The influence of hedge funds on performance through leverage

(1)

The influence of hedge funds on

performance through leverage

University of Groningen International Business & Management Specialization International Financial Management

University of Uppsala Economics & Business

Specialization International Financial Management

Name: Aroen Heller (S1547895) Date: 25 October 2011

Supervisor: Halit Gonenc Keywords

Hedge Funds, cumulative abnormal returns, leverage, capital structure, share price, hedge fund activism, target firms.

Abstract

(2)

2 | P a g e

Introduction

Hedge funds have been around for more than 60 years, and in the last decades they have become a strong financial player in the world. With an estimated $2 trillion worldwide, they manage around 5% of the total assets. Hedge funds are known especially for their influence and impact on companies and the potential profit they can make. But not all hedge funds make huge returns; in fact, most of them actually fail within the first year. In the last decades, hedge funds have become more actively involved in the companies that they invest in. This is because they are aware that greater involvement had a positive impact on the returns. It is clear to see that a hedge fund quite often has a seat on the board of their investments as part of their activism, but there are also other ways that hedge funds influence returns, of which most of them are not easily noted.

Hedge funds are long term investment funds that ‘hedge’ their investments for their clients, mostly with the use of leverage, for a compensation fee. In most cases, standard fund regulations and registration requirements do not apply for hedge funds. As a result, hedge funds are generally seen as big investment entities that only want to make high returns, no matter the cost. However, according to recent studies (Klein and Zur, 2009), ( Boyson and Mooradian, 2007), (Boyson and Mooradian, 2009), (Solarz, 2010), hedge funds have a positive impact for the average shareholder, in fact, according to their research, the hedge funds act as shareholder advocates and agents. This will lower agency costs and increase returns.

(4)

4 | P a g e Hwang (2010), Penman et al. (2007), Bhandari (1988) and Fama and French (1992), leverage has an evident effect on the share price of a firm.

Previous research states that hedge funds influence a target company’s performance in a positive way. Therefore, it is to be expected that hedge funds have a significant influence on their investments. Hedge funds should be able to increase a target firm’s performance by advocating for performance related changes. According to previous research, this can be done in numerous ways (Solarz, 2010); where capital structure activism is one of these ways. The capital structure method for influencing share prices has not been researched extensively; therefore it would be interesting to see if hedge funds actually use this technique and whether or not it works. With capital structure activism, the use and the change of leverage within the target firm is meant. But do hedge funds use the capital structures of the companies they invest in as a tool to influence performance? In this paper I explore the role of hedge funds when concerning leverage and changes in leverage of target firms. I analyze whether or not hedge fund activity, which in this case will be the changes in leverage, have had a positive influence on performance, i.e. if hedge funds, in one way or another, have increased the performance of a firms they invest in. As a result, the research question for this paper will be: Does hedge funds capital structure activism cause an increase

in performance for target firms?

This research uses an event window study to measure the performance, with an initial sample size of 239 target firms within a 10 year period (2000-2010). The short term performance will be measured by the use of cumulative abnormal returns (CAR) for a 5 day event window, and the long term performance will be measured by the change in Tobin’s Q for a 2 year period.

(5)

(6)

6 | P a g e

Literature and hypothesis development

Announcements of share filing

There have been multiple studies about the effects of initial share price filings by hedge funds, and this paper will also focus on this effect. This paper focuses, on target firms all over the world, with a majority of the firms situated in the United States or Europe. Brav, Jiang Partnoy and Thomas (2008), Clifford (2008), Klein and Zur (2009), Solarz (2010), Boyson and Mooradian (2007) and Boyson and Mooradian (2009) used this focus in their research for the United States. Their research found empirical evidence for abnormal returns after share filing. Clifford (2008) comes up with possible reasons for these abnormal returns. When there is a large filing, investor interest will increase for the target company. Investors will no longer base their investment decisions a fundamental analysis, but on news and rumors, which Clifford calls noise trading, a phenomenon in which rumors are more important than fundamentals. Kyle and Vila (1991) state that with the rumors of a takeover, the share price of the target company will increase. Brav et al. (2008) find it logical that firms will have higher returns, because the highly motivated and incentivized hedge fund managers will have a positive impact on a target firms’ share price. The hedge funds will also incorporate active shareholder monitoring, which will decrease agency cost.

