Why Maturity Matters in an Expected-Loss Comparison
An a priori expected ultimate loss is an estimate of the total loss for a period, made before all of that period’s claims have emerged. Paid or incurred losses observed so far are partial, maturity-dependent views of that total. Comparing a partial amount directly with an ultimate estimate can therefore make ordinary development look like unfavorable or favorable experience.
The comparison needs two distinct inputs: the selected expected ultimate and the share of that ultimate expected to have emerged by the evaluation age. The second input changes as the period matures; it does not, by itself, revise the first. This distinction is useful when assessing periods at different ages and is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
Maturity is not simply the number of months since a period began. It is represented here by an expected emergence proportion for a specific measure—paid or incurred—and evaluation age. A period with 40% expected paid emergence is less mature on a paid basis than one with 75% expected paid emergence. Paid and incurred proportions are generally different because incurred losses include case estimates for reported claims as well as payments made.
Expected Loss at the Same Maturity
Before applying the comparison, state its assumptions. The expected ultimate \(U\) is selected on the same coverage, claim, expense, and gross-or-net basis as the observed losses and emergence pattern. The selected proportion describes the expected share of that ultimate represented by cumulative losses of the chosen type at age \(t\). The period and evaluation age are correctly identified. If the observed amount is paid, use a paid emergence proportion; if it is incurred, use an incurred proportion.
On that basis, the expected amount emerged by age \(t\) is:
The corresponding experience comparison is the observed cumulative amount divided by the expected cumulative amount at the same maturity. For actual paid loss \(C_t^P\) and actual incurred loss \(C_t^I\), the ratios are \(C_t^P/(U p_t)\) and \(C_t^I/(U q_t)\), respectively. A ratio of 1.00 means the observed amount equals the amount expected to have emerged to date under the selected assumptions. A ratio above 1.00 means it is higher; a ratio below 1.00 means it is lower.
Why the Expected-to-Date Formula Holds
The emergence proportion expresses the expected cumulative amount as a fraction of the expected ultimate on the same basis. For paid loss, its definition is \(p_t=E[C_t^P]/U\), where \(C_t^P\) is cumulative paid loss by age \(t\). Multiplying both sides by \(U\) gives the expected cumulative paid amount. The incurred relationship follows in the same way using \(q_t\) and cumulative incurred loss.
This identity explains why a-to-date comparison needs an age-specific benchmark. A fixed ultimate estimate can imply a larger expected paid amount at a later age simply because a greater share is expected to have been paid. That is a change in expected emergence, not necessarily a change in the a priori ultimate.
Worked Example: Comparing Paid Loss With Its Expected Maturity
A hypothetical liability portfolio has an a priori expected ultimate paid loss of $1,250,000. At the evaluation age, the selected paid emergence proportion is 40%. Actual cumulative paid loss is $495,000. Calculate expected paid loss to date, the observed-to-expected ratio, and the expected unpaid amount. Explain what the comparison says about the prior.
Method and basis. This is a paid-to-paid comparison. The expected ultimate and actual payments are both stated in dollars on the same basis, and the 40% proportion applies to the evaluation age in the problem.
Expected paid loss to date.
Observed-to-expected ratio.
Actual paid loss is 99% of the amount expected to have been paid by this age, or $5,000 below that expected amount: \(\$495{,}000-\$500{,}000=-\$5{,}000\).
Expected unpaid amount. The expected unpaid share is \(1-0.40=0.60\), so:
This calculation has not changed the a priori ultimate estimate of $1,250,000. It says that paid experience to date is close to the amount expected to have emerged so far. The common mistake is to treat \(495{,}000/1{,}250{,}000=0.396\) as the relevant observed-to-expected comparison. That quotient is 0.396, or 39.6% of the expected ultimate; it is not the ratio to expected paid loss at this maturity. The relevant ratio is \(495{,}000/500{,}000=0.99\).
Comparing Periods at Different Ages
Raw cumulative amounts are not directly comparable when periods have different expected maturity or different expected ultimate loss. First calculate each period’s expected amount to date using its own ultimate and emergence proportion. Then compare each actual amount with its corresponding expected amount. The resulting ratios describe actual experience relative to expectation at each period’s maturity; they do not remove uncertainty in the observed losses or prove that either ultimate is correct.
