1
Hedge fund activism in Europe: Target analysis, effects on stock returns and
firm performance
University of Amsterdam, Amsterdam Business School
MSc Finance: Quantitative Finance
Master Thesis
Bob Keijzers
June 2018
2 TABLE OF CONTENTS Abstract 3 1. Introduction 4 2. Literature 9 3. Hypotheses 16 4. Methods
4.1 Target firm characteristics analysis 17
4.2 Announcement effect 18
4.3 Effects on the long run firm performance 20
5. Data/ Sample
5.1 Sample gathering 22
5.2 Classifying the campaigns 23
5.3 Selection bias 29 6. Results 5.1 Target analysis 30 5.2 Announcement effect 33 5.2 Firm performance 36 7. Robustness 39 8. Conclusion 42 9. Discussion 44 10. Literature list 46 11. Appendix 49 Statement of Originality
This document is written by Student Bob Keijzers who declares to take full responsibility for the contents of this document.
I declare that the text and the work presented in this document are 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.
3
Abstract
This study looks at hedge fund activism in Europe, more specifically it looks at what type of firms get targeted, what the announcement stock returns are and lastly how it impacts underlying firm
performance. Hedge fund activism is a relatively new phenomenon in Europe and is underexposed in scientific literature. Using a handpicked sample containing 376 activist campaigns, I find that target firms are undervalued compared to their industry peers in terms of q and have a significantly lower revenue growth in the year prior to the hedge fund engagement. Contrary to in U.S. (Denes et al., 2015), targets tend to be large in terms of market capitalization. The stock market reacts positively to the announcement of activism, targets significantly outperform a European benchmark with 5.88% in the period 20 days before announcement until 20 days after. In comparison, Brav et al. (2008) find 7.2% outperformance in the U.S. I measure firm performance using ROA and q in [-2, +3 years]. Targets significantly underperform relative to their peers in the 2 years prior to the engagement, however in contrary to Bebchuck et al. (2015) I find no evidence of a significant improvement in firm performance in the 3 years after the start of the activist hedge fund campaign.
4
1: Introduction of the research topic
Hedge fund activism is on the rise in Europe. Over the last few years, it has not been uncommon for financial newspapers to announce in their headlines that a European corporate got targeted by an activist hedge fund. Whereas it has a long standing presence in the United States (Denes et al., 2015) and has also been widely studied in the United States, in Europe relatively few research has been conducted about this phenomenon. This study contributes to this area by studying what type of firms are likely to be targeted by hedge funds and what the effects of hedge fund activism on stock returns and firm performance are, using a handpicked European sample.
Activist hedge funds build up equity stakes in publicly listed firms, which they believe are currently undervalued because the firm’s management is not making the right decisions to maximize
shareholder value. The hedge fund interacts with management, shareholders and sometimes the media and propose changes in order to enhance shareholder value and by doing so creating a return on their invested capital.
A recent example that has drawn a lot of media attention is the AkzoNobel case. In March 2017 the company received an takeover offer from American rival PPG Industries. AkzoNobel rejected the offer and also refused to start negotiation talks, saying the bid significantly undervalued the firm’s value and was therefore not in the interest of their shareholders (Reuters, 2017). Hedge Fund Elliott Management did not share the management’s views, they quickly built up in an equity stake in AkzoNobel and started urging management to start talks with PPG. Elliott used multiple tactics to achieve this, but AkzoNobel did not give in and in the end managed to fight PPG’s takeover attempts off. However, Elliott did not back down and tried to oust chairman Antony Burgmans, claiming he did not properly look after the shareholder’s best interest. Although Burgmans was not forced to step down, he did announcement to quit after his third term as chairman would end in April 2018 (Reuters, 2017). In order to satisfy its shareholders AkzoNobel proposed a strategic plan, which included spinning off its Specialty Chemicals division. Elliott is still a large shareholder and continues to interact with the management.
Activist hedge funds come in different shapes and sizes, proposing a wide range of changes to the target firms and use various tactics to make sure the proposed changes are implemented. Their ultimate goal is always simple create shareholder value, a good return on their invested capital.
This study aims to create a deeper understanding of the effects of hedge fund activism in Europe by studying what kind of firms are likely to be targeted and what the effects on stock prices and underlying firm performance are. A handpicked sample consisting of 376 campaigns in West
5 and market variables and are benchmarked against a peer group, based on industry and geography. A Probit analysis will reveal which variables significantly influence the probability of being targeted by an activist hedge fund. The effect on short term stock returns is measured by the cumulative
announcement return in the multiple periods ranging from [-2, +2 days] to [-30, +30 days], later in this introduction section the conducted tests are explained in more detail. Changes in firm
performance (Return on assets and q, both the raw values and industry-adjusted values) are
investigated in the period 2 years prior to the hedge fund engagement until 3 years after. Outcomes are closely compared to studies focusing on the United States, such as Klein and Zur (2006), Brav (2008) and Clifford (2008).
There are some crucial differences in corporate governance frameworks between (continental) Europe and the United States. Traditionally, the U.S. model focuses strongly on shareholder wealth
maximalization, while the European model stakeholder model gives importance to all stakeholders. As a result the (continental) European market was traditionally seen to be less suited for hedge fund activism as a result of the stakeholder model. Consequently, shareholders have less power/rights and other stakeholders are better protected, thus it may be harder for shareholders to enforce changes in the targeted firms. Inequalities in corporate governance frameworks may result in differences in outcomes of hedge fund activism between Europe and the U.S. and also differences between European countries.
As mentioned by Katelouzou (2013) research focused on the U.S. market has few added value, since a lot of research has already been conducted in that area. The contrary is true for the European market. Previous research only focused on one particular country, such as Belcredi and Enriques (2014) and Bessière, Kaestner and Lafont (2011) or is limited to the action of one particular hedge fund Becht, Franks, Mayer and Rossi (2008). Therefore a more comprehensive study is needed to get a good understanding of the market.
Brav et al. (2008), Denes et al. (2015) and Lazard (2017) point to small and medium sized firms with large institutional ownership and lower dividend payout as typical hedge fund targets in the United States. I will look out accounting variables (ROA, Sales growth, Leverage, Cash/Assets,
Capex/Assets) and market variables (Q, Buy-and-hold return, Market to book ratio, Market capitalization) to find out what type of European firms are likely to be targeted by activist hedge funds. The targeted firms are compared to peer groups, that are formed based on industry and geography criteria. It is investigated whether the means and medians of the targeted firms are significantly different from the peer group in the above mentioned variables. Thereafter a Probit analysis, using a dummy variable that equals one if the target got targeted by an activist hedge fund, is conducted to find out which of the above mentioned variables significantly influence the probability of a firm being targeted. The results of the Probit analysis in combination with the target firm
6 characteristics, give a clear view of the type of firms that are likely to be targeted. The outcomes are compared with prior studies that focus on the U.S.
When the activist hedge fund campaign is announced, either by submitting a regulatory filing or public announcement by the hedge fund itself or other media, the market will react on the news. This market reaction is measured by the cumulative announcement return (CAR). This is an important proxy of how the market regards and values hedge fund activism.
It is calculated by summing up the stock’s daily return in excess of the broader market: CARt = Σ ∗ dARτ with ARi, t = Ri, t − (βi × RMt).
