A study estimates that loan losses would vary widely depending on the policy scenario and how loans are repaid.
Researchers combined loan-level data from the Federal Reserve with estimates of how climate policies could affect different industries. Under the most adverse scenarios, estimated exposures reached 14% of loan portfolios; accounting for historical loan repayment patterns reduced the estimates to 0.5%–2%. The researchers also found that broader economic shocks could raise exposures by 3–5 percentage points.
Carbon emissions explained no more than 60% of the variation in the estimated exposures. Banks that had signed the Net-Zero Banking Alliance had lower estimated exposures than non-signatories, mainly because they reduced lending to industries expected to be hurt most by the transition.
Estimated bank exposures
The researchers estimate that banks’ exposure to climate-transition risk is relatively modest but depends strongly on the scenario and measurement assumptions. In the strictest policies or most adverse scenarios, exposure reached 14% of loan portfolios. When historical loan payoff patterns were included, estimated exposure fell to 0.5%–2%. Aggregate economic shocks could add 3–5 percentage points.
The study also found that commonly used carbon-emissions measures accounted for at most 60% of the variation in banks’ estimated exposures. Banks that signed the Net-Zero Banking Alliance had reduced exposure compared with non-signatories, mainly by cutting lending to industries considered most vulnerable to the transition. Borrowers in those industries disproportionately moved to non-signatory banks, while borrowers in industries expected to benefit from the transition disproportionately moved to signatories.
Evidence and caveats
This is a model-based analysis combining forward-looking estimates from general-equilibrium economic models with loan-level data from the Federal Reserve. The researchers studied loans originated from 2012 through the first quarter of 2023 across 42 banks. The loan data covered a little less than two-thirds of commercial and industrial lending and focused on large banks and selected loan categories, including commercial and industrial, agricultural, property-secured and lease loans.
The estimates depend on the assumptions and scenarios used in the underlying models. The study focuses on transition risks, not physical climate risks, and does not cover all bank assets, such as mortgages. The researchers describe the estimates around climate-transition events as an upper bound on banks’ transition-risk exposures. The evidence that alliance signatories changed lending is an association between signatory status and lower estimated exposure; the excerpts do not establish that signing the alliance itself caused the changes.
// Source
Journal of Financial and Quantitative Analysis · 2026 · DOI: 10.1017/s0022109026103226
Authors: Hyeyoon Jung, João A. C. Santos, Lee Seltzer
Institutions: Federal Reserve Bank of New York, NOVA School of Business and Economics
