Historical Reasonableness Checks

Why Check a Proposed Change Against History?

A rate indication is a prospective estimate, while historical premium and losses describe earlier periods. A historical reasonableness check asks whether a proposed change, when viewed against the experience that led to it, produces a plausible result. It can expose a mismatch in assumptions or a calculation error; it does not establish that the historical experience will repeat.

The comparison must put amounts on a consistent basis. Historical earned premium reflects the rate levels in effect as policies earned. Losses reflect the claim costs and conditions of their own periods. A direct comparison of amounts from different periods may therefore conflate rate changes, changes in exposure or business mix, and changes in loss costs.

Definition: A historical reasonableness check compares historical premium, losses, or underwriting results with values adjusted to a stated common basis, and then assesses whether a proposed rate change is plausible in that context. The check is diagnostic: it is not a substitute for a prospective rate indication.

Put Premium and Losses on Comparable Bases

Suppose earned premium $EP$ was written at an average rate-level factor $r$, where the common reference rate level is 1.00. If the rate change is multiplicative and exposure and coverage are held constant, premium adjusted to the reference level is $EP/r$. If the proposed change is $g$, the corresponding premium at the new level is $(EP/r)(1+g)$.

EP_{\text{reference}}=\frac{EP}{r}, \qquad EP_{\text{proposed}}=\frac{EP}{r}(1+g)

The division follows from how a rate-level factor scales premium. If premium at factor $r$ is $EP$, then the corresponding premium at factor 1.00 is $EP$ multiplied by $1/r$. Applying a further proposed change multiplies that amount by $1+g$. This adjustment does not remove changes in exposure, coverage, or policyholder mix; it only adjusts for the stated rate-level difference under the assumptions just given.

Losses may also require adjustment to a common cost level. If historical losses are adjusted by a stated prospective factor $t$, then $L_{\text{adjusted}}=L t$. The factor must represent the period change being modeled and be applied consistently. Loss trend is not a premium rate-level adjustment: it changes the loss estimate, not historical premium.

Important condition: Use $EP/r$ only when $r$ is the applicable average rate-level factor for the earned premium and the rate change acts proportionally. Use a loss adjustment factor only when its period and assumptions are specified. If exposure, coverage, or mix changes materially, these simple adjustments alone do not make periods fully comparable.

Worked Example: Adjusting a Historical Loss Ratio for a Proposed Increase

A hypothetical portfolio earned $1,500,000 of premium at an average rate-level factor of 0.96. Incurred losses were $990,000. Assume no loss trend adjustment is needed for this mechanical check, and hold exposures and coverage constant. Calculate the loss ratio at rate level 1.00 and after a proposed additional 5% increase.

Method. First adjust premium to rate level 1.00. Then apply the proposed increase and divide the unchanged losses by the resulting premium.

On-level premium is

\frac{\$1{,}500{,}000}{0.96}=\$1{,}562{,}500.

Premium after the additional 5% increase is

\$1{,}562{,}500(1.05)=\$1{,}640{,}625.

The historical loss ratio at rate level 1.00 is $\$990{,}000/\$1{,}562{,}500=0.6336$, or 63.36%. After the proposed increase, the mechanical loss ratio is

\frac{\$990{,}000}{\$1{,}640{,}625} =0.60342857\ldots =60.34\%.

Check and trap. The result is below 63.36%, as expected, because the proposal increases premium while losses are held fixed. A common error is to multiply the historical premium by 0.96 again, or to calculate $1{,}500{,}000(1.05)$ without first adjusting its rate level. Both omit the conversion to the stated reference level. The final answer, rounded to two decimal places, is 60.34%.

Check the Underwriting Result, Not Only the Loss Ratio

A loss ratio omits expenses and other provisions. When the historical information and assumptions are sufficient, a normalized underwriting result can provide a broader check. For this calculation, define the result as premium less adjusted losses and expenses. If expenses include a fixed amount $F$ and a variable proportion $e$ of premium, then expenses are $F+eEP_{\text{reference}}$.

