Why so idle? W s nd Emplo m nt in Crowd d L bor M rk t 5th Ethiopi Economic Upd t 5TH ETHIOPIA ECONOMIC UPDATE WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET December 19, 2016 iii TABLE OF CONTENTS ACKNOWLEDGEMENTS................................................................................................................................v LIST OF ABBREVIATIONS............................................................................................................................vii EXECUTIVE SUMMARY.................................................................................................................................ix RECENT ECONOMIC DEVELOPMENTS AND OUTLOOK....................................................................... 1 The Short View...........................................................................................................................................................1 The Long View: Lessons Learned from Monetary Policy in China (1987–2006)......................................................11 The Future View: Challenges and Outlook...............................................................................................................20 TRENDS AND CONSTRAINTS IN URBAN LABOR MARKETS IN ETHIOPIA.................................... 27 Introduction.............................................................................................................................................................27 A Profile of the Labor Market in Urban Ethiopia......................................................................................................29 Labor Market Trends................................................................................................................................................33 Constraints in Urban Labor Markets: Explaining High Unemployment...................................................................37 Short Summary and Recommendations....................................................................................................................43 Annex 1: Ethiopia: Selected Economic Indicators (High Frequency)........................................................................45 Annex 2: Ethiopia: Selected Economic and Social Indicators (Annual Frequency)....................................................46 Annex 3: Self-employment in Urban Labor Markets in Ethiopia..............................................................................49 Annex 4: Changes in the Composition of the Labor Force........................................................................................50 Annex 5: What Explains Real Wage Trends...............................................................................................................52 REFERENCES.................................................................................................................................................... 57 LIST OF FIGURES Figure 1.1: Economic Activity...................................................................................................................................2 Figure 1.2: Monetary Sector......................................................................................................................................5 Figure 1.3: Fiscal Sector.............................................................................................................................................8 Figure 1.4: External Sector......................................................................................................................................10 Figure 1.5: Inflation, Broad Money Growth and the Exchange Rate in China.........................................................16 Figure 1.6: Drought Assessment..............................................................................................................................23 Figure 1.7: Economic Outlook: Selected Projections to 2018..................................................................................25 Figure 2.1: Unemployment and Wages....................................................................................................................30 Figure 2.2: A profile of Urban Labor Markets..........................................................................................................31 Figure 2.3: Three Types in Urban Labor Markets.....................................................................................................32 iv 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Figure 2.4: Trends in Employment in Urban Ethiopia.............................................................................................34 Figure 2.5: A more educated Workforce but little Structural Change.......................................................................35 Figure 2.6: Nominal and Real Wage Trends.............................................................................................................36 Figure 2.7: Real Wage Trends by sector, Sub-sector and Wage Level.........................................................................36 Figure 2.8: Queueing, Temporary Employment and Unemployment......................................................................38 Figure 2.9: Wage Distribution of Workers with Different Education Levels.............................................................39 Figure 2.10: Job Search, Constraints, and Vacancies..................................................................................................41 Figure A1.1: The Trade-off between Wage and Self-employment, 2014......................................................................49 Figure A2: An Increasingly Educated Public Sector.................................................................................................51 Figure A5.1: Oaxaca Blinder Decomposition of the Real Hourly Wage......................................................................53 Figure A5.2: Labor Costs and labor Productivity in the Manufacturing Sector...........................................................55 LIST OF TABLES Table 1.1: Comparison of Monetary Policy Instruments in Developing Countries.................................................18 Table 1.2: Crop Production All Seasons including Commercial Farms (in Million Quintals).................................21 Table 1.3: Livestock Population Growth in 2015/16..............................................................................................22 Table 1.4: Macro-fiscal Outlook Indicators, 2012 to 2018.....................................................................................26 Table 2.1: City-level Migration Rates and Unemployment (marginal effects).........................................................43 Table A5.2: Earnings and Productivity, Regression Results all Cities.........................................................................56 LIST OF BOXES Box 1.1: Monetary Policy Instruments: Evidence from other Developing Countries............................................18 Box 2.1: What does Decomposing Wage Trends Entail?.......................................................................................52 Box 2.2: Measuring Labor Productivity...............................................................................................................54 v ACKNOWLEDGEMENTS T he World Bank greatly appreciates the close Economist). Gelila Woodeneh (Communications collaboration with the Government of Ethiopia Officer) reviewed the document and provided editorial (the Ministry of Finance and Economic content. Boris Balabanov (Online Communications Cooperation, in particular) in the preparation of Associate) designed the report cover. The report this report. The report was prepared by Michael was prepared under the overall guidance of Albert Geiger (Sr. Economist and Co-TTL), Mesfin Girma Zeufack (Chief Economist for Africa), Kevin Carey (Economist and Co-TTL), Ruth Hill (Sr. Poverty (Acting Practice Manager), and Carrie Turk (Country Economist) and Carolina Mejia-Mantilla (Poverty Director). The peer reviewers were: Vinaya Swaroop Economist). Contributors to the report were: Zerihun (Global Lead for Growth and Lead Economist), Dino Getachew (Research Analyst), Steven Pennings Merotto (Lead Economist), and Chandana Kularatne (Research Economist), and Eyasu Tsehaye (Poverty (Sr. Economist). vii LIST OF ABBREVIATIONS CBE Commercial Bank of Ethiopia OMO Open Market Operations CPI Consumer Price Index PBC Peoples Bank of China CSA Central Statistical Agency PPP Purchasing Power Parity DBE Development Bank of Ethiopia PSNP Productive Safety Net Program EDRI Ethiopian Development Research PV Present Value Institute REER Real Effective Exchange Rate EGTE Ethiopian Grain Trade Enterprise RMB Yuan Renminbi FDI Foreign Direct Investment   SOCB State-owned Commercial Bank GDP Gross Domestic Product   SOE State-owned Enterprise GoE Government of Ethiopia SSA Sub Saharan Africa GTP Growth and Transformation Plan TFP Total Factor Productivity IFC International Finance Corporation ToT Terms of Trade ILO International Labor Organization TVET Technical and Vocational Education IMF International Monetary Fund   Training LEAP Livelihood Early Assessment UEUS Urban Employment and Program Unemployment Survey NBE National Bank of Ethiopia VAT Value Added Tax MAD Mean Absolute Deviation y/y Year Over Year ix EXECUTIVE SUMMARY Recent Economic Developments Strong economic growth continued in 2014/15. While the drought slowed down growth in 2015/16 the economy still expanded by 8 percent, which is a sign of increased economic resilience. The Ethiopian economy grew by 8 in 2015/16 due to the recent drought affecting agricultural production with spillovers on the trade sector. Construction and services sectors account for most of the growth from the supply side. On the demand side, growth is driven by investment followed by private consumption. Economic growth over the past years was accompanied by a reduction of unemployment, although it remained high. Urban unem- ployment in the formal sector declined over the last decade, albeit slowly, and was reduced from 23 percent in 2004 to 17 percent in 2015. The urban economy is dominated by the manufacturing, construction, and services sectors with some unskilled labor migration from rural areas. Chapter 2 of this report will analyze these issues in more detail. Low inflation is another sign of economic resilience and inflation is remarkable stable; it stood at 5.6 percent in October 2016. While inflation rose temporarily above ten percent over the past year, it is remarkably stable given the recent drought situation. Inflation entered to double digits in June 2015 and increased to 11.8 percent one year ago in October 2015; since then it has declined and reached 5.6 percent in October 2016. Reasons for the relatively low inflationary impact are related to high levels of cereal stock reserves that were used during the drought and increased wheat imports. The fiscal deficit was virtually unchanged in 2014/15 and 2015/16. The general government fiscal deficit (excluding SOEs) remained modest at 2.4 percent in 2015/16 similar to the preceding year despite spending to finance drought affected areas. An increase in revenue collection, mainly from non-tax sources, compensated the increase in total expenditure and helped to contain the fiscal deficit. An increase in non- tax revenues was the result of larger than expected collection of state dividends from state-owned enterprises and windfall gains to the Fuel Stabilization Fund. On the expenditure side, the general government spend- ing stance was relaxed only marginally in 2015/16 despite additional spending for drought relief and strong capital budget execution. Exports have had their worst performance in the last decade and the current account balance remained large. The chronic current account deficit (including official transfers) continued to deteriorate in 2015/16. The deficit reached 10.4 percent of GDP in 2015/16 improved slightly from 11.5 percent in 2014/15. This was caused by the large imbalance in import and export of goods and services, which has reached to 19.8 percent of GDP. Goods exports were disappointing due to both volume and price effects in 2014/15—but a slight pick-up in volumes again in 2015/16—and an appreciating real (effective) exchange rate. The downward trend in exports now continues over the last four years. Export of goods dropped by 3.7 percent in 2015/16. x 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET This is a good moment to re-consider the monetary policy strategy in Ethiopia. First, inflation has come down to below 10 percent levels after relatively high inflation rates between 2011 and 2013. Current success in controlling inflation will keep inflation expectations at bay in the medium-term. Maintaining low levels of reserve money and broad money growth within a clear monetary policy strategy that outlines appro- priate forward-looking institutional reforms would further cement the public’s view of successful fighting of inflation in Ethiopia. Second, an improved monetary policy framework is a prerequisite for an exchange rate regime geared towards more competitiveness. This is to be able to effectively combat any inflationary pressures arising from a devaluing currency. A look at the Chinese experience of monetary policy making (between 1987 and 2006) in Chapter 1 aims to extract lessons for strengthening the current monetary policy set-up in Ethiopia for such a longer-term and low inflation strategy. A low inflation environment is important to keep real wages stable and ensure that returns on edu- cation in urban labor markets are positive and to avoid the erosion of the purchasing power of wages for workers of all levels of education. Urban labor markets are discussed in detail in Chapter 2. Inflation is an important part of the wage trends observed over time in urban labor markets. High rates of inflation were a factor in the erosion of the value of real wages in the public sector in the past; and it is because of past inflation that nominal wages for public servants had to be adjusted significantly in 2014/15. Given that the public sector is such a major employer this affects the returns on education in urban labor markets, and also seems to have an effect on private sector wages. Economic Outlook and Challenges It was to be expected that Ethiopia’s strong economic growth would slow down in 2015/16 due to the recent drought. The drought caused by the El-Niño phenomenon affected the economy negatively through reductions in food production. While there was some uncertainty about the actual impact of the drought, preliminary actual figures from the Government for 2015/16 show now that the growth rate was 8 percent, in line with. With this, the growth impact is lower than originally envisaged; according to Government crop data released by the CSA in July 2016 there was a less than expected agriculture production impact; this is due to good crop production during the second season harvest of 2015/16. Positive growth moment will still remain. A decade of remarkable double digit growth rates helped the economy to cope well with the most recent challenges encountered in 2015/16. In fact, the ability to keep growth positive is a remarkable achievement for the Government. In similar occasions in the past, for instance in 1997/98 and 2002/03 the country has experienced negative GDP growth. In 1998 growth dropped to –3.5 percent from 3 percent in the previous year. Similarly, in 2003 it dropped to –2.2 percent from 1.5 percent previous year. Medium-term economic growth can be unaffected from the drought since the rains set in normally again in 2016/17. In addition, the completion of the Addis Ababa – Djibouti rail- way line, significantly eases trade logistics related constraints. The commencement of new industrial parks (Hawassa and Bole-Lemi Phase II) and the increasing capacity in power generation with the completion of transmission lines to neighboring countries (Sudan and Kenya) are also expected to improve the export performance and stimulate growth in the short- to medium-term. On the other hand, potential negative economic effects of the current unrest are a risk to the outlook. Economic growth is still expected to drive further reductions in poverty. Despite the poor rains and the associated lower agricultural production, the overall economic growth effect was modest. And growth Executive Summary xi has been an important driver of poverty reduction in the last decade with each percent of growth reduc- ing poverty by 0.55 percent. Based on this, the proportion of households living below the poverty line of US$1.9 purchasing power parity (PPP) is estimated to decline from its estimated level of 27.2 percent in 2015 to 24.6 percent in 2018. Yet, poverty may fall less than predicted. This is due to the fact that the areas most badly affected by the drought are the poorest parts of the country. The impact of the inadequate rains has not been equal across the country. The areas most affected are the poorest parts of the country where many households are poor or live just above the poverty line. Drought-induced harvest losses in these areas cause large increases in poverty. The poverty head-count ratio estimate based on growth elasticity does not take such regional variations into account. When district-specific data on estimated crop losses are used, the rate of poverty reduction is potentially slower. The overvalued real effective exchange rate contributes to the weak export performance. The real effective exchange rate (REER) has appreciated in cumulative terms by 84 percent since the nominal devalu- ation in October 2010. However, the speed of appreciation has slowed down over the past 6 months. While the appreciation between July and August 2015 was 24 percent (y/y), this has slowed to 8 percent in June 2016 (i.e. slowdown in the rate of appreciation). This is primarily the work of two factors: first, a relative decline in the rate of domestic inflation, and second, the depreciation of the U.S. dollar relative to other cur- rencies since January 2016. Since the Birr is pegged against the U.S. dollar the Birr also remained appreciated against other currencies. Still, the Birr remains overvalued, which is hurting international competitiveness. An overvalued currency does not help to improve export competitiveness and is a concern for the economy, especially with exports falling for three consecutive years. Competitiveness also depends on labor productivity. The analysis undertaken in Chapter 2 shows that increasing labor productivity is key not just for competitiveness, but for making urban labor markets work more efficiently. Productivity of unskilled labor is currently very low, close to the minimum caloric require- ments, so that wages cannot fall to clear the urban labor market. A Functioning Labor Market for Poverty Reduction and Competitive Export Sectors Rapid economic growth since 2004 was driven by public infrastructure investment and supported by a conducive external environment; in addition, a modest shift in labor from agriculture to services and construction explains up to a quarter of Ethiopia’s per capita growth over the past decade (World Bank 2015e). But while structural change contributed to economic growth in the past, it was not sufficiently inclusive and needs to contribute much more to poverty reduction in the future. The urban space plays a key role to advance structural change in Ethiopia, as centers of innovation and industrial development. Well- functioning and efficient urban labor markets are a key ingredient for this transformation to take place, and to ensure that its benefits reach all segments of the population. Yet, unemployment in Ethiopia is by and large an urban phenomenon, particularly in Addis Ababa. Increasing the efficiency of urban labor markets in Ethiopia is not only key for structural trans- formation but also for overall economic development. This Economic Update will describe the nature of urban labor markets in Ethiopia and show how several factors contribute to curb their efficiency. Unemployment levels are high, even among those with limited education. This can be explained in part by low labor productivity in the private sector. Low productivity in the private sector translates into very xii 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET low wage levels for the less skilled, preventing wages from adjusting to clear the market. These issues are aggravated by the high costs of searching for a job in Ethiopian cities and the rural-urban migration influx. Finally, there is evidence that there are not enough opportunity for those with primary and secondary education in urban Ethiopia. Understanding the nature of urban labor markets is important for a successful transition to a manufacturing and service-oriented economy and to further reduce poverty. Urban labor markets differ from rural markets in several aspects: in large urban centers, wage employment is more important than self- employment, underemployment is lower but unemployment rates are high (particularly among the young, moderately educated population). Almost half of wage employees are employed by the public sector. The dominant sector in urban areas is the service sector, and it increases in importance among the educated and also in smaller towns. Manufacturing is more important in larger cities and among the unskilled. Many urban labor market trends are moving in the right direction although there has been little change in the structure of urban labor markets over time. Labor participation is increasing, and unem- ployment and underemployment are falling. Ethiopia’s labor force is becoming increasingly educated. Wage employment is becoming slightly more prevalent, particularly in Addis Ababa. The gender gap in labor participation (13 percentage points) as well as the proportion of wage employment in total employment is similar to countries with similar economic development. New jobs being created equally by the private and public sector and in proportion with the existing sectorial composition of the labor force. Real wage trends in urban Ethiopia have not reflected the increasing educational quality of the workforce, although improvements were observed from 2012 to 2014. Wages are higher for those with more education—in 2014 wages were more than double for those with a degree compared to those with a secondary education—but returns on education (changes in worker compensation related to worker characteristics such as education) have fallen over the last decade. The decline in wages at the beginning of the decade was somewhat compensated for, although not fully, by increasing wages in the last two years (2012–2014). These changes may in part reflect changes in worker productivity, but they are also likely a reflection of changes in the nominal public sector wage structure, which affects private sector wages given its importance as an employer in urban labor markets. Unemployment has fallen but remains high. Evidence suggests this is not fully explained by college- educated workers queuing for high quality jobs. There is evidence that while waiting for a high quality job educated individuals take temporary low-skilled jobs to finance the relatively high costs of searching for a permanent job. The temporary jobs that are taken are those that low-skilled individuals would otherwise take contributing to higher unemployment among low-skilled workers. However the magnitude of the crowding out is relatively small and this phenomenon on its own does not explain unemployment entirely. For those with little education, low labor productivity in unskilled jobs causes wages to be low, barely covering basic needs. This prevents wages from falling to fully clear the market. Higher levels of productivity are needed in low-skilled jobs. Within the unskilled labor segment, increases in productivity are desperately needed in order to ensure that the marginal product of labor increases above the nutrition-based wage. This will ensure that wages can adjust as needed in order to clear the market. Although jobs are being created faster than growth in the urban workforce, not enough jobs are being created for those with primary and secondary education. By analyzing how the demand for labor compares to the supply of labor in Addis Ababa, it becomes clear that there are not enough job opportunities for those with primary and secondary education levels. This might have been aggravated by the increasing Executive Summary xiii migration from rural to urban areas. Existing data suggests that individuals living in cities with a higher migration are more likely to be unemployed, although this may simply reflect local economic conditions. There is a need for both faster job creation and investments in skills for non-graduates. More and better job opportunities for those with secondary and primary education are required. This segment of the market is in desperate need of high-productivity employment creation. Although low-skilled workers are more likely to be employed in manufacturing and construction than high-skilled workers, the service sec- tor is still the primary employer of those with little education; growth in the service sector will be needed for job-growth for non-graduates. Investments in skills of those in the job market are also needed. Many employers report delays in finding employees with the right skills and report needing to invest in training. This will likely require investments in on-the-job training as much as in formal training programs. The high costs associated with job searches also contribute to high unemployment rates. Improving the technology used in job searches can help alleviate the high costs. The nature of the job search—search- ing for vacancies posted on physical job boards at specific points in the city—entails high transportation costs for the unemployed, accounting for almost 25 percent of their monthly expenditure. Increasing access to information on job vacancies throughout the city through the use of technology can reduce the cost of searching. Targeted safety nets and labor market programs can also help reduce unemployment by investing in the skills of the unemployed, funding the cost of search, and improving the quality of matching. Policy Recommendations This Economic Update offers five policy recommendations to enhance urban labor markets: 1. Encourage firm creation and firm growth that creates jobs for non-graduates. This will require a focus on growth in services such as hospitality, as well as in construction and manufacturing, areas as these are sectors that are more likely to hire non-graduates. It will also require a focus on interventions that reduce costs associated with hiring non-graduates, such as certification of skills for secondary workers. 