Bill Sharpe’s Brave New World

If you’re of a certain age, you’ll remember a TV series called The Millionaire, about ordinary citizens upon whom large wealth is suddenly bestowed. And each week, like clockwork, another hapless average Joe or Josephine wound up having their money managing them instead of the other way around. Today a similar, if smaller, drama is taking place in millions of households, as companies switch from traditional paternalistic defined benefit pension plans to employee directed 401(k) structures.

How competently will these nest eggs be invested? (Let’s define competently, say, as the ability to obtain risk-adjusted returns within 100 basis points of the market portfolio, annualized over 30 years.)

Since most investment professionals fall far short of this, the odds that the average retiree will be able to do so must be small indeed. If he is intelligent, industrious, and not a little lucky, he just may come across A Random Walk Down Wall Street, or Common Sense on Mutual Funds, passively manage his assets at low cost, and achieve his goal. Again, not likely, particularly considering the notoriously high expenses of 401(k) plan funds. Odds are his advice will come from a friend, stockbroker, or TheStreet.com. Not large enough to attract professional management, one can only imagine the asset allocation process involved in these accounts.

Pollsters are fond of asking folks to rank the seriousness of various foreign and domestic problems facing the nation. Myself, the number one wake-up-in-the-night-staring-at-the-ceiling worry is the tidal wave of boomers surging towards retirement with no visible means of support besides Social Security. From rebels without a clue to rebels without a plan in just 40 short years.

Enter Bill Sharpe. Nobel Prize recipient, inventor of the Capital Asset Pricing Model (CAPM), and midwife of Modern Portfolio Theory, he rides to the rescue of these forgotten millions with his Financial Engines ("FE") advisory service. Originally set up to provide advice to the beneficiaries of large corporate defined contribution plans, it is now open to the general public. And make no mistake about it—asset allocation advice from Professor Sharpe is like volley instruction from Pete Sampras. Of course, Bill doesn’t actually look over your shoulder in person and pencil out an efficient allocation for you, but he has done the next best thing. For $14.95 per quarter you can log on, describe your personal situation, assets, and available investments to the Java-based expert system, and obtain a roadmap for your finances.

Just how good is this advice? To answer this question Efficient Frontier strapped on Financial Engines and roared off into the future of asset allocation. (And also a bit of disclosure. The author is a principal in a financial advisory firm. If widely successful, FE is not expected to do good things for the average investment advisor.)

Retirement Calculator: A+

The good news is that a substantial part of the service is available for free. Even better, the free part of the site is superb. For starters, you’ll need a Windows 95-based system with a Java-capable browser. It will take you about 15 to 30 minutes to fill out your personal information, including current income and retirement savings, retirement income requirements, etc. Next you will have to enter your current retirement assets. This takes a while, as you’ll have to enter individual fund names or tickers and share/dollar amounts.

After this is accomplished, the program will produce a forecast of how likely it is that your retirement goals will be met, the median, upper and lower 95% confidence limits for your retirement income and nest egg, and the amount of risk (expressed as the worst annual loss) you might expect. This service is light years ahead of any retirement calculator I’ve seen, and if you’re shaky about these sort of calculations, it alone is worth the time spent climbing the program’s substantial learning curve.

Advice Quality: B+

The next step is to ask the program how your allocation might be improved. This is what you’re paying your $14.95 for. In order to evaluate a passively managed approach, I first examined the following asset classes, each represented by a discrete mutual fund. For all but 2 classes (international bonds and precious metals equity) index funds were used. Over a 30 year horizon, Financial Engines predicted the following inflation-adjusted returns, expressed as median, and best case/worst case scenarios (upper/lower 95% confidence limits):