Worked Example: Two Periods With Different Paid Maturity
Two hypothetical property portfolios have different expected ultimates and are evaluated at different ages. Period A has expected ultimate paid loss of $800,000, a paid emergence proportion of 75%, and actual paid loss of $630,000. Period B has expected ultimate paid loss of $1,000,000, a paid emergence proportion of 40%, and actual paid loss of $430,000. Compare the periods on an observed-to-expected basis.
Period A. Expected paid loss to date is \(\$800{,}000 \times 0.75=\$600{,}000\). The observed-to-expected ratio is \(\$630{,}000/\$600{,}000=1.05\). Actual paid loss is $30,000 above the amount expected to date, or 5% above it.
Period B. Expected paid loss to date is \(\$1{,}000{,}000 \times 0.40=\$400{,}000\). The observed-to-expected ratio is \(\$430{,}000/\$400{,}000=1.075\). Actual paid loss is $30,000 above the amount expected to date, or 7.5% above it.
Period B has the lower actual paid amount, but the higher observed-to-expected ratio. On this comparison, B is further above its expected-to-date amount: 7.5% compared with A’s 5%. The result does not establish that B’s ultimate will be higher than its a priori estimate; it is a maturity-adjusted comparison of experience so far.
A tempting but unsuitable comparison is to divide each actual paid amount directly by its expected ultimate. Those quotients are \(\$630{,}000/\$800{,}000=0.7875\) for A and \(\$430{,}000/\$1{,}000{,}000=0.43\) for B. They describe the actual amount paid as a share of each ultimate, but do not compare actual payments with the shares expected to be paid at the respective evaluation ages.
Keep Paid and Incurred Comparisons Separate
Incurred loss commonly combines cumulative paid loss with case estimates for reported claims. Its emergence proportion therefore need not equal paid emergence at the same age. Compare actual incurred with expected incurred to date using an incurred emergence proportion; do not substitute a paid proportion simply because both measures refer to the same period.
Worked Example: Paid and Incurred Views at One Evaluation Age
For a hypothetical coverage, the a priori expected ultimate loss is $900,000. At the evaluation age, the selected paid emergence proportion is 30% and the incurred emergence proportion is 65%. Actual cumulative paid loss is $285,000 and actual incurred loss is $610,000. Calculate and interpret both comparisons.
Paid comparison. Expected paid loss to date is \(\$900{,}000 \times 0.30=\$270{,}000\). The ratio is \(\$285{,}000/\$270{,}000=1.0556\) to four decimal places. Paid loss is $15,000 above its expected-to-date amount.
Incurred comparison. Expected incurred loss to date is \(\$900{,}000 \times 0.65=\$585{,}000\). The ratio is \(\$610{,}000/\$585{,}000=1.0427\) to four decimal places. Incurred loss is $25,000 above its expected-to-date amount.
The paid and incurred comparisons are both above 1.00, but they are not interchangeable measures. The paid ratio compares payments with expected payments; the incurred ratio compares incurred loss with expected incurred loss. In particular, \(285{,}000/585{,}000\) would mix paid actuals with an incurred expectation and would not be a valid comparison. The difference between these two ratios is not, by itself, evidence that the a priori ultimate should be changed.
Tracking a Fixed Prior as a Period Matures
For a single period, the expected cumulative amount usually rises as the selected emergence proportion rises. The a priori ultimate can remain fixed throughout this comparison. Actual cumulative paid loss also usually rises, but actual experience need not follow the selected emergence curve exactly. A ratio below 1.00 at one age can move above 1.00 later, or vice versa.
Worked Example: Reading a Paid Experience Pattern Over Time
A hypothetical accident period has a fixed a priori expected ultimate paid loss of $1,200,000. Selected paid emergence proportions at 12, 24, and 36 months are 30%, 57.5%, and 80%. Actual cumulative payments at those ages are $370,000, $690,000, and $1,008,000. Calculate the expected amounts to date and observed-to-expected ratios at each age.
At 12 months. Expected paid loss is \(\$1{,}200{,}000 \times 0.30=\$360{,}000\). The ratio is \(\$370{,}000/\$360{,}000=1.0278\) to four decimal places.
