Why Maturity Matters Before Credibility
A large volume of experience is not automatically useful for credibility analysis. If many claims from a recent period have not yet been reported, or if reported claims are still developing, the observed result may be systematically different from the result that will ultimately emerge. More exposure does not remove that timing bias.
The key question is whether the observed data measure the intended experience consistently. A frequency based on reported claims, for example, can appear low simply because claims have not had enough time to reach the insurer. Likewise, incurred losses can understate eventual costs if case estimates do not yet reflect later payments or claim developments. Assessing completeness and maturity is therefore a step in deciding whether experience is suitable for credibility analysis, not a substitute for defining the claim measure or exposure base.
Two Different Clocks: Exposure and Claims
Exposure is the amount of risk attached to the experience period, measured using the chosen exposure unit. The exposure must correspond to the period under study: for example, earned vehicle-years for a hypothetical automobile coverage. If a policy is only partly earned by the cutoff date, treating its full term as earned exposure can inflate the denominator and make a frequency appear too low.
Claims have a separate timeline. An incident must occur, be reported, be recorded, and often be investigated and settled. The evaluation date determines which of those events are visible in the data. A period may have finished earning all its exposure while still having claims that are unreported, unpaid, or inadequately estimated.
It helps to be precise about the claim measure. A reported-claim count includes claims reported by the evaluation date, not necessarily every claim that has occurred. Paid losses include amounts paid to date and omit amounts not yet paid. Incurred losses generally combine paid losses with case reserves for reported claims, but they do not automatically include claims that have not been reported. Even for reported claims, case reserves may differ from eventual payments.
| Data element | What can be incomplete or immature? | Possible effect on an observed result |
|---|---|---|
| Exposure | Some of the intended risk period may not yet be earned or included | The denominator may not match the period represented by the claims |
| Reported-claim count | Some occurred claims may not yet have been reported | Observed frequency may be too low |
| Paid losses | Some reported claim amounts have not yet been paid | Observed loss cost may be too low |
| Incurred losses | Some claims may be unreported, or case reserves may be incomplete | Observed incurred loss may be below its eventual level |
When a Simple Completion Adjustment Is Defensible
A simple adjustment can illustrate the effect of incomplete reporting. Suppose a suitable analysis provides a reporting-completion proportion \(p\): the expected fraction of ultimate claims already reported at the evaluation date. If reported claims are representative of all claims, then the reported count is expected to be \(p\) times the ultimate count. This gives the estimate
This adjustment requires more than knowing a percentage. It assumes the completion proportion applies to the period and claim population being analyzed, the exposure base is properly measured, and reporting delay is not systematically related to the types of claims missing. It is an estimate, not a way to identify the exact unreported claims.
A related calculation can be used for losses only when the stated proportion refers to the particular loss measure being adjusted. If reported incurred losses are expected to represent a proportion \(q\) of ultimate losses, then a simple estimate is observed incurred losses divided by \(q\). That assumption is not interchangeable with a claim-count reporting proportion: claim reporting and loss development describe different aspects of maturity.
Worked Examples
Worked Example: Adjusting a Recent Reported-Claim Frequency
A hypothetical insurer has earned 800 vehicle-years in a defined coverage group. By the evaluation date, 72 claims from that period have been reported. A suitable analysis for this group and cutoff indicates that 90% of ultimate claims are expected to have been reported. Estimate the ultimate claim count and frequency. Assume the 90% completion proportion applies equally to claims in the period and that the earned exposure is complete.
Method. Adjust the count for reporting completeness, then divide by the earned exposure. No parametric claim-count distribution is assumed; the calculation uses the stated completion proportion.
First calculate the unadjusted reported frequency:
The reporting-completion proportion is \(p=0.90\), so the estimated ultimate count is:
Then calculate the estimated ultimate frequency using the same 800 earned vehicle-years:
Check. If 90% of 80 ultimate claims are reported, the reported count is \(0.90(80)=72\), matching the observation. The unadjusted frequency of 0.09 is below the estimated ultimate frequency of 0.10. The common mistake is to treat the reported count as complete and skip the reporting adjustment.
Worked Example: Distinguishing Incurred Loss Maturity From Claim Reporting
A hypothetical program has 80 reported claims and 1,000 earned exposure units. Paid losses are $240,000 and case reserves for those reported claims are $160,000, so incurred losses are the sum of those amounts. For this illustration, assume all claims have been reported, and an appropriate assessment indicates that current incurred losses are expected to equal 80% of ultimate losses. Estimate ultimate losses and ultimate loss cost per exposure unit.
