Report No. 38665-ET Ethiopia Urban Labour Markets in Ethiopia Challenges and Prospects (In Two Volumes) Volume I: Synthesis Report March 2007 Poverty Reduction and Economic Management Unit Africa Region) Document of the World Bank CURRENCYAND EQUIVALENTUNITS (as ofNovember 2006) Currency Unit = EthiopianBirr ETB) US$1 = Br8.7 FISCAL YEAR July 8 - July 7 WEIGHTS AND MEASURES Metric System ABBREVIATIONS AND ACRONYMS ADLI Agricultural DevelopmentLedIndustrialization AEMFI Association of EthiopianMicro Finance Institutions ALMPs Active LabourMarket Policies ALMS Addis Labour Market Survey CEM Country Economic Memorandum CGE Computable General Equilibrium CLFS Child Labour Force Survey CSA Central Statistical Agency DG Director General EBDSN EthiopianBusiness Development ServicesNetwork EEF EthiopianEmployers Federation FCSC Federal Civil Service Commission FeMSEDA Federal MSEDevelopmentAgency GDP Gross Domestic Product GOE Government of Ethiopia ICA Investment Climate Assessment IFPRI InternationalFoodPolicy ResearchInstitute ILO InternationalLabour Organization IMF InternationalMonetary Fund IVCAGs Integrated Value Chain Analysis Groups KILM Key Indicatorsof the Labour Market LFS Labour Force Survey LMTC Labour Market Technical Committee MAMS Marquette for MDG Simulations MDAs Ministries, Departments and Agencies MDGs MillenniumDevelopment Goals MFI Micro finance Institution - - 11 .. MOE MinistryofEducation MOFED MinistryofFinanceandEconomic Development MOLSA MinistryofLabourand SocialAffairs MTEF Medium-Term Expenditure Framework MTPRS Medium-TermPayReform Strategy MSEs Micro and Small Enterprises NGO Non-Governmental Organization OECD Organizationfor Economic Cooperation and Development PASDEP Plan for Accelerated SustainedDevelopment to EndPoverty PND Private SectoriNGOsAIonors PPA Participatory Poverty Assessment PRSP Poverty Reduction Strategy Paper ReMSEDA Regional MSEDevelopment Agency SASE Selective Accelerated Salary Enhancement SOEs Statement of Expenses SMEs Small and MediumEnterprise SSA Sub-SaharanAfrica TVET Technical and Vocational EducationTraining UBEUS Urban Employment and Unemployment Survey UN UnitedNations UNECA UnitedNations Economic Commissionfor Africa VAT Value-added Tax WED Wellbeing inDevelopingCountries Vice President : Hartwig Schafer Country Director : Ishac Diwan Sector Director : Sudhir Shetty Sector Manager : Kathie Krumm Task Leaders : Jeni Klugman and Caterina Ruggeri Laderchi -... 111- ACKNOWLEDGEMENTS We would like to thank our primary counterparts, the Ministry of Labour and Social Affairs (MOLSA), and inparticular H.E. Woz. Zenebu Tadesse, State Minister, and At0 Tamiru Jeba, Head, Department of Planning and Programming. The Central Statistical Agency (CSA) provided essential support in accessing and understanding key data sources, and especially Woz. Samia Zekaria, Acting Director General, At0 Yasin Mossa, Acting Deputy D.G, Economic Statistics, At0 Yakob Mudesier, Acting Deputy Director General o f OperationMethodology and Data Processing, At0 Mekonnen Tesfaye, Head, Social Statistics Department, and At0 Kassu Gebeyehu, Manpower, Wages and Salaries Statistics Analysis and Research Expert Team Leader, Social Statistics Department, CSA. The Ethiopian Labour Market Technical Committee (LMTC), chaired by MOLSA, served as focal point for collaboration. Through a series o f consultations they contributed to the agenda and reviewedbackgroundpapers. The LMTC includesmembers from MOLSA, Ministry of Finance and Economic Development (MOFED), Ministry o f Education (MOE), Ministry of Trade and Industry, Central Statistical Authority (CSA), Ethiopian Federal Civil Service Commission (FCSC), Ethiopian Employers Federation (EEF) and Confederation o f Ethiopian Trade Union (CETU) . This study also benefited from the close coordination with the LO, both in Geneva and inAddis. Ajoint two-day workshop was heldinDecember 2005 to discuss the emergingresults o f the E O "Flexicurity Study" and this study. The team would like to thank inparticular Sabina de Gobbi, George Okutho, Larry Kohler, Zafar Shaheed, Muhammed Muqtada, Marva Corley, and Ato Bedada Urgessa. Generous finding from the German Trust Fundprogram on "Job Creation, Core Labour Standards, and Poverty Reduction" i s gratefully acknowledged, as i s the support o f Achim Johansen, Deputy Head of Division, German Federal Ministry for Economic Cooperation and Development (now at BDI) and HeinWinnubst, First Secretary, Development Cooperation at the Embassy inEthiopia. Support i s also acknowledged from the Swedish and the Danish Consultant Trust Funds. The team was composed of Jeni Klugman and Caterina Ruggeri Laderchi (task team leaders), Emily Kallaur, Eyerusalem Fasika, Taye Mengistae, and Simon Commander. Background papers were prepared in collaboration with the team by Abebe Shimeles, Alula Pankhurst, Arne Bigsten, Bo Rosendhal, Feleke Tadele, Hans Lofgren, Martha Getachew, Mohammed Mussa, Niels-Hugo Blunch, Peter Madsen, Philippa Bevan, and Rahimaisa Abdula. Angela Cipollone and Martin Cumpa provided excellent research assistance. Thomas Pave Sohnesenand Jacob Ladenburgalso provided research support. Administrative support was ably provided by Dora Harris, Lucy Kang'arua, Marjorie Kingston, Monica Wachera Ndungu, and Senait Kassa Y i h . The team benefited from the overall guidance o f Ishac Diwan and Kathie Krumm. Useful suggestions and substantive inputs were received from a range of colleagues, and we would especially thank Louise Fox and the Afiica Labour Market Analysis Group, Arvil Van - iv - Adams, Amy Luinstra, Bill Maloney, Furio Rosati, Gary Thiesen, Gordon Betcheman, Irena Omelaniuk, Jean Fares, Jee-Peng Tan, Keith Hinchliffe, Jemal Mohammed Omer, Lorenzo Guarcello, Luc Christiaensen, Magdi Amin, Marito Garcia, Mesfin Girma Bezawagaw, Gizaw Molla, Bob Rijkers, Jennefer Sebstad, and Pieter Serneels. Finally, but not least, we are grateful to our peer reviewers, Pierella Paci and Tamar Manuelyan-Atinc, for valuable advice. - v - CONTENTS EXECUTIVE SUMMARY ............................................................................................... ix 1. INTRODUCTION ......................................................................................................... 1 Summary Outcomes .................................................................................................. 3 2 SETTINGTHE STAGE-BASIC INDICATORS AND TRENDS ............................. . A SnapshotofUrbanLabour Markets ...................................................................... 5 5 Recent Trends inJob Creation ................................................................................................ ................................................................................ 12 A NoteonData Sources andComparability ........................................................... Some Stylized Facts 14 18 3 WHY SO LITTLE JOB CREATION? . ........................................................................ The Transition Process: Expectations Versus Reality............................................. 21 21 The Pace of Reallocation: IndividualWorker Mobility .......................................... 23 Constraints on Jobs Growthinthe Private Sector................................................... The Impact ofPublic Sector Employment andRemuneration Policies..................27 32 The institutional framework for the labour market ................................................. 35 The impact of the business climate ......................................................................... 36 TheFunctioning ofthe Informal Sector .................................................................. 38 4 UNEMPLOYMENT .................................................................................................... . Levels and Trends ................................................................................................... 45 ...................................................................................... 45 Youth Unemployment ............................................................................................. Profile ofthe Unemployed 48 50 Regional Perspectives ............................................................................................. 51 Determinants o fUnemployment 51 KeyImplications ..................................................................................................... ............................................................................. 55 5. MIGRATION, URBAN LABOUR MARKETS AND POLICIES ............................. A SnapshotofInternalMigration............................................................................. 57 ................................. 57 Policy Challenges.................................................................................................... The Impact ofMigration on GrowthandPoverty Reduction 64 67 6. ADDRESSINGTHE POLICY CHALLENGES ......................................................... 70 The Urgencyofthe Labour MarketAgenda ........................................................... 70 Enhancingthe Private Sector Contribution to Employment ................................... IntegratingVulnerable Groups................................................................................ 73 78 Conditions for anEffectiveCivil Service Education and Training........................................................................................... ............................................................... 83 87 Enabling Migration to Contribute to Growth and Poverty Reduction.................... 88 REFERENCES ................................................................................................................. 91 .vi. ListofBoxes Box 1:Key Definitions .................................................................................................................... Box 2: Primary Labour Market Data Sources Used for this Study................................................ 6 19 Box 3: The 2001 Investment Climate Assessment Survey, and the Labour Market ..................... Box 4: Costs o f Formalization....................................................................................................... 43 22 Box 5: MeasuringUnemploymentRates and Trends UsingDifferent Definitions 46 Box 6: The A L M S Survey ............................................................................................................. ....................... 53 Box 7: Pushand Pull Factors inRural and UrbanAreas - Qualitative Evidence ......................... 61 Box 8: The Impact o f Migration on Wellbeing inRural Areas inEthiopia ................................... Box 9: The Importanceof Remittances inRural Portfolios........................................................... 64 Box 10: China: Institutional Reforms Help Reap the Potential o f Internal Migration ..................6567 Box 11:Perception on Constraints to Internal Migrationinthe PPA ............................................ 69 Box 12: The M A M S Model ........................................................................................................... Box 13: Are Active Labour Market Policies (ALMPs) Worthwhile? ........................................... 72 82 Box 14: Donors' Support o f the Tanzania Medium Term Pay Reform Strategy: the Selective Acceleration o f Salary Enhancement program (SASE) ................................................................. 86 ListofFigures Figure 1: Sectoral Composition o f GDP .......................................................................................... 1 Figure 2: MedianEarnings by Employer. 2005 ............................................................................. 10 Figure 3: MedianEarnings by Sector, 2005 .................................................................................. Figure 4: Average Monthly Earnings Relative to the Poverty Line, 2005 .................................... 10 11 Figure 5: Underemploymentby Gender, 2005 (percent "Available and Ready to Work More Hours") .......................................................................................................................................... Figure 6:Trends inUrban Employment by Industrial Sector, 1984-2004 ..................................... 12 13 Figure 7: Civil Service Growth, Federal and Regional, 1975-2003 .............................................. 14 2005 ............................................................................................................................................... Figure 8: EducationalLevels inTotal Population and inthe Labour Force (percent illiterate), 16 Figure 9: 1994-2004 Transitions for Selected Workers (percentage o f group in2004) ................24 Figure 10: MedianWage by Educational Level and Type o f Employer, 2005 ............................. 29 Figure 11:Annual Rate o f Separation from EthiopianFederal Civil Service 31 Figure 12: Comparisons o f Value Added per worker (US$, by firm size) .................................... ............................... 36 Figure 13: Perceived Constraints to FirmExpansion, 2004 .......................................................... ........................................ 37 Figure 15: Regional Averages, Adult vs.Youth Unemployment.................................................. Figure 14: Trends inthe EducationLevel o f Informal Sector Workers 40 46 Figure 16: Labour Market Status, Age 15+ ................................................................................... 47 Figure 18:Unemployment by education level (percent unemployed), Age 15-24 years ..............49 Figure 17: Unemploymentrate by Age and Gender, 2005 ............................................................ Figure 19: Unemploymentby education level (percent unemployed), Age 25+ years ................50 Figure 20: UrbanYouth Unemployment Rates in SelectedAfrican Countries (year varies) 51 Figure 21: Unemploymentby Regionand Age, 2005 ................................................................... ........50 52 Figure 22: A Comparisono f the Incidence o f Migration Based on DHS Surveys (Percentage o f Urban Women o fReproductive Age who are Recent Migrants, by Region) ................................ 58 Figure 23: A Percentages o f Urban Women o f Reproductive Age who are Recent Migrants, by City Population Size and Location, Rest of the World .................................................................. 59 Figure 24: Percentages o fUrban Women o f Reproductive Age who are Recent Migrants in Ethiopia, by City Population Size and Location ............................................................................ 59 .vii. Listof Tables Table 1:Trends inUrbanActivity Rates. by Age Group and Gender (percent) ............................. 8 ................................. Table 3: Trends inUnemployment Rates by Age and Gender (percent) ......................................... Table 2: Trends inEmployment Rates. by Age Group and Gender (percent) 8 9 Table 4: Share o f HighSkilled Workers innon-Professional Occupations (percent) ...................17 Table 5: Annual Accession rates, by Firmtype, 2001................................................................... Table 6: Transition Rates inHungary: 1994 - 2003 ...................................................................... 23 25 Table 7: Rate o f Absorption from Unemployment into Different Sectors .................................... 26 Table 8: Rate o f Entryinto Unemployment (as a share o fVarious Groups at Beginning of Period) ....................................................................................................................................................... 26 Table 9: Average Monthly Earnings. by Type o f Employer andBroad Occupation ..................... 30 Table 10: Changes inKey Indicators at FirmLevel. 1998/9 -2000/1 .......................................... 33 Table 11: Share of FirmsReportingConstraint as Severe or very Severe .................................... 34 Table 12: Private Sector Focus Group Results: Perceptions o fProgress ...................................... 37 Table 13:Prevalence o f Informal Employment by Sector (as percent o f total employment inthe sector) ............................................................................................................................................ 39 Table 15:Transition Probabilitiesout o f Unemployment ............................................................. Table 14: Marginal Tax Rate for Presumptive Income Tax on Business ...................................... 43 Table 16: Share o f Unemployed inDifferent Durationo f Unemployment Groups (percent) .......47 Table 17: Primary Activities o f the Unemployed inthe Past Week (percent) ............................... 48 Table 18: Ratio o f Salaries for SelectedProfessions Relative to Civil Service, 2005 ...................54 84 VOLUME 11:TABLE OFCONTENTS 1. UrbanLabour Markets inEthiopia: Clarifying the Metrics 2 Youth Employment inEthiopia . 3. A portrait of InformalityinEthiopia: Diagnostics andPolicy Implications 4 Time to Move on? Migration. UrbanLabour Markets andPolicies . 5. Mobility and Earnings inEthiopia's UrbanLabour Markets 1994-2004 6. A Simulation PerspectiveonEthiopia's Labour Market up to 2015 7 Ethiopia - Labour Market InformationSystem . .viii. EXECUTIVESUMMARY The development challenge facing Ethiopiai s immense. Its people are among the poorest inthe world, the economy is growing too slowly andpopulationgrowth is rapid. Over the past decade Ethiopia has made significant progress on a number o f fronts. The economy recovered strongly following a severe drought in 2002/03 yet overall performance remains sluggish, even in the more dynamic urban sectors. While performing better than agriculture, the secondary and tertiary sectors contribution to total growth was only about 60 percent, much less than the African average o f about 80 percent. Structural transformation has been slow, both interms o f the traditional-rural to modern- urban transition, and of the transition from a socialist economy to one that is market- based. Agriculture has not been a dynamic source o f rural job creation and its low productivity i s associated with persistent poverty in rural areas and the growing attractiveness o f towns and cities. Ethiopia's second generation PRSP (PASDEP), recognizes that fostering growth requires addressing not only the barriers to productivity in rural areas on which policy attention has been focused but tackling the bottlenecks which are preventing other parts o f the economy to fully contribute to growth. This renewedfocus on growth andjob creation inurbanareas is particularly timely given signs o f growing urban poverty and the rapid growth in the urban labour supply. Such growth has been fast and accompanied by several major shifts on the supply side: 0 A demographic shift. With an estimated 44 percent o f the population below the age o f 15, Ethiopia now faces the opportunities and challenges o f integrating a rising number o f youth into the labour market. There is the potential for a demographic dividend, especially as large investments ineducation are paying off and new entrants to the labour market have higher educational levels. Yet this will require significantly accelerating the pace o fjob creation inthe economy. 0 A geographic shift. Internal migration, while limited by the current institutional framework, has been rising. The current situation is one o f disequilibrium, and even without removing some o f the constraints to mobility it seems likely that migratory pressures will continue to accelerate. 0 A qualitative shift. The literacy rate of Ethiopia's workforce is rising, albeit from very low levels, consistent with Government's strategic objectives. The latest data available show significant improvements inprimary enrollment rates, which rose by over 20 percent over the period 1999-2004. Nevertheless, enrollments, at 74 percent, are significantly below the Sub-Saharan Africa (SSA) average, and are particularly low for girls (52 percent versus an SSA average o f 81 percent). Against this background andthe challenges that Ethiopia faces, this study: - ix - 1. provided a basic assessment of key labour market indicators and their recent trends, based on the 1999and 2005 LFS. 2. analysed the different segments of urban labour markets and investigated whyjob creation in urban areas has been sluggish despite recent good performance on growth. 3. reviewed some of the policy options facing the Government to capitalize on its reform effort and achieve higher rates ofjob creation. Labourmarkettrends: 1999 -2005 The data show a largely unchanged labour market situation between 1999 and 2005, with the notable exception o f improvements inunemployment: Participation rates remained stable at around 65 percent, with increases for women offset by declines for men. International comparisons suggest that Ethiopian activity rates for older groups are comparable with those o f Uganda and Kenya, but noticeably lower for youth. As expected, activity rates tend to rise with educational levels, though for youth, the distribution i s bipolar, with higher activity rates at both ends o f the education distribution. The employment rate remained stable at around 55 percent though there were some changes. In particular, for women improvements in employment were recorded for both the young and the adult cohorts, though overall (and particularly older women) they continued to be less likely than me to be employed. In terms of composition, to date employment is dominated by self-employment (42 percent o f workers), with government employees a distant second (18 percent), followed closely by private organizations (16 percent). The Government is the most skill intensive sector. Both public and private services have been driving net employment creation in urban areas. Since 1999 there has been a strong growth in "public administration" and "other household services." Manufacturing has continued to add jobs, albeit at a reduced pace in the previous 5 years. Within the services category financial services grew spectacularly (by more than 300 percent) since 1994, though from a very low base. Unemployment declined from 14.8 to 13.5 percent, drivenby the drops for youth and older males. However, youth continue to experience relatively much higher unemployment than for older cohorts. Women, and particularly older women, are much more likely than men to experience unemployment. The Government relaxes the "active search" criterion when measuring unemployment, on the grounds that long unemployment durations discourage large numbers o f workers. Between 1999 and 2005, passive unemployment declined even more than active unemployment (the share o f population inpassive unemployment declined by 38 percent, versus a decline o f 9 percent in active unemployment), partly in favor o f employment and even more o f inactivity. The shift in passive - x - unemployment was particularly visible for the youth, which i s compatible with higher school enrollment. (d) Median wages for urban paid employees are generally low (Br 300 a month, or US$34 in 2005), although significant differences are found by type o f employer, sector andworker characteristics. The employers that pay higher medianwages are parastatals and government. Interms o f industrial sectors, median wages are highest in the financial, real estate, and transport sectors, and for service providers and government workers; wages are lowest for domestic employees. Some stylizedfacts for policv makers: The size and quality of urban labour supply is increasing. Inparticular the skills profile o f the urbanworkforce has been rising. By 2005 three quarters o f youth had at least 4 years o f education, while for the older cohort this was true for only half. Inline with the overall decline in illiteracy and growing supply o f skills, the educational composition o f employment i s changing, with lower education levels accounting for a smaller share o f employment. An important part o f the rise in education is that women are catching up with men in terms o f education (the decrease in their share of illiterates is more marked than for the group as a whole), but females on average remains significantly less skilled. Further, adult women with some level o f education are significantly less likely to participate inthe labour market active than the average. The smallest gaps in activity rates are for the small but growing group o f women with at least some higher education. And women continue experience the disadvantage which comes with traditional gender roles. Having children younger than 10, in particular, significantly lowers the probability o f women beingemployed. Supply trends are likely to increase labour market participation as better educated people are more likely to be active. Yet increased labour force skill and female participation also raise challenges, particularly interms o f youth expectations for better paying jobs. An important role in this context is played by migration. While the current system discourages rural-urban migration, there i s evidence o f growing migratory pressures. To date most o f migration is relatively short distance, with both step-migration and moves to urban areas directly from adjacent urban areas being quite common. The available evidence shows that those who migrate for work related reasons have a better employment performance than the others. While migrants do not seem to "compete" directly with non-migrants and to drive down their wages, except for those with very low skills, they do seem to be in competition with each other. These findings do indeed raise concerns for a possible scenario o f increasing migration due to distress in rural areas, resulting inincreasingpoverty among migrants (and ultimately also among the long term resident population) inurbanareas. Education pays, particularly for women and in urban areas outsideAddis. Returns to education are relatively high, with the high skilled premium equal to 90 percent the - xi - earnings o f the illiterates (completed primary results in a 16 percent premium over the illiterates). While there is a gender wage gap o f around 22 percent, controlling for individual characteristics, women do obtain much higher returns to education thanmen at the upper end o f the educational spectrum. In urban areas outside Addis, returns to education are significant and higher thaninAddis for all levels o f education. Some groups in the labour force face specific challenges. Over 80 percent o f jobs found by youth is inthe informal sector (including unpaid family labour), though they are better off when highskilled. Women are another group with specific challenges -despite an improved performance in terms o f employment and activity rates, they experience higher unemployment and are found mostly in less productive and informal activities which can be traced to traditional gender roles, and therefore experience lower pay for equal characteristics. Significant heterogeneity characterizes different types of urban areas. Larger towns, including Addis, are characterized by lower employment rates andhigher unemployment. Differences across urban areas partly reflect the different economic structures as, not surprisingly, smaller towns' economies are also more oriented towards the primary sector, and are more integrated with the surrounding rural areas which are characterized byhigher employment rates. Whv so little iob creation? Despite overall growth in the economy the urban sector has not yet become a dynamic engine for growth, employment creation andpoverty reduction. Two main factors seem to account for this picture. Firsto f all, segmentation. Labour markets remainbroadly separated into three segments: the relatively privileged public sector; the private formal, and the large informal economy, which i s mainly small scale and more `survival' than entrepreneurial. Panel evidence over the period 1994-2004 shows signs o f weakening since 2000 but still significant segmentation affecting the experience andprospects o f individual workers. Secondly, Ethiopia's transition agenda remains unfinished. Expectations that reforms would lead to a gradual reallocation o f resources from low-productivity sectors to more dynamic activities have been partlyrealized as reforms appear to have had some impact. New impetus, particularly on reforms that can energize the private sector, i s needed as outlined below. There is some good news: 0 Panel evidence reveals significant increases in mobility, albeit from a very low base-yet fewer than 50 percent o fthose unemployedin 1994 had ajob in2004 Evidence from the Investment Climate Assessment (ICA) shows signs o f restructuring though overall hiring and separation rates are low by market economy standards. Lookingto overview the main segments o fthe labour market, the highlights are: The civil service has expanded significantly due to decentralization-analysis o f its compensation reveals that changes are needed: for positions below professionals - xii - compensation is muchhigherthan inother sectors, for highlevel positions there are signs o f internal "brain drain" particularly towards NGO and international organizations. Job security remains a key advantage o fjobs inthe civil service. Parastatals continue to account for a significant share o f manufacturing employment, and is characterized by larger firms and older equipment. The analysis reveals signs o f labour hoarding, and working conditions (particularly remuneration) are more favorable than in the private sector. Panel evidence shows higher mobility out of the parastatals over the last years, yet privatization seemed to runout o f steam in2002. The informal sector seems to be a residual or last resort in that flows are largely from the ranks o f the unemployed and new entrants. This sector had been actively repressed during the Derg, and there is a sense that despite legal reforms the legacy o f previous policies might still be felt. Analysis o f firm level data show that, the "dynamic" component o f the sector is very small, and marginally more concentrated in the trade/hotels/restaurants sector than in manufacturing. Most firms do not grow and most respondents say that they resort to in informal activities out o f necessity. Constraints to business expansion (andpresumably formalization) include uncertainty on the regulatory environment, constraints in accessing factor markets (particularly land), and to some extent harassment by local authorities. The formal private sector has not been a dynamic engine o f growth, and in general Ethiopian firms are not internationally competitive. Labour market institutions do not appear to be a major barrier to doing business and firms (at least those already in operation) do not view the combination o f labour taxation andregulation- such as hiring and firing costs, unionization, etc. -as exerting a major negative impact. In terms o f overall business climate, however, access to land and the tax regime appear to be the major concerns voiced by the firms, although focus groups in 2004 revealed significant improvements in the business climate. While the personal tax system is now working well, the share o f firms reporting taxes as a major constraint inthe ICA is higherthan in comparator countries (72 percent vs. 48 percent inUganda and 68 percent inKenya) and the number o fdays spent dealingwith tax officials very high. The Dolicv agenda As segmentation and the sluggishpace o fjob creation result from a variety o fbarriers, an agenda to tackle these challenges is correspondingly broad. The report discusses the scope for five main areas o f intervention: Enhancing the private sector contribution employment through improving the business climate and business support programs. Important actions to improve the business climate include identifjmg measures to increase tax collection without creating disincentives to private sector development; strengthening competition by reducing the regulatory and policy obstacles to competition in key sectors (e.g. trade, logistics and capital markets), and effectively operationalizing the Trade Practices Commission and Proclamation and extending its purview to state owned enterprises; and supporting activities in the informal sector by removing constraints directly related to government actions/regulations o f the sector and o f the land and credit (increasing security o f tenure, facilitating private sector activity in microfinance). Additionally, as most o f those who - Xlll - ... transit out o f unemployment find work in the informal sector, business development schemes aimed at supporting entrepreneurship, and inparticular access to finance, could have an impact. Indeed there seems to be significant unmet demand for the training and BDS schemes currently available. Integrating vulnerable groups. The evidence shows that women and young people experience relatively greater difficulties in securing productive employment than adult men, suggesting that targeted active programs need to be designed to overcome the specific barriers they face (cultural attitudes, traditional allocations o f labour, difficulties in the school to work transition). Strengthening Technical and Vocational Education Training (TVET) and ensuring that it develops in line with market needs, and targeting women (particularly in the informal sector) are two important priorities in this area. Further, developing active labour market interventions to act as safety nets, while no substitute for dynamic private sector job creation, can help alleviate the negative effects o f unemployment for the most vulnerable. While Ethiopia has an extensive food for work program, it is only active inrural areas. Its extension to urbanareas may be the best mechanismto directly expand opportunities for the unskilledunemployed. Creating an effective civil service. Current remuneration and hiring practices appear to have two main consequences: relatively good conditions for the low skilled could distort the lower end o f the labour market, make private firms less competitive andpossibly also encourage "wait" unemployment; and the need to respond to a visible brain drain problem at the highend which, while not quantified inextent, appears to be undermining government capacity at the senior policy level. The current trend based on international experience is to disaggregate pay and employment reforms and sequence them sectorally (agency-by-agency reform) and/or by cadre (ie., creating a senior executive service). In the short term, however, it is possible to introduce (with donor support) simple and focused measures to help retain the key staff who have been identified as essential to senior level management andpolicy functions. Enabling migration to contribute to growth and poverty reduction. In the case o f rural development, migration needs to be seen in a context o f overall income diversification, favoring at least shorter range migration and connectivity with towns. Several policy elements currently pursued are in line with these objectives including strengthening the management, capacity and planning o f rural towns, which play a key role in providing services and access to markets for the rural population; and scaling up the provision o f rural infrastructure such as roads and electricity. Urban authorities, faced with the prospect o f a growing population, need to consider how to create an enabling environment which can turn the challenges posed by increased mobility into opportunities. This involves developing new alternatives for service delivery, providing urban infrastructure and considering access to land and credit by the newcomers. Particularly important in this area are policies aimed at facilitating flows o f people and information, and deregulation (for example interms o f access to services andjobs related to official residence) to prevent the creation o f a new urbanunderclass. - xiv - Strengthening monitoring and evaluation of labour market developments. The inclusion o f employment creation as a strategic priority has been an important innovation in PASDEP and offers an opportunity to enhance the value and policy impact o f labour market information. Monitoring and reporting on PASDEP should build upon an integrated Labour Market Information System. Collaboration among all stakeholders, a rationalization o f the data collection activities and improved coverage o f rural areas are important priorities inthis area. - xv - 1. INTRODUCTION 1.1 The development challenge facing Ethiopia is immense. Its people are among the poorest in the world, the economy is growing too slowly and population growth i s rapid. 