Deriving The Increased Limits Factor

Why an Increased Limits Factor Is a Ratio

A policy limit changes the amount the insurer may pay on a claim. To compare the expected loss cost at two limits, calculate the expected payment under each limit using the same underlying claims and rating assumptions. The increased limits factor (ILF) expresses the higher-limit expected loss relative to the expected loss at a selected base limit.

The ratio is meaningful only if the two expected losses are measured on a shared basis. The severity distribution, exposure definition, valuation date, and other coverage terms must be consistent; the limit being compared is the intended difference. This tutorial considers per-claim limits and no deductible. All examples are hypothetical, use claim amounts at a common valuation date, and involve no discounting or mortality contingency.

Definition: Let $X$ be a nonnegative claim severity with a finite expected limited loss. At per-claim limit $u$, the insurer’s payment is $\min(X,u)$, and the limited expected loss is $L(u)=E[\min(X,u)]$. For a higher limit $u$ and a positive base limit $b$, the increased limits factor is $$ \operatorname{ILF}(u;b)=\frac{L(u)}{L(b)} =\frac{E[\min(X,u)]}{E[\min(X,b)]}, $$ provided $L(b)>0$. The severity distribution and all loss-basis assumptions are the same in numerator and denominator.

The numerator and denominator are expected dollars per claim on the selected severity basis. Thus, an ILF is dimensionless: it indicates the relative expected loss cost at the higher limit, not the premium or the dollar increase in expected loss. If the base limit itself is the reference, its ILF is 1.

Why the Shared Basis Matters

The formula compares two versions of the payment for the same modeled claim. At each limit, cap the claim severity at that limit, take the expected value, and then divide the higher-limit result by the base-limit result. No claim should be counted differently in the numerator and denominator except as required by the change in limit.

For a portfolio, the same logic applies when claims share a common severity distribution and frequency basis. Suppose the number of claims per exposure unit is $N$, with finite mean $E[N]$, and severities are identically distributed as $X$, independent of $N$. Conditional on $N$, expected payments at limit $u$ equal $N E[\min(X,u)]$. Taking expectations gives:

E\!\left[\sum_{j=1}^{N}\min(X_j,u)\right] =E[N]E[\min(X,u)].

At the base limit, the corresponding expression is $E[N]E[\min(X,b)]$. If the same claim frequency and exposure basis apply at both limits, the common $E[N]$ cancels in the ratio. This cancellation is not a license to mix bases: if the frequency, exposure mix, severity assumptions, or valuation adjustments differ between the two calculations, the ratio no longer isolates the effect of the limit.

The derivation is therefore direct. Let $C(u)$ denote expected capped claim payments per exposure unit. Under the assumptions just stated, $C(u)=E[N]L(u)$. Then:

\frac{C(u)}{C(b)} = \frac{E[N]L(u)}{E[N]L(b)} = \frac{L(u)}{L(b)} = \operatorname{ILF}(u;b), \qquad C(b)>0.

This result applies to per-claim limits with a common claim-count and severity basis. It does not by itself handle aggregate limits, interactions among coverage terms, or a change in the population’s mix of risks. Those situations can change the payment structure or the underlying loss distribution, so they require a calculation on the actual applicable basis.

Calculating an ILF from a Severity Distribution

For a discrete severity distribution, find the payment at each limit for every possible severity, multiply each payment by its probability, and add. Do this separately at the base and higher limits. The final division comes only after both limited expected losses have been calculated.

Worked Example: A Three-Point Severity Distribution

A hypothetical liability portfolio has claim severity $X$ with the following distribution: $P(X=\$8{,}000)=0.50$, $P(X=\$20{,}000)=0.30$, and $P(X=\$50{,}000)=0.20$. Calculate the ILF for a $\$50{,}000$ per-claim limit relative to a $\$20{,}000$ base limit. The limits are the only terms changed. There is no discounting or mortality contingency.

Method. Compute the expected capped payment at each limit using the stated severity distribution. At the base limit, the three claim payments are $\$8{,}000$, $\$20{,}000$, and $\$20{,}000$. Therefore:

L(\$20{,}000) =0.50(\$8{,}000)+0.30(\$20{,}000)+0.20(\$20{,}000) =\$4{,}000+\$6{,}000+\$4{,}000 =\$14{,}000.

