Printer Friendly

Informed Trading of Mutual Funds: Evidence from Fund-Underwriter Relationships.

The mutual funds employed by banks and fund families have two distinct but contrasting features that stand-alone mutual funds do not have. First, mutual funds benefit from being in a larger group because doing so not only provides better access to resources but also facilitates the formation of networks and business connections within the group. These connections enable the affiliated funds to obtain material, nonpublic information and capitalize on information to which they would not have access otherwise (e.g., Massa, 2003; Massa and Rehman, 2008; Bodnaruk, Massa, and Simonov, 2009; Ivashina and Sun, 2011).

On the other hand, affiliated mutual funds are by nature less independent. Thus, banks and fund families can pressure their affiliated funds to act in favor of their own businesses, thereby creating potential conflicts of interest between fund managers and fund investors (e.g., Hao and Yan, 2012; Golez and Marin, 2015). Such a conflict of interest arises because the affiliated funds' managers serve two principals--fund investors and fund owners (i.e., shareholders of the fund's asset management firm)--and their goals are not always aligned. (1) In this paper, I investigate how the investment activity of mutual funds that are employed by investment banks is related to the underwriting activity of the investment banks and whether this relationship helps the funds deliver superior performance due to the information advantage or whether it costs the funds due to the conflict of interest.

Investment banks that serve as the lead underwriters in public offerings can collect and analyze substantial amounts of private information about their clients by having direct access to management and most of the firm's books. Because the asset management divisions of the underwriters could obtain private information under their roof before taking a stake in an equity, they can outperform the other mutual funds or investors who do not possess such information. Previous studies provide evidence of the information advantage for affiliated funds. Massa and Rehman (2008) show that after banks provide loans to firms, lending bank-affiliated funds increase their positions in the borrowing firms more than in unaffiliated funds and that these positions provide positive abnormal returns, thus supporting the information advantage hypothesis. Bodnaruk et al. (2009) also find that takeover bidder-affiliated funds take positions in the target firms prior to the announcement of the deals.

On the other hand, for initial public offerings (IPOs), the deal completion is important for the underwriters representing their client firms, as it enables the underwriters to build their reputation and develop long-term business relationships (e.g., Dunbar, 2000; Krigman, Shaw, and Womack, 2001; Griffin, Harris, and Topaloglu, 2007; Ritter and Zhang, 2007). Therefore, underwriters may allocate undersubscribed shares to their affiliated funds at the expense of fund investors, as stated in the so-called dumping ground hypothesis. (2) Ritter and Zhang (2007) test the dumping ground hypothesis, which posits that lead underwriters allocate IPOs to their affiliated funds so that more deals can be completed when there is less demand for the IPOs, thus creating conflicts of interest between fund managers and fund investors. They find little evidence to support the hypothesis.

Hao and Yan (2012), however, provide evidence consistent with the conflict-of-interest hypothesis. They show that investment bank-affiliated funds underperform unaffiliated funds because they tend to hold worse-performing clients longer and disproportionately larger amounts than nonclients. Investment banks have an incentive to buy and hold their clients' stocks to help win future underwriting business, although previous studies provide strong evidence that underwriting clients underperform in the long run. Therefore, whether mutual funds take advantage of the private information on their underwriting clients or simply support the clients at the expense of fund investors is an open empirical question.

Using firms with class-action lawsuits, I test which of the two hypotheses--information advantage or conflict of interest--dominates. Typical class-action lawsuits document alleged economic losses suffered by investors during the class period in which the sued firm committed the fraud. The end of the class period is the date on which the firm's wrongdoing is disclosed. In my sample, stock prices plummet by an average of 6.73% on the disclosure day and approximately 18% a day later. Therefore, consistent with the information advantage hypothesis, underwriter-affiliated funds have a huge economic incentive to sell sued firms before the end of the class period if they are already aware of the ongoing issues with the underwriting clients who violate securities laws and may face class-action lawsuits. In contrast, the conflict-of-interest hypothesis suggests that the lead underwriters pressure their affiliated funds not to close out the positions on clients to maintain a good relationship with the firms and guarantee future business. Furthermore, this is likely to occur when firms recover from their loss quickly or even exceed their pre-lawsuit-period performance once the lawsuit is proven unworthy.

My findings support the information advantage hypothesis. Underwriter-affiliated funds significantly decrease their positions in sued firms underwritten by their banks before the firms' misbehavior is disclosed to the public. The underwriter-affiliated funds hold on to the sued firms, which is worth approximately $4.27 million, on average, at the beginning of the quarter before the end of the class period, and they sell more than 17% of the initial positions during the quarter, indicating approximately 19 times more reduction in a position relative to nonaffiliated funds. (3) The findings remain strong after controlling for firm characteristics--size, book-to-market, firm performance, analyst coverage, and analyst consensus--and for mutual fund characteristics--investment objectives, total net assets, expenses, turnover, lag return, and age.

It is possible that underwriter-affiliated funds not only reduce their stakes in the sued firms but also do the same for all the firms located in the same industry, which would generate a spurious correlation between the underwriting relationship and the investment activity of the underwriter-affiliated funds. To rule out this possibility, I compute the changes in holdings for control firms that are similar in size and book-to-market ratio and belong to the same industry and examine whether the changes in holdings of sued firms and their control firms differ significantly among mutual funds. Consistent with the information hypothesis, the net holding changes are significantly greater for underwriter-affiliated funds than they are for nonaffiliated funds. Another possibility is that the affiliated funds reduce their holdings in IPO firms and seasoned equity offering (SEO) firms after the lockup period regardless of private information. I match the sample of sued firms with nonsued public offering firms underwritten by the same lead underwriter within the past three years. (4) The results show that the underwriter-affiliated funds reduce their holdings significantly more in sued firms than in nonsued firms.

The informed trading of underwriter-affiliated funds implies that the fund-underwriter relationship is a conduit for underwriting client information. Therefore, I further investigate whether information about a firm's fraudulent activity channels to the underwriter-affiliated funds through their investment banks (i.e., underwriting activity). If a lead underwriter knew the misconduct of its clients and acted in the interest of its shareholders, the underwriter would protect itself from a huge loss due to fraud and would thus sell shares of the clients in advance. Using institutional holdings data from Thomson Reuters 13f filings, I examine the stockholding changes of underwriters in sample firms directly. (5) The results show that lead underwriters decrease their holdings of sued client firms significantly, relative to other 13f-filing institutional investors. 13f filings are consolidated stock holdings for each firm at the institutional level. To avoid a situation where underwriters' holding changes are driven by holding changes of their affiliated funds, I exclude all the equity positions of the affiliated funds and run the regressions again. (6) The results still hold.

In addition, for lawsuits regrading IPOs or SEOs, plaintiffs assert claims against underwriters as codefendants, such as violations of Section 11 and 12(a)(2) of the Securities Act of 1933 arising from alleged misstatements in the equity offering documents filed with the US Securities and Exchange Commission (SEC) for a firm's IPO or SEO. Because of the direct involvement of underwriters in the lawsuits, they could foresee the timing of an exit from sued firms more accurately. For underwriter-involved lawsuits, I examine whether the underwriting relationships affect the holding changes in sued firms by affiliated funds. As expected underwriter-affiliated funds reduce positions in those firms by a greater amount, indicating strong support for the information hypothesis.

Finally, I test the effect of informed trading by mutual funds on the lawsuit outcomes. The underwriter-affiliated funds with private information could predict the outcomes of a firm's potential litigation. Cheng et al. (2010) show that securities class-action suits with institutional investors as lead plaintiffs are less likely to be dismissed and tend to have larger monetary settlements than do class actions with individual lead plaintiffs. Their findings imply that institutional investors are more likely to serve as lead plaintiffs if they can produce outcomes such as large settlements. Similar to the study, I propose that affiliated funds will opt out of sued firms in advance if they anticipate a certain outcome--for example, a lower probability of dismissal or a large settlement amount. My results support the hypothesis of affiliated mutual funds' informed trading where sued firms sold by the underwriter-affiliated funds are less likely to be dismissed and tend to have larger settlements.

This paper adds to the growing body of literature on affiliation and trading on private information from business connections. First, affiliation is an important channel for information sharing between institutions (e.g., Massa, 2003; Massa and Rehman, 2008; Bodnaruk et al., 2009; Ivashina and Sun, 2011; Griffin, Shu, and Topaloglu, 2012; Kedia and Zhou, 2014). I contribute to this stream of research by showing that affiliation becomes even more important in the event of a securities class action, especially when the institutions that are connected to the litigated firm through their business relationships suffer significant economic losses from the client's fraudulent activity.

Second, the conflict of interest within large Wall Street investment banks is a controversial issue. Because the investment banking business is the centerpiece of banks' operations and revenue, they are willing to sacrifice the other divisions to win the underwriting mandate (e.g., Ber, Yafeh, and Yosha, 2001; Hong and Kubik, 2003; Hao and Yan, 2012; Berzins, Liu, and Trzcinka, 2013). Therefore, affiliated mutual funds typically offer price support for their underwriting clients (e.g., Schultz and Zaman, 1994; Ellis, Michaely, and O'Hara, 2000). However, my results show that the interests of fund investors are not severely compromised. Finally, I show the effect of institutional trading on litigation outcomes. Specifically, underwriters with nonpublic and material information on underwriting clients anticipate the potential outcomes of the firms' misbehavior and release the information by reducing their holdings in these firms before the end of the class period.

The remainder of this paper is organized as follows. In Section I, I discuss the data and sample. In Section II, I provide evidence of the effect of fund-underwriter relationships on stock holdings of affiliated funds. Section III explores underwriting relationships as an information channel for affiliated funds. In Section IV, I further discuss and explore the results with various robustness specifications. Section V concludes the paper.

