by Rob Bennett
The Old School safe withdrawal rate (SWR) studies say that the SWR for retirees heavily invested in stocks is 4 percent. That is, someone who retirees with $1 million in his portfolio can take out $40,000 to cover living expenses each year with virtual certainty that he will not run out of money for 30 years (taking a 65-year old retiree to age 95).
The New School studies come to very different conclusions. The New School studies say that the SWR varies with changes in the valuation level that applies on the day the retirement begins. Retire with a $1 million portfolio at a time when the P/E10 level is low and you can take out $90,000 each year with virtual certainty that your retirement will not fail. However, if you retire with $1 million at a time when the P/E10 level is high, you might be pushing it to use that portfolio to cover annual living expenses of anything above $20,000.
The numbers come from analyses of the historical stock-return data. But let’s leave that aside for the moment. Forget that the safe withdrawal rate is the product of a numerical calculation. Ask yourself -- Which numbers make more sense? Looking at the numbers from that perspective opens the door to many powerful insights about how stock investing works in the real world.
Let’s start with the most basic question: Should the safe withdrawal rate change or should it be stable?
Does it seem strange to you that smart people believed for many years that the SWR was a single number? It seems exceedingly strange to me. If the SWR is always the same number, stocks always offer the same long-term value proposition to investors. Can that be? Can valuations really have no effect? I don’t just believe that the idea that the SWR is stable is wrong. My personal view is that the idea is zany.
If it really is zany, that’s a big deal. Because this is an idea with far-reaching implications.
Say that the SWR is always 4 percent. Does that sound like a low number or a high number?
It is a remarkably low number. The SWR calculations assume that the portfolio may be reduced to zero over the 30-year time-period. Thus, an asset class providing a zero return for 30 years running would provide a SWR of 3.3 percent. The 4 percent SWR reported in the Old School studies for stocks is only a little better than the SWR that would apply for an asset class offering a return of zero. And yet stocks provide an average long-term return of 6.5 percent real! Why is the SWR for stocks so low?
It’s because of the power of compounding. The data shows that failed retirements are always caused by losses suffered in the first 10 years of the retirement. The retirement that does well or even just okay for 10 years is always going to survive 30 years because even small positive returns achieved on a portfolio large enough to finance a retirement (typically $1 million or more) translate into substantial dollar gains. In contrast, big losses suffered on such large portfolio amounts translate into devastating setbacks when the compounded effect of those losses are taken into consideration.
The studies are teaching us that the price attached to volatility is huge in the retirement context. A super-safe asset class like IBonds often offers a SWR greater than that offered by stocks (remember, IBonds paying 0 percent would provide a SWR of 3.3 percent). Retirees pay a big price for the price volatility that is characteristic of stocks. If there were something we could do to diminish that volatility, we could show investors how to achieve safe retirements at earlier ages than are generally viewed as possible today.
There is something we can do.
It is not precisely right to say that price volatility is the problem. Upward price volatility of course causes no difficulties; it is only the possibility that stock prices will move down sharply that brings the SWR for stocks so low. Also, it is only sharp price drops experienced in the first 10 years of the retirement that cause a problem, for the reasons noted above. If retirees could be sure that they would not experience sharp price drops in the first 10 years of retirement, they could safely retire at earlier ages.
Guess what? Sharp price drops do not occur randomly. Price drops that are deep enough and lasting enough to cause a problem only occur starting from times of high valuations. That is of course just what you would expect to be the case. Regression to the Mean sends stock prices down when they rise too high.
What if stock prices are not high at the time an aspiring retiree is planning to initiate his retirement? The data says that the retiree can safely employ a much larger withdrawal rate. That retiree is not going to experience deep price drops in the first ten years of his retirement. So his safe withdrawal rate, properly calculated, is a much bigger number. When prices are extremely low, the SWR can rise to as high