Smoothing Surplus Across Years

Why Smooth Surplus Across Years?

Annual surplus available for allocation can vary substantially as experience changes. A participating insurer may choose to distribute less than the available amount in a strong year and use the resulting balance to support allocations in a weaker year. The purpose is to moderate year-to-year changes in allocations, not to create additional surplus. Because the balance is carried from one year to another, smoothing also affects which policyholder groups receive amounts arising from different years’ experience.

The analysis below treats annual surplus as already determined and available for allocation. It does not revisit how surplus emerges or how it is divided between policyholders and shareholders. All examples are hypothetical. Unless stated otherwise, amounts are in thousands of dollars, there is no interest credited to the smoothing account, and no mortality or decrements are modeled.

Definition: Let $E_t$ be the surplus available for allocation in year $t$, before any smoothing transfer. Let $C_t$ be the amount retained from that year, $L_t$ the amount released from amounts retained earlier, $D_t$ the amount allocated in year $t$, and $H_t$ the smoothing-account balance at the end of year $t$. With no interest, $$ D_t=E_t-C_t+L_t,\qquad H_t=H_{t-1}+C_t-L_t. $$ Here $C_t$ and $L_t$ are nonnegative, and a release cannot exceed the balance available before release. The account is an accounting record of the timing of allocations; it is not itself new surplus.

These equations describe the basic transfer mechanics. They do not specify who is entitled to a release, whether a target allocation is appropriate, or whether a proposed transfer is consistent with contractual participation rights. Those questions depend on the applicable allocation rules and the groups covered by them.

Worked Example: Equalizing Two Annual Allocations

A hypothetical participating block has $120,000 of surplus available in Year 1 and $40,000 in Year 2. Management considers a target allocation of $80,000 in each year. Assume no interest, no other allocations, and that the amounts are all available for policyholder allocation. Calculate the retained amount, the release, each annual allocation, and the closing account balance.

Method. In Year 1, compare available surplus with the target. The amount above the target is retained: $$ C_1=120-80=40,\qquad L_1=0. $$ Thus the Year 1 allocation and closing balance are $$ D_1=120-40=80,\qquad H_1=0+40-0=40. $$

In Year 2, the target exceeds current surplus by $80-40=40$. The account has $40,000 available, so the full shortfall can be released: $$ C_2=0,\qquad L_2=\min(40,40)=40. $$ Therefore, $$ D_2=40-0+40=80,\qquad H_2=40+0-40=0. $$

Check. Total surplus over the two years is $120+40=160$, and total allocation is $80+80=160$. The closing balance is zero, so no amount remains undistributed. The common mistake is to treat the Year 1 retained $40,000 as an additional Year 2 surplus: it is already part of the original $160,000 and must be counted only once.

Account Roll-Forward and Conservation

The account equations yield a useful reconciliation. Summing the annual allocation equation through year $n$, and using $H_0=0$, gives:

\sum_{t=1}^{n}D_t = \sum_{t=1}^{n}E_t-H_n.

This follows because each amount retained is either still in the closing balance or has been released and included in an allocation. In particular, if the balance is fully released by the end of the period, cumulative allocations equal cumulative surplus available for allocation. If a balance remains, cumulative allocations are lower by exactly that amount.

If the smoothing account earns interest at effective annual rate $i$, the timing convention must be stated. For example, if the opening balance earns interest during year $t$, before current-year contributions and releases, the roll-forward is $$ H_t=(1+i)H_{t-1}+C_t-L_t. $$ The surplus and allocation amounts must then be measured consistently with that convention. The examples here use $i=0$, so the simpler roll-forward applies.

Why Weighted Smoothing Reduces Annual Variance

A useful way to study volatility is to represent an allocation as a weighted combination of current and prior annual surplus. Suppose, for illustration, that $$ D_t=(1-\alpha)E_t+\alpha E_{t-1}, $$ where $0\leq\alpha\leq1$. A larger $\alpha$ places more weight on the prior year. This is a stylized smoothing rule; actual use requires a feasible account balance and an explicit treatment of the first and final years.

Assume annual amounts $E_t$ have the same variance $\sigma^2$, and that $E_t$ and $E_{t-1}$ are independent. By the variance formula for a weighted sum of independent random variables,

\operatorname{Var}(D_t) =(1-\alpha)^2\operatorname{Var}(E_t) +\alpha^2\operatorname{Var}(E_{t-1}) =\big((1-\alpha)^2+\alpha^2\big)\sigma^2.

The variance factor is less than one when $0<\alpha<1$, because $$ 1-\big((1-\alpha)^2+\alpha^2\big)=2\alpha(1-\alpha)>0. $$ Thus this weighted rule reduces the variance of an individual year’s allocation under the stated assumptions. It does not eliminate variability, and allocations in adjacent years are no longer independent: both can reflect some of the same annual surplus. The result is about annual volatility, not a guarantee that every policyholder group receives a more stable amount.

Worked Example: Measuring the Effect of a Weighted Rule

Suppose annual surplus amounts are independent and identically distributed as $E_t\sim N(100,400)$, where amounts are in thousands of dollars. Thus the mean is $100,000 and the standard deviation is $20,000. A hypothetical smoothing rule uses $\alpha=0.25$, so $D_t=0.75E_t+0.25E_{t-1}$. Assume no interest and sufficient retained amounts for this illustrative weighted calculation. Find the mean and standard deviation of the smoothed allocation, and calculate the allocation when $E_{t-1}=80$ and $E_t=120$.

