Why a Claim’s Dates Must Be Kept Distinct
A claim record can contain several dates, and each may answer a different question. The loss date identifies when the covered event occurred; the policy-effective date identifies the policy period; the report date identifies when the insurer first received notice; a payment date identifies when money was paid; and the valuation date identifies the snapshot date for the data. Treating these dates as interchangeable can put a claim in the wrong triangle row, put a payment in the wrong calendar period, or include a transaction that had not occurred by the valuation date.
The central rule is to choose the date that matches the triangle’s stated origin basis, then use the appropriate transaction and valuation dates for the cell. A date can be important without determining the row. For example, the payment date determines when a payment belongs in calendar time, but it does not by itself determine the accident-year row of an accident-year triangle.
What Each Date Does
| Date field | Typical role | Does it normally determine the origin row? |
|---|---|---|
| Loss date | When the loss event occurred; used for accident-year origin | Yes, for an accident-year triangle |
| Policy-effective date | When the policy term began; used for policy-year origin when that is the stated convention | Yes, for a policy-year triangle |
| First report date | When the insurer first received notice; used for report-year origin | Yes, for a report-year triangle |
| Payment date | When a payment transaction occurred; used for its calendar period and valuation inclusion | No, not by itself |
| Valuation date | The as-of date for the snapshot of claim information | No |
A triangle’s origin basis must be stated before assigning records. The same claim can belong to different origin years in different triangles, because its loss, policy-effective, and first report dates need not fall in the same year. For a given triangle, however, the rule is consistent: use its designated origin date for the row. The development age or cell position is then determined under the triangle’s period convention; the payment date does not replace the origin date.
For a paid triangle, payment transactions are included only if their payment dates are on or before the valuation date. A claim can therefore have an assigned origin row even when it has no payment in the valuation snapshot. Conversely, a payment made by the valuation date contributes only to the row determined by the triangle’s origin basis. The record should not be reassigned to a new origin row merely because a later payment occurs.
Formal Assignment Rule
Suppose claim \(c\) has an origin date \(o(c)\), and the triangle uses origin periods such as calendar years. Let \(r(c)\) be the period containing \(o(c)\). For a paid triangle, let payment transaction \(j\) have amount \(p_j\) and payment date \(t_j\). At valuation date \(V\), its contribution to origin row \(r\) is:
The “otherwise” case is essential: the contribution is zero both when the payment is after valuation and when the claim belongs to a different origin row. For a cell that also specifies a calendar period, include the additional condition that the payment date falls in that calendar period.
For a cumulative paid amount in row \(r\), sum the eligible transactions for claims assigned to that row:
where \(\mathbf{1}\{\cdot\}\) equals 1 when all conditions inside the braces hold and 0 otherwise. If a cell is restricted to payment calendar period \(k\), the indicator also requires that \(t_j\) fall in period \(k\). This expression describes assignment and inclusion; it does not prescribe how the triangle’s claim amounts are otherwise defined.
Worked Examples
Worked Example: One Claim, Three Possible Origin Years
A hypothetical insurer records a claim with policy-effective date October 1, 2022, loss date December 29, 2023, first report date January 8, 2024, and a payment of $7,500 on March 15, 2024. The data are valued at December 31, 2024. Determine the claim’s origin year under three triangle definitions and identify the payment’s calendar year and valuation status.
Step 1: Use the correct date for each origin definition. For an accident-year triangle, use the loss date, December 29, 2023, so the origin year is 2023. For a policy-year triangle under the policy-effective-year convention, use October 1, 2022, so the origin year is 2022. For a report-year triangle, use the first report date, January 8, 2024, so the origin year is 2024.
Step 2: Assign the payment to calendar time. The payment occurred March 15, 2024, which is in calendar year 2024. The payment date does not change any of the three origin years.
Step 3: Apply the valuation cutoff. March 15, 2024, is on or before December 31, 2024. Thus the $7,500 payment is included in the snapshot. It contributes to the 2023 accident-year row in an accident-year triangle, the 2022 policy-year row in a policy-year triangle, or the 2024 report-year row in a report-year triangle, subject to the cell’s calendar-period definition.
Check. The payment is included once under any one specified origin basis, and its payment calendar year is 2024. Assigning the claim to 2024 solely because it was paid in 2024 would be wrong for both the accident-year and policy-year triangles. The method avoids that origin-date substitution.
Worked Example: Applying a Valuation Cutoff to Multiple Payments
A hypothetical claim has policy-effective date July 1, 2023, loss date May 20, 2024, and first report date June 2, 2024. It has two payments: $12,000 on September 10, 2024, and $3,000 on February 1, 2025. Find its accident-year, policy-year, and report-year origins, then calculate its cumulative paid amount at December 31, 2024.
Step 1: Assign the origin years. The loss date is in 2024, so the accident-year origin is 2024. The policy-effective date is in 2023, so the policy-year origin is 2023 under the stated convention. The first report date is in 2024, so the report-year origin is 2024.
Step 2: Test each payment against valuation. The September 10, 2024 payment is on or before December 31, 2024, so it contributes $12,000. The February 1, 2025 payment is after the valuation date, so it contributes $0 to this valuation snapshot.
