Hedge fund activism in R&D-intensive industries and company performance

Hedge fund activism in R&D-intensive industries and company performance Abstract This thesis investigates the differences in the effect of hedge fund activism on companies long-term performance between R&D-intensive industries and other industries. Using a dataset of activist hedge fund events, the effect on company performance following the five year period after intervention is measured. Panel regression has shown no significant differences in long-term company performance, whereas the comparison between increases (decreases) in company performance between the R&D-intensive industry and other industries does show significant differences. The overall results from this thesis show some evidence of activist hedge funds short-termism in the R&D-intensive industry, but also show evidence of an improvement of long-term performance in general. Jop Last 10751645 Economie & Bedrijfskunde Financiering & Organisatie Dhr. dr. S. R. Arping 31 januari 2018 1

Statement of Originality This document is written by Student Jop Last 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. 2

1. Introduction Hedge fund activism has played an important role in the corporate governance debate and the media. One who reads financial newspapers can find various articles about hedge funds behaving as activist shareholders. For instance, Elliott Management recently got Alexion to let them fill a board seat (Al-Muslim, 2018). This way Elliott Management has more control as a shareholder and this is a typical example of what hedge funds do to become activist in a company. Many experts have expressed their concerns about this phenomenon, claiming hedge fund activism is focused on short term profits, not on long term company growth. Lipton (2013) says that "activist hedge funds are reportedly outperforming many other asset classes," the value they capture is "appropriated from fellow stockholders with longer-term investment horizons." They expropriate companies of their wealth by increasing dividends, buying back shares, increasing leverage or by selling off company assets. Empirical research has investigated these claims and have contradictorily discovered these claims are untrue. They report long term positive results for the company, measured by Tobin s Q and ROA (Bebchuk, Brav & Jiang, 2015). If these hedge funds are short-termist, what implications could that have on firms which rely heavily on investing in the long term by Research and Development (R&D)? Brav, Jiang, Ma, & Tian (2016) concluded that R&D expenditures drop, but output measured by patents and citations rises. This means activist hedge funds could make the company s R&D department more efficient. But is this also the case for R&D-intensive firms? Do hedge funds use different strategies when the target company is in a R&D-intensive industry? This bachelor thesis investigates the long term effects of hedge fund activism in R&Dintensive industries as compared to other industries. Using panel-data regression, the effects of activism are measured with an extended version of the dataset used by Brav, Jiang, Partnoy and Thomas (2008). Using a time window of 5 years after a hedge funds 13D filing with the SEC, long term performance can be measured. Also, it will be investigated whether activist hedge funds use a different approach to R&D-intensive firms. Section 2 of this thesis discusses the relevant literature of the topic of hedge fund activism. Section 3 describes the data and the methodology that is used in this research. Section 4 shows the results of the regression and in Section 5 the conclusions are drawn. Finally, Section 6 discusses the limitations of this research and some suggestions for future research on this topic. 3

2. Literature review The topic of hedge fund activism is frequently discussed in the literature. This literature review starts with a brief definition of hedge funds and hedge fund activism, followed by a summary of all empirical research on this topic and its outcomes. Afterwards there is a short justification of why this research adds to the existing literature. This chapter concludes with the formulation of the hypothesis. There is no exact definition of a hedge fund. According to Partnoy and Thomas (2007), hedge funds generally have four characteristics. First is that they are pooled, privately organized investment vehicles. Secondly, they are administered by professional investment managers. Thirdly is that they are not widely available to the public. Lastly, they operate outside of securities regulation and registration requirements. The last characteristic is applicable to hedge funds because they have a relatively small number of individual and institutional investors. However, hedge funds are obligated to hand in a certain filing with the Securities and Exchange Commission (SEC) when they become activist shareholders in a company and obtain a share of over 5% in the company. This so called Schedule 13D filing should also state what the purpose of transaction is under Item 4. According to Brav et. al (2008), the purpose can be divided in 5 categories: (1) General undervaluation/maximizing shareholder value, (2) Change capital structure, (3) Alter business strategy, (4) Sell the target company and (5) Improve Governance. These goals can be achieved by using several activist tactics. In order to give an impression, the categorized tactics from Brav et. al (2008) are summarized. Hedge funds can communicate with the board, seek board representation without confrontation, make shareholder proposals or publicly criticize the company, threaten to launch or actually launch proxy contests to replace the board, sue the company and take control of the company by for example with a takeover bid. These tactics serve to achieve one or multiple of the purposes mentioned above. Empirical research To address the short-termist concerns posed in various articles, researchers have made attempts to validate these claims empirically. Brav. et al (2008) investigated the returns to hedge fund activism and found that there was a short term positive abnormal return around the announcement data of 7% to 8%. They conclude that there is no reversal of this positive return by testing if there are negative abnormal returns in the two-year period after the hedge funds 13D filing, which was not the case. Boyson and Mooradian (2011) obtain similar results. Klein and Zur (2009) also find positive abnormal returns and conclude that hedge funds are successful in achieving the goals stated in the initial 13D filing. In 60% of the 4