1.2 Over the past several years at least, Ethiopia's economy has made some major strides forward - growth has been strong, the private sector contribution to output and investment has expanded, some key exports have taken off, and inflation has been contained. The composition of recent growth reflects good performance across sectors, as does the increasing importance of the private and non-agricultural sectors as the driver o f economic activity (Figure 1). Private investment's share o f GDP now exceeds 15 percent, and accounts for about 59 percent o f total investment. Private consumption and investment represent about 80 percent o f gross domestic expenditure. I Figure 1: Sectoral Composition of GDP 100% 1 0 % 4 , , , , , ( , , , , , , / s , , , , , , , , , , , . . . . . . . . . . . . . . . . . . . . C V s - g -Agriculture$ ) - g - -g- g g g & g E .Industry -Senices Source; MOFED. 1.3 This report focuses on a central element o fEthiopia's challenge: the urbanlabour market.' We adopt this limited focus for several reasons, most notably because the dual nature of the economy, a sharper em hasis on the problems, perspectives and environment o f the non-agricultural sector i s useful. Moreover, it is inurban areas that P the growth in labour supply will bring about the most visible changes. This in no way detracts from the need to develop the market and service-providing role played by rural towns and agricultural development. 'The Ethopia CSA defines urban areas as "localities with 2000 or more inhabitants" though inpractice (i) all administrative capitals (regional, zonal and wereda capitals), (ii)localities with urban dweller's associations not included in(i) (iii) localities not included ineither (i) (ii) inhabitants are and all or whose primarily engagedinnon-agricultural activities are considered as urban. 'World Bank (2005a), (2005b), and (2006). 1 1.4 The urbanemphasis o f this study i s strategic. The recently finalized Government strategy-PASDEP (Plan for Accelerated and Sustained Development to End Poverty)-aims to supplement the existing Agricultural Development Led Industrialization strategy with a new emphasis on urban development, and adopts job creation as one o f the priorities, cutting across all the elements needed for pro-poor growth. This shift i s in line with international as well as Ethiopian evidence on the complementarities between urban and rural development, as well as the impacts o f cities on growth andpoverty reduction. 1.5 While Ethiopia is still a predominantly agrarian society, wherein an estimated 85 percent o f the population lives inrural areas, and some 80 percent make a living through agriculture-over the last decade growth per capita in the non-agricultural sector has been more than two and half times higher than in agriculture (averaging 2.3 rather than 0.9 percent ann~ally).~The share o f agriculture in GDP declined by 7 percent over the period. 1.6 The rebalancing suggested in PASDEP implicitly recognizes that urban agglomerations o f all types across the urban spectrum (from small rural towns to large cities) are where the more dynamic secondary and tertiary activities take place. The agglomeration o f people and firms incities, when supported by a conducive institutional environment, leads to "increased efficiency in flows o f goods and services, matching o f workers to jobs, flourishing o f markets, and spillovers o f knowledge and information" (Kessides 2005, xiv). Yet inEthiopia, despite a better performance than agriculture, the 4 contribution o f the secondary and tertiary sectors to growth has been about 60 percent, much less than the African average o f about 80 percent. Fostering growth requires therefore addressing not only the barriers to productivity in rural areas, but also the bottlenecks which are preventing the non-agricultural economy from fully contributing to growth. 1.7 The emphasis on unleashing the productivepotential o f the urban economy goes handinhandwith a focus onjobs creation, particularly inurban areas. The pressures o f a growing labour supply4and o f high levels o f unemployment make this challenge very visible, particularly as inurbanareasno safety nets system exists. 1.8 Focusing on growth andjobs inurban areas i s also central for poverty reduction. While the incidence o fpoverty i s higher inrural areas, and the majority o f the poor lives in rural areas, the case of Ethiopia is notable in Africa (together with Nigeria) in the relative similarity o f the urban and rural poverty rates - at 37 and 45 percent respectively. This is surprising as in the majority o f African countries the greater economic opportunities offered by urban areas translate into much larger gaps inpoverty incidence (Kessides 2005). While urbanization in itself can improve welfare outcomes It is very difficult to arrive at reliable estimates of sectoral growth, particularly as population numbers are basedon projections inthe absence o f a recent Census. 4Population growth since 1990 has been estimated at around 2.6 percent. The population 15+ in urban areas has grown by an estimated 827,000 between 1999 and 2003, and 93 percent o f which has been for the age group 15-30. This has occurred while fertility inurban areas declined particularly inAddis Ababa where it is now estimated to be below replacement level. 2 over the mediumterm, inthe meantime-as experienced by many developing nations in LatinAmerica andAsia-urban poverty canbecome a new and more visible plight. 1.9 InEthiopia the recent Poverty Assessment (World Bank 2005a) identified signs that poverty in urban areas was on the rise, including a growing incidence among household heads involved in non-agricultural activities. This can be traced to trends showing an increase in the numbers o f self-employed in urban areas, largely in low- productivity activities and in the informal sector. While more analytical work on poverty and urbanization i s underway, the importance o f creating functional cities with a supportive economic environment characterized by affordable basic services and without barriers to the economic inclusion o f the poor i s clearly part o f the growth and poverty reduction agenda inurban Ethiopia. 1-10 Focusing on the urban agenda does not detract from but rather complements the attention due to the rural agenda and the plight o f rural poverty. The importance o f rural-urban linkages for poverty reduction has been long recognized (IFPRI 2005) and recent evidence from Ethiopia (Dercon and Hoddinott 2005) illustrates these various linkages. As expected, local market towns play a major role in terms o f selling and buyingo fproducts (as well as inthe delivery o fbasic services) while larger urban areas and Addis offer opportunities for economic diversification through migration and the receipt o fremittances. Thus, boththe connectivity between urbanand rural areas (roads, rural telephony) and the functioning o f urban markets-of the labour market in particular-affect rural households' returns to labour and skills. 1.11 One o f the main linkages between urban and rural areas, where labour markets are central, i s internal migration. While mobility in Ethiopia has historically been low, there are signs that internal migration i s on the rise. Analyzing the performance o f migrants in urban areas, and the impact o f internal migration on growth and poverty reduction is therefore central to understanding an important part o f the structural transformation process ahead. SummaryOutcomes 1.12 The performance o f Ethiopia's urban labour markets has been disappointing. The headlines, which are detailed inthe report, are dramatic, andinclude the following: Openunemployment has been persistentlyhighand average duration is long, though recent trends suggest improvedperformance. There i s significant segmentation-two relatively privileged sectors inthe public and formal private sectors, a massive informal sector and a large stock o f unemployed. Individual transitions across these states have increased over time, but remain relatively limited. Formal sector employment in urban areas is dominated by the state and manufacturing sector employment remains among the lowest inthe world. 3 The majority o f those who are working inurban areas are engaged ininformal sector activity, typically as a last resort but also as a persistent state. Average wages are low, especially for the unskilled and inthe informal sector - but productivity i s also very low. Women are especially disadvantaged inthe labour market-and typically face worse outcomes with higher levels o f unemployment, lower wages, and a greater concentrationinthe informal sector. Many youth seemto enter the labour market through low qualityjobs inthe informal sector or into unemployment. The challenges are growing, as the urban labour supply i s growing fast even with falling fertility rates. 1.13 Inthis context, a much better understanding of the functioning of urban labour markets, key constraints and implications for policy makers, is a priority. This report contributes to the agenda in several key respects. We establish basic metrics, by exploiting the rich data sources that are available, looking at individual decisions (like labour force participation), firm data, and sectoral and aggregate outcomes. We explore where and what types o fjobs have been created, what has been driving those dynamics, and how labour market outcomes have responded to changes inthe policy agenda, such as economic restructuring and privatization. 1.14 Throughout, this report seeks to highlight the policy dimensions o f our findings. The discussion is significantly shaped by the need to accelerate the structural transformation o f the Ethiopian economy, both in terms o f the traditional-rural to modem-urban transition, and the transition from a socialist to a market-based economy. The centrality o f the private sector to a buoyant labour market entails a close focus on constraining factors and options for more proactive policies, especially relating to informal operators. 1.15 The structure o f this report is as follows. Volume Isynthesizes the emerging findings and policy implications while Volume I1presents a series o f thematic chapters which summarize the underlying background work. Inthis volume the next chapter sets the stage for the analysis by clarifying the metrics o f the key labour market indicators. Chapter 3 looks at the structure o f urban labour markets and what has hindered their ability to generate jobs despite the acceleration o f growth inthe last few years. Chapter 4 focuses on the challenge o f urbanunemployment, while Chapter 5 looks at the effects o f migration on urban labour markets. The final chapter in this volume reviews the emerging policy agenda. 4 2. SETTINGTHE STAGE -BASICINDICATORSAND TRENDS 2.1 The Government's poverty reduction strategy identifies urban development and employment creation as strategic priorities, and this chapter attempts to set the stage for this challenge byproviding botha snapshot andrecent trends inurbanlabour markets in Ethiopia. Establishing the basic metrics o f the labour market is a major contribution o f this report. Significant efforts have been made to reconcile the variety o f data sets, definitions and other measurement issues. A series o f stylized facts is developed to help characterize the functioning o f labour markets here, and a more detailed analysis o f the data i s provided invol. 11, Chapter 1. A Snamhot of Urban Labour Markets 2.2 How well are labour markets working in Ethiopia? People and policymakers typically look at employment opportunities and wages to reflect the health o f the labour market and indeed o f the economy overall. A GOE inter-ministerial technical committee has identified a list o f 14 indicators to be monitored to inform policy making, andwork is underway to design a labour market information system (see vol. 11, Chapter 7). 2.3 Key labour market indicators - activity rates, employment and unemployment rates, wages - help to take the pulse in urban areas. Throughout this study and with a more disaggregated analysis (see vol. 11, Chapter 1) we will complement the insights that these basic indicators offer, particularly concerning the quality o f employment (we will look at wages and dynamism o f various sectors, and the inability o f many o f the existing jobs to support urban livelihoods, in line with worldwide evidence on the working poor). Our approach i s designed to resonate with national policy makers, while using international definitions as far as possible (Box 1). While a range o f data is available (see Box 2) we focus here on the two LFS surveys to identify trends over time, inorder to maximize comparabilityacross sources. 5 Box 1:Key Definitions Inthis reportwe adopt theKILM(Key Indicatorsofthe LabourMarket) definitions developed by the ILO to identify the main labour market indicators, unless otherwise specified. There are different names for eachindicator, which we use interchangeably: Activity Rate or Labour Force ParticipationRate or ParticipationRate - the share o f the population aged 15+ either engaged in, or available to undertake, productive activities. It thus captures the idea o f labour supply for all productive activities according to the 1993 UNSystem o f National Accounts, though to the extent that some sectors (e.g. informal activity or where women predominate) are not fully reflected in employment data, labour supply i s underestimated. Employmentrate the share ofthe employed over the total populationagedover 15 years. The - employed are defined as all persons 15+ who were either in paid employment (including apprentices), or in self-employment including those working as employers o f others, own- account workers, members of producers' cooperatives, and those temporarily not at work but with a formal job attachment over the last week. This measure captures the ability of an economy to createjobs. The definition adopted focuses on the population 15+ (rather than 10+ as adopted by the CSA) inorder to derive a clearly monitorable indicator which can be a policy target. Including the those aged 10-14 years creates an implicit conflict between employment Unemployment rate - the share of the unemployed over the labour force (employed + and education policy objectives. unemployed) population aged 15+. According to ILO conventions this means, during the reference period, being: (a) "without work", i.e. not inpaid employment or self-employment, (b) "currently available for work"; and (c) "seeking work", i.e. having taken specific steps to seek such employment. Relevant to Ethiopia is the recognition that "in situations where the labour market i s largely unorganized or of limited scope, where labour absorption is, at the time, inadequate, or where the labour force is largely self-employed, the criterion of seeking work" may be relaxed. The CSA definition i s based on a "relaxed definition" which leads to higher unemployment rates. However inthis study we adopt the internationally comparable definition basedon all three criteria. Activity rates 2.4 At about 65 percent, economic activity rates in Ethiopia (Table 1) are low compared to some other African countries (for example, 85 percent in Kenya and 79 percent inUganda; Leibbrandt and Mlatsheni 2004). In general, participation is higher among men than women and among adults than youth, and increases with educational acquisition. Male participation i s roughly in line with other African countries (which have strikingly similar rates in general) and female participation i s below average. This reflects the economically marginalized position of Ethiopia's women, who have extremely heavy domestic work burdens and often little education. 2.5 Recent trends show a decline in the activity rate o f male youth, but an increase among young and adult women; at the same time there has been a decline in activity among those at the higher end o f the educational distribution (with the exception of the most educated-those with at least some tertiary education). The decline among young males reflects the positive effects o f rising school enrollment. Meanwhile, activity rates rose significantly among women aged 20-24 years and 40-49 years, which may reflect better economic conditions in2005 than in 1999 (the two data points). 6 2.6 Analysis o f the determinants o f activity reveals that women in general (when controlling for other characteristics) are not more likely than men to be inactive, though this is thecase for the age group 25+. This effect hasbecome less markedbetween 1999 and2005. As female inactivity is likely to be at least partly driven by traditional gender roles inthe households, the role o f family composition has been explored inthe analysis. Individuals living in households with more adults are more likely to include inactive adults-in 1999 this effect was stronger in urban areas outside o f Addis, while in 2005 the reverse was true. Women livingwith children younger than 15 years are more likely to be inactive, an effect which is not significant inAddis but i s inother urban areas. 2.7 Education, as shown also by the descriptive analysis, i s correlated with patterns o f inactivity with the more highly educated less likely to be inactive. Our analysis confirms this finding-for example, holding all else constant, in 2005 those in the "completed general education" category were 17 percent more likely to be active than all other education groups. Interestingly, however, in 1999, having completed general education had also a significant impact in reducing inactivity among adults, an effect which in2005 was not significant. Employmentrates 2.8 Partly as a reflection o f low activity rates, measured employment rates are low, particularly for women (Table 2). The average employment rate i s 56 percent. About 41 percent o f youth are employed. Fewer than 40 percent o f the employed are women, albeit with significantly smaller employment gaps for younger age groups than for older ones. Welcome new evidence suggests that female employment outcomes may be improving: since 1999 employment rates have been steady overall, but have risen both for young and adult women. Unemploymentrates 2.9 The official rate o f urban unemployment in2005 was 21 percent (this is based on the population age lo+, while this study looks at those age 15+). Using the international definition, measured unemployment i s lower than official numbers but still high at 14 percent. Standard international definition requires active job search, while Ethiopia relaxes the active job criterion on the ground that long durations o f unemployment discourage workers from actively searching (see Box 5 inChapter 4, this volume). That Ethiopia, together with South Africa, is one o f the only countries in Sub-Saharan Africa officially reporting unemployment at more than 20 per~ent,~is partly due to the national definition ofunemployment. 2.10 Urbanunemployment rates (Table 3) havebeen consistentlyhighandduration is protracted, with distinctive patterns by age cohort, gender and education. Adult male unemployment fell by one percentage point (fiom 9.1 to 8.1 percent) fiom 1999 to 2005, and stagnated around 13 percent for adult women. The gender unemployment gap therefore increased slightly over the period for the adult cohort. The median duration o f South Africa previously reported the relaxed definition, but has moved to the internationally comparable one. For 1999 those stood respectively at 36.2 and 23.3 (Kingdon and Knight 2006). 7 Table 1: Trends in UrbanActivity Rates, by Age Groupa d Gender ~~~~~~~~~ I I 1999 2005 I Total sarnplr 64.8 65.3 Male 76 73.3" I Fenlrtlc I 47.6 I 499s I I ce: For all table UBEGS2003 and 20 INote: * S i ~ i ~ ~ ~a n ~ Ii from ~1999 figure at ~ f ~ ~ 95 f ~ ~ n ~ percent, though results should be interpreted with caution. 8 IAge group 2% Io t ~saniple l 1 ___.. 10.0 10 6 -I f\ldlt 9.1 Y I* 2 11 While finding a job is a c h a ~ ~ e even~ having a job does not ~ u a ~ the ~ ~ ~ e e n ~ , ~ means to escape poverty. ~ ~ d wages" nfor urban paid errlployees are ~ e t i e r low~ ~ ~ a a ~ or US$34 in ZOOS), aIthaugh s ~ ~ ndiffe ~ ces ~ nfound across ~ ~ are t ers that pay higher sectors, and far service providers and g o ~ e ~ ~ers;c wages~ are lowest for ~ 6Gross earnings from main ~ ~and i n c~l ~ ~bulauses, overtime, allowances and other ~ i i ~~ i p ~ ~ benefits ~ n fLFS 2005). Eaniings are reportedonly by "paid ~ ~ ~ ~ o y(48epercent of urban e ~ p l o y ~ dAs .most e s ~ ' ~ infnrntlrl e ~ t ~ b l ~ s are~selfe e nt ~p l ~ ~(notd ''paid e~pIo~ees'~~, ~ ~ n i ~ reposted i ~ ~ wages~Llkelg ~ f ~ a ~ overstate thc averagc. 9 2.12 Those with little ~d~catjoilface p a ~ ~ ~ ~lowawages. An illiterate person l r ~ y earns only Br 4 per day ~ o i ~ with~Br~34efor someone with at least some higher ~ d ~ d u c a t ~ o~~ . ~ l ~analysis c ~ o a ~ thatee~ a ~~ isrise~ with ~ d ~ c having ~ ~ ~ : ~ ~ t ~ ~ s a ~ some p~~~~~ e~~~~~~~~~ associated with wages 16 percent highcr than those of is ~ ~ l ~ t ~workers, aid higher edu~~rion 90 percent increases over illiterate workers r a t e with (although there is a puzzling drop in the premium from non-complete general education-70 percent-to general complete-58 percent). The skills premium is particularly strong inthe case o f higher education for men, while for women the biggest increase i s for those with more than general ed~cation.~ 2.13 Skills gaps partly explain gender differences. At the national level, the average male adult has completed 1.8 grades, the average female less than one grade. Even among economically active young women about a quarter are illiterate (compared to 9 percent for men) and therefore only able to access low-payingjobs. Thus workers with less than four years o f education are likely to be living inpoverty. Figure 4 compares monthly earnings against urban poverty lines (upper and lower). This comparison i s likely to underestimate the incidence o f working poor in urban areas as these refer to individual consumption needs at an austere level-Br 150 or USD$20 per month-obviously breadwinners with dependents find the challenge o f sustaining themselves andtheir families muchmore difficult. Figure 4: Average Monthly Earnings Relative to the Poverty Line, 2005 1000 , 900 800 700 600 500 Formal 400 0 Informal 300 200 100 0 Source: LFS 2005, own calculations, and World Bank,2005a. Note; The dashedline indicates the lower (Br 1 lo); the dotted the upperpoverty line (Br 163). These are unweightedaverages from World Bank, 2005a. 2.14 Low wages are associated with underemployment, particularly in the informal sector. Figure 5 shows that the percent o f workers who are "available and ready to work more hours" i s highest among youth and prime-age adults, and higher for males than females, 'As expected, returns to education are higher for women once marital status control for selectivity, since women are generally less likely to be employed, even at highlevels of education. 11 Figure 5: Underemploymentby Gender, 2005 (percent "Available and Readyto Work iII.'/, More Hours") 40 - n n 2 35 - 30 - 42 I 25 - 54 2 0 - 1 5 - 1 0 - Female 5 - 0 7 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60+ Total Age (years) 2.15 To sum up, urban areas in Ethiopia are characterized by persistently high unemployment (although the typical duration is decreasing), and substantial wage differentials between the formal and informal sector, and across types o f jobs. Significant heterogeneity characterizes the performance o f different groups, with women andyouth facing particular challenges. The following sections describe recent trends, to set the stage for possible explanations inthe next chapter. RecentTrends inJob Creation 2.16 Over the decade to 2004 the urban economy8 grew at an annual rate o f approximately 2 percent per capitaagWithin services the highest rates o f growth were registered in banking and insurance (9 percent), health and education (8 and 7 percent respectively) and in transport and communication (7 percent). Within industry, construction grew at 8 percent while manufacturing growth was limitedto 3 percent. 2.17 Both the public and private services have been driving net employment creation in urban areas with "public administration, social, cultural, recreational, personal and household services" growing by an estimated 42 percent between 1999 and 2005, against 9 percent o f the 1994-1999 period. It is possible to disaggregate these trends only from 1999 - this reveals strong growth in the "public administration" and all subsectors o f "other household services." Manufacturing has continued to add jobs, albeit at a reduced pace, growing by 30 percent versus 37 percent inthe previous period For detailed analyses on the economic performance of the country as a whole and sectoral trends in employment creation see also two recent ILO studies by Demeke et a1 (2005) and De Gobbi (2005), as well as the World Bank Country Economic Memorandum (2007). This estimate is obtained comparing the real growth in industry and services and assuming only "natural" population growth (at 2.5 percent) and no migration. 12 (Figure 6). Within the services category financial services grew spectacularly by more than300percent since 1994, albeit from avery lowbase. Figure6:Trends inUrbanEmploymentbyIndustrialSector, 1984-2004 1200000 1000000 800000 600000 400000 200000 0 1984 1994 1999 2005 --cprimary -m- construction +electricity, gas and water supply +trademnufacturing &transport +financial and business services -PA and services* Note: *social, cultural, recreational, personal and household services. Sources: 1984, 1994 Demeke, Guta Ferede 2003 (based on Census sources), 1999, 2005. Own calculations basedon LFS. 2.18 By analyzing national accounts and employment data jointly, it is possible to disaggregate the productivity and employment components underlying sectoral growth. In aggregate terms it appears that the 1.9 percent per year growth in value added in agriculture and allied activities has been driven almost entirely by growth in employment with insignificant growth in productivity. In contrast, in industry the effects o f the almost 3 percent annual growth in employment have been eroded by productivity falling by 0.6 percent per annum. In services, the 2 percent annual growth registered has been mostly driven by increases inproductivity. 2.19 Disaggregated analysis shows that in most sectors that experienced more than 3 percent annual growth between 1999 and 2005, productivity declined while employment rose significantly. Among the high growth sectors, the exceptions are in transport and communication, where the annual growth in employment was negligible, and in the health sector, where employment actually shrunk while productivity increased significantly. 2.20 Between 1999 and 2004, the public sector and self-employment contributed similar amounts to urbanemployment, growing respectively by 7 and 8 percent intotal. The net public sector contribution, which was almost half o f total net job creation, reflects offsetting trends-a strong increase inthe number o f government employees and a fall inthe number o fparastatal employees. 13 2.21 The significant expansion o f the civil service by over 60 percent during the 1990s was driven by decentralization. Growth was especially rapid at local levels, doubling since 1992 (Figure 7). Figure 7: Civil Service Growth, Federal and Regional, 1975-2003 450,000 400,000 - e 8 350,000 - 300,000 - f 250,000 - -$!200,000 - -Total Public Sector 0 150,000 - -Federal W 100,000 - -A- Regional 50,000 0 Source: EthiopianFederal Civil Service Commission. Some Stylized Facts Some stylized facts are presented here to help to put the h c t i o n i n g o f the labour market incontextbeforeproceedingto explainingobserved trends inthe next chapter. 1. Thesize and aualitv of labour supplv havebeen increasing 2.22 Ethiopia's urban labour supply has been growing at a sustained annual pace o f 3.5 percent, partly due to internal migration. Inthe absence o f a recent census it i s hard to quantify, but there are concerns about absorption into the urban economy, and downward pressures on wages o f the local unskilledpopulation (see vol. I, 5). Chapter 2.23 The education profile o f the labour force has rapidly risen, but the averages remain low. From 1996 to 2004 urbanadult literacy rose from 66 to 74 percent. Still, in 2005 only about one in five in the urban workforce had secondary or higher education and average levels o f schooling o f urban labour force participants are still fairly low (about 8-9 years o f schooling for males, and about a year less for females, with a larger gender gap among older cohorts-but these averages exclude those who have not had any formal schooling). Sectoral studies reveal evidence o f specific skills gaps which create bottlenecks for job creation and rising productivity. , 14 2. Education pays, particularlv for women and in urban areas outsideAddis 2.24 Returns to education in Ethiopia are high and increase at higher levels. Taking illiterates as the reference group, the impact o f education on earnings ranges from 26 percent for those with grade 1-4 to 130 percent for the highest skilled. Women, who on average are paid 22 percent less than men with the same individual characteristics, obtain much higher returns to education than men at the upper end o f the educational spectrum. Although returns are about the same for primary education, returns to completed general education are 95 percent for women (to 77 percent for men), and returns to higher education are 147 percent for women (132 percent for men). lo 2.25 Differences inreturns also emerge when splitting the sample between Addis and other urban areas. In other urban areas, returns to education are significant and higher than in Addis for all levels of education-for example, returns to grade 5-8 are 32 percent in Addis and 49 percent elsewhere. Interestingly, returns from non-formal and grade 1-4 education are not statistically different from returns to no education (e.g. the illiterate) in Addis: suggesting that more than minimal education is needed to make a difference to earnings. Regional differences between Addis and other regions are statistically significant and o f the order o f 10-20 percent for most regions. Somali and Gambella enjoy higher earnings, while SNNP, Amhara, Oromia, Tigray and Benishangul have lower earnings. 