Selecting the Premium Base

Why the Premium Base Matters

A rate indication compares expected costs with the premium available to cover them. Even when the loss estimate is appropriate, the comparison can be misleading if its premium denominator represents different coverage, risks, or time periods. Selecting the premium base is therefore a matter of matching the measures before interpreting their ratio.

The premium base is not simply whichever premium total is easiest to retrieve. It is the premium measure chosen to represent the same covered experience as the losses, expressed on a basis appropriate for the indication. In particular, a current-rate-level comparison uses premium corresponding to the experience being analyzed, restated to the rate level used for the comparison. The detailed mechanics of premium earning and rate-level adjustment are separate issues; here the focus is on choosing the measure those calculations should represent.

Definition: The premium base for an indication is the premium measure used as the denominator when comparing the selected losses with premium for the experience being analyzed. It should represent the same coverage, relevant population, and experience period as the losses, and its rate-level basis should be appropriate to the indication.

The premium base is distinct from the exposure base. An exposure base measures the amount of insured risk, such as vehicle-years or building-years. Premium reflects the rates applied to that exposure, along with the selected premium basis. The earlier material on selecting the exposure base addresses what quantity of risk to measure; selecting the premium base asks which premium amount represents that risk for the loss comparison.

What Must Match

A useful selection begins by describing the losses precisely. Identify the coverage, the population or segment, and the period to which they relate. Then select premium that corresponds to those same dimensions. A total premium figure can be arithmetically correct and still be the wrong denominator.

Dimension	Question to ask	Potential mismatch
Coverage	Does the premium include only the coverage represented by the losses?	Using package premium for losses from one coverage only
Population or segment	Does premium cover the same risks or business segment?	Including a different territory or class of risks
Experience period	Does the premium correspond to the experience underlying the losses?	Pairing losses from one period with premium from another
Rate level	Is the premium expressed at the level appropriate for the comparison?	Using recorded premium at historical rates as if it were at current rates
Premium basis	Are inclusions, deductions, and other definitions consistent?	Comparing losses on one basis with premium on a materially different basis

Matching does not require that every data label be identical. For example, different accounting or data-recording conventions can still identify comparable risks and coverage. The actuary must understand what each measure represents and select a premium measure that corresponds to the losses, rather than relying on the label alone.

Earned premium is associated with coverage provided during a period, while written premium is associated with policies written during a period. They can therefore describe different sets of risk and different timing. The appropriate choice depends on how the loss experience is defined and which premium measure represents that same experience. The later material on earned and written premium examines their construction and use in more detail.

Current-Rate-Level Comparison Base

An indication may compare losses from historical experience with premium expressed at a specified rate level. In that setting, the relevant premium base is not necessarily the amount originally recorded: it is the premium measure that represents the selected experience on the rate-level basis used for the comparison. The rate-level adjustments themselves must be based on appropriate data and assumptions; selecting the premium base means first identifying which premium is meant to be represented.

This distinction matters because the same exposure can generate different premium under different rates. If a rate change occurred during or after the experience period, historical premium and premium at a current rate level are not automatically equal. A raw ratio using historical premium answers a different question from a ratio using premium at the comparison rate level.

\text{Loss ratio on a stated basis} = \frac{\text{losses on that basis}}{\text{premium on the corresponding basis}}.

This ratio is meaningful only when the numerator and denominator are appropriately paired. The formula does not repair a mismatch in coverage, period, or rate level. It simply calculates a ratio from the amounts supplied.

Why the Denominator Must Represent the Same Experience

The reason for matching is visible when an experience is divided into compatible groups. Suppose group $j$ has losses $L_j$ and corresponding premium $P_j$. If all groups use the same coverage and measurement basis, the aggregate loss ratio is

\frac{\sum_j L_j}{\sum_j P_j} = \sum_j \left(\frac{P_j}{\sum_k P_k}\right)\frac{L_j}{P_j}.

