Why Period Basis Matters
An unpaid claim liability is an estimate of amounts still owed for claims, but it can be grouped in more than one way. A claim may arise from an event in one calendar year, be covered by a policy that began in another year, and first be reported in a third year. The period basis determines which of those dates controls its grouping.
These classifications answer different questions. Accident-period grouping organizes claims by when the insured event occurred. Policy-period grouping organizes them by the policy cohort that provides coverage. Report-period grouping organizes them by when the insurer first received notice. The classification does not, by itself, change the estimated amount owed on a claim.
The date that determines the responding policy depends on the coverage contract. For an occurrence form, the occurrence date is central to determining which policy applies. For a claims-made form, the coverage trigger generally depends on the claim being made or reported under the contract’s terms, subject to any applicable reporting provisions. Once the responding policy has been established, the claim can be assigned to its policy period. Do not infer coverage solely from the policy effective date if the problem supplies a coverage determination or contract rule.
For the calculations in this tutorial, assume each listed claim has already been determined to be covered by one identified policy, each claim is counted once, and every unpaid amount is measured at the same stated valuation date. Let \(U_j\) denote the unpaid liability assigned to claim \(j\). A period total is the sum of \(U_j\) for claims whose relevant date falls in that period.
Here \(B\) denotes the chosen basis, such as accident, policy, or report; \(D_j^{(B)}\) is the date used for claim \(j\) under that basis; and \(p\) is a particular period. The formula is a grouping rule, not a separate claim valuation method.
Assigning Claims to Periods
A reliable classification starts with the claim-level record. Preserve the occurrence date, the responding policy and its effective date, the first report date, and the unpaid amount. Then select one basis and group by the corresponding date. Keep the basis explicit in schedules and summaries: a heading such as “2025 claims” is ambiguous unless it says whether 2025 means accident year, policy year, or report year.
Use the established coverage determination and identify the policy that responds. Do not substitute the report date for a coverage decision.
For accident period, use the occurrence date; for policy period, use the specified policy-period convention; for report period, use the first-notice date.
Place each claim in exactly one period for the selected basis, then add its unpaid amount to that period’s total.
When the same complete claim inventory and unpaid amounts are used, the sum across all periods must equal the total unpaid liability under every basis.
Worked Example: One Inventory, Three Period Bases
Method and assumptions. This is a claim-level grouping exercise. A hypothetical insurer has five covered claims, with coverage already determined. All unpaid amounts are in dollars and measured at one common valuation date. Assign each claim to calendar-year accident, policy, and report periods, then calculate the totals. The method avoids treating the three dates as interchangeable.
| Claim | Occurrence date | Responding policy effective date | First report date | Unpaid amount |
|---|---|---|---|---|
| A | December 28, 2023 | July 1, 2023 | January 10, 2024 | $12,000 |
| B | January 15, 2024 | January 1, 2024 | February 1, 2024 | $8,000 |
| C | December 30, 2023 | July 1, 2023 | March 15, 2024 | $5,000 |
| D | February 2, 2024 | January 1, 2024 | December 20, 2024 | $9,000 |
| E | November 20, 2023 | January 1, 2023 | November 25, 2023 | $6,000 |
Accident periods. Claims A, C, and E occurred in 2023, so the 2023 accident-period total is \(12{,}000+5{,}000+6{,}000=\$23{,}000\). Claims B and D occurred in 2024, giving a 2024 total of \(8{,}000+9{,}000=\$17{,}000\).
Policy periods. Using the stated effective-date convention, policies effective in 2023 cover claims A, C, and E for this classification. Their policy-period total is \(12{,}000+5{,}000+6{,}000=\$23{,}000\). Claims B and D are assigned to policies effective in 2024, totaling \(8{,}000+9{,}000=\$17{,}000\).
Report periods. Claim E was first reported in 2023, so the 2023 report-period total is \(\$6{,}000\). Claims A, B, C, and D were first reported in 2024, totaling \(12{,}000+8{,}000+5{,}000+9{,}000=\$34{,}000\).
Check. Each basis gives the same overall liability: \(\$23{,}000+\$17{,}000=\$40{,}000\) for accident and policy periods, and \(\$6{,}000+\$34{,}000=\$40{,}000\) for report periods. The composition differs because the dates differ.
