Why the Measure of Loss Matters
A loss estimate for an accident period can describe different things. It may be a forecast of the period’s eventual total, a record of payments made so far, an incurred amount that also includes case reserves, or an estimate that projects observed losses to ultimate. These measures are related, but they are not interchangeable. Comparing them without accounting for their different meanings and development stages can make experience appear unusually favorable or unfavorable when the comparison is simply not like-for-like.
This distinction is central to interpreting expected claims methods. An a priori expected ultimate loss is an estimate formed from assumptions about the period, such as its exposure and expected loss level. Emerged paid and incurred losses reflect claims recorded or settled by an evaluation date. A development-based ultimate estimate uses emerged data and a development assumption to estimate the eventual total. The estimates can be compared, but the comparison should be made on a consistent loss basis and with an understanding of what has and has not emerged.
“Ultimate” refers to the total loss for the accident period after its claims have fully emerged under the chosen definition of loss. It does not mean the total amount already paid. The definition must also specify the loss basis being measured—for example, whether allocated loss adjustment expenses are included and whether the amount is gross or net of reinsurance. Comparisons are meaningful only when the measures use consistent definitions.
Paid, Incurred, and Not-Yet-Emerged Loss
At an evaluation date, paid loss records cash paid on claims. Incurred loss adds case reserves: estimates of amounts expected to be paid on claims that have been reported but are not fully settled. Incurred loss does not, by this definition, include a separate estimate for claims that have not yet been reported or for other amounts not represented in case reserves. Those potential amounts are part of what must be allowed for in an ultimate estimate.
If the eventual ultimate is \(U\), then \(U-\text{paid loss}\) represents the amount ultimately paid after the evaluation date, under the same loss definition. By contrast, \(U-\text{incurred loss}\) is the difference between ultimate and the current incurred amount. It is not the total unpaid amount: current case reserves are already included in incurred loss. These relationships are accounting and estimation relationships, not a claim that the current incurred amount is already known to be correct.
| Measure | What it represents | What it does not establish by itself |
|---|---|---|
| Expected ultimate | An a priori estimate of the eventual total for the period | How much has actually been paid or reported by the evaluation date |
| Paid loss | Cumulative claim payments made by the evaluation date | The full value of reported claims or the eventual total |
| Incurred loss | Paid loss plus current case reserves | The eventual total, including all future claim emergence |
| Developed ultimate estimate | A projection from emerged paid or incurred loss to an estimated ultimate | A known final outcome; it depends on the development assumption |
Worked Example: Separating Expected, Paid, and Incurred Amounts
A hypothetical accident period has an a priori expected ultimate loss of $2,400,000. At the evaluation date, cumulative paid loss is $1,080,000 and case reserves are $720,000. Assume that the selected expected paid emergence is 45% of ultimate and the expected incurred emergence is 75% of ultimate. Calculate incurred loss and the paid- and incurred-based developed ultimate estimates. Then interpret the amounts.
Method. First calculate incurred loss from the accounting definition. Then divide each emerged measure by its own expected emergence proportion. This method avoids treating paid loss and incurred loss as if they were the same measure.
Step 1: Calculate incurred loss.
Incurred loss is $1,800,000.
Step 2: Project paid loss to ultimate. Under the stated 45% paid emergence assumption, the paid-based estimate is:
Step 3: Project incurred loss to ultimate. Under the stated 75% incurred emergence assumption:
Both development-based estimates equal the $2,400,000 a priori expected ultimate. That agreement is consistent with the stated assumptions; it does not prove that the eventual outcome will be exactly $2,400,000. As a further check, the difference between expected ultimate and current incurred loss is $600,000. The difference between expected ultimate and paid loss is $1,320,000, equal to the $720,000 case reserves plus the $600,000 difference between ultimate and incurred.
Developed Losses: Projection, Not a New Type of Observed Loss
A developed ultimate estimate is not an additional amount of claim payment or a new accounting balance. It is an estimate calculated from observed loss data and an assumption about how losses emerge. In a simple emergence model, let \(U\) denote ultimate loss and let \(q_t\) be the expected proportion of ultimate emerged by development age \(t\), on a specified basis. If the observed cumulative loss at age \(t\) is \(L_t\), the projection is \(L_t/q_t\), provided the emergence proportion is appropriate for that measure and the model is suitable.
For cumulative paid losses, \(q_t\) must describe paid emergence. For incurred losses, it must describe incurred emergence. A paid proportion should not be applied to incurred loss, or vice versa. In chain ladder terminology, an age-to-ultimate development factor can be viewed as the reciprocal of an assumed emergence proportion in this simplified setting. Actual selected development factors may be based on observed loss development patterns and need not be treated as literal fixed proportions in every context.
Why the Simple Development Formula Works
The projection follows directly from the emergence assumption. Suppose, for the specified loss basis and development age, the model says that the emerged amount is the fraction \(q_t\) of ultimate. If the relationship is represented as \(L_t=q_tU\), solve for \(U\) by dividing both sides by \(q_t\), which must be positive:
In practice, \(L_t\) is observed while \(U\) is unknown, so the corresponding estimate is \(\widehat{U}_t=L_t/q_t\). This derivation explains both the method and its limitation: the projection depends on the relationship between emerged loss and ultimate being appropriate. If the selected emergence proportion is unsuitable for the accident period, or the observed data are atypical, the resulting estimate may be biased. A development-based estimate therefore remains an estimate, not a final observation.
Worked Example: Comparing Paid and Incurred Development Estimates
For a hypothetical accident period, cumulative paid loss at the evaluation date is $600,000 and case reserves are $300,000. A selected paid age-to-ultimate factor is 1.50, while a selected incurred age-to-ultimate factor is 1.20. The a priori expected ultimate is $1,000,000. Calculate both developed estimates and compare them with the expectation.
