Somali Poverty and Vulnerability Assessment Findings from Wave 2 of the Somali High Frequency Survey Report No. AUS0000407 April 2019 Report No. AUS0000407 Somali Poverty and Vulnerability Assessment Findings from Wave 2 of the Somali High Frequency Survey April 2019 © 2019 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpreta- tions, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to: World Bank Publications The World Bank Group 1818 H Street NW Washington, DC 20433 USA Fax: 202-522-2625 Cover photo: © Miranda Grant/Human Geographic. Acknowledgments This report was led by Utz Pape (Senior Economist, GPV01) and Wendy Karamba (Economist, GPV01). Substantial contributions to the overall report were made by Gonzalo Nunez (Consultant, GPV01) and Philip Wollburg (Consultant, GPV01). Chapter 1 ‘Poverty Profile’ was written by Gonzalo Nunez. Chapter 2 ‘Spatial Variation in Living Standards’ was prepared by Makiko Watanabe (Senior Urban Specialist, GSU13) and Nastassia Leszczynska (Consultant, GSU13). Contributions were made by Olivia D’Aoust (Urban Econ- omist) and Zishan Karim (Senior Urban Specialist, GSU13). Chapter 3 ‘Drought Impact’ was contributed by Philip Wollburg. Chapter 4 ‘Displacement’ was written by Andrea Fitri Woodhouse (Senior Social Devel- opment Specialist, GSU03), Verena Phipps (Senior Social Development Specialist, GSU07) and Ambika Sharma (Consultant, GPV01). Chapter 5 ‘Social Protection’ was prepared by Zaineb Majoka (Consultant, GSP01) with guidance from Maniza Naqvi (Senior Social Protection Specialist, GSPGL). Chapter 6 ‘Remit- tances’ was contributed by Sonia Plaza (Senior Economist, GFCAE) with inputs from Philip Wollburg. The team is grateful for inputs from Stephen Shisoka (Consultant, GPV01) and administrative support from Martin Buchara (Program Assistant, GPV01). The team is also grateful for inputs and comments from Pierella Paci (Practice Manager, GPV01) and Nobuo Yoshida (Lead Economist, GPV01) as well as the peer reviewers John Randa (Senior Economist, GMTA3) and Kinnon Scott (Senior Economist, GPV04). The Somali High Frequency Survey (SHFS) team was led by Utz Pape and comprised Gonzalo Nunez and Philip Wollburg. The survey was implemented by Altai Consulting in coordination with the respective statistical authorities. The team would like to thank the Director General Abdirahman Omar Dahir from the Directorate of National Statistics, Ministry of Planning, Investment and Economic Development of the Federal Government of Somalia for the close collaboration. Funding for the survey was received from the Somalia Knowledge for Results Trust Fund of the Multi-Partner Fund with additional contributions from Environment and Natural Resources (ENR) Global Practice to over-sample coastal areas and from the Somalia Country Management Unit to over-sample IDP host communities. Vice President Hafez M. H. Ghanem Country Director Bella Bird Senior Director Carolina Sanchez Practice Manager Pierella Paci Task Team Leaders Utz Pape, Wendy Karamba Acknowledgments iii Contents Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Abbreviations and Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Chapter 1  Poverty Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Monetary poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Inequality and vulnerable population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Demographic characteristics and labor markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Quality of dwellings and access to services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Multidimensional deprivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 Chapter 2  Spatial Variation in Living Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 Urban-rural comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 Inter-urban comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Policy recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Chapter 3  Drought Impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 The 2016/17 drought and its effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62 Drought impact on welfare and livelihoods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Policy recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Chapter 4  Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73 Displacement profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Poverty and hunger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Access to infrastructure and quality of dwellings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Health and education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Employment and livelihoods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Social cohesion, justice, and security . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Policy recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Chapter 5  Social Protection . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 Sources of vulnerability at macro Level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Inadequate risk management capacity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Experience and impact of shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Resilience building with social safety nets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Policy recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116 Chapter 6  Remittances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119 International mobility patterns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 Remittances at the macroeconomic level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 The development impact of remittances at the microeconomic level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Contents v Remittance markets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129 Policy recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 Appendix A  Figures and Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Appendix B  Intra-Urban Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Appendix C  Estimating the Drought Impact with a Difference-in-Differences Model . . . . . . . . . . . . . . . 151 Appendix D  Regression Results for Each Type of Shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 Appendix E  Methodology for Reduced Coping Strategy Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Appendix F  Displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Appendix G  Data Gaps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173 List of Figures Figure 0.1: Coverage of household surveys in Somali regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 Figure 1.1: Somali households by type of population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Figure 1.2: Cross-country comparison of poverty in 2017 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Figure 1.3: Cross-country comparison of poverty and GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Figure 1.4: Poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 1.5: Map of poverty incidence from satellite estimates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 1.6: Poverty gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Figure 1.7: Poverty severity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Figure 1.8: Child poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Figure 1.9: Youth poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 Figure 1.10: Food consumption poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Figure 1.11: Experience of hunger in past 4 weeks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Figure 1.12: Cross-country comparison of poverty and inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Figure 1.13: Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Figure 1.14: Consumption distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Figure 1.15: Livestock ownership . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Figure 1.16: Number of livestock owned . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Figure 1.17: Female headed households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure 1.18: Labor force participation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Figure 1.19: Reasons for inactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 1.20: Cross-country comparison of literacy rate and GDP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 1.21: Literacy by age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Figure 1.22: Literacy rate by group (aged 15+) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 1.23: Cross-country comparison of net primary school enrollment and GDP . . . . . . . . . . . . . . . . . 23 Figure 1.24: Net school enrollment rate by age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 1.25: School enrollment by level and age . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 1.26: Net enrollment of primary school-aged children . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 1.27: Net enrollment of secondary school-aged children . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 1.28: Reasons for not attending school for children of primary age (6–13) . . . . . . . . . . . . . . . . . . . 26 Figure 1.29: Reasons for not attending school for children of secondary age (14–17) . . . . . . . . . . . . . . . . 26 Figure 1.30: Households more than 30 minutes away from the nearest school . . . . . . . . . . . . . . . . . . . . . 27 Figure 1.31: Average household expenditure on education per member enrolled . . . . . . . . . . . . . . . . . . . 27 Figure 1.32: Educational level . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Figure 1.33: Population without formal education . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Figure 1.34: Type of floor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Figure 1.35: Type of roof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Figure 1.36: Type of cooking source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Figure 1.37: Access to improved sanitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Figure 1.38: Access to improved water sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 vi  Somali Poverty and Vulnerability Assessment Figure 1.39: Cross-country comparison of access to improved sanitation and GDP . . . . . . . . . . . . . . . . . 30 Figure 1.40: Cross-country comparison of access to improved water sources and GDP . . . . . . . . . . . . . 30 Figure 1.41: Access to electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 1.42: Cross-country comparison of access to electricity and GDP . . . . . . . . . . . . . . . . . . . . . . . . . . 31 Figure 1.43: Households more than 30 minutes away from the nearest market . . . . . . . . . . . . . . . . . . . . . 31 Figure 1.44: Households more than 30 minutes away from the nearest health clinic . . . . . . . . . . . . . . . . 31 Figure 1.45: Number of multidimensional deprivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Figure 1.46: Deprivations in various dimensions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Figure 1.47: Nonmonetary deprivations by poverty status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34 Figure 2.1: Poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 2.2: Food poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Figure 2.3: Hunger over the past four weeks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Figure 2.4: Access to electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Figure 2.5: Access to piped water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Figure 2.6: Source of potable water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 2.7: Access to improved sanitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 2.8: Primary school enrollment rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 Figure 2.9: Distance to health facilities (>30 minutes) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Figure 2.10: Dwelling type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Figure 2.11: Living arrangement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40 Figure 2.12: Area occupied by IDP settlements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Figure 2.13: New IDP settlements in Baidoa . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Figure 2.14: New IDP settlements in Kismayo . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41 Figure 2.15: Access to bank accounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Figure 2.16: Households that saved money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42 Figure 2.17: Main sources of income . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Figure 2.18: Perception of employment opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Figure 2.19: Safety from crime and violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Figure 2.20: Dispute resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Figure 2.21: Trust in institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 Figure 2.22: Payment of taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Figure 2.23: Institutions that collected taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45 Figure 2.24: Poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Figure 2.25: Poverty gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Figure 2.26: Hunger . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Figure 2.27: Food poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 Figure 2.28: Access to electricity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 2.29: Access to piped water . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 2.30: Access to improved sanitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 Figure 2.31: Solid waste management . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Figure 2.32: Primary school enrollment rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 Figure 2.33: Literacy rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Figure 2.34: Distance to health facilities (>30 min) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Figure 2.35: Satisfaction on health service quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49 Figure 2.36: Access to market, public transport, internet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Figure 2.37: Tenure status . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Figure 2.38: Access to improved housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50 Figure 2.39: Legal recognition of land and housing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Figure 2.40: Access to bank accounts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Figure 2.41: Households that saved money . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 Figure 2.42: Main sources of income . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Figure 2.43: Perception on employment opportunities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52 Figure 2.44: Safety from crime and violence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Figure 2.45: Freedom of movement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Contents vii Figure 2.46: Dispute resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53 Figure 2.47: Trust in institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Figure 2.48: Payment of taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Figure 2.49: Institutions that collected taxes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 Figure 2.50: Distribution of IDPs and urban population . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Figure 2.51: IDPs’ access to services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Figure 2.52: IDPs’ access to key facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56 Figure 2.53: IDPs’ perception of safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Figure 2.54: Urban IDPs’ access to services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Figure 2.55: Urban IDPs’ access to key facilities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 Figure 2.56: Urban IDPs’ perception of safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58 Figure 3.1: Rainfall and NDVI anomaly and overview of rainy seasons, all regions . . . . . . . . . . . . . . . . . . . 62 Figure 3.2: 2016 Gu precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Figure 3.3: 2016 Deyr precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63 Figure 3.4: 2017 Gu precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Figure 3.5: 2017 Deyr precipitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Figure 3.6: Population facing food insecurity, all regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 Figure 3.7: Internal displacement due to drought . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65 Figure 3.8: NDVI deviation, 2016 Deyr season . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 3.9: NDVI deviation, 2017 Gu season . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 3.10: Distribution of drought exposure, Overall, Wave 1, Wave 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 Figure 3.11: Illustration of difference-in-differences approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68 Figure 3.12: Drought effect along the income distribution, rural areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Figure 3.13: Drought effect on hunger and food consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70 Figure 3.14: Simulation of income shock among rural households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Figure 3.15: Correlates of drought-impacted rural households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 Figure 4.1: Number of displacements occurring by month, Jan 2016—Apr 2018 . . . . . . . . . . . . . . . . . . . . 74 Figure 4.2: Regional distribution of IDPs, SHFS sample and UNHCR PRMN data . . . . . . . . . . . . . . . . . . . 76 Figure 4.3: Population structure for IDP, non-IDPs and refugees by gender and age . . . . . . . . . . . . . . . . 77 Figure 4.4: IDP profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Figure 4.5: Urban/rural composition of IDPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 Figure 4.6: Trends in traveling to current location, for IDPs and refugees . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 4.7: Original location relative to current location for IDPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 4.8: Reason for leaving original location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79 Figure 4.9: Reason for arriving at current location . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Figure 4.10: Years since IDP displacement and arrival in current location . . . . . . . . . . . . . . . . . . . . . . . . . . 81 Figure 4.11: Conflict events and dates of displacement of conflict-driven IDPs . . . . . . . . . . . . . . . . . . . . . 81 Figure 4.12: Rainfall anomalies, Gu-Deyr seasons, and displacement dates of climate-driven IDPs . . . . 82 Figure 4.13: Dates of displacement for Somali refugees in Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 Figure 4.14: Return intentions of IDPs and refugees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Figure 4.15: Trends in revisiting the original residence location for IDPs . . . . . . . . . . . . . . . . . . . . . . . . . . . 83 Figure 4.16: Push factors for IDPs and refugees who don’t want to move . . . . . . . . . . . . . . . . . . . . . . . . . 83 Figure 4.17: Pull factors for IDPs who want to move . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Figure 4.18: Return timeline for IDPs and refugees that intend to move . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 Figure 4.19: Legal identification and access to documentation restitution mechanisms . . . . . . . . . . . . . 84 Figure 4.20: Poverty headcount ratio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85 Figure 4.21: Poverty gap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 Figure 4.22: Hunger incidence in the last four weeks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87 Figure 4.23: Access to improved housing, now and before displacement . . . . . . . . . . . . . . . . . . . . . . . . . . 88 Figure 4.24: Access to improved drinking water, for IDPs, refugees and residents . . . . . . . . . . . . . . . . . . 89 Figure 4.25: Access to improved sanitation for IDPs, refugees and residents . . . . . . . . . . . . . . . . . . . . . . 89 Figure 4.26: Number of households sharing a toilet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 viii  Somali Poverty and Vulnerability Assessment Figure 4.27: Households more than 30 minutes from services . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Figure 4.28: Access to electricity to charge mobile phone . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 Figure 4.29: Under 15 minutes to network reception point . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 Figure 4.30: Births in health facilities, for IDPs, hosts, refugees, and residents . . . . . . . . . . . . . . . . . . . . . 91 Figure 4.31: Births attended by skilled health staff, for IDPs, hosts, refugees and residents . . . . . . . . . . 91 Figure 4.32 Adult literacy rate by gender, IDPs, refugees, and residents . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Figure 4.33: School enrollment among the school-aged . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 Figure 4.34: Labor force participation for IDPs, refugees and urban and rural residents . . . . . . . . . . . . 93 Figure 4.35: Changes in employment activity after displacement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93 Figure 4.36: Proportion of women perceived to be allowed to work outside the home . . . . . . . . . . . . . 94 Figure 4.37: Reasons for economic inactivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94 Figure 4.38: Main employment activity for IDPs, hosts, refugees, and rural residents . . . . . . . . . . . . . . . 94 Figure 4.39: Main source of income for IDPs, hosts, and residents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95 Figure 4.40: Main source of income for refugees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96 Figure 4.41: Average remittances for IDPs, hosts, and residents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Figure 4.42: Perceptions of safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Figure 4.43: Perceived relations of IDPs with surrounding community . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97 Figure 4.44: Perceptions of refugees among host communities in Ethiopia . . . . . . . . . . . . . . . . . . . . . . . 98 Figure 5.1: Distribution of losses incurred due to 2017 drought by sector . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Figure 5.2: Coping strategies in response to the 2017 drought . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Figure 5.3: Incidence of reported shocks among Somali households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Figure 5.4: Incidence of shock by population type . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Figure 5.5: Difference in incidence of shock by age and gender of household head . . . . . . . . . . . . . . . . 109 Figure 5.6: Difference in incidence of shock between poor and non-poor households . . . . . . . . . . . . . . 110 Figure 5.7: Negative effects of shocks on household welfare . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 Figure 5.8: Risk mitigation strategies in response to each shock . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 Figure 5.9: Adoption of risk mitigation mechanisms by welfare levels . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Figure 5.10: Adoption of risk mitigation mechanisms by location and head’s gender . . . . . . . . . . . . . . . 114 Figure 5.11: Reduced Coping Strategy Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 Figure 6.1: Incidence of remittance receipt and sending . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 Figure 6.2: Average annual value of remittance received and sent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123 Figure 6.3: Remittance-receiving households are in the top 60 percent consumption . . . . . . . . . . . . . . 124 Figure 6.4: Remittances more important for the bottom 40 percent . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Figure 6.5: How do international remittances impact consumption? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 Figure 6.6: Do international remittances impact enrollment? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128 Figure 6.7: Remittance cost as a proportion of sending US$200 to Somalia. . . . . . . . . . . . . . . . . . . . . . . 131 Figure A.1: Population pyramid . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Figure A.2: Poverty measures by gender of the household head . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.3: Poverty measures by remittances status of the household . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.4: Poverty measures by displacement status of the household . . . . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.5: Poverty measures by drought affected status of the household . . . . . . . . . . . . . . . . . . . . . . . 142 Figure A.6: Adult equivalent measure of poverty incidence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 Figure A.7: Age dependency ratio by quintile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Figure A.8: Households deprived in each dimension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147 Figure A.9: Households deprived in living standards dimension by group . . . . . . . . . . . . . . . . . . . . . . . . . 148 Figure A.10: Households deprived in educational dimension by group . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Figure A.11: Households deprived in water and sanitation dimension by group . . . . . . . . . . . . . . . . . . . . . 148 Figure C.1: Hunger in December 2017. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Figure C.2: Humanitarian Response 2017, beneficiaries targeted and reached . . . . . . . . . . . . . . . . . . . . . 155 Figure C.3: Outbreak of communicable diseases 2017, all regions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Figure C.4: Drought effect along the income distribution, urban areas. . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Contents ix List of Tables Table 1.1: Inequality decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Table 1.2: Average real consumption per capita (daily 2017 US$) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Table 1.3: Demographic attributes of poor households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Table 1.4: Factors associated with school enrollment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 Table 1.5: Multiple deprivations and demographic attributes of poor households . . . . . . . . . . . . . . . . . . . 34 Table 3.1: Drought impact on poverty and consumption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Table 4.1: Skills Profile Survey (SPS) 2017, Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75 Table 4.2: Age dependency ratios and household size by gender of household head . . . . . . . . . . . . . . . 77 Table 5.1: Incidence of types of shocks among poor and non-poor households . . . . . . . . . . . . . . . . . . . . 108 Table 5.2: What household characteristics affect the probability of reporting shocks? . . . . . . . . . . . . . . 111 Table 6.1: Selected economic indicators, 2015–2018 (percent of GDP) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 Table 6.2: Frequency remittances are received by households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 Table 6.3: Characteristics of remittance-recipient households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 Table 6.4: Counterfactual without remittances . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126 Table 6.5: Impact of international remittances on educational and health expenditure . . . . . . . . . . . . . . 127 Table 6.6: Housing conditions and remittances receipts among Somali households . . . . . . . . . . . . . . . . 128 Table 6.7: Remittances facilitate financial inclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 Table A.1: Accessibility rate of urban and rural areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 Table A.2: Demographic attributes of poor households by population group . . . . . . . . . . . . . . . . . . . . . . 144 Table A.3: Child poverty and key household characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 Table A.4: Poverty incidence and key household characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Table A.5: Poverty gap and key household characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Table A.6: Youth poverty and key household characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145 Table A.7: Hunger and key household characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 Table A.8: Education of the household head . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146 Table B.1: Urban non-settlement and settlement IDPs and have better access to services than rural IDPs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 149 Table B.2: Urban IDPs are consistently worse off in terms of access to services compared to other urban households . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150 Table C.1: List of control variables for difference-in-differences regression . . . . . . . . . . . . . . . . . . . . . . . . . 152 Table C.2: IPC Phase Classification Reference Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 Table C.3: Difference-in-differences results, consumption and poverty, full sample. . . . . . . . . . . . . . . . . . 156 Table C.4: Robustness of results across various specifications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158 Table C.5: Difference-in-differences results with restricted sample. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159 Table C.6: Difference-in-differences results, consumption and poverty, overlapping sample. . . . . . . . . . 159 Table C.7: Difference-in-differences results, hunger. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 Table C.8: Difference-in-differences results, food consumption. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 Table D.1: What household characteristics affect the probability of reporting shocks? . . . . . . . . . . . . . . 165 Table E.1: Reduced Coping Strategy Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167 Table F.1: Camps with Somali refugees in the SPS 2017 sampling frame . . . . . . . . . . . . . . . . . . . . . . . . . . . 169 Table F.2: Number of refugee and host community households interviewed by stratum . . . . . . . . . . . . . 170 Table F.3: Sampled population by country of nationality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 x  Somali Poverty and Vulnerability Assessment List of Boxes Box 1: The Somali Pulse website shares some of the world’s least represented voices . . . . . . . . . . . . . . xxx Box 2: Wave 1 and 2 of the Somali High Frequency Survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Box 3: Measures of poverty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Box 4: Poverty estimates from satellite images for inaccessible areas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Box 5: A remote monitoring system tracks migration patterns of nomads . . . . . . . . . . . . . . . . . . . . . . . . . 17 Box 6: Multiple deprivations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 Box 7: Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 Box 8: Intra-urban comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Box 9: The World Bank’s response to the drought . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 Box 10: Assessing the robustness of the difference-in-differences estimates . . . . . . . . . . . . . . . . . . . . . . 70 Box 11: Data on Somali refugees in Ethiopia comes from the Skills Profile Survey 2017 . . . . . . . . . . . . . . 75 Box 12: Where are the IDPs? Timing of survey sampling and interpretation of spatial results . . . . . . . . 76 Box 13: Drivers of displacement in Somali regions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 Box 14: What is vulnerability? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102 Box 15: Data caveats for vulnerability analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Box 16: Social protection systems in Kenya and Ethiopia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 Contents xi Abbreviations and Acronyms AE Adult equivalent AML/CFT Anti-Money Laundering and Combatting the Financing of Terrorism AWD Acute Watery Diarrhea CBS Central Bank of Somalia DFID Department for International Development Government of the United Kingdom DINA Drought Impact and Needs Assessment FAO Food and Agriculture Organization (UN) FEWSNET Famine Early Warning Systems Network FSNAU Food Security and Nutrition Analysis Unit GDP Gross domestic product HIPC Heavily Indebted Poor Countries ICRC International Committee of the Red Cross IDPs Internally displaced persons IPC Integrated Phase Classification for Food Insecurity KYC Know Your Client MODIS Moderate Resolution Imaging Spectroradiometer MTB Money Transfer Business MTO Money Transfer Operators NDVI Normalized Difference Vegetation Index OECD Organisation for Economic Co-operation and Development OLS Ordinary least squares PESS Population Estimation Survey of Somalia PPP Purchasing power parity RRF Recovery and Resilience Framework RSP Remittance Service Provider SDGs Sustainable Development Goals SHFS Somali High Frequency Survey UNOCHA United Nations Office for the Coordination of Humanitarian Affairs UNDP United Nations Development Programme UNHCR United Nations High Commissioner for Refugees UNICEF The United Nations Children’s Fund USGS United States Geological Survey WHO The World Health Organization  xiii Executive Summary Poverty and vulnerability in Somalia will impede The incidence of poverty of 69 percent is 19 per- future development without appropriate poli- centage points higher than the unweighted average cies. This report overviews poverty and vulnera- of low-income Sub-Saharan African countries of bility in Somalia to inform long-term development 51 percent in 2017. Somalia’s Gross Domestic Prod- and resilience policies and programs. The report uct (GDP) per capita of US$500 in 2017 and high describes poverty in Somalia in detail, including poverty incidence is in line with low income coun- geographical variation, based on the 2nd Somali tries, as shown by the relationship between poverty High Frequency Survey. The report analyzes the and GDP per capita across Sub-Saharan Africa. livelihoods impact of the recent drought, and esti- mates effects of future droughts, emphasizing effects on precarious livelihoods. The report also Poverty is widespread and deep, particularly discusses general shocks, including conflict and among rural residents, internally displaced climate, and the extent to which they have con- persons (IDPs) in settlements, and children tributed to displacement. Formal safety nets and informal remittances can support resilience. The Poverty is widespread and deep, particularly report discusses and recommends policies and for rural households and for IDP settlements. strategies to protect the poor and vulnerable while According to the 2nd Somali High Frequency Sur- opening paths to escape poverty. vey, almost three-fourths of the population in rural areas, IDP settlements, Mogadishu, and among Somalia is one of the poorest countries in Sub- nomads are poor. Poverty is deepest in rural areas Saharan Africa. Nearly 7 of 10 Somalis live in pov- and IDP settlements. To raise living standards, an erty, the 6th highest rate in the region, only after estimated US$1.64 billion per year is needed if the Democratic Republic of Congo, Central African perfectly targeted to the poor (ignoring admin- Republic, Madagascar, Burundi, and South Sudan. istrative and logistics costs). A significant group ■  Poverty is among the highest in Sub-Saharan Africa 100 95 90 Poverty incidence (% of population) Poverty incidence (% of population) 85 80 SSD 75 70 MWI SOM 60 65 50 55 RWA 40 45 TZA 30 UGA 35 20 ETH 25 10 0 15 0 500 1,000 1,500 2,000 2,500 3,000 SSD BDI MDG CAF COD SOM MWI GNB MOZ RWA LBR SLE BEN TGO MLI NER TZA TCD BFA UGA SEN GIN ETH ZWE COM GDP per capita (US$ PPP) Regional average Authors’ calculations based on the SHFS 2017–18, World Bank Macro Poverty Outlook and World Bank Open Data.  xv ■  Poverty is high and deep for households in rural areas and IDP settlements 100 50 Poverty gap (% of poverty line) 80 40 Poverty incidence (% of population) 60 30 40 20 20 10 0 0 Mogadishu Other urban Rural IDPs in settlements Nomads ur u se R n em al om s s ed H H itt es s Ps Ps af ted ed N ent ad ce er h ba ad H H ttl ur m c th is ct -ID ID ht ec re an an he d O gad fe e de ug ff on d itt ro t a al a ve m o N M he M D gh ei re ou e in ec d al R eive dr Ps m Fe ot ID c re N ot N Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. of non-poor are vulnerable to falling into poverty, Poverty extends beyond lack of money to representing that about 10 percent of the popu- nonmonetary deprivations across multiple lation is within 20 percent in terms of total daily dimensions consumption above the poverty line. In addition to monetary poverty, most Somali Children and households that do not receive households suffer other nonmonetary depri- remittances are disproportionately poor. Chil- vations. Almost 9 of 10 Somali households are dren below 14 years of age represent nearly half deprived in at least one dimension: monetary, elec- of Somalis, and 73 percent of them are poor. tricity, education, or water and sanitation. Nearly 7 Children from poor households face challenging of 10 households suffer in two or more dimensions. conditions—for example, they have no electric- Nomadic populations suffer the most, while urban ity or are deprived of education—which present dwellers experience the least. Poor households strong obstacles to escaping poverty. ■  IDPs in settlements, rural, and nomadic households face high deprivations across multiple dimensions 100 80 Percent of households 60 40 20 0 Overall Mogadishu Other Rural IDPs in Nomads urban settlements None One Two Three Four Five Authors’ calculations based on the SHFS 2017–18. xvi  Somali Poverty and Vulnerability Assessment are slightly more deprived than non-poor ones in for not attending among children aged 6–13, a gen- access to electricity and education. der gap emerges for 14–17 year-olds; male enroll- ment is significantly higher after controlling for age, Access to services is limited, particularly for poverty status, and other household characteristics. rural residents, IDPs in settlements, and nomads. The main reasons for not attending school at this Improved water and sanitation is critical for health, age are lack of money for boys and having to work school performance, and productivity, but only 5 of or help at home for girls. Nomads and girls face the 10 households have access to improved sanitation, biggest challenges. and 8 of 10 to improved water sources. Only 5 of 10 households have electricity. Access to services is Distance from schools, rather than cost, is the somewhat higher in urban areas. Poor households primary barrier to primary school enrollment. are less likely to have access to improved sanita- Schools are at least 30 minutes walking distance tion and electricity. Markets and health clinics are for one of three Somali households, a distance far—more than 30 minutes walking distance—for negatively associated with primary enrollment. 34 to 40 percent of Somali households and for On average, households spend about 3 percent of most nomads. the US$1.90 poverty line on education per house- hold member enrolled. Expenditure on education Overage school enrollment is common, with stark weakly correlates with enrollment, however. geographical and gender disparities in enrollment rates. Nearly 27 percent of children enrolled in pri- Gender and regional disparities in access to edu- mary school are older than 13 years, and more than cation reflect educational outcomes of Somalis. 55 percent of those enrolled in secondary school are Education is key for increasing welfare and break- not between the ages of 14–17 years. Somali children ing the poverty cycle. Only half of Somalis read start primary school late as parents believe children and write, with literacy more common among aged 6–9 are too young to attend. Enrollment of younger generations, urban populations, and men. children aged 6–13 is only 33 percent and highest Similarly, the share of rural residents, IDPs in set- in urban areas. In Mogadishu and other urban areas, tlements, and nomads without formal education is enrollment among primary school-aged children 1.6, 2.6, and 2.5 times, respectively, higher than that aged 6–13 is about twice the enrollment rate in rural of urban residents. Older Somalis are less likely to areas and IDP settlements, and more than six times have formal education than younger Somalis, and that of nomadic children. Geographical disparity in women are less likely than men. Furthermore, chil- enrollment for secondary school-aged 14–17 year- dren are more likely to be enrolled in school when old children is also pronounced. While there are no household heads are literate. Despite higher enroll- gender differences in enrollment rates and reasons ment rates in urban areas, completion of primary ■  IDPs in settlements, rural, and nomadic households lag in access to key services Access to improved sanitation 100 100 Access to electricity (% of households) (% of households) 80 80 60 60 40 40 20 20 0 0 hu n al ts s or r hu n al ts s or r oo oo ad ad ba ba ur en ur en Po Po is is -p -p om om ur R ur R ad ad em em on on er er og og N N ttl ttl N N th th M se M se O O in in Ps Ps ID ID Overall average Overall average Authors’ calculations based on the SHFS 2017–18. Executive Summary xvii ■  Women across all population groups have lower literacy and educational attainment 100 100 Percent of population without 90 population aged 15+) Literacy rate (% of 80 80 formal education 60 70 60 40 50 40 20 30 0 20 Mogadishu Other Rural IDPs in Nomads s s s s s s s s + s –1 ar –2 ar –2 ar –3 ar –3 ar –4 ar –4 ar –5 ar ar 55 15 ye 20 ye 25 ye 30 ye 35 ye 40 ye 45 ye 50 ye ye urban settlements 4 9 4 9 4 9 4 9 4 –1 10 Women Men Overall average Women Men Overall average Authors’ calculations based on the SHFS 2017–18. education is limited—only 11 percent of those aged Mogadishu) compared to 69 percent nationally, 15 or more who were previously enrolled did not 72 percent in rural areas, and 76 percent among complete the primary school. IDPs. The only exception is Mogadishu, where pov- erty is higher than nationally or than rural areas. Cities consistently provide better access to ser- Urban areas generally provide higher standards vices—except for land and housing—and more of living and better access to services than stable income than rural areas. Access to electric- rural areas, except for access to land and ity, water, improved sanitation, health, education, housing improved housing, and Internet is consistently higher in urban areas irrespective of poverty levels, Somali cities tend to have lower monetary pov- whether IDP or female-headed households. Rural erty and better services than rural areas. Poverty areas fare better than urban in land and housing averages 64 percent across urban areas (including tenure: due to land scarcity and high land values, ■  Urban areas provide better services than rural areas 100 Access to water (% of households) 100 Access to electricity (% of households) 80 80 60 60 40 40 20 20 0 ur u N ban ur l C NW an C l ur l tra n ur l bb SW ban d ral n om s s N ishu N an l C W n lu l en ban ur l bb SW n d al n Ps W ra a a N rura tra ura a P ad N ish a ba ba ba ba tra ur SW l rur SW rur an rur N ru b en b an ru b ID ID r ur ur ur r r ur ad ad E E l tra E E W og og N N M M en en al al C Ju Ju Overall average Overall average Authors’ calculations based on the SHFS 2017–18. xviii  Somali Poverty and Vulnerability Assessment urban households are less likely to own. Somali cit- areas, which facilitates entry of external assistance. ies provide more wage employment and access to Subnational governments are nascent in Kismayo, remittances, and since urban jobs are less climate- Baidoa, and Central urban areas, which have only dependent, they provide more stable income than recently liberated from Al-Shabaab. Much of their agriculture or family businesses. rural territories remain under Al-Shabaab control. Despite better conditions in cities, cities struggle with hunger, high absolute poverty of 64 percent, Urban IDPs have more access to services than nonmonetary poverty of 41 percent, and ensur- rural IDPs, but lag behind non-IDP households ing universal access to services. Many cities have not coped with constant and large influxes of IDPs. Urban IDPs, though worse off than urban non-IDP Pressure on land, housing, and services is increas- households, fare better than rural IDP households. ing with 75 percent of IDPs already residing in Irrespective of whether IDPs live in settlements or cities. not, they have better access to electricity, improved housing, and improved sanitation than rural IDPs. Mogadishu and North East and North West cit- However, urban IDPs still have less access to elec- ies provide better access to services compared to tricity, piped water, improved sanitation, improved Baidoa, Kismayo, and Central urban areas. While housing, dwelling ownership, and Internet com- poverty is higher in Mogadishu than all urban pared to non-IDP urban households. Moreover, areas except Baidoa, access to basic services such urban IDPs suffer lower enrollment, literacy, and as electricity, water, sanitation, improved hous- employment rates, and tend to live further from ing, education, and health is higher in Mogadishu. primary schools and food markets. Many urban Kismayo has the lowest poverty yet fares poorly IDPs, deprived of their former livelihoods, assets, on services. Strikingly, access to water, literacy, and social networks, are disadvantaged in educa- enrollment, and employment are significantly bet- tion and access to good jobs. ter in IDP settlements than in Kismayo. Baidoa has high poverty, and correspondingly low access to services. North East and North West cities fare Urban households in IDP host communities are relatively well on access to services, while Central no worse off than other urban households urban areas lag. North East and North West cities, which have been relatively free of violent conflict, There are no significant differences between have relatively high access to services; 86 percent urban households in communities that host IDPs of NW urban residents report feeling “very safe.” (urban host) and those in communities that Public institutions are more established in these do not (urban non-host). While hosting IDPs is ■  Significant regional inter-urban variation exists in access to services 100 Access to services (% of households) 80 60 40 20 0 Mogadishu Kismayo Baidoa NE urban NW urban IDPs Central Cities Other urban areas Electricity Water Improved housing Improved sanitation Authors’ calculations based on the SHFS 2017–18. Executive Summary xix thought to constrain access to services, jobs, and household-level shocks such as injury, death, or housing, survey data show that urban host and unemployment. Shocks contribute to extreme non-host households have similar poverty profiles poverty and vulnerability, constraining economic and access to services. This suggests that either opportunities and livelihoods, damaging assets, effects of hosting IDPs have not yet materialized, and limiting access to farms, fishing, and pastoral- or that hosting IDPs does not deteriorate service ist routes. The persistent cycle of shocks increases access, as services are provided to IDPs dwelling in Somalis’ vulnerability to future shocks as there is settlements. This situation may change if IDPs pro- limited public and private insurance. long their urban stay and/or support from humani- tarian agencies declines. About 66 percent of Somali households report experiencing at least one type of shock in the Continued influx of IDPs causes urban sprawl, past 12 months. Due to the 2017 drought, most hindering service provision in new settlements. reported shocks are related to fluctuation in cli- Seventy-five percent of IDPs in Somalia have set- mate and its impact on livelihoods and the econ- tled on public and private lands in and around cities. omy. In an agro-pastoralist economy, household Most returnees are thought to also have settled in welfare is closely linked with changes in weather. cities. Without secure land tenure, IDPs are vulner- Poorer households are more likely to experience able to eviction. Over 109,000 IDPs living in infor- more than one type of shock. The impact of shocks mal settlements across the country were forcefully is magnified when a household experiences mul- evicted between January and August 2017 alone; tiple shocks simultaneously. 77 percent were around Mogadishu. Many IDPs shift to city outskirts, causing urban sprawl and making Low education, agricultural dependence, unem- service provision difficult and costly as new settle- ployment, low wealth, and large household size ments are disconnected from urban infrastructure make households more vulnerable to shocks. networks. Spatial fragmentation also inhibits IDPs’ Household characteristics affect shock impacts. access to jobs and prevents cities from reaping Households headed by an illiterate person are scale and agglomeration benefits. 12 to 24 percent more likely to report experienc- ing drought and loss of crops and livestock than households headed by someone with some educa- Many Somalis are vulnerable and unable to tion. Households depending on agriculture for their protect resources against shocks main source of income are more likely to report water shortages and loss of crops and livestock, Somali households are vulnerable to shocks such but they are less likely to report high food prices. as natural disasters and epidemics, as well as to Households receiving humanitarian aid were more ■  Drought is the most reported shock among Somali households 80 60 households Percent of 40 20 0 Drought or irregular rain Loss of crop or livestock Water shortage for cattle or farming High food prices Reduction in income Theft Conflict Other natural By type of shock Any shock Authors’ calculations based on the SHFS 2017–18. xx  Somali Poverty and Vulnerability Assessment likely to have reported experiencing a shock, The recent drought exacerbated vulnerability implying that humanitarian aid was well-targeted. and threatened millions of Somali lives Almost all Somali households that experience Somalia’s severe drought triggered a humani- shock report a negative impact on income, assets, tarian crisis as half of Somalis faced acute food or food resources. Households experiencing theft insecurity in 2017. Four consecutive seasons of or conflict report loss of assets such as valuables, poor rains between April 2016 and December 2017 land, or livestock. Most Somalis rely on livestock caused severe drought across the country, exac- and farming for their livelihood, so loss of crops erbating food insecurity for 6.2 million Somalis. or livestock and water shortage reduce household About 2.4 million people needed humanitarian income. Similarly, high food prices decrease pur- assistance to avert loss of livelihoods and reduce chasing power and real income of households. acute malnutrition, and 866,000 people required ■  Food insecurity rose with each successive season of poor rains Wave 1 55 120 Wave 2 (% of population) Food insecurity 50 (% of average) Rainfall/NDVI 100 2016 Gu season 45 80 2016 Deyr season 40 2017 Gu season 60 35 2017 Deyr season 40 30 Rainfall anomaly Feb-17 Feb-18 Dec-17 Oct-17 Dec-16 Oct-16 Feb-16 Apr-17 Jun-17 Aug-17 Aug-16 Jun-16 Apr-16 NDVI anomaly Acute food insecurity Source: FEWSNET, WFP-VAM, and authors’ calculations based on the SHFS 2017–18. ■  Drought-related displacement peaked in mid-2017 350,000 1.2 Monthly drought displacement 300,000 1.0 Cumulative (millions) 250,000 0.8 200,000 0.6 150,000 0.4 100,000 50,000 0.2 0 0 6 6 6 17 17 7 7 7 17 7 17 17 7 7 7 -1 -1 -1 -1 r-1 -1 l-1 -1 -1 -1 n- b- n- g- p- ct ov ec ar ay ct ov ec Ju Ap Ja Fe Ju Au Se O O M N D M N D Displacement due to drought Cumulative drought displacements Source: UNHCR (2018a). Executive Summary xxi urgent food assistance to avert famine. Slightly Drought increased the likelihood of being improved rains in late 2017 to early 2018 eased poor and hungry for the most vulnerable rural drought conditions, but food insecurity remains a households serious concern. Highly drought-exposed rural households are The drought exacerbated vulnerabilities, threat- 24 percent more likely to be poor and more likely ened livelihoods, and displaced almost 1 million to be hungry. In rural areas, higher drought expo- Somalis. Lack of water and pasture led to high live- sure decreased consumption by 19 percent, corre- stock deaths and low birth rates, reducing herds sponding to a 24 percent increase in probability by 25 to 75 percent the first half of 2017. Somalis of being poor. The drought impacted relatively were forced to deplete household assets and food wealthier rural households most: while higher stocks to cope with rising food and water prices drought exposure had no significant impact on as weak demand for agricultural labor reduced consumption for the poorest 10 percent of rural wages. Drought reduced water for hygiene and households, exposure reduced consumption by sanitation and increased water contamination. 17 percent for rural households at the twentieth With drought threatening livelihoods, households percentile, and between 20 and 30 percent for the were forced to leave in search of government and top 80 percent of rural households. international assistance. The 2016 to 2017 drought displaced about one million of today’s Somalis. ■  Drought effect on consumption along the income distribution, rural areas 20% Drought impact on consumption 0% –20% –40% –60% 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Income percentile Drought effect Smoothed drought effect 95% confidence interval Source: Authors’ calculations based on the SHFS 2017–18. ■  Drought has been the major cause of internal displacement in recent years 350 300 Number of people 250 (thousands) 200 150 100 50 0 Jan-16 Feb-16 Mar-16 Apr-16 May-16 Jun-16 Jul-16 Aug-16 Sep-16 Oct-16 Nov-16 Dec-16 Jan-17 Feb-17 Mar-17 Apr-17 May-17 Jun-17 Jul-17 Aug-17 Sep-17 Oct-17 Nov-17 Dec-17 Jan-18 Feb-18 Mar-18 Apr-18 Conflict Drought Source: UNHCR-PRMN, Jan 2016–Apr 2018. xxii  Somali Poverty and Vulnerability Assessment As hunger rose across all Somali regions, rural IDPs remain among the most vulnerable households in highly drought-exposed areas were groups, thus improving rural and urban access most severely affected. Higher drought exposure to services and livelihoods can strengthen their led to a 16 percent decrease in food consumption, viability and support voluntary return or local accompanied by a 17 percent increase in the prob- integration ability of experiencing hunger in December 2017. The drought had no significant effect on poverty IDPs face greater poverty and worse living con- and hunger for urban households. ditions than other residents. Although about 70 percent of Somalis are poor, IDPs are especially Internal displacement has grown rapidly in marginalized: over 3 in 4 IDPs live on less than $1.90 per day, and more than half of IDP households recent years, mainly due to drought face hunger. IDPs largely share essential amenities such as toilets, thereby crowding water, sanitation, Internal displacement has grown rapidly in recent and hygiene (WASH) facilities in settlements. IDP years, mainly due to drought. Four consecutive settlements are also further from essential facili- poor rainy seasons, along with ongoing conflict ties such as schools, health centers, and markets. and violence from armed non-state actors, caused Expanding access to basic services, including displacements to surge from late 2016 to late 2017. health and education, is essential to improve resil- Over half of IDPs are under the age of 15 and less ience in IDP communities. On average, IDP house- than 1 percent are above 64, driving high depen- holds receive about half the remittances of urban dency ratios: IDP households average dependency households. ratios larger than one, indicating that each working- age member provides for at least one child. IDPs also have lower human capital, leading to Poverty-alleviation policies and strategies for lifelong welfare gaps. School-age IDPs are less Somalia must address displacement-related vul- likely to attend school than urban residents. Adult nerabilities and IDPs’ needs. IDPs are less likely than urban residents to read ■  IDPs have greater poverty incidence than residents ■  Only 1 in 3 school-aged IDPs is enrolled in school 6–17 100 80 6–17 100 80 aged aged 80 population 80 60 population 60 population population 60 60 40 40 of 40 of 40 Percent Percent of 20 of 20 20 Percent 20 Percent 0 0 0 0 IDP 2016 host non-host resident resident resident IDP IDP violence event protracted Protracted once multiple headed headed IDP host non-host resident resident IDP IDP violence event headed headed protracted Protracted once multiple 40 60 Poor Non-poor IDP 2016 host non-host resident resident resident IDP IDP violence event protracted Protracted once multiple headed headed IDP host non-host resident resident IDP IDP violence event headed headed protracted Protracted once multiple 40 60 Poor Non-poor Bottom Top Bottom Top Overall Non-settlement Settlement Overall Camp Non-camp Urban Urban Overall Non-settlement Settlement Displaced Overall Camp Non-camp Displaced IDP Urban Urban Climate Climate Displaced Displaced IDP Climate Climate Woman Man Woman Man Displaced Displaced Urban Rural National Urban Rural or or Woman Man Woman Man Urban Urban Displaced Displaced Urban Rural National Urban Rural Settlement or or Urban Urban Not Not Settlement Conflict Conflict Not Not Conflict Conflict Overall IDP Overall IDP Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. Executive Summary xxiii ■  Crowding of toilets squeezes out access to improved sanitation, especially in IDP settlements and urban centers 100 80 Percent of households 60 40 20 0 Overall IDP Urban host Urban non-host Urban resident Rural resident National resident Non-settlement IDP Settlement IDP Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Bottom 40 Top 60 Poor Non-poor Overall IDP Unadjusted for sharing Adjusted for sharing Source: Authors’ calculations based on the SHFS 2017–18. ■  Urban livelihoods today are different from IDPs’ pre-displacement livelihoods 100 Percent of households 80 60 40 20 0 Urban host Urban non-host Urban resident Rural resident Before Current Before Current Before Current Before Current Before Current Non-displaced Overall IDP Camp IDP Non-camp IDP Conflict IDP Climate IDP (current) Salaried labor Remittances Small family business Agriculture Trade, property income Aid or zakat Other Source: Authors’ calculations based on the SHFS 2017–18. xxiv  Somali Poverty and Vulnerability Assessment ■  Most IDPs do not intend to return 100 Percent of households 80 60 40 20 0 IDP Refugee Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Overall IDP Don’t want to move Original place of residence New area Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. ■  Most IDPs arrived in the current location for security, regardless of the background to displacement 100 80 Percent of households 60 40 20 0 P m DP P ce t he d ed d ed m ce le To 0 60 N oor r en oo e Pr cte 4 ID D tip lim len ed on M ead ad ct tle nt I tI ev m p P -p ul ra ra ll en tto on ra o ac d e -s me h ot ot vi pl ce ve at Bo an an pr or D pla on ttle O om ot ct et C is Se N fli W D on is C N Better security Water for livestock Home/land access Education/health access Source: Authors’ calculation based on the SHFS 2017–18. and write. Educational outcomes for IDPs are Urban IDP livelihoods differ significantly from pre- closer to rural outcomes even though three in four displacement livelihoods. IDP livelihoods before IDP households are in urban areas. Gaps in educa- displacement consisted of a mix of salaries, small tional attainment are crucial since half of Somalis businesses, and agriculture, whereas IDP urban liveli- are less than age 15. As the young mature, lags in hoods today depend largely on salaries, remittances, educational attainment for IDPs will lead to persis- and aid. Many IDPs are now employed in helping tent, lifelong gaps in education, employment, and with nonagricultural businesses as they adjust to the overall well-being. employment landscapes of new locations. Executive Summary xxv Displacement has a very negative effect on well- an absence of formal and informal safety nets. being. IDPs displaced by climate events are poorer Household reliance on self-insurance, or choosing and have worse housing quality than those dis- to do nothing, in case of conflict or theft implies a placed by conflict. IDPs experiencing protracted lack of access to formal conflict resolution mech- displacement—mostly in urban areas—have bet- anisms and regulatory frameworks. A negligible ter access to health care. IDP households headed share of households has access to formal or mar- by a woman get only one-sixth the remittances of ket mechanisms. This adds to the vulnerability of IDP households headed by a man. Wealthier IDPs households, especially those in marginalized com- are more confident than poorer IDPs of being relo- munities. Wealthier households also lack access to cated within the next year. formal safety nets, which makes them vulnerable to shocks, albeit less than poorer counterparts. Most IDPs prefer to stay in their current loca- tion and only a few have revisited their original Social safety nets and social protection sys- residence. About 7 in 10 IDPs want to remain in tems are needed to build risk management and their current location, and only 2 in 10 intend to risk coping capacity of vulnerable households. return to their original place of residence. Over 9 A social safety nets system includes both income in 10 have not visited their original residence since and consumption smoothing to build resilience and being displaced. A majority of IDPs cited security enable households to anticipate and/or recover as the reason for preferring their current location, from shocks. A cash transfer can help reduce pov- with 8 in 10 IDPs reporting feeling “safe” or “very erty. Globally, countries tend to spend between 2.5 safe” currently. IDPs also perceive positive social and 5 percent of GDP on such programs. In con- relations with host communities, with 9 in 10 IDP trast, Sub-Saharan countries on average spend only households saying that they have good dealings 1.6 percent of GDP on social safety nets. Somalia with their surrounding communities. spends even less at 0.8 percent of GDP in 2016, even though it receives 16 percent of GDP (US$1.2 bil- lion) in humanitarian aid. Using some resources to In the absence of formal safety nets, self- implement a well-targeted safety net would reduce insurance is a primary coping strategy for many poverty. Households receiving cash transfers could Somali households use them for productive investments, savings, and other income-generating activities. Somalis are vulnerable to various covariate and idiosyncratic shocks, which contribute to pov- erty, vulnerability, and displacement. Almost two International remittances represent a sizable in three Somali households (66 percent) reported share of household consumption, especially for experiencing at least one type of shock in the past the bottom 40 percent 12 months. Of those who experienced a shock, half of households reported experiencing drought and Remittances are the major source of external one in four reported loss of crops or livestock and development finance for Somalia. Somali migrants shortage of water for farming or cattle. One in five and refugees outside Somalia doubled between households experienced high food prices. Two of 1990 and 2017 to total more than 2 million. Dur- five Somali households experienced multiple types ing 2015–2017, Somali diaspora sent home about of shocks within a year. The negative impact of an official US$1.3 billion per year, but remittances each shock is greater if a household experiences may be significantly larger when considering unre- multiple types simultaneously. Poorer households corded flows. Remittances represent 20 percent of are more likely to experience more than one type GDP, about the same amount as grants to Somalia, of shock. Somali households that have experi- and more than three times foreign direct invest- enced a shock report higher food insecurity, lower ment (FDI). Remittances may be countercyclical, wealth, fewer savings, and lower access to coping as relatives and friends often send more during mechanisms; they are also more likely to resort to economic downturns, disasters, conflicts, or other negative coping strategies. negative shocks. Households mostly rely on self-insurance to Households receiving international remittances cope with shocks. This indicates inadequate risk are less likely to be poor. Only 58 percent of management and mitigation systems, as well as remittance-recipient households in Somalia are xxvi  Somali Poverty and Vulnerability Assessment ■  Incidence of remittance receipt and sending 30 Percent of households 22 20 18 16 13 11 10 11 10 8 8 7 6 6 5 7 6 5 4 3 2 0 0 Urban Rural IDPs living in IDPs living outside Nomads settlements of settlements Remittances overall International remittances Internal remittances Sending remittances Overall average Source: Authors’ calculations based on the SHFS 2017–18. ■  Average annual value of remittances received are almost twice those sent $1,000 $800 $600 $400 $200 $0 Overall Urban Rural IDPs inside IDPs outside Nomads settlements settlements Remittances overall International remittances Internal remittances Sending remittances Source: Authors’ calculations based on the SHFS 2017–18. ■  International remittances are more important for the bottom 40 percent 2.0 1.5 US$ 1.0 0.5 0 Bottom 40% Top 60% International remittances US$ pc pd Consumption expenditure US$ pc pd Source: Authors’ calculations based on the SHFS 2017–18. Executive Summary xxvii poor, compared to 71 percent of non-recipient Alleviating poverty and mitigating vulnerability households. Somali households are both remit- in Somalia require accelerating economic tance senders and receivers, but the incidence of growth, improving services, managing receipts tends to be higher for urban households, urbanization, and investing in resilience and while IDPs living in settlements are least likely safety nets, including cost-effective remittance to receive international remittances. Despite the transfers three-fold gap in incidence, those in IDP settle- ments who do receive international remittances receive almost the same amount as urban recipi- Economic growth-creating opportunities, espe- ents. On average, recipients of international remit- cially for youth, is fundamental to sustainable tances receive about US$743 per year—above the poverty reduction, vulnerability mitigation, and 2017 average per capita Somali income of US$535. conflict avoidance. Somalia has a large youth Internal transfers are also important for both urban bulge, so youth must be able to find jobs to contrib- and rural dwellers, as well as for IDPs living outside ute to economic growth. The need for sustainable settlements. work for IDPs is especially urgent given chang- ing livelihood structures and lack of safety nets. Households receiving international remittances Policies to encourage business and entrepreneur- have higher incomes, consumption, and expendi- ship to create jobs are needed to avoid idle youth ture on education. International remittances aver- from resorting to conflict. Furthermore, enhancing age 34 percent of total household income, nearly access to domestic markets can increase inclusiv- as high a proportion of income from salaried labor ity, spur economic activity, and accelerate poverty at 35 percent for households that receive them. reduction. Domestic remittances also comprise 23 percent of total income for households that receive them. Improving service provision—especially Remittances are relatively more important for the education—is crucial to improve human capital bottom 40 percent as income from remittances rep- and reduce inequality that disproportionately resent 54 percent of their total consumption, while affects girls/women, IDPs, and rural and nomadic remittances represent about 23 percent of total households. Policies should aim to improve access consumption for the upper 60 percent. International to education and increase enrollment while consid- remittance-receiving households are more likely to ering disparities and specific needs of vulnerable have higher expenditures on education compared to groups. Increasing access to education for children non-recipient households. Households that receive and youth will allow more productive opportunities international remittances also have substantially later in life and enhance standard of living. Build- higher enrollment rates than non-recipients. ing more schools is one alternative, but further analysis is needed given the complexity and cost Remittance markets in Somalia remain relatively of designing and implementing educational poli- underdeveloped and costly but can leap-frog cies. The challenge of increasing enrollment rates with mobile technologies. Forty-six percent of will continue to grow given Somalia’s demograph- domestic remittances go through mobile money, ics and young population. While access is still a big while 47 percent go through money transfer challenge, and a crucial first step, policies to reduce operators and informal channels, such as hand- drop-out and increase levels of educational attain- carried during visits home and Hawala. Mean- ment are also needed. while, 87 percent of international remittances are channeled through money transfer operators, and Somalia cities need investment in land manage- 12 percent via mobile phones. Due to anti-money ment and coordinated infrastructure. Cities mostly laundering regulations, costs of remitting money need proper land administration systems and land to Somalia have increased, while the number of use planning to control growth and provide secure service providers has declined, reducing competi- tenure to IDPs. Coordinated infrastructure invest- tion and encouraging informal channels. So, while ments aligned with planning would create synergy remittances provide a lifeline for the poor, send- across different types of infrastructure. City invest- ing money to Somalia is costly: from the United ments need to be customized to address each city’s Kingdom to Somalia, costs are more than twice the needs. Detailed city-level assessments are needed SDG target of 3 percent, and for sending from Aus- to understand urbanization constraints and solu- tralia costs are almost three times the SDG target. tions, which consider IDP needs to facilitate their xxviii  Somali Poverty and Vulnerability Assessment ■  Simulation of income shock among rural households 100 90 Percent of population 80 76 70 65 60 50 40 30 20 Poverty change 10 Core consumption (2017) 0 Income shock 0 1 2 3 4 Daily core consumption expenditure per capita (US$) Source: Authors’ calculations based on the SHFS 2017–18. integration. Political economy must be considered agricultural insurance, enabling households to in crafting and implementing policies to foresee diversify income, and improving access to roads opportunities, risks, winners, and losers of policies, and clean water. and anticipate challenges to implementation. It is critical to strengthen government institutions by Cash transfers can help build resilience, espe- channeling development assistance through them cially for poor households with limited access rather than parallel structures. State and municipal to formal and informal insurance. Protecting governments, ultimately accountable for providing vulnerable groups and creating income oppor- services to constituents, can participate more. tunities are crucial to prevent childhood poverty from progressing into adulthood. Poor households Within cities, the needs of IDPs not living in set- most vulnerable to shocks experience the highest tlements should be addressed along with the IDPs welfare impact. High vulnerability tends to make in settlements. Area-based targeting can ensure them risk averse, hence having access to insurance equity among vulnerable urban population groups. and other risk mitigation can help poor household Assistance has focused on urban IDPs living in set- invest with less fear. On average across countries, tlements deemed most deprived, but urban IDP household consumption can increase by US$0.74 not living in settlements are equally deprived of for each dollar transferred. In resource-constrained services. Moreover, they consistently fare worse on environments such as Somalia, short- to medium- development outcomes compared to other urban term humanitarian assistance might be needed to households. Because non-settlement IDPs are dif- complement social safety nets. ficult to track, it is important to use area-based interventions on poor urban areas with high con- Remittances, crucial to resilience and investment centrations of non-settlement IDPs. Group-based in Somalia, would benefit from policies facili- approaches only focus on IDPs in settlements. In tating their flow. Mobile licensing and increasing pursuing poor area-based approaches, develop- competition will decrease costs, as will the intro- ment must align with urban development plans. duction of new products, interoperability among service providers, and establishment of open infra- Investment in resilience is needed to prevent structure to collect digital payments. A barrier to livelihood loss for vulnerable rural households, lowering remittance fees arises from anti-money especially due to likely future droughts. A con- laundering and combatting financing of terror sumption shock of the same magnitude as the requirements. Somalia is working on complying 2016/17 drought is estimated to increase rural pov- with AML/CFT and establishing digital identifica- erty from 65 to 76 percent. Investing in resilience tion to “de-risk” for international banks. Remitters and rural market access would help these house- could benefit from new financial products such as holds avoid livelihood loss. Measures may include micro-insurance and direct payments of tuition. Executive Summary xxix Box 1 ■  The Somali Pulse website shares some of the world’s least represented voices A poverty incidence of 69 percent summarizes the country’s poverty level, yet it does not reveal the daily struggle of the population. Somalia has suffered from armed conflict and several humanitarian crises. The recent drought severely affected the lives of millions and exacerbated existing vulnerabilities. Securing livelihoods has become more and more difficult with 69 percent of the Somali population now living in poverty. Poverty esti- mates are useful for comparisons and analyses to inform policies and programs. However, an abstract poverty number does not depict the suffering that people go through to make ends meet. Wave 2 of the Somali High Frequency Survey (SHFS) used hand-held devices to collect data. At the end of the quantitative survey, respon- dents were asked to voluntarily record a quick message. The Somali Pulse website contains hundreds of video testimonials recorded with tablets during fieldwork to capture the voice of the people and give a face to the data. The website presents insights from the World Bank’s SHFS, as well as video testimonials—with subtitles in English— reflecting the dire situation on the ground and what it is like to live in poverty in Somalia. The videos depict the sense of powerlessness, the pain of hunger, the stress of hopelessness, and the feelings of disappointment that express Somalis’ experiences. The opportu- nity to voice the struggle of respondents is a first step to empowerment of the world’s least represented voices, allowing them to tell the world of what their lives are like. It is also an inspiration to continue finding innovative ways for helping them and millions like them to escape poverty. The Somali Pulse website can be found in the following link: http://www.thesomalipulse.com xxx  Somali Poverty and Vulnerability Assessment Introduction Somalia is on the path to political and security addition to environment shocks, while its reliance stabilization after more than two decades of on imports for food and fuel leaves the country civil war and conflict. Since the disintegration of and its population at risk of spikes in import prices. the central authority in 1991, the remaining power With the Somali economy largely dependent on vacuum was filled by warring local factions. The climate-sensitive activities such as agriculture and country suffered from armed conflict and several minerals, negative climatic events quickly disrupt humanitarian crises linked to the conflict, as well as these sectors, as well as the livelihoods they sup- to drought and deprivation. In the 1990s, Somalia port, and easily translate into humanitarian crises. witnessed the emergence of regional administra- Such shocks often divert attention from long-term tions. Somaliland self-declared independence in institutional strengthening to averting humanitar- 1991 and Puntland declared itself a semiautono- ian crises. Real GDP growth fell to 1.8 percent in mous region in 1998. An interim central state, the 2017 from 2.4 percent in 2016 due largely to severe Transitional Federal Government, was established droughts.3 The impacts extended beyond envi- in 2004 to bring stability, but political instability ronmental and economic impacts to having deep continued to plague the southern regions. Follow- health and social impacts involving food security, ing the end of the interim mandate of the Tran- nutrition, and public safety. sitional Federal Government, Somalia completed its political transition with the establishment of Widespread poverty and food insecurity is a the Federal Government of Somalia in 2012. Since recurring developmental issue. Most of the popu- then, the period has been relatively more stable. Al lation remains poor and is vulnerable to a range of Shabab’s territorial footprint has narrowed, espe- shocks, including repeated cycles of devastating cially in the urban areas of southern Somalia, which droughts such as the one in 2017. Following at least are now the capitals of the newly-formed Federal three successive seasons of below normal rainfall Members States. After completing the first national in most areas, the ensuing drought triggered a electoral process in decades, a new government humanitarian crisis that left more than 5.4 million was ushered in in 2017 opening an opportunity for Somalis (almost half of the population) in need of longer term stability and sustainable development. assistance, mostly in rural areas and IDP settle- ments.4 While a famine was averted in 2017, there Opportunities to ensure a development trajectory are 1 million children projected to be malnourished face many challenges since the country remains a and an additional 1 million displaced, resulting in fragile state subject to multiple shocks. The coun- total displacement of 17 percent.5 A confluence try remains extremely fragile due to conflict.1 Insur- of factors, including conflict and insecurity, natu- gency, although more restrained in recent years, ral disasters, limited safety nets, and high levels of remains a threat to the political progress. Limited unemployment, are among many that contribute government resources and capacity, asymmetric to poverty, food insecurity, and vulnerability. federal structures, and a fragile security situation limit the government’s ability to govern effec- Displacement is a key feature of modern Somali tively.2 Somalia has a highly concentrated export history linked to multiple drivers, including recur- base dependent on primary commodities (live- rent exposure to internal conflict and environ- stock), leaving it vulnerable to market dynamics in mental hazards. More than 1.1 million Somalis live in 1  See the Fund for Peace (2018) “Fragile States Index 2018;” 3  World Bank (2018b). Institute for Economics and Peace (2017a) “Global Peace 4  UNOCHA (United Nations Office for the Coordination of Index 2017;” Institute for Economics and Peace (2017b) “Global Humanitarian Affairs) (2018b). Terrorism Index 2017.” 