Where Ri,t is the return of the target firm’s stock and RM,t is the market return. Where ERi,t is Expected Return for stock i. The expected return is obtained using a market model (CAPM) with a European stock index as benchmark.
The CAR is calculated in the following time intervals: 2, +2 days], 10, -1 days], 30, +5 days], [-20, +20 days], [-30, +30 days]. The announcement effect can be influencing by various campaign characteristics, it is crucial to get a clear view of the factors that drive and influence CAR. In the following paragraphs I will further elaborate at which campaign characteristics and their effect on CAR I will focus in this study.
In the U.S. campaigns the announcement is usually a 13D filing with the SEC. In this filing the hedge fund states how large the stake is they acquired and what their intentions are. Although the filing requirements vary among the difference jurisdictions in Europe, most don’t require hedge funds to state their objective. Therefore the moment a hedge fund declares its equity stake and its intention are usually different. This creates a possibility to separately study the stock returns for both types of announcement, which has never been attempted before (research question 2A).A regression using a dummy variable that equals one when the campaign announcement was via a regulatory filing tests whether the announcement type significantly influences the announcement stock return. This will show how the market reacts to different types of announcements.
The corporate governance landscape varies between European countries, however is not known whether these inequalities are reflected through the number of campaigns launched in each countries and the cumulative abnormal returns in the period of the announcement (research question 2B). I will show the mean CAR for each country group (groups are formed based on similarity in corporate governance systems) and test whether the means are significantly different from zero. In specific a comparison between the United Kingdom and continental Europe is interesting, the rationale for this comparison will be further discussed in the Literature section. In a regression I test whether the CAR
7 is significantly different for campaigns in United Kingdom and Ireland opposed to campaigns in continental Europe.
The corporate governance regulations vary across countries in Europe, it is essential for hedge funds to have a good understanding of the regulations and framework, which can be costly. Over recent years both the European Union and the individual countries have implemented new guidelines and regulations, which gives more rights and better protection to shareholders (Skadden, 2017). These policy changes makes it easier for shareholders (activists) to exercise their rights. The impact of changing corporate governance policies on CAR is captured in research question 2C.
As stated previously, activist hedge funds may pursue a wide variety of objectives. These objectives are categorized and I display the mean CAR and test whether the mean is significantly different from zero. Furthermore a regression is testing whether the objective type significantly influences CAR in a regression (research question 2D), a comparison with Brav et al. (2008) is made.
There are multiple types of activist hedge funds active on the European market, geographic focus, assets under management (AUM) and the investment approach varies among these hedge funds.As these differences have been underexposed in previous studies that look at cumulative abnormal return, this study will accurately classify the types of hedge fund and identify differences in announcement returns (research question 2E).
Announcement returns show how the market values hedge fund activism, however this is not necessarily related to a corresponding change in underlying firm performance. As put forward by Katelouzou (2013) hedge funds are often accused of being short-term oriented without improving the firm’s (long-term) underlying performance. To test this, this study investigates the effect hedge fund activism has on firm performance 2 years prior the hedge fund engagement until 3 years after the initial engagement. ROA (defined as EBITDA/Assets) and q (defined as (Market Capitalization+ Book value of debt)/ (Book value of equity + Book value of debt)) are measured to accurately capture effects on firm performance. The changes in ROA and q in the [-2, +3 years] interval are
benchmarked against an industry and geography matched peer group.
The rest of this paper proceeds as follows, section II reviews the most relevant literature, based on prior research hypotheses are stated in section III. Section IV and V contain an explanation of respectively the Methods and the Sample. In the last methods are results of the study are discussed.
My thesis will focus on the effect of hedge fund activist campaigns in Europe on cumulative abnormal stock return and underlying firm performance. This setup has not been attempted in Europe before.
8 In summary, this thesis aims to answer the following research questions:
1) What type of firms do activist hedge funds target?
2) How does the market react to the announcement of activism?
a. What is the announcement return depended on the type of announcement? b. Are there differences in announcement returns among countries?
c. Are announcement returns time-varying?
d. Does announcement return vary across pursued objective types? e. Are market reactions depended on the type of hedge fund activist? 3) How does activism impact underlying firm performance?
9
2 Literature
The literature review section is build up as follows, first a general introduction of hedge fund activism and its underlying economic theories is given. Then, relevant research about each of the three main topics for this study is discussed:
1) What type of firms get targeted by hedge funds 2) The announcement stock returns
3) Long run changes in underlying firm performance
2.1 Hedge fund activism
Shareholder activism dates back in the 1980’s, when pension funds and mutual funds began engaging with management to protest management decisions or propose alternative strategies themselves. Bethel et al. (1998) and Karpoff (2001) are among the first to investigate these activist block holdings in the United States. Early research showed no evidence that activism significantly increases share price or firm performance. In the 90’s the phenomenon activist hedge fund emerged. Hedge funds are different from other institutional investors and mutual funds in the sense that are less regulated and have more freedom in terms of the use of leverage, derivates and the amount of capital they invest in an individual stock. On the top of that hedge funds have less conflict of interests (Brav et al., 2008). Cheffins et al., (2013) and Denes et al., (2014) both give an extensive summary of the history of hedge fund activism and conclude that in contrary to other activist investors activist hedge funds accomplish more favorable in terms of share price increases or improvements in firm performance.
To understand hedge fund activism, it is important to zoom in on the theory behind the value creation of activism. As agency theory suggests, the management of a publicly listed firm is chosen by the shareholders and has the task to maximize shareholder value (Denes et al., 2014). A good corporate governance structure should be in place to make sure management has the right incentives to act in shareholders’ best interest. In reality, the incentives of shareholders and management are hard, if not impossible, to perfectly align. As a result management may pursue strategies that are not optimal for shareholder wealth. Corporate governance regulations and frameworks, both on governmental and corporate level, are in place to give power and rights to shareholders. Simply put, activist hedge fund use their rights and power as shareholders to enforce management to act in their best interest by maximizing shareholders value.
In essence, it can be stated that activist hedge funds aim to solve a corporate governance problem by proposing changes to management that increases shareholder value. Roughly speaking campaigns of a hedge fund activists can have one of more of the following four objectives: governance related, balance sheet related (for example returning capital to shareholders or balance sheet restructuring),
10 business strategy related, which includes M&A related objectives and governance related. The
ultimate objective is always simple, earn a good return on investment. Hedge funds search for publicly listed firms in which they believe they can increase the valuation by interacting with management, media and other stakeholders. Hence, many researchers in this field pose the question: what type of firms are likely to be targeted by hedge funds?
2.2 Target analysis
Klein and Zur (2006) compare a sample of 140 hedge fund targets in the United States to a control group of 140 firms (based on industry-size-and-book-to-market). The authors look at profitability variables, discretionary spending, cash and debt capacity and firm size. They find that targets of hedge fund activist have more cash on their balance sheets (13.8%) than the control group (12.0%). Also, hedge funds tend to engage relatively smaller firms, both in terms of market capitalization and total annual revenue. Klein and Zur (2016) further investigate which variables influence the probability of being targeted using a Logit analysis with a dummy variable, which equals one if the firm got targeted by a hedge fund and otherwise zero. In the Logit analysis ROA is positively related to the dependent variable, Cash/Assets is positively significantly related as well, however the industry-adjusted Cash/Assets is not significant.
