Why Territorial Experience Needs a Common Basis
A territory comparison is meaningful only if the experience being compared is measured on a consistent basis. One territory’s losses may be less mature because claims have had less time to develop. Another territory’s experience may come from an earlier period, when costs were lower. Comparing the raw amounts without addressing those differences can make territories appear more or less costly for reasons unrelated to their underlying risk.
Loss development and trend address different timing effects. Development adjusts losses from their current evaluation level to an estimate of ultimate losses. Trend adjusts losses from the historical period in which they occurred to a common prospective period. For territorial comparisons, both adjustments must be applied using selections and time periods that are appropriate and consistent across the territories.
Separate the Development and Trend Questions
A loss development factor is selected for losses evaluated at a particular age. For example, a factor of 1.10 applied to losses evaluated at 12 months estimates ultimate losses as 110% of the evaluated amount. The factor should match the loss definition, evaluation age, and development basis of the data. A factor selected for 12-month incurred losses is not automatically appropriate for 24-month paid losses.
Trend is applied over the time from the historical period being represented to the target period. In many ratemaking applications, an accident-year loss cost is associated with the accident period’s midpoint, and a prospective loss cost with the prospective period’s midpoint. The time \(t_j\) is the elapsed time between those two midpoints. It is not automatically the number of years between the evaluation date and the effective date of the proposed rates.
The displayed formula assumes a constant annual trend rate over the interval and a multiplicative trend model. If the rate varies by year, use the product of the annual factors covering the relevant periods instead of applying a single rate for the whole interval. Trend should reflect the relevant loss-cost changes; exposure counts are not multiplied by a loss trend factor. If the underlying measure is a loss cost per exposure, the exposure denominator is already part of that measure.
Worked Example: Adjusting Two Accident Years to One Prospective Period
A hypothetical insurer compares two territories using loss costs per exposure. Territory A has a 2023 accident-year loss cost of $120, evaluated at 12 months, and a selected 12-to-ultimate development factor of 1.10. Territory B has a 2024 accident-year loss cost of $100, evaluated at 12 months, and a selected factor of 1.20. Trend is 5% per year for both territories. The target is the 2025 accident-year midpoint. Treat each accident-year midpoint as July 1 of its year. Calculate each adjusted loss cost and Territory A’s ratio to Territory B.
Method. Develop each loss cost to ultimate, then trend from its accident-year midpoint to the target midpoint. The 2023 midpoint is two years before the 2025 midpoint, so \(t_A=2\). The 2024 midpoint is one year before the target, so \(t_B=1\).
For Territory A, development gives \(120(1.10)=132\). The two-year trend factor is \((1.05)^2=1.1025\), so \(132(1.1025)=145.53\). Territory A’s adjusted loss cost is $145.53 per exposure.
For Territory B, development gives \(100(1.20)=120\). The one-year trend factor is \(1.05\), so \(120(1.05)=126\). Territory B’s adjusted loss cost is $126.00 per exposure.
The adjusted ratio is \(145.53/126=1.155\). Thus, on these selected assumptions, Territory A’s adjusted loss cost is 1.155 times Territory B’s. The calculation is checked by dividing the two adjusted values: \(145.53 \div 126=1.155\).
Common mistake avoided: using two years of trend for both territories because the prospective date is in 2025. The correct periods are measured from each historical midpoint: two years for A, but one year for B.
Why a Common Adjustment Can Preserve a Ratio
When two loss costs receive exactly the same multiplicative adjustment, their ratio is unchanged. This is useful as a diagnostic: if two territories have the same evaluation age, the same selected development factor, the same historical period, and the same trend assumptions, applying those common factors changes the dollar levels but not the relative comparison.
A preserved ratio does not prove that the adjusted costs are correct. It shows only that a common factor cannot change their relative ordering or ratio. The selected factors and time periods still need to be appropriate for the experience.
Worked Example: Comparing Territories With Different Trend Periods
Two hypothetical territories have historical loss costs of $90 and $105 per exposure. The first is evaluated at 12 months and has a selected development factor of 1.15; its historical midpoint is two years before the common prospective midpoint. The second is evaluated at 24 months and has a selected development factor of 1.05; its historical midpoint is one year before the same target. Use a 4% annual trend rate for both. Calculate the adjusted costs and the first territory’s ratio to the second.
Method. Use the territory-specific development factors and the midpoint-to-midpoint trend periods given in the problem.
For the first territory, the developed cost is \(90(1.15)=103.50\). The two-year trend factor is \((1.04)^2=1.0816\). Therefore, \(103.50(1.0816)=111.9456\), or $111.95 per exposure rounded to cents.
For the second territory, the developed cost is \(105(1.05)=110.25\). Its one-year trend factor is \(1.04\). Therefore, \(110.25(1.04)=114.66\), or $114.66 per exposure.
Using the unrounded first adjusted cost, the ratio is \(111.9456/114.66 \approx 0.9763\). Rounded to four decimals, the first territory’s adjusted cost is 0.9763 times the second’s.
Check. The first adjusted amount is slightly smaller than the second, so a ratio just below 1 is consistent with the calculated dollar amounts.