5  UNOCHA (United Nations Office for the Coordination of 2  World Bank (2018d). Humanitarian Affairs) (2018b). Introduction 1 FIGURE 0.1  n  Coverage of household surveys in Somali regions 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. internal displacement and 900,000 are refugees in 22,000 km of road network remains concentrated the near region.6 Recurrent and persistent shocks in urban areas, posing significant connectivity have threatened personal safety and severely con- challenges with other areas, especially since much strained livelihoods and food security, thus playing of the internal transportation is by vehicle. a significant part in past and current displacement. Most internally displaced persons move to urban Remittances are central to the Somali economy areas for protracted periods, settling on public and provide a lifeline to some segments of the and private lands in the outskirts of cities. In the population but not the most vulnerable. The absence of security of land tenures, the risk of Somali economy receives an estimated US$1.3 bil- forced eviction is high, exacerbating existing vul- lion in remittances a year, equivalent to 20 percent nerabilities among IDPs associated with the loss of of GDP.8 The inflow of remittances outweighs for- assets, livelihoods, and social structures. eign direct investment providing resources to the national economy. Somalia has maintained a trade Somalia is urbanizing rapidly due to large-scale deficit for many years, but ample remittances and forced displacement and economic migration grants have been able to partially offset this defi- that have driven large numbers of Somalis toward cit. The economy has also been able to weather the urban areas. This accelerated pace of urban- drought and terrorist attacks in 2017 thanks in part ization, estimated at 4 percent, is placing a strain to the remittance inflows. on the existing physical and social infrastructure.7 Despite these challenges, urban areas fare better The World Bank implemented the second wave of than the rest of the country in terms of access to the Somali High Frequency Survey in 2017/18. The basic services, public infrastructure, and devel- survey achieved greater geographical and popula- opment outcomes. Accessibility of the country’s tion coverage compared to Wave 1 of the Somali High Frequency Survey (SHFS) conducted in 2016 and the Somaliland Household Survey (SLHS) car- ried out in 2013. The SLHS and Wave 1 of the SHFS 6  UNHCR (United Nations High Commissioner for Refugees) (2017). 7  The CIA World Factbook estimates 3.79 percent for the period of 2010–2015. 8  International Monetary Fund (2018). 2  Somali Poverty and Vulnerability Assessment generated much needed indicators, but their geo- IDP settlements. North West includes the pre-war graphic coverage was limited while also excluding regions of Awdal, Sanaag, Sool, Togdheer, and nomads. Further, SLHS did not cover settlements Woqooyi Galbeed. North East includes the regions of internally displaced persons (IDPs).9 For the first of Nugal, Bari, and Mudug. Central includes the time, Wave 2 included the Somali nomadic popula- regions of Galgaduud, Hiraan, and Middle Shabelle. tion and many households in insecure areas. Wave Jubbaland includes Gedo and Lower Juba.10 South 2 targeted almost 6,400 households distributed West includes Bay, Bakool, and Lower Shabelle. among rural and urban areas in Central regions, Mogadishu includes all the households located in Jubbaland, Puntland, Somaliland, and South West the capital excluding IDPs. The population is fur- as well as urban areas in Banadir. The sample also ther grouped according to livelihood types: urban, featured nomads and households in IDP settle- rural, IDPs, and nomads. IDP settlements consist ments located in urban areas in the above geo- of settlements located in Mogadishu, North West, graphic areas, as well as households in IDP host North East, Central, Jubbaland, and South West. communities. This report is based on the most recent and The specific context of insecurity and lack of statis- first extensive household survey, Wave 2 of the tical infrastructure in Somalia posed several chal- SHFS. Caution should be taken to avoid compar- lenges for implementing a household survey and ing results obtained from Wave 1 and Wave 2 of measuring poverty. First, in the absence of a recent the Somali High Frequency Survey. Due to differ- census, no exhaustive lists of census enumeration ences in sampling frames, inferences about the areas along with population estimates existed, cre- population from which the sample was drawn are ating challenges to derive a probability-based repre- not comparable. Therefore, the report focuses on sentative sample. Therefore, geospatial techniques results using Wave 2, given the improved sampling and high-resolution imagery were used in the SHFS frame and greater survey coverage, both in terms to model the spatial population distribution, build a of geographical and population coverage. probability-based population sampling frame, and generate enumeration areas to overcome the lack The poverty and vulnerability assessment pre- of a recent population census. Second, while some sents a picture of Somali welfare conditions with areas remained completely inaccessible due to inse- the objective to inform policies and programs curity, even most accessible areas held potential aimed at building resilience and reducing pov- risks to the safety of field staff and survey respon- erty. Somalia emerged from a long civil war, still in dents, so that time spent in these areas had to be the process of graduating from fragility. Recurrent minimized. To address security concerns, the SHFS natural shocks, like the most recent drought, have adapted logistical arrangements, sampling strategy the potential to reverse development progress and using micro-listing, and questionnaire design to limit contribute to fragility. The large number of dis- time on the ground based on the Rapid Consump- placed people is a testament of Somalia’s volatil- tion Methodology. Third, poverty in completely inac- ity. The objective of the poverty and vulnerability cessible areas had to be estimated by other means. assessment is to contribute to a better understand- Therefore, the SHFS relied on correlates derived ing of livelihoods and vulnerabilities in Somalia to from satellite imagery and other geospatial data to inform improved livelihoods and resilience, a core estimate poverty in such areas. Finally, the nonsta- component of any sustainable development path tionary nature of the nomadic population required for Somalia. special sampling strategies. The methodology is described in detail in the accompanying background The report is organized into six chapters. The paper ‘Estimation of Poverty in Somalia Using Inno- first chapter presents an updated profile of mon- vative Methodologies’. etary and nonmonetary dimensions of poverty for the Somali population, including the nomadic Somali regions are identified according to dis- population. The second chapter explores in more tinct geographical areas: North West, North East, detail spatial variation, with a focus on urban- Central, Jubbaland, South West, Mogadishu, and ization. The third chapter examines the impact of the 2016/17 drought on livelihoods to identify 9  Wave 1 of the SHFS covered Puntland, Somaliland, and South Central. 10  Note: Middle Jubba was not surveyed due to insecurity. Introduction 3 the populations at risk and the factors that pro- and vulnerabilities, the fifth chapter focuses on tected households against its negative effects. social protection as a means of promoting equity The fourth chapter provides an in-depth analysis and building resilience against the effect of shocks of the internally displaced populations to identify on livelihoods. Similarly, the sixth chapter exam- displacement-related needs and to inform durable ines remittances and their role for livelihoods and solutions. As a reaction to the analysis of poverty resilience. 4  Somali Poverty and Vulnerability Assessment CHAPTER 1 Poverty Profile KEY MESSAGES Nearly 7 of 10 Somalis live in poverty, making of poverty. People in households that do not receive Somalia one of the poorest countries in Sub-Saharan remittances have a poverty rate that is 9 percent- Africa. About 69 percent of the population lived in age points higher than those in recipient households. poverty in 2017. Somalia has the sixth highest poverty Poverty is also deeper for non-recipient households. rate in the region, only after the Democratic Republic The negative correlation between poverty and receiv- of Congo, Central African Republic, Madagascar, ing remittances is confirmed by other poverty mea- Burundi, and South Sudan. Poverty incidence is lower sures such as the food consumption poverty and an in other urban areas, excluding Mogadishu, com- adult equivalent measure of poverty. Remittances can pared to nomadic households, IDPs in settlements, serve as a mechanism to smooth consumption in the and those in rural areas and Mogadishu. Nearly half of event of negative shocks and improve welfare condi- the population is not even able to meet the average tions, yet these transfers do not necessarily reach the consumption of food items, confirming the dire living ones most in need. Protecting vulnerable groups and standards of most Somalis. creating income generating opportunities is crucial to prevent childhood poverty from translating into pov- Poverty is both widespread and deep, particularly erty in adulthood. Targeting dedicated social protec- for households in rural areas and IDP settlements, tion programs can be a good alternative to reach the highlighting substantial challenges to overcoming most vulnerable and address the general lack of resil- poverty. While almost three-fourths of the popula- ience mechanisms. tion in rural areas, IDP settlements, Mogadishu, and among nomads live in poverty, according to survey Women are less likely to be the head of the house- estimates, poverty is deeper in rural areas and IDP hold and to participate in the labor market. Women settlements. The average poverty gap in Somalia is represent nearly half of the adult population, but only 29 percent, indicating that the average consump- 4 of 10 Somali households are headed by a woman. tion level of a poor Somali is about 71 percent of the 58 percent of men participate in the labor market international poverty line. Rural residents and IDPs compared to 37 percent of women. The gender gap in settlements are relatively worse off since they is primarily driven by a larger number of women stay- have the largest poverty gap (34 percent). To bring ing at home and caring for their families compared to the poor in the population out of poverty and up to men. Even though 64 percent of the Somali house- the poverty line, a transfer of around US$1.64 billion holds perceive that most or all women can work out- per year would be required under a perfect target- side the home, the gap in employment between men ing scheme and ignoring administrative and logistical and women is substantial (20 percentage points). costs. In addition to the high levels of poverty, a sig- Increasing participation of women in the labor mar- nificant proportion of non-poor Somalis are vulner- ket will be important to accelerate economic growth able to falling into poverty should they experience an and raise the living standards of Somali households. unexpected decrease in consumption levels. Around Removing barriers to work is a crucial step to tackle 10 percent of the population have a total daily con- gender inequalities. sumption expenditure within 20 percent above the poverty line. Overage enrollment is common, with stark geo- graphical and gender disparities in enrollment rates. Children and households that do not receive remit- Nearly 27 percent of children enrolled in primary tances are disproportionately poor. Children aged school are older than 13 years, and more than 55 per- 0–14 years represent nearly half of the total popula- cent of the population enrolled in secondary school tion, and 73 percent of them are poor according to are not aged 14–17 years. Somali children start primary survey estimates. Children from poor households are school late as most parents believe children aged 6–9 likely to grow up in challenging conditions, for exam- are too young to attend school. The perception of ple without electricity and deprived in the education parents is not associated with the fact that some chil- dimension, which ultimately hinders their path out dren would have to walk a long distance to school, —continued Poverty Profile 5 KEY MESSAGES—continued nor with the household’s own perception of safety for also less likely to have formal education compared to walking during the day. The net enrollment rate of the men. Furthermore, enrollment is associated with the population aged 6–25 years is 33 percent, and highest educational level of older generations as children are in urban areas. In Mogadishu and other urban areas, more likely to enroll in school in households with a net enrollment among primary school-aged children literate household head, after controlling for other (6–13 years) is around twice the enrollment in rural factors that affect school enrollment. In urban areas areas and IDP settlements and more than six times where access to education is higher, 11 percent of the the enrollment of nomadic children. The geographi- population aged 15 years or more were previously cal disparities in enrollment for the population of sec- enrolled but did not complete the primary level. While ondary school age (14–17) are likewise pronounced. access is still a big challenge for most Somalis and Moreover, there are no gender differences in both a crucial first step, additional policies to reduce the net enrollment rates and reasons for not attending dropout rates and increase the levels of educational among children aged 6–13 years. However, for chil- attainment will have to be considered. dren aged 14–17 years, a gender gap emerges as male enrollment is significantly higher after controlling for Some improvements in educational outcomes can age, poverty status, and other household character- be seen across generations. Despite large gender istics. The main reason for not attending school at and geographical disparities in terms of access and this age is the lack of money for boys, while having to availability of education, younger generations tend work or help at home for girls. Policy efforts should to have better educational outcomes as they more improve access and aim to increase enrollment rates likely to have formal education and to be literate. The while considering the disparities and needs of differ- government should try to explore and learn from the ent vulnerable groups. drivers behind the improvements seen in younger generations, to ultimately inform policies aimed at Distance from schools rather than the costs of achieving better educational outcomes for the Somali schooling affects the enrollment of children. For 1 population. out 3 Somali households, schools are at least 30 min- utes walking distance. Being more than 30 minutes The nomadic population are at disadvantage and away from school is negatively associated with enroll- face the biggest challenges to improve their edu- ment for primary school-aged children and the overall cational outcomes. The net enrollment rate of the enrolled population. On average, households spend nomadic population is 12 percent for both primary around 3 percent of the poverty line on education per and secondary school-aged children respectively. household member enrolled. Expenditure on educa- Only one in five can read and write and around 80 tion is weakly correlated with net enrollment and is percent do not have any formal education. Nomadic only significant for the overall enrollment rate but households reported the lack of schools nearby as the not for those of primary or secondary age. Increasing first or second reason for not attending school. Thus, access to education for children and youth will allow access seems to be the main barrier with 73 per- them to attain more productive opportunities later cent of households being far—more than 30 minutes in life and enhance their standard of living. Building away—from the closest school. more schools is one alternative, yet further analysis is needed given the complexity and cost of designing Inequalities in access to key services are large across and implementing policies aimed at improving access population groups, with rural residents, IDPs in set- to education. The challenge of increasing enrollment tlements, and nomads left behind for the most part. rates will continue to grow given the demographic Improved water and sanitation are critical for health, structure of Somalia and its overall young population. as inadequate sources for drinking water and poor hygiene affect school performance as well as produc- Gender and regional disparities in access to educa- tivity. However, only 5 of 10 households have access tion are reproduced in educational outcomes of the to improved sanitation and 8 of 10 to improved water Somali population. Education is a key tool for increas- sources. Also, only 5 of 10 households have electricity. ing the levels of welfare and helping to break the pov- Access to these services is higher in urban areas, with erty cycle. Only 1 of 2 Somalis can read and write, with the share of households with access relatively smaller literacy being more common among younger genera- for rural residents, IDPs in settlements, and nomads. tions, urban population, and men. Similarly, the share Poor households are also less likely to have access of urban residents without formal education is 1.6, 2.6, to improved sanitation and electricity. Markets and and 2.5 times lower than that of rural residents, IDPs health clinics are far—more than 30 minutes away— in settlements, and nomads respectively. Women are for more than a third of Somali households (34 to 6  Somali Poverty and Vulnerability Assessment 40 percent) and for most of the nomads. Enhancing are deprived in at least one dimension of educa- access to markets can increase productivity and tion, water, sanitation, or electricity, as well as mon- accelerate the reduction of poverty. etary poverty. The highest levels of deprivations are found among the nomadic population, and the low- Multiple deprivations in education, water, sanita- est in urban areas. Also, poor households are slightly tion, and electricity affect most Somali households more deprived than non-poor ones in educational, and are consistent with monetary poverty. Poverty water, and electricity dimensions. Moreover, mon- extends beyond the monetary component to non- etary poverty is correlated with multiple deprivations monetary deprivations across multiple dimensions. since around 40 percent of poor households are also Somali households are often more deprived in mul- deprived in at least one of the other four dimensions: tiple dimensions. Almost 9 of 10 Somali households education, water, sanitation, and electricity. Profiling the poor and vulnerable is crucial to FIGURE 1.1  n  Somali households by type of inform policies and alleviate poverty. Political population instability and conflict have eradicated the sta- Mogadishu, 10% tistical capacity of Somalia, resulting in a lack of information necessary for the effective design Nomads, 25% and implementation of policies. Such information gaps are currently being filled. Reducing poverty requires identifying and targeting the poor to improve their welfare conditions. Furthermore, the evolution of living standards should be monitored, Other urban, 30% and poverty reduction efforts evaluated.11 Profiling the population living below a minimum threshold IDPs in is a first crucial step for evidence-based planning settlements, aimed at alleviating poverty in Somalia. 14% This chapter presents an overview of poverty in Somalia. It describes the extent of poverty among Rural, 20% Somalis in 2017 using various measures of poverty, analyzes inequality among the population, and Source: Authors’ calculation based on the SHFS 2017–18. profiles the characteristics and living conditions of different groups. The chapter then reviews edu- was below the age of 30, and around 58 per- cational indicators, as well as labor market indica- cent aged 20 or less (Figure A.1 in the Appen- tors and access to services. Finally, it expands the dix). Nearly half of the population are women and analysis beyond monetary poverty to describe the the other half men. Urban households represent socioeconomic realities of Somalis by incorporat- 40 percent of the total (10 percent from Mogadi- ing other types of deprivations, such as water and shu and 30 percent in other urban areas), followed sanitation, living standards, and education. by the nomads with 25 percent, rural households with 20 percent, and internally displaced persons (IDPs) in settlements with 15 percent (Figure 1.1).13 Monetary poverty Provided Somalia decreases fertility rates, it has an opportunity to reap the benefits of a demographic A better future for Somalia depends on the young dividend stemming from a growing working-age and those living in rural areas, IDP settlements, population, but to achieve such gains will require and the nomadic population. A large working-age population in Somalia can accelerate future eco- nomic growth and increase overall welfare condi- tions.12 In 2017, 72 percent of the Somali population 13  Many households in Somalia are nomads or pastoralists, which implies they move from one place to the other in search for pasture, water, and/or food. Mobility is at the center of their livelihood and can involve seasonal concentration and dispersal 11  Baker (2000). of herders and their livestock, according to the availability of 12  Kelley and Schmidt (1999). forage and water in different areas. Poverty Profile 7 FIGURE 1.2  n  Cross-country comparison of poverty in FIGURE 1.3  n  Cross-country comparison of poverty 2017 and GDP 100 95 90 Poverty incidence (% of population) 85 Poverty incidence (% of population) SSD 80 75 70 SOM MWI 65 60 55 RWA 50 40 45 TZA UGA 30 35 20 ETH 25 10 15 0 0 500 1,000 1,500 2,000 2,500 3,000 SSD BDI MDG CAF COD SOM MWI GNB MOZ RWA LBR SLE BEN TGO MLI NER TZA TCD BFA UGA SEN GIN ETH ZWE COM GDP per capita (US$ PPP) Regional average Source: Authors’ calculation based on the SHFS 2017–18, and World Bank Open Data. Source: Authors’ calculation based on the SHFS 2017–18, and World Bank Macro Poverty Outlook. in the region, only after the Democratic Repub- lic of Congo, Central African Republic, Madagas- increased support to younger generations as well car, Burundi, and South Sudan (Figure 1.2).17 The as greater reach to large segments of the popula- Somali population has relatively low levels of eco- tion that live in rural areas, IDP settlements, and nomic activity and income, as reflected by a Gross the nomads.14 Domestic Product (GDP) per capita of US$500 in 2017.18 The high poverty incidence of Somalia is Nearly 7 of 10 Somalis live in poverty, which in line with its low level of income, as suggested makes Somalia one of the poorest countries in by the relationship between poverty and GDP per Sub-Saharan Africa. Poverty in Somalia is wide- capita across Sub-Saharan Africa (Figure 1.3). Alle- spread with 69 percent of the population living viating poverty in Somalia requires accelerating in poverty in 2017 (see Boxes 2 and 3 for more economic growth to increase the income levels details), as defined by having a total daily per cap- and living standards of the population. Reduced ita consumption expenditure lower than the inter- fertility rates and population growth can improve national poverty line of US$1.90 at 2011 purchasing the prospects of economic development and pov- power parity (PPP).15 The incidence of poverty was erty reduction. Somalia has experienced a steady 19 percentage points higher in Somalia compared decrease in fertility rates from 7.7 births per women to the unweighted average of low-income coun- in 1998 to 6.3 in 2016.19 In the same period, annual tries in Sub-Saharan Africa (51 percent) in 2017.16 population growth decreased from 3.4 to 2.9 per- The country has the sixth highest poverty rate cent. These demographic changes could increase 14  The demographic dividend refers to economic growth as a 17  The countries used for regional comparison are all the African result from having a large proportion of working age population low-income countries as defined by the World Bank: Benin, relative to the number of dependents (children and elderly), Burkina Faso, Burundi, Central African Republic, Chad, Comoros, which allows for some resources to be allocated in productive the Democratic Republic of Congo, Eritrea, Ethiopia, Guinea, activities that would have otherwise been used to support the Guinea-Bissau, Liberia, Madagascar, Malawi, Mali, Mozambique, dependents. Niger, Rwanda, Senegal, Sierra Leone, South Sudan, Tanzania, 15  The value of the international poverty line in 2017 was esti- Togo, Uganda, and Zimbabwe. For each country, we include the mated using the 2011 So.Sh./$ PPP, a Somali Consumer Price most recent available year for each indicator. Index increase between 2011 and 2017, and the 2017 nominal 18  GDP per capita estimate from the Macro Poverty Outlook exchange rate between the Somali Shilling and the US Dollar. Indicators, Spring Meetings 2018 in World Bank (2018b). 16  Based on World Bank estimates. 19  According to data from World Bank Open Data. 8  Somali Poverty and Vulnerability Assessment Box 2 ■  Wave 1 and 2 of the Somali High Frequency Survey The infrastructure of the Somali High Frequency Survey (SHFS) offers a modern data collection system that can be used to fill the most important data gaps. In the absence of representative household surveys not much was known about welfare conditions of the Somali population. The World Bank’s Somali High Frequency Survey provides quantitative data to inform essential resilience programs and shape policy.20 The success of this estab- lished survey infrastructure offers an opportunity to implement additional waves of the survey with expanded coverage.21 The World Bank implemented the first wave of the Somali High Frequency Survey in 2016. The survey was administered to urban and rural households in North East, North West, and Banadir, as well as IDP settlements.22 However, the sample was not fully representative of the Somali population as it excluded nomadic households, and households in inaccessible and conflict-affected areas. Wave 2 implemented in 2017 included for the first time the nomadic population and expanded its coverage to include additional urban and rural areas. The survey was administered to households distributed among rural and urban areas in Central regions, Jubaland, North East, North West, South West and urban areas in Banadir. The sample also featured households in IDP settlements and the nomadic population. The data from both waves is not fully comparable due to differences in the sampling frame and accessibility of areas during fieldwork, thus the Poverty Assessment primarily uses data from Wave 2. Data collection is chal- lenging in Somalia due to insecurity in some areas and the lack of an updated and reliable source of information to derive a representative sample. The sampling frame for Wave 1 was based on the 2014 Population Estimation Survey of Somalia (PESS) for urban areas, while for rural areas PESS was combined with a list of settlements from different sources to complement missing rural and semi-urban settlements. Wave 2 used a WorldPop population density layer together with PESS and other existing data sources to create urban, rural, and IDP strata, while considering a security assessment to exclude insecure areas. Therefore, the sampling and accessibility of regions covered in both waves was different in 2016 and to 2017, and it is not recommended to compare the data from both waves of the SHFS without addressing these caveats. the availability of resources within the household p<0.05 vs. IDPs in settlements and nomads, and and help them in feeding, educating, and providing p<0.10 vs. rural areas).24 A higher poverty rate in health care to children. Mogadishu compared to other urban areas might be the result of a larger concentration of IDP popu- Poverty is widespread across Somalia with lower lation and the challenges associated with the dis- incidence found in other urban areas, and simi- placement crisis (see Chapter 4, Displacement, and lar levels among nomads, IDPs in settlements, Chapter 2, Spatial Variation in Living Standards, for and the population in rural areas and Mogadishu. a detailed discussion).25 Urban areas usually benefit from agglomeration effects that result in more economic opportuni- High levels of poverty are more prevalent in the ties and access to services, relative to rural areas.23 North and South West of Somalia according to Poverty incidence is similar (between 72 and 76 poverty estimates from satellite images. Data col- percent) for those living in Mogadishu, rural areas, lection in Wave 2 was restricted to accessible areas IDPs in settlements, and nomadic households due to insecurity. Thus, the survey estimates are (Figure 1.4). Only those living in other urban areas, without considering Mogadishu, have a smaller incidence of poverty (60 percent), than the rest 24  An adult equivalent measure of poverty is consistent with of the Somali population (p<0.01 vs. Mogadishu, this characterization of the poor (see the Appendix A for more details). 25  Banadir concentrates 41 percent of IDPs in settlements and 28 percent of the overall displaced population according to the 20  World Bank (2014). second wave of the SHFS. The share is similar (around 22 per- 21  Pape and Mistiaen (2015). cent) for the overall displaced population with data from the 22  World Bank (2017a). Protection & Return Monitoring Network of the United Nations 23  Lall, et al. (2017). High Commissioner for Refugees (UNHCR). Poverty Profile 9 FIGURE 1.4  n  Poverty incidence FIGURE 1.5  n  Map of poverty incidence from satellite estimates27 100 Percent of population Percent of population 80 60 40 20 0 ur u se R an em al om nts s ed H H itt es s Ps s af ted ed ad ce P er sh ad H H ttl ur m c b ct -ID ID e ht ec re a n an th di he d fe e de O ga ug aff on d itt N al a ve m o N ro t M he M D gh e i re ou e in ec d al R eive dr Ps m Fe ot ID c re N ot N Overall average Source: Authors’ calculations based on the SHFS 2017–18. only representative of accessible areas in Somalia (see Table A.1). Wave 2 filled this critical gap by imputing poverty based on data extracted from satellite images for inaccessible areas (Box 4).26 80–100 This approach allows to have an objective measure 60–80 of poverty for areas where the survey data are not 40–60 available. The satellite estimates indicate that pov- 20–40 erty incidence is highest—more than 80 percent— 0–20 in the North (some districts of Togdheer, Sanaag, and Bari) and South West (some districts of mid- Source: Authors’ calculation from satellite data. dle Juba, Gedo, and Bay), besides a few districts of Note: The poverty incidence of each region does not include IDPs in Mudug and Galguduud (Figure 1.5). settlements. Poverty is both widespread and deep, particularly for households in rural areas and IDP settlements, and IDP settlements (34 percent for both), com- highlighting substantial challenges to overcom- pared to Mogadishu (27 percent, p<0.1) and other ing poverty. The poverty gap can be defined as the urban areas (24 percent, p<0.05). A large share of minimum amount of resources that would have to Somalis living in poverty, together with a large gap be transferred to the poor, under a perfect targeting between their consumption expenditure and the scheme, to eradicate poverty (Box 3). The average poverty line indicate that many of the poor are far poverty gap in Somalia is 29 percent (Figure 1.6), from overcoming poverty and would need a sub- indicating that the average consumption level of a stantial increase in their consumption to bring it poor Somali is about 71 percent of the international to the poverty line. A transfer of around US$1.64 poverty line. While almost three-fourths of the billion per year would be required under a perfect population in rural areas, IDP settlements, Moga- dishu, and among nomads are poor according to survey estimates, poverty is deeper in rural areas The boundaries on the map show approximate borders of 27  Somali pre-war regions and do not necessarily reflect official borders, nor imply the expression of any opinion on the part 26  For a detailed description of the methodology see Pape, U. of the World Bank concerning the status of any territory or the and P. Wollburg (2018). delimitation of its boundaries. 10  Somali Poverty and Vulnerability Assessment Box 3  ■  Measures of poverty Measuring living standards is crucial for poverty reduction efforts to be successful. The international poverty line was introduced in the 1990 World Development Report with the aim of measuring poverty consistently across countries.28 The value of the poverty line has been revised through the years and adjusted to reflect welfare conditions of low-income countries, and it currently stands at a daily value of US$1.90 (2011 PPP) per person. Comparable poverty measures help us to identify poor households, monitor the evolution, and assess the effectiveness of policies. The poverty incidence is the most common poverty measure. The poverty incidence or headcount ratio refers to the share of population that is poor or that have a total consumption lower than the poverty line. It’s derived from the total consumption of the household in food, nonfood, and durable goods; the number of members that comprise the household; and a specific consumption threshold or poverty line. This measure describes the extent of poverty in a country or region. The poverty gap index measures how far poor households are from overcoming poverty, while the poverty severity index measures the level of inequality among the poor. The poverty gap index is the difference between current consumption and the poverty line as a proportion of the poverty line for the poor population. It can be interpreted as the minimum amount of resources that would have to be transferred to the poor, under a perfect targeting scheme, to eradicate poverty.29 The poverty severity index is estimated as the square of the poverty gap. It attributes a larger weight to the poorest among the poor, thus reflecting inequality conditions for the poor. A food consumption measure of poverty considers the total consumption of each household relative to the average expenditure on food items only. Using the total consumption of households, a food consumption mea- sure of poverty identifies those households that cannot afford the average food consumption, even if they were to allocate all their expenditure to food items only. Effectively, the poverty line is scaled down by multiplying for the overall share of food consumption relative to total consumption. targeting scheme and ignoring administrative and by women is only significant for rural areas and IDP logistical costs to bring the poor in the population settlements (p<0.01 and p<0.1 from Table A.2), after out of poverty.30 In line with these results, the aver- controlling for age of the household head, house- age poverty severity index is 15 percent, pointing hold composition, access to services, and sources to inequalities among the poor. These inequalities of income. However, poor households headed by are concentrated in rural areas and IDP settlements men and women have on average the same pov- (Figure 1.7), compared to Mogadishu, other urban erty gap (Figure A.2). Children are also less poor areas, and the nomads (for all the comparisons, at in households headed by women after controlling least p<0.05). for regional differences (p<0.05, Table A.3). Overall, households headed by women have a larger share Poverty has a gender dimension as households of working age members (p<0.05), which might headed by women are slightly less poor. House- explain a slightly higher consumption level among holds headed by women have a poverty incidence this group of households. Any policy or program that is 6 percentage points lower than those headed aimed at reducing poverty should consider the gen- by men (66 vs. 72 percent, p<0.05). The results are der dimension of poverty in Somalia. robust and weakly significant (p<0.1) after control- ling for regional differences (Table A.4). The find- Children are disproportionately affected by pov- ing of overall lower poverty in households headed erty. Children aged 0–14 years are one of the most vulnerable groups, and those from poor house- holds face bigger obstacles to overcome pov- erty in their adult life.31 They represent nearly half 28  Ravallion, et al. (2009). of the total Somali population (49 percent), but 29  Deaton (2006). 30  Corresponds to an annual value for all the regions, including areas not covered in Wave 2 of the SHFS. For these, the same poverty incidence and gap was assumed as in regions covered by the survey. 31  UNICEF (2016). Poverty Profile 11 FIGURE 1.6  n  Poverty gap FIGURE 1.7  n  Poverty severity 50 30 25 Percent of poverty line Poverty severity index 40 20 30 15 20 10 10 5 0 0 Mogadishu Other urban Rural IDPs in settlements Nomads Mogadishu Other urban Rural IDPs in settlements Nomads Overall average Overall average Source: Authors’ calculations based on the SHFS 2017­ –18. Source: Authors’ calculations based on the SHFS 2017–18. 73 percent of them are poor according to survey robust and significant (p<0.05) after controlling estimates. The youth aged 15–24 years represent for regional differences (Table A.4). Among the around 15 percent of the population and 68 per- poor, poverty is also deepest for households that cent of them live in a household whose consump- did not receive remittances (p<0.01, Figure A.3 and tion is less than the poverty line (Figure 1.9). Child p<0.01 from an OLS regressions with fixed effects poverty incidence is similar in Mogadishu, rural in Table A.5). The correlation between poverty and areas, IDPs in settlements and among nomadic receiving remittances is confirmed by other pov- households (Figure 1.8). Compared to other urban erty measures (see Chapter 6, Remittances, for a areas, children are more likely to be poor in Moga- detailed discussion). Food consumption poverty is dishu (13 percentage points of difference, p<0.01) also less likely for households that received remit- and in IDP settlements (16 percentage point dif- tances compared to non-receivers (p<0.01). Youth ference, p<0.05). Moreover, the gap between child are 12 percentage points less likely to be poor in and overall poverty incidence is larger for rural households that received remittances compared households and those living in IDP settlements. to non-receivers (p<0.05) (Figure A.3 and p<0.05 It is partially explained by high poverty rates, but from logistic regressions of Table A.6). Receiving also because households have on average more remittances seems to contribute to the well-being children than the overall average of 2.6 (2.8 and 2.9 of some households. However, they are not immune in rural areas and IDP settlements, respectively). to shocks nor remittances scale with needs. Fur- Breaking the poverty cycle requires improving thermore, remittances are de-centralized and not conditions for children and youth. The challenge targeted to the most vulnerable households. Social will only grow considering the country’s demo- protection programs can reach the ones most in graphic structure. need and help lift the population out of poverty (see Chapter 5, Social Protection, for a detailed Remittances provide a lifeline to some house- discussion). holds, which makes them less likely to be poor or their poverty less deep. Receiving remittances Nearly half of the population is not able to meet can serve as a resilience mechanism to smooth the average consumption of food items, high- shocks and improve welfare conditions. Poverty lighting the dire living standards of most Somalis. is 9 percentage points lower for households that The food consumption measure of poverty corre- received remittances, compared to non-receivers sponds to households that have a total consump- (62 percent vs. 71 percent, p<0.01). The results are tion smaller than the average expenditure on food 12  Somali Poverty and Vulnerability Assessment FIGURE 1.8  n  Child poverty incidence FIGURE 1.9  n  Youth poverty incidence 100 100 90 90 80 80 70 70 Percent of children Percent of youth 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Mogadishu Other urban Rural IDPs in settlements Nomads Mogadishu Other urban Rural IDPs in settlements Nomads Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. items across regions (Box 3). Thus, households the displacement crisis and ensuring this group is considered poor with this threshold are those that integrated into society and the economy. are not able to afford the average food expenditure, even if they were to allocate all their expenditure Experiencing hunger is equally likely for poor to food items and nothing to nonfood and durable and non-poor households. Hunger is likely to be items. Food consumption poverty is 49 percent in present in most households after a severe shock Somalia (Figure 1.10), and is similar in Mogadishu like the drought experienced in Somalia between (43 percent), rural areas (44 percent), for IDPs in March 2016 and December 2017.33 Forty-two per- settlements (50 percent), and the nomads (41 per- cent of poor households reported experiencing cent). Households living in other urban areas are some hunger compared to 38 percent of non-poor, less likely to be poor with this measure than rural but the difference is not significant. Consistent areas (22 percentage point difference, p<0.05) with monetary poverty, households from other and those in IDP settlements (28 percentage point urban areas were less likely to report hunger (22 difference, p<0.05). Food consumption poverty is percent, Figure 1.11) than IDPs in settlements (60 also less likely among households that have not percent, p<0.01), nomads (50 percent, p<0.01), been displaced (46 percent) compared to the rural households (44 percent, p<0.01), and those group of IDPs (55 percent, p<0.05).32 Food con- in Mogadishu (43 percent, p<0.01). Households sumption poverty indicates most Somalis live in receiving remittances have an advantage and thus extreme conditions, and that some vulnerabilities are slightly less likely to report hunger compared seem to be associated to the displacement status to non-receivers after controlling for regional dif- of households. Alleviating poverty in Somalia will ferences (p<0.1, Table A.7). Moreover, those that require addressing significant challenges posed by reported to be affected by the drought are more likely to report some hunger in the past four weeks, Households living outside of IDP settlements were classified 32  as being displaced if they were living in any location because they were forced to leave their usual place of residence due to conflict, violence, human rights violations, and natural or man- Corresponds to experiencing hunger at least 1–2 times in the 33  made disasters. past four weeks. Poverty Profile 13 Box 4  ■  Poverty estimates from satellite images for inaccessible areas Data collection in Wave 2 was restricted to accessible areas, so poverty was imputed for inaccessible areas using data extracted from satellite images. The implementation of the SHFS was challenging due to insecurity. Wave 2 considered a security assessment and excluded insecure areas. Hence, an alternative approach was employed to provide an objective measure of poverty for areas where the survey data are not available. Wave 2 filled this gap by imputing poverty based on satellite imagery for inaccessible areas. The methodology has been used for Nigeria, Tanzania, Uganda, Malawi, and Rwanda.34 These experiences show that image features can explain up to 75 percent of the variation in local-level outcomes, ultimately suggesting that poverty estimates of inaccessible areas are reliable. The data extracted from satellite images corresponds to distance to certain reference points, population, and conflict density, as well as rain and temperature levels. The information used to predict the poverty rate of inac- cessible areas refers to the distance from the center of each geographical unit to bare areas, cultivated areas, major roads, drought areas, health clinics, schools, water sources, waterways, food insecure areas, urban centers, and unsafe areas. In addition, data on temperature, precipitations, conflict density, and population density were also included in the estimation. The pictures below are examples of explanatory variables extracted from satellite images. Distance to cultivated areas Distance to unsafe areas Population density The poverty rate—from survey estimates—was regressed on the data extracted from satellite images for each administrative area, explaining between 56 and 97 percent of the variation. The correlation between poverty and the explanatory variables was different for each population type. For each, a separate linear model was estimated with interaction terms using all the explanatory variables. The final specification was derived from a stepwise regression to maximize the adjusted R-squared and minimize the root mean squared error, considering the information from all accessible areas. Poverty was then predicted and weighted by population in areas where survey data were not available, while excluding inhabited areas. To derive a nationwide poverty rate, survey and satellite estimates were combined. For each pre-war region and population type, the satellite prediction was considered if the accessibility rate of Wave 2 was 90 percent or less, and the survey estimate used if accessibility exceeded this threshold. The adjusted R-squared of the final model for urban areas is 56 percent while 95 per- cent for rural areas. A lower variation explained by the model in urban areas is the result of larger heterogeneity in poverty rates combined with the lack of higher spatial frequency in the data available for urban areas. For a detailed description of the methodology see Pape, U. & P. Wollburg (2018). 34  Xie, et al. (2015)and Jean, et al. (2016). 14  Somali Poverty and Vulnerability Assessment compared to households not affected (p<0.01 from Drought Impact, for a detailed discussion).36 logistic regressions of Table A.7).35 The drought Efforts aimed at building resilience are crucial to resulted in higher food prices, low purchasing protect vulnerable groups from food insecurity power and displaced an additional 1 million people, and malnutrition. leading to acute food insecurity (see Chapter 3, FIGURE 1.10  n  Food consumption poverty incidence Inequality and vulnerable 80 population 70 Percent of population 60 For its level of poverty, inequality is relatively 50 low in Somalia compared to other low-income Sub-Saharan countries. The Gini index, a measure 40 of inequality, was 34 percent for Somalia in 2017 30 (Figure 1.13). Due to the high levels of monetary 20 deprivation shared by most households, consump- tion is relatively homogenous among them. For 10 similar levels of poverty as in Somalia, other low- 0 income Sub-Saharan countries tend to have higher Mogadishu Other urban Rural IDPs in settlements Nomads levels of inequality. For example, Malawi and South Sudan, which have a poverty incidence of 69 and 82 percent respectively, have around a 12 percent- age points higher Gini than Somalia (Figure 1.12). Inequality is highest in rural areas and lowest in Overall average Mogadishu. The Gini index is 41 percent in rural areas, 34 percent in other urban areas, and 26 per- cent in Mogadishu (Figure 1.13). Urban areas might Source: Authors’ calculations based on the SHFS 2017–18. benefit from agglomeration effects that bring FIGURE 1.11  n  Experience of hunger in past 4 weeks 70 Percent of households 60 50 40 30 20 10 0 hu an ral nts ds or or H H es es Ps DPs d cte cte d a dis r urb Ru eme oma Po n-po e d H ed H t a nc tanc n - ID I ffe affe g e ttl N No a d ad t i mi t o a Mo Oth se he he e m e N h t h t in le le dr dr ug ug P s e ma Ma e ive eive t dro Dro ID F c c No t re Re No Overall average Source: Authors’ calculations based on the SHFS 2017–18. 35  Corresponds to households that self-reported to be affected by the drought or shocks associated to it, like fire, severe short- UNHCR (United Nations High Commissioner for Refugees) 36  age in water for cattle or farming, livestock death or disease, (2018a); Famine Early Warning Systems Network (FEWSNET and high food prices. (2017)). Poverty Profile 15 FIGURE 1.12  n  Cross-country comparison of poverty FIGURE 1.13  n Inequality and inequality 50 60 40 GINI Index (0–100) 55 RWA 30 GINI Index (0–100) 50 SSD 20 MWI 45 UGA 10 40 TZA 0 35 Mogadishu Other urban Rural IDPs in Nomads SOM settlements ETH 30 Overall average 15 25 35 45 55 65 75 85 Poverty incidence (% of population) Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18, and World Bank Open Data. TABLE 1.1  n  Inequality decomposition more economic opportunities and larger access to Theil GE(1) inequality index services, ultimately leading to more homogenous consumption among the urban population, relative By population to the rural population.37 Additionally, the support Decomposition type By region from donors is more likely to reach urban centers Between group 0.002 0.005 due to insecurity and inaccessibility of some rural areas, which can also help level the consump- Within group 0.208 0.205 tion of the urban population. Compared to rural Total 0.210 0.210 households, those in Mogadishu are more likely to have water at home (p<0.01), electricity (p<0.01), Source: Authors’ calculations based on the SHFS 2017–18 improved sources of drinking water (p<0.01), a mobile money account (p<0.05), and a larger share who live just less than 10 minutes walking (one way) to the closest market (p<0.01). other urban, IDPs in settlements, and nomads) Inequality stems largely from differences within largely explain inequality in consumption rather regions and population groups, rather than from than the differences between regions or types of differences between them. The Theil index— population. Alleviating poverty in Somalia entails another measure of inequality which decomposes providing sufficient economic opportunities for total inequality into the proportion explained by individuals to improve their income levels. Nev- differences within and between groups—indicates ertheless, to achieve shared prosperity special that around 98 and 99 percent of total inequal- attention should be placed in generating higher ity corresponds to within the group component consumption growth for households at the bottom (Table 1.1).38 Differences between households from of the distribution. the same region or population group (Mogadishu, Households in the top 60 percent of the consump- tion distribution spend nearly three times more than those in the bottom 40 percent. The aver- Lall, et al. (2017). 37  The Theil Index measures inequality based on an entropy 38  age daily real consumption per capita in Somalia is measure. The index presented in this chapter corresponds to US$1.26. Overall, households in the top 60 percent GE(1), which is also referred to as Theil’s T index. of the total consumption distribution consumed 16  Somali Poverty and Vulnerability Assessment Box 5  ■  A remote monitoring system tracks migration patterns of nomads The second wave of the SHFS extended the coverage to consider the nomadic population despite the chal- lenges of including them in a household survey. Nomads make up around a third of the Somali population, yet only sporadic and non-systematic data are available about their welfare conditions, patterns, or needs. Including the nomads in a household survey with traditional methodologies is challenging as they change location con- stantly and thus they could move in and out of the surveyed area.39 Wave 2 filled this critical gap by collecting systematic data to account for the large nomadic population by combining information on water points with a series of Key Informant Interviews and a listing exercise. Wave 2 also introduced a new approach to track the migration patterns of nomads, providing invaluable information for policy efforts. The second wave of the SHFS established a remote monitoring system to track the migration patterns of the nomadic population. It consisted of autonomous position trackers successfully dis- tributed to a group of 197 nomads. These devices will send the location of the nomads for a period of two years to a secure cloud-based server. The pictures below are an example of the position of these nomadic households and their migration patterns captured in real time. This innovation will enhance future sample designs and ensure nomads are accurately represented in surveys. It will also improve our understanding about their patterns and routes, as well as provide invaluable information for emergency assistance and service delivery to this population.   2.8 times more than households in the bottom the largest differences between rural and urban 40 percent (an average of US$1.70 and US$0.62, areas, as well as between IDPs in settlements and respectively). Consistent with inequality measures, nomads, are found below the poverty line (Fig- the disparities for those two groups are larger ure 1.14). Additionally, the difference in consump- among rural households (3.4 times more), and in tion among groups in the top of the distribution is other urban areas (2.8 times more). Contrary to relatively small. this, the difference between households in the top 60 percent and bottom 40 percent is lowest in A large share of the Somali population has con- Mogadishu, where the former group only consumes sumption levels just above the poverty line, and 2.1 times more than the latter (Table 1.2). In terms thus is susceptible to fall into poverty in case of of the overall distribution of total consumption, an adverse shock. The Somali population is at con- stant risk of experiencing a negative shock to their income and consumption levels due to recurrent droughts, among other shocks. Around 10 percent 39  Himelein, et al. (2014). Poverty Profile 17 TABLE 1.2  n  Average real consumption per capita (daily 2017 US$) Bottom 40% Top 60% Overall average Overall 0.62 (0.59, 0.64) 1.70 (1.57, 1.83) 1.26 (1.16, 1.37) Mogadishu 0.71 (0.67, 0.76) 1.48 (1.39, 1.57) 1.17 (1.10, 1.24) Other urban 0.67 (0.63, 0.71) 1.87 (1.76, 1.99) 1.39 (1.25, 1.53) Rural 0.51 (0.47, 0.55) 1.71 (1.25, 2.18) 1.23 (0.87, 1.60) IDPs in settlements 0.55 (0.50, 0.60) 1.48 (1.16, 1.81) 1.10 (0.84, 1.37) Nomads 0.68 (0.63, 0.73) 1.67 (1.40, 1.95) 1.28 (1.06, 1.50) Source: Authors’ calculations based on the SHFS 2017–18. Note: 95% confidence intervals reported in parenthesis. FIGURE 1.14  n  Consumption distribution 100 100 90 90 80 80 Percent of population Percent of population 70 70 60 60 50 50 40 40 30 30 Poverty line (US$1.9 PPP) Poverty line (US$1.9 PPP) 20 20 Food consumption poverty line Food consumption poverty line 10 10 0 0 0 2 4 6 8 0 2 4 6 8 Daily consumption expenditure per capita (US$) Daily consumption expenditure per capita (US$) Urban Rural IDPs in settlements Nomads Source: Authors’ calculations based on the SHFS 2017–18. of the non-poor population have a total daily con- International and humanitarian aid can be con- sumption expenditure within 20 percent from the strained by the local capacity to efficiently deliver poverty line.40 The urban population is more vul- services. Hence, a social safety net program can be nerable since 12 percent of them are in this range a good alternative to support and build resilience (10 percent in Mogadishu and 13 percent in other among the non-poor and vulnerable segments of urban areas), compared to 10 percent of the rural the population. population, and 9 percent and 7 percent of the nomads and IDPs in settlements, respectively. Every nomadic household owns at least one goat, The population clustered above the poverty line sheep, camel, donkey, cattle, or chicken, and they is susceptible to fall into poverty in case of an tend to own more than non-nomadic households. unexpected decrease in their consumption levels. Pastoralist or nomadic livelihood involves rais- ing livestock and moving constantly according to seasonal variations in search of water and pasture 40  The standard international poverty line of US$1.90 at 2011 (Box 5). Every nomadic household owns some PPP corresponds to US$1.40 per day per person in 2017. 18  Somali Poverty and Vulnerability Assessment FIGURE 1.15  n  Livestock ownership FIGURE 1.16  n  Number of livestock owned 100 40 Average number of items owned Percent of households 80 30 60 20 40 10 20 0 0 hu n al ts s hu an al ts s ad a ad ur en ur en is rb is rb om R om ad R m ru ad m ru e e og N og e N ttl e ttl th th M se M se O O in in Ps Ps ID ID Goats Sheep Camels Goats Sheep Camels Donkeys Cattle Chickens Donkeys Cattle Chickens Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. type of livestock, compared to 53 percent of the headed by men and 42 percent by women (Fig- rural households, 35 percent of IDPs in settlements, ure 1.17). There are large gender differences across 23 percent in other urban areas, and 19 percent in regions; households are more likely to be headed Mogadishu. Chickens are the only type of livestock by women in Mogadishu (52 percent), other urban owned by a similar share of rural and nomadic areas (52 percent), and in IDP settlements (54 households, 18 vs. 11 percent, respectively (p>0.1, percent), compared to rural households (37 per- Figure 1.15). Among owners, nomadic households cent, p<0.05 vs. Mogadishu, and p<0.01 vs. IDPs in also own a larger number of livestock than non- settlements and other urban areas) and nomadic nomadic households, nearly twice for cattle, sheep, households (23 percent, p<0.05 vs. rural areas). goat, and camels (Figure 1.16). Wealth in the form In terms of gender composition within the house- of livestock represents an advantage against other hold, poor and non-poor households have a similar populations living in IDP settlements, and urban proportion of male and female members within the and rural areas. Yet, relying primarily on livelihood household (Table 1.3). based on livestock makes them more vulnerable to climate-related shocks. For example, the drought In Somalia, the education of the household head led to low birth rates and livestock deaths, rep- is strongly correlated with age, gender, and resenting a loss of between 25 and 75 percent of receiving remittances, but weakly correlated with their herds in the first six months of 2017.41 poverty. Education allows people to access better economic opportunities and improve their overall well-being. Households headed by men are more Demographic characteristics likely to have some formal education, compared to those headed by women (37 vs. 28 percent, and labor markets p<0.05).42 The results are robust and significant (p<0.01) after controlling for age, poverty status Four of 10 Somali households are headed by and other household characteristics (Table A.8). a woman. Women represent nearly half of the Also, older household heads, those not receiving adult population, yet less likely head a household. Around 58 percent of the Somali households are 42  Corresponds to formal education, including incomplete pri- mary, complete primary, incomplete secondary, complete sec- 41  FSNAU and FEWSNET (2018). ondary, university, and other. Poverty Profile 19 FIGURE 1.17  n  Female headed households the education outcomes of the poor might allow them to engage in better income-generating eco- 70 nomic activities and enhance their consumption 60 levels. Percent of households 50 A larger number of household members and 40 dependents are salient characteristics of the 30 poor. Consistent with cross-country observations, 20 poor households tend to have more members and a higher dependency ratio.44 The typical Somali 10 household has 5.4 members, with 5.9 members 0 among the poor and 4.5 among the non-poor. Overall, poor households have around 1.6 more hu n al ts s or or ad a ur en Po o is rb -p om R ad m members than non-poor households across Soma- ru on e og N e ttl N th M se lia (p<0.01, Table 1.3). The results are robust and O in significant (p<0.01) after controlling for other rel- Ps ID evant household characteristics. Household size Overall average is larger in Mogadishu, urban areas, and among the nomadic population (p<0.01, Table A.2). The Source: Authors’ calculations based on the SHFS 2017–18. Somali population is predominantly young, imply- ing that a large share is not of working age. There are 1.3 dependents in every household for every member of working age.45 On average, there are remittances and those that were displaced are 0.5 more dependents in poor households across less likely to have some formal education (p<0.01 the country (p<0.01), and the dependency ratio for each). There are some differences that are negatively associated with the consumption quin- weakly significant between poor and non-poor tiles (Figure A.7). This finding is smaller and weakly households (p<0.1). One plausible explanation for significant after controlling for other household this finding is that besides having some years of characteristics (p<0.1). Having more members and schooling, the quality of the education matters as dependents among poor households is explained well as the returns to education in the labor mar- by a larger number of children in poor households ket. The share of poor and non-poor households relative to non-poor. Overall, poor households have with at least one employed member is 72 and 73 1.1 more children than non-poor (p<0.01), and there percent respectively, suggesting they have similar are no differences between poor and non-poor in access to the labor market. terms of number of elderly within the household. Poor households have a smaller proportion of Men are much more likely to participate in the literate members. The international evidence labor market than women. Somalia has traditional indicates that educational outcomes tend to be gender roles which are reflected in the profile of associated with poverty.43 Overall, the proportion the population in the labor market. Overall, nearly of literate members in the household is nearly 6 5 in 10 Somalis aged 15–64 years are economi- percentage points lower in poor households com- cally active in the previous week, either employed pared to non-poor (p<0.1, Table 1.3). The difference (45 percent) or unemployed but actively looking is only significant for urban areas (Mogadishu p<0.1 for work (2 percent). Participation rates are simi- and other urban p<0.01), and not in rural areas, IDP lar across the urban, rural, IDP in settlements, and settlements, or for nomads (Table A.2). This might nomadic population, and most of the inactive pop- be explained by large spatial differences in terms ulation are not enrolled in school (Figure 1.18). In of availability and access to education. Improving terms of participation by gender, 58 percent of the Banerjee and Duflo (2007). 