Do Mutual Funds (Window) Dress Up for All Occasions? - A Comparison of SEC and Non-SEC Portfolio Disclosures

Abstract

Given the extensive amount of data available, the mutual fund industry has been used as a laboratory to study performance, risk-taking, agency theory, and behavior of financial agents. Most recent mutual fund studies utilize portfolio holdings data, which are commonly obtained from the Thomson Mutual Fund Holdings database (Thomson). In this paper, we document a wide disparity between holdings data that are reported in Security and Exchange Commission (SEC) filings and Thomson (non-SEC). When examining the same sample of equity funds from 1996 to 2008, we find only forty-eight percent of the 77,582 portfolios available through Thomson and SEC filings appear in both data sources. The remainder of the filings is sent to either Thomson only (thirty-two percent) or the SEC only (twenty percent). All told, Thomson is missing 30% of these funds’ SEC disclosures, or 15,357 portfolios. After hand collecting the missing SEC portfolios, we examine a number of potential issues that may result from this informational disparity. We hypothesize that the most likely reason for the disparity in the timing of the filings is related to administrative convenience or continuity when providing holdings data to Thomson. Since fund managers may not know these non-SEC portfolios are being disclosed or view them as less important, we find more evidence SEC portfolios are window-dressing when compared to non-SEC portfolios. We also, for the first time, are able to estimate the cost of window dressing because of the heterogeneity in portfolio window dressing for the same fund. We estimate that the cost of window dressing to investors is low both in terms of direct costs due to excess trading and indirect costs due to decreased fund monitoring ability. Finally, we discuss the implications of the portfolio disclosure heterogeneity on academic research.