Report No: AUS19442 Poverty & Global Practice, Africa Standard Disclaimer: This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Copyright Statement: The material in this publication is copyrighted. Copying and/ or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978-750-4470, http://www. copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. This report was prepared by Utz Pape (TTL; Economist, GPV01) with substantial contributions from Mario di Filippo (Consultant, GPV01), Gonzalo Nunez (Consultant, GPV01) and Philip Randolph Wollburg (Consultant, GPV01). The chapter about Child and Youth Poverty was written by Pamela Dale (Social Protection Specialist; UNICEF) and Gonzalo I. Nunez, funded by UNICEF. The team is grateful for inputs and comments from Pamela Dale, Luca Parisotto (Consultant, GPV01), Menaal Ebrahim (Consultant, GPV01), Syedah Iqbal (Consultant, GPV01) and Verena Phipps (Senior Social Development Specialist, GSU07) as well as the peer reviewers Kevin Carey (Lead Economist, GMF07) and Kinnon Scott (Senior Economist, GPV04). Vice President Makhtar Diop Country Director Bella Bird Senior Director Carolina Sanchez-Paramo Practice Manager Pierella Paci Task Team Leader Utz Pape CONTENTS CONTENTS i EXECUTIVE SUMMARY vii INTRODUCTION xxi PART I: OVERVIEW OF POVERTY 1 1. Monetary Poverty 1 2. Multidimensional Deprivation 12 3. Evolution of Welfare Conditions in the North West Region 25 PART II: DEEP DIVE INTO SELECTED TOPICS 34 4. Remittances 34 5. Child and Youth Poverty 49 6. Social Protection 59 7. Conclusions 70 REFERENCES 75 APPENDIX 78 A. Figures and Tables 78 B. Lower Poverty Incidence in the North East Region 88 C. Sample and Data Collection 90 D. Consumption Aggregate 101 E. Rapid Consumption Methodology 109 F. Labor Statistics 115 G. Remittances 122 LIST OF FIGURES - EXECUTIVE SUMMARY Figure 0.1 Coverage of Somali household surveys including consumption modules. viii Figure 0.2 Poverty incidence in Somali regions (% of population). ix Figure 0.3 Inequality and poverty within Somali regions. x Figure 0.4 Evolution of poverty in the North West region. xi Figure 0.5 Population without education. xii Figure 0.6 School attendance. xii Figure 0.7 Characteristics of recipient households. xiv i | Contents Figure 0.8 Value and incidence of remittances. xiv Figure 0.9 Poverty and hunger among recipients and non-recipients. xv Figure 0.10 Child poverty by region. xvi Figure 0.11 Youth poverty by region. xvi Figure 0.12 Child school attendance. xvii Figure 0.13 Youth school attendance. xvii Figure 0.14 Impact of SSNs on poverty incidence. xix Figure 0.15 Cost of SSNs in all the Somali regions. xix Figure 0.1 GDP per capita, Sub-Saharan low income countries. xxii Figure 0.2 Coverage of household surveys in Somali regions. xxiii Figure 0.3 Coverage of the SHFS. xxv - PART I: OVERVIEW OF POVERTY Figure 1.1 Cross-country comparison of poverty incidence. 3 Figure 1.2 Cross-country comparison of poverty and GDP. 3 Figure 1.3 Cross-country comparison of poverty gap. 3 Figure 1.4 Cross-country comparison of poverty gap and GDP. 3 Figure 1.5 Poverty incidence. 4 Figure 1.6 Poverty gap. 4 Figure 1.7 Poverty severity. 6 Figure 1.8 Extreme poverty. 6 Figure 1.9 Impact of a consumption shock on poverty. 7 Figure 1.10 Poverty and inequality between regions. 9 Figure 1.11 Consumption distribution. 9 Figure 1.12 Poverty measures by gender of the household head. 10 Figure 2.1 Multidimensional deprivation by category. 13 Figure 2.2 Multidimensional deprivations. 14 Figure 2.3 Literacy rate in Sub-Saharan low-income countries. 15 Figure 2.4 Educational attainment (primary) in Sub-Saharan low-income countries. 15 Figure 2.5 Educational attainment (secondary) in Sub-Saharan low-income countries. 16 Figure 2.6 School enrollment (primary age) in Sub-Saharan low-income countries. 16 Figure 2.7 Literacy. 16 Figure 2.8 Educational attainment, primary. 16 Figure 2.9 Net primary school enrollment. 16 Figure 2.10 Net primary school enrollment, by gender. 16 Figure 2.11 Mean household expenditures in education. 17 Figure 2.12 Net primary school enrollment by gender of household head. 17 Figure 2.13 Labor force participation in Sub-Saharan low-income countries. 18 Figure 2.14 Employment in Sub-Saharan low-income countries. 18 Figure 2.15 Employment. 18 Figure 2.16 Labor force participation. 18 Figure 2.17 Own household work. 19 Figure 2.18 Employment by gender. 19 Figure 2.19 Labor force participation by gender. 19 Figure 2.20 Poverty headcount ratio, by inactivity reason. 20 Contents | ii Figure 2.21 Access to improved source of water in Sub-Saharan low-income countries. 21 Figure 2.22 Access to improved sanitation in Sub-Saharan low-income countries. 21 Figure 2.23 Access to improved source of water. 21 Figure 2.24 Access to improved sanitation. 21 Figure 2.25 Quality of the roof. 22 Figure 2.26 Quality of the floor. 22 Figure 2.27 Average annual health expenditures. 24 Figure 2.28 Child birth in hospital or clinic. 24 Figure 3.1 Poverty incidence 2013-2016. 26 Figure 3.2 Poverty gap. 28 Figure 3.3 GINI coefficient. 28 Figure 3.4 Distribution of consumption in urban areas. 29 Figure 3.5 Distribution of consumption in rural areas. 29 Figure 3.6 Households that received remittances. 29 Figure 3.7 Value of remittances for receivers. 29 Figure 3.8 Poverty and remittances. 30 Figure 3.9 Households headed by a woman. 31 Figure 3.10 Household size. 31 Figure 3.11 Literacy rate. 31 Figure 3.12 School attendance. 31 Figure 3.13 Labor force participation. 32 - PART II: DEEP DIVE INTO SELECTED TOPICS Figure 4.1 Incidence of remittances. 36 Figure 4.2 Per capita value of remittances. 36 Figure 4.3 Remittances per capita in selected countries. 36 Figure 4.4 Incidence and value of remittances, by income and urban, rural, IDP status. 37 Figure 4.5 Remittances by gender and education of the household head. 38 Figure 4.6 Remittances value by gender and education of household head. 38 Figure 4.7 Characteristics of recipient and non-recipient households. 39 Figure 4.8 Enrollment rate by recipient status and income quintile. 40 Figure 4.9 Literacy rate by recipient status and income quintile. 40 Figure 4.10 Remittance receipt compared to previous year. 40 Figure 4.11 Reasons for change in remittances value. 40 Figure 4.12 Poverty incidence by recipient status. 42 Figure 4.13 Poverty incidence by value of remittances received. 42 Figure 4.14 Cumulative distribution of consumption, urban. 43 Figure 4.15 Cumulative distribution of consumption, rural. 43 Figure 4.16 Difference in consumption between recipients and non-recipients. 44 Figure 4.17 Hunger and lack of money to buy food. 45 Figure 4.18 Meals on previous day, children and adults. 45 Figure 4.19 Poverty incidence with & without remittances. 45 Figure 4.20 Poverty incidence by change in remittances value previous year. 45 Figure 4.21 Main source of income, regional breakdown. 46 Figure 4.22 Main source of income, income quintiles. 46 iii | Contents Figure 4.23 Labor market statistics by recipient status. 47 Figure 5.1 Children and youth in the total population. 50 Figure 5.2 Children and youth in the poor population. 50 Figure 5.3 Child poverty by region. 51 Figure 5.4 Youth poverty by region. 51 Figure 5.5 Extreme child poverty by region. 52 Figure 5.6 Extreme youth poverty by region. 52 Figure 5.7 Child poverty by gender of household head and remittances status. 52 Figure 5.8 Youth poverty by gender of household head and remittances status. 52 Figure 5.9 Child poverty by household characteristics. 53 Figure 5.10 Youth poverty by household characteristics. 53 Figure 5.11 Child deprived in each dimension. 53 Figure 5.12 Youth deprived in each dimension. 53 Figure 5.13 Poverty incidence, school attendance and migration by number of children. 54 Figure 5.14 Child school attendance by region. 55 Figure 5.15 Youth school attendance by region. 55 Figure 5.16 Reasons for not attending school. 55 Figure 5.17 Child school attendance by household characteristics. 56 Figure 5.18 Youth school attendance by household characteristics. 56 Figure 5.19 Child school attendance by education of household head and literacy of adults in the household. 57 Figure 5.20 Youth school attendance by education of household head and literacy of adults in the household. 57 Figure 5.21 Water and sanitation for child. 57 Figure 5.22 Water and sanitation for youth. 57 Figure 5.23 Housing conditions of child. 58 Figure 5.24 Housing conditions of youth. 58 Figure 6.1 Impact of SSNs on poverty incidence. 66 Figure 6.2 Impact of SSNs on poverty gap. 66 Figure 6.3 Cost of SSNs in all the Somali regions. 68 Figure 6.4 Coverage, leakage and under-coverage for PMT. 68 Figure 6.5 Cost and impact of poverty reduction efforts through SSN programs. 68 - APPENDIX Figure A.1 Educational attainment, secondary. 80 Figure A.2 Educational attainment, tertiary. 80 Figure A.3 Inactivity reasons for women. 80 Figure A.4 Inactivity reasons for men. 80 Figure A.5 Average annual expenses on electrical devices. 81 Figure A.6 Poverty incidence with total and comparable consumption aggregates. 82 Figure A.7 Household headed by a woman by income quintile and recipient. status. 83 Figure A.8 Total imputed daily consumption value by recipient status. 84 Figure A.9 Labor force participation by recipient status. 85 Figure A.10 Unemployment by recipient status. 85 Figure A.11 Experience of hunger in past 4 weeks by recipient status. 85 Contents | iv Figure A.12 Child deprived in one dimension. 87 Figure A.13 Youth deprived in one dimension. 87 Figure A.14 Child deprived in two dimensions. 87 Figure A.15 Youth deprived in two dimensions. 87 Figure C1. Examples of duplicate GPS. 92 Figure C.2 Thiessen test polygons with bold boundaries representing the known enum. area boundaries. 92 Figure E.1 Relative bias of simulation results using the rapid consumption estimation. 112 Figure E.2 Relative standard error of simulation results using the rapid consumption estimation. 112 Figure F.1 Labor force. 117 LIST OF TABLES - PART I: OVERVIEW OF POVERTY Table 1.1 Total average real consumption (per capita, per day in 2016 US$). 8 Table 1.2 Household demographic attributes: size and age dependency ratio. 10 Table 1.3 Real consumption (per capita, per day in 2016 US$). 11 Table 4.1 Main sources of income for households. 47 - APPENDIX Table A.1 Household demographic attributes: number of children and adults. 78 Table A.2 Selected poverty indicators. 79 Table A.3 Access to improved source of water and sanitation, percentage of population. 79 Table A.4 Consumption items excluded from each survey to obtain a comparable consumption aggregate. 81 Table A.5 Average consumption (per capita, per day in US$). 82 Table A.6 Difference in educational spending per school-aged child between recipients and non- recipients. 82 Table A.7 Difference in share of males, household head excluded 83 Table A.8 Remittances share of total consumption 83 Table A.9 Changes in daily per capita consumption for recipients. 84 Table A.10 Full List of Sources of Income by Income Quintile and Regional Breakdown. 86 Table A.11 Estimated logit for proxy means test. 88 Table B.1 Differences in poverty incidence. 89 Table B.2 Most consumed core food items. 90 Table B.3 Key non-monetary indicators of well-being. 90 Table C.1 Sample properties of the SHFS. 100 Table C.2 Total number of households by PESS region and analytical strata. 100 Table C.3 Sample size calculation, number of replacement and final sample. 101 Table D.1 Summary of unit cleaning rules for food items. 105 Table D.2 Conversion factor to Kg for specific units and items. 106 Table D.3 Summary of cleaning rules for currency. 107 Table D.4 Threshold for non-food item expenditure (US$). 107 v | Contents Table D.5 Consumption of durable goods (per week in current US$). 107 Table D.6 Median consumption of durable goods (per week in current US$). 108 Table D.7 Median depreciation rate of durables goods. 109 Table E.1 Item partitions based on SLHS13 and the pilot in Mogadishu. 113 Table E.2 Laspeyres deflator by analytical strata. 115 Table F.1 SCO 08. 117 LIST OF BOXES - EXECUTIVE SUMMARY Box 1 Innovations to overcome data collection challenges xxiv - PART I: OVERVIEW OF POVERTY Box 2 The international poverty line 5 Box 3 Creating comparable poverty estimates for 2013 and 2016 27 - PART II: DEEP DIVE INTO SELECTED TOPICS Box 4 Real-time tracking of market prices 61 Box 5 Proxy means testing (PMT) 65 Contents | vi EXECUTIVE S U M M A RY 1. Somalia is emerging from 25 years of political instability and economic difficulty but hard data is lacking for evidence-based planning. The civil war and ongoing conflict that started in 1991 fragmented the country, undermined political institutions, and created widespread vulnerability. The conflict has eroded the statistical infrastructure and capacity, leaving policy makers and donors to operate in a statistical vacuum due to the lack of reliable data. In the absence of representative household surveys not much was known about poverty. The lack of information poses a threat to the design and implementation of policies and programs needed to support economic resilience and development as well as assistance in the event of shocks. 2. The region is currently facing a severe and prolonged drought, leaving about half of the population at acute risk, mostly in rural areas and IDP settlements. Food security in the region has been deteriorating due to poor rainfall between October, 2016, and March, 2017. With expected rain levels staying below average in the April to June 2017 season, more than 6 million people will remain acutely food insecure. Geographically, the drought is most severely affecting the southern pre- war regions of Bay and Bakool, as well as rangeland in the North East, leading to crop loss and livestock deaths. Output is expected to decline by 10.6 percent in 2017. In combination with high prices for staple foods, households’ purchasing power is compromised. More than a quarter of a million people have already been internally displaced as a consequence of the drought. vii | Executive Summary Figure 0.1: Coverage of Somali household surveys including consumption modules. SLHS, 2013 SHFS, Wave 1 (2016) SHFS, Wave 2 (2017) Region covered Region covered Region covered Region not covered Region not covered Region not covered Note: The boundaries on the map show approximate borders of Somali pre-war regions and do not necessarily reflect official borders, nor imply the expression of any opinion on the part of the World Bank concerning the status of any territory or the delimitation of its boundaries. Source: Authors’ calculation. 3. The World Bank’s Somali High Frequency affected areas. This report provides the first Survey provides quantitative data to inform poverty-centered profile of the Somali population essential resilience programs to avoid human based on this dataset going beyond but comparing disaster in future expected droughts. In 2013, with the results from North West in 2013. It a household budget survey was implemented characterizes the poor and their livelihoods, with a by the World Bank but covering only the Somali focus on social protection and remittances, before population in the North West. To overcome the the onset of the current crisis. The second wave of lack of data, the World Bank then implemented the Somali High Frequency Survey is planned for the first wave of the Somali High Frequency Survey summer, 2017 with expanded coverage including in Spring, 2016. The survey is representative of nomads. It will offer a second snapshot capturing 4.9 million Somalis, and does not cover nomadic the impact of the crisis on livelihoods and inform people and Somalis living in inaccessible conflict- resilience programs for the future. Executive Summary | viii SOMALIA IS ONE OF THE LEAST DEVELOPED COUNTRIES IN SUB SAHARAN AFRICA 4. The Somali population lags behind most Figure 0.2: Poverty incidence in Somali regions low-income African countries in availability (% of population). and access to basic infrastructure. Access to basic infrastructure such as water, sanitation systems, electricity lines and roads would substantially increase the level of development in all Somali regions, particularly in rural areas. Only 58 percent and 10 percent of Somalis have access to an improved source of water and improved sanitation respectively, compared to an average 69 and 25 percent in low-income Sub- 60-70 Saharan countries. Improvements in access to 50-60 water and sanitation are key for economic and 40-50 30-40 social development. Water and sanitation are 20-30 essential for the individual’s health, as well for 10-20 their productive activities, such as agriculture. Not covered by Inadequate water and sanitation services increase SHFS 2016 children’s exposure to waterborne diseases. In addition, low accessibility to such services affects Note: The poverty incidence of each region includes IDP settlements. The boundaries on the map show the time children need to employ to satisfy their approximate borders of Somali pre-war regions and do basic water and sanitation needs. By affecting not necessarily reflect official borders, nor imply the children’s health and time allocation, low expression of any opinion on the part of the World Bank quality water and sanitation services negatively concerning the status of any territory or the delimitation of its boundaries. Source: Authors’ calculation. influences their educational attainment. 5. Poverty is widespread with every second 6. Inequality is lower than in low-income Sub- Somali living in poverty in 2016 before the onset Saharan countries. The Gini index, measuring of the current shock. Poverty, defined as having inequality as the dispersion in consumption a total consumption expenditure lower than the expenditure among the population, is 37, international poverty line of US$1.90 at 2011 compared to an average value of 42 in low- PPP, also varies considerably across the Somali income Sub-Saharan countries. Inequality in low- population, ranging from 26 to 70 percent. Regional income Sub-Saharan countries ranges from 33 differences in poverty between the North East (27 (Mali) to 56 (Central African Republic), with 16 of percent) and the North West (50 percent) are much 26 countries having an inequality index between larger than urban/rural variation (45/52 percent). In 35 and 49. Within the Somali population, urban areas, poverty ranges from 26 (North East) to inequality is more pronounced for urban than 57 percent (Mogadishu). In rural areas, poverty ranges rural households. When taking into account urban from 34 percent (North East) to 61 percent (North and rural areas separately, poverty and inequality West). Poverty incidence is highest in IDP settlements are positively correlated: The North East region, where seven out of ten people are poor, while more where poverty incidence is lowest, has the lowest than 1.1 million Somalis, roughly 9 percent of the level of inequality, followed by North West and population, considered internally displaced. Mogadishu, and poverty and inequality in IDP ix | Executive Summary settlements are higher than in any other subgroup live in dwellings of lower quality, including lack of (Figure 0.3). access to improved water and sanitation facilities. This relationship between monetary poverty and 7. Poor households are more likely to be non-monetary indicators of deprivation holds deprived beyond monetary poverty, and less both within and across regions. Poor households likely to participate in the labor market. The further have poor labor market outcomes with low poor are more likely to be illiterate, to have labor force participation and high unemployment. lower levels of educational attainment, and to Figure 0.3: Inequality and poverty within Somali regions. 40 IDP Settlements Mogadishu North West 35 GINI index (0-100) North East North West 30 North East 25 20 0 20 40 60 80 Poverty incidence (% of population) Urban areas Rural areas Source: Authors’ calculation. 8. Improving active labor market participation, inactivity caused by illness and sickness, which in particular among women, will be important are among the prime causes for inactivity among to achieve sustained economic development. Somali men. Improved political stability can With poverty strongly correlated with unwanted address the threat of insecurity, another major labor market outcomes, the different reasons for reason for inactivity. Among women, household inactivity need to be addressed by a comprehensive work is the main barrier to better labor force approach. Better access to healthcare can reduce participation and employment outcomes. 5 OUT OF SOMALIS ARE POOR 10 Executive Summary | x POVERTY IN THE NORTH WEST FELL BETWEEN 2013 AND 2016 DESPITE A REDUCTION IN REMITTANCES BUT POOR RURAL HOUSEHOLDS ARE AT RISK OF BEING LEFT BEHIND 9. The Somali North West region records moderate 2016, down from 57 percent in 2013, compared welfare gains between 2013 and 2016, with to rural areas with 64 percent in 2016, down from poverty incidence declining around 5 percentage 69 percent in 2013. Rural households are not only points in urban and rural areas, but a majority poorer but their average shortfall from the poverty remains poor. Trends in poverty can only be line is also larger at 24 percent than in urban studied for the North West region, home to just over areas at 19 percent in 2016, leaving them further a quarter of Somalis, where a survey measuring away from overcoming poverty. Yet, average rural poverty was conducted in 2013. Poverty incidence shortfall decreased from 29 percent in 2013, more decreased for both urban and rural households, than in urban areas whose shortfall in 2013 was but remains more widespread in rural areas: in 20 percent, implying that reduction in monetary urban areas, poverty incidence was 52 percent in poverty was somewhat larger in rural areas. Figure 0.4: Evolution of poverty in the North West region. 100 90 Poverty incidence (% of population) 80 70 69 64 60 57 53 50 40 30 20 10 0 2013 2016 Urban areas Rural areas Source: Authors’ calculation. 10. The decrease in rural poverty is unlikely to be largely (23 percentage points) for receivers of associated with remittances, while in urban areas remittances and moderately for non-receivers (4 poverty increased among recipients. Between percentage points). The share of poor households 2013 and 2016, poverty incidence increased receiving remittances was similar in 2013 and 8 percentage points among urban households 2016 but the average amount received declined. that received remittances, and decreased 9 The urban increase in poverty among recipients percentage points among urban non-receivers. might be explained by a mixing effect with some In rural areas, poverty incidence decreased urban receivers graduating from poverty not xi | Executive Summary requiring remittances anymore and other urban and 2016. The increase in the literacy rate in poor households starting to receive remittances. urban areas is likely to be associated with higher The reduction in rural poverty is unlikely to be levels of education, as the share of people with associated with remittances as a similar number no education in urban areas decreased from 44 of households received remittances, which on percent to 41 percent during the same period. In average were smaller. Furthermore, the urban- rural areas, non-poor households maintained a rural gap in terms of share of households receiving similar literacy rate (around 47 percent), yet poor remittances decreased for poor and non-poor households experienced a decreased in literacy households between 2013 and 2016. of 6 percentage points (from 41 percent to 35 percent). A larger share of the rural poor does not 11. The rural poor are increasingly left behind have any education in 2016 (65 percent) compared in terms of education relative to non-poor and to 2013 (54 percent). Changes in the levels of urban populations between 2013 and 2016. education could be associated with a different Literacy increased by 10 percentage points among composition of the population in urban and rural the urban poor (from 48 percent to 58 percent) areas. The rural poor in the North West seem to be and 6 percentage points for the urban non-poor increasingly excluded in terms of education which (from 56 percent to 62 percent) between 2013 complicates their path out of poverty. Figure 0.5: Population without education. Figure 0.6: School attendance. 70 65 70 66 62 61 60 54 55 60 57 57 54 % of population aged 6-25 51 52 % of total population 50 47 50 44 44 40 40 37 40 30 30 20 20 10 10 0 0 Urban: Urban: Rural: Rural: Urban: Urban: Rural: Rural: Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor 2013 2016 2013 2016 Source: Authors’ calculation. Source: Authors’ calculation. 12. In order to reduce inequality and poverty, rural non-poor population, while it decreased access to, and availability of, key services, around 8 percentage points (from 52 percent particularly education, must be improved for to 44 percent) among the rural poor. Providing poor households. Worse educational levels access and means to reap the benefits from among the rural poor are probably caused by education, among other basic services, is crucial lower school attendance. Between 2013 and to achieve positive labor market outcomes and to 2016, school attendance increased in urban ultimately lift these households out of poverty. areas, remained relatively constant for the In 2016, nearly half of the school-aged Somali Executive Summary | xii population did not attend school due to illnesses, absent teachers, the lack of resources, and having to help at home. Attendance is more likely for boys than girls, and similar between households headed by a men and a woman. The emphasis should be on poor and vulnerable households, since their educational achievements are lower, and low achievement tends to be transmitted across generations. Sustained differences in terms of education between poor and non-poor Only 1 in 2 households, together with higher unemployment in rural areas, may continue to increase the school-aged gap. Thus, these challenges must be addressed soon with programs targeted at the rural poor Somalis attend that provide access and incentives to improve educational outcomes and create employment school opportunities. REMITTANCES ARE IMPORTANT AND IMPROVE SOCIO-ECONOMIC OUTCOMES BUT OFTEN DO NOT REACH THE ONES MOST IN NEED 13. Remittances make important contributions of those households would fall into poverty. In to welfare, with 1 in 5 Somali households fact, households that receive less remittances receiving remittances and many recipients than in the previous year are more likely to be relying heavily on these transfers. Remittances poor, suggesting households struggle to adjust to are a critical source of income for one fifth of such income shocks. A qualitative study supports Somali household who receive them, being the the notion that remittances income is critical to main source of income for more than half of households. Many recipient households rely on a recipient households. With an average annual single sender and would not know how to afford value of US$233 per capita among recipients, basic consumption and services without this these transfers make up around 37 percent of source of income. Thus, while remittances boost household expenditure on average. This suggests the welfare of households fortunate enough that recipients rely heavily on remittances and, to receive them, the lack of other means for consequently, are vulnerable to losing this generating income puts them at risk of falling into source of income. Without remittances, many poverty in case of losing their remittances income. 1 in 5 Somali households receive remittances xiii | Executive Summary Figure 0.7: Characteristics of recipient households. Figure 0.8: Value and incidence of remittances. 70 350 Q5 (top quintile) Current US$ per capita, per year 60 300 50 250 % of households Q4 Urban 40 200 Q3 Q2 Rural 30 150 IDP 20 100 Q1 10 50 (bottom quintile) 0 0 Female Enrollment Literacy rate 0 10 20 30 40 headed rate % of households that receive remittances Recipients Non-recipients Source: Authors’ calculation. Source: Authors’ calculation. 14. Recipient households are typically urban, households mainly use remittances as a top-up wealthier, headed by women, and better for their income from work. educated, but their labor market behavior does not differ much from that of non-recipients. 26 15. The Somali labor market does not provide percent of households headed by women receive many opportunities to substitute for the receipt remittances, compared to 17 percent of households of remittances. The fact that so many households headed by men. Wealthier and urban households rely on remittances as their main source of are more likely to receive remittances and they income is testament to a lack of opportunities receive higher amounts. Recipient households in the Somali labor market. It further suggests are more likely to enroll their children in school that households cannot simply take up work and spend more on education, especially poorer or work more hours to offset a decrease in recipient households. Through remittances, poor remittances, providing additional evidence for recipients can offset much of their educational their vulnerability. While remittances are a critical disadvantage compared to non-poor households. source of income for recipients, poor access to The effect of receiving remittances on labor decent work opportunities affects many Somali market behavior depends on whether household households, recipients or not. Measures to members use these funds to top up income from improve access are key to achieve sustainable work or to substitute work activities, if they can welfare in the long term. rely on income from remittances. The latter use of remittances income implies lower labor force 16. With recipients less vulnerable to poverty participation (full substitution) and fewer hours and hunger, remittances serve as a resilience on the job (partial substitution). Despite the fact mechanism. Poverty incidence is 18 percentage that remittances are the main source of income for points lower in recipient households (recipients: many recipients, there is no significant difference 37 percent; non-recipients: 55 percent). Recipients in labor force participation and hours worked have higher consumption levels, experienced between recipients and non-recipients. Thus, hunger in the past month half as often as non- Executive Summary | xiv recipients, and are less likely to lack money to the poorest households, only around 7 percent buy food. Remittances are providing families with receive remittances. Many of the recipient IDP the resources to cushion poverty and hunger. This households further suffered from a reduction may become critical in adverse situations like the in the value of the remittances relative to the ongoing drought, where households’ purchasing previous year, which can be hard to compensate. power has declined. The amounts received are not effective in reducing poverty for recipient IDP households because 17. Remittances are neither very prevalent nor they are too small relative to the poverty gap: the effective in reducing poverty among the most average daily per capita value of remittances for vulnerable households that are located in IDP poor IDP households is only 13 percent of their settlements. While IDP households are among consumption shortfall relative to the poverty line. Figure 0.9: Poverty and hunger among recipients and non-recipients. 60 50 40 % of households 30 20 10 0 Recipients Non-recipients Recipients Non-recipients Recipients Non-recipients POVERTY HUNGER LACK OF MONEY FOR FOOD Often (>10x) Sometimes (3-10x) Rarely (1-2x) Source: Authors’ calculation. 18. Remittances showcase how cash transfers evidence of beneficial welfare outcomes and provide an effective means of resilience resilience derived from remittances receipt shows to adverse shocks, but they remain largely that they are an apt means for households to deal unavailable to the most vulnerable populations, with such adverse shocks. But recipients’ high making the case for social protection programs reliance on remittances leaves these households to build resilience more broadly. The total value vulnerable to the volatility of diaspora incomes of remittances received should be interpreted with and the uncertainties around sending money to caution. The reported value is lower than stated by the region. Policies directed at facilitating and other sources, possibly due to under-reporting but de-risking remittances transfers can reduce this still reveals general patterns. Recipients are better kind of vulnerability but cannot reach far enough. protected from both monetary and non-monetary With 15 percent of the poor and only 7 percent forms of deprivation, leaving them less at risk in of IDP households receiving remittances, access the face of shocks like the ongoing drought. This to such assistance excludes many people who xv | Executive Summary most need it. This general lack of resilience mitigate the most urgent shortfalls in basic needs, mechanisms can be addressed through more in particular in the current crisis. formal and predictable forms of cash transfers to EVERY SECOND SOMALI CHILD DOES NOT GO TO SCHOOL. ESPECIALLY FOR CHILDREN IN POOR HOUSEHOLDS, THIS CAN CREATE A LIFETIME POVERTY SPELL 19. Like in many parts of the world, Somali conditions in the North East region, the lowest children are particularly likely to be poor. 58 child and youth poverty incidence are found in percent of children (0-14 years) and 46 percent of that area. Child and youth poverty is substantially youth (15-24 years) live in households with total lower in small households, households with an consumption expenditure below the poverty line. educated household head, and households that In line with the general finding of better welfare receive remittances. Figure 0.10: Child poverty by region. Figure 0.11: Youth poverty by region. Overall average Overall average 100 100 Poverty incidence (% of children) Poverty incidence (% of youth) 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 hu an l n l hu n l ts n l ts ra ra ra ra ba ba ba en en b Ru Ru Ru Ru is s di Ur Ur Ur m Ur m ad tle a tle NE NW NE NW NE NW NE NW og og et et M M PS PS ID ID Source: Authors’ calculation. Source: Authors’ calculation. 20. Almost 4 out of 5 children are deprived in at areas. Along with the lack of access to information, least one dimension. 79 percent of children and consumption deprivation is more relevant for 85 percent of youth are deprived in at least one youths in Mogadishu and urban areas of North West. dimension, while 47 and 54 percent are deprived in Lack of access to an improved water source is the two dimensions or more, respectively. Deprivation second most common deprivation in rural areas of is concentrated in rural areas of North West and IDP North West and North East and in IDP settlements. populations. For children, consumption deprivation is the most common type of deprivation in urban 21. Nearly half of Somali children and youth areas and IDP camps, while the lack of access to do not currently attend school, and school improved water source is most prevalent in rural attendance is less likely in poor households. Executive Summary | xvi Education is a powerful tool to improve the points, respectively, and recipient households wellbeing of future generations. However, 47 spend more on education than non-recipients, percent of the children and 45 percent of youth particularly among the poorer households. School do not attend school, with attendance lower in attendance is further 30 percent less likely for IDP settlements. Moreover, poor children are less children and youth when the head of their likely to attend school (46 percent) compared household has no education. The most common to children living in non-poor households (63 reasons for not attending school are illnesses, percent). Thus, children from poor households absent teachers, lack of resources, and, among face bigger obstacles to overcome poverty in the youth, having to help at home. Efforts aimed their adult life. Children and youth that live in at increasing educational outcomes should be households that receive remittances have a higher aimed at these constraints to attendance. school attendance by 13 and 17 percentage Figure 0.12: Child school attendance. Figure 0.13: Youth school attendance. 80 Overall average 80 Overall average School attendance (% of children) School attendance (% of youth) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 r r r n r e e e n e e e n n o o o o at t at at t at tio tio io io ra ra Po Po Po Po er er er er at at ite ite ca ca n- n- lit lit lit lit uc uc du du tl tl No No ts ts ts ts ed ed ul ul :e :e ul ul ul ul ad ad ad ad o o Ad ad Ad ad :n :n 1 1 he he 2+ 2+ ad ad HH HH He He HH HH Source: Authors’ calculation. Source: Authors’ calculation. 22. Poor children and children in IDP settlements on health and future productivity, and thus, on often grow up in an environment of poor sanitary future poverty status. Regional disparities and conditions, with adverse consequences for their dire conditions, especially in IDP settlements and health and future productivity. Less than half in rural areas in North West, make it more difficult of children and youth drink water from a piped to lift households out of poverty. source. Children and youth living in rural areas are much less likely to treat the water they use 23. Breaking the intergenerational poverty from an unprotected water source. Most children cycle requires improving conditions for and youth in IDP camps and rural parts of the children and youth, especially with respect to North West rely on other water sources. Water education. In the current environment, children and sanitation conditions can have large impacts are disadvantaged relative to older generations, xvii | Executive Summary with children from poor households facing a addressed now with dedicated and specific particularly severe disadvantage to overcoming programs to create enabling environments and poverty and deprivation. This disadvantage will opportunities for disadvantaged children and likely translate into poverty in their adult lives. youth. Priority should be given to programs which In light of the overwhelmingly young Somali aim to break the intergenerational transmission population, this will become an extraordinary of poverty by addressing low levels of education, development challenge. Barriers to educational poor health, and poor housing conditions. opportunities and basic services must be A SOCIAL PROTECTION PROGRAM COULD REACH THE ONES MOST IN NEED, AND HELP BREAK OUT POVERTY OVER GENERATIONS, BUT WOULD COME AT A HEFTY COST 24. The absence of effective, resilience-building have a limited impact on the most vulnerable. social protection programs exacerbates the effect Remittances are de-centralized and not targeted of shocks on livelihoods, putting millions of to the most vulnerable households. Often they Somalis at risk in the current severe drought. The are distributed within clan networks, excluding data collected in 2016 shows that a large number exactly those that have lost their social support of vulnerable households lack access to effective network. Still, nearly 43 percent of the Somali mechanisms for coping with shocks. The result population is poor and does not receive any of this has manifested in the many people at risk remittances. Furthermore, remittances are in early 2017 following several severe droughts. volatile and, thus, do not necessarily scale with Recurrent natural shocks like these droughts caused needs. For example, the change in regulations for by El Niño will continue to test the resilience of the international bank transfers to Somalia created Somali population in the future. In the aftermath uncertainty around remittances at the time of the of the current shock, designing a well-targeted and emerging drought. effective social protection program that can work in the local context will be one of the overarching 26. A transparent social protection program objectives to avoid repeated famines and, more like a direct cash transfer can help to reach generally, to open up a sustainable path to poverty the most vulnerable population. While reduction and shared prosperity. donor support for the Somali population is considerable, local capacity to efficiently absorb 25. Remittances can help to smooth shocks the investment and deliver services are limited. and improve welfare conditions, yet they In addition, political economy challenges can Natural shocks continue to test Somali’s resilience Executive Summary | xviii weaken the effectiveness of programs or delay considerable impact on poverty would require their implementation. Therefore, a transparent substantial funding. Using observable household social protection program can be a good characteristics to target poor households, a alternative to reach the most vulnerable. While uniform annual transfer of US$ 157 per capita direct cash transfers have limitations, especially to all eligible households would reduce poverty where services are unavailable rather than just by 19 percentage points. Poverty among the inaccessible, simulating the cost and impact of most vulnerable households in rural areas and such programs serves as a benchmark helping IDP settlements would decline by 26 and 22 to better understand fundamental trade- percentage points, respectively. As for any offs that will also apply to alternative social targeted programs, there would be some leakage: protection programs. Cash transfers are only one 27 percent of poor households would be excluded alternative, and further analysis is needed given while 31 percent of non-poor households would the complexity of designing and implementing a be included into the program. The costs of such a social protection program. program, US$ 1.7 billion, representing around 22 percent of GDP, is high but of similar magnitude 27. A large targeted social protection program as net official development assistance and aid to reduce poverty by 19 percentage points, (US$ 1.3 billion in 2015). This benchmark gives from 51 to 32 percent, would come at a high an idea about the effort and resources needed to cost of US$1.7 billion. Given widespread and have substantial impact on poverty. deep poverty, a social protection program with Figure 0.14: Impact of SSNs on poverty incidence. Figure 0.15: Cost of SSNs in all the Somali regions. 