2. Increase labor productivity in the low-skill population segment by addressing constraints faced by firms in accessing capital (financial and physical) to ensure that the marginal product of labor increases above the nutrition-based wage. 3. Invest further in job training and technical training programs to build the skills of those in the job market: both for low-skilled workers to increase their productivity and for those with higher levels of education to increase their skill base. 4. Introduce targeted urban safety nets and labor market programs to invest in skills of low-skilled employ- ees and the unemployed, and provide financial support to enable their job search. 5. Enhance the use of ICT to provide information on job vacancies throughout the city and reduce the cost of job search. 1 RECENT ECONOMIC DEVELOPMENTS AND OUTLOOK 1 The Short View When considering the last dozen years since 2004, real GDP growth averaged 10.5 percent. This translated into an average per capita (in dollar terms) growth Strong economic growth continued in 2014/15, but the of 7.9 percent, which is equivalent to the annual drought slowed down Ethiopia’s growth to 8 percent in per capita growth rate needed for Ethiopia to reach 2015/16. Inflation was temporarily back into double digits over the past year—after more than two years in single middle-income status by 2025. digit territory—but declined again to 5.6 percent in October The slowdown in the economic activity is 2016. While tight monetary policy continued, which kept mainly explained by lower agricultural produc- non-food price inflation in single digit, food prices are also on the declining trend. On the fiscal side, the general tion and associated negative spillovers on other government fiscal policy stance continued to be cautious. sectors. The drought caused by the El-Niño phenom- Improvements in tax collection pushed government revenues enon caused lower crop production during the main and grants up and the general government spending stance was relaxed only marginally. Goods exports were (meher) harvest season. Main season grain production disappointing (again) in 2014/15 on account of both volume declined by 1.3 percent (Figure 1.2), however, recovery and price effects while volume increase in 2015/16 did not in the small (belg) season production compensated the help lower export earnings. Exports thereby continue their downward trend seen over the last three years. The real decline and led to the overall crop production growth effective exchange rate continued to appreciate, which hurts of 2.4 percent in 2015/16; still, this is significantly the export performance. lower than the 6.2 percent growth in 2014/15. Yet, actual crop production is much better than what Real Sector was originally expected at the time of the drought. The drought also affected other sectors indirectly, for Strong economic growth continued in 2014/15.1 instance, trade and hotels. While the drought slowed down growth in 2015/16 From the supply side, the GDP growth in the economy still expanded by 8 percent, which 2015/16 is explained by construction and the is a sign of increased economic resilience. After services sectors. The construction sub-sector con- double digit growth in 2014/15, Ethiopia’s economic tributed 2.1 percentage points to growth, while the growth slowed down in 2015/16 due to the recent manufacturing contribution relatively increased to drought. Real GDP grew by 8 percent in 2015/16 0.9 percentage points. Within services, trade and compared to 10.4 percent growth in 2014/15. Still, hotels, at 2.2 percentage points, were the leading sub- Ethiopia’s economy was among the fastest growing sectors to drive GDP growth (Figure 1.1.1). Services in the world showing how well the economy passed growth is also reflected in the growth of the transport through adverse shocks. The growth nevertheless falls and communication sub-sector with a contribution short of the Government’s own target set out in the of 0.7 percentage point of GDP. Ethiopian Airlines Growth and Transformation Plan II (GTP II), which projected at 11.4 percent. Overall, the five year GTP 1 The Ethiopian Fiscal Year ranges over 12 months from July 8 to July 7. This note draws upon official national accounts data produced by I period (2010/11 to 2014/15) achieved a very high the Government of Ethiopia. The growth rates quoted are expressed in growth rate of 10.1 percent per year, on average. factor prices. 2 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 1.1: Economic Activity 1) Real GDP Growth (Supply Side) 2) Crop Production Growth, percent 11 10.3 10.4 9.9 20% 8.7 1.1 1.1 1.5 0.6 9 8.0 15% 0.7 0.6 1.7 3.4 1.1 7 0.5 2.2 10% 3.8 0.7 2.2 0.2 5 1.9 2.2 5% 0.3 0.0 2.2 1.5 0.1 1.3 0.7 3 0.7 0.8 2.1 0% 0.5 3.1 2.3 2.6 0.9 1 2.2 –5% 0.9 –1 –10% 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2011/12 2012/13 2013/14 2014/15 2015/16 Agriculture Manufacturing Construction Other industry Trade & hotel Transport & commun. Other services GDP Cereal Pulses Oilseeds Grains 3) Ethiopian Airlines Activity Growth 4) Real GDP Growth (Demand Side) 40 20% 35 15% 30 25 10% 20 5% 15 10 0% 5 –5% 0 –5 –10% 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2011/12 2012/13 2013/14 2014/15 2015/16 Private Cons Investment Public Cons Net export Cargo Passengers Stat disc GDP 5) Urban Unemployment Rate, percent 25 22.9 20.6 20.4 20 18.9 18.0 17.5 17.4 16.8 16.5 15 10 5 0 Apr-04 Mar-05 May-09 May-10 Mar-11 Mar-12 Jun-13 Apr-14 Apr-15 Source: 1.1; 1.4: MOFED, 1.3: Ethiopian Airlines, 1.2; 1.4: CSA. Recent Economic Developments and Outlook 3 growth continued with passenger traffic up by 19.5 Africa (SSA) average of 11 percent, yet better than and cargo services by 12.4 percent in 2015/16 fol- South Africa (25 percent), Namibia (30 percent) and lowing an expansion of its network and improved Botswana (18 percent). More policies are needed to capacity. (Figure 1.1.3). translate the growth dividend into employment gen- On the demand side, investment followed by eration activities in urban areas. The urban economy private consumption accounted for most of GDP is dominated by the manufacturing, construction, and growth in 2015/16. (Figure 1.1.4). Total investment services sectors with some unskilled labor migration contributed 5.9 percentage points to GDP growth in from rural areas. Chapter 2 of this report will analyze 2015/16, while private consumption growth contribu- these issues in more detail. tion was 1.5 percentage points. The contribution of public consumption increased significantly (1.0 per- Monetary Sector centage point) compared to the previous year which could be a result of the Government’s increased spend- Inflation is remarkable stable given the recent ing to mitigate the effects of the drought. On the other drought and even declining; it stood at 5.6 percent hand, the growth contribution of net export dragged in October 2016. While inflation rose temporarily down growth with –0.8 percentage points of GDP as above ten percent over the past year, it is remarkably a result of lower export earnings against a fast increase stable given the recent drought situation. Inflation in imports. It is noteworthy that the demand side data entered to double digits in June 2015 and increased consists of large statistical discrepancies in 2014/15 to 11.8 percent one year ago in October 2015; since and 2015/16, a factor that was absent in earlier years. then it has declined and reached 5.6 percent in It is not clear what is driving this. October 2016. Food inflation, which constituted Within investment, both private and public about 53 percent of the average household consump- sources drove growth rates in 2015/16. This is evi- tion basket, continued to be a major driver of inflation denced by a variety of indicators: (1) Private invest- at 3.4 percent, down sharply from its three-year peak ment grew by 25 percent despite public investment of 16.2 percent in October 2015 (Figure 1.2.1). The declined by 6 percent.2 (2) Domestic credit stock food price inflation was originally expected to rise (including credit from the Development Bank of sharply due to expected reductions in crop produc- Ethiopia (DBE)) grew by 25 percent in 2015/16. tion arising from the major drought earlier this year. (3) Private transfers (which is a source of private Reasons for the relatively lower de facto inflationary consumption and investment), increased by 0.7 per- impact are related to the stabilization of prices through centage points of GDP in 2015/16, and the level large-scale import of wheat, better balancing of sup- has reached to its historical average (8.3 percent vs. ply and demand through strategic cereal reserves, and a decade average of 8.2 percent). (4) Foreign direct tighter monetary policy (see Chapter 1C for details). investment (FDI) improved increasing from 3.4 to The global slowdown in commodity prices also con- 4.2 percent of GDP in 2015/16 and now reached to tributed to lower inflation of tradable goods. US$3.0 billion. Some food-related commodities still show Economic growth over the past years was some inflationary pressures, and non-food prices accompanied by a reduction of unemployment, although it remained high. Urban unemployment in 2 Data availability limit sectoral decomposition of growth and jobs creation. However, there is evidence from the 2014 Poverty Assessment the formal sector declined over the last decade, albeit that over the past decade every percent economic growth reduced poverty slowly, and was reduced from 23 percent in 2004 to by 0.55 percent. But the recent Systematic Country Diagnostic (SCD) found that advances in shared prosperity are less clear cut because the 17 percent in 2015 (Figure 1.1.5). However, the rate of bottom 10 percent of the population have actually not benefitted from unemployment is large compared to the Sub-Saharan growth and poverty reduction and became poorer from 2005 to 2011. 4 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET increased moderately. There were some commodi- inflation (Figure 1.2.3). Nonfood inflation is usually ties with inflationary tendencies, including increases affected by the lagged impact of monetary policy and in sweets, bread and cereals, meat, dairy products inflation expectations. On the other hand, the real and non-alcoholic beverages, which recorded an deposit rate remained in negative territory while the inflation rate more than 9 percent and above. On the real lending rate tends towards zero, following the ris- other hand, oils and fats and the ‘other foods’ cat- ing overall inflation trend (Figure 1.2.4). egory showed deflationary tendencies over the past But broad money is growing relatively faster, five months through October 2016 (Figure 1.2.2). and the growth of credit to state-owned enterprises In contrast, non-food inflation was around an aver- (SOEs) is the major contributor to that effect. age of 8.0 percent in the last twelve months against Looking at broad money growth, it shows a moderate 7.0 percent in October 2015 (the recent lowest rate). declining pattern from 30 percent in November 2014 The relative slowdown in non-food inflation is poten- to 20 percent in June 2016, though it still remained tially supported by the decline in the global price of high. Net domestic credit growth played a leading role fuel since January 2015 and a lagged effect of tighter in broad money growth; while public sector credit monetary policy over the past year. growth continues to be the main driver (increasing by A low inflation environment is important to 22 percent), credit to the private sector picked up to keep real wages stable and ensure that returns 33 percent in June 2016 but its share remained still on education in urban labor markets are posi- low. Domestic credit to the SOEs increased by 18 per- tive. Urban labor markets are discussed in detail in cent (year-on-year) in June 2016 (Figure 1.2.5). The Chapter 2. Inflation is an important part of wage share of public enterprises in total outstanding domes- trends observed over time in urban labor markets. tic credit slightly moved to 61 percent while the share High rates of inflation were a factor in the erosion of private sector credit is 30 percent at the end of June of the value of real wages in the public sector in the 2016. The share of net central government credit in past; and it is because of past inflation that nominal total domestic credit increased by 2 percentage point wages for public servants had to be adjusted signifi- to 10 percent (Figure 1.2.6) as the money printing cantly in 2014/15. Given the public sector is such a continues in order to finance the budget deficit (see major employer—as will be shown in Chapter 2— next section, Figure 1.3.2), which has a potential to this affects the returns on education in urban labor create inflationary pressure with some lags. markets, and also seems to have an effect on private Keeping inflation low in the future requires sector wages. In addition, private sector wages take continued monetary discipline in a coherent, for- about five months to adjust to rapid increases in ward-looking monetary policy framework. Current prices and there is limited room for wages to fall at success in controlling inflation will keep inflation the low end of the market, so high inflation can result expectations at bay in the medium-term. Maintaining in higher unemployment. low levels of reserve money and broad money growth Tight monetary policy, measured by reserve within a clear monetary policy strategy that outlines money growth, is in line with the NBE annual tar- appropriate forward-looking institutional reforms get and helped to keep non-food price inflation low. would further cement the public’s view of success- Reserve money (the nominal anchor) growth is by and ful fighting of inflation in Ethiopia. Section B of this large consistent with the NBE annual growth target of Chapter will look at the Chinese experience of mon- 16 percent. The reserve money growth (annual aver- etary policy making (between 1987 and 2006) with age of about 10 percent through June 2016) was quite an aim to extract lessons for strengthening the current low compared to the estimated nominal GDP growth monetary policy set-up in Ethiopia for such a longer- (18 percent in 2015/16), which helped to stabilize term and low inflation strategy. Recent Economic Developments and Outlook 5 FIGURE 1.2: Monetary Sector 1) Inflation (y/y, %) 2) Major Food Items Inflation, (y/y,%) 20% 30% 25% 20% 15% 15% 10% 10% 5% 0% 5% –5% –10% Jun-13 Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 Mar-15 Jun-15 Sep-15 Dec-15 Mar-16 Jun-16 Sep-16 0% Jan-13 Apr-13 Jul-13 Oct-13 Jan-14 Apr-14 Jul-14 Oct-14 Jan-15 Apr-15 Jul-15 Oct-15 Jan-16 Apr-16 Jul-16 Oct-16 Bread and Cereals Meat Milk, cheese and eggs Fruit General Food Non-food Vegetables Sugar, jam, honey, chocolate 3) Broad Money, Reserve money & Inflation (%, y/y) 4) Real Interest Rates (%) 43% 20 38% 33% 10 28% 23% 0 18% –10 13% 8% –20 3% –2% –30 –7% –12% –40 Jun-10 Sep-10 Dec-10 Mar-11 Jun-11 Sep-11 Dec-11 Mar-12 Jun-12 Sep-12 Dec-12 Mar-13 Jun-13 Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 Mar-15 Jun-15 Sep-15 Dec-15 Mar-16 Jun-16 Sep-09 Dec-09 Mar-10 Jun-10 Sep-10 Dec-10 Mar-11 Jun-11 Sep-11 Dec-11 Mar-12 Jun-12 Sep-12 Dec-12 Mar-13 Jun-13 Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 Mar-15 Jun-15 Sep-15 Dec-15 Mar-16 Jun-16 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Reserve money In ation (y/y) Broad money (M2) Real maximum lending rate Real deposit rate 5) Broad Money Growth (M2, y/y, %) 6) Composition of Domestic Credit Stock (%) 90% 100% 90% 70% 33 30 28 30 80% 37 36 40 37 37 50% 70% 9 8 10 60% 9 30% 11 50% 21 10% 41 37 32 40% 30% 58 62 64 61 –10% 52 20% 42 –30% 27 29 10% 21 –50% 0% Sep-11 Dec-11 Mar-12 Jun-12 Sep-12 Dec-12 Mar-13 Jun-13 Sep-13 Dec-13 Mar-14 Jun-14 Sep-14 Dec-14 Mar-15 Jun-15 Sep-15 Dec-15 Mar-16 Jun-16 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Net Foreign assets Net Domestic credit SOEs CG credit Private credit Other items (net) Broad money (M2) Source: 2.1–2.2: CSA, 2.3–2.4: CSA and NBE, and 2.5–2.6 NBE. Note: 1.6: Monetary survey data is used, which excludes DBE in private credits. 6 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Fiscal Sector drought, non-metallic minerals (–24 percent) as a result of low demand for cement from the construction sec- In FY 2015/16, the general government fiscal policy tor, and machine and spare parts (–36 percent) due stance continued to be cautious. The general govern- to low foreign exchange allocations that constraint ment fiscal deficit (excluding SOEs) remained modest input imports. Meanwhile, external grants declined by at 2.4 percent in 2015/16 similar to the preceding 0.2 percentage points of GDP in 2015/16. year despite additional spending to finance drought The general government spending stance was affected areas. The deficit level is far lower than the relaxed only marginally in 2015/16 despite addi- government target of 3.0 percent. An increase in rev- tional spending for drought relief. Total expenditure enue collection, mainly from non-tax sources, com- increased from 17.8 to 18.4 percent of GDP from pensated the increase in total expenditure and helped 2014/15 to 2015/16. This is the result of increased to contain the fiscal deficit (Figure 1.3.1). Looking at in both recurrent expenditure and capital spending. the financing side, the deficit was covered by exter- In the middle of the fiscal year, the federal govern- nal and domestic borrowing (1.7 and 1.6 percent of ment approved Birr 18 billion (or US$0.85 billion, GDP, respectively), and through the repayment of or 1.2 percent of GDP) additional budget, of which cash balances and residuals (totaling 1.0 percent of 82 percent (or US$0.7 billion, or 1.0 percent of GDP). A large portion of domestic financing relied GDP) was devoted to support the rural population on borrowing from non-bank sources through the sale affected by the drought. The additional budget is of T-bills. Direct advances issued from the National financed mainly from the Fuel Stabilization Fund Bank to the central treasury slowed down to 1.1 per- (to the tune of Birr 12.5 billion) and the remain- cent GDP in 2015/16 from 1.5 percent of GDP in ing comes from profit and residual surplus of public 2014/15 (Figure 1.3.2). financial institutions. Under recurrent spending, pro- Improvements in non-tax revenue collection poor and other sectors (excluding defense and debt pushed government revenues and grants up to 16.0 service) increased by 0.2 percentage points of GDP. percent of GDP in 2015/16. This is an increase of On the other hand, the capital spending-to-GDP ratio 0.6 percentage points from 15.4 percent in 2014/15 rose by about 0.4 percentage points to 9.4 percent in (Figure 1.3.3). The general government revenue per- 2015/16. Capital expenditure accounted for more formance showed improvement mainly because of an than half of total general government expenditure but increase in non-tax revenues. This increase was 1.1 per- showed a relative increase in most of the economic and centage points of GDP in 2015/16 and the result social sector spending. Among other things, agricul- of larger than expected collection of state dividends ture, natural resources/water, and education spending from state-owned enterprises and windfall gains to the increased by 0.3 percentage points of GDP each while Fuel Stabilization Fund3 (in the context of declining urban development spending increased by 0.2 percent- oil prices since January 2015). Collection from for- age points. Yet, road construction spending declined eign trade taxes showed no change to 2014/15 at 4.1 by 0.6 percentage points of GDP in 2015/16. while direct taxes increased by 0.1 percentage point to 4.7 percent of GDP. Domestic indirect taxes, however, 3 The Fuel Stabilization Fund is a resource collected from the price dif- ferential between the global oil price and domestic prices of fuel. However, declined by 0.4 percentage points of GDP mainly on there were cases in which the domestic fuel price was kept low relative account of low outturn of VAT collection from locally to global price and the Fund subsidizes through borrowing from the Commercial Bank of Ethiopia (CBE). Currently, the Fund has benefit- manufactured goods (declined by 16 percent). Products ted from global oil prices declines and has accumulated a large sum of that contributed to the declined VAT collection include money, which is not fully passed through to consumers. The Fund was utilized to cover part of the cost for civil salary increases (Birr 3.0 billion) fuel products (–5 percent) on account of low oil prices, in 2014/15 and to finance part of the supplementary budget to support cotton products (–16 percent) possibly linked to the drought affected areas (Birr 12.5 billion) in 2015/16. Recent Economic Developments and Outlook 7 Generally, the gap between recurrent and a weak external environment; and the supply shock capital spending has narrowed further since last from the drought meant foregone agriculture produc- year to a now almost equal distribution. The gap tion to export as well as additional food imports. As a has narrowed down to 8.9 vs. 9.4 percent of GDP in result, debt sustainability indicators have deteriorated 2015/16 compared to 7.4 vs. 10.1 percent of GDP in on account of poor export performance over the past 2013/14, respectively (Figure 1.3.4). The public sec- three years and faster-than-anticipated disbursements tor salary increases contributed to the adjustment of of non-concessional loans, contracted during FY13 and the balance between capital and recurrent expenditure FY14. The present value (PV) of debt to export ratio where low levels of recurrent expenditure raised some continued to deteriorate in FY 2015 (Figure 1.3.6); concern over the public sector efficiency, especially in addition, the indicator that shows the debt servic- in regions. Keeping the balance between recurrent ing capacity in FY16 came very close to the policy and capital budget is important to meet the running dependent threshold. To enhance debt sustainability, it cost of additional capital expenditure of public sector remains essential to promote the growth and diversifica- projects, to fund the operations of the exiting produc- tion of exports. Ensuring an appropriate pace of public tive assets and to ensure effective service delivery in borrowing—especially from external, non-concessional general. Evidence showed that Ethiopia’s r-coefficient4 sources—is also critical to ensuring that public invest- was falling over time indicating underfunding of recur- ment does not undermine debt sustainability. rent costs, which raised concern of the sustainability of public sector services [PER 2015]. External Sector Looking forward, the federal government bud- get deficit5 envisages a marginal improvement to 3.3 The chronic current account deficit continued to be percent of GDP in 2016/17.6 The expenditure budget high in 2015/16 mainly due to trade imbalances, but increase is similar to the increase in revenues as a share showed a slight improvement. The current account of nominal GDP and as a result the deficit remained deficit (including official transfers) reached 10.4 per- the same as last year. Federal government revenue and cent of GDP lower than 11.5 percent in 2014/15. This grants reached 11.9 percent of GDP. Total federal gov- was caused by the large imbalance in import and export ernment expenditure increased to 15.2 percent of GDP. of goods and services, which registered a deficit of 19.8 The federal recurrent spending budget accounted for percent of GDP (Figure 1.4.1). The trade deficit was 25 percent while the allocation for federal capital spend- driven by poor goods & services export performance ing budget constituted 39 percent and the remaining (dropped by 4.1 percent) and goods and services 36 percent was allocated towards the regional subsidy. imports (increased by 2.9 percent). The trade deficit The Federal government budget deficit (3.3 percent was partially offset by private transfers, one of the larg- of GDP) is expected to be financed through domestic est external resources for Ethiopia that increased from financing of 2.0 percent of GDP and external borrow- 7.6 to 8.3 percent of GDP, official transfers continued ing (mainly concessional) of 1.4 percent (Figure 1.3.5). to decline from 2.3 to 1.4 percent of GDP in 2015/16 Ethiopia’s external debt risk remained at ‘moder- due to fiscal stress in developed countries. The current ate’ while vulnerabilities increased in 2015/16. The account deficit (10.4 percent of GDP) was financed vulnerabilities of the debt risk arose from the worsening in the fundamentals related to the drought —despite 4 r-coefficient is a measure of the annual real O&M expenditure required the government’s ability to successfully cope with the per unit of additional capital expenditure on a public sector services drought situation—and the difficult global price envi- project or programs. 5 This excludes the regional budgets from own sources. ronment for key commodity exports. Exports contin- 6 Applying a nominal GDP growth of 17.8 percent for FY17, which is ued to underperform relative to projections owing to an estimate by IMF(IMF 2016 Article IV report). 8 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 1.3: Fiscal Sector 1) General Gov’t Deficit, % of GDP 2) Direct NBE Advances to GoE, % of GDP 20 19 0 1.6 18 18 18 18 1.5 18 17 17 17 17 16 1.4 15 16 15 –1 16 15 –0.9 1.2 1.1 1.1 14 1.0 –1.2 –1.2 –1.2 –1 12 1.0 0.8 10 –1.6 –1.6 –1.6 –2 0.8 8 –1.9 –2.0 –1.9 0.6 0.6 –2.2 –2 6 –2.4 –2.4 0.4 0.3 4 –2.6 –3 2 0.2 0 –3 0.0 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Revenue and grants Overall balance incl. grants (RHA) Total Expenditure Primary de cit (RHA) 4) General Government Spending, % of GDP 3) General Government Revenue, % of GDP 20 18.8 18.2 17.8 17.8 18.4 15 14.3 14.4 15.2 18 17.5 14.0 13.8 13.8 16.6 13.4 2.7 16 2.0 1.2 1.6 12 2.7 2.0 2.3 14 10.4 9.4 10.3 9.0 12 10.5 10.1 4.3 4.1 4.1 9.8 4.4 9 4.6 4.5 10 4.6 8 6 3.8 4.0 3.7 6 3.7 2.8 3.0 3.1 8.4 8.7 8.9 4 7.9 6.9 7.2 7.4 3 2 3.9 3.8 3.9 4.2 4.4 4.6 4.7 0 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 0 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Recurrent Capital Total Expenditure Direct tax Domestic indirect tax Foreign trade tax Non tax revenue Domestic Revenue 6) PV of Debt-to-Export ratio DSA 2015 DSA 2016 5) FY17 Federal Budget, % of GDP 350 c. PV of debt-to-exports ratio 350 c. PV of debt-to-exports ratio 20 325 325 300 300 15 15 15 275 275 15 14 250 250 11 11 11 12 225 225 10 200 200 150 150 125 125 5 100 100 75 75 –3.1 –3.0 –3.4 –3.3 50 50 0 25 25 0 0 2015 2020 2025 2030 2035 2016 2021 2026 2031 2036 -5 Total Total Overall de cit revenue spending incl grants Baseline Historical scenario 2013/14Bdg 2014/15Bdg 2015/16Bdg 2016/17Bdg Most extreme shock Exports Threshold Source: 3.1–3.5: MOFED, 3.6: WB-IMF DSA. Recent Economic Developments and Outlook 9 largely by external borrowing especially by SOEs and trend, it is important to note that the high growth rate improved FDI inflows (4.2 percent) as well as draw- between 2005 and 2009 was to a large extent due to down of the foreign exchange reserve accumulations an extraordinary increase in sesame (oil seeds) export of 1.3 percent. Foreign exchange reserves reached to in 2009 of 230 percent. Most recently, overall export US$3.5 billion at the end of August 2016 supported values to China have declined by 52 percent y/y in the by a central bank deposit from Saudi Arabia of US$1 first half 2015/16 (vs. the first half 2014/15). billion in December 2015 (Figure 1.4.2). But the exter- Continued growth in goods imports contrib- nal sector is still vulnerable to terms of trade shocks, uted to the deteriorating current account balance in which need to be carefully managed. 