I. Data, Sample, and Descriptive Statistics

A. Data and Sample

I examine the stock holdings of mutual funds for firms that are sued in securities class actions from 2005 to 2016. (7) The data are obtained from several resources. The Center for Research in Security Price (CRSP) Mutual Fund Database provides mutual fund characteristics, such as total net assets (TNAs), fees, investment objectives, and fund returns. The Thomson Reuters mutual fund holdings database (S12) provides quarterly and semiannual stock holdings for US mutual fund holdings and CDA/Spectrum 13f filings for overall institutional holdings. (8)

My sample begins in 2005 because in May 2004, the Investment Company Act of 1940 mandated that individual mutual funds make quarterly disclosures of their portfolio holdings in Forms N-CSR and N-Q with a delay of no longer than 60 days. (9) I also limit the analysis to funds that generally hold diversified portfolios of US equities. I include only mutual funds that have a self-declared investment objective of "aggressive growth," "growth," "growth and income," "income," or "balanced" at the beginning of that quarter. I exclude all other funds, including all other international funds, bond funds, gold funds, real estate funds, and other sector funds, because these types of funds generally hold and trade minimal quantities of domestic equities, if any. To link between mutual fund holdings data and mutual fund characteristics data, I calculate the weighted average of the individual fund characteristics based on the total net assets of a fund with multiple classes and merge them through MFLINK from WRDS. (10)

SDC Platinum New Issues database provides IPO/SEO-related data. I exclude American depositary receipts, closed-end funds, units, real estate investment trusts, limited partnerships, financial firms, and utilities. For each public offering, I identify the lead underwriters, offer date, and firm identification, thereby matching the lead underwriter with their affiliated mutual funds. For firm characteristics, CRSP provides individual stock prices, returns, trading volumes, and shares outstanding. I obtain accounting figures from the financial statements on COMPUSTAT. Finally, Institutional Brokers' Estimate System (I/B/E/S) offers brokerage house information, including the announcement dates of recommendations, reporting analysts, number of analysts following, brokerage houses employing the analysts, the level of the consensus recommendation, and each analyst's recommendations.

B. Securities Class-Action Lawsuits

A securities class action is a case brought pursuant to Federal Rule of Civil Procedure 23 on behalf of a group of individuals who purchased the securities of a company during a specified period called the class period. The complaint contains allegations that the company and/or people inside the company violate(s) one or more of the federal or state securities laws. The Securities Class Action Clearinghouse (SCAC) offers detailed information relating to the prosecution, defense, and settlement of federal class-action securities fraud litigation." Among the firms with the lawsuits, I consider only firms that are publicly traded on the New York Stock Exchange (NYSE), American Stock Exchange (AMEX), and NASDAQ from 2005 to 2016.

Each securities class-action lawsuit provides important dates, including the class period starting (CPS) date, the class period ending (CPE) date, and the filing date (FD). CPS is when a firm's misconduct starts, and CPE is when the wrongdoing is revealed to the public. CPS and CPE are important in that anyone who purchased the defendant's stock between these dates is eligible to participate in the legal process. A negative stock market reaction to alleged firms on CPE typically triggers investors' legal resolution. FD is the date on which investors file a lawsuit against the firm.

C. Descriptive Statistics

To determine whether mutual funds change their stock holdings of sued firms prior to the disclosure of alleged fraudulent activity, I first identify the class period of each firm. Doing so provides an important event window during which informed investors (e.g., underwriters and underwriter-affiliated funds) possibly reduce their loss by trading based on information about a firm's misconduct. Figure 1 illustrates the time line of events associated with lawsuits and changes in stockholdings. If mutual funds informed on sued firms are willing to capitalize part of potential losses due to litigation, it is likely that they will reduce stock holdings of the firms prior to the CPE on which the public discovers the sued firm's wrongdoing, thereby punishing its stock price. Specifically, I investigate the informed trading of mutual funds holding sued firms from Q-2 to Q-\, where quarter Q is the quarter of the CPE.

Figure 2 shows the stock performance around the CPE. The stock performance of the sample firms is measured as cumulative abnormal returns (CARs) using the Fama-French (1993) threefactor model.

[[R.sub.i,t] = [[alpha].sub.i] + [[beta].sub.1] [RMRF.sub.t] + [[beta].sub.2] [SMB.sub.t] + [[beta].sub.3] [HML.sub.t] + [[epsilon].sub.i,t]] (1)

where [R.sub.i,t] is the daily return of a sued firm ;' in excess of the one-month Treasury (T-)bill return, RMRF is the excess return on a value-weighted market portfolio, and SMB and HML are returns on zero-investment factor-mimicking portfolios for size and book-to-market, respectively. Day 0 is the day of the CPE. CARs decline from 250 trading days prior to the CPE, and there is a significant average drop of 25% to 30% in abnormal returns around the CPE. Therefore, mutual funds could avoid huge economic losses if they reduce their stock holdings of sued firms before the CPE.

Table I provides the summary statistics of my sample. From 2005 to 2016, there are 6,904 public equity offerings, including both IPOs and SEOs. Among them, only 387 publicly traded firms are accused of committing fraud within three years of an equity offering. I then filter the sample using additional criteria. First, I exclude utility and financial firms to avoid any distortions caused by regulated industries. Second, I exclude firms with missing variables on CRSP, COMPUSTAT, I/B/E/S, and Thomson Reuters Mutual Fund Holdings. Third, I delete stocks with a public offering price of less than $5. The final sample consists of 198 unique firms and, thus, lawsuits.

Panel A reports the total number of sued equity offerings and our sample--the number of lawsuits and the type of equity issuance--each year from 2005 to 2016. Each sample lawsuit (firm) is classified as settled, dismissed, or ongoing. (12) Panel B provides the number of days related to a relevant event, such as a CPE date, a filing date, or an equity offering date. The class period, from CPS to CPE, is, on average, 364 days--that is, approximately one year. The minimum class period is one day. (13) It takes approximately 82 days, on average, for investors (plaintiffs) to file a lawsuit after the CPE. There are eight firms sued on the same date when their fraudulent activities are disclosed; a firm's lawsuit filing date is the same as its CPE date. The latest underwriting activities are slightly more than a year (466 days) before CPE. The minimum duration from an equity offering date to a CPE date is zero. (14)

Panel C presents the descriptive statistics for sued firms from 2005 to 2016. Size is calculated as price times the number of shares outstanding and is reported in millions. The average firm size is about $3 billion (median $1.1 billion). Book-to-market uses a firm's common stock amount relative to its size. Trading Volume measures the three-month moving average of the monthly trading volume, reported in millions. Analyst Coverage is the average number of analysts following a sample firm. Analyst Consensus is the mean value of analyst recommendations. These variables are measured at the beginning of the quarter prior to the CPE (i.e., at the beginning of Q-1). Firm Performance is calculated as abnormal returns over the past six months of a quarter prior to the CPE, using the Carhart (1997) four-factor model. (15)

Panel D compares mutual funds according to their affiliations. IB-affiliated funds are associated with investment banks, while UW-affiliated funds are related to investment banks that served as the lead underwriters of sample firms within the three years prior to the end of the class period. Nonaffiliated funds are independent mutual funds. I include only funds with TNAs of more than $1 million, are older than one year in age, and portfolios in which stocks account for more than 85%. (16) The TNA of IB-affiliated funds are $939 million, while that of nonaffiliated funds is $623 million. Along with Net Asset Value (calculated as a fund's underlying assets minus its liabilities, divided by the number of shares outstanding), IB-affiliated funds are, on average, larger than nonaffiliated funds. They could take advantage of being part of a bigger institution, which would allow them to attract investors more easily (e.g., cross-selling, advertisement, investment resources).

Turnover is aggregated sales or purchases of securities, divided by the average 12-month TNAs of the fund. Expense Ratio is the proportion of a fund's total operating expenses that shareholders pay for the fund's operating expenses. Lag Quarterly Return is the past-quarter raw returns. The turnover ratio of UW-affiliated funds is significantly less than other funds. The age of the funds, on average, is 24.19 years. Overall, UW-affiliated funds tend to be smaller and younger and to trade less actively. (17)

II. The Effect of Fund-Underwriter Relationship on Stockholdings

Investment banks collect and process firm-specific information during the issuance of public securities (Leland and Pyle, 1977; Campbell and Kracaw, 1980; Beatty and Ritter, 1986; Chemmanur and Fulghieri, 1994). The costly activities could be exploited in various ways, such as underwriting, lending, providing analyst reports, and trading securities if the underwriting arm of an investment bank is willing to share private information about its clients within the other divisions. Previous studies have found that banks could tip specific information to their affiliated institutions, such as mutual funds, to the extent that they do not attempt to violate any federal regulations (Massa, 2003, Massa and Rehman, 2008; Ivashina and Sue, 2011). I further explore the idea of information sharing between funds and underwriters in the context of equity offering and class-action lawsuits.

Although I focus on equity trading, the information transfer for trading purposes not only occurs in equity markets but could also occur in other financial markets such as bonds or derivatives markets. Therefore, my findings represent the lower bound of the informed trading by affiliated funds.

A. Stockholdings of Mutual Funds in Sued Firms

My primary interest is to examine whether fund-underwriter relationships help the affiliated funds deliver superior performance due to the information advantage or whether they cost the funds due to the conflict of interest. The information advantage hypothesis posits that underwriter-affiliated funds lower their positions in client stocks prior to the CPE, relative to nonaffiliated funds. On the other hand, the conflict-of-interest hypothesis posits that underwriter-affiliated funds do not decrease or even increase their holdings in the client firms more than nonaffiliated funds because the affiliated funds are pressured by their parent banks to support the clients for future underwriting business. To test these competing hypotheses, I investigate the changes in the holdings of mutual funds whose client firms face class-action lawsuits.

Stockholding changes are calculated in two ways: the change in the percentage of a firm's outstanding shares held by a mutual fund (HC1) and the change in the percentage of a firm's outstanding shares held by a mutual fund in relation to all the shares that are held by all mutual funds during the same period (HC2). (18) I calculate the trading measures only for cases in which a mutual fund owns the stock at any time over the past four quarters leading up to the quarter of the CPE. I do not require a fund to hold the stock either at the beginning or at the end of the quarter of the CPE. (19) The construction of the holdings and trading measures is meant to avoid concerns over the measures that may be affected by price changes. Because a firm's illegal practices could trigger a huge price drop or price inflation, I consider only the quantity effect. (20)

Table II reports univariate test results of stockholdings of affiliated and nonaffiliated mutual funds in sample firms over the previous four quarters leading up to the quarter of the CPE (quarter Q). I compare three different types of mutual funds in terms of affiliations such as underwriter-affiliated funds (UW-affiliated), investment bank-affiliated funds (IB-affiliated), and nonaffiliated funds (Nonaffiliated). The UW-affiliated funds are a subset of the IB-affiliated funds and are considered to possess superior information on the underwriting stocks. In Panel A, for mutual funds' quarterly positions in sued firms, H1 and H2 are calculated as the percentage of a firm's outstanding shares held by a mutual fund and the percentage of a firm's outstanding shares held by a mutual fund out of all the shares of the firm held by all mutual funds during the same period respectively. All three types of mutual funds hold the sued firms in an economically significant amount. UW-affiliated funds tend to hold a smaller amount than do the other two types of mutual funds. It is possible that the lower holdings are due to the smaller TNAs of the UW-affiliated funds.

