Additivity of Premium, Benefit, and Expense Values

Why Additivity Matters

A state-contingent contract may involve premiums received, benefits paid, and expenses incurred at different times and in different states. It is often convenient to value each component separately: this makes the assumptions visible and helps explain how the net value is built. That separation is valid only when each component is valued using the same contract timeline, state model, interest basis, and sign convention.

Throughout this tutorial, cash flows are viewed from the insurer’s perspective. A premium received is positive; a benefit or expense paid is negative. The net present value (NPV) is therefore the present value of premiums less the present values of benefits and expenses. A different perspective is possible, but the signs must be changed consistently.

Definition: Let $C_k^P$, $C_k^B$, and $C_k^E$ be the insurer’s signed premium, benefit, and expense cash flows at time $k$. Thus, $C_k^P$ is nonnegative, while $C_k^B$ and $C_k^E$ are nonpositive. The net cash flow is $C_k=C_k^P+C_k^B+C_k^E$. At effective annual interest rate $i$, with $v=1/(1+i)$, the expected present value of component $J$ is $\operatorname{EPV}(J)=\sum_{k=0}^{n}v^kE[C_k^J]$. The net expected present value is $\sum_{k=0}^{n}v^kE[C_k]$.

Here $E[\cdot]$ is expectation under the specified state model and any other assumptions governing the contract cash flows. A cash flow may depend on the state at a payment date, on a transition during an interval, or on another event represented in the model. The key requirement is that the individual component values and the combined value use the same probability law for those events.

The Additivity Result

For a finite contract, suppose the present values of the cash-flow components have finite expectations. At every payment time, the net cash flow is the sum of its components. Discounting is multiplication by the same deterministic factor $v^k$. Linearity of expectation then gives the result directly.

\begin{aligned} \operatorname{EPV}(\text{net}) &=E\left[\sum_{k=0}^{n}v^k \left(C_k^P+C_k^B+C_k^E\right)\right]\\ &=\sum_{k=0}^{n}v^kE[C_k^P] +\sum_{k=0}^{n}v^kE[C_k^B] +\sum_{k=0}^{n}v^kE[C_k^E]\\ &=\operatorname{EPV}(\text{premiums}) +\operatorname{EPV}(\text{benefits}) +\operatorname{EPV}(\text{expenses}). \end{aligned}

This identity does not require the component cash flows to be independent. For example, a premium may be received only when a benefit is not payable, or an expense may occur when a benefit is paid. Dependence affects the possible outcomes and the probabilities assigned to them; it does not alter the linearity of expectation. The identity does require that the expected values exist and that the cash flows being added refer to the same contract and valuation basis.

Specify the state model.
State the initial state or initial-state distribution, transition probabilities, and contract horizon. Use this same model for each cash-flow component.

Put every component on its actual payment date.
Identify whether a payment is made at time zero, at the beginning or end of an interval, or on some other specified date. Apply the matching discount factor.

Use consistent signs.
From the insurer’s perspective, record premiums as positive receipts and benefits and expenses as negative outflows.

Calculate and combine.
Find each component’s expected present value under the common assumptions, then add the signed values. As a check, calculate the expected net cash flow by date and discount it directly.

In a discrete state model, state probabilities at each payment date can be used to find expected payments that depend only on the state at that date. For a payment triggered by a transition, the relevant probability is the probability of that transition under the same model, not merely the probability of occupying its destination state at some later date. The earlier material on transition-probability valuations provides ways to obtain those probabilities; additivity concerns how their resulting cash-flow values fit together.

Worked Valuations

Worked Example: Separate Values for State-Dependent Annual Cash Flows

A hypothetical two-year contract begins with a policyholder in state A. At each year-end the policyholder is in state A or state D, with transition matrix $P=\begin{pmatrix}0.8&0.2\\0.3&0.7\end{pmatrix}$, where rows and columns are ordered A, D. At each year-end the insurer receives a premium of $100 in state A, pays a benefit of $300 in state D, and incurs an expense of $10 in either state. The effective annual interest rate is 10%. Calculate the signed EPV of each component and the net EPV.

Method. Find the state probabilities at each year-end, value each component at those dates, and add using the insurer-perspective signs. The expense applies in both states.

