Uncategorized

Why I Doctored the Numbers Generated By My Retirement Calculator

Valuation-Informed Indexing #297

by Rob Bennett

I wrote last week about The Retirement Risk Evaluator, the retirement calculator that I developed with John Walter Russell. I believe that the Risk Evaluator represents a big advance over all existing retirement calculators. It is the first retirement calculator to include an adjustment for the valuation level that applies on the day the retirement begins. Instead of reporting a constant safe withdrawal rate of 4 percent (which would make sense if stock returns really did play out in a random walk, as was once believed), it reports a safe withdrawal rate that drops to as low as 2 percent at times of super high valuations and rises to as high as 9 percent at times of super low valuations.

I was faced with a bit of an ethical dilemma when I gave the article a final review before pushing the “Send” button. Since I know that Buy-and-Holders feel uncomfortable when I say that the numbers generated by their non-valuation-adjusted calculators are “wrong,” I thought that it might be helpful for me to point out a way in which the numbers generated by my own calculator are wrong as well.

The early version of the article contained a statement saying that the safe withdrawal rate can drop to as low as 1.6 percent when valuations are super high. At the last minute, I changed the wording to say what I say above, that valuations can drop as low as 2 percent when valuations are super high.

Huh? Calculators generate numbers. Numbers are hard, objective things. Either the correct number is 1.6 percent or 2.0 percent. I should know which it is since this is my calculator we are talking about. How could I be uncertain about which number to use?

Russell did the research on which the calculator is based. The number he came up with as the safe withdrawal rate that applied at the top of the bubble (January 2000) was 1.6 percent. That’s the number that I believe is the true safe-withdrawal-rate when the P/E10 level is 44, as it was in January 2000.

The reason why I changed the number listed in the article to 2.0 is that the article included a link to the calculator and readers that go to the calculator to check it out will find that the calculator reports a safe withdrawal rate of 2.0 percent at times when the P/E10 level is 44. I have accused the Buy-and-Holders of getting the numbers wrong in their retirement calculators. But my own retirement calculator gets the numbers wrong too!

When Russell showed me the version of the final version of the Risk Evaluator on the day before we were going to publish it, I was surprised to see that it reported the lowest safe withdrawal rate as 2.0 percent rather than 1.6 percent. I asked him how this happened. He said that the 1.6 number was an outlier that was based on very few data points because we have never before in U.S. history seen valuations as high as those that applied in early 2000 (it was a P/E10 value in the low 30s that brought on the Great Depression). He said that he thus could not have as much confidence in that number as he had in the numbers that benefitted from more support in the historical data.

I expressed concern about using any number other than the 1.6 number that results from a simple statistical analysis. My background is journalism. I see it as my job to report facts and then add discussions of context that may cause readers to place less emphasis on some facts; making an adjustment that changed the 1.6 number to 2.0 seemed too much like spin to me. I told Russell that I had no problem with warning readers that there was less justification for having confidence in the 1.6 number but it seemed manipulative for the developers of the calculator to change it.

He pointed me to the words of Robert Shiller in his book Irrational Exuberance that: “The recent values of the price-earnings ratio, well over 40, are far outside the historical range of price-earnings ratios. f one were to locate such a price-earnings ratio on the horizontal axis, it would be off the chart altogether. It is a matter of judgment to say, from the data shown in Figure 12.3, what predicted return the relationship suggests over the succeeding ten years; the answer depends on whether one fits a straight line or a curve to the scatter, and since the 2000 price-earnings ratio is outside the historical range, the shape of the curve can matter a lot. Suffice it to say that the diagram suggests substantially negative returns, on average, for the next ten years.”

I acknowledged defeat at that point. We used the 2.0 number in the calculator.

We still don’t know who was right. There was a major crash in the 10-year period following a retirement that began in January 2000. But the 2008 crash was not long-lasting enough to cause retirements that called for a 4 percent withdrawal to fail. The other side of the story is that the quick return to high valuation levels may only be signaling that we will experience another major crash before bringing this bull-bear cycle to an end. That next crash may cause retirements that called for a 2 percent withdrawal to fail. It’s a worse-case scenario (that’s what a safe withdrawal rate is intended to identify). But it could still happen given what we have seen over the first 16 years of the 30-year time-period that began in January 2000.

I’ll tell you what kills me. Shiller and Russell and I were all feeling social pressure to report a higher safe withdrawal rate. None of us experienced worries that we would be faulted for reporting a number too high. The concern was that we would be criticized for reporting a number too low if we reported the number generated by a statistical analysis without hedging (in Shiller’s case) or adjusting (in Russell’s). We did the “safe” thing by reporting that the safe withdrawal rate was higher than 1.6 percent.

Or at least we did the safe thing for people writing about investing at a time when prices remained insanely high. Did we do the safe thing for our readers, for the people planning retirements that want to know the true worst-case scenario before they hand in resignations from their jobs? I have my doubts to this day. I have my name on a calculator that says that the safe withdrawal rate in early 2000 was 2.0 percent. My personal belief is that it was 1.6 percent. A part of me wishes that I had had the courage to stick to my guns on this one.

But I am human. Like everybody else that writes about stock investing and wants to be liked by his or her readers, people who have made plans for their financial futures rooted in a belief that bull market gains are real.

Rob Bennett’s bio is here.