Panel B documents quarterly holding changes of mutual funds on sued firms, HC1 and HC2, over the past four quarters leading up to the quarter of the CPE. Most mutual funds, on average, either increase or decrease stockholdings in sued firms leading up to quarter Q-2. More importantly, during the quarter prior to the CPE, from Q-2 to Q-1, only UW-affiliated funds significantly decrease their positions in sued firms. IB-affiliated funds also lower their positions slightly, but the reductions are not statistically significant. UW-affiliated funds lower their percentage holdings by about 0.04%, approximately five times greater than the average magnitude of the changes in the holdings of IB-affiliated and nonaffiliated funds combined.

Looking at the holding changes after the revelation of a firm's misconduct, interestingly all the mutual funds, including UW-affiliated funds, tend to lower their stockholdings in sued firms. This implies that institutional investors become aware of the alleged frauds of the sued firms on the disclosure date (i.e., CPE). They would start selling their shares in the firms, causing a huge drop in price. For a further reduction in the stockholdings of the UW-affiliated funds, it is possible that the affiliated funds recognize the severity of an event and foresee the future of the firm, such as potential litigation, and may thus further reduce their holdings.

B. Informed Trading by Underwriter-Affiliated Funds before the End of the Class Period

I now investigate the effect of fund-underwriter relationships on stockholdings in the multivariate setting. Because funds belonging to the same institution or family could make coordinated trades in a particular stock, standard errors are possibly correlated within the institution in a given time. Therefore, all specifications are adjusted for heteroskedasticity and clustered at time and at the fund's family or institution level (i.e., affiliation). (21) I estimate the following model:

[[DELTA][HC.sub.f,i,c] = [alpha] + [beta] * [UWIB.sub.f,i,c] + [gamma][IB.sub.f,i,c] + [delta] * [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c]], (2)

where [DELTA][HC.sub.f,i,c] is the percentage change in the holdings of fund f in the stock of sued firm i before the end of the class period c. The variables of interest are UWIB and IB. First, IB is a dummy variable that takes a value of one when the fund is related to an investment bank, and zero otherwise. UWIB equals one if the fund is related to an investment bank that served as the firm's lead underwriter for a stock offering within the three years prior to the end of the class period and it is zero otherwise. While the coefficient y on IB shows only the effect of mutual funds related to an investment bank, the coefficient 6 on UWIB measures the incremental effect of mutual funds affiliated with an underwriting investment bank over the IB effect (i.e., superior information).

Controls is a vector of control variables, including firm characteristics such as firm performance, trading volume, size, book-to-market, analyst coverage, and analyst consensus recommendation, as well as fund characteristics including TNAs, expenses, turnover, lag return, and age. The summary statistics of the control variables are described in Section I, and the detailed definitions are provided in the Appendix. All specifications also include year and mutual fund investment objective fixed effects.

The coefficient on UWIB is predicted to be negative and significant ([beta] < 0) according to the information advantage hypothesis. On the other hand the conflict-of-interest hypothesis expects the same coefficient not to be negative or even positive, which suggests that the lead underwriter of sued firms pressures its affiliated funds into price support. For IB, I control the effect that stems from it being part of an investment bank that offers ample resources, better research facilities, and potential connections to their portfolio companies. In Table III, the first two columns report the estimation results. (22) Consistent with the information advantage hypothesis, the coefficients on UWIB are -0.032 and -0.184 and are highly significant. The numbers indicate that after controlling for firm and fund characteristics, UW-affiliated funds decrease their positions in the stocks of sued firms 0.032% and 0.184%, on average, more than the control funds do. (23) The coefficients on IB are statistically insignificant, implying that simple IB affiliation does not lead to an information advantage.

Among controls, Initial Position is negative and significant, which suggests that funds with larger positions in sued firms tend to lower their positions. Additionally, Expenses are highly significant across the specifications, which suggest that mutual funds with higher management expenses sell more shares. It is possible that mutual funds with more resources require higher operating expenses and have better access to firms (e.g., Wermers, 2000). Turnover is another highly significant variable. The higher the turnover is, the lower the stockholdings of sued firms are. A potential explanation is that funds focused on the short term may be better informed. (24)

The results so far support the information advantage hypothesis; however, it is possible that UW-affiliated funds not only reduce their positions in the stocks of sued firms but also do so for all the stocks in the same industry. This would generate spurious relations between the affiliation and trading activities of mutual funds. To avoid this possibility, I examine whether UW-affiliated funds decrease their holdings of sued firms in excess of the decrease in holdings of nonsued control firms. I construct a group of the control firms held by the same funds in the same industry and with similar characteristics (size, book-to-market, and firm performance) as the sued firms. Then I investigate the extent to which a decrease in stockholdings of UW-affiliated funds on sued firms exceeds that of the control firms of the same funds. The model specification is as follows:

[[DELTA][HC.sub.f,i,c] - [DELTA][HC.sub.f,j,c] = [alpha] + [beta] * [UWIB.sub.f,i,c] + [gamma][IB.sub.f,i,c] + [delta] * [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c]], (3)

where [DELTA][HC.sub.f,i,c] ([DELTA][HC.sub.f,i,c]) is the percentage change in the holdings of fund / in the stock of sample firm i (control firm j) before the end of the class period c. The control firms are nonsued firms that are held by fund f during the same period and belong to the same industry (i.e., the same two-digit standard industrial classification [SIC] code) as the sued sample firm. Each sued firm has more than one control firm held by the same fund around the CPE. I compute the equal-weighted average holdings of mutual funds in the control firms before the CPE and subtract them from the percentage change in holdings of sued firms by the same funds. (25) All other control variables remain the same as in Equation (2). As reported in Columns (3) and (4) of Table III, the coefficients on UWIB are negative and statistically significant and confirm our earlier findings, thus supporting the information advantage hypothesis.

C. Matching-Firm Analysis for Underwriting Clients

Next, I turn to matching-firm analysis for underwriting clients. Mutual funds that hold IPO or SEO firms may sell their shares of the firms as their lockup periods expire, possibly producing spurious relations between selling activities of mutual funds and their affiliations. To address this concern, I investigate whether UW-affiliated funds decrease their positions in the stocks of the IPO or SEO firms, regardless of whether they possess private information on the accusations.

In Table IV, Panel A presents comparative statistics for the sued firms and matched nonsued equity offering firms (i.e., nonsued firms that were taken public by the same lead underwriters as the sued firms within the past three years). Of the 198 sample lawsuits (i.e., sample firms), 78 firms have matched control firms. The control firms tend to have a larger book-to-market, lower trading volume, and fewer analysts following.

To compare trading activities by mutual funds with sued firms versus those with nonsued control firms, I run the following estimation model:

[[DELTA][HC.sub.f,i,c] = [alpha] + [beta] * [SUE.sub.i] + [delta] * [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c]], (4)

where[[DELTA][HC.sub.f,i,c] is the percentage change in the holdings of UW-affiliated fund/in the stock of sued firm (or control firm) i before the end of the class period c. The control firms are nonsued firms that are held by the same UW-affiliated fund/during the same period and were underwritten by the same investment bank as the sued firm within the past three years. The variable of interest is SUE, which equals one if a firm has been subject to class-action lawsuits and is being held by UW-affiliated funds and is zero otherwise. All other control variables remain the same as in Equations (2) and (3).

The coefficients on SUE are -0.022 and -0.186, statistically significant at the 1% level. It indicates that UW-affiliated funds reduce their positions significantly more in sued underwriting client firms than nonsued underwriting client firms. The results support the information advantage hypothesis, suggesting that the previous findings of UW-affiliated funds selling sued firms are due not to the expiration of the IPO lockup period but to the nonpublic, material information exploited from underwriting relationships.

D. Effect of Informed Trading by Mutual Funds on Litigation Outcomes

For class-action lawsuits, not every litigation goes to trial or is settled with a large amount. Allegations that have no grounds or merit are likely to be dismissed, which will resolve the uncertainty around possible legal and financial liabilities of the sued firms. Because falsely accused firms can quickly recover from the damages caused by wrongful accusations, there is no incentive for UW-affiliated funds to sell shares of these firms. If mutual funds are truly informed, they could predict the outcomes of lawsuits. Accordingly, I posit that UW-affiliated funds are less likely to sell shares of the firms that are accused of wrongdoing on insufficient grounds.

The following probit regression is to determine whether shorting sued firms by UW-affiliated funds increases the probability of the cases not being dismissed:

[Case_[Dismiss.sub.ic] = [alpha] + [beta] * UWlB_[Short.sub.f,i,c] + [gamma] * IB_[Short.sub.f,i,c] + [delta] * [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c]], (5)

where Case_Dismiss equals one if the case is dismissed, and zero otherwise. The variable of interest is UWIB_Short, which equals one if an UW-affiliated fund lowers its position on a firm and is zero otherwise. IB Short takes a value of one if an IB-affiliated fund reduces its position on a firm and is zero otherwise.

In addition to the control variables in the previous equations, I add a set of variables to control for lawsuit characteristics based on the literature, such as the types of accusations and institutional plaintiff. Cheng et al. (2010) find that lawsuits with institutional investors as lead plaintiffs are less likely to be dismissed and tend to have larger monetary settlements than lawsuits with individual lead plaintiffs. Institutional investors only engage in such cases when the likelihood of winning is high, the potential damage (and thus the expected settlement) is large, and the accused firms are important to them. Therefore, this leads to an endogeneity problem if the institutional lead plaintiff is omitted as a control variable due to the possible correlation with the litigation outcomes. Additionally, institutional investors are more likely to serve as the lead plaintiff for certain types of lawsuits. Generally Accepted Accounting Principles (GAAP) violation, for example, could lead to a larger settlement that encourages institutional investors to join in the suit as a lead plaintiff. (26)

Excluding 56 ongoing cases leaves 142 lawsuits. Among them, 79 were dismissed and 63 were settled. The information advantage hypothesis predicts that the sued firms short by UW-affiliated funds are less likely to be dismissed ([beta] < 0). In Table V, the first two columns report the results of the probit regressions. The coefficients on UWIB_Short are -0.216 and -0.177, without and with controlling for lawsuit characteristics, respectively. This suggests that selling shares of a sued firm by UW-affiliated funds lowers the likelihood of a case being dismissed. With respect to economic significance, the results show that the selling of shares by affiliated funds reduces the probability of dismissal by approximately 50%. (27) Among other control variables, the institutional lead plaintiff lowers the probability of a case being dismissed.

I now turn to the settlement amount. I replace the dependent variable in Equation (5) with the logarithm of the total settlement amount. The dismissed cases remain in the sample with a zero-settlement amount. (28) All variables are as previously defined.

