Interpolating ILFs Between Listed Limits

Why Interpolation Is Used

An increased limits factor (ILF) table may list factors only for selected policy limits, while a rating analysis calls for a limit between two of those entries. Interpolation provides an estimate at the requested limit without requiring a new severity calculation. The estimate depends on the interpolation method: a linear interpolation on the limit scale treats the factor, or the underlying limited expected loss, as changing linearly between the two listed limits.

Before interpolating, confirm that the table entries are on the same loss basis, use the same coverage terms, and are normalized to the same base limit. If the question specifies an interpolation method, use it. Unless another method is stated, this tutorial uses linear interpolation with respect to the dollar limit, not with respect to the logarithm of the limit.

Definition: Let $M(u)$ be the expected limited loss at policy limit $u$, calculated using the loss basis and coverage terms underlying the table. For a common base limit $b$, the increased limits factor at limit $u$ is $I(u)=M(u)/M(b)$, provided $M(b)>0$. For a ground-up loss $X$ with no other coverage modification, $M(u)=E[\min(X,u)]$, assuming this expectation is finite.

Linear Interpolation on the Limit Scale

Suppose the requested limit $u$ lies between listed limits $u_1$ and $u_2$, where $u_1<u<u_2$. The interpolation weight is the fraction of the distance from $u_1$ to $u_2$ that has been traversed in dollars. The interpolated factor is the same weighted fraction of the change between the two listed factors.

q=\frac{u-u_1}{u_2-u_1}, \qquad \widehat{I}(u)=(1-q)I(u_1)+qI(u_2) =I(u_1)+q\bigl(I(u_2)-I(u_1)\bigr).

Here $q$ is between 0 and 1 because $u$ is bracketed by $u_1$ and $u_2$. The estimate is a weighted average of the two factors, with more weight on the endpoint closer to the requested limit. The calculation is an interpolation only when the requested limit is inside the listed interval. Applying the same formula outside that interval is extrapolation and is not justified by the two endpoints alone.

Worked Example: Interpolate a Factor Between Adjacent Limits

A hypothetical commercial property program has a table with a $100,000 limit ILF of 1.000 and a $200,000 limit ILF of 1.400. Find the ILF at $150,000 using linear interpolation on the limit scale. Assume the factors use the same coverage and loss basis and the same base limit. No time value of money is applied (interest rate 0%); mortality is not applicable; no parametric severity distribution is assumed because the table factors are supplied.

Step 1: Find the interpolation weight. The distance from $100,000 to $150,000 is $50,000, and the full interval is $200,000 − $100,000 = $100,000. Thus $q=50{,}000/100{,}000=0.5$.

Step 2: Interpolate the factor. The factor increases by $1.400-1.000=0.400$ across the interval. At halfway, the estimated increase is $0.5(0.400)=0.200$, so the interpolated ILF is $1.000+0.200=1.200$.

Check and method. Equivalently, $0.5(1.000)+0.5(1.400)=1.200$, which lies halfway between the endpoint factors. The method is linear interpolation of ILFs. The common mistake is to use a weight based on the target limit divided by the upper limit; the weight must instead measure the target’s position within the bracket.

Why Factor and Limited-Loss Interpolation Can Agree

A factor is a limited expected loss divided by the limited expected loss at the base limit. If both endpoint factors use the same base and are not rounded, linear interpolation of the factors gives exactly the same answer as interpolating the endpoint limited expected losses and then dividing by that base limited expected loss.

\begin{aligned} (1-q)I(u_1)+qI(u_2) &=(1-q)\frac{M(u_1)}{M(b)} +q\frac{M(u_2)}{M(b)}\\ &=\frac{(1-q)M(u_1)+qM(u_2)}{M(b)}. \end{aligned}

The last expression is the linear interpolation of the two limited expected losses, normalized by the same base value. This identity is the justification for comparing the methods: they differ in presentation, not in result, when they use the same endpoints, the same dollar-limit weight, and a common exact normalization. It does not establish that the true expected loss varies linearly between the endpoints. Linear interpolation is an estimate imposed between the listed values.

Key condition: Interpolating factors and interpolating limited expected losses produce the same result if the endpoint factors are $M(u_1)/M(b)$ and $M(u_2)/M(b)$, the same $M(b)$ is used throughout, and both calculations use the same interpolation weight. Rounding, a changed base, or different loss bases can break the numerical equality.

Worked Example: Interpolate Limited Expected Losses, Then Normalize

A hypothetical liability table uses a $100,000 base limit. The limited expected losses are $50,000 at the base, $81,000 at $250,000, and $93,000 at $400,000. Find the ILF at $310,000 by interpolating limited expected losses and normalizing. Assume all amounts use identical coverage terms and loss definitions. No time value of money is applied (interest rate 0%); mortality is not applicable; the supplied limited expected losses require no assumed parametric severity distribution.

Step 1: Calculate the position in the interval. The interval is $400,000 − $250,000 = $150,000. The target is $310,000 − $250,000 = $60,000 above the lower endpoint. Therefore $q=60{,}000/150{,}000=0.4$.

Step 2: Interpolate the limited expected loss. The endpoint increase is $93,000 − $81,000 = $12,000. The estimated increase at the target is $0.4($12{,}000)=$4{,}800$. Thus the interpolated limited expected loss is $81,000 + $4,800 = $85,800.