U_{\text{reference}} =EP_{\text{reference}}-L_{\text{adjusted}} -\left(F+eEP_{\text{reference}}\right)

This definition gives an underwriting result in dollars on the chosen basis. It does not by itself account for investment income or prove that the selected rates are adequate. The expense assumptions and loss adjustment must match the period and purpose of the comparison.

Worked Example: Normalizing an Underwriting Result

A hypothetical book has $2,000,000 of earned premium at rate level 0.97 and $1,200,000 of incurred losses. For a prospective comparison, apply an 8% loss adjustment. At rate level 1.00, expenses are estimated as $80,000 fixed plus 15% of on-level premium. Calculate the normalized underwriting result.

Method. Adjust premium to rate level 1.00, adjust losses by the stated prospective factor, calculate expenses on on-level premium, and subtract losses and expenses from premium.

On-level premium is

EP_{\text{reference}} =\frac{\$2{,}000{,}000}{0.97} =\$2{,}061{,}855.67.

Adjusted losses are

L_{\text{adjusted}} =\$1{,}200{,}000(1.08) =\$1{,}296{,}000.

Variable expenses are $0.15(\$2{,}061{,}855.67)=\$309{,}278.35$. Total expenses are $\$80{,}000+\$309{,}278.35=\$389{,}278.35$. Therefore,

U_{\text{reference}} =\$2{,}061{,}855.67-\$1{,}296{,}000-\$389{,}278.35 =\$376{,}577.32.

Check and trap. The result is positive because premium exceeds adjusted losses plus expenses. A common mistake is to apply the 15% variable expense provision to the original $2,000,000 rather than to on-level premium. That would mix rate levels and understate the expense amount by $0.15(\$61{,}855.67)=\$9{,}278.35$. The normalized underwriting result is $376,577.32, rounded to the nearest cent.

Use Multiple Periods Carefully

A multi-period comparison should keep each period’s premium and loss adjustments visible. A portfolio’s total loss ratio is total losses divided by total premium, not the unweighted average of the individual period loss ratios. If periods have different volume, an unweighted average gives each period equal influence regardless of its premium size.

The next example adjusts two years’ earned premium to one rate level and their losses to a common prospective loss-cost level. The proposed increase is then applied to the adjusted premium. The calculation is deliberately mechanical: it assumes the adjusted historical exposures remain representative and does not model changes in business mix.

Worked Example: Comparing Two Historical Periods With a Proposed Change

A hypothetical portfolio has the following experience. In 2023, earned premium was $4,800,000 at average rate-level factor 0.92 and incurred losses were $3,360,000. In 2024, earned premium was $5,200,000 at factor 0.96 and losses were $3,796,000. For a common prospective loss-cost level, apply two years of 6% annual loss trend to 2023 losses and one year of 6% trend to 2024 losses. Assume the common premium reference level is 1.00. Assess the loss ratio after a proposed 6% increase, and compare it with a stated target loss ratio of 65%.

Method. On-level each period’s premium, trend each period’s losses to the specified prospective level, pool the adjusted amounts, and then apply the proposed premium change.

The adjusted premiums are

EP_{2023}=\frac{\$4{,}800{,}000}{0.92}=\$5{,}217{,}391.30, \qquad EP_{2024}=\frac{\$5{,}200{,}000}{0.96}=\$5{,}416{,}666.67.

Their total is $\$10{,}634{,}057.97$. The loss trend factors are $1.06^2=1.1236$ for 2023 and $1.06$ for 2024. Thus the adjusted losses are

L_{2023}=\$3{,}360{,}000(1.1236)=\$3{,}775{,}296, \qquad L_{2024}=\$3{,}796{,}000(1.06)=\$4{,}023{,}760.

Total adjusted losses are $\$3{,}775{,}296+\$4{,}023{,}760=\$7{,}799{,}056$. The pooled loss ratio at reference rate level 1.00 is

\frac{\$7{,}799{,}056}{\$10{,}634{,}057.97} =0.7334036\ldots =73.34\%.

