Abstract: 

We use a novel loan-level dataset containing borrower-specific probability of default to accurately measure lenders’ expectations. The analysis is based on a learning model where bankers endowed with diagnostic expectations receive noisy signal about firms’ fundamentals and estimate their probability of default. The evidence suggests that banks could be subject to expectational distortions: (i) intermediaries tend to overreact to both micro and macro news, overestimating (underestimating) borrowers’ defaults after negative (positive) signals; (ii) the degree of overreaction is heterogenous among banks; (iii) overreacting bankers decrease (increase) interest rates more than rational ones and the probability of issuing a new loan rises (fall) when bankers receive positive (negative) signals. We rationalize these results with the structural estimation of a model of banking competition where banks’ profits depend on borrowers’ creditworthiness.



Jacopo Tozzo CV



for further information contact angela.baldassarre@unibocconi.it