The question of how to safely generate income from a retirement portfolio is one of the most challenging in financial planning. In the days when people had traditional pensions, their employers simply promised them a constant inflation-adjusted income for the duration of their retirements. As we have moved away from traditional pensions and into self-directed savings plans such as 401(k)’s and IRA’s, investors and advisors must create their own customized income plans. New research from Morningstar highlights what appears to be a better approach to creating a stable income stream from an investment portfolio.
Basics of Safe Withdrawal Rate Estimates
The most common ‘rule of thumb’ is called the ‘4% rule.’ Assuming that an investor has a portfolio that is allocated 60%–70% in stocks and 30%–40% in bonds, a range of studies have found that, historically, planning for an income equal to 4% of your account balance at retirement is a pretty safe bet. If you retire with $1 million, you can plan to draw $40,000 per year in the first year and inflate that annually to keep up with inflation. Analyzing history to come up with a withdrawal rate that has historically had little or no chance of totally depleting a portfolio leads to what is referred to as the ‘Safe Withdrawal Rate’ (SWR). The ‘4% rule’ is the best-known result from SWR studies.
There are a number of problems with the ‘4% rule’ that have been explored in literally hundreds of articles. First, the ‘4% rule’ is based on a very simple allocation to stocks and bonds and many studies demonstrate that you can do better with a more diversified portfolio that includes real estate, commodities, and other asset classes. Second, the 4% rule ignores the specific market conditions when you retire. If you happen to retire at the end of a bull market, the 4% rule is far too optimistic. If you happen to retire at the end of a bear market, the 4% rule is too pessimistic.
While the original research on the ‘4% rule’ looked at historical data, a range of tools generate ‘forward-looking’ projections of portfolio risk and return. These differ from historical data in that they are intended to correct for biases in the specific period of history that is used in a backward-looking study. Forward-looking models are not perfect, but they can be adjusted to reflect a future that is different from the past. For example, I don’t know of any credible expert who believes that we can expect double digit returns from the stock market as a whole over the coming decades. Over the 20-year period through October 2012, however, the average annual return for the S&P 500 (including reinvested dividends) is 12.2% per year. Over the period from 1950–2009, the average annual total return for the S&P 500 is comparable to this level. It would be questionable, however, to simply assume that the average in the future will be this high. The best recent thinking on this problem suggests that the S&P 500 may have an expected return of around 8% per year, and some credible experts are estimating considerably lower. While the confidence level on such predictions is low, it seems far more reasonable to start your projections with a conservative equity risk premium.
An Alternative Approach
A recent research paper by Morningstar’s head of retirement research, David Blanchett, explores how best to determine a sage annual income in an interesting way. The WSJ did a short summary, but the full article is available here. The gist of the paper is that the best way for a retiree to draw the maximum safe level of income is to manage to a goal of limiting the probability of running out of money by a specified age (which the Morningstar article refers to as the Probability of Failure (PoF)). For example, you might have a target that you want no more than a 10% probability of running out of money by age 90. You can calculate this type of probability using Monte Carlo Simulation tools. I developed a software tool that performs this function (Quantext Portfolio Planner) and there are others. In this approach, you need to periodically run such a tool and determine the maximum income draw to maintain an acceptable probability of failure. If the market has declined substantially since the last time you ran the calculation, your safe income draw will be lower. If the market has rallied, however, your safe income draw will be higher.
As an aside, I will note that the term ‘Probability of Failure’ may strike some readers as jarring. Nobody likes to think that they have any probability of failure in terms of being able to provide for themselves. Obviously, it would be great if people could plan a retirement with zero probability of failure. This is possible if you have sufficient assets to live on the income generated by TIPS, a strategy that has long been championed by Zvi Bodie. If you run through a set of simulated outcomes and find the estimated probabilities of failure vs. income too unsettling, a 100% TIPS strategy may be the way to go.
The Morningstar paper’s conclusion is consistent with an article that I wrote in 2008. In that article, I went through an almost identical process of managing to a constant probability of failure through time as proposed by Blanchett. An investor changes his or her income draw to maintain a constant probability of failure as he or she ages. That has always seemed like a logical process on the basis of my work. Blanchett, however, has done the harder work of robustly demonstrating that this is an optimal approach.
In real life, the devil is in the details with regard to how to implement this sort of adaptive income draw based on maintaining a constant probability of failure (e.g. probability of running out of money). We need to come up with expected risk and return for every asset that is held in a portfolio and, accounting for the correlations between them, estimate the income draw that is consistent with a specific target probability of failure.
A Simple Example
I will end this piece with a simple example to demonstrate how the process of managing to a constant probability of failure works. The discussion above makes this sound complex, but it is actually a straightforward analysis.
Imagine that we have a 65-year-old retiring now and he has $1 million in assets. He is invested in a low cost portfolio which is simply allocated to 60% VFINX, an S&P 500 fund, and 40% VBMFX, an aggregate bond fund. When I run this scenario through Quantext Portfolio Planner, with the assumption that he will draw $40,000 in 2013 and then increase that by 3% per year to keep up with inflation, I get the following probability of failure:
Our investor has a 5% chance of totally depleting his portfolio by age 85 and a 10% chance of totally depleting his portfolio by age 88. In my experience, these are pretty good odds.
In the simulation results, the portfolio has an expected return of 5.8% per year. The trailing 3-year average annual return is 10.4%, however.
If our investor is comfortable with drawing $40,000 a year on his $1 million portfolio given the probabilities of failure shown above, he will plan to manage to these statistics in the future. Next year, for example, he will change his income, as needed, to maintain a 10% probability of failure by age 88.
If our model investor feels that he is willing to take a higher risk of failure, he can draw more income. If he feels that this chance of failure is too high, he can reduce his income draw.
Our investor might try to increase his safe withdrawal rate by creating a more diversified portfolio so as to generate a higher expected return at the same level of risk. This would allow him to draw more income at the same probability of failure or the same level of income with a lower probability of failure. Folio Investing’s Target Date Folios, for example, are engineered to provide higher expected return at a range of different risk levels by strategically combining a wide range of asset classes.
Managing your portfolios to maintain a constant probability of failure as you age is a good way to approach the problem of estimating the income that you can afford to draw from a portfolio. The idea of estimating a safe withdrawal rate at retirement based on an acceptable probability of failure is well-established, so why wouldn’t it also make sense to periodically recalculate this figure as you age?
The main problem with this approach, of course, is how well the analytical tools can estimate the expected return and risk of your portfolio. There is little question that estimates of the expected future return of the S&P 500 are very approximate. The problem, however, is that there are no better alternatives if you are not willing to (1) purchase a fixed annuity with an inflation protection rider, or (2) purchase TIPS to cover your basic living expenses.
The views set forth in this blog are the opinions of the author alone and may not represent the views of any firm or entity with whom he is affiliated. The data, information, and content on this blog are for information, education, and non-commercial purposes only. The information on this blog does not involve the rendering of personalized investment advice and is limited to the dissemination of opinions on investing. No reader should construe these opinions as an offer of advisory services.