The authors also looked at the number of years that the 4 percent spending pattern was sustained. These figures give us one angle on the magnitude of the failure rather than just the frequency.

Ninety-five percent of the simulations using a constant 60 percent equity allocation lasted more than 27.7 years, but the 30 percent to 60 percent glidepath lasted the full 30 years, says Nerd’s Eye View. Kitces and Pfau also ran simulations using reduced-return expectations to reflect the popular notion that equity and bond returns would be below historic averages going forward. This, too, resulted in about a 2½-year increase in the life of the portfolio at a 95 percent confidence level.

Increasing equity exposure may seem counterintuitive at first blush, but the more one considers the issue, the more sense it makes. If a retiree starts off with good markets this is all mute, of course, because they should be fine.

By reducing risk at the beginning of the retirement, less damage is done in a bad market and the more time is provided to benefit from market recovery. Further, the  longer the market struggles, the more stocks are bought at the relatively lower valuations and the better the performance when markets advance. In essence, increasing equity exposure over time during an early retirement phase marred by poor market conditions can bear some mathematic similarity to dollar-cost averaging.

As the authors put it, it’s a matter of heads you win, tails you don’t lose.

Clearly, having elderly clients who are aggressively weighted in equities can be problematic, so what are planners supposed to do with this information? I think there are a few takeaways.

First and foremost, stocks will need to remain a significant portion of the portfolios of most retirees. With interest rates low, it is not particularly difficult to convey this message.

Second, it should still be considered unwise to be 100 percent out of equities or 100 percent in equities. A range of “balanced” portfolios is likely to remain a prudent choice. Combinations that started with 20 percent to 40 percent equities and finished the 30-year period between 60 percent and 80 percent seem to provide similar results on the frequency and magnitude of failure. The portfolio values at the end of 30 years were predictably much more varied. 

Third, if a rising equity glidepath has a positive dollar-cost averaging-like effect, it seems logical that a declining glidepath would have a negative reverse dollar-cost averaging effect. Indeed, many of the results show the rising equity glidepath beats the constant allocation method, which beats reducing equity exposure gradually as a client ages. Most media coverage I have seen about the study give the impression that the rising glidepath is clearly superior. I disagree.

Reducing the equity allocation over time does not appear to be as damaging as one might assume from the perspective of frequency and magnitude of failure. Starting at 60 percent and reducing to 30 percent produces an identical historical success rate as sticking with 60 percent equities throughout retirement. At the fifth percentile for the number of years supported, the declining path actually adds 1/10 of a year based on the historic record (Nerd’s Eye View).

Fourth, I’m not ready to implement a rising glidepath because the data presents important contradictions to ponder. For instance, the best worst-case shortfall in the “Lower Future Returns” and 4 percent withdrawal rate scenario (Table 4) resulted from gliding from 20 percent to 70 percent equities; however, the success rate applicable to that approach was only 54 percent. All but one combination of starting and ending allocations that began with at least 50 percent equities, whether rising or declining from there, had higher success rates. Even a constant 40 percent allocation succeeded 58 percent of the time.