In contrary to Klein and Zur (2006), Clifford (2008) finds that target firms perform worse in terms of profitability than their industry peers. Also Clifford (2008) refute the finding by Klein and Zur that target firms hold more cash, which contradicts the claim that hedge funds are short term focused by extracting cash from target firms.
Brav et al. (2008) utilizes a similar approach to Klein and Zur (2016) in order to find which firms are likely to be targeted. They find relatively small firm with low market multiples (q), low dividend payout ratios and sales growth are more likely to be targeted. In a Probit analysis these same variables are significantly influencing the probability of being targeted, also leverage does not have a
significant effect, while dividend yield has a significant negative effect. They point to small and medium sized value firms with large institutional ownership and lower dividend payout as typical hedge fund targets. However they also note that hedge funds are taking on increasingly large targets in terms of market capitalization.
Bethel et al. (1998) use a U.S. sample of 146 activist block purchases obtained from SEC 13D filings. They explore the likelihood of a firm being targeted using a Logit analysis and find that large
diversified and poorly performing firms most likely to be targeted and also, contrary to their expectations, that firms with employee stock ownership plans were not less likely to be targeted.
11
2.3 The announcement effect
When an activist hedge fund announces a campaign by publicly engaging a firm or submits a
regulatory filing, other investors in the market will react to this. If for example the market expects that the proposed changes will increase shareholder value and the hedge fund will succeed in persuading management to implement the change, the share price will rise. The announcement effect is therefore an important proxy for how the market values hedge fund activism. A lot of researchers have
attempted to measure this announcement effect on stock returns, however almost all of the prior research is solely focused on the United States.
Klein and Zur (2006) examine a sample of U.S. campaigns between January 2003 and December 2005, collected using 13D SEC filings. The authors use a market model to estimate expected return and thereafter calculate CAR in a [-30, +5 days] and [-30, +30 days] interval. They find a 1% significant outperformance of 7.3% in the first interval and 1% significant outperformance of 10.3% in the second interval. Their results are benchmarked against a control group, for both CAR intervals the mean of the sample is significantly higher, also the Z-statistic for difference in medians is significant in both cases.
Brav et al. (2008) use SEC 13D filings from the EDGAR database to construct a U.S. sample of activist hedge fund campaigns, a total of 1059 campaigns are examined. The authors test what type of firms are targeted by hedge funds, how the market responds to the announcement of activism, whether hedge funds succeed in implementing their objective and lastly how activism effects firm
performance.
Their campaigns are classified comprehensively. Brav et al. (2008) categorize hedge fund objectives as follows: Balance sheet related, Business strategy related, Sale of the entire company and
Governance related. Each objective is perceived differently by the market, depending on the
possibility to create shareholder value and expected success rate. In order to compare my results to the papers of Brav et al. (2008) I have closely mimicked their classification of the sample. In their study objectives related to business strategy and governance are the most sought after, in specific ‘sale of the company to a third party’ and ‘board independence and fair representation’ are the most popular objectives. Roughly half of the campaigns are classified as hostile in their sample.
Brav et al. (2008) find a 7.2% CAR, using a [-20, +20 days] CAR. Moreover they conduct a
regression using that same [-20, +20 days] CAR as dependent variable, the independent variables are the five hedge fund objective types and a dummy variable which equals one if the hedge fund
engagement is deemed as hostile, along with the control variables, the ln(Market cap), Long term debt and average pre-announcement returns. My study will measure the CAR for each of the objective
12 types, as well as the effect of hostility on CAR. Their regression yields that objectives related to business strategy, sale of the company and general objectives generate significantly higher cumulative abnormal returns, while balance sheet and governance related campaigns do not significantly
influence CAR. The hostile engagement dummy positively effects CAR at 10% significance level.
Bethel et al. (1998) use a U.S. sample of 146 activist block purchases obtained from SEC 13D filings from 1980 to 1989. In terms of announcement returns, Bethel et al. use a [-30, 30] days interval to calculate abnormal stock returns using a market model and find a significant positive CAR of 14.2%.
Clifford (2008) looks at value creation of hedge funds activism, measured in terms of announcement returns both in the short and long term, using a U.S. SEC 13D filing sample from 1998-2005. The author concludes that activist hedge fund campaigns have a positive impact on excess stock return in a [-2, +2] days interval using the CRSP value-weighted index to estimate expected returns. In specific the CAR of active blocks is 3.39%, the value of the mean is significant at 5% level. Clifford also finds that the positive impact in strong in aggressive campaigns compared to less-aggressive campaigns.
Table I
Summary the CAR in prior research
Interval (days) Sample period CAR
Brav et al. (2008) [-20, +20] 2001-2006 7.2%
Klein and Zur (2006) [-30, +30] 2003-2005 10.3%
Clifford (2008) [-2, +2] 1998-2005 3.4%
Bessler et al. (2015) [-15, +15] 2000-2006 4.4% Bethel et al. (1998) [-30, +30] 1980-1989 14.2%
Becht et al. (2010) [-20, +20] 2000-2008 4.4%
Few research has been conducted measuring the CAR for European campaigns. What follows is a summary of the main findings of European studies.
Bessler et al. (2015) use a German sample of activist hedge fund campaigns from 2000-2006 to examine what the effects on announcements return, long run stock performance and firm performance are. In their introduction they highlight that recent regulatory changes have made the German
corporate governance landscape more favourable for hedge fund activist.
The authors use the methods proposed by Schwert (1996) to calculate cumulative abnormal return and find a significant 4.43% CAR in the [-15, 15] days interval. Aggressive engagements yield a higher announcement return than non-aggressive engagements. In discussing their results the authors are cautious towards aggressive engagements, suspecting them to be primarily short term oriented.
13 Becht et al. (2008) utilize a sample of campaigns from British investor Hermes, which only included campaigns in the United Kingdom. The scope is their research is therefore narrow, also as discussed earlier the corporate governance landscape is the U.K. is different from continental Europe. They find positive short run stock returns associated with the announcement of the campaigns of Hermes.
2.4 Long run firm performance
Many critics of hedge fund activism state that hedge funds are looking for short term gains at the excess of long term shareholders and/or bondholders (Cheffins and Armour, 2011). For example by extracting cash from the firm by demanding stock repurchases or dividend payments. Various researchers have tried to address this issue using analysis that focus on the effects of activism on underlying firm performance.
Katelouzou (2013) aims to debunk several myths that haunt hedge fund activism, one of which is the short term myth, referring to multiple papers she concludes that there is no reason to assume that hedge fund activist are short term oriented.
Brav et al. (2008) match the targeted firms with a peer group based similar industry/size/book-to-market and find on that targeted firms tend to underperform their peers prior to the engagement in terms of EBITDA/Assets, EBITDA/Sales, Dividend Yield and Leverage. In one of his video lectures Alon Brav explains that in theory he would expect to find a U-shaped pattern. In the year prior to the engagement Brav expects target firm to underperform their peers, possibility as a result of poor choices made by the management. The resulting undervaluation makes the firm an attractive target. Once the hedge fund is on board and actively engages with the management to maximize shareholder value, the underlying firm performance is expected to improve, hence the U-shaped pattern. However the underperformance is only significant for EBITDA/Assets. The improve in firm performance after the engagement relative to the peer group is not significant. In their paper the authors conclude that hedge fund activism has a positive reaction by the market slight improvement in firm performance.
