Trending and On-Leveling Historical Experience

Why Historical Experience Must Be Restated

Historical treaty losses are not automatically a sound estimate of prospective cost. Loss amounts may reflect an earlier cost environment, a different volume or mix of insured exposures, and treaty terms that no longer apply. Premiums may also have been collected at a different rate level. Before using historical experience for a prospective indication, the actuary must make the comparison basis consistent.

This tutorial focuses on four adjustments: loss trend, exposure changes, coverage changes, and premium rate level. The adjustments address different causes of change. Applying them in the wrong order, or applying the same change twice, can materially distort the indicated cost.

Definition: Trending adjusts historical loss amounts from their historical cost level to the prospective cost level. On-leveling adjusts historical premium to the prospective rate level. Exposure adjustment restates the quantity or mix of insured business. Coverage adjustment restates losses to the treaty terms being priced.

Let $L_h$ denote historical treaty losses, $P_h$ historical earned premium, $R_h$ the historical rate-level index, and $R_p$ the prospective rate-level index. A rate-level index is a relative measure of price, with a larger index representing a higher rate level. The calculations below assume the index is appropriate for the business being analyzed.

Trend and Rate-Level Adjustments

If the annual loss trend is a constant rate $g$, and a loss amount is $t$ years before the prospective period, the compounded trend factor is $(1+g)^t$. Thus, a historical amount $L_h$ becomes $L_h(1+g)^t$. For varying annual trends, multiply the applicable annual factors rather than using one constant rate.

T(t)=\prod_{j=1}^{t}(1+g_j), \qquad L_h^{\mathrm{trended}}=L_hT(t)

The time interval must run from the period represented by the historical losses to the prospective period, using the same time convention as the selected trend. For example, a loss amount from three years earlier receives three annual factors, not two. Trend is not a substitute for adjusting the quantity of insured exposure.

Under the simplifying assumption that historical and prospective business have the same exposure mix, on-level premium is historical premium multiplied by the ratio of prospective to historical rate level:

P_h^{\mathrm{on}}=P_h\frac{R_p}{R_h}

An adjusted historical loss ratio is then trended losses divided by on-level premium. If the exposure mix is stable and the prospective premium is appropriate for the projected business, multiplying this ratio by prospective premium gives an indicated prospective loss amount. The derivation and assumptions are important: the loss ratio method uses premium as the exposure base, so a separate exposure multiplier is generally not added afterward.

Worked Example: Trend Losses and On-Level Premium

A hypothetical reinsurer reviews one year of historical experience. Treaty losses were $100,000, earned premium was $800,000, and the historical rate-level index was 0.80. The loss period is one year before the prospective period. Assume a 4% annual loss trend, a prospective rate-level index of 1.00, stable exposure mix, and prospective premium of $1,250,000. Calculate the adjusted loss ratio and indicated prospective losses.

Method. Trend the losses, on-level the premium, calculate the adjusted loss ratio, and apply it to prospective premium. The stable-mix assumption permits the loss-ratio method.

Step 1: Trend historical losses. The period is one year earlier, so the trend factor is $1.04^1=1.04$. Therefore:

L_h^{\mathrm{trended}}=\$100{,}000(1.04)=\$104{,}000

Step 2: On-level historical premium.

P_h^{\mathrm{on}}=\$800{,}000\left(\frac{1.00}{0.80}\right) =\$800{,}000(1.25)=\$1{,}000{,}000

Step 3: Calculate the adjusted loss ratio.

\frac{\$104{,}000}{\$1{,}000{,}000}=0.104=10.4000\%

Step 4: Apply the ratio to prospective premium.

0.104(\$1{,}250{,}000)=\$130{,}000

The indicated prospective treaty losses are $130,000. As a check, the adjusted ratio is 10.4% of the on-level historical premium, and the same ratio applied to prospective premium gives the stated amount. The common mistake is to trend losses but leave historical premium at its old rate level; that would divide $104,000 by $800,000 and incorrectly produce a 13% ratio.

Exposure and Coverage Changes

An exposure adjustment is needed when the historical experience does not represent the prospective quantity or mix of risk, and the chosen method does not already account for that difference. In an exposure-normalized method, divide adjusted historical losses by a historical exposure measure, then multiply by the corresponding prospective exposure measure. This is valid only when the exposure measure is suitable and the relationship between exposure and expected losses is adequately represented by the method.