Solarz (2010) found that investors do not mind the involvement of a hedge fund, especially an activist hedge fund. According to Solarz, hedge funds are classified as activist hedge funds when they announce any activist intent, which could be buyout offers, proxy fights, private financing, law suits or board seats. Since activist hedge funds outperform passive hedge funds with 3.8% or more on the long run, were the activists outperform the passive hedge funds by 18.4%. This is the reason why investors welcome activist hedge funds.

(7)

7 | P a g e reported 1.7% when using the (-2, +2) window. The remarkable thing is that Boyson and Mooradian used the same model twice for their research in 2007 as well as 2009. The first time they found a 9% abnormal return and the second time a 4.8% abnormal return, which is almost half of the first one.

In accordance with previous studies, it is expected that when an initial share filing of a substantial amount is done by a hedge fund, which is a minimum of 5%, the target share price will increase. This leads to the following hypothesis:

Hypothesis 1:

(8)

8 | P a g e

Leverage

The debt-to-asset ratio (leverage) and its effect has been widely studied in the last decennium. It has been used to measure the performance, the financial health as well as the expected returns (Fama and French 1992) of a firm. In this study, the debt-to-asset ratio will be used to determine the effect of the debt ratio on the share price of a firm. This study will also research the difference between returns of low leverage target firms compared to high leverage target firms.

(9)

9 | P a g e capital adjustments to smooth dividends. These financially constrained firms are riskier and earn higher expected returns. These studies also state that this increase in leverage will only result in an increase in share price until a certain point. When leverage is too high, the risks are too great and this will have an adverse effect on the share price. Welch (2004) looks at it the other way around, he believes that stock returns influence leverage. Poorly performing stocks have lower debt ratios. This can be switched however, when a low return means a low debt ratio, it can also be seen as a low debt ratio is related to a low stock return. Gomes and Schmid (2007) state that growth options are associated with the debt issues, and therefore, the fundamentals will be less risky of highly levered firms. According to most textbooks, who state that less risky firms should have lower returns, the highly leveraged firms should have lower returns. Their research, however, shows that higher leveraged firms have higher stock returns.

(10)

10 | P a g e more exposed than low cost firms. This is because it is not optimal for a high cost firm to reduce its debt to a level where the exposure will be equal to a low cost firm. In other words, they do agree with the fact that leverage increases risk, but because it increases risk, high risk firms lower their leverage in order to become less risky. In their research, George and Hwang (2010) used the three-factor model from Fama and French (1993). They found a negative relation between leverage and return, while Fama and French used this same model and found a positive relation. Solarz (2010) found that for activist hedge funds that have a higher abnormal return than passive hedge funds, leverage actually goes down, it goes up for passive hedge funds.

Hypothesis 2:

(11)

11 | P a g e

Hedge fund activism

A number of studies have been done whether or not there is a difference for the return when a hedge fund is perusing activism. Some of these studies are done by Brav, Jiang Partnoy and Thomas (2008), Clifford (2008), Klein and Zur (2009), Solarz (2010), Boyson and Mooradian (2007) and Boyson and Mooradian (2009). The empirical evidence of all of the above studies suggests that there is a higher return for hedge funds when they use activism as their strategy. Clifford (2008) finds a 8%-21% larger return. According to Boyson and Mooradian (2009), the more active and aggressive a hedge fund is towards the target firm, the higher returns will be. Brav et al. (2008) state that hedge fund activism is fundamentally important, although it remains poorly understood. Brav et al. also state that regulators and critics are not sure whether hedge fund activism will create value for hedge funds and investors or destroy it. According to Brav et al. hedge fund activism will create rather than destroy value due to the fact that hedge funds can act as informed monitors, which in term will lead to decreasing agency costs. Their findings, however, state that hedge funds do create value in the long term. This is supported by research done by Boyson and Mooradian (2008) and Klein and Zur (2008). Solarz (2010) find excess returns of 13.5% over a 2 year time period for active hedge funds and 1.5% for passive hedge funds.

