Underwriting-Year Grouping of Insurance Results

Why Group Insurance Results by Underwriting Year?

An insurance result can be grouped according to when coverage began or according to when a transaction was recorded. Those choices answer different questions. An underwriting-year view follows a cohort of policies from the year its coverage began; a calendar-year view describes activity assigned to a particular calendar reporting period. The same claim may therefore belong to one underwriting year but appear as a payment or expense in a later calendar year.

Underwriting-year grouping is useful when evaluating the results of a set of policies written into coverage during a defined year. It keeps the policy cohort together even when its claims are reported, paid, or otherwise recognized later. Calendar-year accounting, in contrast, organizes activity by accounting period. Neither grouping is inherently more correct: each describes the results using a different organizing date.

Definition: An underwriting year is the calendar year in which a policy's coverage begins. An underwriting-year group consists of policies whose coverage begins in that year. Premiums and claim results associated with those policies are attributed to their underwriting-year group, even if the related transactions occur in later years. In this tutorial, the coverage start date is the grouping rule.

A policy's coverage can extend beyond the end of the calendar year in which it begins. The policy remains in its original underwriting-year cohort; the anniversary of January 1 does not move it into a new cohort. Similarly, a covered claim is associated with the policy that provided coverage, not simply with the calendar year in which the claim is paid.

In practice, an insurer must specify the basis used to assemble a result, including which premium and loss measures are being compared. For the examples below, a stated premium amount is assigned to the policy's underwriting year, and a covered loss is assigned to the underwriting year of the policy providing coverage. Calendar-year paid-loss totals are explicitly identified as payments made in that calendar year. These conventions make the comparison precise without treating the different measures as interchangeable.

Underwriting Year and Calendar Year Compared

Question	Underwriting-year grouping	Calendar-year grouping
What determines the group?	The year coverage begins for the policy	The calendar accounting or reporting period
What does the group describe?	Results associated with a cohort of policies	Activity assigned to a period, potentially involving many cohorts
Where does a later claim payment appear?	In the underwriting-year cohort for the policy that covered the loss	In the calendar year in which the payment is assigned or recorded
Can results change after the year ends?	Yes. Later information can change the estimated results of the cohort	Yes. Period-end estimates and adjustments may also change reported accounting results

The distinction is not that one grouping includes claims and the other does not. It is the date used to assign activity to a group. A calendar-year paid-loss total can include payments on policies from several underwriting years. Conversely, an underwriting-year loss result can include claims whose payments happen after that underwriting year has ended.

How Cohort Assignment Works

The assignment can be expressed formally. Let $U(p)$ be the calendar year in which policy $p$'s coverage begins. Let $P_y$ denote the premium assigned to underwriting year $y$, and let $L_y$ denote the loss result assigned to that same cohort under the stated loss measure. Each policy is placed into exactly one underwriting-year group because it has one coverage start year.

P_y = \sum_{p:U(p)=y} P(p), \qquad L_y = \sum_{p:U(p)=y} L(p).

The sums make the grouping rule explicit: include a policy's associated amount when its coverage start year equals $y$. A transaction date does not, by itself, change $U(p)$. Calendar-year totals instead collect the amounts assigned to a calendar reporting year, which may draw on policies with different values of $U(p)$.

Identify the coverage start date.
Use the policy's coverage beginning, not the date a later claim payment occurs.

Assign the policy to an underwriting year.
The calendar year containing the coverage start date determines the cohort.

Keep associated results with the cohort.
Attribute the policy's premium and covered claim results to its underwriting year, following the stated measurement basis.

Build a calendar-year view separately.
For a calendar-year measure, use the accounting or reporting period specified for the activity, such as the year a payment was recorded.

The cohort totals reconcile to the total across all included policies: each policy appears in exactly one underwriting year, so summing the underwriting-year amounts includes each policy amount once. This is a partition of the policy population, not a claim that the cohort totals equal calendar-year transaction totals. The latter are grouped using a different date and may include activity from several cohorts.

\sum_y P_y = \sum_p P(p), \qquad \sum_y L_y = \sum_p L(p),

provided the same policies, amounts, and measurement basis are included on both sides. This condition matters: comparing a calendar-year paid-loss total with an underwriting-year incurred-loss estimate does not create a valid reconciliation, because the measures differ as well as the grouping dates.