[Log [Settlement.sub.ic] = [alpha] + [beta] * UWIB_[Short.sub.f,i,c] + [gamma] * IB_[Short.sub.f,i,c] + [delta] * [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c]], (6)

I posit that the sued firms that are sold by UW-affiliated funds tend to settle their cases for larger amounts. Therefore, the coefficient on UWIBShort is expected to be positive ([beta] > 0).

The last two columns of Table V present evidence that the cases for firms where UW-affiliated funds lower their positions result in larger settlement amounts, consistent with the information advantage hypothesis. IB-affiliated funds have no association between their selling and settlement amounts. Having an institution as lead plaintiff is likely to increase the settlement amount. In this subsection, I find that UW-affiliated funds could predict the outcomes of potential lawsuits. They tend to sell shares of sued firms whose cases are unlikely to be dismissed and tend to be settled with a greater amount.

III. Underwriting Relationships as Information Channels

Thus far, my findings indicate that UW-affiliated funds reduce their stakes in sued firms before the public recognizes the sued firms' misbehavior. The affiliated funds obtain the nonpublic, material information on a sued firm from fund-underwriter relationships. This section aims to investigate whether underwriting relationships are information channels behind the informed trading of UW-affiliated funds.

A. Informed Trading by Underwriters

If underwriters discover the alleged activities of their clients, they may exploit the private information prior to the information being made public, specifically before the CPE, because the fallout of sued firms is severe. On the other hand, underwriters may sacrifice their short-term trading profits for long-term business relationships, implying increased stakes in sued firms. Therefore, I run similar regressions for underwriters and their equivalent counterparts--nonunderwriting institutions.

Table VI presents holding changes in sued firms by underwriters. Similarly, HC1 measures ownership changes for institutional investors (i.e., underwriters and nonunderwriting institutions) from quarter Q-2 to quarter Q-1, where quarter Q is the quarter of the CPE. Panel A reports univariate test results of the quarterly holding changes. Underwriters with private information on accusations tend to lower their holdings of sued firms, while nonunderwriting institutions do not change their positions. Underwriters apparently attempt to capitalize on potential loss from lawsuits by exiting the sued firms before bad news arrives at the market.

13f filings aggregate equity positions among related institutions, including mutual funds. Thus, selling or buying a particular stock could be driven by a fund instead of underwriters themselves. To separate investment activities of underwriters from their funds, I introduce a new trading measure, PROP, calculated as a difference in the ownership changes between banks and their affiliated funds during the same quarter. The measure is significantly negative for the underwriters but is small and insignificant for the rest of the 13f investors.

Panel B shows multivariate test results of quarterly holding changes. The variable of interest is UWIB, which equals one if a firm is held by its underwriter and zero otherwise. Another dummy variable is IB, which equals one if a firm is held by an investment bank and zero otherwise. The coefficients on UWIB are -0.527 and -0.485, negative and statistically significant, indicating that the lead underwriters themselves are more likely to sell the shares of their clients than are nonunderwriting institutions before the CPE.

B. Underwriter-Involved Lawsuits

Next, I restrict the sample of lawsuits to those involving IPOs and SEOs, merger and acquisitions (M&As), SEC Section 11, and lawsuits with an investment bank named as a codefendant. (29) The aim of this subsample analysis (the sample of underwriter-involved lawsuits) is to see whether it supports the information advantage hypothesis, with an emphasis on the underwriting relationship as an information channel. Underwriters play a critical role as advisers in IPOs, SEOs, and M&As. Thus, they are heavily engaged in the transactions and have better access to management. If the fund-underwriter relationship is the source of information, underwriter-involved lawsuits could provide their affiliated funds with greater information advantage than a typical lawsuit.

Table VII reports the results of the subsample analysis, summarizing the coefficients on UWIB and IB across regressions of the two dependent variables (HC1 and HC2) with style and year fixed effects. Consistent with the information advantage hypothesis, the coefficients on UWIB, -0.058 and -0.345, in the first two columns are statistically significant at the 1% level, indicating that UW-affiliated funds decrease their percentage of ownership by 0.058% or 0.345% more than nonaffiliated funds do. The amount of reduction in equity positions is much greater for underwriter-involved lawsuits than the amount for the entire sample. It suggests that underwriters and their affiliated funds are more likely to have accurate information about accusations and exploit them prior to public disclosure.

The last two columns of Table VII replicate the regression results in Table III for lawsuits involving underwriters. I compare the stockholdings of sued firms with those of nonsued control firms in the same industry. The results remain the same.

IV. Discussion and Robustness

So far, my findings support the information advantage hypothesis. In this section, I discuss and explore the results with various robustness specifications.

A. Mutual Funds with Nonzero Holdings in Sued Firms

I have considered mutual funds that hold sued firms at least once over the past four quarters leading up to the quarter of disclosure (CPE) because mutual funds' selective holding (or non-holding) of some sued firms may contain information. I now restrict the sample to the mutual funds that must hold sued firms during the quarter before the CPE. (30) All the funds (UW-affiliated, IB-affiliated, and Nonaffiliated) hold a positive position in sued firms at the beginning of the event quarter. Given the nonzero stockholdings of sued firms, funds that attempt to capitalize on a loss due to fraud disclosure must sell their shares (active selling). These active trading activities explicitly reveal the information about the potential quality of the sued firms.

Panel A of Table VIII presents the regression results. Among the 198 lawsuits, the sample consists of 8,128 holding changes (i.e., nonzero holdings). The coefficients on UWIB are -0.051 and -0.299 for only sued firms and -0.051 and -0.197 for both the sued firms and control firms, respectively. The coefficients are statistically significant at the 1% level and are slightly greater than the comparable coefficients in Table III, whose sample includes mutual funds with zero holdings. Overall, the results strongly support the information advantage hypothesis.

B. Fund-Underwriter Clustered Standard Errors

The investment strategy of mutual funds that belong to a family or a parent institution would be different from that of a stand-alone mutual fund. It is likely that trading activities of UW-affiliated funds or IB-affiliated funds are correlated within their families or parent banks. One way to control the correlation in errors is to calculate the clustered standard errors at the level of affiliation. Therefore, I run regressions of Equations (2) and (3) with fund-underwriter or family clustered standard errors. Panel B of Table VIII summarizes coefficients on variables of interests. In both cases--that is, sued firms only and a sample containing both sued and nonsued firms--UW-affiliated funds significantly reduce their holdings in sued firms and lower their positions in the firms in excess of nonsued control firms. The results support the information advantage hypothesis.

C. Trading on Innocent Firms

Among the 198 sample firms, 79 firms (or cases) ended up being dismissed. Those firms that successfully defend themselves and dismiss their cases manage to avoid legal penalties, large settlements, and a loss of reputation (Karpoff, Lee, and Martin, 2008). If UW-affiliated funds correctly predict the lawsuit outcome (i.e., dismissal of one's case), it is not necessary to lower their holdings in dismissed firms because they may recover from the fall due to wrong accusations. Therefore, I run the same regression of Equation (2) again but with only the 79 dismissed firms to determine how the UW-affiliated funds trade those innocent firms.

Panel C reveals the regression results. The coefficients on UW-affiliated funds are negative but insignificant, indicating that the affiliated funds tend to not remove their positions in firms with wrong accusations. This result holds for both measures of holding changes. Overall, these statistics suggest that UW-affiliated funds with private information on underwriting clients can predict the outcomes of potential lawsuits against the clients. This implies that only the UW-affiliated funds could obtain information about the misconduct of sued firms and the severity of the transgression from their underwriters, thus supporting the information advantage hypothesis.

D. Interaction Effects in Regressions

In this section, I perform the robustness analysis with respect to interaction terms. (31) The main variable of interest in my regressions is UWIB, an interaction term between two identification variables, underwriters and investment banks. Balli and Sorensen (2013) provide a robust specification and interpretation of linear regression models with interaction terms. Because of nonlinearity, the interaction terms may spuriously pick up the correlation with other variables. Therefore, I attempt to orthogonalize UWIB to other variables and conduct robustness checks.

I run three different robustness specifications. First, I employ interaction terms with centering where each variable of the interaction terms is demeaned. Centering prevents the interaction (i.e., UWIB) from capturing fund-varying slopes. Second, 1 orthogonalize the interaction term based on the Frisch-Waugh theorem. In practice, each variable of the interaction term is regressed on the other control variables to net out any effect of the variables. Third, I add a set of control variables where IB (a dummy variable for investment banks) is interacted with each of the other control variables. By controlling interactions with IB, I rule out the possibility of having the control variables influence the variable of interest through IB. All test results are shown in Panel D. The coefficients on UWIB are negative and statistically significant at the 5% (Columns (1) and (2)) and 1% (Column (3)) level, respectively. Overall, the evidence is robust and supports the information advantage hypothesis.

V. Conclusion

In this paper, I investigate how mutual funds that are affiliated with lead underwriters use the information obtained through their underwriting relationships. In special circumstances, such as class-action lawsuits being taken against the underwritten firms, underwriter-affiliated funds may benefit by using the private information obtained through their investment banking relationships. Alternatively, they may support the client firms to benefit their investment banks at the expense of the fund investors.

Using a unique sample of securities class-action lawsuits and mutual funds from 2005 to 2016, I find evidence that underwriter-affiliated funds decrease their positions in sued firms underwritten by their banks in the three years prior to the revelation date. The results suggest that the information advantage of underwriter-affiliated funds surpasses the effects of conflicts of interest. This conclusion remains valid, even after controlling for various factors that can affect the affiliated funds' motivations for engaging in trading activities. Moreover, the results of the matching-firm analysis indicate that only underwriter-affiliated mutual funds have this information advantage.

I examine the effect of mutual fund trading on the outcome of lawsuits and find that a decrease 8in underwriter-affiliated funds' stock holdings of sued firms is likely to lower the probability of a lawsuit being dismissed and increase the settlement amount. This implies that the affiliated funds can estimate the damage of a firm's misconduct and predict the consequences.