2014/15, but the rate of increase as slowed markedly Goods exports were disappointing (again) in in 2015/16. Imports of goods increased by 20 percent 2015/16 thereby continuing their downward trend in 2014/15, but growth has slowed to only 1.6 per- seen over the last four years as a result of price effects cent in 2015/16. The primary driver in 2015/16 was while volume recovered moderately. Export of goods the 17 percent increase in consumer goods imports dropped by 3.7 percent in 2015/16 as a result of a drop while capital goods imports stagnated (declined by by 9.6 percent in the export price index, while the 1 percent) that are associated with large public sector volume index increased by 5.4 percent (Figure 1.4.3). infrastructure investment activities. In terms of share Service export declined by 4.5 percent primarily due in GDP, capital goods imports accounted for 9.4 per- to lower performance in travel and government service cent of GDP, and consumer goods imports represented while transportation service increased by 2.0 percent 7.3 percent; fuel imports bill were 1.9 percent of GDP with an improvement in the number of passengers in 2015/16 down from 4.6 percent in 2013/14 due and cargo services of Ethiopian Airlines despite rela- to the declined oil prices (Figure 1.4.5). It is obvious tively lower ticket prices in an environment of lower that the total exports of goods and services cannot global fuel prices. In term of major export products, cover the cost of consumer goods and fuel imports. coffee value declined by 7.4 percent due to a decline While the fuel cost was lower compared to the previous in the price of Coffee Arabica, though the volume of years the drop was insufficient to improve the current exports increased by 8.0 percent in 2015/16. Gold account deficit. Services import marginally decreased exports declined by 9 percent because of low inter- from 4.8 percent to 4.7 percent of GDP in 2015/16. national prices and a reduction in volume exported. The overvalued real effective exchange rate Oilseeds, chat, and leather declined by 6.4, 12.4 and contributes to the weak export performance. The 3.7 percent, due to lower prices, volume decrease, and real effective exchange rate (REER) has appreciated decline in both, respectively. Horticulture (flower and in cumulative terms by 98 percent since the nominal fruits, and vegetables) exports increased as a result of devaluation in October 2010. However, the speed of better prices and volume than last year. Overall, since appreciation has slowed down over the past 6 months. 2010/11, Ethiopian exports have been on a declining While the appreciation between July and August 2015 path in percent of GDP (Figure 1.4.4). was 24 percent (y/y), this has slowed to an average of Declining exports to China, Ethiopia`s major 7.5 percent in March-June 2016 (i.e. slowdown in the export destination since 2004, are cause of par- rate of appreciation). This is primarily the work of two ticular concern. Export to China has been growing factors: first, a relative decline in the rate of domestic sharply since 2004, but recently slowed down. The inflation, and second, the depreciation of the U.S. average growth rate of exports to China was 85 per- dollar relative to other currencies since January 2016. cent between 2001 and 2014 compared to 134 per- Since the Birr is pegged against the U.S. dollar the Birr cent between 2005 and 2009, and just 19.5 percent also remained appreciated against other currencies. between 2010 and 2014. Although there is a declining Still, the Birr remains overvalued, which is hurting 10 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 1.4: External Sector 1. External Current Account Balance, % of GDP 2. Gross Official Foreign Exchange Reserves 20 1.3 4500 3.5 3.2 3.9 2.3 8.2 2.4 4.2 4.8 4.4 15 4000 6.4 5.8 2.5 2.6 3.4 4.2 2.6 3.0 4.1 3.2 2.6 2.3 1.4 3500 10 9.0 8.6 7.5 7.5 7.3 7.6 8.3 3000 2.5 –5 –1.0 –0.2 –0.1 –1.2 2500 –4.3 –10 –4.0 –0.7 2000 –6.5 –5.3 –10.4 2.0 –7.9 –11.5 1500 –15 - 1000 1.5 –14.9 –16.4 –20 –17.9 –17.6 500 –19.2 –21.0 –19.8 –25 0 1.0 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 Jan-10 Jun-10 Nov-10 Apr-11 Sep-11 Feb-12 Jul-12 Dec-12 May-13 Oct-13 Mar-14 Aug-14 Jan-15 Jun-15 Nov-15 Apr-16 Private transfers Official transfers FDI CAB/GDP Foreign Reserve (mill $) Reserve in months of import (right axis) Loans Change in reserves Trade balance G&S 3. Growth in Export of Goods 4. Export of Goods and Services, % of GDP 50 18.0 40.4 16.0 40 38.4 37.1 14.0 30 8.6 23.1 12.0 20 18.7 18.7 6.7 7.3 14.8 10.0 6.5 5.9 10 8.0 4.5 5.6 4.6 6.0 4.0 0 –1.0 4.0 8.1 6.0 6.8 6.5 6.0 5.7 –10 –2.3 4.8 –8.6 –4.7 2.0 4.1 –20 0.0 2004/05 2005/06 2006/07 2007/08 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 2008/09 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Value growth Price growth Volume growth Service export Goods export 5. Import of Goods and Service, % of GDP 6. Nominal and Real Effective Exchange Rate 35.0 210 5.1 30.0 0.4 5.7 6.1 4.8 0.4 4.8 4.4 190 0.7 0.4 0.4 4.7 25.0 4.8 0.8 0.3 4.4 5.0 4.3 4.3 170 Jun-16, 170 4.4 4.0 4.2 20.0 5.2 3.2 4.9 4.5 4.6 1.9 Nov. 2008 148 150 15.0 8.4 7.0 7.2 6.7 7.3 10.0 8.2 7.2 130 5.0 9.6 8.6 8.7 10.7 9.4 110 6.8 7.5 0.0 90 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 Oct-10, 86 70 Dec-99 Jun-00 Dec-00 Jun-01 Dec-01 Jun-02 Dec-02 Jun-03 Dec-03 Jun-04 Dec-04 Jun-05 Dec-05 Jun-06 Dec-06 Jun-07 Dec-07 Jun-08 Dec-08 Jun-09 Dec-09 Jun-10 Dec-10 Jun-11 Dec-11 Jun-12 Dec-12 Jun-13 Dec-13 Jun-14 Dec-14 Jun-15 Dec-15 Jun-16 Capital goods Consumer goods Fuel Intermediate goods Others Service imports Source: 4.1–4.5: NBE and MOFED, 4.6: WB based on IMF source. Recent Economic Developments and Outlook 11 international competitiveness. An overvalued currency Ethiopia faces a number of challenges in the implementation does not help to improve export competitiveness and of monetary policy as it implements a strategy that relies on is a concern for the economy, especially with exports public investment to close a large infrastructure gap aimed falling for three consecutive years. at triggering an export-led growth model. China has faced similar challenges since the 1980s. This section reviews Maintaining a competitive exchange rate is the monetary policy experience of China when it was at an important component of maintaining external a similar level of development following a similar growth competitiveness. Compared to its structural and strategy during the period from 1987 to 2006. The note finds that China had great success in stabilizing inflation once it aspirational peers,7 Ethiopia’s REER saw substantial implemented institutional reforms in the mid-1990s. After a appreciation and is overvalued, causing a steady loss large devaluation in 1994, the stable peg against the U.S. of competitiveness over the past year. This stands in dollar acted as a nominal anchor. From the late-1990s, China was increasingly active in conducting open market operations stark contrast to the experience of East Asian countries (OMO), though it complemented OMOs with quasi-direct (including China), which pursued systematic under- instruments (window guidance) and rule-based instruments valuation during periods of rapid growth, but broadly (reserve requirements). Other developing countries have relied less on direct instruments than China, but also use a in line with those of Latin American economies, which combination of rule-based and market-based instruments. adopted overvaluation with far less success. A recent World Bank study shows that a 1 percent real deprecia- tion increases total exports by 0.5 percent and reduces exchange sales (IMF 2013). A key reason for the lim- total imports by 0.6 percent. Disaggregation of exports ited use of indirect instruments is the underdeveloped reveals that a 1 percent real devaluation increases man- interbank market and the fact that nominal interest ufacturing and agricultural exports by about 1.06 and rates are centrally fixed (IMF 2014), both of which 0.33 percent, respectively (Haile 2015c). make indirect liquidity management between banks Competitiveness also depends on labor pro- an impossibility. Hence, the NBE currently relies first ductivity. The analysis undertaken in Chapter 2 and foremost on direct monetary policy interventions, shows that increasing labor productivity is key not while trying to develop the institutional foundation just for competitiveness, but for making urban labor for a more indirect monetary policy approach going markets work better. Productivity of unskilled labor forward. The International Monetary Fund (IMF) rec- is currently so low that wages cannot fall to clear the ommended in 2014 that the NBE should focus more urban labor market. on using indirect monetary policy instruments for its monetary targeting approach and noted that flexibility The Long View: Lessons Learned from of nominal interest rates would be a precondition to Monetary Policy in China (1987–2006)8 pursue such a step (IMF 2014). This is a good moment to re-consider the mon- Introduction etary policy strategy in Ethiopia. First, inflation has come down to below 10 percent levels after relatively Ethiopia’s current monetary policy approach is anchored around reserve money growth and characterized by limited use of indirect monetary 7 Ethiopia’s structural peers (Uganda, Tanzania, Mozambique, and Kenya) are countries that share similar structural features, while its aspirational policy instruments. According to the NBE, the cen- peers (Ghana, Zambia, Bangladesh, Cambodia, and Vietnam) are those tral bank’s suit of monetary policy tools is currently that set a good development precedent and that the country may aspire to emulate. For more details see World Bank (2016 forthcoming: ‘Ethiopia: focusing on reserve requirement ratios; sales of foreign Systematic Country Diagnostics’). exchange; T-bills sales through reserve money; and 8 This section is building on work carried out by Steven Pennings (Re- search Economist, DECMG) in 2015 to inform the design of a study tour adjustments to the interest rate. In practice, the focus of the National Bank of Ethiopia to China in June 2015 to gain insights of operations is on reserve requirements and foreign on the transition of monetary policy of the People’s Bank of China. 12 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET high inflation rates between 2011 and 2013. Current banks. State-owned banks, which mainly focus on success in controlling inflation will keep inflation financing large enterprises, constitute 70 percent of total expectations at bay in the medium-term. Maintaining assets of the banking sector and are dominating the credit low levels of reserve money and broad money growth market share of lending. The government-owned CBE is within a clear monetary policy strategy that outlines the dominant commercial bank and accounts for 60 per- appropriate forward-looking institutional reforms cent of total assets of banks as of June 30, 2015. The share would further cement the public’s view of successful of private banks assets to total assets of the banking sec- fighting of inflation in Ethiopia. Second, an improved tor declined from 36 percent in 2008/09 to 30 percent monetary policy framework is a prerequisite for an in Q3 2015. Domestic credit to private sector as a ratio exchange rate regime geared towards more competitive- of GDP was estimated around 18 percent below the ness. This is to be able to effectively combat any infla- Sub-Saharan Africa average of 30 percent in 2013. Total tionary pressures arising from a devaluing currency. outstanding credit of the banking system—excluding In search for lessons to guide Ethiopia in tran- credit to Government and interbank lending—reached sition from a directly controlled monetary policy ETB 205.7 billion/approx. US$10 billion at the end of approach towards a more indirect strategy, this March 2015,10 a 26.2 percent increase compared to the section studies the performance of Chinese mon- previous year. Finally, the banking sector in Ethiopia etary policy in the 20 years from 1987 to 2006. The remains closed to foreign banks. Chinese experience during this period is relevant for Banks are profitable and well-capitalized Ethiopia because of the two countries’ similar level although liquidity is a concern. The Ethiopian of per capita GDP (US$550 in Ethiopia currently, banking sector continues to show favorable aggregate US$400 in China in 1987), and a common public indicators: in March 2015 the average risk weighted, investment-led growth strategy combined with a system-wide capital adequacy ratio was 16.6 percent developing financial market dominated by state-owned (double the minimum requirement); profitability was banks. By the mid-2000s, China could be considered a robust (with return on assets (ROA) at 3 percent and middle-income country, and so its experiences may be return on equity (ROE) at 45 percent, well above less relevant for a low-income country like Ethiopia.9 regional averages); and the non-performing loan ratio was only 2.4 percent. However, no Financial Sector Current Situation in Ethiopia’s Financial Sector Assessment Program (FSAP) has been conducted in Ethiopia and the absence of detailed bank data pre- Ethiopia’s financial sector is dominated by the bank- cludes the assessment of potential vulnerabilities in ing sector (commercial banks). The banking sector individual banks. currently represents more than 92.6 percent of total The minimum deposit rate, regulated by NBE, assets of the financial sector, excluding the assets of the has remained constant for the past five years at Development Bank of Ethiopia (DBE) and National 5 percent. The lending rate to final borrowers is fully Bank of Ethiopia (NBE). Microfinance institutions liberalized, but has been relatively unchanged over the (MFIs) constitute 5.2 percent and insurance compa- same period with minimum and maximum observed nies 2.2 percent of the total financial sector assets. The Ethiopian financial sector consists of two public banks 9 In 2006, China’s GDP per capita was US$2,000, more than 6 times (i.e. the DBE and the Commercial Bank of Ethiopia that of Ethiopia currently (all figures are GDP per capita in constant 2005 U.S. dollars). The end date also reflects information. In the early (CBE)), 16 private banks, 17 insurance companies, 1980s, there is no inflation for China, and prices (and other parts of 33 MFIs, and five capital goods finance companies. production) are sufficiently controlled to make comparisons difficult. Note that aspects of monetary policy may have changed in the 2007–14 The banking sector is operated exclusively by period, but these are not the focus of the study. domestic banks and dominated by state-owned 10 FX Rate as of 3/31/15: US$1 = ETB 20.43. Source: www.xe.com. Recent Economic Developments and Outlook 13 lending rates unchanged at 7.5 and 16.25 percent, 1995–1998 period, and has been low and relatively respectively. For banks to increase their lending in sup- stable ever since. This change in fortunes, and the port of structural transformation, they need to extend ability to keep the inflation rate under control for the average maturity of their deposit base. Longer- an extended period of time was made possible by a term deposits currently only account for 11 percent number of reforms in the mid-1990s which allowed of the deposit base. Banks pay about 10 percent on the People’s Bank of China (PBC) to move towards longer-term deposits, roughly twice the interest rate greater operational independence with a greater focus on savings accounts. However, even these higher rates on inflation as well as significant changes in China’s do not ensure a positive real rate of return on deposits. exchange rate policy (next section). In addition, depending on the actual nominal lend- Important institutional reforms towards more ing rate which currently fluctuates between 7.5 and operational independence included establishing 16.25 percent, the real interest rate might be positive the PBC as its own entity under the state council or negative given the headline inflation of 11.8 percent (1995) and to end the credit plan (1998). First, as of July 2015.11 March 1995 saw the establishment of the PBC as Despite significant improvement in outreach led its own legal entity, reporting to the State Council. by the expansion of the banks’ branch network in the This, and earlier reforms (in December 1993), clari- past year, penetration of regulated financial institu- fied that the PBC’s primary goal was to maintain the tions remains very limited, particularly in rural areas. value of the currency (which would in turn promote The number of bank branches in Ethiopia increased to growth). With this, the PBC was better able to pur- 2,636 as of March 2015 and, according to GTP II tar- sue its monetary policy objectives, such as monetary gets, is supposed to double by 2020. Nevertheless, the growth targets (first set up in 1994).12 According to country lags behind the regional trends with only 2.9 Montes-Negret (1995), over the 1979–1993 period bank branches per 100,000 adults and only 0.3 ATMs monetary policy was subordinate to fiscal policy which per 100,000 adults compared to the SSA average of 4.5 led to monetary financing of the deficit, contributing and 3.4, respectively. While around 80 percent of the to high and volatile inflation.13 Second, 1998 saw the population lives in rural areas, access points remain con- end of the credit plan in China, and a move towards centrated in the capital city; 35.4 percent of total bank open market operations. Prior to this, the credit plan branches and 53.4 of insurance companies’ branches involved centrally allocated, bank-by-bank, quanti- are located in Addis Ababa. Thus, the long distances tative credit controls. Montes-Negret (1995) argues to regulated financial institutions present a barrier to that the growth and investment goals set by the State account ownership for the rural population. Planning Commission often required credit growth larger than the deposit base, leading to excessive Overview of Institutional Monetary Policy Reforms in China 11 According to IMF Article IV Consultation Press Release from Sep- China was able to change from a high and vola- tember 2015, the reported headline inflation was a result of rising food tile to a low and stable inflation environment prices due to below average rainfall, while non-food inflation was ap- proaching 10 percent. between the 1980s and 1990s driven by reforms 12 Reform of the PBC goals occurred i n several steps around 1993–95. in the operational independence of the People’s Up until 1993, the PBC was responsible for both the promotion of economic growth and currency stability (Laurens and Maino 2007). Bank of China (PBC). In the late 1980s and early 13 Note that the PBC independence bill of 1995 was highly controversial: 1990s, China experienced high and volatile infla- one-third of the delegates of the National People’s Congress abstained or voted against the 1995 Central Bank Law (Li 1996). However, inflation tion—reaching 19 percent in 1988 and 24 percent in was also deeply unpopular, contributing to Chinese political instability 1994 (Figure 1.5.1). However, inflation fell over the in the late 1980s. 14 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET lending by the PBC, increasing monetary supply the change in the consumer price index (CPI) over the growth and fueling inflation. previous couple of years and the inflation gap with While the PBC was much more independent the United States, and so aimed to bringing the real after the 1995 reform, it was far from fully legally exchange rate closer to its equilibrium value rather or operationally independent; still reforms had than undervaluing it.15 their effect. From a legal perspective, most policy The devaluation set the stage for a fixed decisions still required the approval of the State exchange rate to serve as a nominal anchor for Council. In fact, only 2 out of 7 monetary policy China over the following decade. The rate was estab- instruments in use in the period considered did not lished at around 8.3RMB/US$. The size of the 1994 require State Council approval (Geiger 2010). More devaluation also helped in stabilizing the RMB: after broadly, state-owned enterprises (SOE) and state- the 1994 devaluation, the RMB actually appreciated owned commercial banks (SOCB) were still under slight against the dollar over the following two years, the influence of government policy (discussed fur- reversing the trend of regular devaluation since 1980. ther below), and this ensured a way for the govern- The stabilization of the RMB no doubt had the effect ment to pursue its priorities, practically limiting the of lowering (and stabilizing) inflation expectations, PBC’s independence. Yet, the reforms had the effect which had been building in response to the inflation of stabilizing money supply growth, which reduced outbreaks of late 1980s and early 1990s. Together with the rate and volatility of inflation. As Figure 1.5.2 the institutional reforms described earlier in this sec- shows, the annual correlation between inflation and tion, the exchange rate policy of the early 1990s was money supply growth is 0.56 over 1987–2006 (0.52 able to lay the foundation for remarkable inflation over 1987–2013), so to a large extent inflation is a stability over the next decade. monetary phenomenon. Policies to Combat the Inflation Outbreaks The Exchange Rate in the Late 1980s and Early 1990s16 Changes in China’s exchange rate policy played an The increase in inflation in China in the late 1980s important complement to institutional reforms. stemmed from the relaxation of some credit quotas And in fact, the early 1990s saw a significant change and the decentralization of decision-making about in exchange rate policy, with two main components: regional lending. Credit and money growth increased (i) a depreciation of the official exchange rate against to around 30 percent in 1986. Banks lent out excess the U.S. dollar by 33 percent14 in 1994 (Figure 1.5.3), reserves, further increasing monetary growth in 1987. and (ii) a harmonization of the market-based “swap The credit plan constrained the PBC’s ability to rein in center” exchange rate and the official exchange rate lending. In 1988–89, with inflation at 25 percent, the (Mehran et al. 1996). In 1993, around 80 percent PBC raised reserve requirements (from 10 to 13 per- of foreign exchange transactions were conducted cent), credit plan quotas were more strictly enforced, in “swap centers” in different regions of China (Xu 2000), and the swap rate was much higher than the 14 Source: World Development Indicators (Series DI: PA.NUS.FCRF). The depreciations reduced the value of 1 RMB from US$0.17 to US$0.12 official rate—and had been so since the mid-late (33%) from 1993–94. Some commentators measure the depreciation in 1980s (Mehran et al. 1996, Chart 12). The large 1994 terms of the increase in RMB/US$ rate: a 50% increase from 5.8 to 8.6. These are equivalent. devaluation was not the first: the Yuan Renminbi 15 The growth rate of the GDP deflator for China was much lower than (RMB) was devalued by 21 percent in December that of the CPI, so movements in the CPI-based real exchange rate and GDP–deflator based real exchange rate will differ. Another key reform 1989, and by 17 percent over 1990–93 in a number in 1994 was the centralization of foreign exchange trading in Shanghai. of smaller steps. These devaluations were smaller than 16 This section draws on Mehran et al (1996). Recent Economic Developments and Outlook 15 interest rates were raised17 and PBC credit to banks of non-performing loans. Recognizing this problem, was reduced. This resulted in a fall in the rate of money reforms in 1994 transformed these four banks into growth to 3 percent in 1990. State-Owned Commercial Banks (SOCB), and also The relaxation of anti-inflation policies in the created three policy banks (for development, agricul- early 1990s was the primary driver for excessive ture, and export promotion). money supply growth and inflation in 1993 and Despite reforms, the State-Owned Commercial 1994. More relaxed anti-inflation measures combined Banks did not become purely commercial (Cheng with other liberalization efforts that increased invest- and Cheng 1998). Among other things, SOCBs pro- ment expenditure, a real estate boom, an increase in vided credit to loss-making state-owned enterprises and food prices (and other administered prices) and PBC favored investment projects. For SOCBs, this meant lending to the public sector resulted in excessive money providing loans regardless of cost in the money market. supply growth and inflation. The PBC responded by For SOEs, the repayment requirements were soft, and reducing central bank credit to banks, recalling loans so the exact interest rate and terms offered were less made “illegally,” recalling overdue loans, increasing relevant for investment and production decisions. As interest rates, and attracting greater deposits. pointed out by Green (2005), the low ability of SOCB employees to price risk meant that banks were unable Monetary Policy Transmission and Financial to turn variations in the cost of finance into changes Sector Development in commercial loan availability. Given that banks often had effective monopolies in their areas of lending, there The transmission of monetary policy through was little commercial pressure to become more respon- indirect instruments requires that: (i) bank credit sive to the cost and availability of funds (Green 2005). be sensitive to the availability and price of credit That period also saw high levels of excess reserves, in the money market, and (ii) economic decisions a key symptom of a weak interbank market. Excess by firms be influenced by the availability and price reserves apply both to SOCBs and also non-state-owned of bank credit.18 In China in the 1990s and 2000s, banks. Green (2005) reports that in the early-mid both components of the transmission mechanism 2000s, excess reserves fluctuated between about 3.5 and were weakened by a lack of separation of policy and 6.5 percent. One of the key instruments of the PBC commercial lending, excess bank reserves, and a lack was to change reserve requirements, but these would of competition. This section first provides some back- have little impact since those reserve requirements ground on the development of financial markets in were not binding. Green (2005) argues that banks had China and associated reforms, and then discusses the excess reserves because: (i) large inflows of liquidity in effect on monetary policy transmission. the mid-2000s related to the [possibly] undervalued Before the start of the transition in the early exchange rate, (ii) a lack of alternative safe investments 1980s, the PBC was the only bank in China, and (for example, corporate bonds), (iii) interest paid on was really an arm of the Ministry of Finance and excess reserves at the PBC, and (iv) regulations that State Planning Commission. In 1983/4 the PBC was established as the central bank, and four State 17 As open market operations had not yet been introduced, this higher interest rate may have applied to direct lending by PBC to state-owned Owned Specialist Banks were established, provided a enterprises as well as lending by commercial banks. mix of policy and commercial loans to different sectors 18 The transmission mechanism of monetary policy can also occur through other channels (consumer borrowing and saving, asset prices, (Cheng and Cheng 1998). However, the combination exchange rate movements, etc.), but Mishra et al. (2012) argue that of policy and non-policy functions weakened account- in developing countries, the bank lending channel is likely to be the dominant one (at least in relative terms). In China during the period of ability, resulting in excess lending (including from study, credit was generally not available to individuals (Green 2005) and the PBC as described above), as well as a high share the exchange rate was fixed, further weakening these secondary channels. 