KENYA POVERTY AND GENDER ASSESSMENT 2015/16 Reflecting on a Decade of Progress and the Road Ahead Pov rt & Equit September 16, 2018 © 2018 The World Bank 1818 H Street NW, Washington DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved This work is a product of the staff of The World Bank. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. 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. Attribution—Please cite the work as follows: “World Bank. 2018. Kenya Poverty and Gender Assessment 2015/16 - A Decade of Progress and the Challenges Ahead. © World Bank.” All 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; e-mail: pubrights@worldbank.org. Photo Credits: World Bank Design: Robert Waiharo TABLE OF CONTENTS Abbreviations...................................................................................................................................................................................................................................................................... i Executive Summary........................................................................................................................................................................................................................................................ iii Kenya made progress in reducing poverty and inequality over the past decade................................................................................................................ iii Despite progress in reducing poverty, several challenges remain................................................................................................................................................. x Women are left behind in many areas.............................................................................................................................................................................................................. xii Accelerating poverty reduction .......................................................................................................................................................................................................................... xvi 1. KENYA IN CONTEXT............................................................................................................................................................................................................................................. 1 1.1 Macroeconomic performance over the last decade .............................................................................................................................................................. 2 1.2 Fiscal policy and economic growth.................................................................................................................................................................................................... 8 1.3 A review of some policies over the last decade........................................................................................................................................................................... 11 1.4 Overview of monetary poverty.............................................................................................................................................................................................................. 13 1.5 Overview of non-monetary poverty................................................................................................................................................................................................... 19 1.6 Institutional context, elections and devolution........................................................................................................................................................................... 23 1.7 Perceptions on democracy, governance and political participation.............................................................................................................................. 27 2. THE EXTENT AND EVOLUTION OF POVERTY AND INEQUALITY IN KENYA .......................................................................................................... 31 2.1 Steady but modest progress against poverty 2005/06-2015/16 ..................................................................................................................................... 32 2.2 The incidence of progress, shared prosperity and inequality.............................................................................................................................................. 42 2.3 What explains the trends in poverty reduction? Poverty decomposition exercises............................................................................................. 49 2.4 Poverty profiles – What are characteristics of the poor in Kenya?.................................................................................................................................... 55 3. GENDER AND POVERTY.................................................................................................................................................................................................................................... 59 3.1 A profile of poverty and gender in Kenya ..................................................................................................................................................................................... 60 3.2 Gender gaps in endowments.................................................................................................................................................................................................................. 62 3.3 Gender inequality in economic opportunities............................................................................................................................................................................. 70 3.4 Voice and agency............................................................................................................................................................................................................................................ 80 4. AGRICULTURE AND RURAL POVERTY.................................................................................................................................................................................................. 83 4.1 The decline in rural poverty has been the main driver of poverty reduction nationally................................................................................... 84 4.2 Diversifying away from agriculture improves livelihoods..................................................................................................................................................... 85 4.3 Non-agricultural employment is becoming increasingly important for rural households ............................................................................. 86 4.4 Farm productivity has stagnated while commodity prices have increased.............................................................................................................. 89 4.5 Increased market participation can further reduce rural poverty .................................................................................................................................. 97 4.6 Conclusions......................................................................................................................................................................................................................................................... 100 5. URBANIZATION....................................................................................................................................................................................................................................................... 101 5.1 Urbanization and poverty ......................................................................................................................................................................................................................... 102 5.2 Diagnostic of urban poverty..................................................................................................................................................................................................................... 109 5.3 Urban labor markets...................................................................................................................................................................................................................................... 116 5.4 Urban informal settlements...................................................................................................................................................................................................................... 123 6. EDUCATION AND POVERTY......................................................................................................................................................................................................................... 127 6.1 Kenya’s education sector ........................................................................................................................................................................................................................... 128 6.2 Enrollment .......................................................................................................................................................................................................................................................... 129 6.3 Learning outcomes........................................................................................................................................................................................................................................ 136 6.4 The supply-side................................................................................................................................................................................................................................................. 139 6.5 Teacher incentives and school governance................................................................................................................................................................................... 143 6.6 Summary and policy options................................................................................................................................................................................................................... 146 7. HEALTH AND POVERTY................................................................................................................................................................................................................................149 7.1 Background.................................................................................................................................................................................................................................................... 150 7.2 Health outcomes and uptake through an equity lens ...................................................................................................................................................... 157 7.3 The supply side: Physical inputs, health professionals, and incentives..................................................................................................................... 168 7.4 Summary and policy implications................................................................................................................................................................................................... 174 8. VULNERABILITY, SHOCKS,AND SOCIAL PROTECTION .....................................................................................................................................................177 8.1 Introduction .................................................................................................................................................................................................................................................. 178 8.2 Vulnerability .................................................................................................................................................................................................................................................. 180 8.3 Shocks and coping strategies in 2005/06 and 2015/16..................................................................................................................................................... 186 8.4 The coverage and impact of social protection programs ............................................................................................................................................... 194 References........................................................................................................................................................................................................................................................................253 LIST OF FIGURES Figure 1: Kenya’s economic and poverty progress .......................................................................................................................................................................... iv Figure 2: Share of income by source for rural households ......................................................................................................................................................... vi Figure 3: Non-monetary dimensions of wellbeing........................................................................................................................................................................... viii Figure 4: Regional patterns in poverty..................................................................................................................................................................................................... xi Figure 5: Poverty and vulnerability in Kenya......................................................................................................................................................................................... xiii Figure 6. Gender gaps in Kenya.................................................................................................................................................................................................................... xv Figure 7: Socio-economic indicators of rural Kenya......................................................................................................................................................................... xvii Figure 8: Urbanization remains a challenge for poverty reduction........................................................................................................................................xviii Figure 1.1: Kenya’s GDP growth from 2005 to 2015............................................................................................................................................................................. 3 Figure 1.2: Annual GDP growth for Sub-Saharan Africa and selected countries, per year and between 2005 and 2015........................ 3 Figure 1.3: Contributions to GDP growth................................................................................................................................................................................................... 4 Figure 1.4: Agriculture and GDP growth..................................................................................................................................................................................................... 5 Figure 1.5: Productivity and economic growth ..................................................................................................................................................................................... 5 Figure 1.6: Demand-side contribution to growth between 2005 and 2015........................................................................................................................ 5 Figure 1.7: Contributions to real GDP growth.......................................................................................................................................................................................... 6 Figure 1.8: Contributions to GDP growth, regional comparison ................................................................................................................................................ 7 Figure 1.9: Contributions to real GDP per capita growth ................................................................................................................................................................ 7 Figure 1.10: Sectoral contribution to change in real GDP per capita productivity.............................................................................................................. 8 Figure 1.11: Productivity contribution to real GDP per capita growth ...................................................................................................................................... 8 Figure 1.12: Spending has consistently exceeded revenue collection ...................................................................................................................................... 10 Figure 1.13: Revenue collection has not kept up with spending pressures ........................................................................................................................... 10 Figure 1.14: The evolution of fiscal deficit..................................................................................................................................................................................................... 10 Figure 1.15: Sectoral contribution to growth in total spending .................................................................................................................................................... 11 Figure 1.16: Employment trends ....................................................................................................................................................................................................................... 13 Figure 1.17: Poverty at the US$ 1.20, 1.90, and 3.20 lines.................................................................................................................................................................... 15 Figure 1.18: Cumulative consumption distribution with shock ..................................................................................................................................................... 15 Figure 1.19: GDP sectoral growth simulation of poverty trajectory at international poverty lines, 2005 to 2015 .......................................... 16 Figure 1.20: Overall GDP growth simulation of poverty trajectory at international poverty lines, 2005 to 2015............................................. 16 Figure 1.21: Real sector growth, 2007 to 2015........................................................................................................................................................................................... 16 Figure 1.22: Share of households by sector of household head occupation, 2005 vs. 2015 ....................................................................................... 16 Figure 1.23: Consistent sectoral elasticities for poverty pass-through23.................................................................................................................................. 17 Figure 1.24: Combination of growth and redistribution needed to eradicate poverty in 2030 ................................................................................ 17 Figure 1.25: International comparison of poverty.................................................................................................................................................................................... 18 Figure 1.26: Poverty headcount against GDP per capita..................................................................................................................................................................... 18 Figure 1.27: Poverty rate against depth at international poverty line......................................................................................................................................... 18 Figure 1.28: Poverty headcount at IPL and LMIC, international comparison.......................................................................................................................... 19 Figure 1.29: Poverty gap at IPL and LMIC, international comparison.......................................................................................................................................... 19 Figure 1.30: International comparison of elasticity of poverty reduction................................................................................................................................. 19 Figure 1.31: Elasticity of poverty reduction against GDP per capita............................................................................................................................................. 19 Figure 1.32: Multi-dimensional deprivations, 2015................................................................................................................................................................................. 20 Figure 1.33: Poverty headcount against access to improved water............................................................................................................................................. 20 Figure 1.34: Poverty headcount against access to improved sanitation................................................................................................................................... 20 Figure 1.35: Poverty headcount against HDI............................................................................................................................................................................................... 21 Figure 1.36: Poverty headcount against literacy rates........................................................................................................................................................................... 21 Figure 1.37: Poverty headcount against adult educational attainment, primary................................................................................................................. 21 Figure 1.38: Poverty headcount against adult educational attainment, secondary........................................................................................................... 21 Figure 1.39: Poverty headcount against under-five mortality.......................................................................................................................................................... 22 Figure 1.40: Poverty headcount against child stunting ...................................................................................................................................................................... 22 Figure 1.41: Perception of democracy in sub-Saharan African countries................................................................................................................................. 27 Figure 1.42: Responsiveness of National Assembly members to citizens in sub-Saharan African countries...................................................... 28 Figure 1.43: Major issues for citizens in Kenya that government should address................................................................................................................ 29 Figure 1.44: Perceived involvement in corruption, 2016 (% of respondents)......................................................................................................................... 29 Figure 1.45: Political intimidation or violence during election campaigns.............................................................................................................................. 29 Figure 1.46: Expressing political views in sub-Saharan African countries................................................................................................................................. 29 Figure 2.1: Total fertility rate (women aged 15-49) .............................................................................................................................................................................. 33 Figure 2.2: Trends in absolute, food and extreme poverty, nationally and by area of residence ........................................................................... 34 Figure 2.3: Urban and rural food poverty basket comparison by rank, 2005/06 and 2015/16................................................................................. 35 Figure 2.4: Trends in absolute, food and extreme poverty by province and NEDI classifications .......................................................................... 37 Figure 2.5: Distribution of the poor by province.................................................................................................................................................................................... 39 Figure 2.6: Poverty depth and severity, nationally and by urban/rural strata...................................................................................................................... 40 Figure 2.7: Poverty depth by province and NEDI classification ................................................................................................................................................... 40 Figure 2.8: Proportion of consumption by use, nationally and area of residence ........................................................................................................... 41 Figure 2.9: Differential changes in price indices..................................................................................................................................................................................... 41 Figure 2.10: GICs nationally, by area of residence and NEDI classification .............................................................................................................................. 42 Figure 2.11: Real consumption deciles (2016 prices), nationally and by area of residence........................................................................................... 43 Figure 2.12: Response rates and consumption among urban households............................................................................................................................. 44 Figure 2.13: Annualized consumption growth, nationally, by area of residence and by province .......................................................................... 45 Figure 2.14: Annualized consumption growth compared to benchmark countries......................................................................................................... 46 Figure 2.15: Gini inequality index nationally, by area of residence and by province ........................................................................................................ 47 Figure 2.16: Gini inequality index for select African countries......................................................................................................................................................... 47 Figure 2.17: Atkinson index and P75/P25 inequality index nationally and by area of residence............................................................................... 48 Figure 2.18: Determinants of changes in poverty – Datt-Ravallion decomposition by area of residence........................................................... 50 Figure 2.19: Determinants of changes in poverty – Datt-Ravallion decomposition by province ............................................................................. 51 Figure 2.20: Contribution to poverty reduction........................................................................................................................................................................................ 51 Figure 2.21: Real sector growth 2007–2015................................................................................................................................................................................................. 52 Figure 2.22: Household size and average education level nationally, by province and NEDI classification ....................................................... 57 Figure 2.23: Access to improved sanitation, water and electricity by province, urban/rural, and NEDI/non-NEDI status ........................ 58 Figure 3.1: Male and female poverty rates by age group, 2015/6 .............................................................................................................................................. 61 Figure 3.2: Male and female poverty rates by marital status, 2015/6 ....................................................................................................................................... 61 Figure 3.3: Poverty and household demographic composition, 2015/6................................................................................................................................ 62 Figure 3.4: Regional differences in gender parity in the education sector............................................................................................................................ 63 Figure 3.5: Male and female literacy by county, 2015/6.................................................................................................................................................................... 64 Figure 3.6: Maternal mortality............................................................................................................................................................................................................................ 64 Figure 3.7: Kenya’s demographic transition............................................................................................................................................................................................... 65 Figure 3.8: Household members fetching water, 2015/6................................................................................................................................................................. 66 Figure 3.9: Kenya and comparators gender gaps in land and housing ownership......................................................................................................... 66 Figure 3.10: ICT access by sex and age, 2014, 2015/6............................................................................................................................................................................ 67 Figure 3.11: Financial inclusion, male and female population (15+), 2014.............................................................................................................................. 67 Figure 3.12: Difficulty to come up with emergency funds, male and female population (15+), 2014................................................................... 68 Figure 3.13: Financial inclusion, Kenya and regional comparison, 2014.................................................................................................................................... 68 Figure 3.14: Percent of population employed by category, 2005/6 – 2015/6 ....................................................................................................................... 71 Figure 3.15: Changes in school-to-work transition, 2005/6-2015/6 ............................................................................................................................................ 71 Figure 3.16: Male and female labor force participation, 2015/6...................................................................................................................................................... 72 Figure 3.17: Female labor force participation, Kenya and comparators .................................................................................................................................... 72 Figure 3.18: Geographic variation in male and female labor force participation, 2015/6............................................................................................... 73 Figure 3.19: Correlates of male and female labor force participation, 2015/6 ...................................................................................................................... 73 Figure 3.20: Male and female employment by broad sector, 2015/6.......................................................................................................................................... 74 Figure 3.21: Share of male/female employment by detailed sector, 2015/6.......................................................................................................................... 74 Figure 3.22: Profits of male-, female- and jointly-run household enterprises, 2015/6 ..................................................................................................... 76 Figure 3.23: Gender differences in agricultural employment vs. parcel management, 2015/6 ................................................................................. 77 Figure 3.24: Acceptance of norms that constrain women’s physical mobility....................................................................................................................... 80 Figure 3.25: Share of women (15-49) who experienced physical violence by marital status, 2014.......................................................................... 81 Figure 4.1: Rural poverty headcount and its decline by province ............................................................................................................................................. 84 Figure 4.2: Geographic distribution of the rural poor in Kenya.................................................................................................................................................... 85 Figure 4.3: Share of income from agriculture and non-agricultural sources in rural Kenya........................................................................................ 85 Figure 4.4: Changes in rural non-agricultural economic activities.............................................................................................................................................. 87 Figure 4.5: Female non-agricultural labor allocation........................................................................................................................................................................... 88 Figure 4.7: Share of income from different sources for poor and non-poor households............................................................................................. 88 Figure 4.8: Non-farm economic activity by ISIC classification....................................................................................................................................................... 89 Figure 4.9: Relationship between crop yield and poverty rates at the provincial level in rural Kenya, 2015/16............................................ 90 Figure 4.10: Poverty and crop yield at the county level in rural Kenya, 2015/16.................................................................................................................. 90 Figure 4.11: Relationship between yield decile and poverty rates in rural Kenya, 2015/16.......................................................................................... 90 Figure 4.12: Proportion of cultivated area by crop category in rural Kenya ........................................................................................................................... 91 Figure 4.13: Maize and bean yield in selected African countries.................................................................................................................................................... 91 Figure 4.14: Heterogeneity in crop productivity across provinces in rural Kenya ............................................................................................................... 92 Figure 4.15: Heterogeneity in crop productivity by gender of household head ................................................................................................................ 93 Figure 4.16: Gender differences in input use in rural Kenya.............................................................................................................................................................. 93 Figure 4.17: Trends in input use by farmers (Tegemeo Panel)......................................................................................................................................................... 94 Figure 4.18: Trends of crop prices and overall prices............................................................................................................................................................................. 97 Figure 4.19: There was an observed reduction in subsistence agriculture in rural Kenya between 2005/06 and 2015/16 ..................... 98 Figure 4.20: Relationship between poverty and market participation....................................................................................................................................... 99 Figure 5.1: Urbanization rates in Kenya and other countries, 1950–2050.............................................................................................................................. 102 Figure 5.2: Poverty headcount ratio and number of poor, 2005/6 and 2015/16............................................................................................................... 103 Figure 5.3: Poverty rates and number of poor in urban areas by province, 2005/6 and 2015/16........................................................................... 104 Figure 5.4: Share of urban poor across counties, 2015/16............................................................................................................................................................... 104 Figure 5.5: County-level urban poverty rates and number of urban poor, 2015/16....................................................................................................... 105 Figure 5.6: County-level urban and rural poverty rates, 2015/16................................................................................................................................................ 105 Figure 5.7: Sectoral decomposition of poverty reduction, 2005/6 and 2015/16 ............................................................................................................. 106 Figure 5.8: Share of recent migrants in urban areas in 47 counties, 2014.............................................................................................................................. 108 Figure 5.9: Wealth index by migration status, 2014............................................................................................................................................................................. 108 Figure 5.10: Share of household expenditure in urban Kenya, 2005/06 and 2015/16...................................................................................................... 110 Figure 5.11: Housing units with non-durable structures in urban areas, 2005/06 and 2015/16................................................................................ 111 Figure 5.12: Access to improved water in provinces by urban/rural area, 2005/06 and 2015/16.............................................................................. 111 Figure 5.13: Access to water in urban Kenya, 2005/06 and 2015/16 .......................................................................................................................................... 112 Figure 5.14: Access to improved sanitation in provinces by urban/rural area, 2005/06 and 2015/16.................................................................... 113 Figure 5.15: Access to improved sanitation in urban Kenya, 2005/06 and 2015/16 ......................................................................................................... 114 Figure 5.16: Access to electricity in provinces by urban/rural area, 2005/06 and 2015/16............................................................................................ 115 Figure 5.17: Access to electricity in urban Kenya, 2005/06 and 2015/16.................................................................................................................................. 115 Figure 5.18: Labor force participation rates in urban Kenya, 2005/06 and 2015/16 ......................................................................................................... 116 Figure 5.19: Unemployment rates in urban Kenya, 2005/6 and 2015/16................................................................................................................................. 117 Figure 5.20: Economic sectors of workers in urban Kenya, 2005/6 and 2015/16................................................................................................................. 117 Figure 5.21: Employment in urban Kenya, 2005/06 and 2015/16 ................................................................................................................................................ 118 Figure 5.22: Job types in urban Kenya, 2015/16 ...................................................................................................................................................................................... 119 Figure 5.23: Commuting modes in urban Kenya, 2005/6 and 2015/16..................................................................................................................................... 121 Figure 5.24: Share of accessible jobs within 60 minutes in Nairobi.............................................................................................................................................. 122 Figure 5.25: Job accessibility and per capita household expenditure in Nairobi................................................................................................................. 123 Figure 5.26: Household consumption and rents in Nairobi’s informal settlement and non-informal settlement areas, 2015/16........ 124 Figure 5.27: Housing quality in African informal settlements........................................................................................................................................................... 125 Figure 5.28: Perceived tenure security in African informal settlements..................................................................................................................................... 125 Figure 5.29: Previous residence of urban households........................................................................................................................................................................... 126 Figure 5.30: Probability of households moving to non-informal settlement areas in Nairobi and Mombasa................................................... 126 Figure 6.1: Public expenditure in education, 2000–2015................................................................................................................................................................. 129 Figure 6.2: GERs in pre-primary, primary, secondary, and tertiary, 2000–2016................................................................................................................... 130 Figure 6.3: NERs and GERs by level, poverty, quintile, and locality, 2015/16......................................................................................................................... 131 Figure 6.4: Changes in primary and secondary enrollment, between 2005/06 and 2015/16, by poverty, quintile, and locality........ 132 Figure 6.5: Gross enrollment rates by grade and year........................................................................................................................................................................ 132 Figure 6.6: GERs in primary and secondary education by county, 2015/16......................................................................................................................... 133 Figure 6.7: Net intake rate and transition by poverty, quintile, and locality, 2005/06 and 2015/16....................................................................... 134 Figure 6.8: Primary gross enrollment by provider, location, and quintile, 2005/06 and 2015/16............................................................................ 135 Figure 6.9: Average and median household per-student expenditure on education by level, location, and provider, 2005/06 and 2015/16..................................................................................................................................................................................................................... 136 Figure 6.10: Knowledge of fourth-grade students across Sub-Saharan African countries, early 2010s................................................................. 137 Figure 6.11: Learning outcomes in mathematics in ten-year-old children by socio-economic background, 2014 ..................................... 137 Figure 6.12: Proportion of twelve-year-old children proficient in mathematics and english, percent, 2014..................................................... 138 Figure 6.13: Physical inputs at the school-level by location and type of provider, primary schools, 2012........................................................... 140 Figure 6.14: Student-teacher ratios in public schools, 2004-2015, and students per classroom in primary schools, 2012....................... 141 Figure 6.15: Cross-country comparison of teacher salaries by level............................................................................................................................................. 142 Figure 6.16: Teachers’ subject knowledge and pedagogical skills by country, early 2010s .......................................................................................... 143 Figure 6.17: Absence from school and absence from class by country..................................................................................................................................... 144 Figure 6.18: Absenteeism rates by type of provider and type of teacher, 2012.................................................................................................................... 144 Figure 7.1: Outpatient visits and institutional deliveries by provider, January 2012 to December 2017............................................................ 150 Figure 7.2: Levels and trends in health expenditure by source, 2004-2014.......................................................................................................................... 151 Figure 7.3: Membership and resources of National Hospital Insurance Fund (NHIF), 2006/07-2014/15............................................................ 152 Figure 7.4: Health outcomes in Kenya vis-à-vis benchmark countries and aggregates, latest year available................................................. 155 Figure 7.5: Annual rate of reduction in selected indicators of childhood health, percent, c. 2000 to 2015..................................................... 156 Figure 7.6: TFR (number of births per woman) and under-five mortality rate (deaths per 1,000 live births).................................................. 156 Figure 7.7: TFRs against under-five mortality, countries (2015) and Kenyan counties (2014)................................................................................... 157 Figure 7.8: Levels in trends in registered deaths by cause, 2011–2015.................................................................................................................................... 157 Figure 7.9: Self-reported instances of sickness or injury during last four weeks prior to the survey as percent of population........... 158 Figure 7.10: Under-five mortality (deaths per 1,000 live births) by quintile, mother’s educational attainment, and location................. 159 Figure 7.11: Stunting rate by quintile, mother’s educational attainment, and location, 2003–2014....................................................................... 159 Figure 7.12: Child health outcomes by county, 2014............................................................................................................................................................................. 160 Figure 7.13: Obesity rates (BMI > 30, share of women aged 15-49) by quintile, educational attainment, and locality, 2003–2014......... 160 Figure 7.14: Selected indicators of health services uptake (%), 2000–2015............................................................................................................................. 161 Figure 7.15: Availability of health facilities and distance to nearest health facility in which a doctor would be on duty, 2015/16..... 162 Figure 7.16: Uptake of curative health services and number of curative visits by quintile and locality, 2005/06 and 2015/16.............. 163 Figure 7.17: Uptake of preventive health services during four weeks prior to interview................................................................................................ 163 Figure 7.18: Uptake of preventive health goods, select indicators, by poverty and quintile, 2015/16................................................................... 164 Figure 7.19: Access to health services and uptake by county, 2014, select indicators..................................................................................................... 165 Figure 7.20: Share of births (of surviving children 60 months and younger) by circumstance, 2005/06 and 2015/16................................ 166 Figure 7.21: Share of deliveries by provider, 2009–2014...................................................................................................................................................................... 166 Figure 7.22: Health insurance coverage, health expenditure and incidence of asset sales in response to hospitalization....................... 167 Figure 7.23: Average shares of in-patient health expenditure by funding source (democratic shares per hospitalized individual).......... 167 Figure 7.24: Infrastructure availability in public and private facilities by type of facility and location (select indicators)........................... 168 Figure 7.25: Drug availability by type of facility, provider, and location..................................................................................................................................... 169 Figure 7.26: Number of health professionals per 10,000 population........................................................................................................................................... 170 Figure 7.27: Salaries of nurses and midwives by country, 2005/06-2015/16.......................................................................................................................... 171 Figure 7.28: Salaries of select health workers in Kenya, 2005/06 and 2015/16...................................................................................................................... 171 Figure 7.29: Adherence to clinical guidelines and absence from health facility by country......................................................................................... 173 Figure 8.1: Poverty and vulnerability in Kenya: 2005/06 and 2015/16..................................................................................................................................... 182 Figure 8.2: Vulnerability rates by county: 2015/16................................................................................................................................................................................ 183 Figure 8.3: Vulnerability rates by poverty status: 2015/16................................................................................................................................................................ 183 Figure 8.4: CDFs of the rural and urban population: 2015/16....................................................................................................................................................... 184 Figure 8.5: Vulnerability rates relative to the average: 2015/16.................................................................................................................................................... 185 Figure 8.6: The prevalence of different shocks over consumption quintiles: 2005/06 and 2015/16.................................................................... 188 Figure 8.7: Prevalence of shocks by urban-rural location: 2005/06 and 2015/16.............................................................................................................. 188 Figure 8.8: Incidence of shocks by poverty status, agricultural households only: 2005/06 and 2015/16.......................................................... 189 Figure 8.9: Shock prevalence for agricultural households only: 2005/06 and 2015/16................................................................................................. 190 Figure 8.10: Geographic distribution of different shocks: 2015/16............................................................................................................................................... 190 Figure 8.11: The severity of losses from shocks: 2005/06 and 2015/16...................................................................................................................................... 191 Figure 8.12: Coping mechanisms over the distribution of consumption: 2005/06 and 2015/16.............................................................................. 192 Figure 8.13: Coping strategies by urban-rural place of residence: 2005/06 and 2015/16.............................................................................................. 192 Figure 8.14: Coping strategies by shock type – Rural households only: 2005/06 and 2015/16.................................................................................. 193 Figure 8.15: Expenditure on social safety nets: 2015.............................................................................................................................................................................. 196 Figure 8.16: Number of households receiving cash transfers: 2013 to 2016........................................................................................................................... 197 Figure 8.17: Coverage and share of beneficiaries by county: 2016............................................................................................................................................... 199 Figure 8.18: Share of beneficiary households by county and program: 2016........................................................................................................................ 199 Figure 8.19: CDFs of consumption by cash transfer program.......................................................................................................................................................... 200 Figure 8.20: The impact of grant receipt on the probability that all school-aged children in the household are enrolled...................... 203 Figure 8.21: The impact of grant receipt on the probability that no school-aged child in the household is working................................ 203 Figure 8.22: The impact of grant receipt on the probability that a household is food secure: HSNP counties only..................................... 204 Figure A.1: TFP growth was a key driver of GDP growth................................................................................................................................................................... 206 Figure A.2: As growth in capital accelerated, growth of labor moderated............................................................................................................................ 207 Figure A.3: Stagnating human capital growth resulted in a moderation of human capital per unit of labor................................................. 207 Figure A.4: The increase in labor force resulted in increasing unemployment and declining labor force participation.......................... 