Inflation Effects on Policy Limits

Why Policy Limits Must Be Compared on the Same Price Basis

A policy limit is a maximum amount payable for a covered loss. If claim severities rise with inflation while the limit remains a fixed dollar amount, a larger share of claims can exceed the limit. If the limit is increased at the same rate as claim severities, the share exceeding it may remain unchanged. The comparison depends on whether the losses and the limit are expressed in the same price basis.

Let $X$ be the loss amount on a baseline price basis, and suppose every loss is multiplied by a deterministic inflation factor $k>0$. The inflated loss is $kX$. A fixed limit stays at the same dollar amount as losses rise; an inflation-indexed limit is multiplied by the same factor $k$. We compare exceedance probabilities, meaning the probability that a loss is greater than the limit, and may also compare the expected amount by which a loss exceeds that limit.

Definition: If $X$ is a baseline loss and $k>0$ is a deterministic inflation factor, then the inflated loss is $kX$. For a current-dollar limit $u$, the exceedance probability is $P(kX>u)$. A fixed limit retains its dollar amount as $k$ changes; an inflation-indexed limit that begins at $u_0$ becomes $ku_0$.

The factor $k$ must refer to the same period and price basis for the loss and the indexed limit. For example, a 10% annual trend over two years gives $k=(1.10)^2=1.21$, not $1.20$. The calculations below assume all losses receive the same multiplicative trend. If claim trends differ by loss, the single-factor formulas do not describe the entire severity distribution.

Equivalent Thresholds and Exceedance Probabilities

For $k>0$, a current limit $u$ can be converted to an equivalent threshold on the baseline loss scale. The inflated loss exceeds the current limit exactly when the baseline loss exceeds that threshold:

P(kX>u)=P\left(X>\frac{u}{k}\right)=S_X\left(\frac{u}{k}\right).

This conversion makes the fixed-limit effect clear. If the baseline limit is $u_0$ and the fixed limit remains $u_0$, then after losses are inflated by $k$, the threshold on the baseline scale falls to $u_0/k$. By contrast, an indexed limit of $ku_0$ converts back to $u_0$.

\begin{aligned} \text{Fixed limit }u_0: \quad &P(kX>u_0)=S_X\left(\frac{u_0}{k}\right),\\ \text{Indexed limit }ku_0: \quad &P(kX>ku_0)=S_X(u_0). \end{aligned}

The second equality follows because $kX>ku_0$ is equivalent to $X>u_0$. Thus, under a common deterministic trend, indexing the limit preserves the original exceedance probability. If $k>1$, then $u_0/k<u_0$, so a fixed limit cannot have a lower exceedance probability than the baseline limit. It has a strictly higher probability when there is positive probability of a loss between $u_0/k$ and $u_0$.

For expected amounts above a limit, write $(z-u)_+=\max(z-u,0)$. The same scaling gives a useful relation:

(kX-u)_+=k\left(X-\frac{u}{k}\right)_+, \qquad E[(kX-u)_+]=kE\left[\left(X-\frac{u}{k}\right)_+\right].

This is a statement about the expected excess amount per loss, not the probability of exceeding the limit. The probability counts losses above the limit; the expected excess also reflects how far above the limit those losses are.

Worked Comparisons

Worked Example: Uniform Losses and a Fixed Dollar Limit

A hypothetical portfolio has baseline losses $X$ uniformly distributed from $0 to $100,000. Over two years, losses trend at 10% per year. The policy limit is initially $80,000. Compare the probability of exceeding a fixed $80,000 limit with the probability of exceeding a limit indexed at the same rate. Also calculate the expected amount above each current limit.

Method and assumptions. The trend is deterministic and applies multiplicatively to every loss. The two-year factor is $k=(1.10)^2=1.21$, so the inflated loss is $Y=1.21X$. The current indexed limit is $1.21($80,000)=$96,800$.

Baseline probability. Since $X$ is uniform on $[0,100,000]$, the probability of exceeding the initial $80,000 limit is $(100,000-80,000)/100,000=0.20$.

