Why Historical Layer Losses Need Adjustment
A burning-cost indication uses historical losses to estimate the cost of a layer for a prospective period. Historical amounts are not automatically comparable with prospective losses: claims may still develop, the cost level may change over time, and the exposure basis may differ across periods. The adjustments must put losses and exposures on a consistent basis before calculating loss cost per exposure unit.
For a per-occurrence layer attaching at \(a\) with limit \(u\), the payment on a ground-up loss \(x\) is the part above the attachment, capped at the limit. When claim-level ground-up data are available, a sound sequence is to estimate each claim's ultimate amount, trend that amount to the prospective date, and then apply the prospective layer payment function. This preserves the effect of the attachment and limit.
This tutorial uses deterministic hypothetical claim records and supplied development and trend assumptions; no interest or mortality assumption, or probability distribution, is needed. In practice, those assumptions should be selected and documented for the data and coverage being priced. The earlier material on development and trend adjustments provides the general setting; here the key issue is applying those adjustments correctly to an excess layer.
Claim-Level Development, Trend, and Layer Application
Suppose claim \(j\) has reported ground-up amount \(x_j\), a development factor \(D_j\) from the claim's valuation age to ultimate, and an annual trend rate \(g\). If its loss date is \(t_j\) years before the prospective date, the indicated prospective claim amount is
For attachment \(a\) and limit \(u\), apply the layer payment function
Sum the resulting payments over the claims, then divide by the exposure units corresponding to those claims. A limit is the maximum amount paid by the layer per occurrence; for example, a limit of $200,000 excess of $100,000 pays at most $200,000 on one occurrence.
Worked Example: Developing and Trending Claims Before Applying the Layer
A hypothetical insurer is pricing a $200,000 per-occurrence layer excess of $100,000. Its three historical claims have the following reported amounts, development factors to ultimate, and times from loss date to the prospective date. Use an annual trend rate of 4%. The claims correspond to 80 exposure units. Find the prospective burning cost per exposure unit.
Method. Develop and trend each ground-up claim, apply the prospective layer to each adjusted claim, sum the payments, and divide by 80. The supplied factors apply to the individual ground-up claims.
| Claim | Reported amount | Development factor | Years to prospective date |
|---|---|---|---|
| 1 | $180,000 | 1.25 | 2 |
| 2 | $320,000 | 1.10 | 1 |
| 3 | $90,000 | 1.50 | 3 |
Claim 1. Its ultimate estimate is \(180{,}000(1.25)=\$225{,}000\). The two-year trend factor is \(1.04^2=1.0816\), so its prospective amount is \(225{,}000(1.0816)=\$243{,}360\). The layer payment is \(\min(243{,}360-100{,}000,200{,}000)=\$143{,}360\).
Claim 2. Its ultimate estimate is \(320{,}000(1.10)=\$352{,}000\). The one-year trend factor is \(1.04\), so its prospective amount is \(352{,}000(1.04)=\$366{,}080\). The amount above the attachment is $266,080, which exceeds the $200,000 limit; its layer payment is therefore $200,000.
Claim 3. Its ultimate estimate is \(90{,}000(1.50)=\$135{,}000\). The three-year trend factor is \(1.04^3=1.124864\), so its prospective amount is \(135{,}000(1.124864)=\$151{,}856.64\). Its layer payment is \(\min(151{,}856.64-100{,}000,200{,}000)=\$51{,}856.64\).
The total adjusted layer loss is \(143{,}360+200{,}000+51{,}856.64=\$395{,}216.64\). Dividing by 80 exposure units gives \(\$395{,}216.64/80=\$4{,}940.208\), or $4,940.21 per exposure unit, rounded to the nearest cent. As a check, the first and third claims are within the layer limit, while the second reaches the limit; the total therefore includes exactly one capped payment.
Common mistake avoided: Applying the limit before development or trend, or applying the trend factor only to the historical layer payment, can change whether a claim reaches the attachment or limit. The layer is applied here to each claim after it is adjusted.
Why the Order Matters
The layer payment function is not generally proportional to the ground-up claim amount. Therefore, multiplying a historical layer payment by a ground-up development or trend factor is not generally equivalent to developing and trending the claim first. The attachment creates a threshold and the limit creates a cap; both make the payment function nonlinear.
To see this, write the layer payment as \(h(x)=\min(\max(x-a,0),u)\). If the ground-up claim is scaled by a positive factor \(c\), the correctly adjusted payment is \(h(cx)\). Directly scaling the old payment instead gives \(c h(x)\). These expressions need not be equal: if \(x\) is below the attachment but \(cx\) exceeds it, then \(h(x)=0\) while \(h(cx)>0\). Likewise, once the adjusted claim exceeds the layer limit, its payment remains capped even if \(cx\) continues to grow.
This justification also explains an important exception: if only aggregate historical layer losses are available, claim-level ground-up amounts cannot be reconstructed from the aggregate alone. In that case, an actuary may use development and trend factors specifically estimated for the layer loss measure. Those factors must match the relevant layer structure and the intended adjustment. A ground-up factor should not be treated as a layer-specific factor without justification.
Using Layer-Specific Factors and Matching Exposure
If the data are already recorded as losses to a particular layer, a layer-specific development factor may be applied to those losses when it was developed for that layer and valuation basis. Similarly, a layer-specific trend factor may be used if it is intended to trend that layer loss measure to the prospective date. Then the adjusted losses can be combined and divided by the corresponding exposure units.
Exposure units should cover the same historical periods and risks as the losses being adjusted. If periods have different exposure counts, the combined indication is total adjusted losses divided by total exposure units—not an unweighted average of period-by-period loss costs. This is an exposure-weighted average of the period costs. The calculation assumes that the exposure measure is appropriate and consistent across those periods; adjustments for broader portfolio changes are outside the scope here.