16 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 1.5: Inflation, Broad Money Growth and the Exchange Rate in China 1) Inflation rate, annual, percent 2) Broad money growth, annual, percent 30 March 1995: 1998: Abolition of credit plan; 60 March 1995: PBC gains its 1998: Abolition of credit plan; PBC gains its own legal Adoption of Open Market Operations own legal status Adoption of Open Market Operations 25 status; formalization 50 Dec 1993: PBC's primary of key goal of goal currency stabilty currency stabilty 20 40 15 2009: Economic Stimulus Package 30 10 20 5 0 10 –5 0 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 1980 1985 1990 1995 2000 2005 2010 Source: World Development Indicators (FP.CPI.TOTL.ZG). Source: World Development Indicators (FM.LBL.BMNY.ZG). 3) Nominal exchange rate, annual, percent 0 1994 Devaluation 1 2 3 4 5 6 7 8 9 10 1980 1985 1990 1995 2000 2005 2010 Source: World Development Indicators (PA.NUS.FCRF). An increase represents an appreciation of the RMB. Source: World Bank, World Development Indicators. Note: (3) An increase represents an appreciation of the RMB. discouraged risky commercial lending. More gener- growth targets moved closely with what would be ally, high excess reserves are often a symptom of a weak implied by output, velocity, and inflation targets (cor- interbank market: banks are not confident that they can relation 0.85). However the level of money growth borrow from the money market if faced with an unex- targets was typically 4–5 percentage points higher, pected withdrawal of deposits (for example). with a maximum deviation of around 8 percentage points in 2003. Geiger (2008) finds that interest Chinese Monetary Objectives and rates tended to move in the direction of the desired Instruments (mid-1990s-mid-2000s) monetary policy action, but were not a key compo- nent of monetary policy. The lack of a competitive One of the most important reforms in the mid- financial market meant that interest rates were not 1990s was the change in the operational target of Chinese monetary policy.19 From 1994–95, the PBC 19 This section draws heavily on Geiger (2008), with data on money targeted money supply growth (M1 or M2). Money growth outturns and targets from Table 5 and 6. Recent Economic Developments and Outlook 17 allocative (and interest rates were only partly liberal- to sterilize foreign exchange purchases (through repos ized). In fact, China often missed its money growth and the issue of central bank bills). With an increased target—especially when it was trying to bring down need for sterilization starting in 2003, the PBC started inflation.20 Nonetheless, overall performance was issuing central bank bills. Previously, sterilization had quite good in terms of bias—average M1 growth was fallen on repo transactions, which continued after almost exactly the average of targeted growth, with 2003, but were not sufficient for the volume of steril- average M2 growth 1.3 percentage points above tar- ization required. One advantage of central bank bills get. The mean absolute deviation (MAD) of planned over repos was that they could be traded, increasing and actual money growth was larger (around 3 per- liquidity in the money market (Green 2005). cent).21 Despite these difficulties, the PBC succeeded Reserve requirements were a major tool of the in achieving low and relatively stable inflation over PBC over 1998–2006. Initially it was used care- the 1998–2006 timeframe. In part, this might be fully with five small movements in 2003–06, but 10 because specific money growth targets helped the adjustments 2007 and eight in 2008 (Geiger 2010). PBC to avoid the excessive money supply growth of Reserve requirements were first introduced in 1984. the mid 1980s/ early 1990s. They were reduced (13 to 8 percent) over 1998–99 as In addition to changes to the operational inflation fell and the PBC moved to OMO to man- framework, the PBC started relying on new instru- age liquidity (Geiger 2008). Reserves were remuner- ments. Laurens (2005) and Buzeneca and Maino ated at 1.89 percent over 1998–2006 (Laurens and (2007) divide monetary policy operations into three Maino 2007). basic types. (1) Direct instruments involve direct admin- Standing facilities and interest rate guidance were istrative control over interest rates/quantities in the also in effect. During the period of study (1998–2006), financial sector, and are a carryover from a planned banks could borrow from the PBC through the auto- economy. These include (in the Chinese context) win- matic collateralized lending facility at the rediscount dow guidance and wage/price controls. (2) Rule-based rate; or lend excess reserves to the PBC through the instruments work indirectly, but still rely on the regula- excess reserves facility at a rate of around 1–2 percent tory power of the central bank. These include reserve (Laurens and Maino 2007). The rediscount rate was requirements, standing facilities and statutory liquid- the main policy rate of the PBC during the period of ity requirements. (3) Market-based instruments (open study. From 2004 the PBC was able to change the redis- market operations) involve the central bank trading count rate (within a range) without permission from in the money market and are the primary instrument the State Council (Geiger 2008). In 2004, ceilings on used by central banks in developed-countries. China’s lending rates and floors on deposit rates were eliminated. approach involves using all three types; some of the However, floors on lending rates and ceilings on deposit most important ones are described below. Other coun- rates were maintained to protect banks’ profitability. The tries are described in Box 1.1. PBC changed these rates in 2006 in response to high Open market operations were originally intro- credit growth (Laurens and Maino 2007). duced in 1993 in China, but were abandoned and then reintroduced in 1998 when the interbank mar- 20 Volatility in money multipliers is a problem in other countries too, ket was more developed (Geiger 2008). Key instru- such as the Bundesbank in the 1970s and 1980s (Geiger 2008) and also ments included repos and reverse repos of T-bills, and in Ethiopia (Aklilu 2009), which (along with volatile velocity) is one reason why many central banks have switched to interest rates as the issuance/receipt of central bank bills. By the mid-2000s, intermediate target. Some authors suspect that interventions to stabilize open market operations were taking place twice a week the exchange rate may have contributed to the missed targets, though there is little evidence of this as foreign exchange activities were sterilized. (an average of 115 transactions/year over 2000–06). 21 Note that it is probably bias rather than MAD, which is more important Open market operations were particularly useful to try for anchoring inflation expectations. 18 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET BOX 1.1: Monetary Policy Instruments: Evidence from other Developing Countries Most developing countries use a combination of instrument types: Rule-based instruments in combination with market-based instruments (Table 1.1).* Direct instruments are still used occasionally, but are less common than in the 1990s. Both standing facilities and reserve requirements are used in all developing countries studied, with the Lombard facility being the most common type of standing facility. SSA peer countries (The Gambia, Uganda and Zambia) discussed in Laurens (2005) also use a mixture of rules-based and market-based instruments. Reserve requirements have been used in all three countries, while liquid asset requirements are not ubiquitous. All of these SSA peers considered have used OMOs, with T-bill transactions being the most common type used. Comparing the instruments used in peer countries with those used in China, one main difference is that quasi-direct controls which were a common instrument in China over 1987–2006 are relatively rare in the 2000–2010 time frame among low income countries surveyed (15 percent in 2004; Table 1.1), and have not been a tool in selected SSA comparison countries since the 1980s–1990s (Laurens 2005). On the other hand, both China and other peers rely on rule-based instruments and OMOs. The Gambia, for example, relies heavily on adjusting reserve requirements (IMF 2013a), just as China did in 2003–08. Many challenges are also similar: The Bank of Zambia was constrained in its ability to undertake repo transactions to reduce liquidity in the mid- 2000s due to a lack of T-bills in its portfolio; and the PBC faced difficulties in sterilization from 2003 using its stock of T-bills. In recent years, there has been a move away from monetary growth rate targeting in SSA and towards a greater emphasis on interest rates as the operational target, and inflation as the final target. For example, in 2011 Uganda adopted an inflation- targeting “lite” monetary policy framework and Kenya (2011) and Zambia (2012) replaced monetary targeting with a policy interest rate target (IMF 2014, 2013b). * Direct instruments involve direct administrative control over interest rates/quantities in the financial sector (Buzeneca and Maino 2007). Rule-based instruments are more indirect instruments, but which still rely on the regulatory power of the central bank (Lauren 2005). Market-based instruments involve the central bank trading in the money market. TABLE 1.1: Comparison of Monetary Policy Instruments in Developing Countries % of Developing Countries Instrument Type Description/Comments Using Instrument (2004) Direct Instruments Interest rate controls, Directed Credit, Specific lend- 15% (down from 54% ing req.; Bank-by-bank credit ceilings are rare in 1998) Rule-Based Instruments 100% Reserve Requirements [Various types] 100 Statutory Liquidity Requirements Must hold fraction of liabilities as liquid assets 46% Standing Facilities (SF), Types: Short term borrowing/lending at penalty rates 100% Lombard facility Overdraft borrowing by banks 85% Rediscount Credit Purchase of marketable securities 69% Deposit Facility Banks can deposit excess funds 31% Interest rate Arrangements Combination of borrow/lending SF 15% Market-based instruments 92% (Discretionary Arrangements) – all types Primary Market Arrangements Auctions of central bank securities/deposits 62% Secondary Market Operation Auctions/Purchases on secondary markets/ repos 62% Source: Adapted from Buzeneca and Maino (2007). Window guidance and wage/price controls This involved the PBC meeting with banks and were used in China in addition to more “standard” telling them to increase or decrease credit growth instruments between 1994 and 2006: (for example), or since 2003, the publication of official notices. For example, the PBC met with  Window guidance/moral suasion was a key compo- commercial banks a number of times in the sum- nent of monetary policy implementation, intro- mer of 2003 to urge them curb excessive real estate duced in 1998 (with the end of the credit plan). lending (Geiger 2008). It is difficult to determine Recent Economic Developments and Outlook 19 the impact of window guidance, but commenta- deflation and falling credit growth, the PBC tried to tors suggest it was an important component of stimulate the economy using indirect instruments. monetary policy. Over 1998–99, the PBC slashed official interest rates,  Price controls. Around 25 percent of the Chinese increased the supply of base money through OMOs, CPI in 2001 consisted of goods where the govern- reduced the required reserve ratio from 13 to 6 per- ment either directly controlled prices, or issued cent and reduced the interest rates paid on reserves price guidance.22 In periods of high inflation, the (Green 2005). However, credit growth continued to government could (and did) freeze prices for these slow over 2000–01. items. For example in March 2004, the National In late 2001, the PBC resorted to window guid- Reform and Development Commission instructed ance and asked banks to increase lending. The gov- provincial officials to freeze prices of controlled ernment also introduced expansionary fiscal policy. As goods/services once inflation reached a certain outlined in Green (2005) credit growth shot up from point (Geiger 2008). It is important to note that below 10 percent to around 35 percent. China now this was only a temporary freeze rather than a per- faced the opposite problem: credit growth was too fast. manent price subsidy—presumably once inflation Around 2003 the PBC raised interest rates, reduced fell below the threshold, the prices of these goods the money supply through OMOs, and increased could be raised. reserve requirements.23 Again, the policies seemed to  Wage controls. Historically, wages were centrally be working slowly and so in the middle of 2003 the set according to region, occupation, industry etc. PBC began to introduce window guidance to reduce In the mid-1980s and early 1990s, these were lending (Green 2005, Geiger 2008). Combined with indexed to inflation, which contributed to the other initiatives discussed above (such as price con- outbreak of inflation in 1988 and 1994 (Geiger trols) credit growth dropped back to below 16 percent 2008). However, starting in 1994–95, reforms by June 2004. were introduced to decentralize wage setting, This experience over 1998–2004 suggests that which made it more closely, connected to produc- market-based instruments may have been less effec- tivity growth and profitability. Geiger (2008) esti- tive than the PBC has hoped due to incomplete mates that around 15 percent of employees in the development of the financial sector. However, the late 1990s/early 2000s fell under the 1994–95 same lack of institutional structure inhibiting indirect wage-setting arrangements. instruments meant that window guidance was both available and effective. An alternative interpretation Challenges in the Face of Weak Transmission is that market-based polices were new and hence more difficult to implement properly (e.g., harder to The Chinese reforms of 1993–98 outlined above signal future actions), whereas window guidance was were designed to replace the credit plan policy essentially a continuation of the credit plan which had with market-based monetary policy instruments been used for many years. Either way, the overshoot- such as OMOs and interest rate adjustments, but in ing of credit growth in the early 2000s suggest that it practice there were many challenges. It was expected is also very difficult to use direct controls to fine-tune that the PBC would be more able to implement these the macro-economy, which is worth noting for other changes given its greater independence and clarity of countries considering similar policies. objectives, and these measures would be more effec- tive due to the separation of policy and commercial 22 Including utilities, goods produced by monopolies or products/services that were deemed to have specific importance. lending. In practice, the transition was both chal- 23 There were also large forex inflows in 2003, which may have also lenging and incomplete. In the late 1990s, faced with contributed to the boom conditions. 20 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Implications and Lessons Learned for Ethiopia`s strong economic growth slowed down in 2015/16 Ethiopia’s Monetary Policy due to the recent drought, but it is expected to rebound quickly in 2016/17. Inflation is expected to remain in single digits. The main lesson from the Chinese experience in Disappointing exports and soaring imports contribute to poor external sector performance but are expected to improve in monetary policy for Ethiopia is that it is impor- the medium-term. Economic growth in 2016, lower but still tant for relying on a full range of instruments substantial, is expected to drive further reductions in poverty. while financial markets develop. For China, this includes quasi-direct instruments (window guidance), crisis, for instance in the 1980s, and in a large drought rule-based instruments (reserve requirements) and in 2002/03. The large impact of droughts is explained open market operations (market-based instruments). by the fact that the agriculture sector is, by far, the Other peer countries (see Box 1.1) have combined biggest employer in Ethiopia, accounts for most mer- market-based instruments with rule-based instru- chandise exports, and is the second largest in terms ments (particularly reserve requirements). The mixed of output. In addition, agricultural growth was an strategy allowed the central bank to control credit and important driver of poverty reduction in Ethiopia monetary growth even in the face of limited financial over the past decade (2004 to 2014): Each percent of market development and a concentrated and uncom- agricultural growth reduced poverty by 0.9 percent petitive banking sector.24 compared to 0.55 percent for each percent of overall The second lesson is that fully effective market- GDP growth based implementation of monetary policy is not Agriculture is also the major sector contribut- possible without complementary reforms. Such ing to growth in Ethiopia. The share of agriculture reforms need to make the banking sector more com- was 38 percent in FY 2015, which contributed about petitive and to separate policy and non-policy lending. 2.6 percentage points to the 10.2 percent GDP growth Knowing this, Chinese authorities separated policy in 2014/15. Crop production accounts for about banks from commercial banks in 1994. However in three-quarters of agriculture value-added;25 this is line with the gradualist approach, the reforms were followed by livestock which accounts for one-fifth incomplete—state-owned commercial banks were still of the agricultural value added. Agriculture is still not fully commercial—which may have reduced the largely rain-fed and dominated by small holders and effectiveness of market-based instruments over the subsistence production. following decade. Despite the poor rains the national meher (main season) harvest for 2015/16 did not drop significantly. The Future View: Challenges and The overall drop in production for meher season of Outlook 2015/16 was estimated at –1.32 percent,26 while the crop production growth for all seasons including commercial Macroeconomic Challenges through the Current Drought 24 Reserve requirements can be used to create a shortage of liquidity, to encourage banks to borrow from the central bank or each other (Laurens Overview 2005). Standing facilities can be used to ensure liquidity (including in payments system) while open market operations and liquidity forecasting Ethiopia is not a stranger to droughts and their are being developed. Laurens (2005, Appendix V) shows that if banks impact is often large due to the importance of the are reluctant to lend to each other in the interbank market, the central bank can act as their counterparty (in Turkey in the 1980s) or can help agriculture sector in the economy and in peoples’ to aggregate buy/sell orders (Thailand). lives. The country has seen more than 15 drought epi- 25 Belg season harvest accounted for 4 percent of total crop production in 2013/14. sodes between 1964 and 2015. They varied in impact, 26 The failure of belg rains and the arrival of El Nino in 2015 resulted in and some of them led to full-fledged humanitarian failed belg harvest and affected meher (main) season harvest. Recent Economic Developments and Outlook 21 TABLE 1.2: Crop Production All Seasons Including Commercial Farms (in Million Quintals) Growth Seasons 2011/12 2012/13 2013/14 2014/15 2015/16 2015/16 Meher 218.57 231.29 251.54 270.4 266.83 –1.3% Belg 6.82 9.53 10.97 8.93 19.09 113.8% Commercial 7.05 9.87 11.92 12.15 12.58 3.5% Total 232.45 250.69 274.43 291.48 298.5 2.4% Source: CSA Crop Production Reports of Various Years. farms posted a 2.4 percent growth. The crop loss in the Livestock Value Added eastern part of the country particularly in Afar, Somali, The livestock value added is estimated based on and Dire Dawa falls between 25 and 66 percent (CSA, CSA’s Livestock and Livestock Characteristics 2016). In the west, Gambella region faced crop loss report for the year 2015/16. The overall livestock of 30 percent. In general, except Benishangul-Gumuz growth in 2015/16 appears to be 1.1 percent. Yet, region all regions faced crop loss of various magnitudes in Afar and Somali regions a 5.4 percent drop each that mainly coming from areas which were already in the in cattle population has occurred. Nationally, cattle margin. Further, poor grazing in Afar and Somali regions, population has grown by 2 percent, while the sheep which are predominantly pastoral areas, led to high ani- population dropped by 1.5 percent. The latter is mal mortality, worsening animal body conditions, and mainly due to large sheep population losses that declining milk productivity (FAO 2015). occurred in most of the regions (for instance Tigray: On the other hand, the production figures 10 percent, Dire Dawa: 13 percent, Benishangul: reported by the CSA for the belg (second) season 6 percent, Amhara: 2.3 percent, and Oromia: 2.4 per- of 2015/16 show a remarkable growth of 113 per- cent). The goat population has shown a nationwide cent. The belg harvest was good owing to the favor- 2 percent growth, despite a 7 percent drop in Tigray, able belg rainfall, which benefited areas conducive to and 4 percent drop each in Somali and Dire Dawa belg production in the country. Further total area cul- regions. Thus, despite the drought that hit the eastern tivated during the belg season of 2015/16 increased half of the country hosting the majority of livestock by 50 percent (596,844 hectares of land) compared population, the country has avoided livestock losses to the 2014/15 belg season.27 Such a remarkable and posted an overall 1.1 percent growth (Table 1.3). growth in total area cultivated for belg production Still, this is a 3 percentage points drop from the pre- certainly has contributed for the large growth in belg vious year. production. Looking at the meher and belg seasons com- Food Inflation and Imports bined, the all season crop production growth in Despite the worst drought in three decades general 2015/16 was 2.4 percent, 3.8 percentage points Inflation remains relatively stable and food infla- lower than the previous year`s growth levels. tion remains relatively controlled. Although infla- Although crop production dropped slightly during tion has been gradually rising since November 2014, the meher season, the remarkable recovery of the belg it only reached double digits in June 2015 and peaked season production, and the growth in the commercial to 11.9 percent in October 2015. However, it stood at farms crop production (3.5 percent) kept the com- bined crop production growth in 2015/16 to remain 27 The growth of total area cultivated in 2014/15 belg season was 11.6 positive at 2.4 percent (Table 1.2). percent. 22 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Livestock Population Growth in TABLE 1.3:  2.5 million metric tonnes of wheat import, 0.3 mil- 2015/16 lion metric tonnes goes to replenish the strategic 2015/16 food reserve which was 0.4 million metric tonnes and   depleted in an earlier emergency response. High lev- Regions Cattle Sheep Goats Total els of cereal stock reserves and large carry-over stock Tigray 0.8 –9.9 –7.2 –4.2 from 2014/15 (Bellmon, 2015) may explain the low Afar –5.4 4.0 4.8 2.1 inflationary impact. Amhara 5.1 –2.3 0.3 1.7 The investigation of the price trends of major Oromia 1.8 –2.4 6.7 1.8 cereals in different local markets shows that the Somali –2.2 0.9 –3.6 –1.8 price trends in the aftermath of the drought were BG 10.1 –6.3 1.8 5.6 not actually unusual y high, except sorghum SNNPR –0.4 2.7 4.9 1.6 prices. Cereal prices spiked for short period between Gambella 0.7 24.8 6.3 4.1 June 2015 and October 2015, and quickly reverted Harari 9.4 45.6 17.5 14.9 back to their pre-drought levels. As of August 2016, DD 5.7 –12.9 –3.5 –4.5 the year on year price increase of all the three cere- Total 2.0 –1.5 2.0 1.1 als investigated here have posted a price decrease, except sorghum that posted a price increase between Source: CSA Livestock and Livestock Characteristics Report 2015/16. 20 to 60 percent. Large scale food imports, sup- ply of wheat to large bakeries at subsidized prices, 5.6 percent in October 2016. The regional inflationary and strategic distribution of food aid to drought trends were also largely similar to the national trend, affected areas may have helped in stabilizing cereal except that in Afar inflation spiked to 24 percent in prices and hence food inflation. Further, the tight September and October 2015, but dropped to 9.2 per- monetary policy regime also partly contributes for cent in September 2016. Similarly, food inflation the price stability. The spatial price difference found reached to 16 percent in October 2015, but dropped to be higher for wheat, followed by sorghum. (See back to 4.3 percent in August 2016 and rose up to Figures 1.6.4–1.6.6). 