Identifying Causal Effects in Experiments with Spillovers and Non-compliance (pdf, arXiv), with Francis DiTraglia, Camilo Garcia-Jimeno and Alejandro Sanchez Becerra. Journal of Econometrics, 2023, 235 (2), pp. 1589-1624
This paper shows how to use a randomized saturation experimental design to identify and estimate causal effects in the presence of spillovers – one person’s treatment may affect another’s outcome – and one-sided non-compliance–subjects can only be offered treatment, not compelled to take it up. Two distinct causal effects are of interest in this setting: direct effects quantify how a person’s own treatment changes her outcome, while indirect effects quantify how her peers’ treatments change her outcome. We consider the case in which spillovers occur within known groups, and take-up decisions are invariant to peers’ realized offers. In this setting we point identify the effects of treatment-on-the-treated, both direct and indirect, in a flexible random coefficients model that allows for heterogeneous treatment effects and endogenous selection into treatment. We go on to propose a feasible estimator that is consistent and asymptotically normal as the number and size of groups increases. We apply our estimator to data from a large-scale job placement services experiment, and find negative indirect treatment effects on the likelihood of employment for those willing to take up the program. These negative spillovers are offset by positive direct treatment effects from own take-up.
We estimate the spatial spillover effects of a program to promote savings groups in Uganda and Malawi. We use geolocation data from a cluster-randomized experiment to compare the outcomes of households in areas where a high proportion of nearby villages were randomly assigned to treatment to those in areas where a low proportion of nearby villages were assigned to treatment. We have four key results. First, take-up of the program is highest in villages where promotion by NGO-employed field officers was concentrated -- promotion by peers was about half as effective, and `organic' replication was rare. Second, we find evidence of large positive between-village spillover effects on total income, food security, and business outcomes in control villages. Third, we estimate positive direct effects of the program on business outcomes and financial inclusion -- these estimates are consistent with previous research (Karlan et al, 2017), suggesting that the presence of spillovers does not bias estimates of the direct effect in this setting. Finally, we estimate that 23-28% of the overall benefits of the program are not captured by an analysis which does not account for spatial spillover effects.
Decentralised provision of local public goods can lead to externalities and inefficiencies. I develop and estimate a spatial network model of communities’ water pump maintenance decisions in rural Tanzania. The model allows me to quantify the importance of cross-community free riding effects and pump maintenance spillovers. I find evidence of both effects and estimate that: (i) water collection fees mitigate free riding; (ii) pump technology standardization across communities reduces maintenance costs; (iii) increased pump functionality improves child survival and school attendance; and (iv) school attendance improves more for girls, who are primarily responsible for water collection.
Work in Progress
Distributional Effects of Cash Transfers (AEA pre-registration), with Stefan Dercon, Robert Garlick, Kate Orkin and Natalie Quinn
Market-Mediated Effects of Cash Transfer Programmes, with Natalie Quinn