Cash transfers and human capital development: Evidence, gaps and potential Sudhanshu Handa on behalf of the Transfer Project

Cash transfers and human capital development: Evidence, gaps and potential Sudhanshu Handa on behalf of the Transfer Project UNICEF Office of Research-Innocenti and UNC Presented at the Transfer Project Annual Workshop: Addis Ababa, April 2016

Cash Transfer programs in sub-saharan Africa: The quiet revolution No CTs After 2004 Prior to 2004 No data Transfer Project 2

Households covered andpercent of population: Government programs 1200000 Not included (due to scale): CSG in South Africa (>11 million recipients) 10.6% 2.0% 5.2% 1100000 1125000 1000000 800000 600000 400000 200000 0.2% 2.0% 3.2% 2.5% 34.8% 4.6% 64000 69000 80000 21.5% 5.4% 46.8% 150000 163000 170000 182000 190000 250000 5.1% 310000 4.6% 455000 0 Percent of population are estimates assuming average household size and one beneficiary per household (child grant and pension designs)

Key features of the African Model Programs tend to be unconditional (or with soft conditions), with exception of Tanzania (conditional on schooling, health) Targeting is based on poverty and vulnerability (OVC, laborconstraints, elderly) Important community involvement in targeting process Payments tend to be manual ( pulling beneficiaries to paypoints) Opportunity to deliver complementary services

Density 0.01.02.03.04 Density Density 0.01.02.03.04 0.02.04.06 Density Unique demographic structure of recipient households: Missing prime-ages Zambia SCT (Monze) Kenya CT-OVC 0 20 40 60 80 Age (years) Malawi SCT 0 20 40 60 80 100 Age (years) Zimbabwe HSCT Malawi SCTP 0 20 40 60 80 Age (years) 0.01.02.03.04 Zimbabwe HCST 0 20 40 60 80 100 Age (years)

Density 0.01.02.03.04.05.06 Density 0.01.02.03.04.05.06 Labor-constrained criterion selects unique households: Example from Zambia 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95100105 Age (years) 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 100 Age (years) Zambia SCT Households Rural Ultra-Poor LCMS 2010

Who gets the cash? Approximately two-thirds of beneficiaries are female Program Female beneficiaries (%) Femaleheaded households (%) Ghana LEAP 44 60 Ghana LEAP 1000 100 11 Kenya CT-OVC 85 85 And three of five beneficiary HH are female-headed Malawi SCTP 84 84 Zambia CGP 99 - Zambia MCTG 75 - Zimbabwe HSCT 68 68 Figures for female-headed households may reflect evaluation sample, rather than beneficiary sample. Zambia studies did not collect information on headship.

% or per capita consumption How much do programs pay? Transfer as share of beneficiary pre-program consumption 40 35 30 A critical threshold 25 20 15 10 5 0

Overview of programs & evaluations connected with Transfer Project Country (program) Targeting (in addition to poverty) Sample size (HH) Methodology LEWIE Youth Years of data collection Ghana (LEAP) Elderly, disabled or OVC 1,614 Longitudinal PSM X 2010, 2012, 2016 Ghana (LEAP 1000) Pregnant women, child<2 2,500 RDD 2015, 2017 Ethiopia (SCTP) Labour-constrained 3,351 Longitudinal PSM X 2012, 2013, 2014 Kenya (CT-OVC) OVC 1,913 RCT X X 2007, 2009, 2011 Lesotho (CGP) OVC 1,486 RCT X 2011, 2013 Malawi (SCTP) Labour-constrained 3,500 RCT X X 2011, 2013, 2015 South Africa (CSG) Child <18 2,964 Longitudinal PSM X 2010, 2011 Tanzania (PSSN) Food poor 801 RCT X 2015, 2017 Zambia (CGP) Child 0-5 2,519 RCT X 2010, 2012, 2013, 2014 Zambia (MCTG) Zimbabwe (HSCT) Female, elderly, disabled, OVC Food poor, labourconstrained 3,078 RCT X 2011, 2013, 2014 3,063 Longitudinal matched casecontrol X X 2013, 2014, 2016

Cash Transfer How does cash affect the household and its members? Moderators Distance/quality of facilities Prices Shocks Infrastructure Services Norms LEVEL 2 LEVEL 1 Young Child Household Consumption Nutrition Illness Food Security Material well-being Investment Crop production Livestock Assets Income Older Child Schooling Material well-being Work HIV risk Mediators Future expectations Attitudes towards risk Information Time-use Use of services Caring practices Labor Income Mental health Adult Care-giver Self-assessed welfare Health

Summary of results based on 7 rigorous impact evaluations Domain of impact Food security, extreme poverty Alcohol & Tobacco Subjective well-being Secondary school enrollment Spending on school inputs (uniforms, shoes, clothes) Health Spending on health Nutritional status Increased fertility Evidence

