Adjusting Exposure Curves for Severity Trend

Why Severity Trend Changes Exposure Curves

An exposure curve describes how expected loss changes as the policy limit changes. Its shape depends on the distribution of ground-up claim severity. When claim severities trend upward, a fixed prospective limit captures a smaller share of expected severity than it did in the historical period. The trend adjustment therefore belongs on the severity distribution before expected losses are limited at prospective policy limits.

This is distinct from loss development. Development adjusts observed claims toward ultimate amounts for a selected valuation basis; severity trend then adjusts those amounts from the historical experience period to the prospective period. The trend factor should reflect the time between those periods and the selected severity trend assumption. Frequency trend is not part of this adjustment to a severity exposure curve.

Definition: Let $X$ be a nonnegative, ultimate ground-up claim severity on a historical cost level, with $0<E[X]<\infty$. Let $T>0$ be the cumulative severity trend factor from that historical cost level to the prospective period. Under a common multiplicative trend assumption, prospective severity is $X_T=TX$. For a prospective policy limit $u$, the limited expected loss is $m_T(u)=E[\min(X_T,u)]$, and the trend-adjusted exposure curve is $L_T(u)=m_T(u)/E[X_T]$.

A multiplicative trend factor is commonly built by compounding annual trend rates. If the selected rate for year $j$ is $t_j$, then the factor for the relevant period is $T=\prod_j(1+t_j)$. The number of compounding periods must match the time between the experience cost level and the prospective cost level. If claim severities need different factors because their dates differ, apply the appropriate factor to each claim or experience segment rather than imposing one factor on all observations.

For any random severity $Z\geq 0$, its limited expected loss at limit $u$ can also be written as $E[\min(Z,u)]=\int_0^u S_Z(x)\,dx$, where $S_Z(x)=P(Z>x)$ is its survival function. The exposure curve divides that limited expected loss by the corresponding unlimited expected severity. The normalization makes the curve a share of unlimited expected loss, not a dollar amount.

How a Multiplicative Trend Transforms the Curve

Under $X_T=TX$, the payment at a fixed prospective limit is not generally $T$ times the historical payment limited at that same dollar amount. Instead, the following pointwise identity determines the correct adjustment:

\min(TX,u)=T\min\left(X,\frac{u}{T}\right).

To verify it, consider the two possibilities. If $X\leq u/T$, then $TX\leq u$, so both sides equal $TX$. If $X>u/T$, then $TX>u$, so both sides equal $u$. Taking expectations gives the trend-adjusted limited expected loss; taking the ratio to unlimited expected loss gives the transformed exposure curve.

m_T(u)=T E\left[\min\left(X,\frac{u}{T}\right)\right], \qquad E[X_T]=T E[X], \qquad L_T(u)=L\left(\frac{u}{T}\right),

where $L(v)=E[\min(X,v)]/E[X]$ is the historical exposure curve. The formula means that at a fixed prospective limit $u$, the trend-adjusted curve uses the historical curve value at the smaller limit $u/T$. A common multiplicative trend shifts the curve horizontally in dollar-limit space; it does not change the underlying shape under the stated assumption.

An increased limits factor (ILF) comparing a higher limit $u$ with a base limit $b$ is the ratio of their limited expected losses, provided both are calculated on the same severity and coverage basis:

\operatorname{ILF}_T(u;b) =\frac{m_T(u)}{m_T(b)} =\frac{L(u/T)}{L(b/T)}.

The common trend factor cancels from the ratio of limited expected losses, but the ILF can still change: trend changes the historical curve points at which the ratio is evaluated. If both limits are increased by the same factor $T$, then $\min(TX,Tu)=T\min(X,u)$; in that different comparison, the exposure-curve shares and the ILF are unchanged. These results assume a common multiplicative severity trend, unchanged coverage, and positive ground-up severities. They do not automatically apply when trend changes the distribution’s shape or when different claims require different trend factors.