Worked Example: Layer-Specific Factors Across Unequal Exposure Periods
A hypothetical treaty has a $100,000 layer limit excess of $250,000. Historical losses are already reported as payments to that layer. The table gives the observed layer loss, a development factor estimated for this layer, a layer-specific trend factor to the prospective date, and matching earned exposure units. Calculate the combined prospective burning cost.
| Period | Observed layer loss | Layer development factor | Layer trend factor | Exposure units |
|---|---|---|---|---|
| A | $72,000 | 1.10 | 1.06 | 24 |
| B | $51,000 | 1.20 | 1.03 | 17 |
Period A. Its adjusted loss is \(72{,}000(1.10)(1.06)=79{,}200(1.06)=\$83{,}952\).
Period B. Its adjusted loss is \(51{,}000(1.20)(1.03)=61{,}200(1.03)=\$63{,}036\).
The adjusted losses total \(83{,}952+63{,}036=\$146{,}988\), and the exposure units total \(24+17=41\). Thus the combined burning cost is \(146{,}988/41=\$3{,}585.07317\ldots\), or $3,585.07 per exposure unit, rounded to the nearest cent.
As a check, Period A's adjusted loss cost is \(83{,}952/24=\$3{,}498\) per unit, and Period B's is \(63{,}036/17=\$3{,}708\) per unit. The combined cost, $3,585.07, lies between these values and gives greater weight to Period A because it has more exposure units.
Common mistake avoided: The unweighted average of the two period costs is \((3{,}498+3{,}708)/2=\$3{,}603\), which is not the combined cost because the periods have unequal exposure. The correct calculation weights each period by its exposure, equivalently dividing total adjusted losses by total exposure.
When Claim-Level Data Are Available
The next example shows why a factor applied to the historical layer payment can fail even when the arithmetic is straightforward. The records below are individual reported ground-up amounts, so the prospective layer can be applied directly after claim development and trend.
Worked Example: A Claim Moves Across the Attachment After Adjustment
A hypothetical insurer is pricing a $250,000 layer excess of $150,000. One historical ground-up claim is reported at $140,000. Its development factor to ultimate is 1.20, and its loss date is two years before the prospective date. Use a 5% annual trend rate. Find the prospective layer payment and explain why multiplying the historical layer payment by the development and trend factors is wrong.
Method. First develop the ground-up amount: \(140{,}000(1.20)=\$168{,}000\). The two-year trend factor is \(1.05^2=1.1025\). The prospective ground-up amount is \(168{,}000(1.1025)=\$185{,}220\).
The prospective layer payment is \(\min(\max(185{,}220-150{,}000,0),250{,}000)=\$35{,}220\). This claim is above the attachment and below the layer's $250,000 maximum payment.
The historical reported amount of $140,000 is below the $150,000 attachment, so the historical layer payment is zero. Multiplying that zero by the development and trend factors would still give zero; it would miss the prospective payment of $35,220. The adjustment moved the claim across the attachment, which is precisely why the ground-up claim should be adjusted before the prospective layer is applied.
Common mistake avoided: A zero historical layer payment does not establish that the prospective layer payment is zero. The ground-up amount must be developed and trended when the claim-level information and prospective layer are available.
Practical Sequence and Checks
A disciplined calculation helps prevent mixing loss measures, periods, and exposure bases. The sequence below applies to claim-level records; if only layer aggregates are available, use factors appropriate to those layer losses instead.
Determine whether the data are ground-up claim amounts or payments to a specified layer. Confirm whether the development and trend factors apply to that measure.
For claim-level data, multiply each reported amount by its applicable development factor. Keep the valuation age and factor basis consistent.
Apply the stated annual trend over the time from each claim's loss date to the target date. Use the correct time unit and elapsed period.
Use the intended attachment and limit on the adjusted claim amount, then sum claim payments or adjusted layer losses.
Divide by exposure units that correspond to the loss periods. For multiple periods, divide total adjusted loss by total matching exposure.
Before accepting an indication, check that each claim's layer payment is between zero and the layer limit, inclusive. Check that the trend exponent represents elapsed years rather than the number of calendar labels, and that the development factors do not already include the trend being applied separately. Finally, verify that the loss and exposure totals cover matching experience periods.
Check Your Understanding
For each problem, use the stated factors and prospective layer structure. Amounts are hypothetical; round money per exposure unit to the nearest cent when needed.
- A claim is reported at $140,000, has a development factor of 1.20, and is one year from its loss date to the prospective date. Trend is 3% annually. What is its prospective payment to a $150,000 layer excess of $100,000? (Answer: $73,040)
- Historical payments to a specified layer total $48,000. A development factor of 1.25 and a layer-specific trend factor of 1.04 apply to those payments. There are 20 matching exposure units. Find the prospective burning cost per unit. (Answer: $3,120.00 per exposure unit)
- A claim is reported at $80,000, has a development factor of 1.50, and is two years from its loss date to the prospective date. Trend is 10% annually. Find its payment to a $120,000 layer excess of $80,000. (Answer: $65,200)
- For Period A, layer losses are $30,000 with development factor 1.10, trend factor 1.05, and 10 exposure units. For Period B, they are $42,000 with development factor 1.00, trend factor 1.02, and 20 exposure units. The factors are appropriate to the same layer. Find the combined prospective burning cost. (Answer: $2,583.00 per exposure unit)
- A ground-up claim is below a layer attachment at the historical date but above it after development and trend. Should its historical layer payment simply be multiplied by the development and trend factors? State the correct sequence. (Answer: No; develop and trend the ground-up claim first, then apply the prospective layer, because the adjusted claim may cross the attachment.)