Layer Losses As Differences In Limited Loss

Why Layer Loss Is an Increment

An increased limits factor compares expected losses at different policy limits. A layer comparison asks a related question: how much additional expected loss is covered when the limit rises from one level to another? The connection is exact when the limits refer to the same loss basis. A layer’s expected loss is the difference between limited expected losses at its upper and lower endpoints.

This result is useful both as a calculation shortcut and as a consistency check. It also gives a direct interpretation of changes in ILF: the change from one limit to another measures the expected loss added by the intervening layer, scaled by the chosen base-limit expected loss.

Definition: Let $X$ be a nonnegative ground-up loss, and let $u\geq 0$ be a per-loss limit. The limited loss at limit $u$ is $\min(X,u)$, and its expected value is $L(u)=E[\min(X,u)]$. For an attachment $a\geq 0$ and upper endpoint $b>a$, the payment in the layer from $a$ to $b$ is $\min((X-a)_+,b-a)$, where $(z)_+=\max(z,0)$.

Here, $a$ and $b$ are endpoints measured from ground up. The layer attaches above $a$ and pays at most $b-a$ per loss. Thus the layer from $10,000 to $30,000 has a maximum payment of $20,000, not $30,000. The identity below assumes the same loss definition, valuation basis, and coverage terms are used for both limited expected losses and the layer.

Deriving the Layer Identity

For each possible loss amount $X=x$, the payment in the layer from $a$ to $b$ equals the difference between the loss limited at $b$ and the loss limited at $a$. To see this, consider the three ranges of $x$.

Loss at or below the attachment.
If $x\leq a$, both limited losses equal $x$. Their difference is zero, as is the layer payment.

Loss between the endpoints.
If $a<x<b$, the limited losses are $x$ and $a$. Their difference is $x-a$, exactly the amount paid by the layer.

Loss at or above the upper endpoint.
If $x\geq b$, the limited losses are $b$ and $a$. Their difference is $b-a$, the layer’s maximum payment.

The cases establish the equality for every nonnegative loss value, including the endpoints. Taking expectations preserves equality, provided the expected limited losses exist. Since each limited loss is bounded by its limit, these expectations are finite for finite $a$ and $b$.

\min((X-a)_+,b-a)=\min(X,b)-\min(X,a)

E[\text{layer payment from }a\text{ to }b]=L(b)-L(a)

The result does not require a continuous severity distribution. It holds for discrete, continuous, or mixed losses. A probability mass exactly at $a$ produces no layer payment; a mass at $b$ produces the full layer limit. Those endpoint cases agree with the limited-loss difference.

Worked Calculations

Worked Example: Verify a Layer Using Both Methods

A hypothetical insurer models one ground-up claim amount $X$, in thousands of dollars, with the following discrete distribution: $P(X=0)=0.20$, $P(X=5)=0.30$, $P(X=12)=0.30$, and $P(X=25)=0.20$. Find the expected loss in the layer from $10,000 to $20,000. Also express it as a difference in ILFs using a $5,000 base limit.

Assumptions and method. This is a per-claim severity calculation; no claim frequency is needed. The probabilities sum to $0.20+0.30+0.30+0.20=1.00$. There is no discounting (interest rate $i=0$) and no mortality assumption. First calculate $L(5)$, $L(10)$, and $L(20)$, then use their differences. All amounts in the calculation are thousands of dollars.

Limited expected loss at 5. The limited losses at $X=0,5,12,25$ are $0,5,5,5$, respectively. Therefore,

$$ L(5)=0.20(0)+0.30(5)+0.30(5)+0.20(5) =0+1.50+1.50+1.00=4.00. $$

Limited expected loss at 10. The limited losses are $0,5,10,10$, so

$$ L(10)=0.20(0)+0.30(5)+0.30(10)+0.20(10) =0+1.50+3.00+2.00=6.50. $$

Limited expected loss at 20. The limited losses are $0,5,12,20$, giving

$$ L(20)=0.20(0)+0.30(5)+0.30(12)+0.20(20) =0+1.50+3.60+4.00=9.10. $$

The layer expected loss is $L(20)-L(10)=9.10-6.50=2.60$ thousand dollars, or $2,600.00. As a direct check, the layer pays $0,0,2,10$ thousand dollars at the four possible loss amounts. Its expectation is $0.20(0)+0.30(0)+0.30(2)+0.20(10)=0+0+0.60+2.00=2.60$ thousand dollars.

For the ILF check, the base expected loss is $L(5)=4.00$. Thus $\operatorname{ILF}(10;5)=6.50/4.00=1.625$ and $\operatorname{ILF}(20;5)=9.10/4.00=2.275$. Their difference is $0.650$, and $4.00(0.650)=2.60$ thousand dollars. This method avoids confusing the layer limit of $10,000 with its upper endpoint of $20,000.

Worked Example: A Layer With Losses at Its Endpoints

For a hypothetical policy, a single claim has severity distribution $P(X=2)=0.25$, $P(X=6)=0.25$, $P(X=8)=0.20$, $P(X=10)=0.20$, and $P(X=18)=0.10$, with amounts in thousands of dollars. Calculate the expected loss in the layer from $6,000 to $10,000 and verify the result directly. The stated probabilities sum to $1.00$. There is no discounting ($i=0$) and mortality is not applicable.

Method. The layer attaches at $a=6$ and has upper endpoint $b=10$, so its maximum payment is $10-6=4$ thousand dollars. The limited losses at limit 6 are $2,6,6,6,6$, and those at limit 10 are $2,6,8,10,10$.

