Cancellations, Reinstatements, And Earned Time

Why Actual In-Force Time Matters

A policy’s scheduled term does not tell you how long its coverage was actually in force. A cancellation can end coverage before expiration, a reinstatement can restore coverage after a gap, and a replacement policy can create overlapping records for what is effectively one continuous period of protection. For time-based exposure, the relevant quantity is the time coverage was actually in force—not simply the time between the original inception and expiration dates.

This distinction matters in ratemaking data because exposure is the denominator used to relate experience to the amount of risk observed. Counting dates when coverage was not in force understates the observed loss rate per unit of exposure; counting the same risk twice overstates exposure. The method is to identify the actual in-force intervals first, then measure and combine their durations.

Definition: An in-force interval is a period during which a specified coverage is active. For date-based calculations, represent an interval as \([a,b)\): coverage is in force starting on date \(a\), but not on date \(b\). Its duration is the elapsed time from \(a\) to \(b\). The earned time over a period is the total time in that period during which the coverage was in force. A time-based exposure is earned time expressed in the selected exposure unit, such as policy-years.

Using half-open intervals makes effective dates unambiguous: the cancellation date is the first date coverage is no longer active, and the reinstatement date is the first date it is active again. This convention avoids counting a transition date in both the ending and beginning interval. If a contract or data system defines effective times differently, follow that definition consistently and document it.

From In-Force Intervals to Exposure

For a time-based exposure, first add the durations of the nonoverlapping in-force intervals. Then divide by the length of the exposure unit. For an annual policy, the result is commonly expressed in policy-years. The denominator must match the convention being used: an actual-term fraction uses the number of days in the policy term, while a standardized annual exposure convention uses its specified year length.

D_{\text{in force}}=\sum_{j=1}^{k}D_j, \qquad e=\frac{D_{\text{in force}}}{L}

Here \(D_j\) is the duration, in days, of the \(j\)-th in-force interval; \(D_{\text{in force}}\) is total in-force time; \(L\) is the number of days in the selected exposure unit; and \(e\) is exposure measured in that unit. For example, if \(L=365\), then 182.5 in-force days correspond to 0.5 policy-years. With a leap-year policy term, an actual-term calculation may instead use \(L=366\). A separate standardized convention may prescribe another denominator; the key is not to mix conventions within a calculation.

The reason durations can be added is that exposure measures how long the risk is present. If \(I(t)\) equals 1 when coverage is in force at time \(t\) and 0 otherwise, then total in-force time is the integral of that indicator over the period. The indicator contributes 1 during covered time and 0 during uncovered time. For disjoint intervals, integrating over the full period is the same as adding the durations of each covered interval.

D_{\text{in force}}=\int I(t)\,dt =\sum_{j=1}^{k}\int_{a_j}^{b_j}1\,dt =\sum_{j=1}^{k}(b_j-a_j)

This justification also shows why gaps must be excluded and why overlapping records for the same risk cannot automatically be added: where two records describe one unit of risk at the same time, they do not make the risk present for twice as long. Whether separate concurrent policies represent separate exposure depends on the exposure base and the facts of the insurance arrangement.

Establish the relevant coverage.
Identify the coverage and risk unit whose exposure is being measured. Do not assume that every policy record represents a separate insured unit.

List effective intervals.
Use actual effective dates for inception, cancellation, and reinstatement. Split coverage into separate intervals wherever there is a genuine gap.

Measure and reconcile.
Compute each interval’s duration, add disjoint durations, and check that covered time plus uncovered time equals the full period under review.

Convert to the chosen exposure unit.
Divide by the stated year or term length. State the denominator and retain unrounded values until the final result.

Worked Calculations

Worked Example: Cancellation Before Expiration

A hypothetical annual property policy covers one location from January 1, 2025, through January 1, 2026. The policy is cancelled effective July 1, 2025, and coverage does not continue after cancellation. Calculate its actual-term exposure in policy-years using a 365-day annual denominator.

Method. Represent the covered period as \([{\rm Jan\ 1},{\rm Jul\ 1})\). The policy is in force up to, but not including, the effective cancellation date.

Step 1: Count the in-force days. In 2025, January through June contain \(31+28+31+30+31+30=181\) days. Thus \(D_{\text{in force}}=181\).

Step 2: Convert days to exposure. Using the stated denominator, \(e=181/365=0.4958904\ldots\), or 0.4959 policy-years to four decimal places.

Check. The full policy term has 365 days. The uncovered period from July 1, 2025, to January 1, 2026, has \(31+31+30+31+30+31=184\) days. Covered plus uncovered time is \(181+184=365\).

Common mistake avoided. Using the scheduled expiration would assign a full policy-year of exposure even though coverage ended on July 1. The calculation uses the actual coverage interval, not the original policy term alone.

Worked Example: Cancellation Followed by Reinstatement

A hypothetical annual liability policy runs from January 1, 2025, to January 1, 2026. It is cancelled effective May 1. The insurer reinstates it effective June 15, with no coverage for the intervening period, and it is cancelled again effective October 1. Find the in-force exposure as a fraction of the 365-day term.

Method. Treat the two covered spells as separate intervals: \([{\rm Jan\ 1},{\rm May\ 1})\) and \([{\rm Jun\ 15},{\rm Oct\ 1})\). Reinstatement restores coverage from its effective date; it does not make the gap covered retroactively.

Step 1: Measure the first interval. January 1 to May 1 is \(31+28+31+30=120\) days.

Step 2: Measure the second interval. June 15 to October 1 is \(16+31+31+30=108\) days. June 15 through June 30 contributes 16 elapsed days under the half-open interval convention.