Common mistake avoided: giving both territories two years of trend simply because both are being compared at one prospective date. Their historical midpoints differ, so the elapsed periods differ. The development factors also reflect different evaluation ages and should not be interchanged without a reasoned selection.
Inconsistent Selections Can Create False Differences
A key diagnostic is to ask whether apparent territorial differences come from the observed experience or from the adjustment choices. If territories have comparable loss definitions, evaluation ages, and periods, unexplained differences in selected factors can create an artificial relative difference. Conversely, imposing identical factors when evaluation ages or claim characteristics genuinely differ may leave a real timing difference unadjusted.
A development factor should be selected consistently with the maturity of the losses. If experience is evaluated at the same age on the same basis, a different factor by territory needs support, such as a credible and relevant difference in development patterns. Trend periods should likewise follow the actual historical and target periods, not be selected merely to obtain a desired comparison.
Worked Example: Measuring the Distortion From Unequal Development Selections
Suppose two hypothetical territories each have a historical loss cost of $100 per exposure. Both costs represent the same accident period, are evaluated at the same age, and will be trended two years at 5% to a common prospective midpoint. For this diagnostic, assume the appropriate development factor is 1.10 for both. An analyst instead applies 1.20 to Territory A and 1.05 to Territory B. Calculate the properly adjusted costs, the analyst’s adjusted costs, and the relative distortion.
Method. First establish the common-basis result, then compare it with the result from the inconsistent factors. The two-year trend factor is \((1.05)^2=1.1025\).
With the common factor, each territory has adjusted cost \(100(1.10)(1.1025)=110(1.1025)=121.275\), or $121.28 per exposure rounded to cents. The proper adjusted ratio is \(121.275/121.275=1.0000\).
Under the analyst’s unequal factors, Territory A becomes \(100(1.20)(1.1025)=120(1.1025)=132.30\). Territory B becomes \(100(1.05)(1.1025)=105(1.1025)=115.7625\), or $115.76 per exposure rounded to cents.
The analyst’s adjusted ratio, calculated before rounding, is \(132.30/115.7625=1.20/1.05=1.142857\ldots\), or 1.1429 to four decimals. The unequal development selections have created an apparent 14.29% difference even though the input costs and stated timing are identical. Since the trend factor is common, it cancels from the ratio; the distortion comes from the unequal development factors.
Common mistake avoided: interpreting a difference in adjusted costs as evidence of a territorial risk difference without checking the adjustment selections. If different factors are genuinely justified, explain why; otherwise, use comparable selections.
Practical Checks Before Comparing Results
Before using adjusted experience in a territorial comparison, confirm that each loss cost has been put on the intended basis. The following checks target errors that can materially change the result:
Confirm the loss type, evaluation age, and development basis. Do not apply an incurred factor to a paid measure, or a factor for one maturity to another, without a justified method.
State which period midpoint the historical loss cost represents and which prospective midpoint is the target. Compute elapsed time from those dates.
Use the appropriate rate and time interval. If rates vary by year, multiply the applicable annual factors rather than treating the entire interval as one constant-rate period.
Ask whether territory-specific development or trend selections reflect supportable differences or merely inconsistent judgment. Check the adjusted ratio as well as the adjusted dollar amounts.
Carry unrounded values through ratios and round only the reported result. Early rounding can create small, avoidable discrepancies.
One frequent exam trap is to use the proposed rate effective date as the trend endpoint without considering the period represented by the loss cost. The target is the prospective period specified for the analysis; the correct trend duration depends on the historical period midpoint and that target. Another trap is to apply trend to the exposure denominator as well as the loss cost. When the quantity being adjusted is already a loss cost per exposure, the multiplicative loss trend applies to that cost, not to the count of exposures.
Check Your Understanding
For each problem, identify the relevant development factor and trend duration before calculating.
- A hypothetical territory has a loss cost of $80 per exposure evaluated at the stated maturity. Its selected development factor is 1.15, and the target midpoint is one year after the historical midpoint. Trend is 6% annually. What is the adjusted loss cost? (Answer: $97.52 per exposure.)
- A hypothetical territory’s loss cost is $140 per exposure. The applicable development factor is 1.08, and the trend rate is 3% annually for two years. Calculate the adjusted cost, rounding to cents. (Answer: $160.41 per exposure.)
- Territory X has a historical loss cost of $110, a development factor of 1.10, and two years of 4% trend. Territory Y has a historical loss cost of $100, a development factor of 1.20, and one year of 4% trend. Calculate the adjusted costs and X-to-Y ratio, rounding the ratio to four decimals. (Answer: X = $130.87, Y = $124.80, ratio = 1.0487.)
- Two territories have historical costs of $75 and $100, the same development factor of 1.08, and the same 4% trend for 1.5 years to one target midpoint. Calculate each adjusted cost and the first-to-second ratio. Round costs to cents and the ratio to four decimals. (Answer: $85.91 and $114.54; ratio = 0.7500.)
- Two territories have the same historical cost, evaluation age, and historical period midpoint. One analyst applies development factors of 1.18 and 1.06, respectively, then applies the same trend factor to both. What is the ratio of the resulting adjusted costs, first territory to second, and what should be checked before interpreting it? (Answer: \(1.18/1.06=1.1132\); check whether the unequal development selections are justified, since the common trend factor cancels.)