hedge fund interventions, they achieve their goal. Brav, Jiang and Kim(2015) investigated the effect of hedge fund activism on productivity by looking at plant-level data. This rules out the possibility of a survivorship bias, in which divestiture of inefficient assets is not taken into account and therefore the results tend to be biased upwards. They concluded that productivity increased at the plant-level by capital reallocation, improving efficiency of assets in place and an increase in worker productivity. In the same year, Bebchuk et. al (2015) investigated the long term effects of hedge fund activism. Again, there is no empirical basis for the claims made by opponents of hedge fund activism. Following a five year time window after the 13D filing, Bebchuk et. al (2015) find no negative company performance. Instead, company performance improves every year after the intervention. Brav et. al (2016) also conducted a study about corporate innovation and hedge fund activism. They found that corporate innovation increases after hedge fund activism. Despite a drop in R&D expenditure, innovation output measured by patents and citations rose in the 5-year period after hedge funds became activist. They refute alternative explanations about causality by doing four different tests. Mean reversion, voluntary changes, stock picking and market responses are all rejected as alternative causes of the increase in corporate innovation. There have been several researches discovering that R&D expenditures drop after intervention. Though it might be different in the R&D-intensive industry, since their success is based on their innovative performance. Taking the R&D-intensive industry apart, this research may uncover the differences in activist strategies. As Brav et. al (2016) has revealed there is a positive effect of hedge fund activism on innovation, the R&D-intensive industry might benefit from this to an even larger extent than other companies. This could mean there is a difference in the increase in company performance. This has not yet been thoroughly investigated and makes this thesis relevant. Opponents of hedge fund activism have claimed that the operational changes sought by activist hedge funds are detrimental for companies long term performance. Strine (2010) expresses his concerns regarding activist shareholders who seek short term gains and how it is detrimental for long term performance. The hypothesis of this thesis is that the effect of hedge fund activism in R&D-intensive industries will be positive in the long run. As there have not been any empirical papers confirming the claim that hedge fund activism is bad for company performance, there is expected that hedge fund activism will not have a negative effect on the long term performance of companies. Instead, the effect of hedge fund activism might be having an 5

even larger positive effect on R&D-efficiency in the R&D-intensive industry. This effect is expected to be higher than possible negative long term effects due to hedge fund activism. 3. Data and Methodology The data that is required when investigating hedge fund activism consists of a set of activist events, combined with company data to measure company performance and to control for other factors. The data on activist hedge fund interventions is linked with the date the Schedule 13D was filed at the SEC. A Schedule 13D filing with the SEC is obligated for one who acquires a stake of over 5% in a company with the intent to force changes or to seek control over the target (Brav, et. al, 2008). Item 4 of the Schedule 13D requires disclosure of the purpose of the transaction. To give an impression of what this might look like, an example is given of Millennium Management LLC filing a Schedule 13D for its stake in Alexion Pharmaceuticals, inc. Item 4 of this filing is as follows. Item 4. Purpose of Transaction. The Reporting Persons are engaged in the investment business. In pursuing this business, the Reporting Persons analyze the operations, capital structure and markets of companies, including the Issuer, on a continuous basis through analysis of documentation and discussions with knowledgeable industry and market observers and with representatives of such companies (often at the invitation of management). From time to time, one or more of the Reporting Persons may hold discussions with third parties or with management of such companies in which the Reporting Persons may suggest or take a position with respect to potential changes in the operations, management or capital structure of such companies as a means of enhancing shareholder value. Such suggestions or positions may relate to one or more of the transactions specified in clauses (a) through (j) of Item 4 of Schedule 13D of the Exchange Act, including, without limitation, such matters as disposing of or selling all or a portion of the company or acquiring another company or business, changing operating or marketing strategies, adopting or not adopting certain types of anti-takeover measures and restructuring the Issuer s capitalization or dividend policy. The Reporting Persons employ the services of a number of portfolio managers, each of whom independently employs a separate and distinct trading strategy. A portion of the securities of the Issuer held by the Reporting Persons are held in accounts of the Reporting Persons managed by portfolio managers who engage in event-, risk- or merger-arbitrage or fundamental strategies. Except as set forth above, the Reporting Persons do not have any present plans or proposals that relate to or would result in any of the actions required to be described in Item 4 of Schedule 13D. Each of the Reporting Persons may, at any time, review or reconsider its position with respect to the Issuer and formulate plans or proposals with respect to any of such matters, but has no present intention of doing so. 6