2.26 Using the wage regressions to control for education, other important drivers o f differences in earnings emerge-in particular, employment status is significant for nearly all categories. Relative to civil servants, parastatal workers earn about 8 percent less, private sector workers 34 percent less, and domestic workers 88 percent less. 3. Supplv trends pose maior challenaes 2.27 Supply trends are likely to increase labour market participation as better educated people are more likely to be active inthe labour market: in2005, 95 percent o f those age 25+ with higher education were active compared to 62 percent o f the illiterate. Likewise, average years o f schooling are higher for the active population inthe older age groups, suggesting that highlyskilled workers have longer productive lives. 2.28 The educational profile o f younger generations in urban areas i s changing (Figure 8). Currently, the data show that gender participation gaps narrow significantly at higher education levels. Rising education will help promote female participation, though obstacles remain. lo Another distinguishing feature is that women are more likely than mento self-select into the workforce according to their personal characteristics and those women who would earn low wages (because o f the skill profile, the marital status, the age, number of children) chose not to work. 15 Figure 8: Educational Levels inTotal Population and in the Labour Force (percent illiterate), 2005 I I I 15- 20- 25- 30- 35- 40- 45- 50- 55- 60-1 19 24 29 34 39 44 49 54 59 Source: Own calculations. 2.29 Increased labour force skill and female participation are positive developments which also raise challenges, particularly interms o f the expectations o f youth. The 1994 labour proclamation abolished the job guarantee for university graduates. There was evidence that in the 1990s skilled youth were willing to queue and remain unemployed untilthey got "good jobs," particularly inthe public sector (Semeels 2004). Morerecent evidence suggests that individuals are more willing to take any job available (vol. I, Chapter 4). Panel evidence o f high durations o f unemployment also suggests that queuing i s not likelyto be a successful strategy to access goodjobs. 4. Youth, women and the disabledface specific challenges 2.30 Youth employment rates have fallen significantly, and many youth seem to enter the labour market via low quality jobs or unemployment. Only about half o f Ethiopia's 2.4 million urban youth were employed in 2005. About 80 percent o f employed youth work in the informal sector, and many o f them are unpaid family workers. The better skilled have better access to paid employment: in 2005, only 27 percent o f illiterate youth were inpaid employment against 61 percent o fthe highskilled. 2.3 1 Skills mismatch i s another problem facing youth. Although this generation i s the most highly educated in Ethiopia's history, if youth can not put their skills to use, the private and social returns expected to follow the government's massive investment in education will fail to materialize. It will also come with a risk o f frustration among the increasing numbers o f educated urbanyouth. Inthis context, the change between 1999 and2005 is very encouraging. In the first Labour Force Survey (LFS), about two-thirds o f highly skilled young workers were in non-professional jobs; by 2005, the share had fallen to just over one third (Table 4). Highly skilled adults were also much more likely to be inprofessionaljobs in2005 than they were in 1999. In 2005, 36 percent o f highly skilled 15-24 year olds were in non-professional jobs (compared to 26 percent o f their adult counterparts). 16 Age 1999 2005 Ane 15+ 39.8 27.8 Age 15-24 67.8 36.0 Age 25+ 35.7 26.4 2.32 Labour market outcomes are heavily affected by traditional gender roles. Analysis o f the determinants o f employment suggest that inurbanEthiopia as elsewhere, household responsibilities and particularly child care are significant obstacles to labour marketparticipation. Women are underrepresented among white collar workers -while they represent almost half o f total employment their combined share in"technicians and associate professionals,'' "professionals" and "legislator senior officials and managers" does not exceed 30 percent. 2.33 Women earn less than men: given the same observable characteristics, women are likely to be paid 22 percent less. At the same time, decompositions o f inequality o f earnings across gender and educational levels show greater heterogeneity among women, which means that some women do better, others do muchworse. 2.34 In Ethiopia as elsewhere, the disabled face special challenges because of discrimination and reduced capacity for work. For example, the PPA finds that inurban and rural areas the disabled, being unable to secure jobs, are likely to face downward trajectories inwellbeing. Disability in itself is difficult to define and census and survey data often provide very different estimates o f prevalence. For Ethiopia, the latest CSA estimate (2003) i s o f a prevalence o f 7.6 percent (some 5 million people), compared to Uganda, for example, where 5 percent o f the urban household heads were identified in the census as disabled. International estimates often report higher prevalence rates (10- 15 percent), o f whom typically about one third face serious disability. 5. Thereare sipnificant differences across urban labour markets 2.35 Urban areas as defined in Ethiopia are heterogeneous. "Urban" includes "localities with 2000 or more inhabitants" but also (i)all administrative capitals (regional, zonal andwereda capitals); (ii) localities with urbandweller's associations not included in (i); all and (iii) localities not included in either (i) (ii) inhabitants or whose are primarily not engaged innon-agricultural activities. Ongoing work i s analyzing the differences inthe economic roles betweenurban areas o f different types. 17 2.36 Distinctions are generally drawn between primary cities (with more than 1 million people) and secondary cities (between 100,000 and 1million). The distribution o f the urban population in Ethiopia i s skewed: there i s only one primary city, Addis Ababa, with the next largest city (Dire Dawa) having an estimated 300,000 residents. For analytical purposes here we distinguish between larger towns-i.e. the three largest cities-and other urban areas. Addis Ababa, Dire Dawa and Harar together account for about one-third o fthe total urbanlabour supply. 2.37 Compared to small towns, large towns are characterized by lower employment rates and higher unemployment for all age groups. For example, in small towns, employment rates for 20-24 year olds are 10 percentage points higher than in large towns (57 percent vs. 47 percent). Employment rates are higher among prime-age workers than among youth regardless o f location, but the gap between small and large towns remains (e.g. the employment rate for 40-44 year olds is 82 percent in small towns, vs. 70 percent in large towns). Employment in smaller urban areas shrank between 1999 and 2005, as did activity rates, while unemployment increased. 2.38 Employment insmaller cities i s more oriented towards primary activities, as well as to the hotel and restaurants sector. An upcoming World Bank study on urbanization inEthiopiawill shedmore lightonthese issues andthe role ofcities o fdifferent types in fostering growth and poverty reduction. 2.39 Different regional economic structures are reflected inpronounced labour market differences. Employment rates range from 44 percent in Gambella to 69 percent in Benishangul, and unemployment from 24 percent inAddis to 5 percent inBenishangul. The incidence o f informal employment also varies significantly across regions. However, further analysis is neededto understand the underlying determinants. A Note on Data Sources and Comparability 2.40 In analyzing labour markets in Ethiopia we are in the fortunate position of benefiting from a variety o f data sources, mainly collected by the CSA (Box 2), but the analysis also faces significant comparability issues. These issues inevitably arise when drawing on different surveys, particularly if survey instruments are modified over time, ifnew instrumentsare introduced, andor ifdifferent surveys havedifferent geographical coverage. , 18 Box 2: Primary Labour Market Data Sources Usedfor this Study 1999National Labour Force Survev (LFS) conducted in March 1999by the CSA. The survey was nationally representative and covers all zones except 6 inthe Somali region and 2 in Afar. The data are representative at the national, urban, rural andregional level. The survey covers the population aged 10 and over, recording their labour market status as well as a variety o f socio- demographic variables such as age, gender, migration status and education. The survey did not collect wage data. The survey covers 31,859 households in urban areas, 2,518 o f which in Addis. 2005 National Labour Force Survey (LFS) conducted in March 2005 by the CSA. The survey was nationallyrepresentative and covers all zones except Gambellaregion (with the exception of Gambella town) and 6 zones inthe Somali region and 3 inAfar, mostly inhabitedbypastoralists. The data are representative at the national, urban, rural andregionallevel. The survey covers the population aged 10 and over, recording their labour market status as well as a variety o f socio- demographic variables such as age, gender, migration status and education. The survey instrument has been updated with respect to the 1999 version, the most notable change being that it includes wage data. Other changes, including an updating o f the statistical sampling frame and a more precise characterization o f urban areas which allows a break down o f major urban and other urban areas have also been made. The survey covers 29,623 households in urbanareas, 21,420 inmajor urban areas. Ethiopia Urban Household Socio-economicSurvq, ajoint program of Addis Ababa University and the University of Gothenburg, Sweden. Started in 1994, with four waves, in 1995, 1997, 2000 and 2004. Some 9,000 to 10,000 individuals in 1500 to 1600 households are covered. More than 40 percent of individuals in the first wave were tracked by all subsequent waves, while a muchhigher proportionwas coveredby at least three waves. A sub-sample o f the panel in Addis was re-interviewed inApril-May 2006 to provide additional information on informal activities andunemployment - see Box 6. Urban Informal Sector Survey (ISS) conductedin 2003 by the CSA. The survey uses the official definition of the informal sector (enterprises are informal if they have no book o f accounts; no license; and fewer than 10 employees). The survey covers urban centers only, and 15,035 operators were interviewed. The sample was derivedby sampling enumeration areas, compiling a roster of informal operators by inspecting all households living in the given area and finally randomly drawing statistical units from those rosters. 2.41 Some o f the challenges faced inthis analysis include: 0 Difficulties inreconciling evidence from different survey series - this report uses the LFS surveys as main sources o f information. The trends identified by these surveys differ somewhat from those of the Urban Employment and Unemployment Survey (UBEUS) series by the CSA, a smaller biannual survey. Difficulties include possible seasonality effects, and the different sampling frames adopted. T o the extent that the UBEUS becomes a regular annual or biannual survey, more analysis i s needed to understand how it can be made more comparable with the LFS. 0 Difficulties in reconciling some o f the panel with the cross-sectional evidence. The availability o f a panel is essential to understanding labour markets, yet trends might not be comparable with those of the LFS. The panel i s limited to 19 the seven largest cities inEthiopia, which together with differences inthe survey instrumentmakes indicators hardto compare.'' 0 Difficulties in reconciling administrative and survey data, particularly for the civil service. Administrative data exclude contract and temporary work while providing details on some jobs which-because infrequent-are less likely to be covered by the surveys. 2.42 Furtherwork intriangulating different sources andrefining survey instrumentsis needed, ideally by conducting purposefully designed surveys to test some o f the features o f the instruments now in use. As an example, a module o f the A L M S survey (Box 6) suggests that wages from secondary jobs might not be an important element o f overall earnings. Ifconfirmed, this suggests that the quality o f wage data currently collected i s higherthanexpected. l1For example, in the 2000 wave of the Urban Socio-Economic Survey Sample the share of the 15-64 age group who were in employment was 38 percent against 48 percent in the 1999 LFS; similarly, the share of those unemployed was 17percent against 16percent of the LFS. 20 3. WHY SO LITTLE JOB CREATION? 3.1 The performance o f urban labour markets in Ethiopia has been disappointing, even with the acceleration in growth in recent years. Jobs growth has been slow, and way below what i s needed to productively employ urban residents looking for work; unemployment is thus high. Even for those who have income earning opportunities, these are typically in the informal sector and very low paid. Compounding the challenges o f sluggishjob creation and low pay are the particular disadvantages faced by women andyouth. 3.2 Our hypothesis is that these dismal outcomes can be traced back, at least inpart, to segmentation o f the labour market. Ineffect what we see i s an urban economy that is to a large extent cut into three segments-the relatively privileged public sector; the private formal; and the large informal economy, which is mainly small scale and more `survival' than entrepreneurial. To explore this hypothesis, this chapter seeks to explain the scale and pattern o fjob creation inEthiopia and examine the policy and other factors which are generating the observed outcomes. 3.3 The chapter begins by reviewing the expected evolution o f an economy transiting from state domination to the market. To understand better the Ethiopian experience, we bring insights from a mix o f data sets, including firm level data and a household panel consisting o f four waves spanning 1994 to 2004. This panel offers a unique opportunity to monitor and understand the evolution o f the labour markets inthe 7 largest cities inthe country.12 We then turn to review more specific policy-related and other constraints associated with employment inthe civil service, parastatals, private and informal sectors in turn. Various policy implications emerge and are discussed more fully inthe final chapter. The Transition Process: ExpectationsVersus Reality 3.4 Ethiopia's transition agenda raises expectations o f a gradual reallocation o f resources from low-productivity sectors to more dynamic activities leading over time to a series o f observed labour market outcomes, including: 0 a decrease inunemployment, particularly o f the highskilled; a reduced role o f the public sector, and in particular o f parastatals, in the economy; and 0 a progressive shrinking o fthe informal sector. Addis Ababa, Awasa, BahirDar, Dessie, DireDawa, Jimma, Mekele. 21 The pace o f these reallocations should have increased as reforms gained momentum, so that individual mobility would increase over the period, in terms o f flows into, but especially out of unemployment, and from parastatals. 3.5 We turn now to see what has happened in practice in Ethiopia, both at the aggregate and sectoral levels, as well as from individual worker perspectives. 3.6 Growth in private employment appears to have accelerated. While the public sector continues to account for the bulk o f urban formal employment, the private sector has registered strong growth in recent years. However, it i s difficult to fully reconcile available data on trends. The panel(see 3.7 above) shows that the private sector accounted for nearly half o f employment by 2004.13 However care i s needed inusingthe panel to capture overall trends since as we saw inthe last chapter, the aggregate data show that the share o fpublic employment has risen. 3.8 Insights are provided by the ICA sample for 2000/01 - 1999/00 (Box 3), focusing on the accession ratel4and the separation ratet5in 2000/01. We find that the overall accession rate was about 16.7 percent, whereas the separation rate was 10.5 percent. Box 3: The 2001 Investment ClimateAssessmentSurvey, andthe Labour Market The ICA firrhlevel survey conductedby the World Bankin2001with data for 1998/99- 2000/2001provides empirical insightinto: 0 Scale, ownership & performance o f firms 0 Linkfromtheseattributes to thebusiness& financing environment, includingthe impact of labour regulation 0 Identification of characteristics o f firms performing welVpoorly 0 Labour market dimensions o fperformance, including gross flows 0 Disaggregated skill composition of employment over time by sector, ownership, size & other dimensions e Relativereturns to skills, education, tenure & other worker characteristics 0 Relative wages by ownership, sector, size & other firm attributes The individual worker data (n>2500) also permit estimation o f simple wage functions, and estimates ofthe extent ofjob mobility; returns to training, unemployment incidence & duration. The caveats include the relatively small sample size o f firms (n=427), and the partial coverage o f economy (limited to manufacturingand the formal sector). 3.9 These levels o f gross flows are small relative to OECD and many transition economies. There is significant variation across industries disaggregated by skill type, l3Unless otherwise noted, results on labor market transitions and other findings f i o m the panel are drawn from Mengistae, Shimeles, and Bigsten 2006; see vol. 11, Chapter 5. l4The accession rate is defined as the number o f new employees hired inone year divided by the number of workers at the end ofthe previous year. l5The separation rate i s the number of employees who left employment (because of retirement, dismissal or resignation) relative to the number of workers at the end of the previous year. 22 with state firms having markedly smaller gross flows (Table 5). The pattern is different among the highly skilled, however, with government-owned firms having a higher accession rate o fmanagers andprofessionals. 3.10 The story is quite similar when looking at separation rates inthe ICA. Managers at government-owned firms had a separation rate about twice as high as those at domestic private firms (15 percent vs. 7 percent), which is indicative o f restructuring (see below). However, unskilled production workers had a separation rate o f only 4 percent from government-owned firms, compared to 17 percent from domestic privately-owned firms. Table 5: AnnualAccessionrates, by Firmtype, 2001 Domestic Foreign Central Private Private Government Ownership Ownership Ownership Managers 0.06 0.09 0.12 Professionals 0.15 0.21 0.2 Skilled production 0.18 0.19 0.06 workers Unskilledproduction 0.23 0.33 0.02 workers Non-productionworkers 0.08 0.08 0.06 Source: ICA, 2001 (see Box 3). The Pace of Reallocation: IndividualWorker Mobilitv 3.11 An initial impression o f the situation in the late 1990s can be gleaned from the analysis o f individual worker experiences using the I C A dataset, where the evidence points to low turnover rates. Based on retrospective reporting, we find that over the three years to 2001, about half o f workers stayed in same job and one-third changed either once or twice. At the same time, about 23 percent o f workers who had public sector work experience (including the army) moved to the private sector. Very few (0.8 percent) o fworkers had shifted from private to public sector. 3.12 Evidence from the panel provides complementary and updated insights on the extent o f economic change and restructuring through the analysis o f the experiences o f individual workers over the decade. What emerges i s an overall picture o f change in appropriate directions, but from an extremely low base, and to an extent that has been very limitedto date.16 3.13 There has been extensive mobility out o f the public sector, from both the civil service and parastatals (Figure 9). Over the decade as a whole, there was significant outward mobility from the civil service-less than half o f those who had been working for the government were still doing so at the end o f the decade. About one in five had l6There are several labor market states that we can explore using the panel-employment inthe public sector, employment in the private sector (of which informal), unemployed, and inactive. The latter group may either have retired over the period, re-entered education, or withdrawn from the labor force, discouraged. 23 moved out ofthe ~vcre~ n e m p ~ o y $(some d retrench~en~~. F han onc in five had made a successful ~ ~ ~ ~to sthei t ~ o ~ sector-~~ow~ver~o f thcsc almost ha1f were relying on i n f o ~ i awork. l 3.14 The ex~e~enccof parastatal workers over the decade are s s ~ ~ ~of~ ~ t ~ ~ i e s i ~ n i ~ c arn~t s t ~ c t ~ ~sincenby,2004 fewer than one in five were in thc same state as r i ~ 1994. They also had a much higher ~ r o b a b ~ ~ofi tleaving the ~aboL~rEbrcc a~togethcr y (35 percent rather than 21 percent for civil s e ~ a ~coupled~with another 10 pcrccnt ~ ~ s , who repo~edt h e ~ s e ~ ~ase su n e ~ p ~ o y eind 2004. It is notable, however, that. very i few-only one inten-found tvork in the privatc scctor, and about twice as manyjoined the civil service. self-ernployed 3.1S The experience of the the decade o f the panel is clear and ~ ~ ?A very s~ i ~~ ~ ~n ~ ~c a .~ t ~ ercent-of those ~ ~ in 1`3'34 e ~ were still jobless a decade late af 22 percent had with market altogether, Thi the entire ten yea d. For those who did Gndjabs, the ~ ~path was i ~ ~ n ~ \vork-which we tends to be an option o f last resort-only 17 percent go private sectorjobs, and 12 percent d a public sector p ~ ~ ~ t ~ o n . 3.16 It is interest^^^ to contrast the ~ ~ ~ ~ i experiencen o p ~ a with Hungary's (Table 6). Hungary has been a successful refomicr tvhich has faced similar cha~len~es~ ~ h i o p i a to on the state ~ n ~ e front, as l c~ h o ~higher ~ ~ ~ h income levels allow e ~ ~ e welfares ~ ~ ~ e ~ ~ ~ r o ~ i s i o nU. n c ~ ~ l o ~ine thist case has been even higher: over almost a decade 80 n percent o f the ~ ~ e ~ ~ i p ~remaine d~ ~ ~ e ~ p ando overall , Icss than 11 percent o y ~ ~ e d ana aged to obtain a job. 24 ployed State 0.10 96.54 1.30 0.56 1.SO 100 piayee Private 0.39 0.52 96.94 0.82 1.33 z ao ernpioyee 0.74 3.13 7.29 80.14 8.69 100 OLF 0.27 0.67 f .03 0.93 97.11 Total 7.58 20.30 31.42 4.52 36.i8 Source: C o r n ~ ~ n2006,r 1.8 n ~ ~ ~ l~ ~~ ~r ~~ ~r ~ ~ i ~ n ~ . ~ e l 3.17 In Eth~op~a,the fact that few of the une~ployedobta~ne~o v e ~ i ~jobs~ over g ~ n t the ten year period casts doubt on whether a s i ~ ~ ~part c ~ofnu ~r e ~ p ~ o y r nreflected e n t i n d ~ ~ ~ ~ dprepared to wait for a public sector position (as was found for the mid- u a l s 1990s; Serneels 2002)~ The odds of actually securing such ajob are so lirnitedthat this would not be a rational or s~ista~na~leoption for indi~r~du~l~ to pursue. And as we f e r n in Chapter 4 of this volume, a f o ~ l o ~ survey~of the u n e ~ ~ ~ l o ineAddis in May ~ ~ - p y d 2006 f o ~ n that three-fou~~swould accept %nyjoh." d 3.18 ~ n f o ~sector a c ~ ~alsor tendsyto be a p e ~ s ~state ~For~the vast majority. a l ~ ~ ~ s t ~ The survey and qua~itativeevidence suggests that i~d~vidualsremain active in the ~~~0~~~ sector only inthe abse11ce of 0 shows very few sector workers or otherjobs (F A snlali but n from j ~ to public ~ ~ ~ ~ sector jabs, 3.19 The ~ r a ~ ~ i~~ i o~ n t sho~ i ~ e ~ s i g n ~ ~since ~ ~ l ~ c ~ 1397. This can be seen across the le while fewer than 7 at a d r a n ~ a ~rate, SO ~ c t in 1997 had some babitify o f ~ r ~ s ~ t i o ~ from u n ~ ~ p ~ oto~~~~~o~~~~~~~ five times higher bctwccn 1997 and e n t was it was between the initial waves, These relatively high rates of t ~ a ~ s ~ toutnof i o u i i e ~ p ~ oduring~the~period 199~-2000were more than sus~a~ned o u g h2004, at ~ e i t ~ 41 percent. 3.20 W i l e the higher flows out of u n e ~ ~ ~ ointo~ee ~n t~ ~ ~are ~~o ~r ne~ ant ~t b ~ e ~ o with s i ~ ~ i ~ cratest ofjob creation, r e ~ ~ t few~(only ~I3ypercent) foundjobs in the a n ~ ~ e formal private sector. The i n f o ~sector was the fargest single source ofjobs for those ~ 1 ~ ~ n e ~ ~ ~whoyfound jobs. However, fl0tc.s out of u n e ~ ~ ~into I o e d l othe~ p~u ~nl i~c sector rose over the period, mainly into the civil. service (Table 7). 25 2 10 I 13 I n f o r ~ sector a ~ 4 19 I7 Pubticsector t 8 11 0.6 I 8.6 I u ~ e ~ are GO ~ ~ ~ ~ e n ~ le only 2 percent o f th ployed in 1997, 18 percent of those in ycd in and rno~ern~n~sthis o f ~~~~~ n ~ ac o n~~ ~ ni over~the next four years (Table 8). ~ e d ~ ~ ~ Table 8: Rateof Entryinto W ~ ~ m p l o {as~aeshare ofVarious Groupsat ~ ~ ~ i ~ofn i n g ~ n ~ Period) 3.25 ~ ~ n ~also with~thee~ n~ s~~ r ~ c storyii~s the finding that, in the second and j ~ r u r g third periods, job toss rates doubled in statc owned ~ n t e ~ ~ and snearly did so in s e g ~ ~ e agencies,ealbeit~from ~ n ~ tvery~low~levels, ~ ~ ~ n i ~ y ~in the~ firstiperiod note ~ a single public ~ ~ ~worker became ~ in ~ ~ p ~eo5ypercent (8 percen~~ those ~ ~ r e d , of who were is or king in a p ~ ~ s in~ 1997r ~2~~~~were ~ ~ ~ e ~ p Ibyy2000 ~ ~ ~ ~ ~ 4 } . a ~ ~ o e d Likewise we can observe that 8otr.s into parastatal ~ ~ p ldec~inedover time, tas o ~ ~ ~ discussed below. 26 3.24 The experiences o f those in the informal sector have been disappointing. A significant share transited into unemployment and relatively few moved to formal jobs. The probability o f transition from informal to formal sector jobs nonetheless did rise, from the negligible base o f less than one percent (1994-97) to 15 percent in 1997- 2000.17 At the same time, the probability o f transition from informal to formal private sector jobs exceeded that o fpublic sector jobs. 3.25 That the informal sector i s one o f last resort i s consistent with the panel finding that flows into the informal sector are largely from the ranks o f the unemployed or new entrants. Consistent with the hypothesis that informal activities are typically pursued in the absenceo fother options, we note that: More than one in ten people who were out o f the labour force in 2000 entered the informal sector or self-employment by 2004, and about 20 percent o f those unemployed in2000 were self-employed or ininformal wage work by 2004. The probability that a formal sector worker in 2000 would be found in self employment or in informal paid jobs in 2004 was low: 6 percent for a government worker, 8 percent for a public enterprise employee, and 12 percent for an employee o f a private firm. 3.26 The overall impression i s that the policy reforms over the past decade have had some impact. This i s evident in employment restructuring and hence the nature and extent o f individual transitions, consistent with a shift towards a more market-based economy. Yet the evidence on stagnant unemployment and large inactivity rates, as well as low-productivity and low-growth prospects in informal activities, suggest that the reform agenda remains unfinished. In the following sections we will focus on different segments o f the labour market to understand constraints faced to help inform a revitalized policy agenda. 3.27 These outcomes reflect individual preferences in the labour market, but also powerful influences o f government policy which have shaped both the size and structure o f the public sector as an employer, and the business climate for recruitment, pay and firing decisions and the prospects o f smaller operators. Structural conditions, geography and institutions also matter. The focus here i s on the key policy determinants o f the observed outcomes, to set up the discussion and conclusions inthe final chapter. TheImpactof PublicSector EmploymentandRemunerationPolicies 3.28 The policy motivation for our review o f public sector employment and pay policies in this chapter i s grounded in its relative size, and the effects o f public sector employment and pay policy on urban labour market trends. A somewhat distinct issue-the need for Government to be able to recruit and sustain the staff needed to deliver on its development agenda-is addressed in the concluding chapter. The basic questions explored here are whether the public sector is a good place for individuals to 17Indeed the contrast between the corresponding transition probabilities for the periods 1994-1997 and 1994-2004 largely reflects this increase. 27 work, and whether the wages and terms o f employment in the public sector push up costs inprivate sectors. 3.29 Accounting for almost one in five o f urban workers, and almost two-thirds o f those with formal sector jobs, the public sector has a major influence on urban labour markets. '*It is a major player in the skilled segment o f the market, accounting for about 68 percent o f employment among those with higher education. And with nearly three-fourths o f civil servants having at least completed general education, according to the LFS the skills profile o f the civil service i s much higher than that o f other employment categories. 3.30 The Ethiopian civil service has expanded significantly over the past decade as part o f the decentralizationpolicy. As a result o f this bigpush, the share o f the public sector intotal wage employment is highrelative to regional comparators. The expansion was concentrated at decentralized (regional and local) levels, so that by 2004, only 11 percent o f civil servants were Federalrather than regional employees. 3.3 1 Compared to other employees, civil servants enjoy relatively highremuneration. An index o f average public sector pay shows a steady rise since the devaluation o f the early 1990s, aside from temporary dips in 1995 and 2003. Real wages in 2005 were more than double the 1998 level, which in part reflects increases at the top end o f the wage scale. 3.32 Traditionally, the civil service pay structure in Ethiopia was compressed - reflecting the influence o f minimum pay for the unskilled, and constraints at the upper end. Looking at more recent trends we can see: 0 Internal pay relativities were stable in the latter part o f the 1990s, then became more compressed in the first half o f the present decade. A comparison between 1994/95-2001/02 and 2001/02-2004/05 shows that the salary relatives o f the professionals and administration staff and the sub-professionals and professionals were similar. During the recent period, the ratio o f average base salaries between manual and professionals declined from 1:7.2 to 1:5. 0 Internal wage compression has been circumvented to the extent that the expansion in the civil service was accompanied by an expansion in numbers employed in the higher brackets. The Remuneration and Conditions o f Service Commission l9 highlighted the reclassification o f positions as a response to lack o f wage competitiveness at higher levels. At the same time, frequent reorganizations, allowances, manipulation o f performance appraisal and transfers policies, have arguably distorted the civil service pay-scale. 18Unless otherwise specified, `Government' here refers jointly to service delivery workers and the civil service proper. The wage data provided by surveys does not allow disaggregation. It is noted that social service delivery institutions are allowed to implement different pay scales. l9Civil Service Remuneration Task Force Report (2005). 28 3.33 Wage relativities-mean and median-by skill category and type o f employer (Figure 10 anddetailed inTable 9) show that: For most key categories o f skills-with the exception o f professionals and managers-average Government pay is significantly higher thanwhat i s offered bythe private sector. Parastatal wages are more favourable than those offered by the private sector - for technicians, clerks and also the unskilled, which may inturnput private firms at a disadvantage inseeking to attract needed employees. Neither the public nor private sectors offer better pay thanNGOs, 2o although the overall impact o fNGO conditions on the market i s unlikely to be large, given the relatively very small number of jobs in the sector (only 2 percent o f total employment in2003). Figure 10: MedianWage by EducationalLevel and Type of Employer, 2005 ;ource; Own calculations, LFS 2005. 