The initial state distribution is $(1,0)$. After one year the distribution is $(1,0)P=(0.8,0.2)$. After two years it is $(0.8,0.2)P=(0.8(0.8)+0.2(0.3),\,0.8(0.2)+0.2(0.7))=(0.70,0.30)$. The discount factors are $v=1/1.10=0.909091$ and $v^2=1/1.10^2=0.826446$.

The expected premium receipts are $80 at year 1 and $70 at year 2. Thus the signed premium EPV is $80(0.909091)+70(0.826446)=72.7273+57.8512=\$130.58$.

The expected benefit outflows are $60 at year 1 and $90 at year 2. Their signed EPV is $-60(0.909091)-90(0.826446)=-54.5455-74.3802=-\$128.93$.

The expense is $10 at each year-end regardless of state. Its signed EPV is $-10(0.909091)-10(0.826446)=-9.0909-8.2645=-\$17.36$.

Adding the three component values gives $130.58-128.93-17.36=-\$15.71$, using the rounded displayed amounts. Using unrounded values, the net EPV is $-\$15.7025$, or −$15.70 to the nearest cent.

Check. The expected net cash flow is $80-60-10=\$10$ at year 1 and $70-90-10=-\$30$ at year 2. Its EPV is $10(0.909091)-30(0.826446)=9.0909-24.7934=-\$15.7025$, agreeing with the component calculation. This check avoids mistakenly treating the expense as applying only in state A.

Worked Example: Payments Triggered by Different End States

A hypothetical one-year contract starts in state A. At year-end the probabilities of being in states A, D, and X are 0.75, 0.20, and 0.05, respectively. The insurer receives a $120 premium if the ending state is A, pays a $500 benefit if it is D, pays a $1,000 benefit if it is X, and incurs a $20 expense in every outcome. The effective annual interest rate is 5%. Find the separate signed EPVs and the net EPV.

Method. Use each ending-state probability for the cash flow specified in that state, discount every year-end amount by $v=1/1.05=0.952381$, and then combine the signed component values. The stated state probabilities are the complete probability information needed for these one-year payments.

The expected premium is $0.75(120)=\$90$, so its EPV is $90(0.952381)=\$85.71$.

The expected benefit outflow is $0.20(500)+0.05(1{,}000)=100+50=\$150$. Its signed EPV is $-150(0.952381)=-\$142.86$.

The expense is payable in every outcome, so its signed EPV is $-20(0.952381)=-\$19.05$.

The net EPV is $85.7143-142.8571-19.0476=-\$76.1904$, or −$76.19 to the nearest cent. Directly, the expected net year-end cash flow is $90-150-20=-\$80$, and $-80(0.952381)=-\$76.19$.

Check. The probability-weighted expected benefit is $150, not $500 or $1,000: each benefit is payable only in its specified ending state. This avoids applying a benefit amount to the wrong state or treating mutually exclusive benefits as simultaneous payments.

Worked Example: A Time-Zero Expense and Premiums in Advance

A hypothetical two-year contract starts in state A and has the same transition matrix as in the first example. The effective annual interest rate is 10%. The insurer receives a $50 premium at time zero and another $50 at time 1 only if the policyholder is then in state A. At the ends of years 1 and 2, it pays $200 if the policyholder is in state D. An issue expense of $12 is incurred at time zero, and a further expense of $5 is incurred at the end of each year in either state. Calculate the signed component EPVs and check their sum.

Method. Place each cash flow at its stated time before discounting. In particular, the time-1 premium depends on the state at time 1, while the time-0 premium and issue expense are not discounted.

From the first example’s transition calculation, the probabilities of state A at times 1 and 2 are 0.80 and 0.70, and the probabilities of state D are 0.20 and 0.30. The discount factors are $v=0.909091$ and $v^2=0.826446$.

Premiums have expected amounts $50 at time zero and $0.80(50)=\$40$ at time 1. The premium EPV is $50+40(0.909091)=50+36.3636=\$86.36$.

Expected benefit outflows are $0.20(200)=\$40$ at time 1 and $0.30(200)=\$60$ at time 2. The benefit EPV is $-40(0.909091)-60(0.826446)=-36.3636-49.5868=-\$85.95$.

