Integrated Pension Benefit Profit and Reserve Case

Why Pension Profit Tests Need State-by-State Cash Flows

A pension arrangement can have different cash flows depending on whether a member remains active, becomes disabled, retires, withdraws, or dies. Retirement may trigger a pension or a commuted benefit; disability may trigger a separate income benefit; withdrawal may produce a refund; and death may trigger a survivor benefit. A single average claim amount does not preserve these distinctions. A useful projection tracks the member’s state, the transitions between states, and the cash flow associated with each transition.

The calculations below use an annual, discrete-time model. At each year-end, a member moves to one state, and the transition probabilities from a given starting state sum to one. Benefits are paid at year-end. When calculating annual profit, contributions and expenses are paid at the start of the year, and the opening reserve accumulates interest over the year. These timing assumptions matter: changing the payment timing changes both reserves and profit.

Definition: Let $S_t$ be the member’s state at time $t$, with states such as active $A$, disabled $D$, retired $R$, withdrawn $W$, and dead $M$. The one-year transition probability from state $s$ to state $j$ is $p_{sj}=P(S_{t+1}=j\mid S_t=s)$. The probabilities across all possible destination states $j$ sum to one for each starting state $s$. A state-specific reserve is the value, at a specified date, of future net obligations conditional on the member being in that state at that date.

State-Specific Reserves from a One-Year Projection

For a one-year prospective valuation with no future contributions, let $b_{sj}$ be the benefit paid at year-end if the member moves from state $s$ to state $j$. Let $V_j$ be the reserve immediately after that payment date, conditional on being in state $j$. Under an effective annual interest rate $i$, discounting the expected year-end obligations gives the reserve in state $s$.

V_s=\frac{\sum_j p_{sj}(b_{sj}+V_j)}{1+i}.

This expression is conditional on the state at the valuation date. It is not appropriate to apply one transition row indiscriminately to every state: a disabled member and a retired member can have different transition probabilities and different benefits. If contributions or expenses occur in the year being valued, they must also be included, with signs and timing consistent with the chosen reserve definition.

Worked Example: Reserves for Active, Disabled, and Retired Members

A hypothetical pension plan has one year remaining in the projection. It pays a year-end benefit of $8 if the member is disabled at the end of the year, $20 if retired, $5 if withdrawn, and $30 if dead; an active member receives no benefit. There are no further obligations after this year, so all destination-state reserves are zero. The effective annual interest rate is 5%. Assume the following one-year transition probabilities: from active, $p_{AA}=0.80$, $p_{AD}=0.10$, $p_{AR}=0.08$, $p_{AW}=0.01$, and $p_{AM}=0.01$; from disabled, the probabilities to $A,D,R,W,M$ are $0.10,0.70,0.10,0.05,0.05$; from retired, the probabilities to $R,M$ are $0.95,0.05$. These are finite-state Markov transition distributions for the year, including mortality through the transitions to $M$. Calculate the reserve in each starting state.

Method. Use the one-year prospective reserve formula, multiplying each destination benefit by its transition probability, summing, and discounting. The transition probabilities in each stated row sum to one.

Active state. The expected year-end benefit is $0.80(0)+0.10(8)+0.08(20)+0.01(5)+0.01(30)=0+0.80+1.60+0.05+0.30=2.75$. Thus $V_A=2.75/1.05=2.6190$, or $2.62 to the nearest cent.

Disabled state. The expected benefit is $0.10(0)+0.70(8)+0.10(20)+0.05(5)+0.05(30)=0+5.60+2.00+0.25+1.50=9.35$. Thus $V_D=9.35/1.05=8.9048$, or $8.90 to the nearest cent.

Retired state. The expected benefit is $0.95(20)+0.05(30)=19.00+1.50=20.50$. Thus $V_R=20.50/1.05=19.5238$, or $19.52 to the nearest cent.

The reserves differ because the starting states have different transition distributions. The retired-state reserve is not obtained by reusing the active-state transition probabilities. The calculation also makes the treatment of death explicit: the $30 death benefit is included only on transitions to $M$.

