List randomization for sensitive behavior: An application for measuring use of loan proceeds

Policymakers and microfinance institutions (MFIs) often claim to target poor entrepreneurs who then invest loan proceeds in their businesses. Typically in non-research settings these claims are assessed using readily available but unverified self-reports from client loan applications. Alternatively, independent surveyors could directly elicit how borrowers spent their loan proceeds. That too, however, could suffer from deliberate misreporting. We use data from the Peru and the Philippines in which independent surveyors elicited loan use both directly (i.e., by asking how individuals spent their loan proceeds) and indirectly (i.e., through a list-randomization technique that allows individuals to hide their answer from the surveyor). We find that direct elicitation under-reports the non-enterprise uses of loan proceeds.