Bebchuck, Brav and Jiang (2015) research the effect of hedge fund campaigns on long run firm performance using a U.S. sample, building further on previous research these authors have done both individually and collectively. The authors are testing whether activist hedge fund are short term oriented as stated in previous literature. Using the CAPM and Fama–French–Carhart asset-pricing model in the [-36, 60 months] interval, they find that campaigns have positive effects on long run firm performance, rather than negative as previously suggested in the literature. Bebchuck et al. (2015) chose ROA (defined as EBITDA/Assets) and q (defined (Market value of equity + Book value of debt) / (Book value of equity + book value of debt)) to measure changes in firm performance from the engagement year up until 5 years after, looking at 2,000 engagements in the period 1994-2007. They
14 first display raw value of these variables, followed by industry adjusted values. They find significant improvements in both ROA and q and therefore conclude that the claim that hedge fund activism has an adverse effect on long term shareholder wealth is false.
Bethel et al. (1998) test on effect on industry-adjusted firm performance using ROA as benchmark in [-1, +3] years interval, looking at block holdings (both active and passive) in the period 1980-1989 and conclude that there are significant improvements.
2.5 Hedge fund activism: The European Case
There are four main factors contributing to European firms being an attractive targets for hedge funds. As stated, hedge funds look for undervalued firms and currently European firms are undervalued relative to their American peers. Secondly, as stated in the Lazard’s Review of Shareholder Activism (2017) hedge funds are increasingly targeting globally active large cap U.S. firms. European
headquartered global firms are not very different from their U.S. peers and can therefore be targeted in a similar manner. A prominent example is Nestlé, which got targeted by American hedge fund Third Point in 2017. Another contributing factor is that shareholder bases of European firms are increasingly starting to look like American firm’s shareholders bases with less family ownership and more
(American) institutional ownership. As the activist hedge fund usually only buy 1-10% of the firm’s total outstanding equity, the support of institutional shareholders can be crucial of the success of the campaign. Also, as a more general factor, over time search costs, dealing costs and communication costs have lowered as a result of advancing information technology. This has aided the rise of hedge fund activism and because these costs have lowered the number of potentially profitable campaigns has grown (Cheffins and Armour, 2011).
Katelouzou (2013) uses a non-U.S. sample of hedge fund campaigns to measure their effect on corporate governance and firm performance and find abnormal stock returns and improvement in firm performance. Bessler, Drobetz, Holler (2015) emphasize the differences between the U.S. market and the German (continental European) market. German has a bank-based system, while the U.S. has a market based system. In German corporate control is shifting from banks to other institutional investors, which opens the doors for hedge fund activist. Moreover, they point to new corporate governance regulation both in the EU and German as a contributing factor to the rise of hedge fund activism in Germany. Their study Bessler, Drobetz, Holler concludes that hedge fund campaigns are able to generate shareholder value in both the short run and the long run.
The rise in hedge fund activism in Europe can be contributed to multiple factors. Whereas the amount of money being allocated to hedge funds grew, the number of potential targets (publicly listed firms) in the United States remained roughly the same. Therefore hedge fund that were traditionally focused
15 on the United States, have started looking at other markets (Katelouzou, 2013). This pushes hedge funds to look for attractive opportunities outside of the U.S. market. As a result many hedge funds have entered the European market over the past decade.
Because of inequalities in corporate governance landscape among European countries, different outcomes of hedge fund activism might be expected. The CMS Law report (2017) emphasizes the EU is increasingly harmonizing member countries’ corporate governance law. This creates a more transparent market for activism. In particular the EU Directive Shareholder Rights Directive 2007/36/ EC1 protects the voting rights of shareholders at general shareholder meeting and states under which circumstances shareholders can call for general meetings. EU Directive Transparency Directive 2004/109/ EC advocates using the same regulatory filing requirements across member states. Bessler et al. (2015) highlight that recent regulatory changes have made the German corporate governance landscape more favourable for hedge fund activist. As an example the corporate governance landscape in United Kingdom is generally regarded as more suitable for activism than the corporate governance landscape in Germany (Skadden, 2017). Whereas the regulatory framework in the United Kingdom is focused on shareholder rights, whereas the most countries on the European continental are focused on the rights of all stakeholder. The corporate governance policy changes combined with changes in other factors such as changing shareholder basis justify the question how these difference could be reflected through the number of campaigns launched in each countries and the cumulative abnormal returns in the period of the announcement, which related to research question 2B in this study. Several papers are hinted that certain activist hedge fund’s characteristics are influencing the announcement returns. Cheffins and Armour (2011) and Denes et al. (2015) touch upon this topic while summarizing research done on hedge fund activism. Large, well-known hedge funds might trigger a stronger reaction on the stock market, building on their reputation. However there are no papers that try to empirical test the relationship between hedge fund type and announcement returns.
There are multiple types of activist hedge funds active on the European market. Some are solely focused on one country, others are active throughout all of Europe and lastly hedge funds (mostly American hedge funds) that are active globally (Cheffins and Armour, 2011). Moreover there are strong difference in how hedge funds approach management, shareholders and the media, some are collaborative in their approach, others can be characterized as hostile. The Assets under Management (AUM) and public reputation varies across hedge funds as well. Although several papers point it this, it has not been properly studied yet. In order to capture these differences research question 2e has been formulated.
16
3. Hypotheses
Based on the existing literature, certain expectations of the outcomes of this study have been formed. These expectations have been formulated as hypotheses.
1. What type of firms do activist hedge funds target?
H0:Target firm are expected to be less profitable in terms of return on assets, have lower revenue growth and market multiples (measured by q and market-to-book ratio) compared to their industry and geography based peers.
2A. What is the announcement return depended on the type of announcement?
H0: The type of announcement is not expected to have a significant impact on cumulative abnormal return.
2B. Are there differences in announcement returns among countries?
H0: United Kingdom is expected to have a relatively higher number of campaigns than the continental Europe, however the average cumulative abnormal return per campaign is not expected to be
significantly different.
2C. Are announcement returns time-varying?
H0: The cumulative abnormal return is expected to be slightly increasing over time.
2D. Does announcement return vary across pursued objective types?
H0: Objectives related to business strategy and sale of the target company are expected to higher cumulative abnormal return compared to the other objectives. Governance related objectives are expected to have lower cumulative abnormal return associated with them.
2E. Q: Are market reactions depended on the type of hedge fund activist?
H0: Hedge fund reputation is expected to be in significant positive effect on cumulative abnormal return. Well-known fund with assets under management of over $4 Billion are expected to achieve higher announcement returns.
3. Q: How does activism impact underlying firm performance?
H0: Hedge fund activism is expected to significantly decrease return on assets and q prior the engagement and significantly increase return on assets and q after the engagement.
17
4. Methods
Based on the research question, three topic are studied, namely:
1) Target characteristics 1) Announcement effect 2) Firm performance
The research methods of each of these three are discussed in detail in their individual subsection.
4.1 Target firm characteristics analysis
The objective of the target firm characteristics analysis is to get a clear view of what type of firms are likely to be targeted by hedge fund activism. Two methods are utilized to accomplish this. Firstly, relevant financial metrics of target firms are benchmarked against industry and geography peers. Secondly, a Probit analysis, using a target dummy, will reveal which variables significantly influence the probability of being targeted by activist hedge funds.