80 2.0 Poverty incidence (% of population) 71 1.7 70 1.8 Billion current US$ per year 6361 1.6 1.5 60 51 52 1.4 4947 50 44 45 46 1.2 43 40 1.0 40 38 1.0 0.871 32 33 0.8 0.7 0.7 30 25 28 26 0.5 0.6 20 0.6 0.5 20 18 0.4 0.4 0.3 0.3 0.20.2 10 0.2 0.1 0.1 0 0.0 Overall Urban Rural IDP Overall Urban Rural IDP Settlements Settlements Current PMT: Avg. poverty gap PMT: Twice avg. PT: Twice avg. poverty gap poverty gap PT: Avg. poverty gap PMT: Twice avg. PMT: Avg. poverty gap PT: Avg. poverty gap poverty gap PT: Twice avg. poverty gap Source: Authors’ calculation. Source: Authors’ calculation. 28. A smaller transfer amount is less costly reduction to US$ 871 million. However, such a but cannot lift the very poor out of poverty. transfer amount will only reduce poverty from 51 Reducing the transfer amount to US$ 80 per percent to 44 percent. 12 percent of those that capita will half the overall costs for poverty would be lifted out of poverty with a uniform xix | Executive Summary annual transfer of US$ 157 would remain poor. Furthermore, the poverty line is an approximate cost-of-living standard and should be treated as such. Thus, the exact amount for a transfer should be derived from a contextual analysis of needs and their costs. Also, the targeting approach needs to emerge from a discussion of the objective of a social protection program. Targeting only the very poor with a larger transfer can be more suitable depending on the objective. 29. Protecting the poor in times of a shock like a drought is more expensive than just lifting poor households out of poverty. Building resilience is important to protect protective assets from being sold in times of a shock. A 10 percent consumption shock across all households would increase the Any social protection costs of a social protection program to reduce program with poverty to the same level of 32 percent from US$ 1.7 billion to around US$ 2.0 billion. It is worth considerable impact on noting that the 10 percent shock increases the poverty would require cost of a comparable social protection program by substantial funding 17 percent. This large elasticity is due to a large number of households that were almost poor in 2016 but are likely to be pushed into poverty by a shock like the current drought. A MORE IN-DEPTH ANALYSIS WILL PROVIDE EVIDENCE FOR MORE SPECIFIC RECOMMENDATIONS TOWARDS POVERTY REDUCTION 30. A Somali Poverty Assessment is planned to recommendations with respect to poverty provide more in-depth analysis to better inform reduction programs. A more in-depth analysis policies and programs. This poverty profile along the lines proposed in the chapters’ key focuses on descriptive statistics to provide a messages is planned, taking advantage of the snapshot of poverty and other socio-economic second wave of the Somali High Frequency indicators. The analysis is used to make general Survey that is implemented in summer, 2017. Executive Summary | xx INTRODUCTION 1. Somalia is emerging from more than urban areas, three types of livelihood zones make two decades of political instability. After up the vast majority of the landmass: pastoral independence in 1960, Somalia transitioned and agro-pastoral livelihood zones inland, and towards an autocratic regime that finally collapsed fisheries zones on the coast. In the South, the Juba in 1991. The following civil war wiped out the and Shabelle rivers provide irrigation for more central state and created a power vacuum that was sustained agriculture. quickly filled by local warring factions. Between 1995 and 2000 Somalia witnessed the emergence 3. A vibrant but largely informal private sector of regional administrations. Somaliland self- is the result of the long absence of a functioning declared independence in 1991, followed by state. During the period of civil conflict and in Puntland in the northeast declaring itself a the absence of a central government, the Somali regional administration in 1998. In this period, economy continued to grow at a moderate pace.1 security improved and economic development This performance can be explained by statelessness accelerated slightly, while internal displacement following the collapse of the previous predatory increased. The first significant central state regime:2 The lifting of state constraints on private institution, the Transitional Federal Government enterprise led not only to improved economic (TFG), was formed in 2004 but political instability performance but also to private sector provision and violence continued especially in the southern of services which would otherwise be provided regions. After several setbacks and the expiration by the public sector. Several economic activities of the transitionary mandate of the TFG, the including telecommunications, money transfer Federal Government of Somalia (FGS) was finally businesses, livestock exports, and localized created in 2012 followed by a relatively more electricity services grew well during this period. stable period. After peaceful elections in 2016, a The disintegration of the state did not result in new Government was formed in 2017 committed a complete economic collapse in part due to to embark on a development trajectory. the large scale out-migration of skilled Somalis who sent back part of their earnings –created in 2. The prolonged period of instability created much more productive foreign environments– as a highly vulnerable population of 12 million remittances. Remittances grew from a negligible people. According to the 2012 UNFPA Population amount in 1990 to about 24 percent of GDP Estimation Survey (PESS), 42 percent of Somalis in 2015.³ Informal institutions based on clan live in urban areas, 23 percent live in rural areas, networks provided the functions of secure 26 percent are nomadic, and 9 percent –just property rights and contract enforcement. above 1 million– internally displaced. Outside of 1 Estimates indicate that the Somali nominal GDP in 2015 was US$5.9 billion. In 1990, GDP was estimated at US$1.03 billion. These estimates imply an average annual growth rate in excess of 4 percent during the 25 year period. 2 The positive impact of ‘statelessness’ on the economy has been well documented See for example, Leeson, Peter T. J of Comp. Econ. 2007; Powell, Benjamin et al. J of Econ. Behav. and Org. 2008. 3 http://www.worldbank.org/en/country/somalia/overview. xxi | Executive Summary 4. Somalia’s gross domestic product is estimated capita income is on average 20 to 40 percent at US$6.2 billion in 2016, equivalent to US$503 higher than GDP per capita, as large inflows of per capita.4 In current per capita dollar terms, remittances allow households to top up own- among Sub-Saharan, low-income countries, generated income as measured by GDP per Somalia’s economy is larger than The Gambia, capita. According to the most recent World Bank the Democratic Republic of Congo, Liberia, estimates, Somaliland’s GDP was US$1.6 billion Madagascar, Malawi, Niger, Central African in 2012, while authorities in Puntland put its GDP Republic, and Burundi, making it the 9th poorest at US$1.3 billion in 2010.5 country of the region (Figure 0.1). Somalia’s per Figure 0.1: GDP per capita, Sub-Saharan low income countries. 1,000 GDP per capita, current US$ 800 600 400 200 0 15 15 15 15 15 15 15 15 15 16 15 15 15 5 01 20 20 20 20 20 20 20 20 20 20 20 20 20 I2 E A N LI A E A O N M D DG R BD ZW SL BF NE TZ UG BE GI CO TG M SO M Average regional value Source: SHFS 2016 and World Bank dataset. 5. Diaspora remittances are central to Somalia’s 6. The region is currently facing a severe and economy, outweighing both international aid prolonged drought, leaving about half of the flows and foreign direct investment.6 Remittances population at acute risk of famine, mostly in rural are estimated at between US$1.2 and US$2 billion areas and IDP settlements. Food security in the today, equivalent to 23 to 38 percent of GDP. region has been deteriorating due to poor rainfall Remittances as a source of income have been in the October-December 2016 season. Low levels important in cushioning household economies, of rainfall are forecast for the April to June 2017 creating a buffer against shocks (drought, trade season. According to the World Food Programme, bans, inter-clan warfare). Remittances fund direct in January 2017 around 3 million people were not consumption, including education and health, and consuming the minimum food requirements, while some investment, mostly in residential construction, 3.3 million more were in need of assistance to avoid and allow Somalia to sustain its high consumption the crisis. According to the Famine Early Warning rates and to finance a large trade deficit. Systems Network (FEWS NET) and Food Security and 4 Idem. 5 Somalia Economic Outlook. 6 FAO (2013). Executive Summary | xxii Nutrition Analysis Unit (FSNAU), famine (IPC Phase the last Somalia-wide representative Survey. 5) is likely if the rain levels are below the average Existing data sources are mostly limited to in the April to June 2017 season. Geographically, food and nutrition survey conducted by FSNAU the drought is most severely affecting the southern and FAO, and a range of other smaller surveys pre-war regions of Bay and Bakool, as well as implemented by organizations operating in rangeland in the North East, leading to crop loss Somalia. In 2012, the first nationwide Population and livestock deaths, and output is expected to Estimation Survey (PESS) was implemented decline by 10.6 percent in 2017 according to preparing for a census. Somaliland carried out World Bank internal estimates. In combination a household budget survey (SLHS) in 2013, with high prices for staple foods, households’ which generated much-needed indicators, purchasing power is compromised. In addition, including poverty estimates, but the sample is 257,000 people have been internally displaced not representative and did not cover nomads as a consequence of the drought.7 and Internally Displaced Persons (IDPs) (Figure 0.2). The lack of data impedes the design and 7. In the absence of representative household implementation of policies and programs surveys not much was known about poverty. needed to support economic development and The Somalia Socioeconomic Survey 2002 was assistance in the event of severe shocks. Figure 0.2: Coverage of household surveys in Somali regions. SLHS, 2013 SHFS, Wave 1 (2016) SHFS, Wave 2 (2017) Region covered Region covered Region covered Region not covered Region not covered Region not covered Note: The boundaries on the map show approximate borders of Somali pre-war regions and do not necessarily reflect official borders, nor imply the expression of any opinion on the part of the World Bank concerning the status of any territory or the delimitation of its boundaries. Source: Authors’ calculation. 8. The World Bank implemented the first wave IDP settlements. The geographical coverage has of the Somali High Frequency Survey in 2016. been improved compared to the SLHS in 2013 The survey was administered to 4,117 households (Figure 0.2). However, the sample still is not fully distributed among rural and urban areas, and representative of the Somali population as it 7 http://reliefweb.int/report/somalia/somalia-drought-response-situation-report-no-1-24-march-2017. xxiii | Executive Summary excludes nomadic households and households 9. Somali regions have been aggregated into in insecure areas (Appendix).8 Therefore, the distinct geographical areas: North West, presented data should be interpreted with respect North East, Mogadishu and IDP Settlements. to the urban and rural as well as IDP population North West includes the pre-war regions of covered by the survey. Extrapolations towards the Awdal, Sanaag, Sool, Togdheer, and Woqooyi overall Somali population should only be made Galbeed. North East includes the regions cautiously given that the nomadic population is of Nugal, Bari, and Mudug. IDP settlements likely to be different from the urban, rural and include all settlements of internally displaced IDP population with similar reservations for the persons located in Mogadishu, North West and population living in insecure and, thus, not covered North East. Finally, Mogadishu includes all the areas. Even for the safer areas, new solutions had households located in the capital with the to be developed to overcome challenges created exception of IDP settlements.9 In addition to by the fragile context and weak data infrastructure geographical regions, the Somali population including the absence of a sample frame (Box has been further divided into three livelihood 1). The success of this established survey types: urban, rural, and internally displaced infrastructure offers an opportunity to implement settlements (IDPs). The Somali High Frequency additional waves of the survey with expanded Survey is representative of 4.9 million Somalis. coverage. Wave 2 will include for the first time the The nomadic people and Somalis living in Somali nomadic population as well as additional inaccessible conflict-affected areas amount to a urban and rural areas. The survey, funded by the population of 6.5 million that was not surveyed Somali Multi Partner Trust Fund, is expected to be by Wave 1 of the SHFS (Figure 0.3).10 administered in the summer of 2017. Box 1: Innovations to overcome data collection challenges Data collection is the Somali regions is The survey was implemented using tablets challenging due to insecurity in some as survey devices (CAPI). Interviews were areas. Face-to-face time is limited to about conducted using SurveyCTO Collect on the 60 minutes while a full consumption tablet with data transmitted to a secure questionnaire usually takes 90 to 120 minutes. server in a cloud computing environment. Also, limited field access makes monitoring of GPS tracker helped to track all devices using data quality difficult. These challenge were a web interface, Barcode Scanner allowed overcome by a newly developed methodology to use barcodes for the identification to collect consumption data in 60 minutes, of enumerators, and a parental control and with the design of a remote real-time data application provided a safe contained monitoring system. working environment for enumerators. 8 Nomadic households represent about one third of the Somali population. 9 Wave I of the SHFS covered the following pre-war regions: Awdal, Banadir, Bari, Mudug, Nugaal, Sanaag, Sool, Togdheer, and Woqooyi Galbeed. 10 The pre-war regions not included in this study are: Bakool, Bay, Galgaduud, Gedo, Hiraan, Lower Juba, Lower Shabelle, Middle Juba and Middle Shabelle. While the survey also did not include all Somali IDPs, the surveys IDP population was deemed representative of all IDPs. Executive Summary | xxiv The new solutions were tested in a pilot remote data quality management, on-the- survey in Mogadishu. Implementing these fly processing and analytics. The newly innovations in the Somali High Frequency developed Rapid Consumption methodology Survey ensured high data quality despite was applied to estimate poverty based on limitations for field monitoring, as the short 60-minute interviews. The success of infrastructure offers a modern data collection this established survey infrastructure offers system that can be used to fill the most an opportunity to implement additional important data gaps. This set-up enabled waves of the survey with expanded coverage. 10. The poverty profile presents the first Somali- North West region between 2013 and 2016. Part wide assessment of welfare conditions. The II analyzes in detail selected topics: the role and poverty profile is structured in the following way: dynamic of remittances, child and youth poverty Part I explores the monetary and non-monetary and social protection measures to increase dimensions of poverty in Somali regions, as well resilience and reduce poverty. as the evolution of welfare conditions in the Figure 0.3: Coverage of the SHFS. 3.50 3.00 Population in millions 2.50 2.00 1.50 1.00 0.50 0.00 Share of population covered in SHFS Population not covered Source: Authors’ calculation. xxv | Executive Summary PART I OVERVIEW OF POVERTY 1. MONETARY POLICY KEY MESSAGES Poverty is wide-spread with every second Somali living in poverty, and almost 1 in 3 facing conditions of extreme poverty. Poverty varies considerably across different segments of the Somali population, ranging from 26 to 70 percent, with regional disparities exceeding differences between urban and rural areas. Widespread poverty and a moderate poverty gap of 22 percent implies many Somalis are far from overcoming poverty. Somalis living in IDP settlements face most widespread and deepest poverty. 7 out of 10 internally displaced live in poverty and 1 in 2 live in extreme poverty, placing them among the poorest populations in Sub-Saharan low-income countries. Inequality is lower than in most low-income African countries, but many non-poor are at risk of falling into poverty in case of an adverse shock to consumption. With a Gini index of 37 percent, Part I: Overview of Poverty | 1 inequality is considerably below the 42 percent average. Low inequality is owed to homogeneous levels of consumption, leaving even non-poor Somalis not very far from the poverty line. As a result, a 10 percent adverse shock to consumption would raise the poverty rate by 6 percentage points. A more comprehensive analysis will be included in the Somali Poverty Assessment relying on data from Wave 1 and Wave 2 of the SHFS. The analysis will consider adult equivalent measures of monetary poverty considering within household economies of scale. The analysis will also expand on the profile of the vulnerable population, and the impact of the drought on livelihoods. This will help to draw a more robust and comprehensive picture of poverty. 11. 1 in 2 Somali people are poor, with almost far from overcoming poverty. The overall one third facing conditions of extreme poverty. poverty gap for the Somali poor is 22 percent 51 percent of the Somali population lives in of the poverty line or 7,383 Somali Shillings conditions of poverty (Figure 1.5), as defined a day, where the poverty gap index measures by having a total daily per capita consumption the average gap between total consumption expenditure lower than the international poverty expenditure of the poor and the poverty line, line of US$1.90 at 2011 PPP, which equals 34,341 as a percentage of the poverty line. The poverty Somali Shillings per day per person in 2016 gap of 22 percent suggests that many of the (Box 2).11 Further, 31 percent of Somalis have a poor are far from the poverty line and need total daily per capita consumption expenditure a significant increase in their consumption of less than US$1.25, expressed at 2011 PPP, to move out of poverty, reflective of the fact leaving them in conditions of extreme poverty. that many Somalis live in extreme poverty. The At 31 percent of the total Somali population, the severity of poverty, estimated at 11.4 percent, is share of the extreme poor makes up a full 60 further testament to disparities in consumption percent of the poor population. Hence, while a among the poor population.12 As a theoretical large share of the Somali population is poor, a benchmark for addressing this situation: if the majority of the poor face extreme poverty, having poor could be perfectly targeted, an annual to overcome a formidable consumption shortfall subsidy of around US$1.3 billion would be if they are to escape poverty (Figure 1.8). necessary to lift all the Somali poor out of poverty (see Chapter 6. Social protection). 12. Widespread poverty, combined with a moderate poverty gap, leaves many Somalis 11 We compute the value of the international poverty line in 2016 Sh. using the 2011 So.Sh./$ PPP, the Somali Consumer Price Index increase between 2011 and 2016, and the 2016 nominal exchange rate between the Somali Shilling and the US Dollar. 12 The poverty severity index is defined as the average squared poverty gap. 2 | Part I: Overview of Poverty Figure 1.1: Cross-country comparison Figure 1.2: Cross-country comparison of poverty of poverty incidence. and GDP. 80 80 Poverty incidence (% of population) Poverty headcount (% of population) Average Regional Poverty MWI 60 60 SOM TZA SSD 40 40 ETH 20 20 0 0 200 400 600 800 1,000 O 8 E 1 LI 1 N 1 H 2 M 10 04 W 006 R 10 F 7 R 9 A 14 D 4 TG 200 SL 201 M 01 GI 01 ET 201 CA 200 NE 200 TC 201 CO 20 20 LB 0 BF 20 2 2 2 I2 I GDP per capita (2015, US$) BD M Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. Figure 1.3: Cross-country comparison of Figure 1.4: Cross-country comparison of poverty gap. poverty gap and GDP. 40 40 Poverty gap (% of poverty line) Poverty gap (% of poverty line) MWI Average Regional Poverty 30 SOM SSD 20 20 TZA 10 ETH 0 0 200 400 600 800 1,000 N 1 H 2 M 10 04 LI 1 B 09 D 03 N 1 F 2 OZ 08 R 08 O 07 N 11 ET 201 SE 201 GI 01 M 201 CA 01 CO 20 20 GM 20 TC 20 M 20 LB 20 TG 20 BE 20 2 2 GDP per capita (2015, US$) D CO Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. Part I: Overview of Poverty | 3 13. At 51 percent, the poverty rate is in line with income average of 20 percent, where once again the regional average of low-income countries there are large differences underlying the regional across Sub-Saharan Africa. The unweighted average (Figure 1.3 and Figure 1.4). average poverty headcount rate of low-income countries in Sub-Saharan Africa, based on the 14. Poverty varies considerably across the latest available estimates from World Bank Somali population, ranging from 26 to 70 Open Data, is 51 percent, equal to the Somali percent, with regional disparities exceeding overall poverty headcount rate (Figure 1.1). those between urban and rural areas. 1 in This relationship also holds when controlling 2 people in North West and 57 percent in for countries’ GDP per capita (Figure 1.2).13 Mogadishu are below the poverty line, making However, there is considerable variation in it about twice as likely to be poor there than poverty underlying the Sub-Saharan low-income in North East at 26 percent (Figure 1.5). This country average. In monetary terms, Somalis relationship also holds for the poverty gap (Figure are considerably better off than the poorest 1.6). Indeed, poverty in North East is more similar countries in the sample, Burundi and Malawi, of to poverty in neighboring Ethiopia (34 percent) whose population 78 percent live in poverty. In than to other Somali regions. With a poverty rate contrast, the Somali poverty rate is 17 percentage of 52 percent and a poverty gap index of 20 points higher than that of neighboring Ethiopia percent, the rural population is poorer than the (34 percent), and 30 percentage points higher urban population, at 45 percent and 17 percent, than that of Zimbabwe (21 percent). In a similar respectively. However, this difference is less fashion, the Somali poverty gap index at 22 pronounced than the differences across regions. percent is in keeping with the Sub-Saharan low- Figure 1.5: Poverty incidence. Figure 1.6: Poverty gap. 100 Overall average 60 Overall average Poverty incidence (% of population) Poverty gap (% of poverty line) 80 45 60 30 40 15 20 0 0 u n l an l hu an l ts n l ts ra ra ra ra ba sh ba en en b b Ru Ru Ru Ru s di di Ur Ur Ur m Ur m a tle a tle NE NW NE NW NE NW NE NW og og et et M M PS PS ID ID Source: Authors’ calculation. Source: Authors’ calculation. 13 The countries used for regional comparison are all the African low-income countries as defined by the World Bank: Benin, Burkina Faso, Burundi, Central African Republic, Chad, Comoros, Democratic Republic of Congo, Eritrea, Ethiopia, Gambia, Guinea, Guinea-Bissau, Liberia, Madagascar, Malawi, Mali, Mozambique, Niger, Rwanda, Senegal, Sierra Leone, South Sudan, Tanzania, Togo, Uganda, and Zimbabwe. For each country, we include the most recent available year for each indicator. 4 | Part I: Overview of Poverty 15. Poverty is most widespread and deepest groups of the Somali population (Figure 1.10). IDP in IDP settlements. Almost 3 in 4 people are household members are thus among the poorest poor in IDP settlements and 1 in 2 are extremely populations, compared to other low-income Sub- poor, which is equivalent to two thirds of poor Saharan African countries (Figure 1.1 and Figure people, and reflected in an average poverty gap 1.3), and they are at a particularly high-risk before of 36 percent (Figure 1.5, Figure 1.6 and Figure the onset of the current shock and likely in need 1.8). Consequently, inequality is higher among of urgent assistance. the displaced population than among the other Box 2: The International Poverty Line The international poverty line was introduced the national poverty lines to real US$ using in the 1990 World Development Report, with the 2011 PPPs; and 3) Computing the simple the purpose of measuring absolute poverty in a average of the 15 national poverty lines, consistent way across different countries. Using resulting in a value of US$ 1.88 per person data on 33 national poverty lines for the 1970s per day, rounded up to US$ 1.90. and 1980s (for both developed and developing economies), Ravallion, Datt, and van de Walle The increase in the value of the international proposed a line of US$ 0.76 a day at 1985 PPP. poverty line, from US$ 1.25 to US$ 1.90, can That value represented the predicted poverty be mostly attributed to the lower U.S. dollar line for the poorest country in the sample. purchasing power relative to the purchasing power of the currencies of poorest countries. Subsequently, they proposed a higher This is equivalent to saying that US$ 1.90 in line of US$ 1.02 a day, which found more 2011 real terms buys approximately the same international consensus since it was more basket of goods that could be bought by US$ representative of the poverty lines in low- 1.25 in 2005. income countries and it became the US$1 a day line. Throughout the years, the poverty For the Somali population, poverty is line has been revised three times, as new set estimated using the standard international of PPPs have become available. First from US$ poverty line. As the poverty line is defined 1 to US$ 1.08 at 1993 PPPs, then to US$ 1.25 at US$ 2011 PPPs, it must be converted to at 2005 PPPs, and finally to its current value, the currency used to measure consumption in US$ 1.90 at 2011 PPPs. The US$ 1.25 line was the survey. First, US$ 2011 are converted into originally defined as the unweighted average Somali Shilling in 2011 using the regression- of the national poverty lines for the fifteen based PPP estimate for Somalia. Second, poorest countries (see Ravallion et al. 2009). the change in purchasing power per Somali The computation of the current international Shilling is considered by estimating inflation poverty line was obtained in a similar fashion from 2011 to 2016. Third, the poverty line is by: 1) Taking those national poverty lines converted back to US$. The resulting poverty considering inflation to 2011; 2) Converting line is 1.47 US$ (2016) per day per person. Part I: Overview of Poverty | 5 16. Large disparities in poverty emerge when North East region is supported by other welfare comparing different Somali regions. These indicators (see Chapter 2. Multidimensional disparities exceed differences between urban deprivation and Appendix B. Lower poverty and rural areas. Almost 3 in 4 people live in incidence in the North East region for a detailed poverty in IDP camps, with an average poverty discussion). Poverty in rural areas is both more gap of 36 percent. Poverty in North West and widespread and deeper than in urban areas, Mogadishu is about twice as high and twice as but this difference is less pronounced than the deep as poverty in North East (Figure 1.5 and difference between regions. Figure 1.6 ). Lower poverty incidence in the Figure 1.7: Poverty severity. Figure 1.8: Extreme poverty. 40 Overall average 100 Overall average Extreme poverty (% of population) 80 Poverty severity index 30 60 20 40 10 20 0 0 u n l n l u n l ts n l ts ra ra ra ra ba ba sh ba sh ba en en Ru Ru Ru Ru di di Ur Ur Ur m Ur m a tle a tle NE NW NE NW NE NW NE NW og og et et M M PS PS ID Source: Authors’ calculation. Source: Authors’ calculation. ID INEQUALITY AND VULNERABILITY TO SHOCKS 17. With a sizeable share of the non-poor just percent is within 20 percent of the poverty line, above the poverty line, many are vulnerable to implying poverty is highly elastic.14 Being just fall into poverty in case of adverse shocks. A above the poverty line and thus barely out of sizeable part of the Somali population consumes poverty, these segments of the population are just enough to be currently considered non- defined as ‘vulnerable’, and are prone to fall back poor: The total daily consumption expenditure into poverty in case of an unexpected decrease of around 10 percent of the non-poor is within in consumption (Figure 1.9). Consequently, a 10 percent of the poverty line, while that of 19 10 percent shock to consumption leads to an 14 An increase of 10 percent in the poverty line is equivalent to a 9 percent decrease in households’ total consumption, while a 20 percent increase in the poverty line is equivalent to a 17 percent decrease in their consumption. The consumption elasticity is equal to approximately 0.5, meaning that a 2 percent increase in the value of the poverty line results, on average, in a 1 percent increase in the poverty headcount. 6 | Part I: Overview of Poverty increase in poverty of 6 percentage points (57 in Sub-Saharan Africa, like Rwanda (50 percent) percent), and a 20 percent shock implies an or the Central African Republic (56 percent). On additional 4 percentage points increase (61 the contrary, inequality levels are similar to least percent). This finding is of particular significance unequal countries in the comparison sample, in the current crisis, where several seasons of such as neighboring Ethiopia (33 percent). insufficient rains and widening droughts are Relatively low levels of inequality are owed to affecting the purchasing power and food security rather homogenous levels of consumption across of large parts of the population, making these the Somali population, with many poor and most scenarios indeed realistic.15 of the non-poor having moderate expenditure levels. In fact, the vast majority of the Somali 18. Inequality is lower than in most low-income population, around 79 percent, lives on less African countries, as Somalis generally share a than US$3.10 2011 PPP per day. Of course, one relatively homogenous level of consumption. of the implications of moderate inequality owed Inequality among the Somali population, as to homogenously low levels of consumption is measured by the Gini index, is 37 percent (Figure significant parts of the population are just above 1.10 and Figure 1.11). Of note, this is significantly the poverty line and thus corresponds to the high lower than the most unequal low-income countries vulnerability to shocks discussed earlier. Figure 1.9: Impact of a consumption shock on poverty. 90 Poverty incidence (% of population) 80 70 60 50 40 30 20 10 0 ge u t t s s ts as es ea ea sh en ra -E -W Ar Ar di m ve rth a rth n l e og la ra ba ttl No Ru No al M Se Ur er P Ov ID Poverty Incidence in 2016 10% shock 20% shock Source: Authors’ calculation. 15 According to internal World Bank estimates, the current drought is estimated to affect Somali total production by 10.6 percent. 51% of Somalis, live on less than US$1.90 2011 PPP per day Part I: Overview of Poverty | 7 Table 1.1: Total average real consumption (per capita, per day in 2016 US$). Q1 Q5 Top/bottom (Bottom (Top quintile Region quintile) Q2 Q3 Q4 quintile) ratio Mogadishu 0.54 0.92 1.20 1.84 3.58 6.6 North East 0.92 1.58 2.09 2.79 4.90 5.3 North West 0.61 1.01 1.43 2.07 3.65 6.0 Urban 0.62 1.09 1.58 2.29 4.09 6.6 Rural 0.65 0.97 1.38 1.93 3.31 5.1 IDP Settlements 0.33 0.62 0.91 1.35 2.53 7.6 Overall average 0.52 0.94 1.38 2.05 3.76 7.2 Source: Author’s calculation. 19. Poverty and inequality are positively 1.11 and Table 1.1). This relationship between related. A clear trend emerges when comparing poverty and inequality notably hinges on the inequality across regions and livelihoods: poorer consumption levels of the poorest: in regions areas are also more unequal. Poverty is least where poverty is widespread, inequality is high widespread in the North East, where inequality because the poorest consume so little that they is also lowest with a Gini index of 32 percent. are much worse off than wealthier households Here, households in Q5 (the top 20 percent in (Table 1.1). However, while some certain Somali terms of consumption expenditure) consume regions are demonstrably more unequal than around 5 times more than households in Q1 others, these variations are within a rather small (the bottom 20 percent). In stark contrast, IDP range, especially when compared to the variation settlements are poorest and at the same time in inequality in the sample of low-income Sub- most unequal, where the Gini index is 38 percent Saharan African countries. As such, even the high and Q5 households have more than 7 times inequality found in IDP settlements is still below higher consumption than Q1 households (Figure the average of this comparison group. Poorer areas are more unequal, and regional disparities exceed differences between urban and rural areas 8 | Part I: Overview of Poverty Figure 1.10: Poverty and inequality Figure 1.11: Consumption distribution. between regions. 40 IDP Settlements 100 Mogadishu North West 80 35 GINI index (0-100) % of population North East North West 60 30 North East North East 40 Mogadishu 25 North West 20 IDP Settlements 20 0 0 20 40 60 80 0 1 2 3 4 5 Poverty incidence (% of population) Average real consumption (per capita, per day in 2016 US$) Urban areas Rural areas Source: Authors’ calculation. Source: Authors’ calculation. 20. Inequality in urban areas is higher than in have almost identical consumption expenditure rural areas, driven by wealthy urban individuals. in urban and rural areas (rural: US$0.65, urban: While rural areas are overall poorer than urban US$0.62), members of Q5 households (the top areas (poverty headcount rural: 52 percent, urban: 20 percent) in urban areas consume 24 percent 45 percent), their consumption levels are more more than in rural areas (rural: US$3.31, urban: homogeneous and hence inequality is lower US$4.09; Table 1.1). Of note, overall trend of (Gini rural: 33 percent, urban: 36 percent). This a positive correlation between poverty and disparity is driven by the wealthier individuals. inequality also holds for urban areas and rural While Q1 household members (the bottom 20 areas individually. percent in terms of consumption expenditure) THE CHARACTERISTICS OF POOR HOUSEHOLDS 21. Poor households have more household areas (household size poor: 6.2, non-poor: 4.7; members than non-poor households. In many Table A.1 in the Appendix). In IDP settlements the economies poverty increases with household difference in household size between the poor size, as an increasing household size is usually and the non-poor (household size poor: 5.7, non- indicative of a higher number of dependent poor: 5.1) is much smaller than elsewhere and household members. The average Somali not statistically significant. In part, this may be household has 5.3 members (Table 1.2). The due to limited statistical power, given that most difference in household size between poor and IDP households are poor. Further, this is plausibly non-poor households is statistically significant a reflection of disrupted household structure both across regions and between rural and urban marking IDP settlements. Poor households also Part I: Overview of Poverty | 9 have a higher number of dependents than non- poor household, while poor households in IDP poor households (Table 1.3). The age dependency settlements have three times as many children ratio, defined as the ratio of children and old age as non-poor IDP households. This implies that dependents to working age population, is 1.7 in children are disproportionately affected by poor households compared to 1.1 for non-poor poverty, an issue which Chapter 5. Child and households. On average, a poor household has youth poverty will explore in depth. twice as many children (aged 0-14) as a non- Figure 1.12: Poverty measures by gender of the household head. 60 54 49 50 % of households 40 30 23 20 20 13 11 10 0 Poverty Headcount Poverty Gap Poverty Severity Female Household Head Male Household Head Source: Authors’ calculation. Table 1.2: Household demographic attributes: size and age dependency ratio. Household size Age dependency ratio Region All Poor Non Poor All Poor Non Poor North East 5.1 6.5 4.8*** 1.5 2.3 1.4 Urban 5.0 6.5 4.7*** 1.5 2.4 1.3 Rural 5.3 6.5 5.0* 2.0 1.8 2.0 North West 5.7 7.1 4.8*** 1.3 1.7 1.1 Urban 5.81 7.4 4.8*** 1.3 1.7 1.0 Rural 5.16 6.0 4.3*** 1.5 1.7 1.2 Mogadishu 4.8 5.5 4.0*** 1.4 1.9 0.9 Urban 5.3 6.5 4.6*** 1.4 1.9 1.1 Rural 5.2 6.1 4.6*** 1.6 1.7 1.6 IDP Settlements 5.5 5.7 5.1 1.3 1.6 0.7 Overall average 5.3 6.2 4.7*** 1.4 1.7 1.1 *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. 10 | Part I: Overview of Poverty 22. Households headed by a woman are less disaggregation of the overall average reveals poor. Just under half of Somali households are considerable heterogeneity across regions and headed by a woman, and those households are along the rural-urban-IDP divide. Households 5 percentage points less likely to be poor overall headed by women are considerably less poor (poverty incidence female household head: 49 in rural areas and in the North East. In contrast, percent, male household head: 54 percent; they are poorer in urban areas and poorer than Figure 1.12). One plausible explanation for this households headed by men in IDP settlements. finding is that households headed by women Households in IDP settlements are also much are more likely to receive financial remittances, less likely to be headed by a woman in the first arguably because working-age men may have place: 6 in 10 households are headed by a left to work elsewhere, a theme which Chapter woman in rural areas, compared to 5 in 10 in 4. Remittances will further explore. In addition, urban areas and 3 in 10 in IDP settlements. Table 1.3: Real consumption (per capita, per day in 2016 US$). Share of households Total average consumption headed by a woman Region Household Household Difference (percent) head: men head: woman (% points) North East 61.0 2.4 2.5 4 Urban 59.6 2.5 2.5 1 Rural 69.2 1.5 2.3 51*** North West 56.2 1.76 1.75 0 Urban 56.0 1.84 1.79 -2 Rural 57.2 1.35 1.52 13** Mogadishu 36.0 1.6 1.6 -4 Urban 51.0 1.9 2.0 4 Rural 60.9 1.4 1.8 31*** IDP Settlements 31.4 1.2 1.0 -21*** Overall average 47.7 17 18 9 *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. 