The expense EPV includes $-\$12$ at time zero and $-\$5$ at each year-end: $-12-5(0.909091)-5(0.826446)=-12-4.5455-4.1322=-\$20.68$.

The component sum is $86.3636-85.9504-20.6777=-\$20.2645$, or −$20.26 to the nearest cent. As a direct check, the expected net amounts are $50-12=\$38$ at time zero, $40-40-5=-\$5$ at time 1, and $0-60-5=-\$65$ at time 2. Their EPV is $38-5(0.909091)-65(0.826446)=38-4.5455-53.7190=-\$20.2645$.

Check. The second premium is received at time 1, not time zero, and is payable only in state A. The method therefore avoids both discounting it at the wrong date and treating it as certain.

Conditions and Common Exam Traps

Additivity is a general property of expected values, not a special property of a particular transition matrix. Its useful actuarial application is that premiums, benefits, and expenses can be valued separately and then combined, provided the values remain comparable. This means using a common initial-state assumption, transition model, valuation horizon, interest basis, and timing convention. If a component has a distinct payment condition, that condition must still be represented correctly within the common model.

For example, suppose premiums cease on disability while expenses continue, or a benefit is paid only on a transition into a state. Separate valuation remains valid, but the premium and benefit calculations must use their respective indicators and probabilities. It is not valid to use state-occupancy probabilities for one component and transition probabilities for another when that substitution changes the payment rule.

A frequent error is to add magnitudes of outflows to premium receipts, or to apply inconsistent signs. Under the insurer-perspective convention, the signed benefit and expense values are negative. In the first worked example, the incorrect calculation $130.58+128.93+17.36=\$276.87$ treats every component as a receipt. The correction is $130.58-128.93-17.36=-\$15.71$ using the displayed rounded values; with unrounded values it is −$15.70 to the nearest cent. The sign convention must be applied before the components are combined.

Another common error is to calculate each component under a different timing or probability basis and then invoke additivity. Linearity cannot repair an incorrect component value. A useful final check is to form the expected net cash flow at each payment date and discount those amounts directly. Agreement between the direct calculation and the signed component sum is a strong check for omitted flows, incorrect state probabilities, and timing errors.

Key result: Under one common state model and valuation basis, the expected present value of the net cash flow equals the sum of the signed expected present values of premiums, benefits, and expenses. The components need not be independent; their payment conditions, signs, and dates must be represented consistently.

Check Your Understanding

Use the insurer-perspective convention: premiums are positive, and benefits and expenses are negative. Round dollar answers to the nearest cent.

A one-year contract has ending-state probabilities 0.60 for A, 0.25 for D, and 0.15 for X. It receives $80 in state A, pays $400 in state D, pays nothing in X, and incurs a $12 expense in every state at year-end. The effective annual rate is 4%. Find the signed net EPV. (Answer: −$61.54)
A two-year contract starts in A and has transition matrix $P=\begin{pmatrix}0.7&0.3\\0.2&0.8\end{pmatrix}$, with states A and D. At each year-end it receives $90 in A, pays $250 in D, and incurs $8 in either state. At 5% effective annual interest, find the net EPV. (Answer: −$83.45)
A contract receives $120 and incurs an expense of $15 at time zero. At time 1, it is in state A with probability 0.70 and D with probability 0.30; it then receives $40 in A, pays a $200 benefit in D, and incurs a $10 expense in either state. The effective annual rate is 4%. Find the net EPV. (Answer: $64.62)
At the end of one year, two equally likely outcomes have the following cash flows: in outcome 1, premium $100, benefit $0, expense $10; in outcome 2, premium $100, benefit $160, expense $30. At 10% effective annual interest, calculate the three signed component EPVs and the net EPV. (Answer: premiums $90.91; benefits −$72.73; expenses −$18.18; net EPV $0.00)
Explain whether the additivity identity requires the premium and benefit cash flows to be independent. State one condition that must hold for the identity to be used as a valid valuation comparison. (Answer: Independence is not required; the components must be valued under the same probability model, timing, and interest basis, with consistent signs.)

What You'll Learn