207 Figure A.5: County allocation of ordinary government revenues................................................................................................................................................ 210 Figure A.6: Transfers to county governments, 2016–17..................................................................................................................................................................... 210 Figure A.7: Share of transfers to counties.................................................................................................................................................................................................... 211 Figure A.8: Change in allocation of transfers by share of urban population......................................................................................................................... 211 Figure A.9: Development expenditure share of total expenditure............................................................................................................................................. 211 Figure A.10: Absorption rates of county budgets..................................................................................................................................................................................... 212 Figure A.11: Personnel costs by county.......................................................................................................................................................................................................... 212 Figure A.12: Share of county own revenues................................................................................................................................................................................................. 213 Figure A.13: Cumulative annual growth rate of personnel costs.................................................................................................................................................... 213 Figure A.14: Own revenues as a share of actual county expenditure.......................................................................................................................................... 213 Figure A.15: Average annual increase in own-source revenues...................................................................................................................................................... 213 Figure B.1: Map of NEDI counties..................................................................................................................................................................................................................... 216 Figure B.2: Distribution of the log of population density by cluster type.............................................................................................................................. 217 Figure B.3: Occupational sector of household head by area of residence............................................................................................................................. 217 Figure B.4: Source of food consumption by area of residence...................................................................................................................................................... 217 Figure B.5: Household characteristics by area of residence............................................................................................................................................................ 217 Figure B.6: Asset ownership by consumption quintile, Nairobi.................................................................................................................................................... 218 Figure E.1: Number of urban poor and urban poverty rate by county, 2015/16............................................................................................................... 233 Figure E.2: Cash transfer during the last three months in 15 cities, 2013............................................................................................................................... 233 Figure E.3: Expenditure share on housing in urban Kenya.............................................................................................................................................................. 234 Figure E.4: Expenditure share on housing in urban Kenya by county, 2015/16................................................................................................................. 234 Figure E.5: Comparison of health indicators in Kenya, 2000 to 2014........................................................................................................................................ 235 Figure E.6: Number and share of unemployed population in urban area by county, 2015/16................................................................................ 236 Figure E.7: Unemployment rate in urban area by sex and county, 2015/16........................................................................................................................ 236 Figure E.8: Unemployment rate in urban area by the youth and county, 2015/16......................................................................................................... 236 Figure E.9: Comparison of economic sectors in urban Kenya by county, 2015/16.......................................................................................................... 237 Figure E.10: Duration of residence in 47 counties, 2014...................................................................................................................................................................... 238 Figure E.11: Previous residence of recent migrants in 47 counties, 2014.................................................................................................................................. 239 Figure E.12: Previous residence of recent migrants in 47 countries, 2014................................................................................................................................ 240 Figure E.13: Cumulative distribution of the duration of residence in Nairobi and Mombasa...................................................................................... 241 Figure H.1: Consumption levels of vulnerable households, relative to the poverty line: 2015/16.......................................................................... 250 Figure H.2: The prevalence of shocks by poverty and vulnerability status: 2005/06 and 2015/16......................................................................... 250 LIST OF TABLES Table 1: Access to basic services by poverty status........................................................................................................................................................................ v Table 2: Sectoral decomposition of poverty reduction (Ravallion-Huppi)...................................................................................................................... vi Table 3: Monthly earnings in Ksh, by gender..................................................................................................................................................................................... xiv Table 1.1: List of ongoing major projects................................................................................................................................................................................................. 12 Table 1.2: Key monetary poverty Indicators........................................................................................................................................................................................... 14 Table 1.3: Revenue-sharing among counties in Kenya.................................................................................................................................................................... 24 Table 2.1: Absolute poverty headcount rate, nationally, by area of residence.................................................................................................................. 32 Table 2.2: Poor and total populations, nationally, by area of residence and by NEDI classification..................................................................... 32 Table 2.3: Comparison of noncomparable and comparable 2005/06 poverty rates.................................................................................................... 36 Table 2.4: Theil inequality index - decomposition by urban/rural location and province......................................................................................... 49 Table 2.5: Sectoral decomposition of poverty reduction (Ravallion-Huppi)...................................................................................................................... 53 Table 2.6: Sectoral decomposition of poverty reduction (Ravallion-Huppi) - alternative definition................................................................... 54 Table 2.7: Household characteristics by poverty status................................................................................................................................................................... 56 Table 3.1: Primary and secondary enrollment rates and gender parity, 2005/6 and 2015/6................................................................................... 63 Table 3.2: Male and female wage employment by employment status, 2015/6............................................................................................................ 75 Table 3.3: Male and female monthly earnings, in current Ksh, and male-to-female ratio, 2015/6....................................................................... 75 Table 3.4: Descriptive differences between male- and female-run household enterprises, 2015/6................................................................... 76 Table 3.5: Descriptive differences in input use between male and female decision-makers in agriculture, 2015/6................................ 78 Table 4.1: Decomposition of poverty by income classification.................................................................................................................................................. 86 Table 4.2: Determinants of maize yield, FEs model, 2000–10...................................................................................................................................................... 95 Table 5.1: Recent male migration by origin and destination....................................................................................................................................................... 107 Table 5.2: Median nominal wage by economic sector in urban Kenya, 2015/16............................................................................................................ 120 Table 5.3: Average share of accessible jobs in Nairobi..................................................................................................................................................................... 122 Table 5.4: Poverty rates in informal settlement and non-informal settlement areas, Nairobi 2015/16............................................................. 123 Table 7.1: OLS regression of log salary (incl. allowances) on binary indicator of employment in public sector for auxiliary nurses; nurses and midwives; and medical and clinical officers, 2005/06 and 2015/16...................................................................... 172 Table 7.2: Outcomes for select standardized patient cases in Nairobi, urban China, and India............................................................................. 173 Table 7.3: Primary outcomes for standardized patient cases by sector................................................................................................................................. 174 Table 8.1: Profiles of the poor and the vulnerable: 2005/06 and 2015/16........................................................................................................................... 185 Table 8.2: Coping strategies by poverty status: 2015/16................................................................................................................................................................ 194 Table 8.3: Social Protection Programs in Kenya.................................................................................................................................................................................... 196 Table 8.4: Profile of beneficiary households versus non-beneficiary households (by poverty status)............................................................... 201 Table A.1: Poverty trajectory simulation, sectoral and non-sectoral growth...................................................................................................................... 206 Table B.1: Sampling framework ..................................................................................................................................................................................................................... 216 Table B.2: Response rates by county............................................................................................................................................................................................................ 218 Table C.1: Correlates of labor force participation, probit (coefficients).................................................................................................................................. 222 Table C.2: Oaxaca-Blinder decomposition of gender gaps in monthly earnings, summary..................................................................................... 223 Table C.3: Oaxaca-Blinder decomposition of gender gaps in monthly earnings, descriptive statistics ........................................................... 224 Table C.4: Oaxaca-Blinder decomposition of gender gaps in monthly earnings, OLS (coefficients).................................................................. 225 Table C.5: Correlates of household enterprise profits, OLS (coefficients)............................................................................................................................. 226 Table D.1: Determinants of beans yield, FEs Model............................................................................................................................................................................ 230 Table E.1: Nominal monthly salary in urban Kenya............................................................................................................................................................................ 231 Table E.2: Comparison of dwelling characteristics between informal settlement and non-informal settlement areas in Nairobi....... 232 Table E.3: Comparison of access to services between informal settlement and non-informal settlement areas in Nairobi............... 232 Table F.1: GERs and NERs in secondary and primary education by county........................................................................................................................ 242 Table F.2: Determinants of transition from seventh into eighth grade of primary and from primary into secondary............................ 245 Table G.1: Regression results from LPMs – effect of free deliveries in public facilities on uptake by provider (N = 28,154)................. 247 Table G.2: Regression results from LPMs – effect of free deliveries in public facilities on uptake by provider, urban and rural (N = 28,154)..................................................................................................................................................................................................................... 247 Table G.3: Effect of institutional delivery and assistance on neonatal mortality (odds ratios/t-values) (N = 19,080)................................ 248 Table H.1: Coping strategies by poverty status for agricultural households only: 2015/16 ..................................................................................... 251 LIST OF BOXES Box 1.1: The Big 4 policy agenda....................................................................................................................................................................................................................... 9 Box 1.2: The international poverty lines........................................................................................................................................................................................................ 14 Box 1.3: Public expectations from devolution........................................................................................................................................................................................... 25 Box 1.4: Key features of the 2010 Kenyan Constitution....................................................................................................................................................................... 26 Box 2.1: Kenya Integrated Household Budget Survey (KIHBS): A commendable effort.................................................................................................. 33 Box 2.2: Measuring poverty: Computing the poverty lines, the consumption aggregate and classification of peri-urban households.......................................................................................................................................................................................................................... 35 Box 2.3: Nairobi nonresponse rates – dealing with data issues...................................................................................................................................................... 44 Box 2.4: Inequality measures................................................................................................................................................................................................................................ 48 Box 2.5: What does decomposing changes in poverty entail?....................................................................................................................................................... 50 Box 5.1: Definition of urban areas..................................................................................................................................................................................................................... 103 Box 5.2: Decomposition analysis........................................................................................................................................................................................................................ 106 Box 5.3: Job accessibility......................................................................................................................................................................................................................................... 121 Box 5.5: Profile of residents moving to/from informal settlement neighborhoods........................................................................................................... 126 Box 6.1: Free primary education and the quality of education...................................................................................................................................................... 129 Box 6.2: Are private schools more productive?......................................................................................................................................................................................... 139 Box 6.3: Are higher public-sector wages efficient?................................................................................................................................................................................ 145 Box 7.1: Promises and perils of the devolution of health services................................................................................................................................................ 153 Box 7.2: What works to boost skilled birth assistance for safer childbirth?............................................................................................................................. 164 Box 8.1: Concepts of risks, shocks and vulnerability.............................................................................................................................................................................. 179 Box 8.2: Measuring vulnerability using cross-sectional data............................................................................................................................................................ 181 Box 8.3: Measuring the prevalence of and responses to shocks in KIHBS data.................................................................................................................... 187 Box 8.4: Findings from impact evaluations of the OVC and the HSNP programs............................................................................................................... 198 Box 8.5: Evaluating the impacts of Kenya’s cash transfer programs using cross-sectional data and propensity score matching ...... 202 APPENDICES Appendix A: Chapter 1 additional materials ................................................................................................................................................................................................ 206 Appendix B: Chapter 2 additional materials ................................................................................................................................................................................................ 216 Appendix C: Chapter 3 additional materials ................................................................................................................................................................................................ 220 Appendix D: Chapter 4 additional materials ................................................................................................................................................................................................ 229 Appendix E: Chapter 5 additional materials ................................................................................................................................................................................................ 231 Appendix F: Chapter 6 additional materials ................................................................................................................................................................................................ 242 Appendix G: Chapter 7 additional materials ................................................................................................................................................................................................ 246 Appendix H: Chapter 8 additional materials ................................................................................................................................................................................................ 250 ACKNOWLEDGEMENT The World Bank greatly appreciates the close collaboration with the Government of Kenya (GoK), particularly the Kenya National Bureau of Statistics (KNBS) in the preparation of this report. This report was prepared by a team led by Utz Pape (Senior Economist) and Carolina Mejia-Mantilla (Economist)1, with the guidance of Johan Mistiaen (Program Leader) of the Africa Region in the Poverty and Equity Practice. The team consisted of Marina Tolchinsky (Chapter 1), Christine Achieng Awiti (Chapter 1), Nduati Maina Kariuki (Chapter 2), Isis Gaddis (Chapter 3), Habtamu Fuje (Chapter 4), Shohei Nakamura (Chapter 5), Simon Lange (Chapters 6 and 7) and Arden Finn (Chapter 8). The authors received substantive contributions from Haseeb Ali, Paolo Avner, Stephan Dietrich, Yuka Karasawa, Angelo Martelli, Saurabh Naithani, Stephen Okiya, Vera Sagalova, Aaraon Thegeya and the Tegemeo Institute of Agricultural Policy and Development (Egerton University). The report was prepared under the supervision of Diarietou Gaye (Country Director for Kenya, Rwanda, Uganda, and Eritrea) and Pierella Paci (Practice Manager). The peer reviewers were Markus Goldstein, Gabriel Demombynes and Prof. Michael Chege. The report benefitted from excellent comments from G N V Ramana, Ruth Karimi Charo, Frederick Masinde Wamalwa, Emma Mistiaen, Jishnu Das, Richard Chirchir, Mark Pancras and Evelyn Mwangi. 1 With equal contributions. ABBREVIATIONS Currency Equivalents (Exchange Rate Kenyan Shilling Effective as of Sept 28, 2018) US$1.00 = Ksh 100.956 AEZ Agro-Ecological Zone KNUT Kenya National Union of Teachers CBN Cost of Basic Needs LMIC Lower Middle-Income Class CDF Cumulative Density Function LPM Linear Probability Model CEC County Executive Committee LSMS Living Standards Measurement Study CPI Consumer Price Index MP Member of Parliament CRA Commission on Revenue Allocation NEDI North & Northeastern Development Initiative DHS Demographic and Health Survey NER Net Enrollment Rate DPT Diphtheria, Pertussis, and Tetanus NGO Nongovernmental Organization EAC East African Community NHIF National Hospital Insurance Fund ETP Extra Teacher Program NSBDP National School-Based Deworming Programme FAO Food and Agriculture Organization NSNP National Safety Net Programme FBO Faith-Based Organization OCOB Office of the Controller of Budget FPE Free Universal Primary Education ODM Orange Democratic Movement FTSE Free Tuition Secondary Education OPCT Older Persons Cash Transfer GER Gross Enrollment Ratio OVC Orphans and Vulnerable Children GIC Growth Incidence Curve PNU Party of National Unity GoK Government of Kenya PPA Participatory Poverty Assessment HDI Human Development Index PPP Public-Private Partnership HSNP Hunger Safety Net Program PSM Propensity Score Matching ICLS International Conference of Labor Statisticians RCT Randomized Control Trial IDS Institute for Development Studies SDI Service Delivery Indicators IEBC Independent Electoral and Boundaries Commission SPS Social Protection Secretariat ILO International Labour Organization STEM Science, Technology, Engineering and Mathematics IPC Infection Prevention and Control STI Sexually Transmitted Infections IPV Intimate Partner Violence TFP Total Factor Productivity ITN Insecticide-Treated Bed Net TFR Total Fertility Rate KCPE Kenya Certificate of Primary Education TSC Teacher Service Commission KCSE Kenya Certificate of Secondary Education TVET Technical and Vocational Education and Training KDHS Kenya Demographic and Health Survey UFS Urban Food Subsidy KES Kenya Economic Survey UHC Universal Health Coverage KHHEUS Kenya Household Health Expenditure and Utilisation Surveys UNDP United Nations Development Program KICD Kenya Institute of Curriculum Development UNESCO United Nations Educational, Scientific and Cultural Organization KIHBS Kenya Integrated Household Budget Surveys VIP Ventilated Improved Pit KNBS Kenya National Bureau of Statistics WDI World Development Indicators KNOCS Kenyan National Occupation Classification Standard WHO World Health Organization KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead i ii KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead EXECUTIVE SUMMARY KENYA MADE PROGRESS IN REDUCING Households in the bottom 40 percent of the POVERTY AND INEQUALITY OVER THE PAST distribution experienced notable consumption DECADE growth. The annualized consumption growth for T he substantive economic growth of the last decade has brought Kenya into the low middle- income country category in 2014. For the period of the bottom 40 percent, also known as the shared prosperity indicator, was 2.86 percent per year for the period from 2005/06 to 2015/16, and particularly high focus of this report, 2005/06 to 2015/16, growth in for the rural households. Even amongst the poor, those Kenya averaged 5.3 percent, higher than the 4.9 percent at the bottom of the distribution experienced higher observed for sub-Saharan Africa as a whole (Figure 1a). consumption growth: households in the bottom 20 Overall, growth was powered by the service sector, percent of the distribution experienced annualized which now accounts for almost half of the nation’s growth rates of around 3 to 4 percent. This trend was more GDP. The remarkable expansion of telecommunication marked in rural areas, which lead to a more pronounced and mobile-based financial services shifted the poverty reduction amongst rural households compared economic paradigm of Kenya to an extent rarely seen to their urban counterparts. However, compared to its in developing economies. Moreover, the country was regional peers, with the exception of Ethiopia, Kenya capable of bouncing back from the violent political has been less successful in boosting shared prosperity outbreak that followed the 2007 presidential election, (Figure 1d). from the effects of the 2008/09 global financial crisis, and from the harsh drought conditions experienced Consistent with this pro-poor pattern of economic by most of the African Horn in 2011, aggravated by the growth, inequality declined in Kenya. The Gini increase in the international price of oil. index, generally not affected by the upper tail of the distribution, fell from 0.45 in 2005/06 to 0.39 in Poverty incidence declined, benchmarked against 2015/16, indicating that Kenya made considerable both the national and the international poverty lines, progress in reducing inequality. Similar trends are but remains high relative to other lower middle- observed with the Theil index, the 75/25 ratio and the income countries. The proportion of the population Atkinson index. This places inequality in Kenya at a living beneath the national poverty line fell from moderate level when compared to other countries in 46.8 percent in 2005/06 to 36.1 percent in 2015/16, the region (Figure 1e). Now, while the fall in inequality showing a modest improvement in the living standards contributed to poverty reduction, most of the decline is of the Kenyan population, considering the ten year attributable to economic growth rather than a change gap (Figure 1b). Given the high dependence of the in the distribution of resources. This is consistent with agricultural sector on rainfall, the decline was higher in the low coverage of the social protection programs years of good weather and lower in years of drought. in the country, and the fact that fiscal policy as a Similarly, poverty under the international poverty line whole has little incidence on the level of poverty in of US$ 1.90 a day declined from 43.7 percent in 2005/06 Kenya, as shown by a recent fiscal incidence (World to 36.8 percent in 2015/16. At this level, poverty in Bank 2018b). This suggests that a more focused effort Kenya is below the sub-Saharan Africa average and on redistributive policies, such as social protection is amongst the lowest in the East African Community programs and equality of opportunities, can help (Figure 1c). However, it is approximately twice as high accelerate poverty reduction going forward. the average for its middle-income group. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead iii Executive Summary Figure 1: Kenya’s economic and poverty progress a) GDP growth from 2005 to 2016 b) Poverty headcount rate under the national poverty line 10.0 60 8.0 50.5 50 46.8 Proportion of the population 6.0 40 38.8 36.1 32.1 29.4 Percent 4.0 30 2.0 20 0.0 10 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 -2.0 0 National Rural Urban -4.0 GDP per capita growth GDP growth 2005/06 2015/16 c) Poverty headcount rate under the international poverty line d) Shared prosperity indicator 70 12 9.8 60 60 Poverty headcount (% of population) 8 consumption for bottom 40 percent Annualized % change in mean 49 50 41 4.6 4.3 40 37 4 3.5 35 2.9 30 0 20 -0.4 14 10 -4 Kenya Rwanda South Uganda Tanzania Ethiopia 0 (2006 - (2005 - Africa (2005 - (2007 - (2005 - Rwanda Tanzania SSA Kenya Uganda Ghana 2016) 2010) (2005 - 2012) 2011) 2010) 2013 2011 2013 2015 2012 2012 2010) e) Measuring inequality: Gini index f ) Distribution of the poor population across rural and urban areas 0.8 20 0.63 0.6 15 Number of poor (in millions) 0.50 0.42 0.41 0.39 Gini index 0.4 0.38 10 12.6 0.33 14.3 0.2 5 3.8 2.3 0.0 0 Kenya Ethiopia Ghana Rwanda South Tanzania Uganda 2005/06 2015/16 (2015/16) (2010) (2012) (2013) Africa (2011) (2012) (2011) Urban poor Rural poor Source: KNBS; own calculations based on KIHBS 2005/06 and KIHBS 2015/16 and World Bank open data catalogue. iv KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Most of the poverty decline is attributable to the and electricity services is much lower for poor progress observed in rural areas. Poverty declined households (Table 1). In this sense, Kenya should considerably in rural Kenya, from around 50 percent in continue to expand the coverage of this basic services 2005/06 to 38.8 percent ten years later, resulting in a to all segments of the population, while ensuring their decline in the number of rural poor from 14.3 million to quality at the same time. 12.6 million people. This contrasts with the stagnation of poverty in urban areas, where no clear decline in the Off-farm diversification played an important role in reducing poverty poverty headcount is observed as the 2.7 percentage points reduction is not statistically significant from zero. The evidence suggests that off-farm diversification Importantly, the number of urban poor increased during has been important for poverty reduction in this period, with cities concentrating a larger fraction of Kenya. Households whose agricultural income was the poor than they did before (Figure 1f ). This is partly supplemented by non-agricultural activities, mainly explained by the relative increase in food prices, which in small-scale services, account for slightly more than is known to affect the urban poor while benefitting rural a third of the poverty reduction (33.5 percent), the food net-producers. Housing costs have also increased highest share. This highlights the importance of off- in medium- and small-sized towns, reflecting that farm diversification in poverty reduction over the last urban growth has exacerbated a shortage in the supply ten years. While the agricultural sector has not been of affordable housing. It seems that cities, particularly as dynamic as the service or the industrial sector, it secondary cities, are not providing sufficient economic played a notable role in reducing poverty. Households opportunities for individuals to improve their income for which the main source of income is agriculture level and participate in the overall economic progress. (including both crop income, livestock income, and earning of wage workers in the agricultural sector) Poor households remain constrained by demographic account for 27.6 percent of the overall reduction characteristics, low human capital, and low coverage (Table 2). Finally, households engaged exclusively of basic services. Poverty incidence is higher for in non-agricultural activities, including services, households headed by women, the elderly and those manufacturing and construction, contributed with with low educational attainment levels. This suggests about 21 percentage points. that the poor are constrained when accessing income generating opportunities. Moreover, poor households While agriculture remains the main source of tend to be larger, and have higher dependency ratios; income for rural households in Kenya, the share of demographic factors that usually hinder poverty income from non-agricultural employment and non- reduction. In addition, coverage of water, sanitation agricultural employment has increased significantly Table 1: Access to basic services by poverty status 2005/06 2015/16 Significance Significance Non-Poor Poor Non-Poor Poor (wald-test) (wald-test) Access to services             Improved drinking water 70.2% 51.9% *** 80.4% 65.6% *** Improved sanitation 56.4% 37.7% *** 72.2% 47.8% *** Main source light (electricity) 23.0% 4.0% *** 49.9% 18.9% *** HH electricity access 26.5% 4.5% *** 52.0% 20.7% *** HH has garbage collected 10.7% 2.9% *** 21.7% 6.0% *** Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead v Executive Summary Table 2: Sectoral decomposition of poverty reduction (Ravallion-Huppi) Pop. share in Absolute Percentage Source of income period 1 change change Non-agricultural income only 31.64 -2.16 21.19 Agriculture income only 39.79 -2.81 27.63 Mixed - agricultural & non agricultural income 28.57 -3.41 33.51 Total intra-sectoral effect   -8.37 82.33 Population shift effect   -1.68 16.49 Interaction effect   -0.12 1.19 Change in headcount rate   -10.17 100.00 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. in the last decade. Income from crops and livestock income in 2005/06 to 21 percent in 2015/16. This as well as wages in the agricultural sector, declined diversification of income, in which households from 64.0 percent in 2005/06 to 57 percent in 2015/16 complement agricultural income with income derived (Figure 2a). Wage income from service employment is from non-agricultural activities (particularly in services the second most important source of income in rural and trading activities), along with an increase in the areas, increasing from 15 percent of rural household share of labor time allocated to non-agricultural Figure 2: Share of income by source for rural households a) Share of income by source for rural households b) Non-agricultural labor allocation in rural Kenya 60 Percent of labor time on non-agricultural activities 8 6 8 9 40 15 21 Percentage 4 7 20 64 57 0 2005/06 2015/16 Kenya Central Coast Eastern North Nyanza Rift Western Agriculture Industry wage Services wage Eastern Valley Transfers Enterprise income 2005 2015 c) GDP growth rates, by sectors d) Growth of mobile payments, by agents and transaction value 14 250,000 400 350 200,000 9 300 150,000 250 Percent 200 4 100,000 150 100 50,000 50 -1 0 0 Mar-07 Dec-07 Sep-08 Jun-09 Mar-10 Dec-10 Sep-11 Jun-12 Mar-13 Dec-13 Sep-14 Jun-15 Mar-16 Dec-16 Sep-17 Jun-18 -6 2007 2008 2009 2010 2011 2012 2013 2014 2015 Agriculture GDP Services Industry Agents (left) Value (KSh billions; right) Source: KNBS; own calculations based on KIHBS 2005/06 and KIHBS 2015/16 and World Bank open data catalogue. vi KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary activities between 2005/06 and 2015/16 (Figure 2b), sectors of the economy.2 Similarly, many farmers have have been important to the reduction of poverty in rural shifted to bean production in recent years, as the Kenya. Thus, it is important to support rural households country benefited from favorable bean and maize in their effort to diversify their income. Investments in prices from 2011 to 2016. Farmers that shifted to bean human capital, skills formation, as well as attracting production are less likely to be classified as poor. High non-agricultural economic activities into rural areas, are commodity prices, like those observed between 2010 key areas of actions in which the Government of Kenya and 2016, is generally beneficial for Kenya’s net-selling should focus. farmers. However, this is at the expense of the urban poor, as poor urban households spend a large share of Those deriving income from non-agricultural their income on food and are therefore vulnerable to activities benefitted greatly from the expansion rising food prices. This factor may have contributed to of mobile money (thanks to M-PESA) throughout the divergence in poverty reduction between urban the country, particularly in remote previously and rural areas. uncovered areas. The penetration of mobile phones not only brought market efficiency gains associated Non-monetary wellbeing also improved, some issues still pending to be solved with a reduction in transaction costs (which likely also benefited those engaged in agriculture). Through The progress in the wellbeing of the population as mobile money, it also became a platform for service evaluated by monetary measures was accompanied delivery rather than just a communication tool, by the progress in several non-monetary dimensions changing Kenya’s economic paradigm as some have of poverty. Kenya’s Human Development Index pointed out (Jack and Suri, 2013, 2014; Suri 2017). (HDI), a combination of education, inequality, and life Mobile money, used by 18 million people in Kenya in expectancy indicators, gained 0.07 points in the decade 2017, increased the households’ financial resilience leading to 2015 reaching 0.55. This is the highest HDI in and savings, allowing them to: i) invest productively, ii) the East Africa Community, and a relatively high level move out of agriculture or complement that income given the county’s poverty headcount. with that of other businesses, and iii) improve their consumption levels, while also make risk sharing more Enrollment rates at all levels have increased, driven effectively. The way in which the expansion of mobile by higher enrolment of children from poor families. money transformed Kenya’s economy is an example for The government has invested substantial resources in other African countries, and the factors that enabled its recent years to increase enrollment rates, particularly expansion (including investment in infrastructure, the at the primary level with the introduction of universal regulatory environment and the participation of the primary education in 2003. As a result, primary private sector, among others), can provide important education is nearly universal with a net enrollment of lessons for low and low-middle income countries 85 percent in 2015/16, including 78.8 percent for the around the world. poor (Figure 3a). Enrollment in secondary education increased more gradually, and between 2005/06 and Other factors that likely benefitted Kenyan 2015/16 the net enrollment rate increased by more than households, particularly those in rural areas, are 20 percentage points reaching 42.2 percent (Figure 3b). the penetration of motorbikes (boda bodas), high Similarly, enrollment in tertiary education has increasing commodity prices and increased productivity in rapidly after 2009, and according to the 2015/16 KIHBS, the production of bean crops. Boda bodas helped to the gross enrollment rate is about 15.2 percent. lower the transaction costs of trading agricultural and non-agricultural goods as well as services, enhancing 2 Estimates suggest that in 2008 there were a total of 130,000 motor cycles registered in Kenya. By 2017, this number is likely to have the income rural households engaged in all different reached one million. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead vii Executive Summary Figure 3: Non-monetary dimensions of wellbeing a) Enrollment in primary education, 2015/16 b) Enrollment in secondary education, 2015/16 120 120 90 90 Percent Percent 60 60 30 30 0 0 Total Poor Non-poor Bottom 20% Bottom 40% Top 20% Rural Urban Total Poor Non-poor Bottom 20% Bottom 40% Top 20% Rural Urban By poverty By quintile By locality By poverty By quintile By locality Gross enrollment ratio Net enrollment rate Gross enrollment ratio Net enrollment rate c) Under-five mortality rate (per 1,000 live births) d) Multi-dimensional deprivations, 2015/16 180 80 150 120 60 % of households deprived 90 60 30 40 0 Total Bottom 20% Bottom 40% Top 20% Primary and lower Secondary and higher Rural Urban 20 0 Consumption Adult education Primary School Improved water Improved sanitation Access to electricity By quintile of wealth By mother's By location index highest level of educational attainment 2003 2008/09 2014 e) Average number of curative visits per person per year (total population) f) Distance to health facility where a doctor is available, in kilometers, 2015/16 6 40 5 30 4 3 20 2 10 1 0 0 Top 20% Non-poor Urban Poor Bottom 20% Bottom 40% Top 60% Total Urban Total Rural Rural Bottom 40% T60%, urban B40%, urban T60%, rural B40%, rural By poverty By quintile By locality By quintile By locality By quintile and locality 2005/06 2015/16 Mean Median Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16, WDI data. viii KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Kenyans have experienced significant gains in a among the richest quintile, compared to the poorest range of population health indicators in the last ten quintile, at almost half that level. Rural-urban disparities to fifteen years. Mortality among children below the are pronounced, with a 20 percentage point difference age of five has declined from 114.6 deaths per 1,000 in enrollment (Figure 3b). This is a reflection of low live births in 2003 to only 52 in 2014, a remarkable transition rates between primary and secondary achievement (Figure 3c). Moreover, the gains were school stemming from financial constraints and late widely shared, as the under-five mortality gap between enrollment in primary school. While learning outcomes poor and non-poor children declined. This progress of Kenyan children compare favorably to peer countries, is largely attributed to increased uptake of low-cost, the education system often fails to equip students with high-impact interventions (such as malaria nets) basic skills. Learning assessments suggest that Kenyan and declining fertility. Similarly, Kenya has also made children quickly fall behind the standards set by the substantial gains in reducing child stunting and it now national curriculum: only about half of the children in has one of the lowest stunting rates in the region. As fourth grade master the basic tasks that second-graders of 2014, nearly 1 out of every 4 children under the should be able to accomplish (e.g., read and understand age of 5 is stunted in Kenya, down since 2003, when a paragraph). Regional disparities in learning outcomes 35.6 percent of Kenyan children were stunted. In are pronounced and mirror those in enrollment. addition, improvements in uptake of both curative and Finally, while well-paid and knowledgeable by regional preventive services were also often more pronounced standards, Kenya’s teachers lack pedagogical skills and among the poor. are absent from class too often, suggesting that teacher incentives are not always aligned with student learning. However, Kenyan remain deprived in many of the dimensions. When looking at poverty as a Despite the progress, there are still pronounced multidimensional challenge, along the lines of the socioeconomic gradients in health access and components of the upcoming multi-dimensional some health outcomes warrant action. Children poverty index by the World Bank, households are often from poor families are less likely to be vaccinated deprived beyond the monetary dimension. The most and poor mothers are less likely to give birth with a common type of deprivation is access to services, qualified health provider present. In fact, in all domains notably sanitation and electricity: 40.7 percent of -outpatient care, inpatient care, and preventive care- households lack access to improved sanitation3 and and across almost all age groups, the poor are less likely 64 percent lack access to electricity in 2015/16. Fewer to use health services (Figure 3e exemplifies this point households, around 28.2 percent, are deprived of access by showing the average number of curative care visits to an improved source for drinking water (Figure 3d).4 per person per year). They also often have to overcome greater distances to access health care, particularly Regardless of the positive trends, geographic in rural areas. These gaps in access remain large and and socio-economic disparities in net secondary significant and are a major cause for concern (Figure enrollment remain a challenge and learning 3f ). In addition, maternal mortality ratio remains high assessments suggest that Kenyan children often lag at 510 deaths for every 100,000 live births, close to the behind the curriculum. Net enrollment in secondary average for low-income countries and only somewhat education at 56 percent remains substantially higher lower than the regional average. 3 Improved sanitation is defined as a toilet with a flush, a ventilated improved pit latrine or a latrine with a slab. 4 Improved drinking water sources are defined as a piped water system, public tap, borehole, protected dug well, bottled water or water from rainwater collection vendors. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead ix Executive Summary DESPITE PROGRESS IN REDUCING POVERTY, to healthcare and up-take rates, particularly in terms SEVERAL CHALLENGES REMAIN of children who are treated for illness, vaccination Progress has been slow rates and child-birth delivered by a skilled provider. For example, vaccination rates vary from more than H owever, progress is slow and Kenya is not on track to eradicate extreme poverty by 2030. Even though Kenya has experienced moderate GDP growth 90 percent in the Central region to about 44 percent in Mandera in the Northeast and only 36 percent in West Pokot, part of the NEDI counties. Limited access to in the last decade, transmission of growth into increased healthcare coupled with extremely high fertility rates, consumption of households is low. At 0.57, the country’s results in the highest maternal mortality rates of the elasticity of poverty reduction to economic growth – country. In addition, coverage of improved sanitation how much economic growth translates into poverty and electricity, and to a lesser extent, access to improved reduction – is low, below that of Tanzania, Ghana and water, is lower. While the government has implemented Uganda; and weaker than expected given its level of some measures to improve the connectivity and overall GDP per capita. To eradicate extreme poverty by 2030, wellbeing of the population in these areas, a substantive, an annual poverty reduction rate of 6.1 percent would sustained and cross-sectorial effort is required over the be necessary, despite the fact that in the last decade it medium term. has been 1.6 percent. If the trends observed in the last decade continue, the poverty rate will remain above Another source of spatial inequality is the growing 25 percent in 2030. To accelerate the pace of poverty inequality within cities, as the urban population in reduction, Kenya will require a far more inclusive Kenya increases over time. Within Nairobi, poverty is economic growth coupled with a sharper focus on highly concentrated in informal settlements, where the targeted poverty-reducing policies. living conditions are far worse, not only in comparison to the rest of the city but also in comparison to informal Stark spatial disparities remain settlements in other major African cities. Nearly a third Kenya is characterized by stark regional differences. of informal settlement residents in Nairobi are poor, The wellbeing of the population in the NEDI (North compared to 9 percent of the population living outside & Northeastern Development Initiative) group of informal settlement areas. Mean per capita monthly counties, which includes all counties in the North consumption of informal settlement residents (Ksh Eastern province, lags considerably behind the rest 10,377) is nearly 40 percent lower than that of non- of Kenya. In the NEDI counties, 68 percent of the informal settlement residents (Ksh 16,688), as shown population live in poverty compared to 36.1 percent in Figure 4f. Moreover, the living conditions in informal at the national level (Figure 4a). Moreover, these settlements, in terms of housing, access to services, counties saw little progress between 2005/06 and environmental problems, and health, are extremely 2015/16 and remain prone to food insecurity, as precarious. Informal settlement residents also live far shown by the food poverty and extreme poverty away from jobs, constraining their access to economic indicators (Figure 4b,c). Poor households of the NEDI opportunities. It also remains difficult to move out of a counties also lie far below the poverty line and the informal settlement, exacerbating the spatial poverty prevalence of vulnerability is highest in the counties trap in informal settlements. Mandera, Garissa, Samburu, and Turkana while rates are significantly lower in the central counties, Vulnerability is prevalent particularly in Nyeri, Kirinyaga and Nairobi. Although vulnerability and poverty rates fell over A sustained, multi-sectoral effort is required to raise the last decade, over half of Kenya’s population is the living standards of the population of these areas. currently vulnerable to falling into poverty in the near Educational enrollment rates are much lower for these future. Vulnerability rates5 fell faster in rural areas than counties, particularly in secondary education. In terms 5 Households are considered to be vulnerable if their predicted probability of being below the poverty line at any stage within the next two years is of health services, they present lower rates of access greater than 50 percent. x KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Figure 4: Regional patterns in poverty a) Absolute poverty b) Food poverty 74.0 76.2 80 65 68 80 70.0 68.0 57.6 53 51 55 Proportion of the population 51 Proportion of the population 60 54.2 60 50.6 45 48 45 49.0 47.2 44.2 39 42 40.5 41.4 42.5 38 36.7 32 36 33 40 31.8 31.1 40 32.6 28 30 21.3 20 24.3 22 16.7 16 20 20 0 0 h h rn rn t t DI DI DI DI za za lle ift lle ift l l bi bi rn rn ste rt ra ra as as Ea rn ste rt Ea rn W y W y Ea No te te iro iro Va R Va R NE NE NE NE an an ste ste Ea No nt nt Co Co es es Ny Ny Ce Na Ce Na n- n- No No 2005/06 2015/16 2005/06 2015/16 c) Extreme poverty d) Percentage of children age 12-23 months that received all basic vaccinations 80 60 Proportion of the population 47 50 40 32 27 24 23 23 20 20 20 17 12 14 11 6 7 6 6 3 3 0.6 0 (90,100] (80,90] r h lle ift t rn DI DI za l rn bi ra as ste rt (70,80] Ea n W y Va R ste NE NE an te nt iro Ea No Co (60,70] es Ny Ce n- (50,60] Na (40,50] No (30,40] [20,30] 2005/06 2015/16 e) Access to improved sanitation f ) Household consumption in Nairobi’s informal settlement and non- informal settlement areas, 2015/16 100 95 Consumption .00015 71 78 72 71 Proportion of households, % 75 66 66 68 61 51 58 .0001 Kernel density 49 44 50 51 50 43 41 33 33 36 .00005 19 20 25 0 0 0 10000 20000 30000 40000 50000 l rn t rn l y rn Na a No bi DI DI na ra as lle z iro ste ste te an NE NE nt Co tio Va es Ce Ny n- Ea Ea Per adult-equivalent monthly consumption Na W ft Ri rth No 2005/06 2015/16 Informal settlement Non-informal settlement KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xi Executive Summary they did in urban areas between 2005/06 and 2015/16, poor, non-beneficiary households. The programs are but the current urban-rural differences are still very effective in fostering food security, improving school large – 43 percent in urban areas, and 57 percent in rural enrolment and reducing the probability of children areas (Figure 5a). Poverty and vulnerability are highly working. Despite recent efforts by the government, correlated, but over one third of non-poor Kenyans are these programs have limited geographical coverage classified as vulnerable. Vulnerability rates vary widely and remain small in scale (Figure 5f ). by county, being highest in the north and east of the country (Figure 5b), and by household characteristics, WOMEN ARE LEFT BEHIND IN MANY AREAS K with high vulnerabilities particularly among those that enyan women are disproportionately affected are engaged primarily in agriculture, and those with by poverty during the core productive and low educational attainment (Figure 5c). Many of these reproductive years, especially if they experienced non-poor but vulnerable households are clustered just a marital dissolution. As in other African countries, above the poverty line, meaning that even a moderate Kenyan women are more likely to live in poor shock could push them below the line. households than men, starting in their mid-20s and continuing until their 50s (Figure 6a). Moreover, women When faced by shocks, many poor and rural who are separated, divorced or widowed are more likely households often resort to coping strategies with to be poor (compared to men), face higher prevalence adverse implications for future wellbeing. The rates of physical violence (compared to other women) overall prevalence of both economic and agricultural and are disproportionately affected by HIV/AIDS. Kenya shocks declined between 2005/06 and 2015/16. is also among the few African countries with gender However, the incidences of certain kinds of shocks inequality in formal inheritance rights – i.e. the Law affecting agricultural households went up. Agricultural of Succession Act. Gender gaps exist also in terms of households were far more likely to report crop losses access to ICT and financial services, though levels of from preventable causes such as crop diseases or access are high by regional standards. pests in 2015/16 than they were in 2005/06 (Figure 5d). The most common response of poor households Girls and women continue to be disadvantaged in after experiencing a shock is to reduce consumption, education and health in some regions. Girls have while for the richest households the most common lower enrollment rates and educational outcomes than response is to use savings (Figure 5e). The inability of boys in Northeastern Kenya and the coast – but boys’ poor households to cope with adverse shocks and disadvantages emerge in parts of Central and Western their limited financial resilience has severe long-term Kenya (Figure 6b). Girls dropping out of secondary school implications, particularly when they are forced to cut are more likely to be married and to have given birth spending on food, education and health, curbing than girls still attending school.6 Despite improvements human capital accumulation. in girls’ education, adult women are twice as likely to be illiterate as adult men (Figure 6c), reflecting historical Kenya expanded its social protection programs, but gender inequalities, which continue to put women at coverage and scale remain limited. Over the last few a disadvantage in terms of labor market opportunities. years, Kenya expanded its social protection programs, Even though maternal mortality declined since 2005, spending about 0.27 percent of GDP in 2015, well Kenyan women face a staggering lifetime risk of 1:42 of below the average of 1.6 percent of GDP in low- and dying due to complications of pregnancy or child birth middle-income countries. The programs are generally (Figure 6d). well targeted: only 23 percent of grant-receiving households had at least one resident member who 6 Secondary drop out is defined as having attended secondary school Form 1-3 during the last school year, but no longer attending school was employed. This is in contrast to 48 percent in poor, during the current school year. Note that there are only few cases of secondary drop outs captured by the KIHBS N=70), which limits the non-beneficiary households, and 54 percent in non- analysis that can be performed. xii KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Figure 5: Poverty and vulnerability in Kenya a) Poverty and vulnerability from 2005/06 to 2015/16 b) Geographic variation of vulnerability in 2015/16 80 70 60 Percentage of population 50 40 30 20 10 0 Poor Vulnerable Poor Vulnerable Poor Vulnerable Vulnerability rate >90% 80.1% to 90% 75.1% to 80% 70.1% to 75% 65.1% to 70% 60.1% to 65% Total Urban Rural 50.1% to 60% 35.1% to 50% 2005/06 2015/16 10% to 35% c) Relative vulnerability rates, by household characteristics d) Shock prevalence for agricultural households only 90 40 80 35 70 30 60 25 Percentage Percentage 50 20 40 15 30 10 20 5 10 0 Agriculture Manufacturing Services Construction Rural Urban No education Primary education Secondary education Tertiary education Female Male Drought or ood Crop disease/pest Livestock death/theft Severe water shortage HH business failure Loss of salaried employment End of regular assistance/aid Large food price rise Large agri. input price rise Dwelling damaged/destroyed Household Household head Agricultural shocks Economic shocks National vulnerability rate 2005/06 2015/16 e) Coping mechanisms in 2015/16, by poverty quintile f ) Number of households receiving cash transfers Used savings 800 43% 600 29% Reduced consumption Thousands 26% 25% 400 Help from family 17% 14% Borrowed 11% 200 9% 8% Sold assets 9% Help institution 11% 0 2013 2014 2015 2016 Quintile 1 Quintile 5 OVC OPCT HSNP PWSD Total Source: Own calculations based on KIHBS 2005/06, KIHBS 2015/16 and Kenya’s Single Registry for Social Protection. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xiii Executive Summary In the Northeast, women often have a lower Reducing the gender gap can unleash Kenya’s participation in the labor market because of productive potential household work. In 2015/16, Kenya had a female Women’s productivity can be increased by abolishing labor force participation rate of 71 percent for the discriminatory practices in women’s access to core working-age population (15-64 years), compared productive assets. Gender biased legislations, such as with a male labor force participation rate of 77 percent the differential treatment of male and female surviving (Figure 6e). However, there are significant regional spouses under the Law of Succession Act, should differences – female labor force participation is high be eliminated. Savings products with an element of in Central and Western Kenya, but much lower in the illiquidity and soft commitment can increase women’s Northeast (Figure 6f ). Due to traditional gender roles, savings to unlock investments into productive assets. women spent a significant amount of time on unpaid Information campaigns as well as mentoring programs care work within the household. Every child aged 0-5 can help to overcome sectoral segregation locking years reduces women’s probability to be in the labor women into low-productivity jobs. Technological force by over 2 percent. change has the potential to disrupt traditional patterns of sectoral segregation, such as Uber and other ride- There are gender gaps in access to productive hailing services opening up opportunities for women in resources and sectoral segregation. In line with the traditionally male-dominated sectors like transportation. international experience, male wage workers earn 30 percent higher wages and salaries than female Closing the gender gap and creating equal wage workers. This is likely explained by the fact opportunities for boys and girls requires, among that women are disproportionately employed in other interventions, targeted investments in agriculture and services, while men have a higher education and health. Programs subsidizing the share of employment in the industrial sector. Also, direct or indirect cost of education can be effective in profits of male-run household enterprises are about increasing enrollments and educational performance of twice as high as profits of female-run enterprises boys and girls. Increased secondary school enrollment and households for which women are the primary among adolescent girls may also delay fertility decisions. decision-makers in agricultural activities achieve In health, further initiatives to increase access to and lower yields (maize, beans) than other households affordability of reproductive health care services are (Table 3). Only 12 percent of women aged 20-49 years important to reduce maternal mortality, especially in report owning any land on their own, compared with Kenya’s arid and semi-arid regions. Public investments in 39 percent of men. Also, Kenya is among the few services for care can reduce time constraints of women. African countries with gender inequality in formal Scaling up care services for children, however, requires inheritance rights, for example with respect to the innovative approaches, combining public and private Law of Succession Act. sources of funding. Table 3: Monthly earnings in KSh, by gender Male Female Ratio male-to-female Mean 18,276 14,075 1.30 10th percentile 3,000 2,000 1.50 Median 10,000 6,500 1.54 90 percentile th 43,300 35,000 1.24 Source: KIHBS 2015/16. xiv KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Figure 6. Gender gaps in Kenya a) Poverty rates and gender-poverty gap b) Female and male poverty rates by marital status, 2015/6 60 50 50 40 40 Poverty rate 30 Percent 30 20 20 10 10 0 0 Monogamously Polygamously Separated or Widowed or Never 0-4 5-9s 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ married or married divorced widower married cohabitating Age Poverty, male Poverty, female Males Females c) Gender parity index for gross primary enrolment rates6 d) Gender parity index for gross secondary enrolment rates7 (1.4,1.5] (1.3,1.4] (1.2,1.3] (1.2,1.3] (1.1,1.2] (1.1,1.2] (1,1.1] (1,1.1] (.9,1] (.9,1] (.8,.9] (.8,.9] (.7,.8] [.7,.8] (.6,.7] [.5,.6] e) Literacy rates, by gender and county f ) Maternal mortality ratio. 1 .8 .6 .4 .2 0 Nairobi Nyeri Mombasa Kiambu Kisumu Nandi Nakuru Machakos Trans Nzoia Kisii Nyandarua Vihiga Uasin Gishu Bungoma Kirinyaga Taita Taveta Siaya Muranga Embu Nyamira Kericho Homa Bay Migori Elgeyo Marakwet Bomet Baringo Kajiado Makueni Lamu Tharaka Nithi Kakamega Kitui Busia Meru Laikipia Tana River Isiolo West Pokot Kwale Samburu Mandera Garissa Wajir Marsabit Turkana Narok Kili (3000,4000] (2000,3000] (1000,2000] (800,1000] (600,800] (400,600] (200,400] [0,200] Male Female Based on 2009 Census. 7 The gender parity index is defined as the ratio of female to male enrollment rates. A value above (below) unity indicates that girls have higher (lower) levels of enrollments. 8 Ibid. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xv Executive Summary g) Changes in employment, by gender h) Gender gap in labor force participation 80 60 Percent 40 20 0 (40,50] (30,40] (20,30] Male Female Male Female (10,20] (0,10] 2005/06 2015/16 (-10,0] (-20,-10] (-30,-20] (-40,-30] Wage Enterprise Any employment [-50,-40] Source: Own calculations based on KIHBS 2005/06, KIHBS 2015/16, KNBS (2012) and Global Findex 2014. ACCELERATING POVERTY REDUCTION Policies aimed at increasing the adoption of improved agricultural inputs by small farm holders would help Improve the productivity of the agricultural sector and to increase their income and help to further reduce enhance access to markets in rural areas poverty. Extension services programs and educational I ncreasing agricultural productivity remains a potential pathway out of poverty for many households. In Kenya, more productive farmers are less campaigns, together with a competitive inputs markets, are some alternatives. likely to be poor (Figure 7a,b). This correlation between Similarly, agricultural commercialization is also farm productivity and poverty constitutes promising associated with better living conditions. For farmers, evidence that an enhancing agricultural yields a higher degree of commercialization is associated with could lead to a reduction of povertys. However, little higher living standards, as can be observed in Figure progress has been made in terms of raising agricultural 7e. Thus, investments in infrastructure and access to productivity in the last ten years. This is especially true for the production of maize, Kenya’s main food information and communication technologies, so that staple, and commercial crops such as coffee. Increased farmers can more easily reach their clients and can more efficiency in the production of beans appears to be the easily buy the inputs for agricultural production, are an only exception. As a result, agricultural productivity has important policy area to focus in order to accelerate the not been contributing to poverty reduction in rural reduction in poverty. Kenya, a marked difference from the experience of other countries in the region, such as Ethiopia. Policymakers may need to allocate more resources to enhance farmers’ productivity and make sure Technology adoption is the main factor associated that the current spending is efficient and providing with higher productivity, according to analysis using the highest returns. Around 2 percent of total public farm level data. Farmers that applied chemical fertilizer, expenditure was allocated to agriculture in 2016/17, for example, experienced a 20-25 percent increase in even though the sector accounts for 25 percent and maize yield. Moreover, farmers who planted improved 60 percent of the country’s GDP and employment, maize seeds experienced 26-32 percent higher respectively (World Bank 2018). This prevents the productivity compared to those that used traditional country from investing effectively in smallholder low-yield seeds. Despite the yield-enhancing effects of agriculture and provide services to improve basic fertilizer and seeds, the share of farmers adopting these crop yield. There is also a need to asses if the current inputs has not changed much between 2000 and 2010. spending is efficient, taking into account that spending xvi KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Figure 7: Socio-Economic indicators of Rural Kenya a. Maize yield and poverty rate, 2015/16 b) Bean yield and poverty rate, 2015/16 2500 600 Eastern Rift Valley 2000 Rift Valley 500 Yield (kg/hectare) Yield (kg/hectare) Western 1500 Central Central Eastern Nyanza Nyanza 400 1000 Western North Eastern Coast North Coast Eastern 500 300 15 35 55 75 15 25 35 45 55 Poverty rate , % Poverty rate, % c) Poverty and the sale of farm produce in rural Kenya 80 60 Poverty Rate, % 40 20 0 0 0.5 1 Proportion of harvest sold 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16, WDI data. on public goods in this context (e.g. research and More and better jobs, along with infrastructure development, extension services, etc.) has been investment, is required in urban areas proven to be more productive than spending on Many workers remain in volatile and low quality jobs private goods (e.g. fertilizer subsidies). In addition in urban areas, despite a decline in unemployment. there is space to reform the input subsidy program by Unemployment rates dramatically dropped in urban ensuring that the program is targeting small farmers areas (Figure 8e), in tandem with an increase in labor and facilitating technology adoption among them. force participation rates. However, a large fraction of the Moreover, investment in irrigation schemes have a urban poor, women, and youth are unemployed.10 The high rate of return9 and could reduce dependence on existing jobs in urban areas are casual and do not offer rainfall. The fact that food security is one of the Big Four long-term security. Nearly 90 percent of construction priority areas outlined by the government (together jobs in Nairobi are casual work, resulting in 41 percent with manufacturing, affordable housing and universal of the poor being casual workers as opposed to 9 healthcare) is a positive sign and the concrete policies percent for the non-poor (Figure 8d). These jobs do not that will be proposed should be scrutinized carefully. provide long term security, and may not conduce to better job opportunities in the future. 10 In Nairobi, for example, more than 20 percent of the poor are 9 World Bank, 2018. unemployed. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xvii Executive Summary Figure 8: Urbanization remains a challenge for poverty reduction a) Economic sector of urban workers b) Urban unemployment rates in 2005/06 and 2015/16. 100 50 45 45 90 40 80 35 70 30 28 27 27 60 62 24 69 25 22 78 75 73 20 50 80 19 19 19 20 16 15 16 40 15 11 13 13 12 10 10 30 15 10 9 8 8 8 7 7 7 8 9 10 5 20 4 5 9 9 0 10 17 10 10 Non - Poor Non - Poor Non - Poor Non - Poor Poor Poor Poor Poor 15 All All All All 6 8 6 8 0 Poor Non-Poor All Poor Non-Poor All 2005/06 2015/16 All urban Nairobi Mombasa Other urban Agriculture Manufacturing Construction Other Services 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06, KIHBS 2015/16, DHS 2014 and Cities Baseline Survey 2013. It is important to leverage the potential of main form of motorized transport in Nairobi, a worker urbanization for poverty reduction through more can reach only 4 percent (within 30 minutes) and 25 and better quality jobs. Manufacturing and high value- percent (within 60 minutes) of existing jobs, while added services jobs are still lacking in urban Kenya in Greater Dakar, for example, it allows access to 52 despite the fact that they can play an important role percent of existing jobs within 1 hour of travel. Thus, in providing economic opportunities, especially for the investments to lower the transportation costs and young urban population. The urban poor face several shortening the distances between the individuals and challenges in terms of job availability and accessibility, the economic opportunities is necessary. Moreover, in particularly in informal settlement areas. More and some areas investments in physical infrastructure lag good quality jobs in the manufacturing sector can help behind the needs of the urban population. While the to improve the incomes of the urban poor, if paired with share of urban population with access to improved investments in transport infrastructure and skills. Some sanitation facilities and electricity increased during of the areas of focus should be competitiveness and the last decade, the share of those with improved capabilities. Industrial enclaves can help address some water access dropped in some areas, indicating that of the structural bottlenecks that affect manufacturing urbanization outpaced infrastructure provision. competitiveness and help attract foreign direct investment. Worker capabilities can be enhanced by Broader affordable housing can reduce housing costs prioritizing literacy, numeracy and ICT skills and by in urban areas, relaxing the budget constraints. The improving the training programs in collaboration with high costs in terms of food and housing are curbing the the private sector. As one of the Big 4, manufacturing purchasing power of the less well off. Targeted policies has great potential to help improve the livelihoods of to ensure affordable housing can help them to escape the urban poor. poverty, which will be hopefully part of the set of policies implemented under the Big 4 umbrella. In the Improved connectivity through investments in case of informal settlements, localized interventions are infrastructure and the provision of high quality required to ensure that informal settlements function public services is also crucial. High transportation as a place of opportunity rather than as a poverty trap, costs squeeze the budget of urban households, limiting including better service provision. access to economic opportunities. Using minibus, the xviii KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary Improving the provision of education and health attendance and, eventually, student’s outcomes. Along Increasing secondary school enrollment among with greater local oversight, schools could be given the poor requires demand-side interventions. While more resources and greater independence on how to enrollment in secondary has increased among the use them. Increasing the capitation grant, along with poor, significant gaps persist. The evidence presented greater autonomy to school committees to recruit, in this report and numerous academic studies suggest retain, and promote teachers, has the potential to that increasing enrollment in secondary education in improve teacher performance and to lower school Kenya requires primarily demand-side interventions drop-out. The potential of a greater involvement of aimed at loosening the financial constraints that less private providers could also be explored. well-off households face. Cash transfers have already proven effective in increasing enrollment rates. In terms of the provision of universal health coverage Similarly, encouraging on-time primary enrollment and (UHC) one of the ‘Big Four’-priorities, it must be the supporting the primary-to-secondary transition noted that the incidence of catastrophic health could also contribute to raise the enrollment rates at expenditures has decreased recently, which may the secondary level. disincentivize voluntary enrollment going forward. Only around 20 percent of the population are covered Enhancing the quality of education and aligning the by health insurance, with large differences between teachers’ incentives with student learning requires a the poor and the better-off and between rural and series of interventions combined with a close scrutiny urban areas. At the same time, there is evidence that of the recently introduced monitoring and evaluation the incidence of catastrophic health expenditures system. Greater reliance on contract teachers to initially has declined over time and that households rarely fill vacant positions, subsequently moving to an ‘up-or- resort to adverse coping strategies, such as selling out’ promotion system, in which the best-performing their assets, to finance healthcare. This is in line with contract teachers are promoted to public school the removal of user fees in 2013 for a range of public teachers, may have large potential benefits. Contract health services, including birth deliveries, and with the teachers have average levels of subject and pedagogical overall improvement of living standards and health knowledge and lower rates of absenteeism, without amongst Kenyans. The implication maybe that those in being paid a premium. In any case, the system the informal sector have little incentive to voluntarily requires all teachers to be systematically and regularly insure, making it harder for the government to expand evaluated, for benefits to be tied to performance, and a health insurance coverage. credible threat of discontinuation of employment. The effectiveness of the recently introduced monitoring Similarly, given that the poor are more likely to and evaluation systems should be closely monitored. depend on public health services than the rich, While they have the potential to improve teacher effort, recent disruptions in supply during labor disputes it is not clear whether head masters and deputy head between the government and public-sector unions masters are best placed to monitor teacher presence disproportionately affect the less well-off. The string and performance. of recent health worker strikes in the public sector that culminated in major walk-outs at the end of 2016 and The quality of education would benefit from the in mid-2017, resulted in disruptions that likely affect involvement of local stakeholders, particularly the poor disproportionately. Health workers’ salaries in parents, and from enhanced school governance. Kenya remain high by regional standards, despite their Empirical evidence suggests that the local knowledge recent sluggish growth in real terms. This is particularly of stakeholders, particularly parents, may play a key true for health workers in the public sector, which earn role in monitoring teachers at the school-level. Getting a substantial premium, in part because of a lengthy list local stakeholders involved may help improve teacher of allowances that account for a significant portion of KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xix Executive Summary their total pay. There should be a more open, informed Devolution has the potential to address some of the and transparent debate on adequate remuneration of development challenges, and its implementation public health workers, in order to help prevent these should be carefully monitored so that the necessary disruptions in the future. adjustments can be incorporated. Devolution has the potential to address the wide spatial variation in Finally, the sustainability of health financing, wellbeing across counties and regions and improve particularly of priority programs, should be a the accountability in service delivery. Decentralization priority. A recent report has highlighted funding gaps seems like the right path to address these inequities, in all five priority programs analyzed (World Bank but counties have various degrees of institutional 2018) and despite a falling share, healthcare financing capacity and economic development and must be in Kenya still relies significantly on donors. One provided with the resources required (both human and vehicle to increase revenues is through an increase financial). At the same time, outcomes in all sectors of memberships and contributions to the National should be closely monitored and counties should be Hospital Insurance Fund (NHIF). One alternative to hold accountable for their performance. In the coming increase funding would be by introducing ‘health years, as more data becomes available and enough taxes’ on food and drinks that contain high amounts time has passed for the effects of decentralization to be of saturated fat, sugar, salt, or other unhealthy apparent, more studies and research should focus on ingredients, which would also address the problem of the effects of devolution. rising obesity among urban, better-off Kenyans. The decade-long gap between the two most recent Expand Social Protection programs and provide the household consumption surveys makes it difficult to foundations for devolution to work monitor poverty and analyze the impact of policies. Expanding assistance to vulnerable households While Kenya’s most recent household consumption through existing or new social protection programs survey was implemented in 2015/16, the previous can reduce vulnerability. The effort that has been survey dates back to 2005/06. Without more regular made to coordinate and harmonize social protection data collection, it is very difficult to monitor progress programs, combined with the creation of a registry of in terms of poverty reduction, and to assess the impact beneficiary households means that the country is well of policies and programs. An improved monitoring placed to expand assistance to vulnerable households, system should be put in place, ideally one that provides which would benefit greatly from this potential information at the county level and that can inform the expansion. Furthermore, specialized programs to ongoing devolution process in Kenya. The Government mitigate shocks can reduce vulnerabilities. For example, of Kenya’s plans to establish a continuous household the introduction of emergency cash programs can have survey by the KNBS are a good step in the right direction the potential to offset some of the negative effects to design and implement policies based on evidence. of shocks such as droughts and floods, and protect vulnerable households from resorting to negative scoping strategies with long-term impacts like selling productive assets. xx KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Executive Summary ORGANIZATION OF THE REPORT Chapter 4 analyzes rural livelihoods, and explores T his report is organized as follows. Chapter 1 various factors that might have contributed or provides an overview of macroeconomic drivers of hindered the reduction in rural poverty. More economic growth and its fiscal implications. The trends specifically, it examines the role of diversification into of poverty (under the international poverty line) are non-farm employment to differentiate the contributions compared with other countries alongside indicators of on income-diversification and agricultural income. An non-monetary deprivations to provide an international analysis of rural-urban migration sheds light on the role benchmarking. Kenya’s context is discussed in an analysis of migration for poverty reduction. In the second part, of the political economy, with a focus on the two central the chapter delves into agricultural production and themes of political competition and devolution. This is productivity by analyzing its trends and its potential complemented by an analysis of Kenyans’ perceptions, impact on poverty reduction. The analysis concludes embedded in an international comparison. with a discussion of commodity prices and how they affected rural poverty. Chapter 2 first documents the progress made by Kenya in terms of the monetary measures of poverty, The linkages between urbanization and poverty during the period of focus of this report, 2005/16 to with a particular focus on the challenges faced by 2015/16. It analyzes the trends in terms of the national the urban poor are examined in Chapter 5. It reviews poverty headcount rate, other related indicators (such Kenya’s urbanization trends and examines how the as the depth and severity of poverty) and the incidence geographic patterns of poverty changed during the of food and extreme poverty, as officially defined by the last decade. In so doing, it assesses the contribution Kenya National Bureau of Statistics (KNBS). The chapter of urbanization to poverty reduction in the country. It then turns to examine the incidence of consumption also assesses urban poverty from both a monetary and growth, and how this is reflected in terms of an array a non-monetary perspective, in view of its geographic of inequality indicators. It also examines the factors heterogeneity. Thirdly, the chapter analyzes urban labor behind Kenya’s success in reducing poverty, relying markets to figure out opportunities and challenges on decomposition analysis and the finding of various faced by the urban poor. Finally, it takes a closer look studied on the impact of mobile money in the at informal settlements—mainly in Nairobi—, where wellbeing of the population. The chapter concludes urban poverty is concentrated, showing a stark contrast by providing a profile of the poor, in an attempt to in living conditions between informal settlement identify the factors that may be limiting their economic and non-informal settlement areas and the limited opportunities and overall wellbeing. residential movements between them. A synthesis of what is known about the gender- Recent developments in Kenya’s education sector poverty nexus in Kenya is presented in Chapter 3. and their relationship to poverty and equity are It starts with a basic profile of poverty and gender. analyzed in Chapter 6. It takes stock of the recent Next, following the framework of the 2012 World trends in access to education services as well as Development Report on Gender (World Bank 2011) it their quality and examines the incentives in place to then proceeds to analyze gender gaps in endowments, produce quality education for all. The chapter provides gender inequality in economic opportunities and background information on Kenya’s education system, gender differences in voice and agency. Within each while analyzing current levels and recent trends in of these sections, the chapter also provides a brief access and enrollment and their links to poverty discussion of possible policy options to narrow – and and equity. It then shifts the focus from access and ultimately close – gender gaps and promote a more enrollment to learning outcomes and then analyzes equitable society. inputs into the educational production and their distribution, including physical inputs and the ability of KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead xxi Executive Summary teachers to deliver quality education. Finally, it discusses in access and uptake of health services that persist school governance, especially teacher incentives. today, including significant geo-spatial variation. In addition to the analysis of various data sources11 Finally, it shifts the focus towards providers and inputs including administrative data, household surveys, and into health production, including provider knowledge school assessments, the analysis draws heavily on and physical inputs. The analysis relies on a wide array recent academic studies. of microdata sets and administrative data as well as a review of academic studies. Chapter 7 analyzes levels and trends in health outcomes, uptake of services, and health equity. With the aim of understanding how to address It provides background information on recent vulnerability in Kenya and make sure that the developments and initiatives in Kenya’s health sector, country enters a sustainable path of poverty including the devolution of health service delivery to reduction, Chapter 8 examines and analyze changes the counties, the removal of user fees, health workers in the vulnerability profiles for Kenya in 2005/06 and strikes, and universal health coverage (UHC), one in 2015/16. Moreover, it analyzes and compares the component of the ‘Big Four’-agenda. It documents welfare shocks that affected households in 2005/06 the rapid pace at which Kenya in recent years made and 2015/16, as well as which coping strategies were progress in health outcomes, particularly under-five adopted in the face of these shocks. Finally, the chapter mortality, and in the uptake of certain health goods assess the coverage and effectiveness of Kenya’s social and services. While these improvements have often safety net programs, while also measuring their impact been equitable, the chapter also documents inequities on different measures of household welfare. 11 The report relies on the 2005/06 and 2015/16 Kenya Integrated Household Budget Surveys (KIHBS), the 2012 Service Delivery Indicators (SDI) for education, a facility-based survey of teachers, students, and schools, and the Uwezo data, annual learning assessments. In addition, data from the World Development Indicators (WDI) and the Kenya Economic Survey (KES) was used. xxii KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead CHAPTER I KENYA IN CONTEXT SUMMARY Since 2005, Kenya has experienced resilient economic growth despite several shocks, contributing to a steady, though moderate, reduction in poverty. Economic and political shocks in the past decade have included electoral violence, drought, and an overhaul of the centralized political system. Perceptions of democracy and trust in the government have suffered over the years, following contested elections and corruption concerns. As Kenya begins its next five-year development strategy, a larger emphasis on redistributive policies and the devolution process is necessary to bring the country closer to eradicating poverty by 2030. Kenya’s economic growth has exceeded average growth in sub-Saharan Africa in the past decade. Growth averaged 5.3 percent in the 2005 to 2015 period, primarily driven by the services sector on the supply-side and household consumption on the demand-side. In particular, the mobile phone revolution contributed to an expansion of the financial services sector by increasing access to credit and providing services to previously unbanked households. The country faced two major economic shocks between 2005 and 2015. The first was due to electoral violence in early 2008 that compounded the effects of the global financial crisis. The government’s quick policy actions through a stimulus package helped restore growth in 2009 and 2010. A second shock hit the country in 2011 after an increase in international oil prices, combined with a drought in the Horn of Africa that reduced agricultural production in the region. The government’s development policy has been guided by Vision 2030, Kenya’s long-term development plan. Policies were designed to increase aggregate demand, with a focus on supply-side investments in infrastructure projects such as rail transport and renewable energy. While revenue was volatile between 2005 and 2015, the pace of public spending steadily increased and consistently exceeded revenue collections. This put pressure on the fiscal deficit, which increased from 4.7 percent of GDP in 2005/06 to 8.2 percent in 2015/16. Education spending was the largest beneficiary of social sector spending and had a stable upward pace, largely the result of a free universal primary education (FPE) policy. In the past decade, Kenya has experienced a moderate reduction in poverty. As of 2015, about one third of the Kenyan population lives below the international poverty line of US$ 1.90 a day. Poverty declined from 43.6 percent in 2005 to 35.6 percent in 2015. Poverty reduction has been driven by improvements among the poorest of the poor, and particularly among households engaged in agriculture. Agricultural households remain vulnerable to climate and price shocks, as growth in the sector has a strong impact on household consumption. Kenya compares favorably in monetary and non-monetary poverty with peer countries, but is not yet on the same level as other lower-middle income countries. At the lower-middle income line of US$ 3.20 a day, both the rate of poverty and the depth of poverty are worse in Kenya than in countries having similar levels of wealth per capita. More than two thirds of the Kenyan population lives below the US$ 3.20 a day line. Poor households are often deprived on multiple dimensions, with the most common being access to services such as improved water and sanitation. Kenya lags behind peer countries in access to improved water sources, but performs fairly well on education and health indicators. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 1 Kenya in Context Politics and political institutions in Kenya were until very recently influenced by centralized power residing in the presidency and executive branch. A strong political consensus emerged from the need to devolve powers away from the executive and the central government, with a view to making Kenya’s democracy and development more inclusive. Political and civil society efforts culminated in a constitutional referendum in 2010, which led to a “big bang” political and fiscal decentralization that devolved power to 47 counties created from the former eight provinces. Crucial issues however must still be resolved in order for devolution to have its full impact and for the citizenry to trust the process. So far, inherent disparities between counties determine developmental outcomes even if fiscal allocations are equitable. The disputed presidential elections in 2017 renewed the focus on democratic institutions in Kenya and perhaps on the current shortcomings of devolution as implemented. The Afrobarometer Survey captures key perceptions of Kenyans on democracy, the nature of governance, and on participatory politics. Support for democratic norms and processes remained high even through the political volatility and electoral disputes over the past decade – concluding with the 2017 elections. Perceptions in 2008 however did reflect disillusionment regarding the true extent of democracy. Responses in 2016 show low levels of trust in public officials such as the police, government workers and members of parliament (MPs) who are themselves seen to be involved in corruption to some degree. 1.1 MACROECONOMIC PERFORMANCE 2011). Low-income households were affected the most OVER THE LAST DECADE (19.6 percent overall inflation year-on-year) compared 1.1.1 Resilient economic growth to high-income households (14.5 percent overall inflation year-on-year), given smaller expenditure E conomic growth between 2005 and 2015 remained resilient, despite several challenges. The Kenyan economy recorded an average annual real shares for food and transportation for the latter group. The effects of the shocks in 2011 continued into 2012, causing a dip in annual economic growth to 4.6 percent growth rate of 5.3 percent between 2005 and 2015. before rebounding to 5.9 percent in 2013. Overall growth was volatile, including both years of high growth (6.9 percent in 2007 and 8.4 percent in Real GDP per capita growth mirrored economic 2010) and years of low growth (0.2 percent in 2008). The economy faced two major shocks in this period. First, growth (Figure 1.1). GDP per capita growth rose from electoral violence in early 2008 compounded the initial 2.8 percent in 2005 to 4.0 percent in 2007, then fell to effects of the global financial crisis, reducing annual -2.5 percent in 2008. Low growth in the agriculture economic growth to 0.2 percent. The government sector following post-election violence in 2008 was the took quick policy action through a stimulus package, main driver of the decline in per capita growth in 2008. which contributed to an increase in annual growth to Per capita growth peaked at 5.5 percent in 2009. This 3.3 percent in 2009 and 8.4 percent in 2010. A second can be attributed to a recovery in the agriculture sector, dual shock affected the economy in 2011 when implementation of the government economic stimulus international oil prices increased by 37.4 percent while and a recovery in the tourism sector. Since 2009, per a drought in the Horn of Africa reduced food output.12 capita growth has been moderate, reaching 3.2 percent The escalation in food and fuel prices led to an increase in 2016. in overall inflation (18.9 percent year-on-year as of Q3 12 Kenya Economic Update Edition No. 5. 2 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.1: Kenya’s GDP growth from 2005 to 2015 10.0 8.0 6.0 Percent 4.0 2.0 0.0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 -2.0 -4.0 GDP per capita growth GDP growth Source: KNBS. 1.1.2 Kenya’s performance vis-à-vis the region 2009 from 5.4 percent the previous year. The growth Although economic growth in Kenya exceeded rate in Kenya however took an upward turn in that average growth in Sub-Saharan Africa, Kenya’s year, reaching 3.3 percent thanks the introduction performance lagged behind that of its peers in of a stimulus package designed to counteract the East Africa (Figure 1.2). In Sub-Saharan Africa, annual shock. Even though the performance was better than growth averaged 4.9 percent between 2005 and 2015, average for Sub-Saharan Africa, Kenya’s growth was 0.4 percentage points lower than growth in Kenya. consistently below that of its East African peers, namely However, Kenya’s growth was, on average, lower than Rwanda, Tanzania and Uganda (9.3, 6.6 and 6.7 percent that of Sub-Saharan Africa between 2006 and 2008.13 respectively). Higher growth in these East African Suffering the effects of the 2008 financial crisis, Sub- countries can be explained by the lower base of their Saharan Africa’s growth dropped to 2.9 percent in economic development compared to Kenya. Figure 1.2: Annual GDP growth for Sub-Saharan Africa and selected countries, per year and between 2005 and 2015 10 12 10.6 8 10 9.3 8 GDP growth (%) GDP growth (%) 6 6.7 6.6 7.0 6 5.3 4.9 4 4 2 2 0 0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Sub-Saharan Kenya Uganda Tanzania Rwanda Ethiopia Ghana Africa Kenya Sub-Saharan Africa Source: World Bank – Mfmod. 13 Growth in the sub-Saharan region pre-2008 financial crisis was driven by high commodity prices. Since Kenya’s main exports are horticulture, tea and coffee, Kenya did not benefit very much from the commodity price boom. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 3 Kenya in Context 1.1.1 Sectoral supply-side growth growth in the manufacturing subsector slowed down Growth was primarily driven by the services sector, in 2012, following the drought in 2011. The drought fueled particularly by ICT and financial services. The had a moderating effect on the production of hydro- mobile phone revolution increased the number of power, which in turn increased production costs in mobile subscribers to 40.2 million in 2015/16 (from manufacturing due to the use of imported backup 17.4 million subscribers in 2008/09), while Internet thermal-generated power. subscriptions jumped to 26.8 million in 2015/16 (from 1.8 million subscribers in 2008/09).14 The mobile Performance of the agricultural sector was dependent phone revolution also contributed to an expansion of on rainfall. Agriculture, which contributes about 23 the financial services sector, driven by the ability to percent to GDP and employs the bulk of the working provide financial services to previously unbanked population, is also the sector that has contributed the households in the form of the mobile payment least to GDP growth, at 0.8 percent on average (Figure system M-PESA, including credit facilities to mobile 1.3). Performance in this sector was highly correlated phone subscribers. with adequate rain, and years of low rainfall exhibit low growth rates growth rates, such as 2011 when rainfall Kenya’s relatively well-developed financial sector was low and growth was merely 2.4 percent (Figure spurred growth through an increase in access to 1.4).16 Low agricultural production affects food prices.17 credit.15 Credit growth to the private sector averaged Another factor that negatively affected the agricultural 19.6 percent between January 2006 and December sector, mainly after the 2008 financial crisis, was a 2015, comparable to credit growth in regional peers hampered demand for horticultural products in the Uganda and Tanzania. Increased lending across sectors euro area. In addition, Kenya’s loss of competitiveness was broad-based, with households/personal loans and within the East Africa region, a consequence of lower the construction sector having a higher concentration. productivity, has seen regional demand for agricultural Credit to the private sector is an important measure products weaken.18 of the depth of financial systems, and consequently Figure 1.3: Contributions to GDP growth an important driver of short run growth. However, 4 credit growth to the private sector declined to a low of -1.3 percent in 2017. Interest rate caps introduced 3.0 Contribution to GDP growth (%) 3 by legislation complicate this declining credit growth, effectively weakening the private sector. 2 The industrial sector grew by 5.8 percent in 2016, mainly due to construction. The industry sector 1.1 contributed 1.1 percent, on average, to GDP growth 1 0.8 annually between 2005 and 2015. The construction sector recorded an average growth of 10.2 percent, 0 compared to only 3 percent for the manufacturing Agriculture Industry Services subsector. The manufacturing sector experienced a Source: KNBS. slowdown during the period of shocks (2008 and 2011). In 2008, uncertainty due to post poll violence saw 16 It is estimated that a 100mm decline in rainfall would reduce GDP growth by 0.5 percentage points. a decline in output in the manufacturing subsector, 17 For example, the year of the drought recorded annual inflation of 14 with growth slowing to 1.1 percent in 2008 and the percent. This was 7 percentage points higher and almost twice as high as the upper limit of the government target rate of 5 percent +/- 2.5 percentage subsector shrinking by -1.1 percent in 2009. Similarly, points. The high inflation levels were mainly driven by food inflation as the price of foodstuffs such as maize increased due to the drought. 14 Communications Authority of Kenya. 18 The Kenya Economic Update Edition 15 notes that Kenya’s exports to the 15 Beck and Fuchs (2004) note that for a country of its size, Kenya has a East African Region declined from a growth rate of about 29.5 percent in relatively well developed financial sector. 2007 to -8.9 percent in 2013. 4 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.4: Agriculture and GDP Growth Increasing infrastructure spending, coupled 2010 with increased private sector investment, drove 8 investment growth, which in turn supported 7 2007 economic growth. Investment contributed 1.5 percent 2006 6 2011 2013 2015 2005 to GDP growth, second to private consumption. The GDP growth (%) 2004 5 2012 1995 2001 government ramped up spending on investment to 1996 4 ease supply-side constraints, making capital the main 2009 1998 3 2003 1994 contributor to economic growth (Figure 1.5). Examples 2 1999 of infrastructure projects include the Thika Highway, 1991 1 2002 2000 the Northern and Southern bypasses and the Standard 2008 1998 1997 0 Gauge Railway. Infrastructure projects such as the -1 1992 ones undertaken by the government aim to increase -6 -4 -2 0 2 4 6 8 10 12 efficiency and reduce production costs, thereby Agriculture growth (%) creating incentives for domestic production. Source: KNBS. 1.1.2 Demand-side growth analysis Growth in the value of imports, which was much Consumption was the main driver of demand-side faster than growth in the value of exports, widened growth, with household consumption contributing the current account balance.21 The growth rate the largest share to GDP growth.19 With an annual of exports slowed from 19.0 percent in 2005 to 5.5 average growth of 4.1 percent between 2005 and 2015, percent in 2015. As a result, the share of the value of household consumption was the largest contributor to exports declined from 24.8 percent of GDP in 2005 GDP growth. A strong financial services sector, which to 16.7 percent of GDP in 2015. In contrast, the value improved access to credit for households, coupled of imports increased, averaging a growth rate of 9.4 with high remittances, supported consumption percent between 2005 and 2015. Consequently, the growth. Additionally, the government stimulus contribution of net exports to GDP averaged -1.3 program introduced in 2009 led to increased growth percent between 2005 and 2015 (Figure 1.6). in consumption, which contributed 5.7 percentage points to the GDP growth of 8.4 percent in 2010.20 Figure 1.5: Productivity and economic growth Figure 1.6: Demand-side contribution to growth between 2005 and 2015 10 5 4.1 4 8 3 6 Growth (%) 2 Percent 1.5 4 1 0.8 0 2 -1 0 -1.3 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 -2 Private Government Investments Net Capital Labour TFP GDP growth consumption consumption exports Source: KNBS and World Bank. Source: KNBS. 19 Note that final household consumption is calculated as a residual and is likely to include errors and omissions – KNBS. 20 The government stimulus was introduced in 2009 to counter the dual shock of post-election violence and the slowdown in demand for exports due to the global financial crisis. 21 The analysis uses values of exports and imports rather than volumes. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 5 Kenya in Context 1.1.3 Drivers of growth decline in human capital during this latter period was Labor was a key driver of real GDP growth. In the due to an increasing labor force (increase in population five-year period prior to 2005, labor contributed 62.5 age 15+)24, without a matching increase in human percent to growth. Labor’s contribution was the capital levels.25 largest, followed by total factor productivity (TFP) with 26.0 percent, and capital with 11.5 percent. However, TFP was a key driver of growth in countries with lower the trend was reversed in subsequent periods, per capita income. For Rwanda and Tanzania, whose with capital contributing 34.3 percent on average average per capita GDP between 2000 and 2015 was between 2010 and 2015, as the contribution of labor USD 567 and USD 712 respectively, TFP was a key driver declined to 43.4 percent during the same period of growth, contributing an average of 38.4 percent (Figure 1.7).22 Government policy increased spending to growth for Rwanda and 46.1 percent for Tanzania. on infrastructure and led to capital becoming a key Capital was the second most important contributor contributor to growth. to GDP growth in Rwanda, while for Tanzania, labor’s contribution to growth followed the TFP contribution Despite labor contributing the largest share to real to growth. In contrast to Rwanda and Tanzania, TFP was GDP growth, this contribution declined as human on average the lowest contributor to GDP growth for capital per unit of labor declined (Figure 1.7).23 The Ghana (24.4 percent), Kenya (25.7 percent) and Uganda contribution to GDP growth from human capital per (26.3 percent) between 2000 and 2015. While labor was unit of labor averaged 14.5 percent in the five-year the second most important contributor to GDP growth period between 2000 and 2005, but became negative for Kenya and Uganda, capital was the second most averaging -2.0 percent between 2005 and 2010. The important source of growth for Ghana (Figure 1.8). Figure 1.7: Contributions to real GDP growth 70 50 40.7 42.3 40.5 60 40 50 30 25.7 Percent 40 Percent 19.1 20 30 14.5 15.2 10 20 10 0 -2.0 0 2000-2005 2005-2010 2010-2015 -10 2000-2005 2005-2010 Capital Labor TFP Capital stock Labor TFP Human capital per labor Source: KNBS, Barro and Lee, and WDI. 24 Kenya’s youth population (15+) increased significantly in the period between 2000 and 2010. The government has put in place several programs through the Ministry of Sports and Culture that could take 22 The results are derived from the Long Term Growth model, a World Bank advantage of the increase in population/labor to increase growth. analytical tool. 25 Lucas and Mbiti (2012) note that education outcomes did not change 23 This analysis uses Barro and Lee’s definition of human capital, defined as significantly in Kenya even with increased access to education through returns to education per year of schooling. the Free Primary Education programs. 6 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.8: Contributions to GDP growth, regional Figure 1.9: Contributions to real GDP per capita growth comparison 2000 - 2015 2 60 50 Percentage points 40 1 Percent 30 20 10 0 0 Labor Labor Labor Labor Labor TFP TFP TFP TFP TFP Capital Capital Capital Capital Capital Stock 2005-2015 2005-2010 2010-2015 -1 Kenya Tanzania Uganda Ghana Rwanda Productivity Employment rate Participation rate Demographic change Source: KNBS and WDI. Source: KNBS and WDI. 1.1.4 Drivers of per capita GDP Growth26 asset and a binding constraint to development. The Productivity remains a key driver of per capita GDP country could benefit from demographic change growth, with potential for Kenya to reap benefits through an increase in working age population and from the demographic dividends.27 From 2005 to 2015, therefore the potentially larger labor force. However, if productivity contributed 1.75 percentage points, which this demographic dividend is not utilized by more jobs, was 81.1 percent of the total GDP per capita growth. the demographic change can have adverse effects on However, productivity’s contribution to per capita GDP productivity and per capita growth. growth was higher in the first half of the period at 1.60 percentage points, compared to 1.91 percentage Intersectoral reallocation was a key driver of GDP per points in the second half of the period (Figure 1.9). capita productivity as labor moved from agriculture This decline in productivity occurred in spite of an to services. Agriculture makes up about a quarter of the increase in the employment rate. One explanation for economy, contributing 25 percent to GDP. However, the decline in productivity could be due to the quality almost 60 percent of the labor force remains in the of jobs created, as jobs requiring only a low skillset are agriculture sector. Between 2005 and 2010, productivity unlikely to increase productivity substantially. in the agriculture sector declined, contributing -0.22 percentage points to per capita GDP growth. Lewis, The second half of the 2005 to 2015 period took in his structural adjustment model, points out that as advantage of a demographic change in Kenya. more labor (a variable resource) is put to work on land (a During this period, the population aged 15+ grew, fixed resource) – in this case agricultural land – marginal leading to an increase in GDP per capita growth of returns to labor will decrease.28 Since marginal returns 0.66 percentage points (Figure 1.9). However, the to other sectors are high, a wage premium in other demographic change implied that the contribution sectors relative to the agriculture sector can emerge. from the participation rate to GDP per capita between Between 2010 and 2015, the reallocation of labor 2010 and 2015 declined, since the population aged between sectors effectively increased productivity in 15+ increased much faster than the number of jobs the agriculture sector. Its contribution to GDP per capita created. Vision 2030, Kenya’s economic blueprint, growth reached 0.39 percentage points. Consequently, notes that rapid population growth can be both an 28 The Lewis structural change model of growth and development defines a dualistic economy where labor is defined as a variable factor input and land as a fixed factor. Initially, labor is concentrated in the 26 This study uses the shapely decomposition method to analyze the key agriculture sector. However, labor reallocates to the center to work in drivers of per capita GDP growth. the manufacturing sector which is more productive and offers a much 27 Productivity is defined as output per worker and is calculated by dividing higher wage premium. Consequently, productivity increases in both output by the total labor force. agriculture and manufacturing with the reallocation of labor. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 7 Kenya in Context the contribution of productivity in the services sector 1.2 FISCAL POLICY AND ECONOMIC to GDP per capita growth declined to 0.50 percentage GROWTH V points between 2010 and 2015 (Figure 1.10). ision 2030, the country’s development blueprint, outlines government spending policy. Vision 2030 Within the region, productivity has been the main has three key pillars: economic, social and political. driver of GDP per capita growth. Productivity was Government spending falls under the economic pillar.29 a key driver of growth contributing between 80 to The plan is implemented in five-year periods, with the 97 percent to GDP per capita growth between 2005 first period covering 2008 – 2012. Implementation of the and 2015 (Figure 1.11). However, compared to its second period (2013 – 2017) is complete. Preparations for peers, Kenya’s productivity contribution to GDP per the third period are under way, with a focus on the “Big 4” capita growth was the lowest at 80.9 percent, while in priorities of food security, affordable housing, enhanced Rwanda productivity accounted for most of GDP per manufacturing, and UHC (Box 1.1). capita growth at 97.9 percent. Demographic change was the second most important driver of GDP per capita growth, at 20.1 percent for Kenya, 12.2 percent 1.2.1 Revenue vs. expenditure for Uganda, 6.8 percent for Ghana and 3.2 percent for Growth in revenue was erratic, increasing in the Rwanda, an indication that the economy benefitted first half of the 2005 to 2015 period and declining from the increase in the working age population. afterwards. Revenue collection peaked in FY2009/10 However, even as the economy benefitted from the at 21.9 percent of GDP, followed by a decline to 17.2 demographic dividend, growth in job creation did not percent of GDP in FY 2015/16. The main source of match the growth in the working age population, as revenue was income tax, which accounted for almost demonstrated by the declining employment rates. The half of revenue collection (an average of 8.1 percent of employment rate contribution to GDP per capita growth GDP in the ten years prior to FY2015/16). Income tax was negative at -6.3 percent for Kenya, -4.0 percent for comprises personal income tax and corporate income Ghana and -2.1 percent for Rwanda. In contrast, the tax. The second most important source of revenue was employment rate was the second most important VAT, averaging 5.0 percent of GDP in the ten years prior driver of per capita GDP growth for Tanzania explaining to FY2015/16. Other sources of revenue include import 5.3 percent of the GDP per capita growth. and excise duties. Figure 1.10: Sectoral contribution to change in real GDP Figure 1.11: Productivity contribution to real GDP per per capita productivity capita growth 2 2005 - 2015 100 1 Percentage points 90 Percent 0 80 -1 2005-2015 2005-2010 2010-2015 Agriculture Industry 70 Services etc. Intersectoral reallocation e ect Kenya Uganda Tanzania Ghana Rwanda Source: KNBS and WDI. Source: KNBS and WDI. 29 Vision 2030 is the GoK development plan. It is aimed at ensuring Kenya achieves middle income status by the year 2030. 8 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Box 1.1: The Big 4 policy agenda The GoK has announced four key priorities to advance Vision 2030 over the next five years. Known as the Big 4, these priorities are food and nutrition security, affordable housing, increased share of manufacturing, and UHC. Food and nutrition security. The agriculture sector is a key driver of Kenya’s economy, contributing about 50 percent to GDP. Low productivity in the sector, in combination with a growing population, leads to a structural food deficit and poses risks to food security in the country. The sector is characterized by low yields, particularly in grain crops, and vulnerability to climatic shocks. The government intends to invest in sustainably exploiting national water resources through water towers and river ecosystems and to address the distribution, wastage, storage and value-addition of agriculture commodities. Affordable housing. With an estimated housing shortfall of 2 million units, the housing situation in Kenya is expected to deteriorate as urbanization continues. Each year, 500,000 new residents move to urban areas, often residing in informal settlements. Over the next five years, the Government plans to inject capital into the housing sector and provide affordable housing to 500,000 new households. Policy reforms that lower the costs of construction and increase access to mortgages are further intended to increase the affordability of housing. Enhancing manufacturing. The manufacturing sector holds great potential for high job creation, as witnessed by the impressive poverty reduction in countries in Asia. For this to occur, Kenya’s manufacturing firms need to be competitive both domestically against imports and globally in exports, especially within East Africa. Competitiveness challenges in the sector have resulted in a declining share of manufacturing output in GDP. The government aims to increase the share of the manufacturing sector in GDP from 9 percent to 15 percent in the next five years through reductions in power tariffs for manufacturers. UHC. Kenya is in a favorable position to rapidly expand health coverage given the strong institutional foundations and political will. Health insurance is currently concentrated in the formal sector, where contributions are automatically deducted from salaries. However, 70 to 80 percent of the population remains without health insurance coverage, with most of the uninsured in the informal sector. The government aims to achieve 100 percent universal coverage for all households by reforming and expanding the National Hospital Insurance Fund (NHIF). Source: Kenya Economic Update, April 2018. Official website of the presidency of Kenya, April 2018, www.president.go.ke. In contrast, spending increased over this period, GDP in FY2014/15, a reflection of government policy to consistently outpacing revenues. From FY2005/06 increase infrastructure development in a bid to remove to FY2015/16, the government increased deficit supply-side constraints. However, as the government spending (Figure 1.12).30 Recurrent spending was the began fiscal consolidation, development spending main driver of government expenditure, averaging declined in FY2015/16 to 8.2 percent of GDP. about 17.1 percent of GDP over the period. Wages and salaries were the largest component of recurrent Growth in expenditure was faster than growth in spending, with interest payments picking up during revenue, widening the fiscal deficit (Figure 1.13).32 the latter half of the period to 3.2 percent of GDP in The fiscal deficit has been on an upward trajectory, FY2015/16.31 Development spending nearly doubled widening by 3.5 percentage points from -4.7 percent of from 4.5 percent of GDP in FY2005/06 to 8.7 percent of GDP in FY2005/06, to -8.2 percent of GDP in FY2015/16 (Figure 1.14). The -8.2 percent deficit is more than 30 The pane above the dotted line means that spending is higher than revenue (deficit budget), while any points on the dotted line would mean spending equals revenue collections (balanced budget). 32 On the red dotted line, a percentage point change in spending will equal 31 Domestic interest payments make up the larger share of interest a percentage point change in revenue collection, with both variables payments at 2.6 percent of GDP. measured as a percent of GDP. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 9 Kenya in Context Figure 1.12: Spending has consistently exceeded Figure 1.13: Revenue collection has not kept up revenue collection with spending pressures Source: The National Treasury. Source: The National Treasury and World Bank. Figure 1.14: The evolution of fiscal deficit The education sector has been the largest 2005/06 2007/08 2009/10 2011/12 2013/14 2015/16 beneficiary of social sector spending. Education 0 spending maintained its momentum throughout the ten-year period, mirroring government spending. The -2 largest increase in education expenditure as a share -2.7 of the increase in total government spending was in -4 FY2009/10, after which it slowed to its lowest level in Percent -4.2 -4.7 -4.8 FY2013/14 (Figure 1.15).33 However, even as education -4.9 -6 -5.2 -5.6 spending tended to increase with government -5.8 -6.4 spending, the rate of change in the increases was -8 generally low, an indication that education is not likely -8.2 to have any fiscal risk effects. The stable increase in -9.1 education expenditure is attributable to free primary -10 education, as the government employs more teachers Source: The National Treasury. to cater for increasing demand. double the East African Community (EAC) target of 3.0 percent. The government has embarked on a fiscal Implementation of the new constitution led to consolidation plan that should see the deficit decline in moderated health expenditures at the national the medium term to -3.0 percent of GDP in FY2020/21. level. Growth in government spending has trickled down at a slower pace to the health sector compared 1.2.2 Sectoral analysis in government spending to the education sector (Figure 1.15). Following Growth in government spending was uneven. Growth devolution of the health sector in FY2013/14, the in government spending declined from 22.0 percent in momentum of health expenditures at the national FY2009/10 to 3.0 percent in FY2013/14 (Figure 1.15). level slowed down significantly. Full devolution of the The slowdown in spending in FY2013/14 coincided health sector effectively means local governments with the entrance of a new administration and the are responsible for all health care provision. However, implementation of the 2010 Constitution. However, the national government transfers money to a growth in spending accelerated to 14.0 percent in consolidated fund, without specifically earmarking FY2014/15, but decreased moderately to 12.0 percent the amount that should go to the health sectors at the year after. 33 Momentum is defined as the increase in education spending due to an increase in total government spending. See also Merotto et. al. (2015). 10 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.15: Sectoral contribution to growth in 1.3 A REVIEW OF SOME POLICIES OVER THE total spending LAST DECADE P 25 Growth in government expenditure (percent) olicies were designed to increase aggregate demand, in turn contributing to economic growth 20 in Kenya. Kenya’s vision 2030 set a growth target of 10 percent per annum. While the target growth rate has not 15 been achieved, both supply- and demand-side policies aimed at increasing growth have been a recurrent 10 theme in the budget statements over the last 10 years. This section focuses on infrastructure development, 5 use of renewable energy, domestic production, job creation and income inequality reduction, and analyses 0 2009/10 2010/11 2011/12 2012/13 2013/14 2014/15 2015/16 if these policies could also enhance pro-poor growth. Health Education Infrastructure Other expenditure Source: The National Treasury. Supply-side enhancing infrastructure projects, the devolved units. It is unclear if the decline in health such as rail transport and renewable energy, were spending at the national level is substituted at the prioritized. In 2014, exemptions on import duty for devolved units. railway products as well as import duty on machinery, spares and inputs for direct and exclusive use in the Relative increases in social protection spending development and generation of solar and wind energy are large, due to low base effects. In absolute terms, were introduced. During the same period, imports growth in social protection spending increased to a of railway inputs and machinery increased. A more peak of 13.0 percent in FY2010/11. Momentum of social efficient transportation network – attributable to the protection spending, defined as the growth rate of total construction of the US$ 3.6 billion Standard Gauge spending multiplied by the share of social protection Railway – and a stable supply of electricity are crucial spending, was at less than 1.0 percent. This indicates a in reducing the cost of production and fostering scant allocation to social protection expenditure within competitiveness of the manufacturing sector. the overall budget increases. Kenya has abundant clean energy potential which In contrast, not only did infrastructure spending remains untapped. In order to provide incentives have a relatively high growth rate, it also gained to support local production of clean energy, duty momentum. At its peak, infrastructure spending remission was granted on inputs for the production of increased by 74 percent (FY2009/10), accounting for solar panels in FY2013. Geothermal (290MW) and wind more than half of total growth in spending. This reflects (361MW) energy projects were commissioned (Table government policy priorities, namely the national 1.1). In addition, to encourage usage of environmentally development plan, which emphasizes improving friendly vehicles which aimed at reducing carbon infrastructure. Infrastructure spending momentum emission and noise pollution, battery operated vehicles picked up pace from 3.0 percent in FY2009/10 to 5.0 were exempted from duty. However, there are no data percent in FY2015/16, indicating that as the budget to support an increase in imports of environmentally increases, a larger share of spending is allocated friendly vehicles. towards infrastructure. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 11 Kenya in Context Table 1.1: List of ongoing major projects Project value Project name Type Distance (US$ Millions) Standard Gauge Railway Phase 2A Railway 120 Km 1,500 Lamu Port Southern Sudan and Ethiopia Corridor (LAPSSET) Port, Roads, Rail, Pipeline ... Nairobi Mombasa Expressway Road 473 Km 2,300 Northern Corridor Transport Improvement Project Roads Capacity Project value Public Private Partnerships (PPPs) MW (US$ Millions) Thika Power Thermal 87 146 Triumph Thermal 82 156.5 Gulf Power Thermal 80 108 Orpower Geothermal 150 558 Lake Turkana Wind 300 847 Longonot Geothermal 140 760 Kinangop Wind 61 150 Rabai HFO 90 155 Kipevu HFO 74 85 Mumias Bagasse Co-gen 32 50 Source: PPP Unit, National Treasury; Kenya Railways. Growth in domestic production and industrial growth import duty exemptions were granted on television are not only key in ensuring overall growth in the cameras, digital cameras and video camera recorders economy, but are also important for job creation. In while a 100 percent grant was proposed to investment this regard, policies that enhance domestic production deduction on capital expenditure incurred by film spiked from -0.6 percent in 2012 to 5.6 percent in 2013, producers on purchase of any filming equipment. The the growth was attributable to low base effects and purpose of the exemptions was twofold: i) the film a successful election period. Some pro-poor policies industry has traditionally had a very low performance on the supply side included the removal of the sugar in Kenya, with the introduction of exemptions aimed development levy, duty exemptions for raw materials at creating incentives to promote the industry, and ii) used in the manufacture of sanitary towels, as well as the film industry has potential to create employment duty exemptions on all synthetic yarns, acrylic yarn and for the youth, who are the majority of the population polyester yarn. Additionally, duty was eliminated for in Kenya. Indeed, employment in the modern sector basic commodities that make up the largest share of has increased in recent years (Figure 1.16). A second the consumption basket for poor households. Further, area envisaged as creating potential employment such interventions also increase the competitiveness for the youth was the transport sector, in particular of exports to the region, which in turn has positive the motorcycle taxi. Motorcycles are a relatively new implications for macroeconomic indicators such as mode of transportation in the cities, which create jobs the current account balance and the exchange rate. often for youth as motorcycle taxi drivers. Motorcycles However, exports to the region have declined despite have the advantage of being much faster than motor the export-friendly interventions. vehicles given traffic congestion. In FY2009, duty on motorcycles of between 50cc and 250cc were zero Non-traditional sectors were identified as a potential rated, potentially contributing to an increase in the source of employment. In FY2009 and FY2010, VAT and informal sector employment. 12 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.16: Employment trends While the national poverty lines are critical to analyze 180 poverty dynamics and distribution within the country, they are not comparable across countries. 160 Kenya’s national poverty line is derived from the Cost of Basic Needs (CBN) method.36 The CBN method Index=2009 = 100 140 stipulates a consumption bundle deemed to be 120 adequate for “basic consumption needs”, and then estimates what this bundle costs in reference prices. As 100 basic consumption needs are usually different across 80 countries, the poverty rate measured by the national poverty line is not comparable across countries. 60 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Therefore, this section uses the international poverty Modern sector employment Informal sector employment line defined at US$ 1.90 using 2011 purchasing power Total employment parity (PPP) international dollars (Box 1.2). Chapter 2 Source: KNBS. provides a detailed assessment of poverty trends at the The social pillar of Kenya’s Vision 2030 places national poverty line. emphasis on improved quality of life for all Kenyans. An important condition for a higher standard of living 1.4.1 Monetary poverty trends at the international and therefore quality of life is an increase in income. poverty line Policies to enhance income equality over the period About 1 out of 3 people in Kenya live below the included reduction of taxes on basic commodities international poverty line. The daily consumption typically consumed by poorer households, such as expenditure for 36.8 percent of the population is second-hand clothing. A reduction of import duty below US$ 1.90 in 2011 PPP. For 66.2 percent of the from US$ 0.3 per kg to US$ 0.2 per kg on second-hand population it is below US$ 3.20 in 2011 PPP (Box 1.2). The poverty rate has moderately reduced over the past clothing was implemented in FY2010. Similarly, in decade at both international poverty lines, dropping FY2015, all imported farm inputs used in the processing nearly seven percentage points at the US$ 1.90 line and and preservation of seeds for planting were exempted. three percentage points at the US$ 3.20 line between 2005 and 2011 (Figure 1.17). Poverty reduction has 1.4 OVERVIEW OF MONETARY POVERTY34 been steady over the past decade, except for a shock P overty incidence declined from 46.8 percent in 2005/06 to 36.1 percent in 2015/16, using Kenya’s official national poverty lines. The KNBS released the to consumption in the years following the 2008 global economic crisis (Figure 1.19). most recent poverty statistics in March 2018, based on Increased consumption for the poorest of the poor the KIHBS 2015/16. KIHBS 2015/16 closes an important has driven poverty reduction in the past decade. The data gap, as the previous survey collecting expenditure rate of extreme poverty under the threshold of US$1.20 a day in 2011 PPPs has decreased by 7.3 percentage data to estimate poverty was implemented 10 years ago points since 2005 to reach 13.7 percent in 2015 (Figure in 2005/06.35 1.17). The reduced poverty at the US$ 1.90 international poverty line reflects these improvements. The depth of poverty can be measured by the poverty gap index, representing the average deficit between the 34 This section is derived from the Poverty Special Focus of the Kenya Economic Update, April 2018. total consumption of the poor and the international 35 The KIHBS 2015/16 utilized a two-stage stratified cluster sampling method with the objective of providing data for poverty estimates at poverty line. Using this measure, the depth of poverty national and county levels as well as for urban and rural areas. The sample at the US$ 1.90 line decreased from 16.2 percent of the included 24,000 households from 2,400 clusters distributed to urban and rural strata for each of the 47 counties in Kenya based on the 2009 poverty line in 2005 to 11.6 percent in 2015 (Table 1.2). Census. The survey was implemented over 12 months from September 2015 to August 2016 to take into account seasonal effects. Source: KNBS 36 Ravallion, Martin. 1994. “Measuring Social Welfare With and Without (2018): “Basic Report on Well-Being in Kenya”. Poverty Lines.” The American Economic Review 84 (2): 359–364. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 13 Kenya in Context Table 1.2: Key monetary poverty Indicators37 Poverty headcount Poverty gap 2005 2015 2005 2015 US$ 1.20 2011 PPP poverty line38 21.0 13.7 6.7 3.6 US$ 1.90 2011 PPP poverty line 43.7 36.8 16.2 11.6 US$ 3.20 2011 PPP poverty line 69.2 66.2 33.3 28.4 Source: KIHBS 2005, KIHBS 2015, authors’ calculations. Box 1.2: The international poverty lines The international poverty line is defined in absolute terms as a threshold of being able to purchase a fixed basket of goods that meets basic needs across countries. The concept of an international poverty line was first introduced in the 1990 World Development Report. The objective was to measure poverty in a consistent way across countries, using a poverty line that reflected conditions of poverty in poor countries, while also considering real purchasing power across countries of all incomes. To decide on an international poverty line, the World Bank analyzed data from 33 national poverty lines from both developed and developing countries in the 1970s and 1980s. The threshold of US$ 1 a day was agreed upon and became the first international poverty line. Over the years, the poverty line has periodically been adjusted as new purchasing power parity (PPP) measures became available. The new measures reflected both changes in relative price levels across countries, as well as changes to methodologies. The poverty line increased from US$ 1 a day at 1985 PPPs to US$ 1.08 at 1993 PPPs, then to US$ 1.25 at 2005 PPPs, and finally to its current level of US$ 1.90 at 2011 PPPs. The increase in the international poverty line can be mostly attributed to changes in U.S. dollar purchasing power relative to the purchasing power of the local currencies in the poorest countries. Essentially, the increase in the poverty line says that US$ 1.90 in 2011 real terms would buy about the same basket of goods that US$ 1.25 bought in 2005. The World Bank introduced an additional set of international poverty lines in 2016, taking into account the relationship between national poverty lines and the wealth of the country. These lines are defined as the median national poverty line for each grouping of countries by their GNI per capita, using the World Bank classification of countries as low-income, lower middle-income, upper middle-income and high-income. The World Bank now reports poverty rates for countries using the new lower middle-income and upper middle-income poverty lines. The poverty line for lower middle-income countries is US$ 3.21 per day and for upper middle-income countries, it is US$ 5.48 per day. In addition to these poverty lines, this section also uses a US$ 1.25 2011 PPP poverty line to further distinguish between the poor living below US$ 1.90 and the poorest living below US$ 1.25. To allow for international comparisons, poverty in this section is estimated using the current international poverty line and the lower middle-income class (LMIC) poverty line. Since 2014, Kenya has been classified as a lower middle-income country. Its current GNI per capita of US$ 1,380 puts it at the bottom of the LMIC grouping.39 As the poverty lines are defined using US$ 2011 PPPs, this is converted to the local currency used to measure consumption for both survey years 2005 and 2015. First, US$ 2011 are converted into Kenyan Shilling in 2011 using the PPP estimate for Kenya (35.43). Second, the change in purchasing power per Kenyan Shilling is adjusted for by considering inflation or deflation to the survey period as measured by the national CPI. 37 Poverty estimates in this section are preliminary. The official source for World Bank estimated poverty headcounts is PovcalNet. For the estimation for poverty in this section, the poverty line was adjusted using the 2011 PPP estimate and inflated or deflated to the survey period. The official consumer price index (CPI) used for 2011 was 121.1654. For the KIHBS 2005, the weighted average of the official CPI for the survey period was 73.2557. For the KIHBS 2015 survey period, it was 166.299. Poverty was estimated with a per capita aggregate for consumption expenditure. The aggregate was not spatially deflated and excludes rent, unlike the aggregate used in the Poverty Special Focus of the Kenya Economic Update, April 2018. Thus, poverty estimates in this section differ slightly from those in the Economic Update. 38 The US$ 1.20 line is not an international poverty line. It is included in this section for the purposes of distinguishing the poorest in extreme poverty (Box 1.2). 39 Source: World Bank Open Data Catalogue. 14 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context To further distinguish the poorest of the poor, a poverty line of US$ 1.20 in 2011 PPP is included in this section. This line is based on the share of food consumption in total expenditure. On average, Kenyans spend 63 percent of their total daily consumption on food consumption. Starting with the US$ 1.90 international poverty line as a threshold for total consumption, this translates into daily per capita food consumption of US$ 1.20 in 2011 PPP. Those living below US$ 1.20 a day cannot afford the minimum food consumption calories even if they were to cut out all non-food consumption. As the food share specific to Kenya is used to derive this line, it is not suitable for international comparisons. It is only used in this section to distinguish the poorest in extreme poverty. Well-being has stagnated for households living To estimate the relationship between household between the US$ 1.90 and US$3.20 poverty lines. The consumption and growth at the sector level, the percentage of the population consuming between evolution of poverty from 2005 to 2015 is simulated US$1.90 and US$3.20 increased by 3.9 percentage based on sectoral growth rates, while assuming no points between 2005 and 2015 (Figure 1.17). This is not redistribution beyond that resulting from differences surprising as increases in consumption of the very poor in sectoral growth. Consumption expenditure per have pushed them above the US$ 1.90 poverty line household from KIHBS 2005 is augmented based on while in the same period not as many (net) households the growth rate of the household head’s sector of increased consumption beyond US$ 3.20. Therefore, still economic activity. The poverty rate per sector in KIHBS many households have a certain degree of vulnerability 2015 provides the anchor to estimate the growth- to fall back into poverty measured at the US$ 1.90 level. consumption pass-through parameter of that sector.40 A 10 percent consumption shock would push a fifth of In other words, the pass-through parameter ensures households currently between US$ 1.90 and US$ 3.20 that sectoral GDP growth transmitted to household below the US$ 1.90 a day threshold, raising the poverty consumption growth is consistent with the observed headcount by six percentage points (Figure 1.18). changes in poverty between 2005 and 2015. The pass- Figure 1.17: Poverty at the US$ 1.20, 1.90, and 3.20 lines Figure 1.18: Cumulative consumption distribution with shock 80 100 Poverty headcount (% of population) 80 Percent of population 60 25.5 29.4 60 40 40 22.7 23.1 20 20 21.0 0 13.7 0 100 200 300 400 500 0 2005 2015 Average per capita monthly consumption, US$ 2015 PPP Poverty under US$3.20 USD a day Poverty under US$1.90 USD a day US$ 1.90 Poverty Line Consumption, KIHBS 2015 Poverty under US$1.20 USD a day US$ 3.20 Poverty Line Consumption, 10% shock Source: KIHBS 2005, KIHBS 2015, authors’ calculations. Source: KIHBS 2015, authors’ calculations. 40 Occupations are categorized into three broad categories: (1) agriculture; (2) manufacturing; (3) services. Assumptions about sectoral pass-through parameters for these sector groupings are drawn from the sectoral decomposition of poverty analysis between 2005 and 2015. Parameters are assumed to be constant over years. For households without reported household head occupation, average GDP growth is applied. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 15 Kenya in Context Figure 1.19: GDP sectoral growth simulation of poverty Figure 1.20: Overall GDP growth simulation of poverty trajectory at international poverty lines, 2005 to 2015 trajectory at international poverty lines, 2005 to 2015 80 80 70 70 Poverty headcount (% of population) Poverty headcount (% of population) 60 60 50 50 40 40 30 30 20 20 10 10 0 0 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Poverty, under US$1.20 a day Poverty, under US$1.90 a day Poverty, under US$1.20 a day Poverty, under US$1.90 a day Poverty, under US$3.20 a day Poverty, under US$3.20 a day Source: KIHBS 2005, authors’ calculations. Source: KIHBS 2005, authors’ calculations. through parameter indicates the fraction of sectoral (Figure 1.21). From 2011 to 2015, growth averaged GDP growth that translates into private household 4.1 percent. Most household heads are engaged in consumption. While a large pass-through parameter agriculture, followed by services and then industry suggests that high GDP growth helps to improve (Figure 1.22). Households engaged in agriculture benefit consumption of households, it also flags the risk from the highest pass-through rate, especially for those that high GDP volatility translates into consumption consuming less than US$1.20 a day (Figure 1.23). For volatility, making households vulnerable to shocks that these households, real consumption increases by 0.75 affect GDP growth. percent for each one percent growth in the agriculture sector. The flipside of a high pass-through rate is the Agricultural GDP growth largely translates into vulnerability to shocks. The industrial sector has the consumption growth, exposing agricultural smaller pass-through rate, indicating a protection households to shocks in agricultural GDP. In the against shocks of GDP growth but also implying that years following the slow-down of growth in 2008, households in this sector participate less in sectoral the agriculture sector experienced a strong rebound GDP growth. Figure 1.21: Real sector growth, 2007 to 2015 Figure 1.22: Share of households by sector of household head occupation, 2005 vs. 2015 60 12 10 50 8 Percentage of households 6 40 4 30 2 0 20 -2 10 -4 -6 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 Agriculture Industry Services Agriculture GDP Services Industry 2005 2015 Source: KNBS. Source: KIHBS 2005, KIHBS 2015, authors’ calculations. 16 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.23: Consistent sectoral elasticities for poverty Figure 1.24: Combination of growth and redistribution pass-through23 needed to eradicate poverty in 2030 0.8 3.5 0.7 Annual pace in inequality reduction (Percent) 3 0.6 2.5 Pass-through rate 0.5 2 0.4 1.5 0.3 0.2 1 0.1 0.5 0 Agriculture Industry Services 0 0 2 4 6 8 10 12 Sectoral pass-through, 1.20 Sectoral pass-through 1.90 Sectoral pass-through, 3.20 GDP pass-through, 1.90 Annual growth in household consumption (Percent) Source: Authors’ calculations. Source: KIHBS 2015, authors’ calculations. Kenya is not on track to eradicate poverty by 2030, 1.4.2 Monetary poverty in international comparison and higher and more inclusive growth, as well as pro- Kenya’s poverty rate is below the average in sub- poor policies, are needed. In order achieve a poverty Saharan Africa and is amongst the lowest of its East rate below 3 percent by 2030, the poverty rate must African peers.42 The poverty rate at the US$ 1.90 a day decrease at least 33.8 percentage points. However, line in Kenya is nearly half the poverty rate of Rwanda in Kenya’s annualized poverty reduction rate was 1.6 2013 (60.4 percent). However, it is higher than poverty percent between 2005 and 2015. Assuming this rate is in Uganda (34.6 percent) and Ghana (13.6 percent), maintained for the next 15 years, the poverty rate will both measured in 2012 (Figure 1.25). When considering remain above 25 percent in 2030. To meet the 3 percent GDP per capita in constant PPP terms, poverty in Kenya goal in 2030, an annual poverty reduction rate of 6.1 is in line with expectations given the trend of poverty percent would be necessary. Without any reduction to GDP per capita in sub-Saharan Africa (Figure 1.26). in inequality, real household consumption would Kenya’s ratio of poverty to GDP per capita is close to need to increase on average by 11.3 percent per year that of the sub-Saharan Africa aggregate. Ghana and to achieve this objective. With the observed growth- Uganda both have lower ratios of poverty to GDP per consumption pass-through of 0.25, this would imply capita. However, it is important to note that Kenya has an unrealistically high annual GDP growth of about 45 the most recent estimate for poverty (2015), which may percent. Thus, high growth must be complemented by bias its performance in comparison to countries with stronger inclusive growth, increasing the pass-through older poverty estimates such as Ghana and Uganda parameter, and a reduction in inequality through pro- (both 2012). poor policies (Figure 1.24). 42 Four countries were selected for the international comparison due to geographic proximity, comparable population size and/or level of 41 This figure shows the sector elasticity assumptions for the trajectory of wealth: Ghana (GHA), Rwanda (RWA), Tanzania (TZA), and Uganda (UGA). poverty simulations at the US$ 1.20, 1.90, and 3.20 per day poverty line The aggregate for Sub-Saharan Africa is also included as a regional thresholds. For each threshold simulation, different sectoral elasticities benchmark. Tanzania has a GDP PPP per capita ($2,583) comparable were assumed. The pass-through rate is generally highest for poverty to that of Kenya ($2,926), while Ghana ($3,980) is relatively wealthier. under the US$ 1.20 level, indicating that growth has a larger impact Rwanda ($1,774) and Uganda ($1,687) are both relatively poorer than on consumption of the very poor. The pass-through rate of overall Kenya. In terms of population, Tanzania (55.6 million) and Uganda (41.5 GDP growth, in the US$ 1.90 poverty line simulation, is included as a million) are similar in size to Kenya (48.5 million), whereas Ghana (28.2 benchmark. million) and Rwanda (11.9 million) are notably smaller. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 17 Kenya in Context Figure 1.25: International comparison of poverty Figure 1.26: Poverty headcount against GDP per capita 70 100 90 60 80 Poverty headcount (% of population) Poverty headcount (% of population) 50 70 Rwanda 2013 60 40 50 Kenya 2005 Tanzania 2011 30 40 SSA 2013 30 Uganda 2012 Kenya 2015 20 20 10 10 Ghana 2012 0 0 Rwanda Tanzania SSA Kenya Uganda Ghana 6.0 7.0 8.0 9.0 10.0 11.0 2013 2011 2013 2015 2012 2012 Log GDP per capita, constant PPP Source: KIHBS 2015, World Bank open data catalogue, authors’ calculations. Source: KIHBS 2015, KIHBS 2005, World Bank open data catalogue, authors’ calculations. The depth of poverty at the international poverty When considering Kenya’s LMIC status, poverty is line is consistent with expectations. The relationship relatively high. Poverty in Kenya is higher than the between the poverty headcount and the poverty gap aggregate for LMIC countries, both at the US$ 1.90 in Kenya conforms to the trend for sub-Saharan African and US$ 3.20 lines (Figure 1.28). Ghana provides an countries (Figure 1.27). Kenya’s poverty gap is close to appropriate benchmark as it has a similar GNI per capita that of Uganda (10.3 percent), but is notably higher to Kenya (US$ 1,380). The poverty headcount in Ghana than in Ghana (4.0 percent). The improvement in the at the LMIC line (34.9 percent) is 28.8 percentage points poverty gap since 2005 suggests that many of the poor less than that in Kenya. Poverty in Kenya is also much are close to reaching the US$ 1.90 a day consumption deeper at the LMIC line than it is at the international threshold. This reflects Kenya’s notable reduction in poverty line. The poverty gap at the LMIC line is 27.5 poverty below US$ 1.20 a day since 2005. percent, compared to 11.3 percent at the international poverty line. Kenya’s depth of poverty at the LMIC Figure 1.27: Poverty rate against depth at international line is substantially higher than Ghana and the LMIC poverty line aggregate (Figure 1.29). 45 40 Kenya has a relatively weak relationship between 35 Poverty gap (% of poverty line) poverty reduction and GDP growth. Between 2005 30 and 2015, annualized GDP per capita growth in Kenya 25 Rwanda 2013 was 2.75 percent, while the annualized reduction in the 20 poverty rate was 0.7 percentage points, or 1.58 percent. SSA 2013 Kenya 2005 15 Tanzania 2011 This gives Kenya an elasticity of poverty reduction to 10 Kenya 2015 GDP growth of 0.57, meaning that for every 1 percent Uganda 2012 increase in GDP per year, the poverty rate decreases 5 Ghana 2012 by 0.57 percent. This elasticity is lower than the sub- 0 0 20 40 60 80 100 Saharan aggregate (0.74), as well as Tanzania, Ghana Poverty headcount (% of population) and Uganda (Figure 1.30). Kenya’s ratio of GDP per Source: KIHBS 2015, World Bank open data catalogue, authors’ calculations. capita to elasticity is in line with the sub-Saharan Africa aggregate (Figure 1.31). 18 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.28: Poverty headcount at IPL and LMIC, Figure 1.29: Poverty gap at IPL and LMIC, international international comparison comparison 70 30 60 25 Poverty headcount (% of population) Poverty gap (% od poverty line) 50 20 40 15 30 10 20 5 10 0 0 Kenya 2015 Ghana 2012 LMIC 2013 Kenya 2015 Ghana 2012 LMIC 2013 Poverty rate at US$1.90 PPP per day line Poverty rate at US$1.90 PPP per day line Poverty rate at US$3.20 PPP per day line (LMIC) Poverty rate at US$3.20 PPP per day line (LMIC) Source: KIHBS 2015, World Bank open data catalogue, authors’ calculations. Source: KIHBS 2015, World Bank open data catalogue, authors’ calculations. Figure 1.30: International comparison of elasticity of Figure 1.31: Elasticity of poverty reduction against poverty reduction GDP per capita Rwanda Kenya Sub- Uganda Ghana Tanzania Log GDP per capita, constant PPP Saharan Africa 7.0 7.5 8.0 8.5 0.0 0.0 % change in poverty per 1% change in GDP Rwanda -0.2 -0.2 % change in poverty per 1% change in -0.4 -0.4 GDP per capita PPP -0.6 Kenya -0.6 Sub-Saharan -0.8 Africa -0.8 -1.0 Uganda -1.0 -1.2 -1.2 Ghana -1.4 Tanzania -1.4 -1.6 -1.6 Source: KIHBS 2015, KIHBS 2005, World Bank open data catalogue, authors’ Source: KIHBS 2015, KIHBS 2005, World Bank open data catalogue, authors’ calculations. calculations. 1.5 OVERVIEW OF NON-MONETARY defined as a daily per capita consumption expenditure POVERTY below US$ 1.90 in 2011 PPP, which affects 36.8 percent P oor households are often deprived in multiple of households. In education indicators, nearly one third dimensions. The most common type of deprivation of all households are deprived in adult educational is access to services, notably sanitation and electricity attainment, meaning no adult in the household has (Figure 1.32). 40.7 percent of households lack access completed primary education. Primary school enrollment to improved sanitation43 and 64 percent lack access to is the least common deprivation. Less than one quarter electricity. Fewer households are deprived of access to of all households (23.7 percent) have a child of primary- an improved source for drinking water44 (28.2 percent). school age not currently attending primary school. The second most common deprivation is monetary, 43 Improved sanitation is defined as a toilet with a flush, a ventilated improved pit (VIP) latrine or a latrine with a slab. 44 Improved drinking water sources are defined as a piped water system, public tap, borehole, protected dug well, bottled water or water from rainwater collection vendors. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 19 Kenya in Context Figure 1.32: Multi-dimensional deprivations, 2015 (68 percent) and is in line with its level of poverty 70 (Figure 1.33). Kenya performs much better in access to improved sanitation compared to countries with a 60 comparable poverty headcount (Figure 1.34). 50 % of households deprived Kenya’s performance on human development 40 indicators has improved since 2015, but lags behind 30 Ghana. Kenya’s Human Development Index (HDI), 20 calculated by the United Nations Development Program (UNDP) as a combination of education, inequality, and 10 life expectancy indicators, gained 0.07 points in the past 0 decade to reach 0.55 in 2015. This is the highest HDI in Consumption Adult Primary Improved Improved Access to education school water sanitation electricity the EAC, but still behind Ghana (0.58). Kenya’s level of attainment enrollment human development is relatively high given its poverty Source: KIHBS 2015, authors’ calculations. headcount (Figure 1.35), indicating that Kenya performs Kenya has a relatively high level of access to improved better on non-monetary dimensions of poverty. sanitation compared to international benchmarks, but lags behind in access to improved water. The lack Kenya’s adult literacy rate is among the highest in of improved water sources increases the time burden Africa. In 2015, 84 percent of the population aged 15 for women and children, who generally bear the years and over could read and write in any language, responsibility of fetching water. Though progress has a larger proportion of the population than in a country been made in improving access to improved water like Ghana (71 percent), which has a much lower poverty since 2005, Kenya still lags behind other countries in rate (Figure 1.36). The literacy rate has increased by 11 the international comparison. Only 71.8 percent of percentage points since 2005, reflecting the progress Kenyan households have access to improved water in enrollment in Kenya over the past decade. This is in sources. This is below the level of peer countries like line with results from standardized tests suggesting Ghana, Rwanda and Uganda. Kenya’s rate of improved that Kenyan children have somewhat better learning water is close to the average for sub-Saharan Africa outcomes in primary school than children in other Figure 1.33: Poverty headcount against access Figure 1.34: Poverty headcount against access to improved water to improved sanitation 100 100 Ghana 2012 Access to improved sanitation (% of households) 90 90 Access to improved water (% of households) Uganda 2012 80 Rwanda 2013 80 70 Kenya 2015 70 SSA 2013 60 Kenya 2015 Rwanda 2013 60 Kenya 2005 Tanzania 2011 50 50 Kenya 2005 40 40 30 30 SSA 2013 20 20 Tanzania 2011 Uganda 2012 10 10 Ghana 2012 0 0 0 20 40 60 80 0 20 40 60 80 100 Poverty headcount (% of population) Poverty headcount (% of population) Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ calculations. calculations. 20 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.35: Poverty headcount against HDI Figure 1.36: Poverty headcount against literacy rates 0.9 100 90 Kenya 2015 0.8 Adult literacy rate (% population 15+) 80 0.7 Uganda 2012 Ghana 2015 70 Rwanda 2013 0.6 Kenya 2015 Ghana 2012 Kenya 2005 Tanzania 2015 Rwanda 2015 60 0.5 50 HDI Uganda 2015 Kenya 2005 0.4 40 0.3 30 0.2 20 0.1 10 0 0 0 10 20 30 40 50 60 70 80 90 0 20 40 60 80 100 Poverty Headcount (% of population) Poverty headcount (% of population) Source: UNDP HDI. Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ calculations. countries in the region.45 However, significant gender poverty rate (Figure 1.37). However, Kenya’s rate of gaps in adult literacy continue to exist, reflecting adult primary school completion is lower than in gender inequalities in primary education. Ghana and Tanzania. When considering higher levels of educational attainment, Kenya performs worse (Figure In line with increasing enrollment rates, levels of 1.38). Only 14.4 percent of adults aged 25 and older educational attainment among the adult population have completed secondary education. While this also have increased. Over half (57.8 percent) of all Kenyan marks a substantial improvement over 2005 when only adults above the age of 24 have completed primary 3 percent of Kenyan adults had completed secondary education. This marks a notable increase from 2005 school, it is far below rates found in other countries (44.2 percent). Adult primary educational attainment with comparable levels of poverty.46 is high compared with countries that have a similar Figure 1.37: Poverty headcount against adult educational Figure 1.38: Poverty headcount against adult educational attainment, primary attainment, secondary 100 100 90 90 80 80 Completed secondary education Completed primary education 70 Tanzania 2012 70 (% population 25+) (% population 25+) Kenya 2015 60 Ghana 2012 60 Ghana 2012 50 50 Kenya 2005 40 Uganda 2012 40 30 Rwanda 2013 30 Uganda 2012 20 20 Kenya 2015 10 10 Rwanda 2013 Tanzania 2012 0 Kenya 2005 0 0 20 40 60 80 100 0 20 40 60 80 100 Poverty headcount (% of population) Poverty headcount (% of population) Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ calculations. calculations. 46 The results might exaggerate differences, as primary education in 45 Sandefur, Justin. 2018. “Internationally comparable mathematics scores Kenya is eight years but only seven and six years in Tanzania and Ghana. for fourteen African countries.” Economics of Education Review 62 (2018): Kenyan primary school children also score higher on standardized tests 267-286. than Tanzanians. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 21 Kenya in Context Kenya’s net school enrollment rates have improved use of insecticide-treated bed nets (ITNs) that protect over the last decade. The net primary school enrollment children from contracting malaria.48 The decline has rate, the proportion of age-eligible children who are been particularly pronounced among children from currently enrolled in primary, is estimated at 84.6 poorer families and those residing in rural areas; in percent in 2015/16. This is lower than expected given fact, differences in mortality between the bottom 40 Kenya’s poverty headcount. Within the EAC, Uganda percent49 and the top 20 percent and rural and urban and Rwanda both have higher net enrollment rates children were not statistically significant in 2014. Kenya’s (NERs). However, the net secondary school enrollment under-five mortality rate is lower than expected given rate in Kenya is now the highest among countries the country’s level of poverty and is among the lowest of the EAC, at 42.2 percent.47 It more than doubled in sub-Saharan Africa (Figure 1.39). since 2005 (21.0 percentage points) and is in line with expectations given Kenya’s poverty level. Increases in Kenya has also made substantial gains in reducing secondary enrollment in recent years are expected to child stunting; it now has one of the lowest stunting boost educational attainment among young adults in rates in the region. Stunting is defined as a height-for- the near future. age z-score that is more than two standard deviations below the median of a reference population.50 As of Under-five mortality has declined rapidly in recent 2015, nearly 1 out of every 5 children under the age years, particularly among the poor, giving Kenya of 4 (24.4 percent) is stunted in Kenya. While this is the one of the lowest under-five mortality rates in the lowest stunting rate among countries of the EAC, it is region. Mortality among children below the age of five still higher than in Ghana. When considering Kenya’s has declined from 114.6 deaths per 1,000 live births level of poverty, the rate of stunting is lower than in 2003 to only 52.4 in 2014. This decline has been expected (Figure 1.40). The prevalence of child stunting driven mostly by the increased provision and uptake has substantially improved since 2005, when 40.1 of low-cost, high-impact measures, particularly the percent of Kenyan children were stunted. Figure 1.39: Poverty headcount against under- five Figure 1.40: Poverty headcount against child stunting mortality 180 70 160 Prevalence of stunting (% of children under 5) 60 140 50 (deaths per 1,000 live births) 120 Kenya 2003 Rwanda 2013 Under 5 child mortality Kenya 2005 100 40 Uganda 2012 Tanzania 2011 80 Uganda 2016 30 SSA 2013 Tanzania 2015 60 Ghana 2012 Ghana 2014 Rwanda 2015 20 Kenya 2015 40 Kenya 2014 20 10 0 0 0 20 40 60 80 100 0 20 40 60 80 100 Poverty headcount (% of population) Poverty headcount (% of population) Source: USAID Demographic and Health Survey (DHS). Source: KIHBS 2015, KIHBS2005, World Bank open data catalogue, authors’ calculations. 48 The share of children under the age of five that sleeps under an ITN increased from only 4.6 percent in 2003 to 54.3 in 2014. 47 The net secondary school enrollment rate is similarly defined as 49 The statement is based on comparisons across quintiles of a wealth index the ratio of secondary school-aged children who are currently that uses assets to proxy the material standard of living, not consumption enrolled in secondary school to the population of all secondary expenditures. school-aged children. 50 The reference used here is that of the World Health Organization (WHO). 22 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context 1.6 INSTITUTIONAL CONTEXT, ELECTIONS Orange Democratic Movement (ODM) was politically AND DEVOLUTION supported by other ethnic groups, particularly the K enya is a presidential-style democratic republic dominant Luo community. The sitting president Mwai based on a multiparty system in accordance with Kibaki of the PNU was initially declared the winner of a constitution passed in 2010. The president of Kenya is a contentious election. The results were immediately both the head of state and the head of government, and challenged by the ODM, citing voter intimidation and leads the executive branch. Legislative powers rest with a other irregularities. The situation was exacerbated bicameral parliament while the judiciary is independent by the Electoral Commission’s own admission of of these two branches. Although democratic processes, inconsistencies in the process.55 The elections damaged particularly elections, are at times accompanied by Kenya’s image as a relatively-stable country with politically-instigated civil unrest and violence, the country politically mature institutions. 56 57 is considered to have a wider democratic space compared to its neighbors.51 Following major institutional reforms The country undertook reconciliatory measures initiated after the Presidential elections of 2007, there is following the political discord. A power-sharing currently a national government and 47 county-level arrangement with intense support of the international governments that exercise executive and legislative community ended the violence and led to the powers at different levels. formation of the Unity government in 2008, comprising the incumbent PNU and the opposition ODM. The The traditional concentration of power in the constitution was altered to create a new position of executive branch has been a source of political Prime Minister for the opposition’s candidate. The grievance. Since independence, there has been a Afrobarometer Survey conducted in 2008 showed 83 “continuous process of centralization of power” as well percent of Kenyans supporting a constitution that limits as concentration of power in the Presidency.52 This the president to two terms in office, and 77 percent resulted in a sweeping mandate that allowed for, at thought the National Assembly and MPs represent the different times, the redrawing of districts to create new people and should therefore make laws even if the offices for the president’s allies. In addition, new power President or Prime Minister did not agree with them.58 centers at the sub-national level were created, such as the Provincial Administration, that answered directly to A constitutional referendum in 2010 created new the executive.53 The executive was also able to hand out checks on executive power. This process also led to public land to patrons and affiliates. the complete separation of the parliament from the executive under a presidential system of government. The 2007 elections were marked by widespread Political decentralization had always found some political violence and a serious challenge to the degree of support within the diverse communities legitimacy of the electoral system. The frontrunner in Kenya and the country did have some features Party of National Unity (PNU) was widely perceived to of regional autonomy at independence. Successive dominate power and access to resources including land leaders – the founder Kenyatta, followed by President and was led by the Kikuyu community.54 The opposition Moi – centralized state power and influenced key 55 From New York Times coverage of the 2007 elections; Africa: Disputed 51 See Op-Ed “Africa’s Powerhouse” by Kimenyi and Kibe, 6th January 2014; vote plunges Kenya in bloodshed, 31st December, 2007.Article by J. online at www.brookings.edu. Gettleman. 52 Sundet, Geir, Scanteam, and Eli Moen. 2009. Political Economy Analysis 56 Civil unrest over two months recorded over 1,000 dead and up to 500,000 of Kenya. Norwegian Agency for Development Cooperation Report internally displaced, as per Human Rights Watch: see Report titled “Ballots 19/2009. to Bullets, Organized political violence and Kenya’s crisis of governance”, 53 Ibid NORAD; see sub-section on the “Increasing concentration of powers 16th March, 2008. in the Executive”, pg.6. 57 See Commentary titled “Kenya: A country redeemed after a peaceful 54 The grievances related to access to and ownership of land in the past are election” by Mwangi Kimenyi, April 2013, online on www.brookings.edu. interlinked with political competition along ethnic lines and these have 58 See Part II on the Afrobarometer Survey; source: Afrobarometer Survey resulted in violent ethnic conflict in multi-ethnic areas. See Sub-section 2009: “Popular attitudes toward democracy in Kenya: A summary of 2.6 on Prospects and Risks regarding Kenya in ODI (2014). Afrobarometer indicators, 2003-2008. Published 6th June 2009. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 23 Kenya in Context decisions such as the formation of the judiciary and of distribute revenues to the county governments based the parliament.59 The strong provision for devolution on a weighted allocation (Table 1.3). in the new constitution was a “key source of public Table 1.3: First revenue-sharing formula among support for the draft of the constitution”.60 counties in Kenya Parameter Percentage weight Devolution of power was at the core of the new Population 45 constitution and has fundamentally changed Poverty index 20 the structure of government in Kenya. This major Land area 8 undertaking aimed to address deeply-entrenched Basic equal share 25 Fiscal responsibility 2 disparities between regions, allow for the regions Source: Brookings Institution (2013). to have greater autonomy, and rebalance power away from a historically strong central government. The formula for the horizontal sharing of revenues 61 The general elections of 2013 marked the official emphasizes fiscal need. The formula provides launch of the decentralization as the 47 newly historically marginalized counties with higher per formed counties elected their governors and county capita transfers than historically privileged counties assemblies, and a new national senate was established Land area and population are proxies for the costs of to represent the counties.62 service delivery. South Africa places a similar emphasis on fiscal need, taking a more sectoral approach, Devolved governance presents considerable however they accomplish this by directly measuring opportunities to Kenya in strengthening local the costs of service delivery in the education and autonomy over resource allocations. As per the health sectors. On the other hand, India’s approach to constitution, it was agreed that 84.5 percent of the revenue-sharing places an emphasis on fiscal capacity country’s revenues are to be allocated to the national as opposed to need. government while 15 percent will be allocated to the 47 county governments.63 The remaining 0.5 percent was The horizontal formula for revenue sharing has designated as an “equalization fund”. The Commission been highly equalizing, re-allocating revenues to on Revenue Allocation (CRA), created in the 2010 marginalized areas of the country. In particular, referendum, recommends the basis for equitable northern parts of the country have benefitted revenue allocation to the National Assembly, including significantly, with Turkana and Mandera receiving the percentage of national revenue to be divided higher benefits. Reallocation is envisaged to spur between the national and county governments as well growth in these areas and to contribute to improving as the distribution by county. This is not an easy task as living standards and regional economic convergence. any specific allocation criterion is bound to favor some The reallocation of revenues has also led to a decrease counties over others and therefore raise questions of revenues previously allocated to urban areas, about the legitimacy of the process. The National incentivizing these areas to improve on own-revenue Assembly accepted the CRA’s recommendation to collections by leveraging existing infrastructure. 