Method. First calculate incurred losses from paid losses and case reserves. Then apply the stated incurred-loss maturity proportion. The 80% assumption concerns loss dollars, not the proportion of claims reported.
Current incurred losses are:
If this is 80% of ultimate losses, the estimated ultimate loss amount is:
Divide by the 1,000 earned exposure units:
For comparison, the current incurred loss cost is \(\$400{,}000/1{,}000=\$400\) per unit. The stated assumption increases it by a factor of \(1/0.80=1.25\), giving \(\$400(1.25)=\$500\). Check. Eighty percent of $500,000 is $400,000. The common mistake is to divide the number of claims by 0.80 even though 0.80 was specified as a loss-dollar completion proportion.
Worked Example: Comparing Two Periods on a Consistent Basis
A hypothetical coverage group has 1,000 earned exposure units and 100 claims in a fully mature 2024 period. In a later period it has 800 earned exposure units and 68 reported claims by the evaluation date. Assume the later period is 85% complete for claim reporting, that the completion assumption applies to its claims, and that exposure is complete in both periods. Compare estimated ultimate frequencies.
Method. Leave the fully mature count unchanged. Adjust the later period's reported count for its stated reporting completeness before calculating frequency.
For 2024:
For the later period, first estimate the ultimate count:
Then calculate its estimated ultimate frequency:
The two estimated frequencies are both 0.10 claims per exposure unit. If the later period were compared without adjustment, its reported frequency would be \(68/800=0.085\), which would suggest a decline that is explained here by reporting delay. Check. Eighty-five percent of 80 is 68, so the adjusted later count agrees with the given completion assumption. The common mistake is to compare raw reported counts or frequencies across periods with different maturity.
Limits, Judgment, and Exam Traps
A completion adjustment is only as credible as its assumptions. Reporting patterns can vary by coverage, claim type, reporting channel, and time. Severe or complex claims may take longer to report or develop than simple claims. If late-reported claims differ materially in size from early-reported claims, a count completion percentage should not be used as though it were a loss-dollar completion percentage.
The evaluation date also matters. Two periods described as “one year old” need not have the same maturity if one is measured at a different cutoff, uses a different reporting definition, or includes a different mix of risks. Comparisons should use a consistent claim measure, exposure definition, and evaluation basis. If those cannot be aligned, the observed difference may combine underlying experience differences with measurement timing.
Finally, maturity is not the same as credibility. A well-developed period may still have too little exposure or too much random variation to be a stable indication. Conversely, a large volume of immature experience can produce a precise-looking number that is biased by unreported claims or incomplete estimates. Assess the measurement basis before interpreting the amount of experience.
Check that the exposure is earned, assigned to the intended period, and measured consistently with the claim data.
Distinguish reported counts, paid losses, and incurred losses, and state the date through which they are observed.
Ask whether claims may be unreported and whether reported amounts may still change. Use a completion assumption only if it matches the measure being analyzed.
Before considering experience for credibility, place periods on a consistent exposure, claim-measure, and maturity basis, and disclose material uncertainty.
Check Your Understanding
For each problem, state any adjustment and use the specified exposure base.
- A hypothetical group has 600 earned exposure units and 54 reported claims. Assume reporting is 90% complete and the exposure is complete. Estimate ultimate claims and ultimate frequency. (Answer: 60 claims; 0.10 claims per exposure unit.)
- Reported claims have $30,000 paid and $20,000 in case reserves. Assume all claims are reported and current incurred losses are 80% of ultimate losses. Estimate ultimate losses. (Answer: $62,500.)
- A period has 400 earned exposure units and 36 reported claims. Assume 75% reporting completeness. Calculate the estimated ultimate frequency. (Answer: 0.12 claims per exposure unit.)
- A fully mature period has 99 claims for 900 earned exposure units. A later period has 90 reported claims for 1,000 earned exposure units and is assumed 90% complete for reporting. Compare the estimated ultimate frequencies. (Answer: 0.11 for the fully mature period and 0.10 for the later period.)
- A hypothetical portfolio has 1,200 earned exposure units and 114 reported claims. Assume claims are 95% complete for reporting and the exposure is complete. Estimate the ultimate count and frequency. (Answer: 120 claims; 0.10 claims per exposure unit.)