\text{Prospective losses} = \frac{\text{adjusted historical losses}} {\text{historical exposure units}} \times \text{prospective exposure units}

For instance, policy-years may be an appropriate measure for a stable portfolio of similar policies. A simple policy count may be inadequate if insured values, limits, or risk mix have changed materially. In that case, the actuary should use a more suitable exposure base or explicitly model the mix change. A proportional exposure adjustment assumes, in effect, that expected losses change in proportion to the selected exposure measure; it is not a universal rule.

A coverage adjustment addresses a different issue: the historical losses may have been subject to different retentions, limits, exclusions, or other terms. When the underlying claim amounts are available, the preferred conceptual approach is to trend those amounts and calculate what each would have paid under the prospective treaty. For a per-loss layer with attachment $a$ and limit $u$, the payment on loss $X$ is:

Y^L=\min\{\max(X-a,0),u\}

This formula applies to a layer that pays above attachment $a$, up to a maximum payment $u$ per loss. It does not, by itself, account for annual aggregate limits, reinstatements, or other treaty features. Those terms must be handled when relevant. If only aggregate historical treaty recoveries are available, a simple coverage factor may be used only when it is supported by an appropriate analysis; it should not be assumed that changing the layer terms scales all recoveries by the same percentage.

Worked Example: Restate Losses to a New Layer and Exposure Level

A hypothetical portfolio had 1,000 exposure units in a historical year. Three ground-up losses from that year were $700,000, $1,100,000, and $1,600,000. The prospective treaty is $400,000 excess of $600,000, meaning an attachment of $600,000 and a maximum payment of $400,000 per loss. Assume the losses are two years before the prospective period, annual loss trend is 5%, and prospective exposure is 1,200 units. Assume frequency changes in proportion to exposure units, while severity and the mix of business remain otherwise unchanged. Estimate prospective treaty losses using these claims.

Method. Trend each ground-up loss, apply the prospective layer to each trended loss, sum the resulting payments for the historical exposure base, and then scale for exposure. This sequence avoids applying the prospective layer to historical-dollar amounts.

Step 1: Calculate the trend factor.

$$ T=1.05^2=1.1025 $$

Step 2: Trend each ground-up loss.

\$700{,}000(1.1025)=\$771{,}750 $$ $$ \$1{,}100{,}000(1.1025)=\$1{,}212{,}750 $$ $$ \$1{,}600{,}000(1.1025)=\$1{,}764{,}000

Step 3: Apply the prospective layer. For the first loss, the amount above the $600,000 attachment is $171,750, below the $400,000 limit. For the second, it is $612,750, so the payment is limited to $400,000. For the third, it is $1,164,000, also limited to $400,000. The three payments are therefore $171,750, $400,000, and $400,000.

Step 4: Sum and adjust for exposure. The historical-exposure total under prospective terms is $971,750. The exposure factor is $1{,}200/1{,}000=1.20$. Thus:

\$971{,}750(1.20)=\$1{,}166{,}100

The estimate is $1,166,100. As a check, it is 20% above the adjusted total for 1,000 units, consistent with the stated proportional exposure assumption. A common mistake is to multiply the historical recoveries by 1.20 without first restating the claim amounts to the prospective layer; that would preserve the wrong coverage basis.

Why the Loss-Ratio Formula Works

The loss-ratio method can be derived directly from proportional premium rates. Suppose expected losses for a stable mix of business are proportional to exposure quantity $E$, while premium is proportional to both $E$ and the rate level $R$. Write expected losses as $cE$ and premium as $kER$, where $c$ and $k$ are constants for this illustration. Then the loss ratio at a given rate level is:

\frac{cE}{kER}=\frac{c}{kR}

Exposure quantity cancels, but the rate level does not. This explains why a premium denominator must be on-leveled when historical and prospective rates differ. After trending historical losses to the prospective loss-cost level and adjusting premium to the prospective rate level, the adjusted ratio can be applied to prospective premium. This reasoning depends on comparable exposure mix and appropriate premium data; it does not justify ignoring material changes in business composition or treaty coverage.

Theorem: With a stable exposure mix, a suitable premium base, and no unmodeled coverage change, the loss-ratio method estimates prospective losses as the adjusted historical loss ratio multiplied by prospective premium. If material exposure mix or coverage changes are not reflected in that ratio, adjust them separately using a consistent method.

