Collaborative Research: Fingerprinting to Reduce Risky Borrowing

University Of Maryland, College Park, College Park MD

Investigators

Abstract

The research will quantify the impacts of improved personal identification of borrowers through a randomized controlled trial on technology to collect digital fingerprints of microloan borrowers. This alternative means of establishing personal identity, will allow for testing dynamic repayment incentives, which involve conditioning future credit access on a borrower?s past repayment performance. This study is designed to measure how changes in borrowers? behavior due to increased incentives to repay loans affect their profits, consumption, investment, and household wellbeing. In addition, this study will examine how lenders respond to the new technology, by measuring how loan officers interact with clients, how microfinance institutions adjust their borrowing requirements, and the local availability of credit. Impacts will be measured using administrative data from microlenders in combination with two years of survey data on borrowers and loan officers. Our results could lead private and non-profit lenders to adopt fingerprint identification technology of their own accord. In addition, our findings could help shape the policies of governments and of development institutions such as the World Bank, which could (for example) provide technical assistance to help foster the establishment of fingerprint-based credit reference bureaus that could further enhance the social benefits of the technology. We implement a randomized controlled trial involving microfinance customers and loan officers at five different microfinance instiutions. Credit officers play important roles in screening customers and enforcing dynamic incentives for loan repayment, and rely on their personal assessments and knowledge of customers in the absence of reliable ways to identify new or repeat borrowers. We randomize access to fingerprinting technology for screening borrowers and checking credit histories at the credit officer level, stratified by MFI. This strategy also generates variation in the fraction of credit officers in a given geographic area who are using the new technology. We use this variation to estimate the effect of fingerprinting on behaviors of three important sets of actors: borrowers themselves, credit officers, and MFIs. Our main estimates will come from OLS regressions, which are unbiased because of the random assignment of credit officers to the fingerprinting technology, but will include district and MFI (strata) fixed effects to improve precision. At the borrower level, we focus on repayment outcomes and on measures of household welfare. On the supply side, we measure the time allocation and decision-making of credit officers to see whether they reallocate their effort to monitoring borrowers who were not fingerprinted and are therefore not subject to increased repayment incentives, and to measure the evolution of decisions about access to credit at the extensive and intensive margins. We use the regional variation in the intensity of fingerprinting to see whether MFIs reallocate staff time or loanable funds to areas in which they have comparative advantages (more access to fingerprint technology than their competitors within the market, and more access than they themselves have in other markets). Ultimately, this project seeks to deepen our understanding of the impacts of a pervasive form of information asymmetry in many developing countries by studying dynamic behavior of borrowers and lenders, and measuring supply side responses to improved borrower identification. Answers to these questions will provide a fuller picture of the economic impacts of reducing a common but so far understudied type of information asymmetry.