803) (0.0519) (0.0567) Eastern -0.408*** -0.407*** -0.0958* -0.0946* -0.0963* 0.0888*** 0.0864*** (0.142) (0.144) (0.0527) (0.0516) (0.0504) (0.0317) (0.0298) Public n.a. 0.377 0.380 0.327 0.334 (0.445) (0.447) (0.451) (0.458) Tradable -0.0612 0.0973 (0.297) (0.104) ê50% female employees -0.179*** (0.0534) Proportion of female workers -0.164*** (0.0546) Constant 5.687*** 5.373*** 5.434*** 8.891*** 8.143*** 8.143*** 8.047*** 8.155*** 8.139*** (0.175) (0.281) (0.363) (0.159) (0.0612) (0.0606) (0.106) (0.0561) (0.0581) Observations 4,137 4,137 4,137 9,787 9,787 9,787 9,787 5,935 5,928 R-squared 0.006 0.035 0.035 0.014 0.070 0.070 0.070 0.055 0.053 Sector dummies NO YES YES NO YES YES YES YES YES Location dummies NO YES YES NO YES YES YES YES YES Year dummies NO NO NO NO NO NO NO NO NO Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Employment 2005 2016 (log) (1) (2) (4) (5) (6) (1) (2) (3) (4) (5) (6) (7) (8) Mission age category=age 1 to 5 0.145*** 0.156*** 0.156*** 0.149*** 0.108*** 0.132*** 0.132*** 0.130*** 0.130*** 0.109*** 0.102*** 0.128*** 0.125*** Age 6 to 9 (0.0346) (0.0284) (0.0286) (0.0371) (0.0356) (0.0307) (0.0164) (0.0160) (0.0161) (0.0127) (0.0133) (0.0217) (0.0166) 0.330*** 0.276*** 0.276*** 0.219*** 0.194*** 0.246*** 0.216*** 0.217*** 0.217*** 0.202*** 0.206*** 0.198*** 0.179*** Age 10 to 19 (0.0753) (0.0480) (0.0478) (0.0459) (0.0539) (0.0477) (0.0364) (0.0367) (0.0367) (0.0371) (0.0394) (0.0386) (0.0323) 0.330*** 0.244*** 0.243*** 0.197** 0.162* 0.384*** 0.314*** 0.312*** 0.311*** 0.237** 0.249** 0.274*** 0.218*** Age 20 to 29 (0.0834) (0.0607) (0.0602) (0.0785) (0.0814) (0.0911) (0.0878) (0.0859) (0.0859) (0.0963) (0.103) (0.0765) (0.0706) 0.512*** 0.458*** 0.457*** 0.405*** 0.310*** 0.388*** 0.330*** 0.329*** 0.329*** 0.281*** 0.263*** 0.320*** 0.260*** Age 30+ (0.149) (0.119) (0.118) (0.110) (0.107) (0.111) (0.0797) (0.0790) (0.0792) (0.0830) (0.0871) (0.0859) (0.0751) Missing sector category=manufacturing 0.924*** 0.923*** 0.759*** 0.876*** -0.208 -0.212 -0.212 -0.0732 -0.0821 -0.235 -0.315** Agriculture (0.258) (0.258) (0.235) (0.176) (0.158) (0.158) (0.158) (0.181) (0.187) (0.148) (0.134) Mining, utility, 0.556 0.438 0.536 0.429 -0.0914 -0.105 -0.0925 0.115 0.0950 -0.156 -0.273 construction (0.342) (0.350) (0.383) (0.342) (0.215) (0.216) (0.227) (0.328) (0.328) (0.220) (0.191) -0.633*** -0.633*** -0.647*** -0.779*** -0.638*** -0.640*** -0.640*** -0.460*** -0.486*** -0.624*** -0.673*** Commerce (0.124) (0.124) (0.0906) (0.0690) (0.0833) (0.0830) (0.0830) (0.0502) (0.0539) (0.0801) (0.0781) -0.130 -0.207 -0.241* -0.250** -0.416*** -0.424*** -0.411*** -0.220*** -0.233*** -0.439*** -0.465*** Services (0.162) (0.138) (0.127) (0.114) (0.0942) (0.0930) (0.0843) (0.0715) (0.0784) (0.0825) (0.0796) Missing location category=western 0.106** 0.109** 0.0331 -0.0869 -0.234*** -0.233*** -0.233*** -0.0775* -0.0909** -0.247*** -0.197*** Southern (0.0518) (0.0509) (0.0476) (0.0599) (0.0325) (0.0321) (0.0319) (0.0411) (0.0442) (0.0305) (0.0331) 0.294*** 0.297*** 0.243*** 0.132 0.136*** 0.136*** 0.136*** 0.185*** 0.168*** 0.146*** 0.170*** Northern (0.0527) (0.0509) (0.0552) (0.0834) (0.0357) (0.0352) (0.0348) (0.0296) (0.0342) (0.0348) (0.0266) (continued on next page) 63 64 Employment 2005 2016 (log) (1) (2) (4) (5) (6) (1) (2) (3) (4) (5) (6) (7) (8) 0.0763 0.0781 0.0176 -0.0985 0.129*** 0.130*** 0.130*** 0.130*** 0.145*** 0.118*** 0.121*** Eastern (0.0539) (0.0533) (0.0515) (0.0641) (0.0385) (0.0370) (0.0368) (0.0316) (0.0354) (0.0302) (0.0267) 0.427*** 0.428*** 0.434*** 0.449*** 0.396*** 0.326*** Public n.a. n.a. n.a. (0.0745) (0.0751) (0.0794) (0.0811) (0.0588) (0.0577) -0.128 0.0241 Tradable sector (0.105) (0.0684) ê50% female -0.337*** employees (0.0374) Proportion of -0.470*** female workers (0.0607) Normalized Herfindahl 0.513*** 0.0621 index by sector (0.188) (0.108) & time 0.0246*** 0.0879*** Sales (log) (0.00593) (0.0113) 0.752*** 1.152*** 1.279*** 1.203*** 1.340*** 0.790*** 1.326*** 1.325*** 1.301*** 1.411*** 1.459*** 1.316*** 0.501*** Constant (0.0890) (0.141) (0.178) (0.104) (0.0940) (0.0548) (0.0872) (0.0869) (0.111) (0.0523) (0.0601) (0.0915) (0.115) Observations 9,896 9,896 9,896 5,838 4,245 11,104 11,104 11,104 11,104 6,781 6,773 10,038 10,029 R-squared 0.023 0.168 0.169 0.194 0.223 0.030 0.149 0.153 0.153 0.194 0.184 0.152 0.198 Sector dummies NO YES YES YES YES NO YES YES YES YES YES YES YES Location NO YES YES YES YES NO YES YES YES YES YES YES YES dummies Year dummies NO NO NO NO NO NO NO NO NO NO NO NO NO Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 Labor cost per 2005 2016 worker (log) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (6) (7) (8) Missing age category=age 1 to 5 0.330 0.314 0.312 0.351 -0.0353 0.0231 0.0722 0.0710 0.0714 0.0540 0.0531 0.0426 0.0297 Age 6 to 9 (0.222) (0.194) (0.192) (0.251) (0.147) (0.129) (0.114) (0.114) (0.114) (0.154) (0.152) (0.101) (0.0545) 0.376 0.310 0.312 0.183 -0.192 -0.0795 0.0140 0.0156 0.0177 -0.0682 -0.0632 0.0292 -0.0512 Age 10 to 19 (0.296) (0.308) (0.309) (0.362) (0.251) (0.0688) (0.0619) (0.0618) (0.0625) (0.0753) (0.0751) (0.0636) (0.0445) 0.504 0.577 0.577 0.404 -0.114 0.372 0.431 0.431 0.433 0.378 0.356 0.460 0.118 Age 20 to 29 (0.508) (0.413) (0.413) (0.343) (0.210) (0.324) (0.372) (0.373) (0.373) (0.373) (0.357) (0.382) (0.303) 1.137*** 0.848** 0.848** 0.916* -0.0841 0.804*** 0.774*** 0.775*** 0.773*** 0.749** 0.745** 0.792*** 0.338*** Age 30+ (0.342) (0.334) (0.335) (0.459) (0.350) (0.191) (0.178) (0.179) (0.179) (0.290) (0.299) (0.169) (0.0800) Missing size category=1 to 9 employees 10 to -0.0827 0.110 0.109 0.0574 -0.854*** -0.656** -0.439** -0.444** -0.446** -0.211 -0.177 -0.431* -0.911*** 19 employees (0.327) (0.279) (0.279) (0.305) (0.139) (0.264) (0.212) (0.207) (0.205) (0.258) (0.263) (0.214) (0.138) 20 to -0.616 -0.516 -0.539 -0.721 -1.771*** -0.831 -0.691** -0.706** -0.706** -0.254 -0.228 -0.524 -1.285*** 49 employees (0.611) (0.473) (0.475) (0.443) (0.219) (0.515) (0.310) (0.309) (0.309) (0.462) (0.475) (0.370) (0.189) 50 to 0.468 0.177 0.165 -1.706 -2.352** -1.061* -1.017 -1.014 -1.031 -0.869 -0.800 -0.955 -1.942*** 249 employees (1.155) (1.119) (1.117) (1.758) (1.084) (0.617) (0.646) (0.645) (0.651) (0.712) (0.703) (0.831) (0.514) 250 to -2.471*** -2.870*** -2.926*** -2.761*** -3.076*** -3.394*** -3.195*** -3.193*** -3.222*** -3.092*** -3.071*** -3.586** -3.784*** 499 employees (0.631) (0.520) (0.526) (0.682) (0.181) (0.809) (1.059) (1.056) (1.079) (1.121) (1.106) (1.400) (0.706) 1.957*** 1.620*** 1.633*** 1.723*** 2.240*** 2.280*** 1.807*** -3.610*** 500+ employees - - - - - (0.207) (0.308) (0.303) (0.324) (0.433) (0.440) (0.329) (0.391) Missing sector category=manufacturing 0.834 0.829 1.083 0.333 0.724** 0.725** 0.725** 0.524 0.517 0.681** 0.328* Agriculture (0.601) (0.602) (0.640) (0.320) (0.310) (0.310) (0.311) (0.322) (0.329) (0.300) (0.174) Mining, utility, 1.460*** 1.314*** 1.764*** 0.277 1.627*** 1.614*** 1.521*** 1.012** 1.028** 1.590*** 0.564* construction (0.297) (0.392) (0.388) (0.233) (0.355) (0.356) (0.383) (0.441) (0.444) (0.367) (0.322) 1.097*** 1.097*** 0.926*** 0.308* 0.820*** 0.816*** 0.816*** 0.596*** 0.561*** 0.760*** 0.607*** Commerce (0.178) (0.178) (0.220) (0.167) (0.206) (0.207) (0.208) (0.196) (0.189) (0.184) (0.144) 1.063*** 0.974*** 0.902*** 0.286 0.787** 0.778** 0.694** 0.698*** 0.683*** 0.753** 0.577*** Services (0.235) (0.298) (0.230) (0.186) (0.292) (0.289) (0.276) (0.257) (0.251) (0.278) (0.190) (continued on next page) 65 66 Labor cost per 2005 2016 worker (log) (1) (2) (3) (4) (5) (1) (2) (3) (4) (5) (6) (7) (8) Missing location category=western -1.169*** -1.163*** -1.292*** -0.643*** -0.308 -0.307 -0.310 -0.397 -0.409 -0.355 -0.106 Southern (0.260) (0.262) (0.301) (0.128) (0.328) (0.327) (0.329) (0.305) (0.294) (0.323) (0.221) -1.077*** -1.059*** -1.366*** -1.016*** -0.818*** -0.816*** -0.813*** -0.734*** -0.762*** -0.900*** -0.557*** Northern (0.317) (0.343) (0.274) (0.121) (0.0707) (0.0714) (0.0740) (0.0631) (0.0622) (0.0825) (0.0431) -0.0743 -0.0691 -0.485*** -0.430*** -1.021*** -1.022*** -1.020*** -1.070*** -1.077*** -1.045*** -0.930*** Eastern (0.0765) (0.0849) (0.0835) (0.114) (0.183) (0.178) (0.180) (0.227) (0.229) (0.166) (0.148) 0.347 0.337 0.292 0.306 0.318 -0.0172 Public n.a. n.a. n.a. (0.358) (0.344) (0.397) (0.406) (0.371) (0.253) -0.162 -0.167 Tradable (0.321) (0.220) ê50% female -0.800 -0.150** employees (0.689) (0.0681) Proportion of 0.0104 female workers (0.141) Normalized -0.222 Herfindahl index by sector & time (0.357) 0.642*** 0.641*** Sales (log) (0.0507) (0.0285) 5.237*** 4.667*** 4.827*** 5.167*** 0.708 7.876*** 7.514*** 7.513*** 7.679*** 7.640*** 7.586*** 7.592*** 1.174*** Constant (0.197) (0.206) (0.272) (0.261) (0.433) (0.235) (0.214) (0.215) (0.290) (0.188) (0.185) (0.209) (0.277) Observations 1,463 1,463 1,463 1,054 930 5,743 5,743 5,743 5,743 3,716 3,713 5,516 5,516 R-squared 0.010 0.056 0.056 0.074 0.561 0.013 0.078 0.079 0.079 0.061 0.059 0.085 0.435 Sector dummies NO YES YES YES YES NO YES YES YES YES YES YES Location NO YES YES YES YES NO YES YES YES YES YES YES dummies Year dummies NO NO NO NO NO NO NO NO NO NO NO NO Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 ANNEX D: ADMINISTRATIVE MAP OF SIERRA LEONE KOINADUGU BOMBALI KAMBIA PORT LOKO KONO TONKOLILI 1 KAILAHUN MOYAMBA BO 2 KENEMA BONTHE PUJEHUN 1 – WESTERN AREA URBAN 2 – WESTERN AREA RURAL 67