Why Overlapping Benefits Need Careful Rules
Ancillary benefits can be described by separate formulas: a refund on withdrawal, a vested pension after sufficient service, and a survivor benefit following death. The formulas are not automatically separate payments. A plan may make one benefit an alternative to another, or it may permit benefits at different times. Valuing each formula independently and then adding the results can count the same event or entitlement twice.
The safe approach is to begin with the plan’s eligibility and payment rules. For each possible event path, determine which benefits are payable, when they are payable, and whether any benefit is conditional on another not being payable. Only then calculate the expected present value.
Represent the Payment Rules Before Valuing
A useful way to make the rules precise is to split possible outcomes into mutually exclusive paths. For example, at a particular exit date, the member may withdraw, die, retire, or remain in the plan. These outcomes cannot all happen at that date. Death can then be split into further mutually exclusive categories, such as death with an eligible spouse and death without one.
For each path, write down the actual benefit payable. If the plan says “refund if unvested; vested pension if vested,” the refund and pension are alternatives, not two payments for the same withdrawal. If the plan says a survivor pension is payable instead of a refund on death, the death path must be split according to survivor eligibility. A refund should not also be assigned to the eligible-spouse path unless the rules explicitly provide both.
Once the paths are disjoint, the expected present value can be calculated by multiplying each path’s probability by the present value of the payment or payments on that path, then adding across paths. If \(E_j\) are mutually exclusive paths and \(B_j\) is the present value of all payments on path \(j\), then:
This expression requires a complete set of paths for the outcomes being valued, with no path counted twice. A benefit’s value on a path may itself include several payments if the plan genuinely pays them all.
Worked Valuations
Worked Example: Refund or Vested Withdrawal Benefit
A hypothetical plan values a member’s benefits at the end of one year. The member has already met the service condition for vesting. During the year, the probabilities of withdrawal, death, and retirement are 0.20, 0.05, and 0.10, respectively; the member remains active with probability 0.65. These outcomes are mutually exclusive and exhaustive. At year-end, the plan pays a $24,000 vested benefit on withdrawal, a $18,000 survivor-benefit value on death, and a $36,000 pension value on retirement. The $8,000 refund is payable only to an unvested withdrawing member. Find the expected present value at time 0, using an effective annual interest rate \(i=0.05\).
Method. Assign the plan-permitted benefit to each event path first. Because the member is vested, the withdrawal path receives the vested benefit, not the unvested refund. The member’s one-year discount factor is:
The expected benefit payable at the end of the year is:
The remaining probability, 0.65, corresponds to remaining active; this one-year valuation assigns no exit benefit to that path. Discounting the expected year-end amount gives:
Check. The path probabilities sum to \(0.20+0.05+0.10+0.65=1.00\). As an exam trap, adding the refund to the withdrawal payment would add \(0.20(8{,}000)/1.05=\$1{,}523.81\) to the answer. That is wrong here: the member is vested, and the refund is expressly restricted to unvested members.
Worked Example: A Survivor Pension Already Included in the Payment Stream
Consider a hypothetical pension starting immediately at retirement. It pays $10,000 at each year-end while the member is alive. If the member has died by a payment date and the spouse is alive, the plan instead pays the spouse $6,000 at that date. The member’s annual death probability is 0.10 and the spouse’s annual death probability is 0.05. Assume independent mortality from year to year and between the two lives; deaths during each year occur before that year-end payment. Use \(i=0.05\), and value only the first two annual payments.
Method. At each payment date, identify the state that triggers each payment. The member’s probability of being alive at the end of year 1 is \(0.90\). The probability that the member is dead and the spouse alive then is \((0.10)(0.95)=0.095\). Thus the expected first payment is:
At the end of year 2, the member’s survival probability is \(0.90^2=0.81\). The probability the member has died by then is \(1-0.81=0.19\). The spouse’s probability of surviving both years is \(0.95^2=0.9025\). By independence, the probability of a survivor payment at the second date is:
The expected second payment is therefore:
Discount the two expected payments separately. The first-year discount factor is \(v=1/1.05\), and the second-year factor is \(v^2=1/1.05^2\):
Check. The year-2 survivor probability includes members who died in either year and spouses who are alive at the year-2 payment date. The survivor payments are already included in the expected payments above. Adding a separately calculated survivor-pension value for those same dates would count them twice. This example values only two payments; it does not assume the contract ends after year 2.
Worked Example: Split Death Paths by Survivor Eligibility
A hypothetical one-year plan has mutually exclusive probabilities of 0.08 for death with an eligible spouse, 0.02 for death without an eligible spouse, 0.10 for withdrawal, 0.05 for retirement, and 0.75 for remaining active. These probabilities sum to 1.00. At the end of the year, the eligible-spouse death path receives a survivor-pension value of $14,000; the no-spouse death path receives a $9,000 refund; withdrawal receives a $22,000 vested benefit; and retirement receives a $30,000 pension value. The survivor pension and refund are alternatives on death, not cumulative benefits. Find the expected present value at \(i=0.05\).
Method. Treat death with an eligible spouse and death without an eligible spouse as separate, mutually exclusive paths. The expected year-end value is:
The active path has no payment at this valuation date. Discount the year-end expected value:
Check. The total death probability is \(0.08+0.02=0.10\), but the two death categories have different benefits. Incorrectly adding a $9,000 refund to every death, including eligible-spouse deaths, would add \(0.08(9{,}000)/1.05=\$685.71\), producing a false value of $5,447.62. The correction is to apply the refund only to the no-spouse path.
A Short Justification: Why the Path Sum Works
Let \(Z\) be the present value of all payments under the plan for the period being valued. Suppose the event paths \(E_1,\ldots,E_n\) are mutually exclusive and cover every outcome, and let \(B_j\) be the present value of the payments that actually occur on path \(E_j\). On that path, \(Z=B_j\). In indicator notation, this is:
where \(\mathbf{1}_{E_j}\) equals 1 if path \(E_j\) occurs and 0 otherwise. Taking expectations and using linearity of expectation gives:
The paths need to be mutually exclusive for each outcome to receive one correct path value \(B_j\). If instead overlapping labels are used—for example, “death” and “death with eligible spouse”—summing both labels as separate paths may assign two benefits to the same outcome. Refine the paths until each outcome has one applicable benefit rule. If two payments truly are both payable on a path, include both in that path’s \(B_j\), with their own payment dates and discount factors.
Practical Audit and Common Traps
A compact audit table often exposes an overlap before any arithmetic is done. For each path, record its probability, eligibility condition, benefit payable, and payment time. Verify that the paths are exhaustive for the outcomes being valued and that their probabilities sum to 1. Then check that every formula has been assigned only to paths where its trigger is satisfied.
List exit causes and, where needed, split an exit by eligibility such as vesting or spouse status.
Determine whether each benefit is an alternative, a supplement, or a later payment. Do not infer additivity from the fact that the plan describes benefits separately.
Write down every payment permitted on each path, and discount each amount from its actual payment time.
Check probabilities, eligibility, and payment triggers. Confirm that no payment has been included in two path values or added again as a separate benefit.
Several distinctions matter. A refund and a vested pension may be mutually exclusive at withdrawal, while a vested pension paid during the member’s life and a survivor pension paid after death may both be payable sequentially. A refund might be a top-up rather than a substitute if the plan explicitly says so; in that case the top-up amount and its trigger must be valued as written. Similarly, an eligibility condition should be applied to the correct event date and member status. If the terms leave the relationship between benefits unclear, a valuation should state the interpretation rather than silently assume that all listed benefits are cumulative.
These checks are especially useful in multi-decrement valuations. A benefit attached to one cause of exit must not be charged again under another cause, and conditional eligibility must be reflected in the event probabilities or path amounts. Keeping the event partition separate from the benefit formula makes the assumptions visible and makes the result easier to verify. This method is aligned with the SOA Exam ALTAM (Advanced Long-Term Actuarial Mathematics) syllabus.
Check Your Understanding
For each problem, identify the applicable path or payment rule before calculating.
- A hypothetical one-year plan has a 0.25 probability of vested withdrawal, paying $16,000, and a 0.10 probability of retirement, paying $25,000. These are mutually exclusive; all other outcomes pay nothing during the year. Find the expected present value at \(i=0.04\). (Answer: $6,250.00)
- At year-end, death with an eligible spouse has probability 0.04 and pays a survivor-benefit value of $12,000. Death without an eligible spouse has probability 0.02 and pays a $7,000 refund instead. The paths are mutually exclusive. Find the expected present value at \(i=0.05\). (Answer: $590.48)
- A two-year pension valuation includes the survivor payments in its state-by-state expected cash flows. What additional survivor value should be added for those same dates? (Answer: $0; those payments are already included.)
- At the end of one year, mutually exclusive probabilities are 0.12 for withdrawal with a $20,000 vested benefit, 0.03 for death with an eligible spouse and a $15,000 survivor-benefit value, 0.02 for death without an eligible spouse and an $8,000 refund, and 0.08 for retirement with a $28,000 pension value. All other paths pay nothing at that date. Find the expected present value at \(i=0.05\). (Answer: $5,000.00)
- A plan pays a vested withdrawal benefit of $18,000 if the member is vested and a $6,000 refund if the member is not vested; the two payments are alternatives. At the valuation date, mutually exclusive withdrawal paths have probabilities 0.15 vested and 0.05 unvested. Find the expected present value at \(i=0.04\). (Answer: $3,000.00)