Choosing a Consistent Adjustment Path

There are two useful paths, and they should not be mixed casually:

Loss-ratio path: trend losses, on-level historical premium, calculate an adjusted loss ratio, and apply it to prospective premium. This path generally absorbs changes in exposure quantity through the premium base when mix is stable and prospective premium reflects the projected exposure.
Exposure-normalized path: trend and restate losses to prospective coverage, divide by historical exposure units, and multiply by prospective exposure units. This path estimates losses directly. Rate level is not a multiplier on expected losses; rate level affects premium. If a prospective loss ratio is also needed, divide the loss estimate by prospective premium.

A frequent exam trap is to on-level premium, apply an exposure growth factor to the loss ratio, and then multiply by a prospective premium that already reflects that same exposure growth. This can count the exposure change twice. Another is to apply a rate-level factor to losses merely because premium rates changed. Rate level changes the price charged; it does not, by itself, change the underlying loss amount.

Worked Example: Use the Loss-Ratio Path Without Double-Counting Exposure

A hypothetical treaty had historical losses of $180,000 and earned premium of $1,200,000. Those losses are two years before the prospective period. Assume a 6% annual loss trend, historical rate-level index 0.90, prospective rate-level index 1.08, stable exposure mix, and prospective premium of $2,000,000. The prospective premium already reflects the projected exposure quantity. Calculate indicated prospective losses.

Method. Use the loss-ratio path. Since the prospective premium includes projected exposure, do not multiply the resulting indication by a separate exposure-growth factor.

Step 1: Trend losses two years.

T=1.06^2=1.1236, \qquad L_h^{\mathrm{trended}}=\$180{,}000(1.1236)=\$202{,}248

Step 2: On-level historical premium.

P_h^{\mathrm{on}}=\$1{,}200{,}000\left(\frac{1.08}{0.90}\right) =\$1{,}200{,}000(1.20)=\$1{,}440{,}000

Step 3: Calculate the adjusted loss ratio.

\frac{\$202{,}248}{\$1{,}440{,}000}=0.14045=14.0450\%

Step 4: Apply the unrounded ratio to prospective premium.

0.14045(\$2{,}000{,}000)=\$280{,}900

The indicated losses are $280,900. The arithmetic checks because $0.14045 \times 2{,}000{,}000=280{,}900$ exactly. If a displayed ratio has been rounded, use the unrounded ratio for the final dollar calculation where available. The mistake to avoid is adding a separate exposure increase after applying the ratio to prospective premium when that premium already reflects the increase.

Practical Checks Before Using an Indication

Before relying on a result, confirm the following:

The trend period matches the actual distance from historical experience to the prospective period.
The trend applies to the loss measure being adjusted. For a layer change, trend ground-up losses and then calculate prospective treaty payments where claim data permit.
The historical and prospective exposure measures are comparable, and any proportional scaling assumption is reasonable.
The premium rate-level index relates to the same business and premium basis as the historical earned premium.
Each exposure or rate change is reflected once, not twice, and a rate change is not mistakenly treated as a loss trend.

These adjustments produce a more comparable historical indication; they do not remove uncertainty in the experience or guarantee that the future will match the past. The result is only as reliable as its trend selection, exposure basis, coverage information, and assumptions about portfolio stability. These techniques are aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.

Check Your Understanding

For each hypothetical problem, identify the adjustment path and check that no change is counted twice.

A treaty has $50,000 of losses three years before the prospective period. Assume a constant annual loss trend of 4%. What is the trended loss amount? (Answer: $56,243.20)
Historical premium is $600,000 at rate-level index 0.75. The prospective index is 0.90. Assuming the same exposure mix, what is on-level historical premium? (Answer: $720,000)
A ground-up loss of $950,000 is one year old. Trend is 5%. The prospective treaty is $300,000 excess of $500,000. What is the prospective payment for this loss after trending? (Answer: $300,000)
Adjusted treaty losses are $240,000 for 800 historical exposure units. Prospective exposure is 1,000 units. Under a proportional exposure assumption, estimate prospective losses using the exposure-normalized method. (Answer: $300,000)
Trended losses are $90,000, historical premium is $500,000 at rate index 0.80, and the prospective index is 1.00. The stable-mix adjusted loss ratio is to be applied to prospective premium of $750,000, which already reflects prospective exposure. What are the adjusted loss ratio and indicated losses? (Answer: 14.4000%; $108,000)

What You'll Learn