Impact of Ultra-Poor Graduation Pilots: Early Results from Randomized Evaluations Bram Thuysbaert Yale University and IPA November 15, 2012 With A.Banerjee, E. Duflo, D. Karlan, J. Morduch, and others Preliminary and Incomplete
Pilots in 10 Sites Pilot sponsored by CGAP and Ford Foundation: Ghana, Haiti, Honduras, India(3), Ethiopia, Pakistan, Peru, Yemen Randomized evaluations at 8 sites ( )
Who are the Ultra Poor? A few facts Most had at least 2 meals previous day India: 62% Pakistan: 75% Yemen: 87.5% Ethiopia: 83% But most also faced food shortages in at least some months in the past year Large share of expenses spent on food and fuel: Pakistan: 54% India: 72% Peru: 70% Yemen: 69% Pakistan Yemen In past 12 months, did HH members get enough food?
Who are the Ultra Poor? Vulnerable to shocks India: 29% of selected HH experience 1 or more health shock in past year Pakistan: Medical expenses 12% of HH budget Remote areas More than 1 hour to nearest health center Peru: 24-30% Honduras: 28-58%
Fundamental research question What is the impact of the Graduation model on the ultra poor? Impact evaluation measures: How have the lives of households changed compared to how they would have changed in the absence of the program? Note this is different from “How have their lives changed?” The control group gives us the counterfactual Random treatment assignment ensures we compare similar types of households
Data collection Baseline Consumption surveys 2 nd follow-up survey 1st follow-up survey Collect data on eligible households Many sites have quarterly surveys to monitor changes Survey of same baseline households after 2 years (end of program support) 1 year after 1 st follow-up survey to track more long-term impacts
Sample size Many 100’s of households in a few tens of villages Survey modules Health indicators Education Consumption Income and activities Assets Credit Risk preferences Ongoing qualitative research Data collection
Asset transfers Honduras Pakistan Ethiopia
Preliminary Results
Results from 2 nd follow-up survey: Consumption o 7% increase in food consumption o 20% increase in non-food consumption Improved food security o 14 percentage points increase in share of households that report having enough food every day India – West Bengal (Bandhan) Researchers: Banerjee, Duflo et al.
Livestock holdings o Large effect on animal holdings (years after asset transfer!) Income generation o Increased income from livestock o Increase in profits from micro-enterprises Household durables o Significant increase in HH durables index India – West Bengal (Bandhan)
Savings and borrowing o Suggestive evidence of increase in savings deposited in last 30 days (not statistically significant) o 52.9% increase in total borrowing Health o Increase in health knowledge o Large gains in mental health at 1 st follow-up but no longer at 2 nd follow-up India – West Bengal (Bandhan)
Results from 2 nd follow-up survey: Consumption o No effects on consumption or food security Income generation o Increased income from livestock o But self-employment displaces wage labor o No net effect on income Livestock o Effect on livestock holdings but decreasing over time Savings and borrowing o No effects India – Andra Pradesh(SKS) Researchers: Morduch, Ravi and Bauchet
Pakistan Researchers: Karlan and Pariente Results from 1 st follow-up survey: Consumption o 10% increase in monthly consumption food consumption emergency expenses Improved food security o Increase in share of households that eat twice a day (3 % points) o Decrease in share of households that experience food shortages
Pakistan Income generation o Important increase in agricultural output o Increase in micro- enterprise profits o No clear effects on livestock income Assets o Increase in livestock holdings o Increase in physical assets
Honduras Researchers: Karlan and Thuysbaert Consumption and non -food expenditures o No effects Improvement in food security and in self-assessed economic situation o Adults 5 percentage points less likely to have reduced meals o Improved perception of own economic situation
Honduras Income generation o No effects on overall agricultural output nor business profits o Increase in vegetable production o Increased income from livestock Livestock o 25% increase in total value of livestock owned Savings o Increased savings Mental health o Improvement in happiness index
Conclusion First results very promising Enlarged asset base More self-employment Improved food security Increased food and non-food consumption Improved happiness … But variation across sites Are some areas less suitable for graduation model than others? Choice of livelihoods? Critical questions for evaluation Long term Cost-effectiveness No definitive answers yet More data and analysis coming soon!
Thank You!