Step 3: State the row-specific amount. At the December 31, 2024 valuation, cumulative paid is $12,000 in the 2024 accident-year row, the 2023 policy-year row, or the 2024 report-year row, depending on which triangle is being constructed. The later $3,000 payment is not included in any of these rows at this valuation; at a later valuation it may be included, still under the same claim origin row.
Check. The included amount, $12,000, is less than the eventual two-payment total of $15,000 by exactly the post-valuation payment of $3,000. The method avoids treating a payment after valuation as if it were already in the snapshot.
Worked Example: A Boundary Date and a Different Claim’s Row
At a December 31, 2023 valuation, a hypothetical paid triangle has two claims, both assigned by loss year. Claim A has a loss date of December 31, 2022 and a $4,000 payment dated December 31, 2023. Claim B has a loss date of January 1, 2023 and a $2,500 payment dated December 30, 2023. Determine each claim’s origin year and contribution to the 2022 and 2023 rows at this valuation.
Step 1: Assign origin years from loss dates. Claim A’s loss date is in 2022, so \(r(A)=2022\). Claim B’s loss date is in 2023, so \(r(B)=2023\). The claims fall on opposite sides of the calendar-year boundary even though their payments are both made by the valuation date.
Step 2: Check payment eligibility. Both payment dates are on or before December 31, 2023. Therefore, neither payment is excluded by the valuation cutoff.
Step 3: Apply the row condition as well as the cutoff. Claim A contributes $4,000 to row 2022 and $0 to row 2023. Claim B contributes $0 to row 2022 and $2,500 to row 2023. Each claim’s contribution to the other row is zero because its origin year does not match, even though its payment date is before valuation.
Check. The row amounts sum to $6,500, equal to the two payments, and neither payment is counted in both rows. The method avoids the incomplete rule “include any payment before valuation” by also requiring a match to the row’s origin period.
Why the Assignment Rule Works
A triangle cell is defined by more than the fact that a transaction exists. It is restricted by the chosen origin period and, for a paid measure, by the transaction’s timing relative to valuation. A payment can contribute to a row only when both conditions hold: the claim’s designated origin date falls in that row, and the payment has occurred by the valuation date. If either condition fails, the payment is not part of that row’s valued amount.
This can be seen directly by considering a transaction in two cases. If \(r(c)=r\) and \(t_j\leq V\), it belongs to row \(r\) and is observable by valuation, so its full amount \(p_j\) is included. If \(r(c)\ne r\), it belongs to another origin row, so its contribution to row \(r\) is zero even if paid before valuation. If \(t_j>V\), the transaction had not yet occurred at the snapshot date, so its contribution is also zero. These cases establish the piecewise rule, including all conditions.
The same separation of roles helps with incurred triangles. The origin row is still selected using the triangle’s stated origin basis. The valuation date identifies the as-of snapshot of incurred information; it is not a substitute origin date. The report date may be relevant to a report-year triangle, but it does not turn a loss into a different accident year. The specific claim amount included in an incurred cell depends on the incurred information available under the triangle’s definition at that valuation.
Common Mistakes and Practical Checks
- Using the payment date as the accident date. A payment in 2025 can relate to a loss in 2023. For an accident-year triangle, retain the 2023 origin row; use the payment date for payment timing.
- Using the valuation date as an origin date. Valuation determines the snapshot cutoff. It does not move every claim into the valuation year.
- Applying only the valuation test. “Payment date is on or before valuation” is not enough for a row contribution. The claim must also match that row’s origin period. A payment for a different origin year contributes zero to the row in question.
- Confusing first report with a later report update. For a report-year origin, use the defined report date—typically the first notice date—not an arbitrary subsequent update date.
- Ignoring the boundary convention. A date just before midnight on December 31 and one on January 1 fall in different calendar years. Confirm the data’s date and time-zone convention, then apply it consistently.
A useful audit is to write the triangle basis and valuation date beside the data before assigning records. For every claim, identify the origin date required by that basis. For every payment, separately check its payment date against valuation and, if relevant, the cell’s calendar period. These checks catch date substitutions without requiring any change to the underlying claim amount.
Check Your Understanding
For each hypothetical situation, identify the correct date role and apply the stated valuation rule.
- A claim has a loss date of November 18, 2022, a first report date of January 5, 2023, and a payment date of May 4, 2023. What are its accident-year and report-year origins, and what is the payment calendar year? (Answer: Accident year 2022; report year 2023; payment calendar year 2023.)
- A policy effective date is December 15, 2021, and the loss occurs January 3, 2022. Under a policy-effective-year convention and an accident-year convention, which origin year applies in each triangle? (Answer: Policy year 2021; accident year 2022.)
- A claim assigned to accident year 2023 has a $6,800 payment dated October 10, 2024. At valuation December 31, 2024, what is its contribution to the 2023 accident-year row, and to the 2022 accident-year row? (Answer: $6,800 to the 2023 row; $0 to the 2022 row.)
- A claim assigned to report year 2024 has payments of $5,000 on December 30, 2024 and $1,200 on January 2, 2025. What cumulative paid amount is included at valuation December 31, 2024? (Answer: $5,000.)
- Two accident-year claims have payments of $900 and $1,400, both on or before valuation. Their loss dates fall in 2022 and 2023, respectively. What is the 2022-row contribution from the first payment and the second payment? (Answer: $900 from the 2022-origin claim and $0 from the 2023-origin claim.)