This paper contributes to the debate on the extent to which information advantages and conflicts of interest affect the investment decisions of mutual funds that belong to investment banks. My findings show that under certain circumstances, underwriter-affiliated funds may not want to support the firms with which they have a relationship.
Variable            Definition

H1                  The percentage of a firm's outstanding shares held
                    by a mutual fund
H2                  The percentage of a firm's outstanding shares held
                    by a fund out of all the shares being held by all
                    mutual funds during the same period
HC1                 The change in the percentage of a firm's
                    outstanding shares held by a mutual fund
HC2                 The change in the percentage of a company' shares
                    held by a fund out of all of the shares of the
                    company that are being held by all mutual funds
                    during the same period
UWIB                One if a firm is held by a mutual fund that
                    belongs to an investment bank and the
                    investment bank is the lead underwriter of the
                    firm within the past three years and zero
                    otherwise
IB                  One if a firm is held by a mutual fund that
                    belongs to an investment bank and zero otherwise
UWIB Short          One if an underwriter-affiliated fund lowers its
                    position on a firm and zero otherwise
IB Short            One if an investment bank-affiliated fund lowers
                    its position on a firm and zero otherwise
Firm Performance    Cumulative abnormal returns over the past six
                    months, using the Carhart (1997) four-factor model
Trading Volume      Three-month moving average of the monthly trading
                    volume
Size                Price times the number of shares outstanding
Book-to-Market      Common stock amount relative to the size of the
                    firm
Analyst Coverage    The average number of analysts following a sample
                    firm
Analyst             Mean value of analyst recommendations
Consensus
Lag Return          Prior-quarter quarterly returns
Total Net Assets    Total assets under management
Turnover            Aggregated sales or purchases of securities divided
                    by the average 12-month total net assets of the
                    fund
Expenses            The ratio of a fund's operating expenses to the
                    total assets under management
Fund Age            Time in years since the mutual fund began trading
SUE                 One if a firm is sued under a securities class
                    action and zero otherwise
Case-Dismiss        One if the case is dismissed and zero otherwise
Log_Settlement      The logarithm of the total settlement amount
Institutional Lead  One if a lead plaintiff is an institutional
Plaintiff           investor and zero otherwise


References

Agarwal, V., K.A. Mullally, Y. Tang, and B. Yang, 2013, "Mandatory Portfolio Disclosure, Stock Liquidity, and Mutual Fund Performance," Journal of Finance 70, 2733-2776.

Bajaj, M., S. Mazumdar, and A. Sarin, 2003, "Empirical Analysis of Securities Class Action Settlements," Santa Clara Law Review 43, 101-132.

Balli, H.O. and B.E. Sorensen, 2013, "Interaction Effects in Econometrics," Empirical Economics 45, 583-603.

Beatty, R.P. and J. Ritter, 1986, "Investment Banking, Reputation, and the Underpricing of Initial Public Offerings," Journal of Financial Economics 15, 213-232.

Ber, H., Y Yafeh, and O. Yosha, 2001, "Conflict of Interest in Universal Banking: Bank Lending, Stock Underwriting, and Fund Management," Journal of Monetary Economics 47, 189-218.

Berzins, J., C.H. Liu, and C. Trzcinka, 2013, "Asset Management and Investment Banking," Journal of Financial Economics 110, 215-231.

Bodnaruk, A., M. Massa, and A. Simonov, 2009, "Investment Banks as Insiders and the Market for Corporate Control," Review of Financial Studies 22, 4989-5026.

Campbell, T.S. and W.A. Kracaw, 1980, "Information Production, Market Signalling, and the Theory of Financial Intermediation, Journal of Finance 35, 863-882.

Carhart, M., 1997, "On Persistence in Mutual Fund Performance," Journal of Finance 52, 57-82.

Chemmanur, T.J. and P. Fulghieri, 1994, "Investment Bank Reputation, Information Production, and Financial Intermediation," Journal of Finance 49, 57-79.

Cheng, C.S.A., H.H. Huang, Y. Li, and G. Lobo, 2010, "Institutional Monitoring through Shareholder Litigation," Journal of Financial Economics 95, 356-383.

Dunbar, C, 2000, "Factors Affecting Investment Bank Initial Public Offering Market Share," Journal of Financial Economics 55, 3-41.

Ellis, K., R. Michaely, and M. O'Hara, 2000, "When the Underwriter Is the Market Maker: An Examination of Trading in the IPO Aftermarket," Journal of Finance 55, 1039-1074.

Fama, E.F. and K.R. French, 1993, "Common Risk Factors in the Returns on Stocks and Bonds," Journal of Financial Economics 33, 3-56.

Golez, B. and J.M. Martin, 2015, "Price Support by Bank-Affiliated Mutual Funds," Journal of Financial Economics 115, 614-638.

Griffin, J.M., J.H. Harris, and S. Topaloglu, 2007, "Why Are IPO Investors Net Buyers through Lead Underwriters?" Journal of Financial Economics 85, 518-551.

Griffin, J.M., T. Shu, and S. Topaloglu, 2012, "Examiningthe Dark Side of Financial markets: Do Institutions Trade on Information from Investment Bank Connections?" Review of Financial Studies 25, 2155-2188.

Hao, Q. and X. Yan, 2012, "The Performance of Investment Bank-Affiliated Mutual Funds: Conflicts of Interest or Informational Advantage?" Journal of Financial and Quantitative Analysis 47, 537-565.

Hong, H. and J.D. Kubik, 2003, "Analyzing the Analysts: Career Concerns and Biased Earnings Forecasts," Journal of Finance 58, 313-351.

Ivashina, V and Z. Sun, 2011, "Institutional Stock Trading on Loan Market Information," Journal of Financial Economics 100, 284-303.

Karpoff, M.J., D.S. Lee, and S.G Martin, 2008, "The Cost to Firms of Cooking the Books," Journal of Financial and Quantitative Analysis 43, 581-612.

Kedia, S. and X. Zhou, 2014, "Informed Trading around Acquisitions: Evidence from Corporate Bonds," Journal of Financial Markets 18, 182-205.

Krigman L., W.H. Shaw, and K.L. Womack, 2001, "Why Do Firms Switch Underwriters?" Journal of Financial Economics 60, 245-284.

Leland, H.E. and D.H. Pyle, 1977, "Informational Asymmetries, Financial Structure, and Financial Intermediation," Journal of Finance 32, 371-387.

Massa, M., 2003, "How Do Family Strategies Affect Fund Performance? When Performance-Maximization Is Not the Only Game in Town," Journal of Financial Economics 67, 249-304.

Massa, M. and Z. Rehman, 2008, "Information Flows within Financial Conglomerates: Evidence from the Banks-Mutual Funds Relation," Journal of Financial Economics 89, 288-306.

Ritter, J.R. and D. Zhang, 2007, "Affiliated Mutual Funds and the Allocation of Initial Public Offerings," Journal of Financial Economics 86, 337-368.

Schultz, P.H. and M.A. Zaman, 1994, "Aftermarket Support and Underpricing of Initial Public Offerings," Journal of Financial Economics 35, 199-219.

Wermers, R., 2000, "Mutual Fund Performance: An Empirical Decomposition into Stock-Picking Talent, Style, Transaction Costs, and Expenses," Journal of Finance 55, 1655-1695.

Yan, X.S. and Z. Zhang, 2009, "Institutional Investors and Equity Returns: Are Short-Term Institutions Better Informed?" Review of Financial Studies 22, 893-924.

Hyoseok (David) Hwang (*)

(*) Hyoseok (David) Hwang is an Assistant Professor of Finance in the School of Business at Rutgers University in Camden, NJ.

(1) Golez and Marin (2015) consider fund managers to be double agents. Fiduciary duty mandates that fund investors' interests come before those of management and shareholders of a fund's parent firm. In practice, the opposite is possible.

(2) See, for example, "Wall Street's Dumping Ground," Bloomberg (June 2004).

(3) The dollar amount of the initial positions is calculated as the positive initial positions of the underwriter-affiliated funds times their average total net assets.

(4) Following the empirical design of Hao and Yan (2012), my analyses employ three-year underwriting relationships before the end of the class period; however, I also test equity-offering firms within five years of the event and obtain similar results.

(5) Under the SEC Act of 1934, all institutional investors with security assets of $100 million or more under discretionary management are required to report their holdings each quarter on Form 13f.

(6) Some 13f institutions report a consolidated position of their equity holdings.

(7) My sample ends in 2016 due to data availability for stockholdings of mutual funds in Wharton Research Data Services (WRDS).

(8) The SEC requires an institutional investment manager that exercises investment discretion over $100 million or more in Section 13(f) securities to report its holdings on Form 13F.

(9) Prior to the regulation change, portfolio managers disclose their holdings and investment activities on a semiannual basis (see Agarwal et al., 2013).

(10) I followed the approach of Massa and Rehman (2008). Some mutual funds have multiple classes, and collapsing them into one is suggested to merge the CRSP mutual fund database and Thomson mutual fund holdings database.

(11) See http://securities.stanford.edu/index.html.

(12) The type of specific accusation against a sued firm varies. For example, all the sample firms are accused of general financial misreporting. About 90% of the sample is sued due to the artificial inflation of their stock prices. Violation of SEC1934 Sections 10(b) and rule 10b-5 is also covered by approximately 90% of the overall sample.

(13) A lawsuit against Alios Therapeutics, on July 21, 2011. The plaintiff filed a class-action complaint for breach of fiduciary duties. The class period started on July 20, 2011, and ended on July 21, 2011.

(14) J. C. Penney Company, Inc., issued new equity on September 26, 2013, which is the same date as the CPE date.

(15) See Carhart (1997); [R.sub.i,t] = [[alpha].sub.i] + [[beta].sub.1] [RMRF.sub.t] + [[beta].sub.2] [SMB.sub.t] + [[beta].sub.3] [HML.sub.t] + [[beta].sub.4] [UMD.sub.t] + [[epsilon].sub.i,t], where [R.sub.i,t] is the daily return of the sued firm i in excess of the one-month T-bill return; RMRF is the excess return on a value-weighted market portfolio; SMB, HML, and UMD are returns on zero-investment factor-mimicking portfolios for size, book-to-market, and momentum, respectively.

(16) Following Massa and Rehman (2008), multiple classes of the same fund are collapsed by taking the TNAs-weighted average of the individual class characteristics.

(17) This is consistent with Hao and Yan (2012).

(18) Massa and Rehman (2008) employ the same measures for mutual fund trading.

(19) There could be various reasons for mutual funds to hold or not hold sued firms in any given period. Because both the investing strategies may contain information, the analysis is rather limited if the sample excludes those mutual funds holding zero-sued firms at quarter Q-2, where Q is the quarter of the CPE. I also test the trading measures where a mutual fund is required to own the stock at quarter Q-2 and find similar results. The results are reported in Section IV.

(20) See Massa and Rehman (2008).

(21) I appreciate an anonymous referee for suggesting clustering techniques in a regression setting.

(22) I report the standard errors clustered at time and affiliation. In a robustness test, I also calculate affiliation-clustered standard errors and find consistent results.

(23) These results still hold when considering a five-year underwriting relationship between a lead underwriter and a sued firm.

(24) See, for example, Yan and Zhang (2009).