At 24 months. Expected paid loss is \(\$1{,}200{,}000 \times 0.575=\$690{,}000\). The ratio is \(\$690{,}000/\$690{,}000=1.0000\).
At 36 months. Expected paid loss is \(\$1{,}200{,}000 \times 0.80=\$960{,}000\). The ratio is \(\$1{,}008{,}000/\$960{,}000=1.0500\).
The expected amount to date increases from $360,000 to $690,000 to $960,000 because the expected paid share increases. The expected ultimate remains $1,200,000 at all three ages. The ratios show how actual cumulative payments compare with the selected maturity-specific expectations; they should not be read as three different a priori ultimate selections.
As a check on the cumulative figures, expected incremental payments are \(\$690{,}000-\$360{,}000=\$330{,}000\) from 12 to 24 months and \(\$960{,}000-\$690{,}000=\$270{,}000\) from 24 to 36 months. Actual increments are \(\$690{,}000-\$370{,}000=\$320{,}000\) and \(\$1{,}008{,}000-\$690{,}000=\$318{,}000\), respectively. These increments reconcile to the stated cumulative amounts.
Common Mistakes and Interpretation Limits
- Comparing partial actuals directly with an ultimate. A-to-date amount is not expected to equal the ultimate before full emergence. Use an expected-to-date amount based on the same maturity.
- Using the wrong emergence pattern. Paid actuals require a paid expectation; incurred actuals require an incurred expectation. Confirm the loss scope and basis as well as the measure.
- Treating an observed-to-expected ratio as a new ultimate estimate. The ratio is a comparison of experience to expectation at an evaluation age. Turning it into an ultimate estimate requires additional assumptions and a method; it is not automatic.
- Assuming emergence proportions are exact descriptions of every period. They represent selected expectations. Actual losses can develop differently, and case estimates in incurred data can change.
- Ignoring the denominator. Two periods with the same actual loss can have different comparisons if their expected ultimates or maturity proportions differ.
The calculation is only as meaningful as the alignment of its inputs. If the expected ultimate includes loss adjustment expense but actual losses exclude it, or one amount is gross of reinsurance and another is net, the ratio mixes bases. If an emergence proportion was selected for a different evaluation age or loss measure, the expected-to-date denominator is also mismatched. Such questions should be resolved before interpreting the result.
Check Your Understanding
For each problem, use the emergence proportion that matches the stated loss measure.
- A hypothetical workers compensation period has expected ultimate paid loss of $640,000. The paid emergence proportion is 50%, and actual paid loss is $310,000. What are expected paid loss to date and the observed-to-expected ratio? (Answer: $320,000; 0.9688 to four decimal places.)
- A hypothetical equipment portfolio has expected ultimate loss of $750,000. At the selected age, paid emergence is 32% and actual paid loss is $260,000. Calculate the expected paid amount to date and the excess or shortfall of actual against it. (Answer: $240,000 expected; $20,000 excess.)
- Two hypothetical periods each have actual paid loss of $420,000. Period A has expected ultimate paid loss of $700,000 and paid emergence of 60%; Period B has expected ultimate paid loss of $1,000,000 and paid emergence of 40%. Calculate each observed-to-expected ratio and identify which is higher. (Answer: A is 1.0000; B is 1.0500; B is higher.)
- A hypothetical liability period has expected ultimate loss of $500,000. At evaluation, paid emergence is 25%, incurred emergence is 70%, actual paid loss is $130,000, and actual incurred loss is $350,000. Calculate both observed-to-expected ratios. (Answer: Paid \(130{,}000/125{,}000=1.0400\); incurred \(350{,}000/350{,}000=1.0000\).)
- A hypothetical accident period has a fixed expected ultimate paid loss of $2,000,000. Paid emergence is 45% at 18 months and 72% at 30 months. Actual cumulative payments are $900,000 and $1,440,000 at those ages. Calculate expected paid loss to date and the ratio at each age, then state whether the a priori ultimate changes merely because the age changes. (Answer: At 18 months, expected $900,000 and ratio 1.0000; at 30 months, expected $1,440,000 and ratio 1.0000; the a priori ultimate remains $2,000,000.)