Using granular bank-firm data from the Spanish credit register and U.S. syndicated loans, I examine the impact of banks' sectoral specialization on credit supply decisions in response to monetary policy shocks. Following an expansionary monetary policy, banks significantly increase lending to firms in sectors where they specialize. A 25 basis point rate reduction leads specialized banks to increase credit by 1.2–1.4 percentage points more to firms in their sectors of expertise, with effects peaking after one year. This result holds across both the Spanish lending market and the U.S. syndicated one. The mechanism operates through information advantages: the specialization effect is stronger for opaque borrowers, and specialized banks experience lower default rates following monetary easing. These supply decisions have real effects, as firms with greater exposure to specialized lenders exhibit larger increases in investment and profitability during expansionary periods. The findings show that monetary policy transmission varies systematically across sectors depending on the distribution of bank specialization.

We study how uncertainty about collateral values affects household welfare through refinancing frictions in the mortgage market. We measure collateral valuation uncertainty using idiosyncratic price dispersion in housing transactions and link it to mortgage application and loan performance data. Higher uncertainty reduces refinancing and raises denial rates and collateral-related denial reasons, consistent with appraisal noise making LTV cutoffs bind. Lenders internalize the resulting lower prepayment risk: banks are more likely to retain these mortgages rather than sell or securitize them. Suppressed refinancing generates large, spatially uneven welfare losses through foregone interest savings. Overall, we identify an overlooked supply-side driver of refinancing outcomes.

We develop a model with firm dynamics, imperfect competition, and investment options. Firms borrow using credit lines offered by competitive lenders that can optimally provide “zombie” lending, with an interest rate below the risk free rate. We solve and calibrate the model and show that zombie lending is used in equilibrium to subsidise two very different types of firms: “Stagnant zombies” previously borrowed to fund projects that are no longer feasible. They are offered zombie lending so that they can gradually reduce their debt and avoid default. “High-potential zombies” have options to invest in projects that can increase their profitability, and borrow to pursue them. Zombie lending “buys them more time”, in the hope they will become more profitable in the future and will thus be able to repay the debt. The main insight of the model is that while both types of zombie firms are consistent with the way such firms are usually identified in the empirical literature, they have opposite implications for misallocation and aggregate productivity and welfare. We use the model to propose novel criteria to identify the different zombie types and new testable predictions of their allocative effects: an increase in markup dispersion within industries should increase the share of zombie firms and the share of zombies that in the future recover and reach top quartile performance; sectors where zombies behave like high-potential ones should exhibit positive TFP responses to zombie prevalence, while those dominated by stagnant zombies should show declining productivity. Using firm-level data from Italian firms between 2007–2016, we provide robust empirical evidence confirming these predictions.