23. Monetary poverty is correlated with worse recipients of remittances is 18 percentage points outcomes along other dimensions of welfare, lower than that of non-recipient. Similarly, the while it is lower and less deep for recipients of poverty gap index for recipients is half of that remittances. The Somali poor have worse access of non-recipients, implying that poor recipients to services, poorer educational outcomes, and are closer to overcoming poverty. Chapter 4. are less successful in the labor market. Chapter Remittances further explores the link between 2. Multidimensional deprivation explores non- remittances, monetary and non-monetary monetary dimensions of poverty in detail. poverty, and resilience. In contrast, the poverty headcount rate of Part I: Overview of Poverty | 11 2. MULTIDIMENSIONAL DEPRIVATION KEY MESSAGES Poverty strongly correlates with labor market outcomes, level of education, and access to improved quality of dwellings and infrastructure. People living in North East, where poverty is less widespread and deep, have highest levels of employment, educational attainment, and access to improved water and sanitation systems. People living in IDP settlements, where poverty is most severe, are most deprived in all dimensions. Increasing active participation in the labor market is key to improve welfare and decrease inequality. The most serious obstacles affecting labor force participation are conflict-related insecurity and disability, each of these constraints warranting specific intervention through social protection measures. Investments in basic infrastructure, such as water and sanitation systems, and education, are strongly needed in all Somali regions, particularly in rural areas. The Somali population lags behind most low-income African countries in access to improved water and sanitation, and educational attainment. The planned Poverty Assessment will provide a more in-depth analysis including a focus on the gender dimension of poverty and a detailed education analysis including the identified education - health nexus. The gender analysis will include non-monetary aspects of poverty and estimate the gender impact on poverty by controlling for observables like education. The gender analysis will also investigate in more detail the role of women in the economy given their contributions in the informal sector and subsistence farming that are not well reflected in the labor market statistics. The education analysis will analyze constraints to education as well as estimate returns to education to better understand potential entry points to improve educational outcomes with a focus on the identified linkages between education and health. 24. Monetary and non-monetary poverty are of households have no access to information). strongly related with poor households often Monetary poverty is the second most common deprived in multiple dimensions. For the Somali deprivation, affecting 45 percent of Somali population, lack of access to information is the households.16 Lack of access to an improved most common type of deprivation (71 percent source of water and to education affect 41 and 16 Because household size is larger in poor households, the poverty headcount ratio is 51 percent when counting the single individuals, and 45 percent when considering the single households. 12 | Part I: Overview of Poverty 36 percent of Somali households, respectively in 2 or more dimensions (Figure 2.2). Poverty is (Figure 2.1). For rural households though, lack a strong indicator of non-monetary deprivation. of access to an improved source of water is the Households living in rural areas and IDP most common deprivation, with more than 9 in settlements are also much more likely to be 10 rural households deprived in this dimension. more deprived than households living in North 9 in 10 Somali households are deprived in at East, North West, and Mogadishu (Table A.3 in least one dimension, while 2 in 3 are deprived the Appendix). LITERACY AND EDUCATION 25. The level of literacy and educational income Sub-Saharan countries, while 7 percent attainments of the Somali people is slightly of the population has obtained a secondary lower than those of African low-income education degree compared to 19 percent in countries, after taking into account differences low-income Sub-Saharan countries (Figure 2.3, in GDP. 55 percent of Somali people can read Figure 2.4 and Figure 2.5).17 The literacy rates and write, compared to an average value of 56 presented in the analysis have some limitations, percent for low-income Sub-Saharan countries. since they are non-functional and were self- 16 percent of Somali people have completed reported by interviewed households. primary school compared to 34 percent in low- Figure 2.1: Multidimensional deprivation by category. 100 80 % of households 60 40 20 0 NE Urban NE Rural NW Urban NW Rural Mogadishu Urban Rural IDP Camps Overall Monetary Poverty Education Water Information Source: Authors’ calculation. 17 Among low-income Sub-Saharan countries, Zimbabwe has the highest literacy rate (87 percent), level of primary education (81 percent) and secondary education (61 percent), while Niger, Burkina Faso, and Chad have the lowest level of literacy (19 percent), primary education (5 percent) and secondary education (6 percent), respectively. Part I: Overview of Poverty | 13 Figure 2.2: Multidimensional deprivations. 100 80 % of households 60 40 20 0 NE Urban NE Rural NW Urban NW Rural Mogadishu Urban Rural IDP Camps Overall At least 1 At least 2 At least 3 All 4 Source: Authors’ calculation. 26. Poor Somalis have a lower level of literacy than differences in education between poor and education than the non-poor population, and non-poor, especially for university and and the educational gap between regions and secondary education. People living in North between urban and rural areas is even higher, East, where poverty is less widespread and deep, thus it is mostly driven by a geographical lack have the highest level of literacy and primary of access. 48 percent of the poor can read and education; about 10 percentage points and 3 write, compared to 62 percent of the non-poor percentage points higher than the overall average, (Figure 2.7). 13 percent among poor Somalis respectively. Similarly, people living in Mogadishu have completed primary education, compared have the highest level of completed secondary to 18 percent among the non-poor (Figure 2.8 and tertiary education. Rural areas in North East and Figure A.1 in the Appendix). Only 5 and 3 show particularly high level of literacy and primary percent of the poor have completed secondary education when compared to rural areas in North and tertiary education, respectively, compared West. People living in IDP households, where the to 9 and 8 percent among the non-poor (Figure poverty incidence and gap are highest, have the A.2 in the Appendix). Differences in education lowest literacy rate, 14 percentage points lower between rural and urban areas tend to be larger than the overall average. 48% 62% of the poor can read and write of the non-poor can read and write 14 | Part I: Overview of Poverty Figure 2.3: Literacy rate in Sub-Saharan Figure 2.4: Educational attainment (primary) in low-income countries. Sub-Saharan low-income countries 100 100 80 TZA 80 % of population % of population MWI TZA 60 60 SSD SOM ETH 40 40 SSD MWI 20 20 ETH SOM 0 0 200 400 600 800 1,000 200 400 600 800 1,000 GDP per capita (2015, US$) GDP per capita (2015, US$) Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. 27. Poverty is strongly associated with children in Mogadishu and IDP Settlements. Households enrollment in school, as poor households living in North East spend on education more are less likely to spend on education. Poor than 50 percent and more than 100 percent household spends on average US$ 25 per year than households in North West and Mogadishu, in education, compared to US$ 47 for the non- respectively.18 Disparities in school enrollment poor (Figure 2.11). Only one in two Somali between gender are less pronounced than children (52.9 percent) are enrolled in school between poor and non-poor. On average, school against an average of about 70 percent in low- enrollment is 4 percentage points higher among income African countries (Figure 2.6). About 63 boys, with the lowest gap occurring in Mogadishu percent of children living in non-poor households and North East. Boys living in IDP settlements are enrolled in school, compared to 45 percent have a much lower school enrollment than girls for children living in poor households (Figure (Figure 2.10). School enrollment in household 2.9). Large disparities emerge when comparing headed by a woman is much lower than among enrollment and educational expenditures across male-headed households in Mogadishu and in regions. 6 in 10 children are enrolled in school in IDP Settlements, where poverty is more severe North East and North West, compared to only 4 (Figure 2.12). 18 Differences in non-food expenditures across regions (i.e. expenditures in education, health services, electricity, etc.) may be caused by regional differences in prices, which depend on the relative supply, demand, and degree of tradability for that product/service. Part I: Overview of Poverty | 15 Figure 2.5: Educational attainment (secondary) Figure 2.6: School enrollment (primary age) in Sub-Saharan low-income countries. in Sub-Saharan low-income countries. 100 100 MWI % of children aged 6-14 80 80 ETH TZA % of population 60 60 SOM 40 SSD 40 SSD 20 20 MWI ETH TZA 0 SOM 0 200 400 600 800 1,000 200 400 600 800 1,000 GDP per capita (2015, US$) GDP per capita (2015, US$) Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. Figure 2.7: Literacy. Figure 2.8: Educational attainment, primary. 80 20 % of population % of population 60 15 40 10 20 5 0 0 n l n og ral u n Ca l ps l NE an Ur l NW an ad l u n Ca al ps l ra ra al ra a al ba ba ba ba ish ish M Rur r m m er er b b Ru Ru Ru NW Ru Ru Ur Ur Ur Ur Ur ad Ov Ov NE NW NE NW NE og P P ID ID M Poor Non Poor Overall Poor Non Poor Overall Source: Author’s calculation. Source: Author’s calculation. Figure 2.9: Net primary school enrollment. Figure 2.10: Net primary school enrollment, by gender. % of children aged 6-14 % of children aged 6-14 80 80 60 60 40 40 20 20 0 0 n l n og ral u n Ca l ps l NE an Ur l NW an ad l u n Ca al ps l ra ra al ra a al ba ba ba ba ish ish M Rur r m m er er b b Ru Ru Ru NW Ru ID Ru Ur Ur Ur Ur Ur ad Ov Ov NE NW NE NW NE og P P ID M Poor Non Poor Overall Women Men Overall Source: Author’s calculation. Source: Author’s calculation. 16 | Part I: Overview of Poverty Figure 2.11: Mean household expenditures Figure 2.12: Net primary school enrollment by in education. gender of household head. % of children aged 6-14 80 80 60 60 US$ per year 40 40 20 20 0 0 an NW ral NW an ad l u an Ca al s l NE an Ur l NW an ad l u an Ca ral s l a al ra a al pu pu ish ish M Rur r M Rur er er b b b b b b Ru Ru NW Ru Ru m m Ur Ur Ur Ur Ur Ov Ov NE NE NE og og P P ID ID Poor Non Poor Overall Woman Man Overall Source: Author’s calculation. Source: Author’s calculation. EMPLOYMENT AND PARTICIPATION TO THE LABOR MARKET 28. Labor force participation and employment low because of two survey limitations: First, rate of Somalis are lowest among African low- the labor indicators were obtained from income countries. Only 1 in 4 people of working the household member that responded the age are active labor participants, having either survey on behalf of the other members of the worked or seeking work in the last seven days, household, instead of every person responding compared to an average 76 percent in low- on their own. Second, a substantial part of the income Sub-Saharan countries (Figure 2.13). inactive (20 percent), i.e. those who are not Furthermore, labor force participation ranges seeking employment nor have worked in the between 65 and 88 percent in 21 of the 25 reference period, report “taking care of own countries used for international comparison. household” as the main reason for their status Similarly, only 2 in 10 Somalis are employed, of inactive, which may include economically compared to an average 70 percent in low- relevant activities for the household. Both income Sub-Saharan countries (Figure 2.14). factors may lead to underestimation of labor The employment rate ranges between 57 and force participation and employment. 83 percent in 22 of the 25 countries used for international comparison.19 30. Poor households less often find employment compared to non-poor households. On average, 29. Survey limitations warrant some caution employment among the poor is 9 percentage in the interpretation of the labor indicators. points lower than among non-poor (Figure The reported labor force participation and 2.15). Employment rates in urban and rural employment indicators might be unexpectedly areas, as well as across the different regions 19 The lower and upper bound for labor force participation and employment are obtained by respectively subtracting and adding the standard deviation to the mean value computed for that measure. Part I: Overview of Poverty | 17 are not significantly different, despite lower who take care of their own household, which may levels of poverty among those living in urban include economically relevant activities such as areas. Higher labor force participation is weakly agriculture and livestock farming. Indeed, own indicative of the level of poverty, with labor force household work is highest in rural areas (26 participation being higher among the non-poor percent) and lower in urban areas (18 percent), in all regions except for the IDP settlements and IDP settlements (15 percent), and Mogadishu the rural areas of North East. (11 percent). Overall, own household work is weakly correlated with poverty and does not 31. 1 in 5 adults are outside the labor force vary significantly across regions. In North East and taking care of their own household. Own North West, poor households have slightly higher household work is highest in rural areas. Labor level of own household work, while the relation force participation and employment may be is reversed in urban areas such as Mogadishu and underestimated since they do not include Somalis IDP settlements (Figure 2.17). Figure 2.13: Labor force participation in Figure 2.14: Employment in Sub-Saharan low- Sub-Saharan low-income countries. income countries. 100 100 TZA % of working age population % of working age population ETH ETH TZA MWI MWI 80 80 60 60 40 40 SOM 20 20 SOM 0 0 200 400 600 800 1,000 200 400 600 800 1,000 GDP per capita (2015, US$) GDP per capita (2015, US$) Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. Figure 2.15: Employment. Figure 2.16: Labor force participation. 40 40 % of working age % of working age 30 population population 20 20 10 0 0 l n l n og ral n Ca l ps l NE an Ca al ps u Ur l NW ban ad l u n al al ra a ra a ba ba ba ba ish ish r r M Rur m m er er b Ru Ru Ru Ru NW u Ur Ur Ur Ur Ur R ad Ov Ov NE NW NE NW NE og P P ID ID M Poor Non Poor Overall Poor Non Poor Overall Source: Author’s calculation. Source: Author’s calculation. 18 | Part I: Overview of Poverty 32. The Somali labor market reveals a large Figure 2.17: Own household work. gender gap, as evidenced by an extremely low labor force participation rate among women. 30 Labor force participation among men is 32 % of working age population percent, compared to 18 percent among women. 20 Employment among men is 32 percent, more 10 than three times higher than among women (Figure 2.18). The gender gap in employment 0 (23 percent) is much higher that any regional an NW ral NW an ad l u n Ca l ps l a ra al ba ish M Rur m er b b Ru Ru Ur Ur Ur Ov (4 percent) or urban-rural (3 percent) disparity NE NE og P ID in employment. The North East region has Poor Non Poor Overall the lowest level of employment gap between women and men. Source: Author’s calculation. 33. Gender disparities are also evident among behind inactivity, compared to almost 57 percent Somali men and women outside the labor of inactive men. More than 16 percent of inactive market, as evidenced by different causes for women and 30 percent of inactive women living inactivity. In line with employment disparity, in Mogadishu report “not being allowed by the inactivity status greatly varies between men husband” as the main reason for inactivity. While and women (Figure A.3 and Figure A.4 in the the gap in school enrollment between boys and Appendix). For every second woman, housework girls (aged 6-14) is relatively small (4 percent, is the main reason for being out of the labor force, Figure 2.10), the gender gap in school enrollment compared to 6 percent of inactive men. On the between inactive men and women is indicative of other hand, only 19 percent of inactive women the lack of educational opportunities for Somali report school enrollment as the primary reason women once they reach adulthood. Figure 2.18: Employment by gender. Figure 2.19: Labor force participation by gender. 40 60 % of working age % of working age 40 population population 20 20 0 0 n l n u l u ps n l n l n l ps n l ra ra ra ra ra ra ba ba ba ish ba ba ba ish m m Ru Ru Ru Ru Ru Ru Ur Ur Ur Ur Ur Ur ad ad Ca Ca NE NW NE NW NE NW NE NW og og P P ID ID M M Women Men Women Men Overall women Overall men Overall women Overall men Source: Author’s calculation. Source: Author’s calculation. 34. The reportedly low labor force participation suggest that women are mainly inactive and are of women is in contrast to the role of women working in the household, other studies suggest in the economy. While the reported indicators a much more active role of Somali women in Part I: Overview of Poverty | 19 the private sector.20 They are engaged in the is highly indicative of conditions for the Somali informal sector and micro-enterprises, but population. 3 in 4 Somalis who report insecurity also play a role in agricultural production and due to conflict as the main reason for inactivity livestock activities. Data limitations as explained are poor, compared to 1 in 2 at the national level above might contribute to the discrepancy with (Figure 2.20). the reported findings. Figure 2.20: Poverty headcount ratio, 35. Illness, disability and fear of conflict are by inactivity reason. all important factors that prevent men more than women from participating in the labor 80 force. Insecurity due to conflict is reported by 6 % of population, by 60 inactivity reason percent of inactive men while being negligible for women. Disability or illness is reported by 40 12 percent of inactive men compared to only 20 4 percent for inactive women. Not surprisingly, conflicted-related insecurity affects one in four 0 men’s decision to stay out of the labor force in ict y l d ds o lit ol ho nfl bi bi eh r sc co fo isa Mogadishu. The relatively high prevalence of us y/ In d /d ho rit an ss cu n illness-related reasons for men’s inactivity is sb ne Ow se Hu Ill In particularly compelling and indicative of the need of health services that can target this Overall, inactive Overall group. While disability or illness is not strongly Source: Author’s calculation. correlated with poverty, insecurity due to conflict ACCESS TO INFRASTRUCTURE AND QUALITY OF DWELLINGS 36. Somali households lag behind most low- affecting children’s health and time allocation, low income countries in access to improved source of quality water and sanitation services negatively water and sanitation facilities. Improvements in influences their educational attainment.21 Only access to water and sanitation are key for economic about 60 percent of Somali households have and social development. Water and sanitation are access to an improved source of water, compared essential for the individual’s health, as well for to an average of about 70 percent in low-income their productive activities, such as agriculture. African countries. Somali households rank eight Inadequate water and sanitation services increase lowest among Sub-Saharan low-income countries children’s exposure to waterborne diseases. In in terms of access to improved source of water. addition to that, low accessibility to such services About 10 percent of households have access to affects the time children need to employ to improved sanitation facilities, compared to an satisfy their basic water and sanitation needs. By average of 25 percent in low-income African 20 UNDP (2012) ‘The Role of Somali Women in the Private Sector’, UNON Publishing Services, Nairobi. 21 Impact of access to water and sanitation services on educational attainment, 2016, Ortiz-Correa, Filhoa, Dinarb. 20 | Part I: Overview of Poverty countries. The Somali population rank second in terms of access to improved sanitation (Figure lowest among Sub-Saharan low-income countries 2.21 and Figure 2.22).22 Figure 2.21: Access to improved source of Figure 2.22: Access to improved sanitation in water in Sub-Saharan low-income countries. Sub-Saharan low-income countries. 100 MWI 100 80 80 % of households % of households SOM ETH 60 60 SSD TZA 40 40 MWI ETH 20 20 TZA SOM SSD 0 0 200 400 600 800 1,000 200 400 600 800 1,000 GDP per capita (2015, US$) GDP per capita (2015, US$) Source: Authors’ calculation and World Bank Open Data. Source: Authors’ calculation and World Bank Open Data. Figure 2.23: Access to improved source of water. Figure 2.24: Access to improved sanitation. 100 20 % of working age % of working age 80 15 population population 60 10 40 5 20 0 0 n NW ral NW an ad l u n l ps l NE an Ur l NW an ad l u n Ca al ps l a ra al ra a al ba ba ba ish ish M Rur M Rur r m m er er b b b Ru Ru NW Ru ID Ru Ur Ur Ur Ur Ur Ca Ov Ov NE NE NE og og P P ID Poor Non Poor Overall Poor Non Poor Overall Source: Author’s calculation. Source: Author’s calculation. 37. Access to an improved source of water cause for this deprivation. 70 percent of the greatly varies between urban and rural areas, population living in urban areas has access to an signaling that lack of infrastructure is the main improved source of water, compared to 21 percent 22 Access to an improved water source refers to the percentage of the population using an improved drinking water source. The improved drinking water source includes piped water on premises (piped household water connection located inside the user’s dwelling, plot or yard), and other improved drinking water sources (public taps or standpipes, tube wells or boreholes, protected dug wells, protected springs, and rainwater collection). Part I: Overview of Poverty | 21 for households living in rural areas and forty source of water between urban and rural areas. percent for households living in IDP settlements. Indeed, more than seven in ten people living in Given such a large gap between urban and rural urban households of North East have access to areas, access to an improved source of water is an improved source of water, against about 5 in more strongly correlated to welfare conditions in ten in rural areas; a stark contrast to the North rural areas, where the access is a relatively scarcer West region, where only 52 percent of urban resource. In line with other relevant non-monetary dwellers and 9 percent of people living in rural indicators, such as education and employment, households report access to an improved source households living in North East show a relatively of water (Figure 2.23). low degree of inequality in access to an improved Figure 2.25: Quality of the roof. 100 80 % of households 60 40 20 0 Urban Rural IDPs Urban Rural IDPs Urban Rural IDPs METAL SHEETS PLASTIC SHEET OR CLOTH HARAR (SOMALI TRADITIONAL) Poor Non Poor Source: Author’s calculation. Figure 2.26: Quality of the floor. 80 60 % of households 40 20 0 Urban Rural IDPs Urban Rural IDPs Urban Rural IDPs CEMENT MUD TILES (CERAMIC) Poor Non Poor Source: Author’s calculation. 38. Similarly, large disparities are evident in people living in rural, urban, and IDP settlements. access to improved sanitation facilities, both Only 2 percent of people living in rural areas have between poor and non-poor, and between urban access to an improved sanitation system, compared and rural areas. The largest variation in access to to 13 percent in urban areas. Variation across improved sanitation is observed primarily between Somali regions are statistically negligible. About 22 | Part I: Overview of Poverty 14, 12 and 11 percent of people living in North households. Differences within urban non-poor East, Mogadishu, and North West respectively, households are less pronounced than differences have access to an improved sanitation system. within urban poor households across different Access to an improved sanitation system strongly regions. Expenditures on electricity among non- correlates with poverty. Overall less than 5 percent poor urban households range between US$ 57 of people living in poor households have access (North West) and US$ 61 (Mogadishu) compared to an improved sanitation system, compared to to US$ 13 (North West) and US$ 18 (Mogadishu) 15 percent of non-poor households (Figure 2.24). for urban poor households. As for access to water and sanitation, expenditures on electricity are 39. Dwelling quality weakly correlates with more indicative of welfare conditions in rural poverty. In the vast majority of dwellings among areas, where access to the resource is relatively Somali households, roofs are made of metal sheets. scarcer. The expenditure on electric devices is US$ In urban areas, 85 and 86 percent of poor and non- 2 per person per year among poor households poor households have a metal roof, respectively. In living in rural area, compared to US$ 11 among rural areas and IDP settlements, a metal roof is an non-poor households living in rural areas (Figure indication of being non-poor. 71 percent of non- A.5 in the Appendix). poor households living in rural areas have a roof made of metal sheets, compared to 40 percent among poor households. Analogously, 56 percent 70% of the urban population of non-poor households living in IDP camps have a has access to an improved roof made of metal sheets, compared to 33 percent among poor households (Figure 2.25). In the vast source of water compared to majority of dwellings among Somali households, 21% in rural areas. floors are made of cement, 66 and 72 percent of poor and non-poor households living in urban areas, respectively. Similar to metal roofs, non-poor rural 15% of the non-poor and IDP households often have a cement floor (54 have access to improved percent vs 25 percent for poor households). Poor households much more often have a floor of mud sanitation, compared to 6% (54 percent vs 32 percent for non-poor households of the non-poor. (Figure 2.26). 40. Expenditures in electrical devices are a strong indicator of welfare among Somali households. Expenditures in electrical devices may be used as a proxy for access to electric infrastructure.23 The average expenditure on electrical devices is US$ 31 per person, per year, showing a large variation between poor and non-poor households.24 Non- poor households spend on average US$ 47 on electricity, compared to a mere US$ 9 among poor 23 Differences in expenditures on electric devices across regions may be influenced by regional differences in prices, which depend on the relative supply, demand, and degree of tradability. Furthermore, expenditures in electricity may be underestimated in rural areas, where access to electricity is obtained through power generators to a greater extent than in urban areas. 24 Electrical devices include light bulbs, internet/cable TV, expenditures for electricity, music or video cassette or CD/DVD, electric stove or hot plate, Tape or CD/DVD player, HiFi, Television, VCR, Computer equipment & accessories, Satellite dish. Part I: Overview of Poverty | 23 ACCESS TO HEALTHCARE 41. Access to healthcare is substantially higher (urban) and US$ 12 (rural), about 60-90 percent in urban areas. Hospitals seem to be more more than non-poor households located in North likely located in urban areas, as children born in West and Mogadishu. Expenditures in healthcare urban areas are more likely to have been born in among the poor is similarly distributed. Poor hospitals or clinics compared to children born in households of North East spend about US$ 3.6 rural areas. Similarly, to access to water, sanitation (urban) and US$ 3.3 (rural) per year per person, and electricity, access to hospitals or clinic does while households in North West and Mogadishu not significantly vary between poor and non-poor spend US$ 2 (North West urban), US$ 1.2 (North in urban areas, but does significantly vary when West rural) and 2.6 (Mogadishu). In stark contrast considering rural areas and IDP settlements (Figure to the poor/non-poor divide, there is very little 2.28). This evidence supports the hypothesis that difference in health care expenditures between in areas where the resource is relatively scarcer, poor and non-poor households living in IDP only relatively better-off households are able to camps (US$ 1 for poor, US$ 1.2 for the non- afford it. poor). Consistently with the evidence found in the previous chapter (Figure 1.9 and Table 1.1), 42. Poor households spend significantly less on the relative smaller difference in health care health care than non-poor households. Average expenditures between poor and non-poor in IDP annual expenditures in healthcare are about US$ settlements is indicative of the higher degree 2 and US$ 8 per person among poor and non-poor of affinity between these two groups and the households, respectively (Figure 2.7). Non-poor relatively higher level of vulnerability of non-poor households of North East spend about US$ 11 households living in IDP settlements. Figure 2.27: Average annual health expenditures. Figure 2.28: Child birth in hospital or clinic. 14 100 12 80 % of born children 10 US$ per capita 8 60 6 40 4 20 2 0 0 NE ban l NW an ad l u n al ps l NE an Ur l n al u n Ca l ps l NW ura a al a ra al ba NW a ba ish ish M Rur ID Rur NW r M Rur m m er er b b b Ru Ru Ur Ur Ur Ur Ur R ad Ca Ov Ov NE NE og og P P ID Poor Non Poor Overall Poor Non Poor Overall Source: Author’s calculation. Source: Author’s calculation. 24 | Part I: Overview of Poverty 3. EVOLUTION OF WELFARE CONDITIONS IN THE NORTH WEST REGION KEY MESSAGES Poverty incidence decreased between 2013 and 2016 from 69 percent to 64 percent in rural areas, and from 57 percent to 52 percent in urban areas. The decrease in poverty incidence was similar in rural and urban areas, but poverty remains more widespread in rural areas. In the same period, the poverty gap decreased from 29 to 24 percent in rural areas, and somewhat in urban areas from 20 to 19 percent. Thus, monetary poverty reduction was stronger in rural than in urban areas. The decrease in rural poverty is unlikely to be associated with remittances, while in urban areas poverty increased for recipients. Between 2013 and 2016, poverty incidence increased 8 percentage points among urban households that received remittances, and decreased 9 percentage points among urban non-receivers. In rural areas, poverty decreased largely (23 percentage points) for receivers of remittances and moderately for non-receivers (4 percentage points). The urban increase in poverty might be explained by a mixing effect with some urban receivers graduating from poverty not requiring remittances anymore and other urban poor households starting to receive remittances. The reduction in rural poverty is unlikely to be caused by remittances as a similar number of households received remittances, which on average were smaller. The educational gap has widened for the rural poor between 2013 and 2016. While the population in urban areas became more literate from 2013 to 2016, poor households in rural areas became less literate. The increase in the literacy rate in urban areas is likely to be associated with higher levels of education, since the share of people with no education in urban areas decreased during the same period. In rural areas, non-poor households maintained a similar literacy rate, yet poor households experienced a decreased of 6 percentage points. A larger share of the rural poor does not have any education in 2016 compared to 2013. The rural poor seem to be increasingly excluded in terms of education which complicates their path out of poverty. In order to reduce inequality and poverty, access and availability to key services must be improved for poor households, since current programs leave them behind, particularly in terms of education. Worse educational levels for the rural poor are probably caused by lower attendance to school, which decreased around 8 percentage points for this group. School attendance increased in urban areas and remained relatively constant for the rural non-poor. Providing access and means to reap the benefits from education is crucial to achieve positive labor outcomes and to ultimately lift these households out of poverty. Part I: Overview of Poverty | 25 43. The Somali North West region records Figure 3.1: Poverty incidence 2013-2016. moderate welfare gains between 2013 and Poverty incidence (% of population) 2016. Only for the North West region, there is a 100 previous survey measuring poverty in 2013. Using 90 80 comparable figures in 2013 and 2016 (Box 3), 70 69 64 poverty incidence decrease from 69 percent to 64 60 57 50 52 percent in rural areas, while from 57 to 52 percent 40 in urban areas.25 The magnitude of the decrease in 30 poverty was similar in rural areas than in urban areas 20 10 (Figure 3.1), but poverty remains more widespread 0 in rural areas. More than one in two people live in 2013 2016 poverty in urban areas, as opposed to nearly two in three in rural areas. The annual rate of poverty Urban areas Rural areas reduction was 1.5 percentage points in urban areas Source: Author’s calculation. and 1.8 percentage points in rural areas. 44. Improvements in welfare conditions 45. Inequality also decreased and is now higher between 2013 and 2016 benefited more poor in urban areas than in rural areas. In 2013 the households in rural areas than in urban areas. Gini coefficient, a measure of inequality, was In 2013 the poverty gap was 20 percent in urban estimated at 43 for urban areas and 46 among rural areas and 29 percent in urban areas, while 19 households, while at 34 and 32 in these regions percent and 24 percent in 2016 for urban and during 2016 (Figure 3.3). Inequality decreased rural areas respectively. The rural poor still have from 2013 to 2016 in both urban and rural areas a larger consumption deficit than their urban by 9 percentage points and 14 percentage points counterpart, since on average their consumption respectively. In 2013, inequality was larger in is further away from the poverty line. Between rural than in urban areas, a trend that has been 2013 and 2016, the poverty gap in the North reversed in 2016. Improvements in welfare West region decreased 5 percentage points in conditions between 2013 and 2013 in the North rural areas, and only 1 percentage point in urban West were reflected in smaller poverty incidence, areas (Figure 3.2). Thus, rural poor households as well as in poor households with an average benefited more than the urban poor from the expenditure closer to the poverty line, which improvements in welfare conditions in this period, ultimately must have helped to a less unequal since they had a larger reduction in the poverty distribution of total expenditure. Nonetheless, gap and a similar reduction in poverty incidence. large inequality figures in 2013 relative to 2016 On average, poor rural households have a higher could also be driven artificially by outliers in the expenditure and are closer to the poverty line in consumption aggregate estimates from 2013. 2016, compared to 2013. 25 The data from 2013 presented in this chapter was obtained from the datasets of the 2013 Somaliland Household Survey (SLHS). The estimates were revised to ensure comparability with the data collected as part of the Somali High Frequency Survey in 2016. Box 3 describes in more detail the procedure to arrive at comparable poverty estimates. Moreover, due to the sampling design of the 2013 survey, the analysis is conducted separately for urban and urban areas. 26 | Part I: Overview of Poverty Box 3: Creating comparable poverty estimates for 2013 and 2016 This chapter uses data from SLHS 2013 to items that were not covered in the 2016 understand changes in poverty and other questionnaire (Table A.4 in the Appendix). socio-economic characteristics in the three Poverty incidence was calculated using years to 2016. The SLHS 2013 employed pen- the international poverty line of US$ 1.90 and-paper interviewing (PAPI) and a separate (2011 PPP) deflated to 2013. In 2011, US$ sampling frame for urban and rural areas. In 1 (2011 PPP) was worth 10,731 Somali contrast, the SHFS 2016 was implemented Shillings (2011 PPP). To obtain the amount u s i n g co m p u t e r - a ss i s t e d p e rs o n a l for 2013, inflation has to be considered interviewing (CAPI), a rapid consumption measured at 58.4 percent between 2011 and methodology and a more robust sample 2013. Finally, the average exchange rates of frame for rural households. Furthermore, the Somali Shillings and Somaliland Shillings questionnaires had a number of differences, against the US$ were used with Somali including the consumption module. Therefore, Shilling 20,360.53 and Somaliland Shilling the poverty estimates for 2013 must first 6,733.69 for US$ 1. Thus, the US$ 1.90 (2011 be made compatible with 2016 before PPP) poverty line corresponds to 10,680.11 comparisons can be carried out. Somaliland Shillings per person per day in 2013. Finally, the poverty line was scaled More specifically, the originally estimated to account for consumption items included poverty incidence in 2013 is not comparable in the 2016 but not the 2013 questionnaire with that of 2016 for two main reasons. First, (Table A.5 in the Appendix). The scale factor the questionnaires considered different food, was calculated by estimating the average non-food and durable items. Second, the SHFS consumption in 2016 covered by those items 2016 considers a standard international missing in the 2013 questionnaire. poverty line while SLHS 2013 derived a poverty line based on a needs-based The robustness of this standard methodology approach using an average calorie intake as is shown by comparing that the poverty reference point and an allowance for non- incidence in 2016 is relatively similar when food consumption. considering the total consumption and a standard international poverty line, against the In order to compare 2013 and 2016, the comparable consumption aggregate and the consumption aggregate for 2013 was scaled poverty line (Figure A.6 in the Appendix). adjusted by excluding food and non-food 46. A large share of the population has an respectively consumed less than this amount expenditure level below US$ 2 per day in 2013 in 2016. Slightly higher consumption can be and 2016. The share of population with a daily observed in 2016, mainly in rural households, expenditure level below US$ 2 increases rapidly in line with a decrease in poverty incidence and until this mark. The same pattern is observed poverty gap. Still, consumption is higher for urban between 2013 and 2016 in urban and rural households compared to the rural population. areas (Figure 3.4 and Figure 3.5). Nearly 60 and In addition, there are larger differences in 72 percent of the urban and rural population expenditure levels between 2013 and 2016 for Part I: Overview of Poverty | 27 the population at the top of the distribution or aggregate for some urban and rural households with the highest levels of expenditure. This mainly in 2013. corresponds to outliers in the consumption Figure 3.2: Poverty gap. Figure 3.3: GINI coefficient. 35 50 Poverty gap (% of poverty line) 45 46 30 GINI coefficient (0-100) 29 40 43 25 24 35 34 20 20 19 30 32 25 15 20 10 15 10 5 5 0 0 2013 2016 2013 2016 Urban areas Rural areas Urban areas Rural areas Source: Author’s calculation. Source: Author’s calculation. 