AI for Bank ALM: Interest Rate Risk, NIM Sensitivity, and IRRBB with Claude (2026)

Bank ALM and AI

Asset-liability management is the core discipline of bank treasury. Every decision — from loan pricing to deposit product design to investment portfolio duration — flows through the ALM framework. Claude with ClaudeFinLab helps ALM teams analyze repricing gaps, model NIM sensitivity, stress test liquidity, and draft IRRBB regulatory disclosures — work that previously required specialized ALM software systems and days of analyst time.

Repricing Gap Analysis

• "Build a repricing gap schedule for a $2.8B community bank: Assets — floating rate loans $420M (SOFR+250bps, reprice immediately), fixed rate loans $680M (average remaining fixed term 3.2 years), investment securities $380M (average duration 4.1 years), cash $85M. Liabilities — demand deposits $420M (non-maturity, model as fixed with 5-year assumed maturity), MMAs $340M (reprice with 0-3 month lag), CDs $580M (average remaining fixed term 1.4 years), FHLB advances $280M. Compute repricing gap by bucket: 0-3 months, 3-12 months, 1-3 years, 3-5 years, >5 years."
• "Analyze the repricing gap implications: the bank shows a 12-month cumulative repricing gap of -$180M (liability-sensitive). In a +200bps rate environment (FOMC raised rates 8x by 25bps), what is the estimated impact on NII? Apply beta factors: loan beta 0.85 (loans reprice faster than benchmark), deposit beta 0.55 (deposits reprice slower — customers accept lower yields). Compute the net NII impact and NIM compression."

NIM Sensitivity and NII Forecasting

• "Compute NII sensitivity for the next 12 months under BCBS IRRBB shock scenarios: base case NII $85.4M. Scenarios: (1) +200bps parallel shift; (2) -200bps parallel shift; (3) +300bps parallel shift (upper shock per BCBS); (4) -300bps parallel shift; (5) steepening (short rates +200bps, long rates +100bps); (6) flattening (short rates +100bps, long rates +200bps). Apply repricing assumptions, deposit betas, and prepayment speeds. Present as dollar NII change and % change from base."
• "Model deposit repricing behavior: as FOMC raised rates from 0 to 5.25% (2022-2023), our demand deposits had an effective beta of 22% (paid 0% as rates rose vs moving to market). Now that rates are declining, model deposit repricing on the way down: will customers keep funds in low-yield accounts? Model: 40% of MMAs will reprice down with Fed funds (beta 0.40), 35% will seek higher yields (potential runoff), 25% are sticky. Compute NII for -100bps Fed funds rate cut."

Economic Value of Equity (EVE) Analysis

• "Compute Economic Value of Equity (EVE) for rate shock scenarios: EVE = PV(assets) - PV(liabilities). Base case EVE: $185M. In +200bps scenario: duration of assets 3.8 years (ΔPVA = -$213M × 0.038 × 2 = -$16.2M), duration of liabilities 2.1 years (ΔPVL = -$125M × 0.021 × 2 = -$5.3M). ΔEVE = -$16.2M - (-$5.3M) = -$10.9M. EVE under +200bps: $174.1M. Compute for +100bps, -100bps, -200bps, +300bps, -300bps BCBS standard shocks."
• "IRRBB outlier bank test: compute EVE sensitivity as % of Tier 1 capital. Our EVE under +200bps shock is -$10.9M decline (see above). Tier 1 capital $142M. EVE sensitivity = 10.9/142 = 7.7% — below the 15% outlier threshold. What would our EVE sensitivity need to be to trigger regulatory scrutiny? What are the primary levers to reduce EVE sensitivity (asset duration reduction, interest rate swap hedging, deposit product restructuring)?"

Liquidity Risk and Stress Testing

• "Build a 30-day liquidity stress test following LCR methodology: High Quality Liquid Assets (HQLA) — Level 1 cash $85M, UST/agencies $380M (0% haircut), Level 2A GSE MBS $120M (15% haircut). HQLA after haircuts: $594M. Stressed outflows (30-day): retail deposits with <$250K balance: $420M × 5% runoff = $21M; retail CDs maturing: $85M × 100% runoff = $85M; wholesale unsecured deposits >$250K: $180M × 25% runoff = $45M; undrawn credit facilities: $220M × 10% drawdown = $22M. Total outflows: $173M. LCR = $594M/$173M = 343% (minimum 100% required)."

Interest Rate Swap Hedging

• "Design an ALM hedging strategy using interest rate swaps: the bank is liability-sensitive (EVE sensitivity -7.7% to +200bps). To reduce sensitivity, consider a receive-fixed/pay-SOFR swap on $200M notional, 5-year maturity, fixed rate 4.25%. Fair value hedge on fixed-rate loans under ASC 815. Impact: swap adds $200M of floating-rate liability-equivalent, reducing asset duration. Compute the hedged EVE sensitivity and NII impact. What is the cost of the hedge in basis points of NIM?"

Where to Start

Start with the repricing gap schedule — it is the foundation of all ALM analysis. Collect balance sheet data by repricing bucket from your core system, and ask Claude to compute the cumulative gap, apply deposit betas, and estimate NII sensitivity for +/-200bps. From there, build to EVE analysis for IRRBB disclosure. ClaudeFinLab's portfolio risk MCP server handles the duration and present value calculations automatically.