Robert Völter, Goethe-Universität Frankfurt Comment on „Indirect Effects of an Aid Program: How do Liquidity Injections Affect Non-Eligibles Consumption“ by Manuela Angelucci and Giacomo De Giorgi Robert Völter, Goethe-Universität Frankfurt

The Paper Non-poor are not eligible for Progresa transfers But their consumption is higher in villages where the poor get transfers Indirect Treatment Effect ITE Possible Reasons: More labor income? No. More income from goods markets? No. More other program receipts? No. More money from credit market/ family transfers? Yes. Reduction in savings? Yes. Hypothesis: more liquidity in the community facilitates consumption smoothing Test: compare ITE on households hit /not hit by a shock ITE larger for those hit by a shock

Comments SUTVA not fulfilled; indirect effects on the non-treated (and likely also on the treated) ITE estimates exact or lower bounds? Information about social networks across village borders? Insightful study, checking many possible reasons makes it extensive Randomization saves econometrics Standard Errors often quite large, so sometimes effects are not significant in all three waves Differentiantion of ITE by shock very helpful, ITE on credit given shock present often significant where unconditional ITE is not, this corroborates the insurance hypothesis

Questions Size of transfers compared to the community income? Given 20% of pre-program consumption for the treated households (52%) ca 10%? Variation in liquidity increase? Labor market effects: per capita monthly earnings. Household composition? Could look at wages and hours of adults and children. Are the data panel data or repeated cross sections? Long run? Consumption of own savings gives only one-off increase in consumption Family-transfers may give permanent increase in consumption Insurance should just give consumption smoothing.