Journal Title

Journal ISSN

Volume Title



This dissertation consists of two essays on the FinTech lending market. I explore default in the FinTech lending market using a dataset of both extensive credit and soft information for borrowers from the largest FinTech lender in the United States. In the first essay, I study the default risk in this market over the business cycle. I find that both macro and regional economic conditions play a role in consumer default and should be taken into consideration when assessing credit risk. I show that lenders operating in this market increasingly focus on subprime borrowers, whose default rates are more sensitive to macro and regional economic conditions than those of prime borrowers. Based on estimates from a duration model, I provide counterfactual analyses of what default rates and the associated total losses would look like in different economic scenarios. In the case of a recession, the losses would be 37 percent higher than in the case of an expansion. For the same volume of loans in the recession, doubling the subprime share would lead to an additional 6.2 percent increase in losses. In the second essay, I provide an overview of some of the most common machine learning methods used in modeling default risk and assess to what extent these methods are better than traditional approaches. Using the same datasets as in the first essay, I explore the determinants of default in the FinTech lending market. I apply different machine learning algorithms to predict out-of-sample default. I find that some of the machine learning algorithms, such as extreme gradient boosting and artificial neural networks, marginally outperform logistic regression. Annual income, loan purpose, revolving line utilization, and interest rate are the most important variables predicting default. Macro and regional variables are listed among the top 10 variables explaining consumer default behavior.



FinTech lending, default risk, business cycle, modeling risk