2oThe "NGO" category includes international organizations. 29 Employeeprivate Employee h'GQ Employee domestic 3.34 Yct the a t ~ r a ~ ~ ~ ~of~ eann eemployer and job quality is cleaasly multi- ~ s d~n~ensional~ goes beyond zvages, and includes hours worked, and job a~~r~bu~es, This ~ n c ~ ~uc~c ~ ~~ aghealth~and safery, n o n - z ~ ab~~efits,employer's adherence to t j o i ~ ~ ~ ~ core ~ a ~ ~s~~dards, ~ ~ ~ ~ oaf~I~atianto /~ ~t oa socialusecurity~schemes, t l r ~ ~ ~ r ~ ~~ e ~r ~ & o ~ ~di ~ ~ t lopient~ and ea 3 l ~ ~ ~ s stability and career d e ~ e l ~ p ~ ~ ~ e ~ ~ 3.35 in thc case o f civil servants, both pecuniary and n o n - ~ e c ~ n ~nonr yw~ Etheirjobs. Foremost tl ecurity c o ~ e ~ fa~civil a e oyed it7 the military t by the cniplaye enefits for invali are ~ r o ~ i ~ eThe.r e ~ ~ rage i~ fixed ~ 60tyears, with some e ~ c e ~ and~the ~ ~ s ~ d ~ s i ~ at ~ ~ o basic pe~sionis 30 percentof the salary, and cannot exceed 70percent. 3.36 The key ~ i o n - ~ c c ~benefitr yfor civil servants is job s~curity. There are ~ ~ a I e ~ ~ s l aconstraints on d ~ ~ ~decisions,awhich are larger inthe case of dismissal o f t ~ ~ e ~ s s ~ a public sector e~np~oyee.Prior to dismissal of a civil servant on the ground of See X1,Q studies by Ythcrtew Tsegayek report atid Chapter 3 of the psptr by Tassew ~ ~ ~ l ~ ~ ~ i ~ ~ n a FantuGuta, ~ ~ Fexede (2005). ~ ~ e ~ e 30 unsatisfactory performance, training and job transfers should be offered to improve a civil servant's performance. Moreover lawful dismissal requires a notice period o f between one to three months, depending on the number o f years o f service. Remedies against unlawful contract termination are the reinstatement and the payment o f due wages, or compensation inaddition to severance payment. 3.37 The attractiveness and security o f federal civil service jobs staff i s indicated by overall turnover rates. At an annual average o f about 4 percent the rate o f separation from the civil service does not appear highby international standards (Figure 11). Figure11:AnnualRateof SeparationfromEthiopianFederalCivilService 12% 10% 8% 6% 4% 2% 0% 1995 1996 1997 1998 1999 2000 2001 2002 200: Source: EthiopiaCivil Service Commission. 3.38 The basic upshot appears to be that for the less skilled, the public sector offers attractive opportunities. For illiterate job seekers, for example, who account for almost one in five o f those looking for work, a government job i s a better option than a parastatal, private or informal sector job, due to better enforcement o f minimum wages andnon-wagebenefits. 3.39 Some key policy implications, which are explored further in the final chapter, arise because the bulk o f the civil service enjoys better conditions thanwould be offered by the formal private sector, and certainly by informal sector activity. This inturn can have adverse implications for private sector competitiveness. This should be borne in mindwhen attemptingto redress the braindrain issue at the top end (see Chapter 6). 3.40 Turningmore specifically to the labour market aspects o fparastatals, we observe the following: First, the relative size o f parastatal employment, while not large overall-about 2 percent o f total employment-is a significant share o f manufacturing employment. Interms o fprofiles emerging from the ICA dataset, it is also notable that the larger firms in the sample are mostly public sector and it is these firms that also have older machinery and equipment on average, and hence report low productivity. 31 0 Second, with the privatization process, the significance o fparastatals inthe economy andas employers has diminishedover time. 0 Third, the structure ofwages is relatively compressed. A basic picture can be drawn from the ICA dataset, based on firm data from 2001. When the rates are examined by type o f ownership, we observe lower wage inequality for unskilled production workers in state firms, as well as overall less wage inequality in state owned enterprises. 3.41 Relative to private enterprises, parastatals report higher wages and more job security. Comparisons o f private and public sector enterprises22show that labour is on average twice as productive in the former, which is consistent with private firms using more capital per worker. Public enterprises tend to be larger, employing on average about six times as many workers as private firms, which may also reflect overstaffing. When restricting the comparison to firms that are similar in output, public enterprises employ about 25 percent more workers than do the private firms. 3.42 In this context, recent trends are welcome. As we saw from the panel data, transition rates for parastatal workers were the highest across any type o f employer. However as noted recently (World Bank 2007, vol. 11, Chapter 3), the privatization agenda ranout o f steam after 2002 and is inthe process o fbeing revived. 3.43 Inconclusion, the Ethiopian labourmarket suffers notjust from segmentationbut also from an inappropriate structure. The Government sector remains too large and this has negative consequences for the growth o f the private sector, not least in terms o f its composition between formal and informal components. Constraints on Jobs Growth in the Private Sector 3.44 The formal private sector has not yet made a major contribution to job creation in Ethiopia. Various barriershave affected its dynamism as a source o fjobs, some o fwhich can be traced to inappropriate policies. We focus here on those aspects o f the investment climate which are more directly amenable to government policy influences-firstly, those instruments which affect labour demand, namely employment protection and payroll taxes and other costs associated with hiring labour; and second, the business environment more generally, including governance issues and problems associated with what appears to be public enterprises' preferential access to factor markets and government services. 3.45 Before proceeding however, we review the determinants o f firm performance in terms o f employment and wage decisions. We utilise the I C A to examine a three year period (1998/99 - 2000/01), which reveals that while the sample experienced small improvements inreal sales, median employment change was zero (Table 10). 22 Given that public enterprises are larger than private, we focuson sub-samples of comparable firms based on sue, output and profits. 32 Table 10: ChangesinKey Indicatorsat P h iLevel, 1 ~ 9-82 ~~ ~~ ~ / 1 Real Sates 353 0.91 0.14 6.14 -0.93 108.65 Avcragc Real Wage 147 -0.22 -0.I 8 214.17 -0.91 0.23" `X"Qta1~~~~~~r~~~~ 367 0.26 0*00 f .04 -0.92 15 ~c~~~~~~~ 327 0.21 0.00 1,05 -0.89 15.00 sion analysis of the JCA data shou.s that there are, as expected, clear links t. Our analysis showed that between 1 9 ~ $ ~ ~ 9an- ~ ~ ~ ~ / ~ 1 i ~ ~ r e a sinereal sales o f 1 percent was tcd with an t change af O,O9 percent, This result is robust to the int on o f indust true for public as welt. as private firms. 3.47 T u ~ i now to examine pote~tt~aln ~ u ~ n one s n ~ ~ c p r o ~ ~ a b ~ andy labour l i t ~~ d e ~ ~ a nwe ~begin with taxation. In other trans~t~on~ c ~ n o ~ i the d e s ,additiona1 costs which create a wedge between take-home pay and labour costs have been found to either discour~gerecntitment andJor d a ~ ~ wages. It is also possible that the a d ~ ~ n i s t r a t ~ o n e n o f taxes is ~ ~ dbu~ r~dye ~ ~ (time eo n ~ taken and so on). If these factors are i ~ p o ~ina ~ t ~ t h ~ then~a ~review~ and possibly reform of key taxes with an eye to e n a b l ~ ~labour o a ~ g ~ e ~ ~ axidnprivate sector gowth more generalIy, would be ~ a ~ a ~ The ~ d , t ~ ae ~d~~~ l a b l e e~~jdence i s reviewed below. 3.48 A c c o r d ~ nto~recertt 3MF reviews, there do not appear to be major ~ h o ~ c o n i i ~ ~ s iopia% tax system, Rates o f onal income tax (which are withheld by the e ~ ~ p ~are~not cxc o e r ~ ve by ~ ~ t c ~ astandards~(withl a top niarginal rate o f 35 t ~ ~ a percent kicking in at 5,001 and a tax free threshold up to Br 150). Likc ' corporatc tax t"ias not found to be to tional s ~ ~ d a r (top ~ i a r rate~ ~ a ~ d s ~ of 35 percent), a ~ ~ ~limited~c ~ ~ i i o eoufd be an issue, especially for riew firms. There are n 1 taxes. A tax refom ~ r o g r a mhas been ~ ~ d e for ~ a y ~ sortie time. In 2002 were taken to broaden the tax base and j ~ p r ~ ~ e cieney, including new l e ~ i s ~ a ~onn~ ~ a ~ u ~ - aanddin~ d j o d tification number, 3.49 One part of the tax code, the ~ ~ e s ~ ~tax,pist especially re~evan~ sniaitI ~ ~ v e for businesses. The ~ r e s u n i ~tax~ is~paid by smalf business traders and ~ r ~ f e s s i o n a ~ s ~ j e whose ~ i ~ t ~ ~uo avise~less than Br 100.~~0 r (about US$X 1~ ~These taxpayers} are. ~ ~ not r ~ ~ ~to ~ j r ~ account~ngrecords and the income o f the t ~ p a y e r sis estimated niaintain d ~`presu~ed'')by Tax Aut~iari~~f. The pres~mptj~~axpa~'ersare d ~ ~ ~ dintodttvo ~ e e categories: business and r r a ~ ~ andp ao~~t o ~ i services. We focus herc on the former. ~ e y Under the business service sector c a t e ~ o ~the~p r e s ~ i e dAverage Annual ~ r o ~ t a b ~ l i t y y Rate had 22 rates ranging up to 70 percent ~ e p e n d ~ nupon the nature o f bus~ne~s, g Xn 33 July 2005, presumptive taxation was reformed, and inparticular the maximum profitable rate was reduced from 70 to 10percent, andthe number o frates was reducedto 4. 3.50 On the other hand, while a 35 percent corporate tax rate is not high relative to international norms, it i s significantly above the regional average o f 30 percent. And looking at the factors reported as among the three "most serious" obstacles to expansion by firms in the ICA sample (Table ll), rates are cited the most frequently (74 tax percent) and tax administration the second most frequently (60 percent). However firms (and people) everywhere complain about taxation, so it is important to put these findings inperspective by comparing the results to those of other countries that have conducted an ICA. The comparison suggests that taxes are more o f an issue inEthiopia: the share o f firms considering tax rates a "major" or "very severe" obstacle i s 45 percent in Madagascar; 48 percent in Uganda; 50 percent in Senegal; and 68 percent in Kenya, relative to 72 percent in Ethiopia. Also, Ethiopia ranks third worst out o f 62 countries (behind Tanzania and China) on the number o f days firms report that they spend in meetings with tax officials. Table 11: Share of FirmsReportingConstraintas Severeor very Severe Tax rates Tax administration Access to land 43 Economic policy uncertainty 39 I CorruDtion I 39 I Cost of financing (e.g. interest rates) 38 - administration 37 Macroeconomic instability 35 I Crime. theft and disorder I 32 I Telecommunications 29 Skills andeducation of workers 18 Transportation 15 Anti-competitive practices 10 Business licensing and registration 9 Survey). 3.51 The I C A covers medium and large enterprises. For small and informal firms, taxes may also be an issue. While there i s a tax free threshold on business and profits tax o f 1,800 (annually) for unincorporatedbusinesses, there are also VAT and turnover taxes with the latter applying to all goods and services sold by taxpayers who are not obliged to register for VAT. 34 3.52 A recent survey (May 2006) found that avoiding taxation is often a reason to avoid formal status. A small but significant percentage o f firms in the ALMS (12 percent) agreed with the statement "I don't want my business to grow too muchbecause then Iwill have to pay taxes." This suggests that avoiding taxation could indeed play a role inthe decision o f some firms not to formalize. 3.53 The concluding chapter picks up the implications o f this mixed picture on taxation as an obstacle to growth and employment, and suggests possible next steps in terms o f analysis and follow-up. The institutionalframeworkfor the labour market 3.54 Ethiopia's labour law framework-outlined for the private sector by Proclamation No. 377/2003-ostensibly does provide a series o f protections for workers. These include procedures and payments inthe event o f dismissal (e.g. what i s lawful, and severance and notice requirements). 3.55 Social partners in Ethiopia are weak in terms both o f membership and in experience (ILO 2005). For example, the Confederation o f Ethiopian Trade Unions (CETU) membership o f 181,647 workers (and445 basic unions) represents less than one percent o f the total labour force. Unionisation o f the Ethiopian labour force has been constrained bythe largely agricultural andinformal nature ofthe economy. 3.56 Ingeneral, information from the 2002 ICA suggests that labour regulations and labour relations are not seen by firms as significant impediments to doing business-it seems this i s largely because these provisions are not generally enforced outside o f the public sector. We do nonetheless find that labour laws and regulations are important in some sectors, and matter most: 0 to employment decisions infood, beverages, textiles, and garments, where 40-75 percent o f firms cite laws and regulations regarding firing as reason for labour hoarding; and 0 in terms of union pressure, which appears as an important determinant of employment level in some industries, particularly in food and beverages (37-41 percent o f respondents). 3.57 Fear o f social sanctions is an important employment determinant for about half o f the firms in most industries. While labour regulations are not acknowledged as a major constraint, this factor may well be more important than it might initially appear. For example, the I C A data show that there is labour hoarding inmost firms. The bulk report o f firms report over-staffing in the range 1-10percent (viz employment between 1-10 percent greater than desired) but over 20 percent o f firms reported over-staffing as being greater than 10percent. 35 The impactof the businessclimate 3.58 We turn now to the broader business climate agenda, as this relates to jobs growth. The most obvious point at the outset is that Ethiopian enterprises and workers are not internationally competitive. Low productivity makes Ethiopian firms uncompetitive against Asia, for example. An international comparison with other African countries as well as India and China, shows value added per worker inEthiopia to compare unfavourably (Figure 12). 3.59 The question, o f course, is why this should be the case. Since this is, in a very fundamental sense, the core development question facing Ethiopia that has eluded policy makers and advisers for decades, the answer i s well beyond the scope o f this report. At the same time it is possible to map at least some of the causal policy factors which appear to contribute to the observed disappointing outcomes. 3.60 A focus group exercise in2004 reconfirmed the findings o fthe earlier ICA. The importance o f accessing land for buildingand the overall tax regime were confirmed as the two issues deemed as by far the most severe problems (Figure 13). The focus group results are drawn from discussions with Integrated Value Chain Analysis Groups (IVCAGs) covering five sectors: leather, textiles, floriculture, tourism and construction. There were 12-15 members in each IVCAG, drawn from sectoral associations, with priority given to the leading private enterprises. Hence the results are not representative either o f the sectoral or size distribution o f private sector firms inEthiopia, and cannot be regarded as definitive-however the findings are at least illustrative, including o f perceptions of change over time. Figure12: Comparisons of Value Added per worker (US$, byfirm size) loo00 I 9000-- S I l l a l l so00-- 7000-- 6000-- 5000-- large 4000-- large t&% large '*ge 3000-- S m a l l large small r % 2000-- large sman small 1000-- large small .small Source; EifertandRamachandran (2004). 36 13: Perceived Con ...l.. ............ .......................................................... ............................. S a v s r ~ t yo t p r o h i a m : O n o g t r o b ~ e m $ 0 v e r y s e v e r e p r a t r ~ e m . ' "I_-_x- I ..................... I Sotcrce: WB focus groups, 2005. 3.61 The focus groups also ness owners. In the past three years, sonx to the business (see Table 12). Table f 2: Private Sectar FacusGroup Results: Pcrccptiunsof Progress Clear improvement Mkedprogress Ittsirfficiettt prugress 3.62 More gcneral c ~ ~ s t ~ awhichsaffect firms7 growth are related to the need to ~ n t s ~ p p o ~ o c ~ ~~ a~r ~~structures,~and~to~avoid ~ ~ o ~ arid~rent seeking. The e ti t o o ~ y e ~ ~ a b ~ of the~Ci o~~~~~~iet~t i to~ ~o ~~ is an j~~ s~ j~ oistep~in~athe~right direction, ~ s ~ i o~ ~ t and ~ i ~ n ~ f icases t c a ~ are now in the ~ i p ~ l though~this remains to be e ~ f e ~ ~ i ~ e ~ ~ ~ n e ~ ~ e r a t i o ~ a Thusza~~~~ the serious harriers a~ihichremain, the ~ ~ ~ ~ d . ~are: ~ h l 37 Regulatory and administrative issues, such as difficulties in enforcing contracts (as many as 420 days and 15 procedures are required to enforce a contract) and property registration (Ethiopia ranks among ten worst countries worldwide with 15 procedures to register land); and Progress in staffing the Competition Commission seem to have stalled (no private or civil society Commissioners have been named), and capacity building activities o fthe Secretariat are behind schedule. 3.63 This picture o f the constraints facing the formal private sector points to the need to build on the reforms to date. The playing field remains uneven with respect to the generally larger and often dominant public sector enterprises, and significant problems ininput, output and factor markets are hampering dynamism even of sectors which are experiencing robust growth. A policy agenda to revitalize the private sector i s set out in the final chapter. The Functioningof the InformalSector 3.64 The informal sector is a tremendously important component o f urban labour markets in Ethiopia - not only because o f its sheer size relative to alternative opportunities, but also because o f its distinct characteristics and its importance as a source o f livelihood for the poor. Its significance from individual perspectives i s underlined by its evident persistence as a labour market state. 3.65 The Ethiopian policy regime since 1992 has been more hospitable to entrepreneurship, including for small and informal businesses. Whereas Derg policies were "openly aimed at curtailing (ifnot eliminating) the private sector," there has been liberalization alongside some proactive measures. The latter include the 2002 National Micro and Small Enterprise Development Strategy, and the waiving o f license requirements for firms with capital under Br 5,000. Still some observers have noted that the legacy o fprevious policies (such as the one-man, one-license rule), ifnot the policies themselves, likely persists to some degree (Ageba and Amha 2004, p. 4). 3.66 The informal sector accounts for the majority o f employment in Ethiopia- according to the 2005 LFS, it represented 71 percent of urban employment overall and 81 percent o f youth employment. Several sectors are almost exclusively informal (at least as measured by the number o f employees in the sector). These include domestic work, wholesale and retail trade, hotels and restaurants, and primary production (Table 13). Overall, manufacturing accounts for about 45 percent and tradehotelsh-estaurants for about 38 percent o f informal firms. Some activities appear to be strongly gendered: typically female sectors include food manufactures, services (hotel and restaurants) and trade. Men are far more likely than women to work in real estate, transport, and construction. 38 2005 Male Female Total I I etail Trade, Repair of Vchrcles, Personal& 3.67 I ~ f o r m asector f i m s share many ~ ~ o features. ~ Most (60~ p ~ or ~ of n~ ~ ~ ~ ~ e onal profile is law but ~ n c ~ e a s ~ ~ ~ 14): the ~ r ~ ~ofo ~ j soin~total infamia1e m ~ l o fell~from~50~percent in ~ e I999 to 37 percent ill ective figures for women being even more c, from 64 to 51 3.68 ~ ~ ~ ises tend to employ ~only the operator, or one or two ~ d d j t ~ o ~ 2 ~ 1 o ~ a ~ rcent of whom are p e ~ ~ ~employees, twith the rcst unpaid n e ~ ~ cnt) arc sole than half are ~ o ~ e - ~ ~I ~s feo~~.firms sell nearly a l exclusively to the ~ o ~ emarket,~ for the most part directly to c o ~ ~who know the ~ r ~ s t ~ ~ ~ ~ n operator or hear about the fim via word of mouth ~A~~~ 2006). 39 Figure14: TrendsintheEducationLevel ofInformalSector Workers I 6 0 1 50 -*#2 0 40 e 30 W 2003 Informal Sector 5 n 20 0 2005 LFS 0 10 0 Note: LFS includes all informal sector workers; ISS includes only operators. 3.69 Ethiopia's informal sector is predominantly employment o f last resort, rather than a preferred breeding ground for entrepreneurs. There are several key pieces o f supporting evidence for this claim: 0 when asked why they chose their current activity, more than three-quarters o f respondents to the Informal Sector Survey (2003) reported an "involuntary" rationale-that they had no alternative or because a small investment was needed (withwomen more likely thanmento beinvoluntary). 0 evidence from the Informal Sector Survey and another MSE Survey (Ageba and Amha 2004) indicate that older firms are about the same size or even smaller than newer firms, based on average saleshumber o f employees, meaning that most firms do not grow substantially over time. In fact, according to the MSE Survey, "the additional jobs created by the MSEs that added workers [between 2000-20031 were almost fully offset by reduction o f workers by others, such that net job creation by the sample MSEs over the past three years was almost zero" (Ageba andAmha 2004, p. 18). cluster analysis to assess the degree o f correlation among characteristics associated with more entrepreneurial firms (e.g. keeping some accounts, having a permanent location, higher sales, etc.) showed that the "upper" tier o f more entrepreneurial firms comprised only about 5-20 percent o f firms, with level o f sales beingthe main driver. This contrasts strongly with evidence from Mexico, where there is a much larger dynamic group (Maloney 2004). Thus, although Ethiopia's informal sector i s certainly heterogeneous, as inother countries-and manysuccess stories do exist-it seems that for the vast majority it provides a means o f survival rather than upward mobility. 3.70 The share o f the informal sector with more dynamic activities and prospects, has some specific characteristics: 40 0 higher proportion o fmale operators; e greater skills (both interms o f schooling andvocational training); e greater likelihood o f keeping some accounts, selling on credit, working outside the home, and operating 12months o fthe year; 0 greater likelihood o f choosing the place o f work for a business reason (e.g. close to customers or competitors), rather than due to childcare or affordability considerations; and 0 larger proportion o f (tradekotelslrestaurants) activities firms and a smaller proportion o fmanufacturers, andrelatively more transport firms. 3.71 While having a family based set o f firms i s not a problem per se, as the experience o f Italy among others illustrates, it i s a concern that no significant part o f the informal economy is populated by individuals/firms that are using the lack o f regulation and tax as a form o f experimentation ground. Evidence from other countries suggests that absence o f taxation and the like can helpfully serve as a quasi-subsidy to new start- ups. Whythis is not the case inEthiopia is not clear. Lack o feffective demand mayhave been an issue but the economy was growing strongly since the early 2000s, so the explanation may partly be the harassment and other obstacles associated with government, as described below. Some features o f the sector associated with its low productivity and lack o f dynamism-such as the personal nature o f contractual relationships and the importance o f social networks-are responses to market failures, particularly in factor markets and inputs. These mechanisms can help firms survive at least at some minimal level, but are inherently limitinginterms o f growth potential. 3.72 Informal activities take place in a very constrained environment. For initial capital, informal sector entrepreneurs rely almost exclusively on their own savings, and on help from friends and relatives, which i s often a gift rather than a loan. Very few enterprises access startup capital through official channels. Lack o f market or demand i s the other major constraint that informal operators report-their concerns about domestic price competition likely reflect the low barriers to entry. About a third o f operators are price takers, inthat they charge the going rate for their goods/services, which supports the conventional wisdom of a low degree o f product differentiation (ALMS 2006). EBDSN (2006) notes market saturation by similar products, andthat gains by firms that attempt to innovate are quickly erased as others imitate their success. Female-owned businesses are significantly more likely to cite "lack o f market" as a major issue, as are the less educated andthose who chose their activity because o f low barriers to entry. 3.73 While entry barriers appear to be low and obviously some would-be entrepreneurs overcome these constraints, others remain stuck in unemployment. According to the 2003 UBEUS, by far the most important barrier faced by the unemployed wanting to establish a business was shortage o f finance (roughly 60 percent o f respondents cited this reason). Next in importance was lack or problems o f working place/land (over 15 percent). No other single problem-including lack o f skills-comes close inimportance 41 to these two factors. These results are also consistent with messages emerging from other survey data on constraints in access to finance. More than two-thirds o f unemployed ALMS respondents said that they did not attempt to start a business because o f lack o f capital followed bythose who cited educatiodtraining. 3.74 There may be a negative signaling effect in the labour market associated with working in the informal sector. If this diminishes fbture formal jobs prospects it would limit resort to the informal sector to those who feel better prospects to be remote. A non- trivial share (13 percent) o f unemployed respondents-and a higher share o f the better educated indicated that they would only be willing to take up wage employment in the public or private sector, rather than self employment (ALMS 2006). 3.75 Several investment climate issues have been reported by informal operators as obstacles to expansion, which presumably also limit the transition to formality. Inthe Informal Sector Survey, larger operators and those working in the hotelhestaurant business were more likely to report "harassment by government bodies" and the regulatory environment as major problems. Moreover, nearly 40 percent o f ALMS respondents agreed or strongly agreed with the statement that "government and local administration officials make it difficult for me to conduct my business activities." Qualitative research and anecdotal evidence also suggest that some government policies, or the interpretation thereof by local officials, can dampen informal activity. For example, according to one study, "it i s commonplace to witness police harassment o f informal sector operators, including women" (Zewde and Associates 2002). On a related note, lack o f policy predictability appears to be a major issue. Inthe MSE Survey, about a third o f microenterprises reported that they have to deal with `unexpected changes in rules, laws or policies which materially affect their enterprise,' and about 40 percent "do not believe that the Government adheres to its announced policies and rules" (Ageba and Amhap. 60-1). 3.76 Problems in access to landand working premises were reported to be an obstacle to both start-up and expansion by many firms, especially in Addis Ababa (Informal Sector Survey). Evidence from other survey data provides ample confirmation o f these constraints. Recent migrants to a given area are disproportionately likely to report a problem with access to premises, suggesting that their needs may require special policy attention. 3.77 Respondents to the 2003 MSE Survey were most likely to cite "high taxes" as the biggest regulatory obstacle to expansion, while inefficiedarbitrary tax administration was the third most common complaint. Like larger firms, these enterprises do not consider labour regulations per se to be a significant problem, but they do mention problems related to limited infrastructure, especially lack o f telephone, electricity, and water access. Ingeneral, the concerns o f MSEs reflect those o f large firms cited earlier, though with differences inprioritization. Only about a thirdhad noticed improvements in the policy environment resulting from the reforms o fthe 1990s(Ageba and Amha 2004). 3.78 A major question is why informal firms do not formalize and whether this decision affects their performance. Individuals and firms weigh the costs o f becoming 42 formal against the benefits they expect to enjoy, including access to public goods such as contract enforcement. To the extent that the associated costs are high, or the benefits limited (e.g. small firms in Ethiopia rarely resort to the formal judicial system even if they are formal), operators would be less likely to formalize. In Ethiopia there are two key steps to being an officially recognized entity: registering a firm with the authorities, and obtaining a license. All firms must do the former (though inpractice it may be that manyinformal firms do not), and firms with at least Birr 5,000 inpaid-up capital must do the latter as well (see Box 4). The waiving of the license requirement for very small firms means that this is now irrelevant for many informal sector operators. Box4: CostsofFormalization Under the official GOE definition, informal firms are those with fewer than 10employees, that do not keep a full book o f accounts, and that do not have a license. While the number of employees i s presumably driven by factors other than the choice to formalize, the second two criteria clearly involve tradeoffs incosts vs. benefits. Keeping a book of accounts enables compliance with the tax code; Table 14 shows marginal tax rates for business income above the exempt threshold o f Br 1,800. According to ECC & EBDSN (2004), firms generally perceive the exemptionthreshold to be too low. Many MSEs also believe that the official assessment of tax liability tends to be subjective, so that many firms are over- taxed, particularly since small f m s frequently lack a complete understanding of tax laws. On a more positive note, firms do generally believe that the recent tax reforms have led to improvements inthe process. ITable 14: MarginalTax Ratefor PresumptiveIncomeTax on Business Source: IMF (2006). Inorder to register andlicense abusiness, the World Bank DoingBusinessDatabasereports that one must follow 7 procedures (which take about 32 days), and spend the equivalent o f 65 percent o f annual per capita income. Remaining formal also entails significant costs, since licenses must be renewedperiodically. However, businesseswith capital of less than Br 5,000, though required to register, are not obliged to obtain a license. These procedures represent a major improvement; in fact, Ethiopia was one of the "top ten reformers in startup procedures" inthe world in2004, according to the DoingBusiness Indicators. 3.79 Insum, the incentivesfor verysmall firms to formalize areweak. Three-fourths o f unlicensed firm operators surveyed by the ALMS said that they did not have a license because they "don't need a license'' or it is "easier to operate without a license." Moreover taxes serve as a major deterrent to formalization. The 2002 Investment Climate Assessment found that the tax burden and highrents were the main reasons cited bythe 2,200 businessesthat cancelled their licenses in2001/02. 43 3.80 Yet there are benefits to formalization, especially as firms graduate from micro into small enterprises. Although about 36 percent o f micro-enterprises in the MSE survey saw no benefit to registering/obtaining a license (compared to 24 percent o f small enterprises), nearly half said that they "would not have to hide frodgive bribe to government officials" (Ageba andAmha 2004 p. 64). Other potential benefits recognized byoperators includebeing able to apply for credit andworking premises. 3.81 Our analysis o f the informal sector implies significant challenges for Ethiopian policy makers, particularly as the effects o f repressive policies can be long lasting (Bigsten and Makonnen, 1999). The policy agenda on this front is laid out more explicitly in the concluding chapter. Another dimension that needs to be understood i s the relation between the informal sector and the unemployed. Which are the barriers preventing the unemployed from starting up an informal sector activity or taking up informaljobs while continuing their search? These issues are explored further inthe next chapter. 