Worked Example: Assigning Premium and Claims Across Two Years

A hypothetical insurer has three policies. Policy A begins coverage on December 20, 2024, has premium of $1,200, and covers a loss on December 28, 2024; the $500 claim payment is recorded on February 10, 2025. Policy B begins coverage on January 10, 2025, has premium of $1,800, and covers a loss in November 2025; the $700 payment is recorded in November 2025. Policy C begins coverage on December 15, 2025, has premium of $900, and covers a loss on December 20, 2025; its $300 payment is recorded in March 2026. Find the underwriting-year premium and loss totals, and the calendar-year paid-loss totals for 2025 and 2026.

Method. First assign each policy to the year its coverage begins. Then separately total claim payments by the calendar year in which they were recorded.

Policy A belongs to underwriting year 2024. It contributes $1,200 of premium and $500 of covered loss to that cohort, even though the payment is recorded in 2025. Policies B and C both belong to underwriting year 2025. Their combined premium is $1{,}800+900=\$2{,}700$, and their combined covered losses are $700+300=\$1{,}000$.

Therefore, the underwriting-year totals are: 2024 premium $1,200 and loss $500; 2025 premium $2,700 and loss $1,000.

For calendar-year paid losses, the 2025 payments are Policy A's $500 plus Policy B's $700, or $\$500+\$700=\$1{,}200$. The 2026 payment is Policy C's $300. The calendar-year paid-loss totals are therefore $1,200 for 2025 and $300 for 2026.

Check. The underwriting-year loss totals sum to $\$500+\$1{,}000=\$1{,}500$, the same as all three stated claim amounts. The calendar-year payments also sum to $\$1{,}200+\$300=\$1{,}500$. The totals reconcile, but their allocation differs: the 2025 calendar-year payment total includes a claim from underwriting year 2024.

Common mistake avoided: putting Policy A's $500 loss into underwriting year 2025 merely because it was paid in 2025. Payment timing determines its calendar-year payment grouping here; it does not change the underwriting year of the policy.

Comparing Underwriting-Year Results

A common comparison is a loss ratio, defined as the loss amount divided by the corresponding premium amount. The numerator and denominator must refer to a compatible cohort and measurement basis. The ratio is meaningful only when that alignment is maintained; combining losses from one set of policies with premiums from another can misstate the result.

Definition: For an underwriting year with positive premium $P_y$, the loss ratio is $L_y/P_y$, where $L_y$ and $P_y$ are the loss and premium measures assigned to that same underwriting-year cohort. State whether losses are paid, incurred, or otherwise measured; the ratio can differ under different loss measures.

The formula follows directly from the definition of a ratio: divide the cohort's loss amount by the cohort's premium amount. If a cohort has $600 of losses and $1,000 of premium, then $600/1{,}000=0.60$, or 60%. The interpretation is $0.60 of the stated loss measure per dollar of stated premium. It is not automatically a final assessment of profitability, because other expenses and the choice of loss measure also matter.

Worked Example: Calculating Cohort Loss Ratios

A hypothetical specialty insurer reports two underwriting-year cohorts. For year 2024, the premium assigned to the cohort is $2,400, and the selected incurred loss amount is $1,320. For year 2025, the corresponding amounts are $3,600 and $1,800. Calculate each loss ratio and the combined loss ratio using the stated amounts.

Method. Divide each cohort's loss amount by its own premium, then calculate the combined ratio from the combined numerator and denominator. Do not average the two percentages unless the cohort premiums are equal.

For 2024:

\frac{\$1{,}320}{\$2{,}400}=0.55=55.00\%.

For 2025:

\frac{\$1{,}800}{\$3{,}600}=0.50=50.00\%.

The combined loss amount is $\$1{,}320+\$1{,}800=\$3{,}120$. The combined premium is $\$2{,}400+\$3{,}600=\$6{,}000$. Thus:

\frac{\$3{,}120}{\$6{,}000}=0.52=52.00\%.

Check. The combined ratio, 52.00%, lies between the two cohort ratios, 50.00% and 55.00%, and is closer to 50.00% because the 2025 cohort has the larger premium.

Common mistake avoided: taking the unweighted average, $(55\%+50\%)/2=52.50\%$. That is not the combined loss ratio because the cohorts have different premium amounts. The combined calculation weights each cohort by its premium.

Why Underwriting-Year Results Develop

At a valuation date, an underwriting-year cohort may not have a complete view of its claims. Some claims may not yet have been reported, some reported claims may not yet be fully resolved, and some payments may occur later. As a result, an insurer may report a current estimate for the cohort and revise it as information emerges. The underwriting-year label stays fixed; the measured result can change.