59 See World Bank report titled Devolution without Disruption: Pathways to a successful new Kenya. November 2012. Continued disparities in capacities will shape 60 Ibid; Chapter One: Kenya’s devolution in context. both utilization of resource allocations and their 61 See Working Paper 1 on Kenya Devolution (Overview Note on building public participation in Kenya’s devolved government), February 2015, by ultimate impact. The generalized approach based the Center for Devolution Studies, Kenya School of Government. on an equitable allocation formula may work in 62 The country Executive arm is headed by the County Cabinet comprised of up to ten members known as the County Executive Committee principle, but the actual sector-wide utilization of (CEC). Each member of the CEC is in charge of a county department (a “ministry”). This apex body along with most administrative organs have resources depends largely on preexisting capacity already been created at the sub-counties, wards and village-level and at the county-level to effectively utilize the allocated counties recruit key personnel to staff the administrative units. 63 See Op-ed titled “Devolution and resource sharing in Kenya” by Mwangi funds. Differences in human resources, technical S. Kimenyi, on the Brookings Institution online, October 22, 2013. 24 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context abilities and existing infrastructure, among others, Devolving authority to county governments has greatly impact the actual cost of delivering specific given rise to new political dynamics that policymakers services under the management of the county need to address. The political decentralization process governments.64 Policymaking has to capture this vital in some cases resulted in hastily-drawn boundaries factor in resource allocation. which formed new administrative arrangements. Inter- county competition is growing over the ownership and National agencies resist handing over vital services control of national and regional development projects and functions to the counties given human that straddle county borders. This makes border regions resource challenges. The reluctance includes key more prone to violent disputes and reprisals against services such as the management of urban and rural minority residents from rival counties. High impact roads and rural electrification projects. This is due to policy interventions are needed to address disputes the limited administrative and technical capacity to between counties, particularly land claims, as well as handle these functions in certain counties. The central improved efforts towards ethnic inclusion at the county government also deployed County Commissioners, level governments. The latter is already under way on an even before the county governments were fully ad hoc basis in the form of a “County Inclusion Index” by established, who answer only to the Nairobi.65 Some the National Cohesion and Integration Commission.68 public sector agencies and their employees, such as doctors and teachers, are reluctant to be managed The 2017 general election renewed the focus on by local government units that are deemed less the presidency and put pressure on the electoral qualified than their national peers, even if the terms process and its governing institutions. The political and conditions of their services remain the same.66 decentralization achieved through the comprehensive devolution that Kenya has recently undertaken should Political and fiscal decentralization enjoys wide in theory mitigate the political stakes of the country’s political and popular support. There is now widespread presidential elections, among other accomplishments acceptance of – and big expectations (Box 1.3) (Box 1.3). The events of the presidential election in from – the devolution process. The demand for fiscal August 2017, however, demonstrate that this process autonomy is reflected in speed at which new county remains a contentious and ethnically polarizing governments have assumed major responsibilities event. This calls into question the effectiveness of and received greater funding in health, agriculture, new agencies formed under the 2010 referendum, and local roads/infrastructure.67 The share allocated to such as the Independent Electoral and Boundaries counties in 2013-14 was more than twice the minimum Commission (IEBC), which may not have exercised their 15 percent required by the Constitution. powers to the full extent possible.69 Box 1.3: Public expectations from devolution Citizens will get better public services: • Citizens will have better opportunities to participate in governance. • Women will have better opportunities in devolved governments. • Better transparency and accountability mechanisms will be put in place. • Minority communities will have better opportunities. • The process will lead to a more cohesive and peaceful nation. • Vices such as corruption and impunity will be minimized. Source: Based on Figure 4 “Kenyan’s [sic] expectations of Devolution”, Society for Development (2012 figures) in Center for Devolution Studies Working Paper 1 (2015). 64 See “Devolution and resource sharing in Kenya”, Op-Ed by Mwangi Kimenyi, 22nd October 2013; online at www.brookings.edu. 68 This body was created as part of the post-2007 elections’ reconciliatory efforts. A key objective now is to ensure that minorities within the 65 ODI (2014). counties are included in the governance structures and are marginalized Ibid. 66 in development efforts. 67 Center for Devolution Studies Working Paper 1 (2015). 69 ODI (2014). KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 25 Kenya in Context Box 1.4: Key features of the 2010 Kenyan Constitution The demands for constitutional reform in Kenya gathered pace in the 1990s. The impetus for these demands lay primarily within marginalized communities who objected to the centralized nature of power in the presidency. There is a widespread belief in politically disenfranchised communities that devolving powers away from the central government will end bias in resource allocation, among other gains. A referendum in 2005 failed to garner enough support for constitutional change, but a subsequent referendum in 2010 allowed for a groundbreaking redrafting of the constitution. This made way for the first change to the constitution since independence. Key features include: • The country’s first Bill of Rights that states the right of every citizen to basic services such as clean water, decent housing, sanitation and food. • A guarantee in principle to access to public resources irrespective of any community’s lack of influence at the national level. • A new, decentralized system of 47 local counties established that replaced eight provinces and 46 districts. Each county government consists of an Assembly and an Executive which are both directly elected by their constituents. • Dilution in the president’s appointive powers which are now subject to consultations with various commissions and require approval by the National Assembly. • The creation of an Upper House of Parliament called the Senate, where county governments have equal representation. • Establishment of the National Land Commission with powers to allocate land and to repossess illegally-acquired public land. This entity also restricts the ability of the President’s office to allocate public land to individuals and parties as done before. • Article 40 of the constitution sets out principles governing land policy while Article 68 directs the parliament to revise and rationalize existing land laws. Crucially, it stipulates that the manner in which land is converted from one category to another for acquisition must be regulated. • Chapter 11 establishes mechanisms for political and fiscal devolution and directives to allocate 15 percent of the public revenues towards the 47 counties annually. • Chapter 12 of the constitution establishes the Commission on Revenue Allocation to oversee an equitable resource- sharing between the center and the county governments. • A central government funding system that considers counties’ population size, area and poverty levels, and acknowledges that counties have autonomy over the design and details of local spending plans. Source: Online article titled “New constitution means major changes for Kenya”. Voice of America, August 11, 2010. Online at www.voanews.com; online article titled “How Kenya is changing under new constitution” Daily Nation online, Friday, August 28th 2015. Online at www.nation.co.ke; online Country Profile on Kenya and related article titled “Kenya’s new constitution brings political change”. Oxford Business Group, February 2017. Online at oxfordbusinessgroup.com/overview. The IEBC faced allegations of procedural and [therefore was] invalid”.72 The IEBC was observed inconsistencies and weak oversight for the 2017 to have clearly ignored electoral laws and procedures.73 elections.70 The commission had initially declared the An election re-run in October 2017 was boycotted by incumbent President Kenyatta the winner with over 54 the opposition, which demanded reforms to the IEBC.74 percent of the vote. The main challenger Raila Odinga 72 See Al Jazeera Opinion piece titled “Why did Kenya’s Supreme Court from the ODM within the larger National Super Alliance annul the elections?”, by Nanjala Nyabola, 2nd September 2017. Online at coalition challenged the results citing hacking and www.aljazeera.com. 73 Ibid.; the tallying website on which local and international reporting manipulation of the electronic vote-counting system.71 relied was not public as was earlier promised; IEBC conceded that they did not use an electronic transmission system to record ballots The Supreme Court nullified the results a month after and used text messages and photographs of manually filled forms as the elections and determined that the process “was sources of information; and, the forms used for reporting results from different regions were apparently not all available in time for the official not conducted in accordance with the Constitution announcement. The total cost of the elections at USD 500 million makes it one of the most expensive, spending USD 28 per capita in taxpayer money. 70 Article titled “What next in Kenya election crisis?”, by Dickens Olewe, 11th 74 This re-run was won by the incumbent with 98 percent of the votes October 2017. Online at www.bbc.com. while the turnout was recorded at 39 percent and the re-run suspended 71 Article titled “Kenyan opposition leader to challenge election result in 25 constituencies that were opposition strongholds. The Supreme in court”, Reuters/The Guardian, 16th August 2017. Online at www. Court upheld the results, which allows the President to serve another theguardian.com. five-year term. 26 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context 1.7 PERCEPTIONS ON DEMOCRACY, Figure 1.41: Perception of democracy in sub- Saharan African countries GOVERNANCE AND POLITICAL In your opinion, how much of a democracy is Kenya/other today? PARTICIPATION75 (% of respondents) K enyans show a strong preference for their democracy and democratic processes. Citizens largely support the nature of democracy in their country and have Zimbabwe Uganda 6 22 31 27 23 41 17 13 favorable attitudes towards processes linked to the Tanzania 2 11 42 17 functioning of a democratic republic, according to the South Africa 7 43 33 15 2016 Afrobarometer survey. Kenyans have a higher Nigeria 17 47 25 8 regard for their democratic system compared to other Kenya 5 21 48 15 countries in sub-Saharan Africa (Figure 1.41); 63 percent of Kenyans see their country as a “full democracy” or a 0 50 100 “democracy with minor problems.” Kenyans also have a Not a democracy Democracy, with major problems favorable view of the overall environment for electoral Democracy, with minor problems A full democracy Source: Afrobarometer Surveys’ “summary of Results” Kenya, Round 7, 2016, politics in the country. Question 35: “In your opinion how much of a democracy is Kenya today?”; under Question 35 for Uganda and Zimbabwe and Question 40 for Nigeria, South Africa and Tanzania. Remaining respondents in all countries were under the categories The majority of citizens responded positively to “Did not understand question/Democracy” or “Don’t know/refused”. These categories comprised 10 percent or less of total respondents in all countries except several fundamental democratic rights in place Tanzania (27 percent). and supported key features of a functioning democracy. In terms of the freedom of opposition processes. Afrobarometer surveys conducted in Kenya parties or candidates to speak or hold rallies and for in 2003, 2005 and 2008 show that 57 percent of Kenyans, the respondents to state their views or criticize the averaged across the three surveys, regarded their government, over 60 percent of respondents thought country as a “full democracy” with “minor problems”.76 that there was “somewhat more” or “much more” A majority of Kenyans – 68 percent (again, averaged freedom than before. Over 70 percent disapproved – from the three surveys) – also agreed that “many 50 percent “strongly” – of an election where only one political parties are needed to make sure that Kenyans political party is allowed to stand and hold office. A have real choices in who governs them”. Additionally, large majority, 83 percent, disapproved – 63 percent on average, over 88 percent of Kenyans in the surveys did so “strongly” – of the army governing the country rejected military rule as an alternative to electoral as an alternative. Democracy was “preferable to any politics. Democracy “was preferable to any other kind other kind of government” to 67 percent of Kenyans, an of government” for 80 percent of Kenyans in 2003, 75 opinion shared by respondents in other sub-Saharan percent in 2005, and for 79 percent in 2008. African countries: this statement is supported by 81 percent of Ugandans, 75 percent of Zimbabweans, 66 The 2008 survey shows ratings drop on the perceived percent of Nigerians, 64 percent of South Africans, and true extent of democracy, the satisfaction with 57 percent of Tanzanians. democracy, and the quality of the electoral process. Nearly 50 percent of citizens thought that Kenya Views before the devolution in 2010 show was “not a democracy or a democracy with major comparable support for democratic norms and problems”, a 19-point increase since 2005. 42 percent of Kenyans were “fairly satisfied or very satisfied” with 75 Data in this section is based on the latest Afrobarometer Survey’s “summary of Results”, undertaken in Kenya as Round 7 in 2016 the way democracy worked in Kenya, an 11-point (conducted September-October 2016) by the Institute for Development drop from 2005. Only 20 percent of Kenyans in 2008 Studies (IDS). Additionally, previous Summary of Results for Kenya from Round 6, 2014 and Round 5, 2011, and, the report “Popular attitudes claimed that the previous (2007) elections were largely toward Democracy in Kenya: A summary of Afrobarometer indicators, 2003-2008”. Data on sub-Saharan countries is based on Summary of Results from Nigeria, Round 6, 2015; South Africa, Round 6, 2015; 76 Source: Afrobarometer Survey report “Popular attitudes toward Tanzania, Round 6, 2014; Uganda, Round 7, 2017; and Zimbabwe, Round Democracy in Kenya: A summary of Afrobarometer indicators, 2003- 7, 2017. Online at www.afrobarometer.org. 2008”, 26th June 2009. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 27 Kenya in Context free and fair. The drop in positive perceptions from 2005 Kenyans in 2016 listed corruption as the “most to 2008 regarding elections was likely informed by the important problem facing the country” that should disputed 2007 elections and the following civil conflict. be addressed by the government. This was followed by unemployment, crime and security, and management Kenyans hold a neutral view of elected officials. of the economy.78 Concern over corruption has Citizens generally believe that the President, MPs, steadily risen for citizens since 2011 (Figure 1.43). A Members of County Assembly and the County majority of Kenyans stated that ordinary citizens were Governor are doing an acceptable job: in terms of how “very likely” to get away with paying a bribe or using key representatives had performed in their job over a personal connections for a) avoiding payment of year in 2016, 75 percent of Kenyans “strongly approved taxes that they owed to the government (66 percent), or approved” of the performance of the President, b) avoiding paying a traffic fine or going to court (70 45 percent did so of the MPs, and 47 percent of the percent), and c) registering land that did not belong to Members of the County Assemblies. them (73 percent).79 Moreover, 77 percent of Kenyans thought that those who report incidents of corruption The level of responsiveness from elected public “risked retaliation”.80 officials towards their constituents is a concern. When asked whether MPs tried their best to listen to what people have to say, 83 percent of Kenyans Responses around corruption also indicate notably thought that MPs “never did or did so only sometimes,” low levels of trust in public institutions. Some of these while 15 percent thought “often or always”. This is institutions are mandated with addressing corruption comparable to the perceived responsiveness to and redressing grievances, such as the police. Most constituents in other sub-Saharan African countries Kenyans reported some level of involvement in (Figure 1.42). The responsiveness of Members of County corruption by major government institutions (Figure Assemblies in Kenya was thought to be marginally 1.44). Additionally, when asked how well they thought better77 even as Kenyans gave a more balanced view of the current government was fighting corruption, over how they performed in 2016. 70 percent thought “very badly” or “fairly badly”. Figure 1.42: Responsiveness of National Assembly members to citizens in sub-Saharan African countries How much of the time do you think the following try their best to listen to what people like you have to say: Members of Parliament /National Assembly? (% of respondents) Zimbabwe 69 20 Uganda 75 22 Tanzania 87 11 South Africa 80 16 Nigeria 85 10 Kenya 83 15 0 20 40 60 80 100 Never/Only sometimes Often/Always Source: Afrobarometer Surveys’ “summary of results”: Under Kenya Round 7, 2016, Question 54A.; Nigeria Round 6, 2015, Question 59A.; South Africa Round 6, 2015, Question 59A.; Tanzania Round 6, 2014, Question 59A.; Uganda Round 7, 2017, Question 54B.; and Zimbabwe Round 7, 2017, Question 54A. 78 Question 55, Pt.1: In your opinion, what are the most important problems facing this country that government should address? (1st response). 79 Afrobarometer (2016); Questions 48D to 48F: respondents were to choose from a) Not at all likely; b) Not very likely; c) Somewhat likely; 77 Afrobarometer (2016); Question 54A “How much of the time do you think d) Very likely; additionally, there were categories of responses under the following try their best to listen to what people like you have to say? “Missing”, “Refused” (-to answer) and “Don’t know/Haven’t heard”. Members of Parliament.” and Question 54B “How much of the time do 80 Question 47: In this country, can ordinary people report incidents of you think the following try their best to listen to what people like you corruption without fear, or do they risk retaliation or other negative have to say? Members of County Assembly.” consequences if they speak out? 28 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Kenya in Context Figure 1.43: Major issues for citizens in Kenya that Figure 1.44: Perceived involvement in corruption, 2016 government should address (% of respondents) 30 Government workers 3 42 39 8 25 24 20 Police 3 25 37 28 20 19 % of respondents 15 15 13 Members of County Assembly 4 42 33 12 12 10 10 10 10 Members of Parliament 4 40 36 11 6 6 5 4 President's O ce 7 45 26 9 0 2011 2014 2016 0 50 100 Management of the economy Unemployment Percent Corruption Crime and Security None Some of them Most of them All of them Source: Afrobarometer survey, “summary of results”, First Response on Question Source: Based on Questions 44A to 44J (How many of the following people do “In your opinion, what are the important problems facing this country that you think are involved in corruption?), Afrobarometer Survey, “summary of results”, government should address?, Round 5, 2011, Round 6, 2014 and Round 7, 2016 Kenya 2016. Figure 1.45: Political intimidation or violence during Figure 1.46: Expressing political views in sub-Saharan election campaigns African countries During election campaigns in this country, how much do you personally How often do people in this country have to be careful of fear becoming a victim of political intimidation or violence? what they say about politics? (% of respondents) (% of respondents) Zimbabwe 23 76 2016 55 43 Uganda 34 65 Tanzania 45 51 2014 40 59 South Africa 56 44 Nigeria 35 64 2011 56 43 Kenya 22 74 0 50 100 0 20 40 60 80 100 A lot/somewhat A littlebit/not at all Never/Rarely Often/Always Source: Afrobarometer survey, “summary of results”, Round 5, 2011, Question 54; Source: Afrobarometer Survey “summary of results” Kenya Round 7, 2016, Question Round 6, 2014, Question 49; and Round 7, 2016, Question 40. 42A. In your opinion, how often, in this country: Do people have to be careful of what they say about politics? Question 42A. in Uganda and Zimbabwe surveys, Question 51A. in Nigeria, South Africa and Tanzania. Kenyans are also more cautious with respect and associating with political organizations (Figure to political participation. A large proportion of 1.46). The percentage of respondents indicating a respondents are concerned with intimidation or cautionary attitude towards associating with political violence during political campaigns in the country organizations has risen considerably over the past (Figure 1.45). A majority, 74 percent, also thought that decade. Attitudes in 2011 and 2014 indicate fewer they “often or always” had to be careful of what they inhibitions related to joining a political organization.81 say about politics. Citizens show inhibitions on other crucial dimensions of a participatory democracy as compared to other sub-Saharan African countries, 81 According to the Afrobarometer “summary of results” responses, 84 percent of Kenyans thought they were “somewhat free/completely free” seen in responses on expressing political views to join any political organization that they wanted to in 2014 (Question 15B.) and 82 percent thought the same in 2011 (Question 17b.). KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 29 CHAPTER 2 THE EXTENT AND EVOLUTION OF POVERTY AND INEQUALITY IN KENYA SUMMARY Kenya recorded steady progress against poverty between 2005/06 and 2015/16. The proportion of the population living beneath the national poverty line fell from 46.8 percent in 2005/06 to 36.1 percent in 2015/16. Most of the poverty decline is attributable to the progress observed in rural areas, where poverty declined from around 50 percent in 2005/06 to 38.8 percent ten years later. This contrasts with the stagnation of poverty in urban areas, particularly outside Nairobi. As Kenya urbanizes, cities are not providing enough economic opportunities for individuals to improve their income levels and maintain their standards of living. The country also experienced shared prosperity, with substantial consumption growth for households in the bottom 40 percent of the distribution. The annualized consumption growth for the bottom 40 percent has been a satisfactory 2.86 percent per year between 2005/06 and 2015/16, a pattern more pronounced in rural areas. Consistent with this pro-poor pattern of economic growth, inequality declined in Kenya, as confirmed by several inequality measures. While this helped to contribute to poverty reduction, most of the reduction is attributable to economic growth; which means that going forward efforts to reduce inequality can help accelerate poverty reduction. The evidence suggests that off-farm diversification has been important for poverty reduction in Kenya. While a robust agricultural sector and a dynamic services sector contributed to the wellbeing of rural households, most of the poverty reduction is accounted by households whose agricultural income is supplemented by non- agricultural activities (small-scale services). There is compelling evidence that the enabling factor was mobile money. M-PESA increased the households’ financial resilience and savings, allowing them to: i) invest productively, ii) move out of agriculture or complement that income with that of other businesses, and iii) improve their consumption levels. Kenya is characterized by stark regional differences, both in terms of monetary and non-monetary poverty indicators. The wellbeing of the population in the North & Northeastern Development Initiative (NEDI) counties (which includes all counties in the North Eastern province) lags considerably behind the rest of Kenya.82 Moreover, these areas have seen little progress between 2005/06 and 2015/16, remain prone to food insecurity, and present very low levels of educational attainment, access to improved sanitation and, to a lesser extent, access to improved water. While the GoK has implemented some measures to improve the connectivity and overall wellbeing of the population in these areas, substantive, sustained and cross-sectorial efforts will be required moving forward. Poor households remain limited by demographic characteristics, low human capital, and low coverage of basic services. Poverty is associated with female and older household heads, and low levels of educational attainment. This suggests that the poor are constrained when accessing income generating opportunities. Moreover, poor households tend to be larger, and have higher dependency ratios; demographic factors that usually hinder poverty reduction. In addition, coverage of WASH services and household electricity is much lower for poor households. In this sense, Kenya should continue to expand the coverage of this basic services to all segments of the population, while ensuring their quality. 82 NEDI group of counties: Mandera, Lamu, Wajir, Garissa, Tana River, Marsabit, Samburu, Turkana, West Pokot and Isiolo (a map displaying the NEDI counties is included in Appendix B). KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 31 The Extent and Evolution of Poverty and Inequality in Kenya This chapter first documents the progress made by the levels and progress of poverty indicators for the Kenya in terms of the monetary measures of poverty, NEDI counties. during the period on which this report focuses, 2005/16 to 2015/16. It analyzes the trends in terms of the national 2.1.1 Progress in the incidence of poverty poverty headcount rate, other related indicators (such Kenya has seen a steady reduction in the poverty as the depth and severity of poverty) and the incidence rate between 2005/06 and 2015/16 but progress of food and extreme poverty, as officially defined is modest. Over that period and consistent with by the KNBS. The chapter then turns to examine the the overall robust economic growth observed83, the incidence of consumption growth, and how this is country has been able to reduce the share of people reflected in terms of an array of inequality indicators. living below the national poverty line by more than ten It also examines the factors behind Kenya’s success in percentage points. The national poverty headcount reducing poverty, relying on decomposition analysis rate went down from 46.8 percent in 2005/06 to 36.1 and the finding of numerous studies on the impact of percent in 2015/16 (Table 2.1), which corresponds to mobile money in the wellbeing of the population. The an annualized rate of poverty reduction of 2.6 percent. chapter concludes by providing a profile of the poor, in Despite this successful reduction in the incidence of an attempt to identify the factors that may be limiting poverty, the absolute number of poor declined only their economic opportunities and overall wellbeing. marginally, from 16.6 million in 2005/06 to 16.4 million ten years later (Table 2.2). A first look at the absolute 2.1 STEADY BUT MODEST PROGRESS number of the poor in Kenya reveals that the number of AGAINST POVERTY 2005/06-2015/16 R Table 2.1: Absolute poverty headcount rate, nationally, by educing the share of the population living under the area of residence poverty line is an important measure of progress Percentage Annualized 2005/06 2015/16 for any country. This section analyzes how monetary point change change poverty has evolved in Kenya between 2005/06 and National 46.8 36.1 -10.7 -2.6 2015/16, looking closely at the spatial disparities both Rural 50.5 38.8 -11.7 -2.6 in terms of the urban and rural divide and of the marked Urban 32.1 29.4 -2.7 -0.9 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 provincial differences. It also pays special attention to Table 2.2: Poor and total populations, nationally, by area of residence and by NEDI classification Population living in poverty Annualized Distribution of poor (%) Percentage (Millions) percentage point change 2005/06 2015/06 change 2005/06 2015/06 National 16.6 16.4 -0.1 100 100 - Rural 14.3 12.6 -1.3 86.2 76.9 -9.3 Urban 2.3 3.8 5.1 13.8 23.1 9.3 Non-NEDI 14.3 13.2 -0.8 85.9 83.1 -2.8 NEDI 2.4 3.2 2.9 14.1 16.9 2.8 Total population (Millions) Annualized Distribution of poor (%) Percentage percentage 2005/06 2015/06 2005/06 2015/06 point change change National 35.5 45.4 2.5 100 100 - Rural 28.4 32.5 1.4 79.9 71.6 -8.3 Urban 7.2 12.9 6.0 20.1 28.4 8.3 Non-NEDI 32.1 39.9 2.2 90.3 88 -2.3 NEDI 3.4 5.4 4.7 9.7 12 2.3 Source: Own calculations based on KIHBS 2005/06 and 2015/16. 83 Except for the economic slow-down that resulted from the events that followed the general elections of 2007 and the slowdown of agricultural production in 2011, described in detail in Chapter 1. 32 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya people living below the poverty line increased in urban many other countries in Africa, where high total fertility and NEDI counties84, from 2.3 to 3.8 million and from 2.4 rates (TFRs) are undermining growth and poverty to 3.2 million respectively, whereas it decreased in rural reduction, as documented by Beegle and Christiaensen and non-NEDI counties. (forthcoming). In the case of Kenya, the average TFR is estimated at 3.9 children per woman in 2014 (Figure Fertility trends in Kenya have not undermined the 2.1), much lower than the 4.85 estimated for Sub- progress against poverty, as has been the case in Saharan Africa. This also means that fertility declined by many countries in Africa. While the small decline in almost one birth per women over the decade leading the number of poor may not appear as major progress, to 2014, a notable accomplishment. in this sense Kenya presents a better outlook than Figure 2.1: Total Fertility Rate (women aged 15-49) a) Kenya b) Benchmark countries most recent DHS year 6 6 5.4 5.4 5.2 5.2 4.9 5 5 4.6 4.6 4.5 4.2 4.2 3.9 4 3.9 Births per woman 4 Births per woman 3.3 3.1 2.9 3 3 2.6 2 2 1 1 0 0 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Kenya Ethiopia Ghana Rwanda South Tanzania Uganda (2014) (2016) (2014) (2015) Africa (2016) (2016) National Urban Rural (2016) Source: KDHS 2003, KDHS 2009 & KDHS 2014. Source: KDHS 2014, EDHS 2016, GDHS 2014, RDHS 2014-15, SADHS 2016, TDHS 2015-16, UDHS 2016. Box 2.1: Kenya Integrated Household Budget Survey (KIHBS): A commendable effort The analysis of this chapter and most of this report would not be possible without the recent effort by the KNBS to collect the 2015/16 wave of the KIHBS, which comes ten years after the collection of the first wave. Without this effort, it would not be possible to assess with certainty what are the living standards of the Kenyan population along many dimensions, including monetary and non-monetary poverty measures. While both waves are representative at the national, urban/rural, and provincial level, the 2005/06 KIHBS is also representative of Kenya’s 69 districts, and the 2015/16 KIHBS of the 47 counties introduced by the 2010 constitution.85 In addition to reporting statistics by urban and rural areas and by province, this chapter also refers to the NEDI group of counties. These are historically underdeveloped areas and, as will be shown throughout the chapter, lag behind the rest of the Kenya on a wide range of socio-economic indicators. The ten NEDI counties are Mandera, Lamu, Wajir, Garissa, Tana River, Marsabit, Samburu, Turkana, West Pokot and Isiolo (a map displaying the NEDI counties is included in Appendix B). 85 There are two additional differences in the sampling framework of the two waves. Firstly, the 2005/06 survey had 10 households per cluster and an additional 5 replacement households, whereas the 2015/16 KIHBS had 84 However, the number of the poor still grew at a slower pace than the the same number of households per cluster without any replacements. total population, which explains why the proportion of the poor did not Secondly, the 2015/16 KIHBS covered a larger sample: around 21,700 go up. households versus 13,100. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 33 The Extent and Evolution of Poverty and Inequality in Kenya While the reduction in poverty was more pronounced roughly one in ten poor lived in urban areas, by 2015/16 in rural areas, this is where three quarters of the this proportion was close to one in four (Table 2.2). This, poor still live. Poverty incidence in Kenya is still higher in addition to the increase of the absolute number of in rural areas than in urban areas, but it was in rural urban poor, indicates the economics benefits of the areas where the largest decline occurred. During progress observed at the national level are not reaching the ten-year period, rural poverty declined by nearly the poorest households in urban centers, particularly 12 percentage points from 50.5 percent in 2005/06 outside Nairobi, as explored in Chapter 5 of this report. to 38.8 in 2015/16. In contrast, there was little or no progress in urban areas: poverty declined by less Moreover, Kenya has been able to reduce the than 3 percentage points, but the difference is not incidence of food poverty and extreme poverty. statistically different from zero (Figure 2.2a). This Following the KNBS definitions, food poverty is translates into an annualized poverty decline that defined as the share of the population whose food is three times as large for rural Kenya (2.9 percent consumption is below the food poverty line, while versus 0.9 percent). This is explained by an increased extreme poverty is defined as proportion of the diversification of non-farm income sources of rural population whose total consumption (including households, particularly in the services sector, paired food, rent, clothing, energy, health expenditures, with a robust performance of the agricultural sector and education) is below the food poverty line. Both for the better part of the period studied. measures serve as an indication of food security at the household level, and how difficult is for households Poverty is increasingly becoming a concern for to fulfill the minimum caloric requirements. The share Kenya’s urban areas. The distribution of the poor of food-poor people has declined from 44.4 percent population between rural and urban areas changed in in 2005/06 to 32 percent in 2015/16 — a roughly 28 line with the distribution of the total population and percent decline, slightly steeper than the absolute the little progress made in urban areas. While in 2005/06 poverty reduction. Similarly, extreme poverty fell by Figure 2.2: Trends in absolute, food and extreme poverty, nationally and by area of residence a) Absolute poverty b) Food poverty c) Extreme poverty 60 80 60 50.5 50 46.8 50 Proportion of the population Proportion of the population 60 Proportion of the population 38.8 40 36.1 48.3 40 32.1 44.4 29.4 30 40 30 35.1 32.0 29.1 23.0 24.4 19.6 20 20 20 10.7 10 8.6 10 6.0 3.4 0 0 0 National Rural Urban National Rural Urban National Rural Urban 2005/06 2015/16 2005/06 2015/16 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and 2015/16. Note: Lines denote the 95% confidence interval for the statistic. 86 There are two additional differences in the sampling framework of the two waves. Firstly, the 2005/06 survey had 10 households per cluster and an additional 5 replacement households, whereas the 2015/16 KIHBS had the same number of households per cluster without any replacements. Secondly, the 2015/16 KIHBS covered a larger sample: around 21,700 households versus 13,100. 34 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya more than half: from 19.6 percent in 2005/06 to 8.6 the food poverty rate of urban areas as the difference percent in 2015/16 (Figure 2.2 c). In both cases, the between the two years is not statistically significant progress was mainly observed in rural areas. In the (Figure 2.2 b). case of food poverty, it seems there was no change in Box 2.2: Measuring poverty: Computing the poverty lines, the consumption aggregate and classification of peri-urban households Poverty lines The food and absolute poverty lines calculated with the 2015/16 KIHBS follow the Cost of Basic Need (CBN) method outlined in Ravallion (1998). The CBN method defines a consumption bundle required to meet one’s ”basic consumption needs.” The cost of this consumption bundle is then estimated using reference prices for either rural or urban areas. The rural and urban food poverty lines in each survey are determined using the cost of a food basket which meets the 2,250 kilocalorie requirement per adult equivalent. The rural and urban absolute poverty lines are then calculated by adding a minimum allowance for non-food consumption to their respective food poverty lines. While the same methodology had been used in 2005/06 to obtain the food poverty and absolute poverty lines, once the 2015/16 KIBHS was implemented it became evident the changes in the composition and in the relative importance of items within the food consumption basket would require a recalculation of the food poverty line (Figure 2.3). This is not surprising, as ten years later consumer preferences are different and there is larger choice set available to households. To obtain comparable estimates over time, the 2015/16 lines were deflated and revalued at 2005/06 prices. More specifically, the food poverty line is obtained using the 2015/16 basket of food items (and the weights within the basket) at their 2005/06 prices. The non-food component of the line is deflated using the official CPI. Figure 2.3: Urban and rural food poverty basket comparison by rank, 2005/06 and 2015/16 2005/06 2015/16 2005/06 2015/16 1 1 1 1 Beef with bones Beef with bones 2 2 2 2 Sugar 3 Sugar 3 4 4 Fresh cow milk 4 4 5 Fresh cow milk 5 (Unpacketed) (Unpacketed) 6 6 6 6 Bread Bread 7 7 7 7 Sifted maize our 8 Sifted maize our 8 9 9 9 9 Fresh cow milk Fresh cow milk (Packeted) 10 (Packeted) 10 Tomatoes Tomatoes Non-aromatic white rice 13 Non-aromatic white rice 13 Loose maize grain 14 14 Loose maize grain Kale (sukuma wiki) Kale (sukuma wiki) 17 17 18 18 Source: KNBS 2018 There is a minimal difference at the national level between the 2005/06 poverty rates resulting from the noncomparable lines (the 2005/06 poverty line) and comparable lines (using the 2015/16 basket of food items at 2005/06 prices). The absolute and extreme poverty rates calculated using comparable poverty lines are just 0.2 and 0.1 percentage points higher, respectively, than when calculated using the original 2005/06 poverty lines (Table 2.3). Nationally, food poverty is 1.4 percentage points lower due to the drop in the urban food poverty line, which also results in a reduction in the urban extreme poverty rate from 8.3 percent to 6.0 percent. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 35 The Extent and Evolution of Poverty and Inequality in Kenya Table 2.3: Comparison of noncomparable and comparable 2005/06 poverty rates 2005/06 2005/06 Extreme poverty rate (%) (Noncomparable) National 46.6 46.8 Absolute poverty rate (%) Rural 49.7 50.5 Urban 34.4 32.1 National 45.8 44.4 Food poverty rate (%) Rural 47.2 48.3 Urban 40.4 29.1 National 19.5 19.6 Extreme poverty rate (%) Rural 22.3 23.0 Urban 8.3 6.0 Source: Own calculations based on KIHBS 2005/06 and 2015/16. Consumption aggregate The consumption aggregate in both surveys was constructed using the approach outlined in Deaton & Zaidi (2002). The food aggregate uses a recall period of 7 days and comprises food consumption from four sources, namely: purchases, own production, own stock and gifts. Prices were imputed using the cluster-level median for each item since a household may have consumed but not purchased an item and household-level prices may contain outliers. The non-food component of the aggregate includes consumption of energy, education, transport and clothing among other item groups. Housing rent is also included in the non-food component, however only for urban households, wherein the rent is imputed for households that own their dwelling. Over-the-counter medication (items such as cough syrup, painkillers and anti-malaria medicine) is the only form of health expenditure included the non-food aggregate. Lastly, in each survey in order to account for spatial and temporal food price differences, a household-level price deflator based on a Paasche price index was created. Spatial adjustment occurs as the cluster median prices are referenced to the overall rural or urban median prices. Temporal adjustment occurs as each cluster is surveyed in a 2-week period within a year and these prices are then referenced to the median price for the entire survey period. This adjusts for differences in the cost-of-living within urban and rural areas after it is applied to the nominal food and total aggregates. Peri-urban classification Peri-urban households were classified as rural households in the 2005/06 KIHBS survey for the purpose of generating a consumption basket used to create the food poverty lines as well as for the spatial price deflator and the calculation of poverty rates. However, after Kenya’s 2009 Population Census, the KNBS established that the urban category should include peri-urban households. For this report, and after a careful analysis of the characteristics of the peri-urban households in the KIHBS 2015/16, we classify peri-urban households as being rural (as in the 2005/06 KIBHS). As seen in the Appendix B, the socio- economic conditions of these households are closer to their rural counterparts than their core urban counterparts. Thus, using the urban poverty line to identify if these households are poor would not be appropriate and would result in an underestimation of the welfare of these households. 36 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya 2.1.2 Regional patterns in poverty and poverty fact that the economic progress observed during this reduction period is not reaching all areas of the country, and it While poverty fell in every province, there are large validates the recent effort of the government to invest spatial differences in the poverty levels and changes these regions. In addition, female headed households across the different provinces of Kenya. Figure 2.4a in NEDI counties exhibit higher poverty rates (absolute, shows the striking provincial variation in the poverty food and extreme poverty) than in the rest of Kenya. incidence across the different provinces of Kenya: while 70 percent of the population in the North Eastern Food and extreme poverty are highly heterogeneous Province live in poverty, that is true for only 16.7 of the across different provinces. While Nairobi enjoys a population in Nairobi. Moreover, this former province food poverty rate that is close to being only half of the barely saw any progress over the period of focus of this national average (16.1 percent, down from 20 percent study, with poverty declining from 74 to 70 percent in 2005/06) and it has almost eliminated extreme between 2005/16; representing the lowest annual poverty, the North Eastern Province performs drastically reduction rate for all provinces (around 0.6 percent per worse with half of the population being food poor and year). On the contrary, the Eastern and Coast provinces one in four in extreme poverty. Interestingly, these exhibited the largest reductions in the incidence of two extreme cases (the worst- and best- performing absolute poverty (of around 18.8 and 17.1 percentage provinces) have the lowest rates of progress in the points), with annual reduction rates of 4.5 and 3.5 country (Figure 2.4b and c). percent respectively. These two provinces account for around 43 percent of the poverty decline in the country. It is clear that the NEDI counties are prone to food insecurity. Food poverty and, particularly, extreme Poverty incidence in the NEDI counties is significantly poverty, are remarkably high in NEDI counties when higher than in the rest of the country. Remarkably, the compared to the rest of the country. For 55.4 percent poverty rate amongst the NEDI counties in 2015/16 is of the population in these counties food expenditure more than double that of the rest of the country, 68.0 is not sufficient to reach the minimum caloric percent versus 32.6 percent (Figure 2.4a). Moreover, requirement (compared to 29.5 percent for the non- progress has been slow: while the non-NEDI poverty NEDI counties Figure 2.4b). Also, as shown in Figure 2.4c headcount rate fell by 3 percent annually, it only fell for 31.8 percent of the population, even if they devoted by 1.1 percent in the NEDI counties. This reflects the their entire budget into food, this would still not suffice Figure 2.4: Trends in absolute, food and extreme poverty by province and NEDI classification a) Absolute poverty 74.0 76.2 80 70.0 68.0 57.6 Proportion of the population 54.2 60 50.6 49.0 47.2 40.5 41.4 42.5 44.2 36.7 40 31.8 31.1 32.6 24.3 21.3 16.7 20 0 Coast North Eastern Central Rift Valley Western Nyanza Nairobi Non-NEDI NEDI Eastern 2005/06 2015/16 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 37 The Extent and Evolution of Poverty and Inequality in Kenya b) Food poverty 80 65.2 68 52.7 51.5 55.4 51.4 Proportion of the population 60 47.9 45.3 45.4 37.9 42.3 38.8 32.1 35.8 32.9 40 28.5 29.5 22.4 20 16.1 20 0 Coast North Eastern Central Rift Valley Western Nyanza Nairobi Non-NEDI NEDI Eastern 2005/06 2015/16 c) Extreme poverty 80 Proportion of the population 44.6 49.6 60 40 26.7 31.8 24.0 22.7 23.3 19.6 20.2 16.9 20 12.1 13.6 11.1 6.0 6.6 6.1 6.1 2.9 2.8 0.6 0 Coast North Eastern Central Rift Valley Western Nyanza Nairobi Non-NEDI NEDI Eastern 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. (compared to 6.1 percent for non-NEDI). Moreover, the decline in the Eastern province, which as mentioned progress in these counties has been slower than in had a stellar performance in terms of poverty reduction. the rest of the counties. Not being able to attain the Looking at the distribution is worth noting that, given nutrition requirements has severe consequences on its high poverty incidence, the North Eastern province health, productivity and the accumulation of human concentrates a higher share of the poor (close to 7 capital among children, which results in poverty traps percent) compared to the share of the total population that are difficult to overcome. (3.5 percent). The majority of the poor reside in the Rift Valley, It is clear that the national poverty estimates mask followed by the Nyanza and the Western province. stark spatial disparities across the different regions. One third of the poor population resides in the Rift Historically, provincial disparities have been marked in Valley, the most populated province of Kenya, followed Kenya, partly explained by climatic and agro-ecological by Nyanza, accounting for 15 percent of the poor, and differences that affect agricultural productivity, partly the Western province, with 12.7 percent. Overall, as by differences in infrastructure and access to public seen in Figure 2.5, the distribution has not changed services (as will be shown later in the chapter), and much in the past ten years, except for a substantial partly by the differences in political representation 38 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.5: Distribution of the poor by province 2005/06 2015/16 3.6% 4.5% 11.4% 10.4% 14% 14.2% 4.9% Coast 6.8% North Eastern Eastern 11.9% Central 14.5% 17.7% 12.7% Rift Valley Western 7.5% Nyanza Nairobi 8.2% 25.8% 32% Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. and participation in the decision-making process as poverty line. During the period of focus of this study, discussed in Chapter 1 (World Bank 2008). Making inequality amongst the poor declined nearly by half sure that all regions are part of the economic from 8.2 to 4.5 percent. development process and benefit from it will be an important part of sustaining the poverty reduction As with the poverty headcount rate, urban areas effort moving forward. saw less progress in terms of the depth and severity of poverty. Analyzing the poverty gap and poverty 2.1.3 Poverty depth and severity: How far are the severity for the urban and rural population, once again poor below the poverty line and how much inequality amongst the poor is there? it is observed that the decline is steeper amongst rural households. The gap went down from 18.2 to 11.0 Both the depth and severity of poverty have declined percent over the last ten years, while in urban areas the in Kenya. The depth of poverty is represented by how far, on average, the poor fall below the poverty line, and decline was only 1.7 percentage points from 10.6 to 8.9 is expressed as a percentage of the poverty line value. percent. Similarly, rural severity halved from 9.2 percent This is also known as the poverty gap and serves to in 2005/06 to 4.7 in 2015/16, a level similar to that measure the intensity of poverty in a given population. observed in urban areas (Figure 2.6). In short, in terms Between 2005/06 and 2015/16, this measure fell from of how far the poor are below the poverty line and 16.7 percent to 10.4 percent for Kenya as a whole (Figure how much inequality exists amongst the poor, rural 2.6). In other words, if transfers could be perfectly and urban households currently look quite similar. The targeted, it would take a transfer of roughly 10.4 percent same cannot be said of NEDI and non-NEDI counties, (KSh 407) of the poverty line to each poor individual to where a striking contrast arises. Poverty depth in NEDI eradicate poverty. Another alternative indicator is the countries is a staggering 28.7 percent in 2015/16 (Figure poverty gap squared – or severity of poverty – which 2.7), significantly higher than in non-NEDI counties, describes inequality amongst the poor by placing a meaning that the effort needed to lift households out greater weight on individuals who are further from the of poverty in these areas will be considerable. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 39 The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.6: Poverty depth and severity, nationally and by urban/rural strata 20 18.2 16.7 15 11.0 10.6 10.4 10 9.2 8.9 8.2 4.5 4.7 5.0 5 3.9 0 2005/06 2015/16 2005/06 2015/16 2005/06 2015/16 2005/06 2015/16 2005/06 2015/16 2005/06 2015/16 Depth Severity Depth Severity Depth Severity National Rural Urban Source: Own calculations based on KIHBS 2005/06 and 2015/16. Figure 2.7: Poverty depth by province and NEDI classification 80 Proportion of the poverty line 60 40 38.6 33.6 28.7 25.7 20.8 20 18.1 17.7 18.9 16.7 14.8 14.0 12.0 10.7 10.2 8.5 9.0 8.4 5.5 6.4 3.4 0 Coast North Eastern Central Rift Western Nyanza Nairobi Non- NEDI Eastern Valley NEDI 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and 2015/16. 2.1.4 Consumption patterns rural households than urban households. Nonetheless, The share of consumption spent on food has consistent with a lower level of wellbeing, rural increased for households across Kenya. Despite the households allocate more of their consumption to food reduction in poverty, the average share of consumption than urban households. devoted to food has risen by 3.3 percentage points, from 51 in 2005/6 to 54.3 in 2015/16 percent nationally Share of consumption on rent (mainly for urban (Figure 2.8). A contributing factor to this phenomenon households), education and energy increased is that food prices increased at a faster rate compared marginally. The share of consumption spent on rent to non-food prices during that period. As depicted in for urban households87 also increased slightly – from Figure 2.9, while the cumulative inflation (based on 14.1 to 15.1 percent (Figure 2.8). While the increase is the overall CPI) over this period was 134 percent, food not alarming, the housing deficit in urban Kenya is well inflation was significantly higher at 219 percent. The documented, and for the majority of poor households relative increase in food prices likely benefited net-food the housing conditions in which they live, and the producer households and hurt urban households in service accessibility do not correspond to the prices the lower part of the distribution (as will be explored paid (World Bank 2018b). in Chapters 5 and 6 of this report, respectively), which 87 As determined by the KNBS the consumption aggregate for rural helps to explain why poverty declined faster among households does not include rent. 40 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.8: Proportion of consumption by use, nationally and area of residence National 2005/06 Rural 2005/06 Urban 2005/06 30.4% 26.7% 34.3% 39.4% 51.0% 4.9% 5.3% 61.8% 6.6% 5.6% 6.6% 6.9% 6.5% 14.1% National 2015/06 Rural 2015/06 Urban 2015/06 24.3% 22.3% 25.9% 46.6% 7.3% 6.0% 5.0% 54.3% 6.6% 63.8% 7.0% 7.4% 8.4% 15.1% Food Rent Education Energy Others Source: Own calculations based on KIHBS 2005/06 and 2015/16. Figure 2.9: Differential changes in price indices 350 2 319.2 290.0 1.9 300 260.3 1.8 CPI Index (base year (2005 = 100) 239.5 250 Food / overall price ratio 223.3 1.7 203.0 233.8 1.6 200 219.9 168.5 206.3 159.1 193.1 1.5 137.8 182.6 150 167.0 1.4 107.5 112.0 140.7 146.4 100.0 128.5 1.3 100 110.6 106.0 100.0 1.2 50 1.1 0 1 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Food Overall Food / Overall price ratio Source: Own calculations based on KNBS 2017. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 41 The Extent and Evolution of Poverty and Inequality in Kenya 2.2 THE INCIDENCE OF PROGRESS, SHARED 2.2.1 Incidence of progress PROSPERITY AND INEQUALITY Overall, households in the bottom of the distribution, W hile poverty is an important measure of how Kenyan particularly the bottom 20 percent, have experienced living standards have improved, understanding substantial growth in real consumption over the if economic progress has reached all segments of the last ten years. Growth incidence curves (GICs), which population and how the distribution of consumption display annualized consumption growth over the entire has changed over time is also important. This section distribution of the population, reveal that economic takes a closer look at which parts of the distribution have growth in Kenya has been pro-poor from 2005/06 to benefitted the most from economic progress experienced 2015/16 (Figure 2.10a). The lower tail of the distribution, by the country between 2005/06 and 2015/16, focuses on particularly below the 20th percentile, experienced the consumption growth if the bottom 40 percent88 and annualized growth rates of around 3-4 percent. These analyzes changes in consumption distribution in rural and growth rates decline monotonically towards the upper urban areas. tail of the distribution, reaching 2.86 percent at the Figure 2.10: GICs nationally, by area of residence and NEDI classification a) National b) Rural 5 Annualized % change in real consumption Annualized % change in real consumption 4 4 3 3 2 2 1 1 0 0 -1 0 20 40 60 80 100 0 20 40 60 80 100 Share of population ranked, percent Share of population ranked, percent c) Urban d) Non-NEDI Annualized % change in real consumption Annualized % change in real consumption 4 4 3 3 2 2 1 1 0 0 -1 -1 0 20 40 60 80 100 0 20 40 60 80 100 Share of population ranked, percent Share of population ranked, percent e) NEDI 8 Annualized % change in real consumption 7 6 5 4 3 2 1 0 0 20 40 60 80 100 Share of population ranked, percent Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. 88 This group is the focus of the World Bank’s Group goal of shared prosperity. 