Free-riding on mutual fund positioning can be profitable. Funds with the most representative US SEC disclosures (i.e., disclosed holdings that are representative of the fund style) can be outperformed by 5 bps per month, and tracking the best performers offers access to a cheap momentum strategy that beats 90% of active mutual funds. The increased reporting frequency that has been mandated since 2004 has increased opportunities for copycat funds at the expense of mutual fund holders.
What’s Inside?
The authors assess how effective copycat funds are at mimicking active US mutual funds based on the latter’s mandatory quarterly SEC submissions. In 2004, quarterly filing replaced the semiannual regime, in part to minimize window-dressing and portfolio-pumping opportunities. The authors explore whether the SEC’s policy change was effective and truly in fundholders’ interests or whether it led to a reduction in returns to proprietary research. Additionally, by sorting funds into characteristics-based deciles, the authors identify mutual fund types that should make attractive targets for copycat funds.
How Is This Research Useful to Practitioners?
The copycat funds are found to underperform their targets by an insignificant 1 bp per month in the full sample, or 2 bps of outperformance (10% significance) after trading costs and fees, which makes copycat strategies viable. Relative performance is strongest around year-end, when the mutual funds themselves might engage in uneconomic window dressing and tax trading.
By artificially reducing the sampling frequency of holdings from quarterly to semiannually for pre-2004 quarterly reporting funds, the authors find 10 bps of annual gross performance deterioration, but it vanishes after trading costs. When quarterly disclosure was mandated in 2004, representativeness of disclosed holdings (i.e., disclosed holdings that are representative of the fund style) increased, relative net copycat performance increased by a significant 5 bps per month, and tracking error to the target funds dropped substantially from 22 bps to 7 bps per month. The post-2004 regime was considerably more conducive to free-riding.
The authors consider the most fruitful copycat targets. First, they sort the funds into disclosure representativeness deciles and find that replication of the most representative decile yields 5 bps per month above the target fund, representing an 8 bp improvement over replicating the least representative decile. They find that mimicking past winners underperforms targets by 11 bps per month versus 9 bps of outperformance when mimicking past losers.
Crucially, top-decile past performers subsequently outperform their bottom-decile peers by 56 bps per month, and thus, replicating the top decile still beats 90% of actively managed funds. The authors consider two other measures of manager skill: return gap, which measures actual performance against that of the static disclosure portfolio, and active share, which measures how actively the portfolio is managed. Although both measures are correlated with higher subsequent fund performance, the return gap measures the outperformance component not available to copycat funds, whereas active share increases turnover and thus reduces replicability.
How Did the Authors Conduct this Research?
The authors study the period of 1985–2008. Mutual fund returns and characteristics data are from CRSP, fund holdings and objectives from Thomson Financial, and stock returns and prices from the CRSP stock files. Only US mutual funds investing primarily in US equities are included; sector funds, index funds, and duplicate share classes are excluded, leaving 3,046 funds in the sample.
Copycat funds are constructed by investing in a mutual fund’s most recently disclosed holdings with a 60-day lag, the regulatory reporting cut-off. Nonequity positions’ performance is represented by the return of a representative index. Copycat fund trading costs are estimated using the method of Kacperczyk, Sialm, and Zheng (Review of Financial Studies 2008), and expenses are assumed to mirror those of Vanguard’s Total Stock Market Index fund at 20 bps.
The representativeness of mutual fund portfolio disclosures is measured as the tracking error against a characteristics-based benchmark, constructed using the most recently declared holdings. All of the stocks in the portfolio are triple-sorted into 125 size, value, and momentum quintiles. The benchmark for a given fund is then the value-weighted composite of these 125 style benchmarks.
Abstractor’s Viewpoint
This research has ramifications for regulatory design because the authors demonstrate how quarterly holdings disclosure has increased the representativeness of the disclosed portfolio and reduced window dressing and other camouflage trading. But the increased transparency for fund investors permits third parties to free-ride on a fund’s research by running copycat funds. The tracking error and performance of such strategies were reduced dramatically after 2004—as was their relative performance—bringing into question whether fund investors ultimately benefit from increased transparency. From a less altruistic viewpoint, the authors show that those wishing to free-ride on active US mutual funds should focus on good recent performers with a history of reporting representative holdings.