As one might notice, it is not very specific what changes are sought. This is often the case, and that is why this research limits itself to the investigation of the effects of hedge fund activism after the filing only. Extensive research on the differences in the purpose of transaction is not in the scope of this thesis. Professor Alon Brav is a prominent researcher on the topic of hedge fund activism, as pointed out in the literature review. He has made a dataset of activist events from 1994 to 2014 and was willing to share his dataset for this research. This dataset contains over 4000 activist events and around 2800 companies which are subject to hedge fund activism. The date of activism is linked with the date the hedge fund filed the Schedule 13D. Because longterm performance will be measured, the time window of the filings used in this research will be set from 1994 to 2011. Other data about the companies was gathered from Compustat. The data from two years prior to the year of the hedge funds 13D filing and the 5 years after is used. Data on R&D expenditure was not widely available, 42,3% of the companies had missing data. The data was trimmed to the 1 st and 99 th percentile, since there were some obvious outliers. The table below shows some descriptive statistics of the data. Table 1 (1) (2) (3) (4) (5) VARIABLES N mean sd min max Age 15,791 8.521 6.291 0 27 Total Assets 29,493 1,132 2,708 2.133 27,278 Capital expenditures 28,615 46.24 122.6 0 1,286 Dividends 29,776 8.806 30.14 0 324 Dummy variable 1 if R&D-intensive 30,092 0.374 0.484 0 1 Leverage ratio 28,311 0.545 0.262 0.0816 1.340 Market value 29,492 799.4 1,819 1.506 17,067 Tobin s Q 28,890 1.184 1.230 0.0435 7.491 R&D expenditures 15,743 20.54 45.73 0 410.4 Return on assets 29,492-0.0507 0.250-2.059 0.347 Using this data, a panel regression is the appropriate method to measure the company s performance after hedge funds become activist. There is accounted for companyfixed effects and year-fixed effects. The control variables are the natural logarithm of Market Value, the natural logarithm of age and leverage (total liabilities divided by total assets). To measure any difference in long-term performance in R&D-intensive industries, a dummy variable is created, which is 1 if the company is R&D-intensive. To determine which industries are R&D-intensive, the same industries are chosen as in the research of Gerlach, Rønde, & Stahl (2009). Their article investigated labor pooling in R&D-intensive industries 7

and thus is not very related to the topic of this thesis. However, they have stated a list of industries which are perceived R&D-intensive, which will be used to construct the dummy variable. The industries are indicated by certain SIC codes. See the Appendix for an overview of the industries and their SIC codes. To estimate the long term effect of hedge fund activism, a dummy variable is created for every year after the intervention year, in the same fashion as the research of Bebchuk et. al (2015). This in turn can be used to measure the effect of hedge fund activism in R&Dintensive industries by combining the two dummy variables of R&D-intensive firms and each year after hedge fund activism. This will be done by making five interaction variables, for each year after intervention. To measure any overall difference between the R&D-intensive industry and other industries apart from long-term or short-term performance, the same regression will be done with only one dummy variable and one interaction variable. The dummy variable will be equal to 1 in the 5 subsequent years after the hedge fund intervention. The interaction variable will be this time dummy multiplied by the industry dummy. The initial sample in the time period of 1994 to 2011 consists of 1888 firms. 783 firms dropped out of Compustat due to bankruptcy, delistings or takeovers so the panel regression will be unbalanced. Below is a table covering the amount of companies which were available in Compustat for the year of intervention and the five years after. Table 2 Variable HFA year (t) t+1 t+2 t+3 t+4 t+5 Number of observations 1888 1657 1490 1347 1234 1105 Company performance will be measured by Return on Assets (ROA) and Tobin s Q. Here Tobin s Q is defined as the market value of equity divided by total assets. The regression equation with regard to long-term performance will look as follows: Tobin s Q = α + β 1 Leverage + β 2 Ln(Age) + β 3 Ln(Market Value) + β 4 Industry(dummy) + β 5 HFA(dummy) + β 6 HFA t+1 (dummy) + β 7 HFA t+2 (dummy) + β 8 HFA t+3 (dummy) + β 9 HFA t+4 (dummy) + β 10 HFA t+5 (dummy) + β 11 Industry(dummy) HFA t+1 (dummy) + β 12 Industry(dummy) HFA t+2 (dummy) +β 13 Industry(dummy) HFA t+3 (dummy) + β 14 Industry(dummy) HFA t+4 (dummy) + β 15 Industry(dummy) HFA t+5 (dummy) + ε i 8