47. The share of poor households receiving 48. The amount of remittances received remittances is similar between 2013 and 2016, decreased from 2013 to 2016, except for urban while that of non-poor increased in rural areas non-poor households.26 In 2013 non-poor and decreased in urban areas. Nearly 16 percent households received remittance for an average poor and 30 percent non-poor households amount of US$ 632 per capita in rural areas and received remittances in urban areas in 2013, US$ 367 per capita in urban areas, followed by the and only 6-7 percent in rural areas. In 2016, 19 urban poor with US$ 329 and lastly by the rural percent of the urban poor and 23 percent of the poor with US$ 242 per capita. Three years later, urban non-poor were recipients, while 13 percent urban non-poor received an average of US$ 445 and 15 percent of the rural poor and non-poor per capita, rural non-poor US$ 277, while the rural respectively (Figure 3.6). The share of poor poor US$ 238 and the urban poor US$ 227 per households receiving remittances in 2013 and capita (Figure 3.7). The value of remittances per 2016 -urban and rural- is not statistically different. capita that households received decreased slightly For non-poor households, the share of receivers from 2013 to 2016 for the rural poor, around 50 in urban areas decreased by 7 percentage and 30 percent for urban poor and rural non-poor points and it more than doubled in rural areas. households, respectively. In the same period, Households in rural areas are still less likely to urban and non-poor households experienced receive remittances than urban households. The an increase in the value of remittances received, urban-rural gap in terms of share of households which could have helped them not to be classified receiving remittances decreased for poor and as poor in 2016. Poor households still receive, non-poor households between 2013 and 2016. on average, smaller amounts of remittances than 26 These figures should be interpreted with caution as they are lower than those reported by other sources. The information on remittances collected is likely to be under-reported by households surveyed. However, it is expected that the under-reporting is random and not concentrated in a group of households with certain characteristics. 28 | Part I: Overview of Poverty non-poor households. Yet, the gap in the value decreased in rural areas and increased in urban received between poor and non-poor households areas between 2013 and 2016. Figure 3.4: Distribution of consumption in Figure 3.5: Distribution of consumption in urban areas. rural areas. 100 100 90 90 Percentage of the population Percentage of the population 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 2016 2013 10 2016 2013 0 0 0 2 4 6 8 10 12 0 2 4 6 8 10 12 Daily consumption expenditure per capita Daily consumption expenditure per capita (2011 US$ PPP) (2011 US$ PPP) Source: Author’s calculation. Source: Author’s calculation. Figure 3.6: Households that received remittances. Figure 3.7: Value of remittances for receivers. 35 700 632 30 30 600 Annual value per capita % of households that received remittances 25 23 500 455 (2011 US$ PPP) 19 367 20 400 329 16 15 277 15 13 300 227 242 238 10 7 200 6 5 100 0 0 Urban: Urban: Rural: Rural: Urban: Urban: Rural: Rural: Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor 2013 2016 2013 2016 Source: Authors’ calculation. Source: Authors’ calculation. 49. In urban areas, poverty incidence increased that did not receive remittances (62 percent) than for households that received remittances and among urban recipients (41 percent). In 2016 the decreased for non-receivers. In 2013 poverty gap between urban recipients and non-recipients incidence was higher among urban households decreased, since poverty incidence in the former Part I: Overview of Poverty | 29 group was 49 percent while in the latter 53 non-recipient households (69 percent). In 2016 percent (Figure 3.8). Over the last three years, the pattern was reversed as poverty was higher poverty incidence increased 8 percentage points for non-recipients (65 percent) relative to those among households that received remittances, and households that received remittances in rural decreased 9 percentage points among urban non- areas (55 percent; Figure 3.8). Poverty incidence receivers. The share of poor households receiving decreased largely (23 percentage points) for remittances was similar in 2013 and 2016 but the receivers of remittances and moderately for average amount received declined for the urban non-receivers (4 percentage points) in rural poor (Figure 3.7). The urban increase in poverty areas. The reduction in rural poverty is unlikely can be explained by a mixing effect with some to be caused by remittances as the share of urban receivers graduating from poverty not poor rural households receiving remittances requiring remittances anymore and other urban is not statistically different in 2013 and 2016, poor households starting to receive remittances. and households received, on average, smaller amounts per capita in 2016 (Figure 3.6 and Figure 50. In rural areas, poverty incidence decreased 3.7). Furthermore, the urban-rural gap in terms largely for receivers of remittances and of share of households receiving remittances moderately for non-receivers. In 2013 poverty decreased for poor and non-poor households incidence was highest among rural recipients between 2013 and 2016. of remittances (79 percent), followed by rural Figure 3.8: Poverty and remittances. 90 80 79 (% of population) Poverty incidence 70 69 65 60 62 55 50 53 49 40 41 30 2012 2013 2014 2015 2016 2017 Urban: Received remittances Urban: Not received remittances Rural: Received remittances Rural: Not received remittances Source: Author’s calculation. 51. More households are headed by a woman similar shares to those of non-poor households; in urban areas and in poor rural households. In larger for the rural poor, followed by urban non- 2013 around 47 percent of households in urban poor households and then by the urban poor, areas were headed by a woman, while 45 percent with 59 percent, 57 percent and 55 percent of among the rural poor and nearly 56 in rural households headed by a woman respectively non-poor households. In 2016, the percentage (Figure 3.9). Rural and poor households are more of households headed by woman remained often led by a woman in 2016 than in 2013 (13 relatively constant for the rural non-poor (55 percentage points increase). The share in urban percent), while the rural poor and urban reached households -poor and non-poor- also increased 30 | Part I: Overview of Poverty by 8 and 9 percentage points respectively, such by migration patterns as a consequence of the that they are mainly headed by women in 2016. ongoing conflict and droughts (see Chapter 4. Only non-poor households in rural areas have a Remittances). This is supported by the fact that for similar share of households headed by a woman in all the Somali population, there was a difference 2013 and 2016. Poor households are more likely (21 percent) in the portion of adult men (aged 25 to have a woman as the head in rural areas, while to 64) in households headed by women between non-poor households in urban areas. The increase recipients and non-recipient of remittances in in households headed by a woman could reflect 2016 (Table A.7 in the Appendix). a higher absence of men, potentially explained Figure 3.9: Households headed by a woman. Figure 3.10: Household size. 70 8 7.4 7.4 59 6.7 55 57 56 55 7 60 5.9 Average household size 5.9 47 47 6 headed by women 45 5.1 % of households 50 4.7 5 4.3 40 4 30 3 20 2 10 1 0 0 Urban: Urban: Rural: Rural: Urban: Urban: Rural: Rural: Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor 2013 2016 2013 2016 Source: Authors’ calculation. Source: Authors’ calculation. Figure 3.11: Literacy rate. Figure 3.12: School attendance. 66 70 62 70 62 61 58 57 57 % of population aged 6-25 56 54 60 60 52 48 46 47 44 50 41 50 % literate 40 35 40 30 30 20 20 10 10 0 0 Urban: Urban: Rural: Rural: Urban: Urban: Rural: Rural: Poor Non-poor Poor Non-poor Poor Non-poor Poor Non-poor 2013 2016 2013 2016 Source: Authors’ calculation. Source: Authors’ calculation. Part I: Overview of Poverty | 31 52. Households are smaller in 2016 compared to rural non-poor (48 percent and 46 percent), and 2013, except for poor households in urban areas. around 41 percent among the rural poor. In 2016, On average, urban households were composed nearly 2 in 3 of the urban population in the North of 7.4 and 5.9 members in poor and non-poor West were literate (58 percent for the poor and households in 2013, while in rural areas of 6.7 62 percent for the non-poor), while less than half among the poor and 5.1 for the non-poor. In 2016, of the non-poor in rural areas (47 percent), and the households size of urban poor was the largest only around 1 in 3 for the poor in rural areas (35 (7.4) followed by the rural poor (5.9), and then by percent; Figure 3.11) .27 The percentage of literate non-poor households in urban and rural areas (4.7 people increased by 10 percentage points among and 4.3 respectively). The average size of rural the urban poor and 6 percentage points for the households decreased by less than one member, urban non-poor between 2013 and 2016. In rural while poor urban households preserved the same areas, non-poor households maintained a similar size (Figure 3.10). A larger decrease in household literacy rate, yet poor households experienced a size for urban and non-poor households might be worrying decrease of 6 percentage points. The associated with migration prior to 2016, since this increase in the literacy rate in urban areas is likely group is more likely to obtain a job in other markets. to be associated with higher levels of education, This is consistent with a higher value of remittances since the share of people with no education in received in 2016 for these households (Figure 3.7; urban areas decreased from 44 percent to 41 US$ 455 per capita in 2016 vs. US$ 367 per capita percent during the same period. Contrary to this, in 2013). Poor households are still larger than non- a larger share of the rural poor does not have any poor households, and also the average household education in 2016 (65 percent) compared to 2013 size is larger in urban than rural areas of the North (54 percent). Changes in the levels of education West region. could be associated with a different composition of the population in urban and rural areas. The 53. Literacy rates decreased for the rural poor rural poor have been increasingly excluded in the of the North West region. More than half of North West region in terms of education which the urban non-poor were literate in 2013 (56 complicates their path out of poverty. percent), less than half of the urban poor and Figure 3.13: Labor force participation. 35 32 30 23 22 % in labor force 25 20 21 21 20 17 16 15 10 5 0 Urban: Urban: Rural: Rural: Poor Non-poor Poor Non-poor 2013 2016 Source: Authors’ calculation. 27 The literacy rates from SLHS 2013 and SHFS 2016 have some limitations, since they are non-functional and were self- reported by interviewed households, yet their evolution provides reliable insights of the observed patterns during this period. 32 | Part I: Overview of Poverty 54. Worse educational levels for the rural poor of the rural poor. In 2016 labor force participation are associated with lower attendance to school. was 32 percent for the rural non-poor working In 2013 nearly 2 in 3 of the population aged 6-25 age population, between 21-22 percent among attended school in urban areas and among the rural the rural poor and urban non-poor, and nearly 16 non-poor, while only more than 1 in 2 of the rural percent for the urban poor (Figure 3.13).28 Labor poor attended school. In 2016, school attendance force participation decreased between 2013 and was highest in urban areas (66 percent for the non- 2016 by 4 and 2 percentage points for the urban poor and 62 percent for the poor), followed by the poor and non-poor respectively. In rural areas, rural and non-poor population (54 percent), and it increased by 5 percentage points for poor then by the poor in rural areas (44 percent; Figure households and 11 percentage points for non-poor 3.12). Between 2013 and 2016, school attendance ones. Lower levels of labor force participation in increased in urban areas, remained relatively urban areas are driven by less people employed constant for the rural non-poor population, while it (12.9 percent in 2013 and 3.8 percent in 2016). decreased around 8 percentage points among the Rural areas present an increase in labor force rural poor. Access and availability to key services participation due to more unemployed people must be improved for poor households. Providing (0.3 percent in 2013 and 5 percent in 2016), the means to reap the benefits from education, that are still considered in the labor force. Labor among other basic services, is crucial to achieve force participation was higher in urban than rural positive labor outcomes and to ultimately lift these areas in 2013, a trend that has been reversed in households out of poverty. The emphasis should 2016. Also, participation is higher for the non- be on poor and vulnerable households, since their poor than the poor in the North West. Sustained educational achievements are lower, and these low differences in terms of education between poor levels tend to be transmitted across generations and non-poor households, together with a low (see Chapter 5.Child and youth poverty). labor force participation, may continue to deepen the urban-rural divide in this region. In 2016, the 55. Labor force participation decreased in urban main reasons for inactivity among the Somali areas, and increased in rural areas. Overall population were illness and sickness, enrollment in labor force participation is low in the North West school, migration and household work. Generating region. In 2013, around 1 in 5 of the working age employment opportunities and brining people into population was in the labor force in urban areas and the labor force should be a central pillar of any among the rural non-poor, while only 17 percent poverty reduction strategy in the North West. 28 The labor indicators presented in this report have some limitations, as they were obtained from the household member that responded the survey on behalf of the other members of the household, instead of every person responding on their own. However, they were collected in the same way in SLHS 2013 and SHFS 2016, thus their evolution provides reliable insights of the observed patterns during this period. Part I: Overview of Poverty | 33 PART II DEEP DIVE INTO SELECTED TOPICS 4. REMITTANCES KEY MESSAGES Remittances make important contributions to welfare, with 1 in 5 Somali households receiving remittances, but recipients rely heavily on these transfers. Remittances are the main source of income for 16 percent of households, and for more than half of recipients. High reliance on remittances leaves recipients, especially poor recipients, at risk in the face of the volatility of diaspora incomes and the uncertainties around sending money to the region. Recipients are less poor, experience hunger less often, and have better educational outcomes. Recipients are typically urban, wealthier, headed by women, and their members better educated. Poverty incidence is 18 percentage points lower in recipient households (recipients: 37 percent, non-recipients: 55 percent). They also experience hunger in the past month half as often. Educational attainment is higher amongst 34 | Part II: Deep Dive Into Selected Topics recipient households, especially amongst poorer recipients, suggesting poor recipient households can offset much of their educational disadvantage compared to non-poor households. IDP households are most excluded from the benefits of remittances, and IDP recipients are no less poor than non-recipients. IDP households are least likely to receive remittances (7 percent), they receive the lowest amounts (US$149 per capita per year), and in many cases suffered a reduction in the amount of money relative to the previous year. Unlike other recipient households, recipients in IDP settlements are no less poor than non-recipients. There is no difference in the poverty gap, the consumption shortfall relative to the poverty line, between recipients and non-recipients in IDP settlements. This is likely due to the fact that poor households in IDP settlements receive amounts too low to overcome their large consumption shortfall relative to the poverty line. The effect of remittances on labor market behavior is negligible overall. Having an additional source of income through remittances could lead recipients to withdraw from working in a labor market that provides poor opportunities for generating income, thus exacerbating dependency. Overall there is no conclusive evidence for this kind of behavior among recipients, who are usually no less likely to participate in the labor force or work fewer hours. However, a finer breakdown reveals that remittances do crowd out work for some segments of the Somali population, albeit with no clearly discernable pattern. Remittances –and cash transfers more generally– can serve as a resilience mechanism in light of adverse shocks, but access is limited, making the case for a formalized social protection program. Remittances mitigate difficult circumstances, highlighting how cash transfers can build resilience for the poor against shocks. With many Somalis excluded from the benefits of receiving remittances, especially the poor and most vulnerable in IDP settlements, other, more formal and predictable cash transfers programs are a suitable means to mitigate the most urgent shortfalls in basic needs. The emerging role of remittances will be analyzed in more depth in the planned Poverty Assessment. The in-depth analysis will add to the dynamics of remittances and their impact utilizing the second wave of the SHFS. In addition, the descriptive analysis will be extended to gather evidence on the causal link between education and remittances. The analysis will also assess the relevance of remittances in the context of a drought, to inform future policies to create resilience. 56. Every fifth Somali household received of households receiving remittances (both 32 remittances in the last 12 months, but the percent), followed by urban households in the likelihood of receipt varies from 7 to 32 North West (24 percent) and North East regions percent across regions, leaving vulnerable (23 percent), and rural North West region (13 populations, especially IDP settlements, percent). Households in IDP settlements are least relatively excluded. Mogadishu and the rural likely to receive remittances at 7 percent, more North East regions have the highest incidence than 50 below average (Figure 4.1). Part II: Deep Dive Into Selected Topics | 35 Figure 4.1: Incidence of remittances. Figure 4.2: Per capita value of remittances. 35 400 Overall Overall 30 350 Annual per capita value, 25 300 % of households current US$ 250 20 200 15 150 10 100 5 50 0 0 u NE n NW ral al u NW n ts NE n r or NW ral ID NW n em l ts or or ra o ba sh ba ba sh ba Se Rur en en Po Po Po Po Ru Ru Se Ru di di Ur Ur Ur Ur em n- n- a a NE NE No No og og ttl ttl M M P P ID Source: Author’s calculation. Source: Author’s calculation. Figure 4.3: Remittances per capita in selected countries. 131 140 Current US$ per capita 120 107 100 80 54 49 48 60 40 40 28 14 10 20 7 5 3 2 0 15 5 5 15 6 15 5 15 5 5 5 5 15 01 01 01 01 01 01 01 01 20 20 20 20 20 l, 2 ,2 i, 2 ,2 ,2 i, 2 i, 2 2 a, d, n, e, e, a, go da ea al nd aw ga ric bi bw on ni te m in To an ne m Be ru ic al Af Le ba So Gu Rw Za affl M Bu Se n m ra ra Zi er ct ha fli Si Sa on b- /c Su ile ag Fr Source: Authors’ calculation and World Bank Open Data. 57. The annual per capita value of remittances, settlements (US$147; Figure 4.2). Households among those who receive them, is US$233. In in rural areas and especially IDP settlements contrast, the annual per capita value among the thus remain relatively excluded from receiving entire Somali population (both recipients and remittances, and consequently from the benefits non-recipients) is US$48 (Figure 4.3). Recipient that their receipt entails. The annual per capita households in urban areas receive between value for entire population among the entire US$214 (North West) and US$276 (North East), population (counting both recipients and non- significantly more than households in rural recipients) of US$48 places Somalis slightly areas (between US$159 and US$191) and in IDP above the US$40 average in Sub-Saharan Africa, 36 | Part II: Deep Dive Into Selected Topics and in line with conflict-afflicted countries with remittances received by households. The findings a US$49 average. indicate that misreporting is not strongly biased towards groups of households with certain 58. The total value of remittances received characteristics, since there is a correlation between should be interpreted with caution, as this receiving remittances and other indicators. Thus, number is lower than those reported by other the analysis is largely constrained to correlations sources. Drawing from experience in other while results about the level and total value of countries, household surveys like Wave 1 of remittances should be interpreted with caution. the SHFS are likely to under-report the value of THE PROFILE OF RECIPIENTS OF REMITTANCES 59. Urban and wealthy households are more likely recipient households mechanically wealthier to receive remittances and receive higher amounts. the more money they receive from remittances. There is a positive relationship between the average Therefore, an average Q5 (top quintile) household amount received and the probability of receiving is both likelier to receive remittances and to receive remittances. Urban households, which are wealthier a higher amount than an average household in than households in rural areas and IDP settlements, one of the lower income quintiles. Second, it is are at once more likely to receive remittances and, plausible that wealthy households have better contingent on receipt, receive more money (Figure means to send their members away to work and 4.4). The relationship between wealth, urban-rural- transfer remittances in the first place. For example, IDP livelihood, and receiving remittances is arguably wealthy households, both recipients and non- driven by a combination of factors: First, remittances recipients, have better educational attainment than make a direct contribution to household income. poorer households (Chapter 2. Multidimensional Households use this contribution, at least in part, deprivation), placing their members at an advantage to cover basic needs and day-to-day expenses, on labor markets abroad. increasing household expenditure, which makes Figure 4.4: Incidence and value of remittances, by income and urban, rural, IDP status. 350 Current US$ per capita, per year 300 Q5 Urban (top quintile) 250 Q3 Q4 200 Q2 Rural 150 IDP 100 Q1 (bottom quintile) 50 0 0 10 20 30 40 % of households that received remittances Source: Authors’ calculation. Part II: Deep Dive Into Selected Topics | 37 60. Recipient households have better have a 15 percentage point higher enrolment educational attainment. Households headed rate amongst their school-aged children, 14 by members with higher levels of education are percentage points higher literacy rate, and spend more likely to receive remittances (Figure 4.5), 26 percent more on education (Figure 4.7 and while members of recipient households tend to Table A.6 in the Appendix). Similar to the previous receive a better education. However, households findings on income and wealth, the differences led by better educated members do not receive in educational attainment are likely driven by higher amounts of remittances across the board; an interaction of two factors: On the one hand, it is specifically university-educated household receiving remittances improves households’ heads who receive significantly higher amounts means to educate their children. On the other of remittances (Figure 4.6; average amount: hand, better educational attainment could put US$233, household head with university household members in a position to earn a decent degree: US$274). Similarly, recipient households living and thus send remittances in the first place. Figure 4.5: Remittances by gender and Figure 4.6: Remittances value by gender and education of the household head. education of household head. 35 400 Current US$ per capita per year Overall Overall 30 350 300 % of households 25 250 20 200 15 150 10 100 5 50 0 0 ed ed n y y y ty ed ed n y y y ty ar ar ar ar ar ar io io rsi rsi ad ad ad ad nd nd m nd m nd at at ive ive i i he he he he uc uc co co Pr Pr co co Un Un Se Se ed ed Se Se e e e e e e al al al al et et c. c. No No e e m M m M pl pl In In et et Fe Fe m m y/ y/ pl pl co co ar ar m m rim rim In In Co Co eP eP et et pl pl m m Co Co Source: Author’s calculation. Source: Author’s calculation. 38 | Part II: Deep Dive Into Selected Topics Figure 4.7: Characteristics of recipient and non-recipient households. 70 60 % of households 50 40 30 20 10 0 d e hh d hh e ar at de a he in tr in sh a he en es hh y y c nc al lm ra e of al m e te l nd ro m e Li of Ag En Fe pe e ar De Sh Recipients Non-recipients Source: Author’s calculation. 61. Recipient households’ advantage in of their income on education than Q1 non- educational attainment is most important recipients (Table A.6 in the Appendix; 73 percent in poorer households. The enrollment rate difference). Household member literacy follows for recipient households of quintile Q1 (the a similar trend, albeit less pronounced (Figure bottom 20 percent of households in terms of 4.9): in Q1, the literacy rate is higher for members total consumption) is more than double that of recipient households (recipients: 57 percent, of non-recipients in the same quintile (Figure non-recipients: 32 percent), and the advantage 4.8; recipients: 70 percent, non-recipients: 34 wears thinner for wealthier households. Literacy percent). This advantage is progressively receding increases modestly with income, including for wealthier households, and the difference in among recipients. Thus, remittances provide a mean enrollment rates disappears for quintiles means for poor households to mitigate their Q4 and Q5. It is only among non-recipients educational disadvantage compared to non- that members of wealthier households are poor households. This insight justifies further more likely to be enrolled. In the same way, Q1 inquiry into how to foster the nexus between recipients spend a significantly higher fraction remittances and education. IDP households are least likely to receive remittances, they receive the lowest amounts, and recipients in IDP settlements are no less poor than non-recipients Part II: Deep Dive Into Selected Topics | 39 Figure 4.8: Enrollment rate by recipient Figure 4.9: Literacy rate by recipient status and income quintile. status and income quintile. 100 100 90 90 80 80 70 70 % enrolled % literate 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Q1 Q2 Q3 Q4 Q5 Q1 Q2 Q3 Q4 Q5 Recipients Non-recipients Recipients Non-recipients Linear (Recipients) Linear (Non-recipients) Linear (Recipients) Linear (Non-recipients) Source: Author’s calculation. Source: Author’s calculation. Figure 4.10: Remittance receipt Figure 4.11: Reasons for change in compared to previous year. remittances value. Urban recipients Rural recipients IDP recipients Non-recipients 0 10 20 30 40 50 60 70 80 90 100 0 10 20 30 40 50 60 70 80 90 100 Percentage of households Percentage of households Same amount as previous year Change in needs for remittances Less than previous year Change in availability of remittances More than previous year Transferring remittances more difficult Non-recipients in previous year Transferring remittances easier Recipients in previous year Other Source: Author’s calculation. Source: Author’s calculation. 62. Households headed by women are more recipient households are headed by a woman, likely to receive remittances than households compared to an overall average of 48 percent, headed by men. 26 percent of households whether recipients or not. Furthermore, recipient headed by women received remittances, households are more likely to be headed by a compared to 17 percent of households headed woman in higher income quintiles (Figure A.7 in by men (Figure 4.7). Likewise, 57 percent of the Appendix). A plausible explanation for this 40 | Part II: Deep Dive Into Selected Topics finding is that, in households headed by women, recipients received remittances in the previous it is more likely that men have left the household year (4 percent).29 In contrast, half of recipients in to work elsewhere and send remittances, a trend IDP settlements received less money than in the which increases with income. This hypothesis is previous year (Figure 4.10). Of the 26 percent of supported by the fact that recipient households recipients who reported a change in the amount headed by women count fewer adult men among of remittances (17 percent less, 9 percent more), their members (aged 25 to 64) than non-recipient just under half state a change in household needs households headed by women (Table A.7 in the as the main reason (Figure 4.11). To the extent Appendix; 21 percent fewer adult men in recipient that remittances are adaptive to household needs, households headed by a woman). Moreover, these transfers could be particularly important women-led households have significantly fewer when households are affected by adverse shocks adult men regardless of recipient status. Other like the ongoing severe drought. However, for socioeconomic characteristics, such as age of more than half of these recipients, the amount the household head, share of dependents, and transferred changed for reasons unrelated to share of boys and men in the household, do not their present situation, illustrating the limits of correlate significantly with recipient status. remittances as an adaptive shock absorber. In particular, 17 percent of households receive less 63. Remittances have remained relatively money than in the previous year because receiving constant in urban and rural areas but changed remittances has become more difficult, enough for most households in IDP settlements. 3 in 4 to warrant policy efforts towards improving easy recipient urban and rural households received the access to such funds. same amount as in the previous year, and few non- REMITTANCES, POVERTY AND CORRUPTION 64. Recipients of remittances are significantly is most pronounced in urban areas. Most likely, less poor, with considerable differences in this is the consequence of urban households poverty incidence across regions. The poverty receiving higher amounts of remittances than headcount rate of recipients is 37 percent, households in rural areas and IDP settlements, 18 percentage points lower than that of non- as higher amounts of remittances received are recipients, and child poverty among recipients strongly related to lower poverty incidence is 40 percent as opposed to 62 percent among (Figure 4.13). While the difference in poverty non-recipients (Figure 5.7). With a poverty gap of between recipients and non-recipients is sizable, 12 percent, poor recipients are also significantly differences in poverty at the regional level are closer to moving out of poverty than non- larger (Figure 4.21). recipients at 24 percent. The reduction in poverty 29 This estimate is based on respondents retrospectively self-reporting the ordinal change in value of remittances (more, less, or about the same) between the two years before the survey date, February 2016. In contrast, figures reported in Chapter 3: The Evolution of Welfare Conditions in the North West Region are based on survey data on the value of remittances collected three years apart, in 2013 and in 2016. Therefore, Chapter 3 draws on a direct comparison of the value of remittances in 2013 and in 2016. In spite of these methodological differences, both sets of findings consistently point to a net reduction in the overall value of remittances over the years. Part II: Deep Dive Into Selected Topics | 41 65. In IDP settlements, remittances are rare and of the non-poor recipients exceeds that of poor ineffective at reducing poverty. Only 7 percent of recipients by a factor of 1.5 (urban non-poor: IDP households receive remittances and poverty US$276, urban poor: US$186), and by a factor incidence is equally high for recipients and non- of 1.08 in rural areas (rural non-poor: US$178, recipients in IDP settlements, with 69 percent and rural poor: US$166). This vast inequality in 70 percent, respectively (Figure 4.12). In line with the value of remittances among IDP recipients this, poverty depth is 35 percent for recipients and suggests that poor households do not receive 37 percent for non-recipients, with the difference enough to overcome their consumption shortfall not statistically significant. Thus, unlike in urban relative to the poverty line. In fact, the average and rural areas, remittances receipt does not have value of remittances for poor IDP households any discernable effect on poverty. There are few covers only 13 percent of the average poverty recipient IDP households leading to low statistical gap. This deficiency is also reflected in the fact power and limited scope for exploring this finding that just around 1 percent all IDP households, in depth empirically. Yet, several observations and 19 percent of recipients, cite remittances as suggest a plausible explanation: The annual per their main source of income, compared to more capita value of remittances varies widely among than half of recipients overall (Figure 4.21). IDP recipients, and much more so than elsewhere. This discussion reveals that the most at-risk The annual per capita value of the non-poor in households in IDP settlements remain excluded IDP settlements 22 times higher than that of poor from any meaningful benefits stemming from IDP recipients (IDP non-poor: US$438, IDP poor: receiving remittances. US$20). In comparison, in urban areas the value Figure 4.12: Poverty incidence by recipient status. Figure 4.13: Poverty incidence by value of remittances received. 80 100 Poverty incidence (% of population) Poverty incidence (% of population) 90 70 80 60 IDP 70 50 60 40 50 Rural 30 40 Urban 30 20 20 10 10 0 0 u n l n l ts 0 10 20 30 40 50 60 70 80 90 100 ra ra ba ba sh en Ru Ru di Ur Ur em a NE NW Percentile of remittances value per capita NE NW og ttl M Se P ID Percentiles Urban Rural IDP Recipients Non-recipients Overall Non-recipients Linear (Percentiles) Source: Author’s calculation. Source: Author’s calculation. 42 | Part II: Deep Dive Into Selected Topics Figure 4.14: Cumulative distribution of Figure 4.15: Cumulative distribution of consumption, urban. consumption, rural. 100 100 90 90 80 80 70 70 60 Percentile 60 Percentile 50 50 40 40 30 30 20 10 20 0 10 0 1 2 3 4 5 6 0 0 1 2 3 4 5 6 Total imputed consumption (current US$) Total imputed consumption (current US$) Urban recipients Urban non-recipients Rural recipients Rural non-recipients Source: Author’s calculation. Source: Author’s calculation. 66. While overall recipient households meaningfully alter their day-to-day consumption. consume more than non-recipient households, The wealthiest rural households, in contrast, may differences in consumption benefit households be using their remittances income in ways that in urban and rural areas and IDP settlements in do not reflect in their consumption expenditures. different ways. Recipient households in urban areas have the largest percentage increase in 67. The largest increase in total consumption total consumption over non-recipient households accrues to the poorer recipient households. (Table A.9 in the Appendix). In addition, urban In the bottom quintile, recipient households recipients’ consumption increases across all consume 23 percent more than non-recipients three components of total consumption – food, (Table A.9), with the difference most pronounced non-food and assets. In contrast, there is no in non-food consumption and assets (Figure statistically significant increase in nonfood 4.16). The overall consumption surplus for consumption for recipients in IDP settlements recipients wears out for the other income and rural areas. However, recipient households quintiles. Similarly, remittances make up a larger experience a significant increase in assets and share of total expenditure for poorer households, nonfood consumption. Moreover, this analysis even though they receive lower amounts: the of average values conceals an important daily value of remittances relative to daily distributional trend: Rural receivers in the middle consumption is 23 percent for the top quintile part of the income distribution (between the as opposed to 58 percent for the bottom quintile 40th and the 70th percentile) do in fact consume (Table A.8 in the Appendix). Remittances are thus more than non-receivers (Figure 4.14). It is for a critical means especially for poor households the wealthiest and poorest household that the to meet basic day-to-day expenses. Conversely, difference vanishes. A plausible interpretation of poor households are most dependent on this fact is that for the poorest rural recipients, remittances, and would suffer severely from the transfers received are not enough to an adverse shock to this source of income. In Part II: Deep Dive Into Selected Topics | 43 line with this, a previous study finds that many would not know how to afford basic consumption recipient households rely on a single sender and and services without this source of income.30 Figure 4.16: Difference in consumption between recipients and non-recipients. 100 80 Percentage change 60 40 20 0 -20 Q1 Q2 Q3 Q4 Q5 Imputed food consumption Imputed non-food consumption Assets Source: Author’s calculation. 68. Remittance-receiving households sufficient food, as starkly opposed to 14 percent experience hunger less often. Non-recipients of non-recipient households (Figure 4.17, right). are twice as likely to have experienced hunger in This trend is also visible in households’ reported the past month as recipients. This finding holds number of meals on the previous day. Children across income quintiles and regions, as well as under five, in particular, are 2 percentage points along the urban-rural-IDP divide, with the value more likely to have eaten an insufficient number of remittances inversely related to experiencing of meals (Figure 4.18), though the difference is hunger. Also, 2.5 percent of non-recipients have relatively slim and not statistically significant in experienced hunger often (more than 10 times) all specifications. On the one hand, this is due to in the past month, compared to 0.3 percent of low statistical power. On the other hand, there is recipients (Figure 4.17, left). Similarly, 4 percent no information in this statistic about the quality of recipient households lacked money to buy and quantity of the meal. 30 FSNAU (2013) ‘Remittances and Livelihood support in Somaliland and Puntland’. FSNAU, Nairobi. 44 | Part II: Deep Dive Into Selected Topics Figure 4.17: Hunger and lack of money to Figure 4.18: Meals on previous day, buy food. children and adults. 30 18 16 25 Percentage of households Percentage of households 14 20 12 10 15 8 10 6 4 5 2 0 0 Recipients Non- Recipients Non- Adults: 1 meal or less Children: 1 meal or less recipients recipients HUNGER LACK OF MONEY FOR FOOD Rarely (1-2x) Sometimes (3-10x) Often (>10x) Recipients Non-recipients Source: Author’s calculation. Source: Author’s calculation. Figure 4.19: Poverty incidence with & Figure 4.20: Poverty incidence by without remittances. change in remittances value previous year. 100 60 80 (% of population) Poverty incidence Poverty incidence (% of population) 55 56 55 50 51 53 60 47.2 36 40 40 20 30 0 l d ed ts ed d ts al 20 te te en en er ct ct uc uc pi pi Ov du du ed d ci ci de de de Re re d n- 0% % 0% % 10 No 00 00 l, 5 ,5 l, 1 ,1 ts al ts en al er en er pi Ov 0 pi Ov ci ci Re Less than Same amount More than Re previous year as previous previous year year Source: Author’s calculation. Source: Author’s calculation. 