44  The age dependency ratio is defined as the proportion of 45  children and old age dependents to working age population 43  Banerjee and Duflo (2007). (15–64). 20  Somali Poverty and Vulnerability Assessment TABLE 1.3  n  Demographic attributes of poor households Logit regression Household characteristic Poor Non-poor Difference on poverty status Household size 5.9 4.5 1.4*** 0.6*** Age dependency ratio 1.5 1.0 0.5*** 0.2* Number of children 3.0 1.9 1.1*** –0.1 Proportion of men in the household 50.1 49.5 0.7 2.1 Share of households headed by men 60.2 54.7 5.5 2.8 Age of household head 39.9 37.7 2.2*** –0.1 Share of literate household heads 48.9 52.1 –3.2 5.8 Share of literate members in the household 44.1 49.9 –5.8* –6.1* Share of households with improved sources of water 77.4 75.5 1.9 2.6 Share of households with improved sanitation 42.9 51.0 –8.1* –2.5 Share of households with access to electricity 47.0 60.9 –13.9*** –7.9*** Main source of income: Salaried labor 36.0 38.4 –2.4 Reference Main source of income: Agriculture, fishing, and hunting 23.9 21.8 2.1 –4.3 Main source of income: Small family business 12.0 12.0 0.0 –5.7*** Main source of income: Remittances 7.2 7.6 –0.4 0.3 Main source of income: Other 21.0 20.2 0.8 –4.6** Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). The value displayed for t-tests are the differences in the means between poor and non-poor households. The coefficients estimated from the logistic regression correspond to the marginal effects and include region and population fixed effects. The poverty status used in the regression was derived from total core consumption and a rescaled poverty line. FIGURE 1.18  n  Labor force participation men participate in the labor market, compared to only 37 percent of the women (p<0.01). Increas- Percent of population aged 15–64 100 ing participation of women in the labor market will be important to accelerate economic growth and 80 raise the living standards of Somali households. 60 The gender gap in labor force participation is pri- 40 marily a result of a larger share of women staying at home and caring for their families compared to 20 men. Women often tend to engage in unpaid care 0 and domestic work and therefore are less likely to participate in the labor market (Figure 1.19). Even u n em l om s s e e on or r ra oo t ad th ish ba al al en Po se Ru m M -P ur though 64 percent of the Somali households per- ad Fe er og N ttl N ceive that most or all women can work outside M O in the home, the gap in both labor force participa- Ps ID tion and employment between men and women is Active: employed Active: unemployed substantial (21 and 20 percentage points respec- Inactive: not enrolled Inactive: enrolled tively, p<0.01). Changing the perception of women together with removing barriers to work are cru- Source: Authors’ calculations based on the SHFS 2017–18. cial steps to tackle gender inequalities. Poverty Profile 21 FIGURE 1.19  n  Reasons for inactivity FIGURE 1.20  n  Cross-country comparison of literacy rate and GDP 100 100 Percent of inactive and not-enrolled population Literacy rate (% of population 15+ years) 80 90 60 TZA 80 UGA 40 70 MWI RWA 60 20 50 0 SOM ETH 40 u an em al ts s e e or r oo ad th ish al al r en Po 30 b se Ru m M -P om ur ad Fe on SSD er og N ttl N 20 M O in 10 Ps ID 0 In school 0 500 1,000 1,500 2,000 2,500 3,000 III/disabled GDP per capita (US$ PPP) Too young/old Waiting for busy season/on leave Source: Authors’ calculation from survey data and World Bank Open Family and household care Data. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 1.21  n  Literacy by age 65 Education 60 Percent of population literate 55 Around half of the Somalis can read and write, 50 with literacy being more common among younger 45 generations, urban population, and men. The 40 adult literacy rate for the population aged 15 years or more is 50 percent for Somalia. This rate is simi- 35 lar to the unweighted average of low-income coun- 30 tries in Sub-Saharan Africa (49 percent), and is in 25 line with the cross-country comparison after con- 20 sidering the level of GDP per capita (Figure 1.20). 46 6–9 55+ 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45–49 50–54 Younger generations are often more literate, with the highest rate of literacy found among those aged Age group 15–19 years (62 percent, Figure 1.21). The adult lit- eracy rate is 79 percent in Mogadishu, 68 percent Overall average in other urban areas (p<0.01 vs. Mogadishu), fol- lowed by IDPs in settlements (57 percent, p<0.01 Source: Authors’ calculations based on the SHFS 2017–18. vs. other urban), rural areas (45 percent, p<0.1) and by the nomads with the lowest literacy rate (49 percent), relative to poor households (43 per- (16 percent, p<0.01). For all the population groups, cent, p<0.05). literacy is higher for men compared to women (at least p<0.05 for each comparison from Figure Only one-third of primary school-aged (6–13) 1.22). The poor and non-poor population have a children are enrolled, which is very low by inter- similar literacy rate, yet the share of literate house- national comparisons. In Somalia, the share of hold members is higher for non-poor households children of primary school age (6–13) enrolled in school is 33 percent, which is less than half the unweighted average of low-income Sub-Saharan 46  The literacy rates presented in this analysis have some limita- tions, as they are nonfunctional and were self-reported by inter- countries (74 percent, Figure 1.23). For its level of viewed households. GDP per capita, Somalia should have a higher net 22  Somali Poverty and Vulnerability Assessment FIGURE 1.22  n  Literacy rate by group (aged 15+) 100 100 Percent of population aged 15+ 90 90 Percent of population aged 15+ 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 hu n al ts s n- or or ed H HH itt es s Ps s ffe d ed ad a ce P t a cte ur en ad H No Po po is rb m nc ct ID ID om R an ad he ed m ru gh ffe re ta n- e og e d N d it ou t a e No ttl al a ive em th M se M he Dr gh O r u in e ce d ro al e Ps Re eiv m td ID Fe No ec tr No Women Men Overall average Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 1.23  n  Cross-country comparison of net FIGURE 1.24  n  Net school enrollment rate by age primary school enrollment and GDP Percent of population enrolled in school 50 100 MWI RWA Enrollment (% of population aged 6–13) 90 UGA 40 ETH 80 TZA 30 70 60 20 50 10 40 SOM SSD 30 0 7 9 1 3 5 7 9 1 3 5 6– 8– –1 –1 –1 –1 –1 –2 –2 20 –2 10 12 14 16 18 20 22 24 0 500 1,000 1,500 2,000 2,500 3,000 GDP per capita (US$ PPP) Age Overall average Source: Authors’ calculation from survey data and World Bank Open Data. Source: Authors’ calculations based on the SHFS 2017–18. school enrollment rate, yet it has one of the lowest enrollment rates among this group of countries, Many Somali children start primary school late only after South Sudan (Figure 1.23).47 since a large share of the parents think children aged 6–9 years are too young to attend school. Net enrollment rates are low for children aged The net enrollment rate is the ratio of children of primary/ 47  6–9 years, and range between 22 and 30 percent secondary school age who are enrolled in school relative to the (Figure 1.24). For the children aged 10–19 years, the population of the corresponding age group. net enrollment rate increases and hovers around Poverty Profile 23 FIGURE 1.25  n  School enrollment by level and age FIGURE 1.26  n  Net enrollment of primary school- aged children 16 100 14 Percent of enrolled population 90 Percent of children aged 6–13 12 80 70 enrolled in school 10 60 8 50 6 40 30 4 20 2 10 0 0 10–11 12–13 14–15 16–17 18–19 20–21 22–23 24–25 l hu n al s 6–7 8–9 l ts ra ad a ur en is rb ve om R ad ru m O og e N e ttl th M se O Primary Secondary Tertiary in Ps School age ID Primary school Secondary school Girls Boys Overall Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. 40 and 47 percent. This suggests that many Somali FIGURE 1.27  n  Net enrollment of secondary school- children do not start school at age 6. Late enroll- aged children ment seems to be explained by the perception of Somali parents regarding the age at which chil- 100 dren should attend school. Seventy-three percent Percent of children aged 14–17 90 of parents reporting their children are not enrolled 80 in school because they were too young were refer- 70 enrolled in school ring to a child aged between 6 and 9. As a result, 60 27 percent of the children enrolled in primary 50 school are not aged 6–13 years, which corresponds 40 to the typical age for primary school (Figure 1.25). 30 Likewise, nearly half of the population enrolled in 20 secondary school are older than 17. The percep- 10 tion of parents is not associated with the fact that 0 some children would have to walk a long distance ll hu n al ts s ra ad ba ur en is ve to school, which might not be appropriate for om ur R ad em O er og N ttl their age, nor with the household’s own percep- th M se O in tion of safety for walking during the day. The share Ps ID of households with children of primary age that responded they are too young for school is smaller Girls Boys Overall among those located far (more than 30 minutes) from the closest school, compared to those that Source: Authors’ calculations based on the SHFS 2017–18. are located below the 30-minute threshold. Also, the share of households perceiving it was unsafe to walk during the day is similar among those that equally likely to be enrolled across the country. reported children aged 6–9 being too young for Enrollment of children aged 6–13 is highest in urban school and those who had other reasons for not areas (60 percent in Mogadishu and 55 percent in attending school. other urban), followed by similar rates in rural areas and IDPs (30 and 29 percent respectively, p<0.01 Net enrollment of primary school-aged children vs. Mogadishu and other urban) and finally by the is largest in urban areas, yet girls and boys are nomads (12 percent, p<0.01 for all comparisons). 24  Somali Poverty and Vulnerability Assessment TABLE 1.4  n  Factors associated with school enrollment Dependent variable: Net school enrollment Enrollment for Enrollment for population of population of Overall school primary school age secondary school age Independent variables enrollment (1) (2) (3) Male 0.391*** 0.169 0.388** Age N/A 0.228*** –0.069 Primary-school age (6–13) Reference group N/A N/A Secondary- school age (14–17) 0.610*** N/A N/A Tertiary-school age (18–25) –0.779*** N/A N/A Poor household 0.166 0.294 –0.269 Household receiving remittances 0.641*** 0.796*** 0.457 Household headed by men –0.224* –0.314* 0.095 Literate household head 0.414*** 0.551*** 0.401* Household expenditure on education per member 0.006** 0.006 0.007 enrolled School more than 30 minutes away –0.600*** –0.898*** –0.617 Observations 14,646 8,247 2,467 Source: Authors’ calculations based on the SHFS 2017–18. Note: N/A: not applicable. Significance level: 1% (***), 5% (**), and 10% (*). The coefficients were estimated from a logistic regression model with population and region fixed effects. The reported values correspond to the marginal effects. The poverty status was derived from total core consumption and a rescaled poverty line. However, there are no significant differences in net urban) and finally by nomads (12 percent, p<0.01 enrollment between boys and girls at the national for all comparisons). Moreover, at this age girls are level nor for each population group (Figure 1.26). less likely to enroll in school compared to boys, Differences in net school enrollment for the pop- after controlling for regional effects, age, and other ulation aged 6–13 years seems to be driven by factors associated with school enrollment (p<0.01 geographical disparities in terms of access and in Table 1.4). In other urban areas, where net enroll- availability of education. ment is highest, there is a gender gap in school enrollment of 15 percentage points (81 percent for The geographical disparities in net enrollment are boys vs. 66 percent for girls). The nomads with the also present among secondary school-aged chil- lowest net enrollment rates are at disadvantage, dren (14–17 years), but the differences between and together with the girls face the biggest chal- boys and girls are more pronounced. The overall lenges. Policy efforts should aim to increase enroll- share of children of secondary school age (14–17) ment rates while considering the disparities and enrolled is 45 percent (Figure 1.27). Regional dif- needs of different vulnerable groups. ferences are large as secondary school enrollment ranges from 12 to 77 percent. Net enrollment is At age 14 to 17, the main reason for boys for not highest in urban areas (77 percent in Mogadishu attending school is the lack of money and for and 75 percent in other urban), followed by simi- girls it is having to work or help at home. The lar rates in rural areas and IDPs (44 and 36 per- reasons for not attending school vary with the cent respectively, p<0.01 vs. Mogadishu and other age and gender of children. For those of primary Poverty Profile 25 FIGURE 1.28  n  Reasons for not attending school for FIGURE 1.29  n  Reasons for not attending school for children of primary age (6–13) children of secondary age (14–17) 100 100 Percent of children aged 80 Percent of children aged 6–13 not enrolled 80 14–17 not enrolled 60 60 40 40 20 20 0 0 l u n em al N nts s s ys on r r l N Poo oo ra ad irl th dish ba r Bo se Ru G ve e -p om ur l u n em l N nts s s ys or r l a O a oo ra ad irl th ish ba er r og Po Bo se Ru ttl G ve e -p om ur ad M on O O er og in ttl N M Ps O in ID Ps ID Other Work/help at home No schools nearby Lack of money Other Work/help at home Too young No schools nearby Lack of money Too young Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. school age, the main reason given by the parents controlling for regional effects, age, gender, and is that children are too young (52 percent), fol- other factors associated with school enrollment lowed by lack of money (18 percent). The pattern (p<0.01 in Table 1.4). This relationship is robust as is similar for boys and girls (Figure 1.28) as well as the same result is found for children of primary among the population in Mogadishu, other urban school age (p<0.01) and those of secondary school and rural areas, and IDP settlements. Yet, nomadic age (p<0.1). Increasing access to education for chil- households reported the lack of schools nearby as dren and youth will allow them to achieve produc- the second largest issue. For children of secondary tive opportunities in their adult life and enhance school age (14–17 years), the main reason is the lack their consumption levels. This challenge will con- of resources (27 percent), followed by others, and tinue to grow, given the demographic structure of having to work or help at home (19 percent).48 By Somalia and its overall young population. the age of 14–17, girls are more likely to be working or helping at home (Figure 1.29) and not attend- Distance from schools rather than the costs of ing school, compared to boys (15 percentage point schooling affects the enrollment of children. difference, p<0.01). The distance to school and the cost from sending children to school are important factors influenc- School enrollment is associated with the literacy ing this decision. For 1 out 3 Somali households, of the household head in Somalia. Education is schools are at least 30 minutes walking distance a key tool for increasing the levels of welfare and (Figure 1.30), and they are not far for urban house- helping to break the poverty cycle. In Somalia, holds as only 6 to 10 percent of them are beyond net enrollment is associated with the educational the 30-minute threshold, compared to 73 percent level of older generations, as school enrollment for of nomadic households.49 Being more than 30 the population aged 6–25 years is more likely in minutes away from school is negatively associated households with a literate household head, after with enrollment for primary school-aged children and the overall enrolled population (p<0.01, Table 1.4). Another explanation for the low enrollment 48  Other includes too old, the lack of documents to enroll, that parents do not understand how to enroll their children, ill or sick, disabled, pregnant, insecurity, poor quality of schools, and Corresponds to how long it usually takes to walk (one way) 49  other not specified reasons. to the closest school. 26  Somali Poverty and Vulnerability Assessment FIGURE 1.30  n  Households more than 30 minutes FIGURE 1.31  n  Average household expenditure on away from the nearest school education per member enrolled 90 8 Percent of the poverty line 80 Percent of households 7 70 6 60 5 50 4 40 3 30 20 2 10 1 0 0 hu n al ts s or or hu n al ts s or or ad a ad a ur en ur en o Po Po o is rb is rb -p -p om R om R ad m ad m ru ru on on e e og N og N e e ttl ttl N N th th M se M se O O in in Ps Ps ID ID Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 1.32  n  Educational level rates is the cost associated with sending the chil- dren to school. The poor are twice as likely to 100 report lack of resources as the main reason for not Percent of population 80 sending their children to primary and secondary aged 6 or more school (p<0.01). On average, households spend 60 around 3 percent of the poverty line on education 40 per household member enrolled (e.g., tuition, fees, books, and uniforms) (Figure 1.31).50 These costs 20 seem affordable since they represent a small frac- tion of the poverty line. Expenditure on education 0 ll hu n al ts s is weakly correlated with enrollment and is only ra ad ba ur en is ve om ur R ad em O significant for the overall enrollment rate (p<0.05) er og N ttl th M se O but not for those of primary or secondary age. in Ps Efforts aimed at increasing school enrollment ID should address the barriers specific to each group; Other for nomads the availability of schools, in rural areas University and IDPs both availability and accessibility in terms Complete secondary of costs, while for urban areas one needs to better Complete primary/including secondary analyze the reason for prohibitive costs. Incomplete primary No education Gender and regional disparities in access to edu- cation are reproduced in educational outcomes Source: Authors’ calculations based on the SHFS 2017–18. of the Somali population. Low levels of school enrollment are associated with low levels of edu- but did not complete the level and only 7 per- cational attainment. Overall, 60 percent of the cent completed primary but not secondary (Fig- Somalis aged six years or more do not have any ure 1.32). The share of population without formal formal education, 21 percent reached primary education in rural areas is 1.6 times higher than in urban areas (p<0.01), 2.5 and 2.6 times higher for nomads and IDPs in settlements (p<0.01 respec- 50  Corresponds to educational expenses in tuition, fees, sta- tionary, books, school uniforms, and other expenses excluding tively), compared to the same group of urban pop- school meals. It does not include transportation costs, meals, ulation. Also, not having formal education is more and other associated costs from sending the children to school. likely for women compared to men (Figure 1.33). Poverty Profile 27 FIGURE 1.33  n  Population without formal education generations, to ultimately inform policies aimed at achieving better educational outcomes for the 100 Somali population. 90 Percent of population 80 70 60 Quality of dwellings and access 50 to services 40 30 Poor households are more likely to have a floor of 20 mud or wood, less likely to have a roof of metal sheets, and equally likely to use a charcoal or s s s s s s s s s + –1 ear –2 ar –2 ar –3 ar –3 ar –4 ar –4 ar –5 ar ar 55 20 ye 25 ye 30 ye 35 ye 40 ye 45 ye 50 ye ye wood stove. Most Somali households have a floor y 4 9 4 9 4 9 4 9 4 –1 of mud, wood, or other material (43 percent), a 10 15 roof of metal sheets (57 percent) and use a char- Women Men Overall average coal stove (47 percent) or woodstove (20 percent, Figure 1.36). A floor of cement is more common in urban areas, while for nomads and IDPs in settle- Source: Authors’ calculations based on the SHFS 2017–18. ments a floor of mud, wood, or other material (Fig- ure 1.34). The characteristics of the dwellings are There are no significant differences between poor different between poor and non-poor households: and non-poor population. Yet, not having educa- 46 percent of the poor have a roof of mud, wood, tion is more likely for those living in a household or other material, compared to 37 percent of the that did not receive remittances (62 percent vs. 45 non-poor households across Somalia (p<0.05). percent, p<0.01) and for households that live in IDP The poor tend to have a roof made of harar, raar, settlements or outside these settlements but were plastic sheets, and other material (p<0.01), while displaced (88 percent vs. 46 percent, p<0.01). The the non-poor households tend to have a roof of results are significant after controlling for regional metal sheets (p<0.05, Figure 1.35). This information differences, and personal and household charac- on dwelling characteristics can be used to target a teristics (p<0.05). In urban areas where access social protection program, by selecting beneficia- to education is more widespread, 11 percent of ries based on these easily identifiable features. the population aged 15 or more were previously enrolled but did not complete the primary level. Around 5 of 10 Somali households have access to While access is still a big challenge for most Soma- improved sanitation, which is less likely for the lis and a crucial first step, other policies will have to poor and nomadic households. Sanitation is criti- be considered in urban areas to reduce the drop- cal for the health of the members of the house- out rates and increase the levels of educational hold, as poor hygiene conditions can lead to lower attainment in primary and secondary levels. productivity in work. Forty-six percent of Somali households have access to improved sanitation Some improvements in educational outcomes can (Figure 1.37).51 For its level of GDP per capita, be seen across generations. Despite large gender Somalia has high access to improved sanitation and geographical disparities in terms of access relative to other low-income Sub-Saharan African and availability of education, younger genera- countries. Its share of households with access to tions tend to have better educational outcomes. improved sanitation is comparable to countries Not having formal education is more likely for the with a GDP per capita that is 2–3 times higher (Fig- population aged 40 years or more (Figure 1.33). ure 1.39). The share of households with access is Sixty-three percent of Somalis aged 15–19 have highest in urban areas (Mogadishu 69 percent and some formal education compared to 26 percent of those aged 50–54. In line with this, younger gen- erations are often more literate. Fifty-two percent Access to improved sanitation refers to those facilities that 51  of Somalis aged 30–34 are literate, while only 38 are not shared, and are likely to ensure hygienic separation of human excreta from human contact. They include flush/pour percent of those aged 50–54 are literate. The gov- flush (to piped sewer system, septic tank, pit latrine), ventilated ernment should try to explore and learn from the improved pit (VIP) latrine, pit latrine with slab, and composting drivers behind the improvements seen in younger toilet. 28  Somali Poverty and Vulnerability Assessment FIGURE 1.34  n  Type of floor FIGURE 1.36  n  Type of cooking source 100 100 Percent of households Percent of households 80 80 60 60 40 40 20 20 0 0 l hu n em l ts s or r l a oo ra ad a l hu n em l ts s or r ur en l a Po oo ra is rb ad a ve ur en -p Po om R is rb ad ve ru -p om R on O ad ru og on N O e ttl N og N th e ttl M se N th O M se O in in Ps Ps ID ID Cement Mud, wood and other Tiles (ceramic) Charcoal stove Woodstove and other Gas stove Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 1.35  n  Type of roof 100 FIGURE 1.37  n  Access to improved sanitation Percent of households 80 100 Percent of households 60 80 40 60 20 40 0 20 ll hu n al ts s or r oo ra ad ba 0 ur en Po is ve -p om ur R ad em on O hu n al ts s or r er og oo N ad ba ttl ur en N Po th is M se -p om ur R ad em O on in er og N ttl N th Ps M se O ID in Ps ID Metal sheets Harar, raar, plastic sheets and others Overall average Tiles Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. other urban 70 percent), followed by rural areas Almost 8 of 10 households have access to (40 percent, p<0.01 vs. urban areas) and IDPs in improved water sources, but spatial differences settlements (51 percent, at least p<0.05 vs. Moga- are also large. Inadequate sources for improved dishu and other urban), and lowest for the nomadic drinking water increase the water-borne illnesses, population (8 percent, p<0.01). IDPs in settlements which is particularly concerning for children given tend to have a different range of services and the impact health issues can have on their educa- thus they do not rank lowest. Poor households are tional attainment and learning process.52 Seventy- slightly less likely to have access to improved sani- seven percent of Somali households have access tation (43 percent) compared to non-poor house- holds (51 percent, p<0.1). 52  HM Government (2014). Poverty Profile 29 FIGURE 1.38  n  Access to improved water sources FIGURE 1.40  n  Cross-country comparison of access to improved water sources and GDP 100 Percent of households 100 80 Access to improved water sources MWI 60 90 (% of households) 40 80 UGA 20 SOM RWA 70 0 60 SSD hu n al ts s or or TZA ad a ur en Po o is rb -p om R ad m ETH ru on e og N e ttl N 50 th M se O in Ps 40 ID 0 500 1,000 1,500 2,000 2,500 Overall average GDP per capita (US$ PPP) Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18 and World Bank Open Data. FIGURE 1.39  n  Cross-country comparison of access to improved sanitation and GDP The share of households with improved drink- 70 ing water sources is similar between other urban Access to improved sanitation 60 RWA households and those living in IDP settlements (80 and 85 percent respectively), and between 50 SOM rural households and nomads (68 and 67 percent (% of households) 40 respectively). Compared to other urban areas and MWI IDP settlements, access to improved water sources 30 ETH is lowest for the nomads (p<0.1 and p<0.01 respec- 20 tively). Moreover, there are no significant differ- UGA ences between the share of poor and non-poor TZA 10 households with access. Variation in access seems SSD to be determined by the location of the household 0 0 500 1,000 1,500 2,000 2,500 3,000 rather than by their poverty status. GDP per capita (US$ PPP) Half of the households have access to electricity, but access is concentrated among urban residents Source: Authors’ calculations based on the SHFS 2017–18 and World and the non-poor. Fifty-two percent of the house- Bank Open Data. holds have electricity (Figure 1.41). In line with the access to other services, Somalia has a relatively high to improved drinking water sources (Figure 1.38).53 share of households with access to electricity for its Somalia is slightly above average in terms of level of GDP per capita. Countries in Sub-Saharan access, after controlling for GDP per capita, com- Africa with slightly higher GDPs per capita have less pared to other low-income Sub-Saharan African than 15 percent of their households with access to countries (Figure 1.40). Access is almost universal electricity (Figure 1.42). Access to this service also for households in Mogadishu (98 percent, p<0.01). varies considerably across population, with almost universal coverage in Mogadishu (98 percent), fol- lowed by other urban areas (p<0.01 vs. Mogadishu), 53  Access to an improved water source refers to using an then IDP settlements (49 percent), rural areas (32 improved drinking water source, which includes piped water percent, p<0.05 vs. IDP settlements), and finally by on premises (piped household water connection located inside nomads (12 percent, p<0.01 vs. rural areas). More- the user’s dwelling, plot, or yard), and other improved drinking water sources (public taps or standpipes, tube wells or bore- over, poverty is correlated with access to electricity. holes, protected dug wells, protected springs, and rainwater Overall, 47 percent of poor households have access collection). to electricity compared to 61 percent of non-poor 30  Somali Poverty and Vulnerability Assessment FIGURE 1.41  n  Access to electricity FIGURE 1.43  n  Households more than 30 minutes away from the nearest market 100 Percent of households 100 80 90 60 80 Percent of households 40 70 60 20 50 0 40 hu n al ts s or or ad a ur en 30 Po o is rb -p om R ad m ru on e og N e 20 ttl N th M se O in 10 Ps ID 0 hu n al ts s or or Overall average ad a ur en Po o is rb -p om R ad m ru on e og N e ttl N th M se O Source: Authors’ calculation based on the SHFS 2017–18. in Ps ID FIGURE 1.42  n  Cross-country comparison of access Overall average to electricity and GDP Source: Authors’ calculation based on the SHFS 2017–18. 80 Access to electrcity (% of households) 70 FIGURE 1.44  n  Households more than 30 minutes 60 away from the nearest health clinic 50 SOM 100 40 90 30 ETH 80 Percent of households RWA TZA 20 70 UGA 10 60 MWI SSD 50 0 0 500 1,000 1,500 2,000 2,500 3,000 40 GDP per capita (US$ PPP) 30 20 Source: Authors’ calculations based on the SHFS 2017–18 and World 10 Bank Open Data. 0 u an al ts s or or ad h ur en Po po is b om ur R ad em - on households (p<0.01). The difference between these er og N ttl N th M se O two groups of households is also present in other in Ps urban areas (with a 7-percentage point difference, ID p<0.05), rural areas (18 percentage point difference, Overall p<0.05) and IDPs in settlements (30 percentage points difference p<0.01). Nomadic poor and non- poor households have a similar share of households Source: Authors’ calculation based on the SHFS 2017–18. with access to electricity. 30 minutes to walk there (Figure 1.43).54 A similar Markets and health clinics are far (more than 30 share (40 percent) are far from the closest health minutes away) for 34 to 40 percent of Somali clinic or center (Figure 1.44). Due to the pastoralist households and for most of the nomads. Thirty- four percent of the Somali households are far from the closest market as it takes more than 54  Corresponds to how long it usually takes to walk one way. Poverty Profile 31 lifestyle, 8 in 10 nomadic households walk at least Monetary poor households are more deprived 30 minutes to the closest food market (82 per- than non-poor in many nonmonetary dimensions cent) and health clinic (83 percent), respective- except water. The educational dimension considers ly.55 At the national level, there are no differences school enrollment of children and the educational between the share of poor and non-poor house- level of adults in the household. Other dimensions holds located above the 30-minute threshold from include access to improved sources of drinking a food market or health clinic. The accessibility of water, access to improved sanitation, and access markets and health services seems to be associ- to electricity (Box 6). Sixty percent of house- ated with spatial differences and not the poverty holds are deprived in education and 23 percent in status of households. improved drinking water, 56 percent in improved sanitation, and 48 percent in electricity. For all the dimensions, the highest levels of deprivations are Multidimensional deprivations found among the nomadic population and the low- est among urban residents. Also, poor households are slightly more deprived than non-poor ones in Poverty is manifested along various dimensions the educational dimension (p<0.1, Figure A.10). beyond the monetary component as almost 9 of 10 Somali households are deprived in multiple Deprivation in multiple dimensions is consistent dimensions. Due to the lack of data, the Human with monetary poverty. The average number of Development Index has not been constructed for deprivations—excluding the monetary component— Somalia.56 However, the United Nations Develop- for households classified as monetary poor is 2.0, ment Programme (UNDP) estimates life expec- compared to 1.7 of non-poor ones (p<0.01). Mon- tancy at birth to be 56 years in 2015, which is etary poverty is correlated with multiple depriva- similar to the life expectancy of countries that rank tions, since around 40 percent of poor households 178–180 (out of 188) in the Human Development are also deprived in at least one of the other four Index. Deprivation of households is considered dimensions: education, water, sanitation, and along five dimensions: education, water, sanitation, electricity. For other urban areas, multiple depri- electricity, and monetary poverty (Box 6). Overall, vations are consistent with lower monetary pov- 72 percent of households are deprived in two or erty and higher access to services than in other more of these dimensions. Forty-three percent of regions. When considering other dimensions, IDPs households in Mogadishu and 47 percent of other in settlements do not rank last, as with monetary urban areas are deprived in at least two dimen- poverty, due to a larger share of households with sions, compared to 78 percent of IDPs in settle- access to services in IDP settlements. Contrary to ments, 96 percent of rural households and all the this, nomads are more deprived beyond the mon- nomadic population (Figure 1.45). Between 5 and etary dimension since they lack access to most of 11 percent of households are deprived in all the five these key services. dimensions among rural and IDPs in settlements. In addition, Wave 2 of the SHFS collected video tes- Multiple deprivations are correlated with the lit- timonials from households that volunteered. Hun- eracy and gender of the household head. The dreds of video testimonials were recorded during characterization of the monetary poor is similar to fieldwork, capturing the voice of the world’s least that of households deprived in various dimensions. represented people and giving a face to the data. A larger household size and age dependency ratio They have subtitles in English and can be accessed is associated with households deprived in educa- on the Somali Pulse website (Box 1). tion and the total number of deprivations, even after controlling for regional differences and other household characteristics (Table 1.5). Multiple deprivations are also more likely to affect house- holds headed by men in every dimension except 55  The World Health Organization (WHO) and the United Nations water dimension, as well as their total number Children’s Fund (UNICEF) define a water source 30 minutes of deprivations. Households headed by a literate or further as limited (see https://washdata.org/monitoring/­ member are less likely to be deprived in every drinking-water). Also, several studies like Fisseha, et al. (2017) dimension (p<0.01). The strong gender and educa- and Dar and Khan (2011) consider 30 minutes as the cutoff for tional effect of household heads on multiple depri- services being too far. 56  http://hdr.undp.org/en/composite/HDI vations confirms the existing inequalities among 32  Somali Poverty and Vulnerability Assessment Box 6  ■  Multiple deprivations Nonmonetary dimensions of poverty are considered to present a comprehensive profiling of welfare condi- tions. Poverty has more than one dimension and therefore households are assessed along several types of depri- vations, beyond their monetary poverty status, in which a household is classified as poor if their daily per capita consumption expenditure is lower than the international poverty line of US$1.90 at 2011 PPP. The education of children and adults in the household is another key dimension in which households can be deprived. Education is crucial to improve welfare conditions due to its associated externalities and a higher expected income. Households are considered deprived if (i) at least one child (aged 6–14 years) does not attend school, or if (ii) all the adults (aged 15 years or more) in the household have no education. The other dimensions include access to improved water, improved sanitation and electricity. Access to improved sources of drinking water & sanitation are relevant for health outcomes, educational attainment and productive activities. If the household does not have access to improved sanitation and improved sources of drinking water, it’s considered deprived in the particular dimension. FIGURE 1.45  n  Number of multidimensional deprivations 100 Percent of households 80 60 40 20 0 ll hu n al ts s ra ad ba ur en is ve om ur R ad em O er og N ttl th M se O in Ps ID None One Two Three Four Five Source: Authors’ calculation based on the SHFS 2017–18. the Somali population that need to be considered is least deprived in access to improved drinking in poverty reduction efforts. water. Only 2 percent of households in Mogadi- shu do not have access to water, while 20 per- Nomadic households have more nonmonetary cent of households in other urban areas do not deprivations than other Somali households. have access. Poor households are more deprived Nomadic households have a high incidence of in nonmonetary dimensions than the non-poor deprivation in improved sanitation (89.4 percent), (Figure 1.47). While 65 percent of the poor house- education (88.7 percent), and access to electric- holds are deprived in education, only 52.5 per- ity (87.7 percent) (Figure 1.46). Rural households cent of the non-poor households are deprived. and IDPs in settlements also experience high lev- Poor households are 14 percentage points more els of deprivation in these dimensions, while urban deprived in access to electricity than the non-poor residents have better access. Further, Mogadishu households. Poverty Profile 33 TABLE 1.5  n  Multiple deprivations and demographic attributes of poor households Dependent variable: Multiple deprivation Total no. of Total no. of deprivations deprivations excluding including monetary monetary Education Water Sanitation Electricity poverty poverty Independent variables (1) (2) (3) (4) (5) (6) Household size 0.262*** –0.106** 0.066 –0.055 0.069* 0.227*** Age dependency ratio 0.619*** –0.057 0.059 0.261** 0.267*** 0.254*** Household headed by men 0.462** 0.171 0.419** 0.451** 0.476*** 0.493*** Age of household head 0.001 0 0.006 0.008 0.004 0.003 Literate household head –0.866*** –0.511** –0.502*** –1.03*** –0.915*** –0.859*** Observations 6,050 6,050 6,050 6,050 6,050 6,050 Source: Authors’ calculation based on the SHFS 2017-18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Columns 1 to 3 refers to logistic regressions, while for column 4 and 5 to ordered logistic regressions. The coefficients correspond to the marginal effects and include population and region fixed effects. FIGURE 1.46  n  Deprivations in various dimensions FIGURE 1.47  n  Nonmonetary deprivations by poverty status 100 Percent of households deprived households deprived 80 80 70 Percent of 60 60 40 50 40 20 30 0 20 10 hu n al ts s ad ba ur en is om ur R ad 0 em er Education Water Sanitation Electricity og N ttl th M se O in Poor Non-Poor Ps ID Monetary Education Water Source: Authors’ calculation based on the SHFS 2017–18. Sanitation Electricity Source: Authors’ calculation based on the SHFS 2017–18. 34  Somali Poverty and Vulnerability Assessment CHAPTER 2 Spatial Variation in Living Standards KEY MESSAGES Urban areas generally provide higher standards of to cope with the constant and large influxes of the living and better access to services than rural areas displaced, and have been unable to keep up with the except for access to land and housing. Poverty inci- provision of land, housing and basic services that are dence across urban areas is lower at 64 percent com- acutely needed. pared to the overall average of 69 percent, 72 percent in rural areas, and 76 percent among the IDPs. The Significant interurban regional disparities exist as only exception is Mogadishu, where poverty is higher well, which are often greater than the urban-rural than the overall average and is similar to rural areas. divide. While urban areas fare better than rural areas Food poverty is also lower in urban areas. Compared on average, Mogadishu, NE urban, and NW urban pro- to the overall average of 49 percent and rural areas vide better access to services compared to Baidoa, at 58 percent, urban areas fare better on average at Kismayo, and Central urban. Poverty is higher in 41 percent. Mogadishu than all urban areas except for Baidoa, but access to basic services such as electricity, water, Cities consistently provide better access to ser- sanitation, improved housing, education, and health vices and more stable income sources than rural is better in Mogadishu than in other urban areas. areas except for land and housing. Access to elec- Kismayo has the lowest poverty incidence and pov- tricity, water, improved sanitation, health, education, erty gap, yet fares poorly on other services. Strikingly, improved housing, and the Internet, is consistently access to water, literacy, enrollment and employment higher in urban areas irrespective of people’s levels are significantly better in IDP settlements than in of poverty or whether they are IDP or female-headed Kismayo. Baidoa has high levels of monetary and non- households. The only area where rural areas fare bet- monetary poverty and correspondingly low levels of ter than urban areas is the tenure of land and housing. access to services. NE and NW urban fare relatively Due to land scarcity and high land values in urban well in access to services, while Central urban lags. areas, urban households are less likely to own their The likely explanation is that Mogadishu enjoys its land and houses. Somali cities also provide more economic capital and more assistance given its sta- wage labor employment and better access to remit- tus as the capital city. North East and North West are tances. Fifty-two percent of the urban residents rely more developed because they have been relatively on wage labor as their main income source, except free of violent conflict, public institutions are more for Jubbaland, while 42 percent of the rural residents established, and more aid has been flown in. Kismayo, rely on agriculture and family businesses. Since urban Baidoa, and Central urban likely suffer from lower lev- wage labor is not climate dependent and provides els of development as they have only recently been a more stable stream of income, it is less risky than liberated from Al-Shabab, and their subnational gov- agriculture or family businesses. Urban households ernments are still nascent. also have better access to remittances from abroad and better opportunities to borrow money. IDPs in urban areas fare better than rural IDPs in terms of access to services, but still lag behind The relatively better conditions in urban areas com- other non-IDP households. Within cities, urban IDPs pared to rural areas, however, should not mask the that live outside of IDP settlements (non-settlement low base cities are at. The situation is exacerbated IDPs) fare as well as IDPs that live in IDP settlements by the influx of the IDPs. Even though urban areas (settlement IDPs), but both groups are better-off perform relatively better in poverty and access to ser- than rural non-settlement IDPs. Thus, irrespective of vices than rural areas, Somali cities still struggle with whether IDPs live in IDP settlements or not, so long high absolute poverty (64 percent), nonmonetary as they live in urban areas, there is no significant poverty (41 percent), hunger and low levels of access difference in their standards of living. On the other to services. With many new IDPs moving into cities, hand, urban non-settlement IDPs are consistently the pressure on land, housing, and services is increas- worse off than other urban households. Urban non- ing. In many cases, urban centers have ben unable settlement IDPs have less access to electricity, piped —continued Spatial Variation in Living Standards 35 KEY MESSAGES—continued water, improved sanitation, improved housing, dwell- better land management and coordinated infra- ing ownership, and the Internet compared to other structure investments, which are the fundamental non-IDP urban households. Moreover, urban non-­ elements of cities. For Somalia to reap the benefits settlement IDPs suffer from lower enrollment, literacy, of urbanization, the government needs to invest in and employment rates. They also tend to live further two core elements of cities—land and coordinated away from primary schools and food markets. Thus, infrastructure investments. The fundamental ele- urban non-settlement IDPs are worse off than the rest ment in making Somali cities work is to establish a of the urban population as they have likely become proper land administration system and effective land deprived of their former livelihoods, assets, and social use planning. This will allow for a more controlled networks due to displacement and they have more growth of the city and provision of security of tenure limited access to services. Moreover, they are at a dis- to the IDPs, many of whom prefer to settle in cities. advantage in levels of education, which may prevent The other important element is to make coordinated them from finding good jobs. infrastructure investments—rather than ad hoc single sector interventions—aligned with the land use plan To ensure Somali cities can reap the benefits of to take advantage of the synergy across different urbanization, the government needs to invest in types of infrastructure. Somalia’s urban population is growing rapidly Mogadishu and Kismayo are considered two of the partly because of significant forced migration into world’s five most fragile cities in the world.59 urban areas caused by protracted conflicts, inse- curity, and cyclical natural disasters. The current This chapter seeks to examine the spatial variation urban population is estimated at around 5.2 million in standards of living and inform how the govern- people (42 percent) with a growth rate of around ment can better reap the benefits of urbanization 4 percent per annum.57 If the current trend per- in Somalia. The data compare the various aspects sists, by 2030, Somalia will add another 4.5 million of living standards between urban and rural areas, urban residents to its already constrained urban the regional variation across different urban areas, environment, nearly doubling current numbers.58 and among different urban population groups. If managed well, urbanization can help manage risks and contribute to stability in Somalia. Yet, Urban-rural comparison as urban areas fail to keep pace with the rapid urbanization, Somalia’s cities are becoming more fragile. With a greater concentration of people, Monetary and nonmonetary poverty capital, and assets, cities are better equipped to Poverty incidence is lower in urban areas com- provide anonymity and better access to security, pared to rural areas, except for Mogadishu, and services, and jobs than rural areas. Somali cities North West has the largest intra-regional urban- are growing rapidly as they serve as a safe haven rural divide. The poverty incidence across urban for people who seek refuge. Yet, they have not areas (including Mogadishu that has a poverty been able to cope with the increased demands for incidence of 72 percent) is lower at 64 percent land, housing, basic services, and jobs. As a result, compared to the overall average of 69 percent, Somali cities are expanding in a haphazard manner 72 percent in rural areas, and 76 percent among and slums are growing. Large influxes of people the IDPs (p<0.10 vs. rural areas). The intra-regional are also disrupting the social cohesion. Indeed, urban-rural divide is the largest in the North West region at 15 percentage points followed by the World Economic Forum (12 January 2017). ”These are the 59  Fragile Cities in the World and This Is What We Have Learned United Nations Population Fund (UNFPA) (2014). 57  from Them”. (https://www.weforum.org/agenda/2017/01/these- World Bank staff calculation based on UN-Habitat and CIA 58  are-the-most-fragile-cities-in-the-world-and-this-is-what-we- World Factbook. have-learned-from-them/) 36  Somali Poverty and Vulnerability Assessment Box 7  ■ Hypotheses Hypothesis 1: Cities provide more opportunities for the vulnerable. When there is no possibility of staying in rural areas because of conflict, drought or famine, people are likely to move to cities. Cities offer more diverse socio-economic opportunities for the poor. Hypothesis 2: Cities lower uncertainty for the vulnerable. In addition to poverty, famine, health and safety issues, uncertainty is a major driver of concern for the most disadvantaged. Urban populations mostly enjoy a lower level of uncertainty than those living outside of the city: quality of information is higher, and security is better. Prices are more stable, there is more reliable access to food, services, dwelling tenure, easier access to remittances, and easier access to humanitarian aid. Hypothesis 3: Cities allow for risk reduction/diversification for the vulnerable. Since the city offers more options and opportunities, people do not need to rely on weather-dependent jobs such as agriculture. It is also easier to increase household income with a second (or multiple) jobs or better paying and more stable jobs. Risks of violence may be lower as security is better, the reach of government and state security would be higher, and residents can enjoy anonymity in urban areas which can improve their safety. FIGURE 2.1  n  Poverty incidence FIGURE 2.2  n  Food poverty incidence 100 100 Percent of population Percent of population 80 80 60 60 40 40 20 20 0 N th t ur u th as an th t u l C es an So Ce l u l h ra n Ju h W urb l ID bba es an se ur l em n N nts s or s ra tra ura ut st ra in nd ra 0 ad N Eas ish ut nt rba ttl ba or E b N We t ru So We l ru u W rb e om en t r Ps la t r ad st hu th as an or st ral C est n lu l ut ntra an ut st al bb es n in d u al em n om s s tra ura t ad W rba Ju W rba a or Mog So We rur Ps alan rur en Ea dis N th E urb W t ru rb se rb r u u l t a th og N or ttl e M en So e N C th h th or h or or N N N ID Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. South West region at 12 percentage points. The North East region has the lowest divide at 4 per- in urban areas compared to people in rural areas, centage points. Poverty in Mogadishu, however, is although a significant regional disparity exists. higher than the overall average or in rural areas at Between 8 percent (North West urban) and 45 72 percent (Figure 2.1). percent (Central urban) of the households in urban areas report having been hungry in the past four Urban areas also have lower food poverty com- weeks (Figure 2.3) compared to between 24 per- pared to rural areas. Compared to the overall aver- cent (North West rural) and 58 percent (North age of 49 percent and rural areas at 58 percent, East rural) among rural households. Unsurprisingly, urban areas fare better on average at 41 percent lack of financial resources to buy food in the past (Figure 2.2). Mogadishu has the highest food pov- week is higher in rural areas except in the North erty among all the other urban areas, which cor- West urban. Rural households also report higher responds to its high poverty incidence. Fewer levels of difficulty in borrowing money compared people report being hungry in the past four weeks to urban households to purchase food. Spatial Variation in Living Standards 37 FIGURE 2.3  n  Hunger over the past four weeks FIGURE 2.4  n  Access to electricity 100 100 Percent of households 80 Percent of households 80 60 60 40 20 40 0 20 N th ur u th as an th t u l C es an So Ce al u al h ra n ut st ral ID ba es an se ur l em an N nts s or s ra in nd ra ad N Eas ish ut nt rba tr ur or E b N We t ru W rb So We l ru Ju h W urb u b e om en t r Ps la t r ad 0 t or Mog ttl N as shu th ast n l C est n l ut ntr an ut st al bb es n in d u l em n om s s N We rura tra ura a t ad th N th E rba W rba Ju W rba a So We rur Ps alan rur en or So Ce urb se rb b E di r tu u u N al t a or st l og N ttl M en th h th or h Overall average or or N ID Source: Authors’ calculations based on the SHFS 2017–18. Overall average Source: Authors’ calculations based on the SHFS 2017–18. Access to services In all regions, urban households have significantly better access to basic services compared to rural FIGURE 2.5  n  Access to piped water households, irrespective of their poverty status or 100 whether they are IDPs or female-headed house- holds. Specifically, urban households fare better Percent of households 80 in access to electricity, water, and improved sani- tation than rural households. On average, 79 per- 60 cent of urban households have electricity at home 40 compared to 32 percent among rural households. Access to electricity is the highest among North 20 East urban households (99 percent). In rural areas, access is much lower with a minimum of 19 per- 0 or st u u or Ea an or st u l an lu l So en ban tu l Ju We ban in d u l em n ts cent in Central and a maximum of 48 percent in ra tra ra So es ura Ps lan ura h se rba en Ea dis b ru rb ru r r W lr r r the North East (Figure 2.4). The urban-rural divide st C est st a tra og ttl e W M W en in access to electricity is most severe in the North th a C th h th bb th h ut or ut N West. Moreover, access to electricity is correlated N N N ID with poverty, where the poor (the bottom 40 per- cent) have less access compared to the non-poor Overall average households (the top 60 percent). Source: Authors’ calculations based on the SHFS 2017–18. Between 45 percent (Jubbaland) and 96 percent (Mogadishu) of urban households have access to piped water at home. This is considerably higher households (Figure 2.7). While access to improved than rural households’ access, which ranges from sanitation is not correlated with the levels of pov- 1 percent (North West rural) to 43 percent (South erty, urban households are more likely to have West and North West rural) (Figure 2.5). There are access to improved sanitation compared to other no significant differences between the top 60 per- population groups. cent and the bottom 40 percent that have access to piped water in urban areas. For both urban and Urban households have better access to educa- rural households, the main alternatives to potable tion and health than rural households although piped water are boreholes and water trucks (Fig- Central, South West, and Jubbaland lag behind. ure 2.6). Sixty-seven percent of the households Literacy and enrollment rates are significantly have access to improved sanitation in all urban higher among the urban households compared areas compared to only about 39 percent of rural to the rural households. Proportion of households 38  Somali Poverty and Vulnerability Assessment FIGURE 2.6  n  Source of potable water FIGURE 2.8  n  Primary school enrollment rate 100 100 Percent of households Percent enrolled at 90 school age (6–13) 80 80 70 60 60 50 40 40 30 20 20 0 10 N th t ur u th as an th t u al C es ban So Ce al u al h ra n Ju h W urb l ID bb es an se ur l em n N nts s ut st ra in nd ura 0 ad h ut nt rba ttl ba or s r tr rur N Eas dis or E b N We t ru So We l ru e om Ps ala t r W r or st hu th as an th t u l C e ban So Ce al u ral So W al ru n Ju h W urb l ID bb es an se ur l em n om s s en t or s ra ut est ra in nd ura th ga N ent ad ut nt rba ttl ba N Ea dis or E rb N We t ru tr ru or Mo W r Ps ala t r N th u en st a or Mog h r or N th N Overall average Piped water Borehole, protected well Tanker truck, bottled Other Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. East rural, Central rural, IDPs, and nomads all fall below the overall average. Secondary school enroll- FIGURE 2.7  n  Access to improved sanitation ment rates among children aged 14–17 show a simi- 100 lar picture. Average enrollment rate in urban areas is 69 percent as opposed to 46 percent in rural Percent of households 80 areas. North East urban has the highest enrollment 60 rate of 90 percent, while the lowest is in Central 40 rural at 14 percent. This means that to address the structural constraint of overcoming poverty, there 20 is a need to improve educational opportunities for 0 children, particularly in rural areas and among IDPs and nomads. While poverty does not seem to be hu Ea ban l es n l n tu l Ju We an in d u l em n ts s es ra tra ra a Ps lan rura ad W ba So en ba a r en W t ru en t ru ru is b se rb correlated with enrollment in urban areas, belong- om ur th t ur C l ur r ad h ral st s st og N So es ttl ing to the bottom 40 percent is negatively corre- t Ea M W th ID ba h th C ut or th lated with literacy rate. Access to health facilities or b ut or or N N N N is better in urban areas as well, where on aver- Overall average age, 80 percent of the urban residents live within a 30-minute walk to a health facility compared to Source: Authors’ calculations based on the SHFS 2017–18. 37 percent of rural residents and 40 percent of the overall average (Figure 2.9). with literate household members is higher in urban Satisfaction over the quality of primary education areas (68 percent) than in rural areas (41 percent). and health services is higher in urban areas. Sat- On average, households with household members isfaction over the quality of primary education is enrolled in school in urban area are significantly generally higher in urban areas, with the highest higher (54 percent) than in rural areas (32 percent). satisfaction level of 95 percent in Mogadishu com- Disparities in enrollment rates, however, are large pared to the lowest satisfaction rate of 70 percent across regions as well as between urban and rural in Central rural. Satisfaction over the quality of areas. The highest enrollment rate in primary edu- health facilities is marginally higher in urban areas cation is in North East urban with 71 percent among as well, ranging from 81 percent (Central urban) children aged 6–13, while the lowest is 7 percent in to 92 percent, compared to that of rural between Central rural against the overall average of 33 per- 76 (South West rural) and 83 percent (North East cent (Figure 2.8). Curiously, South West rural has rural). a higher enrollment rate than its urban area. North Spatial Variation in Living Standards 39 FIGURE 2.9  n  Distance to health facilities (>30 minutes) FIGURE 2.10  n  Dwelling type 100 100 Percent of households Percent of households 80 80 60 60 40 40 20 20 0 N th u u th as an th t u al C es ban So C al u ral So W al r n Ju h W ur l ID bb es ban se ur l em n N nts s ut est ura in nd ura ad or st h h tr a ttl ba or s r 0 N Ea dis or E rb N We t ru tr ru ut n rb e om Ps ala t r W r en t th ga N as ishu th as an or est ral C est n l ut ntr ban t u al n in d al em an N nts s tra ura ad or Mo W rba Ju W rba So e rur r W t ru Ps la ru N th urb se urb e e om r So Ce l ur E d u W l ID bba st a a t or og e s ttl n or E N M en th h th or h ut N N Own Rent Squatting Other Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.11  n  Living arrangement Dwelling 100 Percent of households 80 Access to land and housing is more constrained in 60 urban areas. Since land is scarce in urban areas, it 40 is not surprising that urban households are statisti- cally less likely to own land and housing compared 20 to rural households. About 42 percent of urban 0 or ast shu th as an or est ral C es an So Ce l ur l a n ut st ral ID bba est n in nd ral em n N nts s households live in rented dwellings compared to tra ura ad ut ntr ba Ju W rba ttl ba b u W b So We l ru Ps la ru e E di om N th E ur tr ur en t r se ur u 12 percent in rural areas. The proportion of renters a og M W is the highest in Mogadishu at 71 percent, where th h th h or or N land values are also the highest (Figure 2.10). South N N West is an exception where a significant portion of rural households live in temporary shelters pro- Apartment/house Shared apartment/house Other vided by aid agencies (Figure 2.11). That said, on average, urban households have better access to Source: Authors’ calculations based on the SHFS 2017–18. improved housing than the rural households.60 Access to land and housing has been further con- absence of security of land tenure, IDPs are highly strained by the recent influx of the IDPs to cit- vulnerable to forced eviction. For example, over ies. Seventy-five percent of IDPs in Somalia are 109,000 IDPs living in informal settlements across thought to reside in urban centers, settling on pub- the country have been forcefully evicted between lic and private lands within and in the outskirts of January and August 2017, 77 percent of which are cities. The majority of the returnees from neighbor- concentrated in Mogadishu.62 Due to forced evic- ing countries such as Kenya and Yemen are consid- tion, many IDPs have shifted to the outskirts of cit- ered to have settled in cities as well. In Mogadishu, ies, causing uncontrolled urban sprawl. Fifty-five areas occupied by IDP settlements increased by 16 percent of IDPs in Mogadishu now reside in the percent between 2013 and 2017. In Kismayo, the periphery of the city. The area occupied by IDP set- IDP-occupied area increased over seven-fold (Fig- tlements in the fringes of Baidoa has increased by ure 2.12 and Figure 2.14), and in Baidoa, it has more 177 percent in 2017. The number of IDPs in Kismayo than tripled (Figure 2.12 and Figure 2.13).61 In the tripled in 2017, and most of them have settled out- side of the city.63 Such spatial sprawl makes service 60  Improved housing is defined as living in apartments, shared apartments, separate houses or shared houses. 62  Norwegian Refugee Council (2018). 61  World Bank (2018c). 63  World Bank (2018e). Urban chapter. 40  Somali Poverty and Vulnerability Assessment FIGURE 2.12  n  Area occupied by IDP settlements 700 660.4 600 570.0 547.7 500 Area (ha) 400 300 176.2 200 128.3 100 29.5 150.8 41.5 54.5 0 Aug-13 Dec-14 May-16 Sep-17 Feb-19 Baidoa Kismayo Mogadishu Source: UNOSAT and UN-Habitat. FIGURE 2.13  n  New IDP settlements in Baidoa FIGURE 2.14  n  New IDP settlements in Kismayo Source: Ipsos. Source: REACH. provision difficult and costly as new settlements better access (16 percent) than rural households are disconnected from the existing urban centers (3 percent). In the absence of the formal bank- and infrastructure networks. Spatial fragmenta- ing sector, mobile banking has filled the void for tion also inhibits IDPs’ access to jobs and prevents Somali households to receive remittances. Sev- cities from reaping the scale and agglomeration enty percent of urban households and 55 percent benefits. of rural households have mobile bank accounts. Surprisingly, South West rural (83 percent) has Access to finance higher access to mobile bank accounts than South West urban (64 percent). The urban-rural divide Only 10 percent of Somali households have access in access to mobile bank accounts is the largest in to a bank account, while the majority of both North East at 41 percentage points and smallest in urban and rural households have access to mobile Central at 5 percentage points. Interestingly, cen- bank accounts. Reflecting the lack of banking sec- tral and southern regions in Somalia, which tend to tor development, very few Somali households have be less developed than the northern regions, have access to a bank account. Urban households have higher access to mobile bank accounts on average Spatial Variation in Living Standards 41 (Figure 2.15). The explanation could be due to lack FIGURE 2.15  n  Access to bank accounts of access to formal bank accounts, where the more 100 developed regions of Mogadishu, North East and Percent of households North West have better access to formal banking 80 than the central and southern regions. 60 Very few households have managed to save 40 money in the past 12 months. On average, only 20 9 percent of both urban and rural households could 0 save money in the past 12 months (Figure 2.16). As u st n or st al C es an So Ce urb l ut ntr an t u ral ID ba st n in d u ral em n om ts s l a more urban households engage in wage labor, it is ad N Eas ish a e a a N We rur tra rur en N th E urb W rb So We l ru Ju h W rb Ps lan ru se rb ad u en t not surprising that more urban households could a t or og N a s ttl M save (13 percent) compared to rural households th th or th h ut b or (3 percent). North East urban has the highest pro- N portion of households that managed to save, while 0 percent managed to save in North West rural and Bank account Mobile bank account North East rural. Rural households in these regions fare worse in terms of savings than the IDPs liv- Source: Authors’ calculations based on the SHFS 2017–18. ing in IDP settlements (9 percent). This is likely because 75 percent of the IDPs live in urban areas, FIGURE 2.16  n  Households that saved money and they are more likely to be engaged in wage labor (even if informal) than rural households. This 100 does not reflect the amount saved however. Percent of households 80 60 Employment 40 Urban households rely on wage labor and remit- 20 tances from abroad while rural households rely on agriculture and small family businesses. On 0 average, urban households are more likely to be st hu th as an or st al C est n lu l ut ntra an ut st al ID bba st n in nd ral em an N nts s tra ura ad W rba e a N We rur So We rur Ps la ru Ea dis N th E urb rb Ju W rb se urb e om employed than rural households. Fifty-two per- r u u t l a og ttl cent of the urban households rely on wage labor M en So e C th h th or h as their main income source, except for Jubba­ or or N N land urban where there is a relatively large share of households (43 percent) that engage in small Overall average family business as the main livelihood. Urban households tend to receive more remittances Source: Authors’ calculations based on the SHFS 2017–18. from abroad than rural households except for the South West region. Twenty-six percent of the rural Perception on living standards households rely on agriculture and fishing while another 16 percent rely on small family businesses Urban and rural households’ perceptions on safety (Figure 2.17). On average 8 percent of both urban do not differ between urban and rural areas. The and rural households count on remittances as their perception that they feel “very safe” from crime main source of income. Perception on employ- and violence is virtually the same between urban ment opportunities is more positive across regions (48 percent) and rural households (49 percent). in urban areas where households report that their North West stands out as an outlier where both employment prospects are “better” or “much bet- urban and rural households feel much safer than ter” than six months before (Figure 2.18). Satisfac- other regions. Perception of safety is lower among tion over employment is equally higher among Jubbaland urban residents compared to the IDPs urban households compared to rural households. (Figure 2.19). 