Default risk surprises in banks' loan pricing
Using one million loans matched to firms' public credit ratings, I show that interest rate differentials unexplained by observables at issuance contain information about future changes in firms' default probability. On average, firms whose default probability increases by 1 p.p. one year after loan issuance were charged an otherwise unexplained 0.04 p.p. rate premium at issuance; the premium doubles for younger and smaller firms, strengthens in crises and at longer maturities. Information available to banks at loan issuance arises as a potential driver of interest rate differentials and of the positive relationship between rates and default risk. Yet, it is not necessary: the same pattern can emerge in a market with search frictions. I bring these forces together in a search model of loan pricing with asymmetric information in which banks see an imperfect public signal on firms' risk types. In equilibrium, the positive rate–risk relationship can arise either when the public signal is sufficiently informative to tell apart borrowers, or, despite an uninformative signal, when banks' pricing strategies induce partial separation. I use a calibrated version of the model to rationalize the empirical findings. I find that banks are informed and separate borrowers in good credit rated markets, and they are uninformed and pool borrowers in bad credit rated ones.
Presented at: Fourth Women in Central Banking Workshop, Banco de México (Nov 2026, scheduled); European Summer Symposium in Financial Markets, Evening Sessions (Jul 2026); Université Paris Dauphine–PSL PhD Workshop (Jun 2026); University of Naples Federico II PhD Workshop (Sep 2025); ICBFS Conference, Piacenza (Sep 2025); Economics Letters Summer School in Macro-Finance, Bocconi SDA (Jul 2025); European Central Bank, internal seminar (May 2025); Paris Macro Workshop (May 2025); The Wharton School, Finance, internal presentation (Fall 2024).
