Spatial Externalities of Social Programs: Why Do Cash Transfer Programs Affect Ineligibles’ Consumption?

While impact evaluations of social policies focus mainly on treatment effects on the treated, less attention has been devoted to the analysis of indirect treatment effects (treatment externalities). This paper analyzes food consumption outcomes of households that do not participate in an income subsidy program yet live in the same village as program participants. Extending the work of Angelucci and De Giorgi (2009), we remove the ’veil of ignorance’ from the distribution of the indirect treatment effect: ’Better-off’ ineligible households benefit more than ’poorer’ ineligible households. Depending on the level of household poverty, the indirect treatment effect arises through different behavioral responses of ineligible households: ’Poorer’ ineligible households reduce a part of the buffer stock previously held to insure against shocks, and receive more informal insurance installments (in-kind transfers) from program participants who are now able to scale-up informal insurance arrangements. ’Better-off’ ineligible households, facing an effective increase in credit resources, additionally finance the increase in food consumption through loans. Our results imply that evaluations that focus merely on program-participants do underestimate the program’s overall effect on poverty. Furthermore, the impact of a cash transfer program on reducing inequality is lower than previously thought. Moreover, the regression discontinuity design estimator is biased if the comparison group is sampled from the same village. Finally, if certain conditions hold, a program’s eligibility borderline may be lowered, hence public budget constraints relaxed, while maintaining the same program effect on poverty.

JEL Classification: E21, H43, I38, O12, O17.

Keywords: Poverty, Program Evaluation, SUTVA, Quantile Treatment Effects