Common mistake avoided. Assigning claims to accident year using the report date would incorrectly produce 2023 of \(\$6{,}000\) and 2024 of \(\$34{,}000\). Those are the report-period totals, not the accident-period totals of \(\$23{,}000\) and \(\$17{,}000\).
Why the Overall Total Reconciles
The invariance of the total follows from the fact that each basis partitions the same claim inventory. A partition places every claim in exactly one group without omitting or duplicating it. Under accident basis, the groups are formed using occurrence dates; under report basis, they are formed using first-report dates. The membership changes, but the set of claim amounts being added does not.
To justify the result, let \(J\) be the complete set of claims and let \(J_p^{(B)}\) be the claims assigned to period \(p\) under basis \(B\). Because the groups form a partition, every claim in \(J\) occurs in exactly one \(J_p^{(B)}\). Therefore, summing the period totals simply rearranges the terms in the claim-level sum:
This equality does not mean the period-level results are interchangeable. A report-period total describes the liabilities attached to claims reported during that period; it does not necessarily describe the events that occurred during it. Similarly, a policy-period total reflects the selected policy cohort, not necessarily the dates when claims were reported or accidents occurred.
Worked Example: A Late-Reported Claim and Its Policy Cohort
Method and assumptions. Consider three hypothetical covered claims under occurrence coverage. The responding policy is supplied for each claim, and policy period is classified by that policy’s effective calendar year. All unpaid amounts are measured on the same valuation date. Calculate the accident-, policy-, and report-period totals. This method avoids moving an occurrence claim into a later policy cohort merely because notice arrived later.
| Claim | Occurrence date | Policy effective date | First report date | Unpaid amount |
|---|---|---|---|---|
| F | November 18, 2022 | January 1, 2022 | April 5, 2023 | $14,000 |
| G | December 20, 2022 | July 1, 2022 | December 28, 2022 | $7,000 |
| H | January 9, 2023 | January 1, 2023 | February 3, 2023 | $11,000 |
Accident-period calculation. F and G occurred in 2022: \(14{,}000+7{,}000=\$21{,}000\). H occurred in 2023, giving \(\$11{,}000\).
Policy-period calculation. F and G have responding policies effective in 2022, so the 2022 policy-period total is \(14{,}000+7{,}000=\$21{,}000\). H belongs to the 2023 policy period, totaling \(\$11{,}000\).
Report-period calculation. G was reported in 2022, for \(\$7{,}000\). F and H were reported in 2023, for \(14{,}000+11{,}000=\$25{,}000\).
Check and interpretation. The totals reconcile to \(\$32{,}000\): accident periods \(21{,}000+11{,}000\), policy periods \(21{,}000+11{,}000\), and report periods \(7{,}000+25{,}000\). F remains in the 2022 accident and policy periods even though it was reported in 2023. The 2023 report-period total is larger because it includes that late-reported 2022 occurrence.
Common mistake avoided. Putting F in the 2023 accident period because it was first reported in 2023 confuses the date of occurrence with the date of notice. Under the stated facts, F is a 2022 accident-period claim and a 2022 policy-period claim, but a 2023 report-period claim.
Interpreting the Basis and Avoiding Errors
The most useful basis depends on the question being asked and on the information available. Accident-period views connect unpaid claims to the timing of covered events. Policy-period views connect liabilities to the policies that provide coverage and can be useful when reviewing policy cohorts. Report-period views describe the claims entering the insurer’s records during each period and can be useful for monitoring reporting patterns. None of these descriptions alone says that claims in one group are more or less complete than claims in another.
A period label should state both the basis and the interval. “2024” could mean accident year 2024, policy year 2024, or report year 2024. A summary that leaves the basis unstated is difficult to interpret and can lead to invalid comparisons. Also verify that claim amounts share the same valuation basis before comparing or summing them; otherwise, differences may reflect inconsistent measurement as well as different grouping.
