Style Drift (still) Happens

he focus of this article is style drift, or the propensity of some mutual funds to migrate from one classification to another within the Morningstar Style Box. For planners and other professionals who utilize mutual funds style drift can present a significant dilemma.

Very simply, the problem with style drift is this: portfolios are often intentionally created to include funds which represent distinctly different equity categories, such as large cap value, mid cap growth, small cap value, etc. If the chosen funds stray from their original style, the portfolio can tilt in unintended ways. Thus, style drift within equity portfolios could be compared to watching the center fielder and right fielder slowly work their way over to left field during the game. By the 6th inning, there are 3 left fielders. Clearly, left field is now over-represented. No coach would intentionally place 3 players in left field and leave center and right field empty. However, this is essentially what happens when funds drift away from their original positions in the portfolio.

This study utilized data over an eight year period from 1995 to 2002 and included 523 U.S. equity funds which had at least 10 years of performance history (as of August 31, 2003) and no more than 20% of their portfolio in bonds, non-U.S. stock, or cash.

The Morningstar Equity Style Box consists of 9 sectors as shown below. The numbering system or style “score” (Large Cap Value = 1, Mid Cap Blend = 5, Small Cap Growth = 9, etc.) was employed by the authors to facilitate quantitative analysis.

Large Cap Value  1	Large Cap Blend  2	Large Cap Growth   3
Mid Cap Value  4	Mid Cap Blend  5	Mid Cap Growth  6
Small Cap Value 7	Small Cap Blend  8	Small Cap Growth  9

The style “score” for each fund was manually retrieved from the "advanced analytics" database within Morningstar’s Principia software. Thus, each of the 523 funds had 8 style scores (one for each of the eight years from 1995 to 2002).

For example, as shown below in Figure 1, T. Rowe Price was categorized by Morningstar as a Large Cap Value fund for each of the eight years and was therefore assigned a score of “1” for each year. With no style deviation (i.e. drift) the standard deviation is, of course, zero. On the other end of the spectrum, Fidelity Select Home Finance was categorized as a large cap value fund from 1995-1998 and then a mid-cap growth from 1999-2002. As a result of such a dramatic shift in style the standard deviation was 2.67. The type of style shift shown by the Fidelity fund is somewhat unusual in that it not only shifted from large cap to mid cap, but also from value to growth. In this study, style drift was identified and measured by calculating the standard deviation of eight annual style scores. In addition, style drift was also measured by the “box” measure. The box measure is the nominal change from one year to the next.

Figure 1. Contrasting Style Scores

The mean style drift measure (i.e. average standard deviation) among this group of 523 equity funds was .35. The median drift score was zero – indicating that at least half of the funds had no style drift at all. In fact, 284 out of 523 funds (54%) had no style drift over the eight-year period. Among the 239 funds that had style drift the average (i.e. the “average” standard deviation) was .77. The highest possible standard deviation (or style drift score) was 4.28, which would have been achieved by a fund fluctuating annually between Large Cap Value(1) and Small Cap Growth (9) over the eight year period.

The distribution of equity styles among the universe of U.S. equity funds is, of course, not uniform. The distribution of the set of funds in this analysis were likewise distributed in a non-uniform fashion, as shown in Figure 2.

Figure 2. Distribution of Funds Being Analyzed in 1995 and 2002

One of the most interesting findings in this analysis was the fact that between 1995-1998 only one fund (among the 523) experienced style shift (Figure 3). Between the end of 1998 and the end of 1999, 164 funds shifted styles. Thus, nearly one-third of the funds had a reclassification within the Morningstar style box within a 12 month period. In calendar year 1999 the average actual style shift (by the “box” measure) was 1.8, or nearly 2 style “boxes”. The median shift was 1.0, or one style box.

Figure 3. What Happened in 1999?

In 1999, two funds had an absolute style shift score of 5, meaning that they shifted both vertically (market cap) and horizontally (growth/value factors). In one case, it was Fidelity Select Home Finance which shifted from a style score of “1” (large cap value) to a score of “6” (mid cap growth). The other fund was Hancock Technology A which shifted from 6 to 1. Five funds had absolute style shift of 4, meaning they shifted horizontally (large to mid, mid to small, mid to large, etc.) and partially vertically (e.g. from value to blend, or from blend to growth). One fund shifted from 1 to 5, one from 6 to 2, and so on.

In 1999, 45 funds shifted vertically only. The most common vertical-only shift was from 1 to 4 (large cap value to mid cap value). The second most common shift was either 6 to 3 (mid cap growth to large cap growth) or 3 to 6 (large cap growth to mid cap growth). This type of style shift appears to have been caused more by the bloated market cap of the late 1990’s than by portfolio altering decisions made by the fund manager. As such, this type of shift is less egregious.

Of the 20 funds in 1999 that experienced a style shift of 2 boxes (e.g. 6 to 8, 5 to 7, 2 to 4) 17 involved a movement on the vertical axis (market cap) and a partial movement along the horizontal axis from blend to growth (or vice versa) or blend to value (or vice versa). Only three funds made a complete horizontal shift (value to growth or growth to value). Changing from value to growth, or vice versa, appears to be a more intentional shift, whereas a change from large cap to mid cap may represent style “slippage” which, in some cases, may be largely unintentional.

The most common style shift in 1999 was a 1 box shift, involving 92 out of the 164 funds which shifted. Of the 92 funds, only 1 fund shifted dramatically (from 4 to 3, from mid cap value to large cap growth). The other 91 funds had a shift which did not involve a change in market cap, but rather from blend to value (or vice versa) or from blend to growth (or vice versa). Inasmuch as Morningstar’s “blend” category is somewhat of a “catch-all” category, one-box shifts involving the blend category may not represent a significant “shift” at all, but rather a change in the style “slant” or “tilt”.

