Why the Loss Basis Matters
A capital estimate begins with a choice about what counts as a loss. Is the question about how large claims could be, how far claims might depart from a plan, or how much the insurer’s surplus could fall? Those questions can use different random variables, even when they describe the same hypothetical year. A number calculated from one basis should not be interpreted as though it came from another.
For a one-period illustration, let \(X\) be a specified loss variable and let \(q\) be a selected confidence level. One common decomposition is expected loss plus unexpected loss, where the unexpected component is the quantile-based capital amount above the mean:
Here \(\operatorname{VaR}_q(X)\) is the \(q\)-quantile of \(X\). For discrete examples, use the convention \(\operatorname{VaR}_q(X)=\inf\{x:F_X(x)\ge q\}\). The decomposition requires both terms to refer to the same loss variable, period, valuation basis and currency. It does not mean that every possible capital framework defines required capital this way; the formula is used here to show how the selected loss basis affects interpretation.
Three Common Loss Bases
The three bases below answer related but distinct questions. Before calculating a capital amount, state the variable in words and, where helpful, as an equation.
| Basis | Example definition | Question it answers |
|---|---|---|
| Total claims | \(X=C\), where \(C\) is aggregate claims for the period | How large could claims be? |
| Deviation from plan | \(X=C-B\), where \(B\) is a fixed planned claim amount | How far might claims depart from the plan? |
| Decrease in surplus | \(X=S_0-S_1\), where \(S_0\) and \(S_1\) are beginning and ending surplus | How much could surplus decline over the period? |
For total claims, the analyst must still specify whether \(C\) includes all relevant claim payments and whether the amounts are gross or net of any specified recoveries. A claims-only variable does not automatically include premiums, expenses, investment results or other changes in surplus.
A deviation-from-plan variable compares actual losses with a stated benchmark. If \(B\) is fixed, a positive value of \(C-B\) means claims exceeded plan, while a negative value means claims were below plan. The plan is not necessarily equal to the expected claims. If it is not, the expected deviation is not zero.
For a surplus basis, define the surplus convention and included cash flows. For example, if \(S_0\) is beginning surplus, \(P\) is premium income, \(I\) is investment income, \(E\) is expenses, and \(C\) is claims, a simplified one-period relationship is \(S_1=S_0+P+I-E-C\). The corresponding loss is \(S_0-S_1=C+E-P-I\). A positive value represents a decline in surplus; a negative value represents an increase. This simple identity assumes the listed items are the only changes in surplus in the hypothetical model.
Worked Example: Total Claims and Unexpected Loss
Worked Example: Total Claims and Unexpected Loss
A hypothetical insurer models one-year aggregate claims \(C\), in thousands of dollars, as follows: claims are 80 with probability 0.50, 120 with probability 0.30, and 200 with probability 0.20. There are no other cash flows in this claims-only model. Calculate expected claims and the unexpected loss at \(q=0.80\). State what the result does and does not represent.
Method. Define the loss variable as total claims, \(X=C\). First calculate its expectation, then find its 80th percentile, and subtract the expectation. This keeps the expected and unexpected components on the same basis.
Expected claims. The probability-weighted amount is
Thus expected claims are 116 thousand dollars. The cumulative probabilities are 0.50 at 80, 0.80 at 120, and 1.00 at 200. Under the stated quantile convention, \(\operatorname{VaR}_{0.80}(C)=120\) thousand dollars.
Unexpected loss.
Check. Expected claims plus unexpected loss equals the quantile: \(116+4=120\). The result describes unexpected loss on the total-claims basis. It is not, by itself, a measure of a decrease in surplus, because premiums, expenses and investment income are not included in \(C\).
Common mistake avoided: Reporting 120 thousand dollars as unexpected loss confuses the total quantile with the amount above expected claims. The unexpected component under this decomposition is 4 thousand dollars.
