Choosing the Prospective Cost-Level Endpoint

Why the Prospective Endpoint Matters

A trend calculation needs two dates: the date representing the historical cost level and the date representing the cost level to which it is projected. The second date is the prospective cost-level endpoint. It is chosen to reflect when the rates being developed are intended to apply—not automatically the start of the rating analysis, the date of the latest data, or the end of the prospective period.

For a block of policies effective over a period, costs do not all arise on one date. A common convention is to represent the block by the midpoint of its policy effective period, assuming policy effective dates are approximately uniform. The convention makes the calculation manageable, but it is still an assumption. If the timing of the relevant exposures is known and is materially different from uniform, the appropriate endpoint should reflect that timing instead.

Definition: The prospective cost-level endpoint is the date representing the cost level sought for the prospective application period. Under a uniform-distribution midpoint convention, it is the midpoint of the period during which the rates are intended to apply. The trend interval is the elapsed time from the date representing the historical cost level to that endpoint.

The date representing historical costs must be specified just as carefully. For example, if losses are summarized for a calendar year and are treated as arising uniformly through that year, the year midpoint represents their average cost date. The trend interval then runs from that historical midpoint to the prospective endpoint. The earlier material on mapping policy effective dates to earned experience helps identify which period the data represent; choosing the endpoint requires a separate decision about when the proposed rates are intended to apply.

From Dates to a Trend Interval

Let $t_0$ be the date representing the historical cost level and $t_1$ the prospective endpoint. Express their difference in years, using a stated and consistent time convention. If the dates are exactly a specified number of months apart, convert months to years by dividing by 12. For dates that do not align to whole months or anniversaries, a day-count convention may be needed; the analyst should state it rather than silently treating every calendar year as the same number of days.

Once the interval is established, a constant annual trend assumption can be applied. If $C(t_0)$ is the historical cost level and the effective annual trend rate is $g$, then, under a constant compounded trend assumption,

C(t_1)=C(t_0)(1+g)^{t_1-t_0}

Here, $t_1-t_0$ is measured in years. A fractional exponent assumes the same constant annual compounding pattern applies within a year. The formula is not appropriate as written if the selected trend changes over the interval, if the time unit used for $g$ differs from the exponent’s unit, or if the selected endpoint does not represent the prospective costs being analyzed.

Justification: For one full year, a cost level grows by the factor $1+g$. Under a constant compounded annual trend, after $n$ full years the repeated factor is $(1+g)^n$. Extending that same constant compounding assumption to a fractional year $s$ gives the factor $(1+g)^s$. Therefore, over $n+s$ years the factor is $(1+g)^{n+s}$, which yields the stated projection formula.

Worked Applications

Worked Example: A Calendar-Year Experience Period and a Later Rate Period

A hypothetical insurer is reviewing a loss cost of $500 from calendar-year 2024 experience. The proposed rates are intended for policies effective throughout calendar year 2026. Assume that each year’s costs are represented by its midpoint and that annual trend is a constant 6%. Find the trend interval and projected cost level. Name the method and the common mistake it avoids.

Method: midpoint-to-midpoint projection with annual compounding. Calendar-year 2024 is represented by July 1, 2024. Calendar-year 2026 is represented by July 1, 2026. The historical and prospective dates are therefore two years apart:

t_1-t_0 =\text{July 1, 2026}-\text{July 1, 2024} =2\text{ years}

The two-year trend factor is:

$$ (1.06)^2=1.1236 $$

Applying it to the historical cost level gives:

\$500(1.1236)=\$561.80

The projected cost level is $561.80. As a check, one year of trend takes $500 to $530, and another year takes $530 to $561.80. The common mistake this method avoids is using the beginning or end of calendar year 2026 without considering that the intended rates apply across the whole year. Under the stated midpoint convention, July 1, not January 1 or December 31, represents that application period.

Worked Example: A Fractional-Year Interval Between Midpoints

A hypothetical coverage program has experience from April 1, 2024 through March 31, 2025, represented by its midpoint, October 1, 2024. Proposed rates will apply to policies effective from October 1, 2025 through September 30, 2026, represented by April 1, 2026. The historical cost level is $825, and the assumed constant annual trend is 4%. Calculate the projected cost level. Name the method and the common mistake it avoids.

Method: fractional-year midpoint-to-midpoint projection. From October 1, 2024 to April 1, 2026 is 18 months, or $18/12=1.5$ years. The stated dates—not the number of calendar-year labels crossed—determine the interval.

t_1-t_0=1.5\text{ years}

The trend factor is:

(1.04)^{1.5} =1.04\sqrt{1.04} \approx 1.060596

Using the unrounded factor for the multiplication:

\$825(1.04)^{1.5} \approx \$874.9917 \approx \$874.99

The projected cost level is $874.99, rounded to the nearest cent. The common mistake this method avoids is counting the interval as two years merely because it touches parts of two calendar years. The specified midpoints are 18 months apart, not 24 months apart.

