Close Close
Popular Financial Topics Discover relevant content from across the suite of ALM legal publications From the Industry More content from ThinkAdvisor and select sponsors Investment Advisor Issue Gallery Read digital editions of Investment Advisor Magazine Tax Facts Get clear, current, and reliable answers to pressing tax questions
Luminaries Awards

Portfolio > Economy & Markets > Stocks

Boxes Are Not Classes

Your article was successfully shared with the contacts you provided.

In the September 2005 Investment Advisor, we showed that investing constrained by characteristics (value-growth and market capitalization) costs investors almost 300 basis points per year in performance. Requiring managers to hold stocks that are limited to certain value-growth and market capitalization characteristics hurts performance because all of the manager’s best stock selections do not necessarily fit into that assigned box.

In presenting these findings at various conferences, many advisors have nevertheless told us why they still think constraining managers is good for their clients. One of their arguments is the notion that by constraining managers to characteristic boxes, the investor or advisor is performing asset allocation. The foundation of that position is the incorrect belief that the various characteristic boxes are asset classes, and that a manager can therefore specialize in a particular asset class. This article, based on our white paper, available at, will show that characteristic boxes are not asset classes, and that allocating among various characteristic boxes is useless at best.

Let’s begin by defining and describing an asset class. To be an asset class, a group of investments should meet three criteria:

Compositionally Unique. This means that an asset’s class should be clear and that different analysts should agree on its classification. It should be obvious to everyone in the market into which class a particular asset is placed. If this is not the case, then the definition of asset class depends on the person or organization performing the classification. Under that scenario, the asset class has little or no commonality across market participants and it would be very difficult to use it as a basis for constructing and managing portfolios. That’s because when an investor wishes to invest in the XYZ asset class, what that investor actually invests in depends on the person or the organization doing the classification.

Low Correlation. Asset allocation among stocks, bonds, real estate, commodities, and cash offers attractive risk/return tradeoffs because those five asset classes have returns that are low in correlation. If, instead, they were strongly correlated, there would be little diversification benefit and there would be just one optimal portfolio; the asset class with the highest return-to-risk ratio. With a strong positive correlation, these investments would just be a linear combination of each other. To rate the distinction of being an asset class and therefore to serve any diversification benefit, an asset’s returns must be low in correlation to other asset classes.

Stable Composition. Finally, the composition of membership of an asset class should be stable over time. Assets should not be changing class frequently, if at all, over time. As an example of this, imagine the case where bonds and stocks constantly moved into each other’s asset class. How useful, then, would the concept’s stocks and bonds be for managing money?

Our evidence shows that the characteristic boxes defined by value-growth and market capitalization fail all three criteria. It is clear that they are not asset classes and that allocating among them is not asset allocation.

We tested the characteristic boxes to see if they met the three criteria, and we found failure on each count.

Failure One: Not Compositionally Unique

Market observers can easily sort assets into the following categories; stocks, bonds, cash, real estate, and commodities. There is not much disagreement. Convertibility aside, stocks are stocks, bonds are bonds. With characteristic boxes, however, the opposite is true: There is tremendous disagreement between classification systems and even within each classification system.

Focusing on the size dimension, five of the best known information services–Morningstar, Morgan Stanley, Standard & Poor’s, Wilshire, and Russell–have very different definitions regarding market capitalization, as shown in the “Wide-Ranging Cap Ranges” chart (below). “Large cap” ranges anywhere from 67% to 92% of the total U.S. market capitalization. The services’ various definitions of “small cap” range from Morningstar’s 3% to Morgan Stanley’s 12% of U.S. market cap. In fact, Russell denies the existence of mid-cap as a separate asset class altogether, designating the smallest 25% (of total market cap) of their large-cap index as “mid cap.”

Focusing on Morningstar (as of 5/31/05) and S&P (as of 6/15/05), there were 1,301 stocks in common. While S&P agreed that every stock Morningstar had as large-cap belonged there, that’s where the agreement ends. Only half of the Morningstar mid-cap stocks were categorized that way by S&P, and a little less than 75% of Morningstar’s small-cap stocks were called small cap by S&P. Over all, there was slightly more than 30% disagreement between the two information services on how individual stocks should be categorized in these supposed “asset classes.”

We also performed a similar cross-service comparison for the value-growth dimension. S&P does not have a “core” category. Among the stocks not in the Morningstar core category, S&P and Morningstar disagree on 20% of the remaining value or growth stocks. That is a remarkably large area of disagreement for supposed “asset classes.”

