Matching Loss and Premium Experience Periods

Why Period Matching Matters

A loss ratio compares losses with the premium associated with the exposure that generated those losses. The arithmetic is simple; choosing the two amounts to compare is not. If a numerator includes losses from one period while the denominator includes premium from a different period, the quotient may be calculated correctly but answer the wrong question.

For example, an actuary may want to summarize two years of experience. The numerator should represent losses for those two years, and the denominator should represent premium for those same two years on a compatible basis. Adding an extra year of losses without adding that year's premium will generally make the ratio appear higher. Omitting losses while retaining their premium will generally make it appear lower.

Definition: A loss experience period is the interval and basis to which the losses in an analysis are assigned. A premium experience period is the interval and basis to which the premium is assigned. For a conventional earned loss ratio, the periods and underlying exposure should be compatible: losses for the selected experience should be compared with earned premium for that same experience, using consistent coverage, business scope, and gross or net treatment.

“Compatible” does not necessarily mean that each loss and each premium dollar can be paired individually. In aggregate ratemaking, losses may be grouped by accident year while premium is aggregated over the corresponding experience interval. What matters is that the periods and bases represent the same intended experience. If a different basis is used, such as written rather than earned premium, the analysis needs a clear rationale and must not be presented as if it were the conventional earned loss ratio.

Identify the Period Basis Before Calculating

The period label describes how experience is grouped, not just the calendar dates printed in a table. In policy-year experience, policies are grouped according to their effective dates. In accident-year experience, losses are grouped according to when the accidents occur. In calendar-year experience, amounts are grouped according to when they are recorded in the calendar year; for paid losses, that is generally when payments are made. These bases can include different policies or claims even when they cover the same dates.

A calendar-year paid loss total, for instance, can include payments on claims arising from accidents in earlier years. Comparing that total with calendar-year earned premium may produce a calendar-year cash-flow or accounting ratio, but it does not turn the losses into accident-year losses. A candidate should state what the ratio measures rather than treating all annual totals as interchangeable.

Name the intended experience.
Specify the years, the loss basis (such as policy year or accident year), and the loss measure being used.

Check the premium basis.
Confirm that the premium covers the same intended interval and business, and use earned premium for a conventional earned loss ratio.

Aggregate only matching periods.
Sum losses and premium over the same selected years before dividing. Do not add a year to only one side of the ratio.

State limitations or differences.
If the bases differ, explain what the resulting ratio represents and why the comparison is appropriate—or correct the mismatch before interpreting it.

The Matched Loss Ratio and Why Pooling Works

For periods $t=1,\ldots,n$, let $L_t$ be the loss amount assigned to period $t$, and let $EP_t$ be the earned premium for the compatible experience in that period. The pooled loss ratio is the total loss divided by the total earned premium:

LR_{\text{pooled}}=\frac{\sum_{t=1}^{n} L_t}{\sum_{t=1}^{n} EP_t}.

This is not generally the simple arithmetic average of the annual loss ratios. To see why, write the annual ratio as $LR_t=L_t/EP_t$, when $EP_t>0$. Then $L_t=EP_t LR_t$, so:

LR_{\text{pooled}} =\frac{\sum_{t=1}^{n} EP_t LR_t}{\sum_{t=1}^{n} EP_t} =\sum_{t=1}^{n}\left(\frac{EP_t}{\sum_{j=1}^{n}EP_j}\right)LR_t.

Thus, the pooled ratio is an earned-premium-weighted average of the period ratios. A period with more premium receives more weight because it represents more premium volume. This identity depends on using the same set of periods in both sums; it does not repair a mismatch in the selected experience.

Worked Example: Pooling Policy-Year Experience

A hypothetical specialty insurer reviews policy-year experience for two years. Policy-year 2022 has losses of $420,000 and earned premium of $1,200,000. Policy-year 2023 has losses of $510,000 and earned premium of $1,500,000. Calculate the pooled loss ratio for these two policy years and compare it with the unweighted average of the annual ratios.

Method: match policy-year losses with policy-year earned premium, then pool totals. The annual ratios are:

LR_{2022}=\frac{\$420{,}000}{\$1{,}200{,}000}=0.35=35.00\%, \qquad LR_{2023}=\frac{\$510{,}000}{\$1{,}500{,}000}=0.34=34.00\%.

The matched loss total is $\$420{,}000+\$510{,}000=\$930{,}000$. The matched earned-premium total is $\$1{,}200{,}000+\$1{,}500{,}000=\$2{,}700{,}000$. Therefore:

LR_{\text{pooled}}=\frac{\$930{,}000}{\$2{,}700{,}000} =0.344444\ldots=34.44\%.

As a check, the premium weights are $1{,}200/2{,}700=4/9$ and $1{,}500/2{,}700=5/9$, when amounts are expressed in thousands. The weighted average is $(4/9)(35.00\%)+(5/9)(34.00\%)=34.44\%$. The simple average, $(35.00\%+34.00\%)/2=34.50\%$, is different because the two years have different premium volumes. Common mistake avoided: averaging annual ratios without considering their premium weights.

How a Period Mismatch Changes the Result

Suppose the loss total for a selected period is held fixed, but the premium denominator is taken from another period. The resulting ratio is not the matched ratio. If $L_A$ is the loss amount for period set $A$, $EP_A$ its compatible premium, and $EP_B$ a different premium amount, then:

\frac{L_A}{EP_B} =\frac{L_A}{EP_A}\frac{EP_A}{EP_B}.