44 4. UNEMPLOYMENT 4.1 The high level and duration o f urban unemployment in Ethiopia raises a number o f questions about measurement, drivers andpolicy implications. The highincidence and depth o f urbanpoverty, and lack o f unemployment benefits, would a priori suggest that unemployment is simply not an option for most Ethiopians. This would seem to hold even if available employment opportunities offer only subsistence earnings, since relatively few families can afford to subsidize their unemployed children for extended periods. Perhaps, then, the reported extent o funemployment is a measurement issue. 4.2 This chapter attempts to address these issues by using the best available quantitative data, complemented inparticular by a very recent follow-up survey to create a profile o f the unemployed. We are especially interested in identifyingthe factors that explain the status and persistence o f unemployment, and how the unemployed survive. We also put Ethiopia in a comparative perspective, and conclude by drawing out some policy messages for the final chapter. Levels and Trends 4.3 Theprevalence andprotracteddurationo funemployment is striking. It appears to be structural rather than cyclical innature, given its persistently high level despite recent robust economic growth-although thismaynow bechanging. Unemployment remained inthe double digits from the 1999to the 2005 LFS, whether abroador narrow definition is used (Box 5). However, while adult female unemployment rates were essentially stagnant, youth and male adult unemployment fell. Open unemployment in Ethiopia is several percentage points higher than the African average for youth, and a couple o f points higher than the adult average as well (Figure 15). 4.4 Most African countries have relatively low unemployment and large informal sectors that absorb unskilled labour (Kingdon, Sandefur, and Teal 2004). In Ghana, Uganda, and Tanzania, the formal-informal wage differential for unskilled workers is fairly low and suggests that "informal wages have fallen to clear the market," such that low wages and underemployment rather than open unemployment are the policy concerns (p. 22). In Ethiopia however, the high public sector wage premium (for parastatals as well as the civil service) probably drives up formal private wages as well. This limits labour demand overall, meaning that the informal sector i s the default option for a large segment o f the population. Yet many low productivity informal sector activities are saturated such that expected earnings are low enough that many remain unemployed or become discouraged workers. 45 Figure15: RegionalAverages,Adult vs. YouthUnemployment 60 M 40 0 lw U 5 10 10 0Y w a d CUrOpL Emlala Urnma H w a d %Ma cnbvnrrr CUrOpL Emlala plcme Urnma b m SO&&to &IO 5UD I M mARIU Source: World Bank (2006b). Box 5: MeasuringUnemploymentRatesandTrendsUsingDifferentDefinitions According to international conventions, two definitions o f unemployment are possible. They vary in the way they classify those who are jobless and available for work, but not taking active steps to find ajob. Their appropriate classification depends on whether these individualsprefer not to work, inwhich case they would be better classified as inactive, or rather whether they can be considered "discouraged workers," i.e. workers who would like to have ajob but believe that an active job search would be hitless. International practice generally followed the first categorization and adopts a "narrow" definition, which considers those jobless who do not actively search out of the labour force (i.e. inactive). The CSA in contrast adopts the second approach and uses a "broad" definition which treats "passive unemployed" workers as unemployed. While our study uses the more common international definition to allow for comparability, insightsmightbe lost by excluding discouraged workers. FollowingKingdonand Knight (2006) we estimate the change inunemployment between the two labour market surveys using the two definitions. Including the non-searching jobless means a higher rate of unemployment overall (they are included in both the numerator and the denominator when calculating the unemployment rate, but excluded fiom both in the narrow definition). The unemployment rate was 14 percent in 2005 using the narrow definition, and 21 percent using the broad definition. Figure 16 below, decomposes the total population into four categories based on labour market status: inactive; employed; passive unemployed; and active unemployed. The active unemployed group, on the far right o f the graph, representsthe narrow definition o f unemployment; adding the passive unemployedgroup shows the total unemployedpopulationper the broad definition. It i s clear fiom the graph that the key change between 1999 and 2005 was a decrease inthe share o f passively unemployed, and a corresponding increase in inactivity (much of which can be attributedto male youth). 46 Figure16: LabourMarket Status, Age 15+ 1 1 - 1 HEmployed 0Passiveunemployd I0 Active unemployed 1999 I 20% 4wh Wh aooh 100% Share of PopulationAge 15+ 4.5 The lack o fjobs makes it easier to become discouraged about the prospects o f an active search. Looking at South Africa, Kingdon and Knight (2006) propose three tests to assess whether the profile o f the "searching" unemployed differs significantly from that o f the "non-searching" unemployed. The results indicate that the non-searching unemployed are no better o f f (and are financially worse off) than the searching unemployed, suggesting inter alia that their decision not to search may reflect a lack o f resources (given low expected returns to searching) rather than lack o f motivation. 4.6 Long-term unemployment i s common, though average duration has shortened considerably. The urban panel supports the finding that it i s becoming more likely that the unemployed in a given year will have found employment by the time o f the subsequent survey wave. As shown inTable 15, a massive 93 percent o f the unemployed in 1994 were also unemployed in 1997; However, the likelihood of transiting from unemployment to inactivity4.e. dropping out o f the labour market-increased from essentially zero between 1994 and 1997 to 23 percent between the final two waves. Still, the total likelihood o f remaining inunemployment or becomin inactive has fallen, and a clear picture o f greater mobility out o funemployment emerges.5 3 Table 15: TransitionProbabilitiesout ofUnemployment Self- Gov't Public Formal Other InitialStatus Employ Worker Enterprise Private Private Unemployed Inactive Status in 2000 Unemployed in 1997 I7.78 I5.56 12 I10.22 I11.56 I51.33 I11.56 in 2004 Unemployed in2000 I 6.82 I8.47 IStatus 2.89 I12.6 I10.54 I35.74 I 22.93 Source: See Vol.11, Chapter 5. . 23While this suggests that unemployment duration is typically long, the survey waves are snapshots in time; it i s possible that a given respondent had one or more jobs between survey waves, but happenedto be unemployed at the time o f the survey. W e attempt to minimize this complication by comparing the waves against each other, rather than comparing unequal periods intime. 47 4.7 The ~ ~ e d~~ ~a ~ r aot ~ o~n n ~ ~ fcl1~ ~ ~b ~~tfromt1999ato 2W5, ~~ s ~p ~~ ~o ~ from n ~ ~ ~ more than 1.5 years to less than 1 yeas. Ac~ord~ngthe LFS, the share of u n ~ ~ ~ ~ ? l o ~ ? e ~ to e ~ ~ e ~ e ~a~spclli nofgsix monthsor less increasedby 65 percent, while the share having c spells e~cc%cdingyears fdl by about half(Table 16). The patte~ii s e ~ s e ~ i ~thea samey 3 i ~ ~ or mcn and omen^ though u n e ~ ~ p ~ o ~tendsn tto last longer for won~en-a~out22 i e percent of u~e~np~oyed women in 2005 had been out of work for more than 2 yeas, co~paredto fb percenr of men. Mo~ever,the gender d~fferencen a ~ o w over the six e ~ year period. Durationof tlne ayment Groups (percent) All Male Femafe 1999 I 2005 1999 1 2005 1 1999 1 2005 4.8 The private sector absorbs most o f ttiose ~ ~ ~ e ~ pwho ~ ae~di a gtoefind job l ~ y o ~ ~ ~ ~ We find~from theepanel .that the in u ~ ~ ~ s ~ n ~ p o ~avenue to a n t joining "otker ~ ~ ~ F a ~ e Thc formal privatc sector ~bsorbsa 7 percentjoin public 4.9 New entrants to the labour force c o ~ ~ the~main~ sourc%ceof the newly s ~ ~ ~ ~ ~ e m ~ follo-tlrede d job losses in the private sector. ~ o ~ ~ ~ c c o ~ ~ ~ ~ ~ ~17 apercent~of the inacti17c pop~ ion in 2000 had joined the u l ~ ~ ~ s 2004. The ~tobab~l~ty of losing ajob in formal private sector was 14 informal sector. Job security was higher inthe public sector, tvith only a 7- ihoodofjob loss. 4.10 The unemplo~entrate varies ~ u b s t ~ ~ tby~ ~ l y i ~bse~ablec~aracter~s~~c~. ~ n e ~ p l oa~~e o~youth n~ ~t is toughly double that of ad~lt~, with y~ungttiomen aged 20-24 having the single highest ~ n c ~ ~ p rateoin~2005catn26~pcrecnt. In relative ~ ~ 48 terms, however, the gender unemployment gap is smaller among youth. Male youth were 76 percent as likely to be unemployed as their female counterparts in 2005, while adult men were 60 percent as likely as adult women to be unemployed. Those with significant formal but less thanhigher education are most likely to be unemployed, and geographical factors appear important. These factors are explored below inturn. 4.11 `The unemployment rate by age group for 2005 shows a spike in the 20-24 year age group (Figure 17). It declines fairly steadily thereafter, with the female rate dipping below the male rate for older groups. The crossover o f the male and female rates after age 50 i s interesting. Although unemployment is lower among the oldest age groups, it remains significant at 8 percent for those aged over 60. Thus labour market participation is still common for this age group (the activity rate for males aged 60+ i s 60 percent, though for women it is only 26 percent). Figure17: UnemploymentratebyAge and Gender,2005 . . 8 25 - 3 2 0 - L CI 3 E --*-Total 1 5 - -+-Male k 0 3 . . . A . . Female 1 0 - 3 5 - 0 15- 20- 25- 30- 35- 40- 45- 50- 55- 6W 20 25 30 35 40 45 50 55 60 Source: LFS 2005, own calculations. 4.12 Education plays a significant role in explaining the incidence o f unemployment. Ethiopian students begin with 8 years o f primary education (subdivided into basic primary, Grades 1-4, and general primary, Grades 5-8), then proceed to the first cycle o f secondary education (Grades 9-10), the completion o f which is considered "General Education"-r they have the option o f applying to a Junior TVET program. Students who pass an exam at the end o f Grade 10 are eligible for preparatory secondary (Grades 11-12), and then can compete for a place in a tertiary institution. Those completing Grade 10 who are not admitted to preparatory secondary, or who prefer, can compete to enroll ina medium-level TVET program (World Bank 2005~). 4.13 Unemployment peaks among those in the middle o f the spectrum o f educational attainment, as shown in Figure 18 and 19, which compare the educational profile o f unemployment by age (youth vs. adults). For youth, female unemployment peaks among those with General Education-for young males, those in the Beyond General category are most likely to be unemployed. Among youth even the most skilled have a very high 49 unemployment rate (for women, those with higher education are about as likely as those with Beyond General to be unemployed). However, the picture changes somewhat for adults, in that those with higher education become the least likely to be unemployed, suggesting that in the long run education pays. The fact that those in the middle o f the spectrum-beyond primary but with no more than General Education-have the highest rates even for adults is cause for concern. It may signal a mismatch between the educational curriculum and the skills demanded in the labour market. Among the illiterate population, who can perhaps least afford not to work, unemployment was about 7-8 percent over the period. Figure18:Unemploymentby educationlevel Figure19: Unemploymentby educationlevel (percent unemployed), Age 15-24years (percent unemployed), Age 25+ years 4s , 4 , , , , , I , Source: LFS (2005), own calculations. Youth Unemployment 4.14 Ethiopia is not alone inAfrica inconfronting a highlevel o f youth unemployment (Figure 20). The demographic transition from highfertility-high mortality to low-fertility and longer life expectancies implies a spike inthe dependency ratio. Young entrants to the labour market, who are generally better educated than their parents and have higher expectations for employment, face difficulty securing jobs in many parts o f the world. Ethiopia has not yet entered its demographic transition, and the fertility rate i s so high that the population continues to grow at a rapid pace. Particularly as land degradation and scarcity act as push factors for migration from rural to urban areas, urban youth unemployment i s becoming an increasingly major concern. 4.15 For youth, who comprise almost one-fourth o f the urban population, unemployment rates range from 16 percent for 15-19 year olds to 23 percent for 20-24 year olds. This reflects much lower activity rates for the younger group (38 percent versus 69 percent labour market participation). Disaggregating by gender reveals even larger differences: males in the younger group have a 14 percent unemployment rate, while the rate for females in the older group i s 12 percentage points higher. Of course, the unemployment rate alone does not fully capture the challenges youth face in securing productive employment, inparticular the quality o fjobs held. 50 Figure20: UrbanYouthUnemploymentRatesinSelectedAfricanCountries(year varies) 30 20 10 Source: Leibbrandt and Mlatsheni, 2004, except Ethiopia own calculations from UBEUS. "Youth" definedas ages 15-24. RegionalPerspectives 4.16 Regional variation inunemployment is significant (Figure 21). The large cities- Addis Ababa and Dire Dawa-have the highest rates. Smaller urban centers have a profile more similar to that o f rural areas, where underemployment rather than unemployment i s a key feature. Also striking i s the variance in differentials between youth and adult unemployment-in Tigray, the youth unemployment rate is about 2.5 times the adult rate, while the relative gap in Afar is fairly small. These differentials suggest that unemployment may be more cyclical in nature in some regions, and more structural in others, but determining these drivers i s beyond the scope o f this report. Statistically significant increases in youth unemployment between 1999 and 2005 were observed in Tigray, Gambella, and Harari, while there was a significant fall in Addis Ababa (from 39 to 34 percent). Determinantsof Unemployment 4.17 Multivariate analysis confirms the descriptive results. The regressions (see details invol. 11, Chapter 1)include variables for age, gender, educational attainment, household characteristics (number o f children and adults), and regional dummies using the 2005 data. As expected, the effect o f being female, all else being equal, i s highly significant and associatedwith a 37 percent higher probability o f unemployment. 51 0 Having completed General Education also stands out as an important correlate o f unemployment; restricting the sample to 15-24 year olds, General Education i s associated with a 78 percent higher likelihood o f being unemployed. The effect is more pronouncedinAddis Ababa than other parts o f the country. 0 Many o f the other education variables are also significant. Interestingly, having had some form o f training is correlated with a 25 percent lower likelihood o f unemployment in Addis Ababa, while it i s not significant in other parts o f the country. This is consistent with the expectation that there are relatively good opportunities for semi-skilled workers in Addis Ababa, and that training may help ease initial entry into thejob market for young people. *I Figure 21: Unemploymentby Regionand Age, 2005 40 Li 15-24 Source: LFS (2005), own calculations. 4.18 Household composition also helps explain unemployment: adding a child under age 10 to a household is associated with an 11 percent lower likelihood o f adult unemployment (though looking only at women shows that it i s linked to a higher likelihood o f unemployment, presumably because women with young children are less likely to find a job that they can balance with childcare responsibilities). Interestingly, increasing the number o f adults in the household i s associated with a higher rate o f unemployment o f the adult household members. This seems to imply that there may be some kind o fjob rationing within the household, though the effect is somewhat smaller for youth. Results from Addis Labour Market Survey (ALMS) 4.19 This section briefly highlights key results o f the Addis Labour Market Study (ALMS) commissioned in 2006 for this report. It attempts to shed additional light on some o f the qualitative questions that could not be answered with the standard labour market surveys (see Box 6 for details). 52 Box 6: The ALMS Survey InApril andMay 2006the World Bank commissioned a small survey to probe further into some o f the peculiarities of the Ethiopian urban labour market, such as the high levels of unemployment and the low levels o f informality. In particular the survey wanted to probe whether some o f these results might be driven by the characteristics o f the statistical instruments used, such as the definition o f work captured (as opposedto productive activities more generally) or the practice o f relying on a key informant at the household level (rather than each of the individuals on which information i s beinggathered). The survey was run on 282 households drawn from the Addis subsample of the UrbanPanel (see Box 2) for a total o f 1,982 individuals. The modules included inthe survey covered: individual labour market behavior including information on primary and secondary jobs; type o f activities undertaken during the week and whether they included any form of direct compensation; unemployment with an emphasis on means of support o f the unemployed, search strategies and desiredcharacteristics o fjobs individualswould like to hold; self-employment and characteristics of the businesses run, including whether they ever tried to formalize their activities and the support networks that facilitate business activities; social capital, including questions on reciprocity networks and organizationshnstitutions individuals rely onto; and subjective well- being. Other household characteristics, such as their welfare status in2004 were also usedinthe analysis as backgroundinformation. 4.20 I s unemployment really as high as the reported figures? The ALMS module on time use asked the unemployed respondents to choose the three activities (from a list) that they had spent the most time doing in the previous week. The list had 22 choices, including inter alia productive activities-several examples are "selling products produced by household business," "construction in the community," and "buyinghelling inputs, running errands for businesses/productive activities of fiends or relatives." The survey thus intended to capture forms o f employment that labour force survey respondents might not consider "proper" jobs. Table 17 shows the results, omitting activities that fewer than 10 percent of men and women selected. Household duties and work for own consumption, and sociaVeducation activities, dominate the list. A negligible share indicated that productive activities were one o f their three main activities; moreover, almost none of the unemployed indicated that they received some form of compensation for the activities they selected. These results reinforce the high measurement o f open unemployment. 53 Table I - - 4.21 The gender ~ ~ s t ~ ~fuactivities nr o ~ ~ o Is several stylized n spend much more time on ho Idwork, though the ce i s still striking. ~ ~ ~ ~ p lmenytenddto sperzd much o c d also on lookirlg for eral, ~ o ~ are rno e n nien to be among the * -se"arcli~n~~'as o "searching)' u ~ ~ m ~ ~ 5 yThese. e d results are related to an r qu~st~othat asked what traits the une~~p~odesired ina n y e d job. ~ h ~a plurality o f the search in^ employed(who we di~propo~~onately l e male) said that it good income vas the most i ~ ~ ~feature, n t ~ a the ~t~~i-searchingwere more Likely to desire i~depe~dent work that is close to hoine. This may be bccausc ~ ~ o needeton ~ balmce m y e ~ ~ ~ owith ~ etheir ~zz ot ~ s e ~r~sponsibii~ties.While they may claim to ~ o l d " for work, the fact that they arc not actively seeking it sugges~sthat their y be ~ ~ n i j t eThis cfi ~ . rization is slightly d ~ e r than~ that of e ~ rs" who also undoub~e~lymake a c ~ ~ p oo~f thennon-sea e ~ 4.22 Given the extent 5f urban p o ~ ~ cdo~the~u n e ~ ~ ~support o y ~ d~henzselves~~ d are not spend^^^ a sig nt amount o f time on "odd jobs" to earn ask how they are survivi short.,it appcars that ~am~lies care o f take their u~~employedi e ~ ~ e r The vast majority (85 percent) of the ~ n e ~ ~ preportythat~ n s . ~ o e living with family is their ~ r means of ~support,a ~ o l~l o ~ ~byd ~ ~ ~ i o n e t ~ y ~ e gi~ s ~ e ~ p / ~ n ~ e!loanafromerelativesifriends"-and r i ~ n c most of these (84 percent) do not expect to be repaid. A ~ t l t o u gtheir families provide for thcm, 13 percent o f the ~ u ~ ~ e m ~ l say etheir f a ~ ~ lthi~k o ~ d i ~ s''1 should wait until Ifind a good job." Perhaps this is part of a h o u ~ e h ol ~~~ v e s~t r a~~ e~~(e.g. ~if ~a member secures formal wage yo ~ ~ ~ ~ p litocoul i e ~~ s~~, a ~lessen a ~ ~ t ~ the~entire l ~ ~ # s e ~ o vud ' s ~ y ~ entage af the searching un ployed (64 pcrc oycd (21 p ~ r say that their~ families want them to "get a job as c ~ ~ ~ 4.23 Almost ~hree-qu"a~ers the ~~e~~~ of art that they would be willing to accept ""any available job." This varies by cducation levef, with skilled workers ntorc likely to ~ n d ~ c thata ~ ethey would only be willing to take wage e ~ ~ p ~ o ~ n e ~ t in a public or private o~ganization(with just a small perce~ltage~ j s ~ ~ n ~ between ~ n ~u i s l ~ 54 the two). Duration o f unemployment does have a statistically significant positive effect on the likelihood o f wanting "any available job," so it does seem that jobseekers become less particular as unemployment becomes more protracted. On indicators o f subjective well-being there were statistically significant differences between the employed and the unemployed, with the latter faring worse on all. This undermines any argument for considering unemployment a result o f "luxury" or a preference for leisure. Perceptions o f the main obstacle to finding employment vary widely, fiom lack of education or experience to lack o f information and contacts, reinforcing the notion that there i s no magic bullet to solve the problem. Kev Implications 4.24 International experience has shown that there are no easy answers for tackling high unemployment, especially for youth. Much more is known about effective early childhood interventions, for example, than best practice interventions aimed at boosting youth employment-though more attentioni s being focused inthis area, including via the creation o f the global Youth Employment Network by the UN, World Bank, and ILO. The World Bank's 2007 World Development Report-entitled Development andtheNext Generation-overs the multipletransitions youth face as they become adults. 4.25 InEthiopia, medianunemployment durationfor young adults isabout ayear, and women experience longer spells o f unemployment than men. Some literature argues that the truly harmful effects o f unemployment (on human capital and individualhocietal welfare) come fiom long-term unemployment, rather than unemployment incidence per se. This implies that poli.cy design should take the heterogeneity o f the unemployed population into account. 4.26 The entrenched youth unemployment inthe Middle East and North Africa region provides a cautionary tale about what can happen with a growing youth labour force and insufficient job creation. The relatively undiversified economies o f many countries inthe region, patronage-based governance, and weak private sector mean that despite dramatic improvements in human capital a large segment o f society i s unable to find productive work. Depending on guaranteed public sector jobs to generate employment for new graduates-a strategy followed by Ethiopia under the Derg but since abandoned-is clearly unsustainable. Yet much o f the employment generated in Ethiopia in the last decade has still been in the public sector as part o f decentralization, rather than in the private sector. Some o f the evidence presented in the preceding chapter provides welcome signs o f new dynamism, which needs to be nonetheless significantly expanded. The task for the government, then, will be to boost private sector confidence and empower economic agents through improvements in the investment climate and send credible signals that political risk i s at an acceptable level (discussion o f these issues is beyond the scope o f this report; see the forthcoming World Bank CEM). 4.27 The empowerment o f women writ large will be critical to reducing the dramatic differential between male and female unemployment. Women are disadvantaged on every measure o f labour market outcomes, to say nothing o f welfare more generally. They are constrained in the types o f jobs they pursue by social norms regarding 55 acceptable female behaviour, and lower asset ownership and household and childcare responsibilities may make it more difficult to become self-employed. While a comprehensive treatment o fpolicy interventions to raise the status o fwomen in society is well beyond our scope here, it i s important to underscore that any employment-related program should be designed with the distributional impact by gender inmind. 56 5. MIGRATION, URBAN LABOUR MARKETSAND POLICIES 5.1 Limited rural-urban migration over the past three decades has contributed to the low urbanization rate o f Ethiopia. According to the most recent national data, 14 percent o f the population (aged lo+) was living in urban areas, and o f those 18 percent had moved there over the four years up to 2005 (LFS 2005). 5.2 Migratory pressures are bound to increase. While the lack o f recent census data does not allow an assessment o f current flows o f internal migration there i s anecdotal evidence supporting the view that migration is increasing. Further increases can be expected due to population pressures on land and environmental degradation in rural areas and the momentum o f the development process itself. Longer term determinants such as "regional imbalances in employment opportunity, improved communications, road and transport networks and changing aspirations o f the younger generation" (Deshingkar 2003, p. 25) are also likely to play a role. 5.3 There are reasons to be concerned about the transformation that increased internal migration flows brings about. To the extent that urban labour markets are segmented, a large influx o f migrants to urban areas would result inincreased competition in the low- barrier low-productivity sectors, typically the informal sector, hrther bringing down wages and increasing underemployment, increasing urbanpoverty and putting pressures on the provision o fbasic services and infrastructure. 5.4 There are elements supporting some o f those concerns, but they need to be evaluated more closely. Inurban areas there are signs o f increasing pressures on access to basic services (World Bank 2005a), and the most recent data show that over the period 1995/96-1999/2000 while poverty decreased in rural areas, it stagnated in urban areas. From the point o f view o f the labour market, it appears that migrants are more likely than non-migrants to be self-employed, suggesting that increased migration might lead to a swelling o fthe informal sector. 5.5 This chapter will build on quantitative and qualitative data sources to scrutinize more closely the evidence on the impact o f migration on the labour market. It will also look at international evidence to highlight the need to create a supportive policy environment to reap the potential benefits o f migration on growth and poverty reduction. These policy conclusions will be explored further inChapter 6. A Snapshot of InternalMigration 5.6 InEthiopia overallmigratory flows appear limitedthough there are signs growing migratory pressures. Between 1999 and 2005 the pace o f migration grew significantly, 57 albeit from a very low base (the share o f migrants who had s ent one year or less intheir current community almost doubled, going from 1.8 to 3.5)J4 Comparisons based on a sample of DHS surveys recording information for female migrants offer a comparative perspective (Figure 22). Figure 22: A Comparisonof the Incidenceof Migration Based on DHSSurveys (Percentage of Urban Women of ReproductiveAge who are RecentMigrants, by Region) 35.0% I 1 30.0% 25.0% 20.0% 15.0% 10.0% 5.0% 0.0% Ethiopia North Sub- Southeast South, Latin Total (not Africa Saharan Asia Central America including Africa West Asia Ethiopia) Source; National ResearchCouncil .2003 and own calculations for Ethiopia (from DHS2000). 5.7 The most common move is within rural areas, and smaller cities play an important part inthe process o f "step-migration." About 40 percent o f recent migrants were living inurbanareas in2005, of whom about 40 percent hadmoved from another urbanareas. Among recent migrants living inrural areas intra-area migration was almost double at 79 percent. As the population i s overwhelmingly found in rural areas, the most common type o fmove was withinrural areas (46 percent), while recent migrants represent a larger share ofthe urbanthan o f the rural population (18 percent rather than 4 percent). Smaller cities play an important role in migration, a pattern common to Northern and Sub- Saharan Africa (Figure 23, 24), consistent with qualitative evidence that much o f migration, particularly within rural areas, i s short distance. 24While the overall share of urbanrespondents is stable between the 1999 and the 2005 LFS (13.5 and 14.2 respectively), the share o f recent migrants inurban areas rose from 15 to 18 percent. 58 Figure23: A PercentagesofUrban Figure 24: PercentagesofUrbanWomen Women of ReproductiveAge who are of ReproductiveAge who are Recent RecentMigrants,by City PopulationSize MigrantsinEthiopia, byCity Population andLocation,Restofthe World Size andLocation 0 . S 1 1 0.3 0 29 0 2 0 16 0 1 0.05 MnhAf&a SubSaharan Southeast&la South. Central. Lainmenct 0 Afea west&la I I Source; National Research Council 2003 andown calculations for Ethiopia(from DHS2000). 5.8 Migrants tend to be young and concentrated inthe most productive age groups. Stark gender patterns in migration behavior are closely associated with different reasons to migrate are visible. Migration rates are highest in the 15-29 age group. Disaggregationby gender shows that women are more likely to be found inthe age group 15-25, while the opposite applies to migrants25+. 5.9 Migrants move for a variety o f reasons, with work-related reasons such as searching for ajob and having a job transfer more prevalent for men than for women, as are reasons o f education and returning home. When disaggregated by intended duration o f migration, however, gender differences for temporary migration disappear, with job- related temporary movesjust as likely for women as for men. Women typically migrate permanently for reasons related to marriage arrangement and dissolution (particularly within m a l areas) and, albeit to a lesser degree, moving along with fmily. Reasons related to distress migration, such as displacement, war, drought and living with relatives (which qualitative evidence shows is a common coping strategy) are equally as likely for men as for women. 5.10 Job related transfers are mostly to urban areas. This is the case for "contracted moves"-i.e. moves part o f an agreement with an employer-though also "speculative moves"-i.e. searching for jobs-are slightly more directed towards urban areas. Not surprisingly, education-related migration i s overwhelmingly directed towards urban areas. 5.11 Economic migration to urban areas has been traditionally analysed by economists in terms of the Harris-Todaro model, which identifies wage differentials as the key drivers o f the migration decision. The lack o f rural wage data prevents us at this stage from runningsuch type o fmodels.25 We resort therefore to reduced form models. 25Two conceptual problems complicate the empirical analysis. One is that while in an aggregate analysis o f flows we can describe wage differentials between recipient and sending areas, we cannot assume that those differences are motivating migration, as migration decisions per se can affect relative wages. 