Why Compare Actual and Expected Emergence?
A reserve estimate is based on assumptions about how claims will develop. At a later valuation, some of that development has become observable: payments have been made, case reserves have changed, and incurred losses have emerged. Comparing the observed emergence with what the prior estimate expected helps identify where experience is departing from the selected pattern.
The comparison is a diagnostic, not a verdict. Actual emergence above expectation can signal faster development, greater claim severity, or a change in case reserving. It does not, by itself, prove that the ultimate loss estimate is inadequate. Likewise, emergence below expectation does not establish that the estimate is conservative. Interpretation requires attention to the loss measure, valuation age, accident year, and assumptions underlying the prior estimate.
Set Up a Consistent Comparison
Let \(U\) be the prior selected ultimate for an accident year. Let \(p_j\) be the prior expected proportion of that ultimate represented by the chosen cumulative measure at valuation age \(j\). For paid losses, the measure is cumulative paid. For incurred losses, it is cumulative incurred, commonly paid plus case outstanding under the applicable reporting definition. The proportions and the ultimate must use a consistent basis, including currency level and treatment of expenses or other adjustments.
Under this fixed prior pattern, expected cumulative losses at age \(j\) are \(U p_j\). The expected emergence from age \(j\) to age \(j+1\) is the difference between the two expected cumulative amounts. If actual cumulative losses at those ages are \(C_j\) and \(C_{j+1}\), actual emergence is their difference.
Here \(E_{j,j+1}\) is expected emergence, \(A_{j,j+1}\) is actual emergence, and \(D_{j,j+1}\) is the signed deviation. For paid losses, actual emergence is ordinarily nonnegative, although corrections, recoveries, or accounting adjustments can make a reported net change negative. Incurred emergence can readily be negative because case reserves may be reduced by more than new payments and reserve additions.
Why the Incremental Formula Holds
The selected pattern assigns expected cumulative amounts \(U p_j\) and \(U p_{j+1}\) at the two ages. The amount expected to emerge between those ages is the later cumulative amount less the earlier cumulative amount. Therefore,
This is an identity for a fixed prior ultimate and pattern. It assumes that both proportions refer to the same measure and valuation basis, and that the prior pattern is the benchmark being tested. If the selected ultimate or pattern is changed after observing experience, the result is no longer a comparison solely against the original expectation. The benchmark should therefore be stated and held fixed for the comparison.
The difference between actual and expected cumulative losses at age \(j\) is \(C_j-U p_j\). Its change from age \(j\) to \(j+1\) equals the interval deviation:
Thus interval deviations explain how the cumulative gap changes. A favorable-looking interval can narrow an earlier adverse cumulative gap, or an adverse interval can widen it. Always distinguish the interval result from the cumulative position.
Worked Examples
Worked Example: Paid Emergence Against a Fixed Prior Pattern
A hypothetical property accident year has a prior selected ultimate paid loss of $2,000,000. The prior selected paid development pattern expects 55% cumulative paid at 12 months, 72% at 24 months, and 84% at 36 months. Actual cumulative paid is $1,100,000 at 12 months, $1,480,000 at 24 months, and $1,700,000 at 36 months. Calculate the expected and actual emergence for each interval, and interpret the deviations.
Method. Use the prior ultimate and the differences between successive cumulative proportions to calculate expected interval emergence. Subtract those expectations from actual interval emergence.
12 to 24 months: expected. The expected proportion emerging is \(0.72-0.55=0.17\). Expected emergence is \( \$2{,}000{,}000 \times 0.17=\$340{,}000\).
12 to 24 months: actual and deviation. Actual emergence is \(\$1{,}480{,}000-\$1{,}100{,}000=\$380{,}000\). Therefore \(D=\$380{,}000-\$340{,}000=+\$40{,}000\). Paid losses emerged $40,000 more than expected in this interval.
24 to 36 months: expected. The expected proportion emerging is \(0.84-0.72=0.12\). Expected emergence is \(\$2{,}000{,}000 \times 0.12=\$240{,}000\).
24 to 36 months: actual and deviation. Actual emergence is \(\$1{,}700{,}000-\$1{,}480{,}000=\$220{,}000\). The deviation is \(\$220{,}000-\$240{,}000=-\$20{,}000\). This interval produced $20,000 less paid emergence than expected.
