Should I Use On-Line Calculators?

A subscriber once asked us, "what is your opinion of the asset allocation calculators available on many other websites?" Based on our previous research, our initial response was "pretty low." However, after a new review of the latest offerings on a number of sites, we'd like to amend that to "very low." We'll illustrate this point with a few examples from some of the sites we researched (the names of which we will withhold for reasons you'll soon understand).

Site #1 makes some interesting claims, each of which is debatable: "Your age is by far the most important aspect of asset allocation." "People with large portfolios...and people who save more each year...can invest more aggressively." "The better your outlook for the economy, the more aggressive you can be with your investments." This site produced its asset allocation recommendations based on our answers to seven questions, including our age, current savings, expected future savings per year, income requirement (expressed as a percentage of the portfolio), tax rate, risk tolerance (expressed by moving a slider from "low" to "high") and our "economic outlook" (for which we could move a slider from "poor" to "good"). Its asset allocation solution divided a portfolio between the following "asset classes": "Large Cap" (we presume this means large capitalization U.S. equities), "Mid Cap", "Small Cap", Foreign Stock, Bonds (presumably some type of U.S. bonds); "Municipals" (tax advantaged U.S. state and local bonds), and cash.

Site #2's calculator used a similar set of questions and sliders to collect our input data, but limited its recommended allocation to just five asset classes: "Large Caps", "Small Caps", International Equity, Bonds and Cash.

Site #3 asked us only four questions: "when do you need the money?", "how much risk can you handle?", "how much wiggle room do you have?" (one possible answer to this question was "if I miss my goal by a year or two I'll still be okay"), and "as the bear market intensified, did you do nothing, see an opportunity to buy more stocks, or sell?". On the basis of our answers, it produced a suggested asset allocation, divided between Large Cap, Small Cap, Foreign Equity, and Bonds. Oh, yes, and it also conveniently provided lists of suggested funds for each asset class. Said suggestions contained a wide variety of actively managed funds, but only two index funds.

Site #4 is run by a mutual fund company. It asked us eight questions to start with, including one about how much time we had to achieve our primary financial goal (the longest time frame available was "more than ten years."). After this, it divided our portfolio between just three asset classes: Stocks (presumably U.S.), Bonds, and Cash.

To its great credit, Site #5 based its asset allocation calculations on the rate of return sought by the investor, and noted that its goal was to produce the mix of asset classes most likely to achieve this return with minimum risk. Once we input our desired rate of return, it produced an asset allocation divided between Cash, Bonds, Large Cap U.S. Equity, Small Cap U.S. Equity, and International Equity. Based on a target return of 8% (which we presume was nominal, but the site didn't specify nominal or real), the site calculator told us that there was a 68% chance that the return in any year would fall between 5.66% and 10.34%. However, as this implies a portfolio standard deviation of 2.34%, we were confused by the next statement, which noted that "it is possible that in any given year you may earn only .06%." Eight percent less three standard deviations of 2.34% (which gives you the 99% probability range) would give you a minimum return of .98%, not .06%. Finally, this site also provided "helpful" links to "high performing stocks and funds that fit in the asset classes in your recommended portfolio."

We could go on with similar examples, but we won't; frequent readers are already in enough pain.

Needless to say, we have a number of problems with the methodology which underlies these and similar asset allocation calculators. First, they confuse tilts within asset classes (e.g., large and small cap companies) with the asset classes themselves (e.g., U.S. equity). Moreover, they present no arguments on the potential advantages and disadvantages of taking such tilts. Second, they generally employ too few asset classes in their solutions, and thereby forego potential diversification benefits.

Third, their recommendations are typically based (site #5 being the exception) on the results of some type of "risk capacity" survey. In other words, their logic proceeds from determining the maximum risk a person is comfortable with, to the translation of this into a maximum portfolio standard deviation (the statistical measure of the dispersion of an asset or portfolio's returns around its mean), to the calculation of an asset allocation that maximizes expected returns subject to this risk constraint. In our experience, the real world doesn't work like this. The majority of people we know start with a set of financial goals they'd like to achieve within a certain time period, along with their current and expected future savings. They then use these starting points to "back into" the minimum compound rate of return they need to earn on their portfolio to achieve their goals. The asset allocation challenge then becomes how to maximize the probability of achieving this rate of return, at the lowest possible risk. If an investor isn't comfortable with the risk implied by the asset allocation solution, he or she must confront some realistic alternatives: accept the risk, reduce his or her goals, stretch out the time period for achieving them, or increase annual savings. Unfortunately, as they are currently structured most online asset allocation calculators skip these practical realities.

Fourth, in our survey of different sites, it was never clear what underlying assumptions were being used in the asset allocation calculations. How were the asset classes defined? Were the assumed future asset class returns, standard deviations, and correlations based on historical averages? If so, over what period were they estimated? Or were the calculator's assumptions based on the outputs of a forward-looking asset pricing model? If so, how does that model work? Or were they based on the combination of these two approaches (theoretically the most defensible solution)? If so, what weights were given to the historical and asset pricing model assumptions? The bottom line is that when you use online calculators you don't know the answers to these critically important questions, and therefore have no way of either judging the quality of the result or comparing them to results produced by other calculators.

Fifth, we have a very strong suspicion that in many cases, the underlying methodology used by these calculators to produce their asset allocation recommendations is mean/variance optimization (a straightforward application of linear programming). The basic problem we have is that MVO is a technique designed to produce optimal solutions to problems involving one year holding periods. Its use for longer period problems, which contain a mix of goals and constraints -- a combination that characterize the situation faced by many users of these calculators -- is much more difficult to theoretically justify (see our blue button "Asset Allocation Methodologies" for a much longer discussion of this subject).

Finally, as we repeatedly point out in our writing on the subject, all asset allocation methodologies are subject to some very important limitations, including non-normal (in the statistical sense) historical returns for most asset classes, and non-stationary (i.e., changing) underlying returns generating economic processes. Unfortunately, none of the online asset allocation calculators we examined disclosed these limitations to potential users. This violates one of our fundamental principles: you should never use an analytical tool unless you also make clear the potential limitations of the results it produces. Sometimes, knowing what you don't know is as important as knowing what you do. In sum, we found all the online calculators we examined had serious limitations.

Download Free Section PDF
Subscribe to The Index Investor