Introduction

This study aims at investigating how the change of information dissemination process would affect the window-dressing behaviors of mutual fund managers. By convention, window-dressing is defined as the portfolio manipulations right before the quarter-end date, when all the fund managers are required to disclosure their holding firms of that date.

Over the past decades, technological progresses largely change the way how information disseminates, and these further influence the information flow of capital markets. For example, the implementation of “Electronic Data Gathering, Analysis, and Retrieval system (EDGAR)” by the U.S. Securities and Exchange Commission (SEC) in 1990s enables a greater and broader access to market information via the Internet. The emerging of social media in 2000s becomes a new channel through which investor sentiment transmitted, as documented in Chen, De, Hu, and Hwang (2014). Or the recent development of blockchain technology that has been adopted by Australia stock exchange in 2017 seems to promise a more transparent and timely information disclosure for the purpose of settlement, audit, and regulating reporting.

However, more transparent information disclosure does not always bring the benefits only, but also the costs. As mentioned in Prat (2005), “the wrong kind of transparency entails a trade-off between damage from information leaks and weaker incentives for the agent.” Therefore, it is of importance for regulator to evaluate the cost of more information disclosure, which mitigates the agency problem but at the expense of the damage to principales when those information is revealed to third party as well. One example regarding to mutual fund holding disclosure is from Frank, Poterba, Shackelford, and Shoven (2004). In their research, primitive funds can suffer after holding disclosure because the hypothetical copycat fund can therefore replicate their portfolio and earn almost the same returns after-expense.

Window-dressing behavior of mutual funds is another example driven by information disclosure. According to SEC regulation amendment in 1985, mutual fund managers are asked to disclose their holding firms two times a year. This disclosure policy results in an alleged agency problem: managers might manipulate their portfolio by buying winners and selling losers right before the disclosure date, in order to mislead their investors about their management skills. In addition to the evidence of window-dressing, literature also documents some possible incentives for managers to engage in window dressing. For example, risky bet incentive(Agarwal, Gay, Ling, 2014), gaming incentive (Carhart, Kaniel, Musto, and Reed, 2002), or copycat-avoiding incentive (Frank, Poterba, Shackelford, and Shoven, 2004).

Based on these literature, the main hypothesis of my proposal is that: the window-dressing effect will be stronger when the dissemination of portfolio disclosure information is more transparent due to better access; for example, via the Internet. The rationale behind this hypothesis is that, a more transparent disclosure can further reinforce the incentive for funds managers to manipulate portfolios; for example, the incentive to avoid copycat funds or to make a risky bet.

Inspired by Gao and Huang (2018), a novel dataset of the staggered implementation of EDGAR in 1990s can be used to examine this hypothesis. In 1993-1995, SEC arranged all the mutual funds into 6 groups, and it required the first group began filing electronically to EDGAR in April 1993 while the last group in November 1995, which means that investors can thereafter access to those information via the Internet. This staggered implementation of EDGAR can be considered as an exogenous shock to information dissemination technologies, which is a nice natural experiment for difference-in-difference (DID) analysis framework. Besides, while Gao and Huang (2018) looks at all listed companies on EDGAR, there is so far no published article nor working paper that investigates the staggered EDGAR implementation of all mutual funds.

To the best of my knowledge, there is a paucity of literature on the relation between information technological changes and window-dressing behaviors of mutual funds. Therefore, the goal of this study is to unravel this relation, and therefore it could help remind policy maker to take the potential cost into account when regulating new disclosure policy, and the potential investors of mutual fund to be aware of the reinforced window-dressing behaviors.