Long-Term Effects after Departure of
Activist Hedge Fund
Abstract
This article reviews the stock performance after a selloff and departure of an activist hedge fund. It uses a comprehensive dataset with interventions by activist hedge funds in the United States over the period 2007-2014. This paper contributes to literature by making a distinction between selloffs and immediate departures by these funds using amendments of Schedule 13D. The approach used in this paper finds control firms that are insignificantly different from event firms to control for underperforming by hedge fund targets prior to the event. Critics argue that the short-term gains after an intervention comes at the expense of the long-term stock performances. However, the results show that target firms experience a higher positive long-term abnormal return compared to their control firms using different approaches to compute abnormal returns. The results suggest that the myopic claims by opponents of activist hedge funds is rejected. This paper shows that the short-term gains do not come at the expense of long-term performances. The claim that hedge funds pump-and-dump their targets is therefore rejected. The paper provides additional results on what variables predict a selloff or departure of an activist hedge fund. In line with the theory of high financial incentives of hedge fund managers, the results show that a higher cash ratio of the target firm predicts a selloff by an activist hedge fund. While pay-out ratio and Tobin Q show opposite results, indicating that both negatively predict a selloff.
Amsterdam Business School MSc Finance
Corporate Finance Master Thesis Kruijenaar, Michiel July 17
Statement of Originality
This document is written by Student Michiel Kruijenaar who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
Table of Contents
I. Introduction ... 3
II. Literature Review ... 6
A. Institutional Activism ... 6
B. Hedge Fund Activism ... 8
C. Empirical Results on Hedge Fund Activism ... 9
D. Hypotheses ... 12
III. Data Sample and Descriptive Statistics ... 12
A. Intervention by Activist Hedge Fund ... 12
B. Departure by Activist Hedge Fund ... 14
C. Summary Statistics ... 16
IV. Methodology ... 18
V. Results Selloff by Activist Hedge Fund ... 21
A. Logistic Regression Results ... 21
B. Nearest Neighbor Matching Control Sample ... 24
C. Short-Term Abnormal Returns ... 26
D. Long-Term Buy-and-Hold Abnormal Returns ... 27
E. Calendar Time Portfolio Regression ... 28
VI. Results Departure by Activist Hedge Fund ... 29
A. Logistic Regressions Results ... 30
B. Nearest Neighbor Matching Control Sample ... 30
C. Short-Term Abnormal Returns ... 32
D. Long-Term Buy-and-Hold Abnormal Returns ... 33
E. Calendar Time Portfolio Regressions ... 34
VII. Conclusion ... 35
Reference list ... 39
I. Introduction
Commentators claim that activist hedge funds are short-term viewed and believe that they destroy shareholder value on the long-term. However, many researches show positive short-term abnormal stock returns after an intervention and no significant negative abnormal returns over a longer horizon (Brav et al., 2010). The commentators dispute the long-term outcomes of many researches by stating that these post-event windows are too short to evaluate the long-term effect of interventions by activist shareholders (George, 2013; Lipton, 2013). They claim that activist hedge funds’ interest is to make short-term profits and leaving the target firm when beneficial while letting the existing management board to clean up the mess. It is interesting to research whether these claims are true and if indeed activist hedge funds follow this pump-and-dump strategy. Therefore, this paper tries to address the following research question: What are the long-term effects on the stock return of a target company after a selloff and departure by an activist hedge fund in the United States?
This paper uses a sample of activist hedge fund interventions provided by professor Brav from the Duke University School of Law. It captures a sample period of activist hedge fund interventions over the period 2007 till 2014. The sample is constructed by using Schedule 13D filings to identify events by activist hedge funds in the United States, a consistent method used throughout literature (Brav et al., 2008; Greenwood and Schor, 2009; Klein and Zur, 2008). These Schedule 13D filings must be filed at the Security and Exchange Commission (hereafter: SEC) when investors hold a beneficial stake of 5% or more in any publicly traded security. The initial filing dates of these Schedule 13D are used as event dates of interventions by activist hedge funds. Accordingly, this paper identifies departures by hedge funds using a similar method as Bebchuk et al. (2015a). They identify an “exit” by an activist hedge fund as the date whenever the fund files an amendment of Schedule 13D in which the fund discloses to hold an equity stake below this threshold of 5%. However, one could argue that these dates are not real departures by activist hedge fund since in many cases the hedge funds may still hold a beneficial stake in the target company, for instance an equity stake of 4.99%. Therefore, this article makes a distinction between selloffs and immediate departures by activist hedge funds. This article defines the approach of Bebchuk et al. (2015a) as selloffs by activist hedge funds. The selloffs could indicate that the hedge fund is becoming less aggressive and therefore important to research. Using the Schedule 13D/A filings this article constructs a sample of 1018 selloffs over the period 2007-2016. This study contributes to literature by identifying immediate departures by activist hedge funds using an adjusted approach of Bebchuk et al. (2015a). A departure of an activist hedge fund is defined
as the date of the amendment of Schedule 13D in which the activist hedge fund discloses an equity stake of 0.1% or lower in the public firm. This procedure finds 131 immediate departures over the period 2007-2016.
Literature (Brav et al., 2008; Clifford, 2008; Denes et al., 2016) has shown that hedge funds tend to target public firms that are statistically different from their peers. Klein and Zur (2008) for instance show that hedge funds target firms that are underperforming compared to their controls by relating the ROA. Many researches fail to control for this difference. This article uses a comparable approach as Cremers et al. (2016) to find control firms that are similar at the time of the activist selloff and departure. It uses logistic regressions to find variables that predict both a selloff and a departure by an activist hedge fund.
The results show that hedge funds are more likely to selloff their stake when there is more cash in the target firm. This outcome is consistent with the theory of high financial incentives of hedge fund managers to get high investment returns. Since one could see holding cash as a measurement of waste, a hedge fund is more likely to selloff her equity stake whenever the target firm holds a substantial proportion of assets in cash. Similar results are found for Tobin Q and payout ratio. A lower Tobin Q or payout ratio significantly predicts a selloff by an activist hedge fund, which indicates that when an activist hedge fund is displeased with their return of their investment, the fund selloffs her stake. The results also show that when a target firm becomes more financially distressed using the Altman (2000) Z-score, an activist hedge fund is more likely to selloff her stake.
Accordingly, this article uses these significant variables like ROA, Tobin Q, LnSize, Z-score, leverage and fiscal year to find matching firms using the procedure of Abadie and Imbens (2006) which is called the Nearest Neighbor Matching approach. The firms are matched with an exact match on fiscal year to control for time varying in stock returns. Nearest Neighbor Matching allows to match on exact variables which is in contrasts of Propensity Score Matching. When finding control firms with PSM, one could match an event firm in 2007 with a control firm in 2012, since PSM gives certain weights to each matching variable. This might give biased results since stock market returns have been highly volatile over the sample period. The control firms that are found, did not experience an intervention by an activist hedge fund and thus no selloff or departure but are insignificantly different from event firms when comparing different firm fundamentals, e.g. ROA, payout ratio, leverage ratio and size. This is an important condition since as mentioned before many researches fail to control for the underperforming of target firms.