6.1 percent in September 2016 at the national level; There is some risk that increased wheat the movement at the regional level has been mixed.28 imports and subsidies put fiscal pressure on the The recent peak in inflation rate may signal the conse- Government. At the current levels, wheat imports quences of the unrest that compelled the government may cost up to US$375 million, which is equivalent to enact state of emergency. to 12 percent of the total annual export earnings of Following the drought situation, wheat imports 2014/15. Following the humanitarian crisis ensuing reached the record high 2.5 million metric tonnes. from the drought, the Government has announced Figure 1.6.3 shows the drought has pushed the wheat 18 Billion Birr (US$0.85 billion) additional budget import to the record high 2.5 million metric tonnes during FY16, of which 82 percent (or US$0.7 bil- (between October 2015 and September 2016), about lion) was devoted to the financial requirement for 1.5 million metric tonnes above the recent aver- drought-affected population (Reporter 2015). age (CIMMYT 2015; USDA and USAID 2016). Compared to the last 5 years average import of close 28 In Addis Ababa, food inflation has increased since January 2015 and reached 30.4 percent in October 2015, but has been dropping since to 1 million metric tonnes in commercial and emer- February 2016 and reached –1.9 percent in August 2016 (year on gency assistance, the current high import induced by year), and stood at 0.46 percent in September 2016. Similarly, Afar and Oromia posted a negative inflation of 0.1 and 0.8 percent, respectively the current drought has helped in containing food in August 2016, and negative 0.3 and 1.75 percent in September 2016, price hikes that would occur otherwise. Out of the respectively. (Figure 1.6.3). Recent Economic Developments and Outlook 23 FIGURE 1.6: Drought Assessment 1) General inflation September 2015 to August 2016 2) Food Inflation September 2015 to August 2016 30% 40% 30% 20% 20% 10% 10% 0% 0% –10% –10% September October November December January February March April May June July August September October November December January February March April May June July August Ethiopia AA Afar Gambella Harari Oromia Amhara B.Gumuz D.Dawa SNNP Somalia Tigray 3) Ethiopia Wheat Import 1965–2016 4) Wheat price Trends in Selected local markets, 3,000 March 2014 to August 2016 1400 Whaet Import ('000 tonnes) 2,500 1300 1200 2,000 1100 1000 1,500 900 800 1,000 700 600 500 2014m3 2014m5 2014m7 2014m9 2014m11 2015m1 2015m3 2015m5 2015m7 2015m9 2015m11 2016m1 2016m3 2016m5 2016m7 0 1961 1965 1969 1973 1977 1981 1985 1989 1993 1997 2001 2005 2009 2013 DD Assela Shashemene Mekelle Bahir Dar AA 5) Maize price Trends in Selected local markets, 6) Sorghum price Trends in Selected local markets, March 2014 to August 2016 March 2014 to August 2016 650 1200 1100 600 1000 550 900 500 800 700 450 600 400 500 2014m3 2014m5 2014m7 2014m9 2014m11 2015m1 2015m3 2015m5 2015m7 2015m9 2015m11 2016m1 2016m3 2016m5 2016m7 2014m3 2014m5 2014m7 2014m9 2014m11 2015m1 2015m3 2015m5 2015m7 2015m9 2015m11 2016m1 2016m3 2016m5 2016m7 AA BD DD AA DD Shashemene Mekelle Mekelle D. Birhan Source: 1.6.1–1.6.2: CSA; 1.6.3: CIMMYT (2015); 1.6.4–1.6.6: EGTE. 24 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Outlook performance and stimulate growth in the short- to medium-term. It was to be expected that Ethiopia’s strong eco- Economic growth in 2015/16, while lower still nomic growth would slow down in 2015/16 due substantial, is expected to drive further reductions to the recent drought. The drought caused by the in poverty. Growth has been an important driver of El-Niño phenomenon affected the economy nega- poverty reduction in the last decade with each percent tively through reductions in food production. While of growth reducing poverty by 0.55 percent. Hence, there was some uncertainty about the actual impact with growth at 8 percent in 2015/16 the poverty reduc- of the drought, preliminary actual figures from the ing potential is still significant. Based on this, the pro- Government for 2015/16 show now that the growth portion of households living below the poverty line of rate was 8 percent, in line with. With this, the growth 1.9 US$PPP is estimated to decline from its estimated impact is lower than originally envisaged; according level of 27.2 percent in 2015 to 24.6 percent in 2018. to Government crop data released by the CSA in July Since regions most badly affected by the drought 2016 there was a less than expected agriculture pro- are the poorest parts of the country, poverty may duction impact; this is due to good crop production fall less than predicted. The impact of the poor rains during the second season harvest of 2015/16. As past has not been equal across the country. The areas most evidence shows, with regular rains again, agriculture affected are the poorest parts of the country where many growth can quickly pick-up in 201729 (Figure 1.7.1). households are poor or live just above the poverty line. But positive growth moment will still remain. A Drought-induced harvest losses in these areas cause decade of remarkable double digit growth rates helped large increases in poverty. The poverty headcount ratio the economy to cope well with the most recent chal- estimate based on growth elasticity does not take such lenges encountered in 2015/16. In fact, the ability to regional variations into account. When district-specific keep growth positive is a remarkable achievement for data on estimated crop losses are used, the rate of poverty the Government. In similar occasions in the past, for reduction is potentially slower than estimated in 2016. instance in 1997/98 and 2002/03 the country has Export growth continues its declining trend, experienced negative GDP growth. In 1998 growth largely on account of falling commodity prices dropped to –3.5 percent from 3 percent in the previous and an overvalued currency. The Birr continues to year. Similarly, in 2003 it dropped to –2.2 percent from be ovrevalued in real terms and this is hurting external 1.5 percent previous year. Medium-term economic competitiveness (The Birr is pegged against the U.S. growth can be unaffected from the drought since the dollar). Over the past three years, exports of goods and rains set in normally again in 2016/17. Still, an overall services have averaged a growth rate of just 0.6 percent dampening effect will set in motion in the medium per year compared to an average of 13 percent over the term as total factor productivity (TFP) is expected to last decade. Good exports continued its disappointing decline in the medium-term following the recent and growth in FY16, declining by 8.3 percent in the first rapid accumulation of capital stock (Figures 1.7.2 and severn months of FY16. Looking forward, the current 1.7.3). And potential negative economic effects of the account deficit is predicted to narrow gradually to current unrest are a risk to the outlook. On the other 7.8 percent in 2018 because of expected rebounding hand, the completion of the Addis Ababa -Djibouti in exports (Figure 1.7.4). railway line, significantly eases trade logistics related 29 Data from the most recent previous drought year in 2002/03 shows constraints. The commencement of new industrial that production growth indeed picked up in the first post-drought year. and the increasing capacity in power generation with A 40 percent increase in meher season production has been observed in 2003/04 following the recovery from the 2002/03 drought. In the same the completion of transmission lines to neighboring period agricultural GDP bounced back from –10.5 percent in 2002/03 countries are also expected to improve the export to 17 percent in 2003/04. Recent Economic Developments and Outlook 25 FIGURE 1.7: Economic Outlook: Selected Projections to 2018 1) GDP Growth (Supply Side), 2009–2018 2) Determinants of Potential Output, 2001–2017 14 30 11.5 12 10.6 10.4 10.2 25 10.0 9.9 10 8.7 8.9 8.6 8.4 20 5.8 8 4.0 4.6 5.8 7.1 5.9 4.4 15 6 5.1 4.7 5.6 1.5 2.8 10 4 1.0 3.0 1.1 2.1 2.2 2.1 2.0 2 4.1 5 3.2 2.5 2.3 2.6 2.3 3.1 2.2 1.7 1.8 0 0.4 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Agriculture Industry Services GDP af factor cost Capital stock Working age population Potential GDP TFP 3) GDP Growth (Demand Side), 2009–2018 4) Fiscal and External Balances, 2001–2018 22 –1 18 –3 14 10 –5 6 –7 2 –9 –2 –11 –6 –13 –10 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2012 2013 2014 2015 2016 2017 2018 GFCF Imports, GNFS Exports GNFS Current Account General Gov't Private consump. Public consump. Real GDP deficit fiscal balance Source: World Bank staff compilation, based on data from the Macro-Fiscal Forecasting Model. However, export performance is set to improve of power transmission lines to neighboring countries in FY17. The commencement of the Addis-Djibouti (Kenya and Sudan), and the expansion in power gen- railway line eases the constraint in trade logistics and eration capacity will increase electricity export, further reduces the transportation cost of moving goods in and boosting the diversification efforts. out of the country. It takes only 10 hours for the new A key challenge to the economy stems from railway to take goods between Ethiopia and Djibouti, rising external imbalances and the risk of debt sus- a huge cut compared to the 3 to 4 days for a truck. tainability to worsen; the latter is currently assessed Further, the recently inaugurated Hawassa Industrial as ‘moderate’ but vulnerabilities are rising, primarily Park and the completion of Bole-Lemi Industrial Park due to lower than expected export performance. The Phase II are set to increase manufacturing exports and reserve position remains low. A key factor of low export contribute to the diversification of the coffee and gold growth is in the continuing appreciation of the cur- dominated export profile of Ethiopia. The completion rency which is hurting external competitiveness. The 26 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET current civil unrest in some regions, if not stopped, market (see Chapter 2), the wage in the private sector for could pose potential harm on the economy and affect a particular job category may be the reservation wage. tourist and FDI inflow—yet the effect is uncertain at A higher reservation wage (due to higher public sector this stage. On the other hand, the negative impact of wages) would adversely affect the unemployment rate. the drought is expected to be temporary in nature. Potential effects of the current unrest in some The impact of the civil servant pay rise in the regions of Ethiopia pose downside risks to the medium-term is not know. First, the increase in itself economy through trade and investment channels. may have a new inflationary impact in the medium-term. The recent unrest, which has intensified since the Yet, the very reason for the pay rise was to account for beginning of the new fiscal year, could pose strains the erosion of the value of real wages in the public sector on economic growth and affect tourist, FDI and trade due to high inflation. Second, given the important sig- flows, especially if it was to continue for a prolonged nalling function of the public sector on the urban labor period. The economic effect is uncertain at this stage. TABLE 1.4: Macro-Fiscal Outlook Indicators, 2012 to 2018 2013 2014 2015 2016 f 2017 f 2018 f Real GDP growth, at constant market prices 10.5 10.3 9.6 8.4 8.9 8.6 Private Consumption 9.8 4.3 3.3 6.5 6.6 7.6 Government Consumption 12.5 17.8 5.7 26.5 5.3 2.5 Gross Fixed Capital Investment 9.4 28.2 18.7 12.0 7.7 8.3 Exports, Goods and Services –0.8 2.8 –7.5 3.6 7.9 6.8 Imports, Goods and Services 4.0 20.2 4.6 14.5 1.9 4.1 Real GDP growth, at constant factor prices 9.9 10.4 10.2 8.4 8.9 8.6 Agriculture 7.1 5.4 6.4 1.1 4.7 5.1 Industry 24.2 17.0 21.8 15.3 13.0 12.1 Services 8.9 13.1 10.0 12.2 10.7 9.8 Inflation (Consumer Price Index) 13.5 8.1 7.7 9.6 8.9 9.2 Current Account Balance (% of GDP) –8.2 –7.9 –12.0 –10.7 –9.3 –8.9 Fiscal Balance (% of GDP) –1.9 –2.6 –2.4 –3.0 –3.2 –3.0 Debt (% of GDP) 36.7 40.8 48.1 54.4 57.4 58.0 Primary Balance (% of GDP) –1.6 –2.2 –2.1 –2.5 –2.6 –2.4 Poverty rate ($1.9/day PPP terms) a,b,c 29.6 28.3 27.2 26.4 25.5 24.6 Poverty rate ($3.1/day PPP terms) a,b,c 67.4 66.1 64.9 63.9 62.9 62.0 Sources: World Bank, Macroeconomics and Fiscal Management Global Practice, and Poverty Global Practice. Notes: e = estimate, f = forecast. a Calculations based on 2008-, 2014-, and 2010-HICES. b Projection using point-to-point elasticity at regional level with pass-through = 0.763 based on GDP per capita constant in constant LCU. c Projections are from 2013 to 2018. 27 TRENDS AND CONSTRAINTS IN URBAN LABOR MARKETS IN ETHIOPIA 2 line decreased considerably from 38.7 percent in A modest shift in labor from agriculture to services and construction explains up to a quarter of Ethiopia’s per 2000 to 29.6 percent in 2011. Poverty incidence has capita growth over the past decade. But while structural also fallen dramatically under the international line change contributed to economic growth in the past, it was of extreme poverty (the US$1.25 PPP poverty line): not sufficiently inclusive and needs to contribute much more to poverty reduction in the future. The urban space plays from 60.5 percent in 1996 to 30.7 percent in 2011. a key role in advancing structural change, as centers of A modest shift in labor from agriculture to services innovation and industrial development. Well-functioning and and construction explains up to a quarter of Ethiopia’s efficient urban labor markets are a key ingredient for this transformation to take place, and to ensure that its benefits per capita growth. World Bank (2015e) shows this posi- reach all segments of the population. Nonetheless, there are tive impact from structural change for the period 2005 several factors that currently curb the efficiency of urban labor to 2013. This illustrates the strong potential of structural markets in Ethiopia. This chapter examines recent trends in change as a driver of economic growth as discussed in the urban labor markets to show how the availability and nature of work has been changing, and what factors limit the efficiency literature (e.g. McMillan et al. 2014). Although Ethiopia of urban labor markets in Ethiopia. A clearer understanding of has experienced high economic growth and some struc- the nature of the labor market helps inform policy interventions tural change in production away from agriculture towards to provide better job opportunities for households, boosting private consumption, and thus, aggregate demand. services, the similar shift in employment has been much more modest. Agricultural employment declined from 80 to 77 percent between 2005 and 2013 but because agri- Introduction cultural labor productivity was so low, this shift gave rise to static efficiency gains as relative labor shares increased Rapid economic growth since 2004 was driven by in construction and services where the average value added public infrastructure investment and supported of a worker is up to five times higher (World Bank 2015e). by a conducive external environment. As shows While structural change contributed to eco- in the World Bank Systematic Country Diagnostics nomic growth in the past, it was not sufficiently (SCD), real GDP growth averaged 10.9 percent annu- inclusive and it has potential to contribute much ally (8.0 percent per capita) in 2004–14, according more to poverty reduction in the future. Structural to official data. This substantially exceeds historical change has had little impact on poverty reduction averages for the country as well as the regional and in Ethiopia given limited movements of labor across low-income averages for the same period. Growth was sectors and locations (World Bank (2016)). In other concentrated in services and agriculture with a recent words, progress among rural, self-employed, agricul- impetus from construction, while contributions from tural households in Ethiopia accounts for the major manufacturing were low (World Bank 2016). share of poverty reduction.30 Hence, looking forward, Economic growth, together with substan- tial improvements in the provision of safety nets 30 Poverty reduction in rural areas accounted for 5.2 and 7.8 percentage and basic services explain the impressive poverty points of poverty reduction during the periods 2000–2005 and 2005–2011 and the contribution of reductions in poverty among those engaged in reduction observed in Ethiopia in recent years. agriculture was similar: 3.8 and 6.9 percentage points respectively. Among The poverty headcount under the national poverty the self-employed it was: 4.8 and 7.5 percentage points respectively. 28 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET the Systematic Country Diagnostic argues that faster However, in urban areas, 15 percent of all households structural change is needed for several reasons. First, report an adult member—male or female—as unem- average and marginal productivity in agriculture is ployed. Unemployment increases in large cities as the much lower than in other sectors suggesting that size of formal labor markets increase. In particular, sectoral transitions will increase returns to labor. there are high unemployment rates in Addis Ababa; Second, the process of structural change offers the 23.5 percent of households report an unemployed potential shift into export-oriented light manufactur- adult, versus 11 percent of households in other urban ing activities. Third, urban demand plays an impor- areas. In other words, urbanization in Ethiopia can tant role in achieving agricultural growth, as would only bring about its structural impact if it is accompa- growth in agro-processing industries. Fourth, further nied by sustained increases in nonfarm employment. agglomeration as a result of urbanization will bring Nonetheless, there are several factors that cur- positive externalities that will encourage growth and rently curb the efficiency of urban labor markets help reduce poverty. Finally, structural transformation in Ethiopia, leading to very high levels of unem- can help reduce vulnerability. More inclusive struc- ployment. In rural areas, fewer than five percent of tural change is needed to ensure that non-agricultural all households have an unemployed adult. However, growth reduces poverty going forward. in urban areas, 17 percent of all households report Fostering faster and more inclusive structural an adult member—male or female—as unemployed. change in Ethiopia requires a healthy interplay Unemployment increases in large cities as the size of between three main actors: firms, workers and the formal labor markets increase. In particular, there are government. Productive and competitive firms are key high unemployment rates in Addis Ababa; 24 percent to success because firms are able to organize labor in of households report an unemployed adult, versus 11 the most efficient manner, thereby raising productivity percent of households in other urban areas. A clearer (World Bank 2016). Workers in turn should be healthy understanding of the nature of the labor market helps and possess the appropriate skills that firms need.31 The inform policy interventions to provide more and better government, particularly urban planners, play a pivotal job opportunities for households, help reduce poverty, role in this overall process as they plan the urban space, and allow the positive effects of urbanization unfold support investment in economic and social infrastruc- with increasing rates of non-farm employment. ture and manage the provision of public services includ- This chapter examines recent trends in urban ing education, training, health, and water/sanitation. labor markets to show how the availability and This support sets the foundation for private investment nature of work has been changing, and explores what in urban areas, which is essential for urban job creation. factors limit the efficiency of urban labor markets The urban space plays a key role to advance in Ethiopia. Evidence suggests that unemployment is structural change in Ethiopia, as centers of innova- partly explained by low labor productivity in the pri- tion and industrial development. Well-functioning vate sector. For those with little education, low labor and efficient urban labor markets are a key ingre- productivity in unskilled jobs causes wages to be low, dient for this transformation to take place, and to barely covering basic needs. If wages were any lower ensure that its benefits reach all segments of the than their current levels it would not be worthwhile for population. Yet, unemployment in Ethiopia is by and large an urban phenomenon, particularly in Addis 31 Yet, skills shortages in Ethiopia constitute a key constraint to growth and improved productivity in the manufacturing sector despite the Ababa and positive effects of urbanization can only country has made significant progress in expanding access to education. unfold with increasing rates of non-farm employment In the short run, the provisuion of Techncial and Vocaltional Education Training (TVET) and second-chance general education programs ccould (World Bank 2015b). In rural areas, fewer than five be used to bridge the gap of skills supply to the manufacturing sector. percent of all households have an unemployed adult. See the 4th Ethiopia Economic Update (World Babnk 2015c) for details. Trends and Constraints in Urban Labor Markets in Ethiopia 29 workers to take the work as they would not be able to out of work and available for work.33 This is high in cover their basic needs. This prevents wages from falling comparison to countries like Rwanda and Uganda, to fully clear the market. In addition, there is evidence where urban unemployment is around 7 percent and that while waiting for a high quality job educated indi- 9.5 percent respectively. In Addis Ababa the unem- viduals take temporary low-skilled jobs to finance the ployment rate is 24 percent (Figure 2.1.1). As a result relatively high costs of searching for a permanent job. Addis Ababa is home to 28 percent of the urban unem- The temporary jobs that are taken are those that low- ployed in Ethiopia. The 15 other major cities for which skilled individuals would otherwise take contributing the UEUS is representative are home to 16 percent to higher unemployment among low-skilled workers. of the unemployed and the remaining 56 percent of However the magnitude of the crowding out is rela- the urban unemployed live in smaller urban centers. tively small and this phenomenon on its own does not In addition, unemployment rates are particu- explain unemployment entirely. larly high for those with secondary education but no college degree (Figure 2.1.2). Given that those A Profile of the Labor Market in Urban with secondary education are not a large share of the Ethiopia32 urban labor force, those with primary or no education comprise the largest share (51 percent) of the unem- ployed pool. Unemployment rates are lowest among Ethiopia is urbanizing and as it does, the structure of work college graduates, but still 14 percent of the unem- is changing. Unemployment rates have been decreasing but remain high and increase with city size. Wages increase ployed are graduates. A quarter of these unemployed with education, particularly for those with post-secondary graduates live in Addis Ababa, 19 percent in other education, but women fetch 28 percent lower wages than major centers and 56 percent in smaller urban centers. men on urban labor markets. Despite rapid improvements in educational outcomes in recent years, the urban labor force Wages increase with education, particularly for remains predominantly comprised of low skilled workers. An those with post-secondary education, but women’s important trait of urban labor markets is the importance of wages are 28 percent lower than men’s wages. This the public sector, which has an important signaling function for the rest of the urban labor market. is shown in Figure 2.1.3. Not only are hourly wages lower for women but they also work fewer hours than men, which compounds the monthly earnings gap. Although unemployment has been decreasing, Women are no more likely to look for additional work it remains the most salient trait of the Ethiopian elsewhere than men (Figure 2.1.4). urban labor markets. Before exploring what factors Despite rapid improvements in educational may be contributing to this phenomenon by limiting outcomes in recent years, the urban labor force the efficiency of the labor market, first it is necessary to understand what characterizes this market and how it 32 A series of Urban Employment and Unemployment Surveys (UEUS) has evolved over the last few years. This section high- collected by the Central Statistical Agency (CSA) from 2003 to 2014 lights the low level of education of the labor force, the is, unless otherwise stated, the data used for this analysis. The data was standardized across survey rounds using the I2D2 protocols. Much of importance of the public sector in wage employment the analysis presents national trends for urban areas in Ethiopia using and the how self-employment is more or less attractive this data. The data is also used to construct city-level statistics for the 16 cities of which the data is representative: Mekelle (Tigray), Assayita (Afar), to individuals with different levels of education; while the Gondar Town (Amhara), Dessie Town (Amhara), Bahir Dar (Amhara), next section explores how real wages have not reflected Bishoftu (Oromiya), Adama (Oromiya), Jimma (Oromiya), Shashemene (Oromiya), Jijiga (Somali), Asossa (Benishangul-Gumuz), Awassa the increasing educational quality of the workforce, and (SNNPRs), Gambella Town (Gambella), Harari, Addis Ababa, Dire Dawa. how detrimental high inflation can be for real wages. 33 According to the 2014 Central Statistical Agency report on the UEUS, an individual is classified as unemployed if he is not engaged in a produc- Unemployment rates are high and increase tive activity (in wage, self, or family employment) and is available for with city size. Seventeen percent of the labor force is work the next month if the opportunity arises. 