IBNR Reserve Calculation with Claude AI: Loss Development Triangles and P&C Actuarial Workflows (2026)
How P&C actuaries use Claude AI for IBNR reserve calculations, chain-ladder loss development triangles, tail factor selection, Schedule P analysis, and actuarial reserve opinion documentation. Practical workflows for insurance reserving teams.
Claude for IBNR Reserve Calculations
Incurred but not reported (IBNR) reserve estimation is core actuarial work for every P&C insurance company and reinsurer. The analytical process — building loss development triangles, selecting age-to-age factors, applying tail factors, and calculating IBNR by line of business — is well-established methodology. The time-consuming parts are different: reviewing triangles for anomalies, documenting methodology selection decisions, writing the actuarial memorandum that explains the reserve estimate to management and auditors, and producing the Schedule P exhibits in the format regulators expect.
Claude accelerates the documentation and review layer of the reserve process. It reads a loss development triangle, identifies development patterns and outliers, recommends methodology, calculates the IBNR under multiple methods, and drafts the actuarial memorandum narrative. The Insurance & Actuarial templates cover the full reserve workflow including the Actuarial Reserve Adequacy Reviewer and reserve opinion documentation tools.
Loss Development Triangle Analysis
The chain-ladder method starts with organizing cumulative loss data into a triangle by accident year and development period. The analytical judgment begins when you look at the age-to-age factors: are they stable? Are there diagonal effects (industry-wide shifts affecting a particular calendar year)? Are any accident years developing differently from historical patterns? Claude examines the triangle data systematically and flags the issues that require actuarial judgment before the calculation proceeds.
- "Analyze this commercial auto liability loss development triangle and identify anomalies. Cumulative paid losses ($000s) by accident year and development period (12, 24, 36, 48, 60 months): AY 2020: 4,820 / 7,105 / 8,890 / 9,445 / 9,680 AY 2021: 5,340 / 8,210 / 10,450 / 11,820 / — AY 2022: 6,180 / 9,450 / 12,340 / — / — AY 2023: 7,290 / 12,650 / — / — / — AY 2024: 8,840 / — / — / — / — Calculate: (1) age-to-age factors for each development period, (2) volume-weighted average, 3-year weighted average, and 5-year weighted average factors, (3) selected development factors with rationale for each selection, (4) flag any anomalies — specifically: AY 2023's 12-to-24 development appears significantly higher than prior years; analyze whether this represents a true shift in loss emergence or a data anomaly, (5) cumulative development factors (tail factor: use 1.015 for 60-to-ultimate)."
- "Workers compensation IBNR calculation using chain-ladder method. I will provide the incurred loss triangle (paid + case reserves). Incurred losses ($000s): AY 2019: 12,450 / 15,890 / 17,240 / 17,890 / 18,120 / 18,230 AY 2020: 11,200 / 14,650 / 16,340 / 17,100 / 17,480 / — AY 2021: 13,800 / 17,450 / 19,650 / 20,420 / — / — AY 2022: 15,600 / 20,100 / 22,450 / — / — / — AY 2023: 17,200 / 21,890 / — / — / — / — AY 2024: 19,400 / — / — / — / — / — Development periods: 12, 24, 36, 48, 60, 72 months. Tail factor: 1.020 (72 to ultimate, based on industry benchmarks for workers comp medical). Calculate: (1) selected development factors (justify each), (2) cumulative development factors, (3) projected ultimate losses by accident year, (4) IBNR by accident year (projected ultimate minus current diagonal), (5) total IBNR reserve needed, (6) compare IBNR as % of earned premium if AY earned premiums were: 2019: $28M, 2020: $26M, 2021: $32M, 2022: $36M, 2023: $40M, 2024: $45M."