The results show that firms which experienced an activist hedge fund selloff, have higher buy-and-hold returns compared to their controls. The outcomes show an abnormal return of 6.7% over the 12-month window after a selloff and significant at the 5% level. The abnormal return experience an increase to 12.3% over the 24-month window and the result is highly significant at the 1% level. Additional robustness tests show similar results. The tests use calendar time portfolio regressions, where it creates two portfolios. One portfolio consists of firms that experienced a selloff by an activist hedge fund and the other portfolio is constructed with their control firms. The calendar time portfolio regressions show that the portfolio that consist of firms that experience selloffs by activist hedge funds, have a positive and highly significant abnormal return over the 12-month and 24-month event window. Since existing literature has shown that these results might be driven towards the underperforming of the target firms, it is critical to compare these outcomes to the control firms. The portfolio regressions of the controls show insignificant abnormal returns for both event windows. The results indicate that after an activist selloff, target firms do not experience a decline in stock performance. These outcomes suggest that the claim that short-term gains come at the expense of long-term gains is rejected.
Using the subsample of firms that experience an immediate departure of an activist hedge funds, this article finds consistent results as above. As mentioned before this article identifies immediate departures when the disclosed equity stake is 0.1% or lower. The results initially report insignificant abnormal returns over the 12-month and 24-month windows after this departure. However, the calendar time portfolio regressions show that the portfolio which consist of firms that experience this departure, have positive and highly significant abnormal returns. This is in contrast of the portfolio with control firms, which show insignificant abnormal returns. These results again suggest that target firms do not suffer from this departure of an activist hedge fund and reject a pump-and-pump strategy by the activist hedge funds.
This paper contributes to literature in a couple of manners. At first, literature is comprehensive about the interventions of activist hedge funds, but barely researches the time after the departure of these hedge funds and what firms’ fundamentals predict this selloff or departure. This research focusses on the departure and selloff by activist hedge funds since critics argue that hedge fund leave when short-term gains are maximized at the costs of long-term performances. Secondly, as mentioned before this article makes a distinction between selloffs and immediate departures by activist hedge funds, an approach which other researches have been unable to do. Thirdly, this article finds control firms that are similar in
terms of firms’ fundamentals which allows this paper to control for underperforming by hedge fund targets. Researchers argue that underperforming of hedge fund target prior to the event might result into abnormal returns, which this article tries to control for. At last, it uses the most recent and comprehensive dataset which can be used in this research regarding the long-term stock performance after a departure or selloff.
The paper proceeds as follows, section II will document the main literature in this topic and accordingly derive the hypotheses for this article. Section III will proceed to describe the data sample and summary statistics that is used to answer the hypotheses. Section IV states the methodology that will be used for testing. Section V will report the results when using the initial sample of selloffs by activist hedge funds. Accordingly, section VI will provide additional analyses on the subsample with only immediate departures of activist hedge funds. At last, section VII concludes the article and will provide recommendations for further research.
II. Literature Review
This section will start with a small overview of agency problems that arise in public companies and how displeased institutional investors use activism to incorporate changes at the target company. Secondly, it compares institutional investors with hedge funds and why hedge funds might be more effective monitors of publicly traded companies. Thirdly, it overviews some empirical researches that has been done in this topic and some criticism on these researches. At last, this section will document the hypotheses that are tested in this article.
A. Institutional Activism
In corporate finance, the separation of ownership and control in public firms raises the possibility of agency conflicts between managers and shareholders (Jensen and Meckling, 1976). To overcome this problem, shareholders can monitor firm’s management and hence increase shareholder value. Grossman and Hart (1980) state that this monitoring is costly for individual investors, since the benefits are enjoyed by all shareholders. According to Shleifer and Vishny (1986), large shareholders could be effective monitors of public companies and increase shareholder value. They state that due to their increased equity stake, large shareholders are willingly to bear the costs of monitoring since the benefits out weight the costs of monitoring and therefore overcoming the free-rider problem as described by Grossman and Hart (1980).
McCahery et al. (2016) report that institutional investors employ both “voice” and “exit” to discipline management and trying to increase shareholder value. Gillan and Starks (2007) document this evolution of shareholder activism in the United States. They describe varieties of how shareholders actively trying to incorporate changes at the target firm. At one end, dissatisfied shareholders can choose to sell their shares, which is known as the “Wall Street Walk”. By selling their stake, shareholders are signaling that they are unpleased with the current control and that the market may want to incorporate changes. This lowest form of activism using “exit” as instrument to change current control has been empirically proven by Parrino, Sias and Starks (2003). They show that the probability of CEO-turnover is increased after a large selloff by institutional investors. They don’t report results of the effect of “exit” on firm or stock performance. At the other end, Gillan and Starks (2007) state that investors can start takeovers and LBO’s to get full control over the target firm. The authors describe that there is a continuum between both ends, that for instance include investors to buy minority stakes to influence managerial decision-making and trying to increase shareholder value.
However, many researches show different outcomes on the theory of Shleifer and Vishny (1986), who argue that large shareholders might be more effective in increasing firm value. In a survey done by Denes et al. (2016), 73 researches regarding the effects of shareholder activism on target firm value are overviewed. The authors show that institutional investors are not effective in influencing the target’s firm value. For instance, Smith (1996) finds that institutional activists are successful in changing governance structure of the target firm and as a result increase shareholder value. However, he finds that when activists pursue a campaign that tries to improve operating performance, the results are statistically insignificant. Karpoff (2001) and Romano (2001) document findings which show that activists are able to implement small changes in the target firm’s structure, yet they state little evidence that activism by institutional investors increases firm value or operating performances. Wahal (1996) studies the target’s firm value after pension fund activism. He shows that pension funds were ineffective in increasing the long-term firm value and neither increase the firm’s operating performances. Brav et al. (2008) state that institutional investor activism has been plagued by regulatory and structural barriers and document that institutional investors are less effective in monitoring target firm’s management. As will explained in the next section, hedge funds suffer less from these regulatory barriers.
B. Hedge Fund Activism
More recent studies (e.g. Brav et al., 2008; Greenwood and Schor, 2009; Partnoy and Thomas, 2006) argue that hedge funds might be more effective monitors and increase shareholder value due to better alignment of the incentives of hedge fund managers with their shareholders. Hedge funds are different in some ways compared to institutional investors, like mutual funds and pension funds. Partnoy and Thomas (2006) define a hedge fund based on four characteristics: (1) they are pooled, privately organized investment vehicles; (2) they are administered by professional investment managers with performance-based compensation and substantial investments in the fund; (3) they are not widely available to the public; and (4) they operate outside of securities regulation and registration requirements.
