Why Exposure Mix Matters
A portfolio can have the same number of exposures in two periods and still have different expected losses. The reason may be that its exposures have shifted among categories with different expected loss costs. For example, a hypothetical insurer could write a greater share of risks in a category with higher expected losses, even if its total number of insured units stays constant.
To estimate this effect, distinguish exposure volume—the total number of exposure units—from exposure mix—the proportions of those units in the categories being considered. Expected losses depend on both. A change in total expected losses should not automatically be described as a volume change.
The expected loss cost for the whole portfolio is the exposure-weighted average of the category costs. The weights are exposure shares, not equal weights and not shares of expected loss. The formula assumes the category costs are appropriate for the periods being compared: changes in trend, coverage, limits, or other relevant assumptions must be handled separately or reflected in the selected category costs.
From Category Shares to Expected Loss
For each category, multiply its exposure count by its expected loss cost. Adding those category expected losses gives the portfolio expected loss. Equivalently, first calculate the weighted average expected loss cost and multiply it by total exposures. These are two forms of the same calculation.
Here \(N_{j,t}\) is the exposure count in category \(j\), and \(E_t\) is expected loss for the whole portfolio. If category counts are given directly, use the sum of category counts for \(N_t\) and divide each count by that total to obtain the shares. If shares are given, check that they sum to one before using them.
Worked Example: Calculating Expected Loss From a Two-Category Mix
A hypothetical commercial insurer has 1,000 exposure units. Category A has an expected loss cost of $120 per exposure and represents 70% of the portfolio. Category B has an expected loss cost of $300 per exposure and represents 30%. Calculate the expected loss.
Method. Calculate the exposure-weighted expected loss cost, then multiply by total exposures.
Category A contributes \(0.70 \times \$120=\$84\) per portfolio exposure. Category B contributes \(0.30 \times \$300=\$90\). Thus the weighted expected loss cost is \(\$84+\$90=\$174\) per exposure.
The expected loss is \(1{,}000\times \$174=\$174{,}000\). As a direct check, the category exposure counts are \(1{,}000\times0.70=700\) and \(1{,}000\times0.30=300\). The category expected losses are \(700\times \$120=\$84{,}000\) and \(300\times \$300=\$90{,}000\); their sum is also $174,000.
Common mistake avoided: Taking the unweighted average \((\$120+\$300)/2=\$210\) would incorrectly assume that both categories contain half the exposures. The given shares, not the number of categories, determine the weights.
Separating Volume and Mix Effects
To compare periods, calculate the expected loss in each period using that period’s total exposures and mix. The difference is the total expected loss change. To describe how much of the difference is associated with volume and how much with mix, use a stated order of calculation.
The volume effect asks what the change in total exposures would do if the starting-period average expected loss cost continued to apply. The mix effect then measures the change in average expected loss cost at the new volume. This order is a convention for allocating the change; state it when reporting the components.
The identity follows by adding and subtracting \(N_1C_0\): \(N_1C_1-N_0C_0=(N_1C_0-N_0C_0)+(N_1C_1-N_1C_0)\). The first bracket is \((N_1-N_0)C_0\), the volume effect; the second is \(N_1(C_1-C_0)\), the mix effect. Thus the two components reconcile exactly to the total change.
If instead the mix is changed first at the starting volume, the mix effect is \(N_0(C_1-C_0)\), and the volume effect is \((N_1-N_0)C_1\). That alternative also reconciles, but generally allocates a different amount to each component. The difference arises because the volume and mix changes interact: a change in average cost applies to a different number of exposures depending on the order.
Worked Example: Separating a Mix Change From a Volume Change
A hypothetical insurer has 1,000 exposures in period 0: 70% are in Category A, with an expected loss cost of $80 each, and 30% are in Category B, with an expected loss cost of $240 each. In period 1, there are 1,200 exposures, with 25% in A and 75% in B. Category costs are unchanged. Calculate both period expected losses and decompose the change using volume first at the starting mix.
Period 0. The weighted expected loss cost is \(0.70(\$80)+0.30(\$240)=\$56+\$72=\$128\). Expected loss is \(1{,}000(\$128)=\$128{,}000\).
Period 1. The weighted expected loss cost is \(0.25(\$80)+0.75(\$240)=\$20+\$180=\$200\). Expected loss is \(1{,}200(\$200)=\$240{,}000\). The total increase is \(\$240{,}000-\$128{,}000=\$112{,}000\).
Volume effect. Exposure volume increases by \(1{,}200-1{,}000=200\). At the starting expected loss cost of $128, the volume effect is \(200(\$128)=\$25{,}600\).
Mix effect. The average expected loss cost rises by \(\$200-\$128=\$72\). At the new volume of 1,200, the mix effect is \(1{,}200(\$72)=\$86{,}400\).
Check. The components sum to \(\$25{,}600+\$86{,}400=\$112{,}000\), equal to the total change. The method avoids attributing all of the increase to greater volume: both the exposure count and the exposure mix changed.
Reading a Mix Effect at Constant Volume
When total exposure volume is unchanged, the expected loss change is entirely attributable to the mix under the assumption that category expected loss costs remain constant. This makes a same-volume comparison a direct way to show the impact of a portfolio shift. It does not imply that the mix shift caused actual losses to change by exactly that amount; the calculation compares expected losses under the stated cost assumptions.
Worked Example: Measuring a Mix Shift With Fixed Exposure Volume
A hypothetical equipment portfolio has 800 exposures in each of two periods. Category A has an expected loss cost of $50, and Category B has an expected loss cost of $250. The starting shares are 50% A and 50% B. The later shares are 40% A and 60% B. Find each period’s expected loss and the mix effect.