- "Diagnostic review of a medical professional liability triangle. This line is known for long-tail development and social inflation effects. Paid losses ($000s): AY 2018: 1,240 / 3,890 / 7,450 / 11,200 / 14,890 / 17,340 / 18,650 AY 2019: 1,580 / 4,340 / 8,120 / 12,650 / 16,780 / 19,450 / — AY 2020: 1,890 / 5,120 / 9,840 / 15,340 / 20,450 / — / — AY 2021: 2,240 / 6,450 / 12,890 / 19,780 / — / — / — AY 2022: 2,890 / 7,890 / 15,450 / — / — / — / — AY 2023: 3,450 / 9,120 / — / — / — / — / — AY 2024: 4,100 / — / — / — / — / — / — Perform: (1) diagonal analysis — calculate calendar year paid loss totals and identify if there's a social inflation signal in recent diagonals, (2) Bornhuetter-Ferguson method setup: to apply B-F you need an a priori loss ratio assumption — recommend what a priori to use for AY 2023 and 2024 given the long-tail development pattern, (3) compare chain-ladder ultimates to B-F ultimates for AY 2022–2024 — which is more credible for the most recent years and why?"
Alternative Reserve Methods
Good actuarial practice uses multiple methods and explains why the selected estimate differs from any individual method's result. The Bornhuetter-Ferguson method is especially important for immature accident years where the chain-ladder can be unstable. The Cape Cod method is useful when you have limited data history. Claude applies multiple methods simultaneously and produces the actuarial summary table comparing them.
- "Apply three reserve methods to this general liability triangle and produce a summary comparison table. Incurred losses ($000s): AY 2022: 8,450 / 12,340 / 14,890 / 15,780 AY 2023: 9,120 / 13,890 / 16,450 / — AY 2024: 10,340 / 15,120 / — / — AY 2025: 12,450 / — / — / — Development periods: 12, 24, 36, 48 months. Tail factor: 1.025. A priori loss ratios (for B-F and Cape Cod): AY 2022 65%, AY 2023 68%, AY 2024 70%, AY 2025 72%. Earned premiums: AY 2022 $18M, AY 2023 $20M, AY 2024 $22M, AY 2025 $25M. Apply: (1) Chain-Ladder method, (2) Bornhuetter-Ferguson method, (3) Cape Cod method. For each method: show projected ultimate losses by AY and total IBNR. Summary table: show three method ultimates side by side by AY, percentage difference from chain-ladder, and your recommended selected ultimate with rationale."
- "Frequency-severity IBNR method for personal auto bodily injury: Instead of developing aggregate losses, separate into claim count and severity development. Claim count triangle (reported claims): AY 2022: 842 / 910 / 935 / 944 AY 2023: 908 / 985 / 1,012 / — AY 2024: 975 / 1,065 / — / — AY 2025: 1,040 / — / — / — Average paid severity triangle ($ per closed claim): AY 2022: $14,200 / $18,450 / $21,890 / $23,400 AY 2023: $15,100 / $19,780 / $23,450 / — AY 2024: $16,800 / $21,450 / — / — AY 2025: $18,200 / — / — / — Apply chain-ladder to both count and severity triangles separately. Calculate projected ultimate claim counts and average severity by AY. Calculate projected ultimate losses = count × severity. Compare to aggregate dollar triangle development. Which approach gives more stable ultimates for the most recent accident years?"
Tail Factor Selection and Long-Tail Lines
The tail factor — the development factor from the last observed period to ultimate — has a disproportionate impact on long-tail lines. For workers compensation, medical professional liability, and general liability, the tail can be 10-20% of total ultimate losses. Getting it wrong by even a few points compounds across all open accident years. Claude evaluates tail factor alternatives including industry benchmarks, exposure-based approaches, and curve-fitting methods, and documents the selection rationale.
- "Tail factor selection for workers compensation indemnity: Our triangle is credible to 84 months of development. After 84 months, the portfolio has minimal history. Available reference points: (1) NCCI Industry Development Factors for workers comp indemnity show 84-to-ultimate factors ranging from 1.018 to 1.025 across states; our primary state is Texas (NCCI factor 1.021). (2) Internal analysis: we have 3 accident years with 84+ months of development, showing actual development of 1.019, 1.023, and 1.016 for the 84-96 month period, and 1.004 cumulative thereafter. (3) Actuarial consultants have historically selected 1.018 for our book. Evaluate the three approaches, recommend a tail factor and document the selection rationale in the format required for the actuarial memorandum. Also: at 1% difference in tail factor, what is the dollar impact on a $180M IBNR estimate?"
Actuarial Reserve Memorandum
The reserve opinion memorandum — whether it's the Statement of Actuarial Opinion required for Schedule P, a management review memorandum, or an external audit support document — requires specific content under ASOP No. 36 and relevant regulatory standards. It must identify the actuary's responsibility, describe the scope of the review, explain the methods and assumptions used, state any significant limitations, and express an opinion on reserve adequacy. Claude drafts the memorandum from the analysis inputs.
- "Draft an actuarial reserve opinion memorandum for a $280M P&C insurance company. Summary of reserve analysis: Year-end December 31, 2025. Lines of business: commercial auto liability ($95M carried reserves), general liability ($112M), workers compensation ($73M). Total carried reserves: $280M. Actuarial indication: $272M (selected ultimates less paid losses to date). Indicated reserve is $8M below carried — management chose to maintain carried at $280M for prudence margin. IBNR as % of total reserves: 45%. Significant uncertainty: GL Social inflation uncertainty — CY 2023–2024 diagonal shows elevated severity vs. prior trend. Draft a complete ASOP No. 36 compliant Statement of Actuarial Opinion for the year-end 2025 reserve review, including: scope of review, methods used by line of business, significant limitations, opinion on reserve adequacy, actuarial central estimate vs. carried, and the IRIS ratio context."
- "Management reserve review memo for audit committee: Summarize the Q3 2025 quarterly reserve review in plain-English format for the audit committee. They are not actuaries. Key findings: total carried reserves $280M, actuarial indication $268M, management is $12M above indication (margin of 4.5%). The favorable development this quarter: $6.2M. The unfavorable development: $2.1M on GL due to late-reported large claims. Reserve adequacy: we believe reserves are adequate based on the current analysis. Write a 2-page executive summary covering: (1) total reserve position and change from Q2, (2) what drove favorable development, (3) what are the key uncertainties (social inflation, reinsurance recoverables, large open claims), (4) whether the margin above indication is appropriate, (5) significant changes in methodology from prior quarter (if any)."
Schedule P Analysis
Schedule P — the statutory loss development exhibit — is the primary regulatory disclosure of a P&C insurer's reserve adequacy. Analysts, rating agencies, and regulators use it to assess reserve development trends. Reviewing Schedule P data for a company (or a competitor) requires understanding which lines are developing favorably vs. adversely, whether favorable development is structural or temporary, and what the IRIS reserve development ratios indicate. Claude reads Schedule P data and produces the analysis.
- "Schedule P Part 2D analysis — homeowners/farmowners line: Prior year reserves carried at 12/31/2024: $145M. Current year development (2025 Schedule P column): $18.2M favorable development on 2024 and prior accident years. Key question: is this favorable development sustainable? Historical Schedule P Part 2 data shows: 2022 12-month reserves $135M, subsequent development favorable $12.4M (9.2%); 2023 reserves $140M, subsequent development favorable $8.1M (5.8%); 2024 reserves $145M. Analyze: (1) development trend — is the favorable development accelerating or decelerating? (2) what IRIS ratio 12 (2-year development as % of surplus) does this imply assuming $420M surplus? (3) is a homeowners book showing consistent favorable development more likely to reflect conservative initial reserving, mix shift toward less complex claims, or claims settlement acceleration? (4) what questions would a rating agency analyst ask about this development pattern?"
Where to Start
ClaudeFinLab's Insurance & Actuarial category has seven templates including the Actuarial Reserve Adequacy Reviewer, which is built specifically for the quarterly reserve review workflow. For actuaries running their annual reserve study, start by pasting the chain-ladder calculation prompt with your actual triangle data — Claude produces the factor selection, IBNR estimate, and method comparison in a format ready to carry into the memorandum. The reserve memorandum template then takes those numbers and produces the ASOP-compliant narrative. If you're reviewing a competitor or acquisition target's Schedule P, the Schedule P Analysis template structures the diagnostic review. All templates work in Claude.ai Pro — no installation required for SKILL.md templates.