In order to accurately capture target firm characteristics, the following variables have been selected.
Accounting variables: ROA (EBITDA/Assets), Revenue growth, Leverage, Cash/Assets,
CapEx/Assets. Market variables: Market capitalization, q, Buy-and-hold return, Market to book ratio.
The accounting variables are on annual basis, measured for the year prior to the engagement. The event year is defined as the nearest start of a new year (e.g. for campaigns announced at 5 July 2013 and 18 June 2014 the event year is 2013, the nearest start of a new year). Scripts written in
programming language Python to correctly process the data. The values of aforementioned variables are extracted from Datastream in the period 2007-2018. Python scripts then select the event year for each of the target firms in the sample, afterwards the correct values for each of the variable are calculated. The raw averages and raw medians of the nine variables are displayed in columns 1-3 of Table VI.
To make the analysis more meaningful the results are benchmarked against a peer group. The peer group is selected based on industry and geography. The Bureau van Dijk Sector Classification is used to select sector peers. Since this study only focuses on European campaigns, it is naturally more interesting to study firm characteristics of target firms compared to European peers. Therefore only European firms are included in the peer group. Using a modified version of the Python scripts is utilized to select the financial data of the peer group in the same year the firm was targeted. Columns 4-5 of Table VI. displays the average difference for the target firm compared to its peers for each of the nine variables. A t-test is conducted to test whether the differences are significantly different from
18 zero, in other words whether the mean of the target firms is significantly different from the mean of the peer group.
Then, secondly a Probit analysis is conducted. The dataset contains 307 targeted firms, this number excludes all banks, insurance companies and investment vehicles. Rationale is that EBITDA/Assets is not a reliable when assessing the profitability of these type of firms. The total number of peers which haven’t been targeted is 4923, therefore the overall dataset used for the Probit analysis contains 5230 firms. The total group of peers is winsorized at 3% level. The dependent variable is a dummy
variable, which equals one if the firm is targeted by an activist hedge fund and zero if the firm hasn’t been targeted. The independent variables are the same nine variables which are used for the target firm characteristics analysis. The results of the Probit analysis are shown in Table VII.
4.2 Announcement effect
The announcement effect is measured by Cumulative Abnormal Return (CAR). CAR will be calculated for the periods: [-20 days, +20 days], [-10days, -1 day], [-30days, +5day], [-2days, +2 days] , [-60days, +60 days] with respect to the initial announcement day. Taking five estimation periods serves two purposes. Firstly it was useful for comparison purposes, as prior studies have used different time frames for CAR, secondly it serves as robustness check. The CAR formula is:
𝐶𝐴𝑅𝑡 = 𝛴 ∗ 𝑑𝐴𝑅𝜏
with ARi,t:
𝐴𝑅𝑖, 𝑡 = 𝑅𝑖, 𝑡 − (𝛽𝑖 × 𝑅𝑀𝑡)
Where Ri,t is the return of the target firm’s stock and RM,t is the market return. Where ERi,t is Expected return for stock i. The Risk Free rate (for this study the German 10 year maturity government bond) is assumed to be zero. This assumption is realistic, since the CAR is calculated over 60 days only.
In other words, Abnormal Return is defined as Actual return minus Expected return. Expected return is measured using a market model, namely the CAPM model. The associated market return is measured by Vanguard FTSE Europe Index Fund ETF. This passively managed ETF measures the investment return of stocks issued by companies located in the major markets of Europe. The ETF does not include Eastern European markets, which are also not represent in the sample. The ETF accurately tracks the market the companies in the sample are active.
The estimation period is 150 days prior to the engagements until 31 days prior to the engagement. The estimation period ends 31 days before the initial engagement, because in the period [-31, 0 days] the
19 hedge funds are building up their stake in the target company and influencing the stock price,
therefore this period is excluded from the estimation period.
In order to calculate Cumulative Abnormal Return (CAR) per campaign, stock price is obtained from Datastream. Stock price time series from 01 September 2006 until 07 June 2018 is obtained for each targeted firm and for the market index. Scripts have been written in programming language Python to correctly select the corresponding timeframe for each campaign using Pandas Dataframes. The script selects the time period 30 days before the engagement until 30 days after the engagements and 150 days before the engagement until 31 days (the estimation period) before the engagements for each campaign and the Vanguard FTSE Europe Index.
The Beta coefficient is calculated as:
𝛽𝑖 =𝐶𝑜𝑣(𝑅𝑖, 𝑡; 𝑅𝑀𝑡) 𝑉𝑎𝑟(𝑅𝑀𝑡)
Consequently, Expected return is calculated as:
𝐸𝑅𝑖, 𝑡 = 𝛽𝑖 × 𝑅𝑀𝑡
Where ERi,t is Expected return for target firm i. The Risk free rate is again assumed to be zero.
Subtracting the Expected return from the Actual return gives the Abnormal return. Summing these Abnormal returns up for each of the five time intervals gives the CAR. The cumulative abnormal stock return is assessed in five periods : [-2 days, +2 days], [-10days, -1 day], [-30days, +5day], [-20 days, +20 days] , 30days, +30 days] for the full sample. The CAR is winsorized at 1% level. The [-10, -1] interval (partially) captures the time period the hedge fund is building up its stake. [-2, +2] captures the short term announcement effect, while [-20, +20] and [-30, +30] captures both the period in which the hedge fund builds up the stake as well as the market reaction to the announcement.
These specific CAR periods have been chosen in order to be able to compare the results with those of prominent U.S. studies. The [-20, +20] days CAR is assessed to compare results with Brav et al. (2008). Clifford (2008) uses multiple intervals, yet only reports [-2, +2] days, the author uses the same estimation period, namely [-150, -31] days, which makes our outcomes easy to compare. Lastly Klein and Zur (2006) use [-30, +5] days and [-30, +30] days. The five time intervals also serve as robust checks, as the periods all represent different assumptions regarding the stake building of hedge fund and the time the market takes to react to the announcement.
To test whether the CAR is varying among countries, hedge fund objective type and over time. The means are showed for each category and t-statistic testing whether the mean is significantly different
20 from zero. The periods include 2010-2012, 2012-2015 and 2016-2018, since the sample ends in June 2018 the last period is slightly shorter than the other periods. The results are shown are in Table XI.
Two regressions are using to establish whether CAR is influenced by certain variables. The first regression tests whether [-2, +2 days] CAR is dependent on the type of campaign announcement, the country of the target firm and type of activist hedge fund. The regression is conducted with CAR in the [-2, +2 days] interval as independent variable and as independent variables three dummy
variables: 1) dummy variable that equals one if the country of the target firm is the United Kingdom (research question 2B), 2) large fund dummy that equals one if the hedge fund is a well-known activist has over $4 Billion assets under management (AUM) (research question 2E) and 3) dummy variable that equals one if the campaign is launched 2016 or later (research question 2C). Market-to-book ratio, announcement type and the natural logarithm of market capitalization are included as control variables. All t-statistics adjust for heteroskedasticity. The results are shown are in Table XII. The following hedge funds are classified as large funds: Elliott Management, TCI Advisory Services, Cevian Capital, Third Point, Trian Partners, Corvex Management, Pershing Square, Greenlight Capital.