69. Without remittances, current recipients of consumption of recipients, and, second, could be as poor as current non-recipients, deducting 100 percent of the value of highlighting dependency. Two simple remittances. In the absence of information on simulations illustrate the effect of remittances how households allocate remittances income on poverty: first, deducting 50 percent of between consumption expenditure and other the value of remittances from the total value uses such as investment, the two cases aim to Part II: Deep Dive Into Selected Topics | 45 serve as benchmarks for how different shocks may amount as the previous year (34 percent), while affect recipients rather than being empirically those who received more are less poor (27 grounded: A 50-percent deduction leads an percent, Figure 4.20). The same relationship increase in poverty of 11 percentage points for holds true for hunger: a decrease in the value recipient households (Figure 4.19; 36 percent of remittances implies experiencing hunger vs. 47 percent), and 2 percentage points for the more often, an increase reduces the incidence entire population (Figure 4.19; 51 percent vs. 53 of hunger. These findings highlight the extent percent). In contrast, a 100-percent deduction to which households depend on remittances makes recipients as poor as non-recipients as a source of income. These transfers are large (Figure 4.19; recipients: 56 percent, non- relative to total consumption expenditure, thus recipients 55 percent), and leads to an overall boosting household welfare and protecting increase in poverty of an additional 2 percentage against the worst forms of deprivation. At points. These results are further supported by the same time, relying on remittances leaves the fact that households which received less recipients vulnerable in the face of volatile money from remittances than the previous year diaspora incomes and the uncertainties around are significantly poorer (poverty incidence: 49 sending money to the region. percent) than those who received the same Figure 4.21: Main source of income, Figure 4.22: Main source of income, regional breakdown. income quintiles. IDP Settlements Q5 (top 20) NW Rural Q4 NE Rural Q3 NW Urban NE Urban Q2 Mogadishu (Urban) Q1 (bottom 20) 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% Salaried labor Family business Salaried labor Family business Remittances from abroad Family assistance Remittances from abroad Family assistance Other Other Source: Author’s calculation. Source: Author’s calculation. REMITTANCES, SOURCES OF INCOME AND THE LABOR MARKET 70. Remittances are important sources of percent), followed by remittances (16 percent), family household income, except in IDP settlements. The assistance (12 percent), and income generated from most common source of income is salaried labor (36 a family business (11 percent). But sources of income 46 | Part II: Deep Dive Into Selected Topics vary across regions and along the urban-rural-IDP households rely on remittances as their main source divide. Urban areas and wealthy households rely more of income. While only 7 percent of IDP households often on salaried labor and on remittances than rural receive remittances, only 1 percent reported relying and IDP households. In contrast, rural households rely on them as the main source of income, a reflection more readily on family assistance from within the of the fact that IDPs receive particularly low amounts country (Figure 4.21 and Figure 4.22).31 Very few IDP in transfers. Table 4.1: Main sources of income for households. Salaried Remittances Family Family Other labor from abroad assistance business Overall 36% 16% 12% 11% 25% Recipients 22% 56% 6% 9% 8% Non-recipients 40% N/A 14% 12% 30% Source: Author’s calculation. Figure 4.23: Labor market statistics by important remittances are in creating welfare. recipient status. At the same time, many households stand to suffer a serious consumption shortfall in case 70 of an adverse shock to their remittances income 60 50 (Table 4.1). The effect of receiving remittances on 40 labor market behavior is therefore all the more 30 20 relevant. Particularly, if the knowledge of having 10 an additional source of income from remittances 0 Labor force Hours worked Unemployment, crowds out work in the labor market as an income participation, per week 7 days (%) generating activity, remittances will exacerbate, 7 days (%) and potentially create, dependency in the first Recipients Non-recipients place. The ensuing paragraph thus explores the relationship between receipt of remittances and Source: Author’s calculation. behavior in the labor market. 71. Remittances are the main source of income for more than half of recipient households, but 72. Overall, receiving remittances does not have reliance on transfers leaves them vulnerable to a large effect on household members’ behavior adverse shocks to remittances income. 56 percent in the labor market. The effect that receipt has on of recipient households rely on remittances as labor market behavior aides an understanding of their main source of income, highlighting how the degree and nature of remittances dependency. 31 Figure 4.21 and Figure 4.22 have been collapsed for presentation. The full list of response options in the Wave 1 questionnaire are: Salaried labor; Remittances from abroad; Savings, investments; Pensions; Family assistance; Revenues from sales of assets; Small family business; Other small family business; Domestic trade; Foreign trade; NGO or foreign aid; None. A full breakdown of sources of income overall, by income quintile, and in regional breakdown can be found in Table A.10 in the Appendix. In Wave 2, response options are updated to consist of: Salaried Labor; Remittances (money and goods from family and friends) from abroad; Savings, interest or other investments; Pensions; Remittances (money and goods from family and friends) from within this country; Revenues from sales of assets; Small family business; Agriculture, fishing, hunting and animal husbandry; Trade in domestic goods / products; Trade in foreign goods / products (export or import); NGO or foreign aid; Property income; Zakat; Other (specify). Part II: Deep Dive Into Selected Topics | 47 The labor market effect depends on how recipients region reveals considerable heterogeneity: labor use the received transfers, as complements or force participation is around 10 percentage substitutes to their usual means of earning a living. points lower for members of recipient households In a labor market with poor opportunities (Chapter in North East urban, North West rural, and IDP 2. Multidimensional deprivation), members in settlements, suggesting that these populations recipient households may be tempted to leave use remittances as substitutes for other income the labor market and live off remittances, which generating activities (Figure A.9 in the Appendix). would imply a lower labor force participation rate, Unemployment is lower in North East urban areas, and, by the same token, a lower unemployment but higher in North West rural areas (Figure A.10 rate. Similarly, recipients may decide to work in the Appendix). There is also considerable fewer hours in the knowledge of having additional heterogeneity across income quintiles, albeit funds from remittances. Indeed, recipients do with no clearly discernable pattern. The sparsity have nominally lower labor force participation, of data on hours of work does not allow for hours worked, and unemployment, but only credible analysis based on a finer breakdown. the difference in unemployment is statistically Overall, there is no strong evidence that receiving significant at the 10 percent level (Figure 4.23). remittances crowds out work in the labor market. However, breaking the labor statistics down by CASH TRANSFERS, RESILIENCE AND SOCIAL PROTECTION 73. With beneficial effects on poverty and the benefits that receiving remittances holds. hunger, remittances show how cash transfers can Poor households are particularly unlikely to be serve as mechanism for resilience. The findings recipients at 14 percent and current recipient are at of this section highlight several ways in which risk of falling into poverty in case of shock to their remittances mitigate difficult circumstances. First, income from remittances. Moreover, households remittances reduce poverty and hunger. The overall in IDP settlements and, to a lesser extent, rural value of remittances is directly related to better households benefit least from remittances: they welfare outcomes, and a change in that value are least likely to receive remittances, receive very affects poverty levels and food security. Second, little money, and in many cases suffer a decline in their positive relation with welfare outcomes is the value of remittances. At the same time, rural particularly pronounced among poorer households, areas, and particularly IDPs, are disproportionally who otherwise lack the means to satisfy basic at risk in the current crisis. Emergency assistance needs. These findings are indicative of the great in the form of direct cash transfers is apt for filling potential that direct transfers hold in insuring the the gaps in access and mitigating the most severe poor against shocks and assisting in overcoming effects of the drought and building resilience to the most urgent deprivations. such crises in the future. Any such intervention is most effective when it is predictable and targets 74. With many households left excluded from vulnerable and excluded populations, namely benefits of remittances, especially the poorest the internally displaced and rural households. and most vulnerable in IDP settlements, an However, any type of social protection program institutionalized social protection program requires fiscal capacity, technical capabilities and offers a promising path for protecting the adequate infrastructure (see Chapter 6. Social poor. 8 in 10 households remain excluded from protection for a detailed discussion). 48 | Part II: Deep Dive Into Selected Topics 5. CHILD AND YOUTH POVERTY32 KEY MESSAGES In line with global trends, children are overrepresented amongst the poor. 58 percent of children and 46 percent of youth live in poor households. In line with the general finding of better welfare conditions in the North East region, the lowest child and youth poverty incidence are found in that area. Child and youth poverty is substantially lower in small households, households with an educated household head and in those that receive remittances. Almost 4 out of 5 children are deprived in at least one welfare dimension. 79 percent of children and 85 percent of youth are deprived in at least one dimension. Deprivation is concentrated in rural areas of North West and IDP settlements. For children, consumption deprivation is the most common type of deprivation in urban areas and IDP camps, while the lack of access to improved water source is most prevalent in rural areas. Education is key to break the poverty cycle, yet nearly half of Somali children and youth do not currently attend school. 47 percent of the children and 45 percent of the youth do not attend school. Children and youth living in households that receive remittances have a higher school attendance by 13 and 17 percentage points respectively. Poor children are less likely to attend school (46 percentage) compared to those living in non-poor households (63 percentage), while children and youth living in households with a head that has no education are 30 percent less likely to attend school. The main reasons for not attending school are illnesses, absent teachers, the lack of resources and having to help at home. Many poor children and youth grow up in challenging water and sanitation conditions possibly impacting their health and productivity, especially in IDP settlements. Less than half of children and youth drink water from a piped source. Children and youth living in rural areas are much less likely to treat the water they use, when the source is unprotected. Most children and youth in IDP camps and rural North West rely on other water sources. Regional disparities and dire conditions especially in IDP settlements and rural North West make it harder to lift households out of poverty in these regions. The Poverty Assessment will explore in more detail how to break the intergenerational poverty cycle. In the current environment, children and youth especially from poor families are disadvantaged across a range of indicators. This disadvantage will likely translate into poverty in their adult lives. In 32 This chapter has been written in collaboration with, and funded by UNICEF Somalia. Part II: Deep Dive Into Selected Topics | 49 light of the overwhelmingly young Somali population, this will become an extraordinary development challenge. Thus, these inequalities and barriers must be addressed now with dedicated and specific programs to create enabling environments and opportunities for vulnerable children and youth. As the Poverty Assessment will explore in more detail, priority should be given to programs which aim to break the intergenerational transmission of poverty by addressing the low educational levels, poor health and housing conditions of children and youth. 75. In line with global trends, children the overall population, while 28 and 26 of the are overrepresented amongst Somali’s poor respectively. Contrary to this, youth (aged poor population, while youth are slightly 15-24) represent around 17 percent of the underrepresented. Children (aged 0-14) total population and slightly less of the poor represent nearly half of the total Somali (15 percent). The same pattern is observed for population (49 percent) and more than half (55 children and youth in IDP settlements, rural and percent) of the poor (Figure 5.1 and Figure 5.2). urban areas, yet more pronounced in urban and Girls and boys represent 25 and 23 percent of IDP areas. Figure 5.1: Children and youth in the Figure 5.2: Children and youth in the total population. poor population. 20 20 18 18 16 16 % of total population 14 14 % of the poor 12 12 10 10 8 8 6 6 4 4 2 2 0 0 e rs e s rs rs rs s s rs s rs ar ar ar ar or or a a a a a a ye ye ye ye ye ye ye ye ye ye m m or or 9 4 9 4 9 4 4 9 4 4 -1 -2 -1 -1 -2 0- 5- -1 0- 5- 25 25 20 15 20 10 15 10 Female Male Female Male Source: Author’s calculation. Source: Author’s calculation. More than 1 in 3 children live in households with conditions of extreme poverty 50 | Part II: Deep Dive Into Selected Topics Figure 5.3: Child poverty by region. Figure 5.4: Youth poverty by region. 100 100 Poverty incidence (% of children) Overall Overall Poverty incidence (% of youth) 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 u n l n l ts u n l n l ts ra ra ra ra ba ba ba ba sh sh en en Ru Ru Ru Ru di di Ur Ur m Ur Ur m tle tle a a NE NW NE NW NE NW NE NW og og et et M M PS PS ID ID Source: Author’s calculation. Source: Author’s calculation. 76. More than half of the children and nearly 77. More than 1 in 3 children and nearly 1 in half of the youth live in a poor household. 58 3 youth live in households with conditions percent of children and 46 percent of youth live in of extreme poverty. 35 percent of children households consuming less than the poverty line. and 27 percent of youth live in conditions of Child poverty incidence is higher in IDP camps, extreme poverty. The profile of poverty and Mogadishu and North West rural (Figure 5.3). extreme poverty for both children and youth is Youth poverty is also higher in IDP settlements similar. Poverty and extreme poverty is highest and North West rural (Figure 5.4). Consistent with in IDP settlements, followed by the rural North the overall trend for lower poverty incidence in West, Mogadishu, North West urban and lastly the North East region, the lowest child and youth by rural and urban North East (Figure 5.5 and poverty incidence are found in this area. Higher Figure 5.6). Moreover, the gap between children poverty incidence rates for children and youth are and youth is smaller for extreme poverty than partially explained by a larger dependency ratio for poverty with a poverty line of US$ 1.9 (PPP in poor households. 2011) per day. Part II: Deep Dive Into Selected Topics | 51 Figure 5.5: Extreme child poverty by region. Figure 5.6: Extreme youth poverty by region. Extreme poverty incidence (% of youth) 70 70 Overall average Overall average 60 60 Extreme poverty incidence 50 50 (% of children) 40 40 30 30 20 20 10 10 0 0 u n l n l ts u n l n l ts ra ra ra ra ba ba ba sh ba sh en en Ru Ru Ru Ru di di Ur Ur m Ur Ur m tle tle a a NE NW NE NW NE NW NE NW og og et et M M PS PS ID ID Source: Author’s calculation. Source: Author’s calculation. Figure 5.7: Child poverty by gender of Figure 5.8: Youth poverty by gender of household head and remittances status. household head and remittances status. 80 80 Poverty incidence (% of children) Overall average Overall average Poverty incidence (% of youth) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 s s ed ed s d ed s ce ce ce ce de ad ad ad an an an an a He He He He itt itt itt itt m m m m e e e e al al al al re re re re m M m M d d No No Fe Fe ve ve i i ce ce Re Re Source: Author’s calculation. Source: Author’s calculation. 78. Child and youth poverty is substantially lower in households that did not received remittances in households with an educated household head (Figure 5.8). Moreover, child and youth poverty is and in those that receive remittances. Child poverty more common in households with a household head incidence is higher in male-headed households and aged 40 years or older and whenever the household in those that did not received remittances (Figure head does not have education (Figure 5.9 and Figure 5.7). Whilst youth poverty incidence is similar in 5.10). The number and the migration status of adults households headed by men and women, it is higher in the household does not seem to be associated 52 | Part II: Deep Dive Into Selected Topics with child and youth poverty incidence. while the lack of access to improved water source is most prevalent in rural areas (Figure 5.11). Along 79. For children and youth, monetary and with the lack of access to information, consumption non-monetary poverty are closely related, yet deprivation is more relevant for youths in Mogadishu consumption deprivation is the first or second and urban areas of North West, while access to an most common type of deprivation. For children, improved water source is the second most common consumption deprivation is the most common deprivation in rural areas of North West and North type of deprivation in urban areas and IDP camps, East and amongst IDP settlements (Figure 5.12). Figure 5.9: Child poverty by Figure 5.10: Youth poverty by household characteristics. household characteristics. 80 80 Poverty incidence (% of children) Poverty incidence (% of youth) Overall Overall 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 n n ea s lts tm t t ag 40 + du on n ea s lts m ts s ag -40 + an no ran t nt t io io tio 40 40 ul m dul n du du i igr 5- ra at t w/ igra w cat 5 ad ca ig ca he ed 2 ig ed ed he d 2 uc a Yo th m du ad du w/ or 2 2 HH /m or or ed HH age no HH d ag w/ or /e e /e m w ad ad u o /o w h 1 HH /1 or or Yo HH w/ HH ut ad a / HH w ad w he w 3 3 he ad ad he he HH HH HH he he HH HH HH HH Source: Author’s calculation. Source: Author’s calculation. Figure 5.11: Child deprived in each dimension. Figure 5.12: Youth deprived in each dimension. Consumption Sanitation Information Consumption Water Education Sanitation Water 100 100 90 90 80 80 70 70 % of children % of youth 60 60 50 50 40 40 30 30 20 20 10 10 0 0 hu l n l ts n u an l an l ts ra ra ra ra ba ba sh en en b b Ru Ru Ru Ru s di di Ur m Ur Ur Ur m tle tle a a NE NW NE NW NW NE NW NE og og et et M M PS PS ID ID Source: Author’s calculation. Source: Author’s calculation. Part II: Deep Dive Into Selected Topics | 53 80. Multidimensional poverty measures indicate to information, in addition to consumption. Child that 79 percent of children and 85 percent of youth poverty is lowest in the North East region, while are deprived in at least one dimension, while youth poverty in urban areas (Figure A.12 and 47 and 54 percent in at least two dimensions, Figure A.13 in the appendix). Multidimensional respectively. Deprivation is concentrated in rural poverty is also more severe in North West rural areas of North West and IDP populations. The and IDP camps regardless of whether we consider number of children and youth deprived in various deprivation in one or two dimensions (Figure A.14 dimensions gives an indication of wellbeing by and Figure A.15 in the Appendix). considering education, water, sanitation and access Figure 5.13: Poverty incidence, school attendance and migration by number of children 70 Poverty incidence (% of population) 60 50 40 30 20 10 0 Poverty incidence Child school attendance HH w/at least 1 adult migrant 0 children in the HH 1-3 children in the HH 4+ children in the HH Source: Author’s calculation. 81. Poverty incidence is higher for households of scarce resources. Household size is a relevant with a larger number of children. Households feature of child development, as households with no children have a poverty incidence of 24 with fewer children can devote more time and percent; for households with 1 to 3 and 4 or more more resources to them, potentially bringing children, this increases to 44 and 67 percent, other benefits in terms of school attendance, respectively (Figure 5.13). This is mainly because educational attainment, productivity and larger households are more often poor than consumption.33 School attendance and migration smaller households, but also since having more of adults in the household does not vary with the children increases expenditure needs in a context number of children in the household. 33 UNICEF (2012), Steelman, et al. (2002), HM Government (2014) and Bird, K. (2007). Education is crucial to interrupt the intergenerational transmission of poverty, yet children and youth in households with low levels of education are less likely to attend school 54 | Part II: Deep Dive Into Selected Topics Figure 5.14: Child school attendance by region. Figure 5.15: Youth school attendance by region. 80 80 School attendance (% of children) Overall average Overall average School attendance (% of youth) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 u n l n l ts u n l n l ts ra ra ra ra ba ba ba ba sh sh en en Ru Ru Ru Ru di di Ur Ur m Ur Ur m tle tle a a NE NW NE NW NE NW NE NW og og et et M M PS PS ID ID Source: Author’s calculation. Source: Author’s calculation. 82. Nearly half of Somali children and youth do not for boys and girls. School attendance is relatively currently attend school, mainly due to illnesses, similar for children and youth, and less frequent for absent teachers, the lack of resources, and in the children in IDP settlements and Mogadishu, and for case of the youth group, having to help at home. IDP and rural youth. Education is a powerful tool to 47 percent of the children and 45 percent of the improve the wellbeing of future generations, thus youth do not attend to school (Figure 5.14 and efforts aimed at increasing school attendance should Figure 5.15). Health issues are the first cause of focus on children and youth with poor health, lack absenteeism for children (32 percent) and youth (35 of resources and those that live in areas where percent) (Figure 5.16). Next, for children, the lack of teachers do not attend to school. Some examples of teacher’s attendance (25 percent) and the lack of these programs include school feeding, conditional money (23 percent), while for youth, helping with cash transfers (see Chapter 6. Social protection), work at home is a greater barrier (17 percent), before conditional in-kind transfers like cooking oil, and teacher’s absenteeism (16 percent) and the lack of community mobilization activities, including child- resources (16 percent). These barriers are similar to-child clubs and community education committees. Figure 5.16: Reasons for not attending school. 40 Child Youth 35 % of children/youth 30 25 20 15 10 5 0 ss s ey y e ay er er rit m ne th lid on h ho cu O ac ck Ho m se at te i /s of In lp nt ss ck He ne se La Ab Ill Source: Author’s calculation. Part II: Deep Dive Into Selected Topics | 55 83. School attendance is more likely for number of adults in the household does not affect children and youth in households that receive outcomes on school attendance. remittances, but there are no differences in attendance by the gender of the household 84. School attendance is nearly 30 percent less head. Children and youth that live in households likely for children and youth in households with that receive remittances have higher school a head that has no education. School attendance attendance by 13 and 17 percentage points is higher for children and youth in households respectively (Figure 5.17 and Figure 5.18). This with a household head with some education (64 provides evidence that remittances might lead vs. 45 percent for children and 65 vs. 46 percent to investments in education, since households for youth) and those with a larger number of that receive them have higher incomes, are less literate adults in the household ( Figure 5.19 poor and are more likely to send their sons and and Figure 5.20). Education is crucial to interrupt daughters to school (see Chapter 4. Remittances). the intergenerational transmission of poverty due Poor children are less likely to attend school (46 to its externalities and a higher expected income. percentage) compared to children living in non- Yet, children and youth in households with low poor households (63 percentage). Thus, children levels of education are less likely to attend school. from poor households face bigger challenges to In the current environment, children and youth overcome poverty in their adult life. Children are especially from poor families are disadvantaged. also more likely to attend school in households This disadvantage will likely translate into poverty with 3 or more adults and an older household in their adult lives. Priority should be given to head. On average, the gender of the household programs which aim to break this cycle of poverty. head does not appear to be a relevant factor In light of the overwhelmingly young Somali impacting school attendance for children and population, this will become an extraordinary youth, while for youth, the poverty status and development challenge. Figure 5.17: Child school attendance by Figure 5.18: Youth school attendance by household characteristics. household characteristics. 80 80 School attendance (% of children) School attendance (% of youth) Overall Overall 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 or or ts itt es s he d ed ed 0 + e lts ts or r itt es s he d ed ed 0 + e lts o ce ce e de e de ag 5-4 40 ag 5-4 40 ul ul po Po po Po m c m c or u or u ad ad re ttan an re ttan an m ad ad m ad ad al a al a n- n- ad 2 ad 2 M he M he or r 2 or r 2 he ged he ed No No ed i ed i iv em iv em e e g al al /3 o /3 o HH ad a HH ad a ce o r ce o r m m w /1 w /1 Fe Fe w w Re N Re N he he HH H HH H H H HH HH Source: Author’s calculation. Source: Author’s calculation. 56 | Part II: Deep Dive Into Selected Topics Figure 5.19: Child school attendance by Figure 5.20: Youth school attendance by education of household head and education of household head and literacy of adults in the household. literacy of adults in the household. 80 80 School attendance (% of children) Overall average Overall average School attendance (% of youth) 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 n n e e n n e e e e io t at io t at at at tio tio ra ra at at er er er er ite ite ca ca uc uc lit lit lit lit du du tl tl ed ed ts ts ts ts ul ul :e :e ul ul ul ul o o ad ad ad ad :n :n ad ad Ad Ad 1 1 he he ad ad 2+ 2+ he he HH HH HH HH Source: Author’s calculation. Source: Author’s calculation. 85. In line with poverty incidence and other Most children and youth in IDP camps and North deprivations, water and sanitation for children West rural rely on other water sources like public and youth are worse in IDP camps and North tap, borehole, protected or unprotected spring, West rural. Less than half of the children and rainwater collection, and tanker-truck, among youth drink water from a piped source (Figure others. Water and sanitation conditions can also 5.21 and Figure 5.22). Children and youth living have a deep impact on health and productivity, in rural areas are much less likely to treat the and thus in income generation opportunities and water they use, when the source is unprotected. future poverty status. Figure 5.21: Water and sanitation for child. Figure 5.22: Water and sanitation for youth. IDP Settlements IDP Settlements NW Rural NW Rural NW Urban NW Urban NE Rural NE Rural NE Urban NE Urban Mogadishu Mogadishu Overall Overall 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% Untreated water Piped drinking water source Untreated water Piped drinking water source Toilet: pit latrine or water closet Toilet: pit latrine or water closet Source: Author’s calculation. Source: Author’s calculation. Part II: Deep Dive Into Selected Topics | 57 Figure 5.23: Housing conditions of child. Figure 5.24: Housing conditions of youth. IDP Settlements IDP Settlements NW Rural NW Rural NW Urban NW Urban NE Rural NE Rural NE Urban NE Urban Mogadishu Mogadishu Overall Overall 0% 20% 40% 60% 80% 100% 0% 20% 40% 60% 80% 100% Floor: Cement or tiles Floor: Cement or tiles Roof: Metal sheets, concrete or tiles Roof: Metal sheets, concrete or tiles Source: Author’s calculation. Source: Author’s calculation. 86. Similarly, housing conditions are poor regions will require dedicated and specific for nearly half of the children and youth, and attention to poverty’s impact on children worse in IDP camps and North West rural. Most and youth. Breaking the poverty cycle requires children and youth live in a dwelling with roof improving conditions for children and youth, of metal sheets, concrete or tiles, and a floor of and the challenge of improving the welfare of cement or tiles (Figure 5.23 and Figure 5.24). In Somali’s young population will only grow in IDP camps and North West rural, more than half light of the country’s demographic structure. of the children and youth live in a dwelling with The first step in this direction is adequate a floor of mud or wood, and less than half with a data collection and analysis to monitor the roof of wood or plastic sheets. Worse conditions conditions of children and youth in poverty. in IDP camps and North West rural makes harder In addition, reducing poverty requires targeted to break the poverty cycle in these regions, as responses to reach children and youth, children and youth face greater challenges to particularly in the areas of social protection overcome in order to have good health, acquire and service delivery (see Chapter 6. Social skills and education, to ultimately benefit from protection) by addressing the low educational income generating opportunities. levels, poor health and housing conditions of children and youth. 87. Successful efforts to address monetary and multi-dimensional poverty in Somali 58 | Part II: Deep Dive Into Selected Topics 6. SOCIAL PROTECTION KEY MESSAGES Large numbers of Somalis are affected by a drought in 2017. Food insecurity and poverty remain acute. Many households live in a state of constant vulnerability and are exposed to shocks that – if not mitigated – quickly become human disasters, putting millions of Somalis at the brink of starvation. Social safety nets are instrumental in reducing poverty, supporting vulnerable households and building resilience. The Horn of Africa is cyclically affected by climate-related events like El Niño. Future droughts and floods are expected. This requires resilience-building as part of a sustainable poverty reduction strategy, with specific focus on the most vulnerable households. In the aftermath of the current shock, designing a well-targeted and effective social protection program will be one of the over-arching objectives to avoid repeating famines and more generally to open up a sustainable path to poverty reduction and shared prosperity. A targeted social protection program could reduce poverty from 51 to 32 percent at a cost of US$1.7 billion. Given wide-spread and deep poverty, a social protection program with considerable impact on poverty would require substantial funding. Using observable household characteristics to target poor households, a uniform annual transfer of US$ 157 per capita to all eligible households would reduce poverty by 19 percentage points. The most vulnerable households in rural areas and IDP settlements would benefit with a poverty reduction of 26 and 22 percentage points respectively. Thus, the program would help to include especially the most excluded households. As for any targeted program, there would be some leakage with 27 percent of poor households excluded and 31 percent of non-poor households included. In addition, the costs with US$ 1.7 billion are large, representing around 22 percent of GDP or 130 percent of official development assistance and aid in 2015. Protecting the poor in times of a shock like a drought is even more expensive than just lifting poor households out of poverty. Building resilience is important to protect protective assets from being sold in times of a shock. A 10 percent consumption shock across all households would increase the costs of a social protection program to reduce poverty to the same level of 32 percent from US$ 1.7 billion to around US$ 2.0 billion. It is noteworthy that the 10 percent shock increases the costs of a comparable social protection program by 17 percent. This large elasticity is due to a large number of households that were almost poor in 2016 but are likely to be pushed into poverty by a shock like the current drought. The Somali Poverty Assessment will explore in more detail the impact on poverty depending on the design of the social protection program. The simulations presented in this report are giving a Part II: Deep Dive Into Selected Topics | 59 general sense for the needs and potential impact of social protection programs. A more detailed analysis will discuss different objectives of a social protection effort as well as its design implications. Simulations will model different targeting schemes including a focus on extreme poverty. Also the amount of a potential transfer will be contextualized and aligned with the objective of the social protection programs. 88. Given widespread poverty in 2016, the in Somali regions has been deteriorating, mainly poverty outlook for 2017 looks grim. Food in rural areas due to poor rainfall in the October- insecurity and poverty remain acute in Somali December 2016 season, and given low levels of regions, as more than half of the population lives rainfall forecasted for the April to June 2017 season. in poverty and 24 percent of the households In January 2017, around 3 million people were not experienced some type of hunger in 2016. consuming the minimum food requirements, while Welfare conditions are critical and fragile, which an additional 3.3 million were in need of assistance complicates the path out of poverty going forward. to avoid famine35 Severe droughts and high food Moreover, the data collected in 2016 showed a prices led to a famine already during 2011.36 In large number of vulnerable households without that year, more than 260,000 people died between access to effective and well-targeted social October 2010 and April 2012.37 Thus, better protection programs. resilience systems are needed. 89. Many households live in a state of constant 91. A severe drought is occurring in 2017. The vulnerability and are exposed to shocks. Integrated Food Security Phase Classification (IPC) The Somali population is highly dependent describes the severity of the crisis, based on a on agriculture, livestock and income from five-level scale of food insecurity: 1) minimal, 2) remittances.34 Non-idiosyncratic shocks can stressed, 3) crisis, 4) emergency and 5) famine. A push poor households deeper into poverty, and risk of famine is likely if the rain levels are below push non-poor households into poverty. Climate- the average in the April to June 2017 season, as related events like El Niño, which refers to the forecasted. Households could lose all their crops warming of sea-surface temperatures in the and livestock. Food production will go down, Pacific Ocean, causes severe conditions in Somali food prices might continue to rise and livestock regions. Previously, it caused massive flooding in prices decrease, and there will be reduced rural some regions during 1997-1998 and in 2006- employment opportunities. As a consequence, 2007, but it has also lead to below average rainfall people will be displaced and households will and droughts. Recurring climate-related events experience the loss of their livelihoods. While represents a great risk and potential source of large-scale humanitarian assistance is needed in shocks for the Somali population every couple such contexts, adding to the supply by importing of years. those products through food assistance programs might not be the best alternative. Prices will drop 90. The Somali population is at risk due to severe further, potentially forcing local producers out of and continuous droughts. The food security outlook business. To avoid stepping into this vicious cycle 34 FAO Somalia (2010). 35 http://www1.wfp.org/countries/somalia. 36 Maxwell & Fitzpatrick, M. (2012). 37 http://www.un.org/apps/news/story.asp?NewsID=44811#.WNz-SRsrLb1. 60 | Part II: Deep Dive Into Selected Topics of aid dependency, food assistance programs their price dynamics is essential for designing should carefully and continuously assess the and implementing life-saving interventions. market dynamics, and whether products should The market surveys of the SHFS provides this be imported or can be sourced from local food information in a timely and ready-to-use manner producers, for example by distributing food (Box 4). vouchers. Hence, the monitoring of markets and Box 4: Real-time tracking of market prices To embark on a sustainable pathway toward vehicle repair), from both the tradable and development, intervention should rely on non-tradable sectors. markets and react dynamically to changes in market equilibria. Since 2016, the market The dashboard provides useful insights into the survey of the Somali High Frequency Survey dynamics of the severe drought that is affecting (SHFS) tracks weekly exchange rate and the Somali population in 2017. Prices have been market price data in near real-time across stable, despite the onset of the crisis, with no the 14 urban locations. The dashboard shortages of products in markets. Thus, urban presents a dynamic and rich set of up-to- markets are functioning and products available date prices for a wide range of different while most of the acute food insecurity is in types of products and services. The items rural areas. Hence, interventions should utilize include livestock, food (cereals, milk), non- the existence and functioning of those markets. food items (clothing, cosmetics), utilities The dashboard and real-time data can be found (electricity), and services (such as motor in the following link: http://www.thesomalipulse.com Part II: Deep Dive Into Selected Topics | 61 92. Even though remittances can help to smooth domestic capacity. The international community shocks and improve welfare conditions, they is an important actor, but the fragmentation of are de-centralized and not targeted to the their programs could limit its impact. There is a most vulnerable households. Poor households multiplicity of small-scale initiatives in the form face more challenges than non-poor ones to of cash and in-kind transfers being implemented successfully send a productive member away and by NGOs and international organizations like receive remittances to support their livelihood. Save the Children, BRCiS, CONCERN, World Vision, Some of these challenges refer to the lack of social ADESO, ACTED, among others.38 Coordination capital, resources to invest, and skills required to mechanisms are being established through the obtain a job. As a consequence, poor households UN’s Cash Working Group to avoid duplicating are less likely to receive remittances. Hence, they efforts. However, these initiatives are also are not always received by the poorest and most constrained by the local capacity to efficiently vulnerable households. Only 15 percent of the deliver services to the most vulnerable groups. poor in Somali regions received remittances. In addition, political economy challenges can Remittances are also susceptible to shocks, weaken the effectiveness of programs or delay besides being volatile and uncertain. For example, their implementation. Therefore, a social safety the change in regulations for international bank net program can be a good alternative to reach transfers to Somalia created uncertainty around the most vulnerable people. Simulating the cost remittances at the time of the emerging drought. and impact of such programs is constructive to better understand fundamental trade-offs that 93. Another source of support comes from donors, will also be valid for alternative programs. yet their coordination is crucial in light of weak SOCIAL SAFETY NETS 94. Programs and policies to escape poverty and works (work with Stipends in Latvia and Programa increase resilience are critical in such contexts. de Apoyo Temporal al Ingreso in Salvador) and A non-contributory social safety net (SSN) can workfare programs (Trabajar in Argentina), mainly serve as an intermediate step between short- targeted at low-skill workers through infrastructure term humanitarian aid and a comprehensive long- projects (Productive Safety Net Program in term development and livelihood strategy, while Ethiopia). Others are aimed at early childhood playing a crucial role in improving resilience and development, like school feeding programs (School increasing welfare. A vast evidence of successful Feeding Program in Kenya), school meals (Bolsa stories from a diversity of low and middle-income Familia in Brazil), maternal-child food (MCH/FP countries has placed SSNs at the heart of the program in Honduras) and food rations (Urban development agenda.39 The details of these voucher program in Gaza), among others. Other programs vary from country to country, depending programs support households with a cash transfers on their needs, constraints and capabilities. Some -conditional and unconditional-, social pensions or of them concentrate on job creation, like public in-kind transfers like food stamps, or other social 38 World Bank (2017). 