Case Reserve Adequacy And Claim Reporting

Why Case Reserves and Reporting Matter

A collateral estimate for a large deductible program depends on more than the deductible and the claims already paid. It also depends on what has happened to each claim, what remains expected to happen, and which claims have not yet been reported. A case reserve that is too low, a closed claim that later reopens, or a late-reported claim can each make an estimate of unpaid deductible obligations unreliable.

The key is to separate two related but different quantities. A gross claim reserve estimates future covered claim payments. An outstanding deductible obligation estimates the insured’s remaining share under the deductible terms, after accounting for reimbursements already credited. A gross reserve does not automatically equal an unpaid deductible amount: the deductible caps the insured’s share for each claim, and prior payments may already have used part or all of that share.

Definition: For a claim with ultimate covered loss $X$ and a per-claim deductible $d$, the insured’s ultimate deductible-layer share, before any aggregate provisions or other contract terms, is $g(X)=\min(X,d)$. If the claim has information $I$ available at the valuation date, and $Q$ is the amount already credited against the insured’s deductible obligation, the estimated remaining obligation is $\max\{E[g(X)\mid I]-Q,0\}$. Here, $Q$ means amounts actually credited under the contract; a billed but unpaid amount may need separate treatment in the program’s records.

This claim-level expression assumes that $X$ is the loss measure used by the contract: for example, indemnity alone or indemnity plus allocated expenses. It also assumes a per-claim deductible and no additional aggregate deductible, aggregate limit, or other interaction. Where those terms apply, the estimate must reflect them rather than applying this simplified expression by itself.

Case Reserves Are Not Deductible Reserves

A case reserve is an estimate of unpaid loss on a reported claim, generally prepared using claim-specific information. For a simple gross-loss view, incurred loss is paid loss plus the case reserve. The reserve can be compared with an estimate of expected future gross payments to assess adequacy, but this comparison is not itself a deductible calculation.

For deductible estimation, the relevant quantity is the expected retained share of ultimate loss, $E[\min(X,d)\mid I]$, less the amount already credited. The cap matters: once a claim’s loss exceeds the deductible, additional loss increases gross incurred loss but does not increase the insured’s per-claim deductible share. Conversely, when a claim is expected to fall below the deductible, the insured’s share is the full claim loss, not the deductible.

\text{Expected remaining deductible obligation} = \max\left\{E[\min(X,d)\mid I]-Q,\;0\right\}.

The expected value must be taken after applying the deductible to each possible ultimate loss. In general, $E[\min(X,d)]$ is not equal to $\min(E[X],d)$. The latter calculation applies the deductible to an average loss, which can hide the mix of claims below and above the deductible.

Why the Deductible Must Be Applied Before Averaging

For a nonnegative loss $X$, the deductible-layer share can also be written as an integral of indicators. For each amount $x$ between zero and $d$, the claim contributes one unit of loss at that level exactly when $X>x$. Thus:

\min(X,d)=\int_0^d \mathbf{1}_{\{X>x\}}\,dx.

Taking expectations and interchanging expectation and integration for this bounded nonnegative quantity gives:

E[\min(X,d)] = \int_0^d E[\mathbf{1}_{\{X>x\}}]\,dx = \int_0^d P(X>x)\,dx = \int_0^d S_X(x)\,dx.

This justifies calculating expected deductible-layer loss from the loss distribution, rather than from the mean loss alone. It also shows why assumptions about the distribution below and around the deductible matter. When information is available for a particular claim, use the distribution conditional on that information, not an unconditioned portfolio average.

Worked Example: Testing a Case Reserve Against the Deductible Estimate

A hypothetical claim has a $25,000 deductible. The insurer has paid $8,000, and the insured has reimbursed that amount, so $Q=\$8,000$. Based on current claim information, ultimate covered loss $X$ is $10,000 with probability 0.20, $30,000 with probability 0.50, and $70,000 with probability 0.30. The gross case reserve is $22,000. Estimate the remaining deductible obligation and assess the gross case reserve. Assume no other contract provisions and no discounting.

Method and assumptions. Apply the deductible to each possible ultimate loss, then take the expectation and subtract the credited reimbursement. For gross reserve adequacy, compare the case reserve with expected future gross loss. The stated probabilities sum to $0.20+0.50+0.30=1.00$.

Step 1: Expected ultimate gross loss.

E[X]=0.20(10{,}000)+0.50(30{,}000)+0.30(70{,}000) =2{,}000+15{,}000+21{,}000 =\$38{,}000.

Step 2: Compare expected future gross loss with the case reserve. Since $8,000 has already been paid, expected future gross loss is $\$38{,}000-\$8{,}000=\$30{,}000$. The $22,000 case reserve is therefore $8,000 below this conditional expected future gross loss.