The second regression has [-20, +20 days] CAR as dependent variable, the independent variables are dummies of the objective types and dummy for campaigns classified as hostile. Announcement type Market-to-book ratio, announcement type and the natural logarithm of market capitalization are included as control variables. All t-statistics adjust for heteroskedasticity. The regression tests research question 2B, the results are shown are in Table IX.
4.3 Effects on the long run firm performance
To measure to effects of hedge fund activism on long run firm performance, the changes in ROA and q are measured over time. Bebchuck et al. (2015) use these same variable to measure long run firm performance, in the end my results will accurately be compared to their results .
As mentioned before, the financial data is obtained from Orbis and Datastream. The data is extracted from the databases in yearly format from 2007 until 2018.
The time period is 2 years prior to the engagement until 3 years after the engagement. I have written scripts in Python to correctly select the right time frame for each campaign.
ROA is defined as EBITDA/Assets and q is defined as (book value of debt + market value of
equity)/(book value of debt + book value of equity). First, the raw values of the three financial metrics in time periods [-2 years, +3 years], the results are shown in Columns 1-3 of Table XII. Of course the raw value are not particularly meaningful without knowing how similar firms that were not targeted
21 performed in the same period. Therefore target firms are benchmarked against a peer group. The peer group is selected based on industry and geography. The Bureau van Dijk Sector Classification is used to select sector peers. Since this study only focuses on Europe campaigns, it is naturally more
interesting to research firm performance of target firms compared to European peers. Therefore only European firms are included in the peer group. Using a modified version of the Python is utilized to select the financial data of the peer group in the same year the firm was targeted. Columns 4-6 of Table XII display the average difference for the target firm compared to its peers. I visually test whether the U-shaped pattern is clearly present by plotting industry-adjusted ROA and q (y-axis) against time (x-axis). The results are shown in Graph II. t-tests show whether the mean for ROA and q in year t-2 and t-1 are significantly different from the mean in the event year and whether the mean for ROA and q in year t+1, t+2 and t+3 are significantly different from the mean in the event year. The results are shown in Table XIIV.
22
5: Dataset/ sample
This section starts with elaborating how the sample has been gathered and continues to describe the sample characteristics in descriptive tables. I identified roughly 550 campaigns, after classification 377 campaigns are included in the final sample.
5.1 Sample gathering
The sample consists of activist hedge fund campaigns in Western-Europe in the period 2010-2018. The entire sample is handpicked using news reports and articles from database LexisNexis, complimented by relevant papers, reports and articles about hedge fund activism in general or individual hedge fund campaigns in specific. Data processing and modification is performed using programming languages Python and R.
The sample and all other required information are gathered in the following steps: identifying the relevant campaigns, coding and classifying the hedge fund engagements, obtaining stock price data and collecting all required (financial) metrics and information. These steps are elaborated in detail in the following paragraphs.
The sample consists of activist hedge fund campaigns in Western-Europe. The countries included in the sample are Norway, Finland, Sweden, Denmark, The Netherlands, Belgium, Luxembourg, Germany, Switzerland, Austria, France, Spain, Italy, Portugal, Ireland and the United Kingdom. Events in which hedge funds buy an minority equity stake in a listed firm and subsequently actively interacts with the management and other shareholders of the firm to propose changes to the firm are included in the sample.
The country are grouped based on similar characteristics with respect to corporate governance framework, the country are paired into the following country groups:
• Nordics (Denmark, Finland, Sweden and Norway) • Benelux (Netherlands, Belgium and Luxembourg) • United Kingdom and Ireland
• Germany and Austria • Switzerland
• France
• Southern Europe (Italy, Spain and Portugal)
The timespan of the sample is from 2010 until present. This time is chosen because it does not overlap with the time periods of past similar studies such as Becht et al. (2010) and therefore produces new
23 unique results. The time period includes the rise in hedge fund activism in Europe that was triggered by several factors stated in the Introduction and Literature section of this study.
In the U.S. investors have to report to the SEC once they acquire an equity stake larger than 5% in a publicly traded company. Therefore these SEC filings provide a good source of data to identify hedge fund campaigns to research focusing on the U.S.
In Europe investors are also required to report significant equity stakes to national regulators, however the specific filing requirements vary per country and are not always publicly available (in English). It is for this reason that filings to the national regulators in the various European countries are not a reliable source of data.
Alternatively, activist hedge fund engagements are hand searched in LexisNexis, using multiple search term, including ‘Firm name + hedge fund + activist’, ‘Hedge fund name + proxy contest’, ‘Hedge fund name + engagement + stake’ and ‘Firm name + activist + management letter’. This is complimented by searching in relevant papers, reports and articles about individual hedge fund campaigns in specific. In total I identified roughly 550 campaigns which in principal met the requirements I set. The selection bias in identifying these campaigns in discussed in section 5.3.
5.2 Classification of the campaigns
After identifying all relevant activist hedge fund engagements, the campaigns are accurately coded and classified. This is done by reading news items, articles, reports and papers regarding the
individual campaigns until extensive and chronological overview of the campaign and all its relevant surroundings. Formal communication from the hedge fund itself (e.g. letters to management), official reactions or statements form the target firm are reliable sources, moreover information in regulatory filings and news reports from respected sources such as Reuters, Bloomberg and Financial Times are used as well. The sample is coded in such a way that all relevant hypotheses and research questions can thoroughly be answered based upon it. The campaigns will be classified in the following categories.
5.2.1 Hedge Fund Objective
Below in Table II the hedge fund’s primary objectives. A relatively high number of campaigns is focused on governance related issues. A lot of these campaigns aim to get an employee/affiliate of the hedge fund on the target firm’s board and/or remove an existing board member, in total over 1/3 of the full sample. The other campaigns are relatively well spread over the objective categories.
24 usually pursued after first obtaining a board seat and therefore not directly observable. Note that hedge fund may pursue multiple objectives, only the primary objectives are shown in the table below.
Table II
Summary of events by hedge fund's objectives
All events Hostile events Non hostile events N % N % N %
1) No specific goal, shareholder value maximalisation 30 10.20% 2 1.37% 27 19.01% 2) Balance sheet related (dividends, capital restructuring)
a. Excess cash, under-leverage, dividends/ repurchases 13 3.74% 7 4.11% 6 3.52% b. Equity issuance, restructure debt, recapitalization 13 3.40% 6 3.42% 6 2.82%
3) Business strategy:
a. Operating efficiency, cost-cutting, tax-related 15 3.06% 5 0.68% 10 5.63% b. Spin off division (diversification discount) 43 12.24% 16 8.90% 27 16.20% c. M&A as target 48 15.31% 36 22.60% 11 7.75% d. M&A as acquirer 9 2.38% 4 2.74% 5 2.11% e. Propose an different growth strategy 12 1.70% 7 1.37% 5 2.11% 4) Sale of the entire company
a. Sale to third party 34 7.14% 23 10.96% 9 3.52% b. Sale to the hedge fund 6 2.04% 6 4.11% 0 0%
5) Governance related
a. Remove takeover defences 2 0.68% 2 1.37% 0 0% b. Remove existing board members 46 12.93% 40 22.60% 6 3.52% c. Gain board representation 80 22.45% 30 12.33% 47 31.69% d. More information disclosure 3 1.02% 2 1.37% 1 0.70% e. Excess executive pay 6 1.70% 4 2.05% 2 1.41% Sum of categories (1) to (5) 360 190 162
5.2.2 Announcement type & Hedge fund reaction
Campaigns can be announced by regulatory filings, public press releases, a simultaneous filing and press release or rumours in the press. The descriptive table of the distribution of the announcement types in the sample can be found in the appendix.
