How Reserve Basis Changes Alter Profit Emergence

Why a Reserve Basis Changes Profit Emergence

A reserve is a liability measured at a particular time using a specified set of valuation assumptions. Changing those assumptions can change the reserve even when the contract, projected cash flows, and experience probabilities are held fixed. Because annual profit includes a change in reserve, a different reserve basis can alter when profit appears without changing the cash flows generated by the contract.

This distinction matters in profit testing. A larger reserve at the end of a year generally means less profit is reported in that year, all else equal. If that reserve is released in a later year, the later profit may be larger. The effect on a discounted profit measure depends on the reserve differences at the beginning and end of the projection, not just on whether one basis produces larger reserves in an intermediate year.

Definition: Let $V_t^b$ be the reserve at the end of year $t$ under reserve basis $b$, for a policy in the relevant state. Let $G_t$ be the net cash flow during year $t$, accumulated to the end of that year at the profit-test account rate, excluding investment earnings on the opening reserve. Under an annual effective account rate $i$, define profit as $$ \pi_t^b=G_t+(1+i)V_{t-1}^b-V_t^b. $$ A comparison of two reserve bases isolates the reserve effect only when $G_t$, the profit-test rate, the projected cash flows, and the experience probabilities are the same under both bases.

The term $(1+i)V_{t-1}^b$ is the opening reserve carried forward with interest. The closing reserve $V_t^b$ is the amount retained to support future obligations, so it is deducted in calculating profit. The reserve basis used to measure $V_t^b$ may employ different valuation assumptions from the account rate used to accumulate the opening reserve. In a controlled comparison, state clearly which rate is used for each purpose.

Reserve Differences and Annual Profit Differences

Write the difference between the two bases' reserves as $\Delta V_t=V_t^B-V_t^A$, and the corresponding profit difference as $\Delta\pi_t=\pi_t^B-\pi_t^A$. Since the cash flow $G_t$ is identical under both bases, it cancels when profits are subtracted.

\Delta\pi_t=(1+i)\Delta V_{t-1}-\Delta V_t.

This equation identifies the two sources of the change in year $t$: the difference in the opening reserves carried into the year, and the difference in closing reserves retained at year-end. A higher closing reserve under basis B, with the opening reserves equal, reduces basis B profit in that year. In a later year, releasing that extra reserve can increase basis B profit.

Key result: If the two bases have the same initial reserve, a reserve difference that arises during the projection shifts profit between years. If they also have the same reserve at the projection endpoint, the present value of the profit differences is zero at the same discount rate as the profit-test account rate. If endpoint reserves differ, the present value generally does not cancel.

Why the Discounted Changes Telescope

The cancellation result follows directly from discounting the annual profit differences. This is an accounting identity; it does not require the reserve differences to follow a particular pattern between the initial and final dates.

Start with the annual difference.
For each year, $\Delta\pi_t=(1+i)\Delta V_{t-1}-\Delta V_t$.

Discount to time 0.
Let $v=1/(1+i)$. Then $v^t\Delta\pi_t=v^{t-1}\Delta V_{t-1}-v^t\Delta V_t$.

Sum over the projection.
Adding the terms from $t=1$ to $n$ cancels each intermediate reserve difference: the closing term for one year is the opening term for the next.

Keep the boundary terms.
The remaining present value is $\Delta V_0-v^n\Delta V_n$. It is zero when both boundary reserve differences are zero.

\sum_{t=1}^{n}v^t\Delta\pi_t =\Delta V_0-v^n\Delta V_n.

The discount rate in this identity must match the rate $i$ used to carry the opening reserve in the profit equation. If a different discount rate is used for the reported net present value, the same cancellation need not hold. Also, a projection ending before the contract's obligations end can have a nonzero endpoint reserve difference. In that case, the endpoint term is economically relevant, not an arithmetic error.

Worked Example: A Reserve Release Moves Profit to the Second Year

A hypothetical two-year contract is in force at the start of both years and terminates at the end of year 2. There is no mortality or other decrement in this example. The annual effective profit-test account rate is $i=5\%$, so $1+i=1.05$. The common year-end accumulated net cash flows are $G_1=\$20$ and $G_2=\$15$. The reserve sequences, in dollars, are:

V_0^A=V_0^B=100,\qquad (V_1^A,V_2^A)=(80,0),\qquad (V_1^B,V_2^B)=(90,0).

Method. Calculate profit under each basis using the annual reserve-profit equation, then compare the signatures and their present values. The cash flows and experience are held fixed; only the reserve basis changes.