The style tilt that started so dramatically in 1999 continued in 2000, 2001, and 2002. What was it about 1999 that may have led to such wide spread style shift or style tilt? One possible explanation is the fact that U.S. equity market weakness began in 1998, or about two years before it was generally recognized by those who only pay attention to the performance of notable equity indicies. Notice in Figure 4 that the returns of each style box (using only the funds in this analysis) were uniformly high during 1995, 1996 and 1997. In 1995, the average across all style boxes was nearly 30% with a standard deviation of return of only 4.7%. With all styles doing well there was little reason for a manager to style shift! In 1996, the average returns fell, but consistently across all style boxes – thus no serious motivation for a manager to look for greener pastures. In 1997, the world was happy again for U.S. equities with an average return of over 25% and very little deviation from that. Then in 1998, portions of the U.S. equity market begin to unravel.

In 1998 the standard deviation of return across the nine style boxes tripled in comparison to 1997. In 1998, value orientation lost its luster as tech mania (i.e. growth orientation) went wild. Returns of large cap value funds “plummeted” in 1998 and 1999. Even more dramatic was the fall of mid-cap value and small cap value funds in the same two years. Even small cap growth was not immune, posting an average gain of a paltry 4.3%. Clearly, 1998 belonged to growth stocks, and large cap growth in particular. The tidal wave of style drift which began in 1999 was the product of several factors.

One of the primary factors was escalating market cap boundaries. With growth stocks creating a bubble that was generating outrageous market caps there were many funds which “drifted” somewhat unintentionally – such as from large cap value to mid cap value. This type of drift would be more likely in a fund with low turnover. Ironically, by not buying “into” the upward drifting market such a fund actually drifts vertically downward because the portfolio market cap is left behind by escalating market cap boundaries. In short, as the buoyant equity market of the late 1990s steadily inflated market cap boundaries, some funds did not keep up. A classic example would be Scudder Dreman High Return Equity A. It was classified as a large cap value fund from 1995 through 1998. In 1999 it was classified as a “4” (mid cap value). Then, in 2002 it was back to large cap value. During this 8 year period, Scudder Dreman averaged less than 20% annual turnover. Thus, the market boundaries of what it took to be labeled “large cap” moved up, and this fund did not. This was a fairly common phenomenon. This type of style drift might reasonably be referred to as market-induced drift. While this explanation does not account for all the style drift that took place between 1999-2002 it was one of the primary variables. One finding is clear: style drift is clearly linked, whether exogenously (changes in broad market cap and/or valuation indicators) or endogenously (fund managers shifting/tilting the portfolio) to volatility of equity returns.

Another type of drift which appears to have taken place might more accurately be referred to as style “tilt”. In 1999, for example, the three style categories that had the largest percentage of drifter funds were 2, 5, and 8 – large cap blend, mid cap blend, and small cap blend. Blend is the common denominator. Now look at the 1999 return patterns in Figure 4 across the different style boxes. In each market cap category (large, mid, small) it was the value funds that we slipping and the growth funds that were having a banner year. Thus, funds in the blend category titled over to growth during 1999. Then in 2000, it was the mid cap blend and small cap blend funds that had the highest percentage of funds leaving that category during the year. In 2000 it was a value orientation that produced better returns which apparently motivated some managers to tilt toward value stocks. Whether or not this type of movement in the style box represents egregious style drift or tactical style tilt is subject to debate. Fundamentally, the argument for or against style drift (or tilt) depends upon whether or not a planner wants to build portfolio with diversity by design (wherein there are not 3 left fielders by the 6^th inning) or a portfolio in which funds are allowed to migrate (i.e. employing a variety of very active managers) in search of better returns.

Interestingly, in 2001 and 2002 the correlation between performance by style box and style drift became fuzzy. Our conclusion is that mayhem begets mayhem. As the market lost direction it became unclear where one should “drift” to in search of a safe haven. Small cap value turned out to be the star in 2001, but this particular category is by far the smallest in terms of number of players (Figure 2) precisely because it is a more difficult niche to participate in. For that reason, it is more difficult for funds to casually drift into small cap value.

Another interesting observation from Figure 4 are the relatively similar arithmetic mean returns among each market cap grouping (large caps, mid caps, small caps) over the eight year period. In addition, growth funds (style boxes 3, 6, 9) had consistently higher standard deviation of return than value funds (boxes 1, 4, 7). This is a reminder that growth investing will certainly produce a higher level of volatility of return but will not certainly produce a higher level of return.

Finally, in Figure 5, are the best of the best among non-drifting funds. The five funds per style box category with the best risk-adjusted performance over the 8 year period are included. The funds listed experienced no style drift between 1995-2002. This doesn’t mean they won’t in the future, but they have at least demonstrated a willingness to stay in their position in “center-field” during a period of pronounced market volatility. This is a notable characteristic, particularly during the bench-clearing melée of 2000-2002!

Figure 4. Average Annual Returns per Style Box

Figure 5. The “No-Drift” All-Stars (1995-2002)

____________________________________________________________________________________
Craig L. Israelsen is an Associate Professor in the Department of Consumer and Family Economics at the University of Missouri-Columbia (http://www.missouri.edu/index.cfm) where he teaches courses in Personal Finance and Family Living. He holds a Ph.D. in Family Resource Management from Brigham Young University. He received a B.S. in Agribusiness and a M.S. in Agricultural Economics from Utah State University. Primary among his research interests is the analysis of mutual funds. He writes monthly for Financial Planning magazine. Darcy A. Kruse is an undergraduate student at University of Missouri majoring in Physics.