Why Losses and Exposures Must Share a Time Basis
A loss cost is a ratio: losses divided by exposure units. When historical amounts are projected, both sides of that ratio must represent the same target period if the result is to be interpreted as a projected loss cost for that period. Trending losses four years but exposures three years may produce a number, but it does not represent a consistent four-year or three-year projection.
Loss and exposure trends need not be equal. A selected loss trend may reflect changes in claim costs, while an exposure trend may reflect changes in the measured amount of insured risk. The essential consistency requirement is about the dates and durations to which the factors are applied—not that the two trend rates match. This tutorial is aligned with the CAS Exam 5 (Basic Ratemaking and Estimated Claim Liabilities) syllabus.
Projecting Losses, Exposures, and Loss Cost
Let \(L_h\) be historical losses and \(E_h\) be historical exposure units. Suppose the selected annual effective trend rates are \(t_L\) for losses and \(t_E\) for exposures. Under a constant compound-trend assumption, an amount is multiplied by its annual trend factor for each year in the trend period. For an elapsed period of \(n\) years, including a fractional number of years when appropriate:
Here \(L_p\), \(E_p\), and \(LC_p\) are projected losses, projected exposures, and projected loss cost. This compound model assumes that the selected annual rates apply uniformly over the period and that the measures remain sufficiently comparable for the trend factors to be meaningful. It does not, by itself, account for changes in coverage, limits, deductible structure, or the composition of the insured population.
When losses and exposures have the same historical date and the same target date, they have the same elapsed period \(n\). Dividing the projected losses by the projected exposures gives:
This identity shows why simply applying the loss trend to historical loss cost can be wrong: the loss cost also reflects any trend in the exposure denominator. If the exposure trend is zero, its factor is 1 and the formula reduces to trending the historical loss cost by the loss factor. If the loss and exposure trend rates are equal, the factors cancel and the loss cost is unchanged, even though projected losses and projected exposures may both increase.
Work From Dates to Factors
A reliable calculation starts by identifying what date each historical amount represents and the common date to which it is being projected. Then calculate the elapsed time for each amount, apply its appropriate factor, and form the ratio only after both amounts have been projected. If different groups start at different dates, each group can have a different elapsed period even though all are projected to the same target date.
State the date the projected losses and exposures are intended to represent.
Measure from each historical amount’s represented date to the target date. Do not infer the period from a label such as “year 1” without checking the dates.
Use the loss trend for losses and the exposure trend for exposure units, each over its elapsed period.
For multiple groups, sum the projected losses and sum the projected exposures. Divide those totals to obtain the combined projected loss cost.
Worked Applications
Worked Example: Combining Groups With Different Starting Dates
A hypothetical specialty insurer has two groups of policies. Group A has historical losses of $40,000 and 800 exposure units, measured two years before a common projection date. Group B has historical losses of $60,000 and 1,200 exposure units, measured one year before that date. The selected annual effective trends are 3% for losses and 2% for exposures. Find the combined projected loss cost, to the nearest cent per exposure unit.
This is a deterministic trend calculation; no interest discounting, mortality assumption, or probability distribution is used. Apply the same selected rates to both groups, but use each group’s own elapsed period.
Group A losses: The loss trend factor is \(1.03^2=1.0609\), so projected losses are \(40{,}000(1.0609)=\$42{,}436\).
Group A exposures: The exposure trend factor is \(1.02^2=1.0404\), so projected exposures are \(800(1.0404)=832.32\) units.
Group B losses: The loss trend factor is \(1.03^1=1.03\), so projected losses are \(60{,}000(1.03)=\$61{,}800\).
Group B exposures: The exposure trend factor is \(1.02^1=1.02\), so projected exposures are \(1{,}200(1.02)=1{,}224\) units.
Combine the projected amounts: Total projected losses are \(\$42{,}436+\$61{,}800=\$104{,}236\). Total projected exposures are \(832.32+1{,}224=2{,}056.32\) units.
Calculate and check the loss cost: \[ \frac{\$104{,}236}{2{,}056.32}=\$50.69055\ldots \] Rounding only the final result to the nearest cent gives $50.69 per exposure unit. As a check, \(\$50.69(2{,}056.32)=\$104{,}234.86\), which is within the difference expected from rounding the quotient to cents.
Method and trap avoided: Project each group to the common date, then divide the sums. Averaging the two group loss costs without weighting by projected exposures would not generally give the combined loss cost.
Worked Example: Recognizing a Mismatched Projection Period
A hypothetical portfolio has historical losses of $75,000 and 1,500 exposure units at the same historical midpoint. The instructions call for projecting losses four years but exposures only three years. The selected annual effective trends are 5% for losses and 2% for exposures. Calculate the ratio produced by those instructions, then explain why it is not a consistent four-year projected loss cost.
This is a deterministic calculation; no interest discounting, mortality assumption, or probability distribution is used. Because both historical amounts have the same starting date, projecting them to a common four-year endpoint would require four years of trend on both.
Follow the stated instructions literally: The four-year loss factor is \(1.05^4=1.21550625\). Projected losses are \(\$75{,}000(1.21550625)=\$91{,}162.96875\).
