The pass-through of loan-loss-provisioning on mortgage lending: Evidence from a regulatory change

Abstract Macroprudential policy and instruments are at the center of the discussion when (re)thinking about banking regulation. In particular, sectoral macroprudential devices have gained momentum to curb specific credit markets. However, there is scarce evidence on how, and by how much, these policies affect the credit markets they are intended to target. In this paper, we analyze one such regulation. We study how loan-loss-provisioning (LLP) affects the mortgage lending market. We do so by leveraging a policy change in Chile, that raised provisions, coupled with rich administrative data for the census of all real estate transactions, interest rates on mortgage loans, and bank balance sheet data. We find that the LLP regulation lowered the loan-to-value (LTV) ratio of granted loans by 2.8% and 9.8% for the mean and median borrowers, respectively. Banks with weaker balance sheets showed more extensive adjustments. We also find that the pass-through of higher expected provisioning cost was quantitatively very small, reaching an upper bound of 0.2% at most. We argue that this evidence supports the notion that under imperfect information, banks do not compete only in terms of interest rates, but also in lending standards, and this latter channel can dominate the former.