Introduction
A member who works beyond a plan’s normal retirement age may receive a larger pension for two separate reasons: additional credited service may increase the pension earned, and the plan may increase the pension because its commencement is postponed. These adjustments are not interchangeable. A plan can provide either, both, or neither, and the amount depends on its specific terms.
The key calculation is to identify which benefit has been earned by normal retirement age, determine whether work after that date adds further accrual, apply any limit on that accrual, and then apply the plan’s rule for delayed commencement. The resulting annual pension is distinct from its actuarial value: the value also depends on interest, mortality, payment timing, and whether payments are conditional on survival.
Representing a Late-Retirement Rule
A benefit formula should show the accrual and postponement components separately. For example, suppose the annual pension at age \(x\) is determined by an accrual rate \(r\), pensionable service \(s_x\), and a final-average-pay measure \(F_x\). Before applying any late-commencement adjustment, the formula is
This formula is only an illustration of a plan design; it does not imply that every plan uses final-average pay or a constant accrual rate. If the plan limits credited service to \(s_{\max}\), then the service used in this formula may be \(s_x=\min(s_x^{\text{actual}},s_{\max})\). A stated limit can apply to service, the pension amount, or some other measure, so it must be read as written rather than assumed.
If the plan applies a postponement factor \(g(t,R)\), the late pension may be represented as \(B_t=B_t^{\text{accrued}}g(t,R)\). This representation is appropriate only if the plan applies its postponement factor to the full accrued pension. Some plans instead apply it only to the pension earned at \(R\), or specify a different coordination rule for new accrual. The benefit formula must reflect that distinction.
Worked Example: Continued Accrual With a Service Limit
Consider a hypothetical final-average-pay plan with normal retirement age 65. Its annual pension is 1.5% of final-average pay for each year of credited service, subject to a maximum of 30 years. A member has 27 years of service at age 65, works until age 68, and commences the pension then. The three salary amounts used in the age-65 final average are $72,000, $75,000, and $78,000; the three amounts used in the age-68 final average are $78,000, $81,000, and $84,000. There is no separate postponement increase.
For this example’s valuation, assume an effective annual interest rate \(i=5\%\). At each age, annual death probability is \(q=0.10\), independently from year to year, so the one-year survival probability is \(p=0.90\). Payments form an annual life annuity-due, with the first payment at commencement if the member is alive. Thus the number of years survived has a geometric survival pattern, \({}_kp_x=0.9^k\).
Step 1: Find the age-65 final average.
The pension earned at age 65 is \(0.015(27)(75{,}000)=\$30{,}375\) per year.
Step 2: Apply the late-service limit. Three additional years of work would bring actual service to \(27+3=30\) years, which is exactly the plan maximum. Thus credited service at age 68 is \(\min(30,30)=30\) years.
Step 3: Find the age-68 final average and annual pension.
The annual pension commencing at age 68 is therefore $36,450. The increase from $30,375 reflects both the plan’s revised final-average pay and three more credited years; no separate postponement factor applies.
Step 4: Value the pension at commencement. Here \(v=1/1.05=20/21\), so \(vp=6/7\). For an annual life annuity-due under these assumptions, the value per dollar of annual pension is \(\sum_{k=0}^{\infty}(vp)^k=1/(1-vp)=7\). At age 68, the conditional value is \(36{,}450(7)=\$255{,}150\).
Check. The service used is 30, not 27 or 31, and multiplying the annual pension by the annuity factor gives $255,150. A common mistake is to add a further service year or apply an extra postponement increase even though the stated plan has no such increase.
Plan Increases and Actuarial Equivalence
A plan may specify a fixed increase for each year of postponement. For example, it might increase a pension already earned at \(R\) by a stated compound percentage for each complete year deferred. That contractual increase is a plan rule; it need not make the deferred pension actuarially equivalent to a pension commencing at \(R\).
To see the distinction, consider a hypothetical annual life annuity-due with constant annual mortality and interest. Let \(p\) be the probability of surviving one year and \(i\) the effective annual interest rate. Assume mortality is independent between years and independent of interest. Conditional on being alive at commencement, the expected present value at that date of one dollar per year is
provided \(vp<1\). If commencement is postponed \(n\) whole years, the value at age \(R\) of a pension \(B_n\) beginning at age \(R+n\) is \((vp)^n B_n\ddot{a}\): the member must survive the \(n\)-year deferral, and each year of discounting and survival contributes a factor \(vp\). The age-\(R\) value of an immediate pension \(B_0\) is \(B_0\ddot{a}\). Equating these values gives the actuarially equivalent annual pension:
This result depends on the stated assumptions and payment timing. A different mortality pattern, annuity form, or payment convention can produce a different factor. In particular, an increase specified by a plan should not be replaced by this equivalence formula unless the plan calls for actuarial equivalence on that basis.
Worked Example: Applying a Stated Postponement Increase
A hypothetical plan provides an annual pension of $30,000 at normal retirement age 65. If commencement is postponed, the plan increases that pension by 5% compounded for each complete year deferred. The member defers for two years, and there is no additional service accrual. For valuation, assume \(i=4\%\), constant annual death probability \(q=0.08\), and independent annual survival, so \(p=0.92\). The pension is an annual life annuity-due.