30 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 2.1: Unemployment and Wages 1) Unemployment rates by city size 2) Unemployment rates by education level 25% 24% 25% 22% 20% 20% 18% 17% 17% 16% 16% 15% 15% 11% 10% 10% 5% 5% 0% 0% Average Addis Other major All other No Education Primary Secondary Post-secondary/O urban Ababa cities urban 3) Wages and Education, 2014 4) Male premium (avg. 2003–2014) 25 35% 4.0 30% 3.5 Real Hourly wage (2003 Birr) 20 25% 3.0 15 2.5 20% 2.0 10 15% 1.5 10% 1.0 5 5% 0.5 0 0% 0.0 No Education Primary Secondary Post-secondary Real Real hourly Decreased More hours monthly wage probability worked wage premium of looking (hours, right for another job axis) Source: World Bank staff calculations using UEUS 2003–2014. Notes: 1) Other major cities includes Mekelle, Assayita, Gondar, Dessie, Bahir Dar, Bishoftu, Adama, Jimma, Shashemene, Jijiga, Asossa, Awassa, Gambella, Harari and Dire Dawa. 4) Controlling for employee education. remains predominantly comprised of low skilled employs 10 percent of urban workers, and manufac- workers (Figure 2.2.1). More than half of the urban turing employs 12 percent of urban employees. Those labor force has primary school education (36 per- with higher levels of education are more likely to be in cent) or no education (15 percent). This contrasts with the service sector, in part reflecting the fact that more countries such as Cambodia and Vietnam, countries educated workers are more likely to be employed in the of successful economic development through light public sector (Figure 2.2.4). Service sector employees manufacturing, where the median education of an are more likely to be self-employed. urban worker is secondary education. The services An important trait of urban labor markets is the sector is the most dominant employer, particularly for importance of the public sector. One in five urban the educated. Three-quarters of urban employees are workers are employed by the public sector, and these engaged in the service sector. This is true across cities are the workers with the highest levels of education. of all size, although the sector employs fewer indi- (Figures 2.2.5 and 2.2.6). Sixty-two percent of public viduals in Addis Ababa (Figure 2.2.3). Construction sector employees have post-secondary education in Trends and Constraints in Urban Labor Markets in Ethiopia 31 FIGURE 2.2: A Profile of Urban Labor Markets 1) Education 2) Wage and self-employment by city size 100% 100% 4% 15% 13% 11% 10% 14% 90% 21% 23% 24% 80% 80% 70% 26% 37% 39% 36% 29% 27% 45% 60% 60% 50% 40% 43% 40% 72% 30% 36% 35% 37% 49% 53% 20% 20% 41% 10% 16% 15% 15% 14% 0% 0% Working age Labor force Employed Unemployed Urban Addis Major Other population participants Ethiopia Ababa city urban No education Primary Secondary Post-secondary Wage emp. Self emp. Other 3) Sector of employment by city size 4) Sector of employment by education 100% 100% 90% 80% 80% 70% 75% 71% 75% 77% 60% 60% 50% 40% 40% 30% 20% 10% 12% 11% 20% 9% 12% 15% 11% 10% 10% 0% 0% Urban Addis Major Other No Primary Secondary Post- Ethiopia Ababa city urban education secondary Services Construction Manufacture Mining Agriculture 5) Public employment by city size 6) Public employment by education level 100% 100% 12% 28% 80% 80% 51% 53% 27% 58% 62% 60% 60% 50% 40% 40% 42% 28% 32% 20% 25% 20% 20% 21% 23% 22% 21% 18% 11% 0% 0% 2% Urban Addis Major Other Private Public Ethiopia Ababa city urban No education Primary Secondary Post-secondary Public Private (wage) Private (rest) Source: World Bank staff calculations using UEUS 2014. 32 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 2.3: Three Types in Urban Labor Markets Individual ability is shown on the x axis, and on the y axis Enterpreneurship the value of different activities. The blue line indicates the value of employment for individuals of differing levels of ability. The red line indicates the value of entrepreneurship and the black line indicates the value of searching for Value Employment employment (i.e. unemployment) for different levels of ability. The current value of unemployment is zero at any point in time but it secures employment in the future, so the Unemployment total value is positive and increasing with ability. aH aL Ability Type Low levels of education Secondary and tertiary education High levels of entrepreneurial ability Strategy These individuals determine These individuals realize that they These individuals realize they can earn they are better not searching for could earn more in the long run if higher returns if they choose to be self- work as it is costly and the wages they were employed rather than self- employed rather than employed. They would not compensate the cost. employed, even though they know this are successful entrepreneurs and employ They choose self-employment. will require spells of unemployment others. while they look for work. Outcome Petty trading, shoe-shining (“ne- Employed in blue- or white-collar jobs Owners of medium-sized enterprises cessity self-employment”) OR unemployed looking for work (“opportunity entrepreneurs”) Source: Poschke (2014). comparison to only 12 percent of those employed in Those in employment are evenly split between the private sector. wage and self-employment, but in larger cities wage Moreover, public sector employees are a large employment comprises a larger share of employees. share of wage employees: from 1 in 2 wage employees On average, 49 percent of urban employees are in in smaller urban centers to 1 in 3 wage employees in wage-employment, 39 percent are in self-employment Addis Ababa. Although public sector workers do not and 11 percent are unpaid family workers, or appren- dominate the urban labor force in any city, this group tices (Figure 2.2.2). The share of wage employment is a significant share of wage employees, particularly for increases with city size, in Addis Ababa, more than small cities. As cities increase in size the private sector 70 percent of workers are in wage employment. The becomes larger and the public sector is a less impor- cross-sectional UEUS data used in this analysis pro- tant source of wage-employment. In Addis Ababa for vides a view at one point in time, but longitudinal example, 50 percent of the workforce are private sector studies in Addis Ababa show that many individuals wage-employees compared to the 23 percent that are move in and out of work and types of work (Franklin public sector employees (Figure 2.2.5). However, even 2014). Usually, self-employment is less attractive for in large cities the magnitude of the public sector as an individuals with secondary and tertiary education employer is significant. Ethiopia also has a much larger but more attractive for those with low level of edu- share of wage employees in the public sector than many cation and high levels of entrepreneurial ability (see of its peers. One in five wage jobs in Uganda, Rwanda, Figure 2.3, also Annex 3 presents a more detailed dis- and Ghana are in the public sector. cussion of self-employment). Trends and Constraints in Urban Labor Markets in Ethiopia 33 Labor Market Trends urban areas) and rural-urban migration (World Bank 2015b). Although rates of rural-urban migration have been relatively low, it has been encouraged by a large Real wage trends in urban Ethiopia have not reflected the welfare premium experienced by migrants to urban increasing educational quality of the workforce. Until 2012 there was a substantial reduction in the returns to education areas (de Brauw et al. 2013) and high rates of agricul- and other worker characteristics observed over the last tural growth, which have helped households finance decade. These changes are likely a reflection fixed nominal migration (de Brauw 2014). public wages in the presence of very high inflation rates. High Underemployment, while also too prevalent, rates of inflation were a factor in the erosion of the value of real wages in the public sector in the past; and it is because of has been falling. Underemployment rates were as high past inflation that nominal wages for public servants had to 52 percent in urban Ethiopia in 2003 and this fell to be adjusted significantly in 2014/15 (see Chapter 1). Given 43 percent in 2014 (Figure 2.4.3). Although there that the public sector is such a major employer this affects the returns on education in urban labor markets, and also has been progress, this is still a staggeringly high rate seems to have an effect on private sector wages. of underemployment. Underemployment has fallen much faster in Addis Ababa than elsewhere, suggest- Growth in Employment in Urban Ethiopia ing that those that are in work are more able to fulfill their desired hours. In 2014 underemployment fell to Labor market trends in urban Ethiopia have been 31 percent In Addis Ababa from 52 percent in 2003. improving over time. Figure 2.4.1 shows that the labor force participation rate has increased consis- A more Educated Workforce, but Little tently since 2004 from 55.6 percent of the population Change in the Nature of Employment to 63.9 percent of the population. Women’s labor force participation rate is lower than men’s, by about Although, many labor market outcomes show 13 percentage points, but this difference has been improving trends, there has been no observable consistent across time implying equal growth rates. structural change in the nature of work in urban The male wage premium has fallen over time from Ethiopia. The share of wage and self-employment has 40 percent in 2003 to 28 percent in 2014, suggesting stayed constant in urban Ethiopia: 48 percent of the some progress over time in addressing the earnings gap workforce was wage employed in 2003 and 49 percent between men and women. in 2014 (Figure 2.5.1). This is potentially concerning Although unemployment rates are high, there given evidence suggests that it is wage employment, has been clear progress with unemployment rates not self-employment, that leads to the emergence of decreasing in all cities since 2004. The average unem- a middle class in developing countries (Banerjee and ployment rate across urban Ethiopia was 22 percent in Duflo 2008). Similarly, there has been almost no 2003 and fell to 16 percent in 2014 (Figure 2.4.2). The change in the composition of labor employed in ser- unemployment rate in Addis Ababa was 33 percent vices, construction and manufacturing (Figure 2.5.2). in 2003 and fell to 24 percent in 2014. This suggests that the construction boom has not The supply of labor in urban areas has been resulted in a change in the sectoral composition of the increasing, driven by demographic changes and workforce with other sectors increasing their employ- migration; the supply of jobs has increased to ment at an equal rate. It also suggests that the structural match this. The urban population in Ethiopia change observed by Martins (2015) is not a result of increased from 9.8 million in 2000 to 18.4 million in the changing nature of employment in urban areas, 2014, according to World Bank (World Development but rather a shift in the labor force from rural to urban Indicators). This has been driven by natural urban Ethiopia (as suggested in the World Bank’s Ethiopia population growth (aided by low mortality rates in Urbanization Review). The share of private sector 34 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FIGURE 2.4: Trends in Employment in Urban Ethiopia 1) Labor force participation rising for men and women 2) Unemployment falling 75% 35% 33% 69.5% 70.7% 29% 30% 29% 70% 67.6% 30% 27% 65.7% 66.9% 65.2% 27% 25% 65% 63.2% 63.1% 63.9% 25% 23% 23% 24% 60.6% 60.7% 21% 19% 60% 58.3% 59.6% 60.3% 59.7% 20% 22% 18% 18% 17% 17% 57.6% 55.6% 19% 55% 53.5% 54.0% 57.5% 15% 18% 14% 14% 15% 16% 54.5% 53.1% 53.0% 13% 50% 10% 50.6% 45% 5% 40% 0% 2003 2004 2006 2009 2010 2011 2012 2014 2003 2004 2006 2009 2010 2011 2012 2014 LFP LFP (Male) LFP (Female) Average urban Addis Ababa Other major cities All other urban 3) Underemployment is high but falling 55% 52% 48% 50% 49% 46% 47% 45% 43% 42% 41% 45% 43% 43% 40% 39% 35% 31% 31% 30% 25% 2003 2004 2006 2009 2010 2011 2012 2014 Average urban Addis Ababa Other major cities All other urban Source: World Bank staff calculations using UEUS 2003–2014. jobs has been constant, indicating that the new jobs in 2003, 4 percent had a post-secondary degree while have been created by the private and public sector in this increased to 13 percent in 2014. The public sector proportion to their share of labor force (Figure 2.5.3). has absorbed much of the increase in educated work- However, the workforce has become more ers. The share of public employees who are graduates educated over time as young entrants to the labor increased from 24 percent in 2003 to 62 percent in market attain more years of education. The share of 2014. See Annex 4 for a more detailed discussion. employees with post-secondary qualifications more than doubled from 9 percent in 2003 to 23 percent in Real Wages are Rising, but not Enough to 2014 (Figure 2.5.4). The share of those with primary Compensate for Earlier Declines education and secondary education stayed the same, but there was a sharp reduction in the share of employ- Real wages have not risen despite an increasingly ees with no education. The share of employees with no educated workforce: monthly and hourly real wages education fell from 26 percent to 15 percent. The pool fell until 2012, with some recovery since then. The of unemployed has also become increasingly educated: recovery in the hourly wage appears stronger than the Trends and Constraints in Urban Labor Markets in Ethiopia 35 FIGURE 2.5: A More Educated Workforce but Little Structural Change 1) No change in the balance between wage 2) No change in the sectoral composition of labor, and self-employment, all urban Ethiopia all urban Ethiopia 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% 2003 2004 2006 2009 2010 2011 2012 2014 2003 2004 2006 2009 2010 2011 2012 2014 Wage emp. Self emp. Other Agriculture Mining Manufacture Construction Services 3) The share of private sector jobs has been 4) Educational composition of the labor force, constant, all urban Ethiopia all urban Ethiopia 100% 100% 28% 10% 26% 11% 28% 9% 15% 17% 18% 19% 23% 80% 80% 28% 27% 25% 26% 27% 60% 60% 35% 36% 37% 35% 35% 34% 40% 40% 35% 35% 20% 20% 28% 27% 26% 22% 21% 21% 19% 15% 0% 0% 2003 2004 2006 2009 2010 2011 2012 2014 2003 2004 2006 2009 2010 2011 2012 2014 Public Private (wage) Private (rest) No education Primary Secondary Post-secondary/Other Source: World Bank staff calculations using UEUS 2003–2014. increase in monthly wages on account of the fall in for temporary and casual workers, have been most the number of hours worked recorded between 2012 important in increasing average wages. Examining the and 2014. Average real hourly and monthly wages distribution of real wages in the private sector shows in the private sector remained almost constant. Real little change until 2012. From 2012 to 2014 there wages increased in both the public and private sector was a reduction in those paid less than 1 Birr (2003 after 2012, particularly hourly wages (Figures 2.6.1 prices) an hour and an increase in those paid between and 2.6.2). 1 and 3 Birr an hour (Figure 2.7.2). In 2014, earn- The increase in real wages post 2012 is observed ings were about 40 percent lower for temporary and in all sectors within the private sector, and all sub- casual workers compared to permanent and contract sectors within manufacturing. Hourly earnings in workers. Over the entire period 2006–2014, earnings manufacturing declined from 1.44 Birr (2003 Birr) per of permanent/contract workers declined, while those hour in 2006 to 1.24 Birr per hour in 2012, and then of temporary/casual workers increased (Figure 2.7.3). increased to 1.78 Birr per hour in 2014 (Figure 2.7.1). The trend followed by real wages may in part be Changes at the lower end of the income distribution, explained by labor productivity in the private sector, 36 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Nominal and Real Wage Trends FIGURE 2.6:  1) Monthly real wage, all urban Ethiopia 2) Hourly real wage, all urban Ethiopia 400 3.0 356.9 354.4 336.2 2.5 2.4 350 289.8 297.1 2.5 300 293.1 282.7 262.7 2.0 2.0 1.9 2.0 1.8 250 200 1.5 200.0 200.0 202.6 196.1 1.5 150 185.9 195.4 182.3 187.4 1.4 1.3 1.3 1.0 1.3 1.1 100 0.5 50 0 0.0 2003 2004 2006 2009 2010 2011 2012 2014 2006 2009 2010 2011 2012 2014 Mean Median Source: World Bank staff calculations using UEUS 2003–2014. FIGURE 2.7: Real Wage Trends by Sector, Sub-Sector and Wage Level 1) Average hourly real wages by sector 2) Distribution of real wages over time 3.5 1.0 3.0 0.8 2.5 2003 Birr Density 0.6 2.0 0.4 1.5 1.0 0.2 0.5 0.0 0.0 2006 2009 2010 2011 2012 2014 0 2 4 6 8 10 Manufacture Construction Services 2006 2012 2014 3) Real wage trends in manufacturing 3.0 2.5 2.0 1.5 1.0 2006 2008 2010 2012 2014 Year of survey All employees Permanent/contract workers Temporary/casual workers Source: World Bank staff calculations using UEUS 2003–2014. Trends and Constraints in Urban Labor Markets in Ethiopia 37 but also reflects changes in public sector wages. Real unusual for a low-income country. What underpins wage trends have not reflected the increasing educa- the high rates of unemployment that characterizes the tional quality of the work-force with a substantial reduc- urban labor markets? tion and then some increase in the returns to education Several hypotheses have been posed about and worker characteristics observed over the last decade potential labor market imperfections in urban suggested by Oaxaca-blinder decomposition analysis. centers, which may help explain high unemploy- Real wages paid by manufacturing firms do, broadly, ment rates. Some have argued that, as in other African reflect changes in worker productivity. However, a countries, high unemployment in Ethiopia is related deeper examination of both employment and firm data to a queuing phenomenon for good quality jobs, suggests that in Ethiopia, wage trends wages do not fol- which creates distortions that prevent the labor market low average labor productivity within sector or across from clearing (Serneels 2007, Krishnan 2000). Other cities (see Annex 5 for a detailed analysis). Real wage plausible hypothesis that could explain this include: trends are likely a reflection of fixed nominal public high search costs in the process of looking for a job wages in the presence of very high inflation rates. This (Franklin 2014), and high reservation wages among affects private sector wages given the importance of the potential entrants to the labor market. In addition, public sector as an employer in urban labor markets. others associate high unemployment and poverty rates to an oversupply of labor in comparison to the num- Constraints in Urban Labor Markets: ber of jobs being created, and to the growing flows of Explaining High Unemployment migration to urban centers from rural areas (Bimerew 2015; Asmame 2011). Background This section explores whether the labor data in Ethiopia is consistent with these various hypoth- eses in order to arrive at an improved understand- Several factors contribute to curb the efficiency of urban ing of the nature of the problem. Some hypotheses labor markets in Ethiopia. Unemployment has fallen but are tested more thoroughly than others due to data remains high, and the evidence suggests this is partly constraints, and in some cases, evidence from previ- explained by low labor productivity in the private sector. While waiting for a permanent job, educated individuals ous studies on particular subjects are also included. take temporary low-skilled jobs that low-skilled individuals would otherwise take. Low productivity in the private sector translates into very low wage levels for the less skilled, Is Unemployment Explained by Queueing preventing wages from adjusting to clear the market. These for Permanent Jobs? issues are aggravated by the high costs of searching for a job and the rural-urban migration influx. Finally, there is High rates of urban unemployment in low and evidence that there are not enough opportunity for those with primary and secondary education in urban Ethiopia. middle income economies are often attributed to college educated workers queuing for highly desired, permanent jobs. In the past, unemployment in urban As mentioned earlier, unemployment rates are per- Ethiopia has also been related to a queuing phenomenon sistently high in urban Ethiopia and especially in for good quality jobs (Serneels 2007, Krishnan 2000). Addis Ababa, at 17 percent and 24 percent respec- Whilst queuing may explain unemployment tively in 2014.34 The rates have fallen in recent years, among graduates in urban Ethiopia, the majority consistent with the impressive record of economic growth that the country has exhibited during the last 34 Following the official definition of unemployment in Ethiopia: some- decade. Nonetheless, such high rates of unemploy- one not engaged in a productive activity in the past 7 days, and available ment, particularly among low skilled workers, are for work if the opportunity arises next month. 38 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET of urban unemployment is among those without unemployment among those without college a college degree, for whom queuing is an unlikely degrees. The size of the effect of temporary employ- explanation. The unemployment rate is 11 percent ment of educated workers on unemployment in other among college educated workers, but twice as high for groups is not large enough to fully explain the high those with secondary education, and still 16 percent unemployment rates observed. Additionally, if no for those with no education at all (Figure 2.1.2). other labor market frictions were present, real wages Queuing can also have impacts on workers in the unskilled sector could fall to clear the market without college degrees: in order to finance con- of unskilled workers. This does not happen, suggest- tinued search for a permanent job, the educated ing that other frictions are preventing this segment of take unskilled temporary jobs for which they are the labor market from clearing. The evidence points overqualified. Job search is an expensive activity in to the presence of other labor market constraints that urban Ethiopia and tracking surveys show that those also contribute to high unemployment rates. looking for permanent white-collar employment take temporary employment in construction and other Do the Unemployed have Unrealistic sectors in order to finance their continued search Expectations about Wages? (Franklin 2014). This phenomenon may explain some of the high turnover rates reported in manufacturing Unemployment can result if individuals looking for jobs (Blattman and Dercon 2015; World Bank 2015c). a job have a reservation wage below which they are When educated workers take unskilled tem- not willing to work and this wage is above the labor- porary jobs to fund search for permanent jobs market clearing wage. Reservation wages can be above they crowd out unskilled workers who would oth- the labor-market clearing wage as a result of unrealistic erwise be well suited for these jobs. This is shown expectations. If reservation wages did exist, and if they in Figure 2.8. A ten percentage point increase in the were somewhat uniform for those of similar education city-wide temporary employment rate of the most levels, the wage distribution among similarly educated educated increases the unemployment rate by 1.6 per- individuals would exhibit a sharp drop in the propor- centage points for those with primary education, and tion of individuals employed below a given wage level. by 1.7 percentage points for those with secondary There is some evidence of reservation wages education. No significant effect is observed for those emerging among more educated workers—par- with no education, a group for whom unemployment ticularly among those with secondary education. rates are much lower. Figure 2.9 shows the evolution of the real hourly wage However, this crowding-out cannot, on distribution of workers belonging to the following its own, explain relatively high rates of urban education levels: no formal education, some or com- plete primary education, some or complete secondary FIGURE 2.8:  Queueing, Temporary education, and some or completed post-secondary Employment and Unemployment education. The graphs clearly indicate a decline in real wages for these groups from 2006 to 2011. For 0.40 the more educated, there is an increasing drop-off in Percentage Points 0.16 0.17 the number of workers working below a given wage 0.08 0.00 –0.09 over time, which supports the existence of reserva- tion wages among this group. This effect is stronger –0.40 among those with secondary education than among Uneducated Primary Secondary Post-secondary those with post-secondary education. Unemployment Source: World Bank staff with UEUS 2003–2014. rates are higher for those with secondary education Trends and Constraints in Urban Labor Markets in Ethiopia 39 FIGURE 2.9: Wage Distribution of Workers with Different Education Levels 1) Real Hourly Wage, 2006–2014 2) Real Hourly Wage, 2006–2014 No Education Primary (some or completed) 1.5 0.8 0.6 1.0 Density Density 0.4 0.5 0.2 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 kernel = epanechnikov, bandwidth = 0.0920 kernel = epanechnikov, bandwidth = 0.1629 3) Real Hourly Wage, 2006–2014 4) Real Hourly Wage, 2006–2014 Secondary (some or completed) Post-Secondary/Other 0.6 0.3 0.4 0.2 Density Density 0.2 0.1 0.0 0.0 0 2 4 6 8 10 0 2 4 6 8 10 kernel = epanechnikov, bandwidth = 0.3370 kernel = epanechnikov, bandwidth = 0.4992 2006 2011 2014 Source: World Bank staff with UEUS 2003–2014. suggesting that the reservation wage of individuals in may be workers willing to work for a lower wage, this group may be unrealistically high. firms won’t offer lower wage as they know it will not Wages of less educated workers are also clus- be enough for the workers to perform normal physi- tered, but the level they are clustered around is cal activities. This helps explain why the wages for the very low. For 2014, the daily nominal wage for the unskilled labor segment do not fall to clear the mar- 25th percentile of individuals with no education was ket. It also helps explain why unemployment among approximately 10.5 Birr per day, which is not very dif- unskilled workers has fallen in the presence of rising ferent from the nominal poverty line a day for Ethiopia real wages, as has been the case from 2012 to 2014. of 13.7 Birr per day (in 2014 prices). As reference wages rise there is more room for wages The clustering of wages around a very low level to fall at the low end of the market while not falling for less educated workers indicates that wages can- below the nutritional minimum. not fall because they need to meet minimum nutri- In the private sector, low wages largely reflect tional requirements, not that reservation wages are Ethiopia’s very low rates of labor productivity, some too high. Wages cannot fall any further to clear the of the lowest in the world. Ethiopia has some of the market for unskilled jobs, because they are already lowest rates of labor productivity in the world (World clustered below the food poverty line. Although there Bank 2015c), which requires wages to be similarly low 40 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET in order for firms to be competitive. Real wages paid The cost of traveling to job boards is not trivial by manufacturing firms do, broadly, reflect changes in for the unemployed, who live a median distance of worker productivity, although by no means perfectly. 