Thinking Inside the Box

by Craig L. Israelsen
Reprinted from Financial Planning Magazine, January, 2006

ithin the Morningstar style box, is there variance in the market capitalization of equity funds within each individual box? Yes. Are there small large-cap funds and large large-cap funds? Yes. Does market-cap variance make any meaningful difference in performance of the funds within the box? Sometimes.

To investigate the issue of style box market-cap variance and its impact on performance, nearly 500 U.S. equity mutual funds were analyzed over a three-year period ending 9/30/05. The results are reported in "Inside the Boxes: 2002-2005". Raw data were obtained from Morningstar Principia. Moreover, Morningstar's 3-by-3 style box methodology was utilized in this study.

Let's dissect the 80 U.S. equity large-cap value funds in the upper leftmost box of "Inside the Boxes: 2002-2005". Each fund had a "High" rating in Morningstar's new Equity Style Consistency rating (increasing our confidence that the funds in the study had experienced minimal style drift); a beta of .50 or higher (with its appropriate "best-fit" U.S. equity index), a best-fit R-squared of 50 or higher; at least 75% of its portfolio in U.S. stocks; and at least a three-year return. Index funds were omitted from this study. Only "distinct" funds were included (meaning that only the primary share class of multiple share class funds was included). Using beta and R-squared as filters was for the purpose of including only those funds that had a positive correlation to the style box in which they were placed.

The 80 large-cap value funds which met all the filters of this study were then divided into quartiles according to the market-capitalization of each fund. In the case of large-cap value funds, there were 20 funds per quartile. The average market capitalization of the funds in the 1st quartile (or the 20 funds with the largest market-cap) was $54.8 billion. The average three-year return (as of 9/30/05) of those same 20 funds was 15.46%. The 20 large value funds in the 4th quartile had an average market-cap of $14.3 billion and an average return of 19.58%, which is higher than the 1st quartile mean return at the 99% confidence level.

Among the large-cap value funds in the 2002-2005 period, there was a market-cap difference of about $40 billion between the funds in the 1st quartile and those in the 4th quartile (see "Gap in the Box"). Associated with that market-cap differential was a statistically significant difference in average three-year return of over 400 basis points in favor of the large-cap value funds with smaller market capitalization.

The same phenomenon is observed among large-cap blend (top row middle box) and large-cap growth funds (top row, rightmost box). In both of those style boxes, the 4th quartile funds (large-cap funds with the smallest market-cap) had significantly higher average returns than funds in the 1st quartile (large-cap funds with the largest market-cap).

Among mid-caps, there was less variance in both the market-cap and three-year performance between the funds in the 1st and 4th quartiles. Among mid-cap value and mid-cap blend funds, there was no significant difference in return between the funds in the 1st and 4th quartiles. However, mid-cap growth funds with smaller market-cap did have a significantly higher return than mid-cap funds with larger market-cap.

Among small-cap funds the results were mixed. In the small-cap blend box, the average three-year returns for funds in the 4th quartile were statistically higher than among the funds in the 1st quartile. Average three-year return among small-cap value funds and small-cap growth funds showed no statistical difference when broken out by largest and smallest market-cap quartiles. The small number of funds per quartile in the mid-cap and small-cap categories is one cautionary note when considering the results.

The average three-year return of funds in the 4th quartile (smallest market-cap funds) was significantly higher than the average return of the funds in the 1st quartile in five out of the nine style boxes (boxes shaded in green). In three other cases the 4th quartile return was higher, but not significantly. In only one case was the return of the 1st quartile higher than the 4th quartile, and the difference was not statistically significant. The most distinct difference in performance between the 1st and 4th quartiles was among large-cap funds. In general, the performance of U.S. equity funds with smaller market-cap within each style box had higher returns during the three-year period ending 9/30/2005.

As shown in "Gap in the Box", the variance between 1st and 4th quartile average market-cap was sizable among large-cap funds, ranging from $40 billion among large value and large blend to $30 billion among large growth funds. The market-cap variance was considerably smaller among mid and small-cap funds. This makes sense inasmuch as the market-cap upper limit boundaries are defined (though flexible over time) for the small and mid-cap categories, whereas there is no upper limit on the market capitalization boundary for large-cap stock funds. This may well explain why large-cap funds showed distinct performance differentials between the 1st and 4th quartiles whereas among mid-cap funds and small-cap funds the performance differential between funds in the 1st and 4th quartiles was less consistent.

Understandably, over this particular three-year period (9/30/2002 to 9/30/2005), mid-cap and small-cap equities (across all three styles of value, blend, and growth) outperformed large-cap equities as shown by maroon colored bars in "Advantage Small." Over a different time period, such as the three-year period ending September 30, 2000 in which large-cap indexes (value, blend, and growth) outperformed their mid and small-cap counterparts (as noted by the blue bars in "Advantage Large"), it would seem reasonable to assume that funds in the 1st quartile of each style box (i.e., those with larger market-cap) would outperform funds in the 4th quartile (i.e., those with smaller market-cap).