Worked Example: Trending an Exponential Severity to a Fixed Limit

A hypothetical portfolio has historical ground-up ultimate severity $X$ exponentially distributed with mean $100,000. The selected annual severity trend is 5% for two years, compounded. Calculate the prospective limited expected loss and exposure-curve value at a prospective limit of $100,000. Use the exponential model, whose survival function is $S_X(x)=e^{-x/\theta}$ for mean $\theta$, and round dollar amounts to the nearest cent.

Step 1: Calculate the cumulative trend factor. The two-year factor is $T=1.05^2=1.1025$. The prospective mean severity is therefore $1.1025($100,000)=$110,250$.

Step 2: Apply the prospective limit to the trended distribution. For an exponential severity with mean $\theta$, limited expected loss is $\theta(1-e^{-u/\theta})$. Here $u/\theta=$100,000/$110,250=400/441$, so

m_T(100{,}000) =110{,}250\left(1-e^{-400/441}\right) \approx \$65{,}739.68.

Step 3: Normalize to obtain the exposure curve. Divide by the prospective unlimited mean:

L_T(100{,}000) =\frac{65{,}739.6816\ldots}{110{,}250} =1-e^{-400/441} \approx 0.5963.

The check is that the limited expected loss, $65,739.68, is below the $100,000 limit and the exposure-curve value is between 0 and 1. Method and trap avoided: trend the severity distribution first, then limit it. Multiplying the historical limited expected loss at $100,000, $100{,}000(1-e^{-1})\approx \$63{,}212.06$, by 1.1025 gives an incorrect $69,691.29. That calculation trends payments already capped at the fixed limit, effectively overlooking that more losses reach the cap after severity trend.

Trend Can Change ILFs by Limit

The effect of trend on an ILF depends on the severity tail and on the limits being compared. In the following example, the severity distribution is unchanged except for a common multiplicative trend. The limits are held fixed in prospective dollars, so their corresponding points on the historical exposure curve move.

Worked Example: Comparing ILFs Before and After Severity Trend

A hypothetical liability portfolio has historical ground-up severity exponentially distributed with mean $100,000. Apply a cumulative severity trend factor of 1.20. Calculate the ILF for a $300,000 limit relative to a $100,000 limit, first without trend and then at the prospective cost level. Round limited expected losses to the nearest cent and ILFs to four decimal places.

Without trend. The historical mean is $100,000. At the $100,000 limit, the limited expected loss is $100{,}000(1-e^{-1})=\$63{,}212.06$. At $300,000, it is $100{,}000(1-e^{-3})=\$95{,}021.29$. Thus the historical ILF is $95{,}021.29/63{,}212.06\approx 1.5032$.

With trend. The prospective mean becomes $1.20($100,000)=$120,000$. The two limited expected losses are

m_T(100{,}000)=120{,}000(1-e^{-100/120}) \approx \$67{,}848.22,

m_T(300{,}000)=120{,}000(1-e^{-300/120}) \approx \$110{,}149.80.

The prospective ILF is $110{,}149.80/67{,}848.22\approx 1.6235$. Equivalently, the respective exposure-curve values are $1-e^{-5/6}\approx0.5654$ and $1-e^{-2.5}\approx0.9179$; their ratio is the same ILF. The result is consistent with the curve transformation $L_T(u)=L(u/1.20)$. Method and trap avoided: compare both prospective limited expected losses at the same prospective dollar limits. Do not assume that the trend factor cancels from the ILF just because it multiplies severity; it cancels from each loss ratio only after the limit is applied, and trend changes the relevant curve points.

Empirical Curves: Trend Claims Before Limiting

For an empirical curve with ground-up claim amounts $x_1,\ldots,x_n$ and equal weights, the prospective limited expected loss at limit $u$ is estimated as $n^{-1}\sum_{j=1}^n\min(Tx_j,u)$. If claims have unequal weights, use the same weights consistently in both the limited and unlimited averages. The ground-up values must be suitable for the intended loss basis; trending capped amounts as though they were uncensored ground-up severities can understate the tail.

Worked Example: Trending an Empirical Curve at Two Limits

A hypothetical portfolio has four ground-up ultimate claim amounts, in thousands of dollars: 40, 80, 160, and 240. Assume equal weights and a common trend factor of 1.25 to the prospective period. Calculate the prospective exposure-curve values at limits of $100,000 and $200,000 and the ILF for $200,000 relative to $100,000. Report dollar amounts in thousands and round the final curve values and ILF to four decimal places.