Step 3: Add and convert. Total in-force time is \(120+108=228\) days. Therefore \(e=228/365=0.6246575\ldots\), or 0.6247 policy-years to four decimal places.

Check. The uncovered intervals are May 1 to June 15, or 45 days, and October 1 to January 1, or 92 days. Thus \(120+45+108+92=365\) days, which reconciles to the full term.

Common mistake avoided. Counting January 1 through October 1 as one continuous covered period would include the 45-day lapse. The two intervals must be measured and added separately.

Worked Example: A Replacement Record With Overlapping Dates

A hypothetical annual policy covers one vehicle from January 1, 2025, to January 1, 2026. The vehicle is replaced, and the replacement policy takes effect August 20. The old policy record shows an expiration of September 1, but the old coverage was actually cancelled effective August 20; the September 1 date is an uncorrected scheduled end date. Assume the replacement policy continues through January 1, 2026, and only one vehicle is to be counted. Calculate exposure using a 365-day year.

Method. Use the actual coverage intervals, not the stale date in the old record. The in-force coverage is continuous from January 1 to August 20 under the old policy and from August 20 to January 1 under the replacement policy.

Step 1: Measure the old policy interval. January 1 to August 20 is 231 days: \(31+28+31+30+31+30+31+19=231\). Equivalently, August 20 is 231 elapsed days after January 1.

Step 2: Measure the replacement interval. The full year has 365 days, so August 20 to January 1 is \(365-231=134\) days.

Step 3: Add and convert. Total in-force time is \(231+134=365\) days. Exposure is \(365/365=\mathbf{1.0000}\) policy-year.

Check. The policy records meet on August 20, so there is no uncovered gap or overlap in the actual coverage intervals.

Common mistake avoided. Counting the old record through September 1 would give 243 days; adding the replacement’s 134 days would produce 377 days, or \(377/365=1.0329\) policy-years. That overcounts the 12 days from August 20 to September 1 because the old scheduled end date does not reflect actual coverage.

Interpreting Cancellations, Reinstatements, and Replacements

The calculation depends on the coverage facts and on what exposure unit is being counted. A cancellation effective on a stated date normally ends the in-force interval at that date under the convention used here. A reinstatement begins a new interval on its effective date. If the insurer reinstates coverage retroactively, use the documented effective interval rather than treating the original lapse as automatically covered. The date a transaction is processed is not necessarily the date its coverage takes effect.

A replacement can create two kinds of records: a record for the old policy and one for the new policy. If the second policy supersedes the first for the same exposure unit, the old interval should end when the new coverage takes effect. If both policies genuinely provide concurrent coverage and the exposure base counts both contracts, that is a different situation; do not apply a blanket rule that all overlap must be removed. Establish whether the data represent duplicate records or distinct, simultaneously active coverage before summing.

Situation	Interval treatment	Calculation check
Cancellation without reinstatement	End the in-force interval on the cancellation effective date.	Covered time ends before the remaining scheduled term.
Cancellation followed by reinstatement	Use separate intervals if a genuine uncovered gap exists.	Covered time plus the gap and any other uncovered time equals the period reviewed.
Replacement with continuous coverage	Join the old and new intervals at the effective replacement date when they represent one continuous risk.	There is neither a gap nor a duplicated period for that exposure unit.
Concurrent active coverage	Determine whether the exposure base counts one risk or multiple covered contracts.	Do not infer the right count from overlapping dates alone.

A useful exam check is to compare total covered time with the maximum possible time in the period. For one exposure unit in a 365-day period, a result above 365 days signals an overlap, an inconsistent date convention, or an incorrect duration. It is a diagnostic, not a universal rule: multiple insured units or a base that counts separate concurrent contracts can legitimately produce more exposure than one unit-year.

The most common traps are counting the cancellation effective date as a full covered day when the stated convention excludes it, treating reinstatement as if it erased a lapse, and relying on a scheduled expiration that was superseded by an actual cancellation. Avoid these by writing the intervals before doing arithmetic, stating the denominator, and reconciling covered and uncovered time to the total period.

Key takeaway: Earned time follows the actual periods coverage is in force. Split the timeline at each effective cancellation, reinstatement, or replacement; add only the durations appropriate to the exposure base; and use one clearly stated time-unit convention throughout.

Check Your Understanding

For each problem, identify the actual in-force time before converting it to exposure. Use the denominator stated in the problem.

A hypothetical annual policy in 2025 runs from March 1 to November 1, then cancels effective November 1. What is its exposure as a fraction of a 365-day year? (Answer: \(245/365=0.6712\) policy-years, to four decimal places.)
A leap-year policy runs from January 1, 2024, to January 1, 2025. It cancels effective March 1 and is reinstated effective April 1, with no coverage during the gap. What is its actual-term exposure? (Answer: \(335/366=0.9153\) policy-years, to four decimal places.)
A hypothetical policy runs from April 1, 2025, to April 1, 2026. The old policy ends and its replacement begins on August 1, with no gap or overlap. Using the 365-day term as denominator, what is the exposure? (Answer: \((122+243)/365=1.0000\) policy-year.)
For one risk, two records describe intervals January 1 to March 1 and February 15 to April 1, 2025. Assume these dates are duplicate records and exposure should count the covered timeline only once. What is exposure on a 365-day basis? (Answer: \(90/365=0.2466\) policy-years, to four decimal places.)
A 2025–2026 annual policy is in force February 1 to May 10, then reinstated May 20 and cancelled effective December 31. Using a 365-day denominator, calculate exposure. (Answer: \((98+225)/365=0.8849\) policy-years, to four decimal places.)

What You'll Learn