Cumulative check. Expected cumulative paid is $1,100,000 at 12 months, $1,440,000 at 24 months, and $1,680,000 at 36 months. Actual less expected is therefore $0, +$40,000, and +$20,000 at those ages. The interval deviations, +$40,000 and −$20,000, change the cumulative gap from $0 to +$40,000 and then to +$20,000. The arithmetic reconciles.
Common mistake avoided. The 24-to-36-month expectation is $240,000 from the fixed prior ultimate and pattern, not a new forecast obtained by multiplying the actual 24-month paid amount by a separately selected factor. Re-basing on actual experience answers a different question.
Worked Example: Interpreting Negative Incurred Emergence
A hypothetical liability accident year has a prior selected incurred ultimate of $800,000. Its prior expected cumulative incurred proportions are 70% at 12 months, 82% at 24 months, and 90% at 36 months. Actual cumulative incurred losses are $555,000, $650,000, and $625,000, respectively. Calculate interval emergence and deviations, and explain the negative actual emergence in the second interval.
12 to 24 months: expected. The proportion increase is \(0.82-0.70=0.12\), so expected emergence is \(\$800{,}000 \times 0.12=\$96{,}000\). Actual emergence is \(\$650{,}000-\$555{,}000=\$95{,}000\). The deviation is \(\$95{,}000-\$96{,}000=-\$1{,}000\).
24 to 36 months: expected. The proportion increase is \(0.90-0.82=0.08\), so expected emergence is \(\$800{,}000 \times 0.08=\$64{,}000\). Actual emergence is \(\$625{,}000-\$650{,}000=-\$25{,}000\). The deviation is \(-\$25{,}000-\$64{,}000=-\$89{,}000\).
Cumulative comparison. Expected cumulative incurred is $560,000, $656,000, and $720,000 at 12, 24, and 36 months. Actual less expected is −$5,000, −$6,000, and −$95,000. The cumulative gap changes by −$1,000 in the first interval and −$89,000 in the second, matching the interval deviations.
The $25,000 decrease in incurred losses can occur if reductions in case reserves exceed newly paid amounts and reserve additions. It is not a negative payment; it is a net decrease in the incurred measure. The large negative deviation warrants investigation of case-reserve changes, claim closures, and data adjustments. It does not alone establish that the prior ultimate should be reduced by $89,000.
Common mistake avoided. Treating incurred emergence as necessarily positive would hide the reserve releases in this example. The correct calculation uses the signed change in cumulative incurred: $625,000 minus $650,000 equals −$25,000.
Worked Example: Comparing Accident Years at Different Maturities
A hypothetical portfolio has two accident years. For Year A, the prior selected ultimate is $1,200,000; the expected paid proportion rises from 50% to 65% over its comparison interval, and actual paid emergence is $230,000. For Year B, the prior selected ultimate is $2,000,000; its expected paid proportion rises from 25% to 45%, and actual paid emergence is $300,000. Calculate each interval deviation and the combined result. The intervals begin at different valuation ages.
Year A. Expected emergence is \(\$1{,}200{,}000 \times (0.65-0.50)=\$180{,}000\). Its deviation is \(\$230{,}000-\$180{,}000=+\$50{,}000\). Relative to its prior ultimate, the deviation is \(\$50{,}000/\$1{,}200{,}000=4.1667\%\).
Year B. Expected emergence is \(\$2{,}000{,}000 \times (0.45-0.25)=\$400{,}000\). Its deviation is \(\$300{,}000-\$400{,}000=-\$100{,}000\). Relative to its prior ultimate, the deviation is \(-\$100{,}000/\$2{,}000{,}000=-5.0000\%\).
Combined result. Total expected emergence is \(\$180{,}000+\$400{,}000=\$580{,}000\). Total actual emergence is \(\$230{,}000+\$300{,}000=\$530{,}000\). The combined deviation is \(\$530{,}000-\$580{,}000=-\$50{,}000\), or \(-\$50{,}000/\$3{,}200{,}000=-1.5625\%\) of the combined prior ultimates.
The portfolio total is below expectation, but this masks opposite accident-year results. The intervals also occur at different maturities: their dollar deviations are not a direct comparison of like-aged development. The maturity-specific expected proportions make each calculation internally consistent; interpretation should retain each year’s valuation age and investigate the separate results before relying on the aggregate.
Common mistake avoided. Comparing only the $50,000 and $100,000 deviations without scaling or noting valuation age can imply a misleading ranking. The normalized deviations add context, but do not remove differences in maturity or claim mix.