Why a Fixed Plan Changes the Level, Not the Unexpected Component
When the plan amount is fixed, total claims and deviations from plan are related by a constant shift. That relationship explains why the two bases can give different expected losses and quantiles while producing the same unexpected amount under the stated formula.
The equality does not make the two variables interchangeable for every purpose. Expected deviation from plan can be positive or negative, and the plan itself may be chosen for budgeting or performance monitoring rather than as an estimate of expected claims. The variable must still be named so that the result is interpreted correctly.
Worked Example: Claims Relative to a Plan
Worked Example: Claims Relative to a Plan
A hypothetical portfolio has a fixed annual claims plan \(B=100\) thousand dollars. Actual claims \(C\), in thousands of dollars, are 70 with probability 0.40, 110 with probability 0.40, and 170 with probability 0.20. Calculate expected deviation from plan, its 80th percentile, and unexpected loss. Compare the result with the claims basis.
Method. Define \(D=C-B=C-100\). Do not assume that the plan equals expected claims; calculate the expected claims first.
Expected claims and expected deviation.
So expected claims exceed the plan by 6 thousand dollars, and \(E[D]=106-100=6\) thousand dollars. The possible deviations are \(-30\), \(10\), and \(70\), with probabilities 0.40, 0.40, and 0.20. Their cumulative probabilities are 0.40, 0.80, and 1.00, so \(\operatorname{VaR}_{0.80}(D)=10\) thousand dollars.
Unexpected loss.
On the claims basis, \(\operatorname{VaR}_{0.80}(C)=110\) and \(E[C]=106\), giving \(110-106=4\) thousand dollars as well. The plan-relative quantile is 10 thousand dollars, not 110, because it measures claims above the plan rather than total claims.
Check. The fixed plan shifts the claims quantile from 110 to \(110-100=10\), and shifts the mean from 106 to \(106-100=6\). Both calculations give an unexpected amount of 4 thousand dollars.
Common mistake avoided: Calling 10 thousand dollars the unexpected loss uses the plan-relative quantile but fails to subtract the expected deviation. The correct unexpected amount is \(10-6=4\) thousand dollars.
Measuring Loss Through a Change in Surplus
A surplus-based loss includes the effects of the cash flows specified in the surplus model. This makes it useful when the question concerns the effect of a period’s experience on surplus, rather than claims in isolation. It also means that the outcome depends on how those cash flows vary together. In particular, claims and investment income occurring in the same scenario should be combined scenario by scenario, not summarized separately and then treated as independent unless independence is an explicit assumption.
With the simplified relationship \(S_1=S_0+P+I-E-C\), define \(L_S=S_0-S_1\). Then \(L_S=C+E-P-I\). A higher claim amount increases this loss, while higher premium or investment income reduces it. Because \(L_S\) can be negative, its expected value or quantile need not be positive. If the loss variable is redefined—for example, by measuring loss relative to a different target surplus—the meaning of its expected and unexpected components changes accordingly.
Worked Example: Loss From a Change in Surplus
A hypothetical insurer begins the year with surplus of 500 thousand dollars. It receives fixed premiums of 180 thousand dollars and incurs fixed expenses of 20 thousand dollars. In three mutually exclusive scenarios, claims and investment income are: claims 80 and investment income 10, with probability 0.50; claims 120 and investment income 0, with probability 0.30; and claims 200 and investment income \(-30\), with probability 0.20. All amounts are in thousands of dollars. Calculate the loss based on the decrease in surplus, its expected value, its 80th percentile and unexpected loss.
Method. Use the specified one-period surplus equation \(S_1=S_0+P+I-E-C\), then define \(L_S=S_0-S_1\). Keep each scenario’s claims and investment outcome paired as given.
Ending surplus by scenario. The fixed part of ending surplus before investment income and claims is \(500+180-20=660\). Therefore \(S_1=660+I-C\). In the first scenario, \(S_1=660+10-80=590\). In the second, \(S_1=660+0-120=540\). In the third, \(S_1=660-30-200=430\).