42 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya 40th percentile and 2 percent for the 70th percentile of when looking at the 40th percentile of the distribution the population. The rates become negative at the very (Figure 2.10b). Pro-poor consumption growth is also top of the distribution, but this might be related to the observed in urban areas, although the growth rates fact that the 2015/16 KIBHS suffered from very high are less spectacular when compared to their rural nonresponse rates in households at the top of the counterparts. Given that the nonresponse problems of distribution in Nairobi, as explained in detail in Box the 2015/16 KIBHS mainly affected households at the 2.3. Given this nonresponse issue, the data are likely top quintile of the consumption distribution (See Box underestimating the consumption levels and thus 2.3), the conclusion that economic growth benefitted the growth rates for the top two deciles in Nairobi the bottom of the distribution in urban settings still (see Figure 2.11a). However, this issue does not remains true. affect the bottom part of the distribution and given the rather steep decline of the GIC up to the 80th In NEDI counties, households at the lower end of percentile, it is clear that economic progress over the distribution also experienced a much higher the past ten years has benefitted the poor, and even consumption growth. Looking at GIC for NEDI counties among the poor, it has disproportionally benefitted separately, it is worth mentioning that households in the poorest of the poor. the bottom of the distribution experienced substantial annualized real consumption growth. Growth for the While consumption growth in rural areas was higher 10th percentile was close to 8 percent while at the 40th for the poor, consistent with the impressive decline percentile, it was around 3.5 percent. Nonetheless this of poverty incidence, all households along the was not translated into a substantial poverty decline, distribution experienced consumption growth since as expected, given how far below the national poverty 2005/06. Despite varying performance, no percentile line are the poor in these counties. Moreover, it is only in rural areas experienced negative real consumption for these counties that we do not observe a decline growth, and the average annualized change is roughly in real growth at the very top of the distribution, and 1.5 percent p.a. for rural households. The highest growth consumption growth for households at the very top rates took place for the poorest ten percent of the was above the average (which is represented by the population at around 4 percent, while this rate halves horizontal red line in Figure 2.10e). Figure 2.11: Real consumption deciles (2016 prices), nationally and by area of residence a) National b) Urban c) Rural 10 10 10 9 9 9 8 8 8 7 7 7 Declie Decile Decile 6 6 6 5 5 5 4 4 4 3 3 3 2 2 2 1 1 1 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 0 15,000 30,000 45,000 60,000 0 5,000 10,000 15,000 20,000 Mean consumption per decile (2016 KShs) Mean consumption per decile (2016 KShs) Mean consumption per decile (2016 KShs) 2015/16 2005/06 2015/16 2005/06 2015/16 2005/06 Source: Own calculations based on KIHBS 2005/06 and 2015/16. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 43 The Extent and Evolution of Poverty and Inequality in Kenya Box 2.3: Nairobi nonresponse rates – dealing with data issues The KIHBS 2015/16 survey had an irregularly elevated level of nonresponse among households in Nairobi: only 3 out of 4 households (76.9 percent) in the capital successfully completed the questionnaire, whereas the response rate was between 81.9 and 96.5 percent in the rest of the country (see Appendix B)89. The high nonresponse rate (both at the item and household level), coupled with the non-replacement of unsuccessful interviewed households, likely caused the survey to not accurately capture the upper end of the consumption distribution. Survey response probabilities usually fall with rising incomes/consumption and if this is not adequately addressed in the sampling strategy, reported mean consumption and inequality measures are likely to be underestimated. Fortunately, the nonresponse can generally be expected to leave all poverty measures widely unaffected (Korinek, Mistiaen, and Ravallion 2006). Figure 2.12 below shows the response rate and median consumption by county for all urban households, where each scatter point is weighted by the proportion of total urban households the county represents. The linear trend line shows that in counties with higher median consumption, response rates tended to be lower and it is expected that the same occurs at the household level. Thus, most likely, the nearly 25 percent nonresponse rate in Nairobi was concentrated among wealthier households. Detailed analysis of asset ownership patterns by consumption quintile provides further evidence for the hypothesis that the missing data stems disproportionally from the upper tail of the distribution (Appendix B). For all of considered assets (house, fridge, sofa, car and washing machine) ownership falls dramatically between the 2005/06 and 2015/16 surveys within the top quintile (and in some instances, within the top two quintiles), which is unlikely to occur. According to the data, house ownership fell from 21.4 percent in 2005/06 to 8.8 percent in 2015/16 in the top quintile and car ownership declined 36.8 to 22.7 percent (Appendix B). Unfortunately, it is then likely that the consumption level for the top two deciles in Nairobi is underestimated. Thus, the staggering decline of almost 60 percent in the real consumption of the 10th decile (wealthiest ten percent of the population of Nairobi), as well as the 10 percent decline for the 9th decile is likely overstated. As mentioned, while the poverty estimates are likely unaffected, this does affect the inequality estimates. For that reason, the national and urban inequality measures most likely will overestimate the reduction in inequality, despite the fact that inequality did decline over the period of interest, as consumption growth was more prominent among the poorest households both in rural and urban areas. Household survey data in emerging countries is widely Figure 2.12: Response rates and consumption among known to underestimate top levels consumption urban households and inequality (Assouad, Chancel, and Morgan 2018). 12000 Nairobi Embu Nonresponse, both item and household nonresponse, is Tharaka Nithi Median consumption per county (2016 Kshs) Kirinyaga a crucial factor contributing to this challenge (Medeiros, Meru Nyeri Mombasa 10000 de Castro, and de Azevedo 2016). In countries like Brazil, Kitui Machakos Makueni Baringo India and South Africa, tax records have been used to i) Nakuru Kiambu Lamu Nyamira Trans Nzoia Kericho Kwale Laikipia verify that household survey data was indeed not properly UAsin Gishu Samburu Kisumu 8000 Migori capturing the income and consumption levels of the West Pokot Elgeyo Marakwet Kakamega Nyahururu top part of the distribution, and more importantly, ii) to Busia Homa Bay estimate more accurate inequality estimators through a 6000 Marsabit Garissa Vihiga Tana River combination of imputation and reweighting techniques. It Mandera Wajir is important to further study if similar techniques could be 4000 implemented in the case of Nairobi and Kenya in general, Turkana 70 75 80 85 90 95 in order to obtain a more accurate measure of inequality Response rate (%) - urban in the country. Source: Own calculations based on KIHBS 2015/16. 89 Unfortunately, these data are not available for the 2005/06 KIBHS. 44 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya 2.2.2 Shared prosperity Eastern and Coast provinces – also saw the greatest Kenya is making satisfactory progress in fulfilling the increases in consumption amongst the bottom 40 shared prosperity goal: promoting the consumption percent. The respective annualized rates at 4 and 4.5 growth of the bottom 40 percent of the distribution. percent were above the national average of 2.86 for The annualized growth rate of Kenya’s bottom the period 2005/06 to 2015/16. Nairobi, an entirely 40 percent of the population was 2.86 percent urban province, saw the lowest consumption growth for the period between 2005/06 and 2015/16. for the bottom 40 percent, at an annualized growth Consistent with the GICs shown in the previous rate of 1.3 (Figure 2.13) (this is not at all affected by the section, consumption growth amongst the rural nonresponse issue). Growth and economic progress in bottom 40 percent was 2.5 times higher than for the Kenya was less broad-based in the urban areas, which urban counterpart (2.4 percent versus 0.9 percent). might help to reduce the urban-rural gap but is not Diversification of income sources off-farm, together consistent with an outlook in which cities are centers of with high food prices during this period, benefitted progress for everyone. rural households more than urban households in this distribution bracket. The resulting rural shared While Kenya’s shared prosperity growth indicator is prosperity premium (calculated as the difference low when compared to other sub-Saharan African between the growth rate of the bottom 40 countries over comparable periods, economic percent and the average growth rate for the whole progress has been concentrated in this lower distribution) is estimated at around 1 percentage segment of the distribution. While the annualized point (Figure 2.13) This number is likely to be close to real consumption growth for the bottom 40 percent the shared prosperity premium for Kenya as a whole in Kenya was 2.9 between 2005/16 and 2015/16, most over this period.90 countries in the region have experienced higher growth amongst households in this segment. It reached 4.6 In all provinces in Kenya real consumption growth for percent for Rwanda between 2005 and 2010, 3.51 in the bottom 40 percent was positive and higher than Uganda over the period 2005 to 2012 in Uganda and for the top of the distribution. However, there were an astonishing 9.76 percent for Tanzania between 2007 marked differences across provinces. Those provinces and 2011 (Figure 2.14). However, economic growth in that saw the largest reduction poverty – mainly the Kenya has been markedly pro-poor, and the estimated Figure 2.13: Annualized consumption growth, nationally, by area of residence and by province 5 4.5 4.0 Annualized % change in mean consumption 4 3 2.9 2.8 2.9 2.6 2.6 2.4 2.4 2.2 2 1.9 1.8 1.5 1.7 1.6 1.2 1.1 1.2 1.3 1.3 0.9 1.0 1.0 1.0 1.1 1.1 1 0.8 0 National Urban Rural Coast North Eastern Central Rift Western Nyanza Nairobi -1 Eastern Valley -2 -3 Bottom 40% Total pop. Premium Source: Own calculations based on KIHBS 2005/06 and 2015/16. Note: due to the non-response issues in Nairobi, not all categories report all three indicators. 90 If taking the average consumption growth of 1.1 at face value, the shared prosperity premium is 1.8. However, given that the consumption growth of the top deciles is underestimated (because of the nonresponse rates in Nairobi), the true premium should be much lower than that. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 45 The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.14: Annualized consumption growth compared to benchmark countries 12 Annualized % change in mean consumption 9.8 10 9.1 8 6 4.6 4.3 4.4 4 3.4 3.6 3.5 2.9 1.8 2 1.1 1.2 1.4 0.7 0.7 0 -0.4 -0.9 -2 -1.8 -4 Kenya Rwanda South Africa Uganda Tanzania Ethiopia (2006 -2016) (2005 - 2010) (2005 - 2010) (2005 - 2012) (2007 -2011) (2005 - 2010) Bottom 40% Total pop. Premium Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 shared prosperity premium (again defined as the improvement for an indicator that is usually very stable difference between the growth rate of the bottom 40 over time. This suggests that redistribution contributed percent and the average growth rate for the whole positively to the substantial poverty reduction distribution) of one percentage point is higher than observed in Kenya’s rural areas during this period. In in all benchmark countries except for Rwanda (with a terms of provincial heterogeneity, inequality declined premium of 1.2 percentage points). faster in the Coast province (from 0.43 to 0.38), in the Central region (from 0.38 to 0.34), and to a lesser extent 2.2.3 Inequality indicators in the North Eastern and Rift Valley provinces (Figure 2.15). The level of inequality in Kenya as measured by Inequality in Kenya declined between 2005/06 and the GINI index is moderate and comparable to that of 2015/16, as confirmed by different measures. The nonresponse problem identified for Nairobi does Tanzania, Uganda and Ghana, and is much lower than affect the precision of some of the measures of South Africa’s index of 0.63 (Figure 2.16). inequality at the urban level using the KIBHS 2015/16, and thus, at the national level. However, the collection Alternative measures of inequality confirm an of the evidence presented in this section indicates improvement in Kenya’s distribution. The Atkinson that inequality in Kenya has declined at the national index, which at high levels of the inequality aversion level since 2005/06, in line with the pro-poor pattern parameter gives more weight to the lower consumption of economic growth described by the incidence levels making the measure less sensitive to issues at the curves of Section 2.2.1 and contributing to the poverty top of the distribution, confirms that the Consumption reduction observed. distribution has improved. At an inequality aversion parameter (∑=1), the Atkinson index declined from 0.3 The decline in the Gini index indicates an in 2005/16 to 0.23 in 2015/16 (Figure 2.17b). This means improvement in the distribution of resources in that in 2015/16, Kenya should be willing to forgo Kenya. The Gini index, which is generally not heavily 23 percent of its consumption to achieve a uniform affected by the upper tail of the distribution (Cowell consumption distribution. Another measure that is not and Flachaire 2002), fell from 0.45 in 2005/06 to 0.39 affected as much by the nonresponse issue is the ratio in 2015/16, indicating that Kenya made considerable of consumption at the 75th and 25th percentile. Under progress in terms of reducing inequality (Figure 2.15). The this measure inequality also declined, albeit the drop Gini index in rural areas (unaffected by the nonresponse is less pronounced and the levels and changes in rural issue) declined from 0.37 to 0.33, a significant and urban areas resemble each other. The consumption 46 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.15: Gini inequality index nationally, by area of residence and by province 0.6 0.45 0.46 0.44 0.43 0.42 0.39 0.38 0.38 0.38 0.38 0.38 0.4 0.37 0.37 0.35 0.35 0.34 0.35 0.34 0.33 0.33 0.33 Gini Index 0.31 0.2 0.0 National Urban Rural Coast North Eastern Central Rift Western Nyanza Nairobi Eastern Valley 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 Figure 2.16: Gini inequality index for select African countries 0.8 0.63 0.6 0.50 0.41 0.42 Gini index 0.39 0.38 0.4 0.33 0.2 0.0 Kenya Rwanda South Africa Uganda Tanzania Ethiopia Ghana (2015/16) (2013) (2011) (2012) (2011) (2010) (2012) Source: World Bank Poverty & Equity Databank. level of the 75th percentile went from being 2.7 times in the past ten years (Table 2.4). Moreover, in 2015/16, higher than that of the 10th percentile in 2005/06 to 2.5 differences across rural and urban households help times higher in 2015/16 (Figure 2.17a). explain about one fourth of the overall inequality. Thus, about three quarters of the inequality can be attributed Inequality in Kenya is primarily explained by to differences within rural and urban households. differences within urban and rural areas (and within Interestingly, these fractions have remained constant provinces), rather than by differences between these over time. Nonetheless, as will be seen later on, the groups. Analysis of the Theil index allows for a better urban-rural divide in non-monetary living conditions understanding of the nature of inequality and how it has and access to services is large, with rural areas lagging changed over time. More specifically, it helps determine behind in access to WASH services and electricity in how much of the inequality in the country is rooted particular. The analysis of the contribution to inequality within particularly groups and how much is attributed from differences within and across provinces produces to differences between these groups. Consistent with comparable results, and it is mostly inequality within all measures described so far, under the Theil index, provinces that helps explain inequality in Kenya. inequality went down by one third from 0.42 to 0.28 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 47 The Extent and Evolution of Poverty and Inequality in Kenya Box 2.4: Inequality measures While poverty measures absolute deprivation with respect to a given threshold, inequality is a relative measure of poverty indicating how little some parts of a population have relative the entire population. In the context of monetary poverty, equality can be defined as an equal distribution of consumption / income across the population. This means that each share of the population owns the same share of consumption / income. The Lorenz Curve compares graphically the cumulative share of the population with their cumulative share of consumption / income. A perfectly equal consumption / income distribution is indicated by a diagonal. The other extreme is complete inequality where one individual owns all the consumption / income. These two (theoretical) extremes define the boundaries for observed inequality. The Gini coefficient is the most commonly used measure for inequality. A Gini coefficient of 0 indicates perfect equality while 1 signifies complete inequality. In relation to the Lorenz Curve, the Gini coefficient measures the area between the Lorenz Curve and the diagonal. The Theil Index measures inequality based on an entropy measure. A parameter α controls emphasis to measure inequality for higher incomes (larger α) or lower incomes (smaller α). The Theil index with parameter α=1 is usually called Theil T while using α=0 is called Theil L or log deviation measure. Relative and absolute consumption / income differences can be used to compare inequality dynamics over time. Usually, percentiles are used to compare incomes of different groups. For example, p90/p10 is the ratio (for relative incomes) or difference (for absolute incomes) of the average consumption in the 90th and 10th percentile. Given the nonresponse issues in Nairobi, we opt for the p75/p25 ratio, which is the average consumption ratio in the 75th and 25th percentiles. Finally, the Atkinson index introduces value judgements about the degree of inequality aversion prevalent in the society, which is expressed by the choice of an inequality aversion parameter. The higher this parameter, the more emphasis is placed on the lower tail of the distributions and the changes experienced there. Source: World Bank’s Poverty Handbook. Figure 2.17: Atkinson index and P75/P25 inequality index nationally and by area of residence a) Atkinson index (∑=1) b) P75/P25 ratio 0.4 3 2.7 2.5 2.5 0.30 2.4 0.3 0.29 2.2 2.1 2 0.23 P75/P25 ratio 0.21 Index 0.2 0.19 0.17 1 0.1 0.0 0 National Urban Rural National Urban Rural 2005/06 2015/16 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 48 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Table 2.4: Theil inequality index - decomposition by urban/rural location and province By urban / rural By province 2005/06 2015/16 2005/06 2015/16 Between group 0.11 0.07 0.10 0.05 Within group 0.30 0.21 0.32 0.23 Total 0.42 0.28 0.42 0.28 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. 2.3 WHAT EXPLAINS THE TRENDS IN have a large impact in terms of poverty reduction, in POVERTY REDUCTION? POVERTY addition to other benefits in terms of human capital DECOMPOSITION EXERCISES accumulation explored in detail on Chapter 8 of this U nveiling the main drivers behind the observed report. It is nevertheless likely that the nonresponse changes in the poverty incidence of Kenya is an problem experienced in Nairobi during the collection important objective of this report. This section makes of the KIHBS 2015/16 survey is overestimating the use of several decomposition exercises to help shed some contribution of the redistribution effect. light on what were some of the main factors behind the ten-percentage point decline observed during the The contribution of economic growth to poverty ten-year period focus of this report. More specifically, it reduction is more marked in rural areas. As shown examines the role of growth versus redistribution, the in Figure 2.18a, poverty reduction is mainly driven by economic growth, accounting for three quarters of the progress in urban / rural areas and provinces and the almost twelve percentage point reduction in the share population shift amongst them, the relative importance of the rural population living beneath the poverty line. of the household’s sources of income (in the sectorial The results of the growth-inequality decomposition sense), and the role of mobile money.91 for urban areas show that the decline in inequality (redistribution effect) drove the entirety of the reduction 2.3.1 The role of growth and redistribution in poverty between 2005/06 and 2015/16.92 However, While both economic growth and redistribution these results are affected by the nonresponse problem contributed to Kenya’s poverty reduction, the former in Nairobi. To partially address this problem, the same helps explain almost two thirds of the decline decompositions have been conducted in a sample observed. Consistent with the common view that excluding the top twenty percent of households and overall economic growth is usually accompanied by are presented in Figure 2.18b. This scenario shows that an increase in the living standards of the population, it is likely that economic growth did contribute to the growth accounts for almost 60 percent of the poverty reduction of poverty in urban areas. reduction observed in Kenya for the period 2005/06 to 2015/16. The remaining 40 percent (the interaction Consistent with the pattern observed at the terms explains less than 1 percent), is attributable to the national level, the growth effect was larger than the redistribution effect (Figure 2.18a). This is an important distributional effect in each of the Kenyan provinces. result, as further efforts to improve redistribution and Economic growth accounts for the majority of the decline further reduce inequality would likely accelerate observed in the provinces (except for Nairobi), with the poverty reduction for Kenya in the medium term. magnitude ranging from 5.5 percentage points (almost Redistributive policies such as the expansion of three quarters of the overall reduction) in Rift Valley to social protection programs at a national level, would 17 percentage points in the Coast province (which 91 For the role of mobile money, we mostly review the extensive recent 92 Actually, the results point out that the redistribution effect alone would economic literature on the link between access to this financial services have reduced poverty by 10.8 percentage points between 2005/06 and and poverty reduction. For most of these studies it is possible to identify 2015/16, resulting in a poverty incidence of 21.3 percent in 2015/16 a causal link either by the use of randomized control trial (RCT) or the rather than the observed 29.4 percent, had it not been because the robust econometric identification strategies. growth effect hindered poverty reduction. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 49 The Extent and Evolution of Poverty and Inequality in Kenya Box 2.5: What does decomposing changes in poverty entail? In this chapter the results of two decomposition methods are presented. The first method is the Datt-Ravallion approach, which isolates the growth and redistribution effects associated with the decline in poverty over the period of analysis. Conceptually, this decomposition is based on the idea that that a measure of monetary poverty can be expressed as the product of mean consumption and a parameterized Lorenz curve. Keeping the Lorenz curve constant gives the distribution neutral growth that would drive the average increase in consumption across the population, for instance, raising the levels of consumption of all households by the same rate. The other part is derived from holding the mean consumption constant (a mean-preserving redistribution) to capture the change in the shape of the consumption distribution driven by, for instance, a faster growth in the consumption of the poorest relative to the consumption growth of the richest (Datt and Ravallion 1992). The second is the Ravallion and Huppi (1991) decomposition method, that quantifies how much poverty reduction among mutually exclusive groups or movement between these groups accounts for national poverty reduction. More specifically, the analysis decomposes changes in poverty over time into “intra-group effects” (poverty changes within sectors, within provinces, or within urban and rural areas, while assuming no changes in the distribution of the population across groups), “inter-group effects” (allowing for changes in the distribution of the population between groups keeping poverty rates constant) and an “interaction” term that can be interpreted as a measure of the correlation between the population shifts and the intra-group changes in poverty. Under both methods, a counterfactual scenario is used and estimates are made as to what would have happened to poverty had the counterfactual scenario occurred. By defining a counterfactual scenario, the changes that have been important to overall poverty reduction can be quantified, be it a distribution-neutral consumption growth, the amount of poverty reduction that took place within a sector (as if the distribution across sectors had not changed), or the amount of poverty reduction that took place as a result of people moving between groups. Figure 2.18: Determinants of changes in poverty – Datt-Ravallion decomposition by area of residence a) All households b) Without top 20% of urban households 10 10 Percentage point change in headcount rate Percentage point change in headcount rate 5 5 8.1 0 0 -1.5 -2.8 -6.3 -9.9 -8.7 -5 -5 -8.7 -10.8 -4.3 -10 -3.0 -10 -0.9 -3.0 -15 -15 National Rural Urban National Rural Urban Growth Distribution Growth Distribution Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. corresponds to nearly the entirety of the reduction, as offsetting the growth effect. Thus, for these two redistribution contributes mere 0.1 percentage points provinces, the decline in poverty has been hindered by to the overall decline, Figure 2.19). In the case of two the changes in the consumption distribution, despite provinces, the North Eastern (poorest) and Central the fact that inequality declined in both provinces, as (second-wealthiest) provinces, the distributional effect measured by the Gini index (Figure 2.15). actually contributed to an increase in poverty, partially 50 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.19: Determinants of changes in poverty – Datt-Ravallion decomposition by province 15 10 5 8.8 2.6 Percentage point change in 0 2.1 -5.5 headcount rate -6.5 -8.9 -7.7 -5 -8.5 -13.4 -2.0 -2.9 -10 -3.2 -17.0 -15 -17.0 -0.1 -1.8 -20 -25 Coast North Eastern Central Rift Western Nyanza Nairobi Eastern Valley Growth Distribution Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. 2.3.2 The role of poverty reduction within prices through most of the period analyzed and greater geographical areas versus internal migration commercialization of agricultural production, increased Unsurprisingly, poverty reduction amongst rural the wellbeing of households engaged in agricultural households accounted for almost all the poverty production. Moreover, the fact that rural households reduction observed at the national level. The Ravallion- experienced an increased off-farm diversification of Huppi decomposition exercises allow a decomposition income activities (as showed later in this section) also of Kenya’s change in poverty over time into changes helped to reduce poverty in rural Kenya. amongst urban households and rural households, assuming no migration between the two sectors, Internal migration, specifically rural-urban migration, as well as changes due to population shifts among is usually associated with economic progress, them (internal migration). Unsurprisingly, poverty access to better job opportunities and better living reduction amongst rural households accounted for conditions. Rural-urban migration is an inherent aspect 87.6 percent of poverty reduction observed in Kenya of the economic development process all around the during the period 2005/06 to 2015/16 (Figure 2.20a). A world and can in principle support poverty reduction. robust performance of the agricultural sector after the When migration is driven by “pull” forces that, for economic slow-down of 2008, together with high food instance, attract migrants from rural areas looking for Figure 2.20: Contribution to poverty reduction a) Rural/urban b) Provincial - 4.2% Coast -7.0% 3.3% 3.4% 14.8% North Eastern 14.3% 1.2% Eastern Rural 13.7% Central Urban 5.1% Rift Valley Population shift e ect Western Interaction e ect 13.7% 28.9% Nyanza Nairobi Population shift e ect 87.6% 17.4% Interaction e ect 7.9% Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 51 The Extent and Evolution of Poverty and Inequality in Kenya better and higher paid economic activities, as well 2.3.3 Structural pattern of poverty reduction as better returns to their endowments, it is usually The agricultural sector contributed to poverty accompanied by poverty reduction (Kenya Urbanization reduction. Several studies have determined that Review, World Bank 2016). However, migration can growth in the sectors in which the poor are employed also be motivated by “push” factors, where migrants is more poverty reducing than growth in other sectors are escaping from conflict, political turmoil, natural (Loayza and Raddatz 2010; Christiaensen, Chuhan-Pole, disasters, or a particular shock affecting their place of and Aly Sanoh 2013). In line with these findings, although residence. In these cases, internal migration can result the agricultural sector has not been as dynamic as the in the physical, social and human capital depletion, and services or the industrial sector (Figure 2.21), it played can lead to a higher incidence of poverty. a notable role in the reduction of poverty in Kenya in the decade leading up to 2015/16. When looking at the For Kenya, internal migration contributed moderately contribution of different sectors to poverty reduction, to poverty reduction throughout the ten-year period each household is first attributed to the sector from being studied. According to the analysis of the two which it draws at least 50 percent of its total income. waves of the KIHBS, the population shift among rural- Those households which do not rely on any one sector urban households did contribute to poverty reduction as their main source of income (meaning no source of in Kenya throughout the period of analysis. Migration income accounts for more than 50 percent) are classified between rural and urban areas accounted for about as diversified. Households for which the main source 14 percent of the overall national reduction in poverty of income is the agricultural sector (including crop (Figure 2.20a). Despite the fact that migration between income, livestock income, and earnings of wage workers urban and rural areas is prevalent in Kenya, this modest in the agricultural sector) account for around 33.85 percent contribution may be partly explained by migration of the overall national poverty reduction (Table 2.5). This selection or the fact that those who migrate are usually contribution stems from the fact that agriculture remains better off or have higher levels of physical or human a source of livelihood for around 60 percent of the labor capital, as is discussed in Chapter 5 of the report based force, and the robust growth of the sector observed on data from the DHS. throughout the period analyzed, thanks to high food and commodity prices. However, agricultural productivity In terms of the provincial contribution to poverty remains low, particularly the production of grains, which reduction, the Eastern province accounted for hinders the transformative potential of the sector to boost almost one third of the overall poverty reduction. As the incomes of poor households. expected, the extent to which the different Kenyan Figure 2.21: Real sector growth 2007–2015 provinces contributed to the overall decline in poverty varies with the progress experienced by the province, 14 as well as changes in the share of the national population and the share of the poor population 10 living there. The Eastern province, for which poverty incidence declined from 50.6 percent in 2005/06 to 6 Percent 31.8 percent in 2015/16, is responsible for over one fourth of the overall poverty reduction (28.9 percent). 2 Other important contributors were the Rift Valley, the -2 most populated province, and the Coast Province, which experienced a large decline in poverty. On the -6 other hand, Nairobi and the North Eastern Province 2007 2007 2007 2007 2007 2007 2007 2007 2007 contributed only 3.4 and 1.2 percent of the decline, Agriculture GDP Services Industry respectively (Figure 2.20b). Source: World Bank 2018b 52 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Similarly, households deriving their income from classified into agricultural, non-agricultural, and mixed the services sector account for almost 30 percent of households, where agricultural and non-agricultural Kenya’s poverty reduction. The expansion of modern are defined as deriving at least 90 percent of the services, particularly financial intermediation and total income from either sector, and mixed being mobile communication (as a result of the introduction everything else in between. Households depending of innovative solutions like M-PESA), has stimulated overwhelmingly on agriculture income account the demand for more traditional services, and the for 27.63 percent of poverty reduction, while non- sector as a whole is responsible for the bulk of the agricultural households account for almost 21.19 economic growth of Kenya. Between 2006 and 2013, percent (Table 2.6). Interestingly, the contribution the sector accounted for 75 percent of GDP growth.93 of diversified households was around 33.51 percent, Not surprisingly, households that report earning the showing that an important factor for poverty majority of their income from services (comprising reduction has been the ability of households engaged those in wage employment and the majority of in agriculture (and sometime deriving the majority of those engaged in a non-agricultural enterprise) help their income from this activity) to complement their explain about 29 percent of the decline in the national incomes through non-agricultural activities. The ability poverty rate. Moreover, the inter-sectoral movement of agricultural households to engage in activities such of households across these classifications accounts for as petty trading, kiosk retailing, operating taxis and 16.4 percent of the decline (Table 2.5), suggesting that running local enterprises, reduces their vulnerability to the structural transformation in the country, mainly climatic and price shocks, and increases their ability to from the agricultural sector towards the services sector, generate income. has aided poverty reduction. 2.3.4 The role of mobile money The evidence suggests that off-farm diversification Access to mobile phones in Kenya increased has been important for poverty reduction in Kenya. dramatically over the last 15 years, transforming Given that close to 70 percent of those households the economic paradigm. In 1999, the Kenya-based engaged in agriculture have additional sources of mobile service provider Safaricom projected that income from non-agricultural activities, additional the total mobile phone market would reach three decompositions analyzing the diversification of million subscribers by 2020 in Kenya. However, by income sources were undertaken. Households were 2009 Safaricom alone had over 14 million subscribers Table 2.5: Sectoral decomposition of poverty reduction (Ravallion-Huppi) Pop. share in period 1 Share of total change Source of income Absolute change (percent) (percent) Agriculture 49.18 -3.32 33.85 Industry wages 5.84 -0.20 2.05 Service wages 24.49 -1.76 17.91 Non-agr. enterprise 7.30 -1.12 11.41 Transfers 5.99 -0.38 3.87 Diversified 7.19 -1.18 11.99 Total intra-sectoral effect -7.95 81.08 Population shift effect -1.61 16.40 Interaction effect -0.25 2.52 Change in headcount rate -9.81 100.00 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. 93 According to the World Bank’s Country Economic Memorandum (2016). KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 53 The Extent and Evolution of Poverty and Inequality in Kenya Table 2.6: Sectoral decomposition of poverty reduction (Ravallion-Huppi) - alternative definition Pop. share in period 1 Share of total change Source of income Absolute change (percent) (percent) Non-agricultural income only 31.64 -2.16 21.19 Agriculture income only 39.79 -2.81 27.63 Mixed - agricultural & non-agricultural income 28.57 -3.41 33.51 Total intra-sectoral effect   -8.37 82.33 Population shift effect   -1.68 16.49 Interaction effect   -0.12 1.19 Change in headcount rate   -10.17 100.00 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. (Aker and Mbiti 2010). As in many countries in increased per capita consumption levels and lifted Africa, mobile phones represented the first modern around 2% of Kenyan households out of poverty (Jack telecommunication infrastructure of any kind, and Suri 2016).94 particularly for those in rural areas. Moreover, the adoption among firms, mainly in urban centers, More effective risk-sharing has been a key factor appears to be correlated with the poor quality of the in improving financial resilience. Through mobile landline services. While it was clear that the reduction in money, households are able to share risk more the communication costs would bring efficiency gains efficiently with relatives, friends and other associates, in all economic markets, as information about prices helping them to smooth consumption over time and and quantities reaches agents faster than before and increase their savings. Mobile money users report better market coordination is possible (Jensen, 2007; having access to more credit and emergency-related Aker, 2008; Aker, 2010; Klonner and Nolen, 2008), transfers than nonusers, suggesting that both explicit the impact of mobile phones has gone well beyond credit and informal insurance arrangements can be that thanks to introduction of mobile money (more more effectively sustained (Suri and Jack, 2013). M-PESA specifically M-PESA). As a growing body of rigorous users are more likely to receive and send (internal) academic literature has shown recently, mobile phone remittances (Jack, Ray, and Suri 2013), and in case of a penetration in Kenya has shifted the economic paradigm, negative shock, M-PESA users receive a larger amount constituting a platform for service delivery rather than of funds from a wider network than non-users (Jack and being a simple communication tool. Suri 2014). Beyond the economic efficiency gains from Mobile money has also contributed to increased improved communication, mobile phones, through employment, savings, and productive investment. mobile money, have been shown to increase per The introduction of mobile money has been shown to capita consumption and reduce poverty among have a positive effect on local employment (Plyler et users. As of 2014, 97 percent of Kenyans reported al., 2010; Mbiti and Weil 2011), and contributes to an having an M-PESA account and by 2015 there were improved business environment and access to trade 65,569 registered M-PESA agents in the country. This credit (Plyler, Haas, and Nagarajan 2010; Beck et al. 2015). service, which allows individuals to transact money Using mobile money appears to increase savings not only without having access to a formal bank account, among the “unbanked” but also among those with access has expanded the economic possibilities of the population, by increasing their financial resilience to regular banking services (Morawczynski and Pickens and savings, and allowing them to move out of 2009), and creates new savings incentives for smallholder agriculture and into business. A recent study shows farmers (Kikulwe, Fischer, and Qaim 2014). In turn, that through these mechanisms access to M-PESA 94 The authors use US$ 1.25 per day as a measure of extreme poverty. 54 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya households are able to make productive investments: 2.4 POVERTY PROFILES – WHAT ARE mobile money has allowed users to accumulate greater CHARACTERISTICS OF THE POOR amounts of capital (Plyler, Haas, and Nagarajan 2010), IN KENYA? and allowed farmers to engage in more commercially- oriented farming, increase sales of harvested products, P rofiling the characteristics of the poor is helpful in identifying what factors are limiting their economic opportunities. Moreover, comparing poor and non- and realize higher profits per acre of production (Kikulwe, Fischer, and Qaim 2014). poor households along different dimensions, such as demographics, human capital, economic activities Gender outcomes have also improved due to and asset ownership, can pin down specific policy mobile money. The introduction of mobile money, in actions that may help raise their living standards and conjunction with M-PESA has helped to increase female overcome poverty. empowerment in Kenya. This was particularly true in rural areas, where the technology made it easier for Household living in poverty have older household women to obtain remittances from relatives (other than heads and are more likely to be headed by a their husbands) and friends increasing their financial woman. Poor households tend to have slightly older independence (Morawczynski and Pickens 2009). household heads. The average age in 2015/16 among Similarly, the effects of M-PESA on consumption and poor households was 47 years versus 42 for non-poor poverty accrue particularly to women: the magnitude households (see Table 2.7). This age gap between poor of the effect for female-headed households was more and non-poor households has remained constant than twice as high as the average effect (Suri and Jack since 2005/06. Female headed households are more 2016). More specifically, the evidence suggests that likely to be poor, even after all other characteristics of mobile money allowed women to increase their savings the household are taken into account in a multivariate and smooth consumption, and induced changes in regression analysis (see column Significance -Model- in occupation choice. Financial inclusion helped them to Table 2.7). This is particularly true for households headed graduate from subsistence agriculture (into sales/small by widows and divorcées (or separated). As explored in business) and to reduce their reliance on multiple part- detail in Chapter 3, marital rupture frequently entails a time occupations. loss of economic means for women. In addition to that, the proportion of households headed by a woman has More recently, the M-PESA platform has facilitated increased slightly between 2005/06 and 2015/16 for targeted development interventions in education both poor and non-poor households. This is important and health. Recognizing that the transition from as age and gender reflects the economic opportunities primary school to high school is costly and often leads of the household head, which matter significantly for to dropout, an intervention encouraging parents of the total income of most households. primary school leavers to open a mobile banking account through the M-PESA platform substantially Poor households tend to have a larger size and increased financial savings and high school enrolment higher dependency ratios. In terms of demographic (Jack and Habyarimana 2018). The M-PESA platform characteristics, poor households have 1.75 household has also been used for provision of conditional cash members more (a considerable gap) and a larger transfers and vouchers covering the full cost of dependency ratio95 than non-poor households (see giving birth in a medical facility, which appears to be Table 2.7). While household size decreased by about highly effective in increasing institutional deliveries one person for both poor and non-poor households among poor rural women (Grépin, Habyarimana, and between 2005/6 and 2015/16, the dependency ratio Jack 2017). did not decline for poor households. Regression analysis 95 Which imply a lower share of working age male and female members aged 15-65. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 55 The Extent and Evolution of Poverty and Inequality in Kenya Table 2.7: Household characteristics by poverty status 2005/06 2015/16 Non- Significance Significance Non- Significance Significance Poor Poor Poor (Wald-test) (Model) Poor (Wald-test) (Model) Demographic                 Age of head 42.6 47.9 *** *** 42.1 47.1 *** *** Female head 27.4% 31.3% *** *** 31.2% 35.7% *** *** HH size 4.4 6.2 *** *** 3.5 5.2 *** *** Dependency ratio 36.7% 47.6% *** *** 33.1% 47.4% *** ** Urban 30.9% 16.9% *** *** 39.9% 27.5% *** *** Education                 Ave. years sch. (15+) 7.7 5.3 *** *** 9.1 6.1 *** *** HH head levels               No education 14.3% 32.7% *** Reference 8.8% 27.7% *** Reference Some or complete 43.4% 51.3% ***   42.5% 52.3% ***   primary Some or complete 39.0% 15.9% *** ** 41.6% 19.5% ***   secondary Some or complete tertiary some or 3.2% 0.1% *** *** 7.1% 0.4% *** *** complete Sources of income                 Non-agricultural income 55.5% 44.2% *** Reference 71.1% 68.1% *** Reference only Agriculture income only 12.6% 17.4% ***   2.5% 4.4% *** ** Mixed - agricultural & 31.9% 38.5% *** *** 26.5% 27.4%   *** non agricultural income Access to services                 Improved drinking water 70.2% 51.9% *** *** 80.4% 65.6% ***   Improved sanitation 56.4% 37.7% *** *** 72.2% 47.8% *** *** Main source light 23.0% 4.0% *** *** 49.9% 18.9% *** * (electricity) HH electricity access 26.5% 4.5% *** *** 52.0% 20.7% *** *** Assets                 HH owns radio 58.1% 51.2% *** *** 51.8% 40.6% *** *** HH owns cell phone 27.9% 5.8% *** *** 90.1% 77.8% *** *** HH owns kerosene stove 53.0% 23.2% *** *** 42.3% 22.9% *** *** HH owns charcoal jiko 62.6% 40.3% *** *** 61.7% 44.2% *** *** HH owns mosquito net 40.0% 25.7% *** *** 68.8% 66.1% ** *** HH owns fridge 5.5% 0.5% *** *** 8.2% 0.7% *** *** HH owns sofa 56.8% 29.2% *** *** 62.3% 40.2% *** *** Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16. Note: Column wald test shows significance values from a wald test of differences between the means. Column Model shows significance values from a log-linear probability model (LPM) (with log of consumption as the dependent variable) controlling for all variables shown along with province dummies. *, **, and *** indicate significance level at 10%, 5%, and 1%. Robust errors used. 56 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead The Extent and Evolution of Poverty and Inequality in Kenya Figure 2.22: Household size and average education level nationally, by province and NEDI classification a) Household size b) Average years of schooling (age 15 and older) 7 14 6.2 6 12 11.3 5.2 5 4.7 10 9.8 4.5 4.3 4.4 9.0 Number of people Number of years 4.0 3.9 3.9 3.9 8.7 4 8 8.3 7.9 7.8 7.6 8.1 7.6 7.6 7.1 5.1 3.2 3.0 6.8 6.5 6.4 6.8 5.8 6.5 3 6 2 4 3.6 2.5 2.1 1.4 1 2 0 0 l t rn l y rn za bi DI DI rn na ra as t l l n Ce n l y rn Na a n- i DI DI lle rn No irob na ra ra as iro ste te lle an NE NE z ba r nt ste Co tio ste te an NE NE Va Ru nt ste Co tio es Va Na Ce Ny Ur n- Ea es Ea Ny Na Ea W ft Ea Na No W ft Ri rth Ri rth No 2005/06 2015/16 2005/06 2015/16 No Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 confirms that a larger share of dependents within a the country when it comes to education outcomes, household is associated with lower consumption, with only 2.54 and 3.64 average years of schooling even after considering all other relevant household respectively (Figure 2.23b). Improving the education characteristics. When looking at the evolution of these outcomes of the poor is necessary for them to access dimensions at the provincial level (and under the NEDI better income-generating economic activities and and non-NEDI classifications), it is clear that the lack of enhance their consumption levels. a demographic transition in the North Eastern province and in the NEDI counties has slowed down the pace of The proportion of households that solely depend poverty reduction in these regions (Figure 2.23a). on agriculture for their income is much lower than before, but still remains higher amongst the poor. As expected, poverty is associated with lower levels Back in 2005/06, 17.4 percent of poor households of educational attainment. In 2015/16, the average depended exclusively on agriculture for their income years of schooling (of household members 15 years (12.6 percent for non-poor households), but ten years old and above) for non-poor households is three years later this proportion declined to 4.4 percent (2.5 higher than for poor households. Similarly, only 19.9 percent of non-poor, see Table 2.7), illustrating the percent of household heads in poor households have off-farm diversification that propelled the consumption completed at least secondary education, compared to growth among the Kenyan population. As expected, 48.7 percent for their non-poor counterparts (see Table the regression analysis points out that household 2.7). Of greater concern, the gap in the educational consumption increases with diversification and attainment between poor and non-poor households decreases with agricultural dependency. Interestingly, increased between 2005/06 and 2015/16, suggesting the latter was not the case for 2005/06, showing that that poor households still face notable barriers to the structural pattern of poverty has evolved since then. access Kenya’s the education system, as will be discussed in Chapter 6. Regression analysis shows that Access to basic services tends to be lower amongst every additional year of education at the household poor households. Although between 2005/06 and level (for those 15 and older) increases consumption 2015/16 access to sanitation, water and electricity by 2.9 percent, consistent with the idea that individuals improved for the poor, they continue to have much with higher levels of education have higher paid jobs. lower access rates than non-poor households. Not surprisingly, the North Eastern Province and the While three out of four non-poor households have NEDI counties lag considerably behind the rest of access to improved sanitation, only one in two KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 57 The Extent and Evolution of Poverty and Inequality in Kenya poor households do. Moreover, while access for both Ownership of basic assets is limited for poor groups increased since 2005/6, progress was more households with the exception of mobile phones: pronounced amongst non-poor households. In the almost 80 percent of poor households have one. case of improved water, the coverage rate for the In general, poor households are characterized by poor is 66 percent, also considerably lower than for limited ownership of assets when compared to non- non-poor at 80 percent. In terms of electricity, access poor households. They are less likely to own a radio remains low for non-poor households at 52 percent, (41 percent versus 52 percent), a stove (23 versus 42 and it is even lower for the poor, with a coverage rate of percent) and a refrigerator (1 percent versus 8 percent), 21 percent (Table 2.7). Access to improved sanitation is to provide some examples (see Table 2.7). One notable remarkably low in the North Eastern province and NEDI exception is the ownership of mobile phones: between counties when compared to the rest of the country. At 2005/06 and 2015/16 ownership of mobile phones by the same time, access has worsened dramatically in poor households increased from 6 to 78 percent. This is the Western province: while in 2005/06 65.8 percent relevant, given the importance of mobile phones and, of the province’s population had access to improved more specifically, of mobile money in transforming sanitation, this number was only 40.9 in 2015/16 calling the livelihoods of Kenyan households discussed in a for an urgent policy action on this front (Figure 2.23a). previous section of this chapter. Figure 2.23: Access to improved sanitation, water and electricity by province, urban/rural, and NEDI/non-NEDI status a) Improved sanitation b) Improved water 99 100 97 100 95 85 84 82 71 78 78 76 Percentage of households, % Percentage of households, % 72 71 75 68 70 75 66 66 68 65 64 66 61 58 51 58 44 50 51 49 50 50 43 41 33 36 33 19 20 25 25 0 0 t DI t n l l za l za rn Ny n DI DI rn bi rn No obi l l rn rn ey ey ra as ra ra DI na na as ba r iro NE te te ste ste an an ste NE NE ste Ru nt nt Co all all Co NE tio tio ir Ur es es Ce Ce Na Na Ny tV tV n- n- Ea Ea Ea Ea Na Na W W No Rif Rif rth rth No No 2005/06 2015/16 2005/06 2015/16 c) Access to electricity 100 91 81 Percentage of households, % 75 62 52 50 46 44 36 29 22 23 25 17 16 17 0 DI Ea t n l l za Ny rn DI No bi rn l rn y ra ra na as ba lle iro NE te ste an NE Ru ste nt Co io Va Ur es Ce Na n- t Ea Na W ft Ri th r No 2005/06 2015/16 Source: Own calculations based on KIHBS 2005/06 and KIHBS 2015/16 58 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead CHAPTER 3 GENDER AND POVERTY SUMMARY Kenyan women are poorer than men during core productive and reproductive years, especially if they experienced a marital dissolution. As in other African countries, Kenyan women are more likely to live in poor households than men starting in their mid-20s and continuing until their 50s. Moreover, women are more likely than men to reside in a poor household if they are separated/divorced (31 vs. 24 percent, p<0.01) or widowed (38 vs. 25 percent, p<0.01). Women who have experienced a marital dissolution also face higher prevalence rates of physical violence than other women and are disproportionately affected by HIV/AIDS. In education, girls continue to be disadvantaged in parts of Kenya but there is also an emerging pattern of boys’ underperformance. Kenya has achieved significant increases in primary and secondary enrollments of boys and girls since the early 2000s, which has been accompanied by a trend towards greater gender parity. At the subnational level, girls have lower enrollment rates than boys in Northeastern Kenya and the coast – but boys’ disadvantages emerge in parts of Central and Western Kenya. Similar patterns are evident for learning outcomes, where advantages for girls are even stronger. Despite these improvements in girls’ education, adult women are twice as likely to be illiterate as adult men, reflecting historical gender inequalities in the education sector. Kenyan women face daunting health challenges and bear the brunt of care work within the household. Despite some decline in maternal mortality since 2005, Kenyan women face a staggering lifetime risk of 1 in 42 of dying due to complications of pregnancy or child birth, which is high even by regional standards. Maternal mortality is most severe in the Northern/Northeastern parts of Kenya, areas with extremely high fertility rates and poor access to reproductive health care. And due to traditional gender roles, women spend a significant amount of time on unpaid care work (for children, elderly and the sick or disabled) within the household. Women are much less likely than men to own property and gender biases linger in parts of Kenya’s legal system. Only 12 percent of women aged 20-49 years report owning any land on their own, compared with 39 percent of men. Kenya is among the few African countries with gender inequality in formal inheritance rights – i.e. with respect to the Law of Succession Act. Gender gaps exist also in terms of access to ICT and financial services, though levels of access are high by regional standards. In 2015/6, 71 percent of working-age women participated in the labor force, compared with 77 percent among men. There has been a significant increase in male and, particularly, female employment over the past decade. For men, this increase was driven by a rise in wage employment, while for women it reflects both rising wage employment but also increasing employment in (farm and non-farm) household enterprises. Female labor force participation is linked to religious norms, education, marital status and the presence of young children in the household. Among women, being of Muslim or other non-Christian religion reduces the probability of participating in the labor force by about 30 percent (relative to being Catholic). Women who are widowed, separated/divorced or polygamously married are significantly more likely to participate in KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 59 Gender and Poverty the labor force than women who are monogamously married. Every child aged 0-5 years reduces women’s probability to be in the labor force by over 2 percent. On the other hand, education, even at the primary level, increases women’s probability to participate in the labor force. Gender inequality in earnings is substantial and cuts across all segments of the labor market. Male wage workers earn 30 percent higher wages/salaries than female wage workers and profits of male-run household enterprises are about twice as high as profits of female-run enterprises. Similarly, households where women are the primary decision-makers in agriculture achieve lower yields (e.g. for maize and beans) than households where men are the primary decision-makers. Gender inequality in earnings reflects a variety of different factors, including gender gaps in access to productive resources and sectoral segregation by gender (i.e. women being disproportionately engaged in low-paying sectors). Gender equality is central to Kenya’s vision of becoming a approach to measuring poverty, the primary source “middle-income country providing a high-quality life to all of information are household surveys and the key its citizens by the year 2030.” (Government of Kenya 2007). indicator is a money‐metric measure of welfare based No society can develop sustainably without transforming on consumption (or income) data collected for the the distribution of opportunities, resources, and choices household as a whole. This approach masks within- for males and females so that they have equal power to household differences in consumption along gender, age shape their own lives and contribute to their families, and other dimensions. communities, and country. We use a lifecycle approach to obtain a better This chapter provides a synthesis of what is known about understanding of gender differences in poverty in Kenya, the gender-poverty nexus in Kenya. It starts with a basic even with the existing constraints (i.e. poverty status being determined at the household-level). A recent profile of poverty and gender in section 3.1. Following the collaborative effort between UN Women and the World framework of the 2012 World Development Report on Bank (Munoz Boudet et al. 2018) analyzes whether Gender (World Bank 2011), it then proceeds to analyze life events – such as the transition from childhood to gender gaps in endowments (section 3.2), economic adolescence, adulthood, and elder years; or marriage, opportunities (section 3.3), and voice and agency (section divorce and widowhood – affect men and women 3.4). At the end of each section, the chapter also provides differently in terms of their probability to live in poor a short discussion of possible policy levers to narrow – households The study further develops a demographic and ultimately close – gender gaps and promote a more taxonomy that categorizes households by the number equitable society.96 and sex of adult household members (e.g. 2 adults of opposite sex, single adult male/female households, etc.) to examine the relationship between poverty and 3.1 A PROFILE OF POVERTY AND GENDER the household’s demographic composition in a way IN KENYA that goes beyond a comparison of male- and female- O ne of the key challenges towards an understanding of poverty and gender is that poverty is typically measured at the household level. In the standard headed households. Following this approach and using the data of the 2015/6 KIHBS, this section presents a profile of poverty and gender in Kenya. 