Why Exposure Growth Changes the Comparison
An aggregate claim liability can rise because an insurer covers more exposure, because the selected liability per unit of exposure rises, or because both occur. Those explanations are not interchangeable. A larger portfolio can require a larger reserve even when the reserve level per exposure is unchanged; conversely, an increase in reserve per exposure can be obscured by a shrinking portfolio.
To interpret a change, compare reserves on a consistent basis and select an exposure measure that corresponds to the claims represented in those reserves. For an accident-year reserve, for example, earned exposures associated with that accident year may be more relevant than the current number of policies in force. The appropriate measure depends on the business and the reserve being analyzed.
The identity \(R_t=E_tq_t\) separates scale from the average amount associated with each exposure unit. It does not, by itself, explain why \(q_t\) changed. Differences in claim maturity, valuation basis, coverage, or assumptions can affect the comparison. A change in the composition of the business can also matter, but this tutorial treats the exposure volume as the scale measure and does not attempt to adjust for composition.
A Consistent Decomposition of the Reserve Change
For two periods, label the earlier period 0 and the later period 1. The total change is \(R_1-R_0\). One exact decomposition first values the exposure increase at the earlier period’s reserve per exposure, then applies the change in reserve per exposure to the later period’s exposure:
The first term is the exposure-volume effect under the stated ordering. The second is the per-exposure effect, evaluated at the later exposure volume. These are accounting components of the change, not necessarily separate causal effects.
The order matters for the allocation between components. The equally valid decomposition \((E_1-E_0)q_1+E_0(q_1-q_0)\) uses the later per-exposure amount for the volume effect and the earlier exposure for the per-exposure effect. Both sum to \(R_1-R_0\), but they can assign different amounts to each component. State the convention used when reporting a decomposition.
Check that the two reserves use the same gross or net basis, coverage scope, valuation convention, and relevant claim population.
Use the exposure measure that represents the claims included in each reserve, with the same unit and time basis in both periods.
Calculate \(q_0=R_0/E_0\) and \(q_1=R_1/E_1\), then apply and label one decomposition convention.
Report the arithmetic separately from conclusions about adequacy. A per-exposure difference is a diagnostic that needs explanation, not proof of a reserve error.
Worked Comparisons
Worked Example: A Larger Portfolio and a Higher Reserve per Exposure
A hypothetical insurer compares two year-end gross claim liabilities for the same coverage and reporting basis. The earlier period has 8,000 earned exposure units and a selected reserve of $12,000,000. The later period has 10,000 earned exposure units and a selected reserve of $16,000,000. Decompose the change using the earlier-period reserve per exposure for the volume effect. State what the result says about adequacy.
Method. Calculate reserve per exposure in each period, then use \((E_1-E_0)q_0+E_1(q_1-q_0)\). This avoids attributing the full aggregate increase to a change in reserve level.
Earlier reserve per exposure: \(q_0=\$12{,}000{,}000/8{,}000=\$1{,}500\). Later reserve per exposure: \(q_1=\$16{,}000{,}000/10{,}000=\$1{,}600\).
The reported aggregate change is \(\$16{,}000{,}000-\$12{,}000{,}000=\$4{,}000{,}000\). The volume effect is \((10{,}000-8{,}000)\times \$1{,}500=2{,}000\times \$1{,}500=\$3{,}000{,}000\). The per-exposure effect is \(10{,}000\times(\$1{,}600-\$1{,}500)=10{,}000\times \$100=\$1{,}000{,}000\).
Check: \(\$3{,}000{,}000+\$1{,}000{,}000=\$4{,}000{,}000\), matching the reported change. Exposure grew by \(2{,}000/8{,}000=25\%\), and reserve per exposure grew by \(\$100/\$1{,}500=6.6667\%\). The aggregate liability grew by \(33.3333\%\), consistent with the compounded factors \(1.25\times1.066667\approx1.333333\).
Interpretation and trap. The later reserve is higher partly because there are more exposures and partly because the selected reserve per exposure is higher. The $1,000,000 component does not establish that the later reserve is inadequate: the comparison does not identify the reason for the per-exposure change or determine the correct liability.