The regression equation used to research the overall difference in the R&D-intensive industry and other industries will look slightly different: Tobin s Q = α + β 1 Leverage + β 2 Ln(Age) + β 3 Ln(Market Value) + β 4 Industry(dummy) + β 5 HFA(dummy) + β 6 HFA t+(1 5) (dummy) + β 7 Industry(dummy) HFA t+(1 5) (dummy) + ε i Note that in both regressions, an alternative regression will be made with Return on Assets as the dependent variable instead of Tobin s Q. All other variables remain the same. Apart from that, the research is extended by investigating the possible differences in activist hedge funds tactics in R&D-intensive firms. The changes in R&D expenditure, leverage, capital expenditures, assets, Tobin s Q, ROA and dividends are compared to companies which are not R&D-intensive. 4. Results This section discusses the main results of this thesis. First, the panel regression of the company s long term performance after hedge fund activism will be presented and discussed. After that, the additional analysis regarding accounting variables as well as performance measures between R&D-intensive and non R&D-intensive companies is disclosed. The table on the next page shows the results of the panel regression. Four regressions were done in total. The first two regressions include Tobin s Q (Q for short) as a dependent variable, the last two use Return on Assets (ROA). The difference between the first and second regression (third and fourth regression) is the inclusion of firm fixed-effects. The control variables are the natural logarithm of Market Value (LnMV), Age (LnAge) and Leverage. The Market Value is calculated by multiplying the share price with the shares outstanding. Age is calculated by subtracting the IPO (initial public offering) year from the panel data year. Leverage is calculated by dividing total debt over total assets. Then, there are two sets of dummy variables. The way they are set up is explained above the table. 9

Table 3 This table reports the coefficients and robust standard errors (between brackets) of the first linear panel regression. In the first two columns, Tobin s Q is the dependent variable and in the last two columns, Return on Assets is the dependent variable. The sample contains Compustat data from 1991-2016. The variable HFA(dummy) is a dummy variable which is 1 in the year the company is targeted by a hedge fund (the year the Schedule 13D was filed). The variable Sector(dummy) is also a dummy variable and is 1 if the company is in a R&D-intensive industry. The variables HFA(t+1) through HFA(t+5) are dummy variables and are equal to 1 in the year after hedge fund intervention for HFA(t+1), HFA(t+2) is equal to 1 two years after intervention, etc. The variables of interest are the interaction variables Sector(dummy)*HFA(t+i)(dummy). These variables are interaction variables and are 1 for companies in a R&D-intensive industry i years after intervention. Constants are excluded in this table. (1) (2) (3) (4) VARIABLES Q Q ROA ROA Leverage -0.372*** -0.689*** -0.364*** -0.146** (0.0894) (0.0713) (0.103) (0.0593) Ln(Age) -0.143*** -0.138*** 0.0815** 0.0552*** (0.0434) (0.0292) (0.0384) (0.0137) Ln(Market value) 0.520*** 0.419*** 0.0725*** 0.0568*** (0.0216) (0.0158) (0.0162) (0.00860) HFA(dummy) -0.0789*** -0.106*** 0.00672 0.00570 (0.0297) (0.0276) (0.0127) (0.0137) Sector(dummy) 4.028*** 0.801*** 0.0434-0.158*** (0.277) (0.0652) (0.185) (0.0301) HFA(t+1)(dummy) 0.00321-0.0165 0.00641 0.0184 (0.0350) (0.0321) (0.0114) (0.0122) HFA(t+2)(dummy) 0.0685* 0.0480 0.00859 0.0218* (0.0370) (0.0341) (0.0112) (0.0115) HFA(t+3)(dummy) 0.130*** 0.103*** -0.00878 0.00553 (0.0384) (0.0355) (0.0217) (0.0214) HFA(t+4)(dummy) 0.129*** 0.107*** 0.0309* 0.0431** (0.0368) (0.0341) (0.0187) (0.0179) HFA(t+5)(dummy) 0.127*** 0.112*** -0.00369 0.00398 (0.0387) (0.0360) (0.0142) (0.0135) Sector(dummy)* HFA(t+1)(dummy) 0.0411 0.0386-0.0307-0.0602 (0.0756) (0.0718) (0.0468) (0.0420) Sector(dummy)* HFA(t+2)(dummy) 0.0325 0.0291-0.0422-0.0629** (0.0782) (0.0741) (0.0355) (0.0314) Sector(dummy)* HFA(t+3)(dummy) 0.109 0.117-0.00883-0.0232 (0.0776) (0.0750) (0.0427) (0.0358) Sector(dummy)* HFA(t+4)(dummy) -0.0313-0.0378-0.0258-0.0498 (0.0681) (0.0657) (0.0415) (0.0342) Sector(dummy)* HFA(t+5)(dummy) -0.0361-0.0393 0.0228 0.00648 (0.0847) (0.0817) (0.0447) (0.0341) Year FE YES YES YES YES Company FE R-Squared YES 0.82 NO 0,71 YES 0.31 NO 0.06 Number of Companies 1,201 1,201 1,204 1,204 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 The combination of hedge fund activism and being an R&D-intensive company is captured by the interaction variables. These variables reflect the possible difference in the effect of hedge fund activism in R&D-intensive industries and non R&D-intensive industries. If one or more variables are significantly different from zero, this means that there is an additional effect in the R&D-intensive industry after hedge fund intervention. The dummy variable HFA has a negative coefficient in regression (1) and (2). This 10