96 Underlying this chapter are several data sources, including the KIHBS of 2015/6 and 2005/6, DHS of 2014, 2008/9 and 2003, Global Findex database for 2014 (Demirguc-Kunt et al. 2015), and other country-level databases (e.g. WDI – World Bank 2017b; Women, Business and the Law – World Bank 2015b). These data sources provide a rich information base to analyze gender gaps in different sectors and their link to poverty in Kenya, but there are still important data gaps. Appendix C1 of this chapter hence provides suggestions for possible tweaks to the KIHBS instrument that would help to fill key gaps in data and knowledge about gender inequality in Kenya. 60 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty 3.1.1 Gender differences in poverty through the and van de Walle 2018). Conversely, among the never lifecycle married population (which here includes children), As in other African countries, Kenyan women are female poverty rates are lower than male poverty rates. more likely to live in poor households than men starting in their mid-20s and continuing until their 50s Figure 3.2: Male and female poverty rates by marital status, 2015/6 (Figure 3.1). Women are hence poorer than men during core productive and reproductive stages of life. This 50 46 pattern suggests that care responsibilities for children 43 40 38 combined with constraints in economic opportunities 38 34 are major vulnerability factors for women. 29 31 30 28 Percent 24 25 Figure 3.1: Male and female poverty rates by age group, 2015/6 20 60 10 50 0 Poverty rate (%) 40 Monogamously Separated Widow Never Polygamously married or living married or divorced or widower married together 30 Males Females 20 Source: KIHBS 2015/6. Note: Cross-tabulation of individuals’ poverty status (assigned at the household- level) and individual-level characteristics (marital status). All differences between 10 males and females are statistically significant at 1 percent, except for the polygamously married population (significant at 10 percent). 0 0-4 5-9 10-14 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ 3.1.2 Gender differences in poverty using a Age group demographic taxonomy of households Poverty, male Poverty, female There is a large gender difference in poverty Source: KIHBS 2015/6. Note: Cross-tabulation of individuals’ poverty status (assigned at the household- amongst households with only a single adult (and level) and individual-level characteristics (age, sex). possibly children). Households comprising one adult female are twice as likely to be poor (35 percent) than Gender differences in poverty also emerge from a households comprising one adult male (18 percent) comparison of male and female poverty rates – i.e. (Figure 3.3). This reflects, among other things, that the probability of living in a poor household – by women living on their own are much more likely than marital status (Figure 3.2). Gender gaps are relatively men to care for children. Poverty rates are highest, at 40 small among the married population, though still percent, among households comprising only children/ favoring men. Women, however, are much more likely seniors and among households comprising 2 adults than men to be poor if they are separated/divorced of same sex or 3+ adults, typically multigenerational (31 vs. 24 percent, p<0.01) or widowed (38 vs. 25 households. Almost 42 percent of Kenya’s poor live percent, p<0.01). These findings are consistent with in these multigenerational households. Poverty rates other studies showing that, for African women, marital are somewhat lower (35 percent) for households rupture frequently entails a loss of economic means and comprising 2 adults of opposite sex, which are in most support that are acquired through, and conditional on cases nuclear families. However, due to their prevalence marriage—including access to productive assets (such in the population, these households still account for as land) and the marital home (Kevane 2004; Djuikom about 40 percent of Kenya’s poor population. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 61 Gender and Poverty Figure 3.3: Poverty and household demographic composition, 2015/6 Poverty rates by demographic household composition Distribution of the poor population by demographic household composition 50 40 40 40 35 35 30 Percent 20 18 10 1 adult female, 13% 0 Only 1 adult 1 adult 1 adult 2 adults Only male female male, 1 of same children 2 adults of same sex or 1 adult male, children 1 adult adult sex or 3+ and/or 3+ adults, 42% 1 adult female, 40% and/or male female adults seniors seniors Source: KIHBS 2015/6. Note: Household taxonomy is based on the number of male and female adults (18-64 years), irrespective of the number of children (<18 years) and elderly (65+ years). Left: Share of population below the poverty line by demographic taxonomy. Right: Distribution of poor population by demographic taxonomy. 3.2 GENDER GAPS IN ENDOWMENTS Girls also perform better than boys in Math, English and T he focus of this section is gender differences Kiswahili, especially in earlier grades of primary school in endowments. This includes human capital (Uwezo 2016). endowments – education and health – but also time “Previously the community preferred withdrawing availability and access to physical and financial assets. a girl child from school during times of economic Gender gaps in endowments not only matter in their own stress. After the introduction of free primary right, but also contribute to gender inequality in economic education, the situation has changed and all opportunities (highlighted in section 3.3) and are hence children have equal opportunity to attend critical for poverty reduction efforts. school.” (Namwitsula community, Busia) 3.2.1 Education Gender gaps in the education sector, however, Kenya has achieved significant increases in primary differ markedly across regions (Figure 3.4). Gross and secondary enrollments since the early 2000s, enrollment rates are higher for girls than for boys especially among girls. The 2005 Participatory Poverty in parts of Central and Western Kenya, but in most Assessment (PPA) already provided a glimpse of this other areas – especially the Northeast and Coast – the societal transformation, as illustrated below by a traditional patterns of higher enrollments among boys quotation from a community in Busia.97 Ten years on, still hold. In terms of learning outcomes (here math the trend towards higher girls’ enrollment is clearly proficiency) girls’ advantages are more widespread – visible. Between 2005/6 and 2015/6, gender parity in consistent with the results at the national level – but gross enrollments, defined as the ratio of female to show a broadly similar geographic pattern. These male enrollment rates, increased at the primary (from regional differences, which may reflect differences 0.95 to 0.97) and secondary (from 0.89 to 0.95) levels. across regions in broader development, female labor And since girls are less likely than boys to attend school force participation, religious and social norms, are over-aged (for the level at which they are enrolled), currently not well understood and would merit further NERs are even higher for girls than for boys (Table 3.1). in-depth analysis. 97 The communities interviewed for the 2005 PPA often commented that a greater emphasis on girls’ education came in the wake of Kenya’s FPE policy introduced in 2003. However, Lucas and Mbiti (2012a) argue that while FPE boosted primary school completion rates of girls and boys, it had larger effects on boys. These results suggest that FPE was not the primary driver for greater gender parity in Kenya’s school. 62 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Table 3.1: Primary and secondary enrollment rates and gender parity, 2005/6 and 2015/6 Primary Net Gross   Female Male Gender parity Female Male Gender parity   index index 2005/6 0.82 0.81 1.01 1.16 1.22 0.95 2015/6 0.85 0.84 1.01 1.06 1.09 0.97 Secondary Net Gross   Female Male Gender parity Female Male Gender parity   index index 2005/6 0.21 0.19 1.09 0.39 0.44 0.89 2015/6 0.44 0.41 1.08 0.73 0.77 0.95 Source: KIHBS 2005/6 and 2015/6. Note: The gender parity index is defined as the ratio of female to male enrollments. Figure 3.4: Regional differences in gender parity in the education sector Gender parity index Gender parity index Gender parity index Gross Primary Enrollment Rates Gross Secondary Enrollment Rates Uwezo - Math proficiency (1.4,1.5] (1.3,1.4] (1.2,1.3] (1.6,1.8] (1.1,1.2] (1.4,1.6] (1.2,1.3] (1,1.1] (1.2,1.4] (1.1,1.2] (.9,1] (1,1.2] (1,1.1] (.8,.9] (.8,1] (.9,1] (.7,.8] (.6,.8] (.8,.9] (.6,.7] (.4,.6] [.7,.8] [.5,.6] [.2,.4] Source: KIHBS 2015/6 and Uwezo 2014 data. Note: The gender parity index is defined as the ratio of female to male enrollment/proficiency rates. A value above (below) unity indicates that girls have higher (lower) levels of enrollments/proficiency. Girls and boys, when they drop out of school, about fertility and schooling are typically made jointly often do so for different reasons. Girls dropping out (see Ozier 2016; Duflo, Dupas, and Kremer 2015a). of secondary school are more likely to be married and to have given birth, than girls still attending While girls are enrolled in greater numbers in Kenyan school.98 Asked directly about the main reason why schools than ever before, adult women continue to a household member stopped attending secondary be disadvantaged in educational attainment and school (before secondary completion), “school cost” is literacy compared with adult men. At the national the most commonly cited reason for boys, followed by level, illiteracy is almost twice as high among women ”lack of interest.” For girls, the reason most commonly aged 15+ (18 percent) than among adult men aged mentioned is ”pregnancy”, followed by ”school cost”. 15+ (10 percent), and no county, except Nairobi, has However, the causality between pregnancy and achieved gender parity in literacy among this age group dropping out of school may run both ways, as decisions (Figure 3.5). This reflects historical gender inequalities in 98 Secondary dropout is defined as having attended secondary school the education sector, which continue to put women at Form 1-3 during the last school year, but no longer attending school a disadvantage in terms of labor market opportunities. during the current school year. Note that there are only few cases of secondary dropouts captured by the KIHBS N=70), which limits the analysis that can be performed. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 63 Gender and Poverty Figure 3.5: Male and female literacy by county, 2015/6 100 80 60 Percent 40 20 0 Nairobi Nyeri Mombasa Kiambu Kisumu Nandi Nakuru Machakos Trans Nzoia Kisii Nyandarua Vihiga Uasin Gishu Bungoma Kirinyaga Taita Taveta Siaya Muranga Embu Nyamira Kericho Homa Bay Migori Elgeyo Marakwet Bomet Baringo Kajiado Makueni Lamu Tharaka Nithi Kakamega Kitui Busia Meru Laikipia Tana River Isiolo West Pokot Kwale Samburu Mandera Garissa Wajir Marsabit Turkana Narok Kili Male Female Source: KIHBS 2015/6. Note: Respondents who report being able to read or write in any language or attended secondary school or above are considered as literate. Population aged 15+ years. 3.2.2 Health and fertility Women are disproportionately affected by the HIV/ Kenyan women face a staggering lifetime risk of 1 AIDS epidemic. While prevalence rates have declined in 42 of dying due to complications of pregnancy from about 7 percent in 2006 to just over 5 percent or child birth. While the maternal mortality ratio has in 2016 (of the total population aged 15-49 years), declined from 728 to 510 (maternal deaths per 100,000 women make up more than 60 percent of the share live births) between 2005 and 2015, it remains high of the population (15+) living with the disease (World by regional standards (Figure 3.6a). Geographically, Bank 2017b). Moreover, the 2008 Kenya Demographic maternal mortality is highest in the Northern/North and Health Survey (KDHS; the latest to include HIV/AIDS Eastern parts of Kenya (Figure 3.6b). These areas testing) shows that widows and divorced/separated of high maternal mortality also perform poorly in women are at particularly high risk, with prevalence rates terms of the share of live births being delivered by a that at the time were more than five times (widows) or skilled provider or in a health facility (Muraguri 2015), twice (divorced/separated women) as high as those of suggesting that lack of access and/or poor affordability the total female population (KNBS et al. 2010). Similar of reproductive care services play an important role demographic patterns have been observed for other (see chapter 7 on health). countries in Africa (Djuikom and van de Walle 2018). Figure 3.6: Maternal mortality a) MMR - 2015 model estimates, Kenya and comparators b) MMR - 2009 census estimates by county 600 547 510 398 400 353 319 290 200 138 (3000,4000] (2000,3000] (1000,2000] (800,1000] 0 (600,800] (400,600] SSA Kenya Tanzania Ethiopia Ghana Rwanda South (200,400] [0,200] Africa Based on 2009 Census. Source: WHO et al. (2015) and KNBS (2012). Note: The maternal mortality ratio (MMR) is defined as the number of maternal deaths per 100,000 live births. 64 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Kenya entered the demographic transition earlier noted that women and girls were disproportionately than most other African countries, but the fertility engaged in fetching water for the family and in care decline has somewhat slowed over the past two work – as illustrated below by quotations from two decades. Kenya’s TFR fell from about 8 births per communities from Kilifi. The KIHBS 2015/6 confirms woman in the 1960s to just over 5 births in the late these intrahousehold differences in labor allocations 1990s, a rate of decline that outpaced most other (see Figure 3.8). African countries at the time (Figure 3.7a). Fertility continued to decline throughout the 2000s to reach “Water collection is a responsibility of women about 4 births per woman in 2015, albeit at a slower and girls but in times of water scarcity, men are pace than observed, for example, in Ethiopia and also involved. Men are affected by lack of water Rwanda. From a geographic perspective there is huge in that it stops them from going to work. It is variation in fertility across regions. In 2014, counties socially frowned on for a man to fetch water. like Wajir or West Pokot still had a TFR above 7 births Women have to wait in long queues and do per woman, similar to Kenya’s national average during not have enough time to attend to household the 1980s or present-day Niger, the country with the chores and run their bossiness’s [sic]. In severe highest fertility in the world. At the other end of the water crisis children do not go to school so spectrum, counties like Kiambu, Kirinyaga, Nairobi as to look for water. It also denies them an or Nyeri have a fertility rate of around 2.5 births per opportunity to play. Women usually carry woman, only slightly higher than current-day Mexico water on their heads which they find tedious.” (Figure 3.7b). (Manjengo-Mariakani community, Kilifi) 3.2.3 Time use “Men and women play different roles when Gender differences in time use, related to social family members get sick. The women nurse norms about the division of labor inside the family, the patient by washing them, preparing their are among the most pertinent factors that distinguish meals and feeding them. The men mostly the lives of men and women in Africa (Blackden and provide money to cater for medical costs.” Wodon 2006). In the 2005 PPA, almost every community (Miyani community, Kilifi) Figure 3.7: Kenya’s demographic transition a) TFR, Kenya and comparators, 1960-2015 b) TFR by Kenyan county and comparators, 2014 Total fertiliy rate 8 9 8 6 7 4 6 2 5 4 0 Wajir West Pokot Turkana Samburu Garissa Tana River Migori Trans Nzoia Mandera Homa Bay Marsabit Bungoma Isiolo Baringo Kwale Busia Vihiga Kajiado Kakamega Lamu Bomet Siaya Keiyo-Marakwet Nandi Kericho Kitui Nakuru Laikipia Kisii Uasin Gishu Kisumu Nyandarua Nyamira Tharaka Machakos Makueni Taita Taveta Mombasa Meru Embu Murang'a Nyeri Nairobi Kiambu Kirinyaga Niger Mexico Narok Kili 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Year Kenya Tanzania Uganda Ethiopia Rwanda Kenyan counties Reference countries Source: World Bank 2018c and DHS 2018 StatCompiler (data for 2014) KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 65 Gender and Poverty Figure 3.8: Household members fetching water, 2015/6 3.2.4 Physical and financial assets Kenyan women are less likely than men to own land and housing property. 12 percent of women aged 7% 20-49 years report owning any land on their own, 9% compared with 39 percent of men – a gender gap of 27 percentage points. The gender gap in sole ownership is even larger for housing – 32 percentage points (Figure 26% 58% 3.9a). Since women are much more likely than men to report joint property ownership, gender gaps are much smaller if joint ownership is taken into consideration, but remain in favor of men (Figure 3.9b). Kenya’s gender gaps in property ownership are similar in magnitude to Boys Girls Men (15+) Women (15+) those found in Tanzania, but significantly larger than, for example, in Ethiopia, where there has been an Source: KIHBS 2015/6. emphasis on joint land registration (Melesse, Dabissa, Though Kenya lacks nationally representative data and Bulte 2018). to document these gender differences in time use, case studies confirm that women spend a significant Kenya is among the few African countries with amount of time on unpaid work. A report by Action gender inequality in formal inheritance rights Aid (2013) collected information on time use patterns (World Bank 2015b). As in other African countries, (based on diaries) across three sites in Kenya. The property rights of women in Kenya are shaped by legal study finds that (per day) women spend on average 99 pluralism, which includes vestiges of colonial, modern minutes on collecting fuel or water, and 359 minutes constitutional, customary and religious laws (Deere and on unpaid care work, together almost a full working Doss 2006; Harrington and Chopra 2010). While Kenya’s day (7 hours and 38 minutes). Men, conversely, report 2010 Constitution contains detailed articles in relation spending only 38 minutes on collecting fuel or water, to equality and non-discrimination, gender biases and 167 minutes on unpaid care work, together 3 hours linger in subordinate statutes. In particular, the Law of and 25 minutes. While these data are based on a small Succession Act distinguishes explicitly between male sample and not nationally representative, they give a and female surviving spouses (Republic of Kenya 2015; sense of the time scarcity of Kenyan women. World Bank 2015b). Figure 3.9: Kenya and comparators gender gaps in land and housing ownership a) Sole ownership b) Sole and joint ownership 60 60 Male ownership rate (%) - female ownership rate (%) Male ownership rate (%) - female ownership rate (%) 40 40 20 20 0 0 -20 -20 Dem Rep Congo Dem Rep Congo Mozambique Mozambique Côte d’Ivoire Côte d’Ivoire Burkina Faso Burkina Faso Sierra Leone Sierra Leone Zimbabwe Zimbabwe -40 Comoros Comoros -40 Tanzania Tanzania Namibia Namibia Ethiopia Ethiopia Rwanda Lesotho Rwanda Lesotho Senegal Senegal Uganda Uganda Burundi Burundi Gambia Gambia Zambia Zambia Guinea Guinea Nigeria Nigeria Malawi Malawi Liberia Liberia Ghana Ghana Kenya Kenya Benin Benin Niger Niger Chad Chad Togo Togo Mali Mali Housing Land Housing Land Source: KDHS 2014 and Gaddis, Lahoti, and Li 2018. Note: Self-reported property ownership in population aged 20-49 years. 66 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Gender gaps exist also in terms of access to ICT and lower on various indicators of financial inclusion (Figure financial services, though levels of access are high 3.11) and are more likely to report difficulties in coming by regional standards. Women are less likely to own up with emergency funds (Figure 3.12). However, a phone or to have a subscription to a mobile money access to financial services is still higher in Kenya than transfer platform than men, and the gender gaps in its comparator countries, apart from South Africa increase with age (Figure 3.10). Similarly, women score (Figure 3.13). Figure 3.10: ICT access by sex and age, 2014, 2015/6 a) Mobile phone ownership by sex and age b) Subscription to mobile transfer platform by sex and age 100 100 90 90 80 80 Phone ownership (%) 70 Subscriptions (%) 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 75+ 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 75+ 15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54 55-59 60-64 65-69 70-74 Age Age Phone, male Phone, female M-money transfer, male M-money transfer, female Source: KIHBS 2015/6. Figure 3.11: Financial inclusion, male and female population (15+), 2014 Saved to start, operate, or expand a farm or business Saved at a nancial institution Saved any money in the past year Outstanding mortgage Mobile account Debit card in own name Borrowed to start, operate, or expand a farm or business Borrowed from a nancial institution Borrowed any money in the past year Account at a nancial institution Account 0 20 40 60 80 Percent Male Female Source: Global Findex 2014 (Demirguc-Kunt et al. 2015). KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 67 Gender and Poverty Figure 3.12: Difficulty to come up with emergency funds, male and female population (15+), 2014 Male Female Very Not at all Very possible, possible, possible, 23.0% 17.8% 23.9% Not at all possible, 26.3% Not very possible Somewhat 24.0% Somewhat Not very possible, possible, possible, 34.8% 28.0% 21.2% Source: Global Findex 2014 (Demirguc-Kunt et al. 2015). Figure 3.13: Financial inclusion, Kenya and regional comparison, 2014 Account at a nancial institution, male and female population (15+) 70 60 50 40 Percent 30 20 10 0 South Africa Kenya Ghana Sub-Saharan Uganda Ethiopia Tanzania Africa Male Female Source: Global Findex 2014 (Demirguc-Kunt et al. 2015). 3.2.5 Policies to reduce gender gaps in endowments In the education sector, recent data on enrollments This section lays out possible policy levers to reduce and educational performance paint an uneven gender gaps in endowments. Given the cross- picture – with girls’ and boys’ advantages in different sectoral nature of the analysis, this naturally cannot be parts of the country. Further efforts to improve girls’ exhaustive. Moreover, most gender gaps do not have school enrollment, retention and attainment are still instant solutions but require fundamental changes needed in many parts of Kenya, where gender gaps in social norms about women’s and men’s roles and in the education sector continue to favor boys. But at abilities. The objective of the section is hence rather the same time, emerging boys’ underachievement, modest – to reflect on the previous analysis from a especially in educational performance, also requires policy perspective and to bring in additional empirical attention and should be further analyzed and addressed evidence on what works to close gender gaps in before the pattern becomes deeply entrenched endowments, especially from the growing impact (building, for example, on experiences in Caribbean evaluation literature. countries, which have experienced similar patterns, see Plummer 2010; USAID 2016). 68 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Several studies from Kenya suggest that programs preventative health products, empirical evidence from subsidizing the direct or indirect cost of education multiple developing countries, including Kenya, shows can be effective in increasing enrollment and that demand is highly price-sensitive. This suggests educational performance of boys and girls. Based that subsidies are a policy option to boost adoption, on a randomized evaluation across 328 public primary especially if targeted to women (Meredith et al. 2013). schools in Western Kenya, Duflo, Dupas, and Kremer In terms of HIV/AIDS, Duflo, Dupas, and Kremer (2015a) (2015a) show that a program that subsidized the cost show that the government’s HIV curriculum, which of upper primary education by providing free school emphasizes abstinence until marriage, does not reduce uniforms significantly reduced school drop outs for sexually transmitted infections (STIs) (nor teenage boys and girls. Similarly, Kremer, Miguel, and Thornton pregnancy). A joint program, where the HIV curriculum (2009) and Friedman et al. (2016) find that a merit- is combined with the education subsidy highlighted based scholarship program targeting adolescent girls earlier, reduced the prevalence of STIs among girls, but in Western Kenya increased academic test scores the education subsidy in isolation was more effective in among girls from treatment schools. The scholarships lowering dropouts and teen pregnancies. These results were awarded to the highest-scoring 15 percent of highlight the complexities of individual decision- grade-6 girls in the program schools in each district making around schooling and engagement in different and included financial grants to cover school fees forms of casual versus committed relationships, which and supplies and public recognition at an awards each carry different propensities for STI and early ceremony. Interestingly, the scholarship program had pregnancy. Policies targeting any one of these issues positive spillover effects on boys (who were ineligible should therefore be carefully evaluated for unintended for the scholarship) and on girls with low pretest scores consequences. (who were unlikely to win the scholarship). Public investments in services for care can reduce Increased secondary school enrollment among time constraints of women. Scaling up care services adolescent girls may also delay fertility decisions. for children, however, requires innovative approaches, The education subsidy program highlighted above for combining public and private sources of funding. IFC its impact on reducing dropouts also reduced teenage (2017) shows examples of employer-provided child pregnancies (Duflo, Dupas, and Kremer 2015a). care (including case studies of Safaricom and an Similarly, Ozier (2016), using a regression discontinuity agroprocessing company in Kenya), and discusses approach, shows that secondary school enrolment what policies and regulations the public sector can lowers teenage pregnancies among women. put in place to support private child care provision. Wattanga (2015) discusses an innovative initiative of In health, further initiatives to increase access to and the Nairobi City County to use social impact bonds affordability of reproductive health care services are to fund 97 new early childhood education centers in important to reduce maternal mortality, especially poor parts of the city. in Kenya’s arid and semi-arid regions. Examples of such efforts are the recent government-supported Further empirical work would be needed to “Linda Mama” program providing free maternity better understand how different types of public services. An evaluation of a pilot program in central infrastructure provision affect time use and the Kenya further demonstrates that post-natal follow ups, intra-household allocation of labor. A desktop review where community health workers visited or called new from Asia (ADB 2015) argues that improved access to mothers three days after delivery and administered a water reduces the time women spend fetching water, simple checklist, led to earlier utilization of postnatal but that this often leads to an increase in time spend care and better recognition of potential complications on other unpaid activities, such as caring for children. from pregnancy (McConnell et al. 2016). In terms of Investments in sanitation were found to reduce the KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 69 Gender and Poverty amount of time needed each day to find a place to and use their newly acquired bank accounts to protect defecate and reduce the burden of caring for family their income from such demands. The notion that members who fell sick due to poor sanitation. Access social pressure to share resources affects women’s to electricity was found to increase the number of decisions to save and invest is further supported by hours women spend on paid work, partly due to an two other studies. Jakiela and Ozier (2016) show in a lab increase in the number of waking hours. Transportation experiment that Kenyan women adopt an investment infrastructure reduced travel time for women but strategy that aims to conceal their initial endowments also added new time demands. More research along from relatives, even though this strategy reduces their these lines for Kenya and other African countries expected earnings. Schaner (2017) shows that offering would be important to understand how infrastructure ATM cards for newly-opened bank accounts (which investments affect women’s time constraints and intra- increases the liquidity of savings) increased account household labor allocation, particularly in rural areas. use (especially the number of transactions) of male- and jointly-owned accounts, but not of female-owned A review of Kenya’s legal landscape could help to accounts. This is consistent with the idea that women ensure the consistency of various laws on property prefer savings instruments with lower levels of liquidity, ownership and inheritance with the progressive as this protects their savings from the demands of principles of Kenya’s 2010 constitution. Gender spouses and other family members. In addition, Dizon, biased legislation, such as the differential treatment of Gong, and Jones (2017) show that accounts with soft male and female surviving spouses under the Law of commitment can help to increase women’s savings. Succession Act, should be eliminated. There is evidence Their study offered “labeled” mobile money (M-PESA) from other African countries that land formalization accounts to vulnerable women, who were existing programs promoting joint registration of both spouses users of M-PESA and already had an account. The can potentially improve outcomes for women and initiative encouraged the women to use the “labeled” narrow gender gaps (O’Sullivan 2017). Rwanda’s land account for emergency purposes and specific saving tenure regularization program, for example, which goals (to help mental accounting), and sent weekly registered married women as co-owners of land SMS reminders (nudges) related to their savings goals, by default, significantly improved documentation but did not affect financial access (since all women of informal land rights among married women (Ali, already had an existing M-PESA account). The program Deininger, and Goldstein 2014). However, at the same was found to increase women’s mobile money savings, time, women who are not legally married saw an without crowding out other types of savings. It also erosion of rights, which highlights the complexities of led to a substitution away from informal-risk sharing these interventions. arrangements, but did not reduce the women’s capacity to manage risks. Several recent studies from Western Kenya suggest that savings products with an element of illiquidity 3.3 GENDER INEQUALITY IN ECONOMIC and soft commitment can increase women’s savings OPPORTUNITIES T (O’Sullivan 2017). Dupas and Robinson (2013) show his section turns to gender inequality in economic that interest-free bank accounts with large withdrawal opportunities. It starts with a brief description of fees increased savings of female market vendors, while trends in male and female labor market indicators over no such effects were observed for male bicycle-taxi the past decade, and a portrayal of the current situation drivers. A potential explanation for the high take- based on the 2015/6 KIHBS data. The section then reviews up rates of accounts by women – despite (de facto) gender gaps in three broad segments of the labor market: negative interest rates – is that women face pressures in wage employment, non-farm household enterprises to share their income with family members and friends and in agriculture. 70 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Figure 3.14: Percent of population employed by category, 3.3.1 Labor market trends and current situation 2005/6 – 2015/6 There has been a striking increase in male and, 80 73 particularly, female employment over the past 65 63 decade. Male employment increased from 63 percent 60 51 in 2005/6 to 73 percent in 2015/6, while female 48 employment increased from 51 percent to 65 percent 40 Percent 40 38 38 39 over this period (Figure 3.14). For men, this increase 30 was driven by a rise in wage employment, while for 20 21 14 women it reflects both rising wage employment and also increasing employment in household enterprises, 0 which here includes both farm and non-farm Male Female Male Female 2005/06 2015/16 enterprises.99 Wage Enterprise Any employment Source: KIHBS 2005/6 and 2015/6. Rising employment has transformed the school- Note: Working-age population (15-64 years). Comparable employment definition. to-work transition of Kenyan youth. There is an increasing number of adolescents, male and female, between 2005/6 and 2015/6. Nonetheless, young who are working while still in school. Moreover, the women continue to be significantly more likely than share of the population below the age of 35 who are young men to be neither employed nor in school neither employed nor in school significantly declined (Figure 3.15). Figure 3.15: Changes in school-to-work transition, 2005/6-2015/6 School-to-work transition School-to-work transition Males - 2005/6 1 Females - 2005/6 1 .8 .8 .6 .6 .4 .4 .2 .2 0 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Employment only Employment and school School only Neither Employment only Employment and school School only Neither School-to-work transition School-to-work transition Males - 2015/6 Females - 2015/6 1 1 .8 .8 .6 .6 .4 .4 .2 .2 0 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Employment only Employment and school School only Neither Employment only Employment and school School only Neither Source: KIHBS 2005/6 and 2015/6. Note: Comparable employment definition. 99 See Appendix C2 for a discussion of the comparability of the 2005/6 and 2015/6 KIHBS labor modules. This section uses a definition of employment and labor force participation that is broadly consistent with the labor statistics standards adopted by the 13th International Conference of Labor Statisticians (ICLS) in 1982. The changes adopted by the 19th ICLS in 2013, which reduce the definition of employment to work performed for pay or profit (thus excluding subsistence agriculture) are not yet incorporated in the KIHBS 2015/6 instrument and hence are not considered in this section. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 71 Gender and Poverty In 2015/6, Kenya had a female labor force Within Kenya, there are significant regional participation rate of 71 percent for the core working- differences – female labor force participation is high age population (15-64 years), compared with a male in central and Western Kenya, but much lower in labor force participation rate of 77 percent (Figure the Northeast. Male labor force participation, though 3.16). The labor force comprises those who are (i) also somewhat lower in the Northeast, varies less. employed and at work, those who are (ii) employed As a result, gender gaps in labor force participation but temporarily absent from work, and those who (measured here as the absolute gap, in percentage are (iii) unemployed.100 In the case of Kenya, both points) are most pronounced in the Northeast, unemployment and temporary absence only account followed by the coast and the areas bordering Tanzania for a small share of the labor force. In terms of regional comparisons, Kenya’s female labor force participation (Figure 3.18), where women are much less likely than rate is higher than the Sub-Saharan African average, men to participate in the labor force. These areas but lower than in most other East African countries, map closely with Kenya’s arid and semi-arid lands, except for Uganda (Figure 3.17). where livestock rearing, particularly of cattle, makes Figure 3.16: Male and female labor force participation, 2015/6 Male share employed, unemployed and out of the labor force (%) Female share employed, unemployed and out of the labor force (%) 100 100 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 Employed - at work Employed - absent Employed - at work Employed - absent Unemployed Out of the labor force Unemployed Out of the labor force Source: KIHBS 2015/6. Note: Working-age population (15-64 years). Figure 3.17: Female labor force participation, Kenya and comparators 100 86.1 80 77.0 79.5 74.5 70.0 66.4 62.8 60 Percent 47.8 40 36.6 20 0 Lower middle South Africa Sub-Saharan Uganda Kenya (KIHBS Ghana Ethiopia Tanzania Rwanda income Africa 2015/6) Source: KIHBS 2015/6 and World Bank 2017b. Note: Population aged 15+. 100 To be counted as unemployed, a person must meet the following three criteria: (i) not be presently employed, (ii) available to work and (iii) actively looking for a job (see Appendix C2). 72 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Figure 3.18: Geographic variation in male and female labor force participation, 2015/6 Female labor force participation (%) Male labor force participation (%) Gender gap in labor force participation (40,50] (80,90] (30,40] (80,90] (20,30] (70,80] (70,80] (10,20] (60,70] (60,70] (0,10] (50,60] (50,60] (-10,0] (40,50] (40,50] (-20,-10] (30,40] (30,40] (-30,-20] (20,30] (20,30] (-40,-30] [10,20] [10,20] [-50,-40] Source: KIHBS 2015/6. Note: Population aged 15-64. The gender gap is defined as the absolute difference between male and female labor force participation rates, a value greater (lower) than zero indicates higher (lower) male labor force participation. an important contribution to local livelihoods. On the another non-Christian religion reduces the probability other hand, male and female labor force participation of participating in the labor force by about 30 percent rates are similar in most parts of Central and Western (relative to being Catholic). Women who are widowed, Kenya. separated/divorced or polygamously married are significantly more likely to participate in the labor force Multivariate analysis points to the salience of religious than women who are monogamously married (or living norms, education, marital status and the presence of together). Every child aged 0-5 years reduces a woman’s young children for women’s participation in the labor probability to be in the labor force by over 2 percent. force. Following Klasen and Pieters (2015) we estimate Living in urban areas reduces a woman’s likelihood to the probability of being in the labor force for men and be in the labor force by another 7 percent, perhaps a women conditional on different socio-demographic reflection of greater difficulties in combining child care variables. The results are summarized in Figure 3.19 with labor market activity in urban areas, where most (marginal effects significant at 10 percent at least) – economic opportunities are outside of agriculture. On the complete set of coefficients is reported in Table the other hand, education, even at the primary level, C.1, Appendix C.4. Among women, being of Muslim or increases a woman’s probability to participate in the Figure 3.19: Correlates of male and female labor force participation, 2015/6 (Marginal e ects, only if signi cant at 10 percent at least) Urban Number of children aged 6-14 years Number of children aged 0-5 years Head's education = 4, other Head's education = 3, university Head's education = 2, secondary Head's education = 1, primary (own) education = 4, other (own) education = 3, university (own) education = 2, secondary (own) education = 1, primary Religion = 5, None Religion = 4, Other Religion = 3, Muslim Religion = 2, Protestant/other Christian Marital status = 5, never married Marital status = 4, widow or widower Marital status = 3, separated or divorced Marital status = 2, polygamously married Age in years -0.35 -0.3 -0.25 -0.2 -0.15 -0.1 -0.05 0 0.05 0.1 0.15 Male Female Source: KIHBS 2015/6. Note: Marginal effects after probit estimation (see Table C.1, Appendix C.4). The figure only shows marginal effects significant at the 10 percent level at the minimum. Reference categories as follows: Head’s/own education – no schooling; Religion – Catholic; Marital status – monogamously married. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 73 Gender and Poverty labor force. Interestingly, we do not find any effects Male and female employment in Kenya shows the on women’s labor force participation of the level of traditional patterns of sectoral segregation. Women education of the household head, suggesting that are disproportionately employed in agriculture and income effects are not very important. Religious norms, services, while men have a higher share of employment education and the presence of children are much in the industrial sector (Figure 3.20). Further analysis less important factors for labor force participation at the detailed industry level shows large differences of men. Marital status plays a role, with men who are across sectors in the female intensity of employment separated/divorced or have never been married being (Figure 3.21). The highest female intensities of significantly less likely to participate in the labor force employment are found in the sectors “activities of relative to monogamously married men (essentially household as employers” (i.e. domestic personnel), the opposite pattern as was found for women). These “accommodation and food services” (i.e. the hotel and results are consistent with traditional norms of married restaurant industry) and “human health and social men being the main breadwinner for their families. work”. On the other side of the spectrum, the lowest female intensities are found in “transportation and storage”, “construction”, and “mining and quarrying”. Figure 3.20: Male and female employment by broad sector, 2015/6 Male Female 42.45% 39.7% 42.45% 54.78% 4.77% 17.85% Agriculture Industry Services Agriculture Industry Services Source: KIHBS 2015/6. Figure 3.21: Share of male/female employment by detailed sector, 2015/6 Crop and animal production, hunting and related service activities Wholesale and retail trade; repair of motor vehicles and motorcycles Construction Manufacturing Education Transportation and storage Other service activities Accommodation and food service activities Activities of households as employers, undi erentiated Administrative and support service activities Human health and social work activities Sorted Public administration and defense; compulsory social security by total Professional, scienti c and technical activities employment Mining and quarrying (in descending Information and communication order) Financial and insurance activities Arts, entertainment and recreation Water supply; sewerage, waste management and remediation activities Real estate activities Electricity, gas, steam and air conditioning supply Activities of extraterritorial organizations and bodies 0 20 40 60 80 100 Percent Male Female Source: KIHBS 2015/6. 74 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty 3.3.2 Wage employment Table 3.3: Male and female monthly earnings, in current Ksh, and male-to-female ratio, 2015/6 Far fewer women than men are in wage employment. Ratio male- Male Female Of the total employed population, almost 50 percent to-female of men are paid employees (denoted in this section as Mean 18,276 14,075 1.30 wage earners) in their primary job, compared with only 10 percentile th 3,000 2,000 1.50 30 percent of women. On the other hand, more women Median 10,000 6,500 1.54 than men work as own-account or contributing family 90th percentile 43,300 35,000 1.24 workers (Table 3.2).101 Source: KIHBS 2015/6. Note: Unconditional earnings (cash and in-kind) of wage earners. Not normalized for working hours. There are large gender gaps in monthly earnings among wage earners, with mean wages/salaries to these characteristics) and an interaction between being 30 percent higher for men than for women. the endowment and coefficient effects.102 As shown These gender gaps are even larger at the bottom in Table C.2, Appendix C.4, the average difference of the earnings distribution (and up to the median), between log earnings of male and female employees where men earn about 50 percent more than women in the regression is 0.37, which corresponds to about (Table 3.3). 45 percent higher wages for male wage workers (consistent with Table 3.3 above).103 The endowment Differences in characteristics between male and effect explains about 43 percent of this difference, female wage earners – in terms of age, education, while 65 percent are explained by the coefficient usual number of working hours, industry, occupation effect. In addition, there is a negative interaction and urban-rural location – explain about half of effect, of -8 percent. the gender gap in monthly earnings. We use the Oaxaca-Blinder decomposition to disaggregate Exploring the endowment effect in detail shows that the gender difference in average monthly earnings the largest advantage of male wage workers is their into an endowment effect (reflecting differences in overrepresentation in industries with relatively high characteristics between male and female wage workers), wage premiums (Table C.3, Appendix C.4). In addition, a coefficient effect (reflecting differences in returns males wage workers benefit from being, on average, Table 3.2: Male and female wage employment by employment status, 2015/6 Population aged 15-64 years, primary job Male Female Total Paid employee (outside household) 47.7 27.4 37.9 Paid employee (within household) 2.2 2.3 2.2 Working employer 1.1 0.7 0.9 Own-account worker 35.9 51.7 43.5 Member of producer cooperative 0.1 0.1 0.1 Contributing family worker 11.3 16.1 13.6 Apprentice 0.7 0.9 0.8 Volunteer 0.4 0.3 0.3 Other (specify) 0.7 0.6 0.6 Total 100 100 100 Source: KIHBS 2015/6. Note: Column percentages. 102 The decomposition is implemented in Stata using the oaxaca command 101 While the KIHBS makes a distinction between own-account and with survey settings and default options (see Jann 2008). contributing family workers, the criteria to distinguish between these 103 Since the dependent variable is log-transformed, we follow Halvorsen types of workers in the context of small, family-run enterprises are often and Palmquist (1980) in calculating the percentage difference in earnings not clearly defined (Beegle and Gaddis 2017). as (exp(0.37)-1)*100=44.8 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 75 Gender and Poverty slightly older than female wage workers, (35 years vs. as high as profits of female-run enterprises. Jointly 33 years) and working longer hours per week (52 vs. run enterprises come out in-between (Figure 3.22).104 46 hours). Female wage workers, on the other hand, The unconditional gender gap is similar in magnitude benefit from having slightly higher levels of education to what is reported in Hallward-Driemeier (2013) based and being overrepresented in occupations with higher on an analysis of household enterprise modules for 19 wage premiums. These two effects, however, are just Sub-Saharan African countries. at the margin of statistical significance and cannot Figure 3.22: Profits of male-, female- and jointly-run compensate the other endowment advantages of household enterprises, 2015/6 male workers. .4 The sizable coefficient effect suggests that men also .3 benefit from more favorable returns to characteristics – but further disaggregation of this effect does not Density yield additional insights. First, apart from age (where 2 women benefit from greater returns to experience), none of the disaggregated coefficient effects is .1 statistically significant in isolation. Second, most of the male advantage in the overall structural effect 0 reflects differences in the regression intercept for male 2 4 6 8 10 12 Pro t (Natural log) and female wage workers, which potentially captures Male Female Joint gender-based discrimination in the labor market, but Source: KIHBS 2015/6. Note: Monthly profits (winsorized) in current Ksh (figure shows natural log). also unobservable factors, and is therefore difficult to interpret (Table C.4, Appendix C.4). Compared with male-run enterprises, female-run household enterprises are less likely to be in industry, 3.3.3 Non-farm household enterprises less likely to be formally registered and tend to Gender gaps in earnings carry over to the non-farm employ fewer paid non-household workers. They household enterprise sector, with average profits of are also less likely to be in urban areas and are more male-run household enterprises being about twice concentrated in poor households (Table 3.4).105 Table 3.4: Descriptive differences between male- and female-run household enterprises, 2015/6 Male-run Female-run By sector (%) Agriculture 2.0 1.9 Industry 15.2 9.6 Services 82.8 88.5 Share registered (%) 15.3 9.2 Labor input Average number of household or unpaid workers 1.2 1.1 Average number of paid non-household workers 0.4 0.1 Share in urban areas (%) 51.5 45.4 Share in poor households (%) 15.2 20.0 Source: KIHBS 2015/6. 104 See Appendix C3 for details on the classification of enterprises as male- female or jointly run. 105 The KIHBS 2015/6 data only collect very limited information at the enterprise-level – i.e. its sector, whether the enterprise is registered with the Registrar of Companies, and male and female labor inputs. For this reason, we do not perform a full decomposition analysis in this section. 76 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Differences in profits between male- and female-run 3.3.4 Agriculture household enterprises do not, however, primarily Even though women make up 56 percent of the reflect differences in the distribution of enterprises total population employed in agriculture, they are across sectors, formal registration, labor input or the primary decision maker on only 39 percent urban-rural location. Regressing log profits on a of agricultural plots (Figure 3.23). This reflects, to dummy variable that captures whether the enterprise is some degree, gender differences in land ownership female-run shows that profits of female-run enterprises documented in section 3.2.4, though land ownership are about 52 percent lower than those of male-run and land use rights do not necessarily fall together enterprises (i.e. a coefficient of -0.73, see Table C.5, (Slavchevska et al. 2017; Doss, Kieran, and Kilic 2017; column 1, Appendix C.4). Controlling for enterprise Gaddis, Lahoti, and Li 2018). characteristics (sector, urban, registration and labor input) reduces this to 43 percent lower profits for There are significant differences in input use and female-run enterprises (i.e. a coefficient of -0.57, see cropping choices between male and female farmers Table C.5, column 3, Appendix C.4). In other words, even (Table 3.5). Parcels managed by men are larger, more after controlling for these differences in characteristics, likely to be irrigated and more likely to use fertilizer than female-run enterprises achieve much lower profits parcels managed by women. Likewise, households than male-run enterprises.106 Unfortunately, the KIHBS where the primary decision-maker in agriculture is male data do not make it possible to investigate the role spend significantly more on labor and non-labor inputs of other key enterprise characteristics, such as capital than households where the primary decision-maker is intensity or access to finance, that are often found to female.107 Interestingly, female primary decision-makers contribute to performance gaps between male and appear to be more diversified, cultivating a slightly larger female entrepreneurs (Hallward-Driemeier 2013). number of crops. This is, however, entirely driven by a Figure 3.23: Gender differences in agricultural employment vs. parcel management, 2015/6 a) Agricultural employment b) Agricultural parcel management 43.6% 38.5% 56.4% 61.5% Male Female Male Female Source: KIHBS 2015/6. Note: Agricultural employment shows the male-female composition of the total population employed in agriculture. Parcel management shows the male-female composition of the primary decision-makers (regarding input use and cropping activities) on agricultural parcels. 