Optimal interest rates personalization in FinTech lending

As an emerging business model innovation in the era of big data and the Internet economy, FinTech lending has thrived in many countries and become an important supplement to traditional bank lending. FinTech service providers perform automated credit assessments based on machine learning technologies and provide various customized Internet lending services with personalized interest rates to their users. In this paper, we develop an analytical framework to investigate the mechanism of personalized interest rate pricing, which is the core decision problem in the FinTech lending business model. The objective of the interest rate personalization problem is to choose an optimal interest rate for each loan applicant based on their credit assessment to maximize the total expected revenue under the constraint on total loan capacity. The credit assessment of an applicant is modeled as the posterior distribution of their credit level given the relevant user data possessed by the firm, and its accuracy is determined by the efficacy of the credit assessment system and the informativeness of the user data. We first characterize the solution procedure under a general framework and then solve the optimization problem with a specified credit assessment mechanism. A key result is that, in an efficient loan market, a loan application should not be approved unless the credit assessment accuracy is higher than a dynamic threshold associated with the predicted credit level, which signifies the importance of data and technology in FinTech lending. We anticipate that our optimization model could be implemented in the automated loan processing system by FinTech firms and provide managerial implications for optimally investing in technology and data assets.