5.5 km from the city center, and who state that this Increasing labor productivity is key to improving constrains their search. The cost of a trip to search the functioning of urban labor markets and reducing for work in the town center could be as much as 15 unemployment among the less educated. Aiding firm Birr in total (US$0.8), and on average respondents competitiveness should help address this constraint, took about two trips per week; compared to a median but equally crucial is improving the educational quali- total weekly expenditure of 100 Birr. Consequently, fication of workers and on-the-job skill formation to job seekers devoted around 25 percent of their bud- increase productivity. Mengistae (1999) shows that on get to transport, a sizeable portion considering that the job skill formation can have a large impact on both they spend around 40 percent in food. Not surpris- marginal productivity and wages of workers. ingly, 50 percent of the respondents reported that transportation costs are the main constraint for their Is Job-search too Costly to be Worthwhile? job search, and 84 percent said that these costs were a constraint for their job search (Figure 2.10.1). The High costs of job search are another labor market costs of looking for a job over a sustained period of friction that may contribute to the high level of time may explain why educated job seekers that can- unemployment in Ethiopian cities. A locational not find a permanent work may accept low quality mismatch between the unemployed and the location temporary jobs for some time (Franklin 2014). of firms, together with a lack of an efficient and com- prehensive system of matching potential candidates Are the Right Kinds of Jobs Being Created? with job opportunities, can negatively affect the pace at which employment is created and vacancies are High levels of unemployment could also be the filled. This might be amplified in large and disperse result of mismatch between the types of workers labor markets, such as Addis Ababa. looking for jobs and the types of jobs being created. Among the unemployed, centrally located job A recent report on Ethiopia’s manufacturing sector boards are one of the most common method of points out that the shortage of skills is a key constraint searching for a job. An Oxford-EDRI survey on for firms to grow and increase their productivity, and unemployed youth in Addis Ababa finds 53.1 percent that firms often report struggling to find the technical use this method in the last week, and 75.7 percent in and soft skills they require when recruiting candidates the last 6 months (Franklin 2014). Usually these job (World Bank 2015c). Given the high rates of unem- boards are located in the city center of Addis Ababa. ployment also present in urban areas, this raises the Newspaper ads are also a common search method, question of whether the right types of jobs are being with 16 percent of the sample reporting having used created. To assess the magnitude of this problem, in the last week, although sometimes respondents data on firms hiring plans collected in a 2014 EDRI- reported travelling to the job board to look at the University of Oxford survey of firms of all sectors in most updated newspaper ads. While job boards and Addis Ababa is compared to the pool of unemployed in newspapers are the preferred method of search for the the 2014 UEUS. Firms were asked to report the type of more educated segment of the sample (those with a vacancies they were trying to fill. The education profile diploma or a degree), those with lower educational of the current holders of these types of vacancies was background rely on more informal search methods, used to estimate the education profile of the vacancies such as visiting work sites and asking for help through they were expecting to hire for. This provides a first social networks. approximation as to whether the firms are looking to Trends and Constraints in Urban Labor Markets in Ethiopia 41 FIGURE 2.10: Job Search, Constraints, and Vacancies 1) Constraints to Job Search 100% 84 80% 60% 50 40% 20% 0% Costs of transport is a major Costs of transport is a constrain for job search constrain for job search 2) Number of workers firms expect to 3) Matching vacancies to workers by education, hire by education, Addis Ababa 2014* Addis Ababa 2014 30,000 140,000 25,000 120,000 100,000 20,000 80,000 15,000 60,000 10,000 40,000 5,000 20,000 0 0 Primary Secondary Post Primary Secondary Post Construction Education Finance Unemployed, match Unemployed, no match Manufacturing Tours-hospitality Source: 1) EDRI and Oxford Unemployment Survey, 2013. 2) EDRI and Oxford University survey of firms and World Bank staff using UEUS data, 2014. * The total number is adjusted by sector specific weights for the simple frequencies. Only firms larger than 40 people were interviewed. fill positions with workers that have different profiles education. The construction sector represents the than those that the currently unemployed have. highest demand (40 percent) for individuals with Individuals with post-secondary education are primary education. most in demand by hiring firms, followed by work- The numbers suggests that there are not enough ers with just primary education. Those with second- job opportunities for those with primary and sec- ary education are least in demand. Figure 2.10.2 ondary education in urban Ethiopia. The number shows that there is a large demand for individuals of unemployed per education level (supply) in Addis with post-secondary education in comparison to the Ababa is compared to the number of total vacancies per demand for those with primary education and second- education level (demand) in Figures 2.10.2 and 2.10.3. ary education. Not surprisingly, the financial and edu- The figures show that the number of unemployed cation sectors present the highest demand for skilled individuals with primary and secondary education is labor, accounting for 40 percent and 25 percent of considerably larger than the number of vacancies for the total demand for individuals with post-secondary these education levels. The firm survey only collected 42 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET data from firms with 40+ employees, so it does not which might result in overcrowding and high unem- fully reflect demand, but the magnitude of the gap ployment rates in urban areas. It also predicts that even suggests is unlikely to be fully filled by smaller firms. in a scenario of high unemployment in urban areas, As previously discussed, the market does not clear migrating can still be a rational strategy that maximizes for these workers because of overly-high expectations the expected value of wage income, depending on the about wages among the secondary educated and low wages and probabilities of obtaining employment in levels of labor productivity among unskilled workers. the two labor markets (urban and rural). These findings also concur with other studies There is limited data on migration trends that have shown that lack of skills for technical in Ethiopia with which to test this hypothesis. professions is a considerable cause for concern in Recently, data on migration was collected in the 2007 Ethiopia. In the case of post-secondary education, Population and Housing Census and in the 2013 the supply for labor was only slightly larger than the Labor Force Survey. The relationship between city- demand for labor. The firm survey only collected data level rates of migration and unemployment is explored from firms with 40+ employees, so it does not fully using this data together with subsequent waves of reflect demand, suggesting demand may not be too low the UEUS collected in 2009 and 2014. Specifically, for these workers. This finding would be consistent with city level migration rates for the 16 cities for which the findings of a World Bank assessment on skills and the UEUS is representative are calculated, and the competitiveness in Ethiopia that finds that vacancies relationship between migration and an individual’s for skilled production worker and managerial positions probability of finding a job is estimated. stay open for long periods of time and that 57 percent Existing data suggests that individuals living in of firms identify lack of appropriate applicants as the cities with a higher migration ratio are more likely key reason for leaving vacancies open for a long time to be unemployed, although this may simply reflect (World Bank 2015c).35 However, there is a different local economic conditions. An increase of 10 percent- type of skill mismatch among this group. An even larger age points of the proportion of the city population challenge among this group is that those hired do not that migrated to the city in the last year increases the have the skills needed for the job—both technical and individual probability of being unemployed by 5.5 per- soft skills—and require training (World Bank 2015c). centage points. An identical increase in the proportion This results in firms engaging in substantial training of the city population that migrated to the city in the costs, which in turn makes hiring more costly and curbs last five years increases the probability of unemploy- the rate at which firms create new positions. ment by slightly less, around 1.9 percentage points (Table 2.1). However, it could be the case that local Does Rural-urban Migration Contribute to economic conditions affect both migration rates and Unemployment? labor market conditions, and the correlation between the two reflects this. Therefore, these results should be Many consider that increasing rates of rural migra- interpreted with care and not in a causal sense. tion have contributed to increased rates of urban In addition, on its own migration cannot explain unemployment (Bimerew 2015 and Asmame 2011, high rates of unemployment as wages should fall to among others). The effects of national and interna- match the additional supply of labor with demand. tional migration on labor markets have been widely Other constraints, such as high reservation wages, low studied in the economic literature. The Harris-Todaro rates of labor productivity or high search costs are model predicts that a positive expected value of the needed for high rates of unemployment to be explained. urban-rural wage income differential should result in positive migration flows from rural to urban areas, 35 See also World Bank 2014, page 4 Trends and Constraints in Urban Labor Markets in Ethiopia 43 TABLE 2.1: City-Level Migration Rates and employment is becoming slightly more prevalent, Unemployment (marginal Effects) particularly in Addis Ababa. The gender gap in labor (1) (2) participation (13 percentage points) as well as the proportion of wage employment in total employ- Probability of being Probability of unemployed being unemployed ment is similar to some structural and aspiration Migrant ratio less 0.551*** peers to Ethiopia. New jobs being created equally by than 1 year [0.211] the private and public sector and in proportion with Migrant ratio less 0.188** the existing sectorial composition of the labor force. than 5 years [0.0800] Real wage trends in urban Ethiopia have not Observations 38,491 38,491 reflected the increasing educational quality of the Pseudo R-squared 0.097 0.097 workforce, although improvements were observed City FE Yes Yes from 2012 to 2014. Wages are higher for those with Time FE Yes Yes more education—in 2014 wages were more than Robust standard errors in brackets, clustered by city. double for those with a degree compared to those with a secondary education—but returns on educa- *** p<0.01, ** p<0.05, * p<0.1 tion have fallen over the last decade. The decline in Source: World Bank staff calculations using the 2007 Population and Housing Census, the 2013 Labor Force Survey, UEUS 2009 and 2014. wages at the beginning of the decade was somewhat An individual’s age, gender, marital status, education level and house- compensated, although not fully, by increasing wages hold size is controlled for. City and year fixed effects are also included. in the last two years (2012–2014). These changes may in part reflect changes in worker productivity, but they Short Summary and Recommendations are also likely a reflection of changes in the nominal public sector wage structure, which also affects private Ethiopia’s urbanization implies that understand- sector wages given its importance as an employer in ing the nature of urban labor markets provides urban labor markets. valuable insights for further poverty reduction The phenomenon of queuing for high qual- and for a successful transition to a manufacturing ity jobs may contribute to unemployment among and service-oriented economy. Urban labor markets skilled and unskilled workers but it does not differ from rural markets in several aspects: in large explain unemployment entirely. While waiting for urban centers, wage employment is more important permanent employment educated individuals take than self-employment, underemployment is lower but temporary low-skilled jobs to finance the relatively unemployment rates are high (particularly among the high costs of searching for a permanent job. The tem- young, moderately educated population). Almost half porary jobs that are taken are those that low-skilled of wage employees are employed by the public sector. individuals would otherwise take. The relationship The dominant sector in urban areas is the service sec- between temporary employment and unemployment tor, and it increases in importance among the educated among unskilled workers is present, but not large and also in smaller towns. Manufacturing is more enough to explain much of the unemployment, important in larger cities and among the unskilled. Higher levels of productivity in low-skilled jobs Many urban labor market trends are moving are desperately needed. Wages of those with little in the right direction although there has been little education are bunched around a very low level calling change in the structure of urban labor markets over in to question whether they are able to fall any further time. Labor participation is increasing, and unemploy- to clear the market and still cover the basic needs of ment and underemployment are falling. Ethiopia’s workers. Increasing labor productivity in these jobs labor force is becoming increasingly educated. Wage will allow the marginal product of labor increases 44 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET above the nutrition-based wage, enabling wages can accounting for almost 25 percent of their monthly adjust as needed in order to clear the market. expenditure. Increasing access to information on Although jobs are being created faster than job vacancies throughout the city through the use of growth in the urban workforce, not enough jobs are technology can reduce the cost of searching. Targeted being created for those with primary and secondary safety nets and labor market programs can also help education. By analyzing how the demand for labor reduce unemployment by investing in the skills of the compares to the supply of labor in Addis Ababa, it unemployed, funding the cost of search, and improv- becomes clear that there are not enough job opportu- ing the quality of matching. nities for those with primary and secondary education The following specific policy recommendations levels. This might have been aggravated by the increas- follow from these conclusions on the urban labor ing migration from rural to urban areas. Existing data analysis: suggests that individuals living in cities with a higher migration are more likely to be unemployed, although 1. Encourage firm creation and firm growth that creates this may simply reflect local economic conditions. jobs for non-graduates. This will require a focus on There is a need for both faster job creation for growth in services such as hospitality, as well as in non-graduates and investment in skills. More and construction and manufacturing, areas as these are better job opportunities for those with secondary sectors that are more likely to hire non-graduates. and primary education are required. This segment of It will also require a focus on interventions that the market is in desperate need for high-productivity reduce costs associated with hiring non-graduates, employment creation. Although low-skilled workers such as certification of skills for secondary workers. are more likely to be employed in manufacturing and 2. Increase labor productivity in the low-skill segment construction than high-skilled workers, the service by addressing constraints faced by firms in access- sector is still the primary employer of those with ing capital (financial and physical) to ensure that little education; growth in the service sector will be the marginal product of labor increases above the needed for job growth for non-graduates. Investment nutrition-based wage. in skills of those in the job market is also needed. 3. Invest further in job training and technical train- Many employers report delays in finding employees ing programs to build the skills of those in the job with the right skills and report needing to invest in market, both for low skilled workers to increase training. This will likely require investment in on-the- their productivity and at higher levels of education. job training as much as in formal training programs. 4. Introduce targeted urban safety nets and labor Costly job search also contributes to high market programs to invest in skills of low-skilled unemployment rates and improving the technol- employees and the unemployed and provide finan- ogy of job search can help address this. The nature cial support to enable their job search. of the job search—searching for vacancies posted on 5. Enhance the use of ICT to provide information physical job boards at specific points in the city— on job vacancies throughout the city and reduce entails high transportation costs for the unemployed, the cost of job search. Annex 1: Ethiopia: Selected Economic Indicators (High Frequency) Oct-15 Nov-15 Dec-15 Jan-16 Feb-16 Mar-16 Apr-16 May-16 Jun-16 Jul-16 Aug-16 Sep-16 Oct-16 Inflation (Year-on-Year): % 11.9 10.0 10.0 10.2 8.9 7.5 7.4 7.9 7.5 6.0 5.9 6.9 5.6 Food 16.2 11.5 12.1 12.4 9.2 7.3 6.2 8.4 7.2 4.1 4.3 6.1 3.4 Non-Food 7.3 8.5 7.8 8.1 8.7 7.9 8.7 7.4 7.9 8.3 7.6 7.8 8.2 Monetarnex1y Growth (Year-on-Year): %                         M2 23.8 22.7 21.7 18.8 18.3 18.0 17.5 18.2 19.9 19.1 20.2 Domestic credit 27.5 27.7 26.9 27.4 26.0 24.9 26.9 22.8 24.6 23.4 25.6 Net Foreign Assets –11.4 –24.3 –20.1 –51.4 –32.4 –29.2 –45.2 –36.0 –42.7 –54.3 –58.0 Reserve Money* 5.0 9.0 5.3 9.4 11.3 10.3 11.7 19.9 16.3 27.7 24.6 Gross reserves (Mill. $) 3049 3985 3980 3677 3660 3412 3072 3323 3402 3617 3485 In months of import 1.7 2.2 2.2 2.8 2.6 2.1   2.6 2.4 3.3 Exchange rate                     Exchange rate (Birr/$), pa 20.9 21.0 21.1 21.2 21.2 21.3 21.4 21.6 21.8 21.8 21.9 22.0 22.1 Real Effective Exchange Rate index 164.2 166.0 167.9 171.1 170.3 167.8 166.6 168.5 169.8 annual growth, % 20.6 18.8 17.4 15.8 12.0 7.5 6.2 8.6 7.8 Black market premium (%) 11.4 12.1 15.4 14.3     Trade Deficit, goods, billion US$                     Trade Deficit, goods, billion US$ –1.1 –1.0 –1.4 –1.1 –1.1 –1.1 –1.4 0.0 –1.0 –1.1 –13.9 Export, (billion US$) 0.2 0.2 0.2 0.2 0.2 0.3 0.3 0.0 0.3 0.3 2.8 Import, (billion US$) 1.3 1.2 1.6 1.3 1.3 1.4 1.7 0.0 1.3 1.4 16.7 International Prices                       Crude oil, average ($/bbl) 47.0 43.1 36.6 29.8 31.0 37.3 40.8 45.9 47.7 44.1 44.9 45.0 Coffee, arabica (/kg) 3.4 3.3 3.3 3.2 3.3 3.5 3.4 3.4 3.6 3.8 3.7 3.9 Gold ($/troy oz) 1159.3 1086.4 1068.3 1097.9 1,200 1245.0 1242.3 1261.0 1276.4 1336.7 1340.2 1326.6 World Growth (quarterly: y-o-y) %     Q4     Q1     Q2     Q3   China 6.8 6.7 6.7 6.7 Euro area 1.5 1.5 1.4 1.6 US 2.0 1.9 2.3 2.3 Trends and Constraints in Urban Labor Markets in Ethiopia OECD-Total     2.2     2.4     1.9     2.2   Source: CSA, NBE, Customs, World Bank, OECD-National Accounts. 45 46 Annex 2: Ethiopia: Selected Economic and Social Indicators (Annual Frequency) Average Fiscal year ending July 7 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2006–2016 Income and Economic Growth                         GDP growth at factor cost (annual %) 11.5 11.8 11.2 10.0 10.6 11.4 8.7 9.8 10.3 10.4 8.0 10.3 GDP growth at market price (annual %) 10.8 11.5 10.8 8.8 12.6 11.2 8.6 10.5 10.3 10.4 7.6 10.3 GDP per capita growth (annual %) 7.8 8.5 7.9 6.0 9.6 8.3 5.9 7.8 7.5 7.5 5.1 7.4 GDP per capita (US$) 195 245 328 382 344 357 472 505 573 616   401.7 GDP per capita, PPP (current international $) 726 809 889 949 1054 1165 1257 1377 1505 1626   1135.7 Atlas GNI per capita, US$ 180 220 280 340 380 390 420 470 550 590   382.0 Private Consumption, nominal (annual %) 25.9 26.9 51.8 35.7 15.3 28.6 45.1 15.7 16.2 14.9 14.9 26.5 Gross Fixed Investment ( % of GDP) 32.2 28.2 28.5 29.5 31.6 32.1 37.1 35.8 40.3 39.4 38.5 33.9 Gross Fixed Investment – Public ( % of GDP) 16.0 14.1 14.7 12.3 15.5 18.7 14.3 12.2 16.6 22.5 24.6 16.5 Gross Fixed Investment – Private ( % of GDP) 11.9 10.4 10.0 13.3 11.9 9.2 18.7 20.8 23.7 16.9 14.0 14.6 Money and Prices                         Inflation, consumer prices (annual %, end of year) 11.6 15.1 55.3 2.7 7.3 38.1 20.8 7.4 8.5 10.4 7.5 16.8 Inflation, consumer prices (annual %, period average) 12.3 15.8 25.3 38.7 3.0 17.9 34.7 13.9 8.1 7.7 9.7 17.0 Treasury bill rate (91-days maturity, annual average) 0.0 0.8 0.6 0.9 0.9 1.3 1.9 1.4 1.2 1.2   1.0 Nominal Exchange Rate (End of period) 8.7 9.0 9.6 11.3 13.5 16.9 17.8 18.6 19.6 20.6 21.8 15.2 Real Exchange Rate Index (1990 = 100) 96.2 102.5 107.5 133.9 106.9 96.4 117.1 125.4 128.1 142.8 154.0 119.2 Fiscal                         Revenue (% of GDP) 14.7 12.6 11.9 11.9 14.0 13.4 13.8 14.3 13.8 14.4 15.2 13.6 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Expenditure (% of GDP) 22.0 20.5 18.8 17.1 18.8 18.2 16.6 17.8 17.5 17.8 18.4 18.5 Current (% of GDP) 11.5 9.9 9.1 8.0 8.4 7.9 6.9 7.2 7.4 8.7 8.9 8.5 Capital (% of GDP) 10.6 10.6 9.6 9.1 10.4 10.3 9.8 10.5 10.1 9.0 9.4 10.0 Fiscal balance including grant (% of GDP) –4.5 –3.6 –2.9 –0.9 –1.6 –1.6 –1.2 –1.9 –2.6 –2.4 –2.4 –2.3 Fiscal balance excluding grant (% of GDP) –7.3 –7.9 –6.8 –5.2 –4.9 –4.8 –2.9 –3.5 –3.7 –3.4 –3.2 –4.9 Primary fiscal balance including grants (% of GDP)* –3.7 –2.9 –2.4 –0.5 –1.2 –1.2 –0.9 –1.6 –2.2 –2.0 –1.9 –1.9 Total public debt (% of GDP) 66.8 43.9 38.5 35.5 39.4 37.8 32.7 37.4 44.7 50.2 54.0 43.7 External public debt (% of GDP) 37.3 11.8 10.4 14.8 18.3 22.2 17.9 20.5 22.6 26.2 30.0 21.1 (continued on next page) Annex 2: Ethiopia: Selected Economic and Social Indicators (Annual Frequency) (continued) Average Fiscal year ending July 7 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2006–2016 External Accounts                         Goods Export growth (%, yoy) 18.1 18.7 23.1 –1.0 38.3 37.1 14.8 –2.3 5.7 –8.5 –3.7 12.8 Goods Import growth (%, yoy) 26.4 11.6 32.8 13.4 7.7 –0.2 34.0 3.7 19.7 19.9 1.6 15.5 Merchandise exports (% of GDP) 6.7 6.1 5.5 4.5 6.7 8.6 7.3 6.5 5.9 4.6 4.0 6.0 of which coffee exports (% of GDP) 2.4 2.2 2.0 1.2 1.8 2.7 1.9 1.6 1.3 1.3 1.0 1.8 Merchandise imports (% of GDP) 28.7 26.0 27.0 23.8 27.6 25.8 25.5 24.1 24.7 25.5 23.1 25.6 Services, net (% of GDP) 1.0 0.8 0.5 1.3 1.7 2.4 0.4 1.2 1.3 –0.1 –0.6 0.9 Service exports (% of GDP) 7.2 6.6 5.9 6.0 6.8 8.1 6.5 6.0 5.7 4.8 4.1 6.1 Service imports (% of GDP) –6.3 –5.8 –5.4 –4.7 –5.1 –5.7 –6.1 –4.8 –4.4 –4.8 –4.7 –5.3 Private transfers, net (BoP,% of GDP) 8.0 8.6 8.8 8.3 9.0 8.6 7.5 7.5 7.3 7.6 8.3 8.1 Current account balance before grant (BoP, % of GDP) –13.4 –10.5 –12.2 –9.8 –10.4 –6.5 –10.6 –8.5 –10.6 –13.8 –11.9 –10.7 Current account balance after grant (BoP, % of GDP) –9.1 –4.4 –5.6 –5.0 –4.0 –0.7 –6.5 –5.3 –7.9 –11.5 –10.4 –6.4 Foreign Direct Investment (% of GDP) 2.4 2.4 3.0 2.8 3.2 3.9 2.5 2.6 2.6 3.4 4.2 3.0 External debt, total (% of GDP) 37.3 11.8 10.4 14.8 18.3 22.2 17.9 20.5 22.6 26.2 30.0 21.1 Multilateral debt (% of total external debt) 81.1 51.6 55.7 46.7 48.6 46.0 45.4 45.0 42.1 34.2 36.5 48.4 Debt service ratio (% of goods and NFS) 8.0 7.3 2.9 2.3 2.7 4.5 6.9 9.6 10.3 16.1 16.7 7.9 Population, Employment and Poverty                         Population, total (millions), UN 78.7 80.9 83.1 85.3 87.6 89.9 92.2 94.6 97.0 99.4   88.9 Unemployment Rate (urban) 17.0     20.4 18.9 18.0 17.5 16.5 17.4 16.8   17.8 Poverty headcount ratio at national poverty line           29.6           29.6 (% of population) Poverty headcount ratio at $1.25 a day (PPP)           30.7           30.7 (% of population) Poverty headcount ratio at $2 a day (PPP)           66.0           66.0 (% of population) Inequality – Income Gini           29.8           29.8 Population Growh (annual %) 2.7 2.7 2.7 2.6 2.6 2.6 2.6 2.5 2.5 2.5   2.6 Trends and Constraints in Urban Labor Markets in Ethiopia Life Expectancy 57.3 58.4 59.4 60.4 61.3 62.1 62.8 63.4 64.0     61.0 (continued on next page) 47 48 Annex 2: Ethiopia: Selected Economic and Social Indicators (Annual Frequency) (continued) Average Fiscal year ending July 7 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2006–2016 Others:                         GDP (current LCU, billions) 133 173 250 338 386 515 747 867 1,061 1,298 1,528 663.3 Nominal GDP (current US$, billions) 15 20 27 32 30 32 43 48 55 64 72 40.0 a Doing Business (rank) 101 97 102 116 107 104 111 124 148 146   115.6 Logistics performance Index   2.3     2.4   2.2   2.6     2.4 b Human Development index ranking 170 169 169 171 157 174 172 173 173 175   170.3 CPIA (overall rating) 3.4 3.4 3.4 3.4 3.4 3.5 3.4 3.4 3.5     3.4 Economic management 3.5 3.5 3.3 3.7 3.7 3.7 3.5 3.5 3.5     3.5 Structural policies 3.2 3.2 3.2 3.2 3.2 3.2 3.2 3.2 3.2     3.2 Policies for Social Inclusion and Equity 3.6 3.7 3.6 3.6 3.6 3.7 3.7 3.7 3.7     3.7 Public Sector Management and Institutions 3.3 3.3 3.3 3.2 3.2 3.3 3.4 3.4 3.5     3.3 Source: CSA, NBE, Customs, World Bank, OECD-National Accounts. Note: a This indicator is ranked out of 175 countries in 2007, 178 in 2008, 181 in 2009 and 183 in 2010 and 2011, 185 in 2012, and 189 in 2013 and 2014. b The HDI ranking in 2001 is in relation to 175 countries; from 2005 to 2008, to 177; in 2009, to 181; in 2010, to 169 countries; and, from 2011–2014 to 187 countries. *Consolidated public sector primary balance includes SOE which is derived from the financing side while the primary fiscal balance includes federal and regional government only. 