To see why, expand the right-hand side:

\sum_j \left(\frac{P_j}{\sum_k P_k}\right)\frac{L_j}{P_j} = \sum_j \frac{L_j}{\sum_k P_k} = \frac{\sum_j L_j}{\sum_k P_k}.

Thus the aggregate ratio is a premium-weighted average of the group ratios, provided each numerator and denominator refer to corresponding experience. If the denominator includes premium from an unrelated coverage or period, the ratio is no longer that weighted average for the selected losses. The arithmetic still works, but the resulting statistic does not describe the intended comparison.

Specify the loss measure.
Write down the coverage, population or segment, and period represented by the losses.

Identify candidate premium measures.
Check what risks, coverage, time period, and premium basis each candidate includes.

Select the corresponding measure.
Choose the premium that represents the same experience and the rate level required for the comparison.

Test the match before calculating.
Confirm that no unrelated premium has been added and that no relevant premium has been omitted.

Worked Applications

Worked Example: Selecting Premium for One Coverage

Method: Match coverage first, then calculate the ratio. A hypothetical insurer is evaluating collision coverage for a defined experience period. The selected collision losses are $ \$312{,}000 $. The corresponding collision premium is $ \$480{,}000 $. The data extract also shows $ \$720{,}000 $ of premium for all coverages combined. Calculate the loss ratio using the appropriate premium base and quantify the result of using the combined premium instead.

Step 1: Match the coverage. The losses are for collision coverage only, so the corresponding premium base is the collision premium, $ \$480{,}000 $. The all-coverages total includes premium that does not correspond to the selected losses.

Step 2: Calculate the matched ratio.

\frac{\$312{,}000}{\$480{,}000}=0.65=65.00\%.

Step 3: Show the mismatch effect. Using all-coverages premium would give

\frac{\$312{,}000}{\$720{,}000} =0.4333\ldots =43.33\%.

The lower result does not show a lower collision loss ratio; it results from adding premium for other coverages without adding their losses. The correct collision loss ratio is 65.00%. The common mistake is to use a broad premium total simply because it is available.

Worked Example: Distinguishing Historical Premium from the Comparison Base

Method: Use premium at the stated comparison rate level, not the unadjusted historical amount. In a hypothetical book of small craft policies, the selected experience represents $900$ exposure-years. Historical premium was $ \$675{,}000 $, reflecting a rate of $ \$750 $ per exposure-year. For this example, the corresponding premium at the comparison rate level is given as $ \$825 $ per exposure-year. Selected losses are $ \$445{,}500 $. Calculate the ratio on the comparison basis and compare it with the ratio using historical premium.

Step 1: Confirm the historical amount.

900 \times \$750 = \$675{,}000.

This agrees with the historical premium supplied.

Step 2: Calculate premium at the comparison rate level.

900 \times \$825 = \$742{,}500.

Step 3: Calculate the ratio using the appropriate comparison base.

\frac{\$445{,}500}{\$742{,}500}=0.60=60.00\%.

Step 4: Calculate the unadjusted historical ratio for contrast.

\frac{\$445{,}500}{\$675{,}000}=0.66=66.00\%.

The historical premium is internally consistent with the stated historical rate, but it is not the premium at the comparison rate level. The appropriate ratio for the stated comparison base is 60.00%. The common mistake is to treat the amount recorded historically as automatically suitable for a current-rate-level comparison.

Worked Example: Rejecting a Premium from the Wrong Period

Method: Match the premium experience period to the period represented by the losses. A hypothetical equipment insurer is analyzing accident-year losses of $ \$384{,}000 $ for a specified coverage. Candidate premium amounts are $ \$576{,}000 $ for the corresponding accident-year experience, $ \$640{,}000 $ for calendar-year experience, and $ \$600{,}000 $ for policies written during the year. Assume the first amount is the selected premium at the required rate level. Calculate the appropriate ratio and the two ratios that would result from the mismatched candidates.

Step 1: Select the matching premium. The losses are organized by accident year, and the corresponding accident-year premium is given at the required rate level. Use $ \$576{,}000 $.

Step 2: Calculate the matched ratio.

\frac{\$384{,}000}{\$576{,}000} = \frac{2}{3} = 66.67\%.

Step 3: Show the calendar-year mismatch.

\frac{\$384{,}000}{\$640{,}000} =0.60 =60.00\%.

Step 4: Show the written-premium mismatch.

\frac{\$384{,}000}{\$600{,}000} =0.64 =64.00\%.

The selected result is 66.67%. The other calculations are arithmetically correct but use premium amounts representing different timing bases. The common mistake is to assume that a premium total for the same calendar year must correspond to accident-year losses.

Common Mistakes and a Practical Check

Before using a premium denominator, ask whether it would still be appropriate if the loss total were described in full. “Premium for this year” is not precise enough: specify the coverage, the business segment, the period convention, and the rate-level basis. A short check often reveals that two totals that look compatible were assembled from different underlying records.

Using package premium for single-coverage losses. Premium for other coverages belongs with their corresponding losses, not automatically in the denominator for one coverage.
Mixing periods because the labels look similar. A calendar-year premium total is not necessarily the right base for accident-year losses.
Assuming recorded premium is already at the comparison rate level. Historical premium may reflect rates different from the level being used in the indication.
Changing the denominator without changing the interpretation. A loss ratio is defined by both its losses and its premium base; changing the base changes what the ratio measures.
Adding totals before checking what they contain. Combined losses and combined premium must each include the same selected coverages and experience. For example, losses of $ \$315{,}000 $ and $ \$45{,}000 $ total $ \$360{,}000 $, not $ \$540{,}000 $; with corresponding premium of $ \$540{,}000 $, their ratio is $ \$360{,}000 / \$540{,}000 = 66.67\% $.

Key takeaway: Select premium by matching the losses in coverage, population, and experience period, then confirm that the premium is on the rate-level basis required for the comparison. A correct division using an unmatched denominator is still the wrong indication input.

Check Your Understanding

For each problem, identify the appropriate premium base before calculating. Amounts are hypothetical.

A specified coverage has losses of $ \$198{,}000 $. Corresponding premium for that coverage and period is $ \$300{,}000 $, while premium for all coverages is $ \$450{,}000 $. Calculate the matched loss ratio and the ratio using all-coverages premium. (Answer: Matched $66.00\%$; all-coverages calculation $44.00\%$.)
A book has $640$ exposure-years. Historical premium was based on $ \$500 $ per exposure-year, while the stated comparison rate is $ \$550 $ per exposure-year. Losses are $ \$211{,}200 $. Calculate the premium base at the comparison rate level and the resulting loss ratio. (Answer: Premium $ \$352{,}000 $; loss ratio $60.00\%$.)
Accident-year losses are $ \$525{,}000 $. The matching accident-year premium at the required rate level is $ \$750{,}000 $; calendar-year premium is $ \$840{,}000 $. Calculate the matched ratio and the ratio obtained from the calendar-year amount. (Answer: Matched $70.00\%$; calendar-year calculation $62.50\%$.)
For one coverage and one experience period, segment A has losses of $ \$180{,}000 $ and premium of $ \$300{,}000 $; segment B has losses of $ \$240{,}000 $ and premium of $ \$400{,}000 $. Calculate the aggregate ratio using both matching segments. (Answer: Total losses $ \$420{,}000 $, total premium $ \$700{,}000 $, aggregate ratio $60.00\%$.)
Collision losses for a defined period are $ \$264{,}000 $. Candidate amounts are current-rate-level earned collision premium of $ \$360{,}000 $, written collision premium of $ \$420{,}000 $, and earned premium for all coverages of $ \$600{,}000 $. Use the stated matching current-rate-level earned premium and calculate the ratio. (Answer: $ \$264{,}000 / \$360{,}000 = 73.33\%$.)

What You'll Learn