Coverage Periods And Earned Exposure

Why Coverage Dates Matter

A policy may be in force for only part of a calendar year, or it may begin before that year and expire after it. To analyze a rating interval correctly, count only the portion of the policy’s coverage that falls inside the interval. That portion is its earned exposure for the interval. It is not necessarily the policy’s full term, nor is it necessarily a whole number of policy-years.

The key operation is an intersection of date periods. First identify the policy’s coverage start and end dates; then identify the rating interval’s start and end dates. The overlap is the time covered by both. Only after finding that overlap should it be converted to an exposure amount using a stated day-count and annualization convention.

Definition: Represent a policy period as the half-open interval \([s,e)\), where \(s\) is the coverage start date and \(e\) is the coverage end date. Represent a rating interval as \([a,b)\). Half-open means the start date is included and the end date is excluded. The overlap begins at the later start date, \(\max(s,a)\), and ends at the earlier end date, \(\min(e,b)\). Its duration is zero if the end is not later than the start.

In this tutorial, dates are measured as elapsed calendar days: the number of days from a start date up to, but not including, an end date. For example, the interval \([{\rm Jan\ 1},{\rm Feb\ 1})\) contains 31 days. This convention prevents a shared boundary between adjacent intervals from being counted twice.

To express exposure in policy-years, this tutorial uses a fixed annualization basis of 365 days. Thus, 365 covered days correspond to 1.0000 policy-years. A different analysis may specify another basis, such as a leap-year adjustment; the important requirement is to state the basis and apply it consistently. The numerator is the covered overlap, not the total policy term.

L_{\text{overlap}}=\max\!\left(0,\ \min(e,b)-\max(s,a)\right), \qquad \text{earned exposure}=\frac{L_{\text{overlap}}}{365}.

Here the subtraction of dates denotes elapsed days. If the coverage and rating intervals do not overlap, the later start is on or after the earlier end, so the overlap duration is zero. If they do overlap, the formula counts from the later of the two starts to the earlier of the two ends. This is why the formula works: those are exactly the dates that belong to both intervals, and no dates outside either interval are included.

A Reliable Calculation Method

A useful method is to write both intervals explicitly before counting days. Determine the overlap start and overlap end separately, check that the end is later than the start, and then count the days between those two dates. Convert to policy-years only after that count is settled.

Write both intervals.
Record policy coverage as \([s,e)\) and the rating period as \([a,b)\), using the same date convention for each.

Find the overlap boundaries.
The overlap starts at \(\max(s,a)\) and ends at \(\min(e,b)\). If the end is not later than the start, there is no covered overlap.

Count elapsed days.
Count from the overlap start up to, but not including, the overlap end. For a span crossing months, add the applicable days month by month or use a verified date-difference calculation.

Apply the annualization basis.
Divide overlap days by the stated number of days per exposure-year. Do not substitute the rating interval’s length or the full policy term unless that is the expressly defined basis.

Check the result.
The overlap cannot exceed either interval. For adjacent rating periods, the sum of the period overlaps should equal the overlap across their combined interval, provided the periods cover that combined interval without gaps.

Worked Example: A Policy That Begins Before the Calendar Year

A hypothetical annual policy covers \([{\rm Nov\ 1,\ 2025},{\rm Nov\ 1,\ 2026})\). Calculate its earned exposure in calendar year 2026, represented by \([{\rm Jan\ 1,\ 2026},{\rm Jan\ 1,\ 2027})\). Use a 365-day annualization basis and report exposure to four decimal places.

Method. Intersect the policy period with the calendar-year interval. The later start is Jan 1, 2026; the earlier end is Nov 1, 2026. Therefore the overlap is \([{\rm Jan\ 1,\ 2026},{\rm Nov\ 1,\ 2026})\).

Count the overlap days. From Jan 1 up to Nov 1, 2026, the elapsed days are \(31+28+31+30+31+30+31+31+30+31=304\). February has 28 days in 2026.

