Cross-elasticities in credit markets

We develop a methodology to estimate quantity cross-elasticities in decentralized markets, and apply it to measure the extent of credit reallocation between firms and banks following shocks. Our identification strategy leverages how not all firms borrow from all banks; indirect connections through the credit network thus provide natural exclusion restrictions. Through theory and simulation, we demonstrate the viability of our approach and show that neglecting reallocation leads to an unpredictable bias in treatment and fixed-effects estimates. By applying our estimator to twenty years of data from the Italian credit registry, we find that credit for firms is substitutable in normal times but complementary during recessions, whereas banks' cross-elasticities do not vary with the business cycle. Such evidence is consistent with the asymmetric effects of credit expansions and contractions documented by the empirical banking literature.