can be interpreted as that the year the activist hedge fund targets a company, it is relatively undervalued in terms of Tobin s Q. What is perhaps a very striking result, is the coefficient of the dummy variable Sector in regression (1). In this regression, firm fixed-effects are included. This means all timeinvariant factors which are firm-specific are included to capture the effect of all the other variables properly. The effect of being a company in the R&D-intensive industry is very positively related to Tobin s Q. The differences in means between the R&D-intensive industry and other industries will be examined later in this chapter. Similar to Bebchuk et. al (2015) some of the dummy variables for hedge fund activism are positive and significant. The positive results for the dummy variables of three, four and five years after hedge fund activism are significant at the 1% level. This can be interpreted as that activist hedge funds increase companies long term performance. In the case of Tobin s Q as dependent variable, the positive long-term effect is even larger with firm fixed effects. However, the interaction variables are almost never significant. Only regression (4) shows a significant coefficient of the interaction variable at the 5% level. The Return on Assets in the R&D-intensive industry is negatively affected by hedge fund activism two years after hedge fund intervention. The coefficient of the dummy HFA(t+2) is positive and significant at the 10% level. This could mean that in general, company performance as measured by Return on Assets increases two years after intervention but in the R&Dintensive industry company performance decreases. Firm fixed-effects are not taken into account in this regression. The possibility that the decrease in performance in the R&Dintensive industry is due to firm-specific factors cannot be ruled out therefore. The insignificance of the interaction variables in the regressions can be explained in several ways. Hedge fund activism in the R&D-intensive industry was expected to differ from other industries, since activist hedge funds are capable of improving R&D-efficiency and the R&D-intensive industry could benefit from this to an even larger extent than other industries. Also it has been claimed that activist hedge fund act short-termist. The results of this research show that there is not enough reason to assume that the effect of hedge fund activism on long-term performance of a company is different in the R&D-industry, as compared to other industries. This can be caused by the two effects stated above, implying that they cancel each other out. An alternative explanation is that both effects are insignificant for the sample used in this thesis. This means that the R&D-efficiency, as measured by Brav et. al (2016), does not contribute to a firms long-term performance. The results of the second regression which made use of one dummy variable for all 5 subsequent years after hedge fund intervention are presented in the table on the next page. 11

Table 4 This table reports the coefficients and robust standard errors (between brackets) of the second linear panel regression. In the first two columns, Tobin s Q is the dependent variable and in the last two columns, Return on Assets is the dependent variable. The sample contains Compustat data from 1991-2016. The variable HFA(dummy) is a dummy variable which is 1 in the year the company is targeted by a hedge fund (the year the Schedule 13D was filed). The variable Sector(dummy) is also a dummy variable and is 1 if the company is in a R&D-intensive industry. The dummy variable HFA(t+1, t+2,..., t+5) is equal to 1 in the 5 subsequent years after hedge fund intervention. The interaction variable Sector(dummy)* HFA(t+1, t+2,..., t+5)(dummy) is equal to 1 for R&D-intensive companies in the 5 subsequent years after intervention. Constants are excluded in this table. (1) (2) (3) (5) VARIABLES Q Q ROA ROA Leverage -0.373*** -0.690*** -0.218*** -0.180*** (0.0894) (0.0713) (0.0213) (0.0174) Ln(Age) -0.140*** -0.135*** 0.0244*** 0.0277*** (0.0433) (0.0292) (0.00888) (0.00615) Ln(Market Value) 0.521*** 0.419*** 0.0567*** 0.0557*** (0.0217) (0.0158) (0.00432) (0.00332) HFA(dummy) -0.0764*** -0.104*** -0.00522-0.00699 (0.0293) (0.0273) (0.00696) (0.00662) Sector(dummy) 4.030*** 0.802*** -0.188*** -0.177*** (0.277) (0.0651) (0.0633) (0.0145) HFA(t+1, t+2,..., t+5)(dummy) 0.0825*** 0.0615** -0.00120-0.000351 (0.0303) (0.0277) (0.00608) (0.00570) Sector(dummy)* HFA(t+1, t+2,..., t+5)(dummy) 0.0280 0.0264-0.0154-0.0191 (0.0572) (0.0542) (0.0131) (0.0122) Year FE YES YES YES YES Company FE R-Squared YES 0.66 NO 0.20 YES 0.56 NO 0.14 Number of Companies 1,201 1,201 1,204 1,204 Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1 This regression shows insignificant results of the interaction variable, implying that R&D-intensive companies are not different from other industries in performance after being targeted by activist hedge funds. Even when the period is split up in two dummy variables (dummy equals 1 in year 1 and two, and another dummy equals 1 in year 3, 4 and 5) to separate short-term from long-term effects, there is still no significant interaction variable (note that the corresponding regression with these variables is not included in the tables). Summarizing it can be concluded that hedge fund activism has a positive effect on long-term company performance. The effects are not different in the R&D-intensive industry and this could mean that the R&D-efficiency gains as pointed out by Brav et. al (2016) are canceled out by the negative effects of hedge funds short-termism. It could also mean that both effects do not significantly affect company performance. To test differences between the R&D-intensive industry compared to other industries, several variables including accounting figures as well as performance measures are compared. Each year after hedge fund intervention the changes in these variables were calculated. The mean 12