59 5.12 The common perceptions o f migrants are o f individuals who are unskilled and with limited training. Further, particularly for women, they are associated with prostitution and begging in urban areas. The data reveal that while migrants are mostly unskilled, they tendto be more educated than non-migrants. In1999, 58 percent o f recent migrantshadno education, which compared very favorably with 79 percent o fthose who have never migrated. Recent migrants have noticeably higher incidences o f grade 5-8 and grade 9-12 schooling than the population who has never migrated, partly o f course because requiring an education beyond basic primarymight have required them to move. Educated migrants are more likely to be men. 5.13 Econometric analysis o f the determinantso f migration points to the heterogeneity o f migrants, with education a key variable in differentiating flows. The likelihood o f being a recent migrant is "bipolar": taking as base case having no education, individuals who have grades 1-12 appear less likely to migrate. Those with more than 12 years o f education, in contrast, appearjust as likely to be migrants as those without any education. Non-recent migrants show a different pattern, with those with grades 1-4 showing a higher likelihood of migration than the non-educated, while those with higher levels o f schooling display lower probabilities. Among the other individual characteristics, gender does not play a significant role.26 5.14 Network effects are important drivers o f migration. The share o f migrants living inthecommunity very significantly affects theprobability ofbeingamigrant. 5.15 Qualitative evidence helps to understand the mechanisms which lead to these characteristics o f the migrants (Box 7).27 Cultural and institutional factors, such as informal rights to land, agricultural practices, marriage practices, affect the migration opportunities o f different types o f individuals. Gender and ethnicity are important dimensions along which migration i s segmented. Further, at variance with economic models that focus on expected wage differentials as the main drivers o f migration, migration i s rarely based on a single motivation. Of the 25 or so reasons for migration to urban areas put forward by the ETPPA respondents "some are individual (for example, being stood-down from the army), but most involve family interactions or events that Secondly, if we try to explain the determinants o f migrationat the individual level, it is difficult to arrive at the "individual specific wage gap" neededina Harris Todaro model. As migrants self-select into migration we cannot impute their wage inthe zone o f origin, as by migrating they reveal that there is something intrinsic to them that makes them different fiom those who stayed. It would be possible to tackle econometrically these problems if individual panel data tracking individuals across the migration experience, or at least different cross-sections providing wage information over given geographical areas were available. Among the data constraints the lack o f disaggregated cost o f living information for different settings (and for different groups) and the difficulties at arriving at estimates o fthe direct costs o f migration should also be noted. 26This finding is at odds with those o fprevious studies of migration inEthiopia which findthat women are more likely to migrate. Our inability to disthguish permanent migrants from seasonal or circular migrants might help explain this difference (with women more likely to be permanent migrants). ''Ideally one should focus on the time o f migration, whereas inpractice the surveys have ex post data (i.e. after migration). A focus on recent migrants goes some way towards addressing this problem, which is the strategy adopted inthis report for quantitative analysis. Unless otherwise specified what follows refers to recent migrants only. 60 place the individual in a weak position for making a livinginrural areas (divorce, loss o f landaccess, death o fa spouse and so on)" (MOFED 2005 pp. 60-61). 5.16 Overall, the employment outcomes o f migrants are better than for non-migrants, though there is much variation in their employment due to their different characteristics and reasons to move. The employment rate o f recent migrants is 70 percent, against 65 percent for non-migrants. Such a difference i s due to the high employment rate among migrant men (85 percent versus 77 percent for non migrants), while the women's employment rate at 55 percent i s the same across groups. In general, migrants who moved for work-related reasons exhibit the highest employment rates. Breaking down women's employment performance, it appears that recent migrants who moved for work- related reasons have employment rates not very different from men's. 5.17 As migrants are relatively better educated, one can expect them on average to have higher probabilities o f being employed than non-migrants. Our analysis shows, in contrast, that it i s not only the difference in observable characteristics (such as years o f education) that drives their employment performance. Among recent migrants this positive effect o f being a migrant is stronger for women, while the opposite is true for those who migrated more than 4 years before. Disaggregating the model by education groups shows that for recent migrants the "premium" on being a migrant in terms o f employment probability is stronger for the low skilled and the high skilled (those with 1- 4 years o f schooling and those with more than 12 years o f schooling). For non recent migrants, this "premium" is significant for the highskilled women and lower skilled men (upto 8 years o f schooling).28 Box 7: PushandPullFactorsinRuralandUrbanAreas - QualitativeEvidence For analytical purposes, reasons for migration are generally classified as push - i.e. factors in sending areas that provide incentives for mobility - and pull - i.e. factors inreceiving areas that provide incentives for mobility. The WED study (Tadele et al. 2005) and the Participatory Poverty Assessment (MOFED 2005) provide insights on these, and highlight the complex motivations driving migration. Such motivations go well beyond the simple differentials in expected wages at the centre o f economic modeling o fmigration. The rural push factors associated with male migration are shortage o f land, landlessness, destitution, the needto raise cash to pay the land tax and debts incurred inpaying for agricultural inputs. Urban areas are associated with the pull of perceptions of better employment opportunities, particularly in the informal sector and in construction, information on jobs from informal sources suggesting that there i s demand for urban workers and better pay than inrural areas, and social support by longterm migrants are identified as important pull factors. Women are overwhelmingly reported to migrate to urban areas to join their spouses, though a distinctstreamo fwomen lookingfor work can also be identified. The former groups participates inthe urban economy by working outside the house and undertakingpetty trading activities to a varying degree, depending on cultural factors linked to their ethnicity and origin. The latter group i s constituted by young single, or divorced or widowed women moving to look for work. Their migration i s related to institutional and cultural factors, such as the plough-based 28The other results are inline with other evidence for Ethiopia (see Volume 11). 61 agriculture o f the north which prevents women from performing key tasks even when they have access to land. Further, the need to move away from the social norms typical of rural areas or from negative perceptions o f women working outside the house are closely associated with the choice of migratingto urbanareas. Inruralareas, there is significantmobility, thoughonly inminimalpart is itlongtermandrelated to labour market behaviors. Muchmobility to rural areas i s seasonal. For example, urban daily labourers who usually work in peri-urban areas in agricultural tasks travel seasonally to rural areas to take advantage of the various cropping seasons and to perform other agricultural tasks. Permanent migration inrural areas, bothvoluntary and involuntary, is driven by land availability and the existence o f previous migrants. Women migrate a lot upon marriage but such migration tends to be short distance. Migration by women to work intowns i s seen as shameful, and risky, though perceptions of female migrants (or commuters to nearby areas) by rural inhabitants can improve iftheir earnings allow themto support their families. Qualitative evidence highlights how migrants rely heavily on contacts and relatives in their area o f destination, who often are the ones who helped them migrating in the first place. Migrants who follow "circular" patterns, by which they go with some regularity to their home communities often convince young men, unmarried women and children, to follow them to urban areas where they canhelp them find ajob. Ethnic basediddirs burialassociations and self-help groups are the key institutions migrants set up to provide mutual assistance for various purposes, including facing unexpected events (funerals, needto provide help to relative inruralareas) or plannedones (organizingjoint trips to go backto areas o f origin for the Meskel holiday). An important element affecting migration decisions is the cost, though we have only indirect evidence on effects. The PPA notes that the cost o f travel i s relatively highincomparison to the ability o f farm households to generate cash. Similarly, evidence from the WED study shows that location decision for migrants to urban areas are influenced by the closeness to home origin. Risingtransport costs are also notedas factors hinderingcontacts with home communities, which are particularly important for those migrants who have left their households in rural areas and move back seasonally to continue to work their land. 5.18 While the local concentration o f migrants is strongly correlated with the decision to migrate, the employment rate o f migrants (rather thantheir concentration) is associated with probability o f employment.29 This finding is in line with qualitative evidence suggesting the importance o fnetworks and personalconnections to obtainjobs, as well as o f the characteristics o f the local economy, and inparticular the existence locally o fjobs that migrants can access (e.g. a growing construction sector).30 5.19 Analysis by gender shows that women migrants are less likely to work than men, and that the effect is particularly strong for non-recent migrants. The positive effect o f 29 Note: these variables referring to migrants in the local area exclude the individual observation in question. 30This finding couldotherwise reflect the characteristics o fthe area-i.e. that it is an area where everybody has a highemployment rate, migrants and non-migrants alike, and where the probability o fbeing employed is thus high. This i s not supported by the data, however, as the employment rate for non-migrants should be equally significant inaffecting the probability for a migrant to work, which i s not the case. Furthermore, the employment rates for migrants andnon-migrants ina givenarea show little correlation. 62 age on employment is similarly more contained for women. For women the highest probability o f working i s associated with no education and with more than a 12thgrade education (effect o f grade 12+ is particularly high for non-recent migrants), while for men with less than 5 years o f primary (also with more than 12 years for non-recent migrants). In terms of life-events, for women marriage is associated with a lower probability o f working (the opposite holds for men). Marriage dissolution is generally associated with a higher probability o f working. For women who have not migrated recently, the negative effects o f marriage are less pronounced and those o f marriage dissolution are more positive. 5.20 The positive performance that migrants have in the labour market i s partly reflected also intheir wages, at least for men.3' Migration is associated on average with a 14 percentage point premium for male migrants. Disaggregated analysis by educational group reveals that such premium applies to recent male migrants who have higher education (more than 12 years o f schooling). 5.21 Closer analysis reveals that migrants and non-migrants face a different wage structure. Returns differ systematically for migrants and non-migrants. Gender differences are also very noticeable. On average migrants have higher returns to skills than non-migrants, though the differences decline for higher levels o f education. This might reflect the correlation o f migration with unobservable individual characteristics which also influence earnings. The pattern differs for women and men. Inthe case o f female migrants, they enjoy higher returns to education than non-migrants at all education levels, and the gap decreases for higher educational levels. Inthe case o f male migrants, for lower educational levels they experience lower returns than non-migrants (for male migrants returns are not significantly different from those o f the non-educated up to 8 years of education) and they experience slightly lower returns thannon-migrants for levels o f education above grade 12. 5.22 The spatial clustering o f migrants i s also related to lower wages, in turn suggesting competition between migrants in the labour market. While this effect i s non- significant on average for the non-migrant population more disaggregated analysis reveals that the individuals with some education (grades 1-4) are those whose wages are negatively correlated with the concentration o fmigrants. 5.23 Does the competition between migrants extend to non-migrants as well? On average there i s a negative but non-significant effect. Disaggregated analysis reveals a negative impact on the wages o f those without education. 5.24 Insummary therefore the available evidence suggests that from a labour market point o f view migration i s not a major concern, but that as pressures on land inrural areas increase, new challenges could follow. To date, however, most o f migration i s relatively 31 The analysis o f wages is conducted with the Child Labor Force Survey database, which collects information also on the overall labor market performance o f the whole family. Unfortunately, this survey does not allow individuals to be identified it i s not possible to identify as migrants, as the question i s only asked at the household level. Compared with data from the LFS of 1999, the CLFS provides lower estimates o f migration. 63 short distance, with both step-migration and moves to urban areas directly from adjacent urban areas being quite common. Those who migrate for work-related reasons have a better employment performance than the others, possibly because o f the importance o f network effects in both motivating migration and in favoring access to jobs. While migrants do not seem to "compete" directly with non-migrants and to drive down their wages, except for those with very low skills, they do seem to be incompetition with each other. These findings do indeed raise concerns for a possible scenario o f increasing migration due to distress in rural areas, resulting in increasing poverty among migrants andultimately also amongthe longterm resident population inurbanareas. The Impact of Mimation on Growth and PovertvReduction 5.25 It is not possible to correlate these findings on the impact o f migration on urban labour market with an estimate o f the overall poverty reduction impact o f migration. The lack o f evidence on wages, returns to capital and underemployment in rural areas makes it difficult to arrive at an overall assessmento fthe effects o finternalmigration on growth andpoverty reduction. 5.26 Qualitative data, however, suggests a positive role o f migration in alleviating poverty. Qualitative evidence from the WED study in line with other (see Box 8) suggests that while the economic opportunities opened to migrantsmight not allow them to escape poverty, they are often able to significantly contribute to improvements in the livelihoods o f sending households and o f sending areas. Migrants who keep contacts with their rural areas o f origin have been found to channel resources directly to the rural areas, through remittances, but also contributing to iddirs in their home area, or contributing to efforts such as fundraising for local roads. 5.27 The same qualitative evidence, however, suggests that things might be becoming more difficult for migrants. This raises concerns for the future if migratory flows increase as expected. Box 8: TheImpactof Migrationon Wellbeing inRuralAreasinEthiopia Several studies review the impacts of remittances in Ethiopia. McDowell (1995) argued that the size of remittances for poor householdscan be very low but it is vital to improve food security since it helps to diversify risks and ensure support in times of harvest. Remittances can also stimulate agricultural intensification where practices allow the head of household(who may be a woman or a less senior man) to employ labour, and remitted earnings are used for purchase of equipment, seeds, fertilizers or draught animals. W o r k (1995) inhis study of Guragemigration argued that the impacts of migration are not merely economic, but that the interaction brings about cultural change through ideas, skills and attitudes, and that migration makesrural villagers more receptive to change in areas such as technology. He found that Gurage migrants are responsible for a dramatic change in the attitudes, values and beliefs of both migrant and non- migrant households. Negative consequences of migration have also been noted. Worku (1995) argues that migration in Gurage is responsible for greater rates of divorce, higher transmission of STDs/HIV-AIDSand the introduction of new habits such as consumptionof locally made alcohol and cigarettes. Migration can also result in a greater work-load for women, and inflation of bridewealthcausedbymiaants' income. 64 5.28 While the evidence for Ethiopia i s still limited-though more work is currently planned in the context o f the ongoing urbanization study-the international experience offers some key findings on the contribution o f migration to growth and poverty reduction. They offer a starting point for the elaboration o f policies to enable the urban economy to reap the benefits o fmigration, while minimizingpotential shortfalls. (9 Migration can increase efficiency andgrowth through improved resource allocations 5.29 Internal migration has long been recognized as central to the processes o f urbanization and structural transformation, which cross-country evidence suggests are associated with higher development levels. Indeed, "as economies develop, the share o f the agricultural labour in total labour declines and converges to a level o f 2-3 percent" (Larson andMundlak, 1997). While there i s significant dispersion inthe degree to which countries have urbanized during their development process, a strong negative association can be found incross-country data between income levels and share o fpopulation inrural areas. 5.30 Migration can therefore play a major role in fostering growth and poverty reduction, by reallocating resources more efficiently both geographically and sectorally across the economy. China offers a spectacular example o f the transforming role o f migration: an estimated 16 percent o f GDP growth over the period 1987-2005 has been contributed by migration. Part of the growth impact i s generated by internal remittances which, given the right conditions (an environment supportive to local business development; infrastructure to support communication and growth), can increase demand for agricultural production and stimulate non-agricultural activities. (ii) Migration can reducepoverty signijkantly, but there are new risks of exclusion 5.31 By opening up more productive employment opportunities and allowing a diversification o f risks, it can significantly contribute to household income. The benefits are not limited to the migrants-for example, to the extent that a dynamic economy develops in urban areas, the influx o f low-wage migrants can prevent a loss o f competitiveness due to rising wage costs. The characteristics and skills o f migrants and evolution o f sectoral demands for labour are important elements o fthis virtuous circle. 5.32 To the extent that there are strong rural-urban linkages some o f the benefits enjoyed by the migrants will filter back to the rural areas. Such linkages include remittances (Box 9), knowledge and ideas that return migrants can bring back, the introduction o f new technologies, and the demand that urban migrants can have for rural products. Box 9: TheImportance ofRemittances inRuralPortfolios A comparative review of empirical work on rural household income portfolios in Sub-Saharan Africa by Reardon (1997) found that on average 15 percent o frural incomes are accounted for by remittances. In China, it has been estimated that remittances sent home by urban migrants compensated sending households for the loss o f labour due to migration (De Brauw et al. 2001, quoted in Fan et al. 2005). The nature and level o f remittances varies widely depending on the 65 accessibility of the home village, employment opportunities, the costs of living, the ease of remitting, andthe `orientation' of the migrants (particularly ifthey intendto return home). Even if income from remittances is small, the limited amounts of additional cash can make huge differences to the options available to peopleto cope with disaster situations. (Sharp et al., 2003). 5.33 At the same time, migration can reconfigure the profile o f poverty and vulnerability in rural and urban areas, posing new policy challenges. Evidence from India shows that internal migration over the 1983-1993 decade contributed to a 2.6 percent decrease in rural poverty, while increasing urban poverty by a smaller amount. Similar patterns were found for the 1990s-suggesting that rural-urban migration contributed to making poverty more visible by shifting it at least in part from rural to urbanareas. It is also possible that those left behindbythe migrants might also be more vulnerable, particularly ifrural-urban linkages are weak. (iii) Country characteristics and institutions mediate the impact of migration 5.34 Country characteristics and institutional features affect the nature o f migratory flows, the success o f the migrants and the linkages that can be kept with home communities. Relevant factors include institutions, geography, transport costs and' information, as well as opportunities to integrate in receiving labour markets. For example in Mexico the poor migrate (internationally in this case) and this reduces national poverty (Stark 1991). The geographic contiguity with a labour market characterized by large demand for low cost low skilled work (the US) is clearly part o f this poverty-reducing outcome. In one study in Western India, seasonal migration was found to be driven by as well as amplifying rural inequality, as it was directed towards urbanrural labour markets strongly segmented along ethnic lines, so that those without access to recruiters' networks or kin can only access the worst paid forms o f jobs. Furthermore, recruitment from villages through networks decreased competition among employers (Mosse et al. 2002). Migration can also be associated with increasing inequality inurban areas, particularly as migratory processes tend to be cumulative, with successive waves o f migrants o f a certain type (location, ethnicity) attracting more o f the same-thereby generating competition and low earnings in sectors where migrants are typically working (e.g. Borjas 2003). (iv) Increased urbanization is an inevitable aspect of the developmentprocess 5.35 It is hardto control migratory flows, particularly ina context ofurbangrowth. As the China experience shows, once migratory pressures build up these flows are hard to contain. Such pressures can be both on the "push" and ``pull" sides. The former often includes population pressures and environmental degradation in rural areas. These key findings are illustratedby China's evolving approach to migration (Box 10). 66 Box 10: China: InstitutionalReforms Help Reapthe Potential of InternalMigration InChina, the hukuo (registration) systemwas introducedin 1958 to prevent urbanmigration by forbidding work and access to basic services outside the area of official residence. Economic reforms introduced fkom 1978 provided a more favorable environment for migration. Duringthe period between 1978 and 2004, income per capita rose from US$153 (constant 2000) to US$1,162, while the share o f the rural population declined from 81 to 60 percent. In 2001 the Government beganreformingthe hukuo, introducingnew fkeedoms o fmovement from rural areas modulated on the basis of the destination, i.e decreasing from movements to small towns to megacities. Today unregistered migrants continue to play a large part inChina's economy with estimates o fthose flows rangingfrom 50 to 120million. Internal migration flows are still male-dominated, owing both to cultural factors and demand in urban areas. In 2000, 37 percent o f migrants were employed in manufacturing, followed by construction, services, and restaurant and commerce (14, 12 and 12 percent, respectively). Small and middle sized cities are significant destinations, with high shares remaining within their province of birth. As in other Asian countries, increases in female migration reflect growing social acceptance and specific demand in some sectors (domestic work for the uneducated and factory work for women with some education). Internal remittances have been estimated to contribute as much as agricultural earnings to rural incomes. Circular migration during the year and highreturn migration drive remittances. Many migrants contribute to non-agricultural activities by returning to townships or small cities intheir area rather than to farming. They are also likely to take leadership positions in their home communities. Labour migration since the late 1990s has been recognized as one o f the national poverty reductionpolicies. The overall net effect o f migration i s an estimated increase inrural household income per capita o f 14-30 percent. There have been several policy initiatives to support the poverty reduction impact o f migration. For example in 2004, the Sunshine program which involves a partnership between the central and provincial governments in poor areas and provinces with highrates of outmigration was launched. It aims to provide vocational training to 10 millionrural laborers who plan to move out of agriculture or want to move to the cities, After a year o f operation 80 percent of the 1.5 milliontrainees are reportedto have foundjobs. New challenges are emerging. While "surplus labour" inrural areas still amounts to 150million people, some manufacturing areas have started experiencing shortages o f manpower, and the increasing sophistication o f products means emerging skills mismatches. Concerns on new vulnerable groups are also emerging: in rural areas, the poor and unskilled still face barriers to migration, and women and the elderly can be left behind. Vulnerable groups in urban areas include jobless migrants with no access to social services. Further policy reform i s needed in continuing the gradual reform o f the hukuo system, making more flexible arrangements for land tenure and continuing efforts to increase education, for China to continue reaping the benefits of rural-urban migrationinthe face o f increasing inequality andemergingnew forms ofpoverty. Sources: Huang and Zhan (2005), Deshingkar (2005). PolicyChallenpes 5.36 The PASDEP agenda includes as strategic priorities both the strengthening of rural-urban linkages and the rebalancing o f the growth strategy. Inthis context there is a new emphasis on small town development and growth poles and on employment creation, particularly in urban areas. Internal migration is not directly addressed by 67 PASDEP though increased migration i s likely to be spurredby sustained urbangrowth, as well as beingto a certain extent integralto the success o fthis development agenda. 5-37 Critically assessing the institutional framework within which migration takes place is an important step, as Ethiopia's regulatory environment has inherited an anti- mobility bias. While significant movements o f people have accompanied the country's development over the lgthand 20thcentury, voluntary movements o f people significantly slowed down in the 1970s. The drastic political and economic reform introduced in the Derg period prevented the consolidation o f land in large estates and kept internal migration in check. For example, the land reform and subsequent measures limited access to rural land only to registered and permanent members o f Peasant Associations. Further, land would be redistributed if someone was absent for more than a year, as was redistributed the land o f resettled persons. The effects o f these regulations inrural areas were compounded by the requirements o f official pass letters to go to the cities, as well as o f official registration with the kebele for urbanresidence. 5.38 Today, while requirements such as local registration and the need for an identity card provide means o f formally controlling internal movements, local perceptions on constraints to mobility also limit migration. The perception o f risk of land redistribution remains high and can stifle diversification o f livelihoods in both rural non-agricultural activities and long term migration. Perhaps equally important for migration decisions are perceptions on the role o f PeasantAssociations andkebele officials incontrolling internal movement (Box 11). The adoption o f a model o f ethnic-based federalism might have also reduced the extent o f interregional mobility, 32by compounding preferences for migration to areas whose language and culture migrants already share and particularly by affecting perceptions on the fairness o f the management and assignation o f housing and other assets, children's education and insecurity. Some discrimination by local authorities in findingjobs and providing business licenses on language or ethnic grounds has beenreported.33 5.39 Sustained flows o f migration without reform in the residence requirements will result inthe creation o f a new underclass o f citizens, with limitedaccess to urban services andat risko flosingtheir livelihoods inrural areas once their landis redistributed. As the experience o f East Asian countries such as China and Vietnam shows, deep pockets o f poverty and vulnerability can therefore emerge inurbanareas. 5.40 Poverty and vulnerability are likely to pose the greatest policy challenges in smaller urban areas, which might be the least able to face them. Such challenges include the creation o f a supportive economic environment for the private sector as well as housing provision and service delivery, which are likely to difficult to provide especially insmaller cities. Internationalexperience (National ResearchCouncil 2003) pointsto the difficulties that are generally found by smaller cities in providing appropriate living 32The Migration, Gender and Health Survey conducted jointly by the Addis Ababa University and Brown University (2000) with a focus on the five most populated regional states (Afar, Oromia, Amhara, SNNPR and Tigray), indicated that permanent intra-regional migration has become more frequent than permanent inter-regional migrationwith the exception of temporary migrationto Addis Ababa. 33Tadele et al. 2005. 68 conditions to their inhabitants who are likely to experience the worst forms o f poverty in urbanareas. Box 11:Perception on Constraintsto InternalMigrationinthe PPA The general case i s for people to be able to move freely-subject, as might be expected, to social constraints (as pertain to adult women in many places) and to carrying personal ID at all times. There was a sense in some reports that local administration kept a very close eye on people's movements and activities, although the degree to which this might inhibit individual initiative was difficult to assess. Inurban sites it was reported that resident registration with the kebele i s compulsory, and movement out o f the kebele to go elsewhere requires an official leaving letter. Letters from the kebele are also required to get access to health services for free (or low fee). How uniformly these requirements are pursued intowns, and the penalties for failure to comply with them are not clear from informationreportedby respondents. Inrural areas it was reported that as people's movements are monitored quite closely by kebele administrations, it i s advisable in many circumstances to have cleared any planned travel with the relevant officials before setting out, to avoid trouble. Furthermore, it i s widely perceivedthat migration beyond a certain duration will result in forfeit o f the land rights of the person concerned by the kebele administration. Insome sites it was also reported that engagement in non-farm activities such as trading in consumer goods would result in forfeit o f a person's rights over land. Limits to migration, therefore, could be seen as part of a widespread perception in rural Ethiopia that engagement innon-farm activities as part o f a mixed livelihood strategy i s not permitted, so that people must choose in an absolute sense whether to stay wholly in agriculture or to engage wholly innon-farmbusiness. Source: Adapted from MOFED2005. 