Monday, April 16, 2012

Rethinking Safe Withdrawal Rates: The Meaning of Failure

My newest column at Advisor Perspectives, "Rethinking Safe Withdrawal Rates: The Meaning of Failure" is now available.  This column attempts to explain why people should give a careful read to articles such as Joe Tomlinson's "A Utility-Based Approach to Evaluating Investment Strategies" from the February 2012 Journal of Financial Planning, and "Spending Flexibility and Safe Withdrawal Rates" from the March 2012 Journal of Financial Planning by Michael Finke, Duncan Williams, and me. 

The idea that retirees should focus on finding a spending strategy that maintains a rather low failure rate, as is par for the course with safe withdrawal rate studies, is not adequate. This narrow focus ignores the lost potential enjoyment from spending more even if it means having to cut back later, it ignores how much flexibility retirees may have to cut their spending at a later date, it ignores the other availability of spending resources outside of the financial portfolio, it ignores the probability of still being alive in late retirement, it ignores any goals to leave a bequest, and it ignores potentially how long the "failure" condition may last. Retirees need to be thinking about a more complete model that incorporates these considerations when developing their retirement income strategies.

Last year, Todd Tresidder wrote a well-received discussion of the history of safe withdrawal rate research called, "Are Safe Withdrawal Rates Really Safe?" In his article, he defines three generations of research. The first generation simply assumed an average annual portfolio return and provides withdrawal rate guidance based on that. The second generation, ushered in by Bill Bengen, incorporated the sequence-of-returns risk associated with volatile portfolios and found that historically a 4% withdrawal rate is much closer to being safe and much lower than advocated in first generation research. The third generation of research generally seeks to incorporate more realism to second generation results by including factors such as market conditions at the retirement date, fees, longevity beyond 30 years, and an acknowledgment that the post-1926 US historical data is not sufficient to provide much confidence about the viability of the 4% rule. 4% is just an educated guess based on limited historical data, and you only get one whack at the cat.

What I wish to suggest is that we are now in the early stages of a fourth generation of research which will broaden the safe withdrawal rate question to include all of these other factors... maximizing the tradeoff between spending more now and less later and understanding the magnitude of potential failures, incorporating bequests, incorporating other sources of income available to retirees even in the event of portfolio depletion, determining how to allocate to bond ladders, SPIAs and other retirement products, and providing some allowance for the issue of survival probabilities to various subsequent ages. There are a lot of researchers working in this area now and I'm writing a longer article which discusses many competing approaches. Again, please have a look at my new column which provides an effort to motivate why this broader view is needed. 

Finally, as I was writing this up, I made another figure which shows a retirement income frontier for retirees. This figure combines three pieces of information for different withdrawal rate and asset allocation strategies that retirees may care about: the spending rate from their assets (higher is better), the failure rate of the strategy (lower is better), and the median expected bequest from the strategy (higher is better). Bequests and failure rates are both calculated by using a simulated age of death for each member of the couple in each Monte Carlo simulation for a 65-year old couple. They are not for 30-year retirements.  For the "risk metrics" approach to retirement planning, a retiree just needs to select the point they believe provides the best balance among these tradeoffs for their own particular situation: