Why Partial-Period Exposure Requires Care
An expected loss calculation must reflect the amount of exposure actually covered during the period being priced or reviewed. If a policy begins after the start of a year, cancels before its expiration, or changes coverage midterm, using a full-year exposure can overstate expected losses. But simply multiplying every annual expected loss by a fraction of a year can also be wrong: the appropriate exposure measure may be payroll, sales, or another quantity that is available directly rather than estimated from elapsed time.
The calculation has two parts: identify the exposure measure used to develop the expected loss rate, then determine how much of that exposure applies during the covered interval. A time fraction is appropriate when exposure accumulates proportionally with time under the stated assumptions. Where actual earned exposure is known, use that amount instead.
Choose the Exposure Basis Before Prorating
An exposure unit is the quantity to which the expected loss rate applies. For example, a rate may be stated per vehicle-year or per $1,000 of annual payroll. These units are not interchangeable. A vehicle-year can often be prorated by covered days if the vehicle is covered throughout those days. Payroll exposure is generally measured by earned payroll; a time fraction is a reasonable estimate only when the problem states or supports an assumption about how payroll accumulates.
The time convention must also be explicit. Under an actual-days-over-365 convention, a covered interval of \(d\) days has time fraction \(d/365\), even if it crosses months of different lengths. A monthly convention, such as counting six covered months as \(6/12\) of a year, is a different assumption and can produce a different result. Use the convention specified by the problem; do not silently substitute one for another.
Why the Time-Proration Formula Holds
Suppose the expected loss rate is \(r\) dollars per exposure unit-year, and assume the expected loss accrues uniformly over the year for each continuously covered unit. The expected loss per unit-day is then \(r/365\). Over \(d\) covered days, the expected loss for one unit is \(d\) times that daily amount. For \(n\) units, linearity of expectation gives:
Thus the fraction \(d/365\) is not an arbitrary discount. It represents the covered share of the annual exposure under the uniform-accrual assumption. The formula does not apply unchanged if expected losses are not proportional to time, if coverage or the applicable rate changes during the interval, or if the selected exposure measure varies in a way that is not captured by elapsed days. In those situations, calculate the exposure by segment or use the directly measured exposure quantity.
Check that the rate is expressed per the exposure unit being counted, such as per vehicle-year or per $1,000 of payroll.
Identify the effective and ending dates, and use the stated day-count or monthly convention consistently.
Prorate by time only when the assumptions support it. Otherwise use earned exposure directly.
Multiply the applicable rate by the exposure. If rates or coverage differ across the term, calculate each segment separately and add the results.
Worked Calculations
Worked Example: Cancellation Before the End of a Policy Year
A hypothetical fleet policy covers two identical vehicles from January 1 through March 31 in a non-leap year, then cancels effective April 1. The expected loss rate is $1,460 per vehicle-year. Use actual days over 365. Calculate expected losses for the covered interval.
Method and assumptions. The interval includes January, February, and March, with cancellation effective on April 1. Because this is a non-leap year, the covered days are \(31+28+31=90\). Assume the rate applies uniformly throughout the covered period and both vehicles are covered for all 90 days.
Step 1: Calculate the time fraction.
Step 2: Calculate vehicle-year exposure.
Step 3: Apply the annual expected loss rate.
Check. The annual rate is $1,460 per vehicle, or \(1{,}460/365=\$4\) per vehicle-day. Two vehicles covered for 90 days produce \(2\times90\times\$4=\$720.00\), agreeing with the exposure calculation. The common mistake is to count the full annual loss for both vehicles, which would give $2,920 instead of accounting for the cancellation.
Worked Example: Midterm Inception with Payroll Exposure
A hypothetical business policy begins July 1 and covers the business through December 31 in a non-leap year. The annual payroll projection is $1,200,000, assumed to accrue uniformly by calendar day over 365 days. The expected loss rate is $18 per $1,000 of annual payroll exposure. Using actual days over 365, calculate the expected losses for the covered term.
Method and assumptions. July 1 through December 31 includes \(31+31+30+31+30+31=184\) days. The payroll projection is an annual amount, not payroll already measured for the covered term, so the stated uniform-accrual assumption permits time-based proration. The annual expected losses corresponding to the full-year payroll projection are:
Step 1: Prorate the annual payroll exposure.