5.41 It will be important to strengthen rural-urban linkages. These include the facilitation o f flows o fremittances andinformation (about jobs, technology and economic opportunities) which can support investment in the rural economy and growth by increasing the efficiency of the allocation of resources throughout the country. They also include strengthening of the infrastructure network inrural areas to facilitate the flows o f goods and satisfy the expected rise inurbandemand. 69 6. ADDRESSINGTHE POLICY CHALLENGES 6.1 The generation on a sustained basis o f decent income earning opportunities -job creation-is the sine qua non o f effective economic policy. In the absence o f jobs, government policy will bejudged as a failure by the population, and economic outcomes will not generate improvedwelfare over time. 6.2 The focus o f this report was on the narrower dimension o f urban labour markets. It is nonetheless a central element in the broader challenge, and indeed arguably more visible and politically potent. The newly heightenedpolicy focus on labour markets as a priority i s extremely welcome. This is not because labour market distortions are in themselves major obstacles for Ethiopia's growth and poverty reduction, but because a range o f policies impact the functioning o f the labour market. Employment creation is therefore an additional lens to evaluate the performance o f the reform agenda. 6.3 This final chapter briefly summarises the major challenges highlighted by this studyandreviews some o fthe policyoptions opento the Government o fEthiopia. The Urpencv of the Labour Market APenda 6.4 There have been a number o f positive developments in the Ethiopian labour market over the period under review. Among the important findings o f this report are the changes inthe quality and quantity o f labour supply due to the increasing skill levels and the substantial improvements in individual mobility over the decade-trends which are consistent with the development aspirations o fEthiopia. 6.5 However these positive trends are overshadowed by pressing emerging needs. These were documented fairly extensively in the preceding chapters, and can be briefly, ifpartially, summarized as follows: The need to accelerate the pace o f job creation in the economy-given the approximately 600,000 people (mainly youth) entering the total labour market each year, and to make inroads into the stock o f unemployment: roughly 550,000 are already sitting unemployed inEthiopia's towns and cities, o f whom about half have beenjobless for more than 12 months. 0 The need to improve labour productivity so as to enable increases inreal earnings, since today evenhaving ajob is not necessarily a route out o fpoverty. Especially for those working informally in self-employment and as unpaid family workers-the needto enable these types o f activities to prosper i s urgent. 70 0 The need to improve the position o f women in the labour market. For the same characteristics (education, experience, etc.), women are paid 22 percent less than men and the unemployment rates, including for more educated women, are much higher than for men. The need for effective programs to cater for youth inthe labour market. These are needed to facilitate school-work transitions, as well as to tackle the stock o f unemployed. And all this is in the context o f the more general challenge o f enabling rapid growth in the economy-which needs to be o f the order o f 6 percent annually in order to meet the poverty MDG. 6.6 To address the central challenge o f increasing the pace o f job creation, the Government has a range o f potential instruments at its disposal-fiom the direct creation o f positions in the civil service and state owned enterprises on the one hand, through improvements in the business environment and investments in skills and people that are conducive to private sector investment and growth on the other. 6.7 The emphasis o f Government policy traditionally tended to focus on the public sector as the route to job creation. Eventhough the Labour Proclamation o f 1975, based on a socialist system, was amended in key respects in 1993, the public sector still dominates the urban labour market in important respects. Reforms such as abolishing recruitment solely through public employment offices and guaranteed employment inthe public sector for college and university graduates, easing the conditions allowing temporary employment, and widening the range o f cases where dismissal i s not unlawful over time have contributed to a better functioning labour market. 6.8 Reducing segmentation, where recent progress needs to be deepened and accelerated - will be key. Segmentation arises where institutional barriers - for example protection o f privileged sectors or firms-act as barriers to entry, but can also constrain overall growth o f demand (due to e.g. high costs). There is evidence that some parts o f the labour market are already well-integrated-in particular at the top end, which is in turn associated with the "brain drain" at senior levels o f the civil service-but overall segmentation across labour market states i s pervasive. 6.9 Inorder to develop a quantitative sense ofthe benefits ofreducing labour market restrictions a set o f labour-focused simulations has been designed with M A M S (Maquette for MDG Simulations) for the period 2005-2015 (see chapter 6, volume I1 for more details on the simulations). 34 M A M S i s a dynamic CGE (Computable General Equilibrium) model and relative to other CGE models it offers a detailed treatment of government services and o f the tax system and other sources o f government receipts. The labour market is disaggregated into three segments on the basis o f educational achievement (see Box 12). 34For technical documentation on MAMS see Lofgren and Diaz-Bonilla (2006a). For other applications o f M A M S to Ethiopia, see Lofgren andDiaz-Bonilla (2006b) and Sundberg and Lofgren (2006). 71 6.10 The simulations have been designed to explore the effects o f a more efficient allocation of labour across the economy and how those would affect other policies. The experiments include the gradual removal of barriers to mobility between urban and rural areas, and a decrease inthe segmentation inurban labour markets. Impacts are analyzed interms ofchanges inthe compositionof GDP and o f employment, as well as effects on growth and on welfare. Box 12: The MAMS Model MAMSis a dynamic CGE(Computable General Equilibrium)model designedfor economy-wide analysis o f the impact o f policies and external shocks on poverty and human development (HD) in developing countries. In the Ethiopian version, production is divided into 11 activities (agriculture, industry, and private services and 8 for different government service functions: education - first primary, secondprimary, secondary, and tertiary -health, water-sanitation, other public infrastructure, and other government). This classificationreflects data limitations and the fact that the key purpose behind the model design was to analyze the impact o f government policy and other economic developments onthe evolution of selectedMDGindicators. The labour market is disaggregated into three types on the basis of educational achievement - workers with less than completed secondary education (the bulk o f the labour force), workers with secondary education but not completed tertiary, and workers with completed tertiary education. The labour market is linked to the educational system. The part o f the populationthat i s enrolled in school i s not in the labour market. Among those not enrolled in schools, a fixed share enters the labour market at the time when their cohort would have entered the second primary cycle. This group consists o f those who either never attended school or had left school before starting the second primary cycle. Others leave the educational system at a higher level, with their educational achievement at the time o ftheir exit determiningtheir labour type. Inthe labour market, the default assumption (from which the various simulations deviate) is that the workers are free to move between the different production sectors. For any given labour type, the model captures observed relative wage differences between employment in different production activities. For example, workers at the lowest educational level in agriculture earn considerably less than if they worked in other activities. For each labour type, a flexible economy-wide wage clears the market in each time period, scaling wages in the different production activities up or down in a manner that retains relative wage differences. For each labour type, the unemployment rate i s fvted (a fixed share o f those in the labour force are not employed). These mean that we are implicitly assuming that the unemployment rates do not change. 6.11 W e find that the elimination o f barriers to the efficient functioning o f the labour market, all else equal, could lead in2015 to national output that i s 2.4 percent higher than inthe base case, with an increase inaverage welfare of about 2 percent. A conservative estimate o f the poverty impact is a 3 percent decline with respect to the base. Removing such barriers would also accelerate the transition out o f agriculture, with the urban economy becoming more dynamic and the expansion of all the non-government urban sectors. 6.12 An interesting aspect o f these simulations is that removing barriers in the labour market would result in increasing returns to secondary and tertiary education outside o f 72 agriculture. This i s good news in the medium term given plans o f accelerating investments in education, suggesting that new cohorts o f better educated workers would face more favorable labour market prospects. 6.13 As segmentation and the sluggish pace o f job creation result from a variety o f barriers, an agenda to tackle these challenges is correspondingly broad. This is laid out below on several fronts: enhancing the private sector contribution to employment, integrating vulnerable groups, creating an effective civil service, and enabling migration to contribute to growth and poverty reduction. Lastly, we will focus on monitoring and evaluation o f labour market developments. Enhancinpthe PrivateSector Contributionto Employment Improvements in the businessclimate 6.14 The most important route to reducing unemployment lies in increasing productivity and growth. A much better record on job creation in the private sector i s central to improved labour market outcomes in Ethiopia. Our analysis suggests that institutional and investment climate factors are very important in deterring entry into the private formal economy and that unless these are addressed, Ethiopia will make little progress inmeeting itsjobs challenge. 6.15 There i s no single magic bullet here, and nor is the agenda for jobs distinct from the more general policy reforms andmeasures needed to stimulate the private sector. The priorities which have emerged from empirical analysis, focus group and other qualitative work suggest a renewedprivate sector strategy needs to be established around three broad themes: building market-supporting institutions, revitalizing sector-level growth strategies, and strengthening economic governance. 6.16 Key elements therein are highlightedhere:35 0 Most important appear to be reducing the regulatory and policy obstacles to competition inkey sectors, such as wholesale and retail trade, logistics and incapital markets; 0 Effectively operationalizing the Trade Practices Commission and Proclamation, including extension of its purview to SOEs, and reviewing mergers and acquisitions that increase the power o f existing large firms; 0 Improving access to finance, so that firms need to rely less on collateral, and a wider range o f instruments, notably equity; Improving property access-through leases and/or ownership-to improve performance on property registration andleasing; and 0 Building on recent administrative reforms which have improved customs, tax administration and foreign direct investment. 6.17 One challenge for the Government is to identify measures to increase tax collections without creating a disincentive to private sector development. Cognizant o f 35The reader is referredto the CEM 2006 Volume 11,Chapter 3 for more. 73 the existence o f outdated and complex tax laws and weak and inefficient tax administration, a tax reform program has been underway for some time, which got renewed momentum in 2001 when the "Revenue Board" was scaled up to become the MinistryofRevenue. The 2005 reforms to the presumptive tax (as mentionedinChapter 3) have been welcomed by firms. Recent work suggests that higher tax collections could be realized through limiting exemptions on imports and rebalancing incentives to investment to accelerated depreciation. 6.18 There appears to be concrete interest on the part o f Government in ensuring positive impacts o f the tax system andtax reforms on investment and growth, saving, and the distribution o f income. Advice and assistance is already being provided by the IMF. On a complementary front, in order to ground a better understanding o f these issues, it may be usehl to undertake more detailed empirical micro-analysis on the conduct o f individuals and households, and both private and state firms, across different sectors in time periods spanning past reforms; and estimate behavioral responses o f economic agents, and calculate associated revenue, efficiency, and distributional effects o f alternative options. This would include, for example, calculating effective marginal tax rates on different agents (households, firms, in different locations, etc.) and in different sectors, and other important parameters such as the cost o f capital to firms. This would help inform understanding o f the distortionary costs o f increased revenue mobilization. Onthis basis, actions could be recommended to improve the efficiency and effectiveness o fthe tax system v i s - h i s private sector growth and employment objectives. 6.19 While the foregoing has mainly to do with the formal sector, these also have a bearing on the success o f informal sector activities. There seem to be two broad sets o f constraints holding down wages and productivity growth in the informal sector. The first, "lack o f demand," is obviously related to low average incomes and extensive poverty (lack o f purchasing power), but also partly reflects the high degree o f competition among relatively undifferentiated goods and services. The second set o f constraints comes from government actions and/or regulations. 6.20 Constraints directly related to government actions/regulations are easier to address inthe short term. There are three main avenues where action could have a sizable payoff: ensuring that informal sector operators can conduct their activities free from official harassment and from arbitraryregulations and/or payments; expanding access to microfinance by facilitating private sector activity therein; and improving access to workplaces - security o f tenure reduces risks and enables informal operators to invest. 6.21 We review each o f these avenues inturn, bearing inmindthe important additional point that emerged from the analysis, that actions and programs responding to these constraints should have a gender-sensitive lens. Further, the role that informal activities have insustaining the poor and vulnerable should be kept inmind: though frequently low productivity and non-dynamic, these activities inurbanareas can at least enable sufficient food intake. Indeed together with low skills, the inability to start-up self-employment 74 enterprises was identified by the PPA as one o f the two defining features of vulnerability inurbanareas. 6.22 Based on the size and significance o f the informal sector in Ethiopia, it i s clear that attempts to repress activity will have adverse impacts on employment prospects in aggregate and at the individual level. The alternative to informal sector activity is not a "proper job" inthe formal sector, but unemployment. 6.23 Freedom to operate. On this point there seems to be a disconnect between official government policy, which recognizes the key role o f the informal sector and SMEs inthe economy, and the actions o f some local officials. Results from the UNECA governance study (2004) also report a perception (inthis case by a panelo f local experts) that the government does not sufficiently support the informal sector. The PPA and anecdotal sources show that petty traders are subject to occasional harassment by officials, although the extent o f such practices is unclear. Inthe survey data, larger and more established informal firms are more likely to report government harassment and regulations as an issue. 6.24 Existing regulations which either deliberately discriminate against informal operators and/or are subject to interpretations which have this impact, are urban zoning regulations. Changes in urbanland use and inparticular on the location o f markets have been reported as adversely affecting the activities o f the small scale self-employed whose low margins make them less able to face increasing costs to access inputs. A more general review o f existing regulations and practices i s needed to inform the design of appropriate reforms. Moreover, Government needs to ensure that officials at all levels are aware o fhow they should treat informal sector firms, and make sure that they do so. 6.25 Microfinance. Lack o f capital surfaces in every survey and inqualitative sources as a major, or even the most important, constraint to would-be and current entrepreneurs. Overall only a small percentage o f informal sector operators rely on formal loans for capital; most draw resources from their own savings, or from friends and relatives. However the microfinance industry i s rapidly expanding, and seems to be having a positive impact not only in terms o f access to credit, but to other financial services (e.g. savings accounts). There are currently a total o f 25 licensed microfinance institutions in Ethiopia, operating in all but 3 regions, with funding coming from own as well as donor resources. According to the Association o f Ethiopian Micro Finance Institutions (AEMFI), the number o f loan clients doubled between 2001 and 2004 to over one million. 6.26 Regulations governing the Ethiopian microfinance industry have been relaxed (including on loan size and interest rates). Yet it does appear that further changes could spark private investment in the sector. According to the National Bank o f Ethiopia existing MFIs satisfy less than 10 percent o f demand for microfinance. MFIs need to deepen as well as broaden their reach, since many small and growing businesses need slightly larger loans than MFIs typically provide, but cannot access commercial credit. Frequently these borrowers would prefer individual loans to dealing with the time 75 commitment and risk involved with group-based loans. There are also concerns at the other end o f the spectrum, since the very poor may not be able to gain acceptance into a group for group-based lending. Further analysis and evaluation on the impact o f existing loans would usefblly inform policy development in this area. Moreover, expanding the reach o f MFIs in towns could be important, since they operate predominantly in rural areas. 6.27 Access to workplaces. Both the formal and informal components o f the private sector have identified the availability o f land, and the associated time and cost o f land related transactions, as one o f the key constraints to growth. With the new Urban Land Lease proclamation issued in 2002, it was expected that there would be significant changes inregionaycity level practices with regardto urbanlandmanagement. However, recent studies indicate that the volume o f informal land transactions is high; leased land plots are actually cheaper than "rent" land and buildings; taxation o f real estate transactions i s high; the allocation o f vacant land is made without auction, for low administratively defined prices or free o f charge; and there are huge differences in the amounts o f payments for similar plots. In this climate, informal sector operators-who have limited capital and at best a weak political voice-would be expected to experience great difficulties. Although there are targeted interventions (e.g. via the Addis Ababa ReMSEDA, see below) that aim to help entrepreneurs find working premises, this i s a second-best (and limited) solution-ideally the legal framework would be more conducive to improved access to secure tenure in general. The upcoming World Bank study on urbanization will look further into these issues and provide specific recommendations. 6.28 Finally, there is the question o f why informal firms do not obtain a license, and whether they consider being unlicensed a problem. The Addis Labour Market Survey found that most firms that did not have a license cited "easier to operate without a license", or "did not need a license" as reasons. A minority cited reasons related to the license process itself. Avoiding taxation i s often a reason to avoid formal status; indeed, a small but significant percentage o f firms in the ALMS (12 percent) agreed with the statement "I don't want my business to grow too much because then Iwill have to pay taxes" (2006 survey). However, there are benefits to formalization-avoiding harassment, access to credit and so on-that these operators are missing out on. Business SupportPrograms 6.29 Beyond removing constraints, governments can also take active steps to promote individual forays into business. Because most Ethiopians who manage to transit out o f unemployment find work in the informal sector, programs aimed at supporting entrepreneurship, and in particular access to finance, could have an impact. While not everyone has the personal qualities and aspirations to be a successful entrepreneur, and the overall business climate influences whether new businesses can succeed, there is some international experience about the types o f business development services (BDS) which work. 76 6.30 International evidence shows that programs to support business development are more successful when they address multiple facets o f the challenge, andnotjust access to finance. O'Higgins (2001) lists the following typical components o fprograms intendedto enable self-employment, and reports that the best results come from programs which cover all (though not via a single organization-microfinance itself, for example, is best providedby a specialized institution and not by an educational organization): 0 Promoting andintroducing the self-employment option; 0 Training inthe necessary skills for self-employment; Mentor support; 0 Access to finance; 0 Access to work space; 0 Business expansion support; and 0 Access to support networks. 6.31 The Government's Micro and Small Enterprises Development Strategy, proclaimed in 1997, recognizes that the MSE sector i s held back by "various policy, structural, and institutional related problems and bottlenecks," including the key constraints noted here (marketing problems, lack o f capital, lack o f working premises, etc.). The strategy led to the creation o f the Federal Micro and Small Enterprises Development Agency (FeMSEDA), which also has Regional bureaus (ReMSEDAs), under the Ministry o f Trade and Industry umbrella. These are the main public sector providers o f training and business development services for MSEs, and there are also numerous private skills/management training institutes and organizations. Services vary widely, with many specialized skills training schools as well as more comprehensive centers. The Addis Ababa ReMSEDA, for example, provides a one-stop shop for entrepreneurs, offering training, microfinance, assistance with registering and accessing workplaces and equipment. Thus far it has served about 200,000 clients and created 150,000 jobs. 6.32 The Ethio-German Micro and Small Enterprises Development Program accounts for much o f the impetus and resources behind BDS in recent years, via public, private, andNGOproviders. An impact study conducted in2004 lookedat 45 businesses that had participated in a BDS program, and found that 23 had subsequently increased their income and 12 had hired new workers. In particular, BDS seemed to improve entrepreneurs' marketing skills, and to help them begin keeping records o f business transactions, inaddition to facilitating access to credit andworkplaces (Zerihun 2004). 6.33 There appears to be significant m e t demandfor training andBDS. About three- fourths o f informal operators surveyed would be willing to pay for some or all o f the cost to attend a training program (Ageba and Amha 2004). The ALMS probed into this issue, and found that about three-fourths o f small business operators had never sought help in runningtheir business. The majority said this was becausethey didnot know where to go for help. Likewise, most o f the unemployed said they were unaware o f any government program inAddis Ababa that could help them find ajob or start a business. This suggests that simply better publicizing the services available (as was recognized in the government's MSE Development Strategy) could make an initial difference. 77 6.34 Expanding the existing types o f interventions also appears to be appropriate, alongside strengthening program monitoring and evaluation. Any significant scale up should be based on careful evaluation o f the evidence to identify what works best and what needs to be adapted in extending existing programs. At a minimum, more systematic monitoring and periodic impact evaluations are needed to assess the effectiveness o f the Government's existingprograms inorder to gauge the value o f future scaling-up. An analysis o fthe employment creation impact o fthe Addis Ababa Integrated Housing Program, which integrates employment creation through support o f SMEs in construction woodwork and metal work, is being undertaken by the World Bank i s following to this report, to contribute to these efforts. IntematinPVulnerable Grows 6.35 Women and young people experience relatively greater difficulties in securing productive employment than adult men, which suggests a need for targeted active programs need to be designed. The fact that women have lower levels o f educational attainment and lower access to capital and assets than men creates disparities that could be addressed through program design. 6.36 A specific obstacle for women which is more intractable lies in their domestic work burden. This is vividly illustrated by the time-use component o f the ALMS, where over two-thirds o f unemployed women said their most time consuming activity in the previous week was domestic responsibilities, while only 9 percent o f men said this was the case. In contrast, more than half o f unemployed men said their primary activity had been social (e.g. meeting up with friends, recreational activities, etc.). Thus women's employment options appear constrained by the need to keep up with chores andtake care o f children. Indeed, most female informal sector operators are home-based and with limited prospects for these reasons. Policy and programmatic interventions should be designed with these concerns inmind. 6.37 Targeting women, as MFIs tend to do internationally also serves to combat gender inequity. For example, evidence on female microfinance borrowers in Bangladesh, where MFIs are well established, indicates significant poverty reduction impacts, as well as better human development outcomes for their children (Littlefield, Morduch, and Hashemi, 2003). In Ethiopia, the percentage of loans held by women varies widely. Some MFIs specifically target women, and women account for 60 percent or more o f the lending portfolio, while other MFIs lend well under 40 percent to women. Increasing efforts to reach women in areas where they appear to be underserved could help reduce gender inequality. 6.38 Reaching youth, who have higher levels o f educational attainment than their predecessors but obviously less experience, entails a different set o f considerations. The consensus in the literature seems to be that preventative interventions (such as keeping children in school longer) are more effective than ex-post interventions such as specific training programs. In the case o f Ethiopia, however, where the human capital base i s 78 extremely low and a largepart o f the population has already missed the chance for formal schooling, other programs are likely to be necessary. Ensuring appropriate targeting will partly depend on the availability o f relevant labour market information (see vol. 11, Chapter 8). 6.39 Though the informal sector provides the main route to employment for most, this often means informal wage work (or unpaid work) rather than self-employment. However, given the limited prospects in the informal sector, programs targeting youth need to be broader than facilitating entrepreneurship, perhaps viajoint initiatives with the private sector to increase opportunities for on-the-job training that enable young people to gain workplace-relevant skills. 6.40 Recognizing the need for a multi-sectoral approach to empowering youth and equipping with the tools for success, the government created the Ministryo f Youth and Sports, which formulated its guiding National Youth Policy in 2004. Currently, the Ministryis working to finalize its strategy to implementthis Policy, which includes plans for interventions in improving healthcare for youth, expanding access to business development support and microfinance, creating new training opportunities, and providing care for vulnerable young people, among other objectives. These will be coordinated with existing services for youth, which mainly fall under the Ministry o f Education (which administers the formal education system) and Ministry o f Trade and Industry(which oversees FeMSEDA). 6.41 Many o f the BDS and training providers referenced above serve a relatively young clientele. The Addis Ababa ReMSEDA in particular specifically targets unemployed youth and women, and informal educational opportunities naturally appeal to new or recent entrants to the labour force. According to the FeMSEDA training institute, which mainly enrolls those between 20-30 years, youth are more likely to take courses in specific skills (e.g. woodworking, metalworking, handicrafts), while the older students are more interested in management and business development courses. Assessing patterns o f demand for training would be help to inform the expansion o f current programs. 6.42 The most visible government effort to improve youth employability lies in the creation o f a formal technical and vocational education and training (TVET) program. The TVET program aims to address concerns that many secondary graduates who do not continue to higher education lack practical skills, and a pilot tracer study o f TVET i s currently underway to determine ifthe new system is effective. The Government plans to continue its massive expansion o f TVET, to eventually accommodate most Grade 10 graduates, while at the same time the size o f upcoming cohorts is increasing (nationwide, there were about 30,000 first level TVET students in government and non-government programs in 2004/05, and nearly 300,000 grade 9 students; MOE 2005). The World Bank's upcoming secondary education study will closely examine issues related to the current program and scaling up. 79 6.43 In addition to lack of a proper education and lack of experience, youth respondents to the ALMS were more likely than older jobseekers to cite a lack o f influential contacts as the main obstacle to getting a desirable job. Policies to strengthen merit-based selection for jobs inthe public sector may warrant consideration, andpolicies aimed at promoting competition in the private sector in general (as discussed above) should indirectly have this effect as well. Active EmploymentPrograms 6.44 A potential set o f tools to tackle unemployment includes job creation and job subsidies, as well as training (Box 13). However, the sheer scale o f Ethiopia's unemployment problem, vis-a-vis government budget constraints, means that active labour market programs alone cannot solve the problem, especially given their other limitations. Low demand for labour in general constrains the potential o f interventions on the supply side (e.g. training and business development support), although these do have an important role to play. Interventions which specifically create jobs such as public works programs, can usefully temporarily raise incomes, but tend not to increase the long-run employability o f the worker. 