Year 1 under basis A: The opening reserve carried forward is $1.05(100)=105$. Therefore, $\pi_1^A=20+105-80=\$45$.

Year 1 under basis B: The opening reserve is also 100, but the closing reserve is 90. Therefore, $\pi_1^B=20+105-90=\$35$. The year 1 difference is $35-45=-\$10$, equal to $-\Delta V_1$, since $\Delta V_0=0$ and $\Delta V_1=10$.

Year 2 under basis A: The opening reserve carried forward is $1.05(80)=84$, and the closing reserve is zero. Thus, $\pi_2^A=15+84-0=\$99$.

Year 2 under basis B: The opening reserve carried forward is $1.05(90)=94.50$, and the closing reserve is zero. Thus, $\pi_2^B=15+94.50-0=\$109.50$. The year 2 difference is $109.50-99=\$10.50$, the interest-accumulated release of the extra year 1 reserve.

The signatures are $(\$45,\$99)$ under A and $(\$35,\$109.50)$ under B. Discounting at 5%, basis A gives $45/1.05+99/1.1025=\$132.65$, to the nearest cent. Basis B gives $35/1.05+109.50/1.1025=\$132.65$, to the nearest cent. As a check, the present value of the differences is $-10/1.05+10.50/1.1025=0$. The initial and terminal reserve differences are both zero.

Common mistake avoided: Treating the higher year 1 closing reserve as a permanent loss of profit. It reduces year 1 profit by \$10, but its release increases year 2 profit by \$10.50; after discounting, the shifts cancel.

State-Weighted Profit Signatures

For a product with multiple states, reserves and conditional profits can differ by state. A profit signature per policy issued is obtained by weighting the conditional profit in each state by the probability of reaching that state. The reserve-basis comparison must use the same state probabilities for both bases if the aim is to isolate the reserve effect.

For example, if state $j$ has probability $p_{t,j}$ at time $t$, the expected annual profit difference per policy issued is $\sum_j p_{t,j}\Delta\pi_t(j)$. Because the profit-difference equation applies separately in each state, it can be applied before or after state weighting, provided the state probabilities and transitions are common to both projections.

Worked Example: Weighting Reserve Effects Across Two States

Consider a hypothetical contract with two states at the end of year 1: active, H, and disabled, D. The transition probabilities from the initial active state are $P(H)=0.8$ and $P(D)=0.2$. There is no mortality in this two-year example, and every policy terminates at the end of year 2. The annual effective profit-test account rate is $i=5\%$. Both bases have an initial reserve of \$50. The common accumulated net cash flows are $G_1(H)=\$30$, $G_1(D)=\$10$, $G_2(H)=\$12$, and $G_2(D)=\$8$. The year 1 closing reserves are $V_1^A(H)=\$40$, $V_1^B(H)=\$45$, $V_1^A(D)=\$20$, and $V_1^B(D)=\$45$; terminal reserves at year 2 are zero in both states.

Year 1, active state: The carried opening reserve is $1.05(50)=\$52.50$. Hence $\pi_1^A(H)=30+52.50-40=\$42.50$, while $\pi_1^B(H)=30+52.50-45=\$37.50$. The difference is $-\$5$.

Year 1, disabled state: The same opening reserve and state-specific cash flow give $\pi_1^A(D)=10+52.50-20=\$42.50$ and $\pi_1^B(D)=10+52.50-45=\$17.50$. The difference is $-\$25$.

Expected year 1 difference: Weight the conditional differences by the transition probabilities: $0.8(-5)+0.2(-25)=-4-5=-\$9$. Equivalently, the expected reserve difference at year 1 is $0.8(5)+0.2(25)=\$9$, and the initial reserve difference is zero.

Year 2 difference: Since terminal reserves agree, the conditional profit differences are the carried reserve differences: $1.05(5)=\$5.25$ in H and $1.05(25)=\$26.25$ in D. Their expected value per policy issued is $0.8(5.25)+0.2(26.25)=4.20+5.25=\$9.45$.

The expected signature difference is $(-\$9,\$9.45)$. Its present value is $-9/1.05+9.45/1.1025=-8.5714+8.5714=\$0.00$, to the nearest cent. The state weighting changes the size of the shift, but the equal initial and terminal reserves still imply cancellation.

Common mistake avoided: Averaging the state reserves without using the state probabilities. The correct year 1 expected reserve difference is $0.8(\$5)+0.2(\$25)=\$9$, not an unweighted average of \$15.