Step 1: Trend each ground-up claim. The prospective amounts are $40(1.25)=50$, $80(1.25)=100$, $160(1.25)=200$, and $240(1.25)=300$. Their mean is $(50+100+200+300)/4=162.5$.

Step 2: Limit each trended claim at $100,000. In thousands, the payments are 50, 100, 100, and 100. The limited mean is $(50+100+100+100)/4=87.5$, and the exposure-curve value is $87.5/162.5=0.5385$.

Step 3: Limit each trended claim at $200,000. The payments are 50, 100, 200, and 200. The limited mean is $(50+100+200+200)/4=137.5$, and the exposure-curve value is $137.5/162.5=0.8462$.

Step 4: Calculate the ILF. The ratio is $137.5/87.5=11/7\approx1.5714$. The check is that each limited mean is at most its limit, the higher-limit mean is no smaller than the lower-limit mean, and the ILF exceeds 1. Method and trap avoided: trend each ground-up amount and then apply the prospective limits. For comparison, limiting the historical claims first would give means of 80 and 120 at those same dollar limits and an ILF of 1.5000, which is not the prospective result.

Practical Checks and Common Errors

A reliable trend adjustment keeps the severity basis, dates, and limits aligned. First establish the relevant historical cost level for the ground-up ultimate claims. Then apply the selected severity trend to the prospective period, and only then evaluate limited expected losses at prospective limits. The same adjustment must be reflected in both the limited and unlimited amounts used to form an exposure curve.

Applying trend after limiting. In general, $T E[\min(X,u)]\ne E[\min(TX,u)]$ when the dollar limit $u$ is fixed. Use the identity $E[\min(TX,u)]=T E[\min(X,u/T)]$.
Using the wrong time interval. A two-year factor is not a one-year factor. Compound the selected annual rates for the actual interval between the historical cost level and the prospective period.
Trending only the numerator. An exposure curve divides limited expected loss by unlimited expected loss. If severity is trended, both quantities must be on the prospective severity basis.
Confusing fixed and trended limits. This tutorial holds policy limits fixed in prospective dollars. If both claim severity and the limits scale by the same factor, their normalized exposure-curve relationship is unchanged under common multiplicative trend.
Ignoring the data basis. A curve based on ultimate ground-up severity is not automatically comparable with one based on limited or otherwise differently adjusted observations. The severities and limits must refer to a consistent basis.

These calculations matter because the exposure curve translates a ground-up severity distribution into expected loss at different policy limits. A trend adjustment can change the share of expected severity captured at each fixed limit and therefore change the ILF, even when the only assumed distributional change is a common scale increase. This treatment of severity trend is aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus.

Check Your Understanding

For each problem, assume nonnegative ground-up severity, unchanged coverage, and a common multiplicative severity trend unless stated otherwise. Round dollar amounts to the nearest cent and probabilities or curve values to four decimals.

A hypothetical portfolio has historical exponential severity with mean $80,000. The cumulative trend factor is 1.10 and the prospective limit is $88,000. Find the prospective limited expected loss and exposure-curve value. (Answer: $55,626.61; 0.6321)
Historical severity is exponential with mean $150,000, and the cumulative trend factor is 1.20. At a prospective limit of $180,000, find the limited expected loss. (Answer: $113,781.70)
Three equally weighted hypothetical ground-up claims are $20,000, $60,000, and $120,000. Apply a trend factor of 1.50 and calculate the prospective limited expected loss and exposure-curve value at a $100,000 limit. (Answer: $73,333.33; 0.7333)
Historical severity is exponential with mean $200,000 and is trended by a factor of 1.10. Calculate the prospective ILF for a $330,000 limit relative to a $110,000 limit. (Answer: 1.9744)
Under a common multiplicative severity trend of 1.15, both ground-up losses and every limit are multiplied by 1.15. How does the exposure-curve value at a particular limit compare before and after this change? (Answer: It is unchanged when the limit is scaled by the same factor as severity.)

What You'll Learn