AI for Insurance Underwriting: Claude Tools for P&C and Life Underwriting Analysis

Insurance Underwriting and AI

Insurance underwriting is the process of evaluating and pricing risk — determining which risks to accept, at what premium, and with what terms. P&C (property and casualty) underwriting focuses on loss frequency, severity, and combined ratio management. Life and health underwriting focuses on mortality, morbidity, and persistency. Actuarial pricing models, loss development triangles, credibility weighting, and reinsurance optimization all require sophisticated analysis. Claude with ClaudeFinLab accelerates underwriting analysis, pricing reviews, portfolio assessment, and regulatory filing support.

P&C Underwriting — Loss Ratio and Combined Ratio

• "Combined ratio analysis: commercial auto insurer with $280M of earned premium. Loss ratio = incurred losses / earned premium = $196M / $280M = 70.0%. Expense ratio = underwriting expenses / earned premium = $70M / $280M = 25.0%. Combined ratio = 70.0% + 25.0% = 95.0%. Below 100% = underwriting profit of $14M. Investment income = $22M (on $420M float). Pre-tax operating income = $14M + $22M = $36M. Operating ratio = combined ratio − investment yield = 95.0% − (22/280) = 95.0% − 7.9% = 87.1%. Compare to industry: commercial auto industry combined ratio 2025: 98.2% (near break-even). Our insurer's 95.0% is materially better — outperformance driven by risk selection and loss control."
• "Loss development triangles: workers' comp insurer evaluating reserve adequacy. Loss development triangle for accident years 2020-2025. Column headers: development periods 12, 24, 36, 48, 60, 72 months. Selected development factors (CDFs): 12-24: 2.850; 24-36: 1.520; 36-48: 1.220; 48-60: 1.085; 60-72: 1.025; tail 72+: 1.015. Ultimate LDF = 2.850 × 1.520 × 1.220 × 1.085 × 1.025 × 1.015 = 6.14 for immature year. For accident year 2025 with $18M in paid losses at 12 months: ultimate loss = $18M × 6.14 = $110.5M. IBNR reserve needed = $110.5M − $18M = $92.5M. Does current reserve of $88M appear adequate? Short by approximately $4.5M."
• "Rate adequacy analysis: homeowners insurer in Florida. Current average premium $4,200. Historical loss ratio target: 65%. Actual 3-year average loss ratio: 78% (elevated by hurricane frequency). Required rate change = actual LR / target LR − 1 = 78% / 65% − 1 = 20.0% average rate increase needed. Factor in: trend adjustment (claim severity trending +8%/year × 2-year development lag = +16.6%); loss elimination ratios (deductible changes reducing expected losses -4%); expense ratio changes (+1%). Total indicated rate change = 20.0% + 16.6% − 4.0% + 1.0% = 33.6%. Florida regulatory cap: +12% per year without prior approval — file for exception or implement over 3 years."

Life Insurance Underwriting

• "Term life underwriting decision: 42-year-old male applicant, non-smoker, $2M 20-year term policy. Health history: BMI 28 (overweight but not obese), blood pressure 138/88 (mildly elevated, on medication), no family history of early cardiovascular disease, cholesterol total 210, LDL 140. Underwriting decision: Standard Plus (preferred non-smoker declined due to elevated BP, but not rated). Premium calculation: base rate for M42 non-smoker 20-year term per $1,000 face = $2.85. Adjusted for Standard Plus vs. Preferred: multiply by 1.15 table rating. Annual premium = ($2.85 × 1.15) × ($2,000,000 / $1,000) = $3.275 × 2,000 = $6,550/year. Compare: preferred rate would be $2.85 × $2,000 = $5,700 and rated is $6,550 — 14.9% higher."
• "Mortality table analysis: 2017 CSO (Commissioner Standard Ordinary) mortality table for pricing 30-year term life. Male age 45, non-smoker. 2017 CSO qx (annual mortality probability) at key ages: 45: 0.00185 (0.185%); 50: 0.00325; 55: 0.00601; 60: 0.01124; 65: 0.01924; 70: 0.03207; 74: 0.05012. Net single premium for $1M 30-year term (pure mortality cost, no expenses): sum of [qx × PV factor × $1M] for ages 45-74. Using 4% discount rate, approximate NSP = $28,400. Add expenses (20% loading) and profit (15% loading): gross premium = $28,400 / (1 − 20% − 15%) = $43,692 per year NSP equivalent. Convert to level annual premium using annuity factor: approximately $1,820/year for a properly priced non-participating term policy."

Reinsurance Structures

• "Quota share vs. excess of loss reinsurance comparison: insurer has $500M catastrophe exposure in hurricane zone. Option 1 — Quota Share: cede 30% of premium and losses to reinsurer. Cede: 30% × $500M exposure = $150M. Reinsurer pays 30% of all losses, insurer retains 70%. Benefit: reduces exposure immediately; cost: cede 30% of profit. Option 2 — Cat XL (excess of loss per occurrence): structure $100M xs $50M cat cover. Insurer retains first $50M of any cat event; reinsurer pays $100M above that (up to $150M per event). Cost: cat XL premium $8M/year. Expected loss: historical frequency 0.5 events/year above $50M × average severity $75M excess layer = expected reinsurer payment $37.5M/year. Compare: QS more expensive in expected value ($15M of premium ceded + profit share vs. $8M XL premium), but XL leaves insurer more exposed to small-medium events."
• "Reinsurance pricing: Lloyd's market facultative quote for a $50M limit on a US commercial property risk. Property: manufacturing facility, $180M TIV (total insured value), sprinklered, occupancy: plastics manufacturing (moderate hazard). Rate calculation: base rate 0.25% of TIV = $450K for full $180M limit. Layer pricing for $50M xs $30M ($30M-$80M layer): using increased limits factors — ILF for $80M vs $30M limit (from severity distribution) = 1.42; layer cost = base rate × ILF factor for layer = $450K × (1.42 − 1.0) / (30/180) = ... Estimated facultative reinsurance rate: 0.35% of $50M limit = $175K. ROL (rate on line) = $175K / $50M = 0.35%."

Reserve Adequacy and Actuarial Analysis

• "Bornhuetter-Ferguson reserve method: commercial liability book. Accident year 2023: earned premium $42M, expected loss ratio 68%, expected losses = $28.56M. Paid losses to date (at 24 months development) = $8.2M. % unreported based on development factors: at 24 months, 60% of ultimate losses remain unpaid (40% paid-to-date factor). BF method: IBNR = (1 − % paid) × expected ultimate = 60% × $28.56M = $17.1M. Reported loss development method: $8.2M × CDF(24 to ultimate) = $8.2M × 2.45 = $20.1M ultimate. Comparison: BF = $28.3M ultimate vs LDF = $20.1M. BF is more credible for immature years with sparse data; LDF method may be more appropriate for mature years. Blend at 24 months: 40% LDF + 60% BF = 40% × $20.1M + 60% × $28.3M = $8.04M + $16.98M = $25.0M blended ultimate."

Related Articles

• Insurance Actuarial AI
• Insurance Carriers AI
• Financial Risk Management AI