This is \(604.931506849\ldots\) units of $1,000 of payroll exposure.
Step 2: Apply the rate.
Rounding only at the final step to the nearest cent, expected losses are $10,888.77.
Check. Prorating the full-year expected losses gives \(\$21{,}600\times184/365=\$10{,}888.767123\ldots\), the same result. Using six-twelfths would instead give $10,800, which is not the stated actual-days convention: the six calendar months in this interval contain 184 days, not half of 365. The common mistake is to replace the specified day count with a convenient month fraction.
Worked Example: Coverage Rate Changes During the Term
A hypothetical one-vehicle policy runs from January 1 through December 31 in a non-leap year. Coverage changes effective July 1. The annual expected loss rate is $730 per vehicle-year from January 1 through June 30 and $1,095 per vehicle-year from July 1 through December 31. Use actual days over 365. Calculate total expected losses.
Method and assumptions. The policy has two segments with different applicable annual rates. January through June contains \(31+28+31+30+31+30=181\) days; July through December contains \(31+31+30+31+30+31=184\) days. Assume each segment’s stated rate applies uniformly within that segment.
Step 1: Calculate the first segment.
Step 2: Calculate the second segment.
Step 3: Add segment expected losses.
Check. The day counts reconcile to \(181+184=365\). If the first rate had applied for the entire year, expected losses would be $730; the increase for the last segment is \((\$1{,}095-\$730)\times184/365=\$184\). Thus $730+$184=$914. The common mistake is to apply one annual rate to the whole term or to prorate the full-year rate first and then apply a second time fraction to one segment.
Practical Checks and Common Traps
Partial-period calculations are often simple arithmetic, but they depend on keeping the rate, exposure unit, and time period aligned. A reliable calculation should make all three visible. In particular, a rate per vehicle-year must be multiplied by vehicle-years, while a rate per $1,000 of payroll must be multiplied by payroll expressed in thousands of dollars.
- Counting dates inconsistently. State whether the effective date is covered and how the ending date is treated. A policy effective January 1 and cancelled effective April 1 covers January 1 through March 31, not April 1.
- Using the wrong time convention. Actual/365, actual/actual, and monthly proration are distinct conventions. Use the one stated or justified for the calculation.
- Prorating an exposure that is already partial. If the problem gives actual payroll earned during the covered term, multiply that payroll exposure by its matching rate. Do not multiply it by a time fraction again.
- Ignoring changes during the term. When a rate, exposure basis, or coverage changes, separate the policy into intervals over which the assumptions are consistent, calculate each interval, and sum.
- Treating time proportionality as automatic. Day-based proration assumes expected loss accrues evenly with time. A directly measured exposure, such as payroll, can be more appropriate when it is available and the rate is based on that exposure.
Expected losses are benchmarks tied to a specified exposure basis and period. Partial-period adjustments preserve that match; they do not change the underlying rate. These calculations are aligned with the CAS Exam 8 (Advanced Ratemaking) syllabus.
Check Your Understanding
For each problem, identify the exposure basis and apply only the proration supported by the stated assumptions. Round dollar amounts to the nearest cent where necessary.
- A hypothetical one-unit policy has an annual expected loss rate of $1,095 per unit-year and covers May 1 through July 31 in a non-leap year. Using actual/365 and uniform accrual, calculate expected losses. (Answer: $276.00)
- A hypothetical fleet has three tractors covered from September 1 through November 30 in a non-leap year. The annual expected loss rate is $730 per tractor-year. Using actual/365 and uniform accrual, calculate expected losses. (Answer: $546.00)
- A hypothetical policy has annual payroll of $900,000, assumed to accrue uniformly by calendar day over 365 days. Coverage runs from October 1 through December 31 in a non-leap year, and the rate is $24 per $1,000 of annual payroll exposure. Using actual/365, calculate expected losses. (Answer: $5,444.38)
- A hypothetical one-unit policy runs for the full non-leap year. Its annual rate is $1,460 from January 1 through September 30 and $2,190 from October 1 through December 31. Assume uniform accrual within each segment and use actual/365. Calculate total expected losses. (Answer: $1,644.00)
- A hypothetical policy covers a business for four months. The rate is $16 per $1,000 of payroll exposure, and actual payroll earned during those four covered months is $480,000. Calculate expected losses without assuming uniform payroll accrual. (Answer: $7,680.00)