(25) See Massa and Rehman (2008).

(26) Bajaj, Mazumdar, and Sarin (2003), for example, show settlement amounts based on alleged damages and allegation types.

(27) The effect of UWIBShort is calculated as the difference in predicted dismissal probability when UWIB_Short changes from zero to one while holding all other variables at their means.

(28) I also test the sample excluding lawsuits dismissed and find similar results.

(29) Section 11 of the Securities Act of 1933 allows investors to file a lawsuit against issuers and underwriters for the price decline below the IPO price due to material omissions in the prospectus.

(30) I appreciate an anonymous referee for suggesting the idea of mutual funds with nonzero holdings.

(31) I appreciate an anonymous referee for suggesting the robustness checks.
Table I. Sample Summary Statistics

This table provides the summary statistics for securities class-action
lawsuits, firms, and mutual funds in our sample from 2005 to 2016.
Panel A presents the number of lawsuits each year. # Public Equity
Offering is the number of equity offerings each year. # Sued Equity
Offering is the number of equity offerings sued within three years. #
Sample is the number of lawsuits each year in our sample. For each
lawsuit, the type of equity issuance is reported. Each case in our
sample is classified as settled, dismissed, or ongoing. Panel B shows
the mean (median) number of days for the class period, the filing
period, and the underwriting period. Panel C reports firm
characteristics and control variables for sample firms during the
quarter prior to the quarter of the class period ending (CPE) date. All
variables are reported as the mean and median values and are computed
using the relevant COMPUSTAT, CRSP, and I/B/E/S data. Size is
calculated as the price multiplied by shares outstanding and reported
in millions. Book-to-Market equals the book value, which is common
equity divided by market capitalization. Trading Volume is the
quarterly average trading volume and is reported in millions. Analyst
Coverage is the average number of analysts following a sample firm, and
Analyst Consensus is the average of analyst recommendations. Firm
Performance is calculated as cumulative abnormal returns over the six
months prior to the quarter of the CPE using the Carhart (1997) four-
factor model. Panel D reports the summary statistics for mutual funds
in terms of affiliation. UW-affiliated funds are defined as mutual
funds affiliated with a lead underwriter of sample firms, IB-affiliated
funds are mutual funds related to an investment bank (excluding UW-
affiliated funds), Nonaffiliated funds are independent mutual funds.
Mutual funds are those with total net assets exceeding $1 million, fund
age older than one year, and a portfolio with more than 85% in stocks.
Multiple classes of the same fund are collapsed by taking the total net
assets-weighted average of the individual class characteristics. Total
assets under management (Total Net Assets) are reported in millions.
Net Asset Value is calculated as a fund's underlying assets minus its
liabilities, divided by the number of shares outstanding. Turnover is
the aggregated sales or purchases of securities divided by the average
12-month total net assets of the fund. Expense Ratio is the proportion
of a fund's total operating expenses that shareholders pay for the
fund's operating expenses. Lag Quarterly Return is the past quarter raw
returns. Fund Age is the number of years since the mutual fund began
trading. Absolute values of the t-statistics are reported.

Panel A. Number of Lawsuits

       # Public Equity  # Sued Equity  # Sample      # Outcome
Year   Offering         Offering       (IPO/SEO)     (Settlement/
                                                      Dismiss/
                                                      Ongoing)

2005     644             33             17 (1/16)    10/7/0
2006     625             24             16 (2/14)     8/7/1
2007     629             31             16 (2/14)     9/7/0
2008     274             25             20 (4/16)    10/9/1
2009     524             17             12 (2/10)     6/6/0
2010     552             31             20 (1/19)     4/13/3
2011     449             31             14 (2/12)     5/8/1
2012     566             27             13 (2/11)     4/2/7
2013     793             36             25 (4/21)     1/9/15
2014     711             41             16 (1/15)     0/2/14
2015     636             43             15 (0/15)     6/6/3
2016     501             48             14 (0/14)     0/3/11
Total  6,904            387            198 (21/177)  63/79/56

Panel B. Number of Days around Class Period End

Event Periods                                Mean   Median    Min.

From Class Period Start to Class Period End  364    272       1
From Class Period End to Filing Date          82     20       0
From Underwriting Date to Class Period End   466    434       0

Event Periods                                Max.     S.D.

From Class Period Start to Class Period End  1,762    334.41
From Class Period End to Filing Date           673    143.91
From Underwriting Date to Class Period End   1,085    297.29

Panel C. Sample Firms

Firm Characteristics   #OBS  Mean      Median

Size (millions)        198   3,241     1,092
Book-to-Market         198       0.32      0.24
Trading Volume         198      36.44     13.96
Analyst Coverage       198      11.42     10.00
Analyst Consensus      198       3.90      4.00
Firm Performance       198       0.01      0.01

Panel D. Sample Mutual Funds and Affiliations

                      (1)             (3)
Fund Characteristics  UW-Affiliated   Nonaffiliated  (1)-(2)

TNA (millions)        704                623         12.71 (***)
Net Asset Value        27.65              26.57       4.49 (***)
Turnover                0.81               0.87       1.69 (*)
Expense Ratio           0.01               0.01       2.11 (**)
Lag Quarterly Return    0.04               0.04       0.97
Fund Age               24.19              25.46       1.69 (*)
#OBS                  351             12,185

                      (1)-(3)
Fund Characteristics  t-stat         (2)-(3)

TNA (millions)        16.53 (***)     8.18 (***)
Net Asset Value        0.99          12.74 (***)
Turnover               1.76 (*)       0.16
Expense Ratio          1.12           7.49 (***)
Lag Quarterly Return   0.67           0.90
Fund Age               2.33 (**)      1.77 (*)
#OBS

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table II. Stockholdings of Mutual Funds in Sample Firms

This table presents initial positions and trading activities of mutual
funds in the four quarters leading up to the quarter of the class
period ending (CPE) date. Panel A reports initial positions of mutual
funds over the past four quarters before the CPE. HI and H2 are
calculated as the percentage of a firm "s outstanding shares held by a
mutual fund and the percentage of a firm's outstanding shares held by a
mutual fund out of all the shares of the firm held by all mutual funds
during the same period, respectively. Panel B reports quarterly holding
changes for sued firms, which are calculated as follows: HC1, changes
in the percentage of a firm's outstanding shares held by a mutual fund,
and HC2, changes in the percentage of a company' shares held by a fund
out of all of the shares of the company that are being held by all
mutual funds during the same period. UW-affiliated funds are the mutual
funds that belong to an investment bank that served as a lead
underwriter of the sample firms within the past three years. IB-
affiliated funds are the mutual funds that belong to an investment
bank. Nonaffiliated funds are independent mutual funds. Absolute values
of the r-statistics are reported.

Panel A. Initial Positions

           (1) UW-Affiliated    (2) IB-Affiliated
Quarter    H1         H2          H1          H2

Q-4          0.145      0.924        0.249       1.512
Q-3          0.163      0.894        0.259       1.472
Q-2          0.184      0.991        0.250       1.394
Q-1          0.148      0.786        0.245       1.383
#OBS       541        541        4,052        4,05

           (3) Nonaffiliated                    t-stat
Quarter    H1         H2           (1)-(2)      (1)-(3)      (2)-(3)

Q-4            0.230      1.552    3.204 (***)  3.135 (***)  1.521
Q-3            0.241      1.573    2.851 (***)  2.962 (***)  1.421
Q-2            0.249      1.604    1.990 (**)   2.512 (***)  0.063
Q-1            0.247      1.622    2.779 (***)  3.718 (***)  0.189
#OBS       8,483      8,483

Panel B. Quarterly Holding Changes

             (1) UW-Affiliated            (2) IB-Affiliated
Quarter      HC1           HC2            HC1            HC2

Q-4 to Q-3    0.018 (**)   -0.030          0.009         -0.040
             (2.037)       (0.462)        (1.579)        (0.997)
Q-3 to Q-2    0.021 (***)   0.097 (**)    -0.008         -0.078 (*)
             (3.687)       (2.551)        (1.501)        (1.936)
Q-2 to Q-1   -0.036 (***)  -0.205 (***)   -0.005         -0.011
             (4.066)       (3.723)        (0.762)        (0.261)
Q-l to Q     -0.023 (***)  -0.065         -0.049 (***)   -0.138 (***)
             (3.351)       (1.617)        (5.934)        (2.737)

             (3) Nonaffiliated
Quarter      HC1            HC2

Q-4 to Q-3    0.011 (***)    0.021
             (3.200)        (0.805)
Q-3 to Q-2    0.009 (***)    0.031
             (2.865)        (1.057)
Q-2 to Q-1   -0.002          0.018
             (0.618)        (0.720)
Q-l to Q     -0.030 (***)    0.070 (**)
             (7.121)        (2.018)

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table III. Determinants of the Change in Mutual Fund Holdings prior to
the End of the Class Period

This table presents the regression results of percentage holding
changes by mutual funds in sample firms for the quarter before the
class period ending date. The first two columns compare the changes in
mutual funds' stockholdings of sued firms. I estimate the following
model: [DELTA][HC.sub.f,i,c] = [alpha] + [beta] (*) [UWIB.sub.f,i,c] +
[gamma] (*) [IB.sub.f,i,c] + [delta] (*) [Controls.sub.f,i,c] +
[[epsilon].sub.f,i,c], where [DELTA][HC.sub.f,i,c] is the percentage
change in the holdings of fund f in the stock of sued firm i before the
end of the class period c. HC1 is the changes in the percentage of a
firm's outstanding shares held by a mutual fund, and HC2 is the changes
in the percentage of a company's shares held by a fund out of all of
the shares of the company that are being held by all mutual funds
during the same period. The variable of interest is UWIB, which equals
one if a firm is held by a mutual fund that belongs to an investment
bank that served as a lead underwriter of the sample firms within the
past three years and is zero otherwise. IB takes a value of one if a
firm is held by a mutual fund that belongs to an investment bank and
zero otherwise. Controls is a vector of control variables consisting of
sample firms and mutual funds characteristics. The details are
described in the Appendix. The last two columns compare the changes in
mutual funds' holdings of the stocks of sued firms with the changes in
their holdings of nonsued control firms. I estimate the following
model: [DELTA][HC.sub.f,i,c] = [DELTA][HC.sub.f,i,c] = [alpha] + [beta]
(*) [UWIB.sub.f,i,c] + [gamma] (*) [IB.sub.f,i,c] + [delta] (*)
[Controls.sub.f,i,c] + [[epsilon].sub.f,i,c], where
[DELTA][HC.sub.f,i,c] ([DELTA][HC.sub.f,i,c]) is the percentage change
in the holdings of fund f in the stock of sued firm i (control firm j)
before the end of the class period c. The control firms are nonsued
firms that are being held by fund f during the same period and that
belong to the same industry (two-digit SIC code) as the sued firm. Each
sued firm has more than one control firm held by the same fund around
the CPE. I compute the equal-weighted average holdings of mutual funds
in the control firms before the CPE and subtract them from the
percentage change in holdings of sued firms by the same funds. All
other variables remain the same. The t-statistics are reported in
parentheses.

