Why Aggregated Data Require Care
The extension method re-rates historical exposures at a selected target rate level. With policy-level records, the calculation can reflect each policy’s applicable rating characteristics. With aggregated data, the same calculation may still be possible, but only if the groups preserve the information needed to apply the correct rate-level adjustment. A total premium alone does not necessarily reveal how that premium is distributed across rating groups.
The key question is not simply whether data are grouped. It is whether the premium in each group can be matched to a single appropriate factor, or whether the group’s premium-weighted average factor can be determined from available summaries. If different rating cells have different adjustments and their premiums have been combined without retaining the mix, the exact extension result generally cannot be recovered.
The group adjustment \(r_g\) may be a simple rate-level ratio if every risk in the group receives the same proportional change. If the adjustment varies within the group, \(r_g\) must instead be the group’s premium-weighted average of those individual or finer-cell adjustments. That average cannot be inferred from the group total premium alone unless the relevant subgroup premium shares are also known.
Conditions for Using Grouped Premium
A grouped calculation is exact when each group has a known historical premium and the group’s appropriate target-to-historical rate ratio is known. The groups may be territories, classifications, coverage forms, or combinations of rating characteristics. They need not match any particular administrative structure; they must separate risks that have different applicable adjustments, or preserve enough subtotals to calculate the adjustments correctly.
The following conditions clarify what the calculation does and does not establish:
- The premium amounts must refer to the experience being on-leveled and must be measured on a consistent basis.
- Each adjustment factor must compare the target rate level with the rate level actually applicable to that group’s premium.
- The group structure must account for relevant differences in rate changes. If policies within a group have different factors, their premium weights or a sufficient summary of those weights must be available.
- The calculation changes the rate level applied to the historical premium. It does not, by itself, adjust for changes in exposure mix or any other change in the risks represented.
When detailed policy records are unavailable, the practical objective is to retain useful subtotals. For example, premiums by territory and rate-change vintage can be sufficient even if individual policy dates are not available, provided every cell has a common factor or an accurately calculated cell-level average. More detailed grouping is not automatically better: it is useful when it preserves distinctions that affect the factor and when the corresponding premium amounts are reliable.
Deriving the Aggregate Factor
The formula follows directly from the definition of a premium-weighted average. For group \(g\), applying the target-to-historical factor \(r_g\) to historical premium \(P_g\) produces target-level premium \(P_g r_g\). Adding these amounts gives the total target-level premium. Dividing by the original total premium gives the multiplicative change from the original total to the target-level total:
Equivalently, if \(w_g=P_g/\sum_h P_h\), then the weights sum to one and
This shows why group premiums, not unadjusted exposure counts, supply the natural weights: the factor describes the proportional change in total premium, and each group’s contribution to that total is its share of the original premium. Exposure counts can be used only when they correctly represent premium shares, or when the premium basis makes that equivalence valid.
Worked Example: Extending Three Premium Groups
A hypothetical insurer has historical premium of $600,000 in one territory, $300,000 in another, and $100,000 in a third. The target-to-historical rate factors are respectively 1.08, 0.96, and 1.12. Calculate the extended premium and aggregate on-level factor. Method: extend each group’s premium, then divide the extended total by the historical total.
For the first territory, extended premium is \( \$600{,}000 \times 1.08=\$648{,}000 \). For the second, it is \( \$300{,}000 \times 0.96=\$288{,}000 \). For the third, it is \( \$100{,}000 \times 1.12=\$112{,}000 \).
The historical premium total is \( \$600{,}000+\$300{,}000+\$100{,}000=\$1{,}000{,}000 \). The extended premium total is \( \$648{,}000+\$288{,}000+\$112{,}000=\$1{,}048{,}000 \). Therefore, \( F=\$1{,}048{,}000/\$1{,}000{,}000=1.048 \). The aggregate premium is 4.8% higher at the target rate level.
Check. The premium weights are 0.60, 0.30, and 0.10. Their weighted factor is \(0.60(1.08)+0.30(0.96)+0.10(1.12)=0.648+0.288+0.112=1.048\), which agrees with the ratio of totals. The common mistake avoided here is taking an unweighted average of the three group factors; the groups do not contribute equally to premium.
What Aggregation Can Hide
Suppose a reported group total combines rating cells that receive different target adjustments. The total premium does not state how much belongs to each cell. Because each cell’s premium is multiplied by a different factor, that missing distribution affects the target-level total. Unless the subgroup premium amounts or their premium-weighted average factor are known, the result is not uniquely determined.
Exposure counts do not generally repair this information loss. Two cells may have the same number of exposures but different average premiums, so their shares of premium differ from their shares of exposure. Weighting rate factors by exposure count can therefore produce a plausible-looking but incorrect aggregate factor.
Worked Example: Recovering the Factor Within a Combined Group
A hypothetical book has one combined historical premium total of $500,000. The group consists of two classes: Class A has 800 exposures and $300,000 of historical premium; Class B has 200 exposures and $200,000 of historical premium. Their target-to-historical factors are 1.10 and 1.30. Calculate the correct aggregate factor and compare it with an exposure-weighted calculation. Method: retain the class premium subtotals and weight by premium.