42  Somali Poverty and Vulnerability Assessment FIGURE 2.17  n  Main sources of income FIGURE 2.19  n  Safety from crime and violence 100 100 Percent of households households 80 Percent of reporting feeling safe 60 80 40 60 20 0 40 or st hu th as an ur l C es an So Ce l u al a n ut st ral ID bba est an in nd ral em n N nts s or st a ad ut ntr rba ttl ba N We rur tra rur Ps la ru N Ea dis or E b W b So We l ru Ju h W urb e 20 om N th ur se ur t en t a or Mog 0 th th h or st hu th as an u l C es an So e l u al a n ut st ral ID bba est an in nd ral em n N nts s or st a ad N ut ntr rba ttl ba N We rur tra rur N Ea dis or E b W rb So We l ru Ju h W urb Ps la ru e om N h ur se ur t en t a or Mog Salaried labor Remittances C th th t h Small family business Agriculture N Other Very unsafe Unsafe Neither/nor Source: Authors’ calculations based on the SHFS 2017–18. Safe Very safe Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.18  n  Perception of employment opportunities 100 FIGURE 2.20  n  Dispute resolution 80 100 households Percent of 60 80 households Percent of 40 60 20 40 0 20 0 or st hu th as an u l C es an So Ce l u al a n ut st ral ID bba est an in nd ral em n N nts s or st a ad ut ntr rba ttl ba N We rur tra rur N Ea dis or E b W rb So We l ru Ju h W urb Ps la ru or st hu N We st r n th t u al C es ban So Ce al u ral So W al ru n Ju h W ur l ID bb es ban se ur l em n N nts s e ut est ra in nd rura om N th ur se ur ad th a a h r a ttl ba or s ur t en t a N Ea dis or E rb tr ru ut nt rb e om or Mog W r N th u en t Ps ala t a or Mog th th h th N N Much better Better The same Worse Much worse Clan elders (Xeer) Religious leaders Informal court Police AMISOM Formal court Source: Authors’ calculations based on the SHFS 2017–18. Community leader Other Trust Source: Authors’ calculations based on the SHFS 2017–18. Urban households tend to rely more on police for dispute resolution, while rural households rely elders for dispute resolution. Only 9 percent of the more on clan elders. On average, 44 percent of all Mogadishu residents and 13 percent of North West urban and rural households rely on police for dis- urban residents rely on clan elders. In terms of pute resolution compared to 39 percent of all urban regions, both urban and rural households in South and rural households that rely on clan elders. A West (39 percent) and Central (45 percent) have more nuanced picture emerges with a closer look, the lowest reliance on police, though regional dis- however. Central rural households have the lowest parity is not as large as rural versus urban (Fig- dependence on the police (12 percent) followed ure 2.20). This signifies that where people rely on by nomads (14 percent). Most of the households in the police, people are less dependent on informal North West urban (77 percent) and Mogadishu (67 institutions such as clan elders and vice versa. percent) rely on the police. In general, rural house- holds tend to rely less on police compared to urban Levels of trust in different government institu- residents. Conversely, the majority of Central rural tions vary along the urban-rural divide as well (78 percent) and nomads (73 percent) rely on clan as along the regional divide. Mogadishu has the Spatial Variation in Living Standards 43 FIGURE 2.21  n  Trust in institutions (20 percent) than among urban households (15 percent), which is unsurprising (Figure 2.21). 100 Percent of households 80 General levels of trust are higher in rural areas compared to urban areas. On average, 71 percent 60 of rural households claim they trust other people 40 compared to 66 percent in urban areas. This is not surprising as rural areas tend to be more socially 20 cohesive while urban areas tend to be a mix of people from different origins. However, when bro- 0 ken down by regions, South West rural has the u th as n th t u al C est an l ut ntr ban ut st ral bb es n in d u al em n om ts s tra ura highest levels of trust (73 percent) followed sur- ad N Eas ish a Ju W rba a N We rur Ps alan rur en N th E urb W rb So We l ru se rb r So Ce l ur ad u t t a prisingly by Mogadishu (73 percent). Jubbaland t or Mog N or s ttl en urban and Central rural have the lowest levels of h th or h or trust for others (59 percent). In terms of regions, N ID Jubbaland has the lowest levels of trust (59 per- Local NGO Clan elders cent), followed by Central urban and rural (61 per- African Union/Unite International NGOs cent). North East urban and rural have the highest Opposition groups Federal Government of Somalia level of trust at 78 percent. Government of Somalil State government Diaspora Other (please specify) Source: Authors’ calculations based on the SHFS 2017–18. Taxes Urban households generally pay more taxes than highest level of trust in the Federal Government of rural households. About 23 percent of all urban Somalia (87 percent) while the overall level of trust households across regions pay taxes of some sort among both urban and rural households is 50 per- compared to only 10 percent of rural households. cent. Urban residents generally have a higher trust The largest proportion of residents that pay taxes for the federal government (62 percent) compared are in North East urban (40 percent) followed by to the rural households (47 percent).64 Trust in the North West urban (34 percent) and Mogadishu federal government is the lowest in Jubbaland (33 percent). Jubbaland urban is an outlier with a (51 percent) and Central (55 percent). The state very low proportion of households claiming to pay government enjoys the highest levels of trust in taxes (1 percent). This is lower than the nomads Jubbaland urban (29 percent) followed by North (2 percent) or the IDPs (8 percent). The highest East urban (26 percent). The lowest is in Moga- proportion of rural households paying taxes are in dishu (2 percent) though this could have been North East (21 percent) while the lowest is in Cen- because they do not have a state government per tral rural (0.8 percent). In terms of regional aver- se, but rather the Banadir Regional Administra- age that combines both urban and rural areas in tion.65 On average, urban residents across regions respective regions, Mogadishu scores the highest have higher trust in the state government (16 per- (33 percent), followed by North East (30 percent). cent excluding Mogadishu and North-West urban) The lowest is in Jubbaland (1 percent) followed by compared to rural residents (12 percent).66 Central Central (6 percent) (Figure 2.22). has the lowest trust in the state government across both rural and urban residents (5 percent) com- More rural households pay taxes to the federal pared to the highest in Jubbaland urban (29 per- government while more urban households pay cent) followed by North East (24 percent). Trust in taxes to the local government. Thirty-seven percent clan leadership is higher among rural households of the rural households across regions pay taxes to the federal government compared to 31 percent of the urban households. Conversely, 59 percent of the 64  This figure excludes that of North West urban residents who urban households across regions pay taxes to the have 0 percent trust in the federal government for political local government (district government) compared reasons. to 36 percent of the rural households. One possible 65  North West is excluded as they see Somaliland government reason could be that local governments and decen- as the only legitimate political representative. 66  Excluding North West rural. tralized tax regimes are more developed in urban 44  Somali Poverty and Vulnerability Assessment FIGURE 2.22  n  Payment of taxes (6 percent).67 Percentage of households that pay taxes to the local government is highest among 100 North East urban (90 percent), followed by North 80 East rural (80 percent) and South West urban (71 Percent of households percent). The lowest is in South West rural (28 per- 60 cent), which is in contrast with South West urban where 71 percent of the households pay taxes to 40 the local government. Sixty percent of North West urban and 89 percent of North West rural house- 20 holds pay taxes to Somaliland government which is no surprise. Few households (0.7 percent) reported 0 that they pay taxes to Al-Shabaab (Figure 2.23). An u st n u l s n l ut ntr an ut st al bb es n in d u ral em n om ts s in-depth political economy analysis is warranted to or st a tra ra ad N Eas ish a W rba Ju h W rba a N We rur So We rur en N rth urb en t ru So Ce urb Ps alan t ru se rb ad u better understand the variation in who people pay al t l or og N a e ttl or E M taxes to. th th C th h o N ID Overall average Inter-urban comparison68 Source: Authors’ calculations based on the SHFS 2017–18. Monetary and nonmonetary poverty FIGURE 2.23  n  Institutions that collected taxes Baidoa has the highest poverty level followed by Mogadishu, while Kismayo has the lowest poverty 100 level. Average poverty incidence across all urban Percent of households who 80 areas is 64 percent, lower than that of rural areas reported paying taxes at 72 percent. Yet there is significant regional vari- 60 ation among different urban areas. Baidoa city has the highest proportion of poor households (84 per- 40 cent), followed by Mogadishu (73 percent), which are both higher than the overall average of 69 per- 20 cent or the rural average of 72 percent. Kismayo, on the other hand, has the lowest poverty incidence 0 of 35 percent (Figure 2.24). A similar pattern fol- hu th as an th t u l C est an l ut ntr ban ut st al ID ba st n in d ral em an N nts s or s ra tra ura ad lows for the poverty gap with Baidoa having the a So We rur is N rth urb N We t ru W rb Ju h W rb Ps la ru se urb e om r So e ur ad u al highest poverty gap (36 percent), higher than that t l og e s ttl n a or E M E en C of IDPs, followed by Mogadishu (27 percent; Fig- th h o b or N N ure 2.25). Kismayo has the highest percentage of households who report that they have experienced Government of Somaliland Gatekeeper hunger in the past four weeks despite the lowest Al-Shabaab poverty incidence (Figure 2.26). Correspondingly, Local militia fewer households in Kismayo along with North Clan representative West urban and North East urban households Representative of the DC/Local gov report lower food poverty incidence. Baidoa has Federal Government of Somalia the highest food poverty incidence at 69 percent, even higher than that of the IDPs (56 percent; Fig- Source: Authors’ calculations based on the SHFS 2017–18. ure 2.27). Gini coefficient is the lowest in Mogadi- shu, meaning there is least inequality, compared to the highest in Central urban. areas compared to rural areas. Interestingly, the pro- portion of households that pay taxes to the federal government is highest among the rural households 67  North West urban and rural households do not pay any taxes in Central (65 percent) followed by urban house- to the federal government due to political issues. holds in Mogadishu (50 percent). The lowest is in 68  All the findings listed in this section are statistically signifi- North East rural (0 percent) and North East urban cant with a p-value<0.05 unless otherwise stated. Spatial Variation in Living Standards 45 FIGURE 2.24  n  Poverty incidence FIGURE 2.26  n Hunger 100 100 Percent of population Percent of population 80 80 60 60 40 40 20 20 0 0 North West Kismayo Baidoa North East Mogadishu IDP settlements Central Baidoa Kismayo Mogadishu North West North East IDP settlements Central Cities Other urban areas Cities Other urban areas Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.25  n  Poverty gap FIGURE 2.27  n  Food poverty incidence 100 100 Percent of poverty line Percent of population 80 80 60 60 40 40 20 20 0 Mogadishu Kismayo Baidoa North East North West IDP settlements 0 Central Mogadishu Kismayo Baidoa North East North West IDP settlements Central Cities Other urban areas Cities Other urban areas Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. Access to services the highest in North West urban (US$23/month) followed by Kismayo (US$20/month) and the low- Virtually all households in urban North East and est in Baidoa (US$11/month). Urban residents have Mogadishu have access to electricity. In other electricity for on average 15 hours a day with Mog- urban areas, access to electricity is more limited, adishu and North West urban having access for for example, with only 61 percent in Central urban the longest time at 20 hours. Baidoa is an outlier and 58 percent in Kismayo having access to elec- where residents have only four hours of electricity tricity. The level of electricity access in Kismayo is a day. This is much worse than for IDPs who have not much higher than that of IDP settlements at access for an average of 14 hours a day. The urban 49 percent (Figure 2.28). Electricity is provided by bottom 40 percent are less likely to have access to private service providers across all urban areas as electricity than the urban top 60 percent irrespec- there is no public sector capacity. The prices are tive of which cities they live in. 46  Somali Poverty and Vulnerability Assessment FIGURE 2.28  n  Access to electricity FIGURE 2.30  n  Access to improved sanitation 100 100 Percent of households Percent of households 80 80 60 60 40 40 20 20 0 0 Mogadishu Kismayo Baidoa North East North West Central IDP settlements Mogadishu Kismayo Baidoa North East North Eest Central IDP settlements Cities Other urban areas Cities Other urban areas Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.29  n  Access to piped water Improved sanitation is available to 9 of 10 urban 100 households with Baidoa as an outlier where only Percent of households 80 6 of 10 households have access. This estimate is 60 lower than that of IDPs (91 percent) (Figure 2.30). This likely reflects the fact that IDPs in camps 40 have access to sanitation facilities provided by the 20 humanitarian agencies. Such facilities are often 0 shared among many households. Yet, both IDPs Mogadishu Kismayo Baidoa North East North Eest Central IDP settlements and residents in Baidoa on average have four fami- lies sharing one facility compared to other urban areas where they have two households sharing. Of these, over 48 percent of the urban residents use septic tanks and 31 percent rely on informal sew- Cities Other urban areas age connection. It is only in Baidoa (37 percent) where the largest proportion of residents are con- Overall average nected to official sewage. Unless septic tanks and informal sewage networks are maintained prop- Source: Authors’ calculations based on the SHFS 2017–18. erly, this situation can cause serious hygiene and health issues. Piped water is available at home for 75 percent Other municipal services, such as roads, solid of the urban residents across all regions, but Kis- waste management, and street lighting are lim- mayo is an outlier with a very low access to piped ited across all urban areas. Solid waste manage- water. Mogadishu has the highest access (96 per- ment is virtually nonexistent. Lack of solid waste cent) followed by Baidoa (92 percent) while only management is a major source of hazard, as solid 25 percent of households in Kismayo do, which is waste scattered across the streets clog existing lower than that of IDPs (39 percent) (Figure 2.29). drainage systems exacerbating the flash floods The low level of access to piped water in Kismayo caused by torrential rain. They also pose serious is in stark contrast with its low poverty incidence. environmental and health issues. Sixty-seven per- In Kismayo (32 percent) and Baidoa (19 percent), cent of the urban households across regions rely the main alternative water source is boreholes, on burning or dumping waste, and only 14 per- whereas in North West urban, 35 percent of the cent rely on a municipal waste management sys- residents rely on water trucks. tem. In Central urban (91 percent) and Baidoa Spatial Variation in Living Standards 47 FIGURE 2.31  n  Solid waste management FIGURE 2.32  n  Primary school enrollment rate 100 100 80 households 80 Percent of Percent enrolled at school age (6–13) 60 40 60 20 40 0 Mogadishu Kismayo Baidoa North East North West Central IDP settlements 20 0 Mogadishu Kismayo Baidoa North East North West Central IDP settlements Cities Other urban areas Municipal Private Community Dumping/burning Other Cities Other urban areas Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. (85 percent), almost all the households rely on burning or dumping (Figure 2.31). Only 46 percent greater than the IDPs (23 percent) and almost of the urban residents have access roads that are eight-fold compared to Mogadishu that has one usable during the rainy season most of the time, of the highest primary school enrollment rates and whereas another 46 percent have roads that are low proportion of households that are above the some of the time usable during the rainy season. 30-minute threshold. Travelling for over half an Eight percent of the urban households on aver- hour in Somalia, and particularly in Kismayo, poses age rarely have access to all-weather roads, and a serious security threat which may dissuade the the situation is worse in Baidoa (15 percent), North parents from sending their children to school. Edu- West urban (12 percent) and Central urban (10 per- cation is important to address the structural cause cent). Street lights, which play an important role of poverty. Indeed, households with household in improving safety, are rare in Somalia. Access to heads that have complete primary or incomplete functioning streetlights ranges from 35 percent in secondary education are less likely to be poor. The Mogadishu to 6 percent in North East urban. secondary school enrollment rate among children aged 14–17 is the highest in North East urban (90 Both primary and secondary school enrollment percent) followed by North West urban (79 per- rates for children aged 6–17 are higher in North cent) and Mogadishu (78 percent). Kismayo again East urban, North West urban, and Mogadishu, has the lowest enrollment rate of 40 percent, which while Kismayo has the lowest enrollment rates. is the same as among the IDPs. Disparities in school enrollment rates are large across regions. Primary school enrollment rates Satisfaction for the quality of primary education among children aged 6–13 are the highest in North is highest in Mogadishu (94 percent) and low- West urban (67 percent) and Mogadishu (62 per- est in Central urban (77 percent). Despite the cent), while Kismayo lags at 23 percent. Kismayo’s low enrollment rate, satisfaction levels are rela- primary school enrollment rate is lower than that tively high among those that send their children of IDPs, which is 26 percent (Figure 2.32). Urban to school in Kismayo (88 percent). The literacy households are less likely to be poor (both mon- rate is the highest in Mogadishu (76 percent) fol- etarily and non-monetarily) if household heads lowed by North East urban (74 percent). Kismayo have some education. The low enrollment rate in has the lowest literacy rate (41 percent) among all Kismayo may be due to the long distance children urban areas, and even lower than that of the IDPs must travel to schools. In Kismayo, over 48 per- (51 percent) (Figure 2.33). Urban households are cent of the children need to travel over 30 min- less likely to be poor if there is a higher proportion utes to get to the closest school, which is much of literate household members. However, there is 48  Somali Poverty and Vulnerability Assessment FIGURE 2.33  n  Literacy rate FIGURE 2.34  n  Distance to health facilities (>30 min) 100 100 Percent of households Percent of households 80 80 60 60 40 40 20 20 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central Cities Other urban areas Cities Other urban areas Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.35  n  Satisfaction on health service quality 100 no significant difference between the poor and the 80 households Percent of non-poor, or whether they are displaced or not, in 60 terms of literacy rate in Kismayo. 40 20 Access to health facilities is poor in Kismayo 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central though satisfaction over their quality of services is high across the board. The proportion of house- holds whose distance to the closest health cen- ters is over 30 minutes is the highest in Kismayo (39 percent)—higher than that of IDPs (28 per- cent)—while North West urban (9 percent) has the Cities Other urban areas best accessibility (Figure 2.32). Despite the dis- tance to the health facilities, residents of Kismayo, Very satisfied Satisfied Neither/nor along with other urban areas, seem to be satisfied Dissatisfied Very dissatisfied with the quality of the health services they are receiving (Figure 2.33). Source: Authors’ calculations based on the SHFS 2017–18. Access to other services differs significantly with the highest at 28 percent, again worse off than the Kismayo worse off than the IDPs in some ser- IDPs at 22 percent (Figure 2.36). vices. While Mogadishu demonstrates the highest proportion of households with access to pub- lic transport (86 percent), other urban areas lag. Land and housing North East urban (27 percent), Kismayo (34 per- cent), and North-West urban (37 percent) have Mogadishu and Kismayo have the highest pro- low access to public transport. Internet access portion of renters, and three-quarters of urban is still not as prevalent at 24 percent on average. residents have registered land certificates. Internet access is almost the same between Kis- Seventy-­one percent of Mogadishu residents reside mayo (20 percent) and the IDPs (19 percent). It is in rented space followed by 56 percent in Kismayo. the highest in North West urban (33 percent). In In Baidoa, where it is more sparsely populated Kismayo, the proportion of households that live than other cities, ownership of the housing is the more than 30 minutes away from a food market is highest among all urban areas at 68 percent, and Spatial Variation in Living Standards 49 FIGURE 2.36  n  Access to market, public transport, Internet 100 80 households Percent of 60 40 20 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central Cities Other urban areas Market distance more than 30 minutes Transport Internet Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.37  n  Tenure status FIGURE 2.38  n  Access to improved housing 100 100 Percent of households 80 80 households Percent of 60 60 40 40 20 0 20 Mogadishu Kismayo Baidoa North East North West IDP settlements Central 0 North West Kismayo Baidoa North East Mogadishu IDP settlements Central Cities Other urban areas Cities Other urban areas Own Rent Squatting Other Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. only 24 percent rent. It is only the IDPs that have a high proportion of households that are squatting in others’ dwelling (17 percent; Figure 2.37). Major- Baidoa (89 percent), and Kismayo (85 percent) all ity of the households in all urban areas live either in scoring high. This is a significant finding as anec- apartments or shared houses. Access to improved dotally it has been assumed that households do housing is available for 99 percent of the residents not possess any legal documents for their proper- in North East urban, followed by 80 percent in ties. It would be useful to understand under which Mogadishu.69 It is only available for 9 percent of administration these registered land certificates the people in Baidoa, which is much lower than the were issued. Central urban, however, seems to rely IDPs (31 percent; Figure 2.38). Most of the urban on customary law (41 percent) as much as regis- residents (75 percent) claim to have registered tered land certificates (52 percent) to establish the land certificates, with Mogadishu (94 percent), tenure. On average, only 16 percent of the urban households across the region rely on decisions by the local government to establish their tenure Improved housing is defined as living in apartments, shared 69  (Figure 2.39). A high proportion of households apartments, separate houses or shared houses. in Baidoa (77 percent) and North West urban 50  Somali Poverty and Vulnerability Assessment FIGURE 2.39  n  Legal recognition of land and housing FIGURE 2.40  n  Access to bank accounts 100 100 Percent of households who reported Percent of households 80 80 being legal dwelling owner 60 40 60 20 40 0 20 Mogadishu Kismayo Baidoa North East North West IDP settlements Central 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central Cities Other urban areas Registered land certificate Cities Other urban areas Decision by local government Bank account Mobile money account By customary law Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. (66 percent) have written formal agreements on FIGURE 2.41  n  Households that saved money tenancy, while only 5 percent in Kismayo do. Only 100 24 percent of the household in Mogadishu have Percent of households written agreements. 80 60 40 Access to finance 20 A high proportion of urban households (69 per- 0 cent) have access to mobile bank accounts. This Mogadishu Kismayo Baidoa North East North West IDP settlements Central is not surprising given the amount of international and domestic remittances received. North West urban has the highest proportion of households that have mobile bank accounts (85 percent) while Central urban has the lowest (55 percent). Interestingly, 54 percent of IDPs have mobile bank Cities Other urban areas accounts (Figure 2.40). Penetration of traditional Overall average bank accounts, on the other hand, is limited, rang- ing from 26 percent in Central urban to 4 percent in Source: Authors’ calculations based on the SHFS 2017–18. Kismayo. This is in line with the lack of availability of commercial banks within the country. However, the proportion of urban households that could Employment save money in the past 12 months was a meager 11 percent on average across all urban areas, with Households mostly rely on wage labor and small North West urban (22 percent) and Central urban family businesses. On average, 50 percent of urban (18 percent) scoring the highest, while Baidoa households across regions rely on wage labor for scores the lowest (22 percent; Figure 2.41). This their livelihood. Mogadishu (64 percent) and North does not reflect the amount saved however. Most West urban (64 percent) have the highest propor- of the households rely on their relatives (64 per- tion of households making their living on wage cent) and friends (24 percent) to borrow money labor. Kismayo is an exception where small family from. People have yet to borrow from private business is the main income source for 50 percent money lenders (2 percent). This could be due to of the households, whereas only 32 percent rely on cultural reasons or since private money lenders are wage labor. In Baidoa, close to 30 percent of the not very popular yet. households rely on a family business. Remittances Spatial Variation in Living Standards 51 FIGURE 2.42  n  Main sources of income FIGURE 2.43  n  Perception on employment 100 opportunities 80 100 households Percent of 80 households 60 Percent of 40 60 40 20 20 0 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central Mogadishu Kismayo Baidoa North East North West IDP settlements Central Cities Other urban areas Cities Other urban areas Salaried labor Remittances Much better Better The same Small family business Agriculture Worse Much worse Other Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. represent a small portion of income sources in Safety and freedom of movement all areas ranging from 3 percent in Baidoa to 14 percent in North East and North West urban (Fig- Kismayo has the lowest proportion of people feel- ure 2.42). Employment satisfaction is high in gen- ing “very safe” from crime and violence closely eral, where 70 percent of the households or more followed by Mogadishu. In Kismayo (35 percent), report to be “very” or “somewhat” satisfied with Mogadishu (38 percent), and Central urban (41 per- their employment. Perception on standard of liv- cent), urban households’ perception that they are ing prospects is also positive in all urban areas, “very safe” is lower than the overall average across with at least 60 percent of the households report- all urban areas. While these are much higher than ing them to be “better” or “much better” than six that of the IDPs (29 percent), it demonstrates that months before. However, perception on employ- people in central southern regions of Somalia still ment opportunities is less positive. North West feel unsafe compared to the more stable North urban (42 percent) has the lowest level of positive West and North East. Baidoa, however, has a rela- perception—even lower than that of the IDPs (44 tively high proportion of people feeling very safe percent)—while the highest in Baidoa (66 percent) (53 percent) (Figure 2.44). Such perceptions of and Kismayo (64 percent; Figure 2.43). safety are not aligned with the people’s perception of freedom of movement. Mogadishu households Across the country, people generally have a high- report the highest level of freedom of movement risk appetite toward economic activities. North among all urban areas compared to IDPs who East urban and Mogadishu households are the report the lowest level of freedom of movement most willing to take risks to invest in high profit but (Figure 2.45). risky business (72 percent) followed by Kismayo (67 percent) compared to the average of 62 per- Trust cent across all urban areas. North West urban has the lowest risk appetite at 40 percent. Interestingly, The more households rely on police for dispute 70 percent of the IDPs are also willing to take risks resolution, the less rely on clan elders. On aver- in risky business. The relatively high-risk appetite age, 56 percent of the households rely on police may be an indication that Somalis are willing to to settle disputes though there is a significant take more risks for higher profit given the longer- regional disparity. It is the highest in North West term uncertainty. urban (77 percent) and Mogadishu (69 percent), while it is less than half of that in Central urban 52  Somali Poverty and Vulnerability Assessment FIGURE 2.44  n  Safety from crime and violence FIGURE 2.46  n  Dispute resolution 100 100 Percent of households reporting feeling safe 80 Percent of households 80 60 40 60 20 40 0 Mogadishu Kismayo Baidoa North East North West IDP settlements Central 20 0 Mogadishu Kismayo Baidoa North East North West Central IDP settlements Cities Other urban areas Very unsafe Unsafe Neither/nor Safe Very safe Cities Other urban areas Source: Authors’ calculations based on the SHFS 2017–18. Clan elders (Xeer) Religious leaders Informal court Police AMISOM Formal court FIGURE 2.45  n  Freedom of movement Community leader Other 100 Percent of households Source: Authors’ calculations based on the SHFS 2017–18. 80 60 confidence in police are lower than the overall 40 urban average and lower than that of the IDPs (65 20 percent). This disparity is likely due to how much 0 outreach the police have, i.e., relatively lower out- Mogadishu Kismayo Baidoa North East North West IDP settlements reach in rural areas than in urban areas, rather than Central a reflection of the levels of trust in police. People’s trust in various state institutions differs across regions. Sixty-two percent of the average Cities Other urban areas urban people believe that the federal government best represents their interests.70 Mogadishu is Overall average the highest with 87 percent followed by IDPs at 63 percent. Other regions hover around the aver- Source: Authors’ calculations based on the SHFS 2017–18. age except for North West urban where none of the households believe that the federal govern- ment represents their interest. This is understand- (33 percent) and Baidoa (37 percent). In Cen- able given that Somaliland is a highly autonomous tral urban (51 percent) and Baidoa (32 percent), state. Instead, 83 percent of the North West urban a higher proportion of households rely on clan residents believe that Somaliland government best elders compared to Mogadishu (9 percent) and represents their interests. Trust in the state gov- North West urban (13 percent). Reliance on reli- ernment is 15 percent on average ranging from 34 gious leaders was at a low average of 8 percent percent in Kismayo to 2 percent in Mogadishu. In (Figure 2.46). On average, 62 percent of the urban Kismayo, there is a strong state government and households have confidence that the police will political independence, which explains that while protect them from crime and violence. This is the highest in North West urban (76 percent) and Baidoa (70 percent), while lowest in North East This average has omitted North West urban households 70  urban (54 percent). In Mogadishu, Kismayo, North where 0 percent voted for the federal government due to the East urban, and Central urban, levels of people’s political contexts. Spatial Variation in Living Standards 53 FIGURE 2.47  n  Trust in institutions FIGURE 2.48  n  Payment of taxes 100 100 Percent of households 80 80 Percent of households 60 60 40 20 40 0 20 Mogadishu Kismayo Baidoa North East North West Central IDP settlements 0 Baidoa Kismayo Mogadishu North West North East IDP settlements Central Cities Other urban areas Other Cities Other urban areas Diaspora Overall average State government Government of Somaliland Federal Government of Somalia Source: Authors’ calculations based on the SHFS 2017–18. Opposition groups International NGOs FIGURE 2.49  n  Institutions that collected taxes African Union/Unite Clan elders 100 Percent of households who Local NGO reported paying taxes 80 Source: Authors’ calculations based on the SHFS 2017–18. 60 40 Mogadishu’s low confidence in its regional govern- ment may be due to the high political turnover, 20 or potentially, people did not associate Banadir 0 Regional Administration as state government (Fig- Mogadishu Kismayo Baidoa North East North West Central IDP settlements ure 2.47). General levels of trust are the highest in Baidoa and the lowest in Kismayo. On average, urban residents’ level of trust in other people is above 60 percent. Eighty-seven percent of the house- Cities Other urban areas holds in Baidoa believe that most people can be Other trusted, while in Kismayo only 47 percent do. Lev- State government els of trust in Mogadishu and North East urban are Government of Somaliland both relatively high at 73 percent. Interestingly, Gatekeeper 67 percent of IDPs believe that most people can Al-Shabaab be trusted, a higher proportion than in North East Local militia urban, Central urban, and Kismayo. Clan representative Municipal/local government A limited number of households currently pay taxes, and among them, the majority pay to the Source: Authors’ calculations based on the SHFS 2017–18. local government. The Somali government cur- rently only collects minimal taxes and fees, such as business tax, customs, cargo fee, and birth 54  Somali Poverty and Vulnerability Assessment certificate fees. Indeed, only 18 percent of the urban households currently pay taxes, ranging Intra-urban comparison from 40 percent in North East urban to 0.3 percent in Kismayo (Figure 2.48). Among them, the major- Variations among urban IDPs, rural ity (55 percent) of the households pay taxes to the IDPs, and settlement IDPs local government with North East urban being the There are no significant differences between highest at 90 percent followed by Baidoa (60 per- urban IDPs, rural IDPs and settlement IDPs in pov- cent). Only 29 percent of the urban households on erty incidence, poverty gap, or food consump- average pay taxes to the federal government with tion poverty incidence. This does not change Mogadishu (50 percent) and Central (48 percent) even when controlling for literacy rate, proportion being the highest. In North West urban, 60 percent of working age household members, gender of of the urban households pay taxes to the Somali- the household head, and share of male members land government (Figure 2.49). Judging from the in the household. The proportion of working age level of trust in different institutions, the amount household members, however, lowers the poverty of taxes paid to different institutions is not cor- indicators across the different IDP groups. There related with the levels of trust. Rather, it seems are also no significant differences across different like people pay taxes to whoever is more able to groups of IDPs in hunger experienced in the last enforce it. Local governments have the advantage four weeks or facing a shortage of money to buy as they are the closest to their constituents. None food. The levels of monetary and nonmonetary of the households responded that they pay taxes poverty are similar across all IDP groups. to Al-Shabab. This may well be the case since Al- Shabaab does not have strongholds in urban areas Urban IDPs are better off than rural IDPs in access anymore. Alternatively, people may not be willing to certain services. Urban IDPs have better access to acknowledge even if they do. to electricity, improved housing, and improved Box 8  ■  Intra-urban comparison While the previous sections focused on the spatial variations, this section focuses on the variations among different population groups that reside in urban areas. Urban residents can broadly be categorized into: IDPs that live in IDP settlements which are all in urban areas (settlement IDPs), IDPs that live outside the IDP settle- ments and are integrated into urban areas such as informal settlements mostly in downtown areas (urban non- settlement IDPs), urban communities that host IDPs in their neighborhood (urban host communities), and urban communities that do not have any IDPs in their neighborhood (urban non-host communities). Given the small sample size, the groups cannot be broken down into different regions. The section thus analyzes whether there are any significant differences across these population groups.71 Urban IDPs versus rural IDPs versus settlement IDPs. This comparison will examine whether urban IDPs, those 1.  who reside outside of IDP settlements and in urban areas, are better off than rural IDPs who live in rural areas or IDPs that live in IDP settlements located in urban areas. Urban IDPs versus other urban households (both host and non-host communities). This comparison exam- 2.  ines whether urban IDPs that do not live in IDP settlements are worse off than other urban non-IDP house- holds, irrespective of whether the households host or do not host IDPs. Urban host communities versus urban non-host communities. This comparison examines whether urban 3.  communities that host urban non-settlement IDPs living in the geographical proximity are worse off than urban communities that do not host IDPs. This section describes the results of OLS and logit regression 71  analyses, controlling for different variables such as access to services. Nonsignificant results are not described unless neces- sary and explicitly stated. Spatial Variation in Living Standards 55 FIGURE 2.50  n  Distribution of IDPs and urban FIGURE 2.52  n  IDPs’ access to key facilities population 100 Percent of IDP households having 100 to walk more than 30 minutes Percent of households 80 80 to reach location 60 60 40 40 20 0 20 Urban Settlement Rural IDPs Host Non-host 0 Health center School Market Urban IDPs Rural IDPs IDPs in settlements Non-IDPs IDP households Urban households Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. there is no significant difference in their standards of living. They are both better off than rural IDPs FIGURE 2.51  n  IDPs’ access to services (Figure 2.51, Figure 2.52). 100 Urban IDPs feel less safe and are more risk averse than settlement IDPs. The only significant differ- having access to the service Percent of IDP households 80 ences between urban IDPs and settlement IDPs are the perception of safety from crime and violence 60 and risk aversion. Urban IDPs, who live outside of IDP settlements, feel less secure. This is intuitive as 40 IDPs residing in organized settlements have gate- keepers that provide certain levels of protection 20 and are served by NGOs or humanitarian agencies. IDPs residing outside any settlements, however, do 0 not have access to any protection network, espe- ty er us ed ta ed et cially as they are removed from their own clan ci rn at ho rov ni ov g n tri W te in tio sa pr structure that normally serves to provide security. ec p In Im Im El This seems to indicate that safety is not a function Urban IDPs Rural IDPs IDPs in settlements of the spatial location but rather on the availability of a protection mechanism (Figure 2.53). Source: Authors’ calculations based on the SHFS 2017–18. sanitation than rural IDPs. However, there are no Variations among urban IDPs and other significant differences between urban and rural urban non-IDP households IDPs in access to piped water, distance to health facilities, distance to primary schools, distance to Although urban IDPs are not any poorer than food markets, literacy, employment, tenure of the other urban households, they are more likely to dwelling, or access to Internet. Urban IDPs are not have experienced hunger and lack money. There better off than settlement IDPs who live in urban are no significant differences between urban IDPs areas in access to services, housing, or employ- and other urban households in poverty incidence, ment. There are no significant differences across poverty gap, and food consumption poverty inci- these two groups of IDPs in access to basic ser- dence. Nonetheless, urban IDPs are more likely to vices, housing, or employment. This means that have experienced hunger in the past four weeks irrespective of whether IDPs live in IDP settle- and lack the money to buy food compared to other ments or not, so long as they live in urban areas, urban households. 56  Somali Poverty and Vulnerability Assessment FIGURE 2.53  n  IDPs’ perception of safety 100 Percent of IDP households reporting safety perception 80 60 40 20 0 Very unsafe Moderately Neither safe Moderately Very safe unsafe nor unsafe safe Urban IDPs Rural IDPs Settlement IDPs Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 2.54  n  Urban IDPs’ access to services FIGURE 2.55  n  Urban IDPs’ access to key facilities 100 100 Percent of urban households having to Percent of urban households walk more than 30 minutes to having access to the service 80 80 reach location 60 60 40 40 20 20 0 0 ty er us ved ta ed et Health center School Market ci rn at ni rov g n tri o W te in tio pr ec p In Im Im Other urban households Urban IDPs El ho sa Other urban households Urban IDPs Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. displacement and they have more limited access to services. Moreover, they are at a disadvantage Urban non-settlement IDPs are consistently in levels of education, which may prevent them worse off than other urban households. Urban from finding good jobs. Accordingly, without any non-settlement IDPs have less access to electricity, concerted support, urban non-settlement IDPs are piped water, improved sanitation, improved hous- likely to remain worse-off across many dimensions ing, dwelling ownership, and internet compared to compared to other urban non-IDP households. other non-IDP urban households. Moreover, urban (Figure 2.54, Figure 2.55). non-settlement IDPs suffer from lower enroll- ment, literacy, and employment rates. They also Urban IDPs face more limited freedom of move- tend to live further away from primary schools and ment compared to other urban households. This food markets. Thus, urban non-settlement IDPs could be due to urban IDPs’ general lack of money are worse off than the rest of the urban popula- to be able to afford public transportation. Their tion as they have likely become deprived of their perception of safety from crime and violence, trust former livelihoods, assets, social networks due to in other people, trust in police, and levels of risk Spatial Variation in Living Standards 57 FIGURE 2.56  n  Urban IDPs’ perception of safety 100 Percent of urban households reporting safety perception 80 60 40 20 0 Very unsafe Moderately Neither safe Moderately Very safe unsafe nor unsafe safe Other urban households Urban IDPs Source: Authors’ calculations based on the SHFS 2017–18. aversion are not significantly different from other to Internet. There are no significant differences in urban households. This signifies that the main bar- literacy and enrollment rates, either. rier urban IDPs face is adequate access to money, services, land and housing, and job opportunities, Urban host communities have marginally higher but they do not face any major social or psycho- trust in police and freedom of movement. How- logical obstacles compared to other urban house- ever, there are no significant differences on the holds (Figure 2.56). other indicators such as perception of safety from crime and violence and trust in other people. Variations among urban host and To conclude, there are almost no differences in the standards of living between urban host and urban non-host communities non-host communities. What can be derived from Urban host communities are marginally poorer these findings is either that the effect of hosting than urban non-host communities. Anecdotally, IDPs has not yet materialized as the duration has urban host communities—defined as communities been too short, as the average period of displace- who live in close geographical proximity with urban ment of IDPs is 2 years, or that hosting IDPs does non-settlement IDPs—are believed to be worse off not result in deteriorated access to services, as than urban non-host communities as they will be the IDPs are self-contained in the settlements and deprived of services, face more competition over have basic services provided for. The picture may limited resources and jobs, and have their social change if the IDPs’ stay in urban areas is prolonged cohesion disrupted with the influx of IDPs. Yet, the and the levels of support from the humanitarian data show that urban host communities and urban agencies declines. non-host communities share similar rates of pov- erty. There are also no significant differences in the poverty gap or proportion of households experi- Policy recommendations encing hunger in the last four weeks. Urban areas provide higher standards of living Urban host and non-host communities also and better access to services than rural areas. appear to have a similar profile in access to ser- But they lag in access to land and housing, which vices. Urban host and non-host communities have has been further constrained by the recent influx very similar levels of access to electricity, piped of the IDPs to cities. Seventy-five percent of the water, improved housing, own dwellings, improved IDPs in Somalia reside in urban centers, settling on sanitation, distance to health facilities, distance to public and private lands within and in the outskirts primary schools, distance to markets, and access of cities. The majority of returnees are considered 58  Somali Poverty and Vulnerability Assessment to have settled in cities as well. In Mogadishu, areas existing urban fabric with the extension of basic occupied by IDP settlements increased by 16 per- services and housing. cent between 2013 and 2017. In Kismayo, the IDP- occupied area increased over seven-fold, and in Investments in cities need to be spatially differ- Baidoa, it has more than tripled.72 In the absence of entiated to better address the regional dispari- security of land tenure, IDPs are highly vulnerable ties. Given the significant regional disparities and to forced eviction. For example, over 109,000 IDPs idiosyncratic development needs across different living in informal settlements across the country urban areas, interventions in cities will need to be have been forcefully evicted between January and prioritized and sequenced according to local con- August 2017, 77 percent of which are concentrated texts. For example, Mogadishu has high monetary in Mogadishu.73 Due to forced eviction, many poverty but relatively good access to services, IDPs have shifted to the outskirts of cities, caus- while Kismayo has very low poverty incidence but ing uncontrolled urban sprawl. Fifty-five percent lacks most of the basic services. Thus, detailed of IDPs in Mogadishu now reside in the periphery assessments at the city level are necessary to bet- of the city. The area occupied by IDP settlements ter understand the symptoms and the drivers of in the fringes of Baidoa has increased by 177 per- constraints to urbanization in each city to derive cent in 2017. The number of IDPs in Kismayo tripled the most appropriate solutions. Political economy in 2017, and most of them have settled outside of must be considered in crafting and implementing the city.74 Such spatial sprawl makes service pro- any policies. It is essential to foresee the oppor- vision difficult and costly, as new settlements are tunities, risks, winners, and losers of any specific disconnected from the existing urban centers and policies, and anticipate challenges to enforce the infrastructure networks. Spatial fragmentation also policies.75 inhibits IDPs’ access to jobs and prevents cities from reaping the scale and agglomeration benefits. Within cities, the needs of non-settlement IDPs should be addressed along with the needs of set- For Somalia to reap the benefits of urbaniza- tlement IDPs. Such assistance should be provided tion, the government needs to invest in two through area-based approaches to ensure equity core elements of cities—land and coordinated among different vulnerable urban population infrastructure investments. The fundamental ele- groups. Much of the attention to date has focused ment in making Somali cities work is to establish on assisting urban IDPs living in settlements as a proper land administration system and effective they were deemed the most deprived. Neverthe- land use planning. This will allow for a more con- less, the data show that urban non-settlement IDPs trolled growth of the city and provision of secu- are equally deprived of access to services as IDPs rity of tenure to the IDPs, many of whom prefer to in settlements. Moreover, they consistently fare settle in cities. The other important element is to worse on all development outcomes compared to make coordinated infrastructure investments. Data other urban households. The urban non-settlement show that cities are better off in terms of infra- IDPs are difficult to track as they have integrated structure and service delivery compared to rural into local areas. It would therefore be important areas. However, the absolute level of infrastruc- to ensure that urban interventions take an area- ture and service delivery is still low across urban based approach that prioritizes areas that have a areas. To make a dent in the soaring demand, cities high concentration of the non-settlement IDPs as need coordinated investments—rather than ad hoc well as the urban poor, rather than a population single sector interventions—aligned with the land group-based approach focusing solely on IDPs in use plan to take advantage of the synergy across settlements, so that all vulnerable urban popula- different types of infrastructure. For example, the tion can benefit from development interventions. government can maximize the benefits of infra- In so doing, it is essential to ensure that any devel- structure investments by coupling road construc- opment is aligned with the broader urban develop- tion that link the new satellite townships and the ment plans. 72  UN-Habitat calculation (2017). 73  Norwegian Refugee Council (2017). 74  World Bank, Somalia Drought Impact and Needs Assessment (2018c). Urban chapter. 75  Lall, et al. (2017) Spatial Variation in Living Standards 59 It is important to continue to help strengthen the assistance and resources through them rather than state institutions, particularly at the subnational through parallel structures. In doing so, more focus level. People’s confidence in state institutions is can be shifted to the subnational governments, relatively high. Given Somalia’s nascent political namely the state and municipal governments, as history, it would be critical for all development they are ultimately accountable for providing ser- partners to continue to help strengthen the gov- vices to their constituents. ernment institutions by channeling development 60  Somali Poverty and Vulnerability Assessment CHAPTER 3 Drought Impact KEY MESSAGES Several consecutive seasons of poor rainfall led to to experience hunger. High drought exposure did not a severe drought in Somalia, as one in two Somalis lead to an increase in poverty or hunger among urban faced acute food insecurity and close to 1 million households. were displaced in 2017. Four consecutive below-­ average rainy seasons between April 2016 and Rural households are vulnerable to further income December 2017 resulted in a severe drought. The shocks. The drought’s impact on rural households drought exacerbated food insecurity among Somalis, indicates that these households are vulnerable to with 6.2 million, half the population, facing acute food income shocks. A renewed shock of the same mag- insecurity in 2017. Lack of water and pasture deci- nitude as the current drought would increase poverty mated livestock herds and threatened livelihoods, as in rural areas by 11 percentage points, from 65 percent 1 million Somalis were displaced due to the drought. to 76 percent. Swift humanitarian interventions averted famine in 2017. Investment in resilience is key to prevent loss of live- lihood of the most vulnerable rural households. Rural The drought affected Somalis in rural areas severely, households relying on agriculture for their income and who were 24 percent more likely to be poor and those lacking access to financial services and infra- 17 percent more likely to experience hunger. The structure are the most vulnerable to income shocks. drought led rural households’ consumption to decline Investing in the resilience and access of these house- by 18 percent, corresponding to an increase of 24 per- holds is key to prevent loss of livelihoods. Measures cent in the probability of being poor. The effect was may include providing insurance products, enabling stronger for wealthier households. Rural drought- households to diversify their sources of income, and affected households were also 17 percent more likely improving access to infrastructure. The Horn of Africa is experiencing a severe High seasonal weather variability and El Niño-La drought, triggering a regional humanitarian crisis Niña events make droughts a recurrent phenome- including elevated levels of food insecurity and non in this region. In Somalia, for example, drought malnutrition.76 At least three consecutive seasons conditions have developed at least 13 times since of poor rains between March 2016 and December 1964, with varying durations and intensities.79 Sev- 2017 resulted in a drought that left 14.6 of 120 mil- eral of these droughts—coupled with prolonged lion facing severe food insecurity as of late 2017.77 conflict and insecurity, governance failures, and Large-scale humanitarian interventions provided inadequate intervention—resulted in famines. This critical relief to affected populations and reduced led the international and donor communities to ini- the risk of famine. Slightly improved rains in late tiate two early warning and monitoring projects, 2017 and early 2018 eased the drought condition, the FAO-managed Food Security and Nutritional but food insecurity in the Horn of Africa remains a Analysis Unit (FSNAU) in 1995 and the USAID- serious concern.78 funded Famine Early Warning Systems Network (FEWSNET) in 1985. These two projects collabo- rate to build resilience and facilitate humanitarian 76  The Horn of Africa comprises four countries: Djibouti, Eritrea, Ethiopia, Somalia. 77  UNOCHA (United Nations Office for the Coordination of Humanitarian Affairs) (2017b); FEWSNET (2017). Centre for Research on the Epidemiology of Disasters (CRED) 79  78  FSNAU and FEWSNET (2018). (2017). Drought Impact 61 FIGURE 3.1  n  Rainfall and NDVI anomaly and overview of rainy seasons, all regions Wave 1 55 120 Wave 2 Percent of average 50 100 2016 Gu season 45 80 2016 Deyr season 40 2017 Gu season 60 35 2017 Deyr season 40 30 Rainfall anomaly Feb-17 Feb-18 Dec-17 Oct-17 Dec-16 Oct-16 Feb-16 Apr-17 Jun-17 Aug-17 Aug-16 Jun-16 Apr-16 NDVI anomaly Source: FEWSNET, WFP-VAM, and authors’ calculations based on the SHFS 2017–18. response in the event of impending drought and central regions (Figure 3.2).81 Subsequently, the food insecurity crises in the Horn of Africa. 2016 Deyr rains performed very poorly with most regions experiencing less than 40 percent of aver- Famines proceeded the droughts in 1992 and 2011, age rainfall (Figure 3.3).82 This particularly severe as armed conflict impeded humanitarian interven- rainfall deficit exacerbated pre-existing food inse- tions. In 1992 and 2011, consecutive seasons of poor curity. The 2017 Gu rains were around normal in rainfall largely concentrated in south and central intensity and duration in northern regions, but alto- Somalia led to droughts and declines in food and gether well below average in southern and central livestock production. Civil conflict and insecurity regions. The 2017 Gu rains thus further aggravated compounded the food shortages, as food resources food insecurity in the affected regions.83 Most were either destroyed or looted and many local recently, the 2017 Deyr rainy season was erratic, markets were disrupted, cutting people off from and total rainfall ranged between 10 to 60 percent food supplies. Informal coping mechanisms were below average in most regions (Figure 3.1).84 eroded, and humanitarian relief operations were impeded from accessing the populations most in One in two Somalis faced acute food insecurity in need. The 1992 crisis started in the immediate after- 2017, while one in four required urgent humanitar- math of the disintegration of the central state in ian assistance. Each successive failed rainy season 1991, which was accompanied by widespread civil in 2016 and 2017 exacerbated the food insecurity and sectarian strife. In 2011, Al-Shabaab and clan among Somalis. By the mid-2017, 6.2 million Soma- militias controlled the most affected regions. The lis, half the population, faced acute food insecurity resulting famines claimed 220,000 lives in 1992 and based on the Integrated Phase Classification (IPC) 260,000 lives between 2011 and 2012.80 for food insecurity (Table A.1).85 Among those fac- ing acute food insecurity, 2.4 million people needed humanitarian assistance to avert loss of livelihoods The 2016/17 drought and reduce acute malnutrition (IPC Phase 3— Crisis). 866,000 people required urgent food and its effects assistance to avert famine (IPC Phase 4—Emer- gency), among whom 300,000 children below five Weather conditions in 2016/17 have been par- ticularly extreme and erratic in Somalia. Somalia has two main rainy seasons; the main Gu rains from April to June and the short Deyr rains from Octo- 81  FEWSNET (2016). ber to December. The drought started with the 82  FSNAU and FEWSNET (2017b). 83  FSNAU and FEWSNET (2017a). 2016 Gu rains, which were below average, erratic, 84  FSNAU and FEWSNET (2018). or shorter than usual, especially in southern and 85  The Integrated Phase Classification (IPC) is a harmonized and internationally comparable system of classification of severity and magnitude of food insecurity, which was first developed in 2004 and revised in 2012. Food insecurity is classified in five phases: Phase 1—Minimal; Phase 2—Stressed; Phase 3—Crisis; 80  Salama, et al. (2012); FSNAU and FEWSNET (2013). Phase 4—Emergency; Phase 5—Famine (Table A.1). 62  Somali Poverty and Vulnerability Assessment were acutely malnourished.86 Humanitarian inter- FIGURE 3.2  n 2016 Gu precipitation ventions averted a famine in 2017. The combina- tion of humanitarian assistance and slightly better rains led to an improvement in food security, with 4.4 million Somalis facing acute food insecurity in early 2018 (Figure 3.6). Vulnerable rural, nomadic, and IDP populations remain at risk given the higher prevalence of hunger. Forty-four percent of rural, 50 percent of nomadic, and 60 percent of IDP households experienced hunger at least on a few occasions in December 2017 (Figure A.1). Extremely wet The drought severely affected livestock, a key Very wet Moderately wet source of livelihood for Somalis. Livestock body conditions worsened atypically in early 2017 as Near normal water stocks and pasture deteriorated, leading to Moderately dry a decline in the market value of livestock and poor Very dry milk production. Low birth rates, high livestock Extremely dry deaths, and distress selling caused pastoralists to lose between 25 and 75 percent of their herds in the first half of 2017.87 Given low livestock supply, livestock market prices increased in the second half of 2017. Improved water availability in July 2017 induced a slow recovery of herd sizes and body Source: USGS/FEWSNET/Funk, et al. (2015). conditions, though several consecutive favorable rainy seasons will be necessary for herd sizes to fully recuperate.88 The drought-related damages and losses in the livestock sector were estimated FIGURE 3.3  n 2016 Deyr precipitation at US$1.6 billion and an additional US$400 mil- lion in losses from reduced livestock exports was expected for 2018.89 Food production fell below average and prices for food and water rose in 2016 and 2017. Cereal harvest was at least 10 percent below average in the southern main crop producing areas for four consecutive seasons. The post-Deyr harvest in 2017 was one of the poorest on record at 68 percent below 1995–2015 average, though harvest yields Extremely wet Very wet were expected to improve somewhat in 2018. Moderately wet Cereal production in the North West also performed poorly, particularly in 2017.90 In combination, the Near normal drought has caused crop production damages Moderately dry and losses estimated at above US$300 million.91 In Very dry the first half of 2017, local cereal prices increased Extremely dry between 32 and 70 percent above the long-term 86  FSNAU and FEWSNET (2017b). 87  FSNAU and FEWSNET (2018). 88  FSNAU and FEWSNET (2018). 89  World Bank (2018c). Source: USGS/FEWSNET/Funk, et al. (2015). 90  FSNAU (2016b); FSNAU (2016a); FSNAU (2017); FSNAU and FEWSNET (2017c); World Bank (2018c). 91  World Bank (2018c). Drought Impact 63 FIGURE 3.4  n 2017 Gu precipitation FIGURE 3.5  n 2017 Deyr precipitation Extremely wet Extremely wet Very wet Very wet Moderately wet Moderately wet Near normal Near normal Moderately dry Moderately dry Very dry Very dry Extremely dry Extremely dry Source: USGS/FEWSNET/Funk, et al. (2015). Source: USGS/FEWSNET/Funk, et al. (2015). FIGURE 3.6  n  Population facing food insecurity, all regions 12 100 75 Population (millions) Percent population 8 1 Minimal 2 Stressed 50 3 Crisis 4 Emergency 4 5 Famine 25 Households reporting hunger, SHFS 0 0 Ju 6 Ju 7 Se 6 Se 7 16 M 7 18 6 7 N 6 6 N 7 7 -1 -1 l-1 l-1 1 -1 -1 1 -1 1 -1 n- n- n- p- p- ay ay ar ar ov ov Ja Ja Ja M M M Source: FSNAU data, authors’ calculations based on the SHFS 2017–18. 64  Somali Poverty and Vulnerability Assessment FIGURE 3.7  n  Internal displacement due to drought 350,000 1.2 Monthly drought displacement 300,000 1.0 Cumulative (millions) 250,000 0.8 200,000 0.6 150,000 0.4 100,000 50,000 0.2 0 0 6 6 6 17 17 7 7 7 17 7 17 17 7 7 7 -1 -1 -1 -1 r-1 -1 l-1 -1 -1 -1 n- b- n- g- p- ct ov ec ar ay ct ov ec Ju Ap Ja Fe Ju Au Se O O M N D M N D Displacement due to drought Cumulative drought displacements Source: UNHCR (2018a). average in central and southern regions and close to 80,000 reported cases of AWD/cholera.94 between 12 and 27 percent above average in north- However, the spread of AWD/cholera slowed con- ern regions. Humanitarian supply alleviated some siderably in the second half of 2017, with no fatali- shortages and imported cereal prices remained ties related to the disease reported since August mostly stable. Local cereal prices stabilized some- (Figure A.3).95 what in early 2018, though they remained up to 17 percent above average. The drought further drove Close to one million Somalis were displaced up water prices in 2017, which remained between 11 between 2016 and 2017. With the drought threat- and 56 percent above average in early 2018.92 ening livelihoods, more and more households have been forced to leave their permanent place of Higher food prices, lower wage levels, and residence in search of assistance from the govern- depleted assets diminished the purchasing ment and international actors. Before the onset of power and coping abilities of Somali house- the drought in 2016, an estimated 1.1 million IDPs holds. Weak demand for labor in the agricultural already lived across Somali regions. The drought sector reduced wage levels in 2017.93 As house- forced an additional 1 million people into displace- hold incomes declined, food stocks and livelihood ment between 2016 and 2017. Drought-driven dis- assets depleted in 2017. Combined with higher placement surged when the effects of the drought food and water prices, this significantly worsened were particularly severe, in the aftermath of the households’ purchasing power. 2016 Deyr and 2017 Gu rainy seasons (Figure 3.7). Lack of clean water and sanitation created con- The humanitarian response to the current crisis ditions for large-scale outbreaks of diseases like was coordinated and swift, reaching up to 3 mil- cholera and measles. Drought conditions reduced lion people through 2017. Humanitarian access the availability of water necessary for proper is better than in previous crises in 1992 and 2011, hygiene and sanitation and increased the risks where conflict and insecurity impeded humanitar- of remaining water being contaminated. These ian efforts and led to famine. Early warning sys- factors contributed to large-scale outbreaks of tems and monitoring enabled government actors measles and acute watery diarrhea (AWD)/chol- and humanitarian partners to intervene and mount era. Inadequate access to health facilities wors- a response program of US$1.2 billion in cash and ened this epidemic. At the end of 2017, there were livelihood support and health, nutrition, and WASH around 20,000 reported cases of measles and 94  UNOCHA (United Nations Office for the Coordination of Humanitarian Affairs) (2017b). 92  FSNAU and FEWSNET (2017a); FSNAU and FEWSNET (2018). 95  UNOCHA (United Nations Office for the Coordination of 93  FSNAU and FEWSNET (2017c). Humanitarian Affairs) (2017b). Drought Impact 65 Box 9  ■  The World Bank’s response to the drought Emergency Drought Response and Recovery Project. The World Bank mobilized US$50 million in grants through IDA’s Crisis Response Window to respond to the crisis in 2017. The World Bank partnered with the International Committee of the Red Cross (ICRC) and the UN Food and Agriculture Organization (FAO), supporting ICRC with US$20 million and FAO with US$30 million. The project’s objective was to address the immediate needs of drought-affected Somalis and support resilient recovery by providing livelihood support and aid the restora- tion of agricultural and pastoral production. The effort was estimated to directly support up to 523,000 Somalis through food in-kind and unconditional cash transfers, as well as 109,800 persons from rural areas through Cash- for-Work and unconditional cash transfers paired with emergency livelihood inputs. It also aimed to provide safe drinking water for up to 656,000 Somalis by rehabilitating water sources and providing water storage and treatment, and improved access to health care. The response further treated, vaccinated, and fed up to 8.5 mil- lion livestock. Drought Impact and Needs Assessment and Recovery and Resilience Framework. The World Bank, along with the UN and the EU, supported the Somali government in carrying out a Drought Impact and Needs Assessment (DINA) and a subsequent Recovery and Resilience Framework (RRF). The goal is to assess and value the impact of the drought on lives and livelihoods in Somalia, identify the root causes of recurrent drought, and develop a strategy for recovery and resilience. interventions.96 The effort reached up to 3 million drought-affected households before and during Somalis per month and contained food insecurity, exposure to drought. water shortages, and the further spread of com- municable diseases (Figure A.3). Funding require- ments for 2018 are US$1.5 billion, of which 85 Measuring the drought’s impact percent are so far unmet.97 The Normalized Deviation Vegetation Index is used to determine households’ level of drought Drought impact on welfare exposure. The Normalized Deviation Vegetation Index (NDVI) is a measure of vegetation health for and livelihoods any given region over time. It is used here to quan- tify drought severity in surveyed areas, as below- The Somali High Frequency Survey provides average NDVI values imply dry conditions and unique data to quantify the drought’s impact on below-average vegetation health. NASA’s MODIS poverty, consumption, and livelihoods. Wave 1 Terra and Aqua platform provides the daily global collected data in February 2016, immediately NDVI data at 250m resolution, which serves as the before the drought, and Wave 2 was implemented source of data for this analysis.98 The percentage in December of 2017 when the drought had taken deviation of the NDVI during the 2016 Deyr and hold of Somali regions, interviewing households in 2017 Gu rainy seasons, relative to the pre-drought severely and less drought-affected regions. Both 2012–2015 average, in a 25 kilometer radius around waves collected high-quality household data, each household, determines each household’s especially information on consumption and pov- level of drought exposure (Figure 3.8, Figure 3.9). erty. The breadth of information and the timing The 2016 Deyr and 2017 Gu rainy seasons are the of data collection facilitated an in-depth assess- evident choice for quantifying drought exposure, ment of the effect of the drought crisis on pov- as weather anomalies in the 2016 Deyr and 2017 Gu erty by comparing outcomes from more and less rainy seasons were the key drivers of the current drought (see above). Households’ level of drought exposure ranges from NDVI values of 6 percent above average to 20 percent below average in 96  FAO (Food and Agriculture Organization of the United Nations) (2012). 97  FAO (2012); Food Security Cluster (2018); UNOCHA (2018c). 98  Schaaf (2015). 