Fund Ticker	Asset Class	Median Return	Best Case	Worst Case
VBISX	Short Bonds	1.78%	2.50%	0.60%
VBIIX	Intermed. Bonds	1.98%	2.98%	0.18%
VBLTX	Long Bonds	2.61%	4.50%	-0.66%
VGSIX	REITs	3.54%	7.99%	-1.12%
VFINX	S&P 500	5.79%	10.99%	1.37%
VEXMX	Wilshire 4500	5.84%	11.53%	0.11%
NAESX	Russell 2000	5.66%	11.76%	-0.75%
DFSCX	CRSP 9-10	5.38%	11.42%	-1.01%
VIGRX	Big Growth	5.65%	11.25%	-0.17%
VIVAX	Big Value	5.36%	10.31%	0.21%
DFSVX	Small Value	5.34%	10.96%	-0.43%
VEURX	MSCI Europe	4.37%	9.76%	-1.21%
VPACX	MSCI Pacific	2.92%	10.11%	-4.47%
VEIEX	Emerging Mkts.	2.87%	9.95%	-4.27%
DFCSX	Cont. Small	3.56%	8.36%	-1.34%
DFUKX	UK Small	2.84%	8.56%	-3.03%
DFJSX	Japan Small	0.62%	10.26%	-9.82%
VGPMX	Prec. Metals	-0.43%	9.79%	-11.06%
BEGBX	Int'l Bonds	1.26%	4.59%	-2.12%

Too reiterate, these are inflation-adjusted returns.

I do have a few quibbles. First, the bond returns are a bit low. After all, 30-year inflation-adjusted treasuries now offer almost 4%. Even taking into account reinvestment risk, a 3.5% real bond return seems reasonably certain. Second, there seems to be no value premium. Decades of research show that value stocks produce above market returns. Perhaps the good professor is just being cautious. Finally, foreign stock, and particularly small foreign stock returns, seem way too low. Consider that UK small stocks currently yield in excess of 3% in dividends. The model seems to be expecting zero real capital appreciation in this area over the next 30 years. Hmm.

But on the whole, the returns assumptions are not too far off, and it is certainly better to err on the side of caution.

I next set up a hypothetical 50 year old investor with a $1,000,000 nest egg to which no further contributions will be made, and a required annual retirement income of $100,000 starting age 65. FE assumes that he and his wife will receive $40,000 annually from Social Security, so they will require $60,000 additional income. Using the above assets, FE spits out the following allocations:

					"Risk Level"
Asset Class	0.34	0.4	0.5	0.6	0.7	0.8	0.9	1.0	1.1	1.2	1.3	1.4	1.5	1.6	1.64
Short Bonds	100%	76%	53%	33%	25%	25%	25%	23%	8%
Intermed. Bonds		9%	21%	31%	25%	21%	12%
Long Bonds					3%			5%	12%	11%	1%
REITs		5%	4%	4%	4%	1%
S&P 500		3%	12%	19%	25%	25%	25%	25%	25%	25%	25%	25%	4%
Wilshire 4500			2%	3%	5%	14%	22%	25%	25%	25%	25%	25%	25%	21%
Russell 2000												8%	25%	79%	100%
CRSP 9-10								1%	2%	3%	4%	8%	21%
Big Growth								2%	4%	7%	11%	25%	25%
Big Value								1%	4%	7%	11%
MSCI Europe				1%	3%	6%	8%	10%	11%	12%	13%	3%
MSCI Pacific		1%	3%	4%	5%	5%	6%	6%	7%	8%	9%	6%
Emerging Mkts.		3%	3%	3%	3%	3%	2%	2%	2%	2%	1%
Cont. Small		1%	1%	1%	1%
UK Small		2%	1%	1%	1%
% Success Rate	<5%	11%	33%	46%	54%	59%	63%	65%	67%	68%	69%	69%	67%	65%	65%
Worst Annual Loss	0%	0%	0%	2%	3%	5%	7%	8%	10%	11%	13%	15%	17%	19%	20%

Pay particular attention to the next to last row, which indicates the chance of successfully meeting his needs. One rational response would be to decrease retirement requirements to $90,000, which would substantially increase your odds of success, or to $80,000, which would make them certain. And in any case, since no benefit accrues from increasing risk level above 1.2, it makes no sense to use a more daring portfolio than this.