Brav et al. (2008) describe a typical hedge fund as a partnership entity that is managed by a general partner. The investors in the entity are limited partners who are passive and have no say in the hedge fund’s investments. Hedge funds face fewer agency conflicts compared to institutional investors based on a couple of characteristics. At first, hedge fund managers’ interests are better alignment with of those investors due to strong financial incentives. Hedge fund managers receive a performance based pay which is usually around 20% of the fund’s annual return on top of the fixed management fees (Brav et al., 2010; Clifford, 2008). Subsequently, hedge funds managers are allowed to hold substantial positions in the fund, which is not allowed for mutual funds by the Investment Company Act of 1940. This compensation structure gives strong financial incentives to hedge fund managers for high investment returns, which limits the agency problem between shareholders and management.
Secondly, hedge funds are lightly regulated since they are only available to a limited number of wealthy investors. This allows them to operate outside the securities regulations and registrations. For instance, hedge funds are not required to hold a diversified portfolio, which is in contrast of some institutional investors. Without this requirement, hedge funds can take larger stakes in less target firms. While holding a larger stake in a target company, a hedge fund manager can more easily influence target’s management to increase shareholder value. Another tool that hedge funds own that comes from a non-diversified portfolio, is explained by Clifford (2008). He argues that due to these non-diversified portfolios, hedge funds can threat to buy-out the company if the fund is displeased with the governance and operating changes made by the target’s management board. Unlike most institutional investors, who cannot buy-out an entire firm since this would affect their diversified portfolio. Clifford (2008) states that hedge funds barely go through with this threat to buy-out
the firm, but the threat itself may be sufficient to incorporate changes by the target’s management.
Thirdly, hedge funds are allowed to “lock-up” capital of their investors for a substantial period of time. In contrast, mutual funds are generally required to sell securities in one day to pay out capital to satisfied investors. Brav et al. (2008) state that this lock-up period creates greater flexibility to hedge funds, without the concern of paying out some of the fund’s capital. Clifford (2008) argues that this lock-up period is required for funds to hold large and illiquid blocks to engage in successful activists’ campaigns. These liquidity concerns are costly for mutual funds, which are less present for hedge funds (Bhide, 1993). Aragon (2007) shows that hedge funds who have higher lock-up restrictions have a 4% to 7% higher excess return. He also finds that higher lock-up restrictions are negatively correlated with the liquidity of the hedge fund’s portfolio. The author concludes that these restrictions allow hedge funds to engage in more illiquid portfolio’s and benefit from this through an illiquidity premium.
C. Empirical Results on Hedge Fund Activism
There has been quite some literature already about the effects of interventions by activist hedge funds on shareholder value and firm’s operating performances. Brav et al. (2010) and Denes et al. (2016) review numerous of researches that have been done in this topic. This study will briefly summarize some of the main findings regarding stock and operating performances of the target company after an intervention.
Greenwood and Schor (2009) construct a dataset using SEC filings from 1993 to 2006. They identify different campaigns of activist hedge funds and whether these campaigns result in higher abnormal returns, e.g. changing capital structure, corporate governance issues, asset sale and block merger. They report a 3.61% abnormal return around the intervention and a 10.26% abnormal return over the 18 months after the intervention. These abnormal returns are the highest when the activist campaign is related to an asset sale or to block a merger.
Clifford (2008) uses a sample over the period from 1998 till 2005 to find interventions by passive and active hedge funds. He makes use of Schedule 13D and Schedule 13G filings to divide hedge funds into active and passive funds. His results show an excess return surrounding the announcement of 3.39% when the investor is an activist hedge fund, compared to ‘only’ 1.64% when the investor is a passive hedge fund. He concludes that activism by hedge funds is creating shareholder value since the difference is statistically
significant. Clifford (2008) states that target firms experience an increase in ROA of 1.22% in the year after an intervention of activist hedge fund. The author shows additional evidence on increases in leverage and dividend payout in the year following the intervention.
Brav et al. (2008) find abnormal returns of 15.9% in 2001 declining to 3.4% in 2006 around the event of an intervention of activist hedge funds. Accordingly, they report different target firm performances one year before and one year after the intervention. They report a significant decrease in CEO contracted pay by comparing the CEO pay of the year of the intervention and one year thereafter. The CEO turnover rate increases by 9.90% comparing one year before the intervention and one year thereafter.
Klein and Zur (2008) find that targets of activist hedge funds earn an 10.2% abnormal return around the announcement date of an intervention. They also document that targets earn an additional 11.4% abnormal stock return during the following year. Accordingly, the authors report evidence that target companies decrease their cash balances, while increasing their leverage and dividend yields in the one-year after the intervention. The authors argue that a suggested motive for hedge fund activism is that hedge funds’ goal is to create short-term gains by reducing the target company’s cash reserves while increasing leverage and dividend payout.
Becht et al. (2015) are among the firsts to investigate hedge fund activism internationally. They use a comprehensive dataset which covers 23 countries in Asia, Europe and North America over the period 2000-2010. They however argue that out of the 1740 interventions by activist hedge funds, a large proportion with 1125 interventions comes from the U.S., which indicates that hedge fund activism is more widely used in the U.S. compared to other countries. They report announcement returns for North America, Europe and Asia which are 7.0%, 4.8% and 6.4%, respectively. The review done by Denes et al. (2016) conclude based on these different researches that all articles show consistent findings. The results show positive and significant abnormal returns around the announcement and additional positive abnormal returns over a long-term horizon thereafter. They conclude that hedge fund activism increases shareholder value.
However, there has still been quite some criticism that dispute the outcomes of these researchers, who state that activism by hedge funds is value-creating. Critics like Martin Lipton (2013), who is the founding partner of law firm Wachtell, Lipton, Rosen & Katz argues that the time periods after the intervention are too short to make conclusions about hedge fund activism. He asks the question; what are the effects of an intervention by an activist hedge fund after 24 months? A similar claim by Harvard Business School professor
William George (2013) and former CEO of S&P 500 component Medtronic, states that activists cloak their demands by saying that they want to increase long-term value, while their real goal is to make short-term profits. He argues that activists bail out when beneficial and leave the company’s management to clean up the mess.
As a response on these “myopic” claims, Bebchuk et al. (2015a) investigate the stock return and operating performance over a period of 5 years after an intervention of an activist hedge fund. Accordingly, they measure abnormal stock returns after the departure of activist hedge funds over a 3-year period. Using an updated data sample of Brav et al. (2008) with a sample period from 1994 to 2007, they don’t report evidence for this myopic claim. The results show positive abnormal buy-and-hold returns over the 5-year period after the intervention. They additional test the abnormal returns after a departure of an activist hedge fund. The authors report no significant negative abnormal returns after the departure of activist hedge fund. They state no evidence that the short-term improvements in performances come at the expense of the long-term performance. During the 5-year intervention period they find no significant declines in operating performances. The claim that activist hedge funds pursue a pump-and-dump campaign is therefore rejected.