Calculate the two expectations, showing each probability-weighted amount:

$$ L(6)=0.25(2)+0.25(6)+0.20(6)+0.20(6)+0.10(6) =0.50+1.50+1.20+1.20+0.60=5.00. $$

$$ L(10)=0.25(2)+0.25(6)+0.20(8)+0.20(10)+0.10(10) =0.50+1.50+1.60+2.00+1.00=6.60. $$

The difference is $6.60-5.00=1.60$ thousand dollars, or $1,600.00. Directly, the layer payments at $X=2,6,8,10,18$ are $0,0,2,4,4$ thousand dollars. Their expectation is $0.25(0)+0.25(0)+0.20(2)+0.20(4)+0.10(4)=0+0+0.40+0.80+0.40=1.60$ thousand dollars.

The zero payment at $X=6$ and full payment at $X=10$ confirm the endpoint treatment. A common mistake is to treat the loss at the attachment as entering the layer; the definition makes the payment there zero.

Connecting Layer Loss to ILF Changes

Choose a base limit $u_0$ for which $L(u_0)>0$. On a consistent per-claim loss basis, define the increased limits factor at limit $u$ by dividing the limited expected loss at $u$ by the limited expected loss at the base limit. The same base must be used when comparing two ILFs.

Definition: For $L(u_0)>0$, $\operatorname{ILF}(u;u_0)=L(u)/L(u_0)$. Therefore, the expected loss in the layer from $a$ to $b$ is $L(u_0)\big[\operatorname{ILF}(b;u_0)-\operatorname{ILF}(a;u_0)\big]$, when both factors use the same loss basis and base limit.

This follows by substituting $L(b)=L(u_0)\operatorname{ILF}(b;u_0)$ and $L(a)=L(u_0)\operatorname{ILF}(a;u_0)$ into $L(b)-L(a)$, then factoring out $L(u_0)$. If the base expected loss is zero, the ILF ratio is undefined; this expression cannot be used.

Worked Example: Convert an ILF Change to Layer Expected Loss

A hypothetical rating analysis provides a per-claim limited expected loss of $375 at a $10,000 base limit. On the same ground-up loss basis, the ILF is 1.28 at $25,000 and 1.76 at $60,000. Find the expected loss in the $25,000–$60,000 layer. Treat the stated ILFs and base expected loss as supplied inputs; no parametric severity distribution is assumed or required. There is no discounting ($i=0$) and mortality is not applicable.

Method. First convert each ILF to limited expected loss. At $25,000, the limited expected loss is $375(1.28)=375+105=$$480. At $60,000, it is $375(1.76)=375+285=$$660.

Subtract the lower-limit expected loss from the upper-limit expected loss: $660-480=$$180. Equivalently, the ILF change is $1.76-1.28=0.48$, and the layer expected loss is $375(0.48)=$$180.00 per claim. The two calculations agree.

The subtraction works because both ILFs use the same $10,000 base and the same loss basis. Multiplying the upper-limit ILF alone by the base loss would give $660, the expected loss limited at $60,000, not the loss in the layer above $25,000.

Interpretation and Common Traps

For fixed $X$, limited loss cannot decrease as the limit increases, so $L(b)\geq L(a)$ whenever $b>a$. Consequently, an ILF built from a positive common base is nondecreasing with limit, and the difference between two such ILFs is nonnegative. These are useful reasonableness checks, not substitutes for confirming the underlying coverage terms.

The layer identity is a statement about expected covered loss under matching definitions. A change in deductible, exclusions, loss adjustment treatment, or valuation basis can change the payment function; then subtracting two limited expected losses is valid only if those limited losses represent the same coverage construction as the layer. The identity by itself also does not add expenses, profit provisions, or other pricing adjustments.

Common exam trap: For a layer from $a$ to $b$, do not subtract the layer limit $b-a$ from a limited expected loss at $b$. The lower endpoint is $a$, so the correct subtraction is $L(b)-L(a)$. In ILF form, subtract factors at $b$ and $a$ that share the same base, then multiply by that base expected loss.

For example, if $L(10)=6.50$ and $L(20)=9.10$, a mistaken calculation might treat the $10,000 layer width as the lower limit and calculate $L(20)-L(10)$ only if the layer actually begins at $10,000. For a layer from $5,000 to $20,000, the correct calculation is $L(20)-L(5)=9.10-4.00=5.10$, not $9.10-6.50=2.60$. Always identify both layer endpoints before selecting the limited expected losses.

Check Your Understanding

For each problem, identify the layer endpoints and keep all calculations on the stated loss basis.

A hypothetical severity distribution has $P(X=0)=0.10$, $P(X=4)=0.40$, $P(X=9)=0.30$, and $P(X=16)=0.20$, with losses in thousands of dollars. Find the expected payment in the layer from $5,000 to $12,000 by calculating the layer payment at each possible loss. (Answer: $2,600.00)
On a consistent per-claim basis, $L(8)=\$3.20$ and $L(20)=\$8.00$, both in thousands of dollars. What is the expected loss in the layer from $8,000 to $20,000? (Answer: $4,800.00)
A $5,000 base limit has limited expected loss of $4.50 per claim. ILFs on that base are 1.40 at $15,000 and 1.85 at $30,000. Calculate the expected loss in the $15,000–$30,000 layer. (Answer: $2.025 per claim)
A common base limit has limited expected loss of $5.00. The ILF is 1.40 at $10,000 and 2.20 at $25,000. Find the layer expected loss from $10,000 to $25,000, and state why the common base matters. (Answer: $4.00 per claim; both ILFs must use the same base and loss basis.)
A hypothetical distribution, in thousands of dollars, is $P(X=0)=0.25$, $P(X=3)=0.25$, $P(X=7)=0.20$, $P(X=12)=0.20$, and $P(X=30)=0.10$. Calculate the expected payment in the layer from $7,000 to $15,000. (Answer: $1,800.00)

What You'll Learn