Mid-Cap Equities

In our June, 2004 issue we examined the arguments in favor and against tilting your equity allocation toward small capitalization companies. We will now look at the pros and cons of tilting toward "midcaps." These are companies whose total market value (capitalization) lies in between those of small cap and large cap companies. We will begin with the theoretical arguments that have been made in favor of midcaps, then look at what the historical data has to say. We will then compare the indexes and related products that can be used to implement a tip toward midcaps.

Arguments in favor of a tilt toward midcaps fall into two categories: those based on fundamental (economic) factors, and those based on market (investor) related factors.

The economic argument is grounded in the notion of a "business lifecycle." When companies are started, they are small, and many fail after a short time. Those that are sufficiently well-managed first learn how to avoid losses, and then move on to the search for a profitable growth idea. Some companies never find this, and at best remain in the "small value" category. These companies' market capitalization primarily reflects the present value of their current cash flow.

A popular way to quantify this is the dividend discount model. It determines the fair market (or fundamental) value of a company by discounting its current dividends to their present value using a rate equal to the required return on equity less the expected dividend growth rate. The required return on equity is assumed to be equal to the risk free bond rate plus an additional "equity risk premium." Mathematically, the model is deceptively simple: Market Value of Equity = Dividends / (Risk Free Bond Yield + Equity Risk Premium - Expected Dividend Growth Rate). The good news is that values for two of the variables in this equation -- the current dividend on a stock (or, indeed, an entire equity market) and the risk free government bond yield are easily obtained online or from a newspaper. The bad news is that the correct values for the other two variables -- the equity risk premium and the future dividend growth rate -- are among the two most contentious issues in finance. However, regardless of the challenges of using this model in practice, it provides an excellent framework for thinking about a lot of investing issues. For example, in the case of a small company that as yet lacks a growth idea, the expected dividend growth rate is basically zero. That means that the company's value is driven by the size of its current dividend, as well as changes in the discount rate (caused by changes in the risk free bond yield and/or the equity risk premium). All else (e.g., industry sector or country) being equal, investors will often demand a higher equity risk premium for a smaller company than for a larger company, on the theory that the former's small size makes it less able to withstand the impact of changing business conditions, and therefore riskier to own.

In contrast, companies that identify a profitable growth strategy see their market capitalization increase due to an increase in investors' (quite uncertain) perception of the rate at which the company's dividends will grow in the future. Broadly speaking, these are "small growth" companies.

The next challenge a company faces is converting its growth options into real cash flows. During this phase of the lifecycle, dividends can be increasing in size, even as their expected future growth rate remains quite high, which causes the company's market capitalization to further increase. This process may be further accelerated by a decline in the equity risk premium required by investors to own shares they perceive to be less risky than those of smaller companies (not only because of their larger size and higher dividends, but also because of less uncertainty about their future growth rate). Because the market values of companies in this stage of the lifecycle are larger than those of small caps but not as big as the largest companies, they are known as "mid-caps."

Some mid-caps will progress into the third stage of the lifecycle, during which increased competition and growing size result in a reduction in investors' perception of both their future growth rates and their business risk. By this time, however, these companies are generating substantial dividends, which causes them to have large market capitalizations.

This lifecycle framework helps to clarify the fundamental argument in favor of tilting toward mid-cap companies. Stated simply, mid-cap companies should deliver higher returns than large cap companies (which also tend to dominate the return of the broad market index), with only slightly higher risk (compared to a tilt toward small caps). A closely related argument is that investors in mid-caps are also well positioned to earn additional premiums because these companies are favorite targets of larger companies making acquisitions.

As you can see, the economic argument for owning mid-caps is consistent with the idea of reasonably efficient financial markets (apart from the occasional excessive acquisition premiums paid by an over-enthusiastic CEO). This argument says that because midcap shares are riskier to own than the overall market, an investor should expect to earn returns that are also somewhat higher (but not as high as those from small cap stocks, which are even riskier than midcaps).

We should also point out that occasionally another argument is offered in support of investing in midcaps. This suggests that midcaps may deliver higher returns with lower risks than small caps, because they receive relatively less attention from investors (who, presumably, are more attracted to the latter's potential for very high returns). While this argument may apply for short periods, its effectiveness over the long-term also requires the existence of permanent obstacles that prevent smart investors from arbitraging away the implied price discrepancies between small and midcap shares (e.g., by buying the latter and selling short the former). Given the intense competition and high rewards that characterize the world of active investment management, we cannot believe that such obstacles exist.

Let's now move on to an examination of the historical data, and see if it agrees with the theoretical arguments we have just outlined. The first problem you confront when trying to do this is the existence of multiple indexes that are intended to measure the performance of midcap companies. We will describe these differences in more detail below. For now, we will note that our quantitative assessment used three indexes that are based on very different underlying methodologies: the Standard and Poor's 400, the Russell Midcap 800, and the DowJones MidCap Index. The second problem you confront is that since most of these indexes were introduced quite recently, you have relatively little data available to work with. Our analysis is therefore based on monthly returns that only cover the June, 1995 to December, 2003 period.

