Window Dressing in Mutual Funds

The following post comes to us from Vikas Agarwal and Gerald Gay, both of the Department of Finance at Georgia State University, and Leng Ling of the College of Business at Georgia College & State University.

In our paper, Window Dressing in Mutual Funds, forthcoming in the Review of Financial Studies, we investigate an alleged agency problem in the mutual fund industry. This problem involves fund managers attempting to mislead investors about their true ability by trading in such a manner that they disclose at quarter ends disproportionately higher (lower) holdings in stocks that have recently done well (poorly). The portfolio churning associated with this practice of window dressing has potentially damaging effects on both fund value and performance.

Our paper attempts to understand the incentives of managers to engage in window dressing by analyzing investors’ reaction to such behavior. These incentives present an interesting enigma: if investors can be misled by window dressing and thus reward such funds with higher fund flows, then why do not all fund managers engage in such activity? In contrast, if investors are not deceived by window dressing and punish such funds with lower flows, then why would any manager engage in it? We develop and test a rationale wherein investors receive conflicting signals regarding managerial ability, notably when a fund’s disclosed holdings at quarter-end (the first signal) do not conform to the fund’s actual performance over the quarter (the second signal). A critical feature of our analysis is regulatory reporting requirements that allow portfolio holdings to be disclosed with a delay of up to 60 days following quarter ends. We contend that a fund’s subsequent performance during this delay period can affect investors’ interpretation of the two conflicting signals. For example, a manager with poor performance during a quarter may decide to window dress and rebalance the fund so as to disclose disproportionately higher and lower proportions of winner and loser stocks. If the manager subsequently performs well during the delay period, investors are less likely to attribute the signal conflict to window dressing and more likely to attribute it to improved security selection. As a result, managers may then benefit from incrementally higher flows. In contrast, if the performance during the delay period is bad, then investors are more likely to attribute the signal conflict to window dressing and thus cause the manager to incur the cost of incrementally lower flows. In essence, our explanation suggests that window-dressing managers are making a risky bet that will pay off if their performance during the delay period turns out to be good.

Since window dressing is unobservable, an important challenge that we address is to develop proxies that can help identify it. We construct two such measures: Rank Gap and Backward Holding Return Gap (BHRG). The first measure Rank Gap captures the discrepancy between a fund’s performance-based ranking and a ranking based on the proportions of winner and loser stocks disclosed by the fund at quarter ends. The intuition underlying this measure is that, on average, a poorly performing fund should have a greater percentage of its assets invested in loser stocks and a lesser percentage invested in winner stocks than that of a well-performing fund. The second measure Backward Holding Return Gap (BHRG) is the difference between (a) the return imputed from the reported quarter-end portfolio (assuming that the manager held this same portfolio at the beginning of the quarter), and (b) the fund’s actual quarterly return. The intuition is that a window-dressing manager upon observing winner and loser stocks over the quarter will tilt portfolio holdings toward winners and away from losers to give investors a false impression of stock selection ability. We also note that since momentum trading similarly involves buying winners and selling losers, we conduct a battery of tests to show that our measures capture window dressing rather than momentum trading.

Using these two measures, we conduct an empirical investigation that involves the analysis of disclosed portfolio holdings of 2,623 equity funds that span the period from September 1998 to December 2008. Our analysis provides several interesting findings. We first investigate the determinants of window-dressing (“WD”) behavior and find such activity is associated with managers who have experienced poor performance and lack trading acumen, which perhaps justifies why certain managers choose to take the risky bet associated with window dressing as they are likely to have the greatest career concerns. We also find that WD managers engage in excessive turnover of stocks associated not only with their WD-related trading, but also with their other trading. Furthermore, we document the magnitude of their significantly higher levels of trade costs and its effect on fund performance. Next, given that WD managers appear to be unskilled, follow non-information-based trading strategies, and incur high levels of trade costs, we conjecture that their future performance should also be poor on average. We find that both short-term and long-term future fund performance are negatively related to window dressing.

Finally, we analyze investors’ reactions to window dressing. Relative to non–window dressers, we find that window dressers receive higher (lower) incremental investor flows if fund performance over the reporting delay period is good (bad). We also find that window dressers exhibit greater dispersion in flows across the two states of good and bad delay-period performance. Together these findings support our explanation that window dressers are making a risky bet on their performance during the delay period where the investor flows are the payoffs of the bet.

The full paper is available for download here.

Both comments and trackbacks are currently closed.