Sizes, Styles and Sectors, Oh My!

Over the past three months, we've compared the different indexes that subdivide the equity market based on size (market capitalization), style (growth versus value), and economic sector. The question we're going to address this month is whether or not it makes sense to use any of these, instead of simply investing in an index that covers the market as a whole (e.g., the Russell 2000, the Dow Jones Total Market Index, or some combination of the S&P 500, 400, and 600 indexes).

Let's start with a basic axiom. At the end of the day, there are only two reasons people choose one investment over another. Either they think its price is less (more) than its fundamental value, or they see other people buying (selling) it. Given the relatively uncertain pricing dynamics in the current market that we have noted above, we think you have to make a pretty good case in one of these areas to justify departing from the "no brainer" option of simply buying the market index. With that in mind, let's look at the arguments for making size, style and sector bets.

Let's start with buying the market index as a whole, which is the alternative against which the size, style, and sector tilts will be compared. First (as is our prejudice), some data. Our benchmarks for the following analysis will be the Dow Jones Total Market Index, the Russell 3000 Index, and the S&P 1500. The first two can be purchased as iShares on the Amex. The S&P can be constructed by purchasing the right amounts of the S&P 500, 400, and 600 indexes. The DJTMI maintains a constant size based market capitalization based weighting. Companies in the top 70% of market cap make up 74% (by market cap) of the DJTMI, companies in the next 20% of market cap make up 21%, and companies in the bottom 10% of market cap make up 5% of the index. By comparison, in the Russell 3000, the top 200 companies (by market cap) make up 70% of the index, the next 800 companies make up 22% of the index, and the bottom 2000 companies (aka, the Russell 2000) make up 8% of the index by market cap.

Within the S&P 1500, the S&P 500 accounts for 90% of market cap, the 400 for 7% of market cap, and the 600 for 3% of market cap. Despite their different composition, all three indexes delivered similar performance over the period for which comparable data is available. Between January, 1994 and September, 2000, the DJTMI had an average annual return of 21.01% and a standard deviation of 17.12% (or 1.23% of return per 1% of risk); the Russell 3000 had an average annual return of 20.83% and a standard deviation of 16.78% (or 1.24% of return per unit of risk), and the S&P 1500 had an average annual return of 19.85% and a standard deviation of 16.63% (or 1.19% units of return per unit of risk).

Could you have done better over the 1/94 to 9/00 period by buying a different mix of market capitalization weightings? Let's look at the large caps first. If you bought the S&P 500 in January, 1994, you would have realized an average annual return of 21.62% through September, 2000, with a standard deviation of 17.06%, or 1.27% of return per 1% of risk. Buying the Russell 1000 would have delivered pretty much the same performance: average annual return of 21.62%, and standard deviation of 17.01%, for the same 1.27% of return per 1% of risk. In short, by buying large caps instead of the market as a whole, you could have realized 3 more basis points in return per unit of risk. If there is an explanation for this, it is probably that institutions using a trend following strategy preferred the large caps because of their liquidity, and were willing to pay up a bit for it.

By way of comparison, buying the S&P midcap 400 would have generated average annual returns of 21.09% and with a standard deviation of 19.76% or 1.07% of return per 1.00% of risk, while buying the S&P smallcap 600 would have generated average annual returns of 14.93% and with a standard deviation of 19.96% or .75% of return per 1.00% of risk. You would have done worse had you chosen the Russell 2000 as your small cap index over this period, earning average annual returns of 14.37% with a standard deviation of 20.89%, or .69% of return per unit of risk. So far, it seems that you would have done a little bit better by buying the large caps instead of the market as a whole, but not by much, as the former make up the large majority of the total market indexes. This is proven out by our optimization software. We used the S&P 500, the Midcap 400 and the Russell 2000 as our asset classes. We first attempted to find an allocation that bettered the 20.83% average return delivered by the Russell 3000, without exceeding its standard deviation of 16.78%. The best we could do was 21.03%, using a mix of 92% large cap and 8% small cap. The results were actually worse when we reversed the test, and tried to take on less risk while matching the R3000's 20.78% return. The best we could do was to reduce standard deviation to 16.71%, by taking on 89% large cap exposure and 11% small cap.

What about a style tilt instead of a size tilt? If you bought the Russell 3000 Growth Index, you would have realized average annual returns of 24.04% over the 1/94 to 9/00 period, with a standard deviation of 20.59%, or 1.17% of return per unit of risk. If you had purchased the Russell 3000 Value Index, you would have realized average annual returns of 17.38%, with a standard deviation of 15.89%, or 1.09% of return per unit of risk. In short, by pursuing a pure growth or value strategy, you would have done worse on a return per unit of risk basis than if you had purchased the market as a whole in the form of the Russell 3000 index. Why? Because by taking a pure style approach you not only give up the returns on the other style, you also give up the diversification benefit that comes from investing in both styles. As is so often the case in life, this is another example of the whole amounting to more than the sum of its parts.

Beyond this, there is another issue with respect to style indexes. In a nutshell, they don't correspond well to the two major logics that drive investment decisions. To begin with, price/book (or any of the other proxies used to separate companies into the growth and value categories) tells you nothing about either (a) whether or not the company is undervalued relative to its fundamentals, or (b) whether or not it has substantial price momentum. For example, a firm can have a low price to book because investors don't fully understand its growth prospects, or because its products are obsolete. Similarly, a company with a high price to book could be wildly overvalued by investors in the middle of a bubble, or it could be spectacularly undervalued because investors don't fully appreciate the value of its growth potential. Or it could have just taken a big write off on its way to bankruptcy, and be the farthest thing from a momentum stock.

Well then, what about combining a size and style tilt. If large cap growth was your taste, and you had invested in the S&P 500 growth index, you would have earned average annual returns of 24.82% over the 1/94 to 9/00 period, with a standard deviation of 19.54%, or 1.27% of return per unit of risk. If you had invested via the Russell 1000 growth index, you would have earned average annual returns of 25.02%, with a standard deviation of 20.56%, or 1.22% or return per unit of risk. On the other hand, if your taste ran to large cap value, and you had invested in the S&P 500 value index, you would have enjoyed average annual returns of 18.35% per year, with a standard deviation of 16.81% or 1.09% of return per unit of risk. If you had invested in the Russell 1000 value index, the comparable figures would have been 17.98%, 16.42%, and 1.10%.

What about midcaps? If you had purchased the S&P 400 growth index, you would have earned average annual returns of 25.70% over the 1/94 to 9/00 period, with a standard deviation of 26.18% or .98% of return per unit of risk. The S&P 400 value index would have generated exactly the same amount of return per unit of risk, with an average annual return of 17.02% and a standard deviation of 17.40%.

If your taste had run heavily to small caps, you wouldn't be happy today. By buying the S&P 600 growth index, you would have received average annual returns of 14.55%, with a standard deviation of 24.43%, or only .60% of return per unit of risk. Buying the Russell 2000 growth index instead wouldn't have helped. In this case, you would have earned average annual returns of 16.23%, with a standard deviation of 28.56%, or just .57% of return per unit of risk. The story on the small cap value side isn't much prettier. Buying the S&P 600 value index would have generated average annual returns of 14.93% with a standard deviation of 17.17%, or .87% of return per unit of risk, while buying the Russell 2000 value index would have generated average annual returns of 12.36% with a standard deviation of 14.59%, or .86% of return per unit of risk. When we combined all six inputs in our optimization software, we once again found that we could achieve very little improvement on the performance of the Russell 3000: a .92% gain in return with the same level of risk, or a .67% reduction in risk for the same level of return. In both cases, the weighting of large caps was about the same as in the overall market index.

What have we learned up to now? Buying the market index as a whole, in the form of either the Dow Jones Total Market Index or the Russell 3000 would have generated, respectively, either 1.23% or 1.24% of return per unit of risk. Of all the eight possible size/style combinations, the only one that exceeds this (and only in the case of one of the two possible indexes in this category) is the S&P 500 Large Cap growth index, with 1.27% of return per unit of risk over the 1/94 to 9/00 period.

Thus far, not exactly an overwhelming endorsement of size and style tilts away from their underlying weights in the market index portfolio.

But couldn't you come up with a weighting of size/style indexes that is different than, and outperforms the market index? Practically speaking, the answer to that question is no. The bottom line is that over the past six years, large cap stocks have been the primary drivers of market index returns (perhaps, as we said, because of their attractiveness to institutional trend followers), and are already quite heavily weighted in the major market indexes. In a nutshell, there is simply very little room to improve risk adjusted performance by overweighting large cap growth relative to its current percentage of the overall market index.

Ah, you say, but what about sector indexes? It is true that had you invested in the S&P Technology Index, you would have realized average annual returns of 40.84% between 1/94 and 9/00. But you would also have experienced a standard deviation of 36.46%, or only 1.12% of return per unit of risk you took -- inferior to the market as a whole. You would have done better, however, with the Dow Jones Health Care Index, which generated average annual returns of 27.57%, with a standard deviation of 20.75%, or 1.33% of return per 1.00% of risk. Then again, had you invested in the S&P Basic Materials Index, you would have realized average annual returns of only 7.23%, with a standard deviation of 21.25% -- that's .34% of return for every 1.00% of risk you took on.

Let's face it: the trend buying was in technology and health care, not basic materials. On the other hand, maybe the latter is more overvalued. Who knows? And that's the point. There is no doubt that we can use historical data and out optimization software to come up with a sector by sector allocation that would have delivered more return per unit of risk than the market as a whole over the 1/94 to 9/00 period.

But hindsight, as they say, is 20/20. Because we already knew what had happened, our backwards looking allocation would capitalize on the bubbles and undervaluations, while limiting the damage from the "low but accurate" valuations and crashes.

The point is that because of the amount of money backing trend following strategies these days, it is very hard to be confident that whatever sector allocation delivered the best performance over the last 81 months will do so over the next 81 months. For example, it is not out of the realm of possibility to guess that technology and health care stocks are in the later stages of a bubble, and that buying these indexes now could very well lead to tears later on, as the momentum traders move into energy and telecommunications. As we said, the market these days seems much more likely to be in a disequilibrium state than it has in the past. And when that is the case, the past is usually a poor guide to the future.

The bottom line is that picking sizes, styles, and sectors within the U.S. equity market are all forms of active management. And winning at active management is very, very hard to do today -- possibly harder than it has ever been, because the past is no longer as good a guide to the future as it once was. Our conclusion is that under these circumstances, discretion is probably the better part of valor. In short, given the nature of market pricing dynamics these days, we are probably better off buying the market index as a whole, and adopting a fully passive indexing approach without any active tilts.

| Sizes, Styles and Sectors, Oh My! | The New Equity Market Dynamics | Recommended Portfolio Performance |