Reserve review

Triangles reconciled to the ledger, last quarter's selections tested against emergence, young years on expected loss ratios rebuilt from rate and trend. Indicated against carried, every selection explained

Customize this Agent

XLSXDOCX16 steps

The problem

Commercial auto liability's 2025 reserve shortfall: $2.0 billion, mostly accident years 2023 and 2024

The newest accident years rest on an expected loss ratio: in Bornhuetter Ferguson often the plan, last year's ratio moved by rate change and an assumed trend; in Cape Cod, older years brought forward on that same trend. When severity outruns the trend, both call the young years adequate until they develop. That is where commercial auto liability's 2025 deficiency sat

The request

Run our Q3 reserve review at 9/30/26: commercial auto liability and general liability, about $700 million of net earned premium. Our data and the Q2 review are attached. Reconcile to the ledger first. Then actual versus expected since Q2, diagnostics, Berquist Sherman where case adequacy or closing speed shifted, paid and incurred chain ladder, Bornhuetter Ferguson and Cape Cod by accident year. Rebuild the expected loss ratios for 2023 to 2026 from our rate index and loss trends fitted to our own data, not from plan, and benchmark them against peers' filed trends and development. Select ultimates, show IBNR against carried by line and accident year with a range, and write the committee memo on what moved since Q2.

CA and GL triangles 9.30.26.xlsx
Large loss listing 9.30.26.xlsx
Earned premium and rate index Q3 26.xlsx
Carried reserves and ledger 9.30.26.xlsx
Q2 26 reserve review.xlsx

How Axe works

ApprovedChanges asked

Your reviewApprove, or say what to change

Approved or changes

Workbook and committee memoWorkbook, memo and peer review

Final reviewApprove the revised files

Revised workbook and committee memoRevised files and peer review

Step

Rebuild the expected loss ratios

Prompt

Sent to the model at this step

The youngest accident years carry most of the IBNR and rest on the expected loss ratio, so build it from our own experience, not from plan. For each line, take every accident year mature enough for its chain ladder ultimate to be credible. Bring its earned premium to current rate level with the rate index, by the parallelogram method for annual policies. Cap its ultimate at the large loss threshold and load large losses back at their long run share. Develop reported counts to ultimate by accident year. Fit annual severity and frequency trends to capped ultimate severities and to ultimate counts per earned exposure, on the adjusted data where the diagnostics found a shift, over several windows, with and without the latest year, and select each with the reason; set closed claim severity by calendar year beside them as a check. Trend each mature year's capped ultimate from its average accident date to that of each target year, 2023 to 2026, net of any premium trend in the exposure base, such as payroll or sales in general liability. Each mature year then gives an indication for each target year: select the expected loss ratio from their weighted average and spread, and set the Cape Cod ratio on the same basis beside it. Restate each target year's ratio to that year's own rate level by multiplying by its on level factor, so it applies to its earned premium and compares with plan and the Q2 a priori. Then benchmark without replacing our data: the fitted severity trend against the loss trends commercial auto and general liability writers select in their latest rate filings on SERFF Filing Access and the incurred severity per reported claim in peers' 10-K development tables, with the BLS CPI for medical care services only for the medical part of bodily injury; the rate index against the renewal rate change peers report for Q2 2026, with the CIAB premium change by line shown as premium, not rate; and our view of 2023 and 2024 against peers' prior year development in their Q2 2026 10-Q filings and AM Best's 13 July 2026 report on 2025 results, which put most of commercial auto liability's further $2.0 billion reserve deficiency in accident years 2023 and 2024. Set each selected ratio beside the plan and the Q2 a priori, with the difference in points and what drives it.

Uses

Valuation date
Book
Premium and rate index
Large losses
Prior review
Reconciled data
Diagnostics
Adjusted triangles
Chain ladder

Produces

Expected loss ratios
Benchmarks

What you get back

XLSXReserve review workbookReconciliation, triangles, actual versus expected, diagnostics, methods, selections, indicated against carried, movement, range

DOCXReserve committee memoIndicated against carried and range, movement since Q2, drivers, expected loss ratios against plan

Peer reviewOpen items first, then each figure marked checked or corrected, with its source

Make it yours

Change the lines, the methods you weight, your large loss threshold or the memo your committee reads. Every step and every prompt can be edited

Customize this Agent

Opens a private copy of this Agent in your workspace, ready to adapt to your own sources

More agent examples

Capital markets analystsBroker opinion of valuePPTXXLSX16 steps

Financial planning analystsBudget variance analysisXLSXDOCX13 steps

Credit research analystsCapital structure analysisXLSX12 steps

Life sciences analystsClinical trial landscapeXLSXDOCX14 steps

Investment banking analystsComparable company analysisXLSXDOCX11 steps