Review:  Spending flexibility and safe withdrawal rates

Bob Veres  | Inside Information | 24 April 2012

 

"Spending flexibility and safe withdrawal rates" by Michael Finke, Wade Pfau & Duncan Williams, Journal of Financial Planning, March 2012

What's missing from the safe retirement withdrawal studies? Last month, in the Journal of Financial Planning, Joe Tomlinson had people smacking their foreheads when he used a utility function to define the willingness of clients to risk running out of money in return for enjoying their retirement portfolio now.

This article follows the same logic, giving retirees a risk aversion coefficient ranging from 1 to 10, and then looking at the implications for portfolio withdrawals and portfolio composition. It starts by defining the risks - a graph shows different stock/bond mixes ranging from 20% to 100% stocks, along with distribution rates ranging from 3% to 9% and then shows the percentage chance, measured in number of years, when the clients would have to live on dog food. Based on this measure, the clients who are willing to risk a 9% distribution pattern (that is, 9% of the initial portfolio, taking out this dollar amount adjusted for inflation in subsequent years) have the best chance of success with a 100% stock portfolio. But the risk-averse client isn't even going to LOOK at that curve which, at best, offers a 25% chance of impoverishment late in life. Further down, you can achieve what appears to be a 7% chance of running out of money with a 60% stock portfolio.

The goal, as Tomlinson pointed out and as these authors make clear, is to optimize the trade-off between spending more now and risking the chances of spending less later.

If you plug in a risk aversion coefficient of 1, you are talking about a risk-taking client who is looking hard at that 9% distribution rate and ignoring the 3% option. On a graph, it takes all the curves on a different path, and you discover that the optimal tradeoff - the maximised certainty equivalence - can be found with a 70% stock allocation and a withdrawal rate of 7%. A close second is the 6% withdrawal rate with a 60% stock allocation, which is very nearly equal to a 100% stock allocation with an 8% withdrawal rate. This is living on the edge.

If the client sitting across the table has a risk aversion coefficient of 4, meaning she is slightly more timid but still above-average in the risk-taking world, you get a remarkably different picture. The optimal portfolio is 30% stocks and the optimal withdrawal rate is 4%. No other combination of withdrawal rates and allocations comes close. The 4% withdrawal rate is preferable for all portfolio mixes, and if the client decides to walk on the wild side and take a 5% distribution pattern, the optimal portfolio is between 40% and 50% stocks.

This, of course, is a remarkable difference between two clients who are on the upper end of the risk-taking scale, which tells you (something the authors don't point out) that this risk aversion coefficient tool is EXTREMELY sensitive, and if you get your client evaluation a teensy bit wrong, there is a high probability that the distribution path and portfolio will be more than a teensy bit off - a pattern advisers have learned to be wary of ever since the introduction of portfolio design optimisers back in the 1980s. The authors add in situations where there are fixed income sources that clients can draw on which of course raises their risk profile because at least some of the core living expenses are being met.

The authors state at the end that a client's willingness to take portfolio risk before retirement is equivalent to a willingness to accept shortfall risk after retirement, but I read through the article several times without finding any evidence that this is true. It's possible that the two are entirely different and not related at all, as FinaMetrica founder Geoff Davey has found (through thousands of questionnaires) that the willingness to risk life by skydiving or bungee jumping is generally unrelated to the willingness to assume downside risk in an investment portfolio. The skydiving adrenaline addict may be a wimp when it comes to a 10% stock market blip, and the person who can tolerate a 45% drop in portfolio value for the long-term higher returns may flinch at the thought of eating cat food after age 75.

That said, it only makes sense that some measure of risk tolerance would find its way into portfolio drawdown practices, just as the concept has long been a core part of pre-retirement portfolio planning. In two articles - Tomlinson's, and now this one - we have come a very long way toward figuring out how to incorporate the risk tolerance concept into retirement planning.  But we have also come hard against the reality that this is a very sensitive measure, and potentially very hard to determine with any precision. The next frontier in these studies is finding a tool that advisers can use, with some confidence, to ascertain a client's utility function or risk aversion coefficient. That, and perhaps we should eventually agree on one consistent thing to call it.

Read the full article >

 

 

Bob Veres is editor of Inside Information, a master study group whose members are the leading practitioners in the US financial planning profession. This article has reproduced with permission from his weekly Media Reviews, a subscription service in which Bob summarises the contents of three or four professional magazines each month, with direct links so you can go straight to the articles that interest you.

More about Inside Information >