This shows directly that the mismatch scales the matched ratio by the ratio of the two premium amounts. If $EP_B$ is smaller than $EP_A$, the calculated quotient is larger; if $EP_B$ is larger, it is smaller. When both the loss period and premium period change, the direction of distortion depends on the amounts included and cannot be inferred from the labels alone.

Worked Example: Accident-Year Losses and a Missing Premium Year

A hypothetical insurer summarizes accident-year experience for 2023 and 2024. Accident-year losses are $600,000 in 2023 and $750,000 in 2024. Corresponding earned premiums are $1,500,000 and $1,800,000. Calculate the matched two-year ratio, then calculate the ratio that would result if the 2023 losses were included but the 2023 premium were mistakenly omitted.

Matched calculation. Total losses are $\$600{,}000+\$750{,}000=\$1{,}350{,}000$. Total earned premium is $\$1{,}500{,}000+\$1{,}800{,}000=\$3{,}300{,}000$. Thus:

LR_{\text{matched}}=\frac{\$1{,}350{,}000}{\$3{,}300{,}000} =0.40909\ldots=40.91\%.

Mismatched calculation. If only the 2024 premium is retained, the denominator is $1,800,000 while the numerator still includes both years' losses:

LR_{\text{mismatched}}=\frac{\$1{,}350{,}000}{\$1{,}800{,}000} =0.75=75.00\%.

The mismatch overstates the matched ratio by $75.00\%-40.91\%=34.09$ percentage points. This does not mean the underlying claim experience suddenly worsened; the calculation compares two years of losses with only one year of premium. Common mistake avoided: assuming that a numerator and denominator with annual labels are compatible without checking which years each includes.

Worked Example: Calendar-Year Paid Losses Are Not Accident-Year Losses

A hypothetical carrier reports $920,000 of losses paid during calendar year 2025 and $2,300,000 of earned premium during calendar year 2025. The paid-loss total includes payments on claims from accidents occurring in 2024 and 2025. Calculate the calendar-year paid-loss-to-earned-premium ratio and state what it does—and does not—measure.

Method: preserve the stated calendar-year basis. Both amounts are reported for calendar year 2025, so their quotient is:

\frac{\$920{,}000}{\$2{,}300{,}000}=0.40=40.00\%.

This is a calendar-year ratio of paid losses to earned premium, as defined by the data. It is not an accident-year loss ratio: the numerator includes payments related to accidents from more than one year, whereas the denominator is calendar-year earned premium. To calculate an accident-year ratio, the losses and premium would need to be assembled on a compatible accident-year basis. Common mistake avoided: relabeling a calendar-year paid amount as accident-year losses merely because the premium is earned premium.

Common Traps and Practical Checks

Adding one more year to only one side. Before dividing, list the years included in the loss sum and in the premium sum. The lists should agree for a matched comparison.
Confusing period basis with period dates. Calendar-year, policy-year, and accident-year totals can cover the same named year while representing different underlying policies, claims, or cash flows.
Using written and earned premium as if they were interchangeable. Earned premium corresponds to coverage provided during the experience period. Written premium is recorded when policies are written; it can relate to coverage extending beyond that period. Use the premium basis appropriate to the analysis and name it accurately.
Taking an unweighted mean of annual ratios. For a pooled ratio, divide total losses by total premium. The equivalent average weights each annual ratio by its share of total premium.
Assuming every mismatch has the same direction. Omitting premium while retaining its losses tends to raise the ratio; omitting losses while retaining their premium tends to lower it. Other mismatches must be evaluated from the actual amounts.

Key takeaway: A loss ratio is interpretable only after the loss and premium periods—and their bases—have been checked. For a conventional earned loss ratio, sum losses and earned premium over the same compatible experience, then divide the matched totals.

Check Your Understanding

For each problem, identify the experience included in the calculation before interpreting the result.

A hypothetical book has policy-year 2021 losses of $360,000 and earned premium of $900,000, and policy-year 2022 losses of $440,000 and earned premium of $1,100,000. What is the pooled two-year loss ratio? (Answer: 40.00%)
For a hypothetical portfolio, accident-year 2023 losses are $600,000 and accident-year 2024 losses are $750,000. Their corresponding earned premiums are $1,500,000 and $1,800,000. What is the matched pooled loss ratio? (Answer: 40.91%)
A report contains accident-year losses of $480,000 for 2022, $600,000 for 2023, and $750,000 for 2024. It contains earned premium only for 2023 ($1,500,000) and 2024 ($1,800,000). What quotient results if all three loss amounts are divided by the two available premium amounts, and what is the mismatch? (Answer: 55.45%; it includes 2022 losses without 2022 premium.)
A hypothetical insurer has losses of $420,000 in 2023 and $580,000 in 2024, with earned premiums of $1,200,000 and $1,800,000, respectively. Calculate the matched pooled ratio, then the incorrect ratio if both years' losses are divided by 2024 premium alone. (Answer: 33.33% matched; 55.56% mismatched.)
A carrier reports $840,000 of losses paid during calendar year 2026 and $2,100,000 of earned premium during calendar year 2026. Some paid losses arose from earlier accidents. What ratio can be calculated from these figures, and is it necessarily an accident-year loss ratio? (Answer: 40.00% calendar-year paid-loss-to-earned-premium ratio; no, it is not necessarily an accident-year loss ratio.)

What You'll Learn