Measuring Causal Impacts on Multifaceted Outcomes with Missingness, with an Application to Welfare Impacts of Mobile Credit

• Activity Number: 35 - Applications of Nonparametric Methods
• Type: Contributed
• Date/Time: Sunday, July 28, 2019 : 2:00 PM to 3:50 PM
• Sponsor: Section on Nonparametric Statistics
• Author(s): Jacqueline Mauro* and Joshua Blumenstock and Katherine Yen and Andrew Linxie
• Companies: and University of California Berkeley and UC Berkeley and UC Berkeley
• Keywords: causal inference; nonparametrics; mobile lending

Abstract

In the absence of a well-established financial sector, new tools for borrowing using mobile phones have arisen in East Africa. The welfare impacts of these high-interest, unsecured loans is not well known; ideally they provide a new avenue to credit to the underbanked, but they may also lead to debt spirals. In order to study the impacts on a multifaceted outcome like welfare, we extend nonparametric multivariate scaled ATE estimators to incorporate missing data, and develop a similar new LATE estimator. These estimators are doubly-robust and fully nonparametric, and allow us to robustly test for impacts of lending on multiple outcomes at once. We use data from a lender in the region to develop proxy measures of welfare which we analyze using these new tools. We compare our measures of welfare to self-reported measures of welfare by conducting interviews in Kenya.