39 Hill, Olinto, Pape, Sherpa, & Sohnesen (2015). 62 | Part II: Deep Dive Into Selected Topics assistance programs such as housing allowances, scholarships, and fees waivers.40 Social safety nets are instrumental in reducing 95. The Government’s development plan supports the implementation of SSN programs poverty, supporting that empower citizens and improve governance. vulnerable households Somalia’s 2017-2019 National Development Plan has a strong focus on tackling poverty and building resilience. and building more resilient communities, with Protecting the poor in times an emphasis on gender and other inequalities. of a shock like a drought is One of the key policy targets is to implement social protection systems, in order to reduce even more expensive than vulnerability and support communities from just lifting poor households internal and external shocks. The introduction of an SSN also represents a direct and transparent out of poverty way to allocate resources, and thus to enable citizens and support strong governance. 96. A well designed SSN can boost productivity the budget constraint of households in a similar in rural areas and upgrade skills in urban areas. way. The simulation represents a simplification In Ethiopia, the Productive Safety Nets Program as only direct impacts are simulated. While the showed that it is feasible to achieve both food analysis concentrates on monetary poverty, other security and land productivity. In rural areas, where indicators are equally relevant as school and health poverty is more acute, a well-designed productive outcomes. Moreover, cash transfers are only one SSN can be used to increase land productivity alternative, and further analysis is needed given through investments in land, and improve access the complexity of designing and implementing a to markets and local roads. Urban centers have social protection program. Also, any SSNs has a more opportunities and requires a different set of short-term component of reaching and supporting skills from rural areas. A SSN in the form of cash vulnerable population, and a long-term component transfer combined with vocational and business on how to build a system and achieve a broader training can aim to create and upgrade skills. This set of goals. Finally, results are partly extrapolated can help to close the gap between basic education to the total Somali population, which should be and marketable skills.41 interpreted with caution given the restricted representativeness of the underlying data. 97. The social protection analysis presented in this chapter concentrates on poverty and monetary 98. The impact of the SSN can be assessed in transfers. A simulation focuses on non-contributory terms of poverty reduction, while the cost of monetary transfer to the poor. An in-kind transfer the instrument and fiscal considerations will might be better suited for some contexts, like rural determine its feasibility. A SSN program can be areas without market infrastructure. However, measured by the number of poor people lifted out these transfers are equivalent to a monetary of poverty, as well as the reduction in the poverty transfer, from a conceptual point of view, in terms gap. Safety nets are usually established for several of their impact on poverty, given that both relax years, such that transfers are stable and predictable 40 World Bank (2012). 41 Hill, Olinto, Pape, Sherpa, & Sohnesen (2015). Part II: Deep Dive Into Selected Topics | 63 for beneficiaries. These program require enough beneficiaries for an SSN program lies in the correct resources to achieve its objectives, and they targeting of the proposed group of beneficiaries. should become part of the fiscal plan for that time That is, including vulnerable and poor households horizon. Any SSN implies a fiscal commitment, and only and excluding those that are not classified as consequently requires a minimum fiscal capacity. such. The inclusion error refers to the erroneous Besides, its implementation requires to develop inclusion of non-poor into the program, while the technical capabilities, administrative capacity at exclusion error corresponds to erroneously leave the central and local levels, as well as an adequate out some poor households. Coverage refers to the infrastructure. Here, the analysis will focus on the proportion of eligible beneficiaries chosen through impact and feasibility to provide foundational the targeting mechanism. The analysis defines evidence for the discussion of social protection coverage relative to both the total population as programs in the Somali context. Additional well as relative to the poor population. Leakage is analysis will be needed to design a specific social defined as the ratio between the total number of protection program. non-poor who may be erroneously targeted and the total number of people targeted by the SSN. 99. The performance and cost-efficiency of an As a result, an effective targeting program should SSN instrument are captured by its coverage and have a low leakage ratio. leakage. One of the main challenges when selecting SIMULATOR OF SSN S FOR SOMALI REGIONS 100. The simulation evaluates two targeting 101. Perfect targeting covers the poor and mechanisms and two transfer amounts: perfect serves as a theoretical benchmark. Using perfect targeting (PT) and proxy means test (PMT), each targeting, only poor households are included with a transfer equal to the average poverty while all non-poor households are excluded from gap and twice this amount. Implementing a the SSN. Perfect targeting is not feasible from an well-targeted SSN program requires choosing the operational perspective, given the difficulty of appropriate group of beneficiaries and the type as identifying poor households, since they will tend well as amount of the transfer. The SSNs analyzed in to under-report their income or consumption this chapter consist of non-contributory and uniform in order to be eligible and receive the transfer monetary transfers such that all eligible individuals or benefits. However, this alternative is a useful receive the same amount. This type of SSNs is used theoretical benchmark against other alternatives, in the simulation, as they have minimal capacity since it provides the maximum possible impact of requirements and are often used in poor countries, a SSN at lowest cost. although they might not represent the best option for the Somali context. Moreover, two amounts are 102. PMT relies on easily identifiable compared in the simulation; a transfer equal to the characteristics to select beneficiaries, yet it average poverty gap (US$0.22 per capita per day or leads to inclusion and exclusion errors. A set US$ 80 per year) and twice the average poverty gap of household characteristics, easily verifiable (US$0.43 per capita per day or US$ 157 per year). and hard to misreport, were obtained for poor 64 | Part II: Deep Dive Into Selected Topics households.42 These characteristics are a proxy fairness since eligible households are identified for the household’s welfare and are correlated through observable attributes. with poverty incidence. Some of them refer to geographic location, size of household, ownership 103. Not surprisingly, PT achieves a higher of durable goods, material of the dwelling, poverty reduction than PMT. The results of the sanitation facilities, access to clean drinking simulation indicate that the best possible SSN water, among others. By using a specific list program (PT) could reduce poverty incidence of characteristics, some households might be by 11 and 26 percentage points, depending on selected as beneficiaries even if they are not the amount considered, while PMT by 7 and 20 poor, while others might not be eligible, even percentage points (Figure 6.1). Similarly, PT reduces if their consumption falls below the poverty poverty gap by 10 and 17 percentage points, line. A key challenge is to determine which depending on the amount transferred, and PMT by characteristics are the best proxies for welfare 9 and 15 percentage points (Figure 6.2). Compared (Box 5). Thus, PMT suffers from the inclusion of to urban and rural areas, PMT is more efficient non-poor and the exclusion of poor households for households living in IDP settlements, as the from the program, due to the challenge of poverty reduction achieved under PMT is closer identifying beneficiaries accurately through a set to the theoretical benchmark of PT. This could of household characteristics. One advantage is be explained by household characteristics being that PMT does not necessarily exclude vulnerable more closely related to poverty in IDP settlements, households, i.e. those slightly above the poverty relative to urban and rural areas. line. Furthermore, it introduces a perception of Box 5: Proxy Means Testing (PMT) In order to identify beneficiaries with PMT, were included: i) geographical location: each household was classified as poor or region and urban-rural-IDP classification; ii) non-poor based on their core consumption household composition: household size and relative to a scaled poverty line. The dependency ratio; iii) characteristics of the poverty line had to be scaled since the total household head: gender and education; iv) consumption aggregate cannot be used characteristics of the dwelling: floor material, because it was constructed using multiple house ownership and type of housing; v) imputation techniques (see Appendix D. Rapid income: members employed and if the Consumption Methodology) and household household received remittances; and vi) characteristics, thus core consumption was hunger: if the household experienced hunger used instead. In line with this, the poverty in the previous month. line was adjusted accordingly using the ratio between the average core consumption and The estimated logit was used to predict the total consumption. probability of households being poor (Table A.11 in the Appendix). They were selected as Then, a logit model was constructed to predict beneficiaries if the estimated probability of the probability of being poor, considering being poor was equal or greater than 0.42. certain identifiable characteristics. In the This threshold was defined to minimize both model, six types of verifiable characteristics exclusion and inclusion (leakage) errors. The 42 Schnitzer (2016). Part II: Deep Dive Into Selected Topics | 65 actual poverty status of households was household characteristics. This provided compared against the list of households the basis to estimate the impact of the selected into the program using the estimated SSN program, the cost and performance, as probability of being poor, based on those measured by coverage and leakage. 104. A social protection program that reduces households could be perfectly identifiable, and poverty to 40 or 25 percent will at least cost considering a transfer equal to twice the average 745 or 1,490 million US$, respectively.43 A poverty gap (Figure 6.1). With this targeting SSN program that provides a uniform transfer mechanism, the reduction in rural areas would to poor households in Somali regions (covered cost 369 or 737 million US$ depending on the and not covered in the survey) can achieve a amount transferred, while 281 or 562 million US$ maximum reduction of 26 percentage points in in urban areas, and 95 or 191 million US$ in IDP poverty incidence. This would be the case if poor settlements (Figure 6.3). Figure 6.1: Impact of SSNs on poverty incidence. Figure 6.2: Impact of SSNs on poverty gap. 80 40 37 Poverty incidence (% of population) 71 Poverty gap (% of poverty line) 70 63 35 61 60 52 30 51 49 46 22 50 44 45 43 47 25 22 22 40 38 20 40 33 20 17 32 28 13 30 25 26 15 11 11 11 10 18 20 8 9 10 20 10 7 6 4 5 10 5 3 2 0 0 Overall Urban Rural IDP Overall Urban Rural IDP Settlements Settlements Current PMT: Avg. poverty gap Current PMT: Avg. poverty gap PT: Avg. poverty gap PMT: Twice avg. poverty PT: Avg. poverty gap PMT: Twice avg. poverty PT: Twice avg. poverty gap PT: Twice avg. poverty gap Source: Author’s calculation. Source: Author’s calculation. 105. Reducing poverty in rural areas is per capita per day, PT achieves a larger poverty more expensive than in urban areas and IDP reduction in urban areas (12 percentage points) settlements. Considering a transfer of US$ 0.22 followed by IDP settlements (10 percentage 43 The simulation of SSNs considers all the population in all the Somali regions, covered and not covered in the SHFS 2016. The reduction in poverty is assumed to be representative of all the regions, which means the SSNs will have the same poverty reduction impact in areas not covered by the survey. The cost was scaled at the rural-urban-IDP level by the share of population not covered in the SHFS 2016, to account for the cost of the SSNs in all the regions. This includes the nomad population, which has been considered in the cost of rural areas. 66 | Part II: Deep Dive Into Selected Topics PMT can reduce 51% 44% 32% poverty TO OR from... points) and then by rural areas (9 percentage 107. For all the Somali regions, PMT covers 37 points). Only when considering a transfer twice percent of them, 73 percent which is poor, and the average poverty gap, poverty reduction is 27 which is not poor, resulting in a leakage of 32 larger in rural areas (32 percentage points vs. 27 percent. Under a PMT targeting mechanism, nearly and 24 of urban and IDP camps respectively). This 1 in 5 poor Somali will not be included in the SSN is explained by a higher poverty gap in IDP camps program, while almost 1 in 3 in the people selected and rural areas relative urban area (37, 20 and 17 as beneficiaries for the program will not be poor in percent respectively), which implies that a larger these regions (Figure 6.4), even though they might transfer is needed to reach or exceed the poverty be vulnerable. Leakage is greater in rural areas, line in rural areas and IDP camps (Figure 6.1 and mainly because some household characteristics Figure 6.2). The latter, combined with the size of the are relatively similar between rural poor and non- poor population in each area (larger share in rural poor households.45 Using a different threshold areas) results in a cost of reducing poverty under to identify beneficiaries will allow to reduce the PT of nearly four and three times higher in rural exclusion error but will come at the cost of a higher and urban areas than in IDP camps, respectively inclusion error and also larger program costs. (Figure 6.3).44 108. The cost of reducing poverty by 1 percentage 106. PMT can reduce poverty from 51 percent point is at least 62 million US$ (PT) but will cost to 44 percent or 32 percent at a cost of 871 106 million US$ for PMT. Any SSN program would or 1,741 million US$, respectively. A feasible cost at least 62 million US$ for 1 percentage point PMT approach can reduce poverty to 44 percent in poverty reduction, but more likely around 106 across all the Somali regions at a cost of 871 million US$ as simulated with PMT. The cost million US$ if the transfer per capita is equal of targeting with PMT is 1.7 times higher than to the average poverty gap. If this amount were PT because the former leads to inclusion and to double, PMT could further reduce poverty exclusion errors, while the latter does not have incidence by 19 percentage points (to 32 these errors, yet PT is not a feasible alternative and percent) at a cost of 1,741 million US$ (Figure only serves as a benchmark. Thus, 27 percent of 6.3). In the latter case, only 1 in 3 people in the poor would not be included in the SSN and 31 urban and rural areas would be classified as percent of the those included would not be poor poor, while still 1 in 2 of them would be poor (Figure 6.4). The exclusion error could be reduced in IDP settlements. but would increase program costs. 44 The poor in all the Somali regions are distributed in the following way: 50 percent in rural areas including the nomads, 38 percent in urban areas and 13 percent in IDP camps. 45 Poor and non-poor households in rural areas are not statistically different for the following household characteristics considered in the PMT model: house ownership, floor material, hunger experienced over the previous four weeks, and if the household has at least one employed member. Part II: Deep Dive Into Selected Topics | 67 Figure 6.3: Cost of SSNs in all the Somali regions. 2.0 1.8 1.7 Billion current US$ per year 1.6 1.5 1.4 1.2 1.0 1.0 0.9 0.8 0.7 0.7 0.6 0.5 0.6 0.5 0.4 0.4 0.3 0.3 0.2 0.2 0.2 0.1 0.1 0.0 Overall Urban Rural IDP Settlements PMT: Twice avg. poverty gap PT: Twice avg. poverty gap PMT: Avg. poverty gap PT: Avg. poverty gap Source: Author’s calculation. 109. Given the depth of poverty in Somali regions, head per day. Analogously, the unit cost of a PMT the unit cost of reducing poverty declines as the targeting scheme is 121 million US$ in case of a transfer amount increases, under both PT and transfer equal to the poverty gap, and 91 million PMT. The unit cost refers to the amount in million for a transfer equal to twice the poverty gap. A US$ for 1 percentage point of poverty reduction. larger transfer amount (US$ 157 per capita per This cost is 17 percent smaller with PT and 25 year) would help the very poor, but it will exceed percent with PMT when the transfer is equal to the expenditure needs to overcome poverty of twice the poverty gap, compared to a transfer of less poor households. However, a transfer amount US$ 0.22 per capita per day (Figure 6.3). Under of US$ 80 per year will only reduce poverty from a PT targeting scheme, the cost of 1 percentage 51 percent to 44 percent, and will maintain in point in poverty reduction is 67 million US$ with poverty 12 percent of the population that could a transfer of equal to the average poverty gap, be lifted out of poverty with a uniform annual while 56 million with a transfer of US$ 0.43 per transfer of US$ 157 per person per year. Figure 6.4: Coverage, leakage and Figure 6.5: Cost and impact of poverty reduction under-coverage for PMT. efforts through SSN programs. 100 60 PMT - avg 90 86 of population) after the introduction of the SSN 78 50 poverty gap Poverty incidence (% 80 73 PMT - twice the 70 63 61 40 Percentage avg poverty gap 60 PT - avg 50 40 30 poverty gap 37 31 37 35 40 28 26 20 PT - twice the avg 27 25 30 22 poverty gap 20 14 10 10 0 0 Overall Urban Rural IDP 0 500 1,000 1,500 2,000 2,500 Settlements Coverage (% poor) Coverage (% population) Total cost (million US$) Under-coverage Leakage PMT PT Source: Author’s calculation. Source: Author’s calculation. 68 | Part II: Deep Dive Into Selected Topics 110. With PMT, the larger the transferred can provide alternatives but are often more amount and the cost, the harder it becomes to complex to implement. further reduce poverty. The poverty reduction efforts depend on the fiscal capacity (Figure 6.5). 111. Protecting the poor in times of a shock The difference in terms of the poverty reduction like a drought is even more expensive than just achieved from a SSN program between the lifting poor households out of poverty. Building theoretical benchmark and PMT widens as the resilience is important to protect assets from being amount transferred increases. A SSN program will sold in times of a shock. Assuming a 10 percent support poverty reduction efforts and increase consumption shock across all households would resilience in light of the vulnerable conditions of increase the costs of a social protection program many households. Yet, it becomes more difficult to reduce poverty to the same level of 32 percent to further reduce poverty with a transferred from US$ 1.7 billion to around US$ 2.0 billion. It is amount beyond twice the average poverty gap noteworthy that the 10 percent shock increases the as some households might need a relatively large costs of a comparable social protection program by transfer to reach the poverty line, and other poor 17 percent. This large elasticity is due to a large households are not included in the PMT targeting number of households that were almost poor in scheme. Larger amounts would further increase 2016 but are likely to be pushed into poverty by costs and make the program less efficient in a shock like the current drought. reducing poverty. A non-uniform transfer program Part II: Deep Dive Into Selected Topics | 69 7. CONCLUSIONS 112. Wave 1 of the Somali High Frequency than urban/rural variation (45/52 percent). In Survey, implemented in February and March of urban areas, poverty ranges from 26 (North 2016, provides critical data on poverty and other East) to 57 percent (Mogadishu). In rural areas, key socioeconomic indicators. In the absence of poverty ranges from 34 percent (North East) to representative household surveys not much was 61 percent (North West). Breaking the poverty known about the state of the Somali population. cycle requires improving conditions for children The lack of information prevents the design and and youth. This challenge will continue to grow implementation of policies and programs needed in light of the demographic structure of Somali to support economic resilience and development regions. Priority should be given to programs as well as assistance in the event of shocks. The which aim to break the intergenerational first wave of the SHFS filled many critical data transmission of poverty by addressing the low gaps, from poverty estimates to components of educational levels, poor health and housing GDP. The data shed light on the circumstances conditions of children and youth. in which Somali households live at a time when the region is amidst a severe drought and an 114. Beyond monetary poverty, many more impending famine. Somalis lack access to basic services, such as education, safe drinking water, and information 113. A large share of the Somali population media. Non-poor households have higher levels lives in poverty, with children and the of literacy, educational attainment and better internally displaced particularly affected. dwelling conditions like quality of water and Poverty is widespread with every second Somali sanitation. 9 in 10 Somali households are deprived living in poverty in 2016 before the onset of in at least one dimension of multidimensional the current shock. 51 percent of the population poverty. 40 percent of Somali households do not are below the international poverty line, and have access to improved sources of water, and 31 percent face conditions of extreme poverty. the drought is further depleting water stocks. In contrast, 58 percent of children live in poor Providing safe drinking water is therefore a policy households, and 1 in 3 in extreme poverty. priority in the current crisis. Investments in basic Poverty also varies considerably across the infrastructure, such sanitation systems, electricity Somali population, ranging from 26 to 70 lines, and roads, are strongly needed in all Somali percent. Households in IDP camps are more likely regions, particularly in rural areas. Overall, the to be affected by poverty. Regional differences Somali population lags behind most low-income in poverty between the North East (27 percent) African countries in most non-monetary correlates and the North West (50 percent) are much larger of welfare. Poverty is widespread across Somali regions, and many households are vulnerable to the effects of the ongoing drought. Access and availability to key services must be improved in order to improve welfare conditions 70 | Part II: Deep Dive Into Selected Topics 115. In order to reduce inequality and poverty, access and availability to key services must be improved for poor households, since current programs leave them behind, particularly in terms of education. The educational gap has widened for the rural poor between 2013 and 2016 in the North West. The percentage of literate people was stable or increased for every group, but the rural poor. Changes in the literacy rate are likely to be caused by changes in the levels of education of the Somali population. A larger share of the rural poor does not have any education in 2016 compared to 2013. Worse educational levels for the rural poor are probably associated by lower attendance to school. Between 2013 and 2016, school attendance increased in urban areas of the North West region, remained relatively constant for the rural non-poor population, while it decreased for the rural poor. This group has reasons for inactivity need to be addressed by been increasingly excluded in the North West a comprehensive approach. The most serious in terms of education which complicates their barriers to labor force participation are gender path out of poverty. Sustained differences in disparities, conflict-related insecurity, and terms of education between poor and non-poor disability, each of these constraints warranting households, together with higher unemployment specific interventions. in rural areas, may continue to deepen the gap. Providing access and means to reap the benefits 117. Many households are highly vulnerable from education, among other basic services, is to the effects of the ongoing severe drought. crucial to achieve positive labor outcomes and to Many of the currently non-poor households live ultimately lift these households out of poverty. just above the poverty line so that in the event In 2016, nearly half of the school-aged Somali of an adverse shock, such as the current drought, population did not attend school due to illnesses, poverty levels can be expected to increase absent teachers, the lack of resources, and having significantly. Specifically, a 10-percent adverse to help at home. The emphasis should be on poor shock implies a 6 percentage point increase in and vulnerable households, since their educational poverty. This simulation is in line with current achievements are lower, and these low levels tend estimates of the effects of the drought which to be transmitted across generations. expect overall output to decrease by 10.6 percent according to World Bank internal estimates. The 116. Access to the labor market, particularly situation is particularly severe for households in for young Somalis and women, is critical to IDP settlements, and children, whom the drought sustainably improve welfare conditions. Poverty is affecting most direly. An unmitigated famine strongly correlates with unwanted labor market would mean a sustained setback to the gradual outcomes. Poor households tend to have lower gains in standards of living of the past years. participation in the labor market and lower employment. Increasing the active participation 118. In the context of the current crisis, of Somalis in the labor market is key to improve monitoring markets is essential for designing welfare and decrease inequality. The different and implementing life-saving interventions. Part II: Deep Dive Into Selected Topics | 71 in light of the current crisis, but the most vulnerable remain excluded. Poverty and hunger are significantly less common among recipients of remittances compared to non-recipients, and they also have better educational outcomes. However, with just above 20 percent of households receiving remittances at all, a large majority of the population cannot fall back on remittances in the current crisis. Moreover, the most vulnerable populations, particularly in IDP settlements, are least likely to receive remittances, receive relatively little money, and in many cases suffer a decline in the value of remittances. At the same time, rural areas, and particularly IDPs, are disproportionally at risk in the current crisis, while the drought has already displaced an additional 257,000 people.47 Donor support through direct cash transfers, targeted to the most vulnerable To embark on a sustainable pathway toward populations, is therefore critical. development, intervention should rely on markets and react dynamically to changes 120. Absence of effective resilience-building in market equilibria. Since 2016, the market social protection programs allows natural survey of the SHFS tracks weekly exchange rate shocks like a drought to become a human and market price data in near real-time across disaster putting millions of Somalis at the brink the 14 urban locations. The market surveys of of starvation. The data collected in 2016 showed the SHFS provides this information in a timely a large number of vulnerable households without and ready-to-use manner.46 The dashboard access to effective and well-targeted social provides useful insights into the dynamics of protection programs. Recurrent natural shocks the severe drought that is affecting the Somali like this drought caused by El Niño will continue population in 2017. Prices have been stable, to test resilience of the Somali population in the despite the onset of the crisis, with no shortages future. In the aftermath of the current shock, of products in markets. Thus, urban markets are designing a well-targeted and effective social functioning and products available while most protection program that can work in the local of the acute food insecurity is in rural areas. context will be one of the over-arching objectives Food assistance programs should carefully and to avoid repeating famines and more generally to continuously assess the market dynamics, and open up a sustainable path to poverty reduction whether products should be imported or can be and shared prosperity. sourced from local food producers, for example by distributing food voucher. 121. Social safety nets are instrumental in reducing poverty and supporting vulnerable 119. Remittances – and cash transfers more households. A targeted social protection program generally – can serve as a resilience mechanism could reduce poverty from 51 to 32 percent at 46 http://blogs.worldbank.org/nasikiliza/how-the-real-time-tracking-of-market-prices-in-somalia-helps-us-respond-to-drought. 47 http://reliefweb.int/report/somalia/somalia-drought-response-situation-report-no-1-24-march-2017. 72 | Part II: Deep Dive Into Selected Topics a cost of US$1.7 billion. Given wide-spread and 123. Wave 2 of the SHFS will provide a timely deep poverty, a social protection program with update in the context of the current crisis. When considerable impact on poverty would require faced with an emergency information is critical to substantial funding. Using observable household shape policies and programs to ultimately support characteristics to target poor households, a the population in the most vulnerable conditions. uniform annual transfer of US$ 157 per capita Policies based on evidence and rapid changing to all eligible households would reduce poverty conditions for the Somali population require by 19 percentage points. The most vulnerable regular updates about the profile of the poor, households in rural areas and IDP settlements to ultimately make an efficient use of national would benefit with a poverty reduction of 26 and international resources. The High Frequency and 22 percentage points respectively. Thus, the Survey managed to close data gaps but, with program would help to include especially the the drought ongoing, the situation is changing most excluded households. As for any targeted rapidly, threatening to further exacerbate poverty. program, there would be some leakage with The survey, funded by the Somali Multi Partner 27 percent of poor households excluded and Trust Fund, is expected to be administered in the 31 percent of non-poor households included. summer of 2017. It will expand the geographical In addition, the costs with US$ 1.7 billion are coverage, include the Somali nomadic population, large, representing around 22 percent of GDP and will provide the most comprehensive update or 130 percent of the annual receipt of official on the status of the current crisis in Somali regions. development assistance and aid in 2015. 124. Video testimonials, part of the innovative 122. Protecting the poor in times of a shock concept for Wave 2, will give voice to the Somali like a drought is even more expensive than just people behind the numbers. As part of data lifting poor households out of poverty. Building collection for Wave 2, survey respondents will be resilience is important to protect protective assets offered to give a short video testimonial reflecting from being sold in times of a shock. A 10 percent on their day-to-day lives. While data is critical for consumption shock across all households would informing policies going forward, the power of increase the costs of a social protection program these testimonials lies in capturing the human to reduce poverty to the same level of 32 percent side of life in difficult circumstances. from US$ 1.7 billion to around US$ 2.0 billion. It is noteworthy that the 10 percent shock increases 125. Wave 2 will provide, for the first time, the costs of a comparable social protection an insight into the conditions facing a large, program by 17 percent. This large elasticity is and vulnerable, Somali nomadic population, due to a large number of households that were including their patterns of movement. Nomads almost poor in 2016 but are likely to be pushed make up around a third of the Somali population, into poverty by a shock like the current drought. and very little is known about poverty, well-being, With expanded coverage, Wave 2 will provide a timely update in the context of the current crisis, as well as insights into the conditions facing a large and vulnerable Somali nomadic population Part II: Deep Dive Into Selected Topics | 73 and needs among this group. Wave 2 will contain will be explored in more detail including the a pilot of interviews attempting to fill this critical identified education - health nexus. The role of gap in a systematic fashion. Through the use of women in the economy will be analyzed given state-of-the-art satellite technology, the team will their contributions in the informal sector and be able to determine the regular routes of nomadic subsistence farming that are not well reflected populations, providing invaluable information, for in the labor market statistics. An education example for emergency assistance and service analysis will analyze constraints to education as delivery to this group, which is among the most well as estimate returns to education to better affected by the current drought. understand potential entry points to improve educational outcomes with a focus on the 126. A Somali Poverty Assessment will utilize identified linkages between education and health. wave 1 and 2 of the Somali High Frequency The emerging role of remittances will be analyzed Survey to delve deeper into selected topics with respect to their dynamics and impact. The and make more specific program and policy Poverty Assessment will explore in more detail recommendations. The poverty analysis will how programs can break the intergenerational consider adult equivalent measures of monetary transmission of poverty. The discussion of social poverty, which is relevant to consider within protection programs will be expanded considering household economies of scale, provide a more different objectives of a social protection effort nuanced profile of the vulnerable population, as well as its design and poverty implications. A and assess the impact of the drought on focus will also be put on IDPs and how durable livelihoods. Also the gender dimension of poverty solutions for them could look like. 74 | Part II: Deep Dive Into Selected Topics REFERENCES • Alatas, V., Banerjee, A., Hanna, R., Olken, B. A., Purnamasari, R., & Wai-Poi, M. (2013). Ordeal Mechanisms in Targeting: Theory and Evidence from a Field Experiment in Indonesia. NBER Working Paper. • Bird, K. (2007). “The intergenerational transmission of poverty: An overview”. Working Paper 99. Manchester: IDPM/Chronic Poverty Research Centre (CPRC). • Bourmpoula, V., Kapsos, S. and Pasteels, J.M (2013), “ILO estimates and projections of the economically active population: 1990-2030 (2013 edition)”, International Labour Organisation, Geneva. • Bruni, L., & Eddy, M. (2012). The Role of a Local Public Goods Transfer in Promoting Stability and Alleviating Poverty in South Sudan. Boston: Harvard University. • Collier, P. (2007), “The Bottom Billion: Why the Poorest Countries are Failing and What Can Be Done About It”, Oxford University Press. • FAO (2010). Strategy Plan of Action 2011-2015: Somalia. • Federal Government of Somalia. National Development plan 2017-2019: Towards Recovery, Democracy and Prosperity • Gabrielle, Thomas, and Micehele Nori. 2007. “Cash Based Safety Nets for Livelihood Support in Northeastern Somalia: A Feasibility Study for Save the Children UK and Horn Relief”. Save the Children UK; Horn Relief. • Gosh, M. E., & Baker, J. L. (1995). Proxy Means Tests for Targeting Social Programs. • Hill, R., Olinto, P., Pape, U. J., Sherpa, M., & Sohnesen, T. (2015). Designing an Urban Safety Net in Ethiopia. Washington: World Bank. • HM Government (2014). “An evidence review of the drivers of child poverty for families in poverty now and for poor children growing up to be poor adults”. • International Labour Office (2008), “Resolution Updating the International Standard Classification of Occupations”, International Labour Organisation, Geneva. • International Labour Office (2012), “International Standard Classification of Occupations: Structure, group definitions and correspondence tables”, ISCO-08, Volume 1. • International Labour Office (2013), “Resolution concerning the statistics of work, employment, and labour underutilization”, adopted by the 19th Conference of Labour Statisticians, Geneva, October 2013, International Labour Organisation. • International Labour Office (2015), “Key Indicators of the Labour Market”, 9th Edition, International Labour Organisation, Geneva. References | 75 • LaGuardia, Dorian, Andrew Pinney, and Leigh Jaschke. 2016. “Impact Evaluation: Shocks & Hard Knocks: Humanitarian Action & Resilience in Somalia.” London: DFID (Department for International Development). • Larson, G., Ajak, P., & Pritchett, L. (2013). South Sudan’s Capability Trap: Building a State with Disruptive Innovation. Boston: Harvard Kennedy School. • Maxwell, D., & Fitzpatrick, M. (2012). “The 2011 Somalia famine: Context, causes, and complications”. Global Food Security, 1(1), 5-12. • Pape, U. and J. Mistiaen (2015), “Measuring Household Consumption and Poverty in 60 Minutes: The Mogadishu High Frequency Survey”, World Bank. • Ravallion, M., Shaohua, C. and Sangraula, P., “Dollar a Day Revisited”, The World Bank Economic Review, vol. 23, no. 2, pp. 163-184, doi:10.1093/wber/lhp007. • Schnitzer, Pascale. (2016). “How to target households in adaptive social protection systems? relative efficiency of proxy means test and household economy analysis in Niger”. Social Protection and Labor discussion paper; no. 1612. Washington, D.C.: World Bank Group. • Steelman, L. C., Powell, B., Werum, R., & Carter, S. (2002). “Reconsidering the effects of sibling configuration: Recent advances and challenges”. Annual Review of Sociology, 28(1), 243-269. • UN, Department of Economic and Social Affairs (2008), “International Standard Industrial Classification of all Economic Activities, Revision 4”, The United Nations, New York. • UNESCO (2012), “International Standard Classification of Education/ISCED 2011”, Paris. • UNICEF (2012). “Children in an Urban World”. The State of the World’s Children. • World Bank (2005). “Conflict in Somalia: Drivers and Dynamics.” Washington DC: World Bank. World Bank (2011), “World Development Report: Conflict, Security and Development”, World Bank. • World Bank (2012). Social Protection Strategy for Africa. Washington: World Bank. • World Bank (2013), “Unlocking Africa’s Agricultural Potential”, Sustainable Development Series, World Bank. • World Bank (2014). South Sudan Jobs and Livelihoods. Washington: World Bank. • World Bank (2014a). South Sudan: Impact of a Continued Internal Conflict on Food Security and Poverty. Washington, DC: World Bank. • World Bank (2014b), “Challenges and Opportunities of High Frequency Data Collection in Fragile States: Lessons from South Sudan” 76 | References • World Bank (2015), “Somaliland Poverty Assessment”, World Bank. • World Bank (2015a), “Innovating Data Collection and Monitoring in Fragile States”, World Bank (forthcoming). • World Bank (2015b). South Sudan: Social Protection Policy Note. Washington: World Bank. • World Bank (2015c). The Fiscal Impact of Declining Oil Prices on South Sudan. Washington, DC: World Bank. • World Bank (2015d). “Aid Flows in Somalia.” Aid Coordination Unit. • World Bank (2016). “Somaliland Social Protection: Stocktaking of Evidence for a Social Protection Policy and Framework.” Washington DC: World Bank. P155120. • World Bank (2017). “Somalia Social Protection: Stocktaking of Evidence for a Social Protection Policy and Framework.” Washington DC: World Bank. • World Food Programme. (2015). Situation Report #76. Juba: World Food Programme. • World Vision (2015). “Empowerment for Better Livelihoods in Nugal, Puntland State: Social Services Mapping.” Hargeisa, Somaliland: Forcier Consulting and World Vision International. References | 77 APPENDIX A. FIGURES AND TABLES Table A.1: Household demographic attributes: number of children and adults. Number of children Number of adults (0-14 years old) (15-64 years old) Region All Poor Non Poor All Poor Non Poor North East 2.6 4.0 2.3*** 2.3 2.4 2.3 Urban 2.6 4.1 2.2*** 2.4 2.3 2.4 Rural 3.1 3.7 2.9** 2.1 2.7 1.9 North West 2.7 3.7 2.0*** 2.9 3.2 2.6 Urban 2.7 3.8 2.0*** 3.0 3.4 2.7 Rural 2.7 3.2 2.0*** 2.4 2.6 2.1 Mogadishu 2.4 3.2 1.6*** 2.3 2.3 2.3 Urban 2.6 3.6 2.0*** 2.6 2.8 2.5 Rural 2.8 3.3 2.4** 2.3 2.7 2.0 IDP Settlements 2.5 3.1 1.3*** 2.9 2.5 3.7 Overall average 2.6 3.4 2.0*** 2.6 2.7 26 *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. 78 | Appendix Table A.2: Selected poverty indicators. Poverty incidence Poverty Gap Poverty Total Gap (% of population) (% of poverty line) Severity (per year, current Region Index million US$) North East 27.2*** 7.9 3.5 49.2 Urban 26 7.5 3.4 40.4 Rural 34 10.1 4.1 8.8 North West 50.0*** 19.2 9.3 229.8 Urban 47.9 18.2 8.9 179.7 Rural 61.1 24.2 11.4 50.1 Mogadishu 57.0*** 23.8 11.9 163.5 Urban 45.0* 17.1 8.4 476.3 Rural 52.5* 19.7 9.1 627.5 IDP Settlements 70.5*** 36.5 22.2 214.6 Overall average 51.4 21.7 11.5 1,318.4 The total monetary value of the poverty gap includes the entire Somali population. *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. Table A.3: Access to improved source of water and sanitation, percentage of population. Variables Number of deprivations Being poor 0.474*** 0.431*** (0.052) (0.046) Rural 0.634*** (0.068) Mogadishu -0.332*** (0.063) IDP 0.288*** (0.079) Constant 1.267*** 1.229*** (0.042) (0.050) Observations 4,064 4,064 *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. Appendix | 79 Figure A.1: Educational attainment, secondary. Figure A.2: Educational attainment, tertiary. 16 16 % of population % of population 12 12 8 8 4 4 0 0 ps n l l u ps n l l an NW ral NW an ad l u NE an Ur l NW an ad l ra al ra al a ra a ba ba ish ish M Rur M Rur m m er er b b b b Ru Ru Ru NW Ru Ur Ur Ur Ur Ur Ca Ca Ov Ov NE NE NE og og P P ID ID Poor Non Poor Overall Poor Non Poor Overall Source: Author’s calculation. Source: Author’s calculation. Figure A.3: Inactivity reasons for women. 80 % of inactive women 60 40 20 0 l ps n l u n l n l ra al ra ra ba ba ba sh m er Ru Ru Ru di Ur Ur Ur Ca Ov a NE NW NE NW og P ID M House Work In school Husband would not allow Does not expect to find work Ill, sick or disabled Source: Authors’ calculation. Figure A.4: Inactivity reasons for men. 80 % of inactive men 60 40 20 0 l l l u ps n n l n ra al ra ra ba ba ba sh m er Ru Ru Ru di Ur Ur Ur Ca Ov a NE NW NE NW og P ID M In School Ill, sick or disabled Insecurity/conflict Housework Does not expect to find work Source: Authors’ calculation. 80 | Appendix Figure A.5: Average annual expenses on electrical devices. 70 60 US$ per capita 50 40 30 20 10 0 an l an l u ps an l l ra ra ra al sh m er b b b Ru Ru Ru di Ur Ur Ur Ca Ov a NE NW NE NW og P ID M Poor Non Poor Overall Source: Authors’ calculation. Table A.4: Consumption items excluded from each survey to obtain a comparable consumption aggregate. Section SLHS 2013 SHFS 2016 Food Coconuts Baker’s vanilla (carfiso buskut) Dried or salted meat Begel Groundnuts shelled Bell pepper Other ‘Roots, Tubers Canned sweetcorn Other vegetables Cardamom (heyl) Cinnamon (qarfo) Clove (dhago yare) Cucumber local (khajaar) Dates - import (timir) Foster Powder Fresh camel Fresh chicken - local Frozen chicken - import Ginger (zanjabiil) Ketchup Mayonnaise Olive oil Parsley - local (kabasr caleen) Pizza Sandwiches Vimto (squash) Non-food Postage stamps or other postal Healthcare expenditures fees Durables N/A Small solar light Appendix | 81 Table A.5: Average consumption (per capita, per day in US$). Urban Rural Total 1.728 1.428 Comparable 1.620 1.344 Scale factor 0.938 0.941 Source: Authors’ calculation. Figure A.6: Poverty incidence with total and comparable consumption aggregates. 100 Poverty incidence (% of population) 90 80 70 63 64 60 53 52 50 40 30 20 10 0 Urban Rural 2016 (comparable consumption and scaled poverty line) 2016 (total consumption and $1.9 USD PPP poverty line) Source: Authors’ calculation. Table A.6: Difference in educational spending per school-aged child between recipients and non-recipients. Q1 Q2 Q3 Q4 Q5 Difference in 0.90*** -0.09 0.01 -0.09 0.22 educational spending These figures are the result of subtracting the log of educational expense of recipients from that of non-recipients and should be interpreted as a percentage change. *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. 82 | Appendix Table A.7: Difference in share of males, household head excluded. Recipient Non-recipient Percentage difference Household head by men 18% 19% -7% Household head by women 8% 11% -20%*** *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. Table A.8: Remittances share of total consumption. Total daily Total daily value of Percentage consumption remittances per capita remittances of total (current US$) (current US$) consumption Q1 0.52 0.30 58% Q2 0.94 0.44 47% Q3 1.38 0.57 41% Q4 2.05 0.67 33% Q5 3.76 0.85 23% Overall 1.73 0.64 37% Source: Author’s calculation. Figure A.7: Household headed by a woman by income quintile and recipient. status. 100 90 80 Percentage of households 70 60 50 40 30 20 10 0 Q1 Q2 Q3 Q4 Q5 (bottom 20) (top 20) Recipients Non-recipients Linear (Recipients) Linear (Non-recipients) Source: Authors’ calculation. Appendix | 83 Table A.9: Changes in daily per capita consumption for recipients. Total Food Non-food Consumption flow consumption consumption consumption of durable goods Urban 0.30*** 0.24*** 0.36*** 0.41*** Rural 0.14 0.06 0.47*** 0.71*** IDP 0.03 -0.05 0.26 0.61*** Q1 0.23*** 0.03 0.90*** 0.47** Q2 0.04 0.03 0.24** 0.32** Q3 0.00 -0.05 0.13** 0.53*** Q4 -0.01 0.03 -0.10 0.27*** Q5 0.04** -0.06 0.11 0.37*** These figures are the result of subtracting the log of consumption of recipients from that of non-recipients and should be interpreted as a percentage change. *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. Source: Author’s calculation. Figure A.8: Total imputed daily consumption value by recipient status. 3.50 3.00 Current US$ per capita, 2.50 per day 2.00 1.50 1.00 0.50 0.00 l n l ts u an ra ra ba sh en Ru Ru b di Ur Ur em a NE NW NW NE og ttl M Se P ID Recipients Non-recipients Overall Source: Authors’ calculation. 84 | Appendix Figure A.9: Labor force participation by Figure A.10: Unemployment by recipient status recipient status. 40 90 35 80 30 70 % in labor force % unemployed 25 60 50 20 40 15 30 10 20 5 10 0 0 u an l an l ts u an l an l ts ra ra ra ra sh sh en en b b b b Ru Ru Ru Ru di di Ur Ur Ur Ur m m a a NE NW NE NW tle tle NE NW NE NW og og t t M M Se Se P P ID ID Recipients Non-recipients Overall Recipients Non-recipients Overall Source: Author’s calculation. Source: Author’s calculation. Figure A.11: Experience of hunger in past 4 weeks by recipient status. 100 90 80 70 % of households 60 50 40 30 20 10 0 Never Rarely Sometimes Often (1-2 times) (3-10 times) (more than 10 times) Recipients Non-recipients Source: Authors’ calculation. Appendix | 85 Table A.10: Full List of Sources of Income by Income Quintile and Regional Breakdown. NW Urban NW Rural NE Urban NE Rural Overall MOG IDP Q1 Q2 Q3 Q4 Q5 Salaried labor 36% 34% 42% 41% 31% 19% 29% 26% 33% 37% 41% 41% Remittances from abroad 16% 30% 18% 17% 18% 10% 1% 8% 10% 17% 16% 22% Savings, investments 1% 1% 1% 0% 0% 1% 0% 0% 1% 1% 1% 1% Pensions 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% Family assistance 12% 8% 10% 10% 22% 20% 18% 18% 14% 9% 11% 8% Revenues from sales of assets 1% 1% 0% 1% 1% 8% 1% 2% 1% 2% 2% 0% Small family business 8% 8% 8% 7% 12% 8% 6% 7% 12% 8% 7% 8% Other small family business 4% 5% 4% 3% 2% 3% 4% 3% 3% 3% 4% 7% Domestic trade 1% 0% 1% 1% 5% 5% 1% 0% 2% 1% 2% 2% Foreign trade 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0% NGO or foreign aid 1% 0% 0% 0% 0% 0% 2% 2% 1% 0% 1% 1% None 21% 13% 14% 20% 9% 27% 37% 34% 24% 23% 17% 12% 86 | Appendix Figure A.12: Child deprived in one dimension. Figure A.13: Youth deprived in one dimension. 100 100 90 90 80 80 70 70 % of children % of youth 60 60 50 50 40 40 30 30 20 20 10 10 0 0 u u an l an l ts an l an l ts ra ra ra ra sh sh en en b b b b Ru Ru Ru Ru di di Ur Ur Ur Ur em em a a NE NW NE NW og og NE NW NE NW ttl ttl M M Se Se P P ID ID Children deprived in at least 1 dimension Overall Youth deprived in at least 1 dimension Overall Source: Author’s calculation. Source: Author’s calculation. Figure A.14: Child deprived in two dimensions. Figure A.15: Youth deprived in two dimensions. 100 100 90 90 80 80 70 70 % of children % of youth 60 60 50 50 40 40 30 30 20 20 10 10 0 0 u u n l n l ts n l n l ts ra ra ra ra sh sh ba ba ba ba en en Ru Ru Ru Ru di di Ur Ur Ur Ur em em a a NE NW NE NW og og NE NW NE NW ttl ttl M M Se Se P P ID ID Children deprived in at least 2 dimensions Overall Youth deprived in at least 2 dimensions Overall Source: Author’s calculation. Source: Author’s calculation. Appendix | 87 Table A.11: Estimated logit for proxy means test. Dependent variable: Poor Explanatory variables Dependency ratio 0.132*** Hunger in the past month: 0.106 Rarely (1-2 times) Male-headed household -0.068 Hunger in the past month: 0.611*** Sometimes (3-10 times) Household size 0.395*** Hunger in the past month: 0.526 Often (more than 10 times) HH has at least one employed -0.631*** House type: Shared house/ 0.417*** member apartment HH head education: -0.234* House type: House -0.016 Incomplete primary HH head education: Complete -0.702*** House type: Hut and other 0.796*** Primary/Incomplete Secondary HH head education: -0.645*** Region: Strata 101 & 105 1.836*** Complete Secondary HH head education: University -1.088*** Region: Strata 201-205, 1103, -0.970*** 1203, 1204 & 1303 HH head education: Other -0.185 Rural 2.075*** Floor material: Tiles(ceramic) -0.597*** IDP Settlement 2.379*** Floor material: Mud 0.483*** NE Urban 1.841*** Floor material: Wood -0.023 NW Urban 1.525*** Floor material: Other 0.494*** NE Rural -0.052 House ownership: Own -0.140 Received remittances -0.435*** House ownership: -0.115 Constant -4.251*** Occupy w/o permission House ownership: -0.572** Occupy w/permission House ownership: Other -0.745 Observations 3,777 Pseudo R2 0.27 *, **, *** indicate significance at the 10%, 5%, and 1% level respectively. 88 | Appendix B. LOWER POVERTY INCIDENCE IN THE NORTH EAST REGION The results from Wave 1 of the SHFS indicate that Somalis in North East regions would be among the the North East region has lower poverty incidence poorer parts of the population. Thus, the findings that in the other surveyed pre-war regions. presented in this report related to North East have Contrary to this finding, anecdotal evidence and been analyzed particularly carefully to assess expert assessments raised the expectation that their validity in a number of ways. Table B.1: Differences in poverty incidence. North West Mogadishu North East Overall Poverty Headcount Rate 53% 57% 25% 46% Quantity of consumption Mean kg of core consumption pc pd per item 0.049 0.054 0.061 0.054 Percentage relative to mean 91% 100% 113% 100% Number of consumption items Average core items per household 13.1 12.3 14.4 13.3 Percentage relative to mean 99% 93% 109% 100% Household size Average household members 5.7 4.5 5.2 5.2 Size relative to mean 109% 86% 99% 100% Replacement rate of EAs 23% 24% 21% 23% To ensure that findings are not artificial due to of the most consumed items, both by quantity some idiosyncrasies of the data collection in North and value, are similar (Table B.2). Lower poverty East, scrutiny was directed at whether low poverty incidence further does not seem to be driven by incidence was driven by (i) consumption quantity household size, as the average household size in per core item, (ii) the number of core items North East is 5.2, compared to 5.7 in North West consumed, or (iii) prices. Households in North- and 4.5 in Mogadishu (Table B.1). Finally, the East consume a higher quantity per capita per rate of replacement of enumeration areas (due item (13 percent more than the overall average) to inaccessibility, missing structures or security) and more items per household (9 percent more is similar in all three regions (Table B.1). Overall, than average; Table B.1). No notable price trends then, North East’s lower poverty incidence emerged, as prices for food items are deflated. appears to be the result of genuinely higher In addition, household consumption profiles of recorded consumption. North East, North West, and Mogadishu in terms Appendix | 89 Table B.2: Most consumed core food items. Value Quantity Region Rank Item Share Rank Item Share North West 1 Sugar 10% 1 Sugar 14% 2 Goat or sheep 9% 2 Macaroni, spaghetti 10% 3 Macaroni, spaghetti 9% 3 Rice, husked 9% Mogadishu 1 Fresh camel 19% 1 Sugar 13% 2 Macaroni, spaghetti 8% 2 Macaroni, spaghetti 9% 3 Milk Powder 7% 3 Millet, flour 7% North East 1 Milk Powder 10% 1 Sugar 14% 2 Goat or sheep 10% 2 Macaroni, spaghetti 9% 3 Sugar 8% 3 Rice, husked 7% In addition, the analysis of other typical well- Somalis in North East participating in the labor being indicators, known to correlate well with force (Table B.3). This trend also holds for access to monetary poverty, support the conclusion that water and sanitation. 70 percent of households in lower poverty incidence in North East is genuine: North East have access to improved water sources, North East is doing better than average in various compared to the average of 58 percent (Table important indicators of well-being. Literacy rate B.3), while 14 percent of North East households in North East is 64 percent relative to an average have access to improved sources of sanitation, of 55 percent (Table B.3), and households in compared to an average of 12 percent in the other North East also outperform the average in other regions. North East’s performance on these non- educational outcomes such as enrollment and monetary indicators of well-being is thus in line educational attainment. The same holds true for with its lower incidence of monetary poverty. labor market outcomes, most critically with more Table B.3: Key non-monetary indicators of well-being. Literacy Primary Primary Labor Access to Access to rate school enrollment force improved improved completion rate participation water sanitation rate North West 57% 16% 62% 32% 45% 12% Mogadishu 58% 16% 39% 37% 96% 12% North East 64% 18% 61% 47% 70% 14% Overall 55% 16% 53% 38% 58% 10% 90 | Appendix C. SAMPLE AND DATA COLLECTION Estimating monetary poverty rates requires fragile country with severe security constraints a sound, reproducible methodology. The for field work and wide spread displacement. methodology starts with the sample design, The sampling methodology was adapted to continues with questionnaire design and the the context by excluding several inaccessible construction of food and non-food consumption areas. The questionnaire design utilized the aggregates, selection of spatial price deflators and Rapid Consumption methodology that can be how to determine the consumption value derived easily and quickly implemented. The choice of from assets, and what process to use to construct deflators and the poverty line were driven by the poverty lines. The appendix describes the data quality. methodology used to estimate poverty for the Wave 1 Somali High Frequency Survey. A household is defined as poor if the per-capita household consumption does not exceed a The chosen methodology balances a trade-off given threshold between feasibility and accuracy. Somalia is a (1) y_i≤z where yi is the nominal per-capita household present the construction of the consumption expenditure and z is the poverty line at the nominal aggregate yi before discussing the choice of the level. In the following, we discuss the selection poverty line z and standard poverty measures. of households i as part of the sample design and SAMPLE AND SAMPLE FRAME The Population Estimation Survey of Somalia Due to the combination of the different data (PESS) was used as sample frame alongside a list sources, the resulting sample frame included of settlements from three different sources (UNDP enumeration areas as well as settlements. While 1997, UNDP 2006 and FSAU 2003) to complement enumeration areas are defined as geographical missing rural and semi-urban settlements. areas with about 50 to 200 households, The combined sample frame was cleaned and settlements often are larger areas with a larger preprocessed before the number of enumeration population. In fact, all rural and a large fraction areas per strata was calculated and enumeration of semi-urban enumeration areas and settlements areas selected proportional to size. Depending did not have boundaries available but were only on the strata, different multi-stage clustering defined by a GPS position. approaches were used to select households. Appendix | 91 Since PESS is also partially based on the same data midpoints for new enumeration areas around sources (especially UNDP 1997 and UNDP 2006) the main duplicate GPS position to ensure that and since some PESS enumeration areas had the larger settlements have the appropriate number same GPS location, several GPS positions were of surrounding enumeration areas.48 very close of each other and, thus, considered duplicates (Figure C.1). Technically, duplicates In a second step, boundaries of enumeration are defined where the distance between the areas without corresponding shape files were GPS position is below 75m. In groups with drawn automatically. First, the GPS positions multiple duplicates, the additional criteria was were used as midpoints of circles with a radius introduced that all GPS positions must have pair- of 200m. Overlapping circles were transformed to wise distances below 200m to prevent large Thiessen polygons where the line connecting the sequential areas of GPS positions. Duplicates overlapping points becomes the new boundary. were merged into one ‘hypothetical’ enumeration The algorithm was tested for areas where PESS area with a tag of the number of duplicates. Those shapefiles were available (Figure C.2). duplicate counts were used to position manually Figure C.1: Examples of duplicate GPS. Figure C.2: Thiessen test polygons with bold boundaries representing the known enum. area boundaries. 48 Note that this was only done for selected duplicate enumeration areas to reduce manual processing. 92 | Appendix SAMPLE STRATIFICATION AND SIZE The sample is designed based on predicted the statistical estimation of indicators. A smaller statistical precision of consumption as well as cost number of households would result in less than considerations. Without political implications, 3 observations for each of the four optional the survey stratifies the sample into four zones, modules capturing consumption data. A including Mogadishu, B including Garowe, C including Hergeiza and D for Sanaag, Sool and A total sample of about 3,800 households is Togdheer. The sample is stratified for each zone sufficient to obtain consumption estimators into economic/political centers, urban centers, with a relative standard error below 1 percent. other urban settlements, rural settlements and – if After rounding the number of enumeration areas existent – IDP camps. The result are 16 strata (star ensuring that 12 households per enumeration marks areas where a micro-listing approach was area, 324 enumeration areas were initially utilized; see below): selected. The 324 enumeration areas are first distributed into the 16 strata. The number of • A: Mogadishu*; IDPs* enumeration areas per strata is determined by • B: Garowe; Urban Centers; Other Urban; Rural; (i) the population of the strata, (ii) the variability IDPs* of consumption within the strata, and (iii) the • C: Hergeiza; Urban Centers; Other Urban; requirement of at least two enumeration areas Rural; IDPs* per strata. Strata with larger population and larger • D: Sanaag Urban; Sanaag Rural; Sool Urban; variability will need a larger sample to retrieve Sool Rural; Togdheer Urban; Togdheer Rural the same relative standard error as a stratum with smaller population and consumption variability The sample employs a clustered design with (Table C.3). Variability is estimated based on the Primary Sampling Unit (PSU) being the previous surveys and a pilot in Mogadishu. The enumeration area. Within each enumeration strata for Mogadishu was later amended by area, 12 households will be selected for an additional 20 enumeration areas to correct interviews. A larger number of households per against a faulty optional module assignment in enumeration area would only marginally benefit the first days of data collection. HOUSEHOLD SELECTION Depending on the strata, different clustering the center of the block. Within each enumeration approaches were used. In strata with more volatile area, one segment was randomly selected and security as well as for IDP camps, a multi-stage within the segment 12 blocks were chosen. In cluster design was employed called micro-listing. each block, all structures were listed before Each selected enumeration area was divided selecting randomly one structure. Within the into multiple segments and each segment was selected structure, all households were listed further divided into blocks. A block is defined as and one household randomly selected for a geographical area where an enumerator can interview. This multi-stage clustering approach see (and list) all households from one location in reduces the time in the field substantially and Appendix | 93 contributes to a lower profile of enumerators, was listed before 12 households were randomly which is paramount in fragile areas. In strata selected for interviews (called full-listing). less volatile, the complete enumeration area DATA COLLECTION AND REPLACEMENTS The survey was implemented using tablets as were approved by the supervisor after a total of survey devices (CAPI). The data collection system three unsuccessful visits of the household. consisted of Samsung Smartphones equipped with SIM cards, mobile data plans, microSD cards Incoming data is processed to create a raw (16 GB capacity), and external battery packs. consistent data set. Interviews with wrongly The phones were secured with Android’s native entered EAs were manually corrected. Interviews encryption and protected by a password. GPS conducted outside sampled EAs were discarded. tracker helped to track all devices using a web For duplicate submissions, only one record is interface (www.gps-server.net), Barcode Scanner kept.49 Sampling weights are added to the final allowed to use barcodes for the identification of dataset and subsequently anonymized at the enumerators and a parental control application strata level. Missing values are recoded into four provided a safe contained working environment different types of missing values: (i) genuinely for enumerators. Interviews were conducted missing values coded as “.”; (ii) respondent using SurveyCTO Collect on the tablet with data indicated “don’t know” coded as “.a”; (iii) transmitted to a secure SurveyCTO server in a respondent refused to respond to the question cloud computing environment. coded as “.b”; and (iv) missing values due to the questionnaire skipping pattern because EAs were replaced if security rendered field work the question does not apply to the respondent unfeasible (Table C.3). Replacements were approved coded as “.z”. by the project manager. Replacement of households 49 Two types of duplicate households are identified. Technical duplicates are defined as duplicate submission of the same interview. They are identified as households with identical GPS data (latitude, longitude and altitude coordinates). Manual duplicates are defined as two interviews conducted with the same household. They are identified by almost identical household rosters. The interview with more information is kept based on manual inspection. 94 | Appendix CLEANING PROCESS OF SUBMISSIONS The total number of interviews submitted through SurveyCTO was 4,590, and the breakdown by zone the following: A B C D 1,06 1,035 2,366 120 The first step corresponds to a cleaning process identifying general issues and inconsistencies with submissions. B 1 empty household record dropped C 3 household records deleted as they were submitted through the web and they were part of a test to monitor scripts before fieldwork 1 submission dropped as it corresponds to a test that a team leader made to check if the GPS of one of his enumerator’s phone was working 1 additional household record dropped as it corresponds to an interview completed by the enumerator to check he had the latest version of the questionnaire Therefore, after making the described adjustment, the number of correct submissions became 4,584, with the following breakdown by region: A B C D 1,069 1,034 2,361 120 Appendix | 95 The second step excludes submissions from EAs and blocks that were not included in the final sample. A 3 submissions were dropped as they belong to a block that was not included in the final sample B 12 submissions dropped, as they correspond to an EA that was not included in the final sample, since it was a replacement EA that was never executed C 3 interviews dropped because the enumerators selected a wrong EA that had been replaced. Therefore, after making the described adjustment, the number of correct submissions became 4,566, with the following breakdown by region: A B C D 1,066 1,022 2,358 120 The next step was to validate the acceptance of submissions, for which six criteria were defined and interviews were dropped that failed to meet at least one of them: The duration of the interview had to exceed a threshold of 30 minutes 1 • 26 submissions were excluded because they were completed in 30 minutes or less Random sound bites check, including respondent and enumerator voices. This criterion will 2 be assumed to hold if a specific interview was not checked on this criterion. • No interview was removed for this reason The interview has GPS coordinates and it was conducted within a buffer area of the 3 correspondent EA • 5 interviews did not have GPS coordinates; and • 5 were also excluded as the GPS coordinates indicate the interview did not take place within the boundaries of the EA 96 | Appendix If the interview was not completed in the first visit, then the household record for the 4 first visit must be valid using the previous criterions (except for the duration), and both household records must contain a matching GPS positions, with a margin of +/- 10 meters • 34 interviews were dropped as they corresponded to a second visit, and the record from the previous visit did not exist or was not valid • 26 additional submissions were not considered, as the GPS coordinates of the first visit did not match with those of the subsequent visit If the interview corresponds to a replacement household, the record of the original 5 household must be valid, except for the duration of the interview • 67 submissions were not considered as the interview corresponded to a replacement household with an inexistent or invalid record for the original household Finally, unsuccessful interviews were discarded; the ones where no one answered the door, 6 there was not a knowledgeable adult present or the respondent did not give permission to continue: • 282 submissions were not successful and thus were also excluded Therefore, at this point, the dataset had a total number of 4,121 submissions, with the following breakdown by region: A B C D 1,031 929 2,045 116 The final step excludes interviews that were incomplete, and thus have several sections without any single response. 4 households did not have any record in the sections corresponding to food consumption, assets and livestock, and thus they were excluded. Therefore, the final dataset includes a total number of 4,117 complete, valid and successful submissions from valid EA and blocks, with the following breakdown by region: A B C D 1,031 929 2,041 116 Appendix | 97 SAMPLING WEIGHTS This subsection describes calculation of sample weights for households in the dataset. The sample design was different for some strata due to security volatility. Thus, the methods differ between micro-listing and full-listing. After the sample weights were calculated as described below, they were scaled to the number of households accessible with GPS from the sample frame. A) Full listing: The sample was drawn in a two-stage process for strata 201-204, 301-304 and 1103-1304. Therefore, the weights were calculated based on the sampling probabilities for each sampling stage and for each cluster in the following way: such that Phij: Probability of selecting household h in EA i of strata j P1: Probability of selecting the EA in stage 1 P2: Probability of selecting the household in stage 2 EAj: Number of EAs selected in strata j Hi: Number of households estimated in the sample frame for EA i Hj: Number of households estimated in the sample frame in strata j HSi: Number of households selected in EA i HLi: Number of households listed in EA i Therefore, the sample weight for each household corresponds to B) Micro-listing: In strata 101, 105, 205 and 305, the sample was segmented in blocks within EAs, in addition to the two-stage, stratified cluster sampling, design.50 Therefore, the weights were calculated based on the sampling probabilities for each sampling stage and for each cluster in the following way: 50 The segmentation step cancels out as exactly one segment is chosen. 98 | Appendix such that Phij: Probability of selecting household h in EA i of strata j P1: Probability of selecting the EA in stage 1 P2: Probability of selecting the Block P3: Probability of selecting the household EAj: Number of EAs selected in strata j Hi: Number of households estimated in the sample frame for EA i Hj: Number of households estimated in the sample frame in strata j BSi: Number of households listed in EA i Bi: Number of blocks in EA i HSi: Number of households selected in EA i HLi: Number of households in EA i Therefore, the sample weight for each household corresponds to Finally, three types of sampling weights were estimated: Unadjusted weights: Considers all submissions (4,117) and scales the weights so that the 1 sum of the sampling weights by analytical strata matches the total number of accessible households with GPS according to sample frame. Adjusted weights: Considers all submissions (4,117) and scales the weights uniformly so 2 that the sum of the weights by analytical strata matches the total number of households according to the PESS (Table C.2). 51 Adjusted weights for consumption and poverty variables: Considers only submissions with 3 consumption data (excludes 53 submissions with missing values in the consumption of food, non-food and durables) and adjusts the weights of the remaining 4,064 submissions according to the following scenarios: • If the number of accessible households with GPS (i.e. the sum of weights) is larger than the total number of households according to PESS by analytical strata, then the weights were scaled downwards uniformly to match the total number of households from PESS, which already reflects the re-allocation of the weights from the 53 submissions excluded • If the number of accessible households with GPS (i.e. the sum of weights) is smaller than the total number of households according to PESS, then the weights were scaled upwards in two steps: i) re-allocating uniformly the weights from the 53 households excluded across the 4,064 submissions; and then ii) assigning the additional weights needed to 51 Usually, the household number from the sample frame should reflect the number of households from the last Census. However, the incomplete sample frame necessitated using different (overlapping) data sources for the sample frame. While the probabilities for selection for duplicates are adjusted for already in the EA selection step, the total number of households did not automatically sum up to the number of households from PESS. Appendix | 99 match the figures from PESS only to those households or submissions in the bottom 25 percent of the total consumption distribution for the respective analytical strata. The bottom 25 percent were taking up the weight of the additional households to reflect the fact that excluded enumeration areas were not randomly chosen but differed from other enumeration areas by inaccessibility due to security and/or infrastructure. As those enumeration areas are expected to be more deprived than the average enumeration area, they were assumed to be similar to the bottom 25 percent. Table C.1: Sample properties of the SHFS. Settlements Other areas Mogadishu NW Urban NW Rural NE Urban NE Rural Nomads Overall IDP Sample Size 4,117 816 643 154 1,405 668 431 0 0 (Households) Covered Households 923,092 187,246 163,444 27,684 281,669 61,086 201,963 0 0 Sample Size 21,026 3,619 3,272 800 7,851 3,294 2,190 0 0 (Individuals) Covered Individuals 4,930,401 895,915 823,041 147,758 1,636,490 315,508 1,111,689 0 0 Population (PESS) 12,316,895 1,280,939 992,207 176,282 1,854,995 384,798 1,106,751 3,186,966 3,333,957 Population Covered 40% 70% 83% 84% 88% 82% 100% 0% 0% Number of 341 67 52 13 118 56 35 0 0 Enumeration Areas Table C.2: Total number of households by PESS region and analytical strata. PESS Region Type Analytical Strata # of HH All IDP All 201,963 Banadir Urban Banadir 187,246 Nugaal Urban Nugaal 23,119 Bari and Mudug Urban Bari and Mudug 140,334 Woqooyi Galbeed Urban Woqooyi Galbeed 123,390 Awdal, Sanaag, Sool and Togdheer Urban Awdal, Sanaag, Sool and Togdheer 158,279 Bari, Mudug and Nugaal Rural Bari, Mudug and Nugaal 27,684 Awdal, Sanaag, Sool, Togdheer and Rural Awdal, Sanaag, Sool, Togdheer and 61,086 Woqooyi Galbeed Woqooyi Galbeed 100 | Appendix Table C.3: Sample size calculation, number of replacement and final sample. 52 Appendix | 52 Note that the number of (accessible) households does not resemble necessarily the number of PESS households due to the merging of multiple data sources. Therefore, sample weights were adjusted 101 accordingly to scale with PESS household estimates. D. CONSUMPTION AGGREGATE The nominal household consumption aggregate is the sum of three components, namely 1) expenditures on food items, 2) expenditures on non-food items, and 3) the value of the consumption flow from durable goods: (1) This section describes in detail the cleaning of the recorded data for each of three components. Subsequently, the construction of the consumption aggregate using the Rapid Consumption Methodology is explained as well as the estimation of the consumption flow for durables and the details on the deflator used to calculate spatial price indices. Moreover, 53 households were assigned a missing value in consumption since 52 of them reported not consuming any food items, and 1 household only reported consuming a non-core food item. CLEANING PROCESS: FOOD Food expenditure data is cleaned in a four-step process. First, units for reported quantities of consumption and purchase are corrected. Typical mistakes include recorded consumption of 100 kg of a product (like salt) where the correct quantity is grams. These mistakes are corrected using generic rules (Table D.1). Then, we introduce a conversion factor to kg for some specific items and units. For example, we recognize that a small piece of bread must have a different weight than a small piece of garlic (Table D.2). The third step consists of correcting issues with the exchange rate selected (Table D.3). Finally, outliers are detected using the six cleaning rules below to correct quantities and prices. • Consumption quantities with missing values for items reported as consumed were Rule 1 replaced with item-specific median consumption quantities. • Missing purchase quantities and missing prices for items consumed were replaced with item-specific median purchase quantity and item-specific median purchase price. Records where the respondent did not know or refused to respond if the household had Rule 2 consumed the item, were replaced with the mean value, including non-consumed records. Records with the same value for quantity consumed or quantity purchased and price are Rule 3 assumed to have a data entry error in the price or quantity and are replaced with the item- specific medians. 