Worked Example: Reconciling a Report-Year Summary
Method and assumptions. A hypothetical claim inventory contains the following five unpaid amounts, all measured at one common valuation date. Coverage and the responding policy have already been established. Calculate report-year totals and compare the 2025 report total with the accident-year total for accidents occurring in 2025.
| Claim | Occurrence date | First report date | Unpaid amount |
|---|---|---|---|
| J | December 30, 2024 | January 5, 2025 | $18,000 |
| K | January 2, 2025 | January 10, 2025 | $12,000 |
| L | December 20, 2024 | December 28, 2024 | $10,000 |
| M | December 31, 2024 | February 2, 2025 | $5,000 |
| N | January 15, 2025 | December 30, 2025 | $15,000 |
Report-year totals. L was reported in 2024, so report year 2024 totals \(\$10{,}000\). J, K, M, and N were reported in 2025, giving \(18{,}000+12{,}000+5{,}000+15{,}000=\$50{,}000\).
Accident-year 2025 total. Only K and N occurred in 2025. Thus, the 2025 accident-year total is \(12{,}000+15{,}000=\$27{,}000\). J and M occurred in 2024 even though they were reported in 2025; together they contribute \(18{,}000+5{,}000=\$23{,}000\) to the 2025 report total.
Check. The report-year aggregate is \(10{,}000+50{,}000=\$60{,}000\). The accident-year aggregate is \(18{,}000+10{,}000+5{,}000=\$33{,}000\) for 2024 and \(\$27{,}000\) for 2025, also \(\$60{,}000\). The difference between the 2025 report total and 2025 accident total is \(50{,}000-27{,}000=\$23{,}000\), exactly the unpaid amounts for J and M.
Common mistake avoided. It would be incorrect to interpret the \(\$50{,}000\) report-year 2025 amount as the liability from accidents occurring in 2025. The claim-level dates show that \(\$23{,}000\) of it is attached to accidents in 2024.
Common Period-Basis Traps
- Using the wrong date. A first-report date determines report-period grouping, not accident-period grouping.
- Assuming the reporting policy is automatically the policy that responds. Coverage must be determined using the contract terms. A reporting date may matter to a claims-made trigger, but that does not justify assigning every claim to the policy in force on its report date.
- Mixing bases in one table. A total grouped by accident year should not be added to a total grouped by report year as though the groups were disjoint.
- Dropping or duplicating claims. The aggregate reconciliation works only when the inventory is complete and each claim appears exactly once within a given basis.
- Reading a changed period total as a changed liability. Reclassification moves claim amounts between groups. Unless the underlying amounts or inventory also change, it does not change the aggregate.
Check Your Understanding
For each problem, state the relevant period basis before calculating. Assume all listed amounts are unpaid liabilities measured on a common valuation date and each claim is included once.
- A claim occurred on December 29, 2024, was covered by a policy effective July 1, 2024, and was first reported on January 8, 2025. To which calendar years is it assigned under accident, policy, and report bases? (Answer: accident year 2024; policy year 2024; report year 2025.)
- Two claims have unpaid amounts of $4,500 and $6,500. Both occurred in 2023; the first was reported in 2023 and the second in 2024. What are the 2023 and 2024 report-period totals, and what is the accident-period total for 2023? (Answer: report year 2023 $4,500; report year 2024 $6,500; accident year 2023 $11,000.)
- A claim of $9,000 occurred in 2022 and was reported in 2023. A second claim of $13,000 occurred and was reported in 2023. Calculate the 2023 report-period total and the 2023 accident-period total. (Answer: report year 2023 $22,000; accident year 2023 $13,000.)
- A complete inventory has unpaid amounts of $8,000 and $12,000 assigned to accident year 2024, and $7,000 assigned to accident year 2025. Under report basis, the first two claims are in 2025 and the third is in 2024. Calculate report-year totals and check the aggregate against accident-year totals. (Answer: report year 2024 $7,000; report year 2025 $20,000; both bases total $27,000.)
- A hypothetical book reports $31,000 for 2025 report year, including $11,000 for claims from 2024 accidents and $8,000 for claims from 2025 accidents. The remaining amount is also from 2025 accidents. What is the 2025 accident-year total represented in this report-year amount, and how much of the report total is from other accident years? (Answer: 2025 accident-year amount $20,000; amount from other accident years $11,000.)