Method. Calculate incurred loss first. Multiply each emerged loss measure by its own selected age-to-ultimate factor. This avoids applying the paid factor to incurred losses or assuming that the two estimates must agree.
Step 1: Calculate incurred loss.
Step 2: Calculate the paid-based estimate.
Step 3: Calculate the incurred-based estimate.
The paid-based estimate is $900,000, or $100,000 below the a priori expected ultimate. The incurred-based estimate is $1,080,000, or $80,000 above it. The two developed estimates differ by $180,000. The difference may reflect different selected development patterns, random variation, or features of the data; these figures alone do not identify its cause. The expected ultimate is a benchmark for comparison, not a constraint requiring either development estimate to equal it.
Interpreting Emerged Experience Against an Expectation
An expected ultimate can be used to assess whether observed development is broadly consistent with the original expectation, but it must be compared with a measure at the same stage of emergence. Comparing an expected ultimate directly with current paid or incurred loss confounds the loss level with the proportion expected to have emerged. A period can have paid loss well below its expected ultimate simply because a substantial share of claims is not yet paid.
A useful comparison is between the actual emerged amount and the amount expected to have emerged by the same age. If an expected ultimate is \(U_{\mathrm{exp}}\) and the relevant expected emergence proportion is \(q_t\), then the corresponding expected emerged amount is \(q_tU_{\mathrm{exp}}\). The actual emerged amount can be compared with that value. Alternatively, the emerged amount can be developed to an ultimate estimate and compared with \(U_{\mathrm{exp}}\). The two approaches answer related but distinct questions: one asks whether emergence so far is in line with expectation, while the other expresses that experience as a projected ultimate.
Worked Example: Reading an Incurred Emergence Signal
A hypothetical accident period has an expected ultimate loss of $1,500,000. At 12 months, the selected expected incurred emergence is 60%. Actual incurred loss is $990,000. Cumulative paid loss is $540,000, and the selected expected paid emergence is 36%. Compare the observed amounts with expected emergence and calculate the two developed estimates.
Method. Calculate expected incurred and paid amounts at 12 months, then project each actual amount using its corresponding emergence proportion. This avoids interpreting an emerged amount as though it were already ultimate.
Step 1: Calculate expected incurred emergence.
Actual incurred loss of $990,000 is $90,000 above expected incurred emergence, a difference of \(90{,}000/900{,}000=10\%\) of the expected amount.
Step 2: Develop actual incurred loss.
This incurred-based estimate is $150,000, or 10%, above the expected ultimate of $1,500,000.
Step 3: Check paid emergence. Expected paid loss at 12 months is:
Actual paid loss equals this expected amount. The paid-based developed estimate is \(540{,}000/0.36=$1{,}500{,}000\), equal to the a priori expected ultimate. The paid and incurred views therefore differ: paid emergence is exactly at its expected level, while incurred emergence is above its expected level. This is a signal to investigate the data and assumptions, not proof that the final loss will be either $1,500,000 or $1,650,000.
Common Mistakes and Practical Checks
- Comparing raw emerged losses with ultimate losses. A paid amount below the expected ultimate does not by itself indicate favorable experience. Compare it with expected paid emergence or first develop it to ultimate.
- Confusing incurred loss with ultimate loss. Incurred loss includes case reserves but can still exclude future emergence and may change as claims develop.
- Using the wrong development assumption. Paid and incurred data have different emergence patterns. Use the factor or proportion selected for the measure being projected.
- Adding case reserves twice. Incurred loss already equals paid loss plus case reserves. Do not add case reserves again when developing incurred loss.
- Treating a projection as an observed final result. A developed estimate depends on its selected development assumption. It is not a statement that the accident period’s final losses are already known.
- Comparing inconsistent loss definitions. Confirm that the expected, emerged, and developed amounts use the same coverage, expense, and gross or net basis.
Check Your Understanding
For each hypothetical situation, identify the relevant measure and calculate the requested values. Compare paid or incurred experience with expected ultimate only after accounting for emergence.
- An accident period has expected ultimate loss of $1,200,000. Paid loss is $480,000, case reserves are $360,000, expected paid emergence is 40%, and expected incurred emergence is 70%. Calculate incurred loss and the paid- and incurred-based developed ultimate estimates. (Answer: Incurred loss = $840,000; both developed estimates = $1,200,000.)
- Paid loss is $420,000 and incurred loss is $700,000. The selected paid and incurred age-to-ultimate factors are 1.80 and 1.25, respectively. Calculate both developed estimates and their difference. (Answer: Paid-based = $756,000; incurred-based = $875,000; difference = $119,000, with the incurred-based estimate higher.)
- Expected ultimate is $2,000,000. At the evaluation age, expected incurred emergence is 55%, and actual incurred loss is $1,210,000. Calculate expected incurred emergence, the developed incurred estimate, and its percentage difference from expected ultimate. (Answer: Expected incurred emergence = $1,100,000; developed estimate = $2,200,000; 10% above expected ultimate.)
- Expected ultimate is $900,000. Paid loss is $270,000 and case reserves are $360,000. Expected paid emergence is 30% and expected incurred emergence is 70%. Calculate incurred loss and both developed estimates. (Answer: Incurred loss = $630,000; paid-based and incurred-based estimates are each $900,000.)
- Expected ultimate is $1,600,000. Actual incurred loss is $1,040,000 and paid loss is $720,000. Expected incurred emergence is 65% and expected paid emergence is 45%. Calculate each developed estimate and state which is above, below, or equal to expected ultimate. (Answer: Incurred-based = $1,600,000, equal to expected ultimate; paid-based = $1,600,000, equal to expected ultimate.)