Credit Markets with New Screening Technologies

We study equilibrium market structure in a model of credit-market competition with correlated or uncorrelated screening errors and endogenous screening technologies. A pool externality arises because rejected borrowers reapply elsewhere, so lenders’ screening and pricing choices affect competitors’ applicant pools. With independent errors, a single market segment emerges, with rates rising smoothly as lenders who choose better technologies lend to more adversely selected borrowers. With correlated screening errors, potentially stemming from similar AI systems and overlapping financial-sector data, ex-ante identical lenders instead specialize across qualitatively distinct segments, generating a hockey-stick interest-rate schedule: segments with uniform rates but heterogeneous borrower quality coexist with segments featuring uniform quality but dispersed rates, echoing traditional banking, fintech, and high-rate indiscriminate lending. We show how technological and regulatory changes affecting screening costs, including frontier AI innovation and various forms of open banking, unevenly affect financial access and can either benefit or harm financially excluded borrowers.