However, different researchers (Bebchuk et al., 2015a; Brav et al., 2008; Clifford, 2008; Cremers et al., 2016) have shown that activist hedge funds target underperforming firms relative to their industry-peers. Cremers et al. (2016) state that since target firms were underperforming, it raises the possibility that target firms may experience an increase in firm value attributable to market mechanisms other than the activism by hedge funds. They mention different mechanisms that could contribute to turn things around at underperforming companies, e.g. key employees, executive management, long-term shareholders and/or other stakeholders. The authors criticize the outcomes of Bebchuk et al. (2015a) since they don’t control for this underperforming. As a results Cremers et al. (2016) use different matching approaches to find underperforming firms that were not targeted by activist hedge funds. Accordingly, they compare different characteristics of the control and target firms, like stock return, Tobin Q and ROA. Cremers et al. (2016) find that the control firms on average improved more than the activist hedge fund targets, by looking at the Tobin Q as measurement of firm performance. Cremers et al. (2016) conclude that other market mechanism thus better improves poorly performing firms than activist hedge funds do. Bebchuk et al. (2015b) have some comments on the paper by Cremers et al. (2016), one is that the paper of Cremer et al. (2016) doesn’t research the impact on stock returns after a departure using the matching method.
D. Hypotheses
Based on these empirical studies and theories, this article is interested in the following hypotheses. At first, after a departure and selloff by an activist hedge fund there is a significant decrease in stock price around the announcement of the event compared to control firms. This would indicate that the market beliefs that hedge fund activism increases firm value, contradicting the “myopic activist claim”. Since the share price should reflect the future profitability of the company, a decrease in stock price would indicate that the firm is less profitable after a departure of activist hedge fund.
Secondly, this paper is interested in the long-term buy-and-hold return after a departure of activist hedge fund. As stated by William George (2013), hedge fund bails out when beneficial and leave management to clean up the mess. This is consistent with the theory that hedge fund managers have high financial incentives and leave the target company when most beneficial. This means that following the myopic claim, one would expect significant decreases in target firm’s abnormal stock return, compared to matched firms that did not experience hedge fund activism. The negative abnormal returns would suggest a pump-and-dump strategy.
III. Data Sample and Descriptive Statistics
This section will give information about the sample used in this article. It will start documenting how the sample of initial interventions of activist hedge funds is constructed. Since this research is interested in the departure of the activist hedge funds, it is essential to first find interventions by activist hedge funds. Accordingly, this section will provide information about how a departure of an activist hedge fund is defined. At last, it will provide summary and descriptive statistics of the target firm at the time of the intervention and time of the selloff and departure.
A. Intervention by Activist Hedge Fund
Literature is quite consistent about defining an intervention of an activist hedge fund. Different researchers (Bebchuk et al., 2015a; Brav et al. 2008; Clifford, 2008; Greenwood and Schor, 2009; Klein and Zur, 2008) use Schedule 13D filings done by hedge funds to determine whether the hedge funds follow an activist agenda or hold a passive investment. The SEC mandates investors to file a Schedule 13D or Schedule 13G when the investors acquire over 5% of any publicly traded security. When the investor intends to be an active investor it must file a Schedule 13D. If not, and the investor holds a passive stake in the
public firm, it must file a Schedule 13G. Literature uses the dates of these filings as event dates of interventions by activist hedge funds.
As explained in the previous section, hedge funds operate outside the securities regulation and registration since they are not entirely available to the public. However, hedge funds are required to file Schedule 13D and Schedule 13G when taking a position above 5% of any publicly traded security, similar like other institutional investors. This is an important condition since else there could exists non-randomly in the sample if hedge funds were able to choose to file Schedule 13D at the SEC. However, it is still hard to find all events in the population since no database keeps track of Schedule 13D fillings done by hedge funds. Thomson One and the Edgar Live databases give information on the filings, but only when searching by individual companies or individual hedge funds. This allows to find events based on well-known hedge funds and therefore presumably biased, since one could argue that hedge funds which are less well-known are probably less successful. This would lead to biasness towards more successful hedge fund activism.
To overcome this problem, I have had correspondence with professor Brav from the Duke University School of Law. Professor Brav keeps track of interventions by activist hedge funds in the United States, together with his colleagues Jiang and Kim. In their article, Brav et al. (2008) first make use of this dataset that initially captures a sample period of 2001-2006. The authors state the method how they constructed a comprehensive dataset of 882 unique target firms. They use a top-down approach by starting with all 11,602 Schedule 13D filings over the period 2001-2006. They were able to manually filter out hedge funds from other investors by making use of Item 2 of these Schedule 13D filings. In Item 2, the filer must disclose her identification and background. Together with internet, news searches and phone calls, the authors were able to divide hedge funds from non-hedge funds. The Schedule 13D filing dates that have been filed by hedge funds are presumed to be the intervention dates of the activist hedge funds. Professor Brav have send me an updated and revised version of this dataset. It captures the period from 1994 till 2014 using the method as described above. It describes events of over 4,000 activist hedge fund interventions in the United States by making use of Schedule 13D. This article focusses on interventions by activist hedge funds that took place in the period 2007 till 2014. Over this period 1786 interventions by activist hedge funds occurred.
B. Departure by Activist Hedge Fund
Accordingly, it is necessary to determine the departures of these activist hedge funds. Bebchuk et al. (2015a) make use of the Schedule 13D/A’s to identify “exits” by activist hedge funds. The SEC mandates investors to file amendments of Schedule 13D when there is a substantial change in the equity stake of the investors. These so-called Schedule 13D/A’s must be filed un till the stake of the investor falls below the 5% threshold. The Thomson One and Edgar databases give information about the filings of these Schedule 13D/A’s. For each intervention by an activist hedge fund, I hand-collected information about the Schedule 13D/A’s. However, one cannot simply use the latest filing of Schedule 13D/A since it not always the case that the latest Schedule 13D/A reports the drop below the threshold of 5%. At first, a hedge fund might change her intentions towards the investment in the target company and becomes a passive investor. As a results these hedge funds must file a Schedule 13G, where it explains to become a passive investor. In this case, the latest Schedule 13D/A filing, is above the 5% threshold and does not capture the departure of an activist hedge fund. So, these events are excluded.
Secondly, in some events the target company terminates their security registration for different reasons, e.g. a takeover, merger, taking private, or bankruptcy. This means that while the hedge fund still had a beneficial stake in the firm, there are no more filings of Schedule 13D/A. Again, the latest amendment might state that the stake was still above the threshold, so these filings must be excluded. Brav et al. (2015) report that in their initial sample, 26% of the target firms are delisted within 2 years after the intervention for these reasons. At last, in many cases the hedge fund still holds the target firm in her portfolio after 2016. This indicates that there has not been a departure of the activist fund which results in a lower amount of departures compared to interventions.
Another reason that one could argue is that hedge funds which drop below the 5% threshold, may still hold a substantial stake in the target firm, for instance 4.99%. This would indicate that the hedge fund hasn’t exit the target firm but may hold a less aggressive investment towards the target firm. Therefore, this article makes a distinction between a selloff by an activist hedge fund and an immediate departure of a hedge fund. The amendments of Schedule 13D require to disclose the latest stake that an activist hedge fund holds when it drops below the 5% threshold. I was able to manually filter out the cases where the filing discloses an equity stake of below 0.1%, which would indicate an immediate departure by an activist hedge fund.