Step 3: Normalize to the base limit. The base limited expected loss is $50,000, so the ILF is $$85{,}800/$50{,}000=1.716$.

Check and method. The endpoint factors are $81{,}000/50{,}000=1.620$ and $93{,}000/50{,}000=1.860$. Interpolating them gives $1.620+0.4(1.860-1.620)=1.716$, the same result. The method avoids treating a limited expected loss as an ILF before normalization; the loss amount must be divided by the base limited expected loss.

Precision, Rounding, and the Stated Method

The equivalence above is exact algebra, but table entries are often displayed at limited precision. Interpolating displayed factors may therefore produce a slightly different result from interpolating more precise limited expected losses. If the problem explicitly instructs you to use the listed factors, use those factors and report the result at the requested precision. Do not silently replace the stated method with a calculation from other data.

When both methods are permitted, use full available precision in intermediate calculations. Round only the final answer. If the problem provides both rounded factors and unrounded expected losses, identify which data and method you are using. A small discrepancy due to rounding is not evidence that the interpolation identity is wrong.

Worked Example: Diagnose a Rounding Difference

A hypothetical fleet table has a base limited expected loss of $20,000. The unrounded limited expected losses are $26,990 at $150,000 and $32,250 at $250,000. The corresponding displayed ILFs are 1.350 and 1.613, rounded to three decimals. Estimate the ILF at $175,000 by each method and explain any difference. Assume the same loss basis and base limit throughout. No time value of money is applied (interest rate 0%); mortality is not applicable; the supplied values are not based on an additional parametric distribution assumption.

Step 1: Find the weight. The target is $25,000 above $150,000 within a $100,000 interval, so $q=25{,}000/100{,}000=0.25$.

Step 2: Interpolate the displayed factors. Using the stated three-decimal table entries, $1.350+0.25(1.613-1.350)=1.350+0.06575=1.41575$. Rounded to three decimals, this gives 1.416.

Step 3: Interpolate the unrounded limited expected losses. The increase between the losses is $32,250 − $26,990 = $5,260. The interpolated increase is $0.25($5{,}260)=$1{,}315$, giving $26,990 + $1,315 = $28,305.

Step 4: Normalize and compare. Dividing by the $20,000 base gives $28{,}305/20{,}000=1.41525$, which rounds to 1.415 to three decimals. The displayed-factor result and unrounded-loss result differ by 0.0005 before rounding because the factors were rounded. This is not a failure of the identity: the two calculations did not use the same-precision endpoint values. If instructed to use the displayed table factors, report 1.416.

A Reliable Interpolation Process

A disciplined sequence helps prevent errors in the interval, the weights, and the normalization. In particular, do not confuse an ILF with a limited expected loss, and do not change bases between endpoints.

Confirm the basis.
Check that both listed values refer to the same coverage terms, loss definition, and base limit. Determine whether the entries are factors or limited expected losses.

Bracket the requested limit.
Identify $u_1$ and $u_2$ such that $u_1 \leq u \leq u_2$. If the requested limit is outside the bracket, the calculation would be extrapolation rather than interpolation.

Compute the dollar-scale weight.
Use $q=(u-u_1)/(u_2-u_1)$. Keep the same weight when interpolating a factor or its underlying limited expected loss.

Interpolate and, if needed, normalize.
Interpolate factors directly, or interpolate limited expected losses and divide by the common base limited expected loss.

Check the result and precision.
For exact consistent inputs, the two methods agree. Retain available precision during the calculation and round only as directed at the end.

Common Mistake: Using the Wrong Weight

Suppose a requested limit is $175,000, bracketed by $150,000 and $250,000, with endpoint factors 1.350 and 1.610. The correct weight is $($175{,}000-$150{,}000)/($250{,}000-$150{,}000)=0.25$, so the interpolated factor is $1.350+0.25(1.610-1.350)=1.415$.

A common incorrect calculation uses $175{,}000/250{,}000=0.70$ as the weight. That gives $1.350+0.70(0.260)=1.532$, which is wrong: 0.70 measures the target relative to zero, not its position between the two tabulated limits. The interpolation weight must be calculated from the lower endpoint and the width of the bracket.

Key takeaway: For linear interpolation on the dollar-limit scale, calculate the target’s relative position within its two bracketing limits. Interpolating ILFs directly is equivalent to interpolating limited expected losses and then normalizing only when the same exact base and consistent endpoint values are used.

Check Your Understanding

For each question, use linear interpolation on the dollar-limit scale unless the wording says otherwise. Assume each set of values uses a consistent loss basis.

A table gives ILFs of 1.000 at $80,000 and 1.360 at $180,000. Find the factor at $130,000. (Answer: 1.180)
A table gives ILFs of 1.480 at $200,000 and 1.900 at $500,000. Find the factor at $350,000. (Answer: 1.690)
The base limited expected loss is $30,000. Limited expected losses are $39,600 at $120,000 and $46,800 at $220,000. Interpolate the ILF at $170,000 by interpolating losses first. (Answer: 1.440)
ILFs are 1.550 at $200,000 and 1.790 at $320,000. Find the ILF at $260,000, and state the interpolation weight. (Answer: weight 0.500; ILF 1.670)
The base limited expected loss is $40,000. Limited expected losses are $48,400 at $125,000 and $62,800 at $325,000. Find the ILF at $275,000 by interpolating losses and normalizing. (Answer: 1.480)

What You'll Learn