Class A’s extended premium is \( \$300{,}000 \times 1.10=\$330{,}000 \). Class B’s extended premium is \( \$200{,}000 \times 1.30=\$260{,}000 \). The extended total is \( \$330{,}000+\$260{,}000=\$590{,}000 \), so the correct factor is \( \$590{,}000/\$500{,}000=1.18 \).
The premium shares are \( \$300{,}000/\$500{,}000=0.60 \) for A and \( \$200{,}000/\$500{,}000=0.40 \) for B. Thus the weighted factor is \(0.60(1.10)+0.40(1.30)=0.66+0.52=1.18\).
The exposure shares, by contrast, are 0.80 and 0.20. Incorrectly using them gives \(0.80(1.10)+0.20(1.30)=0.88+0.26=1.14\). This would imply extended premium of \( \$500{,}000(1.14)=\$570{,}000 \), which is $20,000 below the correct $590,000. Check. The direct class-by-class extension and the premium-weighted calculation both give 1.18. The wrong exposure-weighted calculation fails because the classes’ shares of exposure are not their shares of premium.
If only the $500,000 combined premium were known, with no class subtotals or equivalent mix information, the factor could not be determined exactly. For example, if all of that premium belonged to A, the extended premium would be $550,000; if all belonged to B, it would be $650,000. The same reported total is consistent with different extended totals. Additional detail is necessary to select the correct result.
Worked Example: Grouping Premium by Applicable Rate Level
A hypothetical portfolio summary reports three premium subtotals, each associated with its own applicable historical rate level. The amounts and target-to-historical factors are: $250,000 at 1.20, $400,000 at 1.05, and $350,000 at 0.98. Find the extended premium and aggregate factor. Method: treat each rate-level subtotal as a group and extend each one separately.
The first subtotal extends to \( \$250{,}000 \times 1.20=\$300{,}000 \). The second extends to \( \$400{,}000 \times 1.05=\$420{,}000 \). The third extends to \( \$350{,}000 \times 0.98=\$343{,}000 \).
Historical premium totals \( \$250{,}000+\$400{,}000+\$350{,}000=\$1{,}000{,}000 \). Extended premium totals \( \$300{,}000+\$420{,}000+\$343{,}000=\$1{,}063{,}000 \). Therefore, the aggregate factor is \( \$1{,}063{,}000/\$1{,}000{,}000=1.063 \).
As a check, the premium-weighted factors are 0.25, 0.40, and 0.35, giving \(0.25(1.20)+0.40(1.05)+0.35(0.98)=0.300+0.420+0.343=1.063\). The common mistake avoided is averaging factors without regard to the very different premium amounts associated with the rate levels.
Practical Checks and Exam Traps
Before accepting a factor from grouped data, verify that the group definition aligns with the adjustment being applied. If the factors are constant within groups, the calculation is direct. If a group contains multiple factors, check whether the summarized data permit a premium-weighted average. If not, report that the factor is not identifiable from the available aggregate; do not substitute exposure weights without justification.
- Wrong: unweighted factor average. The arithmetic mean of group factors is appropriate only when group premiums are equal. Otherwise, use premium shares.
- Wrong: exposure weighting by default. Exposure shares and premium shares differ whenever average premium per exposure differs across groups.
- Wrong: treating a combined total as sufficient. A total premium cannot reveal the mix across cells with different factors.
- Check factor direction. Each group factor must be target-level premium divided by premium at the group’s applicable historical level. A factor below 1.00 reduces that group’s extended premium; a factor above 1.00 increases it.
Aggregated extension is useful when the required summaries are available but policy-level records are not. Its reliability depends on whether the aggregation preserves the premium weights and rate distinctions that drive the calculation. This material is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
Check Your Understanding
For each problem, determine whether the summarized information supports an exact extension factor, and calculate it when possible.
- A hypothetical book has $200,000 of premium at factor 1.05 and $300,000 at factor 0.95. Find the extended premium and aggregate factor. (Answer: Extended premium = $495,000; factor = 0.9900.)
- Two rating cells have historical premiums of $120,000 and $80,000, with factors 1.10 and 1.25. Find the premium-weighted factor. (Answer: 1.1600.)
- Class A has 700 exposures with average historical premium of $400 per exposure; Class B has 300 exposures with average historical premium of $800 per exposure. Their extension factors are 1.05 and 1.20. Calculate the correct factor. (Answer: Historical premiums are $280,000 and $240,000; extended total is $582,000; factor = 1.1192.)
- A single combined historical premium of $400,000 contains two classes with factors 1.10 and 1.20, but no class premium subtotals or mix information are available. Can the exact aggregate factor be found? (Answer: No. The class premium shares are unknown; the factor could range from 1.10 to 1.20.)
- A summary has $150,000 at factor 1.04, $250,000 at factor 0.96, and $100,000 at factor 1.15. Calculate the extended premium and factor, and state whether the overall premium increases or decreases. (Answer: Extended premium = $511,000; factor = 1.0220; it increases by 2.2%.)