This method resulted in 1018 selloffs including immediate departures of activist hedge funds over the period 2007-2016. The average (median) holding period of activist hedge funds was 574 (398) days. This is somewhat lower than Bebchuk et al. (2015a) report with 811 days as average and 539 days as median. This is probably due to the difference in sample period. Since Bebchuk et al. (2015a) use a sample set of 1994 till 2007 to find interventions, this article focusses on the period of 2007-2014. That might explain why the median and average of Bebchuk et al. (2015a) is substantially different due to some events with longer holding periods.
Table I
Descriptive Statistics
This table reports the descriptive statistics for the sample of activist hedge fund targets. Panel A summarizes the number of activist interventions per year using the Schedule 13D filings, provided by professor Brav from the Duke University School of Law. Panel B summarizes the number of selloffs including the immediate departures per year using the amendments of the initial filing of Schedule 13D, when the percentage of ownership stake of the activist hedge fund falls below the 5% threshold. Panel C reports immediate departures of activist hedge funds, indicating that the amendments report a stake below 0.1%. The filings are gathered using the Thomson One and Edgar Live database.
Hedge Fund Activist Targets
Hedge Fund Activist Targets Panel A: Year of First 13D Filing
2007 378 2011 185
2008 303 2012 184
2009 151 2013 192
2010 184 2014 209
2007-2010 1016 2011-2014 770
Panel B: Number of Selloffs by Year
2007 46 2012 95 2008 147 2013 126 2009 122 2014 103 2010 117 2015 91 2011 101 2016 70 2007-2011 533 2012-2016 485
Panel C: Number of Departures by Year
2007 2 2012 12 2008 18 2013 17 2009 21 2014 11 2010 13 2015 13 2011 14 2016 10 2007-2011 68 2012-2016 63
Table I shows the number of events per year for the interventions, selloffs and immediate departures of activist hedge funds in Panel A, Panel B and Panel C, respectively. The table shows a lower number of selloffs and departures in the years 2007, 2015 and 2016. This can be explained since this dataset starts following interventions from 2007-2014. This suggests that only a few firms will sell off their equity stake in a target firm within one year, therefore the lower amount of selloff and departures in 2007. Since after 2014 no more interventions are added, it is expected that there will be less selloffs after 2014, which explains the somewhat lower amounts in 2015 and 2016. The table shows a lower amount of interventions than selloffs, this can be explained since some of the events had to be excluded due to change in investment intention, attrition of the target company or since the activist hedge fund still holds an equity stake above the 5% in the target firm after 2016.
C. Summary Statistics
Subsequently, the target firm’s annual fundamentals were collected using the Compustat database and the stock returns are gathered using the CRSP database. Table II reports information about the summary statistics of these target firm’s fundamentals. In columns (1) through (4) it shows different firm characteristics of the fiscal year prior to the intervention of the activist hedge fund, e.g. LnSize, market-to-book ratio, ROA, Tobin Q, sales growth, leverage, cash, payout ratio, KZ-score and Altman (2000) Z-score. Column (5) through (8) reports the characteristics of the target company using the fiscal year’s fundamentals in which the activists hedge fund selloffs their stake in the target company. Accordingly, in columns (9) and (10) it compares the averages of column (2) and (6) and test on significance in difference between both using a t-test. The description of different variables is briefly overviewed in appendix table A1.
What is remarkable about table II is that there is a quite substantial difference in sample size, when comparing column (1) and column (5). In case of the firm’s characteristics at time of the selloff, there is about half of the firms left that initially reported fundamentals at the time of the intervention, this is due to the difference reasons that are mentioned above, like the survivorship of the target company, hedge fund intentions towards the target company and in some events in which the hedge fund still holds an equity stake above 5%.
Table II shows an average (median) LnSize of 6.05 (5.93) million dollars at the time of the activist intervention. This is similar as the results that Klein and Zur (2008) report. They report a mean of 6.85 and a median of 5.33 million dollars. Table II shows an average
Table II
Characteristics of Target Companies
This table shows the characteristics of companies that are targeted by activist hedge funds. The first 4 columns report count, mean, median and standard deviation of the characteristics of the target company of the fiscal year ending prior to the intervention. Columns 5 through 8 report the count, mean, median and standard deviation of the firm characteristics, using the fiscal year in which the activist hedge funds sells off or departures the target firm. Columns 9 and 10 report the difference in average between each firm characteristics and a t-test to test the significance of the differences. The variables are explained as; LnSize is the natural logarithm of total assets; Mv-to-Bv is defined as the ratio of market value of equity to book value of equity; Tobin Q is defined as the ratio of the market value of total assets to book value of total assets; Growth reports the sales growth; ROA is defined as the return on assets using EBITDA/Total Assets; Leverage is defined as the ratio of book value of total debt to the book value of assets; Cash is defined as the ratio of the cash to the book value of assets; CAPEX is defined as the ratio of capital expenditures to the book value of assets; Payout is the payout ratio as total dividends paid to EBITDA; Kscore is the score that the target company gets using the Kaplan and Zingales (1997) index, it measures the degree of financial constraint; Z-score is defined as the Z-score using the Altman Z-Z-score, which measures the degree of financial distress. All variables with exception of LnSize are winsorized at the 2% and 98% cuts. The variables are gathered using the Compustat database.
Activist Hedge Fund Intervention Activist Hedge Fund Selloff Difference
Firm Characteristic Count (1) Mean (2) Median (3) SD (4) Count (5) Mean (6) Median (7) SD (8) Avg. Diff. (9) T-test (10) LnSize 1500 6.05 5.93 1.78 686 6.38 6.32 1.82 0.33 3.98 Mv-to-Bv 1493 2.04 1.46 2.48 685 2.27 1.35 3.27 0.24 1.64 Tobin Q 1500 1.57 1.28 0.94 686 1.63 1.27 1.07 0.06 1.20 Growth 1451 0.08 0.03 0.30 667 0.03 0.01 0.29 -0.05 -3.32 ROA 1497 0.03 0.07 0.18 684 0.04 0.08 0.19 0.01 1.37 Leverage 1491 0.21 0.14 0.23 684 0.25 0.17 0.25 0.04 3.49 Cash 1492 0.15 0.09 0.16 683 0.16 0.10 0.17 0.01 1.46 CAPEX 1496 0.04 0.02 0.05 679 0.04 0.02 0.05 -0.01 -2.49 Payout 1496 0.06 0.00 0.18 679 0.08 0.00 0.20 0.02 2.37 KZ-Score 971 -1.68 0.17 5.94 549 -2.27 0.30 7.24 -0.59 -1.62 Z-Score 1273 2.83 2.46 3.77 568 2.57 2.22 3.65 -0.25 -1.36
value of 1.57 for Tobin Q, while the median is 1.28. These outcomes are similar to Brav et al. (2008) who report a value of 1.54 and 1.24 for the mean and median, respectively. Klein and Zur (2008) report that hedge funds target companies that are financially healthy. They measure this using the Altman (2000) Z-score variable. Table II shows two measurements for financial health, the KZ-score and Altman Z-score. KZ-score is a measurement for financially constraint while the Z-score measures financial distress at target firms. Table II reports at the time of the intervention that target firms have on average (median) a KZ-score of -1.68 (0.17). Which suggests that target firms might be financial constraint. The Z-score reports a value of 2.83 for the mean and a value of 2.46 for the median, indicating that target firm don’t experience financial distress. These outcomes are consistent with the results of Klein and Zur (2008).