                     Sued Firms Only
                     HC1                       HC2

UWIB                     -0.032 (**)       -0.184 (**)
                        (-2.323)          (-2.339)
IB                       -0.004            -0.056
                        (-0.404)          (-0.961)
Initial Position         -0.100 (***)      -0.091 (***)
                        (-6.261)          (-4.768)
Firm Performance          0.153 (**)       -0.081
                         (2.118)          (-0.148)
Trading Volume           -0.012            -0.025
                        (-1.319)          (-0.399)
Size                     -0.009 (**)       -0.065 (**)
                        (-2.247)          (-2.096)
Book-to-Market           -0.029             0.014
                        (-1.595)           (0.106)
Analyst Coverage          0.000            -0.003
                         (0.014)          (-0.622)
Analyst Consensus         0.008            -0.008
                         (1.337)          (-0.187)
Total Net Assets          0.000 (*)         0.003 (*)
                         (1.805)           (1.906)
Expenses                 -2.067 (***)     -21.946 (***)
                        (-2.949)          (-3.282)
Turnover                -0.010 (***)       -0.083 (***)
                        (-2.678)          (-2.854)
Lag return                0.069            -0.117
                         (1.520)          (-0.295)
Age                       0.001 (*)         0.006 (**)
                         (1.824)           (2.009)
Intercept                 0.195             1.727
                         (1.261)           (1.612)
Clustering           Affiliation         Affiliation
                     and Time            and Time
Style fixed effects  Yes                 Yes
Year fixed effects   Yes                 Yes
[R.sup.2]                 0.037             0.029
#OBS                 13,076            13,076

                     Sued Firms vs. Nonsued
                     Firms
                     HC1               HC2

UWIB                     -0.032 (**)       -0.193 (**)
                        (-2.298)          (-2.394)
IB                       -0.004            -0.062
                        (-0.343)          (-0.952)
Initial Position         -0.108 (***)      -0.096 (***)
                        (-5.422)          (-4.968)
Firm Performance          0.157 (**)       -0.032
                         (2.017)          (-0.055)
Trading Volume           -0.013            -0.013
                        (-1.528)          (-0.214)
Size                     -0.010 (**)       -0.051
                        (-2.128)          (-1.511)
Book-to-Market           -0.033 (*)         0.008
                        (-1.799)           (0.055)
Analyst Coverage          0.000            -0.003
                         (0.644)           (-0.593)
Analyst Consensus         0.015 (**)        0.008
                         (2.370)           (0.170)
Total Net Assets          0.000             0.003
                         (1.547)           (1.645)
Expenses                 -2.020 (***)     -19.856 (***)
                        (-2.700)          (-3.038)
Turnover                 -0.010 (**)       -0.072 (**)
                        (-2.275)          (-2.256)
Lag return                0.033            -0.310
                         (0.648)          (-0.734)
Age                       0.001 (*)         0.007 (*)
                         (1.660)           (1.955)
Intercept                 0.164             1.295
                         (1.593)           (1.152)
Clustering           Affiliation       Affiliation
                     and Time          and Time
Style fixed effects  Yes               Yes
Year fixed effects   Yes               Yes
[R.sup.2]                 0.036             0.027
#OBS                 13,076            13,076

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table IV. Comparison of Mutual Fund Holdings in Public Offering Firms

This table provides a matching-firm analysis, comparing sued public
offering firms with nonsued public offering firms. Panel A reports firm
characteristics of sued public offering firms (Sued POs) and nonsued
public offering firms (Nonsued POs). Firm characteristics include
market capitalization (SIZE), book-to-market ratio, average quarterly
trading volume (Trading Volume), a number of analyst following a firm
(Analyst Coverage), mean level of recommendation consensus (Analyst
Consensus), and abnormal returns over the past six months (Firm
Performance). Panel B shows the regression results of the changes in
holdings of underwriter-affiliated funds in the stocks of sued firms
and control firms. I estimate the following model:
[DELTA][HC.sub.f,i,c] = [alpha] + [beta] (*) [SUE.sub.t] + [delta] (*)
[Controls.sub.f,i,c] + [[epsilon].sub.f,i,c], where
[DELTA][HC.sub.f,i,c] is the percentage change in the holdings of
underwriter-affiliated fund/in the stock of a firm i before the end of
the class period c. The firm i is a sued firm or a nonsued firm that is
underwritten by the same lead underwriter as the sued firms within the
past three years and held by an underwriter-affiliated fund/during the
same period as the sued firms. HC1 and HC2 are the dependent variables.
HC1 is the changes in the percentage of a firm's outstanding shares
held by a mutual fund, and HC2 is the changes in the percentage of a
company's shares held by a fund out of all of the shares of the company
that are being held by all mutual funds during the same period. The
variable of interest is SUE, which equals one if a firm is subject to
class-action lawsuits and zero otherwise. Controls is a vector of
control variables consisting of firm and fund characteristics. The
details of the control variables are described in the Appendix. The t-
statistics are reported in parentheses.

Panel A. Sample Firms

                       (1) Sued POs          (2) Nonsued POs
Firm Characteristics   Mean       Median     STD        Mean

Size                   5,630      2,310      9,180      4,350
Book-to-Market             0.27       0.20       0.25       0.35
Trading Volume            57.20      19.90      78.70      33.80
Analyst Coverage          14.60      14.00       6.57      12.09
Analyst Consensus          3.83       4.00       0.53       3.82
Firm Performance           0.01       0.01       0.07       0.02

                                              (1)-(2)
Firm Characteristics   Median     STD         t-stat

Size                   1,890      8,620       1.21
Book-to-Market             0.29       0.43    2.83 (***)
Trading Volume            14.70      67.90    2.72 (***)
Analyst Coverage          11.00       6.59    3.48 (***)
Analyst Consensus          4.00       0.58    0.11
Firm Performance           0.02       0.06    0.96

Panel B. Multivariate Analysis

                      HC1                     HC2

SUE                      -0.022 (***)            -0.186 (***)
                        (-4.730)                (-3.375)
Control variables     Yes                     Yes
Clustering            Affiliation and Time    Affiliation and Time
Style fixed effects   Yes                     Yes
Year fixed effects    Yes                     Yes
[R.sup.2]                 0.007                   0.008
#OBS                  4,231                   4,231

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table V. Effect of Informed Trading by Mutual Funds on Lawsuit Outcomes

This table presents the multivariate regression results regarding the
effect of informed trading by mutual fund on the outcomes of lawsuits.
The sample consists of 142 firms with lawsuits either dismissed or
settled, excluding ongoing cases from 2005 to 2016. The first two
columns show the results of the probit regression testing the effect of
informed trading by mutual funds on the likelihood of the case being
dismissed. The dependent variable is a dummy variable that equals one
if the case is dismissed and zero otherwise. The last two columns
report the results of the ordinary least squares regression testing the
determinants of the settlement amounts. The dependent variable is the
logarithm of the settlement amount. The variable of interest is UMIB
Short, which equals one if a mutual fund affiliated with a lead
underwriter of a sued firm sells shares of the firm and zero otherwise.
IB Short takes a value of one if a mutual fund that belongs to an
investment bank sells shares in a firm and zero otherwise.
Institutional Lead Plaintiff is one if a lead plaintiff is an
institutional investor and zero otherwise. All the control variables
remain the same and are describe in the Appendix. The t-statistics are
reported in parentheses.

                              Likelihood of the Case
                              Being Dismissed
                              (1)                  (2)

UWIB Short                       -0.216 (**)         -0.177 (*)
                                (-2.241)            (-1.755)
IB Short                         -0.033              -0.051
                                (-0.859)            (-1.290)
Institutional Lead Plaintiff                         -0.431 (***)
                                                   (-12.959)
Firm Performance                 -0.314               0.890 (***)
                                (-1.385)             (3.576)
Trading Volume                    0.566 (***)         0.570 (***)
                                (12.174)            (11.275)
Size                              0.038 (***)         0.029 (***)
                                 (7.460)             (4.495)
Book-to-Market                   -0.088               0.277 (***)
                                (-1.255)             (3.372)
Analyst Coverage                 -0.001               0.011 (***)
                                (-0.406)             (4.112)
Analyst Consensus                -0.215 (***)        -0.152 (***)
                                (-9.224)            (-6.271)
Lawsuits                      No                  Yes
Style fixed effects           Yes                 Yes
Year fixed effects            Yes                 Yes
[R.sup.2]                         0.040               0.129
#OBS                          9,905               9,905

                              Settlement Amount
                              (3)                 (4)

UWIB Short                        0.219 (**)          0.138 (*)
                                 (2.048)             (1.788)
IB Short                         -0.000               0.029
                                (-0.011)             (0.774)
Institutional Lead Plaintiff                          0.197 (***)
                                                     (4.075)
Firm Performance                  0.976 (***)        -2.623 (***)
                                 (2.64)             (-5.01)
Trading Volume                    0.668 (***)         0.712 (***)
                                (37.36)             (33.06)
Size                              0.535 (***)         1.365 (***)
                                 (4.30)             (12.42)
Book-to-Market                   -2.458 (***)        -2.802 (***)
                               (-16.31)            (-21.63)
Analyst Coverage                  0.016 (***)        -0.010 (**)
                                 (4.22)             (-2.06)
Analyst Consensus                 0.334 (***)         0.418 (***)
                                 (9.72)             (15.25)
Lawsuits                      No                  Yes
Style fixed effects           Yes                 Yes
Year fixed effects            Yes                 Yes
[R.sup.2]                         0.508               0.623
#OBS                          4.418               4,418

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table VI. Changes in Holdings of Sued Firms by Underwriters

This table presents the regression results of percentage holding
changes in sued firms by 13f institutional investors during the quarter
before the class period ending date. Panel A reports univariate tests
of quarterly holding changes. HC1 is the percentage change in shares of
a firm held by a 13f institutional investor from quarter Q-2 to quarter
Q-1, where quarter Q is the quarter of the class period ending, and
PROP is calculated as differences in the percentage change in shares of
a firm held by 13f institutional investors and their affiliated mutual
funds from quarter Q-2 to quarter Q-1. Panel B shows multivariate test
results of quarterly holding changes. I estimate the following model:
[[DELTA][HC.sub.k,i,c] = [alpha] + [beta] (*) [UW.sub.k,i,c] + [gamma]
(*) [IB.sub.k,i,c] + [delta] (*) [Controls.sub.k,i,c]. +
[[epsilon].sub.k,i,c], where [DELTA][HC.sub.k,i,c] is the percentage
change in the holdings of an institution k in the stock of sued firm i
before the end of the class period c. HC1 and PROP are the dependent
variables. The variable of interest is UW, which equals one if a firm
is held by an investment bank that served as a lead underwriter of the
firm within the past three years and zero otherwise. IB takes a value
of one if a firm is held by an investment bank and zero otherwise. All
the control variables remain the same and are describe in the Appendix.
The t-statistics are reported in parentheses.