107 The KIHBS 2015/6 asks for each parcel which household member makes decision on input use and cropping activities and this is used to determine the primary-decision maker. However, many inputs are collected at the household- (e.g. labor and non-labor input cost) or 106 In this respect, the Kenyan results differ from those reported in Hallward- crop-level (i.e. use of improved seeds). To analyze gender differences, Driemeier (2013) for 19 Sub-Saharan African countries, where controlling we distinguish between household where the primary decision-maker for whether the enterprise is registered reduced the gender gap by is male vs. female based on the share of the household’s agricultural land about half. that is being managed by male vs female household members. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 77 Gender and Poverty Table 3.5: Descriptive differences in input use between male and female decision-makers in agriculture, 2015/6 Parcel-level Household-level Male Female Significance Male Female Significance decision- decision- level of decision- decision- level of maker maker gender gap maker maker gender gap Land size (ha) 0.58 0.49 *** 0.80 0.66 *** Share irrigated 0.06 0.04 *** 0.08 0.05 *** Share using inorganic fertilizer 0.62 0.58 *** 0.65 0.60 *** Total input cost (KSh/year) n.a. 9,337 6,291 *** Total labor cost (KSh/year) n.a. 5,256 4,014 *** Number of cash crops n.a. 0.43 0.33 *** Number of food crops n.a. 1.95 2.23 *** Total number of crops n.a. 2.39 2.55 *** Source: KIHBS 2015/6. Note: *** denotes p<0.01 greater number of food crops, as households having trade through small/large traders and millers, while a man as the primary decision-maker cultivate more households where women are the primary farmers cash crops. Overall, these results are consistent with are more likely to sell through consumers, neighbors World Bank (2013a), which provides a more in-depth and cooperatives. Also, a greater proportion of women analysis of gender differences in agriculture using data (22 percent) than men (7 percent) reported that their collected under the Kenya Agricultural Productivity and spouse kept the revenue from crop sales (dried maize), Agribusiness Project. even in cases where women were managing the production of the crop. The study further showed that Households where the primary decision-maker in female farmers are less likely than male farmers to seek agriculture is female achieve, on average, yields advice from extension service providers. that are 15 percent lower for maize and 8 percent lower for beans than households where the primary- 3.3.5 Policies to reduce gender gaps in economic decision maker is male. Maize and beans are the opportunities two most common food crops. Decomposing these The preceding analysis has shown that Kenyan gender differences in yields using the Oaxaca-Blinder women’s decision to participate in the labor force is decomposition method (as described in section 3.3.2) strongly influenced by cultural and religious norms and a regression model similar to the one used in and there is some evidence that gender norms can chapter 4 on agriculture, shows that gender differences be transformed through programs targeting young in endowments (especially household size, as a proxy adolescents (e.g. Lundgren et al. 2013 for Nepal). for household labor availability, use of certified seeds, However, more rigorous empirical evidence is needed and spending on non-labor inputs) explain more than to understand if such programs work in conservative, 70 percent of the gap in maize yields, but only about 20 percent of the gap in beans yields.108 traditional societies, like Northeastern Kenya, where gender gaps in labor force participation are most Gender differences also emerge with respect to prominent. An ongoing evaluation by the Africa Gender trading channels, decision-making power over crop Innovation Lab of the CHOICES program in Somalia income and the use of agricultural extension services. (P165258) will provide further insights in a culturally As highlighted in World Bank (2013a), households similar context.109 where men are the primary farmers are more likely to 109 This is a program designed to transform attitudes and behaviors of 108 The fact that the KIHBS data do not allow assigning household-level very young adolescent girls and boys aged 10-14 years towards greater inputs to specific crops may also play role in explaining the difference in gender equality, which are perceived as markers of future labor market results for maize and beans. decisions. 78 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty Social protection programs need to pay attention teams. Moreover, the Africa Gender Innovation Lab to the specific vulnerabilities of women who went is evaluating the effectiveness of coding boot camps through a marital dissolution, especially if they are with a focus on gender under the Kenya Industry and also caring for children. As shown in this section, Entrepreneurship Project (KIEP, P161317). A recent female labor force participation is strongly linked gender assessment of the oil and gas sector in Kenya to marital status and the number young of children provides additional recommendations tailored to living in the household. Moreover, section 3.1 showed extractive industries, where women’s participation has that women who went through a marital dissolution traditionally been low (Cardno 2018). (divorce, death of their spouse) are significantly poorer than their male counterparts. The results also reinforce Business training programs hold some promise to the need to invest in care services to allow women with enhance the performance of female-run household children to participate in the labor force. enterprises, though more rigorous empirical evidence would be needed to support the effectiveness of While patterns of sectoral segregation are highly a specific curriculum. Reviews of business training persistent, a few studies suggest that information programs in developing countries have found that, interventions and, possibly, mentoring programs in general, the effectiveness of trainings differs across hold promise. Findings from an evaluation in Western study contexts and curriculums, and is often worse for Kenya of the national vocational training program women than for men (McKenzie and Woodruff 2014). show that information interventions, which emphasize However, a recent evaluation of the International Labour the discrepancies in expected earnings for graduates Organization’s (ILO’s) “Get Ahead” business training of traditionally male-dominated trades (e.g. mechanic) program, has found that the program significantly vs. female dominated trades (e.g. seamstress) can increased the sales and profits of female market encourage women to enroll in male-dominated vendors three years after the intervention (McKenzie professions (Hicks et al. 2016). A study from Uganda and Puerto 2017). The study, which was conducted in of female entrepreneurs who managed to succeed four counties in Western and Eastern Kenya, also did in male-dominated sectors highlights further the not find any evidence of negative spillover effects on importance of mentoring relationships and role models non-treated businesses, as markets as a whole appear (Campos et al. 2015). to have grown in terms of customers and sales volumes as a result of the intervention. Technological change has the potential to disrupt traditional patterns of sectoral segregation. New Empirical evidence from across Africa suggests business models, such as Uber and other ride-hailing that providing access to formal savings products services, can open up opportunities for women in is a promising approach to improve labor market traditionally male-dominated sectors like transportation outcomes of women (Campos and Gassier 2017). In (IFC 2018). Ongoing World Bank activities explore the Kenyan context, Suri and Jack (2016) show that the options to increase women’s participation in Science, rollout of the country’s mobile money system M-PESA Technology, Engineering and Mathematics (STEM) induced women to move out of agriculture into the occupations and may provide additional guidance non-farm enterprise sector, thereby contributing to over the lifespan of this assessment. Specifically, the a reduction in poverty, which was more pronounced project “Women in STEM – Infrastructure” (P166990) among female-headed households. In a similar vein, seeks to collate practical strategies on the recruitment, Dupas and Robinson (2013) find that better access retention and promotion of women in STEM to formal savings products increased productive occupations, specifically in infrastructure sectors, and investments of female entrepreneurs in Western Kenya. to develop a compendium of good practices for project KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 79 Gender and Poverty More research is needed on how to reduce gender 3.4.1 Women’s mobility gaps in agricultural productivity. There is still a lack of Social norms for acceptable behavior often constrain empirical evidence on what interventions are effective women’s physical mobility, i.e. their ability to move in closing these gaps. A few studies suggest that freely beyond the household. In 2014, 22 percent of programs that strengthen women’s property rights over Kenyan women and 19 percent of Kenyan men agreed land and tenure security can increase investment and with the statement that a husband is justified in hitting productivity among female farmers (Goldstein and Udry or beating his wife if she goes out without telling him. 2008; Ali, Deininger, and Goldstein 2014). In addition, an Acceptance of social norms that limit women’s mobility ongoing research project of the Africa Gender Innovation is strongly linked to poverty, with 36 (31) percent of Lab on agricultural labor constraints of female farmers Kenyan women (men) in the poorest quintile agreeing (P166082) might provide useful information, as studies with the above statement, compared with 9 (13) from other Sub-Saharan African countries have shown percent in the wealthiest quintile (Figure 3.24a). Yet, consistently that female farmers’ lack of access to labor social norms are changing rapidly, as the share of the is a key determinant of the gender gap in agricultural population who agreed with the above statement fell productivity (O’Sullivan et al. 2014). by about half between 2003 and 2014 (DHS 2018). Kenyans are also less likely to agree with the above 3.4 VOICE AND AGENCY statement than the population in most other African G ender gaps in endowments and economic opportunities are in many cases a reflection of women’s lack of agency. Agency is the ability to make countries (Figure 3.24b). Constraints on women’s physical mobility curb their decisions about one’s own life and act on them to achieve labor market opportunities and life choices. They not desired outcomes (World Bank 2015a). Differences only directly affect women’s preferences for seeking between men and women’s ability to make these choices, employment outside the home, but also limit women’s usually to the detriment of women, exist in all countries access to education, markets, banks and social networks and cultures. This section zooms in on two expressions and thus affect labor market behavior indirectly of agency –women’s mobility and their freedom from (Chakravarty, Das, and Vaillant 2017). Salon and Gulyani gender-based violence (Klugman et al. 2014). (2010), using data collected in informal settlements in Nairobi in 2004, show that working women are less Figure 3.24: Acceptance of norms that constrain women’s physical mobility a) Share of women/men accepting wife beating b) Share of women/men accepting wife beating, Kenya and by residence and quintile, 2014 comparators if wife goes out without telling her husband if wife goes out without telling her husband 100 40 Percent of population (15-49) Percent of population (15-49) 80 30 60 20 40 10 20 0 0 Guinea Chad Mali Sierra Leone Dem Rep Congo Burundi Ethiopia Niger Gambia Tanzania Congo Uganda Senegal Burkina Faso Zambia Liberia Comoros Cote d'Ivoire Cameroon Nigeria Zimbabwe Rwanda Kenya Gabon Togo Ghana Angola Namibia Lesotho Benin Mozambique Malawi Second Highest Lowest Urban Fourth Rural Middle Total Residence Wealth quintile Male Female Male Female Source: KDHS 2014 (KNBS et al. 2015) and DHS STATcompiler (DHS 2018). Note: Population aged 15-49 years. 80 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Gender and Poverty likely to travel outside their home settlement for work Figure 3.25: Share of women (15-49) who experienced physical violence by marital status, 2014 and, if they do commute, they are less likely than men 70 to use motorized transportation. Mobility constraints 64 can also further increase the time women spend on 60 domestic tasks and hence contribute to time poverty. 50 47 For example, some communities visited for the 2005 40 Percent PPA reported that it was “inappropriate” for women to 32 30 use bicycles or wheelbarrows for fetching water. 24 25 20 12 3.4.2 Gender-based violence 10 In 2014, 45 percent of women aged 15-49 years 0 Never Married/ Divorced/ reported having (ever) experienced physical married living separated/ together widowed violence, while nearly half of all ever-married women Past 12 months Ever experienced at least one form of intimate partner Source: Zumbyte 2018 based on KDHS 2014. violence (IPV, i.e. emotional, physical, or sexual IPV).110 Gender-based violence is a serious violation of women’s to experience IPV, compared to those married younger voice and agency and can lead to reduced mobility, than 15 years, which highlights the importance of less access to economic opportunities and long-term eliminating child marriages. Women who have been in physical and mental health issues – for the women more than one union have about four times the odds themselves, but also their children. of IPV compared with women who have been in just one union, which shows again the vulnerable position Women who have ever been married, and especially of women who underwent a marital dissolution. those who have gone through a marital dissolution, Women who are employed for cash are twice as likely are more likely to have experienced physical violence to experience IPV than women who are not working. than women who have never been married – which In terms of her partner’s characteristics, the risk of a reflects that violence is often perpetrated by current women experiencing IPV declines with the education or former spouses (Figure 3.25). In addition, there is level of her spouse, but increases if her spouse has a strong regional variation – with the highest rates of history of alcohol abuse. Perhaps surprisingly, women’s physical violence being reported in Nyanza, Nairobi education does not have a significant association with and Western regions. the experience of IPV. However, there is evidence from other studies that education may positively affect Multivariate analysis shows that the risk of a women’s attitudes towards domestic violence. For woman experiencing (physical) IPV is linked to example, the evaluation of a merit-based scholarship her age at marriage, whether she remarried and program targeting adolescent girls (discussed in her employment status – though her partner’s section 3.2.1) found that the program led adolescent characteristics play an important role as well. girls to reject the legitimacy of domestic violence Women who marry older than 25 years are less likely (Friedman et al. 2016). 110 This section draws on Zumbyte (2018). It reports standardized measure of gender-based violence (see KNBS et al. 2015 for details). The reported incidence of physical violence declined from 47 percent in 2003 to 39 percent in 2008/9, and then increased to 45 percent in 2014 (DHS 2018). This uneven trend, which may partly reflect reporting behavior, merits further investigation. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 81 CHAPTER 4 AGRICULTURE AND RURAL POVERTY SUMMARY Rural Kenyan households experienced a remarkable decline in poverty over the last decade, independent of their source of income. The proportion of the rural population living below the national poverty line declined from 50.5 percent (14.3 million people) in 2005/06 to 38.8 percent (12.6 million) in 2015/16. Households with diversified income from both agricultural and non-agricultural sources accounted for most of the poverty reduction, followed by agriculture and non-agricultural households. Improved infrastructure, including mobile network access, has raised the welfare of rural households, particularly of those with diversified income. In recent years, mobile network coverage improved substantially in rural areas, enhancing the efficiency of labor and agricultural markets. Improved coverage made it possible for mobile networks to not only serve as a communication tool but also constitute a platform for service delivery in rural areas. This has especially benefited households that rely on both agricultural and non-agricultural income, suggesting that off-farm diversification has been important for poverty alleviation efforts in Kenya over the last decade. Although productivity growth in the production of many crops has been stagnant, increased agricultural productivity remains a potential pathway out of poverty for many households. Little progress has been made in terms of raising productivity in the agriculture sector, especially concerning the production of maize, Kenya’s main food staple, and commercial crops such as coffee. Increased efficiency in the production of beans appears to be the only exception. As a result, agricultural productivity has not been contributing to poverty reduction in rural Kenya, a marked difference from the experience of other countries in the region such as Ethiopia. Nevertheless, more productive farmers are less likely to be poor in Kenya. This correlation between farm productivity and poverty constitutes promising evidence that an improvement in agricultural yields could lead to a reduction of poverty. Agricultural commercialization has helped to improve the livelihoods of Kenya’s farmers. Between 2005/06 and 2015/16, the country’s level of agricultural commercialization increased, and agricultural households sold a higher share of their production. Given that agricultural yields have been stagnant, better access to markets, as a result of infrastructure investments and better access to information and communication technologies, is the likely cause for higher levels of commercialization in the sector. Since farmers that sell a higher share of their products exhibit a lower incidence of poverty, agricultural commercialization is likely having a positive contribution to poverty reduction in Kenya. High commodity prices and increased productivity in the production of bean crops have also contributed to an improvement in the welfare of agricultural households. Many Kenyan farmers have shifted to bean production in recent years, as the country benefited from favorable bean and maize prices in 2011-16. Data suggest that farmers that shifted to bean production were less likely to be classified as poor. However, the increase in crop prices is generally beneficial for Kenya’s net-selling farmers at the expense of the urban poor, as poor urban households spend a large share of their income on food and are therefore sensitive to rising food prices. This may have contributed to the large divergence in poverty reduction between urban and rural areas. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 83 Agriculture and Rural Poverty 4.1 THE DECLINE IN RURAL POVERTY HAS The large drop in rural poverty along with a BEEN THE MAIN DRIVER OF POVERTY stagnating urban poverty rate cannot solely REDUCTION NATIONALLY explained by rural-urban migration. The migration R ural poverty alleviation has been driving Kenya’s of large numbers of poor rural households to urban overall progress in reducing poverty over the last areas can lead to a decline in rural poverty without decade. The country’s national poverty rate declined an actual improvement in livelihoods. Moreover, an from 46.6 percent in 2005/06 to 36.1 percent in inflow of poor households to urban areas can raise 2015/16, driven by a substantial decline in rural poverty, the urban poverty rate. However, while the share of from 50.5 percent to 38.8 percent in the same period Kenya’s population living in rural areas declined by 8 (Figure 4.1). By contrast, urban poverty declined by only percent between 2005/06 and 2015/16, households 2.7 percentage points, from 32.1 percent in 2005/06 to that migrated to urban areas were not from the 29.4 percent in 2015/16. bottom part of the distribution, as will be further explored in the Chapter 5. As a result, factors other While rural poverty has declined across Kenya, the than migration must explain the country’s progress in rate of poverty reduction has varied significantly reducing rural poverty. across provinces. Rural poverty headcount rates varied substantially across provinces in 2005/06, from 31 Some provinces with low rural poverty rates still percent in Central to 74 percent in North Eastern (Figure constitute a large proportion of the rural poor 4.1). Between 2005/06 and 2015/16, the rate of poverty population due to their large population size. For reduction ranged from 23 percentage points in Coast example, while the Rift Valley does not have the to statistically non-significant 3 percentage points in highest rate of rural poverty, due to its large land North Eastern, which remained the province with the size and relatively dense population, the province highest rural poverty rate at 71 percent in 2015/16. accounts for a third of the rural poor population By contrast, Central had the lowest poverty rate at 24 (Figure 4.2). Similarly, a considerable share of the percent in the same period, followed by Eastern at 32 rural poor reside in the Eastern, Western and Nyanza percent and Nyanza at 36 percent. Although Eastern provinces. These three provinces, along with Rift and Nyanza still suffer from poverty levels above the Valley, account for almost 78 percent of the rural poor 2005/06 average, they reduced their poverty rates by population in Kenya. an impressive 20 percentage points and 13 percentage points, respectively, between 2005/06 and 2015/16. Figure 4.1: Rural poverty headcount and its decline by province 80 74 71 71 60 55 52 Percent of population 51 49 50 48 42 42 39 40 36 31 32 24 20 0 Kenya Central Coast Eastern North Eastern Nyanza Rift Valley Western 2005/06 2015/16 Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. 84 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Figure 4.2: Geographic distribution of the Figure 4.3: Share of income from agriculture and non- rural poor in Kenya agricultural sources in rural Kenya Central, Western, 6.7% 8 6 14.9% Coast, 9 8 8.7% 15 21 4 7 Eastern, 13.7% 64 57 Rift Valley, North Eastern, 33.6% 7.0% 2005/06 2015/16 Nyanza, Enterprise income Transfers Services wage 15.5% Industry wage Agriculture Source: Authors’ calculation using KIHBS 2015/16. Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. Note: Agriculture income includes return from crop, livestock, and agriculture wage incomes. On the other hand, the Central and Northeastern Although poverty declined among all rural provinces each account for only 7 percent of the total households, independent of their income source, the rural poor population. Despite the prevalence of a rise in welfare among households with diversified high poverty headcount in Northeastern, the province incomes contributed the most to poverty reduction. accounts for only a small share of the total rural poor Kenya’s rural poverty reduction of 11.7 percentage in the county as it is sparsely populated (Figure 4.3). points between 2005/06 and 2015/16 was mainly The Central and Coast provinces have small rural driven by households that continued to derive their populations and the former also has a relatively lower income from just one source (either agriculture or non- poverty rate. As a result, they account for only a small agricultural activities), contributing 10.4 percentage fraction of Kenya’s rural poor population. points (Table 4.1). The poverty rate fell by a mere 0.8 percentage points for households that changed their 4.2 DIVERSIFYING AWAY FROM source of income (e.g., from exclusively agricultural AGRICULTURE IMPROVES LIVELIHOODS income to mixed or non-agricultural income) in the W hile agriculture remains the main source of same period. While, the remainder 0.5 percentage point income for rural households, the share of drop in the poverty rate was attributed to the interaction income from non-agricultural employment has effect, i.e. resulting from, for instance, a population shift increased significantly in the last decade. As a share into a sector that is greatly contributing to poverty of agricultural household income in rural areas, income reduction. Among the different groups of income from crops and livestock as well as wages declined sources, households with diversified incomes—while from 64 percent in 2005/06 to 57 percent in 2015/16 only representing one-third of the rural population— (Figure 4.3). Wage income from service employment is contributed 40 percent of the 10.4 percentage point the second most important source of income in rural decrease in rural poverty in 2006.16, followed by solely areas, increasing from 15 percent of rural household agricultural households at 31.4 percent and exclusively income in 2005/06 to 21 percent in 2015/16, whereas non-agricultural households at 17.6 percent. the share of wage income in industry increased by a mere 3 percentage points in the same period. The share of rural household income from non-farm enterprises and transfers has remained at basically the same level since 2005/06. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 85 Agriculture and Rural Poverty Table 4.1: Decomposition of poverty by income classification 2006 50.48 Headcount rate 2016 38.76 Source of income Pop. share in period 1 Absolute change Percentage change Non-agricultural income only 18.2 .2.1 17.6 Agriculture income only 48.0 .3.7 31.4 Mixed: agriculture and non-agriculture income 33.8 .4.7 40.0 Total intra-sectoral effect .10.4 89.0 Population shift effect .0.8 6.7 Interaction effect .0.5 4.4 Change in headcount rate .11.7 100.0 Note: Agricultural income includes income from wages from agricultural employment, inferred income from the value of crop sales plus the value of own crop consumption, and income from livestock. A household with only agricultural income is defined as having a share of income of more than 90 percent from agriculture. A household with only non-agricultural Income is defined as having a share of income of less than 10 percent from agriculture. Households with incomes in between are defined as mixed. 4.3 NON-AGRICULTURAL EMPLOYMENT IS activities). As a result, the share of rural households’ BECOMING INCREASINGLY IMPORTANT income from non-agricultural sources increased from FOR RURAL HOUSEHOLDS an average of 35 percent in 2005/06 to 42 percent 4.3.1 Households are allocating more time to non- in 2015/16, with the biggest gains in household agricultural activities income in the provinces of Western (39 percent) and W hile agriculture remains the primary sector of employment for rural households, labor time allocated to non-agricultural activities increased Coast (12 percent). The poverty rate among households that depend between 2005/06 and 2015/16. Rural households in solely on agricultural work is higher compared all provinces, except for Coast, spent an average of less to those engaged in non-agricultural activities. than 45 percent of their labor time on non-agricultural Households engaged in non-agricultural activities on a activities in 2015/16, up from below 40 percent in full-time or part-time basis are less often poor compared 2005/06 (Figure 4.4). By contrast, rural households in to households that focus exclusively on agriculture, the province of Coast allocated an average of only 52 a trend that was already visible in 2005/06 but has percent of their time to non-agricultural activities in strengthened since (Figure 4.6). While it is difficult to 2015/16, up from 40 percent in 2005/06. The increase establish a causal relationship, the strong correlation in labor time spent on non-agricultural activities varied between off-farm diversification and lower poverty between provinces, from an increase of 4 percentage rates is suggestive of the fact that households that points in the province of Nyanza to 15 percentage complement agricultural income with non-agricultural points in the province of Northeastern. Also, there was activities are better prepared to face an adverse virtually no change in the allocation of labor time to agricultural shock such as a drought or low prices, and non-agricultural activities in Nyanza. Compared to smooth consumption. At the same time, households 2005/06, fewer households are exclusively agricultural with higher levels of education are less likely to depend (allocating more than 75 percent of labor to agricultural exclusively on agricultural employment. 86 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Figure 4.4: Changes in rural non-agricultural economic activities a) Non-agricultural labor allocation in rural Kenya 60 53 50 41.5 40 39 Percent of households 40 37 37 35 34 33 31 31 31 31 30 29 28 24 20 10 0 Kenya Central Coast Eastern North Eastern Nyanza Rift Valley Western 2005 2015 b) Distribution of employment by time spent on agricultural and non-agricultural activities 70 63 60 53 50 Percent of households 40 30 20 18 15 13 12 13 12 10 0 [0,0.25] (0.25,0.5] (0.5,0.75] (0.75,1] Proportion of a household's total employment hours spent on agriculture 2005-2006 2015-2016 c) Proportion of income earned from non-agricultural sources 70 62 60 54 50 50 Percent of households 45 42 43 40 38 40 35 37 34 33 35 35 31 33 30 20 10 0 Kenya Central Coast Eastern North Eastern Nyanza Rift Valley Western 2005/06 2015/16 Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. Note: Female employment in non-agricultural activities as a share of total female employment has increased significantly in the Coast (by 20 percentage points) and North Eastern (by 19 percentage points) provinces (Figure 4.5). There has been little change in the remaining provinces since the previous survey was conducted in 2005/06, and the share of female employment in total non-agricultural employment even decreased in Rift Valley, Eastern, and Western. This suggests that most of the increase in non-farm employment has been concentrated among men in Kenya, which can potentially have adverse consequences for intra-household equality. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 87 Agriculture and Rural Poverty Figure 4.5: Female non-agricultural labor allocation Figure 4.6: Rural poverty rate by the proportion of total employment in agriculture 60 70 60 50 48 43 50 40 Poverty rate, % 40 Percent 30 29 29 28 25 27 30 23 23 25 24 23 22 21 20 20 18 20 10 10 0 0 0 0.2 0.4 0.6 0.8 1 Kenya Central Coast Eastern North Nyanza Rift Western Proportion of labor hours spent engaged in agriculture Eastern Valley 2005 2015 2005/06 2015/16 Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. 4.3.2 Wage employment within the service sector has Moreover, wholesale and retail trade is the most increased significantly important non-agricultural industry in terms of Income from wage employment in the services sector employment (Figure 4.8). About 17 percent of rural represents the largest share of non-farm household Kenyan households had at least one family member income in the Kenya (Figure 4.7). While it has increased who worked in wholesale and retail trade in 2015/16, for both poor and non-poor households since 2005/06, a more than threefold increase compared to 2005/06. it constitutes a larger share of the income of non-poor Similarly, employment in transport and communication households. In rural Kenya, the share of wage income also increased threefold, from 2 percent to 6 percent of from the services sector in total household income rural households having one family member employed increased from 15 percent in 2005/06 to 21 percent in the industry in the same period. However, there in 2015/16, reducing the share of agricultural income. was only a slight increase in the employment rate However, agricultural income still remains the most in community, social, and personal services (which important income source for both poor (64 percent) mainly includes public and private sector employment and non-poor (53 percent) households. in education, health, and administration) between Figure 4.7: Share of income from different sources for poor and non-poor households 9 7 7 4 9 8 9 8 11 15 19 24 4 7 4 7 69 64 61 53 Non-poor Poor Non-poor Poor 2005/06 2015/16 Agriculture Industry wage Services wage Transfers Enterprise income Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. 88 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Figure 4.8: Non-farm economic activity by ISIC classification a) Participation of households in non-farm employment by industry b) Proportion of salaried non-farm income by industry 20 19 50 50 Percent of total salaried employment income, % 17 40 15 Percent of households 13 34 33 30 10 7 20 18 6 6 5 12 12 5 10 10 3 3 10 9 10 2 0 0 Mining & Construction Wholesale & Transport, Community, Mining and Construction Wholesale and Transport, Other service Manufacturing Retail Trade Storage and Social and manufacaturing retail trade+food storage & activities Communication Personal Services Service activities communication 2005/06 2015/16 2005/06 2015/16 Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. 2005/06 and 2015/16. Also, employment rates rose in groundnut varieties increased household crop income the mining and manufacturing, construction, trade, by US$130.$254 and improved the chance of a and transportation and communication industries household escaping poverty by 7.9 percentage points during the same period, even though their individual (Kassie, Shiferaw and Muricho, 2011).112 Gains were shares have remained relatively low.111 greater for households with a relatively smaller farm size and for more educated households. Finally, a quasi- 4.4 FARM PRODUCTIVITY HAS STAGNATED experimental study by Davis et al. (2012) demonstrated WHILE COMMODITY PRICES HAVE the importance of learning about improved farming INCREASED practices among small-scale farmers in East Africa. 4.4.1 Higher productivity is associated with lower It showed that farmer field schools contributed poverty rates to increased crop productivity, resulting in higher S everal studies of African countries show a causal link between improved agricultural productivity and reduced poverty rates. A meso-level study of household income and an improvement in farmer welfare. While the productivity and income of female-headed households increased significantly, village-level data in Madagascar shows that communes they increased only marginally for male- that adopted agricultural technologies at a higher rate, headed households. Moreover, the effects were and subsequently had higher crop yields, enjoyed lower concentrated among households with little formal food prices, had higher real wages for unskilled workers, education, presumable because these households and exhibited better welfare indicators (especially lower had the most to gain from such training programs. extreme poverty rates, Minten and Barrett 2008). This This section presents some indicative evidence suggests that an increase in agricultural productivity that this causal relation between agricultural can raise incomes for surplus farmers, reduce prices for productivity and poverty reduction. However, it consumers, and increase employment opportunities should be noted that they represent correlations, and wages for unskilled workers. Similarly, another not necessarily causal, between higher crop yield study in Uganda found that adopting improved and increased household welfare. 111 KIHBS 2005/06 uses ISIC Revision 2 to classify employment by subsector, whereas KIHBS 2015/16 uses ISIC Revision 4. Appropriate steps have 112 The authors of the studies eliminated selection bias on observable been taken to ensure correspondence of industrial classification. differences between adopters and non-adopters. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 89 Agriculture and Rural Poverty Figure 4.9: Relationship between crop yield and poverty rates at the provincial level in rural Kenya, 2015/16 a) Maize yield and poverty b) Bean yield and poverty 2300 600 Rift Valley Eastern 2100 550 1900 Rift Valley Yield (kg/hectare) Yield (kg/hectare) 1700 Western 500 1500 Central 450 1300 Nyanza Central Eastern 1100 400 Nyanza North Western 900 Eastern 350 Coast North Coast 700 Eastern 500 300 15% 25% 35% 45% 55% 65% 15% 25% 35% 45% 55% Poverty Rate Poverty Rate Source: Authors’ calculation using KHIBS 2015/16. Figure 4.10: Poverty and crop yield at the county level in rural Kenya, 2015/16 a) Maize yield and poverty b) Bean yield and poverty 3500 1400 3000 1200 2500 1000 Yield (kg/hectare) Yield (kg/hectare) 2000 800 1500 600 1000 400 500 200 0 0 0% 10% 20% 30% 40% 50% 60% 70% 0% 10% 20% 30% 40% 50% 60% Poverty Rate Poverty Rate Central Coast Eastern Nyanza Central Eastern Nyanza Rift Valley Western Linear (Kenya) Rift Valley Western Linear (Kenya) Source: Authors’ calculation using KIHBS 2005/06 and KIHBS 2015/16. Figure 4.11: Relationship between yield decile and poverty rates in rural Kenya, 2015/16 a) Maize b) Beans 60% 55% 50% 45% 40% 35% Poverty Rate Poverty Rate 30% 25% 20% 15% 10% 5% 1 2 3 4 5 6 7 8 9 10 1 2 3 4 5 6 7 8 9 10 Yield decile Yield decile Central Eastern Nyanza Rift Valley Western Central Eastern Nyanza Rift Valley Western Source: Authors’ calculation using KIHBS 2015/16. 90 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty The evidence suggests that an improvement in farm instance, the poverty rate among households in the productivity could potentially reduce poverty in lowest maize yield decile is 49 percent, compared to Kenya. While agriculture was not the main driver of only 22 percent for those in the highest (10th). poverty reduction in rural Kenya between 2005/06 and 2015/16, an increase in crop yield could significantly 4.4.2 Stagnating productivity means that there is an reduce poverty, as agricultural productivity is strongly unmet potential for rural farmers and negatively correlated with poverty rates at the Almost 85 percent of Kenya’s cultivated land was provincial, county, and household level. Provinces with devoted to growing maize and beans in 2015/16. higher maize and bean yields have generally lower Bean production increased significantly in cultivated poverty rates (Figure 4.9). Similarly, a comparison of areas: from 27 percent of total crop areas in 2005/06 to counties within a given province shows that counties 37 percent in 2015/16 (Figure 4.12). However, there were with higher farm productivity have much lower poverty only minor changes in the share of land allocation for rates (Figure 4.10). all other crop categories. Approximately half of Kenya’s total crop area was devoted to maize production for Most farm households with high crop yields appear both years. The remainder of this section will focus on to have escaped poverty in Kenya. In each Kenyan maize and bean yields, the two most commonly grown province, households in a higher yield decile tend to staple crops in Kenya. have lower poverty rates (Figure 4.11). In Rift Valley, for Figure 4.12: Proportion of cultivated area by crop category in rural Kenya a) KIHBS 2005/06 b) KIHBS 2015/16 2%2% 3% 4% 3% 2% 4% 5% 37% 48% 51% 27% 6% 6% Maize & cereal Tubers & roots Maize & cereal Tubers & roots Beans, legumes & nuts Fruits & vegetables Beans, legumes & nuts Fruits & vegetables Tea & co ee Other cash crops Tea & co ee Other cash crops Other crops Other crops Source: Authors’ calculation using KHIBS 2015/16. Figure 4.13: Maize and bean yield in selected African countries a) Maize yields in selected African countries, 2005–16 6000 5000 4000 Kg/hectare 3000 2000 1000 0 2005 2016 Burundi Kenya Malawi Rwanda South Africa Uganda Tanzania Ethiopia KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 91 Agriculture and Rural Poverty b) Bean yield in selected African countries (2005-2016) c) Bean yield in Kenya since 2000 1800 1,668 1,667 700 1,589 1,530 1600 1,487 1,471 650 1,434 1,416 1,401 1,382 1,347 1400 1262 600 1199.9 1,225 1,185 1,169 1,147 1200 1,069 1,043 550 1,035 1,002 1,001 1,000 Kg/hectare Kg/hectare 997 947.7 968 967 1000 916 891 500 877 846 791 780 762 800 718 700 450 659 622 588 615 585 567 557 600 534 508 484 400 413 370 400 350 200 300 0 1995 2000 2005 2010 2015 2020 Kenya South Africa Tanzania Ethiopia 2000-2009 2010-2016 Source: Author’s calculation based on FAO data. In recent years, Kenya’s agricultural productivity Tegemeo panel household data survey, collected has been low and stagnant compared to that of between 2000 and 2010,113 and both waves of the neighboring countries, except for bean crops. Since KIHBS household data (Figure 4.14). In contrast, bean 2005, maize yields in Kenya have been stagnant at a yield increased by approximately 50 percent between relatively low level compared to many of its neighbors, 2010 and 2016, according to the FAO (Figure 4.13).114 according to cross-country yield data from the Food and Agriculture Organization (FAO) (Figure 4.13). Other There are important differences in yield levels across countries, such as Ethiopia, Malawi, Rwanda, and provinces. Maize yield is multiple times higher in Rift Uganda, have experienced varying levels of productivity Valley than in North Eastern, the latter which has a growth. The level of maize yield in South Africa, which high and persistent poverty rate (Figure 4.14). Maize is indicative of capital- and input-intensive farms, yield is also low in Coast, which is likely explained by illustrates the tremendous potential for Kenyan farmers the high share of non-agricultural employment in the to increase their crop productivity and raise their living province. By contrast, heterogeneity of bean yield is standards. The stagnation in maize productivity over less pronounced, with Eastern and Rift Valley provinces the period 2005-2016 seems to be confirmed by the having relatively higher yields than Kenya’s other provinces. Figure 4.14: Heterogeneity in crop productivity across provinces in rural Kenya a) Maize a) Beans 2500 1000 2,268 2,150 648 2000 1,738 1,683 750 1,532 1,493 Yield (kg/hectare) Yield (kg/hectare) 1,532 1,474 577571 1500 559 1,301 502 535 531 1,260 1,215 485 1,191 885 500 435 383 346 353 388408 383 380 1000 780 767 594 250 500 0 0 Kenya Central Coast Eastern North Nyanza Rift Western Kenya Central Coast Eastern North Nyanza Rift Western Eastern Valley Eastern Valley 2005/06 2015/16 2005/06 2015/16 Source: Authors’ calculation using KHIBS 2005/06 and KIHBS 2015/16. 113 Tegemeo data are of households in some parts of Kenya and not in the entire country. 114 This trend in however not observed in the KIHBS data. 92 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Nationally, female headed households have lower while in bean cultivation, this difference amounts to productivity in both beans and maize crops (Figure over 15 percent. However, there is heterogeneity across 4.15). Female headed households have 10 percent lower provinces, with statistically insignificant differences maize yields compared to male headed households, observed in maize cultivation in the Rift Valley, Eastern, Figure 4.15 : Heterogeneity in crop productivity by gender of household head a) Maize a) Beans 2500 700 2,189 2,144 582 2,005 600 544 546 2000 1,927 519 469 1,325 500 455 1,645 444 381 429 1,490 1,439 382 1500 1,383 368 384 377 Kg/hectare Kg/hectare 1,282 1,250 1,300 400 328 1,219 300 1000 700 590 200 421 500 100 0 0 Kenya Central Coast Eastern North Nyanza Rift Western Kenya Central Coast Eastern Nyanza Rift Western Eastern Valley Valley Male Female Male Female Source: Authors’ calculation using KHIBS 2005/06 and KIHBS 2015/16. Figure 4.16: Gender differences in input use in rural Kenya a) Land area cultivated by households b) Total input costs (excluding labor) per acre 2.9 11,273 3 12,000 2.4 2.3 2.5 2.1 10,000 1.9 2.0 1.8 7,591 2 8,000 1.6 6,040 1.4 1.4 5,554 KSh./acre 5,218 Acres 1.5 1.2 1.2 5,262 6,000 4,403 1.1 1.1 4,318 4,317 3,587 3,043 1 4,000 2,516 2,893 1,886 0.5 2.000 0 0 Kenya Central Coast Eastern Nyanza Rift Western Kenya Central Coast Eastern Nyanza Rift Western Valley Valley Male Female Male Female c) Inorganic fertilizer spent per acre d) Labor costs per acre 8000 6,961 10000 7,500 9000 7000 5,392 8000 6000 7000 5,675 3,976 5000 5,591 KSh./acre 6000 KSh./acre 3,380 3,425 3,546 4000 3,038 5000 2,748 3,967 2,435 4000 3,463 3000 2,095 3,048 1,726 3,070 1,615 2,751 1,353 3000 2,446 2000 1,719 2,114 1,596 1,907 2000 367 314 1,307 1,286 1000 1000 0 0 Kenya Central Coast Eastern Nyanza Rift Western Kenya Central Coast Eastern North Nyanza Rift Western Valley Eastern Valley Male Female Male Female ource: Authors’ calculation using KHIBS KIHBS 2015/16. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 93 Agriculture and Rural Poverty and Western provinces and in bean cultivation in D). More specifically, to investigate the determinants of Western and Eastern provinces. The differences in crop yield, we apply a fixed effects model. In this model, productivity are partly explained by differences in we start with a basic specification where logarithm of the use of yield enhancing inputs. Farm households per acre yield is regressed on fixed effects of household headed by women use inputs less intensively than and a vector of household characteristics. male-headed households, as they spend less on yield- enhancing inputs such as inorganic fertilizer (Figure Technology adoption is the main factor associated 4.16). While these households also have slightly lower with improvements in maize yield. Households that labor costs, reflecting lower labor inputs, differences are applied chemical fertilizer, for example, experienced a only statistically significant in Eastern. 20.25 percent increase in maize yield. Moreover, farmers who planted improved maize seeds experienced 4.4.3 Improved technologies are the key drivers of 26.32 percent higher productivity compared to agricultural productivity those that used traditional low-yield seeds. However, The adoption of improved farming technologies farmers who used both chemical fertilizer and planted and practices can increase agricultural productivity improved maize seeds did not appear to have higher and reduce rural poverty among small farmers. This maize yield relative to those who applied these section examines what factor are associated with high inputs individually. While the application of chemical crop productivity at the household level in Kenya using fertilizer is positively associated with higher bean the Tegemeo panel dataset115 for 2000-10 (see Appendix yield, the yield increase is negligible. Figure 4.17: Trends in input use by farmers (Tegemeo Panel) a) While the share of farm households that apply chemical fertilizer b) The share of households that use improved seeds for maize increased is high, it has increased only moderately since 2000 in the most recent survey round 100 100 79.1 80 76 75 80 70 72 67.6 65.7 68.1 60 60 Percent Percent 40 40 20 20 0 0 2000 2004 2007 2010 2000 2004 2007 2010 Source: Author’s calculation based on Tegemeo Panel Household Survey (2000-2010). 115 The Tegemeo Rural Household Panel Data cover the years 2000, 2004, 2007, and 2010. The data were collected in 22 rural districts across the country. Stratified simple random sampling was used to create the sample of households. After assigning agro-ecological zones (AEZ) to each rural division, 2-3 divisions were selected in each AEZ based on their population size. Villages within selected divisions and households within selected villages were picked through a blind equal chance ballot. A total of 1,446 sampled households were interviewed in 2000, 1,397 in 2004, 1,342 in 2007, and 1,304 in 2010. The rate of household attrition was 9.8 percent between 2010 and 2000. Households that were overlooked during the interview process were not replaced and efforts were made to interview them in subsequent surveys. 94 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Despite the yield-enhancing effects of fertilizer, There is a positive relationship between the adoption the share of households that applied chemical of improved seeds and maize yield. The application fertilizer did not increase much between 2000 and of certified seeds is strongly associated with maize 2010. In the Tegemeo panel, which only covers parts productivity. However, the opposite is true for bean of Kenya, more than 70 percent of farmers apply productivity, a result attributable to the small number fertilizer on their maize plots. However, the share of farmers that use certified seeds for beans (less than 10 percent) compared to maize (close to 70 percent). of farmers that use fertilizer has not changed much since 2000 (Figure 4.17a). By contrast, the share of An analysis of the relationship between crop yield farmers that use improved seed varieties increased and plot size shows that, even after controlling by more than 10 percent between 2000 and 2010, to for technology adoption and other household almost 80 percent of maize farmers in 2010 (Figure characteristics, smallholder farmers are more 4.17b). It is worth noting however, there is very productive than large farmers. Columns 2 and 3 of limited use of improved seeds for other crops. Table 4.2 show the relative productivity of maize Table 4.2: Determinants of maize yield, FEs model, 2000–10 (1) (2) (3) 0.21 *** 0.20 *** Fertilizer adoption (Yes=1) (3.77) (4.16) 0.26*** 0.28*** Improved seed adoption (Yes=1) (6.30) (6.89) .0.00 0.00 0.00 Distance to extension services (.0.21) (0.05) (0.90) 0.05 0.07* 0.06 Cooperative/Group membership (1=yes) (1.32) (1.89) (1.63) Cropped land quartile (the lowest quartile is the reference group): 0.00 0.00 .0.17*** .0.19*** 2nd quartile (.3.80) (.4.22) .0.38 *** .0.41*** 3rd quartile (.9.62) (.9.91) .0.69 *** .0.68*** 4th quartile (.14.20) (.12.93) Effectiveness of fertilizer on improved seed 0.00 No improved seed x Fertilizer used (0.83) 0.00*** Improved seed x Fertilizer used (3.25) 5.64*** 5.79*** 6.43*** Constant (18.07) (19.08) (18.44) Observations 4897 4897 3996 Source: Author’s calculation based on Tegemeo Panel Household Survey (2000-2010). Note: Standard errors in parentheses: * p < 0.1, ** p < 0.05, *** p < 0.01. Note that the dependent variable is logarithm of yield (kg/acre). A vector of household characteristics including: gender, age, age squared and education of household head, household size and dependency ratio. KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead 95 Agriculture and Rural Poverty farmers with plot size in the upper three quartiles and initiatives can help bridge the productivity gap. An compared to those with plots in the lowest quartile. increase in agricultural productivity, as demonstrated The production of large maize farmers in the highest in the previous section, could significantly reduce landholding quartile is 69 percent lower per acre poverty among farm households. That is why the compared to those in the lowest quartile. This inverse announcement of having food security and agricultural relationship between plot size and maize yield is productivity as one of the main four priorities (the Big 4) persistent, and the productivity gap increases from of the GoK is welcome news. 17 percent to 38 percent and 69 percent as plot size quartile ranking increases from 2nd to 3rd and 4th Policymakers may need to allocate more resources quartile, respectively. Similarly, small bean farmers are to enhance farmers’ productivity and make sure more productive than large bean farmers (see Table D.1 that the current spending is efficient and providing in Appendix D). the highest returns. The recently published Kenya Economic Updates noted that only 2 percent of total The inverse relationship between plot size and public expenditure was allocated to agriculture in maize yield is not unique to rural Kenya, as it has 2016/17, even though the sector accounts for 25 been observed in several developing countries and percent and 60 percent of the country’s GDP and confirmed by various studies.116 The relationship employment, respectively. This prevents the country’s is contrary to economic theory, which states that from investing effectively in smallholder agriculture factor productivity must be equal across farms, as and provide services to improve basic crop yield such land would be sold or leased from farmers with lower as extension services, improved seeds and seedlings, marginal productivity to farmers with higher marginal irrigation, etc. There is also a need to assets if the productivity.117 Some of the most common and current spending is efficient, taking into account that plausible explanations for this inverse relationship spending on public goods in this context (e.g. research relate to market imperfections. First, smallholder and development, extension services, etc.) has been farmers face an imperfect labor market and continue proven to be more productive than spending on to excessively use labor on their small plots. Second, private goods (e.g. fertilizer subsidies). In addition an imperfect insurance and crop market forces risk- there is space to reform the input subsidy program by averse small farmers to work more hours than optimal ensuring that the program is targeting small farmers to secure enough food from their plots.118 and facilitating technology adoption among them. Moreover, investment in irrigation schemes have a 4.4.4 Policies that promote investments in high rate of return119 and could reduce dependence productivity-enhancing technologies are vital for farmers on rainfall. Currently, only 2 percent of Kenya’s total arable land is irrigated, compared to 6 percent in Investment in productivity-enhancing technologies Sub-Saharan Africa, and most of the country’s crop such as fertilizer, improved seeds, and agricultural production is rainfed. extension services, as well as irrigation, is critical to increase the productivity and welfare of Kenya’s 4.4.5 An increase in grain prices since 2005 may have farmers. There is a huge potential to facilitate poverty helped reduce rural poverty reduction through increase in agriculture income (crop In the absence of major crop-enhancing productivity and wage income). The fact that Kenya’s current level investments, higher crop prices can reduce the of crop productivity is lower compared to that of its poverty rate among farm households. Due to a lack of neighboring countries, signals that public investment farmgate price data to analyze changes in crop prices, market price data are used as a proxy. An analysis of Barrett et al. (2010) summarizes a list of studies, including Chayanov 116 market price data reveals that crop prices had been (1926) and Sen (1962), that have noted this inverse relationship. In addition, a recent study by Ali and Deininger (2015) also found similar increasing at a similar rate as general prices through results in Rwanda. the period 2005 to 2011. Figure 4.18 shows the nominal Barrett et al. 2010. 117 Barrett et al. 2010; Ali and Deininger 2015. 118 119 World Bank 2018b. 96 KENYA POVERTY AND GENDER ASSESSMENT 2015/16 · Reflecting on a Decade of Progress and the Road Ahead Agriculture and Rural Poverty Figure 4.18: Trends of crop prices and overall prices a) Maize prices, 2006–16 6,000 5,000 4,000 3,000 2,000 1,000 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 Nyanza Coast Nairobi Rift Valley KIHBS implied price change* *Price increase of the average price across all provinces in the first year (2006) using price inflator implied by the numerical change in the poverty line in KIHBS between 2005 and 2015. b) Bean prices, 2006–16 10,000 9,000 8,000 7,000 6,000 5.000 4.000 3.000 2,000 1,000 0 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 * Price increase of the average price across all provinces in the rst year (2006) using price in ator Nyanza Coast * Price increase of the average price across all provinces in the first year (2006) using price inflator Source: Author’s calculation using FEWS NET data. price and the estimated trend if crop prices would have 4.5 INCREASED MARKET PARTICIPATION followed the overall inflation pattern.120 Data show CAN FURTHER REDUCE RURAL POVERTY I that maize and bean prices were significantly above mproving access to markets for rural households has their estimated trends in 2011-15, which coincides been a key policy goal for Kenyan policymakers. with the period directly prior to KIHBS 2015/16. As Easier access to markets allows rural households to higher crop prices generally tend to benefit rural areas improve their productivity by facilitating access to (which produce crops) at the expense of urban areas agricultural i