Annual Profit with Premiums, Expenses, Benefits, and Reserves

For an annual profit test, write all amounts from the insurer’s perspective. Let $V_s$ be the opening reserve for a member in state $s$, $C_s$ the contribution received at the start of the year, and $E_s$ the expense paid at the start of the year. At year-end, if the member moves to $j$, the insurer pays benefit $b_{sj}$ and holds closing reserve $V_j$. With an effective annual interest rate $i$, the conditional profit for that transition is $(V_s+C_s-E_s)(1+i)-b_{sj}-V_j$.

\Pi_{sj}=(V_s+C_s-E_s)(1+i)-b_{sj}-V_j, \qquad E[\Pi\mid S_t=s]=\sum_j p_{sj}\Pi_{sj}.

The profit definition uses a specific timing convention: opening reserve, contribution, and expense are combined at the start of the year and accumulate to year-end; benefits and closing reserves are measured at year-end. If a contribution is instead paid at year-end, it should not be accumulated for a full year. Profit on a particular transition is conditional profit; expected profit averages those conditional results using the transition probabilities.

Set the starting state and timing.
Identify the opening reserve, start-of-year contributions and expenses, interest rate, and year-end payment convention.

List destination states.
For each possible transition, record its probability, benefit, and closing reserve. Check that the probabilities sum to one.

Calculate conditional profit.
Accumulate the opening reserve and net start-of-year cash flow, then subtract the transition benefit and destination-state reserve.

Take the expectation.
Multiply each conditional profit by its transition probability and sum. Keep that expected result distinct from the profit on any one transition.

Worked Example: Integrated Profit from Five Active-Member Outcomes

A hypothetical pension contract has all amounts measured in thousands of dollars. At the start of the year, an active member has reserve $V_A=100$, pays a contribution of $12$, and incurs an expense of $2$. The effective annual interest rate is 5%. At year-end, the probabilities of retirement, remaining active, becoming disabled, withdrawal, and death are $0.10, 0.73, 0.10, 0.05,$ and $0.02$, respectively; these form a finite categorical transition distribution and sum to one. Assume no further mortality within the year beyond the stated death transition. For the respective outcomes, the pairs (benefit, closing reserve) are: retirement $(3,111.5)$, active $(0,113.5)$, disability $(4.5,114)$, withdrawal $(5.5,100)$, and death $(115.5,0)$. Calculate conditional and expected profit.

Method. First accumulate opening funds and the net contribution after expense. Then calculate profit separately for each destination state. All amounts, including the final answer, are in thousands of dollars.

The amount available at year-end before benefits and closing reserves is $(100+12-2)(1.05)=110(1.05)=115.5$.

Retirement: profit is $115.5-3-111.5=1.0$.
Remaining active: profit is $115.5-0-113.5=2.0$.
Disability: profit is $115.5-4.5-114=-3.0$.
Withdrawal: profit is $115.5-5.5-100=10.0$.
Death: profit is $115.5-115.5-0=0$.

The expected annual profit is $0.10(1.0)+0.73(2.0)+0.10(-3.0)+0.05(10.0)+0.02(0)$ $=0.10+1.46-0.30+0.50+0=1.76$ thousand dollars, or $1,760.

As a check, the expected year-end benefit-plus-reserve amount is $0.10(114.5)+0.73(113.5)+0.10(118.5)+0.05(105.5)+0.02(115.5)$ $=11.45+82.855+11.85+5.275+2.31=113.74$ thousand dollars. The accumulated amount $115.5$ less $113.74$ is $1.76$, agreeing with the probability-weighted conditional profits. The common mistake this check prevents is weighting benefits but forgetting to weight closing reserves as well.

From Conditional Results to an Expected Profit Signature

For a single member, the expected annual profit is the probability-weighted average across possible transitions. For a cohort, the same idea can be expressed as expected profit by destination state: multiply the number of members expected to reach a state by the conditional profit associated with that state. This distinction is useful because a profit can be positive on average even when some transitions produce losses, or vice versa. The expected result is not the profit that every member experiences.