At the higher limit, the claim payments are $\$8{,}000$, $\$20{,}000$, and $\$50{,}000$. Thus:

L(\$50{,}000) =0.50(\$8{,}000)+0.30(\$20{,}000)+0.20(\$50{,}000) =\$4{,}000+\$6{,}000+\$10{,}000 =\$20{,}000.

The factor is:

\operatorname{ILF}(\$50{,}000;\$20{,}000) =\frac{\$20{,}000}{\$14{,}000} =\frac{10}{7} =1.429.

The check is that the factor exceeds 1: the higher limit cannot reduce a claim payment. The calculation avoids the common mistake of assigning a payment above the $\$50{,}000$ limit; here the largest modeled loss equals that limit.

An ILF can also be calculated after a common adjustment to the severity basis, such as applying the same trend to every modeled severity before capping. The adjustment must be applied consistently to the entire distribution before both limited expected losses are computed.

Worked Example: Applying Common Severity Trend Before Capping

A hypothetical model has pre-trend severities of $\$10{,}000$ with probability 0.60 and $\$30{,}000$ with probability 0.40. Apply a common 20% severity trend, then calculate the ILF for a $\$60{,}000$ limit relative to a $\$20{,}000$ base limit. The probabilities are unchanged, the stated limits are on the post-trend basis, and there is no discounting or mortality contingency.

Method. Trend each severity by multiplying it by $1.20$. The post-trend severities are $\$10{,}000(1.20)=\$12{,}000$ and $\$30{,}000(1.20)=\$36{,}000$.

At the $\$20{,}000$ base limit, the capped payments are $\$12{,}000$ and $\$20{,}000$. Hence:

L(\$20{,}000) =0.60(\$12{,}000)+0.40(\$20{,}000) =\$7{,}200+\$8{,}000 =\$15{,}200.

At the $\$60{,}000$ limit, both post-trend severities are below the limit. The payments are $\$12{,}000$ and $\$36{,}000$, so:

L(\$60{,}000) =0.60(\$12{,}000)+0.40(\$36{,}000) =\$7{,}200+\$14{,}400 =\$21{,}600.

Therefore:

\operatorname{ILF}(\$60{,}000;\$20{,}000) =\frac{\$21{,}600}{\$15{,}200} =\frac{27}{19} =1.421.

The factor is above 1, as expected. The key check is that neither post-trend severity exceeds the $\$60{,}000$ limit, so the higher-limit expected loss is $\$21{,}600$, not an amount based on a larger capped payment.

Frequency and the Portfolio Loss Basis

When calculating expected losses per exposure unit rather than per claim, frequency can appear in both the numerator and denominator. It cancels only if the same frequency basis applies to both limits. The next example makes that cancellation explicit. The Poisson assumption is used to specify the hypothetical claim count; only its mean is needed for the expected payment calculation.

Worked Example: Checking Frequency Cancellation

For a hypothetical book, the claim count per exposure unit is $N\sim\operatorname{Poisson}(0.20)$. Claim severities are independent and identically distributed, independent of $N$, with $P(X=\$5{,}000)=0.40$, $P(X=\$15{,}000)=0.35$, and $P(X=\$40{,}000)=0.25$. Find the ILF for a $\$30{,}000$ limit relative to a $\$10{,}000$ base limit. All terms other than the per-claim limit are held fixed; there is no discounting or mortality contingency.

Method. First calculate the expected capped payment per claim at each limit. At $\$10{,}000$, the payments are $\$5{,}000$, $\$10{,}000$, and $\$10{,}000$:

L(\$10{,}000) =0.40(\$5{,}000)+0.35(\$10{,}000)+0.25(\$10{,}000) =\$2{,}000+\$3{,}500+\$2{,}500 =\$8{,}000.

At $\$30{,}000$, the payments are $\$5{,}000$, $\$15{,}000$, and $\$30{,}000$:

L(\$30{,}000) =0.40(\$5{,}000)+0.35(\$15{,}000)+0.25(\$30{,}000) =\$2{,}000+\$5{,}250+\$7{,}500 =\$14{,}750.

