The golden years are supposed to be about kicking back taking it easy. But it's hard to take it easy if you run out of money. How do you manage a portfolio that gives the best shot at making your money last? It turns out there are no easy answers.
The conventional wisdom of retirement portfolios is to gradually reallocate away from stocks with age. Imagine a target-date fund whose equity portion continues to sink after retirement. This conventional wisdom fits well with the idea that we tend to become more conservative with age, and that the consequences of a market crash aren't something a 90-year-old should bear.
Unfortunately, conventional wisdom doesn't jibe with simulations that use the traditional retirement shortfall risk methodology. New research by Michael Kitces, partner and research director at Pinnacle Advisors, and American College professor Wade Pfau finds convincing evidence that a retiree has a better chance of making it to the finish line at age 95 with money in the bank by increasing their allocation to stocks as they age.
I'm a sucker for provocative research that questions the status quo. This research certainly has provoked. "Sure. Why not. Lets allocate 90-year old grandma to 90% stocks," tweeted Moshe Milevsky, York University professor and Research contributor, after hearing the results.
Kitces responded that he and Pfau did not actually find that grannies should be allocating 90% of their retirement portfolios to stock (results suggest a more modest rising glide path from about 30% to 60% in later years). In fact, Kitces points out that the flip side of a rising glide path is lower equity exposure than many planners recommend—particularly early in retirement.
The key to rising glide paths is the big impact that negative returns can have on an early retirement portfolio, or the so-called sequence of return risk. Professor Michael Drew at Griffith University in Australia has been studying this retirement risk zone by simulating the impact of a market crash at various stages before and after retirement. His studies provide convincing evidence that bad luck right before or right after retirement has a much larger impact on retirement income sustainability than a decline experience later in life. And the consequence of bad returns declines rapidly throughout retirement.
Simulation Arguments
If you're mathematically inclined, the basic formula used in retirement income simulations helps illustrate why low returns early in retirement are so important. At retirement, you invest in stocks and bonds during the first year and earn a rate of return (r1). The formula for the second year is the net portfolio from the first year which grew/declined by (1+r1), then multiplied by the return in year two. In year three, what you start with is a function of r1 and r2. In year four, r1 is still hanging around like an incurable disease. And, unlike in younger years, a new retiree's assets are much larger and not balanced out by the ability to earn income.
Since the consequence of a bad outcome is so extreme early in retirement, this is the time when it's best to avoid the possibility of a big portfolio drop. Why not err on the side of safety early in retirement and shift assets toward risk during the later years of retirement to grow a legacy and increase the chance you won't run out of money?
It's important here to acknowledge that none of this is consistent with neoclassical economic theory. Paul Samuelson pointed out that if the goal is utility maximization, and that stock and bond returns are completely random, then your optimal portfolio should be the same every year. He later pointed out that the allocation could change during working years if your human capital acted like a bond and was slowly depleted over time. This is the philosophical foundation of the declining equity glide path followed by target-date funds. It also conveniently matches conventional wisdom among advisors (think 100 minus age as a traditional guideline).
Many of the assumptions of neoclassical theory don't hold up to observed reality. Practitioners can also benefit from studies that simulate a range of possible investment returns to achieve a different objective such as portfolio sustainability.
One critique of these simulations can be directed toward any analysis that uses the traditional 30-year shortfall methodology first proposed by William Bengen. A strategy is judged successful only if it allows a smooth inflation-adjusted withdrawal every year for 30 years. There's no accounting for what happens if you run out of money in year 31, and a failure in the 29th year (when you're probably not going to be around to experience it) is just as bad as running out in year 10.
Moshe Milevsky has long pointed out that it's better to use an objective like total lifetime utility that captures the full range of spending outcomes. That means assigning a proper weight to all simulated lifetime outcomes including the really bad ones.
Milevsky gives an example: "You have a choice of two medications. Both are likely to cure you. Pill A is known to have a 90% chance of success in curing you (and a 10% chance of failure). Pill B has a (higher) 95% chance of curing you, and a 5% chance of failure. Which pill do you choose?"
