You can pick your friends, but you need to watch them: loan screening and enforcement in a referrals field experiment

We examine a randomized trial that allows separate identification of peer screening and enforcement of credit contracts. A South African microlender offered half its clients a bonus for referring a friend who repaid a loan. For the remaining clients, the bonus was conditional on loan approval. After approval, the repayment incentive was removed from half the referrers in the first group and added for half those in the second. We find large enforcement effects, a $12 (100 Rand) incentive reduced default by 10 percentage points from a base of 20%. In contrast, we find no evidence of screening.