in column (2) indicates the mean of industries which are not R&D-intensive. The mean reported in column (4) is the mean of the differences in each year for R&D-intensive companies. These are compared and using T-statistics, it can be concluded whether the changes in the variables significantly differ between the R&D-intensive industry and other industries. Table 5 The table below gives the values of the mean difference in the variable between the end of the year and the beginning of the year. One year after hedge fund intervention is referred to as t+1, two years after is referred to as t+2 and so on. Column (2) gives the mean of the changes for the sample of industries which are not R&Dintensive. Column (4) gives the mean of the changes for the sample of industries which are R&D-intensive. (1) (2) (3) (4) (5) (6) Mean Standard Mean T- VARIABLES N non-r&d Deviation R&D-int. Statistic Significance Change in ROA in t+1 1011 0,00794 0,468-0,0424-3,420 *** Change in ROA in t+2 921-0,0108 0,483-0,0228-0,754 Change in ROA in t+3 825 0,0144 0,469 0,016 0,098 Change in ROA in t+4 756 0,0152 0,469 0,00911-0,357 Change in ROA in t+5 679 0,00342 0,479 0,00911 0,310 Change in Tobin s Q in t+1 1038-0,0418 0,69 0,00735 2,295 ** Change in Tobin s Q in t+2 916-0,00218 0,746 0,126 5,200 *** Change in Tobin s Q in t+3 815 0,00699 0,663-0,0281-1,511 Change in Tobin s Q in t+4 746-0,0103 0,624-0,00174 0,375 Change in Tobin s Q in t+5 680 0,0367 0,645-0,0809-4,754 *** Change in capital expenditures in t+1 969-3,246 64,28-1,542 0,825 Change in capital expenditures in t+2 854 0,519 68,92 0,822 0,128 Change in capital expenditures in t+3 779 2,113 69,23 0,698-0,570 Change in capital expenditures in t+4 718 3,039 54,65 1,848-0,584 Change in capital expenditures in t+5 646 6,543 77,36 0,58-1,959 * Change in leverage ratio in t+1 999 0,0486 0,452 0,0277-1,461 Change in leverage ratio in t+2 884 0,0421 0,466-0,0109-3,382 *** Change in leverage ratio in t+3 792 0,00756 0,479-0,00387-0,672 Change in leverage ratio in t+4 724-0,0117 0,476-0,00884 0,162 Change in leverage ratio in t+5 652 0,0102 0,463 0,0553 2,487 ** Change in R&D expenditure in t+1 156-1,235 8,163 0,124 2,079 ** Change in R&D expenditure in t+2 131 0,345 4,461 0,885 1,385 Change in R&D expenditure in t+3 119-0,0645 5,943 2,424 4,568 *** Change in R&D expenditure in t+4 112-0,304 4,701 2,122 5,461 *** Change in R&D expenditure in t+5 98 0,638 4,44 0,948 0,691 Change in dividends in t+1 467-0,629 31,9 0,745 0,931 Change in dividends in t+2 411 1,943 22,87-3,795-5,086 *** Change in dividends in t+3 364 2,07 32,98 0,395-0,969 Change in dividends in t+4 337 1,179 29,69 0,183-0,616 Change in dividends in t+5 313 1,654 22,62 3,003 1,055 Change in total assets in t+1 1041 25,25 853,9 3,17-0,834 Change in total assets in t+2 935 34,66 434,8 20,63-0,987 Change in total assets in t+3 836 48,78 515,2 21,84-1,512 Change in total assets in t+4 766 61,52 620,6 17,88-1,946 * Change in total assets in t+5 688 84,25 588,6 38-2,061 ** Significance: * is at the 10%-level, ** is at the 5%-level and *** is at the 1%-level 13