6.45 Active interventions, while no substitute for dynamic private sector job creation, can help alleviate the negative effects o f unemployment for the most vulnerable. In Ethiopia, longer durations o funemployment are found for those who are looking for their first job, which means that individuals (especially youth) risk being lock out o f productive opportunities. Most interventions tend to be on the supply side, via literacy and vocational skills training programs, for example, and there is some evidence that such programs may have a greater impact in developing than developed countries (where their effectiveness has been questioned). There are a variety o f ways to package this kind o f assistancethat could be considered for Ethiopia, andtailored to fit the local context. In Kenya, for example, the Jua Kali Voucher Program distributes training vouchers to entrepreneurs or employees o f SMEs to allow them to boost their skills. Although the program has not been rigorously evaluated, it is believed to be achieving many o f its objectives, including increasing the amount o f training accessible to informal sector workers and increasing competition among training providers (Adams 1997). While voucher programs also have critics (see Gage12005 for arguments against), at least one useful lesson can be drawn from the Kenyan program-that such projects are better run by the private sector than the government, and that exit strategies are needed so that subsidies do not have to continue indefinitely (Godfrey 2003). 6.46 Labour market interventions can impact both the labour supply (e.g. training) and demand (e.g. improving the investment climate; business development support; public works programs) sides, ideally in mutually reinforcing ways. For example, training programs work better when there are close links with the private sector, since employers have some knowledge o f the kinds o f qualifications that are in demand. Especially in light o femerging shortages inspecific sectors and for certainprofessionprofiles, it would be important to have an integrated approach to vocational training, starting with a bottom-up approach that links firms' needs with the type o f educational system that 80 would best support the creation o f those skills. Singapore offers a very successful example o f integrated management o f a vocational training system, and after having developed it in collaboration with donors, has graduated from receiving foreign assistance. Today it pursues strategic partnerships whereby private firms outsource part o f their production processes to firms inpoorer countries such as Vietnam and help them builduptheir skills. 6.47 Much o f the evidence about the impacts o f ALMPs comes from industrial, rather than developing countries. In countries where the informal sector dominates, implementation capacity is low, and demand for labour i s relatively weak, the ability o f ALMPs to make a major impact on employment and wages is limited. Available program evaluations tend to be less encouraging than those in OECD and transition countries, but the results vary, and there are examples o f youth training programs that have been more successful in developing than industrialized countries. This may arise because, for example, such programs are addressing skilled labour constraints (Betcherman, Olivas, andDar, 2004). 6.48 The factors leading to successful ALMPs tend to apply for all countries (Betcherman, Olivas, and Dar, 2004). Interventions that offer a combination o f services, are demand-oriented and relevant to the workplace, and are well-targeted generally work well. However the evaluations underline the fact that program impacts are usually more positive when the economy is growing, and cannot effectively offset lack o f growth. It i s important to ensure that programs are well targeted, and that vulnerable groups receive special attention (e.g. the disabled and orphaned). 6.49 There may be a role for enhanced employment exchange services to improve the flow o f information in Ethiopia, but because' most hiring tends to be via informal networks the impact i s unlikely to be large. International evidence has suggested that employment services are most beneficial when integrated with career counseling and other support, which is in line with MOLSA's plans to expand its services. Still, the impact i s generally limited, with only a minority o f firms listing their job openings and the absolute numbers o fjob-seekers assisted beingvery small relative to the stocks o f and flows into unemployment. 6.50 Better educated job seekers tend to be more likely to register with employment services-this is certainly true in Ethiopia, where according to MOLSA 82 percent o f those registered had at least junior secondary education in 2001/02. This suggests employment services are probably not a good tool for reaching the unskilled unemployed. Moreover, MOLSA data suggests a mismatch between the registered jobseekers and the postedjobs, since many vacancies remained unfilled for extended periods o f time-this could be related to skills, or perhaps to expectations (e.g. if skilled workers are uninterested injobs located outside major towns, which is supported by some anecdotal evidence). 81 Box 13: Are Active Labour Market Policies (ALMPs) Worthwhile? while the debate over the effectiveness of AMPs is not conclusive, a recent review of 159impact evaluation studies in both developing and industrial countries does draw general conclusions about which interventions work better than others, which can be highlighted asfollows: Employment services include counseling, placement assistance, job matching, labour exchanges, and other related services. They generally have positive impacts on the post-program employment andearnings of participants. Costs are relatively low so the cost-benefit ratio i s often favorable. However, employment services per se are o f limited use insituations where structural unemployment i s high and there i s a lack of demand for labour. The coverage and effectiveness o f these services in developing countries i s less in countries where many labour market transactions are informal. Trainingfor the unemployed. Participants often benefit from these programs interms o f higher employment rates but not in terms o f higher earnings. The few evaluations in developing countries paint a less favorable picture. Programs seemto work best with on-the-job training and active employer involvement. Results are more positive for women than men. Retrainingfor workers in mass layofs most often have no positive impacts. The few successful cases typically include a comprehensive package o f employment services to accompany the retraining, but these are generally expensive. Trainingfor youth. Inindustrial countries, these programs often fail to improve labour market outcomes, and the conclusionhas been that earlier investments inthe education system to reduce drop-outs and other schooling problems are more effective. While there are few studies in developing countries, evaluations in Latin America do find positive impacts for programs that integrate training withremedial education, job searchassistance, and social services. Wage/employment subsidies. These often do not have a positive impact and have substantial deadweight and substitution costs. Targeting and monitoring may help but at the cost o freducing take-up rates. Public works. This can be an effective short-term safety net but public works do not improve future labour marketprospects for participants. Micro-enterprise development/self-employment assistance. There i s some evidence of positive impacts for older andbetter-educated workers. However, take-upi s low. Source: Excerpted from Betcherman, Gordon, Karina Olivas, and Amit Dar (2004). 6.51 Alongside the state employment service, there are also private sector brokers. Each works with a network o f 5-10 job dealers who are in charge o f recruiting. Some of them also offer shelter to job searchers who don't have it. They obtain fees from both the employee and the employer, partly depending on wages. It i s anecdotally reported that wages can vary a lot by neighbourhood and that wages depend not only on the job descriptionbut the employer's ability to pay. There are concerns for migrants, especially female migrants, who are vulnerable to exploitation in the absence o f their own social and other networks. These operators seem to perform a useful hnction for those who lack information about the jobs market and/or do not have networks, but might warrant some monitoring and possibly regulation. 6.52 Public works programs can serve as a safety net for the most vulnerable, and alleviate unemployment, even if they do not guarantee better employment outcomes for participants in the long run. They can also support economic growth by building/maintaining community assets. Some o f the other advantages o f public works programs include their attractiveness to informal sector workers and the relative ease o f 82 targeting, fairly low administrative burden (compared to other come support schemes), and ability to provide quick relief after shocks (Vodopivec and Raju2002). 6.53 One successful public works program in Senegal employed youth in labour- intensive construction and public service provisioning. By sub-contracting the work to SMEs, the program created 80,000 jobs in SMEs in its first phase, including 6,000 permanent jobs (which i s atypical for a public works program). The effectiveness o f the program was attributed to "clear objectives, strong political support, and assistance from development partners" (Sarr 2000, cited inUNECA 2005, p. 26). By integrating training programs with public works employment schemes, youth-oriented programs inparticular might increase the likelihood o f having a longer-term impact - one criticism o f the Senegal program was the lack o f such a component. 6.54 Ethiopia has extensive experience with food-for-work, and more recently broad- based cash-for-work schemes (via the Productive Safety Nets program), but only inrural areas. Adapting and expanding the concept for urban areas should be on the policy agenda, and may be the best mechanismto directly expand opportunities for the unskilled unemployed. It is important to underscore that although unemployment rates are higher among the relatively well-educated, in absolute numbers the unskilled unemployed constitute a very large group (and experience an unemployment rate that i s still high relative to other countries). If linked with training programs, public works interventions may have a better chance o f improving participants' employability in the future, rather than simply offering a temporary source of income. Moreover, along with generating employment, public works programs can help expand and maintain critical infrastructure and public assets in large cities as well as towns. This dovetails nicely with the Government's increasing focus on urban development as a central part o f its growth and poverty eradication strategy. 6.55 An analysis of Ethiopia's existing food security programs by Quisumbing and Yohannes (2004) showed that the programs tend to mirror patterns in the labour market more generally, i.e. with lower female participation and earnings. In order to improve outcomes for women, they recommended careful selection o f the type o f project (men are more likely to work on infrastructure projects, for example) and allowing flexible schedules/providing day care to enable more female participation. They also emphasize the importance o f female involvement in the program planning stages as critical to achieving benefits for women. Conditions for an Effective Civil Service 6.56 The policy implications o f current remuneration and hiringpractices are two fold (see Chapter 3, this volume): 0 that relatively good conditions inthe civil service for the low skilled could distort the lower end o f the labour market, make private firms less competitive andpossibly also encourage "wait" unemployment; and 83 0 the need to respond to a visible brain drain problem at the high end which, while not quantified in extent, appears to be undermining government capacity at the senior policy level. 6.57 Anecdotal evidence suggests that attrition rates are high among skilled personnel and at the high levels o f the structure (department head level and above), although aggregate turnover i s not high (4 percent annually). There i s intense demand for professionals with a masters qualification (plus 5-15 years o f experience) in economics, management, public health and others depending on the specific demand o f the employers, which constitute about 5 percent o f the total civil service employment. 6.58 The data confirm that the inequality inwages across different types o f employers is less variable for the high skilled relative to the low skilled. This suggests that the market for relatively scarce highskills is more integrated (there is a "going rate"), which would be expected to encourage individual mobility inresponse to earnings differentials. 6.59 A recent comparison o f civil service salary scales with those o f parastatals, the private sector and NGOs though small in sample size and not representative suggested that highskilled workers can findhigher earnings outside the civil service. 6.60 Table 18 summarizes the ranges by qualification. For example, parastatals pay 1.3 to 3.8 times as much as the civil service for social science graduates. The private sector also pays more for similar qualifications and experience. NGOs and international organizations, who generally recruit graduates, offer by far the best pay-although obviously the numbers o fjobs in those agencies are limited inboth absolute and relative terms. Table 18: Ratio of Salaries for Selected Professions Relative to Civil Service, 2005 Parastatals Minimum Maximum Social science 1:1.3 1:3.8 Law and engineering: 1:2.4 1:2.9 Private sector Social science 1:1.6 1:1.7 Law and engineering 1:1.6 1:3.3 Medicine 1:2.6 1:1:42 Teaching 1:1.9 1:8.3 National NGOs Social science 1:2.5 1:5.5 Law and engineering 1:3.0 1:4.2 Medicine 1:2.9 1:3.3 International NGOs Social science 15.8 1:10.8 Law and engineering 15.4 1:9.5 Medicine 1:9.1 1:9.7 Source: Mussa (2005). 6.61 Somewhat surprisingly, even for non-salary benefits (allowances and medical insurance), civil service conditions for professionals are also relatively poor. Allowances, per diem and medicalhnsurance coverage are more attractive in the non- 84 public formal sectors than the civil service, at least for professionals. For example, unlike the civil service, some international organizations revise the salary scales to reflect changes inthe cost o f living. 6.62 Civil service reforms, together with new ways o f managing human resources, including performance evaluation and more accurate (and less inflated) definitions o f grades and tasks, will be needed to meet the challenges ahead. Some o f these reforms cannot be implemented rapidly, but other countries' experience has shown that disaggregating pay and employment reforms and sequencing over government agencies can be an effective strategy. 6.63 While the situation may seem to call for across-the-board decompression and rationalization strategies, such approaches have been increasingly criticized across Africa. The current trend based on international experience is to disaggregate pay and employment reforms and sequence them sectorally (agency-by-agency reform) andor by cadre (i.e., creating a senior executive service). 6.64 The stated goal o f the Government i s to reform more deeply the system and make the civil service performance based - however this i s a long term program that will take some years to come to fruition. The current system is based on seniority rather than merit, so that not all those at higher levels warrant higher pay or efforts to retain, and over time younger people need to be attracted and given incentives to excel. As we showed inChapter 3, it appears difficult to lay off civil servants. 6.65 It is nonetheless possible inthe short term to introduce simple, focused measures that could help to retain the key staff who have beenidentified as essential to senior level management and policy functions. At the high end, for the GOE to keep up with the private sector/NGOs/donors (PND), the approach could be as follows: The subset o f GOE senior cadres that are qualified for the PND market should be identified. A simple historical analysis o f the qualifications and characteristics o f GOE staff recently hired by the PND will reveal precisely the constellation o f skills, education and experience that qualify for PND employment (typically heads o f department or equivalent, usually with a Masters degree, and at least ten years o f experience). 0 An anonymous survey o f the terms o f employment given to the former GOE employees, to inform the design o f a competitive package. Develop and introduce a package o f incentives for the select group, and monitor implementation. 6.66 Box 14 provides an example o fhow donors inTanzania are collaborating with the Government to ensure that over the longer term efforts to retain and build senior cadres are sustained. At the same time, as stressed in Chapter 3, the issue o f compensation is broader than pay, and involves life choices, incentives and staff evaluation, training opportunities andlocation incentives, as well as societal andpolitical factors. 85 Box 14: Donors' Support of the TanzaniaMedium Term PayReform Strategy: the Selective Accelerationof SalaryEnhancementprogram(SASE) SASE is the key intervention introduced by the Government of Tanzania to leverage implementation o f its Medium-Term Pay Reform Strategy (MTPRS), adopted in 2000. SASE targets priority-post holders, essentially those inmanagerial, professional and technical positions and aims to commit, motivate and retain critical personnel to introduce and sustain a results- oriented culture in the public service. Salary scales under SASE are pegged to target levels contained in MTPRS. The intention i s that in the medium-term, donor partners will on a declining sliding scale, finance the gap between current andtarget salary levels. At the beginning ofeachyear, donor partners disburse the annual amounts pledgedto SASE(that are within MTEF), to the Consolidated Fund. Resources are allocated to the various Ministries Departments and Agencies (MDA)based on objective and transparent criteria for the selection of candidates. At the end of each year, individuals benefiting from SASE undergo a rigorous assessment procedure and each Ministry reports (using independent sources o f information) on the impact of the scheme, as well as whether those who benefited from the scheme ought to continue being included. The scheme i s at an early stage inits implementation. Currently, just over 1,000 employees in4 MDAs benefit from SASE. Inthe medium-term, a tenfold increase in the numbers on SASE is planned, though the total actual numbers will be reviewed once resource ceilings have been agreedwith the MinistryofFinance andparticipating donor partners. Recent progress reports highlight some difficulties in implementingsuch a scheme. These have included unrealistic planning horizons, the slow rollout with only 4 out o f the 11MDAs targeted benefiting from the scheme, and the "projectization" o f SASE, with some donors targeting particular MDAs. Donors' willingness to support the scheme has also been weakened by the emergence o f gaps between actual pay and target pay (particularly for higher grades), as public employment grew unexpectedly. The scheme is voluntary and despite the availability o f systems and tools provided by the Public Service Commission the management culture within Ministries remains weak. Further, compliance with safeguards procedures i s very time consuming for managers. Source: adapted from hthx//www.estabs,ao.tz/sase.uhp; progress reports on the overall Public Service Reform Programcan be downloaded at http://www.estabs.go.tz/progress-report.php. 6.67 Some recent research has focused on turnover o f staff in service delivery. Very highturn-over rates have beenfound amongteachers inthe first cycle ofprimaryschools, especially in remote areas in which communities contribute not only to school construction, but also to the hiring and paying o f teachers. Because community salaries are not guaranteed, teachers, after gaining some experience, look for alternative job opportunities. On the other hand, some anecdotal evidence on health workers now suggests that the private sector may be saturated and "losing its appeal," and some health professionals may be movingback to the public sector (Serneels and Lindelow 2004). 6.68 The future policy challenges faced by the Government in a context o f scaling up to reach the MDGs are clearly those o f increasing demand for staff, and for staff with higher skills, coupled with the competitionby the private sectors and donors which raises the premium on those very skills that are so scarce. At the same time, as explored in 86 Chapter 3, it is important to ensure that relative good conditions for the average and unskilled worker do not unduly push up private sector costs and reduce Ethiopia's competitiveness. EducationandTraining 6.69 Our study has shown that returns to education inEthiopia are highand increase at higher levels. Relative to illiterates, men with some higher education earn 132 percent more, and for women the premium is even higher at 147 percent. Interestingly, however, returns from non-formal and grade 1-4 education are not statistically different from illiterate in Addis: suggesting that more than minimal education is needed to make a difference to earnings inthe capital. 6.70 Increasing educational attainment o f the Ethiopian population and workforce has had important effects on the labour market. On the one hand, it partly explains the lower activity rates o f youth (who are still in school), although completed general and tertiary education does tend to make individuals active inthe labour market. One welcome trend also is that the high education gaps between men and women have been narrowing over time. 6.71 We also found largereturns to training both interms o f earnings, as well as higher activity and employment rates.36 Only 11percent o f those with training are inactive, and those with training represent about 21 percent o f the active population. At the same time, much o f the training that we observe has been provided to civil servants-about half o f whom have training, against 19 percent o f those working for the private sector and only 4 percent of parastatal workers. The distribution o f those employed with training by sector (primary, manufacturing etc.) likewise shows a predominance o f the public sector (out o f the entire workforce that has received training, 20 percent are public sector workers) and the educational sector (26 percent), and a spread across other sectors. At the same time, looking at the occupational distribution, reveals that the groups most likely to have training are those with high skilled occupation (as many as 90 percent o f professionals have training, against 6 percent o f those inelementary occupations). 6.72 A key question on training relates to relevance to private sector demands. We do findthat the private sector laments lack o f specific skills (supply chain study). Follow-up work being supported by the World Bank will help to explore how the supply chain from basic education to training can be shaped to respondto private sector needs. 6.73 It should be noted that in our sample most (three out o f four) o f the individuals who had received training also had at least general education - thus the results here may not be replicated with the expansion o f training through people being streamed into TVET muchearlier. 36Notethat the data do not allow us to distinguishon the job training (likely to be provided by better employers) fromtrainingwhichprecededthe entry into the labor market. 87 EnablinpMimation to Contributeto Growthand Poverty Reduction 6.74 The contribution o f migration to growth and poverty reduction is strengthened by an overall labour market policy that enables dynamism, and by the overall coherence o f the policy framework. This involves understanding the diverse and complex strategies that people, especially the poor, use to diversify their income sources and reduce risk. Thinking about migration (as more generally about population) needs to be integrated into the planning o f different ministries. Both the rural and the urban development agenda need to consider migration more explicitly. 6.75 In the case of rural development, migration needs to be seen in a context of overall income diversification, favoring at least shorter range migration and connectivity with towns. Several policy elements currently pursued are in line with these objectives including strengthening the management, capacity and planning o f rural towns, which play a key role inproviding services and access to markets for the rural population; and scalingup the provision o frural infrastructure such as roads and electricity. 6.76 Urban authorities, faced with the prospect o f a growing population, need to consider how to create an enabling environment which can turn the challenges posed by increased mobility into opportunities. This involves developing new alternatives for service delivery, providing urbaninfrastructure and considering access to land and credit bythenewcomers. 6.77 Specific policies to be considered with respect to migration include: Facilitating flows of information and people. Strengthening rural-urban linkages through improved information flows on jobs and skills required, better transport facilities that can allow easier movement particularly for migrants that maintain links with their areas o forigin can have a strong impact on their development. Developing migrant-friendly financial services i s an important element in favoring the diffusion o f the benefits o f migration-this could be done by NGOs but also by the private sector as intermediatingremittances can be profitable. Deregulation. Inthe short runa priority is addressing the exclusion o f migrants from goods and services that require registration-this could involve temporary identifications like those now issued in China. Temporary identification could also help reduce the need for a local guarantor required to access certain jobs (typically jobs that involve dealing with property), which hinders the employment prospects o f those migrants who do not possess strong linkages with the non-resident population. Women and children are particularly vulnerable to the denial o f service provision and protection, particularly ifthey do not have strong support networks inthe host area. Civil society. The involvement o fNGOs indealing with migrants has been successful elsewhere (e.g. state o f Rajastan in India) for example in setting up resource centers that provide information on job availability, wage rates and rights, or in providing skills and help in accessing better jobs. In another example, again from India, an NGO has worked closely with local governments o f source villages to develop an 88 informal system o f identity cards for migrants, which gives them some protection against official harassment (Deshingkar 2005). Urban safety nets. This is an area still to be explored, but to the extent that those who migrate under duress have no alternatives they will be present in cities and will require targeted measures, particularly if they have nowhere to return to and have no assets. Overall urban policy framework. Over time a system that facilitates flows to areas whose development is supported by other policies (e.g. secondary cities as part o f a rebalancing of the urban structure) can be put in place. Coherence o f overall policy framework and appropriate resource flows within a decentralized context are very important inthis respect. Monitorinp and Evaluation 6.78 Progress on policy actions and developments inthe labour markets, particularly in urban areas for which data collection is more frequent, need to be part o f the PASDEP monitoring and evaluation framework. Indeed including employment creation as a strategic priority has been an important innovation and an opportunity to enhance the value and policy impact of labour market information. Monitoring and reporting on PASDEP should not substitute but build upon an integrated Labour Market Information System (vol. 11, chapter 7). 6.79 Collaboration o f all the stakeholders interested in labour market policy both within the Government and outside will be needed to optimize the significant efforts in terms o f data collection and reporting that Ethiopia undertakes, andto maximize the gains in feeding into effective policy making. Such collaboration will have to take place at various levels along the spectrum from data production to data analysis andreporting and evaluation o f specific program outcomes. It should therefore involve active participation in producer-user consultations through to access to data in its various forms and in consultation and collaboration on the analysis. 6.80 The design and updating o f PASDEP's monitoring matrix and the identification o f target indicators provide the starting point for this monitoring process. It will be important that the indicators chosen for the monitoring are clear and o f easy interpretation, and that they can be disaggregated by age group and gender to focus on priority groups. Indicators should also be prioritized, to the need to keep the number o f indicators manageable. 6.81 The consultation process around this study has highlighted a variety o f opinions on appropriate indicators for monitoring the labour market policy side o f PASDEP. Internationalevidence suggests two useful criteria to consider inidentifyingindicators: The indicators chosen should offer a clear sense o f progress. Current practices o f monitoring employment for the age group 10+ might not fulfill this criterion, as enrolment i s increasing. Complementing this indicator with an indicator for the age 89 group 15+ would help disentangling the effects o f increased enrollments from those o f labour market developments such asjob creation.37 0 The indicators chosen should be clearly linked to policy. For example, while unemployment figures can steal the headlines, trends in inactivity may be an equally important policy concern that risks neglect. 6.82 Important indicators that fulfill these criteria are the employment rates for the individuals 15 and over and the wage levels o f the employed. Reporting regularly on these indicators, in total and disaggregating by gender, age group, skills, and region (including min and max) would provide a good summary o f how the country i s using its productive potential. 6.83 Reporting on even such a minimal set o f indicators could prove however to be a challenge. As part of ongoing efforts at the CSA to develop a strategic plan for its survey activities and to link them more closely with the PASDEP, monitoring issues o f data coverage and schedule o f collection would need to be addressed. N o w that the CSA i s planningto undertake the UBEUS survey annually, labour markets urban areas could be regularly monitored. The challenge is to strengthen the quality o f wage data (whose inclusion in the survey has been a major step forward), o f considering urban domains capable o f providing more details on different types o f urbanareas (larger cities, smaller, secondary cities and for rural towns), and o f exploring ways o f capturing earnings inthe informal sector. Further, as discussed in Chapter 2, issues o f comparability with the LFS need to be addressed. 6.84 Data coverage for regular monitoring in rural areas remains more problematic. Alternatives rangingfrom the inclusion o f an all purpose nationally representative survey inthe statistical surveyplanto customization ofexisting surveys (e.g. agricultural sample survey) to fully include labour market aspects in additional modules/questions might be considered. Learning opportunities such as the ongoing Rural Investment Climate Assessment (ICA) could offer a platform to start a discussion on rural specificities, on the type o f instruments which might better capture them and on the appropriate tools for analysis. Similarly, the ongoing work for the new I C A (in urban areas) could offer an opportunity for discussion, collaboration and capacity building on firm level surveys, buildingon international experience and on the specificities o f the Ethiopian statistical program andits multiple firm level surveys. 6.85 Overall, it is important to ensure that the various elements o f labour market monitoring aim to reflect closely the informational and analytical needs raised by PASDEP. This should be responsive to emerging demands for more detailed information, such as those related to the scaling up o f existing government programs either directly or indirectly supporting firm development (through the provision o f infrastructure, for example), and those relatedto the new urban agenda. ~~ 37Monitoring separately child labor (as an indicator to be reduced) and employment rates will help in distinguishing these different policy outcomes. 90 REFERENCES Adams, h i 1Van (1997). "Assessment of the Jua Kali Pilot Voucher Program. Mimeo. Available at: http://www1.worldbank.org/education/economicsed/finance/demand/casefl