Reductions on poverty measures 14.0 12.0 10.0 10 10 12 11 10 8.0 6.0 4.0 2.0 1.4 0.0 Ethiopia (24 months) Kenya (24 months) Lesotho (24 months) Malawi (36months) Zambia MCTG (36 months) Zambia CGP (48 months) 0 Zimbabwe (12 months) Increase in household income (%) Reduction in poverty head count (pp) Reduction in poverty gap (pp) Solid bars represent significant impact, shaded insignificant. Impacts are measured in percentage points, unless otherwise specified

Across-the-board impacts on food security Spending on food & quantities consumed Ethiopia SCTP Ghana LEAP Kenya CT-OVC Lesotho CGP Malawi SCTP Zambia MCTG Zambia CGP Per capita food expenditures Per capita expenditure, food items Kilocalories per capita Frequency & diversity of food consumption Number of meals per day Dietary diversity/nutrient rich food Food consumption behaviours Coping strategies adults/children Food insecurity access scale Green check marks represent significant impact, black are insignificant and empty is indicator not collected Zim HSCT

No evidence cash is wasted on alcohol & tobacco Alcohol/tobacco represent 1% of budget share Across 7 countries, no positive impacts found on alcohol/tobacco: Data comes from detailed consumption modules covering over 250 individual items In Lesotho negative impacts on alcohol consumption (possible decrease through decrease in poverty-related stress?) Alternative measurement approaches yield same result: Has alcohol consumption increased in this community over the last year? Is alcohol consumption a problem in your community?

Beneficiaries are happier too: Consistent impacts on subjective well-being 45% 40% 41% ** 35% 30% 25% 20% 20% ** 15% 10% 5% 9% *** 6% ** 12% ** 0% Ghana LEAP (happiness) Zambia CGP (happiness, women) Malawi SCT (QoL score) Kenya CT-OVC (QoL score) Zimbabwe HSCT (SWB scale) Impacts are percentage changes, countries not shown did not collect data on subjective well-being

What about the kids?

School enrollment impacts (secondary age children): Equal to those from CCTs in Latin America 20 18 16 14 12 10 8 6 4 2 0 8 3 7 8 15 8 9 12 6 9 6 10 Primary enrollment already high, impacts at secondary level. Ethiopia is all children age 6-16. Bars represent percentage point impacts; all impact are significant.

Grade 3 math test Serenje District, Zambia More kids in school but school quality still a challenge

Significant impacts on spending on school-age children (uniforms, children s shoes and clothing) 35 30 25 26 30 23 32 20 15 10 5 11 11 5 0 Ghana (LEAP) Lesotho (CGP) Malawi (SCTP) Zambia (MCTG) Zambia (CGP) Zim (HSCT) small hh Zim (HSCT) large hh Solid bars represent significant impact, shaded insignificant. Impacts are measured in percentage points; Lesotho includes shoes and school uniforms only, Ghana is schooling expenditures for ages 13-17. Other countries are shoes, change of clothes, blanket ages 5-17.

Young child health and morbidity Regular impacts on morbidity, but less consistency on care seeking Ghana LEAP Kenya CT-OVC Lesotho CGP Malawi SCTP Zambia CGP Zimbabwe HSCT Proportion of children who suffered from an illness/frequency of illnesses Preventive care Curative care Enrollment into the National Health Insurance Scheme Vitamin A supplementation Supply of services typically much lower than for education sector. More consistent impacts on health expenditure (increases) Green check marks represent positive protective impacts, black are insignificant and red is risk factor impact. Empty is indicator not collected

Budget shares and expenditure impacts on health 10 9 8.7 8 7 6 5 4 3 2 5.8 2.1 3 1.4 5 3.9 1 0 Ghana (LEAP) Kenya (CT- OVC) Lesotho (CGP) Malawi (SCTP) Zambia (CGP) Zambia (MCTG) Zimbabwe (HSCT) NS Ethiopia Solid bars represent significant impact, shaded insignificant. Impacts are measured in percentage points (top figure). Bars represent % of budget share at baseline - Malawi figures represent treatment means.

