This article was posted August 1, 1999.
Financial Engines, Inc., the Web based investment advisory firm founded by Nobel Laureate William F. Sharpe, recently unveiled the consumer version of its retirement planning software. One of the most interesting features of this software is its retirement income forecaster. The user can input a goal of a specified annual income in retirement and the forecaster calculates the retiree's probability of success in attaining that goal. Retire Early used the Financial Engines' retirement planner to calculate the "safe" withdrawal rate for several well known portfolios. While the results were for the most part as expected, the study revealed a few surprises.
Let's look at some portfolios.
This analysis looks at a 40 year pay out period for a portfolio invested at the "efficient frontier", the mix of stock and fixed income securities that results in the maximum 100% survivable inflation adjusted withdrawal rate. For a 40 year pay out period, the efficient frontier is 77% stock and 23% fixed income. The fixed income portion of the portfolio is invested in the Vanguard Prime Reserve money market fund.
The portfolios considered include the following:
Using the Financial Engines Retirement Planner.
In an attempt to mirror a 40 year pay out period, the study assumed a 43 year old male with a goal of retiring at age 45 (the youngest retirement age permitted by the Financial Engines' software.) He has a portfolio value of $1,000,000 at age 43. For simplicity, the retiree receives no Social Security or pension benefits. He has a goal of $50,000 per year in retirement income at age 45 which he plans on withdrawing from his portfolio.
According to IRS Life Expectancy tables, a 45 year old is expected to live 37.7 years, close to the goal of a 40 year pay out period. It is assumed that the Financial Engines' software uses a similar mortality table.
The Financial Engines Retirement Planner reports a variety of results. The most interesting are listed in the table below.
The results of Financial Engines forecast for the five portfolios described above appear in the table below. The "95% safe" withdrawal rate from the Retire Early's Safe Withdrawal Rates for Concentrated Portfolios is also shown at the bottom of the table for comparison.
Discussion of the Results.
A first glance, the most striking result is the low "downside" withdrawal rates. For all but the Bill Gates and Jerry Yang portfolios, Financial Engines' "95% probability of success" withdrawal rate was much lower than the Retire Early study. It's also noteworthy that the 5% withdrawal rate recommended by many financial experts (i.e. $50,000 per year from a $1 million portfolio) has a low probability of success -- at least according to Financial Engines. A 13% probability of sustaining a 5% withdrawal from the Foolish Four Portfolio and 16% probability for the S&P500 portfolio is much lower than expected. The Retire Early Study on Safe Withdrawal Rates calculated a 82% probability of sustaining a 5% inflation-adjusted withdrawal from a S&P500 portfolio with a 40 year pay out period.
It also seems to defy logic that a retiree has only a 16% chance of sustaining a $50,000 per year withdrawal from the S&P500 portfolio but a 36% chance with a portfolio consisting entirely of YHOO stock. This anomaly can be easily explained once you understand how Financial Engines makes its retirement income forecast. They construct a probability distribution, or "bell curve", from the retirement income forecasts for each of 250 economic scenarios. This bell curve is much wider for a single volatile stock than for the S&P500. The width of the bell curve is most noticeable if you look at the "95% probability of success" withdrawals. It's $33,600 for the S&P500 portfolio, but only $8,200 for the YHOO portfolio. The difference between the upside and downside forecasts is ($55,600 - 33,600 = $22,000) for the S&P500, but ($113,000 - 8,200 = $104,800) for YHOO.
Foolish Four problems.
Financial Engines' software concludes that the Foolish Four portfolio would have slightly less risk than the broadly diversified S&P500 index fund. Although the higher probability of success for a $50,000 withdrawal from the S&P500 portfolio (16% vs. 13% for the Foolish Four) would seem to argue that the S&P500 portfolio is less risky.
While we're on the subject of the Foolish Four portfolio, one reader pointed out a shortcoming of this study. The Financial Engines software assumes our retiree is holding a "long-term buy and hold" (LTB&H) portfolio. Of course, the Foolish Four strategy requires us to update the portfolio each year with the "new" Foolish Four. So the Financial Engines program may not properly reflect the Foolish Four performance.
One possible solution to this shortcoming would be to replace the Foolish Four with one of the Dogs of the Dow unit investment trusts. (The Van Kampen Dow Strategic 5 UIT is one possibility.) Unfortunately, the Financial Engines database does not include the Van Kampen Dow UIT. The high fees and commissions on the Van Kampen UIT (nearly 3% annually) would also have to be adjusted to match the performance of buying the Foolish Four through a discount broker.
How do you explain the difference between the Financial Engines and Retire Early results?
There are two possible reasons for the differences between the Financial Engines and Retire Early results.
The Retire Early study is based on over 125 years of historical stock and fixed income returns. The Financial Engines software uses "forward looking" economic scenarios devised by a panel of experts. It's possible, though unlikely, that a significant number of these scenarios call for economic performance much worse than anything we've seen in the past 125 years.
The other difference between the Financial Engines and Retire Early results is that Financial Engines assumes that the retirement portfolio is "annuitized" once the employee retires.
A retiree opting for annuitization would obviously incur additional fees and expenses as well as the lost "opportunity" cost of not being in the equity markets during his retirement years. A 45 year old could buy an immediate life annuity with an annual payment of about 6% of the purchase price. When you factor in inflation for the 40 year pay out period, it's not hard to see the payment adjusted all the way down to 3%. It's likely that "annuitization" explains the bulk of the difference in the "95% probability of success" withdrawal rates.
Should you annuitize your retirement portfolio?
To Financial Engines credit, they say they do not recommend annuitizing your retirement assets when you retire. They are merely using the annuity to make an estimate of your income in retirement.
Unfortunately, unless you're extremely risk adverse and "outlive" the mortality table, a lifetime annuity is unlikely to maximize your retirement income. You should do much better if you continue to hold a portfolio with a significant equity allocation and limit what you pay in fees and commissions. For someone holding Retire Early's S&P500 portfolio, the difference between Financial Engines 3.03% draw and Retire Early's 4.44% draw is the difference between an annuity and holding an equity portfolio in retirement. See, "Should I buy an immediate life annuity when I retire?"
Are Financial Engines' "95% probability" retirement income estimates too low?
While I'm confident that Financial Engines' numbers are calculated correctly, they may be misleading to users that don't understand the effect of annuitizing thier retirement portfolio. You'd hate to see an employee work 5 or 10 years longer because he thinks he can only withdraw 3.03% per year, when he could retire today on 4.44% if he didn't annuitize.
filename = fineng.html
Copyright © 1999 John P. Greaney, All rights reserved.