Step 3: Calculate expected ultimate deductible-layer loss. The deductible-layer amounts for the three possible ultimate losses are respectively $\min(10{,}000,25{,}000)=\$10{,}000$, $\min(30{,}000,25{,}000)=\$25{,}000$, and $\min(70{,}000,25{,}000)=\$25{,}000$. Therefore:

E[\min(X,d)] =0.20(10{,}000)+0.50(25{,}000)+0.30(25{,}000) =2{,}000+12{,}500+7{,}500 =\$22{,}000.

Step 4: Deduct the amount already credited. The estimated remaining obligation is $\$22{,}000-\$8{,}000=\$14{,}000$. The result is positive, so the floor at zero does not change it.

Check and common mistake. The $22,000 case reserve is a gross future-loss estimate, whereas $14,000 is the estimated remaining deductible obligation. They differ because the deductible caps the insured’s ultimate share and $8,000 has already been reimbursed. Applying the deductible to expected ultimate loss would give $\min(\$38{,}000,\$25{,}000)-\$8{,}000=\$17{,}000$, which is wrong: it caps the mean rather than each possible loss. The correct claim-level expectation is $14,000.

Late Reporting and Unreported Claims

A reported-claim inventory cannot, by itself, capture obligations from losses that have occurred but are not yet reported. An estimate for unreported claims needs assumptions about both the number of claims expected to be reported later and the severity distribution for those claims. The severity distribution should reflect the unreported population when reporting delay is related to claim size or type.

For example, if expected claim count is $E[N]$, the probability that a claim has been reported by the valuation date is $q$, and reporting is independent of severity, expected unreported claim count is $E[N](1-q)$. If $X$ denotes the ultimate loss of an unreported claim, the expected deductible obligation from these claims, before any payments, is:

E[N](1-q)\,E[\min(X,d)].

This factorization requires the same reporting probability and conditional severity assumptions to apply consistently. If reporting probability depends on severity or claim characteristics, estimate the count and expected deductible share for appropriate groups, or use the conditional distribution for unreported claims. Multiplying by an overall reporting percentage while using a reported-claim severity distribution can bias the result.

Worked Example: Estimating Obligations for Late-Reported Claims

A hypothetical program expects 12 claims in a period. By the valuation date, 75% are expected to have been reported. For each unreported claim, ultimate severity is $10,000 with probability 0.40 and $40,000 with probability 0.60. The per-claim deductible is $25,000. No payments have been made on the unreported claims. Estimate their expected total deductible obligation. Assume reporting is independent of severity, no aggregate provisions apply, and there is no discounting.

Method and assumptions. First estimate the number of unreported claims, then calculate expected deductible-layer loss per such claim. The expected number is sufficient for the expected total: independence among claim counts is not needed for linearity of expectation.

Step 1: Expected unreported count. The unreported proportion is $1-0.75=0.25$, so expected unreported claims are $12(0.25)=3$.

Step 2: Apply the deductible to each severity. The deductible-layer amounts are $10,000 for a $10,000 loss and $25,000 for a $40,000 loss. Thus expected deductible-layer loss per unreported claim is:

0.40(10{,}000)+0.60(25{,}000) =4{,}000+15{,}000 =\$19{,}000.

Step 3: Calculate the portfolio estimate.

3(\$19{,}000)=\$57{,}000.

Check and common mistake. The expected unreported count is three, and the per-claim retained expectation is $19,000, so the total is $57,000. A common error is to use the $25,000 deductible for every unreported claim, which would overstate the estimate because a $10,000 claim does not reach the deductible.

Claim Status Changes and Reopened Claims

Claim status is an operational label, not a guarantee that the estimated obligation is zero. A claim marked closed may reopen after new medical information, litigation activity, or a late invoice. Likewise, an open claim may have a case reserve that changes substantially when facts develop. A valuation process should use current claim information and retain a way to identify reopenings and status changes after the prior valuation date.

When a claim has both paid loss and expected future loss, combine them to form the ultimate-loss scenarios before applying the deductible. Then subtract reimbursements already credited. This avoids treating a paid amount as if it were still outstanding, and avoids assuming that a claim’s status alone determines whether its deductible is exhausted.

Worked Example: Reopening a Claim With a Paid Reimbursement

A hypothetical claim with a $15,000 deductible was closed after $6,000 of covered loss had been paid and reimbursed by the insured. New information leads to reopening the claim. Conditional on the new information, future covered loss is $4,000 with probability 0.50 or $20,000 with probability 0.50. Estimate the remaining deductible obligation and compare the expected future loss with a $9,000 gross case reserve. Assume no other provisions and no discounting.