This is different from assigning a later transaction to a new underwriting year. A claim from a policy that began coverage in 2025 remains associated with underwriting year 2025, even when a payment is made in 2026. The 2026 payment can appear in a 2026 calendar-year paid-loss total while also contributing to the developing 2025 underwriting-year result.

Worked Example: An Immature Cohort at Valuation

At the end of 2025, a hypothetical insurer evaluates policies that began coverage in 2024. The cohort has $1,000,000 of premium. Its claims to date consist of $400,000 paid, $200,000 of estimated outstanding reported claims, and $100,000 of estimated claims incurred but not reported. Calculate the current incurred loss estimate and loss ratio. If a later review increases the incurred loss estimate by $150,000, calculate the revised ratio.

Method. For this example, incurred losses are defined as paid losses plus estimated outstanding reported claims plus estimated incurred-but-not-reported losses. The premium and each loss estimate are for the same underwriting-year cohort.

The current incurred loss estimate is:

\$400{,}000+\$200{,}000+\$100{,}000=\$700{,}000.

The current loss ratio is:

\frac{\$700{,}000}{\$1{,}000{,}000}=0.70=70.00\%.

The later review increases the loss estimate by $150,000, so the revised estimate is $\$700{,}000+\$150{,}000=\$850{,}000$. With the same stated premium, the revised ratio is:

\frac{\$850{,}000}{\$1{,}000{,}000}=0.85=85.00\%.

Check. The loss estimate increased by 15% of the $1,000,000 premium, so the loss ratio rises by 15 percentage points, from 70.00% to 85.00%. The policy cohort is still underwriting year 2024; only its estimated result has changed.

Common mistake avoided: treating the later $150,000 estimate change as a new underwriting-year loss because it was recognized after 2024. The grouping remains with the 2024 policies; the date of a later estimate update does not reset the cohort.

Common Interpretation Traps

Using the payment year as the underwriting year. A later payment can belong to a policy cohort established years earlier.
Treating a policy as if it changes cohorts at year-end. A policy that begins in December and runs into the next calendar year remains in the underwriting year in which coverage began.
Comparing mismatched measures. A calendar-year paid-loss amount and an underwriting-year incurred-loss estimate differ in both timing and loss basis. Do not interpret their difference as a simple cohort gain or loss.
Assuming an underwriting-year result is final when the year closes. The cohort is defined at coverage inception, but its estimated claims and resulting ratios may develop afterward.
Taking an unweighted average of cohort ratios. A combined ratio uses total losses divided by total premium, not necessarily the arithmetic average of cohort percentages.

Key takeaway: Underwriting year answers, “When did this policy's coverage begin?” Calendar year answers, “Which calendar reporting period contains this activity?” Keep the two groupings distinct, state the measure being used, and associate a policy's results with its coverage-start cohort.

Check Your Understanding

For each problem, apply the stated grouping rule and keep the premium, loss, and timing measures aligned.

A policy begins coverage on December 31, 2023, and its one-year term ends in 2024. A covered loss is paid in 2024. Which underwriting year contains the policy and its covered loss? (Answer: Underwriting year 2023.)
Two policies begin coverage in 2025. Their premiums are $900 and $1,100; their selected loss amounts are $300 and $500. Calculate the 2025 underwriting-year loss ratio. (Answer: $(\$300+\$500)/(\$900+\$1{,}100)=40.00\%$.)
In calendar year 2026, paid claims are $250 from underwriting year 2024, $400 from underwriting year 2025, and $150 from underwriting year 2026. Find the 2026 calendar-year paid-loss total and identify the underwriting years represented. (Answer: $800; underwriting years 2024, 2025, and 2026.)
Underwriting year 2023 has premium of $1,200,000 and loss of $720,000; underwriting year 2024 has premium of $1,500,000 and loss of $1,050,000. Find the combined loss ratio. (Answer: $(\$720{,}000+\$1{,}050{,}000)/(\$1{,}200{,}000+\$1{,}500{,}000)=65.56\%$.)
At valuation, an underwriting-year cohort has $1,000,000 premium, $400,000 paid, $200,000 of outstanding reported claims, and $100,000 of estimated incurred-but-not-reported claims. A later review increases the incurred-loss estimate by $150,000. Find the revised loss ratio and state whether the underwriting year changes. (Answer: $(\$400{,}000+\$200{,}000+\$100{,}000+\$150{,}000)/\$1{,}000{,}000=85.00\%$; the underwriting year does not change.)

What You'll Learn