After 2002, strong preference for one of the 4 styles declined dramatically with Small Value being only a modest favorite through 2006, and Large Growth only modestly thereafter.

The reason for this decline in favored styles is unclear. The rising availability of data on style performance has been cited as a possible cause, but that would seem to increase rather than to decrease the strength and duration of style preference.

Another possibility is the realization starting in 2000 with REITs and in 2002 with Internationals that there were great opportunities for U.S. investors outside of the U.S. stock market. As these markets have flourished (with the aid of a declining US Dollar for Internationals) there may have been less interest in finding under valued U.S. styles and therefore no strongly favored style emerged.

Whatever the cause, the result is a reality. Style Index Investing is by no means dead but it needs to be supplemented by other options, both within and outside of the U.S. market. If, in the future, interest wanes away from REITs and Internationals, SII needs to remain an option and may again become a potent source of outperformance.

What Follows is a detailed description of the options available and a system for choosing between them to assemble a winning portfolio. Discussion and data are limited to Vanguard index Funds and ETFs, but are applicable to similar offerings from other fund companies.

Vanguard has a myriad of index funds, most with ETF counterparts. They cover every conceivable style (Large and Small, Growth and Value, and mixtures) of U.S. Stocks with 10 that are specific to industrial sectors such as Energy, Financials, Healthcare, etc. In addition there are Internationals not specific to any one country, but to regions such as Europe, Pacific, Emerging Markets, and combinations of them.

All are based upon MSCI indexes and their returns mimic those indexes very closely, less a small fraction of one percent for expenses.

To judge the performance of any one index, U.S. or foreign, it is helpful to relate it to a common denominator. In the U.S. this is usually the S&P 500 or the Dow Wilshire 5000. The problem with these indexes is that they are weighted by the capitalization of the stocks that make them up. The top 50 to 100 companies have such huge capitalizations that whether we're dealing with 500 or 5000 total companies, the top 50 to 100 dominate all the rest. In short, both are essentially Large Cap Indexes.

For institutions and large investors dealing with portfolios in the billions of dollars these indexes make sense because their portfolios are mostly limited to large stocks. For the average investor though, small stocks are not a problem, so their menu is essentially without limit. Thus, instead of using a large stock dominated index as the common denominator, we calculate an index weighted equally between Large Growth, Large Value, Small Growth, and Small Value represented by the Growth and Value components of the MSCI Large Cap 300 and the Russell 2000. This serves as the common denominator to which the returns of all the Vanguard indexes ( U.S. and International) are compared.

The bottom portion of the accompanying graphs track the relative performance of these indexes over the past several years. The data is presented as a three month moving average to minimize meaningless "wiggles" in the lines. The slope of a line over any time period represents the total return of that index relative to the common denominator. Thus by comparing slopes we will attempt to identify future outperformers.

The 14 plus year graph tracking SII, REIT, and the 3 internationals shows that if one's portfolio had been limited to just one of these indexes, SII would have yielded the best total return, as it ended up highest on the graph by its end in May, 2008. But if we had perfect foresight, and had switched between investments, still holding only one index at a time, we would have held SII through 1999, switched to REIT in early 2000, and to Emerging Markets in late 2001. If we had done this, by May 2008 the curve representing it would have ended up closer to 12 instead of roughly 3 for staying with SII the whole time. This 4 fold improvement in final result is an enticing goal and even though we won't come close to matching it, trying to get just part of it seems worthwhile.

To repeat, the slope of a line at any one point represents its relative performance. The end point of a line represents the cumulative effect of its slopes over the previous time period.

Lacking perfect foresight, we attempt to predict future winners by comparing their slopes in the present. As will be shown, a certain minimum positive slope is required before we even become interested.

So, over how long a period of time should slopes be compared? Two time periods were tested. Twelve months seemed long enough to eliminate short term reversals in trend, but six months gave better results as most short term trend reversals were still "Bridged Over" and true longer term trend reversals were recognized sooner.

This was particularly important immediately following the tech bubble burst in early 2000, when indexes with very strong up trends suddenly nose dived and the more defensive indexes had not yet established up trends. A way to minimize delays such as this is to stick with the 6 month period, but calculate the percent change in the 3 month moving average ratio from its previous 6 month high and from its previous 6 month low and add them together. This sum can represent less than a 6 month interval depending on how many months back the high or low occurred. In the case of the bubble the "Sum" emphasized the negative decline of the former hot indexes, and the positive rise of the more defensive indexes. The result was a quicker switch out of the losers and into the new winners than if a straight 6 month period had been used.

The "Sum" of % changes from previous 6 month highs and lows was used throughout this study. So as stated above, the time periods may be less than 6 months but never more.

So how high does a "Sum" have to get in the first place before we even become interested and how low can it go before we decide it has really lost favor with investors and should be sold?

The top portions of the accompanying graphs track monthly "Sums" for all 15 indexes (SII, REIT, Internationals, and 10 U.S. individual sectors). For the first 5 there are 13 years of data and for the 10 sectors only 5 years. The cyclical nature of the graphs is in part due to the way we calculate "Sums" which accentuates changes in slope.

A visual comparison of the upper and lower portions of these graphs shows that "Sums" have to get up to around 10 before there is any likelihood of sustained outperformance, and that they can dip to zero or a little lower before becoming sustained under performers.

The upper portions of the graphs for the REIT, Emerging Markets, and Pacific display wider swings up and down than do the graphs for Style Index Investing (SII), Europe and the 10 Industrial Sectors. This suggests that the threshold for selling these "wider swingers" should be at a "sum" somewhere below zero to minimize premature sales and loss of potential profit.