Worked Example: Retirement Versus Continued Active Membership

Consider a hypothetical one-year pension projection in thousands of dollars. The starting active-member reserve is $90$; a contribution of $12$ and an expense of $2$ occur at the start of the year. The effective annual interest rate is 5%. At year-end the member either retires with probability $0.10$ or remains active with probability $0.90$. There are no disability, withdrawal, or death transitions during this interval, so the mortality probability for this interval is zero. The transition distribution is therefore a two-outcome categorical distribution. If retirement occurs, the benefit is $3$ and the closing reserve is $101$; if the member remains active, the benefit is zero and the closing reserve is $103$. Calculate the expected profit and explain why it is not the unweighted average of the two conditional profits.

The accumulated funds before year-end outgo are $(90+12-2)(1.05)=100(1.05)=105$ thousand dollars.

If the member retires, conditional profit is $105-3-101=1$ thousand dollars. If the member remains active, conditional profit is $105-0-103=2$ thousand dollars.

Weight each result by its probability: $0.10(1)+0.90(2)=0.10+1.80=1.90$ thousand dollars. The expected profit is therefore $1,900.

The unweighted average, $(1+2)/2=1.50$, would incorrectly assign probability $0.50$ to each outcome. Retirement has probability $0.10$, not $0.50$. The probability-weighted calculation also reconciles directly: expected year-end benefit plus reserve is $0.10(104)+0.90(103)=10.4+92.7=103.1$, and $105-103.1=1.9$ thousand dollars.

Common Errors in Integrated Pension Cases

The central discipline is to preserve the state, probability, cash-flow timing, and valuation basis attached to each amount. In particular, distinguish assumptions used to calculate transition probabilities from the reserve values assigned to destination states. A reserve can be calculated on a valuation basis that differs from the experience probabilities in the profit projection; if so, state that distinction rather than silently mixing bases.

Using the wrong starting-state transition row. Apply probabilities conditional on the member’s actual state at the start of the year. Retirement probabilities for active members are not the transition probabilities for members already retired.
Omitting a transition benefit or reserve. Each outcome contributes both its benefit and its closing reserve to year-end outgo. A zero benefit does not imply a zero closing reserve.
Mixing cash-flow times. Accumulate only amounts paid or received at the start of the year. A year-end benefit is not accumulated for an additional year before comparing it with a year-end reserve.
Confusing conditional and expected profit. A conditional result belongs to one transition. The expected result is the sum of conditional profits multiplied by their probabilities.
Failing to reconcile. Independently compare accumulated opening funds and net start-of-year cash flow with expected benefits plus closing reserves. A mismatch often reveals a sign, timing, or probability error.

Key takeaway: In an integrated pension profit test, calculate each state-specific reserve and each transition-specific profit on a clearly stated basis. Then use the transition probabilities to form expected profit. This method keeps retirement, disability, withdrawal, and death outcomes distinct while providing a direct reconciliation check.

Check Your Understanding

Use the stated annual timing convention in each problem. Unless otherwise stated, benefits and closing reserves are measured at year-end, and contributions and expenses are paid at the start of the year.

A one-year liability has an effective annual interest rate of 4%. From active state, the probabilities of ending active, disabled, or retired are $0.70,0.20,0.10$. The corresponding year-end benefit-plus-closing-reserve amounts are $4, $16, and $45. Calculate the active-state reserve. (Answer: $10.10.)
An active member has opening reserve $80, start-of-year contribution $9, expense $1, and an effective annual interest rate of 5%. The year-end benefit-plus-closing-reserve amount is $30 with probability $0.20$ and $88 with probability $0.80$. Calculate expected annual profit. (Answer: $16.00.)
In a two-outcome pension projection, retirement has probability $0.10$ and conditional profit $1.00 thousand; continued active membership has probability $0.90$ and conditional profit $2.00 thousand. Calculate expected profit. (Answer: $1.90 thousand.)
A retired member’s one-year transition probabilities are $0.90$ to remain retired and $0.10$ to die. The year-end benefit is $18 if retired and $28 if dead; the closing reserve is $40 if retired and zero if dead. At an effective annual interest rate of 5%, calculate the reserve at the start of the year. (Answer: $52.38.)
A cohort has 200 active members. In a one-year projection, each has expected profit of $1.90 thousand. Calculate the cohort’s expected total profit, without interpreting it as a guaranteed outcome. (Answer: $380,000.)

What You'll Learn