I had the good fortune to attend the Fall Conference of the Retirement Income Industry Association earlier this month in Boston. As I am currently somewhat isolated in Japan working on my research, it was a good opportunity to see how financial planners, financial companies, and insurance companies are approaching the search for solutions to the problem of funding long and uncertain retirement periods.
Though I already sort of knew this, the conference did help to clarify in my mind that there are three basic approaches to drawing a stream of income from one’s savings.
1. Safe Withdrawal Rates / Systematic Withdrawal Plans: This is the area in which I have mostly been researching so far. This approach focuses on the total returns of the retirement portfolio, not distinguishing between sources of income or capital gains (except, perhaps, for tax purposes). The basic question is how much retirees should be able to withdraw each year without running out of wealth. There is no attempt to guarantee an income floor, and research in this area generally tends not to consider other sources of retirement income or annuities, as the basic thrust is getting a handle on how to drawdown wealth from a volatile portfolio in a sustainable manner.
2. The Buckets Approach: This approach moves away from a total returns analysis to provide a mental accounting framework that distinguishes different accounts (or buckets) of the portfolio to be used for different purposes. Of the various versions of this I have seen so far, the most intriguing to me is Stephen Huxley and J.Brent Burn’s Asset Dedication approach, which I discussed here before. Rather than worrying about short-term volatility and then allocating between stocks and bonds to find the happy mix, Asset Dedication instead uses bonds specifically to lock-in future spending amounts over the subsequent horizon of 3-10 years. Then stocks can be left alone to grow without worrying that they will have to be sold after a downturn. Though some sources downplay this point (Asset Dedication is upfront about it), this notion is only true in the case that the length of the lock-in period is allowed to adjust based on stock performance. I still would like to run my own simulations about this to see if there really is something to the approach, or whether it mainly has outperformed due to the higher stock allocation it tends to allow. Though I haven’t done a thorough review of research about buckets yet, most everything I’ve seen has tested the approach using U.S. data since 1926, which I have emphasized repeatedly on my blog is a rather remarkable period in world history for stocks. Dimson, Marsh, and Staunton's “Irrational Optimism” from the January/February 2004 issue of Financial Analysts Journal explains the potential problems with putting too much faith in the idea that future U.S. stock returns will be as exceptional as in the past.
3. Income Floor / Upside Potential Approach: This approach most closely matchs economic theory. The idea is to first focus on an acceptable minimal income floor you will need in retirement, and then pay the costs (and they can be hefty, especially with the currently low interest rate environment) to get this income floor guaranteed. In addition to Social Security and other pensions, this floor could include annuities, guaranteed-income products, TIPS ladders, etc. Once the floor is established, then remaining wealth can be used to invest in volatile assets that provide more upside potential. Though the “income floor / upside potential” wording was not the same, this is basically the approach suggested in Zvi Bodie’s Worry-Free Investing, and I understand that his upcoming book, Risk Less and Prosper: Your Guide to Safer Investing, will also discuss this approach. I saw as well that recently Laurence Kotlikoff added a new module about this to his ESPlanner software. I think we will be hearing much more about this in the future, and it is where the most research is needed. In particular, providing an income floor can be expensive, so finding an optimal level for the income floor, when to start establishing the floor, and how much risk to take with the upside are all important questions.
One closing point, a few days ago here I expressed my enthusiasm for the recent research by Milevsky and Huang, and by Williams and Finke. In part, why I am so enthusiastic about these papers is that I think they effectively provide a link between “safe withdrawal rates” and “income floor / upside potential.” The safe withdrawal rate literature usually focuses on keeping failure rates low, but with a sufficient income floor the acceptable failure rate for remaining wealth can actually be quite high.
In an upcoming blog post, I'd like to write about Robert D. Curtis's "Monte Carlo Mania" in Retirement Income Redesigned: Master Plans for Distribution edited by Harold Evensky and Deena B. Katz, because I think it has a really unique and intriguing way to think about this issue as well.
Ultimately, I think the “income floor / upside potential” approach will come to dominate retirement planning, and this is a good issue for researchers to investigate.