Step 1: Apply the plan’s stated increase. The rule is compound, so the factor is \(1.05^2\), not \(1+2(0.05)\).
The annual pension beginning at age 67 is $33,075.
Step 2: Calculate its value at age 67. Here \(vp=0.92/1.04=23/26\), so \(\ddot{a}=1/(1-23/26)=26/3\). The value conditional on being alive at age 67 is \(33{,}075(26/3)=\$286{,}650\).
Step 3: Calculate its prospective value at age 65. The probability-and-discount factor for surviving and discounting two years is \((23/26)^2\). Therefore the value at age 65 is
The calculation rounds to the cent only at the final step. For comparison, the age-65 value of immediate commencement is \(30{,}000(26/3)=\$260{,}000\). The stipulated 5% annual plan increase does not offset the full value of the two-year deferral under the assumed interest and mortality basis. This does not make the plan rule incorrect; it shows why a plan increase and actuarial equivalence are different concepts.
Check. The postponement factor was applied to the pension earned at age 65, with no new accrual. A common error is to call the 5% increase actuarially equivalent without testing it against the interest and mortality assumptions.
Worked Example: Finding an Actuarially Equivalent Increase
A hypothetical plan asks for a deferred annual pension that is actuarially equivalent to an immediate $20,000 annual pension at age 65. The deferred pension begins at age 67. Assume effective annual interest \(i=4\%\), constant annual death probability \(q=0.05\), independent annual survival, and annual annuity-due payments. Thus \(p=0.95\), and the number of years survived follows the corresponding geometric survival pattern.
Step 1: Set up the values at age 65. The immediate pension has value \(20{,}000\ddot{a}\), where
If the age-67 annual pension is \(B_2\), its age-65 value is \((0.95/1.04)^2B_2\ddot{a}\). Equating the values cancels the same annuity factor from both sides.
Step 2: Solve for the deferred pension.
The actuarially equivalent annual pension commencing at age 67 is $23,968.98, rounded to the nearest cent. Its age-65 value is \(20{,}000(104/9)=\$231{,}111.11\). Substitution into the deferred value gives the same amount, subject only to rounding.
Check. The increase reflects both interest and the probability of surviving to deferred commencement. Omitting mortality would give the interest-only amount \(20{,}000(1.04)^2=\$21{,}632\), which is not equivalent under the stated mortality assumptions.
Coordinating Accrual, Postponement, and Limits
When a member works after normal retirement age, write down the plan’s sequence of operations before calculating. A useful procedure is:
Calculate the benefit earned through service and pay recognized at \(R\), using the plan’s definition of each.
Apply the plan’s post-\(R\) accrual rule and any limit on service or benefit. Do not assume additional years count merely because the member remains employed.
Check which pension amount receives the increase, whether the increase is simple or compound, and how partial years are treated.
Use the stated interest, mortality, survival condition, and payment timing. Keep this valuation separate from the calculation of the plan benefit.
A common exam trap is to combine these stages prematurely. Suppose a plan adds pension for service after \(R\) and separately increases the pension earned at \(R\) for deferred commencement. Applying the postponement factor to the whole final pension may overstate the result if the plan specifies that the new accrual is excluded. Conversely, applying the increase only to the age-\(R\) pension may understate the result if the plan explicitly applies it to the entire accrued amount at commencement.
For a hypothetical plan that increases only the age-\(R\) pension, the representation is \(B_t=B_Rg(t,R)+\Delta B_{\text{service}}\). Here \(\Delta B_{\text{service}}\) is the additional pension earned after \(R\), subject to the plan’s limits. This expression is not a universal rule: it is valid only when the plan says that new accrual is not itself increased for postponement. If the plan gives a different coordination rule, use that rule instead.
Check Your Understanding
For each hypothetical plan, follow the stated rule exactly. Answers are rounded to the nearest cent where needed.
- A member has an annual pension of $24,000 at normal retirement age. The plan applies a 3% compound increase for each complete year of postponement. What annual pension is payable after two years, assuming no further accrual? (Answer: $25,461.60)
- A plan provides 1.25% of final-average pay per credited year, with a maximum of 30 years. A member has 28 years at normal retirement age, works two more years, and has final-average pay of $82,000 at commencement. What annual pension does the formula produce? (Answer: $30,750)
- A $25,000 annual pension at age 65 is deferred one year. Find the actuarially equivalent annual pension at age 66 under annual annuity-due payments, effective annual interest \(i=3\%\), and constant annual death probability \(q=0.05\), with independent annual survival. (Answer: $27,105.26)
- A plan applies a 4% simple increase for each year deferred to the $22,000 pension earned at normal retirement age. It separately adds $1,200 of annual pension for each of three later service years, and does not increase that additional accrual. Find the annual pension at commencement after three years. (Answer: $28,240)
- A $15,000 annual pension at age 65 is deferred for two years. What annual pension is actuarially equivalent under annual annuity-due payments, effective annual interest \(i=5\%\), and constant annual death probability \(q=0.10\), with independent annual survival? (Answer: $20,416.67)