66  Somali Poverty and Vulnerability Assessment FIGURE 3.8  n  NDVI deviation, 2016 Deyr season FIGURE 3.9  n  NDVI deviation, 2017 Gu season Very wet Very wet Very dry Very dry Source: Authors’ calculations based on MODIS NDVI. Source: Authors’ calculations based on MODIS NDVI. FIGURE 3.10  n  Distribution of drought exposure, Overall, Wave 1, Wave 2 –51 –48 –45 –43 –40 –37 –34 –32 –29 –26 –24 –21 –18 –15 –13 –10 –7 –5 –2 1 3 6 9 12 14 NDVI (% deviation) Overall Wave 2 households Wave 1 households Source: Authors’ calculations based on MODIS NDVI and the SHFS 2017–18. Wave 1, and from 4 percent above average to 36 interest before and after exposure to the drought. percent below average in Wave 2, reflecting the Wave 1 captured household outcomes before the overall spectrum of drought severity (Figure 3.10). beginning of the drought in early 2016. Hence, The NDVI measure also correlates significantly with none of the households interviewed in Wave 1 were households’ self-reporting to be drought affected. affected by the drought. Wave 2 captured out- comes after the drought had set in in late 2017. The The drought impact is estimated using a difference- approach relies on the fact that some households in-differences model. The difference-in-difference in Wave 2 were more drought-exposed than oth- approach is used to compare households’ level ers, because the intensity of the drought differed of poverty, consumption, and other outcomes of across Somalia (Figure 3.10). To assess the impact Drought Impact 67 FIGURE 3.11  n  Illustration of difference-in-differences approach Outcome (y) Feb-16 Jan-17 Dec-17 Wave 1 Wave 2 Counterfactual Highly drought-exposed Less drought-exposed Source: Authors’ calculations based on the SHFS 2017–18. of the drought on poverty and consumption, sample consists of urban and rural households in the difference-in-difference approach compares all regions covered in Wave 1 and Wave 2. Geo- how much poverty and consumption changed graphical coverage across waves is different, as between Wave 1 and Wave 2 for households in additional regions were surveyed in Wave 2. The highly drought-exposed areas, to how much pov- lack of complete geographical overlap impedes erty and consumption changed for households in controlling for regional idiosyncrasies of regions less drought-exposed areas over the same period covered in Wave 2 only at baseline. This implies of time. That is, if households in highly drought- that a common trend between these regions and exposed areas experienced a larger increase in others must be assumed rather than controlled for. poverty than households in less drought-exposed As a robustness check, the analysis will include areas, the interpretation is that the drought made a specification of only overlapping Wave 1 and these households poorer (Figure 3.11).99 The validity Wave 2 areas, allowing for a genuine region fixed of this interpretation rests on the assumption that effect. The additional specification restricts the changes in poverty, and other outcomes of inter- analysis to urban households in Mogadishu and est, between Wave 1 and Wave 2 would be similar North West and to rural households only in North for the compared households had the drought not West. This limits the appeal of the additional speci- happened. To make this comparison more cred- fication because it reduces the analysis to estimat- ible, the estimation controls for various observable ing a localized rather than global drought-effect. characteristics of, and factors affecting, households (Appendix C). The drought effect is estimated in a regression, using ordinary least squares (OLS) or Drought impact on poverty, Probit as appropriate (see Appendix C). consumption, and hunger The drought impact is estimated for urban and Highly drought-exposed rural households are rural households in regions covered in Wave 1 and 24 percent more likely to be poor. In rural areas, Wave 2. The analysis focuses on urban and rural an increase in drought exposure of one stan- households. It excludes IDP and nomadic house- dard deviation led to a decline in consumption of holds to make Wave 1 and Wave 2 households 19 percent, where one standard deviation means a credible comparison groups. Large-scale drought- 7 percentage-­point loss in NDVI. The reduction in related displacement implies that IDP populations consumption corresponds to an increase of 24 per- before the drought in Wave 1 were different from cent in the probability of being poor. The drought IDP populations surveyed during the drought in had no effect on poverty and consumption among Wave 2. Nomadic households do not have a per- urban households (Table 3.1). manent place of residence, so a geographical treatment assignment is meaningless. The analysis The drought’s impact on consumption is larger for less poor rural households. Implementing the difference-in-differences model with controls 99  Imbens and Wooldridge (2007). through quantile regressions allows assessing 68  Somali Poverty and Vulnerability Assessment TABLE 3.1  n  Drought impact on poverty and consumption (I) (II) (III) Sample Full urban + rural sample Full urban sample Full rural sample Outcome variable Poverty status Drought impact 0.00635 0.00696 0.238*** S.E. (0.0485) (0.0562) (0.0880) Outcome variable ln (core consumption) Drought impact 0.00478 0.00461 –0.189** S.E. (0.0370) (0.0338) (0.0876) Controls Yes Yes Yes Observations 7,214 5,678 1,536 R-squared 0.348 0.347 0.520 Source: Authors’ calculations based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Poverty status results estimated using Probit, Consumption results estimated using OLS. Drought effect expressed in standard deviations of NDVI loss. Standard errors (S.E.) are reported in the table. FIGURE 3.12  n  Drought effect along the income distribution, rural areas 20% 10% 0% Drought impact on consumption –10% –20% –30% –40% –50% –60% 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Income percentile Drought effect Smoothed drought effect 95% confidence interval Source: Authors’ calculations based on the SHFS 2017–18. the drought’s impact on consumption at differ- is similar to the overall average. With an average ent points along the income distribution. In urban poverty gap of 72 percent, this group is very poor. areas, that impact is around zero at all points of It is unlikely that these households were able to the income distribution (Figure A.4). In rural areas, cope with the drought shock more effectively than the drought affected poorer and wealthier house- wealthier households. Instead, it is more plausible holds heterogeneously: higher drought exposure that the drought affected them to the extent that had no significant impact on consumption for the they could not sustain their livelihoods and were poorest 10 percent of rural households, reduced driven into displacement. consumption by 17 percent for rural households at the twentieth percentile, and by between 20 and Highly drought-exposed rural households are 30 percent for the top 80 percent of rural house- more likely to experience hunger. As levels of hun- holds (Figure 3.12). Varying levels of drought expo- ger rose across all Somali regions (Figure 3.6), rural sure along the income distribution do not explain households in highly drought-exposed areas were these differences, as the median drought inten- most severely affected. Higher drought exposure sity among the poorest 10 percent of households led to a 13 percent decrease in food consumption, Drought Impact 69 FIGURE 3.13  n  Drought effect on hunger and food similar in urban and rural areas. This indicates that consumption drought-exposed urban households, including poor urban households, more effectively coped 40 with this shock than rural households, who were 30 more likely to be poor and experience hunger. The 20 drought further had a larger impact on wealthier Percentage change 10 rural households, while the poorest rural house- 0 holds may have lost their livelihood and become displaced. It affected rural households across all –10 Somali regions, as the impact on poverty and con- –20 sumption was significant and similar in magnitude –30 in different regional specification (Table A.4, Table –40 A.5). The drought’s impact on poverty and con- Urban Rural sumption among rural households shows that they are vulnerable to income shocks. The analysis in Hunger Food consumption Chapter 4 provides further insight into rural house- holds’ vulnerability to shocks. Source: Authors’ calculations based on the SHFS 2017–18. Another income shock could increase rural pov- accompanied by a 19 percent increase in the prob- erty by 11 percentage points. The detailed results ability of experiencing hunger in December 2017. from difference-in-differences analysis allow an Urban households were not similarly affected (Fig- assessment of how a renewed income shock ure 3.13). of the same magnitude as the 2016/17 drought would affect rural households. To model another income shock, the quantile regression estimates Policy recommendations of the drought’s effect on household consumption at different points along the income distribution The drought affected rural households severely, (Figure 3.12) are applied to the 2017 Somali High indicating vulnerability to income shocks. Higher Frequency Survey data. Based on this simulation, levels of drought exposure had no significant con- a renewed income shock could increase rural pov- sumption effect among urban households, regard- erty by 9 percentage points, from 65 to 76 percent less of their level of income. Drought intensity was (Figure 3.14). Box 10  ■  Assessing the robustness of the difference-in-differences estimates The main results were tested for robustness in several ways. The robustness of the results of this chapter were tested for the inclusion and exclusion of control variables, the exclusion of various Somali regions, and with over- lapping Wave 1 and Wave 2 regions only. Inclusion and exclusion of control variables. The drought has a significant effect on poverty and consumption among rural households regardless of which group of the defined control variables is included, and also without any controls (Table A.3). Exclusion of regions. The drought’s effect on poverty and consumption is not driven by any one region. The results hold up in several reduced samples, in which any one region covered in Wave 2 only was excluded at a time (Table A.4). Estimates are of similar magnitude. Overlapping sample only. All presented results hold in the overlapping sample as well. The various estimated drought effects of interest are slightly more pronounced than in the full sample (Table A.5). 70  Somali Poverty and Vulnerability Assessment FIGURE 3.14  n  Simulation of income shock among rural households 100 90 80 Percent of population 76 70 65 60 50 40 30 20 10 0 0 1 2 3 4 Daily core consumption expenditure per capita (US$) Poverty change Core consumption (2017) Income shock Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 3.15  n  Correlates of drought-impacted rural households 40% 36% 30% 20% Probability of being impacted by drought 20% 13% 10% 0% –10% –20% –30% –29% –40% –37% –50% Agriculture Salaried Difficult to Far away Water access labor borrow from in household money market Main source of income Access Source: Authors’ calculations based on the SHFS 2017–18. Note: Coefficients from Probit regression with self-reporting to be impacted by the drought as dependent variable. Regression with controls for drought exposure measured by NDVI, household income, and region. All reported results significant at the 5% level. Vulnerable rural households rely on agriculture drought. Agricultural households and those lack- and lack access to infrastructure and services. ing in access to infrastructure and services are thus Rural households in Wave 2 more often reported particularly vulnerable to income shocks. being impacted by the drought than urban house- holds. Among rural households, those relying on Investment in rural resilience is paramount to avoid agriculture as their main source of income were loss of livelihoods among vulnerable households. 20 percent more likely than average to be impacted The drought made rural households worse off and by the drought, even when controlling for loca- thus likely exacerbated existing vulnerabilities. A tion, income, and households’ level of drought renewed income shock could threaten livelihoods exposure as measured with NDVI. In contrast, rural of the most vulnerable. Investing in resilience is key households relying on salaried labor were 29 per- to reduce vulnerabilities and avoid livelihood loss. cent less likely than average to be impacted. Rural Agricultural households may benefit insurance households without access to water in the dwell- products as well as measures facilitating the diver- ing, agricultural households more than an hour sification of income sources. Investment in infra- away from the nearest food market, and house- structure and basic services could further improve holds who struggle to borrow money in an emer- rural households’ resilience. gency were also more likely to be impacted by the Drought Impact 71 CHAPTER 4 Displacement KEY MESSAGES IDPs and refugees are overwhelmingly young. Over locations. IDPs today rely on a mix of salaries, small one in two IDPs is under 15 years old, and less than family businesses, and aid for household income. The 1 percent are above 64 years old. The large propor- contrast in livelihoods is even more stark for refugees, tion of children drives high dependency ratios—IDP who have gone from a majority reliance on agricul- households have dependency ratios larger than one, ture before displacement to virtually no agricultural indicating that for each working-age member there is income today, instead depending on aid. a child who must be provided for. IDPs receive relatively low remittances, indicating IDPs are poorer and have worse living conditions a lack of safety nets. Only 7 percent of IDP house- than the average Somali resident. Although almost holds rely on remittances as the primary source of 7 in 10 Somali residents are poor, over three in four livelihood. The average IDP household receives half IDPs live on less than $1.90 per day, and more than the remittances of the average urban household. IDP one in two IDP households go hungry. Large numbers households were as likely to rely on remittances after of IDPs must share essential amenities such as toilets, displacement as before, indicating that remittances crowding out the improved WASH facilities in settle- do not serve as a safety net for displacement. ments. Compared to host communities, IDPs in settle- ments are also further away from essential facilities Some IDPs are better off than others. IDPs displaced such as primary schools, health centers, and markets. by climate events are poorer and have worse housing quality than those displaced by conflict. IDPs who are IDPs also have lower human capital than others. in protracted displacement—mostly in urban areas— IDPs of school-going age (6 to 17 years old) are less have better access to health care. IDP households likely to attend than urban residents. Adult IDPs are headed by a woman get only one-sixth the amount less likely than urban residents to be able to read and of remittances that IDP households headed by a man write. The educational outcomes of the IDP popula- get. tion are closer to rural outcomes and lag urban ones. However, most IDP households (three in four) are in Most IDPs report a preference to stay in their current urban areas. These gaps in educational attainment locations, but this would require substantial urban are particularly crucial since half the Somali popula- investment. More than 7 in 10 IDPs want to remain tion is less than 15 years old. As the young population in their current location, and 9 in 10 have not visited matures, there is a risk that these lags in educational their original residence since they were displaced. attainment for IDPs will translate to persistent, life- Intentions to stay are likely motivated by security—a long gaps not only in education, but also in employ- majority of IDPs cited security as the reason for choos- ment and overall well-being. ing their current location, and 8 in 10 IDPs report feel- ing safe or very safe where they currently are. IDPs Urban livelihoods today differ significantly from also perceive positive social relations with host com- IDPs’ and refugees’ pre-displacement livelihoods, munities, with 9 in 10 IDP households agreeing that indicating a need for adjustment as agricultural they have good dealings with their surrounding com- income is squeezed out. IDP livelihoods before dis- munities. However, successful local integration for placement consisted of a mix of salaries, small busi- IDPs would require substantial investment in strained nesses, and agriculture, while urban livelihoods today urban centers, which can currently only offer subpar consist largely of salaries, followed by remittances. living conditions to the displaced. The challenge of Agricultural income has been squeezed out over the ensuring sustainable livelihoods for IDPs, who have course of displacement, and many IDPs are employed come to urban centers and seem to be adjusting in helping with businesses, indicating an adjust- away from agriculture, also needs to be addressed. ment into the employment landscape of their new —continued Displacement 73 KEY MESSAGES—continued Somali refugees in Ethiopia do better than IDPs agricultural pre-displacement livelihoods have been on certain current living conditions but worse on wiped out, to be replaced by aid. The heavy depen- sustainable solutions. While refugees have lower dence on aid and large levels of low participation in poverty rates and poverty gap and better health the labor force places refugees in a situation that may outcomes, they do worse on parameters such as be addressing humanitarian needs but still leaves access to shared sanitation and electricity to charge uncertainty on sustainable developmental solutions, phones. They also have lower adult literacy than IDPs especially for livelihood, education and resettlement/ and urban and rural residents. Their predominantly return. Forced displacement is a massive humanitarian low: over 52,000 Somalia refugees have been sup- and development challenge in Somali regions. ported to return to Somalia since 2014, of whom Over 926,000 people were displaced by drought 29,000 returned between January and June 2017. between November 2016 and October 2017; and Forcibly displaced populations in Somali regions 171,000 were displaced by conflict. This represents are thus a complex mix of IDPs, returnees, and the only the latest wave of forced displacement in the caseload of refugees seeking asylum within the country, adding to a pre-existing caseload of 1.1 mil- country. lion people estimated in 2014, who accounted at the time for almost 9 percent of the total popula- Addressing this challenge is complex and requires tion (FGS 2018). Additionally, over 877,000 Somali development as well as humanitarian policy refugees live in neighboring countries, making responses. The longstanding development defi- them one of the largest refugee populations in the cits and vulnerabilities of Somali regions, includ- world.100 Most Somali refugees reside in Yemen, ing in host communities, render it challenging to Kenya, and Ethiopia.101 Refugee returns to Somalia address the needs of forcibly displaced popula- have increased in recent years, in part due to the tions effectively. The persistent and cyclical nature Government of Kenya’s decision to close Dadaab of the drivers of migration and conflict contribute Refugee Camp in 2016, but the numbers remain to entrenched conditions, which require a develop- mental, resilience-based approach to help affected populations cope with these shocks and stresses, combined with continuing humanitarian assistance 100  UNHCR (United Nations High Commissioner for Refugees) to shore up basic needs. (2018b). 101  Ibid. FIGURE 4.1  n  Number of displacements occurring by month, Jan 2016–Apr 2018 350 300 250 Thousands 200 150 100 50 0 16 6 6 6 16 6 17 7 7 7 17 7 18 8 -1 -1 l-1 -1 -1 -1 l-1 -1 -1 n- p- n- p- n- ar ay ov ar ay ov ar Ju Ju Ja Se Ja Se Ja M M M M N M N Conflict Drought Source: UNHCR-PRMN, Jan 2016–Apr 2018. 74  Somali Poverty and Vulnerability Assessment Box 11  ■  Data on Somali refugees in Ethiopia comes from the Skills Profile Survey 2017 Data about IDPs, collected in the Somali High Frequency Survey (SHFS) is supplemented by data on Somali refugees in Ethiopia from the Skills Profile Survey (SPS) 2017. The SPS was conducted across Ethiopia in regions with high numbers of refugees. The survey population consists of refugees (South Sudanese, Eritrean, Somali, and Sudanese) living in camps in Ethiopia, and Ethiopian host communities within a 5-kilometer radius of a camp. The sampling frame was the list of all refugee camps in the four main regions of the country that host refugees: Tigray and Afar (hosting mostly Eritreans), Gambella (hosting South Sudanese), Benishangul Gumuz (hosting both Sudanese and South Sudanese), and Somali (Somalis). Refugees do not enjoy rights of freedom, nor possibility to work. A total of 871 Somali refugee households were surveyed, along with 303 host community households (Table 4.1). TABLE 4.1  n  Skills Profile Survey (SPS) 2017, Ethiopia Benishangul Stratum Tigray Afar Gambella Gumuz Somali Total Refugees 894 439 1423 871 3627 (438 South Sudanese) (399 South Sudanese) (837 South Sudanese) Host community 412 0 975 303 1690 Source: Authors’ calculations based on the SPS 2017. This chapter seeks to inform such approaches by examining the multiple dimensions of poverty Displacement profile among IDPs in Somali regions, as well as among Somali refugees in Ethiopia. The data highlight Demographic profile and household the micro effects of displacement across several characteristics dimensions, including poverty, health, food secu- rity, education, jobs, gender, housing, and ser- IDPs, non-IDPs, and Somali refugees alike are vices. The analysis considers the heterogeneity of overwhelmingly young and skew slightly male. affected populations, comparing several subsets The demographic structure of IDPs and non-IDPs of IDPs (those living in and out of settlements, dis- is almost identical. About 1 in 2 national residents102 placed by conflict and climate, in male and female- and IDPs, both in and out of settlements, are headed households, recently displaced and in under 15 years of age (national residents: 47 per- protracted displacement, displaced once and mul- cent; IDPs: 51 percent; settlement IDPs: 50 percent; tiple times, and in rich and poor households), as non-settlement IDPs: 51 percent). About 2 in 3 are well as host and non-host communities in urban under 25 (national residents: 62 percent; IDPs: 65 areas, urban and rural residents, and the national percent). The majority of IDPs are thus children population. This information provides a more and youth. IDP and non-IDP households alike have comprehensive picture of displacement-related slightly fewer women than men: women make up impacts and dynamics in Somali regions to better 48 percent of national residents, non-settlement inform development-oriented, area-based solu- IDPs, and settlement IDPs (Figure 4.3). Somali ref- tions. The chapter also compares the situation of ugees in Ethiopia are even younger: 63 percent of IDPs in Somali regions to that of the sizable Somali such refugees are under 15. refugee population in Ethiopia, one of the largest recipient countries for Somali refugees. References to ‘National residents’, the ‘national population’, 102  the ‘urban population’, ‘urban residents’, the ‘rural population’, ‘rural residents’, ‘host communities’, and ‘non-host communi- ties’ in this chapter exclude IDPs and nomads. Displacement 75 Box 12  ■  Where are the IDPs? Timing of survey sampling and interpretation of spatial results The chapter examines IDPs across Somali regions, and is nationally representative; however, the regional dis- tribution of IDPs in the survey sample differs from that of other estimates. According to the SHFS data, IDPs are clustered in Banadir, Bay, Lower Shabelle, Mudug, and Lower Juba. This differs, however, from UNHCR’s current PRMN data, which have IDPs clustered in Banadir, Bay, Lower Shabelle, Hiraan, and Mudug. In the SHFS sample, certain regions with substantial numbers of IDPs, including Hiraan and Sool, (which have 7 percent and 5 percent of the total IDP population, respectively) are under-sampled, while others such as Banadir, Mudug, and Lower Juba are oversampled (for instance, Banadir has 22 percent of the actual population but 28 percent of the SHFS sample, Figure 4.2). FIGURE 4.2  n  Regional distribution of IDPs, SHFS sample, and UNHCR PRMN data Banadir Bay Lower Shabelle Hiraan Mudug Gedo Sool Bakool Middle Shabelle Sanaag Togdheer Galgaduud Woqooyi Galbeed Bari Middle Juba Lower Juba Awdal Nugaal 0 10 20 30 40 50 Percent of IDP population UNHCR—PRMN IDP population HFS sample—All IDPs HFS sample—Settlement IDPs HFS sample—Non-settlement IDPs Source: Authors’ calculation based on the SHFS 2017–18 and UNHCR-PRMN 2016–18. These differences are methodological. The SHFS sample of settlement IDPs was drawn using IDP location data from 2016, before the most recent drought event. The bulk of drought-related displacements (about 1 million IDPs) occurred from January to October 2017, influencing the spatial distribution of IDP households today.103 Further, the SHFS set of non-settlement IDPs were households in the rural and urban samples, who self-identified as having been displaced. Thus, it was not possible to stratify these households by region ex ante. Because of this, the chapter does not cut the sample of IDPs by region. The results on the regional distribution of IDPs (Figure 4.2) are presented here but are compared with that of the latest PRMN data and should be interpreted with caution. These differences do not affect how the broader survey results are interpreted. The survey itself was con- ducted from December 2017 to January 2018, after drought conditions improved, and its findings are nationally representative. The survey results further capture impacts of the drought. The timing of the sampling thus does not affect the accuracy or representativeness of the survey results themselves, which capture the impact of the 103  UNOCHA (United Nations Office for the Coordination of Humanitarian Affairs) (2017a). 76  Somali Poverty and Vulnerability Assessment Box 12  ■ Continued drought, but does mean that the results on the spatial distribution of IDPs presented in Figure 4.2 should be interpreted with caution. Host communities in the survey consist of households living around the IDP camps. Host communities, as defined in the SHFS 2017–18, were households found in areas that surround IDP camps. Thus, the host communi- ties in this survey refer to resident communities surrounding IDP camps, rather than communities that house IDPs within their households or within the resident community. Results in this chapter are interpreted accordingly. FIGURE 4.3  n  Population structure for IDP, non-IDPs and refugees by gender and age 1 1 1 0 1 1 1 1 1 0 50 16 10 Percent of population 19 18 19 12 40 16 16 17 17 7 6 30 7 9 6 7 7 7 8 7 20 33 30 26 27 24 27 23 27 24 10 21 0 Men Women Men Women Men Women Men Women Men Women National Overall IDP Non-settlement Settlement IDP Refugees (excluding IDP) IDP Under 15 years 15–24 years 25–64 years Above 64 years Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. TABLE 4.2  n  Age dependency ratios and household size by gender of household head   Non-settlement IDP Settlement IDP Refugee National Man Woman Man Woman Man Woman Man Woman   headed headed Overall headed headed Overall headed headed Overall headed headed Overall Percentage of 62.7 37.3 100.0 45.6 54.4 100.0 60.7 39.3 100.0 51.7 48.3 100.0 households Dependency ratio 1.5 1.1 1.4 1.2 1.5 1.4     1.2 1.2 1.3 1.2 Household size 5.9 5.5 5.8 5.1 5.7 5.4       5.1 5.0 5.1 Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. As with the national population, every second security often present in more formal settlements, IDP household is headed by a woman. About or because displaced women are separated or dis- 48 percent of IDP households overall are headed connected from family/social networks and have by a woman, which is the same as in the national fewer housing options outside formal settlements. population IDP households living in settlements Somali refugees in Ethiopia are more likely to be are more likely to be headed by a woman (54 per- headed by a man (61 percent). cent) compared to IDP households outside settle- ments (37 percent, <0.01, Table 4.2). This may be IDP and non-IDP households have similar char- because women are seeking the higher levels of acteristics. Households have similar numbers Displacement 77 of dependents for every working age adult: IDP FIGURE 4.4  n  IDP profile households both in and out of settlements have an 100 Percent of overall sample average of 1.4 each, compared to 1.2 nationally. The exception is female-headed IDP households out- 80 side settlements, which have only 1.1 dependents 60 for every working age adult, compared to 1.5 in 40 male-headed IDP households outside settlements (p<0.05). Household sizes are also mostly similar, 20 except that IDP households outside settlements 0 are slightly bigger, with 5.9 people on average C on et DP C or v me t lim io nt ev e an Ot nt N n h ad r pr ad d D Pro trac d is pl ra ed Bo ulti e tto ple To 4 0 60 -p or r ct tle en ed d o d M h he oo e c ot e e o e m nc e ac e e on o at len compared to the overall average of 5.1. Female- D is t t N S ll I fli et m pl ac ct m p N P on -s tle a e ra headed IDP households in settlements are also ve O om larger, with 5.7 people on average, compared to W 5.1 in male-headed IDP households in settlements. Other differences between households are not sta- tistically significant (Table 4.2). Source: Authors’ calculations based on the SHFS 2017-18.104 Displacement profile FIGURE 4.5  n  Urban/rural composition of IDPs 100 Most IDP households are in urban areas and in Percent of households formal settlements. Three in four IDP households 80 overall (75 percent) are in urban areas (Figure 4.5). 60 Six in ten IDPs (62 percent, Figure 4.4) live in for- 40 mal settlements. All such settlement IDPs, in the 20 SHFS sample, are in urban areas (Figure 4.5). 0 Overall IDP National resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Most IDPs have not gone far from home. About 7 in 10 IDP households live in the same districts as they did originally, and fewer than 1 in 10 are in a different region, federated member state, or country. Those who are displaced multiple times are more likely to travel out of their districts than those displaced only once (p<0.01). Households Overall IDP headed by a woman are significantly more likely to stay in their districts than those headed by a man Urban Rural (female-headed households: 61 percent; male- headed households: 86 percent, p<0.01). The lim- Source: Authors’ calculations based on the SHFS 2017–18. ited distances traveled could be linked to limited freedom of movement for women, proximity of available humanitarian resources or secure settle- should be interpreted with some caution, since ments, or possibly due to security risks linked to they run counter to more common understandings traveling long distances from home and outside of forced displacement in Somali regions, in which environments with available clan protection (Fig- displaced populations often experience multiple ure 4.7). displacements, due in part to forced evictions Most IDPs have been displaced only once and have traveled to their current locations with their fami- lies, though this finding should be interpreted with The variable used for the poor and non-poor comparison 104  caution. Approximately four in five IDPs (75 percent groups is a dummy variable for whether the household is poor of non-settlement IDPs and 81 percent of settle- or not, whereas the poverty statistics reported in this chap- ment IDPs) report being displaced once, and only ter are based on a variable which is the probability of being below the poverty line (using 100 imputations of the Rapid a tiny minority of IDPs report being displaced more Consumption Methodology). Thus, minor variations (less than than twice (Figure 4.4). However, these findings 1 percent) in the means of these two variables are possible. 78  Somali Poverty and Vulnerability Assessment FIGURE 4.6  n  Trends in traveling to current location, FIGURE 4.8  n  Reason for leaving original location for IDPs and refugees 100 100 Percent of households 80 80 60 Percent of households 40 60 20 40 0 IDP Refugee Settlement Non-settlement Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor 20 0 With my family Alone With a larger group Overall IDP IDP Refugee Other Drought/famine/flood Source: Authors’ calculations based on the SHFS 2017–18 and SPS Discrimination Increased violence but not conflict 2017. Armed conflict Source: Authors’ calculations based on the SHFS 2017–18 and SPS FIGURE 4.7  n  Original location relative to current 2017. location for IDPs 100 Climate-related events (drought, famine, or Percent of households 80 flood) and conflict are the main causes of dis- 60 placement cited by IDPs. About two in five IDP 40 households (38 percent) are displaced from their original locations because of climate-related 20 events (drought, famine, or flood). About another 0 two in five (40 percent) are displaced because of armed conflict in their village or another village. m nce 60 n- ttle P t o em nt Cl r vi ent om e ev ce M n he nt No hea ed d ra d lac ce ed Bo tiple To 40 n- oor or de ot cte No Se ID nf sett me e po at en d ct m p No P Somali refugees in Ethiopia are also highly likely to ed d o an a ul Protra all im ol tto l er be displaced by armed conflict. (Figure 4.8) r Ov tp sp la a Di sp lic Di W Co The main reason IDPs live where they do is Same district Same region different district improved security. This is true whether their house- Different region Outside country holds are in or out of settlements, displaced by cli- mate events or conflict, headed by men or women, Source: Authors’ calculations based on the SHFS 2017–18. or rich or poor. Over three in five non-settlement and settlement IDP households, and almost four in five households displaced by conflict, report that and/or new cycles of violence.105 Approximately they are in their current locations because of bet- four in five IDPs have traveled with their families ter security, rather than for other reasons such as to their current locations, about 1 in 10 alone, and access to humanitarian assistance or better live- about 1 in 10 as part of a larger group. Refugees are lihoods. These patterns differ slightly for house- much more likely to travel as part of a larger group holds displaced by climate, but even among these, than IDPs. (Figure 4.6). approximately half (53 percent, p<0.05) are in their current locations for better security, and the rest because they can get better access to livelihoods, employment, land and housing, or humanitarian 105  Federal Government of Somalia (2018). Also see UNHCR assistance (Figure 4.9). There are some remain- (United Nations High Commissioner for Refugees) (2016). ing differences in motivation across types of IDPs, Displacement 79 Box 13  ■  Drivers of displacement in Somali regions Although the household survey indicates that most people are displaced either by conflict or climate-related events, in practice, these categories are intertwined. The drivers of displacement in Somali regions are over- lapping, multiple, and complex. Forced displacement in Somali regions is a consequence of decades of internal conflict, insecurity, political uncertainty, human rights violations, and governance failures, compounded by cycli- cal environmental challenges, including periods of acute drought and famine. While survey respondents were asked to indicate one primary driver motivating migration, it is more likely that individuals and households were influenced by several interrelated factors, including both climate and security-related events. Indeed, drought conditions in Somali regions have been known to exacerbate conflict, while the impacts of drought are worsened by conditions of violence and insecurity. The Somalia Drought Impact and Needs Assessment reports that in Somali regions, drought conditions in 2017 have exacerbated conflicts over pasturelands and natural resources, with mediating impacts on food prices and livestock, and highlights the upsurge in communal and political vio- lence in 2017 (particularly in the southern and central regions of the country) which compounded the devastat- ing humanitarian and development impacts of drought and contributed further to displacement dynamics. Source: Federal Republic of Somalia, World Bank, United Nations and European Union. 2018. Somalia Drought Impact and Needs Assessment. Vol II, page 147. FIGURE 4.9  n  Reason for arriving at current location commonly cite better security as the main reason for being in the current location, the remainder— 100 Percent of households unlike IDPs within the country—cite humanitarian 80 assistance as the main driver. 60 40 Most IDPs have been displaced in the last five 20 years, and those outside settlements more 0 recently. IDPs outside settlements tend to have IDP Refugee Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor been displaced more recently than those in set- tlements: Settlement and non-settlement IDPs alike arrived in their current locations about two years ago but on average, non-settlement IDPs have been displaced for about two and one-fourth years, whereas IDPs in settlements have been dis- Overall IDP placed for three years (p<0.01). Non-settlement IDPs are also quicker to settle once originally dis- Better security Water for livestock placed, taking on average four months to do so, Home/land access Education/health access compared to about a year for IDPs in settlements Employment opportunities Join family (p<0.01; Figure 4.10). Humanitarian aid In contrast to Somali refugees, whose numbers Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. spiked after famine in 2011, conflict and climate- driven IDPs within the country have experienced continued and ongoing displacement. The pat- with IDPs in settlements being more likely to cite tern of displacement (Figure 4.11; Figure 4.12) joining family as a reason for being where they are, shows clear peaks, which have increased since and poor IDPs being less likely than non-poor IDPs 2013. These peaks, however, are not as dramatic as to cite security as a motivation (poor: 58 percent, that shown by similar data in other countries in the non-poor: 82 percent, p<0.01), but overall, security region with large-scale displacement.106 Although is the main motivation for all groups of IDPs. These patterns differ somewhat for Somali refugees in Ethiopia. Although such refugees also most 106  For example, see World Bank (2018e). 80  Somali Poverty and Vulnerability Assessment FIGURE 4.10  n  Years since IDP displacement and arrival in current location 8 7 6 5 Years 4 3 2 1 0 P tle t t at nce t he d ed d ed m ce le To 0 60 or r et en en en oo an de Pr cte 4 D tip Po ed on ad ct lI -s m m ev m p -p lim ole M ea ul ra ra al on tle tto on ac ed e h ot ot r v i ve N Set Bo N is ac an pr or O D spl om ot ct C N fli i pl W D on C Displaced from original location Arrived at current location Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 4.11  n  Conflict events and dates of displacement of conflict-driven IDPs 30 400 Percent of conflict-displaced IDPs 350 25 300 20 Conflict events 250 15 200 150 10 100 5 50 0 0 Jan-2009 Apr-2009 Jul-2009 Oct-2009 Jan-2010 Apr-2010 Jul-2010 Oct-2010 Jan-2011 Apr-2011 Jul-2011 Oct-2011 Jan-2012 Apr-2012 Jul-2012 Oct-2012 Jan-2013 Apr-2013 Jul-2013 Oct-2013 Jan-2014 Apr-2014 Jul-2014 Oct-2014 Jan-2015 Apr-2015 Jul-2015 Oct-2015 Jan-2016 Apr-2016 Jul-2016 Oct-2016 Jan-2017 Apr-2017 Jul-2017 Oct-2017 Battle Remote violence Riots/protests Violence against civilians Conflict-displaced IDPs Source: Authors’ calculation using SFHS Wave 2, ACLED (conflict events 2006–2017). the Somali drought displacement pattern shows Most IDPs intend to stay in their current locations spikes between or at the edges of the Gu and Deyr and only a few have revisited their original resi- rainy seasons, the displacement spikes also corre- dence. About 7 in 10 IDPs (70 percent) wish to stay late less clearly to climate and conflict events than in their current locations, and only 2 in 10 (23 per- they do elsewhere. This suggests that displace- cent) intend to return to their original place of resi- ment in Somali regions reflects underlying and dence. A minority intends to move elsewhere. This continual uncertainties related to climate and con- is in stark contrast to refugees, who are more evenly flict, rather than one-off shocks. These patterns divided between wishing to stay (42 percent) and differ for Somali refugees in Ethiopia, whose num- wishing to move on to a new area (45 percent). Few bers clearly spiked after the 2011 famine (48 per- want to return to their original residence (Figure cent arrived in Ethiopia in 2011, Figure 4.13). 4.14). Over 9 in 10 IDPs have not gone back to their Displacement 81 FIGURE 4.12  n  Rainfall anomalies, Gu-Deyr seasons, and displacement dates of climate-driven IDPs 30 450 Percent of climate-displaced IDPs 400 25 350 Rainfall anomaly 20 300 250 15 200 10 150 100 5 50 0 0 Jan-2009 Apr-2009 Jul-2009 Oct-2009 Jan-2010 Apr-2010 Jul-2010 Oct-2010 Jan-2011 Apr-2011 Jul-2011 Oct-2011 Jan-2012 Apr-2012 Jul-2012 Oct-2012 Jan-2013 Apr-2013 Jul-2013 Oct-2013 Jan-2014 Apr-2014 Jul-2014 Oct-2014 Jan-2015 Apr-2015 Jul-2015 Oct-2015 Jan-2016 Apr-2016 Jul-2016 Oct-2016 Jan-2017 Apr-2017 Jul-2017 Oct-2017 Gu season Deyr season Climate-displaced IDPs Rainfall anomaly Rainfall average Source: Authors’ calculation using SHFS Wave 2, VAM (Rainfall anomalies 2006-2017). Note: Rainfall anomaly is the monthly deviation of rainfall from the long-term average. The long-term rainfall average is scaled to 100, thus deviations are seen relative to this ‘100’ threshold. FIGURE 4.13  n  Dates of displacement for Somali refugees in Ethiopia 90 Severe famine in 80 Somalia in 2011 70 Percent of refugees 60 50 40 30 20 10 0 2011 1994 1996 1997 1998 1999 2000 2001 2002 2003 2004 2006 2007 2008 2009 2010 2012 2013 2014 2016 1995 2005 2015 Source: Authors’ calculations based on the SPS 2017. original residences. Those who have returned have health, education, and humanitarian aid, or family. done so mainly to visit family (Figure 4.15). Settlement IDPs are more likely than non-settle- ment IDPs to cite security as a reason for wanting The return intentions of IDPs are strongly moti- to stay where they are (89 percent of settlement vated by security considerations, whereas Somali IDPs vs. 75 percent of non-settlement IDPs, p<0.01), refugees outside the country are more likely to which is likely because higher levels of security are want to stay where they are for health, educa- available in formal settlements compared to out- tion, and humanitarian aid. Almost 8 in 10 non-­ side. Apart from that, IDPs cite similar motivations settlement IDPs, and 9 in 10 settlement IDPS, cite for wanting to stay where they are, whether their security as a motivation for wanting to stay where households are in or out of settlements, headed by they are. Less than half cite other factors, which men or women, displaced by conflict or climate, or include homes, land, livestock, and employment; are rich or poor (Figure 4.16). 82  Somali Poverty and Vulnerability Assessment FIGURE 4.14  n  Return intentions of IDPs and FIGURE 4.15  n  Trends in revisiting the original refugees residence location for IDPs 100 100 Percent of households 80 60 80 Percent of households 40 60 20 0 40 IDP Refugee Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor 20 0 Not gone Visit Check back family property Overall IDP status New area Original place of residence Source: Authors’ calculations based on the SHFS 2017–18. Don't want to move Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. FIGURE 4.16  n  Push factors for IDPs and refugees who don’t want to move 100 Percent of households who do not want to move 80 60 40 20 0 IDP Refugee Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Overall IDP Security Home, land, livestock, employment Health, education, humanitarian aid Family Source: Authors’ calculations based on the SHFS 2017–18. IDPs who do want to move have a broader range or out of settlements, displaced by conflict or vio- of motivations. These include getting better secu- lence, live in households headed by men or women, rity, as well as family ties and improved housing or are rich or poor. Yet at least 6 in 10 IDPs who or access to land, livestock, and employment. Over want to move cite family as a motivation, and— 7 in 10 IDPs who want to move cite security as a apart from the poorest IDPs, who may get better reason for wanting to do so, whether they are in services by being displaced—at least 5 in 10 are Displacement 83 FIGURE 4.17  n  Pull factors for IDPs who want to move 100 Percent of households 80 who want to move 60 40 20 0 Overall IDP Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Security Home, land, livestock, employment Health, education, humanitarian aid Family Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 4.18  n  Return timeline for IDPs and refugees FIGURE 4.19  n  Legal identification and access to that intend to move documentation restitution mechanisms107 100 80 Percent of households 70 that intend to move Percent individuals 80 60 or households 60 50 40 40 30 20 20 0 10 Displaced multiple Not protracted 0 Conflict or violence Displaced once Man headed Top 60 Climate event Non-settlement Woman headed IDP Refugee Protracted Bottom 40 Poor Non-poor Settlement Overall IDP Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Overall IDP Overall IDP Don't know yet More than 12 months Legal identification 6–12 months Less than 6 months Access to restoration mechanisms Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. Source: Authors’ calculations based on the SHFS 2017–18. motivated by homes, land, livestock, and employ- ment. Among the IDPs who want to move, richer IDPs plan to do so sooner than others (p<0.01), which might reflect the lower capacity of poorer families to bear the costs of moving and to deal with uncertain livelihoods. Somali refugees outside 107  World Bank (2018e). Access to legal identification is cal- the country are much less likely to know when they culated at the individual level, whereas access to restoration can move. (Figure 4.18). mechanisms is calculated at the household level. 84  Somali Poverty and Vulnerability Assessment Only a small proportion of Somalis, and an even when comparing IDPs and rural residents, 70 per- lower proportion of IDPs, have legal identification cent of whom are poor (Figure 4.20).108 Poverty is or access to mechanisms to restore documents. also deeper among IDPs than urban residents. The About 17 percent of IDPs have legal identifica- poverty gap measures how much less the average tion, compared to 36 percent of urban residents poor person consumes relative to the international (p<0.01) and 50 percent of host community mem- poverty line: it measures not how widespread bers (p<0.01); similarly, few have access to mecha- poverty is, but how deeply the average poor per- nisms to restore documents. IDPs in households son feels it. In Somali regions, the poverty gap headed by a woman are more likely to have an ID among IDPs relative to the US$1.90 a day inter- compared to those in households headed by a man national poverty line is 35 percent, meaning that (p<0.01). The poorest 40 percent of IDPs are also IDPs below the poverty line typically consume only less likely to have an ID than the richest 60 percent 65 percent of what is consumed by those who are (p<0.05) and have less access to document res- at the US$1.90 a day threshold. This gap is greater toration mechanisms (p<0.01; Figure 4.19). Other than that of urban residents (24 percent, p<0.01) than this, the rate of legal identification ownership and the national population (27 percent, p<0.01), does not differ much according to the displace- but does not differ significantly compared to rural ment circumstances of IDPs. residents (32 percent) (Figure 4.21). Poverty is more widespread and deeper among Poverty and hunger IDPs than non-host communities, but there is no significant difference when comparing IDPs and host communities.109 The poverty headcount The incidence and depth of poverty is greater ratio among IDPs (74 percent) is higher than that among IDPs than urban residents, but about of non-host communities in urban areas (64 per- the same as among rural residents. The poverty cent, p<0.05) (Figure 4.20). The depth of poverty headcount ratio is the proportion of a population who live under the poverty line: it indicates how widespread poverty is. About three in four IDPs 108  National populations reported in this chapter are of national (74 percent) live under the US$1.90 a day (2011 residents, which include urban and rural residents, and exclude PPP) international poverty line. Poverty is more IDPs and nomads. widespread among IDPs than among urban resi- 109  As noted previously, ‘host communities’ in this survey refer dents (63 percent, p<0.05), but there are no sig- to resident communities surrounding IDP camps, rather than nificant differences in the incidence of poverty communities that house IDPs within their households or within the resident community. FIGURE 4.20  n  Poverty headcount ratio 100 Percent of population 80 60 40 20 0 IDP Settlement IDP 2016 Refugee Urban host Urban non-host Urban resident Rural resident National resident Non-settlement Settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Overall IDP Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. Displacement 85 FIGURE 4.21  n  Poverty gap Percent of poverty line 60 50 40 30 20 10 0 IDP Settlement IDP 2016 Refugee Urban host Urban non-host Urban resident Rural resident National resident Settlement Non-settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Overall IDP Source: Authors’ calculations based on the SHFS 2016–18 and SPS 2017. among IDPs is also greater. The poverty gap among among those who have been displaced only once. IDPs (35 percent) is higher than that of non-hosts The poverty headcount ratio among IDPs who have in urban areas (24 percent, p<0.01), but there is no been displaced for less than five years (76 per- significant difference in the poverty gap when com- cent) is significantly higher than that of IDPs who paring IDPs and host communities (Figure 4.21). have been displaced for longer than five years (56 percent, p<0.01)—though notably, all those IDPs in settlements are about as poor as IDPs in the SFHS sample who have been displaced for outside settlements. There is no significant dif- more than five years are in urban areas. The pov- ference in how widespread poverty is when com- erty headcount ratio among IDPs who have been paring IDPs who live in settlements (76 percent) displaced only once (73 percent) is significantly compared to those living outside settlements higher than that of IDPs who have been displaced (73 percent) (Figure 4.20), or in how deep the multiple times (57 percent, p<0.01) (Figure 4.20). poverty gap is (settlement IDPs: 34 percent; non- settlement IDPs: 36 percent) (Figure 4.21). Poverty is somewhat more common among IDP households headed by men than women. Seventy-­ Poverty is much more widespread and deeper five percent of households headed by men live among IDPs displaced by climate rather than under the US$1.90 a day international poverty line, conflict. Over four in five IDPs (85 percent) dis- compared to 64 percent of households headed by placed by climate-related events (drought, famine, women (Figure 4.20, p<0.1). or flood) live under the $1.90 a day international poverty line, compared to only three in five IDPs Somali refugees in Ethiopia are somewhat bet- (61 percent, p<0.01) displaced by conflict (Fig- ter off than IDPs who have stayed within Somali ure 4.20). Poverty is also deeper among climate-­ regions. Although the poverty headcount ratio displaced IDPs under the poverty line, who have a among such refugees is still high (62 percent), the poverty gap of 40 percent, compared to 28 per- poverty gap among such refugees is lower (23 per- cent for poor IDPs displaced by conflict (p<0.05). cent), indicating that they are closer to the poverty This means that IDPs displaced by climate events line (Figure 4.20, Figure 4.21). (drought, famine, or flood) are typically consum- ing only 60 percent of what is consumed at the Hunger is more common among IDPs than hosts, US$1.90 a day international poverty line threshold urban residents, and rural residents. About 55 (Figure 4.21). percent of IDP households went at least once without having food of any kind in the last four Poverty is more widespread among recent IDPs weeks, compared to 17 percent of the host com- than those in protracted displacement, and munity (p<0.01), 25 percent of urban residents 86  Somali Poverty and Vulnerability Assessment FIGURE 4.22  n  Hunger incidence in the last four weeks 100 Percent of households 80 60 40 20 0 Overall IDP National resident Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Overall IDP Source: Authors’ calculations based on the SHFS 2017–18. (p<0.01), and 43 percent rural residents (p=0.107). different from the share of rural residents (18 per- While being inside or outside a settlement had no cent) who have such access. The quality of housing significant relation with hunger, IDPs displaced by is mostly homogenous for different types of IDPs: conflict are more likely to face hunger than those it is low for most groups and does not differ sig- displaced by climate events (p<0.05), despite nificantly whether they are in settlements or not, being less poor. This could indicate that conflict- displaced once or more, or are displaced by con- driven IDPs are in areas that are more difficult flict or climate events. The only exception is the for humanitarian actors to reach. IDPs that are in pre-housing quality of poor and non-poor IDPs: protracted displacement, or displaced more than although they have similar rates of improved hous- once, are also more likely to face hunger than those ing at present, non-poor IDPs had better hous- who have been displaced for less time (p<0.01) or ing than poor IDPs before displacement (p<0.01). displaced once (p<0.01). Poor IDPs are more likely Somali refugees in Ethiopia are likely to live in to be hungry than non-poor (p<0.1, Figure 4.22). improved housing now but are much less likely to have done so before being displaced (Figure 4.23). Access to infrastructure About 8 in 10 IDPs have access to improved drink- ing water, but this does not account for likely and quality of dwellings overcrowding of drinking water access points in settlements, so should be interpreted with cau- About one in four IDPs has access to improved tion. The share of IDPs with access to improved housing, which is much worse than among the drinking water (78 percent) is about the same as national population and host and non-host com- the share of the national population (77 percent), munities, but similar to the share among rural urban residents (85 percent), and non-hosts in residents. Improved housing is defined as living in urban areas (85 percent) who have such access, apartments, shared apartments, separate houses, but is lower than among Somali refugees in Ethi- or shared houses. About one in four IDPs (26 per- opia, 95 percent of whom have such access. The cent) currently has access to improved housing or share of IDPs with such access is higher than had it before being displaced (27 percent). This is among rural residents, only about 56 percent of much lower than the share of the national popula- whom have such access (Figure 4.24, p<0.01), and tion (59 percent), host communities (80 percent), is similar across most types of IDPs, whether they and non-host communities (75 percent) (p<0.01) are in or out of settlements, displaced by climate who have improved housing but is not significantly or conflict, were displaced recently or long ago, Displacement 87 FIGURE 4.23  n  Access to improved housing, now and before displacement Percent of households 100 80 60 40 20 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident National resident Non-settlement Settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Bottom 40 Top 60 Poor Non-poor Overall IDP Origin Now Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. live in households headed by men or women, or residents, and significantly less than among the are relatively rich or poor (Figure 4.24). However, national population, urban residents, host commu- this finding should be interpreted with caution, as nities, and non-hosts in urban areas. After adjust- the survey question on which it is based does not ing for sharing, there are no significant differences account for (nor enable disaggregation for) possi- in improved sanitation access across different ble overcrowding in access points for water, which types of IDPs. Somali refugees in Ethiopia also other analyses has indicated is a serious problem: see a stark difference in access when adjusted for IDPs are reportedly 2.5 times more likely than oth- sharing, indicating that they too face overcrowd- ers to experience problems with water points, ing of toilets, to a greater degree, than even the including overcrowding.110 IDPs (only 20 percent have improved sanitation after accounting for sharing, compared to 50 per- IDPs appear on the surface to have better access cent of IDPs; Figure 4.25). to improved sanitation than rural residents, but this advantage disappears when discounting IDP Toilet crowding is more common among climate- households who share such facilities. Almost 8 in displaced, non-settlement, and poorer IDPs. Hav- 10 IDPs have access to improved sanitation when ing access to toilets is important in stopping disease. including those whose households share such IDPs have two households per toilet, meaning that facilities as well as those who use them exclusively. toilet crowding is more common when compared This is about the same as among the national pop- to the national population, urban and rural resi- ulation and urban residents, and more than among dents, and hosts and non-hosts in urban areas, all rural residents (p<0.01), only 6 in 10 of whom have of whom have fewer than one household per toi- access to such facilities. Yet such facilities are often let. There are also large disparities in toilet access overcrowded and are no longer classified as being among different types of IDPs. IDP households in ‘improved’ if they are shared. When discounting settlements, in protracted displacement, headed those who share, the higher rates of access among by women, and in the top 60 percent of households IDPs disappears. Only half of IDP households have have between one and two households per toilet. their own exclusive access to improved sanitation Non-poor households experience less crowding facilities, which is about the same as among rural than poor households (p<0.05). C ­ limate-displaced and non-settlement IDPs are much worse off, with three or more households per toilet (Figure 110 Federal Government of Somalia (2018). 4.26). This may reflect the rapid recent increase in 88  Somali Poverty and Vulnerability Assessment FIGURE 4.24  n  Access to improved drinking water, for IDPs, refugees, and residents 100 Percent of households 80 60 40 20 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident National resident Non-settlement Settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Bottom 40 Top 60 Poor Non-poor Overall IDP Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. FIGURE 4.25  n  Access to improved sanitation for IDPs, refugees, and residents 100 Percent of households 80 60 40 20 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident National resident Non-settlement Settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Bottom 40 Top 60 Poor Non-poor Overall IDP Unadjusted for sharing Adjusted for sharing Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. drought-induced displacement: existing toilet facil- at similar distances as settlement IDPs for all four ities are likely insufficient to accommodate such a services (Figure 4.27). rapid expansion of migration flows. IDPs have lower access to charged mobile phones Host communities are closer to services than set- with network than non-IDPs within the country, tlement IDPs. Host communities are more likely but somewhat higher access than refugees. IDPs to be less than 30 minutes away to the closest are less likely to have enough electricity to charge health facility (p<0.05), the nearest primary school mobile phones than urban hosts and urban resi- (p<0.1) and the closest market (p<0.05), than set- dents overall (p<0.01 each). Conflict-motivated tlement IDPs. However, there are no significant IDPs have more access than climate-driven IDPs differences between host communities and settle- (p<0.05). The richest 60 percent and the non-poor ment IDP households in how far they are to the are also more likely to have sufficient electricity closest water point. Non-settlement IDPs are also to charge phones (p<0.01 and p<0.1 respectively; Displacement 89 FIGURE 4.26  n  Number of households sharing a toilet Number of households 5 4 sharing toilet 3 2 1 0 Urban non-host Overall IDP Urban host Urban resident Rural resident National resident Settlement Non-settlement Conflict or violence Climate event Not protracted Protracted Displaced once Displaced multiple Woman headed Man headed Bottom 40 Top 60 Poor Non-poor Overall IDP Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 4.27  n  Households more than 30 minutes FIGURE 4.28  n  Access to electricity to charge mobile from services phone 50 100 Percent of households 80 40 Percent of households 60 30 40 20 20 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor 10 0 Urban host Settlement IDP Non-settlement IDP Water point Health School Market Overall IDP Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. Figure 4.28). IDPs are more likely than urban resi- dents to be more than 15 minutes away from the Health and education closest point where they can get mobile phone reception (p<0.01), but about as far as host com- IDPs have less access to health care than urban munities. Non-settlement IDPs are closer to phone residents, and more than rural residents, while network reception than settlement IDPs (p<0.05) refugees have better health care in Ethiopia, but and male-headed households are closer than the rates of access should be interpreted with female-headed ones (p<0.01) (Figure 4.29). Somali caution. IDPs are twice as likely as urban resi- refugees in Ethiopia are less likely than Somali IDPs dents, but about half as likely as rural residents, to have electricity to charge phones, at about one to give birth at home rather than in a maternity in four refugee households. clinic, maternal and child health center, or hospital: 90  Somali Poverty and Vulnerability Assessment FIGURE 4.29  n  Under 15 minutes to network FIGURE 4.30  n  Births in health facilities, for IDPs, reception point hosts, refugees, and residents 100 100 gave birth in last 2 years Percent of women who Percent of households 80 80 60 60 40 40 20 20 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor 0 Overall IDP Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Overall IDP Overall IDP At home Maternity clinic/MCH Hospital Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. almost 4 in 10 IDP women, less than 2 in 10 urban women, but 7 in 10 rural women (p<0.01; Figure 4.30) who have given birth in the last two years FIGURE 4.31  n  Births attended by skilled health staff, have done so at home. These figures, however, for IDPs, hosts, refugees, and residents are higher than expected. IDPs are also much less 100 gave birth in last 2 years Percent of women who likely than urban residents, but more likely than 80 rural residents, to have their births attended by skilled health staff: only half of IDP women who 60 have given birth in the last two years have done so 40 assisted by a nurse, midwife, or doctor, compared 20 to 8 in 10 urban women and 3 in 10 rural women (p<0.01; Figure 4.31). Somali refugees in Ethiopia 0 IDP Refugee Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor have better access to health care—three in four births occurred at hospitals, and more than 9 in 10 were attended by a nurse, midwife or doctor. Access to health care varies greatly across differ- ent types of IDPs. IDP women in settlements are Overall IDP half as likely (p<0.01) to give birth at home com- pared to those outside settlements. Protracted Relative/friend IDPs, all of whom are in urban areas, also have bet- Traditional attendant Nurse/midwife/doctor ter health care access than recent IDPs. Overall, the pattern of disparities across groups suggests that location is an important driver of disparities Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. in access. IDPs have lower levels of literacy and schooling enrollment among those aged 6–17 is also much than urban residents and, like the rest of the pop- lower among IDPs (35 percent) than urban resi- ulation, there are stark gender gaps between men dents (64 percent, p<0.01) and hosts (62 percent, and women in literacy. The literacy rate of IDP p<0.01) (Figure 4.33). The gender gap in literacy adults (52 percent) is lower than that of urban res- is stark, and consistent across groups: the share of idents (73 percent, p<0.01) (Figure 4.32). School adult men who can read and write, compared to Displacement 91 FIGURE 4.32  n  Adult literacy rate by gender, IDPs, Adult literacy and schooling levels vary little refugees, and residents when comparing IDPs and refugees to rural residents, and when comparing different types 100 of IDPs. There are no statistically significant dif- Percent of population aged 15 or more ferences across IDPs, rural residents, and differ- 80 ent types of IDPs, except that school enrollment among those aged 6–17 is somewhat lower among 60 settlement IDPs (31 percent) than non-settlement IDPs (42 percent, p<0.1), and among the bottom 40 percent of IDPs across the income distribution 40 (28 percent) compared to the top 60 percent of IDPs (43 percent, p<0.05). The overall similarities 20 across different types of IDPs, however, suggest that the wider disparities in poverty across differ- ent types of IDPs are primarily because of their 0 present circumstances, rather than educational Men Women Overall Men Women Overall Men Women Overall Men Women Overall endowments. Somali refugees in Ethiopia have an overall adult literacy rate of 43 percent, which is Overall IDP Refugee Urban Rural lower than that of IDPs and residents. (Figure 4.32; resident resident Figure 4.33). Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. Employment and livelihoods FIGURE 4.33  n  School enrollment among the Employment school-aged IDPs participate in the labor force at similar rates 80 to the urban and rural population, while refugees Percent of population 60 in Ethiopia have much lower labor force participa- tion. Almost 5 in 10 IDPs (48 percent) aged 15–64 aged 6–17 40 are economically active, meaning that they have worked (45 percent) or have been unemployed 20 but sought work (3 percent) in the last seven days. This is similar to the economically active share of 0 the urban population (49 percent) and rural pop- Overall IDP Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor ulation (48 percent). Almost two in five inactive IDP working age adults (1 in 10 of all IDPs, whether active or inactive) are enrolled in school. This is somewhat lower than the share of urban inactive adults who are in school (p<0.05), but similar to Overall IDP rural enrollment. The remainder of IDPs in Somali regions, however (4 in 10 IDPs overall) are inac- Source: Authors’ calculations based on the SHFS 2017–18 and SPS tive and unenrolled: they are neither working, 2017. looking for work, nor in school. This is comparable to the share of urban and rural residents in this category (Figure 4.34). Somali refugees in Ethio- women, is 22 percent higher among IDPs (p<0.01), pia, however, have higher levels of inactivity, with 18 percent higher among urban residents (18 per- 60 percent neither in the labor force nor enrolled cent, p<0.01), and 20 percent higher among rural in education. Refugees in Ethiopia are not officially residents (p<0.01) but there are no statistically sig- allowed to work, which explains the low labor force nificant gender gaps in school enrollment for pri- participation rate. mary (ages 6–13) or secondary school (ages 14–17) children (Figure 4.33). There are significant gender gaps between dis- placed men and women in labor force participation 92  Somali Poverty and Vulnerability Assessment FIGURE 4.34  n  Labor force participation for IDPs, gender gap. The gender gap in being neither active refugees and urban and rural residents nor enrolled, however, is smaller among IDPs (a 23 percent gap between men and women) than it is 100 among urban and rural residents (a 33 percent gap Percent of working-age 80 for each). This may be because IDP women have greater access to schools than rural women (9 per- population 60 cent of IDP women aged 15–64 are enrolled, com- 40 pared to 5 percent of rural women), or because 20 male IDPs are much more likely to be neither active nor enrolled than urban men. It may also 0 be because such women lack alternate sources of Men Men income and are required to find work to support Overall Overall Overall Overall Women Women Women Women Men Men the home. Three in 10 male IDPs (31 percent) are neither working, looking for work, nor enrolled in IDP Refugees Urban Rural school, compared to only 2 in 10 urban men (20 percent; p<0.05) (Figure 4.34). Male IDPs are Active, employed Active, unemployed Inactive, enrolled Inactive, not enrolled almost twice as likely as female IDPs to work as salaried labor (51 percent of male IDPs vs. 31 per- cent of female IDPs, p<0.01). Female IDPs are more Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. likely than men to work on their own account (27 percent for women IDPs vs. 18 percent for male IDPs, p<0.1), and to be working as unpaid helpers FIGURE 4.35  n  Changes in employment activity after in family businesses (31 percent for women IDPs displacement vs. 18 percent for male IDPs, p<0.01) (Figure 4.38). 