The following table shows summary data for each of our three indexes, as well as for the Russell 3000, a broad market index (it covers about 98% of total U.S. public equity market capitalization). The data include the following measures: (1) the average annual return for each index. (2) The standard deviation of returns (also known as "volatility"), which measures how widely individual returns are distributed around the average. The higher the standard deviation, the riskier the asset. (3) The skewness of returns, which measures the degree of asymmetry in their distribution. Negative skewness implies higher risk, because it means that more returns fall below the average than above it. (4) The kurtosis of returns, which measures the extent to which returns are grouped close to or far away from the average. High kurtosis implies more returns far away from the average (or "extreme events" as they are sometimes referred to). Whether or not this implies higher risk depends on the skewness measure. If it is negative (that is, if more returns lie below the average than above it), then high kurtosis (that is, a high probability of returns that are far away from the average) implies higher risk, due to the presence of more big unpleasant surprises on the downside than nice surprises on the upside. (We also note that prospect theory suggests that investors don't value these equally, with downside surprises hurting roughly twice as much as upside surprises feel good). And (5) as one measure of return relative to risk, we also include a variable equal to the average return divided by the standard deviation. A higher value for this variable is good, because it shows you are getting more return per unit of risk.

This table illustrates a number of interesting points. At first glance, it looks like two out of the three midcap indexes provide, as theory predicts, higher return and higher risk than the broad market index. In fact, if you only looked at the return/standard deviation measure, you might even conclude that a midcap tilt offered a superior risk/return trade-off compared to the broad market index. However, when the meaning of "risk" is expanded to include skewness and kurtosis, it becomes clear that this may not be the case. In particular, the historical index data suggest that a midcap tilt exposes an investor not only to higher volatility, but also to somewhat greater extreme event risk.

Another technique for evaluating the pros and cons of a midcap tilt is called an Information Ratio, or "IR." The logic behind this is as follows. A tilt toward midcaps and away from the broad market index is a type of active management decision. One undertakes such decisions in the expectation that they will, in exchange for the additional "active risk" taken on, also produce additional active return, or "alpha." The IR is simply a measure that relates the size of alpha to the amount of active risk that was taken on to generate it.

Mathematically, in any given month, alpha is equal to the difference between the return on the midcap index and the return on the broad market index. Some months it is positive, and some months it is negative. The alpha for the overall tilt strategy is equal to the average alpha for the period being study (in this case, annualized from monthly data). The active risk taken on is defined as the standard deviation of the monthly alphas. This is also known as "tracking error" versus the broad index. Armed with that quick summary of active management math, let's look at the Information Ratios produced by our different midcap tilts over the 6/95 to 12/03 period. In the case of the S&P 400, the IR was .60 (quite an impressive outcome). For the Russell 800, it was .31 (still respectable). But for the DowJones MidCap, it was slightly negative, at (.06).

Before moving on to examine the index-specific factors that could account for these different Information Ratios, let's conclude about the overall wisdom of taking a midcap tilt. First, the good news. On the basis of the Information Ratios we found, a midcap tilt appears to make sense. Compared to the results we found in our June, 2004 analysis of small and microcap tilts, we would say that a midcap tilt makes more sense than the former, and at least as much sense as the latter. The bad news is that we can't say this with any degree of confidence, at least in the statistical sense. Due to our very short data series, none of the Information Ratios we found is statistically different from zero (at the 95% confidence level). The other argument against midcap tilts is one we've already made: even if they produce a statistically significant Information Ratio, they appear to do so by taking on more extreme event risk than the broad market benchmark.

Let's now look at the index-specific factors that caused the disparity in the Information Ratios we found. Broadly speaking, there are three ways one can construct an equity index. First, one can set the target number and size range for the companies to include, and have a committee choose them using a loose set of guidelines (e.g., for industry sector representation and liquidity). This is the approach used to construct the Standard and Poor's MidCap 400 Index.

The other two approaches are more mechanical, and build their indexes using clearly defined sets of rules. One of these starts by ranking companies according to some factor (e.g., their market capitalization), and then grouping a fixed number of companies (counting from the top down) into one index, and another fixed number of companies into another. This is the approach used to construct the Russell MidCap 800 Index. It starts with the top 3,000 public companies in the United States equity markets (including the NYSE, AMEX, and NASDAQ). The top 200 companies are included in the large cap index, the next 800 companies comprise the midcap index, and the bottom 2000 companies are the small cap index. Besides the Russell Indexes, the Morgan Stanley Capital International (MSCI) Indexes tracked by many Vanguard mutual funds and ETFs are based on this approach. MSCI assigns the top 300 companies to its large cap index, the next 450 to its midcap index, and the next 1,750 to its small cap index.