The table shows some significant differences between the R&D-intensive industry and other industries. What is a particularly interesting result, is the difference in Tobin s Q over time. In the R&D-intensive industry there is a clear higher increase of Tobin s Q in the first two years after hedge fund intervention. Afterwards, there is a significantly bigger decrease in Tobin s Q in year five. This would confirm the activist claim that activist hedge funds are good for short-term performance and detrimental to long-term performance for the R&D-intensive industry. Looking at R&D-expenditure, the R&D-intensive industry has significantly higher increases than other industries. In the first year after intervention, R&D expenditure rises for the R&D-intensive industry and it decreases in other industries (significantly different from 0 at the 1%-level). For the rest of the years R&D-expenditure drops in almost every year in other industries, where it rises in every year in the R&D-intensive industry. This result could reflect a difference in activist hedge funds strategy between R&D-intensive industries and other industries. Note that these statistics do not necessarily mean causality. The effect is very unlikely to be due to year effects, since the 13D filings are spread out over 18 years. Though various other factors can cause these differences and future research should focus on how to mitigate bias in order to retrieve empirically correct results. A significantly bigger decrease in assets in the R&D-intensive industry could also be the cause of the higher value of Tobin s Q, as well as a higher increase in Tobin s Q in the R&D-intensive industry in the sample period in general, regardless of hedge fund activism. 5. Conclusion This thesis investigated the effect of hedge fund activism in the R&D-intensive industry on long-term performance and the differences of these effects with other industries. Using a sample of hedge fund activist events, data was gathered following a five year time-window after a hedge funds initial announcement of becoming activist by looking at Schedule 13D filings at the SEC. To estimate the long-term effects and possible differences between R&Dintensive companies and other companies, dummy variables were created in order to estimate the effects. Apart from the regressions, the differences between the course of various accounting figures and performance measures between the R&D-intensive industry and other industries in the five year time-window after hedge fund intervention were compared. The results of the panel regression indicate a significant improvement of companies long-term performance overall, but very few differences amongst the R&D-intensive industry. This does not confirm the hypothesis that the R&D-intensive industry has an even larger increase in company performance. It does confirm what we already know from other 14

research. The comparison in differences of various accounting figures and performance measures has shown different results. These suggest that there is a significantly higher increase in company performance in the short-term for R&D-intensive industries and significantly lower increase in the long run. This result supports the claim that activist hedge funds seek short-term gains to the detriment of long-term performance. Apart from the question if hedge fund activism has a different effect on companies long-term performance in R&D-intensive industries, it is investigated if activist hedge funds use different strategies in these industries. The results indicate a strong difference in R&Dexpenditure, showing a decrease in other industries and an increase in the R&D-intensive industry. Taking everything into consideration, it can be concluded that there are differences in companies long-term performance after activist hedge funds intervene between the R&Dintensive industry and other industries. The insignificance in the regression model is likely due to the absence of factors influencing company performance. Future research should therefore focus on investigating what factors to control for when measuring company longterm performance between the R&D-intensive industry and other industries. 6. Discussion Even though this thesis investigated something which has not been investigated yet empirically, there are some limitations towards the implications of this research. This thesis and research about hedge fund activism in general faces some difficulty regarding causality. If a research does not incorporate all factors influencing the dependent variable, the slope of the independent variable will not be accurate. In this discussion causality, biases and its possible solutions, and suggestions for further research will be discussed. Endogeneity is a problem when facing hedge fund activism research. Hedge funds know their ways in the stock market and make a substantial amount of return on their investments. It is hard to establish whether their returns can be attributed to their skill in stock picking or their shareholder activism. And even if it is not attributable to stock picking, is the activism the cause of the change in performance of a company? The outcomes in performance could also be reached had the shareholders made proposals in a non-active way and the company voluntarily implemented these changes. Apart from that, numerous factors can influence corporate performance. Researchers often tried accounting for this bias, many different approaches were used to diminish the bias in results. Both Boyson and Mooradian (2011) and Brav et. al (2008) cover possible endogeneity issues by matching activist targets with a peer which is not targeted by an activist hedge fund. This would not rule out the possibility of hedge funds being excellent stock pickers rather than their activism 15