Fixed limit probability. For the fixed $80,000 limit, the equivalent baseline threshold is $$80,000/1.21=$66,115.70$. Therefore, $P(Y>$80,000)=(100,000-66,115.70)/100,000=0.338843$, or 0.3388 to four decimals.

Indexed limit probability. For the $96,800 indexed limit, the equivalent baseline threshold is $$96,800/1.21=$80,000$. The exceedance probability is therefore $0.20$, unchanged from baseline.

Expected amounts above the limits. The inflated loss is uniform from $0 to $121,000. For a uniform loss on $[0,M]$, with $0\leq u\leq M$, direct integration gives $E[(Y-u)_+]=(M-u)^2/(2M)$: only the interval from $u$ to $M$ contributes to the expectation.

\begin{aligned} E[(Y-80{,}000)_+] &=\frac{(121{,}000-80{,}000)^2}{2(121{,}000)} =\frac{1{,}681{,}000{,}000}{242{,}000} =\$6{,}946.28,\\ E[(Y-96{,}800)_+] &=\frac{(121{,}000-96{,}800)^2}{2(121{,}000)} =\frac{585{,}640{,}000}{242{,}000} =\$2{,}420.00. \end{aligned}

Amounts are rounded to the nearest cent only at the final step. The fixed limit has both the larger exceedance probability and the larger expected excess. As a check, the indexed limit converts exactly to the original $80,000 threshold, so its exceedance probability must match the baseline value of 0.20.

Common mistake avoided: Treating the fixed $80,000 limit as though it had risen to $96,800 would incorrectly give an exceedance probability of 0.20. That calculation applies to the indexed limit, not the fixed one.

Worked Example: Exponential Losses and Two Limit Designs

A hypothetical insurer models baseline claim severity $X$ as exponential with mean $50,000. Losses increase by 20%, so $k=1.20$. The initial limit is $60,000. Find the exceedance probability and expected amount above the limit under (a) a fixed $60,000 limit and (b) an indexed $72,000 limit.

Method and distribution. For an exponential loss with mean $\theta=$50,000$, $S_X(x)=e^{-x/\theta}$ for $x\geq0$. The inflated loss is $Y=1.20X$. For an exponential variable, $E[(X-a)_+]=\theta e^{-a/\theta}$, which follows by integrating $\int_a^\infty (x-a)f_X(x)\,dx$.

(a) Fixed limit. The equivalent baseline threshold is $$60,000/1.20=$50,000$. Hence $P(Y>$60,000)=e^{-50,000/50,000}=e^{-1}=0.3679$, to four decimals.

Using the positive-part scaling relation, the expected excess is $1.20(50,000)e^{-1}=$22,072.77$, rounded to the nearest cent.

(b) Indexed limit. The indexed limit is $1.20($60,000)=$72,000$, whose baseline threshold is $60,000. Thus $P(Y>$72,000)=e^{-60,000/50,000}=e^{-1.2}=0.3010$, to four decimals.

The expected excess is $1.20(50,000)e^{-1.2}=$18,071.65$, rounded to the nearest cent. The indexed result matches the original baseline exceedance probability $P(X>$60,000)=e^{-1.2}$, which checks the comparison.

Common mistake avoided: The fixed limit's threshold is $50,000 on the baseline scale, not $60,000. Using $60,000 for the fixed limit would mistakenly calculate the indexed-limit probability.

Worked Example: Discrete Losses and a Limit Boundary

A hypothetical coverage has baseline losses, in thousands of dollars, with $P(X=60)=0.40$, $P(X=85)=0.35$, and $P(X=110)=0.25$. Losses are inflated by 20%. The original limit is $90,000. Compare a fixed $90,000 limit with an indexed $108,000 limit, including exceedance probabilities and expected amounts above the limits.

Method and assumptions. The inflated losses are $72$, $102$, and $132$ thousand dollars, with the same respective probabilities. Exceedance means strictly greater than the limit. This convention matters if a loss has positive probability exactly at the limit.

