To determine how similar mutual funds are to their benchmarks, an analysis was conducted. All investments categorized in one of the nine domestic equity categories based on a Dec. 31, 2008 Morningstar data feed were included in this study. Note, though, since holdings data is often delayed for mutual funds, holdings data is used based on its effective date from both Dec. 31, 2008 and March 31, 2009 data feeds. Index funds, enhanced index funds and ETFs were removed from the test sample.
Passive overlap was determined by comparing the holdings of the mutual fund (or ETF) to the comparable Russell iShares ETF in the same Morningstar category (e.g., the holdings for Large Blend fund would be compared to the holdings for the iShares Russell 1000 Index Fund). This methodology differs from the methodology employed by Cremers and Petajisto, who compute active share by comparing the mutual fund holdings to 19 different indexes (a group that includes Russell, S&P and Wilshire benchmarks), and determining the active share based on the index with the highest overlap. The author’s reliance on the Morningstar style categories to select a single index allows for a more “apples to apples” comparison.
The actual ETFs, as well as the number of holdings as of Dec. 31, 2008, are included in Figure 2.
Note that the initial overlap tests were also conducted against S&P, MSCI and Wilshire benchmarks, but the results (correlations) among the tests were similar (high) enough that overlap benchmark was reduced to the Russell indexes for simplicity purposes.
Information on the amount of passive overlap by investment category for the respective Russell benchmark is included in Figure 3. In Figure 3, a fund in the 95th percentile would have more passive overlap than all but 5 percent of the funds studied; a fund in the 75th percentile would have more passive overlap than all but 25 percent of the funds studied, etc.