Standardizing Paid Losses for Settlement Rates

Why Standardize Paid Losses for Settlement Rates?

Cumulative paid losses reflect both how many claims have closed and how much has been paid, on average, per closed claim. If claim-handling practices change the speed of closure, two otherwise comparable cohorts can show different paid amounts at the same development age. Comparing those amounts without accounting for closure rates can confuse a settlement-speed change with a change in paid severity.

A paid-loss settlement adjustment expresses the observed paid amount on a selected common claim-closure basis. It replaces the cohort’s observed cumulative closure proportion with a standard proportion while holding its effective paid severity per closed claim fixed. The result isolates the arithmetic effect of the selected closure-rate difference; it does not, by itself, remove changes in claim mix, inflation, coverage, or severity. This treatment is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.

Definition: For a claim cohort at a specified development age, let $N$ be the number of reported claims, $C$ the cumulative number of those claims closed by that age, and $P$ the cumulative paid losses for the same cohort and age. The cumulative closure proportion is $q=C/N$. The effective paid severity per closed claim is $s=P/C$, when $C>0$. A selected standard closure proportion is denoted $q^*$.

The definition of $s$ uses the cohort’s cumulative paid losses divided by its cumulative closed-claim count. It is an effective severity measure for this decomposition; depending on the data definition, cumulative paid losses can include payments on claims that remain open. Consistency matters: $P$, $C$, and $N$ must refer to the same claim cohort, valuation age, coverage basis, and claim-count definition.

Deriving the Adjustment

By definition, $q=C/N$ and $s=P/C$. Multiplying these quantities by the reported claim count gives the paid amount:

Nqs = N\left(\frac{C}{N}\right)\left(\frac{P}{C}\right) = P.

This identity shows why the decomposition works: the reported count multiplied by the fraction closed gives the closed count, and the closed count multiplied by effective paid severity gives cumulative paid losses. To put the cohort on the selected standard closure basis, replace $q$ by $q^*$, keeping $N$ and $s$ unchanged.

P^* = Nq^*s = Nq^*\frac{P}{C} = P\frac{q^*}{q}, \qquad q>0.

Thus the paid-loss adjustment factor is $q^*/q$. If the standard closure proportion exceeds the observed proportion, the factor exceeds one: the standardized paid amount is higher because it represents more claims closed at the same effective severity. If the standard is lower than the observed proportion, the adjusted amount is lower. If $q=q^*$, the factor is one and no adjustment is needed.

Key condition: The ratio formula requires an observed positive closure proportion and a meaningful, consistently measured paid severity. It standardizes closure only. It does not estimate the unpaid balance, predict ultimate losses, or correct for differences in severity, claim mix, or other changes.

Applying the Formula to a Cohort

A calculation should begin with the cohort and development age, not with the adjustment ratio. Establish the reported count, closed count, cumulative paid amount, and selected standard closure proportion on a common basis. Then calculate the observed closure proportion and effective paid severity before applying the factor. This makes it easier to check whether the adjustment changes only the intended component.

Fix the comparison basis.
Specify the cohort, development age, claim-count rules, and cumulative paid-loss basis. Use the same definitions for all inputs.

Measure observed settlement.
Calculate $q=C/N$, then calculate $s=P/C$. Check that $N>0$ and $C>0$.

Select the common closure basis.
Use the stated standard $q^*$, which must refer to the same relevant development age and claim definition.

Adjust and verify.
Calculate $P^*=P(q^*/q)$, or equivalently $Nq^*s$. Agreement between the two calculations checks the arithmetic.

Worked Example: Standardizing One Cohort to a Higher Closure Rate

A hypothetical insurer has 1,200 reported claims in a property cohort at a specified development age. Of these, 504 are closed, and cumulative paid losses are $4,536,000. Standardize the paid amount to a 50% cumulative closure proportion. Show the adjustment factor and check the result by rebuilding the amount from claim count and severity.

Method. Calculate the cohort’s observed closure proportion and effective paid severity, then replace only the closure proportion with the 50% standard.

Observed closure proportion. The number closed is 504 and the number reported is 1,200, so $q=504/1{,}200=0.42$, or 42%.

Effective paid severity. Divide cumulative paid losses by the closed-claim count: $s=\$4{,}536{,}000/504=\$9{,}000$ per closed claim.

Adjustment factor. The standard is $q^*=0.50$, so $q^*/q=0.50/0.42=50/42=25/21=1.190476\ldots$.