Worked Example: Standardizing the Later Reserve to the Earlier Exposure
For a hypothetical liability portfolio on a consistent valuation and gross basis, the earlier estimate is $18,000,000 for 12,000 earned exposure units. The later estimate is $21,600,000 for 13,500 units. Calculate the later reserve at the earlier exposure volume and reconcile that standardized comparison to the actual change.
Method. Use the earlier period as the benchmark. First calculate the earlier reserve per exposure, then apply it to the later exposure volume to obtain the reserve expected from volume alone under that benchmark.
Earlier reserve per exposure: \(\$18{,}000{,}000/12{,}000=\$1{,}500\). The later reserve per exposure is \(\$21{,}600{,}000/13{,}500=\$1{,}600\). At the later exposure volume, the amount using the earlier rate is \(13{,}500\times \$1{,}500=\$20{,}250{,}000\).
The volume-only increase from the earlier reserve is \(\$20{,}250{,}000-\$18{,}000{,}000=\$2{,}250{,}000\). The difference between the actual later reserve and the volume-standardized amount is \(\$21{,}600{,}000-\$20{,}250{,}000=\$1{,}350{,}000\).
Check against the decomposition: the actual change is \(\$21{,}600{,}000-\$18{,}000{,}000=\$3{,}600{,}000\). The per-exposure component is \(13{,}500\times(\$1{,}600-\$1{,}500)=\$1{,}350{,}000\). Adding the volume component gives \(\$2{,}250{,}000+\$1{,}350{,}000=\$3{,}600{,}000\).
Interpretation and trap. The standardized amount is a comparison benchmark, not a replacement estimate. Treating $20,250,000 as the actual later liability would omit the observed change in reserve per exposure. The remaining $1,350,000 identifies a difference to investigate, not its cause.
Worked Example: Correcting for Exposure Duration
A hypothetical insurer reports reserves for the same coverage at two valuations. In the earlier period, 5,000 policies were each exposed for an average of 0.8 years, and the selected liability is $6,000,000. In the later period, 6,000 policies were each exposed for an average of 0.75 years, and the selected liability is $7,200,000. Use policy-years as the exposure unit. Compare the reserves after allowing for exposure duration.
Method. Convert policy counts and average exposure durations to policy-years before calculating reserve per exposure. A policy count alone would not represent equal amounts of exposure in the two periods.
Earlier exposure is \(5{,}000\times0.8=4{,}000\) policy-years. Later exposure is \(6{,}000\times0.75=4{,}500\) policy-years. The corresponding reserve per exposure amounts are \(\$6{,}000{,}000/4{,}000=\$1{,}500\) per policy-year and \(\$7{,}200{,}000/4{,}500=\$1{,}600\) per policy-year.
The aggregate increase is \(\$1{,}200{,}000\). Using the earlier reserve per exposure, the volume effect is \((4{,}500-4{,}000)\times \$1{,}500=500\times \$1{,}500=\$750{,}000\). The per-exposure effect is \(4{,}500\times(\$1{,}600-\$1{,}500)=4{,}500\times \$100=\$450{,}000\).
Check: \(\$750{,}000+\$450{,}000=\$1{,}200{,}000\), the actual change. If policy counts were incorrectly used as the denominator, the apparent amounts would be \(\$6{,}000{,}000/5{,}000=\$1{,}200\) and \(\$7{,}200{,}000/6{,}000=\$1{,}200\), falsely suggesting no change per exposure.
Interpretation and trap. The calculation shows a $100 increase per policy-year on the selected basis. It does not prove that the reserve level is deficient; the comparison still needs investigation of valuation consistency, claim maturity, and other relevant conditions.