Fixed limit. The inflated losses of 102 and 132 exceed $90,000, so the probability is $0.35+0.25=0.60$. The expected amount above the limit, in thousands, is $0.40(0)+0.35(102-90)+0.25(132-90)=4.20+10.50=14.70$, or $14,700.

Indexed limit. The $108,000 limit is exceeded only by the $132,000 loss, so the probability is $0.25$. The expected excess is $0.25(132-108)=6$ thousand dollars, or $6,000.

At baseline, only the loss of 110 thousand dollars exceeds the original 90-thousand-dollar limit, giving probability $0.25$. The indexed calculation agrees. The fixed-limit calculation includes the 85-thousand-dollar loss after it rises to $102,000.

Common mistake avoided: Applying the limit test to baseline losses alone would omit the 85-thousand-dollar loss that exceeds the fixed limit after inflation. The boundary convention also means a loss exactly equal to a limit is not counted under the stated strict-exceedance definition.

Practical Interpretation and Exam Traps

A limit comparison should answer two separate questions: what proportion of losses exceed the limit, and how large the amounts above the limit are on average. Indexing can preserve the first quantity under a common multiplicative trend, but the expected excess still changes in dollars because the losses and limit are both larger. A fixed limit can increase both quantities, but the size of the change depends on the severity distribution near and above the effective threshold.

Keep the price basis consistent. When using baseline losses, divide the current limit by $k$. When using inflated losses, compare them directly with the current-dollar limit.
Do not confuse fixed and indexed limits. A fixed limit $u_0$ gives threshold $u_0/k$ on the baseline scale; an indexed limit $ku_0$ gives threshold $u_0$.
Check the direction. For $k>1$, a fixed limit has a lower equivalent baseline threshold, so its exceedance probability cannot be lower than the original probability.
Respect atoms and boundaries. For a discrete or mixed distribution, $P(X>u)$ and $P(X\geq u)$ can differ. Use the inequality specified by the problem.
State the trend assumption. The equal-probability result for indexed limits relies on the same deterministic factor applying to losses and the limit. It need not hold if their indexing differs.

Key takeaway: Convert the limit to the loss distribution's price basis before calculating. A fixed limit $u_0$ after trend factor $k$ corresponds to baseline threshold $u_0/k$; an indexed limit $ku_0$ corresponds to $u_0$ and therefore preserves the baseline exceedance probability.

These comparisons are aligned with the SOA Exam FAM (Fundamentals of Actuarial Mathematics) syllabus. The central exam check is whether the limit and loss have been placed on the same price basis before evaluating a probability or expected excess.

Check Your Understanding

For each problem, state the threshold on the baseline loss scale before calculating. Monetary answers are in dollars unless stated otherwise.

A hypothetical loss is uniform from $0 to $200,000. It is inflated by 10%, and a fixed limit remains $100,000. Find the probability of exceeding the limit. (Answer: 0.5455)
A baseline loss is exponential with mean $40,000. Inflation is 25%, and the initial $50,000 limit is either fixed or indexed. Find both exceedance probabilities. (Answer: Fixed 0.3679; indexed 0.2865)
A loss in thousands of dollars takes values 30, 70, and 100 with probabilities 0.20, 0.50, and 0.30. It is inflated by 10%. The initial $75,000 limit is fixed in one design and indexed in another. Find the probability of exceedance and expected excess under each design. (Answer: Fixed probability 0.8000 and expected excess $11,500; indexed probability 0.3000 and expected excess $8,250)
A baseline loss is uniform from $0 to $150,000 and is inflated by 20%. Compare exceedance probabilities for a fixed $120,000 limit and an indexed limit based on the same initial $120,000 limit. (Answer: Fixed 0.3333; indexed 0.2000)
A baseline loss is exponential with mean $30,000 and an initial limit of $45,000. Losses increase by 10%. Find the indexed limit that preserves the baseline exceedance probability, then find the exceedance probability if the limit instead remains fixed at $45,000. (Answer: Indexed limit $49,500; fixed-limit probability 0.2557)

What You'll Learn