The www.ValueWalk.com site today published my article entitled The Retirement Risk Evaluator: Part One — Why We Need a New Type of Retirement Calculator.
Juicy Excerpt #1: It is true that the historical record does not today contain a returns sequence in which a retirement plan calling for a 4 percent withdrawal failed within 30 years. However, this reality should offer small comfort to retirees who took withdrawals for retirements beginning at the valuation levels that have applied since 1996. From 1900 forward, there have been only two cases in which we have seen the insane levels of overvaluation that applied from January 1996 through September 2008. In each of those two cases, retirements calling for a 4 percent withdrawal came within a whisker of falling. The fact that 4 percent withdrawals have come close to failing in both of the two cases in which they were attempted for retirements beginning at times of high valuations shows not that this withdrawal rate is safe in these circumstances but that it is risky in these circumstances.
Juicy Excerpt #2: The studies and calculators have not been fixed because fixing them would require a widespread public acknowledgment of the flaws of the popular Buy-and-Hold model for understanding how stock investing works. Buy-and-Hold was developed at a time of widespread confidence in the Efficient Market Theory, which posits that stocks are always priced more or less properly. Yale Professor Robert Shiller’s 1981 research showing that valuations affect long-term return challenges this belief (if stocks were always priced properly, the concept of overvaluation would be meaningless and valuations could not predict long-term returns). From 1981 through today, we have been living through a transition period in which the experts’ confidence in the Buy-and-Hold Model has been greatly diminished but in which most experts have felt too cowed by the continued popularity of Buy-and-Hold to develop alternative models more in accord with our new understanding of the message of the historical data.