Within each service, the break points for cap size are not crisp and distinct, and all the services display some overlap. Fully 50% of the Morningstar “mid-cap” stocks are smaller than the largest “small-cap” stocks in the S&P classification, and a little over 23% of the “large-cap” stocks are smaller than the largest “mid-cap” stocks in S&P.

This is not a case of who is right and who is wrong or which system is best. It is a case of characteristic boxes failing the first criteria for being asset classes. We assert that they are not compositionally unique, but rather are compositionally common or blurred. Therefore, they do not pass the first test of an asset class.

Failure Two: High Correlation

Historical returns data for market capitalization indexes (small, mid, and large) were obtained from four services: Wilshire Dow Jones, Wilshire US, Standard & Poor’s, and Morgan Stanley. The first two services included monthly returns from December 1985 through May 2005; the S&P series included monthly returns from August 1995 through May 2005; and the Morgan Stanley series included daily returns from September 1992 through June 2005. The average correlation was 0.92 for size combinations over all the services, with the lowest average correlation being 0.81 for the large/small comparison.

Clearly if diversification is limited to various size categories, there is little benefit. Instead, the various size categories should be viewed as a linear combination of one another and the decision should be focused on investing in the one size category providing the best risk/return performance rather than diversifying among all three categories. In other words, small, mid and large are not separate asset classes.

Correlations were also calculated for value-growth indexes for the four services. Once again correlations were consistently high, with an average correlation of 0.80, indicating that value and growth stocks move largely in tandem. As with market capitalization, the diversification benefits are low and thus, even when we consider the added dimension of value-growth, characteristic boxes look more like linear combinations of one another rather than unique asset classes. This can be seen in the “High Correlations” table (above) that shows the size/value-growth correlations for the Morgan Stanley series.

Diversifying across 13 widely recognized asset classes, such as stocks, bonds, money markets, international, and so forth, is seven times more effective than diversifying across characteristic boxes. This is because potentially 74% of portfolio volatility can be eliminated by diversifying over the widely recognized asset classes, while only 11% can be eliminated by diversifying across characteristic boxes (see our aforementioned white paper for more details regarding these calculations). Characteristic boxes fail the second criterion–low correlation–which is necessary to be distinct asset classes.

Failure Three: Unstable Membership

If characteristic boxes are asset classes, a stock should not easily move from one asset class to another. Looking at the S&P 1500 Index data over the period June 1998 to June 2005, U.S. equities do not satisfy this criterion. S&P re-categorizes stocks twice a year during the 2nd and 4th quarters. During these quarters, an average of one stock in nine changes categories. Furthermore, 90% of these changes come as a result of changing a growth or value judgment, and 10% came from a size re-classification. Still, the fact that an average of more than 10% of stocks change classification every six months calls into question the usefulness of characteristic boxes as asset classes. Thus, characteristic boxes fail our third and final asset class criterion.

U.S. Stocks Are One Asset Class

Stocks are stocks. The U.S. stock market is one asset class, not six or nine. The boxes are not asset classes because they are not compositionally unique. As we have shown, information services categorize the same stock differently. Characteristic box returns are highly correlated, so there is minimal risk reduction when combining them in a portfolio. Finally, the composition, or membership, of these boxes is not stable over time. They fail all three criteria for qualifying as asset classes, so allocating among characteristic boxes is a fairly useless activity.

Moreover, this calls into question the notion of specialization, such as a manager who specializes in small-cap value or mid-cap growth. First, as we have shown, small cap is largely in the eye of the beholder, as there is no universally agreed-upon definition of small cap. Second, does it take different skills to analyze stocks in those two boxes? Probably not!

If stocks are stocks, it’s all the same and the notion of specializing in a characteristic box disappears. In other words, small-cap growth, for example, is not something that a manager can specialize in because the stocks in that box are just stocks. They are not a separate asset class.

In summary, the current practice of allocating funds across characteristic boxes is an ill-advised exercise. In our next article for Investment Advisor, we will address the issues of risk and style boxes.

Craig Callahan is founder and president of Icon Advisers in Denver and can be reached at [email protected]. Mr. Callahan is also a principal at Athena Investment Services. C. Thomas Howard is a Professor of Finance at the Daniels College of Business in Denver. He can be reached at [email protected].


© 2024 ALM Global, LLC, All Rights Reserved. Request academic re-use from All other uses, submit a request to [email protected]. For more information visit Asset & Logo Licensing.