The table shows an average (median) payout ratio of 0.06 (0.00) at the time of an activist intervention. The sales growth is on average 8% and reports a median of 3%. Capital expenditures are 4% of the total asset on average and the median reports a value of 2%. Which are similar as the 3.2% and 2.3% that Klein and Zur (2008) report for the mean and median, respectively. Looking at the ratio of cash to total asset, table II shows a mean of 15% and median of 9%. Klein and Zur (2008) and Brav et al. (2008) report similar averages of 14% and 18%, respectively.
The results in columns (9) and (10) show that firms at the time of the departure are on average significantly larger, experience lower sales growth, an increase in leverage, decrease in capital expenditures and an increase in payout ratio compared at the time of the intervention. Some of these changes might indicate that indeed hedge fund activism pump up the target companies by increasing leverage and payout ratio, while lowering investment regarding capital expenditures and a decrease in sales growth. However, these outcomes could also indicate that hedge fund are able to increase efficiency at the firms and increase performance. It is therefore interesting to look at the shareholder value over the long-term after this departure.
IV. Methodology
This section describes the methodology that will be used for testing the hypotheses. At first, the section will document the approach to test which variables predict a selloff by an activist hedge fund. Secondly, the technique for finding control firms is explained. At last, this
section will state the approach to test whether selloffs by activist hedge funds causes abnormal returns for the target company.
To find variables that predict a selloff by an activist hedge fund, a logistic function will be used where the dependent variable is a dummy named !"##$%%&', which equals 1 when there is a selloff by an activist hedge fund in the target firm and 0 otherwise. Subsequently, the dependent variable is regressed on different firm characteristics from the latest annual fundamentals prior to the event. Equation (1) shows a baseline regression that will be used in the logistic regressions.
!"##$%%&' = ) + +,-./&'0,+ +23$4567&'0,+ +896!5:"&'0,+ +;<=>$?=&'0, 1 + +A9"="BCD"&'0,+ +EFGH$B"&'0,+ +IJFGH$B"&'0,+ +KLB$M>ℎ&'0,
+ +OPCGℎ&'0,+ +,QRCS$T>&'0,+ +,,PCU"V&'0,+ W&
The variables that are significant from equation (1) are used to find control firms from the entire Compustat population that did not experience an activism event by hedge funds. Using these variables this article creates a control sample with firms that are similar to the hedge fund targets, but were not (yet) an event firm. As explained in section II, literature shows that hedge fund activists tend to target underperforming companies. When matching firms on items like Tobin Q and/or ROA that are significant in predicting a hedge fund departure, the control sample is similar in terms of operating performance. Consistent with Klein and Zur (2008), table II of this article show that hedge fund activist target firms that don’t experience financial distress, but might be financially constrained. When these variables are significant predicting selloffs, then not matching on these variables may lead to differences compared to matched firms. Therefore, this article will use the Nearest Neighbor Matching approach of Abadie and Imbens (2006) to find control companies using different significant variables from equation (1). The Nearest Neighbor Matching procedure is provided as a tool in Stata. The matching approach finds for each treatment firm (selloff by activist hedge fund) at least one control firm that is the best control based on the significant variables. This method is similar as the article of Cremers et al. (2016). They find significant variables that predict an intervention of an activist hedge fund and use these variables to find control firms with the Nearest Neighbor Matching approach, e.g. Tobin Q, LnSize, leverage, ROA and fiscal year.
This procedure allows this article to find the treatment effect of a selloff by an activist hedge fund. As explained in section II, critics of activist hedge funds argue that these funds bail out when most beneficial for the fund. George William (2013) states that hedge funds
leave the company when most beneficial, while existing management has to clean up the mess. This research uses these matched firms to calculate the abnormal returns around the selloff of the activist hedge fund, using the CAAR-method. It compares the returns around the event date of the treatment firm and the control firms and tests whether these are significantly different using equation (2).
P//-&' = -&'XYZ[' \&]^− -&'`a[']ab \&]^ ,Q
'c0,Q
(2)
Equation (2) cumulates the daily average abnormal returns over the [-10, 10] period around the selloff and takes the average over all target firms. The efficient market theory states that current price should reflect all futures cash flows of the company. So, if there is a negative cumulative abnormal return around the announcement of a selloff by a hedge fund activist, then the market values activism as positive. This article tests whether there is a significant difference from zero between the event firms and control firms, using a t-test.
Accordingly, this article is most interested in the long-term stock performance after a selloff by an activist hedge fund. To test the long-term stock performance, this article compares the two-year buy-and-hold return between the event firms and the control firms.
?d/-&' = BHRXYZ[' h&]^&' − ?d-&'`a[']ab \&]^ (3)
Since the target firms and control firms are similar at the time of the selloff, one can test whether the effect of the selloff influences the return of the target company. If the buy-and-hold abnormal returns are negative and statistically significant, then this indicates that activist hedge fund follow pump-and-dump strategy in which the short-term gains that are showed in previous researches come at an expense of long-term shareholder value.
The above methodology is replicated for the smaller sample with immediate departures of activist hedge funds. As mentioned in section III, a subsample is created in which Schedule 13D/A filings reported an equity stake of 0.1% or lower. For these events a dummy variable is created called i"UCB>TB"&', which equals 1 when there is a departure of an activist hedge fund in the target firm and 0 otherwise. It uses the same baseline regression as equation (1) to find variables that predict a departure of an activist hedge fund. Accordingly, it uses Nearest Neighbor Matching of Abadie and Imbens (2006) to find control
firms and tests whether abnormal returns between the event firms and control firms are significantly different.
V. Results Selloff by Activist Hedge Fund
This section gives information on the results based on the methodology as described in section IV. Section A will report results on the logistic regressions, to find variables that are associated with a selloff by an activist hedge fund. Section B will provide information on the Nearest Neighbor Matching control sample. Accordingly, section C will provide information on the announcement returns of an activist selloff. Section D will provide information on the long-term abnormal buy-and-hold return after a selloff by activist hedge fund. At last, section E will provide robustness tests using calendar-time portfolio regressions.