100 Women are much more likely than men to be Percent of employed 80 economically inactive because they are caring population 60 for their families or households. Unpaid care work is not counted in labor force participation statis- 40 tics as being economically ‘active’. Most Somalis 20 (about 7 in 10 of IDPs and non-IDPs alike) believe 0 that most or all women in their communities are Overall IDP Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor allowed to work outside the home, despite a signif- icant minority reporting that only some or almost none are (Figure 4.36). Yet even if social norms per- mit, women are much more likely than men to be unable to work or be enrolled in school because of family and household care responsibilities: among Did not work before No Yes IDPs, 59 percent of women and only 24 percent of men are economically inactive because of family Source: Authors’ calculations based on the SHFS 2017–18. and household care responsibilities (p<0.01). Rural and urban women are also more likely than men to be economically inactive because of family and and employment. IDP women are less likely to be household care (rural women: 64 percent; rural employed than men (p<0.01), and over half of IDP men: 24 percent, p<0.01; urban women: 69 per- women are neither active nor enrolled in school, cent; urban men: 38 percent, p<0.01). IDP men, in compared to less than a third (31 percent) of IDP contrast, are much more likely than IDP women men (p<0.01). Somali refugees in Ethiopia have to be not working because of illness or disability the same pattern in the labor force status of the (p<0.01), the reason cited by 30 percent of IDP genders: women are less likely to be employed or men for economic inactivity. enrolled in education, and more likely to be neither working nor enrolled. Women’s lower likelihood of The employment patterns of IDPs, host communi- being enrolled in education can translate to lower ties and refugees, and of IDPs in and out of settle- employment in the future, leading to a persistent ments, differ. Most employed IDPs work as salaried Displacement 93 FIGURE 4.36  n  Proportion of women perceived to be FIGURE 4.37  n  Reasons for economic inactivity allowed to work outside the home 100 Percent of inactive and not-enrolled population 100 Percent of households 80 80 60 60 40 40 20 20 0 0 Overall IDP National resident Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor Men Women Overall Women Overall Men Women Overall Men IDP Urban Rural Ill/disabled Overall IDP Waiting for busy season/on leave Almost none Some Majority Almost all Too young/old In school Source: Authors’ calculations based on the SHFS 2017–18. Family and household care Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 4.38  n  Main employment activity for IDPs, hosts, refugees, and rural residents 100 Percent of employed population 80 60 40 20 0 Men Women Overall Overall Women Overall Men Women Overall Women Overall Women Overall Women Overall Women Overall Men Men Men Men Men Overall IDP Refugee Urban host Rural resident Settlement Non-settlement Conflict IDP Climate IDP IDP IDP Salaried labor Own business Help in business Own account agriculture Apprenticeship Source: Authors’ calculations based on the SHFS 2017–18 and SPS 2017. labor or labor paid in kind, including in agriculture 22 percent for IDPs, p<0.01), but are less likely to (43 percent), or in non-farm businesses that they be helping in their families’ businesses (14 percent (22 percent) or their households (23 percent) own. vs. 23 percent for IDPs, p<0.1) (Figure 4.38). Settle- These patterns differ from those of host com- ment and non-settlement IDPs also have different munities, who are almost twice as likely to work employment patterns, which may be because set- in their own businesses (40 percent for hosts vs. tlement IDPs are only in urban areas. Settlement 94  Somali Poverty and Vulnerability Assessment IDPs are more likely than non-settlement IDPs to Household income work as salaried labor (p<0.05), and are less likely to farm, hunt, or fish for themselves or help on Most IDPs rely on salaried labor, small family busi- family farms (p<0.01). This difference may partly nesses, or aid/zakat to provide their main source be because all settlement IDPs (100 percent) are in of income, while refugees rely overwhelmingly urban areas, compared to only 35 percent of non- on aid. When examining household income (what settlement IDPs. Somali refugees in Ethiopia also households live on) rather than employment (what largely rely either on salaried labor (56 percent) they do), a more nuanced picture of IDP livelihoods or own businesses (32 percent) for employment emerges, which captures how aid, zakat, remit- (Figure 4.38). tances, trade, property, and other income sources contribute to household livelihoods. In Somali Most IDPs do the same work they did before regions, about two in five IDPs have salaried labor being displaced, but about half of the poorest and as their main income source, and one in five rely those outside settlements have had to change on small family businesses, but almost none rely their main employment. Almost 7 in 10 employed on trade or property income. IDPs are more likely IDPs (67 percent) report the same main employ- to rely on small family businesses than urban resi- ment activities as before being displaced. These dents or host communities (IDPs: 19 percent; urban figures are even higher for IDPs in protracted dis- residents: 12 percent, p<0.05; host communities: placement (87 percent), who may have had more 12 percent, p<0.05). IDPs are also less likely than time than others to re-establish their livelihoods, before to make a living from agriculture (15 percent and settlement IDPs (84 percent). However, non-­ before being displaced, vs. 7 percent after being settlement IDPs and the poorest 40 percent of displaced, p<0.01) (Figure 4.39). Somali refugees IDPs (who, because 69 percent of all IDPs are in Ethiopia have seen a stark shift in livelihoods. under the poverty line, are a subset of the poor), Almost 7 in 10 households relied on agriculture as are more likely to have changed their employment. their primary source of household income before Every second IDP (49 percent) living outside a set- displacement. After displacement, agricultural tlement has had to change his or her main employ- livelihoods have been almost completely squeezed ment activity since being displaced, as have over out, and more than 6 in 10 households depend on 4 in 10 (44 percent) of the poorest 40 percent of aid. A combination of low labor force participation, IDPs (Figure 4.35). especially as refugees are officially not allowed to FIGURE 4.39  n  Main source of income for IDPs, hosts, and residents 100 Percent of households 80 60 40 20 0 Urban host Urban non-host Urban resident Rural resident Before Current Before Current Before Current Before Current Before Current Non-displaced Overall IDP Settlement Non- Conflict Climate (current) settlement IDP IDP IDP Salaried labor Remittances Small family business Agriculture Trade, property income Aid or zakat Other Source: Authors’ calculations based on the SHFS 2017–18. Displacement 95 FIGURE 4.40  n  Main source of income for refugees of conflict-displaced IDPs who do so (23 percent, p<0.01, Figure 4.39). This suggests not that cli- 100 mate IDPs need less assistance, but that they may 80 get less. This may again reflect the recent rapid Percent of households increase in rates of displacement due to the most 60 recent drought, the growth of which has outpaced the ability of humanitarian actors to expand their 40 assistance to fully meet the scale of demand. 20 Average remittance amounts vary consider- ably across different types of IDPs, with set- 0 tlement, protracted, female-headed, and the Before Current bottom 40 percent of IDP households receiv- Agriculture Retail Services ing low amounts. The average annual value of Salaried labor Aid Remittances remittances for all IDP households, whether they Other receive remittances or not, is US$27 per capita, which is about half what urban residents get on average (US$56, p<0.1). There are considerable Source: Authors’ calculations based on the SPS 2017. disparities in how much different types of IDPs get on average. Settlement IDPs receive about an work in Ethiopia, and the complete lack of agricul- eighth of what non-settlement IDPs get, receiv- tural opportunities, can explain why most refugees ing only US$7 on average per capita per year in now rely mainly on aid. Apart from aid, about one remittances, compared to US$59 for IDPs outside in five refugee households get most of their liveli- settlements (p<0.1). Protracted IDPs get less than hood from salaried labor (Figure 4.40). a dollar on average per year, compared to US$37 for recent IDPs (p<0.05). The bottom 40 percent Few IDPs rely on remittances, aid, or zakat, and, of IDPs get US$6 on average, compared to US$39 although they are poorer, climate IDPs are less for the top 60, and women-headed households likely than conflict IDPs to rely on aid or zakat. get far less than men-headed households, getting Less than 1 in 13 IDPs (7 percent) relies on remit- US$7 on average, compared to US$45 for male- tances as their main source of income, and only headed households (p<0.1). These findings are 1 in 10 IDPs overall (12 percent) relies on aid or consistent with earlier surveys and likely reflect zakat. IDP aid dependency is much lower than in the extent to which such households are marginal- other countries in the region, such as South Sudan, ized and disconnected from social networks that where over 7 in 10 IDPs, and 9 in 10 refugees, would otherwise provide such support. This may rely on humanitarian assistance.111 Climate IDPs in be particularly true of minority clans that are dis- Somali regions are also much less likely to rely on connected from social networks and may have no aid or zakat than conflict IDPs (Figure 4.39). This mechanisms of support other than formal settle- is even though they are much poorer (Figure 4.20; ments (Figure 4.41). Figure 4.21), and even though drought, famine, or flood have disrupted their agricultural livelihoods: as might be expected, far fewer climate IDPs Social cohesion, justice, now rely on agriculture, fishing, hunting, and ani- mal husbandry (28 percent before vs. 13 percent and security now, p<0.01) as before being displaced, shifting to salaried labor (25 percent before vs. 41 per- Most IDPs and refugees feel safe where they are, cent currently, p<0.01) to earn an income. Only a and IDPs report good relations with the commu- tiny minority of climate IDPs (3 percent) rely on nities around them. Almost 8 in 10 IDPs (78 per- aid or zakat, which is much lower than the share cent) feel safe (moderately or very) where they are, which is similar among the national popula- tion, but somewhat less than among host com- munity members (92 percent, p<0.05). Somali 111  World Bank (2018e). The comparison, however, should be refugees in Ethiopia report very high levels of interpreted with some caution, since the South Sudan survey was conducted in Protection of Civilian camps only. perceived safety, with about 8 in 10 households 96  Somali Poverty and Vulnerability Assessment FIGURE 4.41  n  Average remittances for IDPs, hosts, and residents 120 amount (USD) 90 Remittance 60 30 0 Overall IDP Not protracted Protracted Top 60 Woman headed Man headed Bottom 40 Rural resident Urban host Urban non-host Urban resident Settlement Non-settlement Overall IDP Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 4.42  n  Perceptions of safety FIGURE 4.43  n  Perceived relations of IDPs with surrounding community 100 80 100 households Percent of Percent of households 60 80 40 20 60 0 40 IDP Refugee National resident Urban host Urban non-host Urban resident Rural resident Settlement Non-settlement Conflict or violence Climate event Woman headed Man headed Not protracted Protracted Displaced once Displaced multiple Bottom 40 Top 60 Poor Non-poor 20 0 on et DP m t lim io t ev e t ea d d -p or r D isp trac ed d ul e Bo ple To 40 60 tle n en en oo e c de de te m nc on set eme on o at len o ct ti I m N P p ed d o an ea Pr tra ll tto ra l M h h t ac e o C v ve pl c an pr r is la N S o O - om ot ct N fli Overall IDP W D C Very safe Moderately safe Neither safe nor unsafe Moderately unsafe Very bad Bad Neither good nor bad Very unsafe Good Very good Source: Authors’ calculations based on the SHFS 2017–18 and SPS Source: Authors’ calculations based on the SHFS 2017–18. 2017. feeling very safe. IDPs displaced by conflict are of IDPs, whether in or out of settlements, in house- less likely to feel safe, as are IDPs displaced mul- holds headed by men or women, in protracted or tiple times: 3 in 10 conflict-displaced IDPs (31 per- non-protracted displacement, displaced once or cent) and almost 4 in 10 IDPs displaced multiple multiple times, or rich or poor (Figure 4.43). times (37 percent) feel very unsafe, moderately unsafe, or neither safe nor unsafe, compared to 15 Somali refugees in Ethiopia have markedly posi- percent of climate-displaced IDPs (p<0.1) and 19 tive relations with their host communities, more percent of IDPs displaced only once (p<0.05) (Fig- so than refugees from other countries. The Somali, ure 4.42). This overall perception of safety among Tigray-Afar, and Beninshangule regions in Ethiopia the IDP population at large is in line with percep- host Somali, Eritrean, and Sudanese-South Suda- tions of host community relations. Almost 9 in 10 nese refugees, respectively. In the Somali region, IDPs (87 percent) think that their relations with the large majorities of the host communities disagree communities around them are good or very good. with statements such as “Ethiopians want refu- This likelihood is consistent across different types gees to return to their homes” and “The arrival of Displacement 97 FIGURE 4.44  n  Perceptions of refugees among host communities in Ethiopia 100 host population 80 Percent of 60 40 20 0 Somali Tigray/Afar Beninshangule Somali Tigray/Afar Beninshangule Somali Tigray/Afar Beninshangule Ethiopians want refugees The arrival of refugees has The arrival of refugees to return to their homes made it more difficult for has brought insecurity people in this community to the area to find work Strongly disagree Slightly disagree Neither agree nor disagree Slightly agree Strongly agree Source: Authors’ calculations based on the SPS 2017. refugees has brought insecurity to the area.” In housing and access to sanitation, and are further the Tigray and Afar regions that host Eritreans the away than others from schools, health centers, and picture is slightly more mixed, although relations markets. They receive low levels of remittances and are overall perceived positively. In contrast, host have few safety nets. They also have less access to community sentiments in the Benishangul Gumuz health care and schooling, which, combined with region (which hosts 75 percent Sudanese and 25 hunger, can translate into persistent, lifelong gaps percent South Sudanese refugees), relations are in well-being. not very positive: most host community members agree with the aforementioned statements (Fig- Advancing durable solutions for displacement- ure 4.44). The more positive perceptions of Somali affected populations in Somali regions is thus a refugees tie with their longer permanence as refu- central challenge for longer-term stability and gees, common ethnic identity, and similar clan sys- development. Displacement is widespread; its tem. These commonalities entail a higher degree deprivations many and deep. Development and of integration, from economic self-reliance and poverty alleviation strategies for Somali regions higher participation in the labor market to better will thus not be achieved without addressing housing conditions and lower poverty incidence. displacement-related vulnerability and ensuring that displaced populations are integrated into society, the economy, and development policy and Policy recommendations planning.  IDPs should be able to choose freely whether IDPs in Somalia are mostly young, poor, and out to return, stay, or settle elsewhere. International of work; often go hungry; have poor housing, standards highlight that durable solutions for water, health, and schooling; and are increas- displaced populations may entail returning sus- ingly concentrated in urban areas. Half of IDPs tainably to places of origin, locally integrating are under the age of 15, half experience hunger, into current communities, or settling in another and three in four live on less than the international part of the country; particularly important is the poverty line of US$1.90 a day, consuming on aver- right of displaced populations to choose freely age about 35 percent less than that. Three in four between these options.112 More specifically, in the are in already strained urban areas. About a third have had to change their livelihoods, many shifting out of agriculture; and 4 in 10 are neither work- For example, Council (1998). Also see the “IASC Framework 112  ing, looking for work, nor in school. IDPs have poor on Durable Solutions” (2010). 98  Somali Poverty and Vulnerability Assessment context of Somali regions, advancing durable solu- basic services, including health and education, tions for displacement-affected populations—­ is also important in enabling communities to including IDPs, returning Somali refugees, and become more resilient. communities—should further reflect: host ­ ■■ Strengthen the viability of urban and peri- ■■ Support for return to communities of origin in urban areas and enable IDPs to better integrate areas where conflict and climate-related events into them. Seventy percent of IDPs express a have abated and where voluntary, safe, and dig- desire to stay in their current locations, which nified return is feasible; are mostly in urban areas. This is consistent with other studies, which indicate that even ■■ Support for local integration for those unwilling when climate-related conditions in communi- to return to areas affected by continuing conflict ties of origin improve, IDPs may feel too unsafe or climate-related events, or other factors; and to return.113 Given that IDPs are concentrated in urban centers and secondary towns, and that ■■ Support as feasible for those currently displaced rapid urbanization is having an impact on exist- in areas of continuing conflict and/or humanitar- ing development deficits, vulnerability, and mar- ian emergency or for those interested in return ginalization in Somali cities, strengthening the even in the context of ongoing instability. viability and resilience of Somalia’s urban and peri-urban areas to enable IDPs to integrate into Providing durable solutions in Somali regions the local economy and become more self-­ reliant requires a broad-based approach led by govern- is critical. This will entail investing in services ment. This entails a combination of area-based, and infrastructure (including housing, shelter, cross-sectoral, multi-stakeholder needs and water and sanitation, and health and education) rights-based policies and investments, in which to help cities better absorb massive population humanitarian and development partners engage growth and provide services for displacement- collaboratively under government leadership. affected populations and host communities Enabling government ownership and leadership alike. There is also a need to empower municipal across any policies and investments is a prior- authorities to plan, monitor, and budget for city ity. Interventions should align with the develop- growth. At the same time, the cities are already ment priorities for durable solutions outlined in sites of innovation, with extensive private sector the National Development Plan, as well as other delivery mechanisms for services, financial government-led efforts, including the Recovery ­ investment, and job creation, which can be fur- and Resilience Framework (RRF) in development ther harnessed. to respond to the most recent drought. Efforts should further build on other ongoing initiatives, ■■ Support rural resilience and recovery to enable including the Durable Solutions Initiative and safe and voluntary return and reintegration: regional initiatives such as the Comprehensive Ref- Although IDPs have mainly moved from rural to ugee Response Framework, the Nairobi Declara- urban areas, investing in rural solutions to sup- tion on Durable Solutions for Somali Refugees and port return and recovery, and to provide oppor- Reintegration of Returnees in Somalia, and ongo- tunities in rural areas, should also be pursued. ing engagement by IGAD’s Regional Secretariat on The survey findings highlight that socioeconomic Mixed Migration and Forced Displacement. and human development indicators of IDPs are often comparable or even better than those of rural residents, highlighting the vulnerabilities Policy and program recommendations and development deficits confronting rural pop- ulations in Somali regions. Improving access to Policy and program recommendations include basic services and investing in socioeconomic the following: infrastructure will be critical in supporting IDPs who wish to return. This will likely require start- ■■ Continue to provide humanitarian assistance up assistance and support to restore livelihoods. to address basic needs and support resilience: With more than half of IDPs reporting hunger, continuing life-saving activities to support basic See UNHCR (United 113  Nations High Commissioner for needs remains critical. Expanding access to Refugees) (2016). Displacement 99 Consideration for investments may include cash should consider key protection provisions to transfers for basic consumption, skills develop- minimize potential exposure to harm, harass- ment, and other forms of livelihood support, ment, or forms of gender-based violence. including inputs for agricultural production or restocking of livestock for pastoralist activities. ■■ Support policy and planning solutions for Interventions may also include consideration for improved access to land, housing, and shelter: developing systems to enable recovery of lost Insecurity of land tenure constitutes a significant assets, land, or repair/restoration of housing. challenge in Somalia, which has had a major influ- ence on the success of housing and resettlement ■■ Promote livelihood and employment oppor- provisions in Somalia to date. Other studies and tunities: Employment and labor force partici- humanitarian reports indicate that forced evic- pation among IDPs is low. Enabling access to tions due to land tenure insecurity are a common livelihoods, employment, and opportunities to feature of urban life and perpetuate cycles of earn an income is critical both for household displacement. As this survey further highlights, stability and resilience, as well as for local eco- lack of access to improved housing for three- nomic development and growth. In urban set- quarters of IDPs also constitutes a major barrier tings, this may include expanding salaried labor to development and resilience across multiple opportunities, for example through public work dimensions. These findings underscore the need schemes or other infrastructure investment for laws, frameworks, and policies to assure both activities. Development investments targeting secure property rights and to identify housing male or youth employment should investigate planning and policy solutions for IDPs and host integrated approaches that combine business communities alike. Addressing land tenure and skills development, vocational training, or cash housing disputes may further require establish- transfers with cognitive and non-cognitive ment and mediation through local-level dispute skills building, which have had demonstrated resolution mechanisms. effectiveness in other high risk contexts and may be appropriate for addressing psycho- ■■ Promote protection and social cohesion: While social ­ challenges—e.g., trauma, depression, survey findings indicate general positive feelings dislocation—that may impede participation in ­ of safety and cohesion by displaced popula- employment opportunities.114 Employment and tions, humanitarian and development program- livelihood initiatives should further investigate ming should necessarily consider interventions gender-differentiated approaches to address to strengthen social cohesion and protection key barriers to women’s economic empower- considerations to minimize potential grievance ment, particularly as linked to entrenched social and monitor or address tensions between dis- norms and expectations for women’s domestic placed and host communities in both urban and care burden. Policies and interventions to enable rural environments. women to engage in economic opportunities For information related to psychosocial impacts of conflict on 114  Somali men and boys and associated implications for engaging in employment, see “The Impact of War on Somali Men” (2015). For evidence of integrated economic empowerment program- ming targeting high risk youth in Liberia see “The Sustainable Transformation of Youth in Liberia (Styl) Program” (2015). 100  Somali Poverty and Vulnerability Assessment CHAPTER 5 Social Protection KEY MESSAGES Somalia is prone to both natural and man-made relatives, or moneylenders; or using other social shocks and has inadequate risk management capac- networks to smooth consumption. Households also ity at both national and household levels. It has led relied on using assets to generate more income; sup- to extreme poverty and vulnerability where Somalis plying more work; or allocating more hours to work have limited economic opportunities and face severe by those who are already employed. Due to inade- constraints on livelihoods through losses to produc- quate risk mitigation capacity at household and com- tive and physical assets, access to farmlands, fishing, munity levels, there was a direct impact on household and pastoralists’ traditional routes for tending live- consumption and wealth, which also led to high levels stock. This cycle of shocks has increased their vulner- of food insecurity. ability to future shocks as there is very limited access to public and private insurance systems. A social protection system can help address the vulnerability experienced by households through Almost two in every three Somali households preventing and mitigating their impact. An efficient reported experiencing at least one type of shock social protection system responds to the needs of the in the past 12 months. Of those who experienced a population both under emergency and normal cir- shock, one in every two households reported expe- cumstances and relies on information on the causes riencing the drought. Households that are more and type of risks that the population is exposed to likely to experience shocks are mostly male-headed, and needs protection against. Hence, it consists of elderly, nomadic, and poor. Wealth plays an essential strategies that ex ante prevents poverty and ex post role in reducing vulnerability to shocks where a one alleviates poverty. Before the risk, it relies on mea- percentage increase in wealth is associated with a 24 sures to prevent its occurrence or at least prepares to 56 percent decrease in probability to experience the households in a way that can help them mitigate drought and loss of crops or livestock, ceteris paribus. its impact. After the shock, it relies on different strate- gies to help the household to cope with it. Ninety-five percent of Somali households experi- enced loss of income or assets because of shocks. Cash transfers are an important social protection The drought in 2017 alone led to pastoralists los- intervention that enable households to increase ing around 70 percent of their average annual investment in productive assets, savings, and other cash incomes while agropastoralists lost around income generating activities. Children in recipient 30 percent. households exhibit higher school attendance rates. Such households also avail health services more and The majority of Somalis relied on self-help or self- pay off their debts. Cash transfers also encourage insurance mechanisms to cope with the shocks. They households to save and build household resilience include selling, pledging, or mortgaging their physi- that can help them to smooth their consumption in cal and productive assets; borrowing from friends, an event of a shock. Somalis are vulnerable to various covariate (i.e., shocks have contributed to the extreme poverty, community level shocks such as natural disasters vulnerability, and displacement in the country. and epidemics) and idiosyncratic (i.e., household A coping mechanism, for most Somalis, in the level shocks such as injury, death or unemploy- absence of government-led support, is to rely on ment) shocks, which have become a threat to informal safety nets and humanitarian assistance. their well-being. Somalia has faced almost three However, the capacity and reach of such mecha- decades of humanitarian crises caused by recur- nisms remain limited. There were 1.1 million inter- rent climatic and conflict related shocks. These nally displaced persons (IDPs) and around 1 million Social Protection 101 Box 14  ■  What is vulnerability? Vulnerability refers to the potential risks and shocks that can negatively impact an individual’s welfare. Usually poor households are also more vulnerable, as they lack access to resources required to protect themselves against shocks, but poverty is not the only predictor of vulnerability. Being vulnerable refers to being prone to uninsured shocks and risks that can threaten one’s livelihood or survival or both. In contrast to poverty, which is an ex post measure of household’s welfare, vulnerability refers to an ex ante risk that can push a non-poor household into poverty or an already poor household deeper in poverty. In this framework, vulnerability depends on exposure to risks and shocks and lack of access to adequate resources and social protection mechanisms to manage these risks. Some groups are vulnerable only when exposed to a shock while others are in a chronic state of vulnerability with their livelihoods always at risk. Poor households are generally more exposed to risks while also less protected from them as they are less likely to be insured against risks and do not have access to formal and informal safeguards. Source: Holzmann (2001); Chaudhuri (2000); Tesliuc and Lindert (2004); and Hoogeveen, et al. (2004). Somali refugees across the borders before the 2017 losses of productive assets, access to farm lands, drought started.115 And 3.2 million Somalis were fishing and pastoralists’ traditional routes for tend- declared severely food insecure at the onset of the ing livestock. drought. Just the drought alone in 2017, caused an estimated US$3 billion in damages and losses in Somalia has been experiencing recurrent climatic Somalia. An effective and responsive social protec- shocks such as droughts and floods because of tion system is based on a thorough understanding its geographic location. The presence of con- of poverty and vulnerability so that relevant risk flict and violence has exacerbated the impact of mitigating and coping strategies can be adopted. climate shocks by affecting access and mobility. Somalia had its first famine in 1991. In 2011, Soma- lia experienced its most destructive drought that Sources of vulnerability put 750,000 people at the risk of starvation and caused almost 260,000 deaths.117 A few years later at macro level in 2015, Somalia was hit by El Niño with below average rains in the following two consecutive Somalia’s history is rife with conflict and vio- years triggering a potential famine. As a result, lence, which often led to physical and economic water reserves have been depleted, directly affect- displacement, combined with loss of life and ing agricultural and livestock sectors. The 2017 productive assets. Most of the armed conflicts in drought had the largest negative impact on agri- Somalia are along clan lines with clan identities culture and livestock accounting for 56 percent of being used to gain control of power and resources total losses (Figure 5.1). such as land and water points. These resources are fundamental for survival in Somalia where agro- Loss in agriculture and livestock sectors directly pastoralism is the main source of livelihood and or indirectly impact welfare of Somalis, as these access to land and water translates into improved sectors form the backbone of the economy and livelihoods and a stable food supply. Force and are the largest source of employment, income, violence are often used to grab land, which ensues and exports. Agriculture and livestock contribute a cycle of conflict.116 Al-Shabab’s increased domi- around 65 percent to Somalia’s Gross Domestic nance in southern Somalia has also worsened the Product (GDP) and represent 93 percent of total security situation. This continuous threat of vio- exports. Around 23 percent of the total population lence often leads to reduced economic opportuni- is agropastoralist and is dependent on a mix of ties and severe constraints on livelihoods through crop production and livestock rearing.118 Both sec- 115  https://reliefweb.int/disaster/dr-2015-000134-som 117  FSNAU (2011). 116  World Bank (2005). 118  World Bank (2018c). 102  Somali Poverty and Vulnerability Assessment FIGURE 5.1  n  Distribution of losses incurred due to 2017 drought by sector 23.34% Environment, clean energy and natural resource management 56.52% Agriculture— livestock 0.14% 2.09% Nutrition Livelihoods and employment 1.46% Health 5.05% Other 0.92% Water supply and sanitation 0.45% Agriculture— fisheries 11.09% Agriculture—irrigated and rain-fed crops Source: World Bank (2018c). tors are heavily dependent on favorable climatic 6.2 million people—more than half Somalia’s conditions. The recurrent droughts caused loss of ­ population—were estimated to be food insecure crop production due to reduced cultivated land (IPC Phases 2, 3, and 4).123 Somalis, especially those area as well as the reduced yields. The livestock living in rural areas, totally lost access to food mar- sector has also suffered due to a severe dearth kets and even those who still had access to mar- of water and unavailability of pasture for the live- kets experienced much higher prices because of stock. It also disrupted the normal migration pat- limited supply. It restricted household’s capacity to terns of pastoralist households that are driven by access and procure food.124 searching for grazing land and water for the live- stock. The drought in 2017 alone led to pastoralists Ever since the civil war of 1991, the governance losing around 70 percent of their average annual structures and institutions have deteriorated cash incomes, while agropastoralists lost around causing political fragility. Only in 2012, the Federal 30 percent.119, 120 Government of Somalia was established but still lacks technical and institutional capacity to deliver Dwindling food supply causes a hike in food goods and services adequately. In the absence of prices, which aggravated food insecurity. The formal institutions and regulatory structures, the increase in cereal prices is closely associated with household is left on its own to cope with shocks. irregular rainfall.121 In 2017, there was a sharp drop in crop production in Somalia due to the drought, It is hard to distinguish between the impact of with maize and sorghum harvests being 75 per- conflict and climatic shocks because the politi- cent lower in 2017 than in previous years.122 About cal economy of the two is closely intertwined. The impact of natural disasters is compounded by the ongoing conflict and insurgency and political 119  World Bank (2018c). 120  Zanini, et al. (2018). 121  World Food Programme (2011). 123  World Bank (2018c). 122  http://www.fao.org/news/story/en/item/470220/icode/ 124  Zanini, et al. (2018). Social Protection 103 instability. In 1991, food aid became part of the war another constraint, with domestic revenue repre- economy, when different clans fought over getting senting only 2.1 percent of total GDP. Households access to it.125 Similarly, due to restrictions on trade also lack access to formal insurance and credit and freedom of movement caused by the current markets.129 insecurity and conflict, Somalis living in southern Somalia had very limited access to humanitarian At the community level, clans have played a cru- funding and other external resources during the cial role in such circumstances through a network 2011 drought. It led to spreading of famine across of informal safety nets by establishing charity all regions of south Somalia. mechanisms (e.g., sadaqah) and sharing livestock as well as their products (e.g., irmaansi).130, 131 Such networks are exclusionary in nature, as the access Inadequate risk management is mostly limited to clan members. Also, since they are informal by definition, they do not have an insti- capacity tutional setup or regulatory framework to ensure maximum coverage and timeliness when respond- An efficient risk management system includes ing to shocks such as a drought. The resources are mechanisms for risk prevention, risk mitigation, limited and cannot cover all vulnerable households and coping strategies designed to alleviate the even within the same clan. negative impact of shocks.126 Its objective is to address the immediate impact of shocks by pre- Most households when exposed to a shock, take paring people to cope and to facilitate investments up loans from formal and informal networks to that can reduce household vulnerability to future smooth consumption during periods of shocks. shocks. It includes, but is not limited to, pooling This increases their risk exposure which is already of resources using both formal and informal net- high due to climatic and conflict conditions. On the works such as market-based arrangements, pub- other hand, informal mechanisms such as remit- lic or government led arrangements, and informal tances, have served as a lifeline. Somalia receives or community-based arrangements. The specifics around US$1.4 billion in remittances every year, of this system will vary by type and scale of risk which is around 23 percent of its GDP.132 It is facili- but will be based on three components: acquir- tated by high mobile phone penetration (around ing knowledge (gathering information on poten- 70 percent), which has made it possible to reach tial risks and their impact), obtaining protection networks that are physically separated whereby (to reduce the likelihood of experiencing risk), and allowing households to tap into those resources. procuring insurance (to transfer resources between Beneficiaries of cash transfers tend to use funds good and bad periods to smooth consumption).127 first to play off existing debt. This in turn makes households more resilient to future shocks.133 Somalia’s vulnerable population has high expo- sure to risk and lacks access to public and pri- Humanitarian programing is another source of vate sector safety nets and insurance systems. assistance for vulnerable households but due to Somalia’s authorities have inadequate capac- lack of resources, access, administrative capac- ity to mitigate risks and to protect households ity, and coordination, households often remain against shocks, due to a lack of institutional setup excluded. In recent years, cash-based transfers required to administer such programs. Humanitar- have become more prevalent and were instru- ian organizations are filling the void.128 However, mental in the response to the drought in 2017. the government is inclined toward transitioning Approximately 3.2 million individuals received from short-term emergency response to a long- cash transfers in October 2017 only, with the help term and stable safety nets program. But it lacks technical and institutional capacity to administer an expansive program. Low revenue generation is 129  World Bank (2018c). 130  Majoka (2017). 131  World Bank (2005). 125  World Bank (2005). 132  https://qz.com/848447/gift-remitting-somalis-in-the- 126  De Ferranti, et al. (2000). diaspora-send-1-4-billion-in-cash-remittances-every-year/ 127  World Bank (2013). 133  https://icai.independent.gov.uk/html-report/effects- 128  World Bank (2017b). dfids-cash-transfer-programmes-poverty-vulnerability/ 104  Somali Poverty and Vulnerability Assessment FIGURE 5.2  n  Coping strategies in response to the 2017 drought 25.0% Recieved humanitarian assistance 51.8% 31.1% Sold farming land Borrowed other assets 37.5% Sold breeding stock 44.6% Borrowed money 48.1% Sold draught animals 43.2% Out-migrated to look for food or work 24.6% Sold milking animals 29.6% Sold household valuables 30.5% Sold farm implements Source: World Bank (2018c). of various NGOs.134 Mostly only those individuals in time of emergency. Loss of productive asset is are targeted that are in acute need, which is IPC 3 a direct shock to income whereas loss of physical and IPC 4 level of food insecurity.135 Some of these asset indicates reduced savings. transfers are one-time only whereas others con- tinue for up to 2 years. However, due to the secu- Another common coping strategy is migrating rity situation, it has remained a challenge to reach for food or work. Because of the drought, people rural parts of central and southern Somalia.136 start moving to places to have better access to safe shelter, food, and water for their own survival Households, with no access to formal or infor- as well as for pasture for their livestock.138 Migra- mal safety nets, resort to coping strategies that tion, usually adopted as a coping strategy, can also are detrimental to their well-being and create a compound issues related to displacement. The vortex of increasing risk and vulnerability which recurrent shocks related to natural hazards and is difficult to exit. Other than the direct negative conflict determine the patterns of migration and impact of a shock on a household’s welfare, it can displacement. IDPs generally belong to marginal- lead to adoption of negative coping mechanisms ized groups, live in informal settlements with poor such as selling or consuming productive assets, conditions, and are more prone to violence and incurring debt, taking children out of school, fore- discrimination.139 going medical care or reducing the share of meals consumed.137 In response to the 2017 drought, Conflict and natural disasters are two of the major households coped by selling their assets such as contributors to hunger and poverty, making them farming land (52 percent), breeding stock (37 per- essential targets for creating resilience.140 Build- cent), draught animals (48 percent), milking ani- ing resilience means to enable households to pro- mals (25 percent), and household valuables (30 tect their assets and level of well-being during a percent; Figure 5.2). Poor households mostly use shock and to bounce back to the level of welfare informal savings arrangements such as buying jew- prior to the shock. With more than half of the elry or saving under the mattress that can be used population living below the extreme poverty line and suffering from food insecurity, it is important 134  https://ocha-dap.github.io/hdx-somalia-cash-v2/ 135  World Bank (2017b). 138  http://www.internal-displacement.org/countries/somalia 136  FSNAU (2017). 139  Ibid. 137  Hoogeveen, et al. (2004). 140  Myers (2017). Social Protection 105 Box 15  ■  Data caveats for vulnerability analysis The data used for this analysis is the second wave of the Somali High Frequency Survey. The survey instrument relies on self-reported information on shocks and risks while certain groups are more likely to report experienc- ing a shock than others. For example, richer household overreport whereas poor households underreport illness episodes. Quantitative surveys are also limited in their ability to capture certain types of shocks. For example, experience of discrimination or corruption can be explored better in qualitative studies. Shocks are only one of many factors that affect household welfare. At the time of data collection, Somalia was experiencing a drought, potentially subduing other types of shocks being reported.141 Source: Dercon and Krishnan (2000); Hoogeveen, et al. (2004). to understand drivers of poverty and vulnerability, of the population both under emergency and nor- identify circumstances where poverty and vulner- mal circumstances and relies on information on ability persists, and map livelihood strategies that the causes and type of risks that the population is are employed to survive and cope. In general, the exposed to and needs protection against. Hence, poor are more exposed to risk with little access to it consists of strategies that ex ante prevents pov- preventative measures. In this context, a poverty erty and ex post alleviates poverty. Before the risk, and vulnerability analysis can inform the govern- it relies on measures to prevent its occurrence or ment policies around risk management and resil- at least prepares the households in a way that can ience building. help them mitigate its impact. After the shock, it relies on different strategies to help the household A social protection system can help in address- cope with it.144 Building a social protection system ing household vulnerabilities through prevention is a high priority on the agenda of the Federal Gov- of shocks and mitigation of their impact. Poverty ernment of Somalia, as emphasized in the National can be transient or chronic in nature, both caused Development Plan 2017–19. by different factors and, thus, also with different remedies, which are important to understand to inform social protection policy.142, 143 An efficient Experience and impact of shock social protection system responds to the needs Both exposure and experience of shock affects the behavior and welfare of vulnerable house- 141  There is no information on certain characteristics of shocks, holds. Exposure to risk can make a household such as exact timing and duration of shock and whether the poor but at the same time, a poor household is household could recover from it. The recall period is 12 months more likely to take decisions that increases its where all the shocks that a household experienced are lumped together. The respondents are not asked to quantify the impact exposure to risks. A vulnerable household will of shock in monetary terms so it is also hard to assess their allocate a large share of its welfare to smooth its impact. There is information on whether the shock had a nega- consumption in response to a shock. This can push tive impact on income, assets, or wealth but without quantify- the household into poverty or further increase its ing the impact, it is not possible to identify households that severity. Similarly, a poor household is less likely to were affected the most. However, this data allows to explore the link between household characteristics and these shocks. save or invest in insuring its productive assets. This As Somalia is prone to recurrent climatic shocks, this analy- further increases its vulnerability to shocks. sis has important insights into how Somalis responded to the drought as well as the measures they took to cope with the Almost two in three Somali households (66 per- shock. cent) reported experiencing at least one type of 142  Jalan and Ravallion (2000). 143  According to the World Bank’s Social Protection & Labor strategy, social protection programs and policies “help indi- viduals and societies manage risk and volatility and protect them from poverty and destitution—through instruments that improve resilience, equity, and opportunity.” 144  Hoogeveen, et al. (2004). 106  Somali Poverty and Vulnerability Assessment Box 16  ■  Social protection systems in Kenya and Ethiopia Ethiopia’s Productive Safety Nets Program (PSNP) targets the poorest 6.5 million beneficiaries, however with a flexible caseload that can include an additional 3.1 million people in the event of a climate shock using a fed- eral contingency budget. This kind of dynamic targeting helps to respond to impacts of recurrent droughts and chronic food insecurity. It has incorporated public works activities that help build climate resilience by providing households with alternative livelihood strategies. In times of crises, it increases the duration of the program by three months. Kenya’s Hunger Safety Nets Program functions in a very similar way in terms of scaling up during climate shocks. It is an unconditional cash transfer targeted at the poorest households but increases its caseload depending on triggers, such as changes in the Vegetation Condition Index. As a result, beneficiary households in Kenya and Ethiopia are more resilient to climate shocks. For example, Hunger Safety Net Program (HSNP) households in Kenya are more likely to save, whereas households engaged in PSNP’s public works program are less likely to engage in distress selling of assets to meet food or cash needs. Source: World Bank (2018f); World Bank (2017c); Hoddinott, et al. (2015). FIGURE 5.3  n  Incidence of reported shocks among Somali households 66% 70 Percent of households 60 50 40 30 20 10 0 Drought or Loss of Water High food Reduction Theft Conflict Other irregular crop or shortage prices in income natural rain livestock for cattle or farming By type of shock Any shock Source: Authors’ calculations based on the SHFS 2017–18. Note: Households were asked to select all the shocks that they experienced in the past 12 months. Percentages indicate share of Somali households that reported experiencing a shock. shock in the past 12 months.145 Due to the 2017 experienced a shock, one in every two households drought, most of the reported shocks are related reported experiencing the drought and one in four to fluctuation in climate and its impact on liveli- households reported loss of crops or livestock hoods and economy. In an agropastoralist econ- and shortage of water for farming or cattle. One omy such as Somalia, household welfare is closely in every five households experienced high food linked with changes in rain patterns. Of those who prices. (Figure 5.3). Two of five Somali households experienced mul- 145  There were 18 categories of shock in the dataset which were tiple types of shocks within a year. The negative collapsed into 8 categories presented in the graphs. Loss of impact of each shock is greater if a household crop and livestock refers to crop failure; crop disease or pest; experiences multiple types of shocks simultane- and livestock death or disease. Reduction in income includes ously as it leads to accumulation of vulnerabilities. loss of remittances or other assistance; job loss or business fail- Poorer households are more likely to experience ure; and loss of a household member or main earner due to illness or accident. Conflict covers both experiencing violence more than one type of shock, but it is hard to con- and land eviction whereas other natural shocks include floods clude anything about the direction of causality, or landslides and fire. i.e., poor households are more likely to experience Social Protection 107 TABLE 5.1  n  Incidence of types of shocks among rural areas reported experiencing a shock in the poor and non-poor households past 12 months (73 and 72 percent respectively). In comparison, only one in three urban households Types of shocks Poor Non-poor report a shock. IDPs living in settlements report experienced in the households households experiencing conflict and violence more than the past 12 months (%) (%) other population subgroups. In Mogadishu, cer- Did not experience 35 38.5 tain militia groups continue to operate in the city a shock even after withdrawal of Al-Shabab and have tried Experienced at least 20.9 24 to maintain control over IDP camps.147 Members of 1 shock these militia groups act as “gatekeepers” at IDP settlement camps in Somalia and charge IDPs in Experienced at least 36.9 34 settlements a certain fee in exchange for providing 2 shocks security. Experienced at least 47.2 39.9 3 shocks Overall, male headed households are more likely Experienced at least 52.3 42.1 to experience a shock than female headed house- 4 shocks holds (70 and 60 percent respectively), but Source: Authors’ calculations based on the SHFS 2017–18. trends vary across types of shocks. One in every two male-headed households reported experi- encing the drought in comparison to one in every three female headed household (Figure 5.5). Simi- multiple types of shocks or experiencing multiple larly, male headed households are more likely to shocks make households poor (Table 5.1).146 experience water shortages for farming and cattle rearing. These shocks are closely related to the agropastoralist livelihood strategy, which is com- mon among Somali households. Unsurprisingly, Who is more vulnerable to shocks? incidence of conflict and violence is higher among Experience of a shock is influenced by various female headed households as women generally demographic characteristics such as location, experience higher levels of violence at domestic, age, and gender. Such information can be used to social, and institutional levels. formulate relevant targeting strategies for safety nets that are inclusive of the vulnerable popula- Households with heads older than 55 years are tion and its needs. For example, natural shocks more likely to experience shock as compared to are more common in certain locations, which can households with younger heads. Other factors be used to make the response more efficient and such as loss of income, lack of livelihood oppor- effective. However, other household characteris- tunities, immobility, loss of networks, and loss of tics, such as gender or age of the household head health and physical strength contribute to their vul- or employment status, also contribute to vulner- nerability. These factors limit their access to cop- ability to shocks, and, thus, increase the risk at the ing mechanisms. Usually child headed households household level. are common in conflict and fragile contexts and are more vulnerable due to social isolation. Child Nomads are most vulnerable to shocks, with headed household are also common in Somalia 98 percent of them reporting at least one type due to continuous violence and displacement, but of shock. Given their dependence on agropasto- there is no systematic data on them.148 ralist lifestyle, they are more likely to experience drought and loss of livestock (Figure 5.4). Three Poor households are more likely to experience a out of every four households in IDP settlements and shock than non-poor households (67 and 64 per- cent respectively). Usually, poor households are 146  The cross-sectional nature of data does not allow to identify households that experienced shock because of being poor in h t t p : //www. i r i n n ews .o rg / Re p o r t /9 6 6 8 6/S O M A L I A- 1 47   contrast to those that became poor as a result of experiencing Mogadishu-IDPs-suffer-extortion-eviction shocks. 148  Ward and Eyber (2009). 108  Somali Poverty and Vulnerability Assessment FIGURE 5.4  n  Incidence of shock by population type 80% 70% 60% 50% 40% 30% 20% 10% 0% Drought Natural Water Crop or High food Reduction in Conflict shortage livestock prices income for cattle loss or farming Urban Rural IDP Nomad Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 5.5  n  Difference in incidence of shock by age and gender of household head 14 14 Difference between elderly and non-elderly headed Difference between male and female headed 12 12 10 10 households in percentage points households in percentage points 8 8 6 6 4 4 2 2 0 0 –2 –2 –4 –4 ht al ge ss es e ct ht l ge ss es e ra t m m ef ur fli ug ug lo lo ic ic ta tu rta co co Th on at pr pr na ro or ro ck ck ho in in n C D sh D od od to to er er s in in es es fo fo er er th th n n liv liv O at O at tio tio h h ig ig W W or uc or uc H H ed ed p p ro ro R R C C Source: Authors’ calculations based on the SHFS 2017–18. Note: Respondents were asked to select the shock that affected them the most. Left: If the bar is above the x-axis, it means male headed households had a higher incidence than female headed households. Right: If the bar is above x-axis, households with heads 55+ experience have a higher incidence of shock than households with younger heads. more likely to experience shocks because they gender, consumption changes in response to lack access to risk management instruments a shock. However, the breakdown of shocks in such as insurance or credit. But the relationship Figure 5.6 shows interesting results where non- between incidence of shock and consumption- poor households report experiencing shortage based poverty remains spurious because unlike of water for cattle and farming and reduction in other household characteristics such as age and income more often. Social Protection 109 FIGURE 5.6  n  Difference in incidence of shock between poor and non-poor households 2.0 and non-poor households Difference between poor 1.5 in percentage points 1.0 0.5 0 –0.5 –1.0 Drought Natural Water Crop or High food Reduction Conflict shortage livestock prices in income loss Source: Authors’ calculations based on the SHFS 2017–18. Note: If the bar is above the x-axis, it means a higher incidence among poor households. Factors determining household often rely on informal networks, such as family and friends, to share risk.150 Remittances have served vulnerability to shocks as a lifeline for Somalis through emergency times Households with a male head or elderly head, where inflow increases during droughts and other nomads, and poor are more likely to experience natural shocks. This can possibly explain house- drought. Female headed households and IDPs in holds that reported receiving remittances in the settlements report experiencing high food prices past 12 months are more likely to report experi- and conflict and violence more than other popula- encing a shock as compared to those that did not tion groups. However, for insights into what makes receive any remittances (See Table 5.2). households more vulnerable to shocks, regres- sion analysis is conducted with whether a house- The impact of household characteristics varies hold experienced a shock or not as a dependent across different types of shocks. If the household variable.149 head is illiterate, the probability of experiencing drought and loss of crops and livestock is 12 to Overall, one percentage increase in wealth 24 percent higher than households whose heads decreases the probability to experience any shock have some education. Unsurprisingly, households by 20 percent, ceteris paribus. Other factors that which depend on agriculture as their main source make households more vulnerable to experienc- of income, are more likely to report water short- ing shocks are lack of education, dependence on age for livestock and farming and loss of crops and agricultural income, unemployment, and house- livestock but less likely to report high food prices. hold size. In low-income countries, households Different types of population are more prone to certain shocks. For example, settlement IDPs are more likely to experience conflict and violence as 149  Probit regression model is used where the dependent vari- compared to the urban households. This is consis- able is whether a household reported experiencing a specific tent with the evidence in literature on prevalence shock or not. It is clustered at regional level and assesses the impact of demographic and geographical characteristics on of violence in IDP settlements. On the other hand, whether a household experienced a shock. Household wealth both nomads and rural households are more likely is used instead of consumption to control for household’s long- than urban households to experience drought but term socioeconomic status. But there is a dummy variable that less likely to experience high food prices. Possibly, compares the top 60 percent of consumption distribution with reliance on agropastoralism makes households the bottom 40 percent. Other characteristics of household head such as literacy, age, and gender are also added. Other controls more prone to experiencing drought as a shock. are location of the household (urban, rural, IDP, or nomad as well as region), whether the household has an employed mem- ber, if the household depends on agriculture as its main source of income, and if the household has received any assistance and remittances in the past 12 months. 150  Jack and Suri (2014). 110  Somali Poverty and Vulnerability Assessment TABLE 5.2  n  What household characteristics affect the probability of reporting shocks? Crop or Other Water livestock High food Any shock Drought natural shortage loss prices Conflict Wealth index –0.112*** –0.080*** –0.006 0.005 –0.028*** –0.005 –0.002 [0.018] [0.021] [0.005] [0.005] [0.005] [0.014] [0.006] Head (no 0.039*** 0.041*** –0.007 –0.008 0.014** 0.001 –0.003 education) [0.014] [0.013] [0.007] [0.006] [0.006] [0.011] [0.005] HH with employed 0.066*** 0.031* 0.000 –0.007 –0.006 0.034** 0.005 member [0.026] [0.019] [0.010] [0.008] [0.011] [0.015] [0.007] HH has agricultural 0.158*** 0.024 0.021** 0.046*** 0.033** –0.051*** –0.027** income [0.039] [0.016] [0.008] [0.009] [0.013] [0.011] [0.013] Male headed HH –0.006 0.021 0.015*** –0.013*** –0.013 –0.014 –0.007 [0.038] [0.023] [0.004] [0.004] [0.016] [0.020] [0.004] HH head age 0.000 0.001 –0.000 0.000 –0.001*** 0.001* 0.000 [0.001] [0.001] [0.000] [0.000] [0.000] [0.000] [0.000] Household size 0.01*** –0.008 0.001 –0.002 0.006*** 0.003** 0.001 [0.004] [0.005] [0.002] [0.004] [0.002] [0.002] [0.001] HH receives 0.111*** –0.017 0.027** 0.030*** 0.018* 0.017 0.001 assistance [0.05] [0.028] [0.012] [0.009] [0.010] [0.012] [0.010] HH receives 0.083** 0.011 0.005 0.010 0.015 0.021 0.010*** remittances [0.04] [0.035] [0.008] [0.008] [0.013] [0.019] [0.003] Household welfare Bottom 40% [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] Top 60% 0.10*** 0.021 0.025*** 0.0009 0.014 0.020 0.003 [0.031] [0.019] [0.005] [0.006] [0.015] [0.016] [0.006] Population type Urban [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] Rural 0.095* 0.143*** –0.016* 0.015 0.004 –0.036** –0.006 [0.051] [0.023] [0.009] [0.015] [0.010] [0.018] [0.009] IDP (settlement) 0.021 0.014 –0.002 0.050** 0.032 0.034 0.063*** [0.054] [0.078] [0.020] [0.022] [0.019] [0.030] [0.024] Nomad 0.298*** 0.256*** — 0.026 0.010 –0.057*** 0.002 [0.054] [0.078] [0.016] [0.013] [0.011] [0.022] Control for region Yes Yes Yes Yes Yes Yes Yes Predicted 0.56 0.33 0.03 0.04 0.05 0.06 0.02 probability No. of observations 3,163 3,170 2,516 2,974 3,032 3,134 2,570 Pseudo R 2 0.26 0.26 0.12 0.18 0.17 0.11 0.14 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*) Dependent variables are dummies for experiencing each type of shock. Results are presented as margins with robust standard errors in parenthesis below. Wealth Index refers to the score based on assets and dwelling conditions, calculated using Principal Factor Analysis and is based on Filmer and Pritchett (2001).151 A higher score reflects households that are better off. Variable for agricultural income means that the main source of household income is agriculture. For results on each type of shock, see Appendix D, Regression results for each type of shock. 151  Wealth Index was created using Filmer and Pritchett (2001), but the choice of asset variables was slightly different. There is a huge asset depletion among households in Somalia, and most assets had less than 10 percent average incidence. The choice of variables on dwelling conditions was based on available information, as the data was collected using rapid survey consumption methodology. Social Protection 111 Households receiving assistance are more likely Similarly, high food prices reflect a decrease to report a shock in the last 12 months, suggest- in purchasing power parity and real income of ing that humanitarian assistance is well targeted. households. Due to the timing of data collection, humanitarian assistance had already reached the affected house- holds, or households are reducing their future risk How do households cope with shocks? exposure by reporting vulnerability knowing that this increases their chances of receiving assistance. Households rely on formal and informal net- works to mitigate the impact of shock and to Households in the top 60 percent of the con- smooth their consumption. These coping strate- sumption distribution are more likely to experi- gies fall into these categories: (i) self-insurance, ence a shock, particularly a reduction in income which refers to selling, pledging, or mortgaging and increase in theft. It may seem counterintuitive their assets; borrowing from friends, relatives, but households in the top half of the distribution moneylenders, or using other social networks to are more likely to have jobs and valuables as com- smooth consumption; (ii) self-help entails using pared to the poorer households. assets to generate more income, supplying more work, or allocating more hours to work by those who are already employed; (iii) informal insur- How do shocks affect households? ance is pooling of risks by tapping into informal networks such as friends, family or clan. In Soma- Ninety-five percent of Somali households that lia’s case, remittances have served as a strong experienced a shock reported a negative impact informal insurance mechanism facilitated by on their income, assets, or food resources. House- mobile money operators; (iv) credit is either from holds experiencing theft or conflict report a loss of informal mechanisms such as family and friends assets such as loss of valuables, land, or livestock or market-based mechanisms such as banks or (Figure 5.7). Conflict and violence also lead to other financial institutions. Since market-based destruction of property and other valuables. High mechanisms are not well developed in Somalia, food prices, loss of crops or livestock, and water the use of informal credit is likely to dominate; (v) shortage have a negative impact on household government help consists of government’s assis- income. Most Somalis rely on livestock and farm- tance, both cash and in-kind, given directly to ing for their livelihood so any shock to these leads households; and (vi) help from NGOs includes all to a direct reduction in household income sources. disaster relief and aid or ad hoc social assistance FIGURE 5.7  n  Negative effects of shocks on household welfare 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Drought Natural Water Crop or High food Reduction in Theft Conflict shortage livestock prices income for cattle loss or farming Income Assets Food production Source: Authors’ calculations based on the SHFS 2017–18. 112  Somali Poverty and Vulnerability Assessment FIGURE 5.8  n  Risk mitigation strategies in response to each shock Conflict Theft Reduction in income High food prices Crop or livestock loss Water shortage for cattle or farming Other natural Drought 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Percent of HHs adopting each strategy Did nothing Self insurance Reduced consumption Informal insurance Obtained credit Help from government Help from NGO/rel institution Source: Authors’ calculations based on the SHFS 2017–18. services. Households that don’t have access to were able to save money in the past 12 months, any of these, respond by doing nothing.152 this leaves them very vulnerable to future shocks. In such cases, households resort to selling their The most common coping strategy is self-­ assets, which leads to even higher vulnerability. As insurance and incurring debt, which indicates lack the resources keep dwindling, there are fewer and of adequate risk management and mitigation sys- fewer ways to cope with future shocks. tems, as well as an absence of formal and informal safety nets (Figure 5.8). If households are relying Reliance on informal risk mitigation mechanisms on self-insurance or choose to do nothing in case is also consistent across different locations and of conflict or theft, it implies a lack of access to male and female headed households. However, formal conflict resolution mechanisms and regula- female headed households are more likely to tory frameworks. This adds to the vulnerability of respond by not doing anything as compared to households, especially those who belong to mar- male headed households. Similarly, urban house- ginalized communities. Only an almost negligible holds are more likely to do nothing after experi- percentage of households have access to formal or encing a shock. It is indicative of absence of formal market mechanisms. risk mitigation mechanisms (Figure 5.10). Poverty and wealth do not influence access to Households who have experienced a shock in different types of mitigation strategies. Generally, the past 12 months are more food insecure than the poor are more exposed to shocks and risks but those who did not experience a shock. Drought is have fewest instruments to cope with them.153 In ranked as the single most common cause of food Somalia, however, the 60 percent wealthiest have shortage, especially in low-income countries and similar access to formal and informal safety nets can also trigger malnutrition and famine, depend- as the bottom 40 percent of the population (Fig- ing on the local context. It affects all four dimen- ure 5.9). The main coping strategy remains resort- sions of food security: availability, stability, access, ing to informal mechanisms and self-insurance and utilization.154 Coping Strategy Index (CSI) or not doing anything in both subpopulations. is based on a household’s behavioral response More than half of Somali households find borrow- to sudden decrease in resources and uses latent ing money from both formal and informal institu- variables such as reducing portion size or taking tions (including friends and relatives) difficult or children out of school to estimate level of food very difficult. As only 8 percent of households 152  Tesliuc and Lindert (2004). http://www.fao.org/crisis/28402-0f9dad42f33c6ad6ebda 154  153  World Bank (2003). 108ddc1009adf.pdf Social Protection 113 FIGURE 5.9  n  Adoption of risk mitigation mechanisms by welfare levels Top 60% Bottom 40% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Did nothing Self insurance Informal insurance Obtained credit Help from government Help from NGO/rel institution Reduced consumption Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 5.10  n  Adoption of risk mitigation mechanisms by location and head’s gender Male Female Nomads IDP Rural Urban 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Self insurance Informal insurance Obtained credit Help from government Help from NGO/rel institution Reduced consumption Did nothing Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 5.11  n  Reduced Coping Strategy Index insecurity.155 Households in rural areas are much 10.5 worse off, potentially because of disruption of 9.3 9.6 markets and loss of crops and livestock due to the 7.8 drought that makes food even more inaccessible and unavailable. 4.7 5.1 Resilience building with social Experienced Did not Urban Rural IDP Nomads safety nets shock experience shock Somali households that have experienced a shock report a higher level of food insecurity, low level Source: Authors’ calculations based on the SHFS 2017–18. 155  Coping Strategy Index (CSI) is one of the tools used to mea- sure the level of food insecurity at the household level, which can then serve as an early warning system by identifying groups where the need for aid is the highest. The CSI tries to quantify behavior of people when they are unable to access sufficient food. We calculate a reduced version of CSI, which is highly correlated with other measures of food insecurity. 