Panel A. Quarterly Holding Changes

                              #OBS     HC1             PROP

Underwriters                  140       -0.595 (***)    -0.552 (***)
                                       (-4.786)        (-4.528)
13f Institutional Investors   27,631    -0.004          -0.005
                                       (-0.857)        (-1.047)

Panel B. Multivariate Regressions

                     HC1                        PROP

UW                   -0.527 (*)                     -0.485 (*)
                     (-1.854)                      (-1.808)
IB                   0.077                           0.075
                     (0.647)                        (0.684)
Initial Position     -0.071 (***)                   -0.069 (***)
                     (-4.143)                      (-4.286)
Intercept            0.454                           0.430
                     (1.238)                        (1.233)
Control variables    Yes                        Yes
Clustering           13f Institution and Time   13f Institution and Time
Year fixed effects   Yes                        Yes
[R.sup.2]            0.045                           0.045
#OBS                 27,771                     27,771

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table VII. Underwriter-Involved Lawsuits

This table provides the regression results for the lawsuits in which
underwriters are closely involved, such as IPOs, SEOs, and M&As related
lawsuits and lawsuits for underwriters accused as codefendants. The
first two columns compare the changes in mutual funds' stockholdings of
sued firms. I estimate the following model: [DELTA][[HC.sub.f,i,c] =
[alpha] + [beta] (*) [UWIB.sub.f,i,c] + [gama] # [IB.sub.f,i,c] +
[delta] (*) [Controls.sub.f,i,c] + [[epsilon].sub.f,i,c], where
[delta][HC.sub.f,i,c] is the percentage change in the holdings of
fund/in the stock of sued firm i before the end of the class period c.
HC1 is the changes in the percentage of a firm's outstanding shares
held by a mutual fund, and HC2 is the changes in the percentage of a
company's shares held by a fund out of all of the shares of the company
that are being held by all mutual funds during the same period. The
variable of interest is UWIB, which equals one if a firm is held by a
mutual fund that belongs to an investment bank that served as a lead
underwriter of the sample firms within the past three years and is zero
otherwise. IB takes a value of one if a firm is held by a mutual fund
that belongs to an investment bank and zero otherwise. Controls is a
vector of control variables consisting of sample firms and mutual funds
characteristics. The details are described in the Appendix. The last
two columns compare the changes in mutual funds' holdings of the stocks
of sued firms with the changes in their holdings of nonsued control
firms. I estimate the following model: [[DELTA][HC.sub.f,i,c] -
[[DELTA][HC.sub.f,i,c] = [alpha] + [beta] (*) [UWIB.sub.f,i,c] +
[gamma] (*) [IB.sub.f,i,c] + [delts] (*) [Controls.sub.f,i,c] +
[[epsilon].sub.f,i,c], where [[DELTA][HC.sub.f,i,c]
([DELTA][HC.sub.f,i,c]) is the percentage change in the holdings of
fund/in the stock of sued firm i (control firm j) before the end of the
class period c. The control firms are nonsued firms that are being held
by fund/during the same period and that belong to the same industry
(two-digit SIC code) as the sued firm. Each sued firm has more than one
control firm held by the same fund around the CPE. I compute the equal-
weighted average holdings of mutual funds in the control firms before
the CPE and subtract them from the percentage change in holdings of
sued firms by the same funds. All other variables remain the same. The
t-statistics are reported in parentheses.

                     Sued Fit         ms Only
                     HC1              HC2

UWIB                    -0.058 (***)     -0.345 (***)
                       (-2.730)         (-3.873)
IB                       0.038            0.117
                        (1.330)          (0.820)
Initial Position        -0.055 (**)      -0.113 (***)
                       (-1.969)         (-3.816)
Intercept                0.005            2.491
                        (0.043)          (1.052)
Control variables    Yes              Yes
Clustering           Affiliation      Affiliation
                     and Time         and Time
Style fixed effects  Yes              Yes
Year fixed effects   Yes              Yes
[R.sup.2]                0.015            0.032
#OBS                 2,590            2,590

                     Sued Firms vs. Nonsued
                     Firms
                     HC1             HC2

UWIB                    -0.045 (***)   -0.333 (***)
                       (-3.168)       (-4.362)
IB                       0.029          0.105
                        (1.455)        (0.878)
Initial Position        -0.082 (**)    -0.122 (***)
                       (-2.300)       (-3.764)
Intercept               -0.024          1.915
                       (-0.167)        (1.451)
Control variables    Yes             Yes
Clustering           Affiliation     Affiliation
                     and Time        and Time
Style fixed effects  Yes             Yes
Year fixed effects   Yes             Yes
[R.sup.2]                0.020           0.031
#OBS                 2,590           2,590

(***)Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.

Table VIII. Robustness Tests

This table provides the robustness test results. HC1 is the percentage
change in shares of a firm held by a mutual fund from quarter Q-2 to
quarter Q-1, where quarter Q is the quarter of the class period ending,
and HC2 is calculated as changes in the percentage of a firm's shares
held by a fund out of all the shares of the firm held by all mutual
funds from quarter Q-2 to quarter Q-1. The variable of interest is
UWIB, which equals one if a firm is held by a mutual fund that belongs
to an investment bank that served as a lead underwriter of the sample
firms within the past three years and is zero otherwise. IB takes a
value of one if a firm is held by a mutual fund that belongs to an
investment bank and zero otherwise. All other control variables remain
the same and are described in the Appendix. Panel A presents the
subsample test results for trading by mutual fund with nonzero holdings
of sued firms. Panel B reports the estimates of regressions using fund
affiliation clustering techniques. Panel C reports regression results
for innocent firms that are wrongly accused and dismissed later. Panel
D reports robustness tests to interaction terms. The t-statistics are
reported in parentheses.

Panel A. Mutual Funds with Nonzero Holdings of Sued Firms

                     Sued Firm Only
                     HC1              HC2

UWIB                    -0.051 (***)     -0.299 (***)
                       (-4.130)        (-40.595)
IB                      -0.018           -0.125
                       (-1.639)         (-1.613)
Intercept               -0.022            0.081
                       (-0.150)          (0.088)
Control variables    Yes              Yes
Clustering           Affiliation      Affiliation
                     and Time         and Time
Style fixed effects  Yes              Yes
Year fixed effects   Yes              Yes
[R.sup.2]                0.042            0.045
#OBS                 8,128            8,128

                     Sued Firm vs. Nonsued
                     Firm
                     HC1              HC2

UWIB                    -0.051 (***)     -0.197 (**)
                       (-3.958)         (-2.152)
IB                      -0.020           -0.128
                       (-1.239)         (-1.448)
Intercept                0.019           -0.239
                        (0.166)         (-0.256)
Control variables    Yes              Yes
Clustering           Affiliation      Affiliation
                     and Time         and Time
Style fixed effects  Yes              Yes
Year fixed effects   Yes              Yes
[R.sup.2]                0.038            0.041
#OBS                 8,128            8,128

Panel B. Fund-Underwriter Clustered Standard Errors

                       Sued Firm Only
                       HC1               HC2

UWIB                       -0.032 (***)      -0.184 (***)
                          (-3.483)          (-3.652)
IB                         -0.004            -0.056
                          (-0.495)          (-1.152)
Intercept                   0.195             1.727
                           (1.372)           (1.431)
Control variables      Yes               Yes
Clustering             Fund              Fund
                       Affiliation       Affiliation
Style fixed effects    Yes               Yes
Year fixed effects     Yes               Yes
[R.sup.2]                   0.037             0.029
#OBS                   13,076            13,076

                       Sued Firm vs. Nonsued
                       Firm
                       HC1               HC2

UWIB                       -0.030 (***)      -0.192 (***)
                          (-3.200)          (-3.822)
IB                         -0.004            -0.062
                          (-0.376)          (-1.193)
Intercept                   0.210             1.317
                           (1.342)           (1.181)
Control variables      Yes               Yes
Clustering             Fund              Fund
                       Affiliation       Affiliation
Style fixed effects    Yes               Yes
Year fixed effects     Yes               Yes
[R.sup.2]                   0.036             0.027
#OBS                   13,076            13,076

Panel C. Trading on Innocent Firms

                      HC1                   HC2

UWIB                     -0.025                -0.204
                        (-1.041)              (-1.570)
IB                       -0.014                -0.119
                        (-1.340)              (-1.494)
Control variables     Yes                   Yes
Clustering            Affiliation and Time  Affiliation and Time
Style fixed effects   Yes                   Yes
Year fixed effects    Yes                   Yes
[R.sup.2]                 0.032                 0.045
#OBS                  5,487                 5,487

Panel D. Interaction Effects in Regressions

                    (1) Centering         (2) Orthogonalized

UWIB                    -0.032 (**)           -0.045 (**)
                       (-2.147)              (-1.986)
IB                      -0.004                -0.007
                       (-0.422)              (-1.054)
Control variables   Yes                   Yes
Clustering          Affiliation and Time  Affiliation and Time
Style dummies       Yes                   Yes
Year dummies        Yes                   Yes
[R.sup.2]                0.037                 0.037
#OBS                13,076                13,076

                    (3) IB Interactions

UWIB                    -0.026 (***)
                       (-2.729)
IB                       0.181
                        (1.259)
Control variables   Yes
Clustering          Affiliation and Time
Style dummies       Yes
Year dummies        Yes
[R.sup.2]                0.037
#OBS                13,076

(***) Significant at the 0.01 level.
(**) Significant at the 0.05 level.
(*) Significant at the 0.10 level.
COPYRIGHT 2019 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2019 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Hwang, Hyoseok (David)
Publication:Financial Management
Article Type:Report
Geographic Code:1USA
Date:Mar 22, 2019
Words:14815
Previous Article:Initial Offer Precision and M&A Outcomes.
Next Article:Identifying and treating outliers in finance.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters |