Allowing for Changes in Limits and Coverage

Why Coverage Terms Must Be Reconciled First

An exposure curve describes how expected covered loss changes with the limit or layer being considered. Historical claim payments, however, reflect the contracts that produced them. A change in limit, deductible, attachment point, or definition of covered loss can change the payment even when the underlying event is identical. Before using historical data to estimate a curve for prospective coverage, identify what the historical payment measures and whether it can be translated to the target terms.

This is a coverage-mapping problem, not merely a matter of relabeling a policy limit. Historical and prospective dollar amounts must also be on a consistent cost level. The earlier material on severity trend addresses that adjustment; here the focus is the contract terms applied to the losses.

Definition: Let $X$ be ground-up loss for one covered occurrence. For a policy with deductible $d$ and a maximum insurer payment $c$, the insurer payment is $$ Y^L=\min\bigl((X-d)^+,c\bigr), \qquad (z)^+=\max(z,0). $$ This formula assumes a per-occurrence deductible, a cap on the insurer's payment, and no other coverage features that change the payment. It does not by itself describe aggregate limits, coinsurance, exclusions, or expenses that erode the limit.

The units and meaning of “limit” need particular care. A contract may state a maximum insurer payment, or it may state the maximum covered loss before applying the deductible. If a total-loss ceiling is $U$ and the deductible is $d<U$, the maximum insurer payment is $U-d$, not $U$. A layer with attachment $a$ and width $w$ has payment $\min((X-a)^+,w)$; its top is $a+w$. Stating these terms explicitly avoids comparing unlike contracts.

Y^L=\min\bigl((X-d)^+,c\bigr) \qquad\text{and, for a total-loss ceiling }U>d,\qquad Y^L=\min\bigl((X-d)^+,U-d\bigr).

The limited expected loss at a payment cap $c$ is $E[Y^L]$. An increased limits factor compares limited expected losses under specified limits on a common coverage and loss basis. If the deductible or coverage definition changes too, the ratio may still be a comparison of two expected payments, but it is not an isolated effect of increasing the limit. The distinction matters when interpreting or applying the resulting curve.

A Direct Translation When Ground-Up Losses Are Known

The cleanest case is one in which the ground-up loss for each historical claim is available, and the target contract terms can be applied directly. The formula for the historical payment is useful for checking what was paid; the target payment must then be recalculated from $X$, rather than inferred by scaling the historical payment.

Worked Example: Changing Both Deductible and Payment Limit

A hypothetical insurer reviews four ground-up occurrence losses, in thousands of dollars: $8, 40, 100,$ and $210$. The historical contract has a deductible of $10$ and a maximum insurer payment of $90$. The proposed contract has a deductible of $25$ and a maximum insurer payment of $175$. Calculate the payment for each loss under both contracts and compare their sample mean payments.

Method. Because each ground-up loss is known, apply the payment formula separately under each set of terms. All amounts below are in thousands of dollars.

For the historical contract, $Y^L=\min((X-10)^+,90)$:

$X=8:\ \min(0,90)=0$.
$X=40:\ \min(30,90)=30$.
$X=100:\ \min(90,90)=90$.
$X=210:\ \min(200,90)=90$.

The historical payments sum to $0+30+90+90=210$, so their sample mean is $210/4=52.5$, or $52,500.

For the proposed contract, $Y^L=\min((X-25)^+,175)$:

$X=8:\ \min(0,175)=0$.
$X=40:\ \min(15,175)=15$.
$X=100:\ \min(75,175)=75$.
$X=210:\ \min(185,175)=175$.

The proposed payments sum to $0+15+75+175=265$, so their sample mean is $265/4=66.25$, or $66,250. The increase is $66.25-52.5=13.75$ thousand dollars per claim in this four-claim sample.

Check. Each target payment is nonnegative and no greater than the target cap of $175$; the claim below the new deductible produces no payment. This comparison reflects both a deductible change and a limit change. It is not an estimate of the effect of changing the limit alone. A common mistake is to multiply each historical payment by a limit ratio; that fails here because the higher deductible also changes which part of each loss is covered.

Translate the Contract’s Definition of Limit

A limit written on the policy may refer to a total covered-loss ceiling rather than the insurer’s maximum payment. Those interpretations produce different payments once a deductible is applied. Convert the contract wording to a payment formula before using the claim data.

Worked Example: Total-Loss Ceiling Versus Insurer-Payment Cap

A hypothetical policy has a deductible of $20,000 and a historical total-loss ceiling of $100,000. The proposed policy has the same deductible and a total-loss ceiling of $160,000. For known ground-up losses of $50,000, $110,000, and $180,000, calculate payments under each contract and compare the sample means.

Method. Since each stated limit is a ceiling on total covered loss, the insurer’s payment cap is the ceiling minus the deductible. The historical payment cap is $100-20=80$ thousand dollars, and the proposed payment cap is $160-20=140$ thousand dollars.

Historical payments are $\min((X-20)^+,80)$:

For $X=50$, the payment is $\min(30,80)=30$.
For $X=110$, the payment is $\min(90,80)=80$.
For $X=180$, the payment is $\min(160,80)=80$.

The historical total is $30+80+80=190$ thousand dollars, and the mean is $190/3=63.333\ldots$ thousand dollars.

Proposed payments are $\min((X-20)^+,140)$:

For $X=50$, the payment is $\min(30,140)=30$.
For $X=110$, the payment is $\min(90,140)=90$.
For $X=180$, the payment is $\min(160,140)=140$.

The proposed total is $30+90+140=260$ thousand dollars, and the mean is $260/3=86.666\ldots$ thousand dollars. Rounded to the nearest dollar, the means are $63,333 and $86,667.

Common mistake and correction. Treating the historical $100,000 total-loss ceiling as a $100,000 insurer-payment cap would produce payments of $30,000, $90,000, and $100,000. That is incorrect because the deductible uses $20,000 of the stated total-loss ceiling. The correct historical insurer-payment cap is $80,000, giving payments of $30,000, $80,000, and $80,000. The final means are consistent with the computed totals: $190/3$ and $260/3$ thousand dollars.

Attachment Points and Coverage Definitions

When an attachment point changes, a claim can move into or out of the covered layer even if the layer width increases. For a simple per-occurrence layer attaching at $a$ with width $w$, the payment is $\min((X-a)^+,w)$. This assumes the same definition of covered loss under both contracts and no other restrictions. An attachment point should not be confused with a deductible unless the contract’s payment mechanics make them equivalent.

Other terms can change what counts against the limit. For example, defense expenses may be included within the limit and erode the amount available for indemnity, or they may be paid outside the indemnity limit. A curve based on indemnity payments alone cannot be assumed to represent total payments under both arrangements. If the data do not record the relevant expenses or contract terms, a direct claim-level translation may not be possible.

Worked Example: Defense Costs Inside Versus Outside the Limit

A hypothetical historical policy has a $100,000 combined limit for indemnity and defense expenses. Assume defense expense of $15,000 is paid first and erodes that limit. A proposed policy instead has a $100,000 indemnity limit, with the same $15,000 defense expense paid outside that limit. There is no deductible. For indemnity losses of $70,000, $95,000, and $130,000, calculate mean indemnity payment and mean total payment under each arrangement.

Method. Under the historical terms, defense uses $15,000 of the $100,000 combined limit, leaving $100-15=85$ thousand dollars for indemnity. Historical indemnity payment is therefore $\min(I,85)$, where $I$ is the indemnity loss.

For $I=70$, indemnity payment is $70$; total payment is $70+15=85$.
For $I=95$, indemnity payment is $85$; total payment is $85+15=100$.
For $I=130$, indemnity payment is $85$; total payment is $85+15=100$.

Historical indemnity payments total $70+85+85=240$ thousand dollars, with mean $240/3=80$ thousand dollars. Historical total payments are $85+100+100=285$ thousand dollars, with mean $285/3=95$ thousand dollars.

Under the proposed terms, indemnity payment is $\min(I,100)$, and the $15,000 defense expense is paid separately:

For $I=70$, indemnity payment is $70$; total payment is $70+15=85$.
For $I=95$, indemnity payment is $95$; total payment is $95+15=110$.
For $I=130$, indemnity payment is $100$; total payment is $100+15=115$.

Proposed indemnity payments total $70+95+100=265$ thousand dollars, with mean $265/3=88.333\ldots$ thousand dollars. Proposed total payments sum to $85+110+115=310$ thousand dollars, with mean $310/3=103.333\ldots$ thousand dollars. Rounded to the nearest dollar, the indemnity means are $80,000 and $88,333; the total-payment means are $95,000 and $103,333.

Check and trap. Under the historical contract, indemnity plus defense never exceeds $100,000; under the proposed contract, indemnity can reach $100,000 in addition to defense. Applying the $100,000 limit to indemnity under both contracts would ignore the historical erosion provision and overstate historical indemnity payments.

When Historical Payments Cannot Identify the Target

A known ground-up loss can generally be evaluated under a newly specified deductible and limit using the payment formula, provided the covered-loss definition and other necessary terms are known. Historical payment data alone are different. If the historical contract capped the payment, several different ground-up losses may all produce that same observed payment. The payment does not reveal how far the underlying loss exceeded the cap.

For example, with a $10,000 deductible and a $90,000 maximum insurer payment, any ground-up loss of at least $100,000 produces a $90,000 payment under the simple formula. A historical payment of $90,000 therefore does not distinguish a $100,000 ground-up loss from a $300,000 loss. If the prospective coverage pays above that historical cap, that missing information matters. Censoring corrections and exposure-curve methods can address such limitations under their assumptions, but changing the stated limit does not recover the unobserved severity by itself.

A short derivation shows why the payment formula has its form. Below the deductible, the insurer pays nothing; above it, the amount before the cap is $X-d$. The payment cannot be negative and cannot exceed $c$, so it is the lesser of the nonnegative amount above the deductible and $c$:

X\le d \ \Longrightarrow\ Y^L=0, \qquad d<X<d+c \ \Longrightarrow\ Y^L=X-d, \qquad X\ge d+c \ \Longrightarrow\ Y^L=c.

These three cases are exactly $\min((X-d)^+,c)$. The derivation also identifies the information problem: when $X\ge d+c$, the observed payment stays at $c$ regardless of the size of $X$.

Practical rule: Before using a historical payment as evidence for a prospective exposure curve, document the loss definition, deductible or attachment, limit meaning and amount, and any features that alter the payment. Translate only when the data and terms support the translation. If an essential feature or the ground-up loss above a historical cap is unknown, state the limitation rather than treating the observations as target-coverage payments.

A Coverage-Reconciliation Sequence

A disciplined sequence makes the comparison auditable and helps separate a change in limits from a change in the risk or the contract.

Specify the target payment.
Write the target deductible, attachment, payment cap or total-loss ceiling, and covered-loss definition in unambiguous terms.

Decode historical terms.
Identify whether the historical limit caps insurer payment or total covered loss, and note any defense-cost, aggregate, or other provisions that affect payment.

Check what the data contain.
Determine whether ground-up losses and the relevant expense or coverage details are available, or whether the observations are capped historical payments.

Translate or qualify.
Apply the target payment terms to eligible losses when the needed information is present. Otherwise identify censoring or coverage mismatch and do not present the raw historical payments as target payments.

This process is aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus. Its central exam trap is to compare numbers that look like limits without first establishing what those numbers cap. A reliable solution writes the payment function, states its assumptions, and checks whether the supplied observations contain the information that function requires.

Check Your Understanding

For each hypothetical situation, identify the correct payment or the information needed before making the comparison.

A policy has a $12,000 deductible and a $48,000 maximum insurer payment. For a ground-up loss of $75,000, calculate the insurer payment. (Answer: $48,000)
A contract has a $90,000 total-loss ceiling and a $15,000 deductible. A ground-up loss is $100,000. Calculate the payment, and state the insurer-payment cap. (Answer: The payment is $75,000; the insurer-payment cap is $75,000.)
A hypothetical layer attaches at $40,000 and has a width of $70,000. For losses of $30,000, $85,000, and $140,000, calculate the three layer payments and their mean. (Answer: Payments are $0, $45,000, and $70,000; the mean is $38,333.33.)
A historical contract has a $50,000 deductible and a $100,000 insurer-payment cap. The observed historical payment is $100,000, and a proposed contract pays above that cap. Can the historical payment alone determine the proposed payment? Explain. (Answer: No. It only establishes that the ground-up loss is at least $150,000 under the stated simple terms; the amount above that threshold is unknown.)
A $120,000 combined limit includes $20,000 of defense expense paid first. A hypothetical indemnity loss is $115,000. Calculate historical indemnity and total payments. Under a proposed $120,000 indemnity limit with defense outside the limit, calculate proposed indemnity and total payments. (Answer: Historical indemnity is $100,000 and total payment is $120,000; proposed indemnity is $115,000 and total payment is $135,000.)

What You'll Learn