Valuation-Informed Indexing #360 on the question of safe withdrawal rate and whether it even exists
By Rob Bennett
Michael Kitces argues in his recent podcast interview with the MadFientist blog that there is a floor on safe withdrawal rates — that the number cannot go below 4 percent. I do not agree.
It’s an odd argument for Kitces to make. He gained some fame in early 2008 for making the argument that the safe withdrawal rate can rise to levels above 4 percent. Until he made his case, the conventional belief was that that safe withdrawal rate is always the same number. I think he is right that the safe withdrawal rate can rise to higher levels and I believe that understanding this advances our understanding of how safe withdrawal rates work in a significant way. But it undermines the case to maintain that the number can change in only one direction — that it can move upwards but not downwards.
Kitces explains that the reason why the rate can rise to higher levels is that it is highly dependent on the valuation level that applies on the day the retirement begins. Since stocks provide an average annual return of 6.5 percent real and since the convention in calculating the rate is to permit the portfolio to diminish to within one dollar of zero over 30 years (so that a return of zero for 30 years would produce a safe withdrawal rate of 3.3 percent), the 4 percent number is a counter-intuitively low number. The reason why it is so low is that the idea behind the safe withdrawal rate concept is to look at worst-case scenarios.
While there have been a small number of times in U.S. history when withdrawal rates above 4 percent would not have worked, the reality is that there have been very few of them. Outside of the onset of the three secular bear markets that we have lived through since 1870 (we are now living through the fourth secular bear market but we of course do not yet know what withdrawal rate is the highest that will work for this time-period), withdrawal rates higher than 4 percent would work fine. So Kitces asks — Why not tell investors that, so long as valuation levels are not as high as those that brought on the three secular bear markets, higher withdrawal rates are entirely prudent?
That’s sensible and balanced thinking. It’s high valuations that cause just about all the trouble that is to be seen in the world of stock investing. It’s one thing to go with a low number because you want to be sure that you are covered in the event that a worst-case scenario turns up but it is something else again to apply the worst-case scenario from times of high valuations to retirements beginning at far less risky times for stock investing.
So the idea that the safe withdrawal rate is usually higher than 4 percent is sound. However, I don’t think that it is sound and balanced thinking to believe that, while valuations can pull the safe withdrawal rate above 4 percent, they can never pull it down lower than that. The safe-withdrawal-rate calculator housed at my site reports a safe withdrawal rate for retirements beginning in January 2000 (when the P/E10 value was 44) of 1.6 percent.
Kitces fully acknowledged the strong correlation between valuations and safe withdrawal rates. But he ignores the fact that in the late 1990s we reached valuation levels higher than any that have ever applied before in concluding that the 4 percent figure that applied in earlier time-periods represent a floor. If no withdrawal of greater than 4 percent worked for retirements initiated when we were at a P/E10 of 33, it seems highly imprudent to assume that 4 percent withdrawals will work for retirements beginning at that price level. My thinking on this question is that the safe withdrawal rate that applied in 2000 MUST be lower than any we have ever seen before (since the valuation level that applied then was so much higher.)
There’s another factor that Kitces ignores in his claim that the 4 percent figure is a floor. It’s not only valuations that determine the safe withdrawal rate. The other big factor is the return sequence that applies. Two different return sequences that produce the same long-term return do not necessarily produce the same safe withdrawal rate; one that takes more money out of the portfolio in the early years of the retirement will generally produce a smaller safe withdrawal rate because it causes the portfolio size to drop to dangerously low levels before the gains needed for it to recover come into play. The 4 percent figure comes from a time-period in which valuations were high but from a returns sequence that was on the lucky side; the reality is that retirements taking a 4 percent withdrawal and beginning in 1929 had only a 50 percent chance of working out (these retirements ultimately survived but they were not safe when they were initiated because we did not know at the time that a lucky return sequence would play out).
So even if valuations were no higher in 2000 than they were in the earlier times that produced the 4 percent number, it would not have been right to say in 2000 that a 4 percent withdrawal was safe. Calculations that take both factors (the valuations factor and the returns sequence factor) into account indicate that retirements beginning at that time and calling for a 4 percent withdrawal had only a 30 percent chance of surviving 30 years. Not safe!
Another oddity that I heard in the Kitces interview is that he took reassurance that 4 percent is a floor for the safe withdrawal rate from data showing safe withdrawal rates of “3 to 4 percent” in other countries. A showing that a 3 percent withdrawal rate works is not a showing that a 4 percent withdrawal is safe!
One of the problems that I believe that all of us involved in safe withdrawal rate analysis face is that the numbers produced at times of high valuations are shockingly low. I believe that our minds kick in with efforts to find ways around what the brutal numbers are telling us. It is my belief that we are better off accepting what the numbers say and drawing the lesson that high valuations are a worse poison than just about any of us realize today. The answer is not to comfort ourselves with arguments that there is a floor to the safe withdrawal rate but to get to work persuading investors of the need to not let valuations get too out of hand.
Rob’s bio is here.