After a 6% increase, adjusted premium is $\$10{,}634{,}057.97(1.06)=\$11{,}272{,}101.45$. The resulting mechanical loss ratio is $\$7{,}799{,}056/\$11{,}272{,}101.45=69.19\%$, rounded to two decimal places. This remains above the stated target loss ratio of 65%.

As a further consistency check, the increase that would equate this adjusted loss ratio to the 65% target, under the same fixed-loss assumptions, is

\frac{73.34036\%}{65\%}-1=0.1283\ldots=12.83\%.

Interpretation and trap. The calculation suggests that a 6% increase alone does not bring the adjusted loss ratio to 65% under these assumptions. It does not establish a final selected change: expense provisions, other assumptions, and the credibility of this historical comparison still matter. A common mistake is to average the two period loss ratios equally. Instead, pool adjusted losses and premium, as done here, so that each period contributes in proportion to its adjusted premium.

Limitations and Practical Checks

Historical comparisons are useful because they provide a separate view of whether a proposal is consistent with observed results after stated adjustments. They are not forecasts by themselves. Interpret any result in light of what the calculation holds constant and what it cannot normalize.

Exposure and mix may change. Dividing premium by a rate-level factor does not adjust for changes in the number or type of risks, limits, deductibles, territory, or coverage.
Loss adjustments require a defined basis. A trend factor should refer to the correct time interval and prospective period. Applying the same number of trend years to every period can misstate losses when the periods differ.
Historical results can be volatile. A few large claims or random variation can make one period unrepresentative. A mechanical check does not resolve that uncertainty.
Accounting measures must be consistent. Earned premium, incurred losses, and expenses must refer to compatible periods and definitions. A comparison can be misleading if the components are measured on different bases.
A target ratio is an assumption. If a comparison uses a target loss ratio, the target must be consistent with the expense and other provisions being contemplated. A loss-ratio check alone is not a full underwriting result.

Before accepting a historical check, verify the reference rate level, the direction of each adjustment, the loss trend interval, the expense base, and the pooling method. Then state the conclusion conditionally: for example, “under the stated fixed-exposure and trend assumptions, the proposed change produces this adjusted loss ratio.” That is more informative than presenting an adjusted number as if it were an unconditional prediction.

Key takeaway: Normalize earned premium for rate level, adjust losses only on a stated and consistent basis, and compare pooled results when combining periods. Treat the result as evidence for evaluating plausibility, not as a replacement for prospective analysis or judgment.

Check Your Understanding

For each problem, keep the stated assumptions and adjustment bases separate. Dollar answers are rounded to the nearest cent when needed; percentages are rounded to two decimal places.

A hypothetical account has $840,000 earned premium at rate-level factor 0.875 and $588,000 incurred losses. With no loss adjustment, find the loss ratio at reference level 1.00. (Answer: 61.25%)
A hypothetical portfolio has $1,250,000 earned premium at rate-level factor 0.95 and $900,000 incurred losses. Apply one year of 4% loss trend, then find the adjusted loss ratio at reference rate level 1.00. (Answer: 71.14%)
An adjusted historical loss ratio is 72.80%. If premium is increased by 8% while adjusted losses are held fixed, what is the resulting mechanical loss ratio? (Answer: 67.41%)
A hypothetical book has $1,400,000 earned premium at rate-level factor 0.98 and $910,000 incurred losses. Apply a 5% loss adjustment. At reference level 1.00, expenses are $50,000 fixed plus 12% of premium. Find the normalized underwriting result. (Answer: $251,642.86)
Two periods have already been adjusted to common bases. Period A has premium of $2,400,000 and losses of $1,680,000 before a 15% loss adjustment; Period B has premium of $3,600,000 and losses of $2,520,000 before a 5% loss adjustment. Pool the adjusted results, then apply a proposed 10% premium increase with losses fixed. What are the pooled loss ratio before the increase and the mechanical loss ratio after it? If the target loss ratio is 68%, what increase would equate the pooled ratio to that target under the same assumptions? (Answer: 76.30% before; 69.36% after; 12.21% increase to reach the target.)

What You'll Learn