Indeed, an analysis of relative returns (relative to the 4 style average) of funds that have previously given a Buy signal with a "sum" of 10.0 or more shows that a sale threshold of -5.0 or lower works best for the wide swingers, while a threshold of 0.0 or lower is better for the "narrow swingers". Results are summarized in Table 1.

TABLE 1

One may conclude that on average for the narrow swingers it pays to sell as soon as the "sum" drops below 0.0, and for the wide swingers it is better to allow a little more leeway and not sell until the "sum" drops to -5.0 or lower.

Using these as guidelines applied to the above transactions and adding the dates of the Buys and Sells, we get the result shown in Table 2.

TABLE 2

From 1/95 through 12/08 investing in these funds or ETF's over the times indicated out performed the 4 Style Average by roughly 10 percent per year. Over the same 14 years, although leadership changed a few times, the S&P 500 and the 4 Style Average ended up a tie. Can we assume that use of this system will outperform by 10 percent per year in the future?

Figure 1 (next page) shows the periods of time over which these funds or ETF's were theoretically held. Much of the time from 1995 through 1999 no funds qualified for purchase, or if they did they yielded a negative relative return. The 10 Industrial sector funds were not available then (or until 2004), and if they had been we don't know if any would have been qualified for purchase. After 1999 the situation reversed in that most of the time at least two and sometimes up to 5 funds qualified for ownership simultaneously. During that period, of the eight funds or ETF's that qualified for purchase at one time or another, six enjoyed significantly positive relative returns, but the other two (Industrials and Pacific) suffered negatively, although not by large percentages.

FIGURE 1 (Click Here)

Under these conditions claiming a 10 percent per year advantage over the S&P 500 for the system seems a bit of a stretch. By making the worst possible choices the advantage would disappear altogether, while very good choices could yield an advantage of somewhat more than 10 percent per year.

Perhaps it is best to think of the system as one that identifies the U.S. Style, Industrial Sector, REIT, or International Fund or ETF that is likely to give superior return and that when two or more are identified simultaneously, investor judgment rather than "sums" should determine where funds are invested.

While this is a far from perfect solution it does go a long way toward providing a much broader menu of choices beyond Style Index Investing alone. In no way does it supersede SII but offers profit opportunities during periods when investors do not have a strong preference for one of the four basic style of stocks, as has been the case since 2002.

Critics of this system will point out that these results are the product of data mining. That is; setting the trading rules to fit the data available over the history period. This is true in that in January 1995 the historical data that preceded it may or may not have yielded exactly the same trading rules and thresholds. On the other hand, 14 years is a fairly long period which included the tech and housing bubbles, and the first 7 years of which saw the REIT and Emerging Markets indexes grossly under performing the 4 style average. Will the phenomenal performance of Emerging Markets and Energy during recent years be repeated by some other indexes in the future? These facts need to be considered by anyone contemplating use of this system.

As of this writing (March, 2009) Four Industrial Sector funds or ETF's are buys. They are Consumer Staples, Healthcare, Telecomm Services, and Utilities, all of which seem suited to resist the negative effects of a severe recession better than the other 6 sectors, REIT, or Internationals. Americans presumably will continue to east, get sick, use the telephone, and turn on the lights.

On the other hand judgment may say to avoid Health Care until the future extent of the Federal role is clarified. Or maybe people will trade down in the services they are willing to pay for to communicate with each other. Or while residential use of electrical power may not go down much, the curtailment of industrial production could hurt utility profits.

These are the types of judgments that need to be made after the mathematics of the system identify the possibilities. With reasonably good judgments combined with this system, an advantage of 10 percent per year over the S&P 500 seems attainable.

If one believes that U.S. central bankers and their counterparts in developed countries are smart and strong enough to hold the time from peak to recovery to closer to 3 years than 7 and 1/2 years, then since we are already 1 and 1/3 years into the current decline (Figure 2 red line), one could make a case for retaining some equities along with fixed income securities. This would position one to enjoy at least part of the rebound from the bottom, and if one could be 100% sure a bottom had been reached (which one never is) the portfolio should then be 100% in equities.

Thus, as of March 2009, the proportion of fixed income in the portfolio is directly proportional to one's estimate of how many more months it will take to reach a market bottom. If it is like the Great Depression we've got another 1 and 1/2 years to go. On the other hand, if it is more like the 1973 - 1976 decline and recovery is only 1/2 year to a bottom.

As indicated above, one can never be sure a bottom has been reached until several months later when an intermediate decline fails to establish a new low. So before one is confident to make a major portfolio switch from fixed income back to equities will probably take at least another 6 months after the true bottom has been reached.

Added thoughts as of January, 2010:

The uncertainties about the timing of a bottom as expressed in March, 2009 and the fact that the broad market has risen more than 50% since then, demonstrates the dangers of trying to predict its direction. The doom and gloom that pervaded the media at that time was a perfect inducement to switch out of stocks and into fixed income investments.

As later events have proved, that would have been the worst possible timing, as the first week of March, 2009 marked the bottom of the current cycle. In addition to reconfirming that trying to time the market is futile, events of the past couple of years shed light on the behavior of six of the ten MSCI sectors that offer an opportunity for profit when market direction is uncertain. (When is it ever otherwise?)

Basically it denotes Consumer Staples, Health Care and Telecommunications Services as Defensive sectors and Consumer Discretionary, Materials, and Industrials as Aggressive sectors. During broad market declines the Defensive sectors either rise or decline far less than the broad market, while during broad market rises the Aggressive sectors rise more that the broad market and far more than the Defensive sectors.

By Using the Relative Strength concept described in this essay it is possible to identify periods when the Aggressive or Defensive sectors will outperform. This is explained in detail in "Living With Risk" by the author.

Ray Schumacher

January 18, 2009