Several comments are in order. First and foremost, once you reach a portfolio risk level that is 100% in equities (at about 1.3 in the above example), further increases in risk dramatically change the relative proportions of the asset classes within the equities portfolio, resulting in portfolios that no one could reasonably describe as "optimal." By the time you reach the highest portfolio risk column on the right, FE puts the entire portfolio into just one asset, the Vanguard Small Cap Index Fund. Ironically, the CAPM, which Mr. Sharpe invented, would instruct us to utilize margin to increase portfolio risk once the allocation has reached 100% stock, rather than concentrate the portfolio in just the highest risk equity asset classes. FE does not incorporate this option, however, presumably because the use of margin is not allowed in 401(k) plans. In fact, within the stock portion of the portfolio, the CAPM would dictate identical equity asset class allocations over the entire range of risk. I spoke to Chris Jones of FE about this, and he explained that the equity asset class percentages move around a bit because there really is no "riskelss" asset in real portfolios. However, I'm not entirely satisfied with such wide swings in relative equity class recommendations. What is clear is that FE's recommendations cannot be relied upon at very high levels of risk.

Next, there is no allocation to US small value stocks, and almost none to small foreign stocks. This is of course an artifact of the low returns projections for these assets. One wonders if this is intentional, so as not to unduly upset clients with unconventional portfolios. Lastly, look at the long row of 25% allocations to the S&P 500 and Wilshire 4500. Mr. Jones tells me that the outputs are not cooked, or "constrained" in any way, but this output certainly walks, looks, and quacks like a seriously constrained duck.

Since FE penalizes funds with high expenses and cash positions, it tends to favor index funds. This is not necessarily a bad thing; most retirement investors would be well served by selling their actively managed funds and indexing. However, this is an optimizer, and it is pretty easy to trick it into doing stupid things when you mix active and index funds.

To demonstrate this, I fed 3 assets into the model: a domestic stock fund, an emerging markets stock fund, and the Vanguard Short Term Corporate Fund as the diluting bond asset. First, I used a domestic index fund (Vanguard 500 Index Trust) and an emerging maket actively managed fund (Templeton Developing Markets A). Not surprisingly, it returned only a few percent contribution to the Templeton fund over most of the range of risk. This is to be expected—the Templeton Fund, already starting with the low emerging markets expected return, is further penalized by its 2% expense ratio.

But switch things around and use an active domestic fund (Tweedy Browne American Value) and the Vanguard Emerging Markets Index Fund, and you get a really biziarre set of portfolios, starting with 100% Vanguard Emerging Markets at the high risk end. This is because the Vanguard Emerging Markets Fund has a higher expected return than the Tweedy Fund. Even the 1.25 risk portfolio is 50/50, and the low .75 risk portfolio still has 11% emerging markets, plus 45% domestic stock, and 44% short bond. These are definitely not your father's portfolios.

At the end of the day FE is an optmizer. And optmizers are like chainsaws; useful for circumscribed tasks, but dangerous in untrained hands. Even Bill Sharpe can't render the technique completely idiot proof—I have no doubt that some folks will do themselves serious harm with this tool. So be careful; set up FE with only index funds in each asset class, as was done above. Where index funds are not available, use the best actively managed fund you can buy for that asset class. But do not mix actively managed and index funds from different asset classes in your analyses. If you do, you will likely get some very strange allocations.

If you fuel the engine exclusively with index funds, you will get allocations which are both reasonable and efficient. It’s a fact of life that nobody much likes anyone else’s allocations, and I'm not wild about many aspects of the engine's outputs. Still, the above portfolios represent a vast improvement over most folks’ 401(k) strategies. FE’s most attractive feature is that if you’re unhappy with your retirement forecast (accompanied by cute little cloudy/sunny weather icons), you can easily adjust your risk tolerance, savings amount, or retirement income up or down. This feature is a facile and effective way of learning the interplay between savings, liabilities, risk, and return. FE gets a well done here as well.

FE Meets the Real World: C+

Just as there is a difference between experiencing an airplane crash in a simulator versus in a real airplane, so too there is a large difference between simulating losing 15% of your money in an applet and living through the real thing. Only time will tell what will happen when millions of investors simultaneously sustain significant losses in their FE-designed portfolios. How many will stay the course? (And how many will join in a class action which might make the tobacco settlement look like lunch at Burger King?) I don’t have the answer to this one, and I don’t think that that FE does either.

The folks from Palo Alto are both intellectual and social pioneers, and I wish them every success in doing well by doing good. If they fail, we will at least have learned something about the nature of the retirement planning crisis which looms ever larger over our aging population. And if they succeed, then they will not be alone for long, and this too will be a good thing for the millions of small investors saving for their futures.

copyright (c) 1999, William J. Bernstein