Adjusted paid losses. Applying the factor gives $\$4{,}536{,}000(25/21)=\$5{,}400{,}000$. The independent reconstruction gives $1{,}200(0.50)(\$9{,}000)=\$5{,}400{,}000$, so the two methods agree.

Common mistake avoided. Using $q/q^*$ would reduce the paid amount even though the standard closure proportion is higher than the observed one. The correct direction is upward: more claims are represented as closed at unchanged effective severity.

What the Adjustment Does—and Does Not—Hold Constant

The adjustment holds $s$, the effective paid severity per closed claim, constant by construction. It therefore answers a limited question: what would the cohort’s cumulative paid amount be if its closure proportion were $q^*$, with its observed effective paid severity unchanged? This is a standardization, not a claim-level prediction that identifies exactly which claims would have closed under different handling practices.

When comparing cohorts, calculate each cohort’s own $q$, $s$, and adjusted amount. A common target $q^*$ makes closure proportions comparable, but it does not make the severities comparable. If cohorts differ in average claim size, limits, deductibles, inflation level, or mix, those differences remain in the adjusted amounts. An adjusted paid amount can consequently remain higher for one cohort even after both are expressed at the same closure standard.

The selected standard should be appropriate to the comparison. It must match the development age and the definition of reported and closed claims. If a problem supplies a standard, use it as given. If a standard is constructed from several cohorts, follow the stated selection rule rather than substituting an unweighted average or an individual cohort’s rate without justification.

Worked Example: Applying One Common Standard to Two Cohorts

Two hypothetical liability cohorts are observed at the same development age on the same coverage basis. Cohort A has 900 reported claims, 540 closed claims, and $6,480,000 in cumulative paid losses. Cohort B has 1,000 reported claims, 450 closed claims, and $4,950,000 in cumulative paid losses. Standardize both to a 50% closure proportion. Compare their standardized paid amounts and explain what remains different.

Method. Apply the same $q^*=0.50$ to each cohort, using each cohort’s own observed effective paid severity.

Cohort A. Its observed closure proportion is $q_A=540/900=0.60$. Its effective paid severity is $s_A=\$6{,}480{,}000/540=\$12{,}000$. The adjustment factor is $0.50/0.60=5/6=0.833333\ldots$, so adjusted paid losses are $\$6{,}480{,}000(5/6)=\$5{,}400{,}000$. Checking directly: $900(0.50)(\$12{,}000)=\$5{,}400{,}000$.

Cohort B. Its observed closure proportion is $q_B=450/1{,}000=0.45$. Its effective paid severity is $s_B=\$4{,}950{,}000/450=\$11{,}000$. The adjustment factor is $0.50/0.45=10/9=1.111111\ldots$, so adjusted paid losses are $\$4{,}950{,}000(10/9)=\$5{,}500{,}000$. Checking directly: $1{,}000(0.50)(\$11{,}000)=\$5{,}500{,}000$.

Comparison. Cohort B’s standardized paid amount is $100,000 higher. The closure proportions are equal after adjustment, but the effective paid severities are not: $11,000 for B versus $12,000 for A. The result does not establish why the severities differ; it only shows that the closure-rate difference has been standardized.

Common mistake avoided. Applying the same adjustment factor to both cohorts would be incorrect because their observed closure proportions differ. Each factor must use that cohort’s own $q$.

Assumptions and Interpretation

The calculation is exact as an algebraic re-expression of the specified cohort totals. Its interpretation as a settlement-rate standardization depends on the data and comparison being suitable. In practice, the analyst should consider the following conditions before treating standardized amounts as meaningfully comparable:

Consistent cohort definitions: reported counts, closed counts, and paid amounts must include the same claims and use consistent rules for reopenings, transfers, and claim status.
Comparable development ages: the standard closure proportion must correspond to the age at which the cohort’s $q$ and $s$ are measured.
Positive closed count: if $C=0$, then $q=0$ and $s=P/C$ is undefined. The ratio adjustment cannot be applied using these definitions; a separate method or additional data is needed.
Severity interpretation: holding effective severity fixed is a standardization assumption, not evidence that claim severity would actually remain unchanged under a different settlement process.
Meaningful common standard: a target outside the relevant observed range may imply a substantial extrapolation. The arithmetic still follows the formula, but the interpretation may be less credible.

A further limitation is that a cohort-level ratio cannot show which particular claims account for differences in closure timing or payments. If the claim population changes substantially between cohorts, the observed effective severity may change because of mix as well as payment behavior. This adjustment alone does not separate those effects.