The results for 1997-2000 time period are shown in "Inside the Box: 1997-2000". As a technical note, the number of funds in each quartile increased due to fact that the Equity Style Consistency rating was not available in Principia in September, 2000 (hence, with one less selection filter the number of funds increased). Funds selected for analysis as of 9/30/2000 were retrieved from the October, 2000 release of Morningstar Principia.

As shown in "Inside the Box: 1997-2000", we observe that in only three of the nine style boxes (denoted by yellow shading) was the average three-year return of funds in the 1st quartile (from 9/30/1997 to 9/30/2000) statistically higher than the return of the funds in the 4th quartile. Those three style boxes were large value, mid value, and small blend. Conversely, funds with the smallest market-cap in the large growth box (noted by green shading) significantly outperformed funds with the largest market-cap during a period in which stocks with larger market-cap generally outperformed.

In summary, during a three-year period in which U.S. equities with smaller market capitalization generated the highest returns (2002-2005), U.S. equity funds with the smallest market capitalization also outperformed in five of the nine Morningstar style boxes. The most noticeable performance differential based on market-cap variance was in the three large-cap boxes (large value, large blend, and large growth).

The inverse effect was not observed to the same extent during a period of time (1997-2000) in which equities with larger market capitalization tended to have higher returns. Equity funds with larger market capitalization (i.e., 1st quartile funds) generated higher three-year returns in only three of the nine style boxes. However, among mid-cap and small-cap funds, the 1st quartile three-year return was higher than the 4th quartile mean return in five of the six boxes, though statistically higher in only two cases.

"Intra-style-box" variance in both market capitalization and performance is most pronounced among the three large-cap style boxes (large value, large blend, and large growth). Generally speaking, variance in market-cap and three-year performance between funds in the 1st and 4th quartiles was less conspicuous among the mid-cap and small-cap style boxes.

Market-capitalization variance can lead to significant performance differences within style boxes-particularly among large-cap funds. So, when you're shopping for a large-cap fund, it's a good idea to determine if it's a large large-cap fund or small large-cap fund.

Inside the Boxes: 2002-2005

Data as of 9/30/2005	Value	Blend	Growth
Large	20 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       54.8       15.46             4       14.3***    19.58***	33 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       60.8       13.23          4       20.8***    17.51***	26 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       49.0       12.89           4       19.9***    16.08***
Mid	12 funds/quartile         Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       6.9        22.32              4       2.4***     23.60ns	5 funds/quartile   Mkt-cap    Average   Average                       Quartile   Mkt-cap   3 Yr Rtn           1        6.4       21.00              4        2.4***    22.42ns	8 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       6.2        19.19           4       2.0***     24.54***
Small	8 funds/quartile          Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.4        22.19             4       0.64***    21.53ns	4 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.3        21.54            4       0.77***    26.21*	8 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.4        21.01           4       0.82***    22.95ns

*** = 99% confidence level

**  =  95% confidence level

*   =  90% confidence level

ns  =  non-statistically different means between 1^st and 4^th quartiles.

Gap in the Box

Advantage Large

Advantage Small

Inside the Boxes: 1997-2000

Data as of 9/30/2000	Value	Blend	Growth
Large	65 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn             1       67.8      10.73        4       15.5       9.00*	65 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       110.9      15.42        4        34.8      16.55ns	63 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       111.7      22.20        4        17.2      32.71***
Mid	28 funds/quartile         Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       9.6       7.18        4       2.2       2.62***	7 funds/quartile   Mkt-cap    Average   Average                       Quartile   Mkt-cap   3 Yr Rtn           1        8.5       14.39        4        2.3       12.47ns	40 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1        8.9       27.69        4        2.2       24.10ns
Small	19 funds/quartile          Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.2        3.09        4       0.25       2.58ns	13 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.3        7.29        4       0.48       0.37***	39 funds/quartile   Mkt-cap    Average   Average                      Quartile   Mkt-cap   3 Yr Rtn           1       1.6       15.37        4       0.64      18.35ns

*** = 99% confidence level

**  =  95% confidence level

*   =  90% confidence level

ns  =  non-statistically different means between 1^st and 4^th quartiles.

____________________________________________________________________________________
Craig L. Israelsen, Ph.D. is an associate professor at Brigham Young University. He teaches family finance in the Department of Home and Family Living. His research interests include mutual fund analysis. He writes monthly for Financial Planning magazine. Learn more about Craig Israelsen at http://familyliving.familylife.byu.edu/faculty/israelsen.htm