We provide the first in-depth examination of exchange-traded funds (ETFs) within actively managed mutual fund (AMMF) portfolios to better understand why AMMFs make substantial investments in passive ETFs. We examine the association between holding ETF positions and AMMF performance, as well as indirect measures of performance, including market timing, flow management, and cash holdings. We find that over one-third of AMMFs take an ETF position between 2004 and 2015. Our results indicate that AMMFs allocating large portions of their portfolio to ETFs perform worse, by between 0.41% and 1.63% annually using various performance measures. These AMMFs also exhibit worse market timing and hold more cash. In contrast, AMMFs that hold ETFs in small amounts have similar characteristics to non-user AMMFs. Therefore, the act of holding an ETF does not signal inferior ability, however, taking large ETF positions does.
Actively managed mutual funds (hereafter AMMFs) are well known for attempting to pick undervalued securities in an effort to generate excess returns. While equity AMMFs typically focus on picking stocks, AMMFs frequently hold positions in passively managed exchange-traded funds (ETFs). Between 2004 and 2015, we find that 37.88% of AMMFs held an ETF in their portfolio. Despite ETF-user AMMFs accounting for over one trillion dollars under management and almost one out of every five dollars managed by AMMFs, current literature has yet to examine the impact that holding a passive investment, such as an ETF, has on performance.
With the intense competition for new money within the AMMF industry (Berk and Green, 2004), the literature observes countless strategies designed to increase AMMF performance or inflows. These strategies include incubation (Evans, 2010), changing AMMF names (Cooper et al., 2005), making concentrated industry bets (Kacperczyk et al., 2005), and switching between market timing and stock picking strategies (Kacperczyk et al., 2014), among oth-ers. Additional research has focused on AMMF ability managing specific assets or position types. Chen et al. (2013) examines AMMF short selling and finds an association between short positions and improved AMMF performance. Koski and Pontiff (1999) provide the first detailed look at how AMMFs utilize derivative positions. Frino et al. (2009) find improved flow management associated with index futures positions, and Cici and Palacios (2015) find options use is associated with income generation and hedging motives.
We examine the impact of a strategy where actively managed portfolios hold substantial ETF positions. The potential benefits of this type of strategy can range from purely increasing performance, to improved cash and flow management, to the ability to time markets by moving into or out of an ETF position. As such, we examine if AMMFs utilizing ETF positions have improved performance, thus justifying holding large proportions of an actively managed portfolio in passive investments.
An AMMF’s most visible and well documented characteristic is performance.2 Through the examination of risk-adjusted performance and excess returns, we find that holding an ETF is associated with significantly lower performance over the subsequent 12 month period. To further our analysis, we follow and divide our sample into low- and highETF-user groups at the median proportion of a portfolio invested in ETFs over 12 month periods. We find high-user AMMFs drive our results, generating the lowest performance and subsequently underperforming non-ETF-user AMMFs by between 0.41% and 1.63% per year.3 These results hold after controlling for fund, style, family, and objective characteristics.
Our findings suggest that large ETF positions provide a strong indicator of subsequent AMMF underperformance. This result, in conjunction with high-ETF-users allocating an average of 12.81% of their portfolios to ETFs, brings into question an AMMF’s decision to hold large ETF positions. Utilizing alternative methodologies for robustness such as a tercile ranking, a style-based ranking, a lifetime cross-sectional ranking, a 24 month ranking with monthly returns, and a matched sample approach, we find quantitatively similar results, and oftentimes stronger evidence and greater magnitude of the underperformance associated with large ETF positions among AMMF portfolios.
To better understand the source of this underperformance, we decompose performance measures into an ETF portion and a nonETF portion. We find that both ETF and non-ETF portions significantly contribute to the underperformance of high-ETF-user AMMFs. In contrast, we find that ETF positions do not make a meaningful or significant contribution to the performance of lowETF-user AMMFs. The negative returns of the ETF portfolio provide evidence that the high-ETF-user AMMF’s lack of skill extends beyond directly selecting securities and into picking indices tracked by ETFs. This indicates that high-user-AMMFs may be unskilled in multiple facets of portfolio management rather than just lacking the ability to select securities.
Next we examine the direct, contemporaneous impact that holding an ETF has on AMMF performance. We find that lowuser AMMFs experience no meaningful difference in performance during periods they hold an ETF relative to periods they do not hold an ETF. Among high-user AMMFs we observe a significant decrease in style excess, objective excess, and benchmark excess performance measures during months they hold an ETF, further suggesting that high-user AMMFs lack the ability to utilize ETF holdings in a manner that improves performance.
We determine the impact that various levels of ETF activeness have on AMMF performance. Using a measure of Active Share calculated for an AMMF’s ETF portfolio, we find that more active ETF positions have a marginal association with decreased performance. Those high-user-AMMFs that make more active ETF bets, thus using ETFs as part of an active investment strategy, generate lower performance. We then examine the impact of ETF type on AMMF performance and find that high-user AMMFs holding traditional index tracking ETFs, though not necessarily tracking the assigned benchmark of the AMMF, generate the majority of our observed underperformance. This is consistent with the result of our ETF Active Share analysis, indicating that AMMFs taking large, index ETF bets outside of their assigned benchmark are the AMMFs generating the lowest performance.
Although high-ETF-user AMMFs significantly underperform, ETFs can provide value through indirect sources of performance. To examine if AMMFs increase market exposure during up markets and decrease market exposure during down markets, we employ the market timing methodologies of Henriksson and Merton (1981) and Treynor and Mazuy (1966). We find that, in general, AMMFs have poor market timing ability. However, high-ETF-user AMMFs are the poorest market timing funds. Relative to non-ETFusers, low-ETF-user AMMFs exhibit similar market timing ability. These results support the association between large ETF positions and a lack of AMMF ability.