Convert to exposure. The earned exposure is \(304/365=0.8328767\ldots\), which rounds to 0.8329 policy-years to four decimal places.

Check and common mistake avoided. The policy also has 61 days from Nov 1, 2025 to Jan 1, 2026, but those days are outside calendar year 2026. Adding those 61 days would count the pre-year portion instead of the overlap. The 304-day overlap is less than both the policy term and the 2026 rating interval, as required.

Worked Example: A Policy That Expires After the Rating Interval

A hypothetical policy covers \([{\rm Dec\ 15,\ 2025},{\rm Jun\ 15,\ 2026})\). Find its earned exposure in the rating interval \([{\rm Oct\ 1,\ 2025},{\rm Apr\ 1,\ 2026})\). Use a 365-day basis and report to four decimal places.

Method. The later start is Dec 15, 2025, and the earlier end is Apr 1, 2026. The overlap is therefore \([{\rm Dec\ 15,\ 2025},{\rm Apr\ 1,\ 2026})\). The policy’s coverage after Apr 1 is not in this rating interval.

Count the overlap days. Dec 15 to Jan 1 is 17 days. January contributes 31 days, February 28 days, and March 31 days. The total is \(17+31+28+31=107\) days.

Convert to exposure. The earned exposure is \(107/365=0.2931507\ldots\), or 0.2932 policy-years to four decimal places.

Check and common mistake avoided. The end date Apr 1 is excluded, so the overlap includes all of March but no day beginning Apr 1. Counting only 16 days from Dec 15 to Jan 1 would omit Dec 31; counting the policy through Jun 15 would include days outside the rating interval. The correct overlap is 107 days.

Worked Example: Splitting Exposure Across Adjacent Rating Periods

A hypothetical policy covers \([{\rm Nov\ 1,\ 2025},{\rm Feb\ 1,\ 2026})\). Calculate its earned exposure separately in calendar quarter 4 of 2025, \([{\rm Oct\ 1,\ 2025},{\rm Jan\ 1,\ 2026})\), and calendar quarter 1 of 2026, \([{\rm Jan\ 1,\ 2026},{\rm Apr\ 1,\ 2026})\). Use a 365-day basis and report each result to four decimal places.

Quarter 4 overlap. The overlap is \([{\rm Nov\ 1,\ 2025},{\rm Jan\ 1,\ 2026})\). November contributes 30 days from Nov 1 to Dec 1, and December contributes 31 days from Dec 1 to Jan 1, for \(30+31=61\) days. Its exposure is \(61/365=0.1671233\ldots\), or 0.1671 policy-years.

Quarter 1 overlap. The overlap is \([{\rm Jan\ 1,\ 2026},{\rm Feb\ 1,\ 2026})\), which is 31 days. Its exposure is \(31/365=0.0849315\ldots\), or 0.0849 policy-years.

Reconciliation. The two adjacent rating intervals together cover \([{\rm Oct\ 1,\ 2025},{\rm Apr\ 1,\ 2026})\). Within that combined interval, the policy is covered from Nov 1 to Feb 1: \(61+31=92\) days. The total exposure is \(92/365=0.2520548\ldots\), or 0.2521 policy-years. The unrounded quarterly exposures also sum to \(92/365\).

Check and common mistake avoided. Jan 1 is the end of quarter 4 and the start of quarter 1. Under the half-open convention it belongs to the second interval, not both. Counting Jan 1 in both quarters would overstate the total by one day; omitting it would understate the total.

Day Counts, Annualization, and Checks

The date intersection and the exposure conversion are separate decisions. The intersection determines which days are covered within the rating interval. The day-count basis determines how those days are expressed as exposure. Keeping these decisions separate makes assumptions visible and helps identify errors.