being the actual cause of the improving performance. Their ability of picking the right company instead of their underperforming peer does not imply that activism induced the improved performance which was found in both the articles. In fact, this result could also confirm that hedge funds are excellent stock pickers. Another way of resolving endogeneity issues was done by Clifford (2008) and made use of Schedule 13G filings with the SEC. These filings are similar to 13D filings but they indicate a stake above 5% with passive investor purposes. Clifford made a control group of passive hedge funds investing in companies and estimated the difference. This thesis included firm fixed-effects in order to rule out other factors from influencing company performance. Moreover, survivorship bias often occurs when conducting researches using Compustat data. This is caused by delistings, mergers/acquisitions and bankruptcy. The company s which survive will be used for the research but one can see that this will not reflect the overall effect. For instance, if a research starts with 500 firms in year 1 and only has available data of 100 firms in year 5, this is not a proper reflection of what happened with the 500 firms. Another form of survivorship bias is that the assets of a company which do well tend to retain in the company whilst poorly performing assets are sold off during hedge fund activism. Brav et. al (2015) try to diminish this by looking at plant-level data. Unfortunately this research has some exposure to survivorship bias since Compustat data is used and due to the scope of this thesis, asset sales cannot be accounted for. This does not make the results negligible. They do provide interesting insights in the differences of hedge fund activism between R&D-intensive and non R&D-intensive firms. Another problem of hedge fund activism research is mean reversion. This phenomenon appears on the stock market, where stock prices tend to move towards the mean over time. Since hedge funds typically target undervalued firms, an increase in the stock price can be due to mean reversion rather than hedge fund activism. Several authors found some support that mean reversion is not the cause of the improved performance. Future research should focus on mitigating the methodological issues presented above. Apart from that, future research could try to find out how the effect of hedge fund activism on the long-term performance of R&D-intensive companies is different from other companies. For instance, this can be done by looking at how R&D-efficiency affects the company s long-term performance. Also companies which are not targeted by activist hedge funds can be used as a control group in order to see how effective hedge fund activism is in the R&D-intensive industry. 16

Reference list Al-Muslim, A. (2018, jan 2) Alexion Agrees to Work With Hedge Fund Elliott on Filling Board Seat. The Wall Street Journal. Retrieved from https://www.wsj.com/articles/alexion-agreesto-work-with-hedge-fund-elliott-on-filling-board-seat-1514933312 Bebchuk, Lucian A., Brav, Alon, Jiang, Wei, (2015). The long-term effects of hedge fund activism. Columbia Law Rev. 115 (5), 1085 1155. Boyson, Nicole M., Mooradian, Robert, (2011). Corporate governance and hedge fund activism. Rev. Deriv. Res. 14, 169 204 Brav, A., Jiang, W., Ma, S., & Tian, X. (2016). How does hedge fund activism reshape corporate innovation? (No. w22273). National Bureau of Economic Research. Brav, Alon, Jiang, Wei, Kim, Hyunseob, (2009). Hedge fund activism: a review. Found. Trends Finance 4, 185 246. Brav, Alon, Jiang, Wei, Kim, Hyunseob, (2015). The real effects of hedge fund activism: productivity, asset allocation, and labor outcomes. Rev. Financ. Stud. (forthcoming). Brav, Alon, Jiang, Wei, Partnoy, Frank, Thomas, Randall S., (2008). Hedge fund activism, corporate governance, and firm performance. J. Financ. 63, 1729 1775 Brav, A., Jiang, W., Partnoy, F., & Thomas, R. S. (2008). The returns to hedge fund activism. Financial Analysts Journal, 64(6), 45-61. Gantchev, Nickolay, (2013). The costs of shareholder activism: evidence from a sequential decision model. J. Financ. Econ. 107, 610 631. Gerlach, H., Rønde, T., & Stahl, K. (2009). Labor pooling in R&D intensive industries. Journal of Urban Economics, 65(1), 99-111. Klein, April, Zur, Emanuel, (2009). Entrepreneurial shareholder activism: hedge funds and other private investors. J. Financ. 64, 187 229. Lipton, M. (2013, March). Important questions about activist hedge funds. In Law School Forum on Corporate Governance and Financial Regulation (Vol. 9). Strine Jr, L. E. (2010). One Fundamental Corporate Governance Question We Face: Can Corporations Be Managed for the Long Term Unless Their Powerful Electorates Also Act and Think Long Term?. The Business Lawyer, 1-26. 17

Appendix R&D-intensive industries SIC code Description 28 Manufacturing - Chemicals and Allied Products 34 Manufacturing - Fabricated Metal Prdcts, Except Machinery & Transport Eqpmnt 35 Manufacturing - Industrial and Commercial Machinery and Computer Equipment 36 Manufacturing - Electronic, Elctrcl Eqpmnt & Cmpnts, Excpt Computer Eqpmnt 37 Manufacturing - Transportation Equipment 381 Manufacturing - Search, Detection, Navigation, Guidance, Aeronautical, and Nautical Systems and Instruments Manufacturing - Laboratory Apparatus And Analytical, Optical, Measuring, And Controlling Instruments 382 384 Manufacturing - Surgical, Medical, And Dental Instruments And Supplies 385 Manufacturing - Ophthalmic Goods 3861 Manufacturing - Photographic Equipment and Supplies 7371 Services - Computer Programming Services 7372 Services - Prepackaged Software 7373 Services - Computer Integrated Systems Design 7374 Services - Computer Processing and Data Preparation and Processing Services 7375 Services - Information Retrieval Services 873 Services - Research, Development, And Testing Services 18