An index that includes a fixed number of companies will, by definition, cover a varying percentage of total market capitalization (e.g., the Russell 3000 covers about 98%). An alternative indexing approach fixes the percentage of market capitalization to be covered, and varies the number of companies it includes. This is the approach used to construct both the Dow Jones and the Morningstar Indexes. This includes companies that make up a fixed 95% of total market capitalization, while the latter covers 97% of the market. Both of these companies include the top 70% of market capitalization in their large cap index, and the next 20% in their midcap index. Dow Jones' small cap index includes the next 5% of market capitalization, while Morningstar's includes the next 7%.

The following table summarizes the differences between different midcap indexes:

One interesting point in this table is the difference between the number of companies in the DowJones and Morningstar Indexes, which theoretically cover the same range of market capitalization percentiles. We suspect that the cause of this difference is the fact that Morningstar employs wider "buffer zones" than DowJones. Buffer zones exist at the borders between different subindexes -- for example, between large and midcap, or midcap and smallcap. They are used to help limit turnover in the companies included in an index. This is important because higher turnover generates higher trading costs (and lower performance) for index fund managers. Buffer zones limit trading by allowing a company to remain in one index even though it no longer quite qualifies (e.g., because its market capitalization has just surpassed or fallen below the cutoff point for index membership. In these cases, small changes in companies' stock prices can have them moving into and out of the index quite frequently. Using buffer zones reduces the trading costs that might otherwise be caused by these moves.

As you recall from the previous discussion of historical performance, the S&P 400 and the Russell 800 delivered the best performance over the relatively short 6/95 to 12/03 period we analyzed, while the Dow Jones product lagged behind. Given the similarity in their construction and market coverage, had comparable historical data been available, the Morningstar product probably would have delivered results similar to Dow Jones'.

It is less clear that this also would have been true for the Russell and MSCI products, since their definitions of "midcap" are somewhat different. However, data on their respective websites shows backtested ten year average returns (through September, 2004) of 12.77% for Russell, and 12.72% for MSCI.

Perhaps the most important question we have yet to address is what could possibly account for the relatively strong performance of the S&P 400 Index, compared to the other midcap products? A recent analysis of this issue (see "The Returns of the S&P 400: Implications for Active Mid-Cap Managers" by Peter Jankovskis) concluded that much of it was due to the migration of companies from the S&P 400 into the S&P 500 Index. Because many more assets under management track the latter compared to the former, speculative investing in companies thought to be due for "promotion" has a very strong impact on their share prices and consequently on the performance of the S&P 400 Index. This brings us back to a fundamental point about the Standard and Poor's indexes: they involve a significantly higher degree of active management than their competitors. As we have noted, companies are included in these indexes not as the result of the consistent, mechanical application of a set of rules, but rather based on decisions by the S&P Index Committee. And as Jankovskis has shown, to some degree these decisions can be self-fulfilling in their results as companies migrate from the S&P 600 to the 400 to the 500 over their lifecycles.

Another logical question to ask is whether midcap indexes also exist in equity markets outside the United States. The answer is that they do, but the range of offerings is narrower. The following table lists these indexes in key currency regions:

Finally, we need to look at midcap index investment products.

In the United States, there are far fewer vehicles that track midcap, as compared to small and large cap indexes. The following table summarizes this limited product offering. Note that it excludes funds that take growth and value tilts within the midcap segment.

Outside the United States, midcap offerings, and especially midcap index products, are also relatively limited. The only fund we know of that tracks the Australian midcap index is found in New Zealand (MOZY). Canada has an iUnit ETF (XMD) that tracks the midcap index there. In the Eurozone, there is a country specific midcap index ETF in German (MDAXEX). State Street (Balzac), Robeco, and Fidelity all offer actively managed pan-European midcap funds. On the other hand, in the UK there are relatively more index funds that track the FTSE-250, including an iShares ETF and a unit trust from HSBC. There are also actively managed funds that invest in the midcap segment of the UK equity market. In addition, Barclays' Global Investors (the company behind iUnits and iShares) has recently licensed the Dow Jones Stoxx midcap indexes in Europe. As a result, the number of midcap index ETF products available in both the Eurozone and UK should increase over the next year. Finally, in Japan all the micap-oriented funds we have identified are actively managed.

To conclude, there is a good theoretical basis for expecting an investment in a midcap index to produce higher returns, but with higher risk than an investment in the broad equity market index. However, a midcap fund's returns and risks should be lower than those delivered by small cap products. Our analysis of the limited historical data available found that it is in line with this theory. When we applied the analytical technique typically used to measure the performance of active investment managers, we found that a tilt toward midcaps produced reasonably attractive (but not statistically significant) Information Ratios. However, we also found that the methodology used to construct the underlying midcap index appears to have a significant impact on these results. Last, but not least, we found that while they are relatively few in number today (in most markets of the world), the range of midcap index tracker products seems likely to grow over the next few years as more investors seek low cost ways to take this tilt in their portfolios.