Letters to the editor

May 2005 Highlights Summary

I received this email from Rob Bennett about my May 2005 Highlights Summary. Rob is the author of Passion Saving, which he will make available from his web site soon.
http://www.passionsaving.com/

John:

Your article summarizing recent findings is of great value. It is very easy for people to lose sight of the forest for the trees. This piece does a fine job of putting things in perspective.

There is one word in Item 11 that I consider an unfortunate choice. You say:"In terms of numbers, varying allocations according to P/E10 historically would have allowed us to increase the amount that we could withdraw SAFELY from 4.0% to 5.0%+ (of the portfolio's initial value plus inflation), when compared to a fixed allocation of stocks and bonds." I do not believe this is so.

It appears to me that what you mean to say here is "to increase the amount that we could withdraw SUCCESSFULLY..." I presume that your finding is that there were no failures at a withdrawal of up to 5 percent, looking backward today at what we know happened in those returns sequences. Looking backward does not tell us what is safe, but only what happened to survive.

It does not appear to me (it is possible that I do not understand precisely what sort of analysis you were doing here) that you have determined that a withdrawal of up to 5 percent was safe with switching, but only that it survived. It's fair to say that it proved successful. It does not appear to me (again, based on my possibly incorrect understanding of what you looked at here) that you have determined that a withdrawal of up to 5 percent was historically safe with switching. To determine what was safe would require a different sort of analysis, one that focused on what was known at the time the hypothetical retirements began, not on the returns sequences that played out in the real world but which we learned of only after the retirements commenced.

Even if I am wrong in my understanding of what you were looking at here, my sense is that this aspect of things continues to be a source of confusion for many and that it would be a good idea to clarify whether your analysis was a survival analysis or a safety analysis. I think it is important for people to understand that the root flaw of the conventional methodology is not that we entered a bubble in the late 1990s, but that the methodology always examined not safety but survival. The bubble greatly exacerbated the problem that existed from the day the conventional methodology was created.

Rob

This is the article.
May 2005 Highlights

HERE IS MY REPLY:

Rob, you are absolutely right.

There is a lot of confusion about Safe Withdrawal Rates and Historical Surviving Withdrawal Rates. Just because a portfolio would have survived does not mean that it was safe. It could have been lucky.

We require that a portfolio have about a 95% chance of success (one-sided probability about a calculated rate) before we declare it SAFE. Only when the withdrawal rate exceeds the High Risk Rate do we declare it to be UNSAFE. (The High Risk Rate has a 95% chance of failure. It is the other confidence limit about the calculated rate.)

I have calculated the Safe Withdrawal Rates with optimized switching using stocks and 2% TIPS. Here is the link.
Safe Withdrawal Rates with Switching

Here is a link to the data.
Safe Withdrawal Rates with Switching Data

I wrote this on May 22, 2005.

Here is a FOLLOW UP:

John:

This is a follow-up to your response to my recent "Letter to the Editor" that I wrote in regard to your most helpful May 2005 Highlights article. You say in your response: "We require that a portfolio have about a 95% chance of success (one-sided probability about a calculated rate) before we declare it SAFE. Only when the withdrawal rate exceeds the High Risk Rate do we declare it to be UNSAFE. (The High Risk Rate has a 95% chance of failure. It is the other confidence limit about the calculated rate.)"

I understand the distinctions you are drawing. What you are doing here is clear enough that I do not find it misleading. But I do not agree with the suggestion that a withdrawal rate becomes "unsafe" only when the odds of it working out (according to the historical data) are less than 1 in 20.

It is perfectly reasonable to refer to withdrawal rates lower than the SWR as "safe." A take-out number that has an 80 percent chance of working out (according to the historical data) is safe enough to fairly be referred to as "safe," in my view. But I do not agree that a take-out number that has only a 60 percent chance of working out is "safe." Any numbers with a chance of working out of too much below 80 percent or so are unsafe, in my assessment. I think that it would be better to refer to the take-out number that has only a 5 percent chance of working out as the "Extreme High-Risk Withdrawal Rate." Using that terminology would take away the suggestion that take-out numbers higher than that are safe.

The fundamental problem here is that the continuum being described is not a continuum in which the mid-point (50 percent likelihood of survival) is generally perceived as acceptable. Most retirees want their retirements to have a far better chance of working out than the odds of picking a coin flip correctly. Safety is achieved only at the upper end of the continuum. As you go to points on the continuum lower than about 80 percent, you are taking on greater levels of risk (and it is useful to calculate these numbers so that people know how much additional risk is being taken on with various higher take-out numbers), but you have already left the concept of retirement safety as it is commonly understood behind.

Rob

Here is my RESPONSE:

Rob, you are right once again.

I need to go back to the terms SAFE and HIGH RISK.

The intermediate zone contains varying degrees of risk. Roughly speaking, the Calculated Rate is equivalent to a coin toss (with an honest coin). I doubt that very many people would accept 50%-50% odds for normal planning. Some might use it when looking at extremes such as living beyond age 100.

How about this? Let me refer to the withdrawal rates above the Calculated Rate as being in the HAZARDOUS REGION. The High Risk Rate corresponds to almost certain failure. Withdrawals below the Calculated Rate are in the PLANNING REGION. The Safe Withdrawal Rate identifies almost certain success.

Have fun.

John Walter Russell
I wrote this on May 23. 2005.