Method. Linearity of expectation gives $$ E[D_t]=0.75(100)+0.25(100)=100. $$ Independence allows the variance terms to be added: $$ \operatorname{Var}(D_t) =0.75^2(400)+0.25^2(400) =0.5625(400)+0.0625(400) =225+25=250. $$ The standard deviation is $\sqrt{250}=15.811$ thousand dollars, or $15,811.39 to the nearest cent. Compared with the unsmoothed standard deviation of $20,000, the reduction is $20-15.811=4.189$ thousand dollars, or $4.189/20=20.945\%$, approximately 20.95%.

For the stated observed amounts, $$ D_t=0.75(120)+0.25(80)=90+20=110, $$ so the allocation is $110,000. The method uses the variance of a weighted sum, not a weighted average of standard deviations. The check is that the expected allocation remains $100,000, while its variance is $250<400$, as the smoothing result predicts.

Aggregate Stability and Policyholder Groups

A stable total allocation does not necessarily mean stable benefits per policy or equal treatment across policy-year groups. If the number of eligible policies changes, the same aggregate amount can translate into different per-policy amounts. In addition, when a release is funded by a prior year’s retained surplus, the group receiving the release may differ from the group whose year generated the retained amount.

A comparison should therefore distinguish at least three quantities: the surplus emerging in each year, the amount allocated in each year, and the allocation per eligible policy or unit of participation. The transfer rule should also identify which groups have a claim on retained amounts. A smooth aggregate series alone cannot establish that the allocation is fair.

Worked Example: A Smoother Total Can Mean Different Per-Policy Outcomes

A hypothetical block has $120,000 available in Year 1 and $40,000 in Year 2. There are 200 eligible policies in Year 1 and 100 in Year 2. Use the $80,000 annual target rule from the first example, with no interest. Compare the allocation per eligible policy with the amounts that would have been allocated without smoothing. Assume the groups are distinct and that the scenario is an arithmetic comparison, not a conclusion about their contractual entitlements.

Without smoothing, Year 1 allocation per policy is $$ 120{,}000/200=\$600, $$ and Year 2 allocation per policy is $$ 40{,}000/100=\$400. $$ With smoothing, Year 1 retains $120-80=40$ thousand dollars and allocates $80,000. Its per-policy allocation is $$ 80{,}000/200=\$400. $$ Year 2 releases $40,000, bringing its allocation to $80,000. Its per-policy allocation is $$ 80{,}000/100=\$800. $$

The aggregate allocations are now equal at $80,000 in each year, but the per-policy amounts are $400 and $800. In this example, smoothing the total allocation increases the difference between the two groups’ per-policy amounts, from $200 to $400. The check is that combined allocations remain $80{,}000+80{,}000=160{,}000$, equal to combined surplus. The key caution is not to infer individual or group stability from aggregate stability.

Common Exam Trap: Confusing a Release with New Experience

Suppose an account opens a year with $30,000, current surplus is $50,000, and the intended allocation is $70,000. The shortfall is $20,000, so a $20,000 release is sufficient: $$ L_t=\min(30,70-50)=20,\qquad D_t=50+20=70. $$ The closing balance is $30-20=10$ thousand dollars.

A wrong calculation would report $D_t=50+30=80$, releasing the entire opening balance simply because it is available. That answer misses the stated $70,000 target and over-allocates by $10,000. The correction is to release only the amount required by the rule, subject to the available balance. A release is a transfer from an existing account balance, not a second measure of current-year experience.

Key takeaway: Smoothing shifts allocation timing across years. Reconcile every retention and release to the account balance, measure volatility at the level that matters (aggregate, per policy, or group), and assess cross-year transfers under the applicable participation rules. Smoothing alone does not establish fairness or financial adequacy.

Check Your Understanding

Use no interest unless a question states otherwise. Treat all amounts as hypothetical.

A year has $90,000 available and a target allocation of $60,000. In the next year, $30,000 is available and the full prior balance can be released. Find both allocations and the ending balance. (Answer: $60,000 and $60,000; ending balance $0.)
Annual surpluses are independent with variance $16$ thousand-dollars-squared. Under $D_t=0.5E_t+0.5E_{t-1}$, find the variance of the smoothed allocation. (Answer: $8$ thousand-dollars-squared.)
An account opens with $25,000, current surplus is $45,000, and the target allocation is $60,000. Find the release, allocation, and closing balance. (Answer: release $15,000; allocation $60,000; closing balance $10,000.)
Surplus is $96,000 in a year with 160 eligible policies and $24,000 in the next year with 80 eligible policies. A $60,000 target is used in both years, with sufficient balance for the release. Find smoothed per-policy allocations in each year. (Answer: Year 1 retains $36,000 and allocates $60,000, or $375 per policy; Year 2 releases $36,000 and allocates $60,000, or $750 per policy.)
Suppose $E_t\sim N(50,100)$ independently across years, and $D_t=0.8E_t+0.2E_{t-1}$. Find the expected smoothed allocation and its variance. (Answer: mean $50$; variance $68$, in the corresponding squared monetary units.)

What You'll Learn