When the Present Value Does Not Cancel

A reserve basis can leave a nonzero difference at the final projection date. This can occur when the projection stops while the contract still has future obligations, or when the two valuation bases assign different values to those obligations. The telescoping result then leaves the discounted endpoint reserve difference in the net present value.

Worked Example: A Nonzero Endpoint Reserve Difference

A hypothetical policy remains in force for a two-year projection, with future obligations continuing beyond the projection date. There is no mortality or decrement during the two projected years. The profit-test account rate is $i=4\%$. The common accumulated net cash flows are $G_1=\$30$ and $G_2=\$25$. The reserve sequences are $V_0^A=V_0^B=\$70$, $V_1^A=\$40$, $V_1^B=\$50$, $V_2^A=\$10$, and $V_2^B=\$30$.

Year 1: The carried opening reserve is $1.04(70)=\$72.80$. Thus, $\pi_1^A=30+72.80-40=\$62.80$, while $\pi_1^B=30+72.80-50=\$52.80$. The difference is $-\$10$.

Year 2: The carried reserves are $1.04(40)=\$41.60$ under A and $1.04(50)=\$52.00$ under B. Therefore, $\pi_2^A=25+41.60-10=\$56.60$, while $\pi_2^B=25+52-30=\$47.00$. The difference is $-\$9.60$.

The present value difference is $-10/1.04-9.60/1.0816=-\$18.49$, to the nearest cent. Here $\Delta V_0=0$ and $\Delta V_2=\$20$, so the boundary formula gives $0-20/1.0816=-\$18.49$, also to the nearest cent. The amount does not cancel because the endpoint reserves differ.

Common mistake avoided: Assuming that any reserve-basis change must leave total discounted profit unchanged. That conclusion requires equal initial and terminal reserves, as well as the matching discount and account rates.

Practical Checks and Exam Traps

When comparing profit signatures, keep the scope of the comparison explicit. If projected cash flows, transition probabilities, or the account rate also change, the observed profit difference is not solely a reserve-basis effect. Separating the causes prevents attributing a change in expected claims or premiums to reserve measurement.

Check the sign. With equal opening reserves, a higher closing reserve reduces current profit: $\Delta V_t>0$ implies $\Delta\pi_t=-\Delta V_t$.
Check the timing. A reserve difference at year-end becomes an opening reserve difference in the following year and is carried forward with interest.
Check the weighting. For state-dependent reserves, use the probability of each state under the common experience model.
Check both boundaries. Cancellation of discounted changes requires the relevant initial and terminal reserve differences to be accounted for.
Check the rate. The telescoping present-value identity uses the same effective rate for reserve accumulation and discounting.

These comparisons are aligned with the SOA Exam ALTAM (Advanced Long-Term Actuarial Mathematics) syllabus: the central task is to distinguish a change in the timing of reported profit from a change in the underlying contract cash flows or in the total value measured over a defined horizon.

Check Your Understanding

For each problem, compare the reserve bases using the stated annual profit convention. Amounts are in dollars unless otherwise stated.

A two-year projection uses $i=3\%$, equal initial reserves, and reserve differences $\Delta V_1=\$8$ and $\Delta V_2=0$. Find the profit differences in years 1 and 2. (Answer: $-\$8.00$ in year 1; $+\$8.24$ in year 2.)
At $i=6\%$, reserve differences are $\Delta V_0=\$4$, $\Delta V_1=\$10$, and $\Delta V_2=0$. Calculate the profit differences in both years and their time-0 present value. (Answer: $-\$5.76$ in year 1; $+\$10.60$ in year 2; present value $+\$4.00$.)
At the end of year 1, a two-state model has probabilities $0.7$ in H and $0.3$ in D. The reserve differences are $\Delta V_1(H)=\$6$ and $\Delta V_1(D)=\$16$; $\Delta V_0=0$, and all terminal reserve differences are zero at year 2. With $i=5\%$, find the expected profit difference in each year and the present value difference. (Answer: year 1 $-\$9.00$; year 2 $+\$9.45$; present value $\$0.00$.)
A three-year projection has $\Delta V_0=0$, $\Delta V_3=\$12$, and a matching account and discount rate of $4\%$. Without knowing the intermediate reserve differences, find the present value of the profit differences. (Answer: $-12/(1.04)^3=-\$10.67$, to the nearest cent.)
At $i=4\%$, reserve differences are $\Delta V_0=0$, $\Delta V_1=-\$7$, and $\Delta V_2=0$. Find the two annual profit differences and the present value difference. (Answer: $+\$7.00$ in year 1; $-\$7.28$ in year 2; present value $\$0.00$.)

What You'll Learn