Conditions for a Useful Comparison
Normalization is only as meaningful as its numerator and denominator. Before interpreting a result, confirm that the reserve amounts are comparable and that each exposure total corresponds to the liabilities being measured. A gross reserve should not be compared with a net reserve without making the basis consistent. Likewise, a liability for one coverage scope should not be divided by exposure for a different scope.
| Check | Question to ask | Why it matters |
|---|---|---|
| Valuation and reserve basis | Do both amounts use a consistent valuation date convention and gross or net basis? | A change in presentation or measurement can look like a change in the liability level. |
| Exposure unit | Are both denominators measured in the same unit and over the relevant claim-producing period? | Policy counts, earned exposures, and policy-years are not interchangeable. |
| Liability scope | Does each exposure denominator correspond to the claims included in its reserve? | A mismatched numerator and denominator produce an uninformative ratio. |
| Comparability over time | Could differences in maturity, coverage, or estimation basis affect the per-exposure result? | The normalized change may reflect more than a change in reserve adequacy. |
For claim liabilities, the exposure period is especially important. A reserve at a valuation date may include unpaid claims from several past accident periods. Dividing that liability by the current in-force policy count may therefore pair a stock of claims from earlier periods with a denominator describing a different population or time interval. Use a denominator that is defensible for the particular comparison and explain the choice.
Reserve per exposure is a descriptive ratio. Even when its movement is stable, that fact alone does not show that either reserve is adequate: both estimates could share the same bias. If it rises, the increase may warrant investigation, but the ratio cannot identify the cause. The earlier material on reserve development and reserve sensitivity provides relevant context for evaluating changes in selected liabilities; normalization by exposure answers a distinct question about scale.
Common Mistakes and Reporting Practice
- Attributing all aggregate growth to reserve strengthening. First quantify the change in exposure volume. In the first example, $3,000,000 of the $4,000,000 increase was the volume component under the stated convention.
- Using a convenient but mismatched denominator. Current policy count may not represent the exposure underlying an unpaid claim reserve. Match the exposure definition to the liability being compared.
- Calling a per-exposure increase “inadequacy.” It is a signal to investigate. Adequacy requires evaluating the selected liability against the relevant claims and assumptions, not just dividing by exposure.
- Failing to state the decomposition convention. The volume and per-exposure components depend on the order used, even though their sum always equals the total change.
- Assuming a ratio removes all differences between periods. It adjusts for the chosen exposure scale, not every factor that can influence a reserve comparison.
Check Your Understanding
For each problem, use the exposure definition stated and show the volume and per-exposure components before checking the total change.
- A hypothetical portfolio has 6,000 earned exposure units and a $9,000,000 reserve in period 0. In period 1 it has 7,500 units and a $12,750,000 reserve. Using the earlier reserve per exposure for the volume effect, calculate both components and the percentage change in reserve per exposure. (Answer: Volume effect $2,250,000; per-exposure effect $1,500,000; reserve per exposure rises from $1,500 to $1,700, or 13.3333%.)
- A consistent-basis reserve changes from $14,400,000 on 12,000 exposure units to $15,600,000 on 12,500 units. Calculate the reserve per exposure in both periods and decompose the change using the earlier-period amount for the volume effect. (Answer: $1,200 and $1,248 per exposure; volume effect $600,000; per-exposure effect $600,000; total increase $1,200,000.)
- In period 0, 4,000 policies each average 0.75 years of exposure and support a $4,800,000 selected reserve. In period 1, 5,000 policies each average 0.8 years and support a $7,000,000 reserve. Using policy-years and the earlier reserve per exposure for the volume effect, calculate the decomposition. (Answer: Exposure is 3,000 and 4,000 policy-years; reserve per exposure is $1,600 and $1,750; volume effect $1,600,000; per-exposure effect $600,000; total increase $2,200,000.)
- A reserve rises from $10,000,000 to $12,000,000, while corresponding exposure rises from 8,000 to 9,000 units. What are the per-exposure amounts and the decomposition using the earlier-period amount for the volume effect? Does the result establish inadequacy? (Answer: $1,250 and $1,333.3333 per exposure; volume effect $1,250,000; per-exposure effect $750,000; total increase $2,000,000; no, the decomposition alone does not establish inadequacy.)
- A hypothetical report shows a gross reserve of $5,000,000 divided by 4,000 earned exposures in one period, and a net reserve of $5,400,000 divided by 4,500 earned exposures in the next. Is the resulting ratio comparison sufficient to conclude that net reserve per exposure rose, and what must be done first? (Answer: No; the bases differ. Put both reserve amounts on a consistent gross or net basis and confirm that the exposure definitions and liability scopes correspond before comparing.)