The target firm’s management reaction is either accommodating, negotiating, resisting or no reaction/ no information. The firm management is more often resisting the hedge fund’s proposed changes than accommodating. The descriptive table of the distribution of the reactions by the management in the sample can be found in the appendix.
5.2.3 Classification of hedge fund tactics
The campaigns in the sample are broken down into Hostile (subcategories 4-7) and Non-hostile (subcategories 1-3). Subcategory 3 (Expresses public proposals/ criticism ) can both be hostile and non-hostile, this is judged on individual basis based on the nature of the interaction with the media/public.
25 Below in Table III the tactics a description of by the hedge funds. Most campaigns that have a
business strategy related objective (objective category 3) use tactics in category 3: public
proposals/criticism. While governance related campaigns often threaten to launch a proxy contest or actually launch one. Very hostile tactics, suing the target firm’s or try to take control are relatively seldomly used.
Table III
Tactics employed by Hedge Fund
N. of events %
1) Communicating with management 34 10.21%
2) Demands board seat, no proxy contest 32 9.61% 3) Expresses public proposals/ criticism 135 40.54%
4) Threatens with proxy contest 67 20.12%
5) Launches proxy contest 52 15.62%
6) Sues the target company 8 2.40%
7) Tries to take control of the entire company 5 1.50%
Total 333 100%
5.2.4 Success of the campaign
The success rate of hedge fund campaigns indicates the number of campaigns in which the hedge fund succeed/failed/partially succeed in implementing their objectives.
The most notable thing visibly below is that the hedge fund is more often victorious than the management. This indicates that hedge fund are relatively successful in achieving their objectives. Governance related objective are the most likely to succeed, while Sale of the entire company is more likely to fail than succeed. In 41.1% of the campaigns the hedge fund succeeds in implementing their objectives, in 22.1% of the cases the hedge funds fail to its their attempts. The success rate is the U.S. is slightly lower 40.6% is successful, 33.6% is unsuccessful (Brav et al., 2008).
Table IV
Summary of Success rate by Hedge fund's objectives
N. of events Successful % Partially Successful % Failure % Ongoing/ No Info % Withdrawn % Balance sheet related
20 25.0% 20.0% 30.0% 25.0% 0.0% Business strategy
related
103 34.0% 9.7% 22.3% 33.0% 1.0% Sale of the entire
company 21 38.1% 4.8% 42.9% 4.8% 9.5% Governance related 114 50.9% 6.1% 16.7% 21.9% 4.4% Sum of categories 258 41.1% 8.5% 22.1% 25.2% 3.1%
26 5.2.5 Country Group
Below in Table V the number of campaigns launched in each of the country groups. The country groups are formed based on the characteristics of the corporate governance structure and their economic and business interdependence. The U.K. market is by far the most active, their market accounts for over the half of the total number of campaigns and is therefore more mature than other European markets. As discussed in the Literature section, a possible explanation is that the U.K. corporate governance is more suited for activism, because it is easier for shareholders to exercise their rights. In a regression it is tested whether CAR is significantly different for U.K. campaigns as opposed to continental European campaigns.The total number of campaigns is continental Europe is 175, the aggregate for the Kingdom is 192.
Table V
Distribution of Targeted Countries
Full sample Hostile subsample Non-hostile subsample N. of Events Percentage N. of Events Percentage N. of Events Percentage
Nordics 29 7,71% 8 27.6% 21 72.4%
Benelux 27 7,18% 10 37.0% 17 63.0%
UK & Ireland 201 53,46% 72 35.8% 129 64.2% Germany & Austria 41 10,90% 12 29.3% 29 70.7%
Switzerland 27 7,18% 10 37.0% 17 63.0%
France 30 7,98% 19 63.3% 11 36.7%
Southern Europe 21 5,59% 5 23.8% 16 76.2%
Total 376 100% 136 36.2% 219 63.8%
5.2.6 Number of Launched Campaigns per Quarter
Hedge fund activism has risen in Europe over the last 10 years, as the trendline in the figure below indicates, the number of launched campaigns is gradually rising over time. In the Results section the mean CAR per time period is displayed. This will reveal if, apart from the number of campaign launched, the mean CAR is also rising over time.
27 5.2.7 Sector distribution
The descriptive table of the distribution of distribution of sectors in the sample can be found in the appendix. The sector distribution is based upon Bureau van Dijk Industry Classification.
28 5.2.8 Most active hedge funds
When looking at the most active hedge fund funds in Table VI, it is visible the campaigns are relatively well-spread across hedge funds, with only four hedge funds with over 10 campaigns in the current sample. Elliott Management is the most active hedge fund with 44 campaigns. Weiss Asset Management, Elliott Management and Third Point are prominent U.S. headquartered hedge funds, while Cevain, TCI and Knight Vinke are European hedge funds. Crystal Amber and Harwood Capital Management is purely focused on the U.K. market. The top 20 most active hedge funds account for more than 50% of the total sample.
Table VI
Hedge Funds and Targeted Countries
Summary of the country breakdown of the campaigns launched by the most 20 hedge funds. The hedge fund are ranked based on the number of campaigns launched.
Rank Hedge fund Benelux France Germany Ireland Italy Nordics Spain Switzerland UK Total
1 Elliott Management 3 2 7 3 2 2 23 44
2 Cevian Capital 1 4 1 8 4 5 23
3 Crystal Amber Advisers 1 17 18
4 TCI Advisory Services 3 2 1 3 11
5 Amber Capital 3 5 2 10
6 Laxey Partners 10 10
7 Knight Vinke 1 2 2 2 2 9
8 Harwood Capital Management 7 7
9 Sherborne Investors 7 7
10 Toscafund Asset Management 1 5 6
11 Wyser-Pratte & Co 1 3 2 1 7
12 Third Point 1 1 1 1 2 6
13 Weiss Asset Management 6 6
14 Damille Investments 5 5
15 GVO Investment Partners 5 5
16 ValueAct Capital Management 1 4 6
17 Artisan Partners 1 1 1 4
18 Centaurus Capital 2 1 1 4
19 GO Investment Partners 1 1 2 4
20 Gyllenhammar Holding 1 3 4
29
5.3 Selection bias
Since the sample is handpicked and self-classified, selection biases will separately discussed in this section and is touched upon in the final discussion of this study.
Identification of campaigns is depended on publicly available information. Large, well-known hedge funds and large cap firms naturally receive more media coverage than small, local hedge funds. I searched extensively to identify these smaller, local hedge funds and their campaigns. However there is still a bias towards large well-known hedge funds and large cap firms in the sample. Likewise long and hostile campaigns receive more media coverage, than non-hostile campaigns that in some cases stay under the radar. Most hedge funds initially engage management privately to discuss their
proposed changes. If this succeeds, the campaign does not get public attention and is therefore harder to identify. If management resists or rejects the proposals, hedge funds will often resort to more hostile tactics to implement their proposed changes, this cases often get media attention and are therefore easier to identify.
