Letters to the Editor
To the Editors:
In the Inherited IRAs section of “Traditional IRAs: The Rules Revisited” (September 2005 AAII Journal), it states that a non-spouse IRA can’t be rolled over. While this is true, it goes on to say “Instead, you are required to withdraw and pay tax on the inherited IRA assets.” While this is the general rule, it is very important for your readers to know that there is another, and oftentimes better, option. The IRA beneficiary can leave the IRA as a Beneficiary IRA and, while this does require an annual withdrawal based on the beneficiary’s life expectancy, most of the monies can remain tax-deferred for many years.
One should of course seek advice from a qualified tax adviser, but the article implied only one option and one that is often very costly to the beneficiary.
To the Editors:
Jonathan Guyton is to be commended for highlighting the problems with the usual method of portfolio withdrawals in retirement (“Withdrawal Rules: Squeezing More From Your Portfolio,” August 2005 AAII Journal). Constant (inflation-adjusted) withdrawals are not realistic because people can, and do, adjust their spending habits with portfolio fluctuations. Moreover, those strategies leave unrealistically large balances at the end of the planning period.
Guyton presents portfolio management rules and withdrawal rules that are innovative and appealing. He evaluates them during a specific time interval, what he calls the “perfect storm” for a retiree—i.e., 1973–2002.
However, readers should be cautious about slavishly following rules derived under these circumstances. Using one specific history, as Guyton does, runs the risk that these rules are “tuned” to a specific data set. (Using a single data set, even a “perfect storm,” is analogous to replaying a game of solitaire over and over with the same card order, slightly tweaking the rules each time until the desired result is obtained.)
Those who develop systems to work on future data (forecasting methods, trading systems, neural networks) have learned to avoid this “over-fitting” problem by dividing the historical data into two parts: one part for training (e.g., optimizing portfolio management and withdrawal rules), and the second part for evaluation and selection. Performance on the second data set is never as good as performance on the training set. However, this two-set method selects systems which are more robust to the inevitable deviations from the historical data set.
Renwick E. Curry, Ph.D.
To the Editors:
I read Jonathan Guyton’s article on portfolio withdrawal rules (August 2005 AAII Journal) with great interest. I found his real-world approach refreshing after years of reading articles (and hearing from investment advisers) focused on “straight-line” withdrawal approaches. I do not believe that most rational human beings act in this manner faced with real-world, “life” volatility. Over the years, the Journal has published several other articles in this vein and I encourage you to continue, as I have found them to be horizon-expanding.
One question that came to mind as I read the article is whether there is any sensitivity regarding the tax-structure of the portfolio. For example, are the rules impacted if an individual is primarily withdrawing from, e.g., an IRA or a 401(k) plan account versus drawing from a taxable account?