Reading Deviations by Measure, Age, and Accident Year
A useful review keeps the calculation and interpretation separate. First compute the interval deviation against the frozen prior expectation. Then ask what the deviation may indicate, and what additional evidence is needed. The same sign can have different meanings for paid and incurred losses.
| Observation | Possible interpretation | What to check |
|---|---|---|
| Paid emergence exceeds expectation | Payments were faster or larger than the pattern implied | Large settlements, payment timing, claim mix, inflation, and changes in handling |
| Paid emergence is below expectation | Payments were slower or smaller during the interval | Settlement delays, payment processing, recoveries, and whether later emergence may be shifted rather than eliminated |
| Incurred emergence is negative | Case reserve reductions or other negative adjustments exceeded additions and paid activity | Case reserve changes, claim closures, reopenings, and data or accounting adjustments |
| One accident year has a large deviation | The result may reflect year-specific experience or an unusual claim | Claim-level detail, mix, coverage, and consistency of the selected ultimate and pattern |
Valuation age matters because expected incremental emergence generally varies by age. A $100,000 deviation in an early interval with substantial expected emergence is not equivalent to the same deviation in a late interval with little expected emergence. Report both the dollar difference and a suitable relative measure, such as deviation divided by prior ultimate or by expected emergence. A ratio to expected emergence can be unstable when the expected amount is small, so include the dollar value and denominator.
Accident-year patterns also matter. A portfolio-wide result can hide one year above expectation and another below it. Review the years separately, then summarize the aggregate with its weights and the range of valuation ages represented. A comparison across differently mature years is informative only when the expected pattern and maturity are made explicit.
Common Traps and Practical Controls
- Changing the benchmark after observing results. Replacing the prior ultimate or pattern with a current selection turns a back-to-prior comparison into a new forecast. Label the benchmark valuation and preserve its assumptions.
- Confusing interval and cumulative results. A negative interval deviation may reduce, but not eliminate, a prior cumulative excess. Show both when the interpretation depends on the accumulated gap.
- Assuming incurred emergence cannot be negative. Incurred losses include estimates of unpaid claims, so reserve revisions can outweigh newly reported or paid amounts.
- Calling every positive deviation adverse ultimate development. More emergence than expected may reflect timing or changes in case reserving, not necessarily a higher ultimate. Examine ultimate estimates and claim-level explanations separately.
- Mixing bases. Do not compare paid actuals with an incurred pattern, or amounts on different inflation, expense, or gross/net bases. Align definitions before calculating.
A practical summary identifies the prior benchmark, measure, accident year, valuation ages, expected and actual interval amounts, signed deviation, cumulative gap, and material explanations. This makes it possible to distinguish a mechanical pattern variance from a plausible change in ultimate loss expectations. The comparison is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
Check Your Understanding
For each problem, use the fixed prior benchmark stated in the problem. Treat incurred changes as signed amounts.
- A prior selected ultimate is $500,000. Expected cumulative paid proportions are 40% at one age and 58% at the next. Actual cumulative paid is $205,000 and then $300,000. Find expected emergence, actual emergence, and the deviation. (Answer: Expected $90,000; actual $95,000; deviation +$5,000.)
- A prior selected incurred ultimate is $900,000. The expected incurred proportion rises from 75% to 83%. Actual incurred rises from $680,000 to $740,000. Find the expected emergence and deviation. (Answer: Expected $72,000; actual $60,000; deviation −$12,000.)
- A prior selected ultimate is $1,600,000, and the expected paid proportion rises from 88% to 93%. Actual paid emergence is $100,000. Calculate the deviation and the deviation as a percentage of the prior ultimate. (Answer: Expected $80,000; deviation +$20,000; 1.2500% of prior ultimate.)
- For one accident year, expected cumulative incurred is $420,000 at the first valuation and $480,000 at the second. Actual cumulative incurred is $430,000 and then $455,000. Calculate the interval deviation and the cumulative actual-minus-expected gap at the second valuation. (Answer: Expected emergence $60,000; actual emergence $25,000; deviation −$35,000; second-valuation cumulative gap −$25,000.)
- Two accident years have expected emergence of $150,000 and $250,000, with actual emergence of $180,000 and $200,000. Their prior ultimates are $1,000,000 and $2,000,000. Calculate the combined deviation and its percentage of combined prior ultimates, and state one limitation of the aggregate. (Answer: Combined expected $400,000; actual $380,000; deviation −$20,000; −0.6667% of combined prior ultimates; the total can mask opposite year-level deviations and may combine different valuation ages.)