How Borrowers Respond to Dynamic Incentives? An Exploration of Repayment Decisions on Digital Credit
中文摘要
本文考察数字信贷市场中的借款人如何应对动态激励。为此，我们利用渐进式贷款政策中的准实验变异来识别其对还款行为的因果效应。我们的实证策略揭示了借款人在借款人与贷款人关系不同阶段上响应的系统性异质性。我们发现，外生获得更大初始贷款的首次借款人的违约概率显著更高——这与道德风险上升和还款负担加重效应相一致。相反，相对于上一借款周期获得更大贷款增量的重复借款人则表现出显著更低的违约率。这些对比鲜明的模式与一个理论框架相一致：在该框架中，有经验的借款人将贷款规模的增加解读为未来信贷扩张的可信信号，从而强化其维持还款纪律的激励。为支持这一机制，我们记录了稳健的战略性还款行为证据，尤其是在连续贷款周期中快速进阶的借款人之中。总体而言，我们的发现凸显了关系动态在塑造借款人行为中的关键作用，并强调了根据借款人经验水平量身定制渐进式贷款政策的重要性，以在多元化客户群体中改善还款结果。本文由 Bruno Biais（金融学）接受。补充材料：数据文件可在 https://doi.org/10.1287/mnsc.2021.02016 获取。
Abstract
We examine how borrowers in digital credit markets respond to dynamic incentives. To this end, we leverage quasi-experimental variation in progressive lending policies to identify causal effects on repayment behavior. Our empirical strategy uncovers systematic heterogeneity in borrower responses across different stages of the borrower–lender relationship. We find that first-time borrowers who receive exogenously larger initial loans are significantly more likely to default—consistent with elevated moral hazard and increased repayment burden effects. In contrast, repeat borrowers who receive larger loan increments relative to their prior borrowing cycle exhibit substantially lower default rates. These contrasting patterns are consistent with a theoretical framework where experienced borrowers interpret loan size increases as credible signals of future credit expansion, thereby reinforcing their incentive to maintain repayment discipline. Supporting this mechanism, we document robust evidence of strategic repayment behavior, particularly among borrowers who advance rapidly through successive loan cycles. Collectively, our findings highlight the critical role of relationship dynamics in shaping borrower behavior and underscore the importance of tailoring progressive lending policies to a borrower’s experience level in order to enhance repayment outcomes across a diverse client base. This paper was accepted by Bruno Biais, finance. Supplemental Material: The data files are available at https://doi.org/10.1287/mnsc.2021.02016 .