114  Somali Poverty and Vulnerability Assessment of wealth, and are less likely to save or have Households receiving cash transfers use them access to formal or informal coping mechanisms, for productive investments, savings, and other and are also more likely to resort to negative cop- income generating activities.162 Similarly, cash ing strategies. In this context, there is a need of transfers lead to an improvement in household social safety nets and a social protection system consumption with an increase in livestock owner- that can build both risk management and risk cop- ship, agricultural assets and inputs, and savings.163 ing capacity of vulnerable households. In Ghana and Zambia, there was an increase in sav- ings by 11 and 24 percentage points, respectively. There is a distinction between risk-management In terms of human capital, households receiving a and risk-coping strategies where the former cash benefit are less likely to take children out of refers to ex ante management of income gen- school.164 Cash transfers also lead to an increase in eration (income smoothing), and latter refers to school attendance but not necessarily in learning dealing with the income or welfare risk ex post outcomes. In the context of health, cash transfers (consumption smoothing).156A social safety nets encourage households to use health services and system relies on a mix of both in its effort to build to improve dietary diversity.165 Cash transfers also resilience, which enables households to antici- lead to reduction of loans and debt repayments. pate and/or recover from the effects of a shock in Cash transfers help households manage risks in a timely and efficient way.157 Under severe condi- a more effective manner by diversifying income tions, households are forced to sell their assets to generating activities and avoiding negative cop- smooth their consumption. For example, in India, ing strategies, such as begging or changing eat- farmers sold their cattle when exposed to shocks ing patterns.166 Households are better able to cope and crises.158 Similarly, in West-African countries, with shocks if they have more human capital and the transactions in livestock sales were respon- assets, have access to jobs, and have diversified sive to income fluctuations related to the semi- livelihoods.167 arid environment.159 In Somalia, which has suffered recurrent shocks, households are experiencing The primary target for cash transfer programs extreme asset depletion where the most common must be poor households as they are typically durable owned by households is a cell phone. more exposed to multiple risks with limited access to formal and informal insurance networks. Poor One of the most effective approaches toward households are most vulnerable to shocks as they risk-coping at household level is to create con- experience the highest marginal impact on wel- ditions where households can participate in vol- fare because of low welfare levels to begin with, untary savings such as building grain reserves.160 but also due to lack of access to risk management This enables households to create capacity for instruments. High level of vulnerability tends to self-insurance. Households with access to formal make them risk averse so that they are less likely insurance and those that have higher income levels to engage in high risk, high return activities. Hence, and savings, are better able to smooth their con- having access to insurance mechanisms and other sumption. Moreover, in rural contexts where there risk mitigating instruments can give them an are credit constraints, households are less likely opportunity to make investments without fearing to save. Similarly, the saving pattern among poor losses.168 households indicates that most of their savings are usually meant to smooth income shocks, but A cash transfer to poor households can help they seldom make long-term investments, which reduce poverty. Globally, countries tend to spend have higher returns.161 In Somalia’s context, safety between 2.5 and 5 percent of GDP on such pro- nets can encourage households to save and build grams.169 In contrast, Sub-Saharan countries on household resilience that can help them to smooth their consumption in an event of shock. 162  World Bank 2018a. 163  Bastagli, et al. (2016). 156  Alderman and Paxson (1994). 164  Davis, et al. (2016). 157  Mitchell and Harris (2012). 165  Hagen-Zanker, et al. (2016). 158  Rosenzweig and Wolpin (1993). 166  Daidone, et al. (2015). 159  Fafchamps, et al. (1998). 167  World Bank (2018f). 160  Lim and Townsend (1998). 168  World Bank (2003). 161  Zeldes (1989); Kimball (1990); Deaton (1992). 169  World Bank (2018b). Social Protection 115 average spend only 1.6 percent of GDP on social safety nets. Somalia spend even less at 0.8 per- Policy recommendations cent of GDP in 2016, even though it received 16 The effectiveness and efficiency of safety net percent of GDP (US$ 1.2 billion) in humanitarian program depends on its ability to reach vulner- aid.170 Using some resources to implement a well- able communities, which remains a challenge targeted safety net could substantially reduce given local clan dynamics and the security situ- poverty. ation in Somalia. Most development partners rely on clan leaders or local partner organizations to The impact of social safety nets on poverty and gain access to communities, which in most cases inequality is influenced by coverage, benefit level, also serve as gatekeepers for information. An benefit incidence, and other design features. On inclusive program will have to break these barriers, average, household consumption can increase by for example by using an objective and transparent US$0.74 for each dollar transferred, though the targeting scheme. impact varies in magnitude across countries.171 In general, countries that have a very high rate of The first step could be to identify geographic coverage coupled with a high benefit level, have a regions that experience higher incidence of greater impact on poverty and inequality. Georgia shocks, followed by a selection of households that and South Africa are two examples that display the qualify as a vulnerable household. Given barriers highest level of poverty reduction, particularly in to access communities, a phased approach simi- the bottom quintile of the distribution. In Georgia, lar to Pakistan’s Benazir Income Support Program the coverage rate is 93 percent of the poorest quin- could be successful. In the first phase, members tile where each household receives a benefit that of Parliament were responsible for selecting geo- constitutes 68 percent of its total welfare. This has graphic areas as well as for compiling a list of ben- led to reduction in headcount poverty by 43 per- eficiary households. It was a subjective approach cent. Similarly, low coverage and low benefit levels to targeting that resulted in very high inclusion will have lower impact on poverty reduction.172 and exclusion errors, measured against poverty criteria. However, once the government estab- Globally, countries that are more prone to natu- lished access with target communities, a targeting ral shocks have low safety net coverage rates.173 strategy based on Proxy Means Test was devel- Low coverage rates might be explained by budget oped and implemented. This has led to reduction constraints, lack of implementation and institu- of leakages and has also helped the government to tional capacity, or other political economy issues. build trust among citizens.175 However, low coverage rates of social safety nets can lead to lack of access to those who need them In a resource constrained environment such as the most. But in most fragile countries, such as Somalia, a social safety net program in the short Afghanistan, the Democratic Republic of Congo and medium term cannot replace humanitarian and Haiti, even though there is a low safety net assistance but only complement it. The objective coverage, humanitarian programming is much of humanitarian assistance is to help households larger.174 In Somalia, currently there is no national smooth their consumption after experiencing the government-led safety net system, but around shock, and hence serves as a risk coping strategy. 20 percent of the population was covered by Somalia has been receiving humanitarian assis- humanitarian assistance in 2017. tance for the past two decades, which by defini- tion is ad hoc and demand based. As Somalia is prone to natural and manmade shocks, a transition to social safety nets can help in building resilience at the household level. In contrast to humanitarian assistance, social safety nets serve as both a risk management and risk coping mechanism. These 170  https://reliefweb.int/sites/reliefweb.int/files/resources/ systems can exist concurrently and complement final_ocha_somalia_humanitarian_bulletin_october_2017v3_ each other. In fact, the presence of humanitarian 002.pdf 171  Ralston, et al. (2017). 172  World Bank (2018f). 173  Ibid. 174  Ibid. 175  Haseeb and Vyborny (2017). 116  Somali Poverty and Vulnerability Assessment programming can ease the fiscal burden of imple- Household-level investments in human capital will menting such programs. Thus, a coordinated directly benefit children, representing nearly half national level program that offers long-term and of the Somali population. A large young popula- reliable cash transfers in Somalia can assist the tion is a huge asset for Somalia that can contribute most vulnerable households that are not being tar- to its growth. But the challenge is to create con- geted by humanitarian assistance or remittances. ducive conditions. In conflict and fragile situations, young men are more likely to engage in violence, Due to political fragility in Somalia, NGOs and substance abuse, and gang activities. Unemploy- INGOs will have to work collaboratively with ment is one of the major factors that motivates Somali government to build technical, institu- young men to join rebel movements.176 In this con- tional, and fiscal capacity. A Multi-Donor Trust text, cash transfers will initially enable households Fund (MDTF) can be set up as was done in Ethio- to make investments in health and education of pia, where different donors committed funds for a children, whereby opening opportunities for them social protection program that was implemented to participate in the growth of Somali economy. by the Ethiopian government. Donors and imple- However, in the medium to long term, social pro- mentation partners also provided continuous tection must include youth focused programming technical assistance to formulate policy and to that targets their specific needs and goes beyond set up relevant institutions and systems. Such an providing only employment. arrangement helps the government to establish its legitimacy. In the medium term, cash transfers can be com- bined with productive inclusion strategies that Even though the poverty impact of a safety net can help diversify livelihood strategies. In Soma- appears small, it can have a profound impact lia, where agriculture, fisheries, and livestock are in the long term on reducing vulnerability and the main contributors to the GDP but at the same building human capital by helping households time most vulnerable to climate shocks, households to invest in health, education, and assets while with diversified livelihood strategies can cope with increasing savings and reducing exposure to risk climate risks better. This will prevent them from by reducing debt repayments. Cash transfer pro- depending entirely on agriculture, fisheries, and grams should be designed for multiple years to livestock for their livelihoods and so when hit by a serve as a reliable source of income for households, shock, they will explore other livelihood avenues to which helps them to smooth their consumption smooth their income and consumption. Even when over time. Given asset depletion and household not combined with labor market activities, cash level loans in Somalia, households first respond by transfers enable households to switch from casual repaying loans and then only after that will invest agricultural labor to on-farm labor.177 in human and physical capital. 176  World Bank (2011). 177  Davis, et al. (2016). Social Protection 117 CHAPTER 6 Remittances KEY MESSAGES Remittances are the major source of external devel- US$505 to US$876 and domestic remittances range opment finance for Somalia. Remittances contribute from US$138 to US$525. Most remittance-recipient to international reserves, help finance imports, and households receive remittances monthly. Households improve the current account position of the coun- receiving international remittances draw a larger pro- try. Somali migrants send on average $1.3 billion per portion of their incomes from salaried labor (35 per- year. These estimates of remittance inflows based on cent) and remittances (34 percent). data reported by the International Monetary Fund are likely below the actual volume of remittance flows to Remittances are associated with reductions in pov- Somalia. There is need to improve data collection and erty and increased access to health and education reporting of remittances, as well as to capture flows services. The proportion of households receiving that take place outside of formal financial channels. remittances tend to be less poor. International remit- tance recipient households are typically urban, In Somalia, remittances were more stable than both headed by women, and have their kids enrolled in FDI and official aid. Sometimes remittances may also school. Urban households receiving international behave countercyclically with respect to the eco- remittances tend to have both higher consumption nomic cycle of the recipient country. Thus, the greater levels and higher enrollment rates for their children. stability of remittance flows and their anti-cyclicality may contribute to the stability of resources received Even though remittances provide a lifeline to the by Somalia. Remittance inflows are more than three poor, sending money to Somalia remains costly. times the size of foreign direct investment and are the Somalia has been affected by “de-risking.” Due to the same size as grants and official aid Somalia received. anti-money laundering and combatting the financ- Remittances contribute to the country’s international ing of terrorism regulations, the costs of remitting reserves, help finance imports, and improve the cur- money to Somalia have increased. According to the rent account position. Remittance Prices Worldwide database, the aver- age cost of sending US$200 from Australia and the Somali households are both remittance receiv- United Kingdom to Somalia has increased. Reducing ers and senders, however the incidence of receipt transaction costs increases the disposable income exceeds the incidence of sending. Remittance- of poor migrants and increases their incentives to recipient urban households receive the largest remit. Policies to foster the use of innovative mobile amount on international remittances, while IDPs liv- money transfer technologies and payment systems, ing outside settlements receive the largest amount as well as the use of digital financial IDs, will facilitate on internal remittances. The average amount of inter- remittance flows and the compliance with Know Your national remittances households received range from Client (KYC) regulations. Remittances have impact at both the household the microeconomic level (accessing new financial level and at the level of the economy, affecting products for micro-insurance, education, food, and macroeconomic management, labor force par- micro and small and medium enterprises). ticipation, and patterns of household expendi- ture. Remittances are associated with increased Remittances are the major source of exter- household investments in education and health. nal development finance for Somalia. During Remittances may play a significant role in alle- 2015–2017, Somali migrants and refugees remit- viating poverty. Remittances are private money ted on average US$1.3 billion per year.178 The true that belong to the households. However, remit- size of remittance flows is believed to be tances can be leveraged at the macroeconomic level (accessing improving credit ratings) and 178  International Monetary Fund (2018a). Remittances 119 significantly larger considering unrecorded flows. Somalia is the fourth top refugee origin country In Somalia, remittances are close to 20 percent with almost 1 million refugees in the world.181 The of GDP.179 Remittances have consequences at number of Somali refugee arrivals to the United both the household level and at the level of the States increased from 6,969 in 2007 to 9,020 in economy, affecting macroeconomic management, 2016. In Africa, around 835,900 Somali refugees labor force participation, education and health are still displaced. Since the beginning of the Vol- outcomes, income distribution, and patterns of untary Repatriation program in December 2014, household expenditure. This chapter discusses the 81,030 refugees were repatriated.182 Out of 81,030 economic implications of migrant remittances for who were repatriated, 1,089 were assisted in March, Somalia and recipient households. namely, 759 from Kenya, 272 from Yemen, 56 from Libya, and two from Gambia.183 International mobility patterns Despite being the home country to millions in the diaspora, Somalia is also host to millions of internally displaced persons and thousands of Somalia has one of the most complex migration refugees, asylum seekers, and migrants. Somalia patterns of any part of the Horn of Africa, in part is estimated to host over 1.1 million internally dis- due to the conflict in the region. It is a country that placed persons, 116,040 returnees, 15,259 refugees, sends migrants and refugees while also receiving 14,885 asylum seekers, and 44,868 migrants.184 A migrant and refugee returnees. Recently, the con- total of 4,293 Somali refugees returned to Somalia flict in Yemen has pushed new Yemeni refugees to in the first three months of 2018. Large numbers Somalia. of migrants and refugees transit through Somalia, particularly Somaliland and Puntland, but no data The stock of Somali migrants and refugees living capture this movement. Somalia is also a destina- outside of Somalia reached more than 2 million tion country for undocumented migratory flows in 2017, having doubled since 1990.180 Somalis pri- due to its extensive borders. Transit migration is marily migrate to Kenya, Ethiopia, the Republic of driven by the same drivers of voluntary or forced Yemen, Libya, the United Kingdom, Djibouti, the migration, including better economic opportuni- United States and Sweden in descending order of ties and security. Estimates suggest that there are popularity. Before the 1990s, Somali migration had at least 20,000 undocumented migrants, mainly been focused on the Arabian Peninsula and the Ethiopian, in Somaliland.185 Persian Gulf, in part because Somalia has historic trade ties with the Gulf States linked to the Somali livestock trade and labor migration. Somali mobil- ity patterns shifted in the 1990s with the eruption Remittances at the of the civil war. After the war, Somalis went to vari- macroeconomic level ous destinations outside Africa. Migrants and refu- gees continue using the migration/refugee routes Remittances have been a lifeline for Somalia. The from the southern, northeastern and northwest- importance of remittances as a means of develop- ern Somali regions to the Gulf of Aden looking ment finance and household income in Somalia for better opportunities and security. Some ven- has sparked substantial interest. Somali migrants ture northwest through Sudan and Libya as tran- sent at least $1.3 billion in remittances in 2017. The sit countries. Several of the youth migrants go to true size of remittance flows, including unrecorded Libya and try to cross to Europe. Libya is a main flows, is believed to be significantly larger. Remit- transit country in which several Somalis are being tances are the most tangible link between migra- abused by smugglers and traffickers. Others head tion and development. south through Kenya and the eastern Africa cor- ridor toward South Africa. 181  UNHCR (2017). 182  UNHCR (2018a). 183  UNHCR (2018b). 179  IMF (2018a). 184  World Bank (2018a); UNDESA (2017); UNHCR (2018a). 180  World Bank (2018a). 185  RMMS (2016). 120  Somali Poverty and Vulnerability Assessment TABLE 6.1  n  Selected economic indicators, 2015– consumption and investment during economic 2018 (percent of GDP) downturns. Thus, they perform the role of a “shock absorber” or insurance for origin countries against 2016 2017 2018 macroeconomic shocks or other shocks.188, 189 Eco- (estimated) (projected) (projected) nomic activity in Somalia is recovering from the Real GDP 2.4 1.8 2.5 effects of the drought in 2016–17. The drought growth impacted Somalia’s economic activity in 2017, Current –6.3 –6.7 –7.2 but sustained international community support, account and remittances helped Somalia avoid a severe balance humanitarian crisis as well as to finance the trade deficit.190 Trade balance –46.2 –50.5 –45.8 Remittances 19.6 20.6 19.5 Large and sustained remittance inflows can Grants 20.8 23.7 19.5 make manufacturing less profitable. Like other External debt 74.5 71.5 … sources of exogenous foreign exchange, such as Nominal GDP 6,887 7,382 7,781 development assistance, remittance inflows can in US$ cause an appreciation of the real exchange rate, making tradable goods production less competi- Source: Somali Authorities and Fund staff estimations and projections. tive overall, and perhaps making low-cost manu- Taken from the 2017 Article IV Consultation. facturing unprofitable. Empirical evidence on the adverse effect of large inflows of foreign exchange is scarce. It is even more scarce with reference to Remittances tend to be relatively stable and remittances. According to Chami (2018), the Dutch may behave countercyclically. The reason is that disease effect is less pronounced in fragile states relatives and friends often send more when the due to the fact there is a small tradeable sector.191 recipient country is in an economic downturn or experiences a disaster (Mohapatra, Joseph, and Appropriately accounting for remittances can Ratha, 2009). In Sub-Saharan Africa, remittances improve the evaluations of external debt sustain- have been more stable than foreign direct invest- ability and creditworthiness. The ratio of exter- ment, private debt, and equity flows. Neverthe- nal debt to exports would be significantly lower less, even small fluctuations in remittance inflows if remittances were included in the denominator. can pose macroeconomic challenges to recipient The International Monetary Fund (IMF) is help- countries, especially those with large inflows. ing Somalia reach debt relief under the Heavily Indebted Poor Countries (HIPC) initiative as soon Remittance inflows in the period 2015–2017 as feasible within established HIPC procedures, stood at about US$ 1.3 billion per year. Remit- including the preparation of a Poverty Reduction tances represented 20 percent of the GDP in 2017 Strategic Paper (PRSP).192 Remittances could be (Table 6.1).186 These inflows are more than three included in the preparation of the debt sustain- times the size of foreign direct investment and are ability analysis as per the latest guidance on debt the same size as grants and official aid Somalia sustainability analysis. received. Remittances contribute to the country’s international reserves, help finance imports, and Remittances can affect economic growth directly improve the current account position. by raising consumption and investment expen- ditures, and by improving the stability of con- Remittances offer some important advantages sumption and output at both the household and from the point of macroeconomic manage- macroeconomic level.193 Remittances tend to be ment in poorer countries. Remittances tend to relatively stable, and may behave be a more stable source of foreign exchange than other sources so that the resulting real exchange rate level may be sustainable.187 Remittances are often countercyclical, helping to sustain 188  Frankel (2011); Chami, et al. (2009); Singh, et al. (2009). 189  Ratha (2007). 190  IMF (2018b). 191  Chami, et al. (2018). 186  International Monetary Fund (2018b). 192  IMF (2018a). 187  IMF (2005). 193  Chami, et al. (2009); Mohapatra, et al. (2009). Remittances 121 countercyclically—because relatives and friends often send more when the recipient country is in The development impact an economic downturn or experiences a disaster of remittances at the (Mohapatra, Joseph, and Ratha, 2009). In Somali, remittances have been more stable than foreign microeconomic level direct investment, private debt, and equity flows. It is well established that the primary economic Nevertheless, even small fluctuations in remit- benefit of migration to recipient households is tance inflows can pose macroeconomic challenges the receipt of remittances, although it can be dif- to recipient countries, especially those with large ficult to separate the effects of remittances from inflows. the overall effect of migration in empirical stud- ies.194 While there is general agreement that bil- Reliable data on remittances are hard to come by lions of dollars in money and goods are remitted as in the case with migration data. Data on remit- to developing countries, there is less consensus on tances are believed to be underestimated in Soma- the growth implications for developing countries. lia. While the IMF is trying to assess the volume of remittances, these data are neither comprehen- Somali households are both remittance receiv- sively reported nor do they capture flows of monies ers and senders, but their incidence of receipt that take place outside of formal financial chan- exceeds the incidence of sending fourfold and nels. Data inaccuracy stems from problems associ- is biased toward urban areas. Fifteen percent of ated with knowing the universe of remitters and Somali households receive remittances while only the intermediaries facilitating the process, enforc- 4 percent send remittances. Urban households are ing data collection, and applying the appropriate more likely to receive international remittances, methodologies to capture the data. Reliable data while IDPs living outside settlements are more on remittances are hard to come by. Some recom- likely to receive internal remittances (Figure 6.1). In mendations to improve remittances data include: contrast to urban populations, the proportions of (i) improve data compilation and methodologies IDP households living in settlements that received for Somalia; (ii) improve coverage of all remittance remittances were found to be generally low. These service providers including mobile phone service ranged from 4 percent for domestic remittances to providers; and (iii) increase resources and build 5 percent for international remittances (Figure 6.1). capacity to improve the accuracy of data compiled by the Central Bank of Somalia. 194  World Bank (2006); McKenzie and Sasin (2007). See Plaza and Ratha (2011) for other benefits, such as the transmission of knowledge, trade, and investment linkages. FIGURE 6.1  n  Incidence of remittance receipt and sending 30 Percent of households 25 20 15 10 5 0 Overall Urban Rural IDPs living in IDPs living Nomads settlements outside of settlements Remittances overall International remittances Internal remittances Sending remittances Source: Authors’ calculations based on the SHFS 2017–18. 122  Somali Poverty and Vulnerability Assessment FIGURE 6.2  n  Average annual value of remittance received and sent $1,400 $1,200 $1,000 $800 $600 $400 $200 $0 Overall Urban Rural IDPs inside IDPs outside Nomads settlements settlements Remittances overall International remittances Internal remittances Sending remittances Source: Authors’ calculations based on the SHFS 2017–18. Both urban and IDP households living outside set- US$876 for international remittances. This could tlements also send remittances. This has been a be also in response to the impact on the drought. feature of the Somali community that send cash IDPs living outside of settlements reported expo- between various locations.195 sure to drought (70 percent) compared to the IDPs living in settlements (46 percent). Equally, rural Not all migrants send remittances home and areas had been more impacted by the drought. not all migrant households receive remittances. However, due to the remoteness and not easy About 23 percent of the households reported access by Money Transfer Operators (MTOs) to receiving remittances from a former household these localities, remittances are not easily sent. member or a friend who is living abroad as its main source of income. Most Somali households receive remittances monthly. The incidence of receiving international Remittance-recipient urban households receive remittances once per month is higher (61.9 per- the largest amount on international remittances cent) than the incidence of receiving domestic followed by IDPs living outside settlements. remittances (42 percent). Somalis reported that The average amount of international remittances they receive internal remittances every other received per household per year is US$743— month (12.6 percent) and for special occasions above the average per capita income of Somalia (11.2 percent). However, only about 6.6 percent of of US$535 for 2017 (Figure 6.2). There are varia- recipients receive international remittances during tions in the average amount of remittances house- special occasions. These facts underscore the role holds received. International remittances range international remittances play in household con- from US$505 to US$876, and domestic remit- sumption (Table 6.2). tances range from US$138 to US$525. Domestic and international remittances are important for Remittances from the Somali diaspora have ­ recipient-receiving households, particularly among emerged as an important source of income. For the urban, non-settlement IDPs, rural households, many households in Somalia, remittances represent and nomads. While it appears that IDPs inside settle- both a sizeable proportion of household income as ments receive large sums of remittances, the share well as a substantial source of fund inflows into the of households receiving remittances is very low local communities. Households receiving interna- tional remittances draw a larger proportion of their Interestingly, the average amount received by incomes from salaried labor (35 percent) and remit- IDPs outside settlements was relatively high at an tances (34 percent). Households receiving internal average of US$430 for domestic remittances and remittances draw 35 percent of their income from salaried labor, remittances (23 percent), and agri- culture (19 percent). While for households that do not received any remittances, salaried labor 195  European Commission (2017). Remittances 123 TABLE 6.2  n  Frequency remittances are received by households Internal International remittances Internal remittances International (frequency) remittances (%) (frequency) remittances (%) Once per week or more 28 5.3 20 2.5 Twice per month 27 5.1 28 3.4 Once per month 221 42.1 504 61.9 Every other month 66 12.6 84 10.3 Once every three months 55 10.5 54 6.6 Once every four months 27 5.1 30 3.7 Twice a year 21 4.0 12 1.5 Once a year 21 4.0 28 3.4 Special occasions only 59 11.2 54 6.6 Total 525 100 814 100 Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 6.3  n  Remittance-receiving households are in the top 60 percent consumption 15 13 10 9 8 5 5 0 Bottom 40% Top 60% Internal remittances (% of households) International remittances (% of households) Source: Authors’ calculations based on the SHFS 2017–18. (38 percent) and agriculture activities (26 percent) percent. International remittances received per are the dominant sources of income. capita per day are almost the same amount for the bottom 40 percent (US$0.35) and the upper 60 Remittances can reduce the level of poverty by percent (US$0.39). However, income from inter- directly augmenting the incomes of recipient national remittances represents 54 percent of the households and by increasing aggregate demand. total consumption for the bottom 40 percent of About 13 percent of Somali households receiving households while they only represent 23 percent international remittances are in the top 60 per- of the total consumption for the upper 60 per- cent of the consumption distribution (Figure 6.3). cent (Figure 6.4). Internal remittances represent Households receiving international remittances 33 percent of the total consumption for the bot- from outside Africa may have high incomes since tom 40 percent of households, while they only these remittances tend to be larger—much larger represent 13 percent of the total consumption for than remittances from domestic sources. the upper 60 percent. Facilitating remittance flows to the bottom 40 percent could have a positive Both internal and international remittances are impact on welfare. A social protection program relatively more important for the bottom 40 targeted to the bottom 40 percent could also alle- viate poverty. 124  Somali Poverty and Vulnerability Assessment FIGURE 6.4  n  Remittances more important for the bottom 40 percent 60 1.8 60 1.8 54 Percent of total consumption 1.6 1.6 Percent of total consumption 50 50 1.4 1.4 40 1.2 40 1.2 US$ value US$ value 1.0 33 1.0 30 30 23 0.8 0.8 20 0.6 20 0.6 13 0.4 0.4 10 10 0.2 0.2 0 0 0 0 Bottom 40% Top 60% Bottom 40% Top 60% International remittances as a International remittances as a percentage of total consumption percentage of total consumption Intl. remittances US$ pc pd Intl. remittances US$ pc pd Consumption expenditure US$ pc pd Consumption expenditure US$ pc pd Source: Authors’ calculations based on the SHFS 2017–18. FIGURE 6.5  n  How do international remittances impact consumption? 90 70 Percent change in consumption 50 30 10 –10 –30 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.9 0.95 Consumption percentile Consumption change of recipients Smoothed consumption change 95% confidence interval Source: Authors’ calculations based on the SHFS 2017–18. International remittance-receiving households There is a strong correlation between households are positively impacted in their level of con- that receive remittances and poverty. The propor- sumption. To assess the importance of receiving tion of households receiving remittances tend to remittances for the poorest households, a quan- be less poor. About 58 percent of the households tile regression is used to measure the effect in receiving international remittances are poor com- consumption for recipients over non-recipients pared to 71 percent of the households that do not along the consumption distribution as a function receive remittances. Table 6.3 shows that recipient of remittances controlling for population type. households are typically urban, headed by women, An increase in remittances is associated with an and have their kids enrolled in school. There is increase in consumption for the lowest quantiles strong correlation between households receiv- (Figure 6.5). This implies that remittances contrib- ing international remittances and preparedness ute to increases in consumption that belong to the to absorb shocks. Not surprisingly, urban house- worse-off group. holds which have higher incidence of international Remittances 125 TABLE 6.3  n  Characteristics of remittance-recipient households Internal International Do not receive remittances remittances remittances Poverty 64% 58% 71% Consumption expenditure (2017 PPP US$ per capita $1.35 $1.41 $1.25 per day) Consumption expenditure (2017 PPP US$ per $7.40 $7.63 $6.66 household per day) Enrollment (6–17 years) 48% 60% 33% Labor force participation (7 days) 54% 46% 47% Female household head 36% 45% 42% Source: Authors’ calculations based on the SHFS 2017–18. remittances tend to have both higher consumption variable the ownership of mobile phones. This vari- level and higher enrollment rates for their children. able will be more appropriate for Somalia where mobile phones and mobile money are highly used. Since remittances may be endogenous, it will be important to address it when estimating the To explore the impact of remittances on poverty, impact of remittances on poverty. Several authors we created a counterfactual of expenditures with- have used different methods to take into consider- out remittances. Using the Kinnon and Soler (2018) ation the value of that migrant had he stayed and methodology, we compared actual, observed pov- worked at home.196 We explained below why this erty levels that would have existed if remittance is not possible in the case of Somalia. The second income had not been available to households. This method is the one used by Lopez, et al. (2007). scenario provides an upper bound estimate of the He constructed a counterfactual and used a two- difference in poverty rates associated with migra- stage Heckman model to correct for selection tion. The calculations show that: (i) using the head bias.197 For the case of Somalia, it is difficult to find count ratio without remittances, poverty would an exogenous variable that propels migration or have been more severe; (ii) the poverty gap index the receipt of remittances in the first stage equa- would have been widened; and (iii) the inequality tion that it is not related to the dependent vari- measured by the Gini coefficient would have been able in the second stage equation. Some authors larger (see Table 6.4). have used the nearest distance to the border to instrument for conflict (as in the case of Pakistan- For Somalia, it is not advisable to impute the Afghanistan border) because of the endogeneity per capita household income of remittances- of conflict and remittances.198 This could not be the receiving households. In a country where migra- best instrument for Somalia. Another method is to tion takes place due to conflict, it is difficult to use an instrumental variable that it is correlated with remittances but exogenous to poverty. In the literature, several instruments have been applied TABLE 6.4  n  Counterfactual without remittances such as rainfall shocks, distance, migrant ethnic networks, ownership of non-agricultural land, and With internal Without internal number of return migrants in the ethnic group with and international and international which the head of household identifies, among oth- remittances remittances ers. Those instruments could be particularly prob- Poverty 69% 71% lematic for the case of Somalia. One possibility is headcount to use a quintile regression using as an instrument Poverty 0.29 0.32 gap index 196  Barham and Boucher (1998). Gini 0.34 0.35 197  Lopez, et al. (2007). 198  Ghorpade (2017). Source: Authors’ calculations based on the SHFS 2017–18. 126  Somali Poverty and Vulnerability Assessment make assumptions on how to convert migrants and reducing the need for child labor (e.g., Ghana). back to household members in the household of Girls’ school attendance and educational attain- origin. It is not possible to assume that only one ment rise from the receipt of remittances (e.g., Pak- adult male had migrated and would need to be “re- istan, Peru). Remittances can contribute to better integrated” into the household to estimate house- health outcomes by enabling household members hold income in the absence of migration. Given the to purchase more food and health care services army conflict, several members of the household and perhaps by increasing information on health have migrated. It is not possible also to assume practices. Some studies found that higher remit- that each remittance-receiving household would tances per capita were associated with greater have retained the same number and gender of access to private treatment for fever and diarrhea. migrants as appeared in the survey data. In addi- Remittances reduce overall child mortality, and tion, it will not be possible to add both the labor remittances and access to knowledge facilitate force participation and the unemployment rate of new treatments for HIV/AIDS and malaria. the population into the equation since the coun- try is still recovering.199 And the assumption on the International remittances may increase expen- demographic characteristics will not be accurate. diture on education and health in Somalia. This section examines the relation between the inter- national remittance inflows and educational and Uses of remittances health expenditures by estimating a linear model. When the probability of receiving international Remittances have been found to have positive remittances is considered, a statistically significant impacts on human development. Evidence from relationship emerges between remittances and Latin America, Africa, South Asia, and other regions educational and health expenditures. International suggests that remittances reduce the depth and remittance-receiving households have a 67 per- severity of poverty, as well as indirectly stimulate cent higher chance to increase expenditures on economic activity. Remittances have also implica- education compared to non-recipient households. tions for human welfare, including poverty reduc- In the case of internal remittances, there is not a tion and promoting shared prosperity. Migration significant effect on education. Moreover, inter- and remittances lead to increased investments in national remittances and health expenditures of health and education. In some countries, remit- households receiving remittances were positively tances contribute to better school attendance, correlated (Table 6.5). higher school enrollment rates, and additional years in school. Remittances may increase expen- Households in the bottom 40 percent that receive diture on education by helping finance schooling international remittances have substantially higher school enrollment than non-recipients. Households receiving international remittances have a positive correlation with higher enrollment 199  Scott and Soler (unpublished). TABLE 6.5  n  Impact of international remittances on educational and health expenditure (i) (ii) (iii) Dependent variable Log (educational Log (health expenditure) expenditure) Land access International remittances receipt 0.637*** 0.249*** 0.004 Standard error (0.183) (0.068) (0.025) Observations 5,132 5,132 6,057 R-squared 0.132 0.129 Source: Authors’ calculations based on the SHFS 2017–18. Note: *** p<0.01, ** p<0.05, * p<0.1. (i) and (ii) from OLS regression, (iii) from Probit regression. Controlling for household size, income, and population type. Remittances 127 FIGURE 6.6  n  Do international remittances impact enrollment? 70 63 Percent of school-aged enrolled 60 52 50 40 35 30 30 20 10 0 Bottom 40% Top 60% Recipients Non-recipients Source: Authors’ calculations based on the SHFS 2017–18. TABLE 6.6  n  Housing conditions and remittance receipts among Somali households Households not receiving Households receiving Households receiving Floor material remittances domestic remittances international remittances Cement 33% 43% 58% Mud 25% 27% 11% Wood 9% 5% 3% Grid access 38% 54% 72% Source: Authors’ calculations based on the SHFS 2017–18. rates (Figure 6.6). Moreover, although these find- status by using propensity score-matching meth- ings do not control for the possible endogeneity ods.200 Although, it was not possible to apply score- of remittance-receiving status, they suggest that matching, the data suggest that Somali households remittances may help raise the level of resources that receive international remittances have houses devoted to education. with cement floors and have better grid access, although possible endogeneity is not controlled International remittances help households cope for (Table 6.6). Stark differences between inter- with droughts. Transfers from friends and relatives national remittance-receiving households and abroad played a key role in reducing the distress non-receiving households in floor material exist, caused by the drought. The remittances received with about 58 percent of international remittance- by IDPs living outside of settlements could be receiving households having floors made from in part due to transfers received to survive the cement, with 33 percent of non-­ recipient house- drought. holds having cement floors. The most visible distinc- tion between international remittance-receiving There is a positive relationship between remit- households and non-­ receiving households was tances and the quality of dwelling and access to the access to electricity. The share of households electricity. Remittances can also enable recipient with access to the grid was 34 percentage points households to build stronger and more resilient higher among the households receiving interna- housing. For example, remittance-receiving house- tional remittances compared to non-receiving holds in Burkina Faso and Ghana were more likely households. to have a concrete house, after controlling for the possible endogeneity of the remittance-receiving 200  Mohapatra, et al. (2009). 128  Somali Poverty and Vulnerability Assessment Remittances can help reduce poverty, increase Remittance markets in Somalia remain relatively access to health and education services, and pro- underdeveloped in terms of their financial infra- mote household savings. Remittance-receiving structure and the regulatory environment, but households are less likely to be poor compared the rapid adoption of innovative money-­ transfer to the households that do not receive remittanc- technologies is transforming the landscape for es.201 Somalis continue to transfer funds to family remittances and broader financial services. Two left behind and invest back home to support the remittance channels are involved: (a) domes- recovery of the country. tic remittances are conducted overwhelmingly through mobile money (46 percent), money trans- fer operators (47 percent), and informal chan- Remittance markets nels such as hand-carried during visits home, and Hawala, and (b) international remittances are largely channeled through money transfer opera- Following the civil war, remittances have been tors (87 percent) and mobile phones (12 percent). mainly sent through the hawala system. 202 Hawala The top three mobile money players are Hormuud, refers to money transfers that occur in the absence Someteland, and Golis. of, or are parallel to, formal banking sector chan- nels.203 Somalis call this informal system “xawilaad” The use of mobile phone has been limited to which is the Somali rendering of the Arabic word domestic money transfers. This is mainly because “hawala.” The xawilaad operates in almost every of concerns about money laundering and terror- part of the world and is operated and used by ist financing related to cross-border remittances. Somalis to send money back home to families and However, these technologies have the potential to conduct business transactions.204 Transfers by to vastly improve access to both remittances and xawilaad are fast and made with great efficiency.205 broader financial services, including low-cost sav- Currently, there are more than 20 Money Transfer ings and credit products, for Somali migrants and Operators (MTOs) in the country that work cross- remittance recipients in the country. border and across regions within Somalia. However, the market is dominated by three main players: Somalia is facing de-risking and Know-Your- Dahabshiil, Amal Express, and Taaj. Interviews Client (KYC) regulation. Migrants send money using conducted in Virginia, United States of ­ America— MTOs and with family members.207 The choice of an area with one of the largest Somali migrant the intermediary is affected by, among other things, populations—indicate that the Somali community costs, trust in the intermediary, and convenience mainly uses two large companies to transfer remit- factors—such as location, hours of operation and tances: Dahabshiil and Amal Express.206 Some of language—and identification requirements. How- these MTOs have become banks, are registered ever, the closure of correspondent relationships companies on the sending side (the United States, with commercial banks due to concerns related to the United Kingdom, Australia, and others), and regulatory compliance (referred to as “de-risking”) are also regulated as money transfer business threatens the sustainability of business transac- (MTB) by the Central Bank of Somalia (CBS). How- tions by many MTOs in Somalia. ever, there are also funds transferred through non- registered “hawalas.” Remittance costs 201  Cuecuecha and Adams Jr (2016); Adams Jr and Page (2005); Acosta, et al. (2006); Yang and Martinez (2006); Lokshin, et al. Costs of remitting money to Somalia have (2010). increased due to the Anti-Money Laundering reg- 202  Hawala or Hewala, also known as hundi, is an informal value ulations. Somalia has been affected by “de-risking,” transfer system based on the performance and honor of a huge network of money brokers, primarily located in the Middle East, North Africa, the Horn of Africa, and the Indian subcontinent, operating outside of, or parallel to, traditional banking, financial The Financial Action Task Force (FATF) defines de-risking 207  channels, and remittance systems. as “the phenomenon of financial institutions terminating or 203  El Qorchi, et al. (2003). restricting business relationships with clients or categories of 204  Horst and Van Hear (2002). clients to avoid, rather than manage, risk.” Somalia has been 205  Montclos and Kagwanja (2000). affected by “de-risking”—the closing of bank accounts of 206  After September 11, one of the largest xawilaad company, Al money transfer operators by banks due to perceived legal, reg- Barakat, was closed down. Page and Plaza (2006). ulatory, sanctions, and AML/CFT risks.  Remittances 129 which refers to the financial institutions terminating UK, and IGAD. In this context, the World Bank has or restricting relationships with clients perceived as selected and appointed “Abyrint AS” to act as the high risk for money laundering or financing terror- “Trusted Agent” to the CBS and assist the authori- ism. High remittance costs represent an unneces- ties in comprehensively regulating and supervising sary burden on Somali migrants and likely reduce money transfer businesses. amounts sent and their development impact. Since the events of September 11, 2001, many countries The Central Bank of Somalia has licensed and reg- have adopted stringent Anti-Money Laundering istered four money transfer businesses and has and Combatting the Financing of Terrorism (AML/ registered nine money transfer businesses under CFT) regulations for funds transfers. Several banks the Money Transfer Business Registration Regu- in the United States (Wells Fargo, US Bank, the lations and Money Transfer Business Licensing TCF bank, and Sunrise Community Bank) and in Regulations passed by the Central Bank in 2014, the United Kingdom have closed the accounts of developed with the support of the World Bank. money services business to avoid incurring penal- The CBS recently concluded on-site examinations ties for not complying with the new regulations.208 of four of the largest MTBs operating. The CBS The account closures have changed how the remit- is working on improving MTBs compliance with tance market works in both the United Kingdom AML/CFT regulations, including reporting require- and the United States, including carrying cash ments. The World Bank is considering support directly. Banks still perceive the remittance sector to the Somalia Financial Reporting Center under as having a high risk for money laundering or ter- a proposed new program that is currently under rorism financing in Somalia. The recent closing of a discussion. Somalia does not yet have a system in bank account of a correspondent bank in Canada place for know your client (KYC) or customer due indicates that Somalia continues to be impacted diligence (CDD) requirements. Some money trans- by de-risking. Know Your Client (KYC) regulation mitters are considering the use of biometric iden- remains a concern, and related issues about the tification for meeting KYC requirements. absence of reliable identification systems need to be addressed. In general, banks consider remittance service providers as entities that do not have adequate The World Bank began working with the UK in controls, do not implement the adequate cus- 2015 to develop mechanisms, in case of severe dis- tomer due diligence, and lack the capacity to ruption of remittance flows between the UK and comply with AML/CFT regulations. The risk fac- Somalia. This work has since evolved to address tors of remittance service providers (RSPs) operat- fundamental issues affecting remittance flows to ing in Somali jurisdictions include that the majority the country. The current activities are focused on of operations are cash transactions; government improving the formalization, transparency, and oversight is weak or lacking; and operations are compliance of the money transfer business sector conducted through agents, which makes it diffi- in Somalia. The World Bank and the Federal Gov- cult to implement the “know your client” norms. A ernment of Somalia are working together to help recent report from the United States Accountabil- support the flow of remittances and to address key ity Office found that money transmitters operating deficiencies in the Somali financial sector affecting in Somalia reported using non-banking channels remittance flows to the country. The remittance such as cash couriers to move funds for cross- crisis highlighted the need for the Central Bank border transfer of remittances.209 of Somalia to start formal supervision of Somali MTOs. In response to the crisis, a larger reform Personal identification in the financial services program of policy change, institutional reforms, will be important for addressing the issues of de- and technical assistance was also developed and is risking, AML/CFT, and KYC requirements. Identifi- now being implemented. The work is coordinated cation will also facilitate the transfer of aid to IDPs through the Somali Remittances Stakeholder and refugees. The objective is to create systems Advisory Council, co-chaired by the Central Bank that are interoperable. The World Bank is devel- Governor and the World Bank and with represen- oping a project for expanding financial and digi- tatives from Somalia, IMF, AfDB, US Treasury, the tal access in Somalia, which includes components 208  Note: HSBC, a banking institution, was fined US$1.9 billion for not complying with money laundering controls in 2012. 209  GAO (2018). 130  Somali Poverty and Vulnerability Assessment FIGURE 6.7  n  Remittance cost as a proportion of sending US$200 to Somalia 12 10 8 6 4 2 0 2014Q4 2015Q1 2015Q4 2016Q1 2016Q4 2017Q1 2017Q4 2018Q1 United Kingdom Australia Source: Remittances Prices Worldwide database (2018Q1). on: (i) expanding access to finance and capabil- of US$200 (inclusive of all fees and charges) ity for micro, small, and medium enterprises; remained at 7.1 percent in 2018 Q1.210 (ii) deepening the regulatory capacity for Central Bank of Somalia, the Somalia Financial Report- According to the Remittance Prices Worldwide ing Center, and the telecommunications regula- database, the average cost of sending US$200 tor; (iii) enhancing connectivity and development from Australia and the United Kingdom to Somalia of government digital services; and (iv) extending have increased. In the United Kingdom, the remit- digital identification coverage and accessibility. tance cost increased from 6.3 percent in the fourth quarter of 2014 to 7.1 percent in the fourth quar- Three effects on the remittance markets to Soma- ter of 2017 (Figure 6.7). In Australia, three major lia are observed due to the AML/CFT regulations. banks, the Commonwealth Bank, the National Aus- First, Banks stopped offering low cost remittance tralia Bank, and the Westpac closed the accounts services. Second, banks closed accounts of MTOs. of MTOs serving Somalia in Australia. Due to the And third, small MTOs also closed since they could closure of the bank accounts, remittance costs not any longer operate without bank accounts. increased from 9.06 percent in the first quarter These developments in the remittance markets of 2016 to 11.2 percent in the first quarter of 2018. increase remittance prices, reduce competition, The costs for sending remittances from the United and encourage the use of informal channels. Kingdom to Somalia are more than twice the SDG target of 3 percent. From Australia to Somalia the The United Nations has recently adopted the costs are almost three times the SDGs target. Sustainable Development Goals (SDGs) targets and indicators for migration. The SDGs include explicit targets to ensure safe, orderly, and regular Remittances and access to finance migration, including through well-managed migra- tion policies (10.7) and reductions in the costs of Somali households that receive internal and remittance transfers (10.c). By 2030, reduce to less international remittances typically have better than 3 percent the transaction costs of migrant access to financial services such as bank accounts remittances and eliminate remittance corridors (Table 6.7). Households receiving remittances with costs higher than 5 percent. The cost of send- within the country tend to be better off in terms of ing money continues to be high and regressive, financial access, in part because households that well above the SDG target of 3 percent. Accord- send out internal migrants are using mobile money ing to the Remittance Prices Worldwide database, and could save using mobile phones that effec- the global average cost of sending remittances tively substitute for formal banking services. 210  Page and Plaza (2006). Remittances 131 TABLE 6.7  n  Remittances facilitate financial inclusion Reducing remittances costs (I) (II) (III) Technological advances including digital pay- ments have increased efficiency and contrib- Bank Mobile uted to reducing remittance costs. On the other access money Savings hand, compliance with AML/CFT requirements Internal remittances 0.082*** 0.014 0.073*** seems to have increased the overall costs of remit- S.E. (0.021) (0.064) (0.020) tances. Promoting policies that reduce entry, such as mobile licensing and increasing competition, International will decrease costs. Thus, reductions in remittance remittances 0.053** 0.047 0.031* costs can be supported by financial and regula- S.E. (0.023) (0.056) (0.017) tory frameworks that facilitate the introduction of Observations 6,058 6,060 6,048 new products, interoperability among MTOs, and the establishment of open infrastructure to col- Source: Authors’ calculations based on the SHFS 2017–18. Note: *** p<0.01, ** p<0.05, * p<0.1. Results from Probit regression lect digital payments. Reducing transaction costs controlling for income, household size, population type. increases the disposable income of poor migrants and increases their incentives to remit because the net receipts of recipients will increase. An impor- tant barrier to lowering remittance fees arises from Policy recommendations the costs associated with implementing AML/CFT requirements. Further development at the national Remittances are private money that belong to level of a risk-based approach to AML/CFT regula- the households. However, remittances can be lev- tion could help reduce these costs. Somalia is work- eraged at the macroeconomic level (accessing ing on complying with AML/CFT requirements and capital markets and improving credit ratings) and establishing a digital identification which could the microeconomic level (accessing new financial facilitate and reduce ‘de-risking’ by international products for micro-insurance, education, food, and banks. Remittance services to Somalia have been micro and small and medium enterprises). This impacted, including banking and trade operations. section outlines policies to leverage remittances for development for Somali people. Policies to foster the use of innovative mobile money–transfer technologies and payment sys- tems will help to reduce costs. Mobile money Improving remittance data transfer systems offer new opportunities for more effective ways of sending money. Although Soma- The Central Bank of Somalia is working to lis use mobile money widely, less than a third sub- improve statistics on remittances. The CBS can scribe to mobile services. Measures that would also improve data collection by expanding the encourage the expansion of mobile phones to be reporting of remittances to all nonbank providers able to undertake international remittances include of remittance services (such as money-transfer (i) harmonizing banking and telecommunications companies, mobile operators) and using surveys regulations to enable mainstream Somali banks to of migrants and recipient households to estimate participate in mobile money transfers and for tele- remittance flows through formal and informal communication firms to offer microdeposit and channels. Having an estimation of the volume of savings accounts, (ii) simplifying AML/CFT regu- remittances will help in the preparation of the debt lations for small-value transfers, and (iii) ensur- sustainability analysis for Somalia once a decision ing that mobile distribution networks are open to on the HIPC initiative has been taken. Improving multiple international remittance service provid- data collection on remittances is also receiving ers, instead of becoming exclusive partnerships attention from the international community: The between international MTO and country-based Sustainable Development Goals and the Global mobile money services. The Central Bank of Soma- Compact on Migration promote improving remit- lia is preparing mobile money regulations. tance data collection as one of its areas of action. 132  Somali Poverty and Vulnerability Assessment Facilitating financial inclusion migrants with better control over the use of remit- tances. In addition, basic savings accounts where Having access to financial products facilitated by remittances can be paid, small savings deposited, remittances contributes to reduce poverty. Access and payments processed should be offered in con- of poor migrants and their families to formal finan- nection with remittances. cial services for sending and receiving remittances could be improved through public policies that encourage expansion of banking networks, pro- vide identification cards to migrants, and facilitate Reaching HIPC decision point the participation of microfinance institutions and for Somalia credit unions in providing low-cost remittance ser- Somalia is eligible for debt relief under HIPC, vices. The issue of identification cards is important which will facilitate the use of funds on programs for both sending and receiving countries. In Soma- that benefit the poor. Facilitating access to con- lia, the issuance of a digital ID for financial ser- cessional financing through reduction in the debt vices will help to comply with KYC regulations. For burden will facilitate Somalia’s reconstruction. A Somalis residing in Kenya, IDs will facilitate access poverty reduction strategy will be prepared that to financial institutions. focuses on expenditures on health, education, and social services. Monitoring of the progress in implementing the poverty reduction strategy and Developing new products the macroeconomic program will be important to ensure the focus on the humanitarian and devel- New financial products such as micro-insurance opment needs of Somali people. Conversations or direct payments of tuition could be offered for with the government are taking place to start the the remitters. Remittance-linked insurance prod- process. ucts could help to protect the downside of at-risk populations. Equally, direct payments provide Remittances 133 References Acosta, P., C. Calderon, P. Fajnzylber, and H. López. Council, U.E.a.S. “Guiding Principles on Internal “Remittances and Development in Latin Amer- Displacement.” 1998. ica.” The World Economy 29, no. 7 (2006): Cuecuecha, A., and R.H. Adams Jr. “Remittances, 957–87. Household Investment and Poverty in Indone- Adams Jr, R.H., and J. Page. “Do International sia.” Journal of Finance and Economics 4, no. 3 Migration and Remittances Reduce Poverty in (2016): 12–31. Developing Countries?”. World development 33, Daidone, S., L. Pellerano, S. Handa, and B. Davis. no. 10 (2005): 1645–69. “Is Graduation from Social Safety Nets Possible? Alderman, H., and C.H. Paxson. “Do the Poor Insure? Evidence from Sub‐Saharan Africa.” IDS Bulletin A Synthesis of the Literature on Risk and Con- 46, no. 2 (2015): 93–102. sumption in Developing Countries.” In Econom- Dar, O.A., and M.S. Khan. “Millennium Development ics in a Changing World, 48–78: Springer, 1994. Goals and the Water Target: Details, Definitions Baker, J.L. Evaluating the Impact of Development and Debate.” Tropical medicine & international Projects on Poverty: A Handbook for Practitio- health 16, no. 5 (2011): 540–44. ners. World Bank Publications, 2000. Davis, B., S. Handa, N. Hypher, N.W. Rossi, P. Win- Banerjee, A.V., and E. Duflo. “The Economic Lives ters, and J. Yablonski. From Evidence to Action: of the Poor.” Journal of Economic Perspectives The Story of Cash Transfers and Impact Evalu- 21, no. 1 (2007): 141–68. ation in Sub Saharan Africa. Oxford University Barham, B., and S. Boucher. “Migration, Remit- Press, 2016. tances, and Inequality: Estimating the Net De Ferranti, D., G. Perry, I. Gill, and L. Servén. Secur- Effects of Migration on Income Distribution.” ing Our Future in a Global Economy. World Bank Journal of Development Economics 55, no. 2 Washington, DC, 2000. (1998): 307–31. Deaton, A. “Household Saving in Ldcs: Credit Mar- Bastagli, F., J. Hagen-Zanker, L. Harman, V. Barca, kets, Insurance and Welfare.” The Scandinavian G. Sturge, T. Schmidt, and L. Pellerano. “Cash Journal of Economics (1992): 253–73. Transfers: What Does the Evidence Say.” A rig- . “Measuring Poverty.” Understanding pov- orous review of programme impact and the role erty (2006): 3–15. of design and implementation features. London: Dercon, S., and P. Krishnan. “Vulnerability, Season- ODI, 2016. ality and Poverty in Ethiopia.” The Journal of Centre for Research on the Epidemiology of Disas- Development Studies 36, no. 6 (2000): 25–53. ters (CRED). “Emergency Events Database, Em- El Qorchi, M., S.M. Maimbo, S.M. Autmainbo, and Dat.” Brussels, Belgium: Université catholique de J.F. Wilson. Informal Funds Transfer Systems: An Louvain, 2017. Analysis of the Informal Hawala System. Vol. 222. Chami, R., E. Ernst, C. Fullenkamp, and A. Oek- International Monetary Fund, 2003. ing. “Are Remittances Good for Labor Markets European Commission. “Somalia: In Pursuit of a in Lics, Mics and Fragile States?” International Safety Net Programme in the Short Term Paving Monetary Fund, 2018. the Way to a Social Protection Approach in the Chami, R., D. Hakura, and P. Montiel. Remittances: Long Term: Issues and Options. Final Report.” An Automatic Output Stabilizer? International Advisory Service in Social Transfers, European Monetary Fund, 2009. Commission, 2017. Chaudhuri, S. “Empirical Methods for Assessing Fafchamps, M., C. Udry, and K. Czukas. “Drought Household Vulnerability to Poverty.” School of and Saving in West Africa: Are Livestock a Buf- International and Public Affairs, Columbia Uni- fer Stock?”. Journal of Development Economics versity, New York, 2000. 55, no. 2 (1998): 273–305. References 135 FAO (Food and Agriculture Organization of the Analysis Unit—Somalia and Famine Early Warn- United Nations). “Integrated Food Security ing Systems Network, 2017b. Phase Classification. Technical Manual Version . “Somalia Food Security Outlook. Despite 2.0.” 2012. Improvements, 2.7 Million People Need Emer- Federal Government of Somalia (FGS), W.B., United gency Assistance through the Lean Season.” Nations, European Union. “Somalia Drought Food Security and Nutrition Analysis Unit— Impact & Needs Assessment.” 2018. Somalia and Famine Early Warning Systems FEWSNET. “East Africa Special Report. Illustrat- Network, 2018. ing the Extent and Severity of the 2016/17 Horn . “Somalia Food Security Outlook. Risk of of Africa Drought.” Famine Early Warning Net- Famine (Ipc Phase 5) Persists in Somalia.” Food work, 2017. Security and Nutrition Analysis Unit—Somalia . “Somalia Food Security Outlook. June 2016 and Famine Early Warning Systems Network, to January 2017”: Famine Early Warning Sys- 2017c. tems Network, 2016. . “Study Suggests 258,000 Somalis Died Due Filmer, D., and L.H. Pritchett. “Estimating Wealth to Severe Food Insecurity and Famine.” Food Effects without Expenditure Data—or Tears: An Security and Nutrition Analysis Unit—Somalia Application to Educational Enrollments in States and Famine Early Warning Systems Network, of India.” Demography 38, no. 1 (2001): 115–32. 2013. Fisseha, G., Y. Berhane, A. Worku, and W. Terefe. Fund for Peace. “Fragile States Index 2018: Issues “Distance from Health Facility and Mothers’ Per- of Fragility Touch the World’s Richest and Most ception of Quality Related to Skilled Delivery Developed Countries in 2018,” 2018. Service Utilization in Northern Ethiopia.” Inter- Funk, C., P. Peterson, M. Landsfeld, D. Pedreros, national Journal of Women’s Health 9 (2017): J. Verdin, S. Shukla, G. Husak, et al. “The Climate 749. Hazards Infrared Precipitation with Stations— Food Security Cluster. “Somalia—Fsc Monthly a New Environmental Record for Monitoring Dashboard, January 2018.” Food Security Clus- Extremes.” Scientific Data 2 (2015): 150066. ter Somalia, 2018. GAO. “Remittances to Fragile Countries: Trea- . “Somalia—Fsc Monthly Dashboard, July sury Should Assess Risks from Shifts to Non- 2017.” Food Security Cluster Somalia, 2017. Banking Channels.” U.S. Government Account- Frankel, J. “Are Bilateral Remittances Countercycli- ability Office, 2018. cal?”. Open Economies Review 22, no. 1 (2011): Ghorpade, Y. “Extending a Lifeline or Cutting 1–16. Losses? The Effects of Conflict on Household FSNAU. “Food Security and Nutrition Analysis: Receipts of Remittances in Pakistan.” World Post Deyr 2010/11.” Nairobi, 2011. Development 99 (2017): 230–52. . “Special Brief: Focus on Post Deyr 2016 Hagen-Zanker, J., F. Bastagli, L. Harman, V. Barca, Early Warning.” Food Security and Nutrition G. Sturge, and T. Schmidt. “Understanding the Analysis Unit—Somalia, 2016a. Impact of Cash Transfers: The Evidence.” ODI Briefing. London: Overseas Development Insti- . “Special Brief: Focus on Post Gu 2016 Early tute, 2016. Warning.” Food Security and Nutrition Analysis Unit—Somalia, 2016b. Haseeb, M., and K. Vyborny. “Reforming Institu- tions: Evidence from Cash Transfers in Pakistan,” . “Special Brief: Focus on Post Gu 2017 Early 2017. Warning.” Food Security and Nutrition Analysis Unit—Somalia, 2017. Health Cluster Somalia. “Health Cluster Bulletin, December 2017.” Health Cluster Somalia, 2017. FSNAU, and FEWSNET. “Quarterly Brief—Focus on Post Gu 2017 Season Early Warning.” Food Himelein, K., S. Eckman, and S. Murray. “Sampling Security and Nutrition Analysis Unit—Somalia Nomads: A New Technique for Remote, Hard-to- and Famine Early Warning Systems Network, Reach, and Mobile Populations.” Journal of Offi- 2017a. cial Statistics 30, no. 2 (2014): 191–213. . “Somalia Food Security Alert. Severe HM Government. “An Evidence Review of the Driv- Drought, Rising Prices, Continued Access Lim- ers of Child Poverty for Families in Poverty Now itations, and Dry Forecasts Suggest Famine Is and for Poor Children Growing up to Be Poor Possible in 2017.” Food Security and Nutrition Adults,” 2014. 136  Somali Poverty and Vulnerability Assessment Hoddinott, J., J. Lind, G. Berhane, K. Hirvonen, Jalan, J., and M. Ravallion. “Is Transient Poverty Dif- N. Kumar, B. Nishan, R. Sabates-Wheeler, et al. ferent? Evidence for Rural China.” The Journal of “Psnp-Habp Final Report, 2014.” Washington, Development Studies 36, no. 6 (2000): 82–99. DC: IFPRI, 2015. Jean, N., M. Burke, M. Xie, W.M. Davis, D.B. Lobell, Holzmann, R. “Risk and Vulnerability: The Forward- and S. Ermon. “Combining Satellite Imagery and Looking Role of Social Protection in a Globaliz- Machine Learning to Predict Poverty.” Science ing World.” Washington, DC: World Bank, 2001. 