Lending Through the Fire: Credit Expansion and Rising Bank Risk After the 2025 California Wildfires

The 2025 Los Angeles (LA) County wildfires destroyed more than 16,000 structures and were followed by more than $20 billion of insurance payments within seven months. We use a difference-in-differences (DiD) design to examine how this acute physical climate shock transmitted through banks’ capital, credit risk, profitability, management efficiency, liquidity, funding, and credit supply. Treated banks are commercial banks headquartered in LA County; control banks are headquartered in counties with no FEMA major disaster declaration over 2022Q1–2025Q4 and have no LA County branch. We estimate entropy-balanced DiD and event-study models with bank and quarter fixed effects on a seven-quarter window (2024Q1–2025Q3) with onset in 2025Q1. To separate the effects of the wildfires from pre-existing trends and from differences between LA County banks and other banks that appear even in a year without a fire, we examine each effect with parallel-trends and placebo-timing diagnostics. The cleanest evidence comes from the single-county sample. Relative to the control banks, single-county treated banks’ non-performing loan ratio rose, and their management efficiency and return on assets deteriorated, while their total and real estate lending grew and the yields on these loans and their cost of deposits fell. We term this coexistence of rising credit risk and expanding intermediation the disaster liquidity paradox. For the full sample, only the increase in non-performing loans and the decline in the cost of deposits pass all diagnostics. The full-sample deposit and lending expansion, however, is of a size that LA County banks also display in a year without a fire, so we present it as a description of the post-wildfire period rather than as a causal effect of the wildfires. Insurance and relief payments sustaining bank funding are a possible explanation for this pattern; because Call Report data cannot trace these flows, the analysis does not directly demonstrate this mechanism. As all treated banks are headquartered in a single county and experienced a single event, the estimates describe this episode and may not generalize to other disasters or banking markets.