berntson wrote:...The linked paper presents evidence that given an equal starting balance, increasing stock allocations in retirement lowers the failure rate...
The problem with many such studies is that they go at it backwards. They start from some arbitrarily chosen withdrawal rate first. They should start with risk tolerance first. One should begin by exploring carefully what "failure rate" is really acceptable to retirees, without anchoring the choice by offering preselected alternatives. Is 5% low enough? It is higher than the chances of rolling snake-eyes. Are most retirees willing to take that chance of watching their portfolio shrink sharply and having to make a wrenching change in plan?
If they are, then, sure, they are.
The characteristics of conservative assets is both low return and
predictability. Predictable is another way of saying they have a steep, sharp, well-defined relationship between withdrawal rate and failure rate. Stay slightly under that they can deliver and the failure rate is very low. Exceed it slightly and the failure rate is very high. The extreme is physical cash under the mattress. $1,000,000 in cash over a 30-year time period it will sustain withdrawals of $33,333 per year with 0% failure rate, while $33,334 per year gives a 100% failure rate.
Stocks and other high-risk, high-return assets have a fuzzy edge. They have almost no upper limit to the amount they can deliver, given good luck. As withdrawal rate increases, failure rate tapers off slowly, with no "cliff edge."
The result is that the effect of adding stocks depends on how aggressive the withdrawal rate chosen is.
Simplistically, if the withdrawal rate is reasonably within what bonds can deliver, then stocks are not needed, do not reduce failure rate, and failure rate is low. If the withdrawal rate outside the range that bonds can deliver, then stocks are needed, do
reduce failure rate--and failure rate is high even with the best possible allocation.
The extreme case is a withdrawal rate of, let's say, 50% per year. For any portfolio of stocks and bonds, the historical failure rate will be 100%. However, if you add lottery tickets to the portfolio, the portfolio will succeed if the retiree wins the lottery. Therefore, with lottery tickets, the failure rate will be less than 100% and it will be accurate to say that "higher allocations of lottery tickets reduce the failure rate."
The reason why nobody seriously suggests this is that
nobody thinks that a 99.999% failure rate is acceptable,
even if it is lower than 100%.
However, statements that "stocks reduced the failure rate" has to be evaluated with regard to what the reduced failure rate was. Kitces and Pfau choose to use the "5th percentile," i.e. 5% failure rate. For example, using 20->80 cut the failure rate from 8.9% to 5%. This is great if 5% is acceptable.
But where do they explain why they chose 5%? Did they interview retirees?
Would the study have shown the same benefits for a reverse glidepath if the acceptable failure rate had been set at "snake-eyes," 2.78%, and the withdrawal level reduced accordingly?
Whenever I have looked at studies that offered a wide choice of withdrawal rates and portfolios, the results I've seen have always been the same. At low withdrawal rates, the failure rate is practically independent of stock allocation. The effect that "stocks reduce failure rate" is not seen. At high withdrawal rates, the failure rates are low
ER with a high stock allocation, yet still "high" relative to some investors' risk tolerance.
So, as always, it boils down to an issue of risk versus return, and the retiree's risk tolerance. A retiree
who is willing to accept higher risk for higher return selects a higher stock allocation, uses a higher withdrawal rate, and experiences a higher failure rate. A retiree
who is risk-averse and is willing to accept a lower withdrawal rate--and it is not
that much lower--does not benefit from a high stock allocation (although their heirs may).
The big problem is that so many published studies start out by
sneaking in an assumption that they already know the retiree's risk tolerance, and that it is relatively high.
This flies in the face of the common observation that many retirees choose to hold their entire retirement savings "portfolio" in a bank account. A common interpretation is that they are fools who do not know what is best for them, but perhaps they are better in touch with their own risk tolerance than some retirement advisors are.
Annual income twenty pounds, annual expenditure nineteen nineteen and six, result happiness; Annual income twenty pounds, annual expenditure twenty pounds ought and six, result misery.