Quantifying the Value of Heterogeneous Preference: An Empirical Matching Model of Peer-to-Peer Lending

This paper focuses on an essential, but previously underexplored, mechanism that underlies the operation of peer-to-peer (P2P) lending markets: lender-borrower matching. We model the formation of lender-borrower matches as an endogenous matching process in which agents strategically choose their partners based on their preferences. Using data from a large online lending site, we empirically investigate the determinants of lender-borrower matching and explore their importance. Our structural estimates suggest that heterogeneous preference leads to various matching patterns across different types of lenders and borrowers. Experienced, hasty, and male lenders exhibit more tolerance toward borrowers with a lower credit score than do their counterparts. Based on these estimates, we perform counterfactual analyses to quantify the economic value of P2P lending. Note that P2P lending differs from traditional lending by following a decentralized matching process under which heterogeneous preference is maintained. We quantify the matching value from the decentralized matching process in P2P lending and compare it to that of its centralized counterpart for which heterogeneous preference is not respected. We find that P2P lending gains value from heterogeneous preference and that a more centralized matching process hurts large and active markets less than for small and inactive markets. Further, an agent-level analysis shows that more experienced lenders and low credit score borrowers are more vulnerable to a credit-score-ruled centralized matching process. Our findings offer managerial implications for credit scoring and information disclosure policy.