The Poisson mean is $E[N]=0.20$ claims per exposure unit. Expected capped payments per exposure unit are therefore $0.20(\$8{,}000)=\$1{,}600$ at the base limit and $0.20(\$14{,}750)=\$2{,}950$ at the higher limit. The ratio is:

\operatorname{ILF} =\frac{\$2{,}950}{\$1{,}600} =\frac{\$14{,}750}{\$8{,}000} =1.84375 \approx 1.844.

The two portfolio amounts are consistent with the per-claim calculation because the same $0.20$ frequency multiplies both. Using different frequency assumptions at the two limits would prevent this cancellation and would confound the limit effect with a frequency change.

Conditions, Checks, and Common Traps

For a higher limit $u\ge b$, each claim’s payment at $u$ is at least its payment at $b$, since $\min(X,u)\ge\min(X,b)$. Taking expectations preserves that ordering, so $L(u)\ge L(b)$. When $L(b)>0$, the resulting ILF is at least 1. This is a useful reasonableness check, not a substitute for calculating both expected losses correctly.

The comparison also requires a well-defined, shared loss basis. In practice, confirm that the two limited expected losses use the same severity distribution, claim population, exposure definition, and valuation assumptions. If adjustments are applied, they must be consistent with the intended comparison. The ratio does not automatically account for premium expenses, profit provisions, or other loading components; those require separate treatment.

Applying only one cap. Calculate $\min(X,b)$ for every outcome in the base calculation and $\min(X,u)$ for every outcome in the higher-limit calculation.
Changing the loss basis unnoticed. A numerator based on one severity distribution and a denominator based on another is not the ILF defined here.
Putting frequency in only one side. If converting to an exposure basis, use the same frequency basis in both calculations. Otherwise it does not cancel.
Dividing in the wrong direction. For a higher limit relative to a base limit, divide higher-limit limited expected loss by base-limit limited expected loss.
Using the ratio as a premium by itself. The ILF is a relative factor. Applying it to a base-limit amount requires that amount to use the same loss and rating basis.

Key takeaway: Derive an ILF by calculating the limited expected loss at each limit on one shared basis, then dividing the higher-limit amount by the base-limit amount. The ratio isolates the limit comparison only when the other assumptions are held consistent.

Check Your Understanding

For each hypothetical problem, use the stated severity distribution and keep all terms other than the indicated limit comparison fixed. Amounts are at a common valuation date, with no discounting or mortality contingency.

A severity model assigns probabilities 0.50, 0.30, and 0.20 to $\$10{,}000$, $\$25{,}000$, and $\$60{,}000$, respectively. Calculate the ILF for a $\$40{,}000$ limit relative to a $\$15{,}000$ base limit. (Answer: $20{,}500/12{,}500=1.640$)
Claim count per exposure unit is Poisson with mean 0.10. Independent severities are $\$4{,}000$, $\$12{,}000$, and $\$30{,}000$, with probabilities 0.50, 0.30, and 0.20. Find the ILF for a $\$20{,}000$ limit relative to a $\$10{,}000$ base limit, showing the expected payments per exposure unit. (Answer: $\$960/\$700=1.371$)
A hypothetical severity distribution has pre-trend losses of $\$8{,}000$ with probability 0.25 and $\$24{,}000$ with probability 0.75. Trend all severities by 25%. Find the ILF for a $\$40{,}000$ limit relative to a $\$20{,}000$ base limit, applying trend before capping. (Answer: $\$25{,}000/\$17{,}500=1.429$)
For a severity distribution with probabilities 0.20, 0.50, and 0.30 at $\$5{,}000$, $\$15{,}000$, and $\$45{,}000$, calculate the ILF for a $\$25{,}000$ limit relative to a $\$10{,}000$ base limit. (Answer: $\$16{,}000/\$9{,}000=1.778$)
A hypothetical severity model assigns probabilities 0.30, 0.40, and 0.30 to $\$0$, $\$8{,}000$, and $\$50{,}000$. Calculate the ILF for a $\$40{,}000$ limit relative to a $\$10{,}000$ base limit, and verify the base limited expected loss is positive. (Answer: $\$15{,}200/\$6{,}200=2.452$)

What You'll Learn