Report Year

Study guide: Exam 5

Report Year (RY) is a data aggregation method that groups claims by the calendar year in which the claim was first reported to the insurer, regardless of when the loss occurred.

$\text{Report lag} = \text{Report date} - \text{Accident date}$

RY is the correct basis for claims-made coverage, because a claims-made policy's trigger is the report date: every claim reported during the policy period is covered by that policy, so the report year cohort and the coverage cohort coincide.
A report year closes to new claims at year end — pure IBNR is zero by construction, since every claim in the cohort is already known. All remaining development is IBNER: re-estimation of claims already reported.
That makes RY triangles develop faster and more smoothly than Accident Year triangles, and it makes RY the natural basis for severity studies: the claim count denominator is fixed, so an average-severity triangle is not distorted by late-emerging claims.
The cost is that a report year is not a homogeneous exposure period. RY 2024 contains old accidents reported late alongside fresh ones; if the reporting pattern shifts, the mix of accident dates inside the cohort shifts with it, and trend measured on an RY basis is contaminated by the change.
For occurrence coverage, report year cannot be matched to earned premium at all — the premium was earned in the accident year, not the report year — so occurrence ratemaking uses accident year and reserves for the unreported claims.

Report Year vs. Accident Year Under Each Coverage Trigger A general liability loss occurs $11/15/2023$ and is reported $2/3/2024$ . The insured held an occurrence policy effective $6/1/2023$ – $5/31/2024$ and, separately, a claims-made policy renewing each $1/1$ .

Which policy year responds, and which cohort does the claim join?

Answer Accident year 2023, report year 2024.

Under the occurrence policy, the $6/1/2023$ policy responds — the trigger is the $11/15/2023$ occurrence date. The claim is part of AY 2023, and at $12/31/2023$ it sat in that year's IBNR.

Under claims-made, the policy in force on $2/3/2024$ responds — the $2024$ policy — provided the accident date is on or after the retroactive date. The claim belongs to RY 2024, and the $2024$ policy's premium was collected to pay it.

The same claim is therefore priced from AY 2023 data on one contract and RY 2024 data on the other.

Why Severities Are Studied on a Report Year Basis For accident year 2024 at $12$ months, $820$ claims are reported with $\$6{,}150{,}000$ of reported losses. By $24$ months, $940$ claims are reported with $\$8{,}500{,}000$ . The $120$ newly reported claims are small, late-emerging claims averaging $\$4{,}000$ .

What happens to the accident-year average severity, and how does a report-year view fix it?
Answer
 $\begin{align*} \text{Avg severity at 12 mo} &= \frac{\$6{,}150{,}000}{820} = \$7{,}500 \\[4pt] \text{Avg severity at 24 mo} &= \frac{\$8{,}500{,}000}{940} = \$9{,}043 \end{align*}$ 
Severity rose $20.6\%$ , but two different things are mixed into that number: genuine development on the original $820$ claims, and dilution from $120$ cheap new claims.

Isolating the original cohort:

$\frac{\$8{,}500{,}000 - 120 \times \$4{,}000}{820} = \frac{\$8{,}020{,}000}{820} = \$9{,}780$

On a report year basis the denominator is frozen at the claims reported in the year, so the triangle measures IBNER on a fixed set of claims and gives the $\$9{,}780$ answer directly. This is why frequency-severity analyses of average severity favour report year, while the frequency side stays on accident year.