No impacts on young child nutritional status (anthropometry) Evidence based on Kenya CT-OVC, South Africa CSG, Zambia CGP, Malawi SCTP, Zimbabwe HSCT However, Zambia CGP 13pp increase in IYCF 6-24 months Some heterogeneous impacts If mother has higher education (Zambia CGP and South Africa CSG) or if protected water source in home (Zambia CGP) Possible explanations Determinants of nutrition complex, involve care, sanitation, water, disease environment and food Weak health infrastructure in deep rural areas Few children 0-59 months in typical OVC or labor-constrained household

percentage point impact Impacts on number of children, ages 0-1, in the household, Malawi, Kenya No fertility incentives! Total children 0-1 year - Incident risk ratio 50 40 1.300 30 1.200 20 10 1.100 0-10 1.000-20 0.900-30 -40 0.800-50 -60 All Girls Boys All Girls Boys Malawi SCTP Kenya CT-OVC 0.700 All Male Female Zambia Malawi & Kenya: DD Probit models predicting Pr(child aged 0-1 in household) Zambia: DD Poisson models estimating number of children 0-1 years in household

Congo, Democratic Republic Zimbabwe Burundi Liberia Eritrea Niger Malawi Central African Republic Madagascar Mali Togo Guinea South Sudan Mozambique Guinea-Bissau Comoros Ethiopia Sierra Leone Burkina Faso Uganda Rwanda Benin Tanzania, United Republic of Zambia Côte d'ivoire Kenya The Gambia Senegal Mauritania Sao Tome and Principe Lesotho Cameroon Chad Sudan Djibouti Nigeria Ghana Cape Verde Congo Brazzaville Swaziland Angola Namibia South Africa Mauritius Botswana Gabon Seychelles Equatorial Guinea Social cash transfer expenditure estimates Scaled up cash transfers are affordable in SSA Plausible simulations show average cost 1.1% of GDP or 4.4% of spending 20% 15% 10% 5% 0% In % of general government total expenditure In % of GDP

Emerging evidence that effect of cash larger depend on supply side factors Example 1: Skilled attendance at birth improved in Zambia CGP, only among women with access to quality maternal health services Example 2: Anthropometry in Zambia CGP improved among households with access to safe water source Example 3: Impacts on schooling enrollment in Kenya CT-OVC are largest among households which face higher out of pocket costs (uniform/shoes requirement, greater distance to school) [program offsets supply side barrier]

What determines type and size of impacts? Predictability of transfers (Allows planning, consumption smoothing) Size of transfer and protection from inflation (Rule of thumb of 20% of mean consumption of target population) Context (Supply of health and education, user fees) Who you target (Labor-constrained; households with more adolescents/ovc and fewer pre-school children)

Evidence, potential, gaps Evidence: Cash transfers are protective they work Potential: Programs are affordable, can contribute to inclusive growth strategy Gaps: Health and nutrition effects on 0-5 years inconsistent Few households with young children targeted are reached under current approaches Health infrastructure not as well developed as schooling, attitudes and other factors at play in demand for health

Summary of results based on 7 rigorous impact evaluations Domain of impact Food security, extreme poverty Alcohol & Tobacco Subjective well-being Secondary school enrollment Spending on school inputs (uniforms, shoes, clothes) Health Spending on health Nutritional status Increased fertility Evidence

For more information Transfer Project website: www.cpc.unc.edu/projects/transfer Briefs: http://www.cpc.unc.edu/projects/transfer/publications/briefs Facebook: https://www.facebook.com/transferproject Twitter: @TransferProjct @ashudirect Email: Ashu Handa, shanda@unicef.org Photo credits: Ghana LEAP 1000 coverphoto, Ivan Grifi (2015) Kids in Malawi, slide 16, Darlen Dzimwe (2014) Math test Zambia MCTG, slide 18, Ashu Handa (2013)

Acknowledgements Transfer Project is a multi-organizational initiative of the United Nations Children s Fund (UNICEF) the UN Food and Agriculture Organization (FAO), Save the Children-United Kingdom (SC-UK), and the University of North Carolina at Chapel Hill (UNC-CH) in collaboration with national governments, and other national and international researchers. Current core funding for the Transfer Project comes from the Swedish International Development Cooperation Agency (Sida) to UNICEF Office of Research, as well as from staff time provided by UNICEF, FAO, SC-UK and UNC-CH. Evaluation design, implementations and analysis are all funded in country by government and development partners. Top-up funds for extra survey rounds have been provided by: 3IE - International Initiative for Impact Evaluation (Ghana, Malawi, Zimbabwe); DFID - UK Department of International Development (Ghana, Lesotho, Ethiopia, Malawi, Kenya, Zambia, Zimbabwe); EU - European Union (Lesotho, Malawi, Zimbabwe); Irish Aid (Malawi, Zambia); KfW Development Bank (Malawi); NIH - The United States National Institute of Health (Kenya); Sida (Zimbabwe); and the SDC - Swiss Development Cooperation (Zimbabwe); USAID United States Agency for International Development (Ghana, Malawi); US Department of Labor (Malawi, Zambia). The body of research here has benefited from the intellectual input of a large number of individuals. For full research teams by country, see: https://transfer.cpc.unc.edu/