A. Logistic Regression Results
Table III shows the outcomes on the logistic equation (1) from the previous section. The dependent variable is a dummy variable which equals one when there is a selloff by an activist hedge fund in the target company and zero otherwise. Column (1) and (2) show the logistic results when excluding the KZ-score and Z-score. Both are initially excluded since the summary statistics in table I show that the number of observations of the Z-score and moreover of the KZ-score is substantially lower than all other variables, due to missing values to calculate both. The table in column (3) through (6) shows highly significant levels for coefficients of the Z-score. However, the inclusion leads to significant difference in the payout ratio. When including Z-score in model (3) and (4), the variable payout ratio becomes insignificant while it is significant in model (1) and (2). While in the first columns the coefficient is insignificant, it becomes highly significant in column (3) and (4). Meaning that while controlling for KZ-score, the payout ratio indeed predicts a selloff by an activist hedge fund.
The ROA is in all six regressions positive and highly significant. The coefficient in column (4) is 0.737 when using year fixed effects. Meaning that a higher ROA predicts a selloff by an activist hedge fund. This is consistent with the summary statistics that showed that at the time of the selloff, the target firms have a higher ROA. What is interesting is that the other performance measure, Tobin Q, has an opposite effect of the prediction of a selloff by an activist hedge fund while it is highly significant in all six models. The table shows in column (4) that the coefficient of Tobin Q is -0.187 using year fixed effects. This indicates that a higher Tobin Q leads to a lower probability of a hedge fund selloff. A paper done by
Table III
Outcomes Logistic Models Predicting Sell Off
This table shows the results of the logistic regressions of a selloff by an activist hedge fund in a public firm. The dependent variable is a dummy variable which equals one when there is a sell off of an activist hedge fund in the target company and zero otherwise. All variables are winsorized at the 2% and 98% levels, with exception of LnSize. The definitions of variables are further explained in Appendix table A1. The table shows regression coefficients for each variable while robust standard errors in parentheses below. ***, ** and * indicate statistical significance at the 10%, 5% and 1% levels, respectively.
(1) Selloff Dummy (2) Selloff Dummy (3) Selloff Dummy (4) Selloff Dummy (5) Selloff Dummy (6) Selloff Dummy ROA 0.456** 0.457** 0.762** 0.737** 0.997*** 0.963*** (0.228) (0.229) (0.298) (0.299) (0.337) (0.340) Tobin Q -0.251*** -0.237*** -0.199*** -0.187*** -0.155*** -0.150*** (0.047) (0.046) (0.051) (0.050) (0.053) (0.053) LnSize -0.089*** -0.093*** -0.070*** -0.073*** -0.092*** -0.093*** (0.017) (0.017) (0.019) (0.019) (0.021) (0.021) Mv-to-Bv -0.011 -0.012 -0.010 -0.010 -0.021 -0.022 (0.017) (0.017) (0.016) (0.016) (0.017) (0.017) Leverage 0.778*** 0.782*** 0.472** 0.488** 0.634*** 0.658*** (0.164) (0.164) (0.212) (0.211) (0.233) (0.233) Growth -0.036 -0.043 -0.038 -0.043 -0.055 -0.062 (0.047) (0.048) (0.049) (0.051) (0.055) (0.057) Cash 0.938*** 0.858*** 0.912*** 0.836*** 0.529* 0.484 (0.234) (0.235) (0.253) (0.254) (0.313) (0.314) Payout -0.757*** -0.777*** -0.299 -0.313 -0.725** -0.746** (0.241) (0.241) (0.264) (0.264) (0.309) (0.309) Capex -0.628 -0.649 -1.222 -1.281* -0.926 -0.884 (0.657) (0.657) (0.750) (0.752) (0.834) (0.838) Z-score -0.049*** -0.048*** -0.050*** -0.048*** (0.016) (0.016) (0.017) (0.017) KZ-score -0.011 -0.011 (0.008) (0.008) Intercept -3.220*** -4.120*** -3.181*** -4.042*** -3.104*** -3.870*** (0.134) (0.208) (0.146) (0.224) (0.164) (0.298) N 48240 48240 38588 38588 32213 32213
Reddy et al. (2015) gives an explanation why these performances measures might give opposite effects. They state that the ROA is a measurement of short-term performance of a company, while the Tobin Q measures the long-term operating performances. The results from table III would suggest that a hedge fund selloff is more likely when there is a high short-term performance and less likely when a higher long-term performance. The results might suggest that hedge funds will sell off their stake when the short-term performances are high, since this is most beneficial time to leave the company. While maintaining their stake when they could still benefit from long-term performance. Cremers et al (2016) report similar results with opposite coefficients of Tobin Q and ROA that predict an intervention.
However, the correlation matrix in appendix table 2A shows high correlation between different performances measures. The correlation for instance between Tobin Q and market-to-book ratio is 0.676. Since high correlation could lead to multicollinearity, this article presents robustness logistic regressions in table 3A of the Appendix. The table uses similar regressions like table III but including different combinations of the performances measures. The coefficients of ROA and Tobin Q are very consistent between different regressions and always highly significant. However, the coefficient of the market-to-book ratio becomes insignificant when controlling for Tobin Q, probably due to the high correlation between both. The robustness results in table A3 shows no further concerns about these performances measures.
Table III also shows that the size of the company predicts a selloff by an activist hedge fund. In all six columns, the coefficient is highly significant and negative ranging from -0.070 to -0.093. This is consistent with findings of Klein and Zur (2008) who show that activist hedge funds target public firms that have a smaller asset base. Reasons for this might be that for hedge funds to have substantial influence on target board, it must increase their stake. This is more easily done for firms that have a smaller asset base. Brav et al. (2008) show that indeed activist hedge funds target firms that have lower market value compared to control firms.
The table shows that when there is more cash in the target firm, a hedge fund is more likely to sell off her stake. Column (1) through (4) show highly significant and positive coefficients. One could see cash as a measure of waste, since it could be used for valuing creating investments. Since hedge fund managers are highly incentivized to generate high returns, it is more likely that a hedge fund selloffs her stake when there is a waste of investment in the target company. The table also reports negative coefficients for the pay-out ratio. When hedge fund managers are displeased with the return their investment generates, it
is more likely that they will sell off their stake. Furthermore, the coefficients of the variable Z-score are negative and highly significant. It indicates that a hedge fund selloff is more likely when the firm experiences financial distress. At last the variables KZ-score, growth, capex and market-to-book value are never significant. This indicates that these variables do not predict a selloff by an activist hedge fund.
In appendix table A3 the marginal effects are reported using probit regressions which are similar as the models in table III. The probit regressions are reported as robustness since the coefficients are more easily interpreted as probabilities. Table A3 shows exactly the same significance levels for each individual coefficient throughout all models. However, the coefficients in table A3 display the probability increase of selloff for each variable. For instance, the coefficient of ROA in model (4) is 0.0130 and significant at the 1% level. This indicates that the probability of a selloff by an activist hedge fund is 1.3% higher when the ROA increases by 1. Other variables can be interpreted in a similar way. So, column (4) shows that a one point increase in cash will lead to a 1.49% increase in probability of an activist selloff.
B. Nearest Neighbor Matching Control Sample
As explained in section IV, the highly significant variables of the logistic models will be used to find control firms, using the Nearest Neighbor Matching method from Abadie and Imbens (2006) provided in Stata. This indicates that the control firms will be matched on the variables, ROA, Tobin Q, LnSize, Z-Score, leverage and fiscal year, since these variables are significant throughout all models. This method allows this article to find control firms that are on average similar to event firms, but did not (yet) experience an activist event. The procedure of Nearest Neighbor Matching in Stata allows to find control firms on variables with an exact match on specified variables. This is critical since the undocumented year fixed effects in table III are highly significant for predicting a selloff by an activist hedge fund for each separate year. This article finds control firms with an exact match on fiscal year of the selloff. It is important to match firms based on the fiscal year since the dependent variable in this article, stock return, was presumable highly volatile in the sample period. Matching an event firm in the financial crises to a control firm outside the financial crisis, may cause a difference in stock return to be unconnected to the selloff by an activist hedge fund. The exact match on fiscal year allows to control for time varying in stock returns. An alternative method for matching is Propensity Score Matching, also provided by Stata. However, Propensity Score Matching does not allow to exactly match on a specified variable, like the
month of the event. Therefore, the Nearest Neighbor Matching method is more appropriate to use in this research.
Table IV shows the tests for different firm characteristics, using the matching method as described above. Column (1) shows the average difference between event firms and control firms, respectively 1 versus 0. Column (3) shows the p-value for the significance in difference between both groups. The table shows that there is no difference in size and ROA of the event group and control group. The Z-score of the treatment group is 0.029 lower than the control but the value is insignificant since the p-value is 0.436. Table IV shows similar results for the other matching variables, e.g. leverage and Tobin Q.
It is interesting to see that not only the variables that are used for matching are insignificant different, but also unmatched variables. Additional variables like cash, market-to-book ratio, capital expenditures, KZ-score, sales growth and pay-out are insignificantly different between the event and control group. This indicates that the Nearest Neighbor Matching method found control firms, that are insignificantly different from the event group. This is an important condition for this article since the objective is to find companies that are similar in terms of firm performance but did (not) yet experience an activist event. When the Table IV
Testing Event versus Control Group
This table shows a test between the event group and the control group on different independent variables. The control group is based on the Nearest Neighbor Matching method by Abadie and Imbens (2006) provided in Stata. The control firms are matched on the variables LnSize, ROA, Tobin Q, Leverage and Z-score and exact match on fiscal year. All initial variables were winsorized at the 2% and 98% cuts, with exception of LnSize. Column (1) measures the average difference between the event firms and control firms, respectively dummy variable 1 and 0. Firm fundamentals are gathered from Compustat.
(1) Difference (1 vs 0) (2) Standard deviation (3) P-value LnSize -0.011 0.018 0.562 ROA -0.002 0.002 0.943 Z-score -0.029 0.037 0.436 Leverage 0.001 0.001 0.644 Cash 0.012 0.008 0.138 Tobin Q 0.012 0.012 0.296 Mv-to-Bv 0.123 0.127 0.334 Capex -0.002 0.003 0.595 KZ-score -0.568 0.379 0.134 Growth -0.041 0.045 0.368 Payout 0.001 0.011 0.931
firms are insignificantly different, an outcome in the stock return will be caused by the selloff by the activist hedge fund.
C. Short-Term Abnormal Returns
This section will report the outcomes of the announcement returns after an activist selloff. These results are trying to answer the first hypothesis of section II.D, which states that after a selloff by an activist hedge fund it is expected to find negative abnormal returns. Table V shows two event windows around the announcement. The first window [-1, +1], cumulates the returns from one day prior to the selloff to one day thereafter. The second window cumulates the returns 10 days prior to the announcement of the Schedule 13D/A to 10 days thereafter. The firms that experienced an activist selloff are matched on control groups based on firm fundamentals as described in previous section and with an exact match on the day of the selloff, to control for stock market volatility.
Table V shows for both windows negative abnormal results but the outcomes are insignificant. The difference between control and treated group is -0.4% when cumulating the return over a 3-day window but insignificant. Similar outcome is obtained over the 21-day window around the event. The firms that experienced a selloff have a 2.5% lower return over this period compared to control firms that were similar in terms of firm fundamentals. The hypothesis of section II expected a negative difference between event and control firms. The results indeed show this negative difference however the results are insignificant. The outcomes suggest that there is no evidence for market reaction around a selloff of an activist hedge funds, which results in rejection of the hypothesis in section II.
Table V
Short-Term Cumulative Abnormal Returns
This table shows the cumulative abnormal returns between public firms that experience a selloff by an activist hedge fund and their control firms. The control firms are matched on the firms’ characteristics LnSize, Tobin Q, ROA, leverage and Z-score and additional with an exact match on date of activist selloff. Firms’ annual fundamentals are gathered from Compustat and daily stock performance from CRSP database. ***, ** and * indicate statistical significance at the 10%, 5% and 1% levels, respectively.
CAR Difference (1) Standard deviation (2) P-value (3)
Window [-1, +1] -0.004 0.008 0.622
D. Long-Term Buy-and-Hold Abnormal Returns
To test whether stock performance of the event firms and control firms differ, this article compares the one-year and two-year stock performance of both groups. Table VI shows the outcomes of the method as described in section IV. The buy-and-hold return is constructed with the monthly stock return provided from CRSP and includes the delisting return when the event or control firm got delisted within the time window. The control firms are matched as described in section D on firm fundamentals prior to the selloff like ROA, Tobin Q, leverage, LnSize and Z-score and with an exact match on the month of a selloff by a hedge fund.
The table shows that the stock of firms that experience a selloff by an activist hedge fund, perform significantly better than of those control firms. Table IV showed that at the time of the selloff, control and treated firms are statistically similar. However, Table VI shows that the stock of the firms that experienced a selloff by an activist hedge fund performed significantly better. When comparing the 12-month buy-and-hold return between both groups, the firms that experienced a selloff by an activist hedge fund had a 6.7% higher stock return with a significant level of 5%.
Accordingly, within 24-months after the selloff by an activist hedge fund, the targeted firms performed significantly better by 12.3% compared to their controls. These results are consistent with existing literature, that activist hedge funds indeed tend to turn things around at targeted firms (Denes et al., 2016). Additional to this literature, this article shows that these firms experience a higher improvement even when the activist fund has left, indicating that Table VI
Long-Term Abnormal Buy-and-Hold Returns
This table shows the long-term buy-and-hold returns between public firms that experience a selloff by an activist hedge fund and their control firms. The control firms are matched on the firms’ characteristics LnSize, Tobin Q, ROA, leverage and Z-score, and additional with an exact match on month of the selloff. The buy-and-hold return include delisting returns, in the case the event or control firm got delisted within the time period. Firms’ annual fundamentals are gathered from Compustat and monthly stock performance from CRSP database. ***, ** and * indicate statistical significance at the 10%, 5% and 1% levels, respectively.
BHR Difference (1) Standard deviation (2) P-value (3)
+12 months 0.067** 0.027 0.014
