The question behind this project

How much will an insurer still need to pay?

An accident happens today, but its bills can arrive for years. We look at past payments to estimate how much money an insurer needs to set aside for what comes next.

Read the analysis
Painting of an analyst reviewing insurance claim folders, damage photographs, repair invoices, and payment records.

Insurance An analytical case study

Estimating outstanding insurance claim payments

How much remains to be paid, and how sensitive is that estimate to the reserving method?

Insurance records and simulated claims

The finding

The newest accident years contain the least observed development. Their ultimate cost depends most on the projection assumptions.

Analytical work

Translate loss triangles into an explanation of unpaid obligations, model sensitivity, and profitability assumptions.

FIG. 01Retained evidence

Claims payments made and estimated payments remaining

Year1510
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997

Development period →

36.1%

of projected ultimate losses already paid

1997 · 1 observed periods
Paid · 36.1%Projected unpaid · 63.9%
Paid Chain-Ladder · all API companies and lines · 1988–1997. Newer years have less observed development. Projected unpaid losses include case reserves and IBNR; this is not an uncertainty interval. API snapshot · 2026-09-05.

01

Read across one accident year

A row groups claims from the same accident year. Moving right adds development time, not new accident years. At the valuation date, the missing corner is unobserved future development; it is not zero loss.

02

Separate what is paid from what is estimated

Paid losses are cash already disbursed. Incurred losses include paid amounts and case reserves. Ultimate losses are projected final costs. Ultimate minus paid is the total unpaid reserve; ultimate minus incurred is the additional provision beyond reported losses.

03

Compare the assumptions, then the estimates

Chain-ladder extends observed development patterns. Bornhuetter–Ferguson combines emergence patterns with an a priori expected loss ratio, giving immature years another anchor. Their difference is model sensitivity, not a confidence interval.

Decision supported

What to do with this evidence

Review the immature years and long-tail lines before using a point estimate for reserve or pricing discussions. Compare methods on the same paid or incurred basis.

What this analysis cannot establish

  • Claim-level severity, frequency, and reporting delays are synthetic demonstrations calibrated to aggregates; they are not observed policyholder records.
  • Combined ratios assume a 30% expense ratio. Profitability conclusions change with that assumption; no stochastic reserve interval is estimated.

Continue the investigation

Inspect the evidence behind the story.