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Trends and Constraints in Urban Labor Markets in Ethiopia 49 Annex 3: Self-employment in Urban The Trade-Off between Wage FIGURE A1.1:  Labor Markets in Ethiopia and Self-Employment, 2014 25 70% Real Hourly wage (2003 Birr) Workers choose between self-employment and 60% 20 Self-employment entering the wage labor market, which may 50% 15 involve periods of unemployment and wage- 40% 30% employment. Those with little education choose 10 20% necessity self-employment rather than searching for 5 10% wage employment because the high probability of 0 0% No education Primary Secondary Post- secondary being unemployed makes job search costly and the wages earned do not compensate them for the cost of looking. The productivity and income of these Wage Expected Self-employment individuals is likely to be lower than it would be if wage (right axis) they were employed, but the cost of being unem- Source: World Bank staff calculations using UEUS 2003–2014. ployed and searching is not worth the gain. Those with moderate levels of education look for and attain employment. They may be in unemployment for education, and correspondingly, the share of those in some time before they secure a job, but the returns to self-employment is higher at lower levels of education. being employed are worth the search. Those with the There are also other factors that may explain the lower highest ability are successful entrepreneurs, operating rates of self-employment among educated workers. businesses that are of a scale such that they employ Those with education that aspire to wage employ- others. This model is summarized in Figure A1.1 ment are often from wealthier families and better (Poschke 2014). able to finance job search (Franklin 2014). There is Thus, self-employment becomes less attractive some evidence (from Tanzania and Ghana) that once as wage rates increase and unemployment falls. The someone is self-employed they are less likely to secure trade-off for individuals of different education level wage employment in the future (“scarring”) and thus in urban Ethiopia in 2014 is presented in Figure 2.4. those with higher expected wages choose not to be Expected wages are much lower for those with less “scarred” (Falco et al. 2014). 50 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Annex 4: Changes in the Composition halved to 21 percent of those with secondary school- of the Labor Force ing securing public sector employment. Addis Ababa has seen some change in the struc- In particular, the educational composition of pub- ture of employment, particularly in the growth of lic sector employees has improved significantly wage employment with respect to self-employment. with a dramatic increase in the share of public The share of wage-employees increased from 62 per- employees who are graduates. Perhaps one of the cent in 2003 to 72 percent in 2014 (Figure A2.3), largest changes that has taken place over time is the representing an almost equal increase in private educational composition of public sector employ- and public sector wage employment—4 percentage ees (Figures A2.1 and A2.2). The public sector has points and 3 percentage points respectfully. Although consistently been the largest employer of those with the sectoral composition of the labor force in Addis post-secondary qualification in Ethiopia. In 2003, Ababa did not change much, there was evidence of 24 percent of public sector employees had a graduate a rapid expansion in the construction sector in the degree and by 2014 this had increased to 62 percent. employment data (Figure A2.4). The construction These graduates displaced individuals with secondary sector increased from employing 9 percent of the schooling (Figure A2.1). In 2003, those with second- workforce, to employing 12 percent of the workforce ary education were quite likely to be employed by the from 2011, a relative increase in its share of employ- state: 41 percent of those with secondary education ment of 33 percent. The public sector increased its were thus employed. By 2014, this had more than share in employment from 20 percent to 23 percent. Trends and Constraints in Urban Labor Markets in Ethiopia 51 FIGURE A2: An Increasingly Educated Public Sector 1) Share of public sector employees, by education level 100% 80% 60% 40% 20% 0% No Education Primary Secondary Post-Secondary Public Private 2) Share of educated workers, by public/private employment 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 2003 2004 2006 2009 2010 2011 2012 2014 2003 2004 2006 2009 2010 2011 2012 2014 Private Public No education Primary Secondary Post-secondary/Other 3) Wage employment increasingly important 4) Little change in the sectoral composition of labor 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% 2003 2004 2006 2009 2010 2011 2012 2014 2003 2004 2006 2009 2010 2011 2012 2014 Wage emp. Self emp. Other Agriculture Mining Manufacture Construction Services Source: World Bank staff calculations using UEUS 2003–2014. 52 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET Annex 5: What Explains Real Wage Trends changing endowments and changing returns. Box 2.1 summarizes the method. There are four possible drivers of real wage trends. Decomposition analysis shows that changes The drivers are: (i) changing characteristics of the work- in the gender, age and educational composition of force, (ii) changes in public sector wages, (iii) sticky workers do not explain the change in wages over wages in the presence of high inflation, and (iv) changes time; neither the decrease in public sector wages in the marginal productivity of labor. Changing aver- prior to 2012 nor the increase in public and private age wages could reflect the changing characteristics of sector wages since then. The characteristics of workers workers in the labor force. As the previous sections have in both the public and private sector have been chang- highlighted, the sectoral composition of work has not ing in a way that would have suggested continual wage changed over the last decade, but the characteristics of increases from 2003 to 2014. However, the returns to workers have, with many more educated workers in these endowments fell prior to 2012, particularly in the workforce in 2014 than in 2003. More educated the public sector, and increased post-2012. As a result workers are paid more and their increasing share in nearly all of the fall and the increase in the public sec- the labor force could thus have resulted in average tor is explained by changes in returns to the workers’ wages rising. Oaxaca-blinder decomposition analysis characteristics. In the private sector, 92 percent of the decomposes differences in average wages into the part increase since 2012 is explained by changes in returns. that can be explained by changes in worker character- What then explains changes in returns on char- istics, called the “endowment” effect, and the part that acteristics? Fixed nominal wages in the public sector can be explained by a changing relationship between in the presence of high inflation could explain the worker characteristics (such as education) and wages, patterns observed for public employees. The falling called “returns.” In decomposition analysis there is also and then rising returns on worker characteristics in always a third effect, which is the “interaction” between the public sector can be explained relatively easily by BOX 2.1: What Does Decomposing Wage Trends Entail? Decomposing wage trends relies on defining a counterfactual scenario and estimating what would have happened to wages had the counterfactual scenario occurred. By defining a counterfactual scenario the changes that have been important to overall wage trends can be quantified. The figure below depicts how this can work for two different counterfactual scenarios. In the Oaxaca-Blinder decomposition, the focus is on a counterfactual of a constant relationship between worker characteristics (called “endowments”) and wages in urban Ethiopia. This counterfactual is used to determine which changes in worker characteristics could have contributed to wage changes, and how much could have changed as a result of a changing relationship between wages and worker characteristics. The latter is sometimes referred to as changes in returns, but really it represents how the conditional correlation between a given characteristic and the wage rate has changed. In all decomposition approaches there is an interaction effect, which can be interpreted as a measure of the correlation changes in endowments and returns. In the decompositions shown here it is quite small. Using counterfactuals to decompose wage trends Change in wages for Counterfactual: Wages Change in endowments Wages people with a wage if no change in 2003 and “interaction effect” in 2012 given endowment in endowments Counterfactual: Change in wages for people Wages Wages Change in endowments wage if only change with a given endowment and in 2003 in 2012 in endowments “interaction effect” Trends and Constraints in Urban Labor Markets in Ethiopia 53 FIGURE A5.1: Oaxaca Blinder Decomposition of the Real Hourly Wage 1.00 0.84 0.71 0.59 0.48 0.31 0.44 0.50 0.08 0.04 0.04 0.01 0.00 –0.02 –0.07 –0.50 –0.29 –0.26 –1.00 –1.50 –2.00 –1.68 –1.98 –2.50 Difference Endowments Returns Interaction Difference Endowments Returns Interaction Public Sector Private Sector 2006–2012 2012–2014 Source: World Bank staff calculations using UEUS 2003–2014. the fact that public sector wages were held constant in is some stickiness in wages it cannot fully explain nominal terms until a whole-scale revision of the public the declining returns observed until 2012. Headey sector salary structure in 2014. Some public agencies et al. (2012) examines the degree to which unskilled had increased salaries prior to then, indicating why an wages (maids, guards, and casual labor) in 120 urban increase would have been identified prior to when this centers and rural towns in Ethiopia adjusted to the revision came into full effect. Public sector wages can price increases observed in 2008 and 2010/11. They have a large influence on private sector wages given the show that in the short run, wages do not adjust, but importance of the public sector as a wage employer. In that in the longer run they do. However they find many cities, the public sector comprises half of wage that it takes 4–5 months for wages to adjust which employment, and public sector wage trends can be suggests considerable hardship for individuals in the expected to also impact wages in the private sector. first months of high food price inflation, but does not Even if public sector wages were not directly explain the decline in returns into 2012. influencing private sector wages, they could have A final possible explanation for the wage been similarly affected by high rates of inflation in trends observed is that it reflects real changes in urban areas in 2008 and also 2010/11. Private sector labor productivity. In a perfectly competitive, non- wages may not be fixed, but if they are slow to adapt segmented labor market, the market-clearing wage then it is quite possible that a fall in wages would have should reflect the marginal productivity of labor. In been observed for a few years until wages adjusted and reality it is difficult to estimate the marginal produc- increased in the long run to compensate for the loss. tivity of labor, so the average productivity of labor, The UEUS does not provide data on wages frequently measured as the valued added per worker, is used as enough to test this, but this question was tested by a proxy. Headey et al, (2012) using monthly data on unskilled Changes in wages in the manufacturing sector wages collected in the monthly retail price survey. are explored, given this is a sector of considerable It takes 4–5 months for wages to adapt to interest for the government of Ethiopia and also higher food prices, suggesting that although there the only sector for which data on both firms and 54 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET BOX 2.2: Measuring Labor Productivity Value added per workers is defined as total output and income from services provided, minus total costs (raw materials, other industrial costs, and non-industrial costs) divided by the number of workers. When cost data was missing, this was treated as 0 costs. However a missing value was kept when the total costs of raw materials and parts were missing. Labor costs and productivity according to the manufacturing census are based on permanent and contract workers only. That is, wage costs and employment of seasonal and temporary workers is excluded. Temporary workers make up about 20 percent of workers, so the true level of labor productivity would be lower than the levels estimated here. However, there is considerable measurement error in estimating the person-months engaged and the wages paid and this makes it justifiable to exclude them. All averages reported are weighted by firms’ number of permanent and contract workers. Labor costs are quite sensitive to outliers, particularly in some years such as 2009/10, and removing the top and bottom percentile by year or in the pooled data produces rather different numbers. Unless stated otherwise, averages presented are calculated after truncating the pooled data (i.e. removing the top and bottom percentiles in the pooled firm-level data). employees are available. Data on firms in Ethiopia However, a deeper examination of both employ- is limited. Data is collected annually on medium- ment and firm data suggests wage trends may not and large-scale manufacturing firms, but these firms be so easily explained. Wages do not follow trends in employ a relatively small proportion of the labor average productivity within sector or across cities. Per force. Nevertheless this data is used to examine wage worker labor cost trends are similar across industries trends and labor productivity within manufacturing reflecting the findings above that the wage changes and to explore whether wages follow average labor were reflected in all sectors (Figure A5.2.2). Labor pro- productivity. Details on how value added per worker ductivity gains have been present in most industries, was calculated are provided in Box 2.2. but have declined since 2010 in textiles, apparel, and Real labor costs reported by manufacturing leather (Figure A5.2.3). This is not reflected in a differ- firms broadly reflect trends in average labor pro- ent wage trend in this sector (Figure A5.2.2). Increases ductivity. The manufacturing firm data exhibits con- in wages have been present in all cities, but this is not siderable noisiness, even after trimming the data for the case for increases in labor productivity, with as many outliers. Taking this into account, Figure A5.2.1 shows of the large cities experiencing falls in productivity as labor productivity falling until 2010 and increased increases since 2012 (Figures A5.2.4 and A5.2.5). The thereafter. Although labor productivity increased scatter plot in A5.2.6, shows that (broadly) productiv- after 2010, gains in recent years have been weaker. ity and wages are positiviely correlated (confirmed in Figure A5.2.2 showed real labor costs falling until the regression results of Table A5.1), but this positive 2011 and then increasing. These figures suggest that trend is not significant, and explains very little of the trends in real wages paid by manufacturing firms do, variation in wage levels (2 percent). This could reflect broadly, reflect changes in worker productivity, but by the quality of the manufacturing census data, but it no means perfectly, with large differences in the ratio does question whether it is real productivity changes between the two values. that underly the observed wage trends. Trends and Constraints in Urban Labor Markets in Ethiopia 55 FIGURE A5.2: Labor Costs and Labor Productivity in the Manufacturing Sector 1) Trends in labor productivity in manufacturing 2) Trends in labor costs in manufacturing Labor productivity in '000 2003 Birr Per worker labor costs in '000 2003 Birr 150 12 10 100 8 50 6 4 0 2006 2008 2010 2012 2014 2006 2008 2010 2012 2014 Value added per worker Output per worker All manufacturing Food products and beverages Firm values weighted by number of workers. Textiles, apparel, leather Other Manufacturing Top and bottom percentiles dropped. Source: Census Firm values weighted by number of workers. Top and bottom percentiles dropped. Source: Census 3) Trends in labor productivity by subsector 4) Real monthly wage trends by city Value added per worker in '000 2003 Birr Monthly earnings by city – all manufacturing workers in 2003 Birr 80 500 60 400 40 300 20 200 0 2006 2008 2010 2012 2014 2006 2008 2010 2012 2014 All manufacturing Food products and beverages Mekele Addis Ababa Awasa Textiles, apparel, leather Other Manufacturing Bahir Dar Debre Zeyit Dire Dawa Nazaret Firm values weighted by number of workers. Top and bottom percentiles dropped. Source: Census Source: UEUS 5) Trends in labor productivity by city 6) Growth in real wages and labor productivity by city Value added per worker by city in '000 2003 Birr Earnings and productivity growth across cities correlation 0.099 annual percentage growth Real monthly earnings – 200 0.2 150 0.1 0.0 100 –0.1 50 –0.2 0 –0.3 2006 2008 2010 2012 2014 –1 0 1 2 3 4 Real value added per worker –  Mekele Addis Ababa Awasa annual percentage growth Bahir Dar Debre Zeyit Dire Dawa Nazaret Source: UEUS and Census Firm values weighted by number of workers. Top and bottom percentiles dropped. Source: Census Source: World Bank staff calculations using UEUS 2003–2014. 56 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET TABLE A5.2: Earnings and Productivity, Regression Results All Cities log real hourly wages log real monthly earnings All paid workers Permanent/contract workers Permanent/contract workers log real per worker 0.013 0.029 0.013 0.024 –0.014 0.004 value added (0.039) (0.038) (0.051) (0.043) (0.043) (0.038) 2009 –0.067 –0.273 –0.236 (0.086) (0.097) (0.087) 2010 –0.043 –0.286 –0.196 (0.089) (0.1) (0.089) 2011 –0.152 –0.413 –0.281 (0.087) (0.097) (0.087) 2012 –0.259 –0.515 –0.417 (0.085) (0.097) (0.086) Constant 0.304 0.37 0.491 0.777 5.874 6.06 (0.125) (0.141) (0.162) (0.155) (0.137) (0.138) N 63 63 62 62 62 62 R-squared 0.002 0.221 0.001 0.432 0.002 0.372 Note: Standard errors in brackets. All regressions control for city fixed effects. Standard errors are reported below the coefficient estimate. 57 REFERENCES Chapter 1: Recent Economic Laurens, B. 2005. “Monetary Policy Implementation Developments and Outlook at Different Stages of Development.” IMF Occasional paper 244. Section B: The Long View Laurens, B. and R Maino. 2007. “China: Strengthening Monetary Policy implementation.” IMF WP/07/14. Aklilu, Y. 2009. “The money multiplier: proximate Li, K-W. 1996. “Some thoughts on China’s 1995 determinants and stability (the case of Ethiopia).” Bank Reform.” Conference Paper presented National Bank of Ethiopia Birritu Magazine 103, at conference on “China’s Economic Reform pp 22–35. and Opening” (Oct 1995 Hong Kong) and the Buzeneca, I and R. Maino. 2007. “Monetary Policy Canadian Economic Association Conference Implementation: Results from a Survey.” IMF (May/June 1996). WP/07/7. Mehran, H., M. Quintyn, T. Nordman, and B. Cheng, E. and Y. Cheng. 1998. “Banking Reform and Laurens. 1996. “Monetary and Exchange the Separation of Policy and Commercial Loans System Reforms in China: An Experiment in in China.” MOCT-MOST: Economic Policy of Gradualism.” IMF Occasional Paper 141 Transitional Economies, 8, 5–21. Mishra, P., P. Montiel, and A. Spilimbergo. 2012. Geiger, M. 2008. “Instruments of Monetary Policy “Monetary Transmission in Low-Income Countries: in China and their effectiveness: 1994–2006” Effectiveness and Policy Implications.”,  IMF UN Conference on Trade and Development Economic Review 60, 270–302. Discussion Papers 187. Montes-Negret, F. 1995. “China’s Credit Plan: an Geiger, M. 2010. “Monetary Policy in China: Overview.” Oxford Review of Economic Policy, 11 (4) Institutions, Targets, Instruments and Strategies.” Xu, Y. 2000. “China’s exchange rate policy.” China PhD Dissertation, Julius Maximilians University Economic Review 11: 262–277. of Würzburg; Available at: http://d-nb. info/1021078328/34 Section C: The Future View Green, S. 2005. “Making Monetary Policy Work in China: A Report from the Money Market Bellmon. 2015. Bellmon Analysis 2015/16: Front Line.” Stanford Center for International Reassessment of crop Production and Marketing Development, Working Paper No. 245. for the year 2014/15. IMF. 2013a. “The Gambia Staff Report for the Bloomberg. 2016. Ethiopia to Miss 11 Percent 2013 Article IV Consultation.” Country Report Growth Target As Drought Hurts Agriculture. 13/289. Addis Ababa and New York. IMF. 2013b. “Sub-Saharan Africa Regional Economic CIMMYT. 2015. Wheat Atlas (http://wheatatlas. Outlook.” May 2013. org) Accessed December 15, 2015. IMF. 2014. “Zambia Staff Report for the 2013 Article Ethiopian Reporter Amharic Edition. December 26, IV Consultation.” Country Report 14/5. 2015. 58 5TH ETHIOPIA ECONOMIC UPDATE – WHY SO IDLE? WAGES AND EMPLOYMENT IN A CROWDED LABOR MARKET FAO. 1997. FAO/WFP Crop and Food Supply Banerjee, Abhijit V., and Esther Duflo. 2008. “What Assessment Mission to Ethiopia. http://www. Is Middle Class about the Middle Classes around fao.org/giews. the World?” Journal of Economic Perspectives 22 FAO. 2010. FAO/WFP Crop and Food Security no. 2: 3–28. Assessment Mission to Ethiopia. http://www. Bigsten, Arne, Taye Mengistae, and Abebe Shimeles. fao.org/giews. 2013. Labor Market Integration in Urban FAO. 2015. Crop prospects and Food Situation. No. Ethiopia, 1994–2004. Economic Development and 4. http://www.fao.org/giews. Cultural Change 61(4): 889–931. FEWS NET. 2015. Ethiopia Food Security Outlook Bimerew, Habtamu. 2015. “Rural-Urban Migration October 2015 to March 2016. www.fes.net/ and its Consequence on Urban living: the case in Ethiopia. Hawassa City Southern Ethiopia.” Global Journal Humanitarian Requirements Document (HRD). 2016. of Human-Social Science: Economics. Volume 15 Ethiopia Joint Government and Humanitarian Issue 4 Version 1.0. Partners` Document, Addis Ababa, Ethiopia. de Brauw, Alan, Valerie Mueller, and Tassew Minot, N., W. James, L. Solomon, K. Leulseged, Woldehanna. 2013. “Does Internal Migration G. Abate G, and R. Shaidur. 2015., The Wheat Improve Overall Well-Being in Ethiopia?” ESSP Supply Chain in Ethiopia: Patterns, Trends, working paper 55. International Food Policy and Policy Options. International Food Policy Research Institute (IFPRI). Research Institute (IFPRI), Washington DC. de Brauw, Alan, 2014. “Migration, Youth and Reuters. 2015., http://www.voanews.com/content/ethi- Agricultural Productivity in Ethiopia.” opia-drought-gdp-finance-minister/3055402.html Background Paper prepared for the Ethiopia UNDP. 2016. Assessing the cost drought in Ethiopia Poverty Assessment. and the case for long-term development approach. Falco, Paolo, Andrew Kerr, Pierella Paci, and Bob Policy Note. Rijkers, 2014. “Working toward Better Pay: UNOCHA. 2016. http://www.unocha.org/top- Earning Dynamics in Ghana and Tanzania.” stories/all-stories/world-must-support-ethiopi- World Bank Publications, The World Bank, ans-frightening-drought-looms; http://www. number 18553. un.org/apps/news/story.asp?NewsID=52382#. Franklin, Simon. 2014. “Work, Unemployment and Vsc5rk8WMrp. Job Search among the Youth in Urban Ethiopia.” USDA and USAID. 2016. GAIN Report Number Mimeo, Oxford University. ET1603. http://gain.fas.usda.gov/. Headey, Derek, Fantu Bachewe Nisrane, Ibrahim Xinshen, D., F. Belay, H. Steven, S. Alemayehu, W. Worku, Mekdim Dereje, and Alemayehu Seyoum Kassu, and Y. Bingxin. 2007. Agricultural growth Taffesse. 2012. “Urban Wage Behavior and Food linkages in Ethiopia: Estimates Using Fixed and Price Inflation: The Case of Ethiopia.” ESSP Flexible Price Models. International Food Policy Working Paper 41 Research Institute Discussion Paper No. 00695 Imbert, Clement. 2012. “Decomposing the Labor Market Earnings Inequality: The Public and Chapter 2: Trends and Constraints in Private Sectors in Vietnam, 1993–2006.” The Urban Labor Markets in Ethiopia World Bank Economic Review 27(1): 55–79. Kibru, Martha. 2012. Employment Challenges in Asmame, Birhan. 2011. “Causes and Consequences Ethiopia. Friedrich Ebert Stiftung, Nairobi. of Rural-Urban Migration”. Mimeo. University Krishnan, P. 2000. “Public Sector Pay and Private of South Africa. Sector Wage Premiums: Testing Alternative References 59 Models of Wage Determination.” Center for the World Bank. 2013. Ethiopia Economic Update II: Laying Study of African Economies, Oxford, Working the Foundation for Achieving Middle Income Status. Paper Series, WPS/2000-7. Africa Region, World Bank, Washington DC. Martins, P. 2015. “Structural change in Ethiopia – an World Bank. 2014. “Skills for Competitiveness and update using 2013 LFS data.” Background Paper Growth in the Manufacturing Sector.” World prepared for World Bank’s Ethiopia Great Run Bank Report: Human Development, Education Report published in 2015. and Private Sector Development, Africa Region, McMillan, M. and D. Rodrik, 2011. Globalization, Washington DC. Structural Change, and Productivity Growth, World Bank. 2015a. Ethiopia Poverty Assessment. Africa NBER Working Paper 17143, National Bureau Region, World Bank, Washington DC. of Economic Research. World Bank 2015b. Ethiopia Urbanization Review: Mengistae, Taye. 1999. The Relative Effects of Skill Urban Institutions for a Middle-Income Ethiopia. Formation and Job Matching on Wage Growth in Africa Region, World Bank, Washington DC. Ethiopia. World Bank Policy Research Working World Bank. 2015c. Ethiopia Economic Update IV: Paper 2104. Overcoming Constraints in the Manufacturing Sector. Mengistae, Taye. 1999. Wage rates and job queues: Africa Region, World Bank, Washington DC. does the Public Sector Overpay in Ethiopia? World Bank 2015d. Rwanda Labor Market Study. World Bank Policy Research Working Paper 2104 Africa Region, World Bank, Washington DC. Poschke, M., 2014. “A Model of Entrepreneurship and World Bank 2015e. Ethiopia’s Great Run: The Growth Employment in Ethiopia: Simulating the Impact Acceleration and How to Pace It. Africa Region, of an Urban Safety Net.” Background Paper pre- World Bank, Washington DC. pared for the Ethiopia Poverty Assessment. World Bank 2016. Systematic Country Diagnostics Serneels, Peter. 2007. “The nature of unemployment (SCD): Priorities for Ending Extreme Poverty and among young men in urban Ethiopia.” Review Promoting Shared Prosperity. Africa Region, World of Development Economics. Vol. 11, No. 1, pp. Bank, Washington DC. 170–186, February. The World Bank 1818 H Street, NW Washington, DC 20433 Phone: (202) 473-1000 Fax: (202) 477-6391