Loss by scenario. Subtract each ending surplus from the opening surplus of 500: \(L_S=500-590=-90\), \(500-540=-40\), and \(500-430=70\), respectively. The scenario losses are therefore \(-90\), \(-40\), and \(70\), with probabilities 0.50, 0.30, and 0.20.
Expected loss.
The cumulative probability is 0.50 at \(-90\), 0.80 at \(-40\), and 1.00 at 70. Thus \(\operatorname{VaR}_{0.80}(L_S)=-40\) thousand dollars, and
Check. Expected loss plus unexpected loss equals the loss quantile: \(-43+3=-40\). A negative loss means surplus increased relative to its opening amount in that outcome. The calculation uses the surplus basis; it is not the total-claims capital amount.
Common mistake avoided: Using only the claims distribution would omit the scenario-specific investment outcomes and the fixed premium and expense cash flows. For this question, the defined loss is the decrease in surplus, so all specified cash flows belong in the calculation.
A Consistent Calculation Process
The following sequence helps prevent a correct calculation from answering the wrong question.
Clarify whether the objective concerns total claims, departures from plan, or a change in surplus.
Write the included cash flows and any benchmark explicitly. For example, use \(C\), \(C-B\), or \(S_0-S_1\), as appropriate.
Find the expectation and quantile of that same variable, then take their difference if using unexpected loss as quantile less mean.
Report the period, currency or units, quantile convention, loss basis and whether the result is expected loss, a quantile, or unexpected loss.
Common Interpretation Traps
- Calling a claims quantile a surplus loss. Claims alone do not describe the change in surplus when other cash flows are part of the question.
- Treating the plan as the expected value. A fixed plan can differ from expected claims, so \(E[C-B]\) may not be zero.
- Subtracting the wrong mean. Unexpected loss is calculated from the mean of the chosen loss variable, not automatically from the mean of total claims.
- Dropping scenario dependence. If claims and investment outcomes are paired by scenario, preserve those pairings when calculating surplus loss.
- Ignoring the sign. Under \(S_0-S_1\), positive values mean a fall in surplus and negative values mean an increase.
The practical lesson is simple: define the loss basis before selecting or applying a capital formula. A well-calculated quantile can still be misleading if its variable does not match the question. Being explicit about the basis makes comparisons meaningful and helps identify which cash flows the estimate includes. This distinction is aligned with the CAS Exam 9 (Risk Management for Actuaries) syllabus.
Check Your Understanding
For each problem, identify the stated loss basis before calculating. Amounts are in thousands of dollars unless noted.
- A hypothetical insurer’s total claims are 50 with probability 0.60, 100 with probability 0.25, and 250 with probability 0.15. Using the 85th percentile and the expected-claims decomposition, calculate expected claims and unexpected loss. (Answer: expected claims 92.5; 85th percentile 100; unexpected loss 7.5.)
- A hypothetical portfolio has a fixed claims plan of 90. Actual claims are 60 with probability 0.25, 90 with probability 0.50, and 180 with probability 0.25. Find the expected deviation from plan and the unexpected loss at the 90th percentile. (Answer: expected deviation 15; deviation quantile 90; unexpected loss 75.)
- A hypothetical insurer begins with surplus of 300, receives fixed premiums of 100, and incurs fixed expenses of 10. Claims and investment income are 40 and 5 with probability 0.50, 90 and 0 with probability 0.30, and 150 and \(-20\) with probability 0.20. Find the surplus-loss outcomes and unexpected loss at the 80th percentile. (Answer: surplus-loss outcomes \(-55, 0, 80\); expected loss \(-11.5\); 80th percentile 0; unexpected loss 11.5.)
- Let \(D=C-75\), where 75 is a fixed plan amount. If expected claims are 82 and the 90th percentile of claims is 120, find the expected deviation, the 90th percentile of deviations, and unexpected loss on either basis. (Answer: expected deviation 7; deviation quantile 45; unexpected loss 38.)