The three-year exposure factor is \(1.02^3=1.061208\). Projected exposures are \(1{,}500(1.061208)=1{,}591.812\) units.
The resulting ratio is: \[ \frac{\$91{,}162.96875}{1{,}591.812}=\$57.2699\ldots \] So the ratio, rounded to the nearest cent, is $57.27 per exposure unit.
Why the ratio is inconsistent: Its numerator represents losses four years after the historical midpoint, while its denominator represents exposures only three years after that midpoint. The ratio therefore does not compare amounts for a single target date. If the intended target is four years after the midpoint, project exposures for four years as well. The consistent four-year loss cost is: \[ \frac{\$75{,}000(1.05)^4}{1{,}500(1.02)^4} =50\left(\frac{1.05}{1.02}\right)^4 =\$56.14699\ldots \] or $56.15 per exposure unit, rounded to the nearest cent.
Method and trap avoided: The literal instructions produce $57.27, but that value must not be described as a four-year projected loss cost. The correction is to align both projection periods with the intended common endpoint.
Worked Example: Using a Fractional Trend Period
A hypothetical equipment-insurance portfolio has losses of $240,000 and 4,000 exposure units at a historical midpoint. The desired projection date is 18 months later. Assume constant annual effective trends of 4% for losses and 1% for exposures. Calculate the projected loss cost to the nearest cent.
This is a deterministic projection under a constant compound-trend assumption; no interest discounting, mortality assumption, or probability distribution is used. Eighteen months is \(18/12=1.5\) years, so both measures must be projected for 1.5 years.
Project losses: The factor is \((1.04)^{1.5}\), approximately \(1.060596058\). Thus projected losses are \(\$240{,}000(1.04)^{1.5}\), approximately \(\$254{,}543.054\).
Project exposures: The factor is \((1.01)^{1.5}\), approximately \(1.015037438\). Thus projected exposures are \(4{,}000(1.01)^{1.5}\), approximately \(4{,}060.149752\) units.
Divide projected amounts: \[ \frac{\$240{,}000(1.04)^{1.5}}{4{,}000(1.01)^{1.5}} =60\left(\frac{1.04}{1.01}\right)^{1.5} =\$62.6929\ldots \] The projected loss cost is $62.69 per exposure unit, rounded to the nearest cent.
Check: The projected losses divided by the projected exposures give the same result, approximately \(\$254{,}543.054/4{,}060.149752=\$62.69\). The 1.5-year factor is appropriate here because the model assumes a smooth, constant compound trend; it should not be used automatically if the selected trend varies materially during the period.
Common Errors and Practical Checks
The order of operations matters when combining groups. Projecting each group first and then adding preserves differences in their starting dates. Applying one trend factor to an already combined historical total is valid only when the amounts being combined share the same starting date and use the same applicable trend factor.
A frequent exam error is to calculate a plausible-looking quotient and overlook that its numerator and denominator describe different dates. For example, using four years of loss trend and three years of exposure trend does not become consistent merely because both factors were applied correctly according to separate instructions. First identify the intended target date; then confirm that each amount has been projected to it.
- Check the period units. Convert months to years when the annual factor is compounded by a fractional exponent; 18 months is 1.5 years, not 18 years.
- Check the trend assigned to each amount. Apply the selected loss trend to losses and the selected exposure trend to exposure units.
- Check the endpoint. For each group, verify that projected losses and exposures represent the same target date.
- Check the aggregation. Add projected losses and projected exposures separately, then divide the totals.
- Check the interpretation. If the loss and exposure periods differ, label the quotient as an instructed calculation, not as a consistent projected loss cost.
Check Your Understanding
For each hypothetical problem, assume constant annual effective trends over the stated periods. Round final dollar loss costs to the nearest cent. No interest discounting, mortality assumption, or probability distribution is used.
- A portfolio has historical losses of $120,000 and 2,400 exposure units. Project both amounts two years using a 3% annual loss trend and a 1% annual exposure trend. What is the projected loss cost? (Answer: $52.00 per exposure unit)
- A hypothetical book has $90,000 in losses and 1,800 exposure units at the same historical date. Project both one year, using a 6% loss trend and a 2% exposure trend. What is the projected loss cost? (Answer: $51.96 per exposure unit)
- Group A has losses of $30,000 and 600 exposure units measured one year before a common target date. Group B has losses of $50,000 and 1,000 exposure units measured two years before that date. With annual loss and exposure trends of 4% and 1%, respectively, calculate the combined projected loss cost. (Answer: $52.44 per exposure unit)
- A portfolio has historical losses of $64,000 and 1,000 exposure units. The instructions project losses for two years at 5% annually but exposures for one year at 3% annually. What ratio do those instructions produce, and why is it not a consistent two-year projected loss cost? (Answer: $68.50 per exposure unit; the numerator is projected two years and the denominator one year)
- Historical losses are $150,000 and exposure units are 2,000. Both amounts are projected three years at the same annual trend rate of 2%. What is the projected loss cost, and what does this show about equal loss and exposure trends? (Answer: $75.00 per exposure unit; the equal trend factors cancel in the ratio)