The Bank's Climate Risk Capital Question
A regional bank has €40bn in loans, of which 22% is to commercial real estate (CRE) and 15% to agriculture. Its climate risk team runs a transition scenario and a physical scenario and finds: under a disorderly transition, CRE values fall 8% and agriculture margins compress 12%; under a hot-house physical scenario, flood losses hit 6% of the CRE portfolio and drought cuts agricultural output 15%. The bank's internal capital model uses a one-year horizon and does not include climate factors. The regulator has asked the bank to demonstrate that climate risk is 'adequately captured' in its ICAAP. Management proposes adding a qualitative overlay to the ICAAP narrative and a small management buffer, arguing that climate risk is 'too uncertain' for quantitative capital modelling.
1. Is the management proposal adequate? What does 'adequately captured' require?
Read the worked answer
A qualitative overlay plus a small buffer is unlikely to be adequate if the bank has quantified material exposures. Regulators (ECB, PRA) expect climate risk to be materially captured in the ICAAP - either through scenario-based capital add-ons, sensitivity analysis, or integration into credit-risk parameters - with a clear statement of limitations. The bank has already produced numbers (8% CRE value fall, 6% flood losses, 15% agricultural output cut) that map to credit losses. 'Too uncertain' is not a sufficient justification for exclusion when the bank can quantify a range. Adequate treatment would include: (1) scenario-based loss estimates translated into PD/LGD adjustments; (2) a capital add-on or buffer sized to the stress results; (3) disclosure of data gaps and model limitations; (4) a plan to improve over time. The overlay can complement but not replace quantification.
2. How should the bank reconcile the one-year capital horizon with climate risk's long time horizon?
Read the worked answer
Climate risk is long-horizon (decades) while regulatory capital is one-year - a genuine mismatch. The reconciliation: (1) recognise that long-horizon risks transmit into one-year risk through repricing, refinancing, and market sentiment - e.g., a CRE loan maturing in a high-carbon, flood-exposed building may face refinancing risk now; (2) use scenario analysis over multiple horizons (1, 5, 10, 30 years) to show the trajectory, even if capital is one-year; (3) identify near-term transmission channels (collateral revaluation, insurance withdrawal, borrower cash-flow stress) and model those in the one-year horizon; (4) treat long-horizon findings as strategic risk feeding the business plan and risk appetite, not only capital. The bank should not pretend the horizons match, but should show how long-term risk is brought into short-term decisions.
3. The CRE and agriculture exposures are concentrated regionally. What additional analysis is needed?
Read the worked answer
Concentration and correlation analysis is essential. CRE (22%) and agriculture (15%) are 37% of the book and likely regionally correlated - the same flood or drought event can hit both, and both are sensitive to the same physical hazards. The bank should: (1) map exposures geographically against hazard layers to identify clusters (e.g., floodplain CRE + drought-prone farmland in the same region); (2) run joint physical scenarios, not separate ones, to capture correlated losses; (3) assess insurance and collateral - flood-exposed CRE may become uninsurable, removing a mitigant; (4) consider second-round effects (agricultural income falls → local CRE retail demand falls). The combined stress could exceed the sum of the individual 6% and 15% figures. This correlation is exactly what a qualitative overlay would miss.