Method and assumptions. Treat the $6,000 paid loss as part of ultimate loss in each scenario, apply the deductible to total ultimate loss, and subtract the $6,000 reimbursement already credited. The two future-loss probabilities sum to 1.00.

Step 1: Construct ultimate-loss scenarios. With future loss of $4,000, ultimate loss is $\$6{,}000+\$4{,}000=\$10{,}000$. With future loss of $20,000, ultimate loss is $\$6{,}000+\$20{,}000=\$26{,}000$.

Step 2: Calculate expected future gross loss.

E[\text{future loss}] =0.50(4{,}000)+0.50(20{,}000) =2{,}000+10{,}000 =\$12{,}000.

The $9,000 gross case reserve is $3,000 below this expected future gross loss.

Step 3: Calculate the expected ultimate deductible-layer share. The deductible-layer amounts are $\min(10{,}000,15{,}000)=\$10{,}000$ and $\min(26{,}000,15{,}000)=\$15{,}000$. Therefore:

E[\min(X,d)] =0.50(10{,}000)+0.50(15{,}000) =\$12{,}500.

Step 4: Subtract the reimbursement already credited. The estimated remaining obligation is $\$12{,}500-\$6{,}000=\$6{,}500$.

Check and common mistake. The reopened status prompts a revised estimate, but it does not erase the $6,000 already reimbursed. The final estimate is $6,500. Treating the claim as if it had no prior payment would overstate the remaining obligation by $6,000; treating the closed status as proof of no remaining obligation would miss the reopened exposure.

Data Controls for a Reliable Estimate

Collateral calculations are only as dependable as their claim data and assumptions. A useful review distinguishes estimation uncertainty from data defects and checks that the amounts in the calculation match the contract’s reimbursement basis. The following controls help identify weaknesses before an estimate is used:

Reconcile claim records. Compare claim counts, paid amounts, case reserves, and reimbursement credits with the relevant claim and financial records. Investigate duplicate claims, missing payments, and unexplained changes between valuation dates.
Review status transitions. Identify recently closed, reopened, denied, or reclassified claims. Confirm that a status change has not removed a claim from a valuation extract while leaving an obligation unresolved.
Test late-reporting assumptions. Compare expected and actual reporting patterns by accident period and, where useful, by claim type or severity group. Consider whether late reports have a different severity profile from claims already reported.
Check deductible treatment. Confirm that the deductible is applied to the contractually defined loss measure, and that all payments already credited are subtracted once—not omitted and not counted twice.
Separate estimates from known amounts. Label paid, reserved, billed, reimbursed, and projected amounts distinctly. A billed but unpaid reimbursement is not the same data item as a payment already credited against the insured’s obligation.

An expected unpaid-loss estimate is an input to collateral analysis, not a guarantee that a chosen collateral amount will be adequate in every outcome. The earlier material on collateral sizing and stress testing distinguishes expected amounts from estimates designed to withstand adverse variation. Poor reserve or reporting data weaken either kind of calculation, so the data checks remain important even when an additional margin is used.

Key takeaway: Estimate the insured’s share by applying the deductible to claim-level ultimate-loss scenarios, then account for reimbursements already credited. For unreported claims, use assumptions appropriate to the unreported population. Treat claim status and case reserves as evidence to review—not as substitutes for estimating the outstanding deductible obligation.

Check Your Understanding

For each problem, assume no discounting or aggregate provisions unless stated. Amounts are hypothetical.

A claim has a $18,000 deductible, no credited reimbursement, and ultimate loss of $10,000 with probability 0.25 or $30,000 with probability 0.75. What is its expected deductible obligation? (Answer: $16,000)
A program expects 20 claims, and 90% are expected to be reported by the valuation date. For each unreported claim, ultimate loss is $12,000 with probability 0.50 or $40,000 with probability 0.50. The deductible is $25,000. Assuming reporting is independent of severity, estimate the expected obligation from unreported claims. (Answer: $37,000)
Two claims each have a $20,000 deductible and no payments. Claim A has ultimate loss $8,000 with probability 0.50 and $32,000 with probability 0.50. Claim B is certain to have ultimate loss $24,000. What is their combined expected deductible obligation? (Answer: $34,000)
A portfolio expects 50 claims, of which 80% are reported by the valuation date. Each unreported claim has ultimate loss $20,000 with probability 0.30 or $50,000 with probability 0.70. With a $30,000 deductible and no payments on unreported claims, estimate their expected total deductible obligation. (Answer: $270,000)
A claim with a $25,000 deductible has $10,000 of covered loss paid and reimbursed. It reopens; conditional on current information, future loss is $8,000 with probability 0.50 or $30,000 with probability 0.50. What is the estimated remaining deductible obligation? (Answer: $11,500)

What You'll Learn