For example, under the fixed 365-day basis used here, a full calendar year that contains 366 days would correspond to \(366/365\) policy-years, not exactly 1.0000. That is a consequence of the selected basis, not an error in the overlap calculation. If a task instead defines exposure using an actual-days-per-calendar-year convention, use the specified denominator for that period. Never silently change the denominator midway through a calculation.

A few checks are especially useful:

Boundary check: the overlap starts at the later start date and ends at the earlier end date.
Length check: overlap days cannot be negative and cannot exceed the length of either interval.
Unit check: divide days by days per exposure-year to obtain policy-years; do not report the day count itself as policy-years.
Partition check: when adjacent rating intervals partition a larger interval, their unrounded overlap days should add to the overlap in the larger interval.
Rounding check: calculate with unrounded day counts and divide once; round only the final exposure to the requested precision.

Earned exposure is a measure of how much coverage time falls within the period being analyzed. Accurate dates matter because an error of one day changes exposure by \(1/365\) policy-years under the convention here. That can affect rates calculated per unit of exposure, especially when the portfolio or rating cell has relatively little exposure.

Common Mistakes

Using the wrong end of the policy. A policy that extends beyond the rating interval earns exposure only through the rating interval’s end.
Using the wrong start of the rating interval. For a policy that began earlier, count from the rating interval’s start, not from the policy’s inception.
Including the end date. With \([s,e)\), the end date is excluded. Adjacent periods share a boundary, not a day.
Confusing overlap days with exposure. A 107-day overlap is not 107 policy-years; on a 365-day basis it is \(107/365\) policy-years.
Changing the day-count basis without saying so. State the basis first, then use it consistently for every policy in the calculation.

Key takeaway: Earned exposure for a rating interval is the duration of the intersection between the policy coverage period and the rating interval, divided by the stated annualization basis. Find and verify the overlap before converting it to policy-years.

Check Your Understanding

For each problem, use half-open intervals and a fixed 365-day annualization basis. Round final exposures to four decimal places.

A hypothetical policy covers \([{\rm Nov\ 1,\ 2025},{\rm Nov\ 1,\ 2026})\). Find its earned exposure in \([{\rm Jan\ 1,\ 2026},{\rm Jan\ 1,\ 2027})\). State the overlap days and exposure. (Answer: 304 days; \(304/365=0.8329\) policy-years.)
A hypothetical policy covers \([{\rm Dec\ 15,\ 2025},{\rm Jun\ 15,\ 2026})\). Find its earned exposure in \([{\rm Oct\ 1,\ 2025},{\rm Apr\ 1,\ 2026})\). Show the month-by-month day count. (Answer: \(17+31+28+31=107\) days; \(107/365=0.2932\) policy-years.)
A hypothetical policy covers \([{\rm Sep\ 15,\ 2025},{\rm Jun\ 15,\ 2026})\). Find its earned exposure in \([{\rm Oct\ 1,\ 2025},{\rm Apr\ 1,\ 2026})\). (Answer: 182 days; \(182/365=0.4986\) policy-years.)
A hypothetical policy covers \([{\rm Nov\ 1,\ 2025},{\rm Feb\ 1,\ 2026})\). Calculate exposure in \([{\rm Oct\ 1,\ 2025},{\rm Jan\ 1,\ 2026})\) and \([{\rm Jan\ 1,\ 2026},{\rm Apr\ 1,\ 2026})\), then check the combined day count. (Answer: 61 days and 31 days, respectively; exposures 0.1671 and 0.0849; combined 92 days, or \(92/365=0.2521\) policy-years.)
A hypothetical policy covers \([{\rm May\ 1,\ 2024},{\rm May\ 1,\ 2025})\). Find its earned exposure in the 2024 calendar interval \([{\rm Jan\ 1,\ 2024},{\rm Jan\ 1,\ 2025})\) under the fixed 365-day basis. Explain why the interval’s leap day does not change the denominator in this problem. (Answer: 245 overlap days; \(245/365=0.6712\) policy-years. The stated convention fixes the denominator at 365.)

What You'll Learn