Artificial Intelligence in Lending: Social and Regulation Implications of FinTech


Student thesis: Doctoral Thesis

View graph of relations


Related Research Unit(s)


Awarding Institution
Award date13 Jul 2022


Artificial intelligence (AI) has been widely adopted in a broad range of industries and has profoundly changed society. As one central application area, the finance industry has been significantly reshaped by algorithm trading, Internet lending, and cryptocurrency, among other AI-related techniques. Ever-increasing data and computational power and advancement in machine learning (ML) necessitate AI as a value-added component in finance. However, this also raises concerns about its regulatory and social implications. This dissertation tackles these issues using the lending industry as a research context.

First, I investigate AI lending’s support for the underprivileged, particularly regarding the adverse impacts of natural disasters, which wreak economic havoc and cause victims financial distress. However, when needy victims apply for loans, lending firms often tighten their lending practices in response to the applicants’ impaired repayment ability. Through collaboration with a leading credit scoring company that offers AI-as-a-Service (AIaaS) for lenders, I could observe lenders’ use of AI services to assess loan applications. Via stochastic frontier analysis (SFA), I found that lenders who use AI services fare better in terms of reducing natural disaster victims’ default rate. Notably, the default reduction effect is more pronounced for underprivileged borrowers with lower credit scores. In this paper, I explore possible underlying mechanisms and discuss the societal implications of the findings.

Second, I investigate the impact of financial regulations on the FinTech industry and explore the potential application of AI toward a better regulation policy. Specifically, I examine how borrowers’ upward mobility was affected by China’s suppressive regulation on P2P lending in 2018, which switched from an “all-in” policy to an “all-shutdown” policy, leading to the massive closure of P2P lending companies and the effective shutdown of the entire industry by 2021. Drawing upon a unique dataset on the Chinese credit market, I show that this one-size-fits-all regulation obstructed borrowers’ upward mobility as reflected in their credit scores and their loan channel choices. Moreover, I found that those borrowers whose upward mobility the regulation nixed tended to commit more crimes. Note that financial regulations often focus on containing risks in financial services, while paying inadequate attention to the adverse effects on the underprivileged; this dissertation advocates the use of AI to help stipulate personalized regulatory policies (PolicyTech). By restricting borrowers’ access to P2P lending based on their AI-predicted financial risk, this dissertation demonstrates the possibility of protecting borrowers’ upward mobility while simultaneously containing financial risk, which has significant theoretical and societal implications.

Overall, the two studies contribute to the literature on the role of AI in consumer lending, with the aim of enhancing borrowers’ social welfare from the perspectives of both lenders and regulatory authorities. Generally, AI has been confirmed as a positive factor that can improve borrowers’ welfare. It relieves the adverse outcomes of natural disasters and has the potential to address the regulatory dilemma of simultaneously protecting lenders and borrowers.