Starting period. The average expected loss cost is \(0.50(\$50)+0.50(\$250)=\$25+\$125=\$150\). Expected loss is \(800(\$150)=\$120{,}000\).
Later period. The average expected loss cost is \(0.40(\$50)+0.60(\$250)=\$20+\$150=\$170\). Expected loss is \(800(\$170)=\$136{,}000\).
Total volume is unchanged, so the volume effect is \((800-800)(\$150)=\$0\). The mix effect is \(800(\$170-\$150)=800(\$20)=\$16{,}000\). The check is \(\$136{,}000-\$120{,}000=\$16{,}000\), which equals the mix effect.
Common mistake avoided: A 10-percentage-point decline in Category A does not mean that expected loss falls, because Category A has the lower cost. The shift is from the lower-cost category to the higher-cost category, so the weighted cost and expected loss increase.
Order of Allocation and Reconciliation
When both volume and mix change, their separate effects depend on the order used, even though the total expected loss change does not. This is not a failure of the calculation. It reflects how the joint change is allocated between two components.
Worked Example: Comparing Two Valid Decompositions
A hypothetical policy portfolio begins with 900 exposures: 60% in Category A and 40% in Category B. Expected loss costs are $100 for A and $250 for B. In the next period, the portfolio has 1,000 exposures, with 30% in A and 70% in B. Costs remain unchanged. Calculate the total change and compare the two allocation orders.
Starting-period expected loss. The starting average cost is \(0.60(\$100)+0.40(\$250)=\$60+\$100=\$160\). Expected loss is \(900(\$160)=\$144{,}000\).
Next-period expected loss. The new average cost is \(0.30(\$100)+0.70(\$250)=\$30+\$175=\$205\). Expected loss is \(1{,}000(\$205)=\$205{,}000\). The increase is \(\$205{,}000-\$144{,}000=\$61{,}000\).
Volume first, then mix. At the starting average cost, the volume effect is \((1{,}000-900)(\$160)=\$16{,}000\). The mix effect at the new volume is \(1{,}000(\$205-\$160)=\$45{,}000\). The sum is $61,000.
Mix first, then volume. At the starting volume, the mix effect is \(900(\$205-\$160)=\$40{,}500\). The volume effect at the new average cost is \((1{,}000-900)(\$205)=\$20{,}500\). The sum is again $61,000.
The volume-first convention assigns $4,500 more to mix and $4,500 less to volume than the mix-first convention. In either case, the components reconcile to the same total increase. The common mistake is to combine the volume effect from one order with the mix effect from the other; those components do not form a valid decomposition.
Checks and Practical Limits
Before interpreting an estimate, verify the inputs and the scope of the comparison. A sound calculation can still be misleading if the exposure units or category definitions are inconsistent across periods.
- Check the shares. Category shares must sum to 1 (or 100%). If they do not, resolve the missing, overlapping, or inconsistent categories before calculating.
- Check the exposure basis. The counts and category costs must use the same exposure unit. For example, do not multiply policy counts by a cost stated per vehicle-year unless those quantities have been reconciled.
- Check cost comparability. The basic decomposition holds category costs fixed. If costs differ between periods due to trend or changed coverage, the comparison must use costs appropriate to each period and separate that issue from the mix effect.
- Check the arithmetic. Independently calculate each period’s expected loss, subtract to get the total change, and confirm that the chosen volume and mix effects sum to it.
These calculations estimate expected losses, not realized losses. Actual outcomes may differ because claims are uncertain. The purpose of the mix adjustment is narrower: it prevents a change in portfolio composition from being mistaken for a change in total exposure volume when comparing expected losses. This distinction is useful when explaining changes in an expected loss estimate or constructing a like-for-like comparison.
Check Your Understanding
For each problem, calculate the requested expected loss or change. Assume category expected loss costs are unchanged across periods unless stated otherwise.
- A hypothetical portfolio has 600 exposures: 75% in Category A at $80 per exposure and 25% in Category B at $200. What is its expected loss? (Answer: $66,000.)
- A portfolio has 500 exposures with 60% in Category A and 40% in Category B. The category costs are $100 and $250, respectively. The next period has 650 exposures with the same mix. Find the expected loss in each period and the volume effect. (Answer: starting expected loss $80,000; next-period expected loss $104,000; volume effect $24,000; mix effect $0.)
- A hypothetical portfolio has 1,000 exposures in both periods. Category costs are $50 for A and $250 for B. The mix changes from 80% A and 20% B to 50% A and 50% B. Find the expected losses and mix effect. (Answer: starting expected loss $90,000; next-period expected loss $150,000; mix effect $60,000.)
- A portfolio changes from 800 exposures with a 50% A, 50% B mix to 1,000 exposures with a 30% A, 70% B mix. Category costs are $80 for A and $200 for B. Decompose the change using volume first at the starting mix. (Answer: starting expected loss $112,000; next-period expected loss $164,000; total increase $52,000; volume effect $28,000; mix effect $24,000.)
- A portfolio changes from 1,200 exposures with 25% in A and 75% in B to 1,000 exposures with 40% in A and 60% in B. Costs are $60 for A and $180 for B. Using volume first at the starting mix, find both expected losses and the two effects. (Answer: starting expected loss $180,000; next-period expected loss $132,000; total change \(-\$48,000\); volume effect \(-\$30,000\); mix effect \(-\$18,000\).)