When Exposure Should Not Receive a Loss Trend

Why the Exposure Component Needs Its Own Treatment

A projected aggregate loss can change for two distinct reasons: the expected loss per unit of exposure can change, or the amount of exposure can change. Those effects should be identified separately before applying a trend. A loss-cost trend belongs on expected loss per exposure unit; it does not automatically belong on the exposure quantity itself.

For example, a hypothetical insurer might expect more vehicle-years next year because it plans to insure more vehicles. That is exposure growth. Separately, expected loss per vehicle-year might rise because repair costs or claim frequency are expected to increase. That is a change in loss cost. If both are expected, project both components. If the only expected change is more vehicle-years, applying a loss trend as well would introduce a change not supported by the assumptions.

Definition: Let $E$ denote the quantity of exposure on a specified basis, and let $C$ denote expected loss per unit of that exposure. Aggregate expected loss is $L=E C$, provided $C$ is the average loss cost appropriate to that exposure basis. An exposure change alters $E$; a loss-cost change alters $C$. Project each component only for changes that actually affect it.

Exposure may be measured as policies, vehicle-years, payroll, sales, insured value, or another basis appropriate to the coverage. A change in exposure count is not the same as a change in the size of each exposure. For example, if the exposure base is insured value, a larger insured value can increase the number of exposure units even when the number of buildings is unchanged.

In a pure premium projection, the central distinction is whether the projected loss cost is being applied to a projected exposure quantity. In a loss-ratio indication, losses and premiums have their own adjustments and bases; the same discipline applies: do not use a loss trend to represent exposure growth and then separately increase the exposure or premium base for that same growth.

A Component-Based Projection

Suppose experience has aggregate losses $L_0$ on exposure $E_0$. The observed loss cost is $C_0=L_0/E_0$. If the expected loss cost changes by a factor $T_C$, and exposure changes by a factor $T_E$, then the projected aggregate loss is

L_1 = E_0 T_E \times C_0 T_C = L_0 T_E T_C.

Here $T_C$ represents only the change in expected loss per exposure unit, while $T_E$ represents only the change in exposure quantity. A factor of 1.00 means no projected change in that component. The factors may reflect different assumptions and need not be equal.

This decomposition is useful only when the units and assumptions are consistent. If the exposure basis itself grows—for example, payroll increases—then exposure measured in payroll units has increased. If the loss cost is expressed per dollar of payroll, it should not also be increased for that same payroll growth unless there is a separate reason for expected loss per payroll dollar to change.

Assumption: The product formula treats the projected average loss cost as applicable to the projected exposure. It is exact for a homogeneous exposure basis under the stated factors, and can also be used with an appropriately exposure-weighted average loss cost. If the mix of exposures changes and different groups have different loss costs, a single unchanged average loss cost may not be appropriate.

Why the Product Formula Holds

The formula follows from the definition of average loss cost, rather than from a special trend rule. For a portfolio with $E$ equivalent exposure units and aggregate expected loss $L$, define $C=L/E$. Then $L=EC$. At the future date, if exposure is $E_0T_E$ and loss cost is $C_0T_C$, substituting those projected components gives the result directly:

$$ L_1=(E_0T_E)(C_0T_C) =E_0C_0T_ET_C =L_0T_ET_C. $$

The algebra also shows the double-counting risk. If one assumed change is already represented by $T_E$, adding that same change to $T_C$ multiplies the aggregate loss by it twice. A factor belongs in both components only when the assumptions support two distinct effects.

Worked Example: More Policies and a Separate Loss-Cost Trend

A hypothetical program has 12,000 earned policy-years and $960,000 of expected losses at the current level. For the prospective period, the actuary assumes loss cost per policy-year will increase by 3% annually for two years, while policy-year volume will increase by 5% in total. Assume the same coverage and exposure definition apply throughout, and the two changes are separate. Project aggregate expected losses.

Method: separate loss-cost trend from exposure growth. The current loss cost is

C_0 = \frac{\$960{,}000}{12{,}000} = \$80 \text{ per policy-year}.

The two-year loss-cost trend factor is

$$ T_C = 1.03^2 = 1.0609. $$

Thus projected loss cost is $ \$80 \times 1.0609 = \$84.872 $ per policy-year. Projected exposure is $12{,}000 \times 1.05=12{,}600$ policy-years. Therefore,

L_1 = 12{,}600 \times \$84.872 = \$1{,}069{,}387.20.

Check. Starting with aggregate losses gives the same result: $ \$960{,}000 \times 1.0609 \times 1.05 = \$1{,}069{,}387.20 $. The calculation avoids treating the 5% increase in policy-year volume as if it were a loss-cost trend. Money is reported to the cent.

Worked Example: A Growing Exposure Base With No Separate Loss Trend

A hypothetical property portfolio has 40 stores, each with $1,500,000 in annual sales. Its current expected aggregate loss is $480,000. Exposure is measured in units of $1,000 of sales. The actuary expects sales per store to increase by 6% annually for two years, and assumes expected loss per $1,000 of sales remains constant. No separate loss-cost change is assumed. Project aggregate expected losses.

Method: project the exposure base, not the loss cost. Current exposure is $40 \times \$1{,}500{,}000=\$60{,}000{,}000$, or 60,000 units of $1,000. Current loss cost is

C_0 = \frac{\$480{,}000}{60{,}000} = \$8 \text{ per \$1,000 of sales}.

The two-year exposure factor is $1.06^2=1.1236$. Projected exposure is $60{,}000 \times 1.1236=67{,}416$ units. Because the assumed loss cost per unit is unchanged, projected aggregate loss is

L_1=67{,}416 \times \$8 = \$539{,}328.00.