Worked Example: A Three-and-a-Half-Year Projection

A hypothetical insurer uses a historical cost level of $500 represented by January 1, 2022. The intended application period is July 1, 2025 through June 30, 2026, represented by July 1, 2025. Assume a constant effective annual trend of 8%. Determine the interval and prospective cost level. Name the method and the common mistake it avoids.

Method: midpoint-to-midpoint projection with fractional-year compounding. January 1, 2022 to January 1, 2025 is three years, and January 1 to July 1 adds six months. Thus the interval is $3.5$ years.

t_1-t_0=3.5\text{ years}

The trend factor can be written as:

(1.08)^{3.5} =(1.08)^3(1.08)^{0.5} =1.259712\sqrt{1.08} \approx 1.309131

Applying the factor gives:

\$500(1.08)^{3.5} \approx \$500(1.309131112) =\$654.565556 \approx \$654.57

The prospective cost level is $654.57, rounded to the nearest cent. The common mistake this method avoids is projecting to the start of the application period, July 1, 2025 in this example, but then treating that date as though it were the historical date or extending the interval by an extra half-year. The historical date is January 1, 2022, and the selected prospective endpoint is July 1, 2025, so the elapsed time is 3.5 years.

A Practical Endpoint-Selection Sequence

A consistent sequence helps separate the choice of endpoint from the arithmetic. The endpoint should be documented before calculating a trend factor; otherwise, a precise calculation can still project to the wrong cost level.

Identify the intended application period.
State the dates during which the proposed rates are intended to apply. Do not substitute the date the analysis is performed.

Select a date representing that period.
Use the stated prospective cost-level date when one is specified. If using the midpoint convention, identify the period midpoint and the assumption supporting it.

Identify the historical cost date.
Use a date consistent with the period and timing assumptions underlying the historical cost level.

Calculate the elapsed time and apply trend.
Express the interval in years, use the trend rate’s matching time unit, and check that the projected date is the endpoint selected in Step 2.

The midpoint convention is a practical representation, not a claim that every policy or claim occurs at the midpoint. If effective dates are strongly concentrated near one end of the application period, a simple midpoint may not represent the average exposure timing well. Likewise, a stated prospective cost date should not be replaced by a midpoint merely because that date is convenient. The date must be chosen to match the intended cost level and the assumptions used in the analysis.

Common Endpoint and Time-Unit Traps

Using the data cutoff as the endpoint. The latest date in the experience data is not automatically the date to which costs should be projected. The target comes from the intended application period.
Using a period boundary without a reason. Projecting to the beginning of a full prospective period can understate the cost level represented by that period when a midpoint convention is intended.
Counting year labels instead of elapsed time. A period from October in one year to April two years later spans 18 months, even though the dates include three calendar-year labels.
Mixing months and years. An 18-month interval is $1.5$ years. If $g$ is an annual trend rate, an exponent of 18 would use the wrong time unit.
Rounding too early. Retain precision in the fractional-year factor and round the final monetary result as required.

This distinction matters because endpoint selection is a modeling decision, while exponentiation is a calculation. A correct formula applied to the wrong endpoint does not produce the intended prospective cost level. Clear dates also make the chosen trend assumption easier to interpret and review. These endpoint and interval choices are aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.

Check Your Understanding

For each problem, state the representative dates, find the elapsed interval, and calculate the projected cost level where requested. Assume constant annual compounding unless the problem says otherwise.

A hypothetical insurer represents experience by July 1, 2025 and a full prospective calendar year by July 1, 2026. At 6% annual trend, a cost level of $700 is projected. Find the interval, factor to six decimal places, and projected cost to the nearest cent. (Answer: 1 year; factor 1.060000; $742.00)
A hypothetical program represents historical experience by October 1, 2023 and its prospective period by April 1, 2026. The historical cost level is $900 and annual trend is 4%. Find the interval and projected cost to the nearest cent. (Answer: 2.5 years; factor approximately 1.103020; $992.72)
A hypothetical insurer represents the historical cost level by February 1, 2024 and the target period by February 1, 2026. If the historical cost level is $1,500 and annual trend is 3%, calculate the projected level. (Answer: 2 years; factor 1.060900; $1,591.35)
A hypothetical coverage block has a historical representative date of July 1, 2024. The intended prospective period is January 1 through December 31, 2027, and the analysis date is January 1, 2027. Under the midpoint convention, which date is the endpoint? At 5% annual trend, project a historical cost level of $1,100. (Answer: Endpoint July 1, 2027; interval 3 years; factor 1.157625; $1,273.39)
A hypothetical insurer has a historical representative date of January 1, 2022 and targets a prospective period represented by July 1, 2025. An analyst calls the interval four years because the dates touch four calendar-year labels. At 8% annual trend, identify the correct interval and project a historical cost level of $500. (Answer: 3.5 years, not 4 years; factor approximately 1.309131; $654.57)

What You'll Learn