Background: The length of retirement is one of the most important assumptions in a financial plan. For married couples, life expectancy calculations typically assume that

mortality rates are not related (i.e., independent). This common assumption ignores the well-documented fact that mortality rates typically increase for the surviving spouse after spousal loss, something known as the “widowhood effect.” In a recently published research in the Retirement Management Journal, I explored the existence of the widowhood effect, leveraging data from the Health and Retirement Study (HRS), and the implications of the observed widowhood effect on retirement projections.

Findings: There is clear evidence of the widowhood effect in the HRS, where mortality rates among surviving spouses increase by approximately 20%, consistent with past research on the topic. Although noted the increase in mortality rates was statistically significant (demonstrated via a series of regressions included in the paper), it’s not necessarily economically significant.

I demonstrate this by estimating various retirement planning periods that incorporate the widowhood effect to varying degrees, based on mortality rates from the Society of Actuaries (SOA) 2012 Immediate Annuity Mortality table. The “independent” model assumes mortality rates for the surviving spouse are not related (i.e., independent), effectively ignoring the widowhood effect. The other two approaches reflect different adjustments to the surviving spouse's mortality rates: a 25% increase (consistent with the general findings) and a 100% increase (a relatively extreme potential change).

Source: “My Heart Will Go On: The Retirement Planning Implications of the Widowhood Effect” by David Blanchett, 2025

Increasing the mortality rate for the surviving spouse decreases the estimated retirement period, as expected, but the effect is relatively muted for reasonable mortality rate increases (25%) and reasonable retirement periods, e.g., targeting a 25% probability of outliving the retirement period.

For example, if mortality rates for surviving spouses are assumed to be independent, i.e., there is no widowhood effect, if the target probability of outliving was 25%, the retirement period would be assumed to last 34 years. Increasing mortality rates by 25%,consistent with the past noted effect, only decreases the assumed retirement period by one year (to 33 years). The widowhood effect would be more important to include if

spousal survivor mortality rates were notably higher (e.g., a 100% increase), but this is not the case.

Bottom line: Overall, the analysis suggests that while the widowhood effect exists, including it in longevity estimates at generally observed levels is unlikely to materially affect a retirement-income plan, especially when considering the underlying uncertainty around all the errors inherent in the forecast. Therefore, assuming mortality rates for spouses are independent is likely a reasonable simplifying assumption in a retirement income plan.

Insights shared by guest contributors are their own and do not represent the views of DCIIA or the RRC. The opinions voiced in this material are for general information only and are not intended to provide specific advice or recommendations for any individual.