Campaigns in Southern Europe (Italy, Portugal and to a lesser extent Spain) were not necessarily difficult to identify, but rather difficult to classify, because of a lack of the availability of English press coverage. Where possible relevant articles have been translated, however if the risk of
misclassification due to faulty translations became too large, these campaigns were removed from the sample. As a consequence the number of campaigns in Southern Europe is not a true representation of the actual number of launched campaigns. As time passes, ongoing campaigns may not be receive media coverage anymore, this is especially the case for non-hostile campaigns. In these cases it is often hard to classify whether the hedge fund succeeded in implementing their objectives.
In order to minimalize the subjective classification of the sample I have closely mimicked the classification prior studies use, primarily Brav et al. (2008) and to a lesser content Klein and Zur (2006) and Clifford (2008). I am well-aware of the selection biases and have taken these in account with formulating research questions and research design.
30
6. Results
6.1 Target firm characteristics
The section contains the target firm characteristics, both raw values and industry adjusted, and the Probit analysis, which assesses which firms are likely to be targeted by activist hedge funds.
From Table VII I derive that ROA is higher for target firm compared to their peers, the difference is not significant however. Sales growth (in the year prior to the engagement), leverage and Cash/Assets are significantly lower for target firms. Market-to-book ratio and q are significantly lower for target firms, indicating that these firms are undervalued relative to their peers.
The observation that Cash/Assets ratio is lower for target firm then for peers contradicts findings of Klein and Zur (2006), but is in line with Clifford (2008). It refutes the claim that activist hedge funds are looking to extract cash from firms and are therefore short term oriented.
The market capitalization of targeted firm is significantly higher than their peers. This finding contradicts the findings of researchers using a U.S. sample. I identify three possible explanations. Firstly, the European market for hedge fund activism is less mature than the U.S. market, Lazard (2017) stated that hedge funds that are active globally are more likely to target European
multinationals, whose shareholders bases and business strategy resembles those of American
multinationals. As the market becomes more mature and the ‘low hanging fruit’ within the European large cap range disappears and hedge funds might start targeting smaller targets. Secondly, the large amount of capital located to hedge funds allows them to pursue increasingly large targets, as pointed out by Bebchuck et al. (2015).
The last and perhaps most important explanation lays within a selection bias. In U.S. campaigns are identified using 13D SEC filings, which are required for equity stakes in U.S. listed firms, regardless of their market capitalization, there is no selection bias here. In Europe regulatory filings are a less reliable data source, instead the sample is hand-picked using LexisNexis news database. Naturally campaigns in small, less known firm will capture less media attention and are therefore harder to identify. This selection bias certainly partially explains why the market capitalization of the targeted firm is significantly higher than the peers.
The results are roughly in line with Brav et al. (2008), they also find that Return on assets is higher (0.02) for target firms, while annual revenue growth (-0.057), q (-0.397) and cash/assets (0.028) are lower. Their peer group is based on industry/size/book-to-market.
Even after winsorizing, the distributions of variables often display fat tails and skewness. As a result extreme observations influence Mean and t-stat. As a robustness check I use the Wilcoxon statistic,
31 which tests whether the median significantly different from zero. In table VII we observe that the Wilcoxon statistic is much lower than the t-stat of the mean. This indicates that the mean is heavily influenced by extreme observations (e.g. Nestlé and Volkswagen).
Table VII
Target firm characteristics
Columns 1-3 summarize the Mean, Median and Standard deviation of five accounting variables and four market variables for the targeted firm in the sample. The variables include: Return on Assets (measured as
EBITDA/Assets), Cash/Assets, leverage (measured as Debt/Assets), Annual revenue growth, Capital expenditures/Assets, Market capitalization (in Millions €), q (defined as (defined as (Market Capitalization+ Book value of debt)/ (Book value of equity + Book value of debt)), Annual buy-and-hold return. Column 4-5 reports the Mean difference with an Industry and Geography matched peer group and the
corresponding t-statistic. The Wilcoxon statistic tests whether the Median is significantly different from zero and is added as a robust check. ** and * indicate statistical significance at the 5% and 10% levels.
Sample Difference
Accounting Obs Mean Median Std. Dev. Mean t-stat Wilcoxon
ROA 239 7.66% 8.70% 0.128 0.012 1.40 3.46 Sales growth 237 0.57% 0.94% 0.206 -0.099 -7.51 -9.25 Leverage 220 25.82% 22.68% 0.179 -0.027 -2.28 -3.31 Cash/ Assets 246 14.89% 11.04% 0.159 -0.036 -3.53 -6.95 CapEx/Assets 240 4.05% 2.85% 4.195 -0.272 -1.01 -4.45
Market Obs Mean Median Std. Dev. Mean t-stat Wilcoxon Market cap 305 8709 1500 22000 6069 5.56 2.22
q 288 1.083 0.953 0.681 -0.368 -8.23 -8.95
Buy-and-hold
return 288 0.108 0.049 0.453 -1.505 -0.57 -2.02 Market-to-book 288 1.081 0.922 0.828 -0.335 -6.77 -7.41
The Probit analysis gives significant negative coefficient for the variables q, growth and the constant. Market capitalization is significant and positively related to the probability of being targeted. The Leverage, Cash/Assets, CapEx/Assets coefficients are positive, yet they are not significantly related to the probability of being targeted. ROA is negatively related to probability of being targeted, yet not significant. The buy-and-hold return coefficient is positive, yet not significant.
Again, the results are roughly in line with Brav et al. (2008), firms with a higher return on assets also have a higher probability of being targeted, while higher annual revenue growth, leverage, q and cash/assets lower the probability of being targeted. Their peer group is based on industry/size/book-to-market. The leverage coefficient they find is not significant, while all the other above mentioned variables are significant at 5% level. Klein and Zur (2006) use a Logit model to calculate probability of being targeted and, in line with my results, find return on assets, leverage and cash/assets positively influences the probability of being targeted.
32 Table VIII
Probit analysis of targeting
Probit analysis of the probability of being targeted by hedge fund activism. The dependent variable is a dummy variable which equals one if the firm got targeted by an activist hedge fund. The independent variables are Market capitalization, q (defined as (Market Capitalization+ Book value of debt)/ (Book value of equity + Book value of debt)), Annual buy-and-hold return, leverage (measured as Debt/Assets), Annual revenue growth, Return on Assets (measured as EBITDA/Assets), Cash/Assets, Capital
expenditures/Assets. All independent variables are measured in the year before the hedge fund got targeted. Coefficients, standard error and z-score. All t-statistics adjust for heteroskedasticity. ** and * indicate statistical significance at the 5% and 10% levels.
Target Dummy Coef. Std. Err. z
Market cap 3.64E-08** 3.57E-09 10.2
q -0.326** 0.062 -5.27 Buy-and-Hold return 0.003 0.003 1.05 Leverage 0.032 0.194 0.16 Growth -2.243** 0.520 -4.31 Return on Assets -0.069 0.274 -0.25 Cash/Assets 0.023 0.252 0.09 CapEx/Assets 0.015 0.014 1.08 Constant -1.282** 0.097 -13.19 Pseudo R2 0.132 Percent targeted 6,24%