102 | Appendix Records that have the same value in quantity consumed and quantity purchased but Rule 4 different units are assumed to have a wrong unit either for consumption or purchase. For both quantities, the item-specific distribution of quantities in kg is calculated to determine the deviation of the entered figure from the median of the distribution. The unit of the quantity that is further away from the median is corrected with the unit of the quantity closer to the median. • Missing and zero prices are replaced with item-specific medians Rule 5 • Outliers for unit prices were identified and replaced with the item-specific median. This includes unit prices in the top 10 percent of the overall cumulative distribution (considering all items), and unit prices below 0.07 US$. The consumption value in US$ was truncated to the mean plus 3 times the standard Rule 6 deviation of the cumulative distribution for each item, if the record exceeded this threshold. All medians are estimated at the EA level if a minimum of 5 observations are available excluding previously tagged records. If the minimum number of observations is not met, medians are estimated at the strata-level before proceeding to the survey level. In addition, medians greater than 20 kg and smaller than 0.02 kg were not considered for quantities, while medians greater than 20 US$ and smaller than 0.005 US$ were also excluded for unit prices. CLEANING PROCESS: NON-FOOD The non-food dataset only contains values without quantities and units. First, we apply the same cleaning rules for currencies (Table D.3) and then the following cleaning rules: Zero, missing prices and missing currency for purchased items are replaced with item- Rule 1 specific medians. Records where the respondent did not know or refused to respond if the household had Rule 2 purchased the item, were replaced with the mean value, including non-consumed records. Prices that are beyond a specific threshold for each recall period (Table D.4) are replaced Rule 3 with item-specific medians. Prices below the 1 percent and above the 95 percent of the cumulative distribution for Rule 4 each item are replaced with item-specific medians Appendix | 103 The purchase value in US$ was truncated to the mean plus 3 times the standard deviation Rule 5 of the cumulative distribution for each item, if the record exceeded this threshold. The item-specific medians were applied at the EA, strata and survey level as described above. CLEANING PROCESS: DURABLES For durables, we also apply the same cleaning rules for currencies (Table D.3), and then the following cleaning rules: Vintages with missing values and greater than 10 years are replaced with item-specific Rule 1 medians. Rule 2 Current and purchase prices equal to zero are replaced with item-specific medians. Records that have the same figure in current value and purchase price are incorrect. For Rule 3 both, the item-vintage-specific distribution is calculated to determine the deviation of the entered figure from the median. The one that is further away from that median is corrected with the item-year-specific median value. Depreciation rates are replaced by the item-specific medians in the following cases: Rule 4 • Negative records • Depreciation rates in the top 10 percent and vintage of one year • Depreciation rates in the bottom 10 percent and a vintage greater or equal to 3 years Records with 100 items or more, and those that reported to own a durable good but did not Rule 5 report the number were replaced with the item-specific medians of consumption in US$ Consumption in the top and bottom 1 percent of the overall distribution were replaced Rule 6 with item-specific medians. Records where the respondent did not know or refused to respond if the household Rule 7 owned the asset, were replaced with the mean consumption value, including non- consumed records. 104 | Appendix The consumption value in US$ was truncated to the mean plus 3 times the standard Rule 8 deviation of the cumulative distribution for each item, if the record exceeded this threshold. All medians are estimated at the EA level if a minimum of 3 observations are available excluding previously tagged records. If the minimum number of observations is not met, medians are estimated at the strata-level before proceeding to the survey level. Table D.5 contains a general overview of consumption of durables, and Table D.6 presents the details. Table D.7 shows the median depreciation rate by item. Table D.1: Summary of unit cleaning rules for food items. Unit Condition Correction Affected Areas53 250 ml tin <=0.03 Multiply by 4 2; 39 Animal back, ribs, shoulder, thigh, head or leg >=7 Divide by 10 4; 35 Basket or Dengu (2 kg) >=10 Divide by 10 1,004; 20 Bottle (1 kg) >=10 Divide by 10 473; 281 Cup (200 g) >200 Divide by 2 447; 24 Faraasilad (12kg) >12 Divide by 12 544; 60 Gram (if item corresponds to a spice) <1 Multiply by 100 115; 5 Gram (if item does not correspond to a spice) <1 Multiply by 1,000 69; 19 Haaf (25 kg) >=25 Divide by 25 357; 921 Heap (700g) >=0.69 Divide by 7 182; 11 Kilogram >=100 Divide by 1,000 68; 4 Large bag (50 kg) >=50 Divide by 50 1; 27 Liter >=10 Divide by 10 3; 32 Madal/Nus kilo ruba (0.75kg) >=7.5 Divide by 10 849; 20 Meals (300 g) >2.1 Divide by 10 366; 208 Packet sealed box/container (500 g) >=5 Divide by 10 340; 16 Piece (large - 300g) >=3 Divide by 10 397; 43 Piece (small - 150g) >=1.5 Divide by 10 95; 5 Rufuc/Jodha (12.5kg) >=12.5 Divide by 10 37; 15 Saxarad (20kg) >=20 Divide by 10 312; 793 Small bag (1 kg) >=10 Divide by 10 110; 8 Teaspoon (10 g) <0.009 Multiply by 10 45; 4 53 The first number indicates the number of affected records reported for consumption while the second number states the number of affected records for purchases. Appendix | 105 Table D.2: Conversion factor to Kg for specific units and items. Items Unit Conversion to Kg Biscuits Piece – large 0.030 Piece - small 0.010 Bread Piece – large 0.400 Piece - small 0.100 Eggs Piece – large 0.070 Piece - small 0.050 Canned fish/shellfish Piece – large 0.420 Piece - small 0.140 Grapefruits, lemons, guavas, limes Piece – large 0.350 Piece - small 0.100 Milk Piece – large 0.750 Piece - small 0.250 Milk powder Piece – large 0.450 Piece - small 0.100 Small bag 1.00 Garlic Piece – large 0.065 Piece - small 0.040 Onion Piece – large 0.150 Piece - small 0.095 Tomatoes Piece – large 0.200 Piece - small 0.110 Bell-pepper Piece – large 0.150 Piece - small 0.080 Sweet/ripe bananas Piece – large 0.110 Piece - small 0.070 Canned vegetables Piece – large 0.400 Piece – small 0.200 Sorghum, flour Cup 0.200 Cooking oats, corn flakes Cup 0.200 Other cooked foods from vendors Small bag 1.00 Purchased/prepared tea/coffee Small bag 0.400 consumed at home Other spices Small bag 0.400 106 | Appendix Table D.3: Summary of cleaning rules for currency. Currency Condition Correction Somaliland shillings Entry in Somaliland shilling Replace currency to Somali shillings Price <=500 Replace currency to Somaliland shillings Price>=500,000 (Thousands) Somali shillings Entry in Somali shilling Divide by 10 Price <=500 Replace currency to Somaliland shillings Price>=500,000 Replace currency to Somali shillings (Thousands) US$ Price >1,000 Divide by 10 Replace currency to Somali(land) shillings Table D.4: Threshold for non-food item expenditure (US$). Recall period Minimum Maximum 1 Week 0.05 30 1 Month 0.20 95 3 Months 0.45 200 1 Year 0.80 1,200 Table D.5: Consumption of durable goods (per week in current US$). SOM Wave 1 SOM Wave 1 Pilot All regions Mogadishu Mogadishu Median 0.74 1.17 1.01 Mean 1.24 1.52 1.91 SD 1.51 1.49 2.62 Appendix | 107 Table D.6: Median consumption of durable goods (per week in current US$). Item SOM Wave 1 SOM Wave 1 Pilot All regions Mogadishu Mogadishu Air conditioner 0.005 0.005 0.041 Bed N/A N/A 0.861 Bed with mattress 0.700 0.746 N/A Car 0.001 0.001 0.001 Cell phone 0.361 0.413 0.430 Chair 0.073 0.072 0.253 Clock 0.028 0.003 0.046 Coffee table (for sitting room) 0.005 0.005 0.106 Computer equipment & accessories 0.020 0.020 2.837 Cupboard, drawers, bureau 0.240 0.240 1.099 Desk 0.047 0.005 0.429 Electric stove or hot plate 0.001 0.001 N/A Electric or gas stove; hot plate N/A N/A 0.012 Electric stove N/A N/A 0.004 Fan 0.069 0.064 0.101 Gas stove 0.007 0.007 0.275 Generator 0.000 0.000 0.000 Iron 0.043 0.035 N/A Kerosene/paraffin stove 0.024 0.007 0.009 Kitchen furniture 0.023 0.015 1.112 Lantern (paraffin) 0.000 0.000 0.002 Lorry 0.000 0.000 0.000 Mattress without bed 0.217 0.212 N/A Mini-bus 0.000 0.000 0.001 Mortar/pestle 0.016 0.009 0.112 Motorcycle/scooter 0.002 0.002 0.006 Photo camera 0.001 0.001 0.595 Radio (‘wireless’) 0.021 0.001 0.016 Refrigerator 0.282 0.018 0.267 Satellite dish 0.117 0.008 0.265 Sewing machine 0.002 0.002 0.732 Small solar light 0.003 0.003 N/A Solar panel 0.000 0.000 0.018 Stove for charcoal 0.032 0.023 0.020 Table 0.042 0.042 0.092 Tape or CD/DVD player; HiFi 0.001 0.001 0.092 Television 0.330 0.278 0.417 Upholstered chair, sofa set 0.019 0.019 2.657 VCR 0.000 0.000 0.000 Washing machine 0.405 0.368 0.557 108 | Appendix Table D.7: Median depreciation rate of durables goods. Item Wave 1 Wave 1 Pilot Wave 1: All SLHS13 All Mogadishu Mogadishu pre-war regions Air conditioner 0.278 0.241 0.210 0.134 0.145 Bed N/A N/A 0.364 N/A 0.088 Bed with mattress 0.172 0.172 N/A 0.172 N/A Car 0.118 0.118 0.111 0.118 0.066 Cell phone 0.188 0.188 0.296 0.188 0.169 Chair 0.149 0.149 0.371 0.149 0.114 Clock 0.204 0.204 0.228 0.204 0.110 Coffee table (for sitting room) 0.279 0.279 0.329 0.279 0.114 Computer equipment & accessories 0.182 0.240 0.364 0.150 0.204 Cupboard, drawers, bureau 0.150 0.150 0.296 0.150 0.098 Desk 0.134 0.134 0.502 0.134 0.108 Electric stove or hot plate 0.262 0.257 0.005 0.252 N/A Electric stove N/A N/A 0.296 N/A 0.138 Fan 0.131 0.131 0.235 0.131 0.134 Gas stove 0.174 0.135 0.296 0.174 0.333 Generator N/A N/A 0.296 N/A 0.127 Iron 0.161 0.161 N/A 0.161 0.110 Kerosene/paraffin stove 0.224 0.224 0.296 0.224 0.210 Kitchen furniture 0.188 0.188 0.393 0.188 0.101 Lantern (paraffin) 0.064 N/A 0.067 0.064 0.114 Lorry 0.154 N/A 0.296 0.154 0.052 Mattress without bed 0.185 0.185 N/A 0.185 N/A Mini-bus 0.153 0.172 0.296 0.153 0.039 Mortar/pestle 0.210 0.210 0.254 0.210 0.114 Motorcycle/scooter 0.172 0.172 0.138 N/A N/A Photo camera 0.134 0.134 0.296 0.122 0.171 Radio (‘wireless’) 0.210 0.210 0.337 0.210 0.134 Refrigerator 0.133 0.133 0.065 0.133 0.096 Satellite dish 0.110 0.110 0.303 0.110 0.097 Sewing machine 0.138 0.114 0.296 0.138 0.134 Small solar light 0.296 N/A N/A 0.471 N/A Solar panel 0.005 0.038 0.296 0.005 0.110 Stove for charcoal 0.226 0.226 0.337 0.254 0.188 Table 0.157 0.157 0.296 0.160 0.114 Tape or CD/DVD player; HiFi 0.172 N/A 0.138 0.172 0.092 Television 0.131 0.131 0.240 0.131 0.099 Upholstered chair, sofa set 0.168 0.168 0.289 0.168 0.101 VCR 0.166 0.488 0.296 0.130 0.092 Washing machine 0.138 0.138 0.171 0.138 0.114 Appendix | 109 E. RAPID CONSUMPTION METHODOLOGY The survey used the new Rapid Consumption by-one while iterating over the optional module methodology to estimate consumption. A detailed in each step. A more sophisticated method takes description including an ex post assessment into account correlation between items and of the methodology is available in a separate partitions them into orthogonal sets per module. document.54 The rapid survey consumption This leads to high correlation between modules methodology consists of five main steps. First, supporting the total consumption estimation. core items are selected based on their importance Conceptual division into core and optional items for consumption. Second, the remaining items are is not reflected in the layout of the questionnaire. partitioned into optional modules. Third, optional Rather, all items per household will be grouped modules are assigned to groups of households. into categories of consumption items (like After data collection, fourth, consumption of cereals) and different recall periods. Using CAPI, optional modules is imputed for all households. it is straight-forward to hide the modular structure Fifth, the resulting consumption aggregate is used from the enumerator. to estimate poverty indicators. Third, optional modules will be assigned to First, core consumption items are selected. groups of households. Assignment of optional Consumption in a country bears some variability modules will be performed randomly stratified but usually a small number of a few dozen items by enumeration areas to ensure appropriate captures the majority of consumption. These representation of optional modules in each items are assigned to the core module, which enumeration area. This step is followed by the will be administered to all households. Important actual data collection. items can be identified by its average food share per household or across households. Previous Fourth, household consumption will be estimated consumption surveys in the same country or by imputation. The average consumption of each consumption shares of neighboring / similar optional module can be estimated based on countries can be used to estimate food shares.55 the sub-sample of households assigned to the In the worst case, a random assignment results in a optional module. In the simplest case, a simple larger standard error but does not introduce a bias. average can be estimated. More sophisticated techniques can employ a welfare model based Second, non-core items are partitioned into on household characteristics and consumption of optional modules. Different methods can be used the core items. The results presented in this note for the partitioning into optional modules. In the uses a multiple imputation technique based on a simplest case, the remaining items are ordered multi-variate normal approximation. according to their food share and assigned one- 54 Pape & Mistiaen (2015), “Measuring Household Consumption and Poverty in 60 Minutes: The Mogadishu High Frequency Survey”, World Bank (2015). 55 As shown later, the assignment of items to modules is very robust and, thus, even rough estimates of consumption shares are sufficient to inform the assignment without requiring a baseline survey. 110 | Appendix Next, the methodology is formalized and assessed using an ex post simulation based on the consumption data from Hergeiza using the Somaliland 2013 Household Survey (SLHS13). Food and non-food consumption for household i are estimated by the sum of expenditures for a set of items where yi f and yi f denote the food and non-food consumption of item j in household i. As the estimation for food and non-food consumption follows the same principles, we neglect the upper index f and n in the remainder of this section. The list of items can be partitioned into M+1 modules each with mk items: For each household, only the core module yi(0) and one additional optional module yi(k*) are collected. The item assignment to the modules are based on the SLHS13 survey with manual modifications specially to treat ‘other’ items correctly.56 The core module was designed to maximize its consumption share resulting in 91 percent and 76 percent of food respectively non-food consumption captures in the core modules (based on SLHS13 consumption; Table E.1). Optional modules are constructed using an algorithm to assign items iteratively to optional modules so that items are orthogonal within modules and correlated between modules. In each step, an unassigned item with highest consumption share is selected. For each module, total per capita consumption is regressed on household size, the consumption of all assigned items to this module as well as the new unassigned item. The item will be assigned to the module with the highest increase in the R2 relative to the regression excluding the new unassigned item. The sequenced assignment of items based on their consumption share can lead to considerable differences in the captured consumption share across optional modules. Therefore, a parameter is introduced ensuring that in each step of the assignment procedure the difference in the number of assigned items per module does not exceed d. Using d=1 assigns items to modules (almost) maximizing equal consumption share across modules.57 Increasing d puts increasing weight on orthogonality within and correlation between modules. The parameter was set to d=3 balancing the two objectives. In each enumeration area, 12 households were interviewed with an ideal partition of three items per optional module.58 The assignment of optional modules must ensure that a sufficient number of households are assigned to each optional module. Household consumption was then estimated using the core module, the assigned module and estimates for the remaining optional modules. 56 Items ‘other’ are often found to capture remaining items for a food category. Using the Rapid Consumption Methodology, this creates problems as ‘other’ will include different items depending on which optional module is administered. This can lead to double-counting after the imputation. Therefore, ‘other’ items are re-formulated and carefully assigned so that double counting cannot occur. 57 Even with d=1, equal consumption share across modules is not maximized because among the modules with the same number of assigned items, the new item will be assigned to the module it’s most orthogonal to; rather than to the module with lowest consumption share. 58 Field work implementation aimed to achieve a balanced partition among optional modules but due to challenges in following the protocol exactly some enumeration areas are not completely balanced. Appendix | 111 where K* := {1,…,k*-1,k*+1,…,M} denotes the set of non-assigned optional modules. Consumption of non-assigned optional modules is estimated using multiple imputation techniques taking into account the variation absorbed in the residual term. Multiple imputation was implemented using multi-variate normal regression based on an EM-like algorithm to iteratively estimate model parameters and missing data. This technique is guaranteed to converge in distribution to the optimal values. An EM algorithm draws missing data from a prior (often non-informative) distribution and runs an OLS to estimate the coefficients. Iteratively, the coefficients are updated based on re-estimation using imputed values for missing data drawn from the posterior distribution of the model. The implemented technique employs a Data-Augmentation (DA) algorithm, which is similar to an EM algorithm but updates parameters in a non-deterministic fashion unlike the EM algorithm. Thus, coefficients are drawn from the parameter posterior distribution rather than chosen by likelihood maximization. Hence, the iterative process is a Monte-Carlo Markov –Chain (MCMC) in the parameter space with convergence to the stationary distribution that averages over the missing data. The distribution for the missing data stabilizes at the exact distribution to be drawn from to retrieve model estimates averaging over the missing value distribution. The DA algorithm usually converges considerably faster than using standard EM algorithms: Figure E.1: Relative bias of simulation results Figure E.2: Relative standard error of simulation using the rapid consumption estimation. results using the rapid consumption estimation. 2.0% 2.0% 1.5% 1.0% 1.5% 0.5% 1.0% 0.0% -0.5% 0.5% -1.0% -1.5% 0.0% TO T1 T2 TO T1 T2 HH EA l HH EA l Al Al FG FG FG FG FG FG n n n n n n io io io io io io pt pt pt pt pt pt um m um um m um su su ns ns ns ns n n Co Co Co Co Co Co Source: Authors’ calculations based on the SLHS13. Source: Authors’ calculations based on the SLHS13. 112 | Appendix The performance of the estimation technique was assessed based on an ex post simulation using the Hergeiza data from SLHS13 and mimicking the Rapid Consumption methodology by masking consumption of items that were not administered to households. The results of the simulation were compared with the estimates using the full consumption from SLHS13 as reference. The simulation results distinguish between different levels of aggregation to estimate consumption.59 The methodology generally does not perform well at the household level (HH) but improves considerably already at the enumeration area level (EA) where the average of 12 households is estimated. At the national aggregation level, the Rapid Consumption methodology slightly over-estimates consumption by 0.3 percent. Assessing the standard poverty measures including poverty headcount (FGT0), poverty depth (FGT1) and poverty severity (FGT2), the simulation results show that the Rapid Consumption methodology retrieves estimates within 1.5 percent of the reference measure (Figure E.1). Generally, the estimates are robust as suggested by the low standard errors (Figure E.2). Table E.1: Item partitions based on SLHS13 and the pilot in Mogadishu. Food Items Non-food Items Number Share Share Share Number Share Share Share of Hergeiza Mogadishu Mogadishu of Hergeiza Mogadishu Mogadishu items items Imputed Imputed Core 33 91% 64% 54% 26 76% 62% 52% Module 1 19 3% 9% 16% 15 7% 9% 12% Module 2 20 2% 14% 14% 15 5% 9% 12% Module 3 15 2% 5% 6% 15 6% 8% 9% Module 4 15 2% 8% 9% 15 6% 11% 15% Source: Authors’ calculations based on the SLHS13. DURABLE CONSUMPTION FLOW The consumption aggregate includes the consumption flow of durables calculated based on the user-cost approach. The consumption flow distributes the consumption value of the durable over multiple years. The user-cost principle defines the consumption flow of an item as the difference of selling the asset at the beginning and the end of the year as this is the opportunity cost of 59 The performance of the estimation techniques is presented using the relative bias (mean of the error distribution) and the relative standard error. The relative error is defined as the percentage difference of the estimated consumption and the reference consumption (based on the full consumption module, averaged over all imputations). The relative bias is the average of the relative error. The relative standard error is the standard deviation of the relative error. The simulation is run over different household-module assignments while ensuring that each optional module is assigned equally often to a household per enumeration. The relative bias and the relative standard error are reported across all simulations. Appendix | 113 the household for keeping the item. The opportunity cost is composed of the difference in the sales price and the forgone earnings on interest if the asset is sold at the beginning of the year. If the durable item is sold at the beginning of the year, the household would receive the market price pt for the item and the interest on the revenue for one year. With it denoting the interest rate, the value of the item thus is pt (1+it ). If the item is sold at the end of the year, the household will receive the depreciated value of the item while considering inflation. With πt being the inflation rate during the year t, the household would obtain pt (1+πt )(1-O) with the annual physical or technological depreciation rate denoted as o assumed constant over time.60 The difference between these two values is the cost that the household is willing to pay for using the durable good for one year. Hence, the consumption flow is: By assuming that δ×π_t≅0, the equation simplifies to where rt is the real market interest rate in period t. Therefore, the consumption flow of an item can be estimated by the current market value pt, the current real interest rate rt, and the depreciation rate π. Assuming an average annual inflation rate π, the depreciation rates o can be estimated utilizing its relationship to the market price:61 The equation can be solved for o obtaining: Based on this equation, item-specific median depreciation rates are estimated assuming an inflation rate of 0.5 percent, a nominal interest rate of 2.0 percent and, thus, a real interest rate of 1.5 percent (Table D.7). For all households owning a durable but did not report the current value of the durable, the item- specific median consumption flow is used. For households that own more than one of the durable, the consumption flow of the newest item is added to the item-specific median of the consumption flow times the number of those items without counting the newest item.62 60 Assuming a constant depreciation rate is equivalent to assuming a “radioactive decay” of durable goods (see Deaton and Zaidi, 2002). 61 In particular, π solves the equation 62 The 2016 HFS questionnaire provides information on a) the year of purchase and b) the purchasing price only for the most recent durable owned by the household. 114 | Appendix DEFLATOR Prices fluctuate considerably between regions, thus we calculated spatial price indices using a common food basket and spatial prices to make consumption comparable across regions. The Laspeyres index is chosen as a deflator due to its moderate data requirements. The deflator is calculated by analytical strata areas based on the price data collected by the HFS. The Laspeyres index reflects the item-weighted relative price differences across products. Item weights are estimated as household-weighted average consumption share across all households before imputation. Based on the democratic approach, consumption shares are calculated at the household level. Core items use total household core consumption as reference while items from optional modules use the total assigned optional module household consumption as reference. The shares are aggregated at the national level (using household weights) and then calibrated by average consumption per module to arrive at item-weights are applied to the relative differences of median item prices for each analytical strata. Missing prices are replaced by the item-specific median over all households. A large Laspeyres indicates a high price level deflating consumption stronger than a lower Laspeyres index. The resulting indices show the fluctuation of prices across regions (Table E.2). Table E.2: Laspeyres deflator by analytical strata. Analytical Strata Deflator All IDPs 0.923 Mogadishu 0.964 Garowe 0.862 Urban Bari and Mudug 1.107 Hergeiza 1.133 Urban Awdal, Sanaag, Sool and Togdheer 0.922 Rural Bari, Mudug and Nugaal 1.013 Rural Awdal, Sanaag, Sool, Togdheer and Woqooyi Galbeed 1.075 Appendix | 115 F. LABOR STATISTICS This appendix describes the construction of key (ILO) Key Indicators of the Labour Market (KILM), labor statistics for Wave 1 of the Somali High wherever possible and sensible given data Frequency Survey. The statistics presented in restrictions. The KILM consist of the 17 most this note follow closely the international standard important indicators of labor market conditions, set as per the International Labour Organisation’s designed to allow for cross-country comparisons. PRELIMINARY DEFINITIONS The labor market indicators at hand rely critically Labor Force Status on a number of preliminary definitions that recur throughout the construction of the higher- Labor force status comprises three mutually level statistics. This section introduces the most exclusive and exhaustive categories: important concepts. 1. Employment, 2. Unemployment, Reference Periods 3. Outside the labor force or inactivity. There are two key reference periods: (a) the short Persons in employment are those who are of observation period defined as 7 days, and (b) the working-age and engaged in activities producing long observation period defined as 12 months. goods or providing services for at least one hour Following ILO guidelines, most statistics are during the past seven days. This includes persons reported for the short observation period. working for pay or for profit and workers who contributed within the family establishment. Working Age and Age Groups Note that this definition deviates slightly from the international standard. This is related to In the SHFS, working age is defined as all persons the concept of ‘contributing family member’. A aged 15 to 64. This definition departs slightly contributing family member works in the family from the ILO definition (15 years and older, no establishment, and is not remunerated directly, top limit). It is referenced against standardized profits accruing to the family. The international age groups of five years, i.e. 15–19, 20–24, 25– standard counts contributing family members 29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, as ‘employed’ only if the family establishment 60–64. Youth labor is defined over age group is a market unit, i.e. it works for pay or profit of 15–24, adult labor as 25+, and elderly labor as some kind. This definition excludes production 55+. In addition, the data of Wave 1 of the High for own-use from the ‘employment’ category. In Frequency survey for Somalia (HFS SOM) includes contrast, the SHFS does not determine whether a information on children of the ages 10 to 14 contributing family member works in an own-use engaging in work activities (ILO, 2013). or in a market-unit family establishment. 116 | Appendix Specifically, in Wave 1 of the SHSF, employment questions: the absence of employment as defined is constructed by determining if a person has above and ascertaining whether the respondent engaged, over the previous 7 days (short reference has been looking for work in the past four weeks period), or over the past 12 months (long reference and is available to take up work (ILO, 2013). period), in one of the following work activities: 1. Working as an apprentice Persons outside the labor force those of working- 2. Working on the household’s farm, raising age who are neither employed nor unemployed, livestock, hunting or fishing according to the preceding definitions. Persons 3. Conducting paid or commissioned work outside the labor force are also referred to as 4. Running a business of any size for oneself or ‘inactive’. But inactive should not be construed for the household as idle, especially in the context of a developing 5. Helping in a household business of any size. economy. The Wave 1 data of the SHFS allows drawing important distinctions within the group The definition further includes persons who are of persons outside the labor force. This group temporarily absent from their work due to training comprises persons who work in the household, or working-time arrangements such as overtime persons in education, and discouraged persons, leave, and paid interns. Note that the definition among others (ILO, 2013). excludes household work. The labor force is the sum of persons in Persons in unemployment are of those of working- employment and in unemployment. It is the age not in employment during the short reference counterpart of the group of inactive persons, period, but seeking employment within the past i.e. the labor force plus the inactive sum up to four weeks, and currently available to take up the entire working-age population (ILO, 2013). employment. In the HFS SOM data, unemployment visualizes the distinctions between labor force, is determined through the combination of three inactivity, and employment status. Figure F.1: Labor force. Working-age Population (15 to 64 years) Labor Force/Active Outside of the labor force/inactive In in Pursuing Household Discouraged Other Employment Unemployment Education work Appendix | 117 LABOUR MARKET INDICATORS This section lays out the indicators presented in this document. Each indicator is presented as the overall average for the sample as well as disaggregated by region, urban/rural, and IDP camps, gender, and consumption quintile, allowing for a detailed analysis of the labor market situation in Somalia. Labor Force Participation and Inactivity The labor force participation rate (LFPR) is the ratio of the labor force to the working age population, expressed as percentages. That is, where LF is labor force, POP is working age population, t is the reference period, a refers to age groups, and s to sex. The LFPR provides an indication of labor supply relative to the population at large (Bourmpoula, Kapsos, Pasteels, 2013). The Inactivity Rate (IR) is the number of inactive persons of working age as a percentage of the working- age population. As such, it is the counterpart to the LFPR, given by 100 minus LFPR. Of particular interest are three groups: Household workers, pursuing education, and discouraged, inactive persons who state they are not looking for work because of unavailability of jobs. All three are determined by asking respondents why they have not been looking for a job in the past four weeks. The size of the inactive population may change over time. For example, as perceived employment prospects change, some people resume looking for work, thereby entering the labor force (ILO, 2015). Employment, Unemployment, Hours of Work The employment-to-population ratio (ER) is proportion of the working-age population that is employed, i.e. with EMP referring to the number of persons in employment, and all other variables defined as before. The employment-to-population ratio is a way to assess the ability of the economy to create employment. Note that in the context of the low-income countries, the ER sometimes decreases in times of growth and development due to concurrent improvements in education and training opportunities (ILO, 2015) The unemployment rate is the number of persons in unemployment as a percentage of the total labor force. With unemployment defined as above, the unemployment rate (UR) is given by 118 | Appendix Youth Unemployment refers to unemployed persons in the 15-to-24 age-bracket: This is figure is complemented by another statistic: The number of 15-to-24-yearolds not in employment, education or training (NEET) as a percentage of the entire youth population. The NEET is a key metric for determining the state of the economy’s youth and their prospects. The NEET is determined in the HFS SOM by means of matching the youth population according to whether they are currently in employment, or currently pursuing education. Long-term Unemployment refers to persons unemployed for 12 months or longer: Two metrics are of interest: first, the long-term unemployment rate; and, second, the incidence of long-term unemployment, that is, the long-term unemployed as a percentage of the total unemployed. Where unemployment as such is not necessarily and indicator of wellbeing or the lack thereof, long-term unemployment can be considered to have a closer relation with well-being in many contexts (ILO, 2013). Long-term unemployment is determined in the HFS SOM by comparing those who state having worked at some point in their lives while not having worked in the past 12 months. Hours of work refers to the total hours of work spent on any work activity during the past 7 days: This is the definition of actual hours worked per week, which includes ‘related hours’, e.g. cleaning of instruments, ‘down time’, and ‘resting time’. Hours worked is central in so far as it is the starting point for number of other indicators, such as time-related underemployment (persons working less than they desire) – for which working hours of 20 or less may be indicative –, part-time employment, and over-employment (persons working 49 hours or more per week) (ILO, 2013). In the HFS SOM, respondents self-report hours worked. Note that this goes both for hours in employment and hours worked in the household. Both metrics are censored at 100 hours per week. Employment by Status, Sector, and Occupation Status in Employment refers to two broad categories of the employed population: employees or salaried workers, and self-employed workers. The self-employed are further distinguished as A B C D employers: self- own-account workers: contributing workers not classifiable employed with self-employed without family workers, by status: workers who employees, employees, cannot be classified in one of the preceding categories, e.g. due to lack of available information. Appendix | 119 Status in employment speaks both to working conditions as well as the state of development of the economy. A large proportion of own-account workers is indicative of a less developed economy, a large agriculture sector, and low growth in the formal economy. Specifically, a large share of contributing family workers suggests low levels of development. In addition, own-account and contributing family workers, typically lacking formal work arrangements, are considered to be in vulnerable employment. This group is less likely to have conditions of decent employment as defined by the Millennium Development Goals (ILO, 2013). In the HFS SOM, status in employment is determined by respondents’ direct self-classification into one of the categories. Employment by sector: In line with the International Standard Industrial Classification of all Economic Activities (ISIC) Revision 4 of 2008, sectors are defined as 1 2 3 4 Agriculture Industry Services Sector not adequately defined. Employment by sector provides an insight into the stage of development of the economy. Economic development has historically been associated with fundamental shifts in the allocation of the labor force, from agriculture, towards industry, and eventually services (ILO, 2015). In the HFS SOM, sectors are collapsed from a list narrower categories as defined by ISIC, Rev.4 2008, according to which each respondent is classified: A B C D Agriculture, forestry Mining and quarrying Manufacturing Electricity, gas, steam and fishing and air conditioning supply E F G H Water supply; Construction Wholesale and Transportation and sewerage, waste retail trade; repair of storage management and motor vehicles and remediation activities motorcycles 120 | Appendix I J K L Accommodation and Information and Financial and insurance Real estate activities food service activities communication activities M N O P Professional, scientific Administrative and Public administration Education and technical activities support service and defence; activities compulsory social security\ Q R S T Human health and Arts, entertainment Other service activities Activities of households social work activities and recreation as employers; undifferentiated goods- and services-producing activities of households for own use U V Activities of Not classified/No extraterritorial occupation organizations and bodies In this classification, category A corresponds to agriculture, categories B-F to industry/manufacturing, and G-U to services (UN, 2008). Employment by Occupation: The International Standard Classifications of Occupations of 2008 (ISCO08) defines the major employment groups, along with suggested levels of skill, as presented in Table F.1. ISCO skill levels are defined as: (1) primary education; (2) first stages of secondary education; (3) completed secondary education, and training not equivalent to a university degree; (4) university degree or equivalent. Employment by Occupation is informative of levels and composition of skills in the economy (ILO, 2008). In the HFS SOM, ISCO-08 occupations are determined via self-classification of respondents aged 15 and older. Appendix | 121 Table F.1: ISCO 08. ISCO08 Major Groups ISCO Skill Level 1 Managers 3+4 2 Professionals 4 3 Technicians and Associate Professionals 3 4 Clerical support workers 2 5 Service and sales workers 2 6 Skilled agricultural, forestry and fishery workers 2 7 Craft and related trade workers 2 8 Plant and machine operators and assemblers 2 9 Elementary occupations 1 10 Armed forces occupations 1+ 2 + 4 11 Non-classifiable workers. - EDUCATIONAL ATTAINMENT AND LITERACY Levels of education and basic literacy are a key metric for the human capital supply in the labor market. Literacy is the ability to read and write a simple sentence about every-day life: In the HFS SOM the ability to read and the ability to write are assessed in two separate questions in order to avoid confusion in regards to the concept (ILO, 2015). The five categories of educational attainment are: No education/Less than primary, primary, secondary, tertiary, and other. This definition is in line with the International Standard Classification of Education (ISCED) of the UN. Note that ‘primary’ includes primary education as well as lower, incomplete secondary; ‘secondary’ includes upper secondary and non-tertiary post-secondary education; and tertiary covers all levels of tertiary education (UNESCO, 2012). In the HFS SOM, educational attainment is determined by means of self-classification of respondents in levels of schooling in line with the education system in Somalia. Of note, the ‘other’ category includes non-formal education as well as the option ‘other’ as chosen by respondents. The ‘tertiary’ category contains first university degree, master’s degree, PhD, and post-secondary technical education. 122 | Appendix G. REMITTANCES In the Wave 1 of the SHFS, data on remittances was collected at the household level, as part of the household characteristics module of the questionnaire. The primary reference period for the receipt of remittances is the past 12 months. Five main question determine the nature and scope of remittances received: 1. If the household received remittances in the past 12 months or not, 2. Amount and currency of remittances receipt in past 12 months, 3. If the household received remittances in 12 months prior to the past 12 months, 4. Change (same/more/less) in the value of remittances between the two periods, 5. Reason for this change. Around 22% of the households (1, 905 out of 4,117) reported receiving remittances in the past 12 months. The cleaning process of this data was done in a three-step process. First, corrections were introduced to of the exchange rate selected (Table D.3). Then, the value of remittances was converted to US$. Finally, the following cleaning rules were used to identify and replace outliers: Records where the respondent did not know or refused to respond if the household had Rule 1 received remittances, were replaced with the mean value, considering all the recipients and non-recipients. Outliers were identified and replaced with the median for the respective analytical strata. Rule 2 This includes values in the top and bottom 1% of the overall cumulative distribution. Appendix | 123