Worked Example: Constructing and Applying a Pooled Standard

A hypothetical analyst is given two benchmark cohorts at the same development age. Benchmark 1 has 800 reported claims and 400 closed claims; Benchmark 2 has 1,200 reported claims and 540 closed claims. A target cohort has 750 reported claims, 315 closed claims, and $2,835,000 in cumulative paid losses. For this calculation, define the common standard as the pooled closure proportion across the two benchmarks. Find the target cohort’s standardized paid amount.

Method. First pool the benchmark counts to obtain the stated common standard. Then calculate the target’s severity and apply the standardization formula.

Standard closure proportion. The pooled benchmark has $400+540=940$ closed claims and $800+1{,}200=2{,}000$ reported claims. Therefore, $q^*=940/2{,}000=0.47$, or 47%.

Target observed closure proportion. $q=315/750=0.42$, or 42%.

Target effective paid severity. $s=\$2{,}835{,}000/315=\$9{,}000$ per closed claim.

Adjustment factor and adjusted amount. $q^*/q=0.47/0.42=47/42=1.119047\ldots$. Thus, $\$2{,}835{,}000(47/42)=\$3{,}172{,}500$. The direct check is $750(0.47)(\$9{,}000)=\$3{,}172{,}500$.

Common mistake avoided. The pooled standard is a ratio of pooled counts, $940/2{,}000$, not the unweighted mean of the two benchmark closure proportions. The unweighted mean would be $(0.50+0.45)/2=0.475$, which is not the standard specified here.

Common Exam Trap: Reversing the Ratio

The most direct trap is reversing the adjustment ratio. To replace observed closure $q$ with standard closure $q^*$, the factor must be $q^*/q$, not $q/q^*$. A quick directional check catches this: if $q^*>q$, the standardized paid amount should increase when severity is held constant.

For the target cohort in the pooled-standard example, the incorrect factor would be $0.42/0.47=42/47$, approximately 0.8936. This would produce $\$2{,}835{,}000(42/47)\approx \$2{,}533{,}404.26$, a decrease despite raising closure from 42% to 47%. The correct factor is $47/42$, giving $3,172,500. Another frequent error is to adjust severity as well as paid losses: doing so changes a second component and no longer performs the stated closure-only standardization.

Key takeaway: For a cohort with positive observed closure proportion, standardize cumulative paid losses using $P^*=P(q^*/q)=Nq^*s$. The formula changes the closure component to the common target and retains the cohort’s effective paid severity.

Check Your Understanding

For each problem, keep the cohort, development age, and claim-count basis consistent. Answers are provided for checking.

A hypothetical specialty insurer observes 600 reported claims, 270 closed claims, and $2,430,000 cumulative paid losses at a specified age. Standardize to a 50% closure proportion. What are the observed closure proportion, effective paid severity, adjustment factor, and adjusted paid amount? (Answer: 45%; $9,000; $10/9=1.111111\ldots$; $2,700,000.)
A hypothetical equipment policy cohort has 1,500 reported claims, 900 closed claims, and $8,100,000 cumulative paid losses. The standard closure proportion is 55%. Calculate the adjusted paid amount and state whether it is above or below the observed amount. (Answer: $q=60\%$, factor $55/60=11/12$, adjusted paid amount $7,425,000, below observed.)
At the same development age, benchmark cohorts have 500 reported and 275 closed claims, and 1,500 reported and 750 closed claims. A target cohort has 800 reported claims, 320 closed claims, and $3,840,000 cumulative paid losses. Use the pooled benchmark closure proportion as the standard. Find the standard and the target’s adjusted paid amount. (Answer: standard $1{,}025/2{,}000=51.25\%$; target severity $12,000; factor $0.5125/0.40=1.28125$; adjusted paid amount $4,920,000.)
A cohort has an observed closure proportion of 48% and cumulative paid losses of $5,760,000. Its selected standard is 42%. What is the paid-loss adjustment factor and adjusted amount? Explain the direction. (Answer: factor $42/48=0.875$; adjusted amount $5,040,000, lower because the standard closure proportion is lower.)
A cohort has 700 reported claims, 0 closed claims, and positive cumulative payments on open claims. Can the formula $P^*=P(q^*/q)$ be applied using these data? State the key reason. (Answer: No. Here $q=0$, so the ratio is undefined, and $s=P/C$ is also undefined because $C=0$.)

What You'll Learn