353, no. 6301 (2016): 790–94. Hoogeveen, J., E. Tesliuc, R. Vakis, and S. Der- Kelley, A., and R. Schmidt. “Economic and Demo- con. “A Guide to the Analysis of Risk, Vulner- graphic Change: A Synthesis of Models, Find- ability and Vulnerable Groups.” World Bank. ings, and Perspectives,” 1999. Washington, DC. Available online at http://site- Kimball, M.S. “Precautionary Saving and the Mar- resources. worldbank. org/INTSRM/Publica- ginal Propensity to Consume.” National Bureau tions/20316319/RVA. pdf. Processed, 2004. of Economic Research, 1990. Horst, C., and N. Van Hear. “Counting the Cost: Lall, S.V., J.V. Henderson, and A.J. Venables. Africa’s Refugees, Remittances and The ‘War against Cities: Opening Doors to the World. The World Terrorism’.” Forced Migration Review 14 (2002): Bank, 2017. 32–34. Lim, Y., and R.M. Townsend. “General Equilibrium “Iasc Framework on Durable Solutions.” Brookings Models of Financial Systems: Theory and Mea- Institution, 2010. surement in Village Economies.” Review of Eco- Imbens, G., and J. Wooldridge. “What’s New in nomic Dynamics 1, no. 1 (1998): 59–118. Econometrics? Lecture Notes 10, Difference-in- Lokshin, M., M. Bontch‐Osmolovski, and E. Glin- Differences Estimation.” Cambride, MA: National skaya. “Work‐Related Migration and Poverty Bureau of Economic Research, 2007. Reduction in Nepal.” Review of Development IMF. “IMF Executive Board Concludes 2017 Article Economics 14, no. 2 (2010): 323–32. Iv Consultation with Somalia, and Completion of Lopez, J.H., P. Fajnzylber, and P. Acosta. The Impact the First Review under the Staff-Monitored Pro- of Remittances on Poverty and Human Capital: gram (Smp).” news release, 2018a. Evidence from Latin American Household Sur- . “IMF Staff Completes Review Visit and veys. The World Bank, 2007. Reaches Staff-Level Agreement on a Third Staff- Majoka, Z. “In Somalia, Resilience Can Be Strength- Monitored Program with Somalia.” news release, ened through Social Protection System.” World 2018b. Bank, 2017. . “World Economic Outlook: Globalization Maxwell, D., and R. Caldwell. “The Coping Strat- and External Imbalances.” Washington, DC: egies Index: A Tool for Rapid Measurement of International Monetary Fund, 2005. Household Food Security and the Impact of “The Impact of War on Somali Men.” news release., Food Aid Programming in Humanitarian Emer- 2015, http://www.logica-wb.org/PDFs/LOGICA_ gencies.” Field Methods Manual, 2nd Edition, The_Impact_of_War_on_Somali_Men.pdf. January, 2008. Institute for Economics and Peace. “Global Peace McKenzie, D.J., and M.J. Sasin. Migration, Remit- Index 2017,” 2017a. tances, Poverty, and Human Capital: Conceptual . “Global Terrorism Index 2017: Measuring and Empirical Challenges. Vol. 4272: World Bank and Understanding the Impact of Terrorism,” Publications, 2007. 2017b. Mitchell, T., and K. Harris. “Resilience: A Risk Man- International Monetary Fund. “Somalia: 2017 Arti- agement Approach.” ODI Background Note. cle Iv Consultation and First Review under the Overseas Development Institute: London, 2012. Staff-Monitored Program—Press Release; Staff Mohapatra, S., G. Joseph, and D. Ratha. “Remit- Report; and Statements by Executive Director tances and Natural Disasters: Ex-Post Response for Somalia, February 2018.” International Mon- and Contribution to Ex-Ante Preparedness,” etary Fund, 2018. 2009. Jack, W., and T. Suri. “Risk Sharing and Transac- Montclos, M.-A.P.d., and P.M. Kagwanja. “Refugee tions Costs: Evidence from Kenya’s Mobile Camps or Cities? The Socio-Economic Dynamics Money Revolution.” American Economic Review of the Dadaab and Kakuma Camps in Northern 104, no. 1 (2014): 183–223. References 137 Kenya.” Journal of Refugee Studies 13, no. 2 Remittances, and Development: The Experience (2000): 205–22. of the Northern Triangle. Edited by Tanida Aray- Myers, K. “Tackling Climate Change to Reduce avechkit, Kinnon Scott, and Liliana Sousa. Forth- Poverty.” Concern Worldwide, https://www coming, World Bank. .concernusa.org/story/tackling-climate-change- Singh, M.R.J., K.-w. Lee, and M.M. Haacker. Deter- to-reduce-poverty/, 2017. minants and Macroeconomic Impact of Remit- Norwegian Refugee Council. “Thousands of Peo- tances in Sub-Saharan Africa. International ple Forcibly Removed.” https://www.nrc.no/ Monetary Fund, 2009. news/2018/january/thousands-of-people- “The Sustainable Transformation of Youth in Libe- forcibly-removed/. ria (Styl) Program.” edited by World Bank, 2015. OECD. “The OECD List of Social Indicators,” 1982. Tesliuc, E.D., and K. Lindert. Risk and Vulnerabil- Page, J., and S. Plaza. “Migration Remittances and ity in Guatemala: A Quantitative and Qualitative Development: A Review of Global Evidence.” Assessment. Social Protection, Labor Markets, Journal of African Economies 15, no. suppl_2 Pensions, Social Assistance, World Bank, 2004. (2006): 245–336. UNDESA. “International Migration Stock: The 2017 Pape, U., and J. Mistiaen. “Measuring Household Revision.” UNDESA United Nations Department Consumption and Poverty in 60 Minutes: The of Economic and Social Affairs Population Divi- Mogadishu High Frequency Survey.” World sion, 2017. Bank, 2015. UNHCR. “Mid-Year Trends 2017.” United Nations Pape, Utz Johann, and Philip Randolph Wollburg. High Commissioner for Refugees, 2017. “Estimation of Poverty in Somalia Using Innova- . “Somalia 1–31 March 2018. Fact Sheet.” tive Methodologies.” 2019. United Nations High Commissioner for Refu- Plaza, S., and D. Ratha. Diaspora for Development gees, 2018a. in Africa. World Bank Publications, 2011. . “Somalia 1–31 March 2018. Repatriation Ralston, L., C. Andrews, and A. Hsiao. “A Meta- Update.” United Nations High Commissioner for Analysis of Safety Nets Programs in Africa.” Refugees, 2018b. Working Paper, 2017. UNHCR (United Nations High Commissioner for Ratha, D. “Leveraging Remittances for Develop- Refugees). “Camp Coordination and Camp Man- ment,” Mpi Policy Brief, June 2007, Washington, agement Cluster. February 2018.” United Nations 2007. High Commissioner for Refugees, 2018a. Ravallion, M., S. Chen, and P. Sangraula. “Dollar . “Horn of Africa Somalia Situation.” https:// a Day Revisited.” The World Bank Economic data2.unhcr.org/en/situations/horn, 2018b. Review 23, no. 2 (2009): 163–84. . “Internal Displacement Profiling in Mogadi- RMMS. “Somalia/Somaliland Country Profile.” shu,” 2016. Regional Mixed Migration Secretariat, Horn of . “Somalia Internal Displacement: Displace- Africa and Yemen, Danish Refugee Council, 2016. ments Monitored by UNHCR Protection and Rosenzweig, M.R., and K.I. Wolpin. “Credit Market Return Monitoring Network (PRMN).” https:// Constraints, Consumption Smoothing, and the unhcr.github.io/dataviz-somalia-prmn/index Accumulation of Durable Production Assets in .html. Low-Income Countries: Investments in Bullocks . “Somalia Situation 2017.” New York, NY: in India.” Journal of Political Economy 101, no. 2 United Nations, 2017. (1993): 223–44. UNICEF. “The Situation Analysis of Children in Salama, P., G. Moloney, O.O. Bilukha, L. Talley, Somalia 2016,” 2016. D. Maxwell, P. Hailey, C. Hillbruner, et al. “Famine United Nations Population Fund (UNFPA). “Soma- in Somalia: Evidence for a Declaration.” Global lia Population Estimation Survey 2014: For the Food Security 1, no. 1 (2012): 13–19. 18 Pre-War Regions of Somalia.” Nairobi, Kenya, Schaaf, Z. “Mcd43a3 Modis/Terra+ Aqua Brdf/ 2014. Albedo Daily L3 Global–500m V006. Nasa Eos- UNOCHA (United Nations Office for the Coordina- dis Land Processes Daac,” 2015. tion of Humanitarian Affairs). “2018 Humanitar- Scott, K., and J. Soler. “Remittances and Their ian Needs Overview: Somalia,” 2017a. Effects on Poverty and Inequality.” In Migration, 138  Somali Poverty and Vulnerability Assessment . “Horn of Africa Humanitarian Outlook. . “Somalia Social Protection: Stocktaking January–June 2018.” United Nations Office for of Evidence for a Social Protection Policy and the Coordination of Humanitarian Affairs, 2017b. Framework.” Washington, DC: World Bank, . “Humanitarian Bulletin Somalia, June 2– 2017b. July 5, 2018.” New York, NY: United Nations, . “Somalia Systematic Country Diagnostic.” 2018a. Washington, DC: World Bank, 2018d. . “Humanitarian Bulletin Somalia, March . “South Sudan Poverty Assessment.” World 2018.” New York, NY: United Nations, 2018b. Bank, 2018e. . “Somalia Humanitarian Response Plan 2018. . “The State of Social Safety Nets 2018.” January–December 2018.” United Nations Office Washington, DC: World Bank, 2018f. for the Coordination of Humanitarian Affairs, . “Unbreakable: Building the Resilience of the 2018c. Poor in the Face of Natural Disasters.”. Washing- Ward, L.M., and C. Eyber. “Resiliency of Children in ton, DC: World Bank, 2017c. Child-Headed Households in Rwanda: Implica- . “World Development Report 2011: Conflict, tions for Community Based Psychosocial Inter- Security, and Development.” Washington, DC: ventions.” Intervention 7, no. 1 (2009): 17–33. World Bank, 2011. World Bank. “Bilateral Migration Matrix 2017.” . “World Development Report 2014: Risk and Washington, D.C.: World Bank Group, 2018a. Opportunity—Managing Risk for Development.” . “Challenges and Opportunities of High Fre- Washington, DC: World Bank, 2013. quency Data Collection in Fragile States: Les- World Food Programme. “Somalia: Food Market sons from South Sudan.” World Bank, 2014. and Supply Situation in Southern Somalia, Octo- . Conflict in Somalia: Drivers and Dynamics. ber 2011.” Geneva: World Food Programme, 2011. World Bank, 2005. Xie, M., N. Jean, M. Burke, D. Lobell, and S. Ermon. . “Global Economic Prospects 2006: Eco- “Transfer Learning from Deep Features for nomic Implications of Remittances and Migra- Remote Sensing and Poverty Mapping.” arXiv tion.” Washington, D.C.: World Bank Group, preprint arXiv:1510.00098, 2015. 2006. Yang, D., and C. Martinez. “Remittances and Pov- . “Macro Poverty Outlook Somalia.” Wash- erty in Migrants’ Home Areas: Evidence from ington, DC: World Bank, 2018b. the Philippines.” International Migration, Remit- . “Social Risk Management: The World Bank’s tances and the Brain Drain, no. 3, 2006. Approach to Social Protection in a Globalizing Zanini, G., S.P. D’Alessandro, V. Phipps-Ebeler, C.M. World.” Washington, DC: World Bank, 2003. Ngumbau, P. Sanginga, J. Seevinck, Y. Cherrou, . “Somali Poverty Profile, June 2017: Findings A. Read, and S. Akester. “Rebuilding Resilient and from Wave 1 of the Somali Hih Frequency Sur- Sustainable Agriculture in Somalia: Volume 1— vey.” Washington, DC: World Bank, 2017a. Main Report (English).” World Bank, 2018. . “Somalia Drought Impact and Needs Zeldes, S.P. “Optimal Consumption with Stochastic Assessment: Synthesis Report.” Washington, Income: Deviations from Certainty Equivalence.” DC: World Bank, 2018c. The Quarterly Journal of Economics 104, no. 2 (1989): 275–98. References 139 APPENDIX A Figures and Tables FIGURE A.1  n  Population pyramid 85+ 80–84 75–79 70–74 65–69 60–64 55–59 50–54 45–49 Age 40–44 35–39 30–34 25–29 20–24 15–19 10–14 5–9 0–4 –10% –5% 0% 5% 10% Men Women Source: Authors’ calculations based on the SHFS 2017–18. TABLE A.1  n  Accessibility rate of urban and rural areas Region Urban areas Rural areas Mogadishu 87% N/A North East 99% 89% North West 98% 97% Central regions 77% 52% Jubbaland 64% 26% South West 50% 34% Source: Authors’ calculations based on the SHFS 2017–18. Figures and Tables 141 FIGURE A.2  n  Poverty measures by gender of the FIGURE A.4  n  Poverty measures by displacement household head status of the household 90 90 80 80 70 70 60 60 Percent Percent 50 50 40 40 30 30 20 20 10 10 0 0 e rty rty rty rty p y e rty ty rty rty p y rit rit nc ga nc ga er ve ve ve ve ve ve ve ve ve de de ov ty rty po po po po po po po se se ci er ci tp ve in in od nt ld th od ld h v ty rty en Po ut Po hi hi le u ty er rty Fo Fo ve Yo al Yo C C va er v ve iv Po Po v ui qu Po Po eq te lt ul u Ad Ad Female headed HH Male headed HH Non-IDPs IDPs Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE A.3  n  Poverty measures by remittance status FIGURE A.5  n  Poverty measures by drought affected of the household status of the household 90 90 80 80 70 70 60 60 Percent 50 Percent 50 40 40 30 30 20 20 10 10 0 0 e rty ty rty rty p y rit nc ga er ce rty ty rty ty p y ve ve ve ve rit ga de ov er er rty en ve ve po po po ve se v v ci tp rty ve po po id po po se in od ld h rty n c ve ut Po hi le in rty od nt ld h Fo rty ve Yo C ut Po a hi e rty ve Fo iv ve Po al Yo C qu ve Po iv Po u te Po eq ul t Ad ul Ad Not received remittances Received remittances Not drought affected Drought affected Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. 142  Somali Poverty and Vulnerability Assessment Adult equivalent measure FIGURE A.6  n  Adult equivalent measure of poverty incidence of poverty 80 Percent of adult equivalent population An adult equivalent measure of poverty considers 70 the age and composition of households, as it rec- 60 ognizes economies of scale within them. Unlike the poverty headcount ratio, an adult equivalent 50 measure of poverty incidence considers the size of 40 the households, not in numbers of persons but in numbers of adult equivalents (AE), acknowledging 30 economies of scale within the household. The scale 20 used from OECD (1982) considers two children to be equivalent to an adult, and the second and sub- 10 sequent number of adults in the household only as 0 a 0.7 fraction of an effective adult equivalent. hu an al ts s ad ur en is rb om R ad m ru e Even though consumption per capita is a widely og N e ttl th M se O used measure for monetary poverty, consumption in Ps per adult equivalent acknowledges economies of ID scale within the household by considering the Overall average size and the age of its members. A child typically requires smaller amounts of food than an adult, Source: Authors’ calculations based on the SHFS 2017–18. and to reflect this the Organisation for Economic Co-operation and Development (OECD) scale measures the size of the households not in num- 11 percentage points smaller for households that bers of persons but in numbers of adult equiva- received remittances (33 percent), compared to lents.211 Forty-two percent of the Somali population non-receivers (44 percent, p<0.01), and 9 percent- are poor under this approach (Figure A.6).212 Adult age points smaller for households not located in equivalent poverty incidence ranges from 39 to IDP settlements or not displaced (39 percent) rela- 58 percent across population groups, but its simi- tive to IDPs in settlements and outside of them (49 lar or not statistically different between Mogadishu percent, p<0.05). Differences between households (40 percent), rural areas (53 percent), IDPs in set- headed by men and women are found for the pov- tlements (49 percent), and the nomadic popula- erty incidence, but not when considering an adult tion (39 percent). Households in other urban areas equivalent measure since households headed by are less likely to be poor compared to those in rural men have more children (2.8) compared to house- areas (18 percentage point difference, p<0.05) holds headed by women (2.5). and living in IDP settlements (14 percentage point difference, p<0.05). Furthermore, incidence is OECD (1982). 211  The OECD equivalence scale is recommended for countries 212  which have not established their own equivalence scale, like Somalia. Figures and Tables 143 TABLE A.2  n  Demographic attributes of poor households by population group Dependent variable: Poverty status Other IDPs in Independent variables Mogadishu urban Rural settlements Nomads Household size 0.88*** 0.64*** 0.27 0.10 1.06*** Age dependency ratio 0.14 0.35* 0.36 –0.55 0.17 Number of children –0.18 –0.18 0.06 0.87*** –0.33 Proportion of men in the household 8.24 –1.31 –13.27 0.69 15.8* Share of households headed by men 4.33 –4.99* 11.28*** 9.97* 4.27 Age of household head –0.03* –0.01 –0.01 0.04** –0.03* Share of literate household heads 4.77 7.69 –5.66 –3.26 –3.95 Share of literate members in the household –14.57* –13.57*** –1.68 –0.41 1.83 Share of households with improved sources –2.23 3.92 6.63 –6.93 3.32 of water Share of households with improved sanitation –3.90 –4.92* 6.80 0.55 –3.96 Share of households with access to electricity –8.96* –12.48*** –15.72*** –13.37*** 12.44** Main source of income: Salaried labor Reference Reference Reference Reference Reference Main source of income: Agriculture, fishing & –9.51 12.14** –10.15** –9.84 7.82 hunting Main source of income: Small family business 0.58 2.47 –15.68** –14.22** 4.54 Main source of income: Remittances –0.08 –1.43 –3.26 –5.99 14.56 Main source of income: Other 4.11 –4.01 –14.90*** –10.60** 11.00 Observations 5,945 888 3,098 466 487 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). The poverty status was derived from total core consumption and a rescaled poverty line. The coefficients were estimated from a logistic regression model with population and region fixed effects. The reported values correspond to the marginal effects. TABLE A.3  n  Child poverty and key household characteristics Dependent variable: Poverty status of children Independent variables (1) (2) (3) (4) (5) Household headed by men 0.409** 0.405** Receiving remittances –0.208 –0.211 Displaced household 0.022 0.003 Household affected by the drought 0.300 0.317 Region and population fixed effects Yes Yes Yes Yes Yes Observations 16,369 16,369 16,369 16,369 16,369 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from a logistic regression. The poverty status was derived from total core consumption and a rescaled poverty line. 144  Somali Poverty and Vulnerability Assessment TABLE A.4  n  Poverty incidence and key household characteristics Dependent variable: Poverty status Independent variables (1) (2) (3) (4) (5) Household headed by men 0.274* 0.276* Receiving remittances –0.417** –0.419** Displaced household –0.001 –0.005 Household affected by the drought 0.009 0.020 Region and population fixed effects Yes Yes Yes Yes Yes Observations 6,092 6,092 6,092 6,092 6,092 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from a logistic regression. The poverty status was derived from total core consumption and a rescaled poverty line. TABLE A.5  n  Poverty gap and key household characteristics Dependent variable: Poverty gap Independent variables (1) (2) (3) (4) (5) Household headed by men 0.023 0.022 Receiving remittances –0.068*** –0.069*** Displaced household 0.036 0.037 Household affected by the drought –0.007 –0.007 Region and population fixed effects Yes Yes Yes Yes Yes Observations 6,092 6,092 6,092 6,092 6,092 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from an OLS regression. TABLE A.6  n  Youth poverty and key household characteristics Dependent variable: Poverty status of youth Independent variables (1) (2) (3) (4) (5) Household headed by men –0.029 –0.024 Receiving remittances –0.524** –0.525** Displaced household 0.489 0.481 Household affected by the drought –0.132 –0.165 Region and population fixed effects Yes Yes Yes Yes Yes Observations 4,866 4,866 4,866 4,866 4,866 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from a logistic regression. The poverty status was derived from total core consumption and a rescaled poverty line. Figures and Tables 145 TABLE A.7  n  Hunger and key household characteristics Dependent variable: Experiencing hunger Independent variables (1) (2) (3) (4) (5) Poor household 0.035 0.014 0.030 0.024 0.021 Household headed by men –0.137 –0.145 Receiving remittances –0.335* –0.364* Displaced household 0.360 0.373 Household affected by the drought 0.547*** 0.548*** Region and population fixed effects Yes Yes Yes Yes Yes Observations 6,063 6,063 6,063 6,063 6,063 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from a logistic regression. TABLE A.8  n  Education of the household head Dependent variable: Household head with some formal education Independent variables (1) (2) (3) (4) (5) Poor household 0.282 0.305* 0.279 0.282 0.304* Household headed by men 0.752*** 0.766*** 0.808*** 0.756*** 0.825*** Age of household head –0.031*** –0.032*** –0.030*** –0.031*** –0.032*** Receiving remittances 0.552*** 0.558*** Displaced household –1.683*** –1.682*** Household affected by the drought 0.188 0.220 Region and population fixed effects Yes Yes Yes Yes Yes Observations 4,279 4,279 4,279 4,279 4,279 Source: Authors’ calculations based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Estimated coefficients from a logistic regression. 146  Somali Poverty and Vulnerability Assessment FIGURE A.7  n  Age dependency ratio by quintile 2.0 Dependents to working age (15–64) ratio 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 Q1 Q2 Q3 Q4 Q5 (bottom) (top) Overall average Source: Authors’ calculations based on the SHFS 2017–18. FIGURE A.8  n  Households deprived in each dimension213 100 Percent of households 80 60 40 20 0 ll hu an al ts s r r o oo ra ad ur en Po is rb ve -p om R ad em u on O er og N ttl N th M se O in Ps ID Living standards Water and sanitation Education Source: Authors’ calculations based on the SHFS 2017–18. 213  A living standards dimension captures the type of dwelling, access to electricity, and source of energy for cooking. A household is deemed deprived along this dimension if they meet at least one of the following three criteria: (i) does not have access to electricity; (ii) the dwelling is not classified as improved housing (apartment or house); and (iii) uses dung, wood, charcoal, or grass as the main source of energy for cooking. Figures and Tables 147 FIGURE A.9  n  Households deprived in living FIGURE A.10  n  Households deprived in educational standards dimension by group dimension by group 100 100 Percent of households Percent of households 80 80 60 60 40 40 20 20 0 0 r Ps ed -p r de HH H an s s ID s fe d r ed H H an s s ID s Ps fe d -p r ed on o oo on o oo itt e ce P af te itt e ce P af te H ad H H N Po N Po m nc ct m nc ID ct D ht fec ht fec he ed d he ed I re ta - re ta - on on ug af ug af e d e d d it d it al ea a ve em al ea ve em N N ro t ro t D gh D gh h h ei d r ei d r ou ou e e al al ec e ec e dr dr R iv R iv m M m M ce ce Fe ot Fe ot re N re N ot ot N N Overall average Overall average Source: Authors’ calculations based on the SHFS 2017–18. Source: Authors’ calculations based on the SHFS 2017–18. FIGURE A.11  n  Households deprived in water and sanitation dimension by group 80 households Percent of 60 40 20 0 r de H H an s s Ps Ps d or ed oo m nce ce ht cte he d H H Po ct -ID ID -p d ug ffe fe d itta e on on itt af ad a a N ve m N D ht he ei re re g ou e e ec d ro al al R ive dr m M ce Fe ot re N ot N Overall average Source: Authors’ calculations based on the SHFS 2017–18. 148  Somali Poverty and Vulnerability Assessment APPENDIX B TABLE B.1  n  Urban non-settlement and settlement IDPs have better access to services than rural IDPs (1) (2) (3) (4) (5) (6) (7) (8) (9) At least Improved Proportion Proportion one Own Electricity Water housing Sanitation enrolled literate employed dwelling Internet Rural IDPs –1.612*** –0.693 –2.855*** –1.824*** 0.196** –0.0151 0.0722 –0.0399 –0.0785 –0.527 –0.907 –0.626 –0.578 –0.0838 –0.057 –0.0822 –0.0977 –0.0575 IDPs in settlements –0.291 –0.374 –0.528 0.622 0.0152 0.0856 0.103 –0.0500 0.0368 (0.431) (0.411) (0.488) (0.457) (0.0730) (0.0681) (0.0707) (0.0753) (0.0588) Poverty incidence –1.137*** –0.178 –0.0396 –0.128 –0.0811 –0.0649 0.0233 –0.0210 0.000978 Intra-Urban Analyses (0.295) (0.390) (0.515) (0.518) (0.0689) (0.0576) (0.0773) (0.0952) (0.0622) Female-headed 0.0749 0.448* 0.342 –0.541 0.00546 0.0158 0.0764 –0.0635 0.0471 household (0.388) (0.234) (0.303) (0.354) (0.0479) (0.0442) (0.0533) (0.0558) (0.0447) Constant 1.048** –0.136 –0.418 2.046*** 0.348*** 0.459*** 0.620*** 0.384*** 0.127** (0.462) (0.432) (0.476) (0.407) (0.0811) (0.0648) (0.0797) (0.0917) (0.0637) N 1,028 1,028 1,028 1,028 732 1,027 1,028 1,028 1,023 Source: Authors’ calculations based on the SHFS 2017–18. Standard errors in parentheses * p<0.10, ** p<0.05, *** p<0.01. Intra-Urban Analyses 149 150  TABLE B.2  n  Urban IDPs are consistently worse off in terms of access to services compared to other urban households (1) (2) (3) (4) (5) (6) (7) (8) (9) Somali Poverty and Vulnerability Assessment At least Improved Proportion Proportion one Own Electricity Water housing Sanitation enrolled literate employed dwelling Internet IDPs –1.893*** –1.342*** –1.310*** –1.163*** –0.309*** –0.264*** –0.120** –0.144** –0.108** (0.310) (0.334) (0.287) (0.369) (0.0685) (0.0514) (0.0584) (0.0663) (0.0419) Poverty incidence –0.491* 0.451 –0.122 –0.248 0.0470 –0.0255 0.0403 0.0323 –0.0511 (0.266) (0.290) (0.205) (0.369) (0.0465) (0.0373) (0.0352) (0.0366) (0.0429) Female-headed –0.0386 –0.114 –0.148 –0.164 –0.0110 –0.0154 0.0862** –0.0474 –0.0377 household (0.213) (0.234) (0.187) (0.320) (0.0367) (0.0360) (0.0426) (0.0499) (0.0470) Constant 2.612*** 1.171*** 1.217*** 3.044*** 0.585*** 0.719*** 0.725*** 0.489*** 0.312*** (0.265) (0.310) (0.218) (0.298) (0.0397) (0.0339) (0.0464) (0.0425) (0.0507) N 4011 4011 4011 4011 2733 4010 4011 4011 3994 Source: Authors’ calculations based on the SHFS 2017–18. Standard errors in parentheses * p<0.10, ** p<0.05, *** p<0.01. APPENDIX C Estimating the Drought Impact with a Difference-in- Differences Model Motivating the difference-in-differences model. outcomes of interest for household i at time t, pri- The difference-in-differences approach is appro- marily the poverty headcount rate, but also other priate where household or individual outcomes indicators of welfare that the drought likely affects, (y) are observed in two periods, before and after such as hunger and enrollment. postt is a binary exposure to a treatment, and where there is varia- variable indicating time period t (Wave 1, Wave 2), tion among households or individuals in the expo- and DroughtIntensityi is the continuous treatment sure to treatment. In the simplest case of a binary variable, indicating the level of drought exposure treatment, there are two groups. The first group is of household i in standard deviations of NDVI devi- exposed to the treatment in the second period but ations from the 2012–2015 average. eit denotes the not in the first. The second group is not exposed error term. b1 is the expected mean change in out- to the treatment in either of the two periods, but come from before to after the drought among the is otherwise subject to the same influences as the control group. The coefficient of the drought expo- treatment group except to the treatment itself. This sure variable, b2, is the estimated mean difference eliminates pretreatment differences in the outcome in outcomes prior to the drought: it represents variables and controls for factors changing over whatever baseline differences existed between time and affecting both groups. The validity of the households before exposure to treatment. b3 is the difference-in-differences approach is contingent difference-in-difference estimator, and hence the on a common trend assumption: that differences in coefficient of interest. Xit is a vector of control vari- outcomes would be similar in both groups if it had ables for household i at time t. not been for the treatment.214 Then the difference between the difference in outcomes over time for A set of control variables addresses potential the treatment group and the difference in outcomes bias of the estimates. Confounding factors affect- over time for the control group can be interpreted ing the outcome variables at the same time as the ˆ3DD as the effect of treatment (Equation (1)). Here, b drought, such as conflict or humanitarian assis- is the difference-in-differences estimator. tance, may violate the common trend assumption. Further, the use of repeated cross-sectional data ˆ – Treatment – y b3DD = ( y – Treatment) does not allow for household-level fixed effects 2017 2016 – – to control for all baseline differences. The model – (y2017 Control – y Control), (1) 2016 may therefore suffer from omitted variable bias. A vector of control variables Xit addresses these Implementing the difference-in-difference issues. An important baseline difference is that model. With repeated cross-sections, this continu- some regions may be more likely to experience ous ­difference-in-difference model is estimated in drought than others. Xit therefore includes the the following equation: medium-term (2002–2013) average NDVI value for each region surveyed, as a proxy for the region’s Yit = b0 + b1 postt + b2DroughtIntensityi propensity to experience drought. Price levels are + b3postt * DroughtIntensityi + b4xit + eit (2) a further potential confounding factor. They are therefore included in the regressions as controls. This equation is implemented using OLS or Pro- Further control variables fall into five categories: bit as appropriate. In Equation (2), Yit denotes regional and population-type controls, household characteristics, dwelling characteristics, exposure to conflict, and humanitarian assistance. 214  Imbens and Wooldridge (2007). Estimating the Drought Impact with a Difference-in-Differences Model 151 TABLE C.1  n  List of control variables for difference-in-differences regression Variable Description Average NDVI Average value of NDVI at the district level, 2002–2013. Price level Price level at the disaggregation of analytical strata. Regional and population type controls Region x type Interaction takes the following values: Mogadishu—urban, NE—urban, NE—rural, NW—urban, NW—rural, Central regions—urban, Central regions—rural, Jubbaland—urban, SW—urban, SW—rural. Type Urban, rural indicator. Household characteristics Household size Number of members in the household. Remittances Household remittances receipt status (Yes/No). Household head age Age of the household head (years). Household head literacy Literacy of the household head (Yes/No) Gender composition Gender composition of the household (share of males). Dwelling characteristics Tenure Tenure status of household (own, rent, other). Dwelling type Type of the dwelling (shared, separate, other). Roof material Roof material of the dwelling (metal sheets, tiles, harar, wood, plastic, other). Floor material Floor material of the dwelling (concrete, tiles or mud, other). Improved sanitation Access to improved sanitation. Conflict controls Conflict fatalities Conflict fatalities in district in past 12 month according to ACLED. Conflict x drought Interaction of drought intensity and conflict fatalities. Assistance controls Assistance in region Percentage of beneficiaries reached through food aid and livelihood inputs in 2017 in region. 152  Somali Poverty and Vulnerability Assessment FIGURE C.1  n  Hunger in December 2017 Urban areas Rural areas Percent of households Percent of households 0%–10% 0%–10% 10%–20% 10%–20% 20%–30% 20%–30% 30%–40% 30%–40% 40%–50% 40%–50% 50%–60% 50%–60% 60%–70% 60%–70% 70%–80% 70%–80% 80%–90% 80%–90% 90%–100% 90%–100% IDPs Nomads Percent of households Percent of households 0%–10% 0%–10% 10%–20% 10%–20% 20%–30% 20%–30% 30%–40% 30%–40% 40%–50% 40%–50% 50%–60% 50%–60% 60%–70% 60%–70% 70%–80% 70%–80% 80%–90% 80%–90% 90%–100% 90%–100% Source: Authors’ calculations based on the SHFS 2017–18. Estimating the Drought Impact with a Difference-in-Differences Model 153 TABLE C.2  n  IPC Phase Classification Reference Table Phase 1 Minimal Phase 2 Stressed Phase 3 Crisis Phase 4 Emergency Phase 5 Famine More than four in Even with any Even with any Even with any Even with any five households humanitarian humanitarian humanitarian humanitarian (HHs) are able to assistance at least assistance, at least assistance, at least assistance, at least Phase name and description meet essential one in five HHs in one in five HHs in one in five HHs in one in five HHs food and non-food the area have the the area have the the area have the in the area have needs without following or worse: following or worse: following or worse: an extreme lack engaging in atypical, of food and other unsustainable Minimally adequate Food consumption Large food basic needs where strategies to food consumption gaps with high or consumption gaps starvation, dealth, access food and but are unable above usual acute resulting in very high and destitution are income, including to afford some malnutrition OR acute malnutrition evident. any reliance on essential non- Are marginally able and excess mortality humanitarian food expenditures to meet minimum OR Extreme loss (Evidence for all assistance. without engaging in food needs only with of livelihood assets three criteria of irreversible coping accelarated depletion that will lead to food food consumption, strategies. of livelihood assets consumption gaps in wasting, and CDR is that will lead to food the short term. required to classify consumption gaps. as Famine). Priority response Urgent Action Required to: Action required Action required objectives to build resilience for disaster risk Protect livelihoods, Save lives and Prevent widespread and for disaster risk reduction and to reduce food livelihoods mortality and reduction protect livelihoods consumption gaps, total collapse of and reduce acute livelihoods malnutrition More thatn 80% Based on the IPC Based on the IPC Based on the IPC Based on the IPC Food consumption and of households in Household Group Household Group Household Group Household Group livelihood change the area are able Reference Table, Reference Table, Reference Table, Reference Table, Area outcomes (directly measured or inferred) to meet basic food at least 20% of the at least 20% of the at least 20% of the at least 20% of the needs without households in the households in the households in the households in the engaging in atypical area are in Phase 2 area are in Phase 3 or area are in Phase 4 area are in Phase 5 strategies to access or worse worse or worse food and income, and livelihoods are sustainable Acute malnutrition: Acute malnutrition: Acute malnutrition: Acute malnutrition: Acute malnutrition: <5% 5–10%, 10–15% OR > usual 15–30%; OR > usual >30% Nutritional BMI <18.5 BMI <18.5 and increasing and increasing BMI <18.5 status* prevalence: <10% prevalence: 10–20% BMI <18.5 BMI <18.5 prevalence: far prevalence: 20–40%, prevalence: >40% >40% 1.5 x greater than reference CDR: <0.5/10,000/ CDR: <0.5/10,000/ CDR: 0.5–1/10,000/ CDR: 1–2/10,000/ CDR: >2/10,000/day Mortality* day day day day OR >2x USDR: >4/10,000/ USDR: ≤1/10,000 USDR: ≤1/10,000/ USDR: 1–2/10,000/day reference day day day USDR: 2–4/10,000/ day Source: Integrated Food Security Phase Classification, Technical Manual v 2.0. *For both nutrition and mortality area outcomes, household food consumption deficits must be an explanatory factor in order for that evidence to be used in support of a Phase classification. For example, elevated malnutrition due to disease outbreak or lack of health access—if it is determined to not be related to food consumption deficits—should not be used as evidence for an IPC classification. Similarly, excess mortality rates due to, murder or conflict—if they are not related to food consumption deficits—should not be used as evidence for a Phase classification. For Acute Malnutrition, the IPC thresholds are based on percent of children under 5 years that are below two standard deviations of weight for height or presence of oedema. BMI is an acronym for Body Mass Index. CDR is Crude Death Rate. U5DR is Under 5 Death Rate. 154  Somali Poverty and Vulnerability Assessment FIGURE C.2  n  Humanitarian Response 2017, beneficiaries targeted and reached 3,500,000 3,000,000 2,500,000 2,000,000 1,500,000 1,000,000 500,000 0 17 Ap 7 M 7 Ju 7 Ju 7 Au 7 Se 7 O 7 N 7 D 7 Ja 7 Fe 8 18 -1 r-1 -1 1 l-1 1 1 -1 -1 -1 1 b- n- g- p- n- b- ar ay ct ov ec Fe M Persons targeted Persons reached Source: Food Security Cluster (2017); Food Security Cluster (2018). FIGURE C.3  n  Outbreak of communicable diseases 2017, all regions 100,000 2,000 90,000 1,800 80,000 1,600 70,000 1,400 Suspected cases 60,000 1,200 Deaths 50,000 1,000 40,000 800 30,000 600 20,000 400 10,000 200 0 0 17 17 7 7 7 17 7 17 17 7 7 7 -1 r-1 -1 l-1 -1 -1 -1 n- b- n- g- p- ar ay ct ov ec Ju Ap Ja Fe Ju Au Se O M M N D AWD/cholera Measles cases AWD/cholera deaths Source: Health Cluster Somalia (2017). Estimating the Drought Impact with a Difference-in-Differences Model 155 TABLE C.3  n  Difference-in-differences results, consumption and poverty, full sample Outcome variable Consumption Poverty Urban + Urban + Population rural Urban Rural rural Urban Rural Post –0.106** –0.086* –0.213*** 0.170*** 0.257*** 0.429*** (0.047) (0.047) (0.066) (0.056) (0.077) (0.072) Drought intensity –0.046 –0.024 0.097*** 0.028 0.032 –0.134*** (0.028) (0.030) (0.029) (0.043) (0.056) (0.050) DD estimator 0.005 0.005 –0.189** 0.006 0.007 0.238*** (0.037) (0.034) (0.088) (0.049) (0.056) (0.088) Average NDVI –1.044*** –0.490 –1.922 1.617*** 0.941* 0.614 (0.356) (0.301) (1.438) (0.496) (0.504) (1.339) Price level –0.192 –0.410** 0.375 0.572*** 0.475* 0.411   (0.165) (0.164) (0.399) (0.167) (0.245) (0.386) Regional controls       NE–urban 0.116 0.285***   –0.569*** (0.078) (0.068)   (0.105) NW–urban –0.022 0.129**   –0.260** (0.072) (0.062)   (0.108) NE–rural –0.219***     (0.062)     NW–rural –0.146*   0.091 0.352*** (0.077)   (0.068) (0.077) Central–urban 0.232*** 0.311***   –0.609*** (0.073) (0.076)   (0.120) Central–rural 0.188   0.800*** –0.229 (0.190)   (0.201) (0.172) Jubbaland–urban 0.521*** 0.473***   –1.162*** (0.098) (0.085)   (0.163) SW–urban 0.403*** 0.295***   –0.539*** (0.099) (0.089)   (0.155) SW–rural 0.248**   1.191*** –0.550** (0.112)   (0.355) (0.271) Household controls             HH head literacy 0.046*** 0.066*** 0.013 –0.049 –0.061* –0.039 (0.016) (0.015) (0.030) (0.030) (0.032) (0.055) HH head age 0.001** 0.001 0.002** –0.001 –0.000 –0.003** (0.000) (0.000) (0.001) (0.001) (0.001) (0.002) Received remittances 0.068*** 0.075*** 0.031 –0.141*** –0.145*** –0.125 (0.014) (0.014) (0.032) (0.021) (0.020) (0.081) Household size –0.058*** –0.056*** –0.056*** 0.082*** 0.084*** 0.068*** (0.003) (0.003) (0.008) (0.005) (0.005) (0.017) Gender composition 0.032 –0.003 0.104* –0.073 –0.029 –0.205** (0.032) (0.032) (0.054) (0.057) (0.062) (0.102) 156  Somali Poverty and Vulnerability Assessment Outcome variable Consumption Poverty Urban + Urban + Population rural Urban Rural rural Urban Rural Dwelling controls             Dwelling tenure: Rent 0.014 0.010 0.028 –0.037 –0.044 0.032   (0.014) (0.016) (0.025) (0.025) (0.028) (0.051) Dwelling tenure: Other –0.043 –0.077** 0.047 0.099** 0.165*** –0.014   (0.028) (0.031) (0.052) (0.046) (0.054) (0.078) Dwelling floor: Tiles or mud –0.005 0.026* –0.151*** –0.016 –0.055* 0.229***   (0.016) (0.015) (0.043) (0.027) (0.029) (0.071) Dwelling floor: Other –0.063*** –0.062*** –0.160*** 0.044 0.064 0.222***   (0.023) (0.024) (0.038) (0.036) (0.040) (0.081) Dwelling type: Separate 0.023 0.024 –0.081 –0.037 –0.038 –0.024   (0.025) (0.021) (0.049) (0.039) (0.039) (0.087) Dwelling type: Other 0.021 –0.002 0.059 –0.038 –0.027 –0.080   (0.022) (0.017) (0.040) (0.031) (0.030) (0.087) Dwelling roof: Tiles 0.014 –0.067 0.530*** 0.104* 0.170** –0.241***   (0.062) (0.042) (0.120) (0.062) (0.074) (0.087) Dwelling roof: Harar –0.048 –0.114*** 0.024 0.073 0.217*** –0.052   (0.030) (0.029) (0.062) (0.051) (0.059) (0.078) Dwelling roof: Raar –0.217** –0.283*** –0.172 0.215 0.414** 0.086   (0.085) (0.083) (0.123) (0.131) (0.192) (0.181) Dwelling roof: Wood –0.035 –0.071** –0.012 0.095* 0.099 0.189*   (0.032) (0.030) (0.057) (0.052) (0.061) (0.105) Dwelling roof: Plastic –0.075** –0.163*** –0.053 0.034 0.301*** –0.085   (0.035) (0.043) (0.066) (0.078) (0.078) (0.075) Dwelling roof: Concrete 0.028 0.058 –0.062 0.063 0.070 0.085   (0.054) (0.072) (0.090) (0.091) (0.111) (0.104) Dwelling roof: Other –0.125 –0.104 –0.251** 0.117 0.072 0.320*   (0.078) (0.093) (0.123) (0.072) (0.093) (0.165) Improved sanitation 0.020 0.026 0.054 –0.058* –0.086*** –0.037   (0.025) (0.030) (0.034) (0.034) (0.033) (0.070) Conflict controls       Conflict fatalities in district –0.000 –0.000 –0.000* –0.000 –0.000* 0.001*** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Conflict x drought 0.000 0.000 –0.000* 0.000 –0.000 0.001** (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) Assistance             Assistance (% of beneficiaries –0.320*** –0.362*** –0.312*** 0.559*** 0.615*** 0.418*** reached)  (0.048) (0.045) (0.087) (0.077) (0.084) (0.129) Observations 7,214 5,678 1,536 7,214 5,678 1,536 R–squared 0.348 0.347 0.520       Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Standard errors in parentheses. Poverty status results estimated using Probit, Consumption results estimated using OLS. Drought effect expressed in standard deviations of NDVI loss. Estimating the Drought Impact with a Difference-in-Differences Model 157 FIGURE C.4  n  Drought effect along the income distribution, urban areas 25% Drought impact on consumption 20% 15% 10% 5% 0% –5% –10% –15% 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Income percentile Drought effect Smoothed drought effect 95% confidence interval Source: Authors’ calculation based on the SHFS 2017–18. TABLE C.4  n  Robustness of results across various specifications (I) (II) (III) (IV) (V) (VI) Sample Full rural sample Outcome variable Poor PPP Drought impact 0.192*** 0.218** 0.187** 0.176** 0.226** 0.238*** S.E. (0.0629) (0.0858) (0.0803) (0.0822) (0.0913) (0.0880) Outcome variable ln (core consumption) Drought impact –0.107** –0.192*** –0.152** –0.143** –0.173* –0.189** S.E. (0.0428) (0.0715) (0.0677) (0.0696) (0.0885) (0.0876) Controls             Regional No Yes Yes Yes Yes Yes Household No No Yes Yes Yes Yes Dwelling No No No Yes Yes Yes Conflict No No No No Yes Yes Assistance No No No No No Yes Observations 1,591 1,591 1,563 1,536 1,536 1,536 R-squared 0.032 0.250 0.370 0.485 0.501 0.520 Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Poverty status results estimated using Probit, Consumption results estimated using OLS. Drought effect expressed in standard deviations of NDVI loss. 158  Somali Poverty and Vulnerability Assessment TABLE C.5  n  Difference-in-differences results with restricted sample Sample Rural, NE excluded Rural, Central excluded Rural, SW excluded Outcome variable Poverty Drought impact 0.197*** 0.213** 0.137** 0.115 0.224*** 0.415*** S.E. (0.066) (0.088) (0.038) (0.052) (0.059) (0.087) Outcome variable ln (core consumption) Drought impact –0.129*** –0.191** –0.051 –0.048 –0.128*** –0.296*** S.E. (0.048) (0.093) (0.038) (0.052) (0.043) (0.080) Controls No Yes No Yes No Yes Observations 1,511 1,456 1,087 1,035 1,319 1,277 R-squared 0.054 0.508 0.029 0.515 0.065 0.564 Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Poverty status results estimated using Probit, Consumption results estimated using OLS. Drought effect expressed in standard deviations of NDVI loss. TABLE C.6  n  Difference-in-differences results, consumption and poverty, overlapping sample Outcome variable Consumption Poverty Sample Urban + rural Urban Rural Urban + rural Urban Rural Post –0.193*** –0.221*** –0.229*** 0.323*** 0.321** 0.515*** (0.058) (0.062) (0.076) (0.108) (0.151) (0.092) Drought intensity –0.036 –0.067** 0.055 0.002 0.070 –0.114 (0.026) (0.032) (0.045) (0.036) (0.062) (0.090) DD estimator –0.042 –0.042 –0.117** 0.127 0.050 0.223* (0.036) (0.043) (0.052) (0.079) (0.096) (0.112) Average NDVI 0.016 0.082 1.155 1.559 1.468 –0.783   (0.536) (0.536) (1.412) (1.094) (1.200) (2.448) Price level –0.127 –0.108 –0.146 0.585*** 0.360 0.963***   (0.220) (0.192) (0.275) (0.219) (0.387) (0.331) Regional controls       NW–urban 0.163** 0.136*   –0.178 (0.066) (0.070)   (0.142) NE–rural             NW–rural 0.079     (0.073)     Household controls             HH head literacy 0.077*** 0.081*** 0.044 –0.090*** –0.079** –0.141*** (0.014) (0.014) (0.049) (0.029) (0.031) (0.038) HH head age 0.001 0.000 0.003* –0.001 0.000 –0.005*** (0.001) (0.001) (0.002) (0.001) (0.001) (0.001) Received remittances 0.074*** 0.076*** –0.026 –0.143*** –0.146*** –0.115** (0.014) (0.016) (0.043) (0.024) (0.026) (0.056) —continued Estimating the Drought Impact with a Difference-in-Differences Model 159 TABLE C.6—continued Outcome variable Consumption Poverty Sample Urban + rural Urban Rural Urban + rural Urban Rural Household size –0.058*** –0.056*** –0.078*** 0.086*** 0.080*** 0.133*** (0.003) (0.003) (0.010) (0.005) (0.005) (0.010) Gender composition 0.010 0.000 0.042 –0.048 –0.021 –0.203*** (0.033) (0.036) (0.073) (0.065) (0.069) (0.069) Dwelling controls             Dwelling tenure: Rent 0.002 0.000 0.038 –0.019 –0.022 0.038   (0.014) (0.015) (0.031) (0.023) (0.023) (0.049) Dwelling tenure: Other –0.048 –0.066* 0.202*** 0.063 0.115** –0.271**   (0.035) (0.036) (0.071) (0.052) (0.054) (0.115) Dwelling floor: Tiles or mud 0.011 0.043** –0.296*** –0.018 –0.055* 0.336***   (0.019) (0.018) (0.065) (0.032) (0.033) (0.077) Dwelling floor: Other –0.064*** –0.082*** –0.235*** 0.085** 0.146*** 0.224**   (0.024) (0.026) (0.071) (0.039) (0.042) (0.092) Dwelling type: Separate 0.010 0.015 –0.021 –0.020 –0.025 –0.022   (0.020) (0.020) (0.072) (0.040) (0.041) (0.074) Dwelling type: Other –0.007 –0.011 0.058 –0.030 –0.032 0.013   (0.019) (0.019) (0.079) (0.028) (0.029) (0.077) Dwelling roof: Tiles –0.062 –0.079 0.382*** 0.162** 0.196** –0.249   (0.050) (0.051) (0.099) (0.069) (0.080) (0.156) Dwelling roof: Harar –0.046 –0.114*** 0.165* 0.114** 0.188*** –0.089   (0.037) (0.032) (0.092) (0.052) (0.061) (0.066) Dwelling roof: Raar –0.233 –0.365*** 0.042 0.308** 0.737*** –0.041   (0.147) (0.067) (0.165) (0.146) (0.092) (0.154) Dwelling roof: Wood –0.084** –0.074** –0.036 0.147*** 0.124* 0.246**   (0.034) (0.031) (0.100) (0.052) (0.063) (0.119) Dwelling roof: Plastic –0.135*** –0.179*** 0.109 0.213*** 0.289*** –0.119   (0.050) (0.044) (0.089) (0.060) (0.079) (0.100) Dwelling roof: Concrete 0.040 0.095 –0.025 0.062 0.032 0.011   (0.062) (0.061) (0.113) (0.071) (0.081) (0.169) Dwelling roof: Other –0.112 –0.091 –0.031 0.115 0.061 0.034   (0.077) (0.093) (0.115) (0.077) (0.094) (0.135) Improved sanitation 0.025 0.017 0.056 –0.054 –0.054 –0.035   (0.027) (0.034) (0.053) (0.035) (0.035) (0.066) Conflict fatalities in district 0.000 0.000 –0.014 –0.000** –0.000* 0.014 (0.000) (0.000) (0.010) (0.000) (0.000) (0.013) Conflict x drought 0.000* 0.000* –0.002 –0.000 –0.000 0.010 (0.000) (0.000) (0.008) (0.000) (0.000) (0.014) Assistance (% of –0.210*** –0.232*** –0.126 0.391*** 0.448*** 0.077 beneficiaries reached) (0.046) (0.041) (0.092) (0.088) (0.108) (0.134) Observations 4,044 3,348 696 4,044 3,348 696 R-squared 0.332 0.349 0.474     Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Standard errors in parentheses. Poverty status results estimated using Probit, Consumption results estimated using OLS. Drought effect expressed in standard deviations of NDVI loss. TABLE C.7  n  Difference-in-differences results, hunger All regions Overlapping regions Outcome variable Hunger Sample Urban + rural Urban Rural Urban + rural Urban Rural Post 0.100* 0.131** 0.115 0.117*** 0.123*** –0.010 (0.057) (0.057) (0.129) (0.033) (0.033) (0.060) Drought intensity –0.063 –0.089* –0.050 –0.085*** –0.118*** –0.038 (0.038) (0.047) (0.059) (0.032) (0.044) (0.030) DD estimator 0.101** 0.096* 0.166** 0.160*** 0.116*** 0.591*** (0.044) (0.053) (0.079) (0.038) (0.038) (0.129) Average NDVI 0.360 0.512 –2.031 1.748* 1.987** 1.022 (0.524) (0.545) (1.410) (0.901) (0.861) (0.805) Regional controls NE–urban –0.030 (0.083) NW–urban –0.225*** –0.098 (0.067) (0.084) NE–rural NW–rural     –0.336**   (0.135) Central–urban –0.015 (0.095)  Central–rural   0.329     (0.233) Jubbaland–urban –0.127   (0.176) SW–urban –0.149   (0.128) SW–rural   0.213     (0.352) Household controls HH head literacy –0.051** –0.032 –0.119** –0.027 –0.017 –0.137***   (0.024) (0.026) (0.058) (0.026) (0.027) (0.038) HH head age –0.001 –0.001* 0.002 –0.001 –0.001 0.000   (0.001) (0.001) (0.002) (0.001) (0.001) (0.002) Received remittances –0.002 –0.033 0.170*** –0.020 –0.033 –0.015   (0.024) (0.025) (0.044) (0.024) (0.025) (0.021) Household size –0.006 –0.001 –0.021 –0.012** –0.009 –0.007   (0.006) (0.005) (0.017) (0.005) (0.005) (0.009) Gender composition –0.003 0.024 –0.019 –0.007 0.012 0.010   (0.051) (0.051) (0.130) (0.048) (0.054) (0.033) —continued Estimating the Drought Impact with a Difference-in-Differences Model 161 TABLE C.7—continued All regions Overlapping regions Outcome variable Hunger Sample Urban + rural Urban Rural Urban + rural Urban Rural Dwelling controls Dwelling tenure: Rent 0.029 0.018 0.074 0.004 0.013 –0.101***   (0.022) (0.020) (0.065) (0.020) (0.019) (0.028) Dwelling tenure: Other 0.212*** 0.116* 0.246** 0.153* 0.109* 0.024   (0.070) (0.060) (0.112) (0.079) (0.063) (0.079) Dwelling floor: Tiles or –0.010 –0.016 0.036 –0.010 –0.006 0.106** mud   (0.031) (0.030) (0.082) (0.028) (0.030) (0.042) Dwelling floor: Other 0.003 0.051 –0.027 0.058 0.053 0.124**   (0.041) (0.038) (0.087) (0.042) (0.043) (0.051) Dwelling type: Separate –0.068 –0.087* –0.063 –0.056 –0.085** 0.156**   (0.054) (0.050) (0.100) (0.034) (0.036) (0.066) Dwelling type: Other –0.036 –0.026 –0.130* –0.036 –0.030 0.065   (0.044) (0.037) (0.071) (0.031) (0.028) (0.054) Dwelling roof: Tiles 0.036 –0.001 0.296** –0.143*** –0.216**   (0.126) (0.125) (0.114) (0.034) (0.089) Dwelling roof: Harar 0.130** 0.174*** 0.140* 0.065 0.093 0.04   (0.058) (0.060) (0.076) (0.057) (0.061) (0.049) Dwelling roof: Raar 0.070 0.099 0.120* –0.052 0.006 0.056   (0.068) (0.077) (0.067) (0.046) (0.072) (0.064) Dwelling roof: Wood –0.059 –0.042 –0.109 –0.077* –0.097 –0.007   (0.064) (0.079) (0.146) (0.046) (0.072) (0.055) Dwelling roof: Plastic 0.091 0.076 0.124 –0.004 0.048 –0.052   (0.063) (0.092) (0.075) (0.065) (0.086) (0.058) Dwelling roof: Concrete –0.053 –0.033 –0.228   (0.104) (0.119) (0.148) Dwelling roof: Other 0.075 –0.000 0.161 –0.010 –0.019 –0.020   (0.077) (0.083) (0.109) (0.060) (0.074) (0.078) Improved sanitation –0.002 0.015 –0.043 –0.031 –0.045 0.013   (0.039) (0.043) (0.052) (0.030) (0.031) (0.035) Conflict fatalities in 0.000 0.000 0.001*** district   (0.000) (0.000) (0.000) Conflict x drought –0.000 –0.000 –0.000   (0.000) (0.000) (0.000) Assistance (% of –0.052 –0.191** 0.198 0.078 –0.010 0.039 beneficiaries reached) (0.094) (0.122) (0.068) (0.099) (0.055) Observations 7,153 5,637 1,516 3,962 3,292 663 Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Standard errors in parentheses. Results estimated with Probit. Drought effect expressed in standard deviations of NDVI loss. 162  Somali Poverty and Vulnerability Assessment TABLE C.8  n  Difference-in-differences results, food consumption All regions Overlapping regions Outcome variable Food consumption Sample Urban + rural Urban Rural Urban + rural Urban Rural Post –0.061 –0.015 –0.200*** –0.088* –0.095* –0.243*** (0.041) (0.052) (0.062) (0.047) (0.057) (0.059) Drought intensity –0.025 0.013 0.086*** –0.012 –0.060* 0.058 (0.023) (0.034) (0.029) (0.022) (0.032) (0.041) DD estimator –0.015 –0.039 –0.164** 0.000 0.064 –0.124*** (0.031) (0.041) (0.071) (0.029) (0.042) (0.046) Average NDVI –0.909** –0.569* –1.371 –0.177 –0.033 1.357   (0.354) (0.338) (1.326) (0.527) (0.583) (1.010) Price level 0.034 –0.293 0.501 0.164 0.394** 0.113   (0.160) (0.265) (0.326) (0.160) (0.181) (0.224) Regional controls NE–urban –0.003 0.155* (0.080) (0.089) NW–urban –0.036 0.125*   0.071 –0.032 (0.062) (0.075)   (0.055) (0.060) NE–rural –0.440*** (0.055) NW–rural –0.074 0.345*** 0.066 (0.065) (0.058) (0.065) Central–urban 0.201*** 0.311*** (0.075) (0.104) Central–rural 0.210 0.918*** (0.147) (0.177) Jubbaland–urban 0.393*** 0.375*** (0.099) (0.101) SW–urban 0.293*** 0.260** (0.092) (0.108) SW–rural 0.240** 1.204*** (0.102) (0.331) Household controls HH head literacy 0.032** 0.054*** –0.001 0.054*** 0.058*** 0.031 (0.013) (0.011) (0.024) (0.013) (0.014) (0.042) HH head age 0.001 0.000 0.002** 0.001 0.000 0.003** (0.000) (0.000) (0.001) (0.001) (0.001) (0.001) Received remittances 0.038*** 0.046*** 0.002 0.049*** 0.056*** –0.047* (0.014) (0.016) (0.026) (0.016) (0.017) (0.028) Household size –0.048*** –0.046*** –0.049*** –0.047*** –0.044*** –0.071*** (0.002) (0.003) (0.007) (0.003) (0.002) (0.009) Gender composition 0.006 –0.035 0.110* –0.014 –0.031 0.051 (0.032) (0.028) (0.058) (0.029) (0.032) (0.062) —continued TABLE C.8—continued All regions Overlapping regions Outcome variable Food consumption Sample Urban + rural Urban Rural Urban + rural Urban Rural Dwelling controls Dwelling tenure: Rent 0.010 0.004 0.015 0.010 0.009 0.009   (0.012) (0.013) (0.023) (0.015) (0.016) (0.028) Dwelling tenure: Other –0.043* –0.060** 0.023 –0.022 –0.038 0.187***   (0.024) (0.028) (0.041) (0.031) (0.030) (0.068) Dwelling floor: Tiles or mud –0.003 0.019 –0.114*** –0.001 0.018 –0.223***   (0.016) (0.016) (0.042) (0.018) (0.018) (0.057) Dwelling floor: Other –0.033* –0.017 –0.128*** –0.048** –0.054** –0.218***   (0.020) (0.023) (0.038) (0.023) (0.026) (0.060) Dwelling type: Separate 0.015 0.015 –0.091** –0.011 –0.002 –0.055   (0.019) (0.017) (0.037) (0.018) (0.018) (0.060) Dwelling type: Other 0.019 0.002 0.019 –0.012 –0.012 –0.010   (0.018) (0.016) (0.030) (0.017) (0.017) (0.060) Dwelling roof: Tiles 0.045 –0.006 0.295*** –0.018 –0.035 0.364***   (0.046) (0.037) (0.073) (0.043) (0.043) (0.094) Dwelling roof: Harar –0.033 –0.077*** 0.015 –0.026 –0.081*** 0.144*   (0.026) (0.028) (0.055) (0.033) (0.029) (0.080) Dwelling roof: Raar –0.168*** –0.204*** –0.126 –0.158 –0.268*** 0.038   (0.064) (0.054) (0.093) (0.106) (0.044) (0.130) Dwelling roof: Wood 0.026 0.011 0.024 –0.008 0.004 –0.003   (0.028) (0.031) (0.050) (0.033) (0.033) (0.090) Dwelling roof: Plastic –0.027 –0.097** –0.016 –0.082* –0.116*** 0.090   (0.030) (0.038) (0.052) (0.043) (0.038) (0.077) Dwelling roof: Concrete 0.064** 0.097*** –0.062 0.046 0.090*** –0.065   (0.027) (0.035) (0.075) (0.041) (0.028) (0.095) Dwelling roof: Other –0.068 –0.063 –0.161 –0.053 –0.047 0.028   (0.073) (0.090) (0.111) (0.072) (0.087) (0.107) Improved sanitation 0.003 –0.000 0.039 –0.022 –0.025 0.013   (0.027) (0.036) (0.032) (0.033) (0.040) (0.046) Conflict fatalities in district –0.000** –0.000** –0.000** –0.000** –0.000*** –0.013 (0.000) (0.000) (0.000) (0.000) (0.000) (0.008) Conflict x drought 0.000 0.000 –0.000* 0.000*** 0.000*** –0.002 (0.000) (0.000) (0.000) (0.000) (0.000) (0.007) Assistance (% of –0.197*** –0.232*** –0.240*** –0.186*** –0.174*** –0.078 beneficiaries reached)  (0.035) (0.041) (0.081) (0.035) (0.034) (0.081) Observations 7,214 5,678 1,536 4,044 3,348 696 R–squared 0.347 0.304 0.591 0.297 0.312 0.461 Source: Authors’ calculation based on the SHFS 2017–18. Note: ***p<0.01, **p<0.05, *p<0.1. Standard errors in parentheses. Results estimated with OLS. Drought effect expressed in standard deviations of NDVI loss. 164  Somali Poverty and Vulnerability Assessment APPENDIX D Regression Results for Each Type of Shock TABLE D.1  n  What household characteristics affect the probability of reporting shocks? Crop or High Other Water livestock food Income Drought natural shortage loss prices reduced Theft Conflict Wealth index –0.080*** –0.006 0.005 –0.028*** –0.005 –0.008 –0.009* –0.002 [0.021] [0.005] [0.005] [0.005] [0.014] [0.011] [0.005] [0.006] Head (no education) 0.041*** –0.007 –0.008 0.014** 0.001 0.002 –0.004 –0.003 [0.013] [0.007] [0.006] [0.006] [0.011] [0.008] [0.003] [0.005] HH with employed member 0.031* 0.000 –0.007 –0.006 0.034** 0.004 0.026** 0.005 [0.019] [0.010] [0.008] [0.011] [0.015] [0.012] [0.006] [0.007] HH has agricultural income 0.024 0.021** 0.046*** 0.033** –0.051*** –0.021 0.018* –0.027** [0.016] [0.008] [0.009] [0.013] [0.011] [0.015] [0.010] [0.013] Male headed HH 0.021 0.015*** –0.013*** –0.013 –0.014 0.003 –0.006 –0.007 [0.023] [0.004] [0.004] [0.016] [0.020] [0.010] [0.009] [0.004] HH head age 0.001 –0.000 0.000 –0.001*** 0.001* 0.000 –0.000 0.000 [0.001] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] [0.000] Household size –0.008 0.001 –0.002 0.006*** 0.003** 0.007*** 0.003*** 0.001 [0.005] [0.002] [0.004] [0.002] [0.002] [0.001] [0.001] [0.001] HH receives assistance –0.017 0.027** 0.030*** 0.018* 0.017 0.028** –0.001 0.001 [0.028] [0.012] [0.009] [0.010] [0.012] [0.013] [0.002] [0.010] HH receives remittances 0.011 0.005 0.010 0.015 0.021 0.018*** –0.005 0.010*** [0.035] [0.008] [0.008] [0.013] [0.019] [0.007] [0.012] [0.003] Household welfare Bottom 40% [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] Top 60% 0.021 0.025*** 0.0009 0.014 0.020 0.014** 0.010*** 0.003 [0.019] [0.005] [0.006] [0.015] [0.016] [0.006] [0.003] [0.006] Population type Urban [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] [Ref] Rural 0.143*** –0.016* 0.015 0.004 –0.036** –0.022** –0.019*** –0.006 [0.023] [0.009] [0.015] [0.010] [0.018] [0.009] [0.004] [0.009] IDP (settlement) 0.014 –0.002 0.050** 0.032 0.034 –0.020 — 0.063*** [0.078] [0.020] [0.022] [0.019] [0.030] [0.021] [0.024] Nomad 0.256*** — 0.026 0.010 –0.057*** –0.047*** — 0.002 [0.078] [0.016] [0.013] [0.011] [0.007] [0.022] Control for region Yes Yes Yes Yes Yes Yes Yes Yes Predicted probability 0.33 0.03 0.04 0.05 0.06 0.04 0.02 0.02 No. of observations 3,170 2,516 2,974 3,032 3,134 2,954 1,713 2,570 Pseudo R 2 0.26 0.12 0.18 0.17 0.11 0.13 0.11 0.14 Source: Authors’ calculation based on the SHFS 2017–18. Note: Significance level: 1% (***), 5% (**), and 10% (*). Regression Results for Each Type of Shock 165 APPENDIX E Methodology for Reduced Coping Strategy Index The Coping Strategies Index records the frequency of times a household has adopted a certain behavior in the past seven days and then assigns each behavior a certain weight. The list of questions is given below. TABLE E.1  n  Reduced Coping Strategy Index In the past seven days, if there have been times when you did not have enough food or money to buy food, how often has your household had to: Weight Rely on less preferred or less expensive foods? 1 Borrow food, or rely on help from a friend or relative? 2 Limit portion size at mealtimes? 1 Restrict consumption by adults in order for small children to eat? 3 Reduce number of meals eaten in a day? 1 Source: Maxwell and Caldwell (2008). Methodology for Reduced Coping Strategy Index 167 APPENDIX F Displacement Along with the data on IDPs and host communities TABLE F.1  n  Camps with Somali refugees in the SPS used from the Somali SHFS 2017–18, the displace- 2017 sampling frame ment chapter draws on data from the Skills Pro- file Survey, conducted in refugee camps and host Region in Nationality of communities in Ethiopia, in 2017. Ethiopia refugees Camp Ken-Borena Skills Profile Survey, Ethiopia Kebribeyah Aw-barre Sample design Sheder Outcomes of Somali refugees in Ethiopia are ana- Somali Somali Bokolmanyo lyzed using the Skills Profile Survey (SPS). The SPS is a household survey administered in and Melkadida around refugee camps in Ethiopia in 2017. It sur- Kobe veyed South Sudanese, Somali, Eritrean, and Suda- nese refugees in Ethiopia, and the Ethiopian host Hilaweyn communities located close to the refugee camps. Buramino About 33 percent of refugee households215 in Ethi- opia are outside camps, and are primarily Eritrean. Source: UNHCR. These households were excluded from the sam- pling frame due to feasibility constraints. The SPS is therefore only representative of refugees all the camps in the sample frame were selected living in camps. The list of refugee camps, sites, in the sample and were surveyed. Within camps, and locations provided by UNHCR-Ethiopia as of EAs were selected using equal probability to make January 2017 was used as the sample frame (Table up the required number of EAs for that camp. In F.1). Four strata were drawn based on four regions total, 82 enumeration areas were selected from Tigray Afar (primarily Eritrean refugees), Gambella each stratum. All the households in the selected (primarily South Sudanese), Benishangul Gumuz EAs were listed, and 12 households were randomly (primarily Sudanese, with a quarter of South Suda- selected and surveyed per enumeration area mak- nese), and Somali (primarily Somali). Somali refu- ing up to a total of 900 refugee households per gees mostly populate the Somali region in Ethiopia stratum. (Table F.2). Since each region hosts a majority of one refugee nationality, the stratification is implic- Households within a 5-km radius of a camp were itly based on refugee nationality (Table F.3). classified as host community households. Areas within a 5-km radius of camps were divided into The sample design uses a multi-stage stratified EAs of 300 by 300 meters using GIS technol- random sample. Camps in each stratum were ogy. Of these, EAs marked as residential by Open divided into Enumeration Areas (EAs) of 150 by Street Maps were included in the sample frame. 150 meters using GIS technology. The number of EAs within a stratum were then selected using EAs to be selected from each camp was obtained proportional probability sampling with the prob- proportional to the size of the camp. In this way, ability of selection of an EA equal to the area of the Enumeration Area outside the camp. In total, 42 EAs were selected for each stratum. Like EAs within camps, all the households in the EAs Household is here defined as all people living in the same 215  selected for host community sampling were listed, dwelling and sharing all meals and finances. Displacement 169 TABLE F.2  n  Number of refugee and host community households interviewed by stratum Stratum Tigray Afar Gambella Benishangul Gumuz Somali Total Refugees 894 439 1,423 871 3,627 Host community 412 0 975 303 1,690 Source: Authors’ calculations based on the SPS 2017. TABLE F.3  n  Sampled population by country of nationality Number of households Percentage of households Nationality surveyed in surveyed population South Sudanese 837 16% Somali 871 16% Eritrean 893 17% Sudanese 1,016 19% Ethiopian (host community) 1,690 32% Other country 10 0% Total 5,317 100% Source: Authors’ calculations based on the SPS 2017. and 12 households were selected randomly and radius of refugee camps. The selection probabil- surveyed per EA making up to a total of 500 host ity P for a household can be decomposed into the community households per stratum. selection probability P1 of the EA and the selection probability P2 of the household within the EA: Conflict in Oromia and Somali regions necessi- tated sampling modifications. In early September P = P1P2 2017, violent conflict in Oromia and Somali regions escalated, rendering some of the camps in Somali As refugee population in the different strata lived stratum inaccessible. The enumeration areas of in different camps, the selection probability P1 of the Jijiga subregion were replaced by enumera- an EA k is calculated as the number of households tion areas in nonviolent areas of Somali stratum. within the EA divided by the number of house- Also, as most refugee camps are in remote areas holds within the stratum multiplied by the number with sparse host population, the final number of of selected EAs in the stratum host households surveyed fell short of the original intended sample of 500 host households per stra- nk * K ˆ tum. However, despite the changes in sampling, the P1 = N survey captured roughly similar number of refugee households of the four main refugee nationalities. where ˆ nk denotes the number of households in EA k (obtained by multiplying the percentage of camp area covered by the EA with the number of Weights households in the camp as information on num- Sampling weights are applied to survey obser- ber of households in an EA was not available prior vations to make them representative of refugee to listing), K is the number of EAs selected in the populations in different regions and of the overall corresponding stratum and N is the total number camp-refugee population in Ethiopia. Weights for of households in the stratum. For host community host populations are constructed to be represen- sampling, as information on number of host house- tative of the host households living within a 5-km holds living within 5 km of camps in a stratum was 170  Somali Poverty and Vulnerability Assessment not available, the selection probability of an EA for P2 = |H| host sampling is calculated as the number of EAs nk selected divided by the total number of EAs in the stratum. where |H| is the number of households selected in the EA and nk denoting the number of listed K households in EA k. Usually the number of house- P1 = holds per EA is 12, while a few exceptions exist due T to invalid interviews. Where K is the number of EAs selected in a stratum and T is the total number of EAs in the correspond- Sampling weights were scaled to equal the num- ing stratum. Replacement enumeration areas were ber of households per strata using the information assigned the sampling weight of the enumeration for number of households provided by UNHCR. area that they were replacing. Due to changes in There was no source of information on number of sample during fieldwork, the number of enumera- host households living within 5 km distance of the tion areas surveyed in each stratum differed from camps. The weights for host community surveys the original sample. The weights were therefore were therefore not scaled. scaled at the end to correct for the change in the value of K. The selection probability P2 for a household within an EA k is constant across households and can be expressed as Displacement 171 APPENDIX G Data Gaps While the knowledge base about the state of earnings, Somalia lacks an agricultural census to the Somali economy and living conditions has help better understand the structure of the sector, improved considerably in recent years, large gaps its production systems, and constraints to produc- remain. A data ecosystem depends heavily on the tivity. Business/establishment censuses—which ability of the government to collect and manage are invaluable tools for information on the struc- statistical and administrative data. This capacity is ture and activities of enterprises, employment, severely lacking, but it is critical to further devel- and contribution of private sector in GDP—are cur- opment as the availability of more credible and rently unavailable. reliable data sources can enable stakeholders to discuss issues based on factual information rather There is a paucity of quantitative data upon than perception. which to evaluate macroeconomic development. The statistical system in Somalia is fragmented A few new surveys are helping to mitigate the and lacks coordination, resulting in statistical data gaps. Production of data on poverty in information that is often incomparable, not nation- Somalia is improving, but regular collection of ally representative, and scattered across various household surveys is needed to continue monitor- national and international stakeholders. Somalia ing poverty and other socioeconomic indicators. lacks a harmonized, comprehensive, nationally More recently, the Somali High Frequency Surveys representative Consumer Price Index (CPI) data have provided invaluable insight into consump- series which is instrumental for poverty measure- tion patterns and poverty among Somalis. Since ment and monitoring. Often, prices are collected the Somali High Frequency Surveys are designed in geographically limited areas and for a limited to capture the core indicators within a short time- set of items, and the CPI is produced differently frame, some information that may normally be depending on the stakeholder producing the esti- captured in a multi-topic survey are not available. mates. A series of market price surveys collecting Thus, information on child anthropometry, fertil- prices across various markets nationally at regular ity, price data, time use, savings, and health need intervals and a full consumption survey are needed to be captured in other surveys. The health sector to support the generation of the CPI. A nationally faces an absence of national surveys and weak civil representative labor force survey for labor market registration and vital statistics. The planned Somali indicators is not available. Health and Demographic Survey (SHDS) will help to fill some of these data gaps. At present, many forms of administrative data are not collected, collected as a limited set of indica- Somalia lacks several censuses that are essen- tors, and/or disputed. Administrative data sources tial for planning and policy making. Somalis have are critical for compiling GDP by production endured over four decades without a population approach, yet the fragmented statistical system and housing census, which would normally provide lacks coordination and resources to compile exist- a basis for a sampling frame, provide information ing records, harmonize data collection methods, for budgeting purposes, and track demographic improve the quality of the records, and utilize the and socioeconomic changes. While the last cen- data for compiling GDP by production approach. sus was concluded in 1986, high levels of fertility, Cadaster or business registers are incomplete and migration, and mortality render any projections challenged. Basic data on trade volumes are only based on the census highly uncertain. United collected at regional levels; there is no common Nations’ efforts have shed some light on popu- approach to classification and no system for aggre- lation estimates with the Population Estimation gation into national estimates. This forces anyone Survey for Somalia (PESS) in 2014. Although agri- seeking to estimate these figures to impute them culture is critical to Somalia’s economy in terms of by piecing together other countries’ data on trade contribution to the GDP, employment, and export with Somalia. Data Gaps 173 Beyond administrative and statistical survey Finally, improving the understanding on the func- data, many biophysical datasets are missing. Par- tioning of key sectors, in particular health and ticularly important for an economy in which more education, as well as helping to build ‘fit for con- than two-thirds of income is derived from natural text’ management information systems, would capital is a better accounting for the status and be important in facilitating future reform and in value of ecosystems and the monitoring of risks defining the appropriate role of the public sector to sustainability so that issues of deforestation, in service delivery. Similarly, the understanding of flooding, overgrazing, and otherwise depleting the internal dynamics of other sectors—from the labor natural capital resources can be monitored and market to transportation—is necessary if the gov- assessed more systematically. Similarly, data on ernment is to be able to better coordinate the vari- fisheries catch and landing are missing. Improved ety of stakeholders active in the economy. External data on hydrometeorological, water availability, actors can be particularly helpful by supporting and factor market conditions would be helpful as the development of the data sources and of key would systems that could deliver that information analytical insights that will help greater transpar- to market participants in urban and rural areas. In ency about the situation on the ground and enable other areas, collection of seismological and other informed policy making. data on hydrocarbons has been contracted out to private sector actors through concession agree- ments, with proprietary clauses limiting the use of the data beyond the immediate concessionaire. 174  Somali Poverty and Vulnerability Assessment