Common mistake and correction. Applying the 6% annual sales increase to exposure and then also applying a 6% annual “loss trend” to loss cost would give $ \$480{,}000 \times 1.1236 \times 1.1236 = \$605{,}988.94 $. That calculation counts the assumed sales growth twice. Under the stated assumption that loss cost per $1,000 of sales is constant, the correct projection is $539,328.00. The answer is checked by multiplying projected exposure by the unchanged $8 loss cost.

Worked Example: A Changing Exposure Count and a Distinct Trend

A hypothetical program has 10,000 exposure units and $500,000 of expected losses. The actuary expects loss cost per unit to increase by 4% annually for two years because of a separately assumed change in claim costs. The number of exposure units is expected to increase by 8% over the same period because of planned portfolio growth. Assume the cost trend does not include the portfolio growth. Project aggregate expected losses.

Method: apply each factor to its own component. Current loss cost is $ \$500{,}000/10{,}000=\$50 $ per unit. The two-year loss-cost trend factor is $1.04^2=1.0816$, so projected loss cost is $ \$50 \times 1.0816=\$54.08 $ per unit. Projected exposure is $10{,}000 \times 1.08=10{,}800$ units. Hence,

L_1=10{,}800 \times \$54.08 =\$584{,}064.00.

Check. The aggregate-loss calculation is $ \$500{,}000 \times 1.0816 \times 1.08=\$584{,}064.00 $. The 4% annual factor is applied for two years to loss cost; the 8% increase is applied once to exposure. The mistake to avoid is applying the 8% portfolio growth to loss cost as well as to exposure.

When Not to Trend Exposure as Though It Were Loss Cost

A loss trend is intended to reflect a change in expected loss cost on a stated exposure basis. It is not a universal multiplier for every quantity in a projection. In particular, exposure should not receive a loss trend merely because the overall loss amount is expected to rise. Instead, identify what the exposure measure counts and what the trend assumption describes.

Count-based exposure grows, but loss cost is stable. If the number of insured units rises while expected loss per unit is assumed unchanged, increase exposure only.
The exposure base grows, but cost per base unit is stable. If payroll, sales, or insured value increases and loss cost is expressed per unit of that base, reflect the base increase in exposure. Do not also apply a trend for that same increase to the loss cost.
Loss cost changes independently of exposure. If expected loss per unit changes for a separately identified reason, trend loss cost, whether or not exposure also changes.
The assumption is already embedded in a component. If a projected exposure measure already reflects a change, do not apply that same change again through an additional loss-cost factor.

The word “trend” can refer to different underlying quantities. A forecast that total losses will rise by 10% is not enough information to determine whether loss cost, exposure, or both should change. The analyst needs to identify the source of the increase and the units of the loss-cost measure. If that decomposition is unavailable, a numerical projection may be computable but is not fully justified by the assumptions.

Exam Traps and a Practical Check

A common trap is to see that both exposure and losses are measured in dollars and assume both should receive the same trend. The units alone do not determine the adjustment. For example, exposure measured in dollars of sales may grow because sales per insured business increase. If loss cost is defined per dollar of sales and is assumed constant, that growth belongs in exposure, not in a second loss-cost trend.

A useful check is to write the projected total as “projected exposure × projected loss cost” before inserting numbers. Label the reason for each factor. If the same assumed change appears in both labels, pause and determine whether it genuinely affects both components or has been counted twice. Also confirm that a rate of growth expressed per year is applied for the correct number of years, while a stated total-period change is applied only once.

Name the exposure basis.
State whether the measure is a count, time-based unit, payroll, sales, insured value, or another quantity.

Define loss cost on that basis.
Divide aggregate expected loss by exposure, or otherwise state the appropriate average loss cost per unit.

Assign each assumed change once.
Use an exposure factor for a change in exposure quantity and a loss-cost factor for a separate change in expected loss per unit.

Reconcile to aggregate loss.
Multiply projected exposure by projected loss cost and check against the equivalent aggregate-factor calculation.

Key takeaway: Projecting exposure is not the same as trending loss cost. Apply loss trend to expected loss per exposure unit and project exposure separately. When one assumed change affects both expressions only because it has been represented twice, the resulting aggregate loss is overstated.

Check Your Understanding

For each hypothetical projection, identify the component that changes and compute the requested amount. Answers are rounded to the cent where needed.

A program has 2,000 exposure units and $140,000 of expected losses. Expected loss cost increases by 5% over the next year, while exposure count increases by 10%. What is projected aggregate loss? (Answer: $161,700.00)
A hypothetical portfolio has $25,000,000 of payroll exposure and $500,000 of expected losses. Payroll is expected to increase by 4% annually for two years, while expected loss per dollar of payroll remains constant. What is projected aggregate loss? (Answer: $540,800.00)
A program has 6,000 exposure units and $720,000 of expected losses. Loss cost increases by 3% annually for two years, and exposure volume decreases by 5% in total. What is projected aggregate loss? (Answer: $725,655.60)
A hypothetical book has 5,000 units and $400,000 of expected losses. For next year, loss cost per unit increases by 2% for a separately identified reason, and exposure count increases by 7%. What are projected loss cost per unit and aggregate expected loss? (Answer: $81.60 per unit and $436,560.00)
A portfolio has 100 buildings, each with $1,000,000 of insured value, and aggregate expected loss of $200,000. Exposure is measured per $1,000 of insured value. Insured values increase by 5% over the projection period, and expected loss per $1,000 of insured value is assumed unchanged. What is projected aggregate loss, and why should the 5% change not also be applied to loss cost? (Answer: $210,000.00; the increase is already reflected in the insured-value exposure.)

What You'll Learn