(12)

12 | P a g e George and Hwand (2010), Johnson (2004), and Penman et al. (2007). This is perhaps reflecting heterogeneity even within the strategy (Solarz, 2010).

Since prior research states (Gomes and Schmid, 2007), (Livdan et al.,2009), (Welch, 2004), (George and Hwang, 2010), (Penman et al., 2007), (Bhandari, 1988) and (Fama and French, 1992), that either a low or a high debt-to-equity ratio will increase returns, it can be expected that hedge funds will try to change the debt-to-equity ratio, this depends on the hedge funds believe whether or not it should be increased or decreased. Either the hedge fund believes that a low debt-to-asset ratio or a high ratio is favorable. But actively managing and changing the ratio will according to research most likely result in higher returns.

Hypothesis 3:

Higher performance is expected from target firms with a change in the debt-to-asset ratio.

(13)

13 | P a g e

Research data

Hedge fund and target firm data

There are a few hedge fund databases available and for all of them a fee has to be paid in order to access the information. Therefore, for this paper, these hedge fund databases could not be used. The databases used in this paper in order to subtract hedge fund data is ZEPHYR and DataStream. ZEPHYR is a database providing deal information, from mergers, acquisitions and other transactions. From this database the deal information from the hedge funds was obtained. The fundamentals were obtained from DataStream; this database has extensive financial information of firms worldwide. Furthermore, publicly available sites with hedge fund information were used to collect and additional information

This paper has a focus on target firms with a minimum ownership of 5% by a hedge fund in the period of 1999 till present.

This results in a sample size of firms targeted by hedge funds of 239 firms. The data is obtained through ZEPHYR.

For the first hypothesis, the full 239 firms from the sample size were used. For hypothesis 2 and 3 only 128 firms were selected and used, this because of the lack of extra information available for the rest of the sample. The necessary extra information in order to calculate leverage, Tobin´s Q, return on equity, profit margin, A/ME, or A/BE was not available for all of these 239 firms. Some firms had just one year of a certain variable missing, due to the lack of information, only the firms with all the necessary information were used. This is order to minimize errors.

(14)

14 | P a g e

Methodology

Event study

This paper will focus on the announcement events of initial share purchases as well as the longer term performance effects. According to literature, Cumulative Abnormal Returns (CAR), are the best to use in this case. This is because CAR is an important variable; CAR is one of the easiest ways to see the impact of an event. In an event study, the focus is on firm specific events that have an impact on the price. According to MacKinlay (1997), event studies are useful because they reflect the immediate effect of an event on a share price. MacKinlay also states that there have to be taken certain steps in order to perform a successful study. An estimation of an event window has to be made, a model has to be chosen and the empirical test results have to be presented.

With multiple related studies also using event studies, Brav et al. (2008) Boyson and Mooradian (2009) and Klein and Zur (2009), this paper will also use this method in order to derive test results and conclusions.

Warner and Brown (1980, 1985) as well as MacKinlay (1997) discuss that the event window should be compromised of at least the event and the day after the event for the event window to be able to capture the effects of the event on the share price. According to MacKinlay (1997), event windows usually consist of eleven, five or three days. Remarkable is that most studies do not use these event windows that Mackinlay stated. A number of studies used a larger event widow (-20 to -30, +20 to +30), smaller event windows were also used (-2, +2). This paper will use the five day window (-2, +2), as multiple other studies also used this same window (Doukas, Gonenc and Plantinga, 2008 and Clifford, 2008).

Research variables

 Leverage (This is the debt-to-asset ratio)

 Tobin´s Q (long term debt plus market value equity divided by (long term debt plus book value equity)

 Return on equity (ROE), is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share.  Profit margin, is calculated as net Income before preferred dividends / net sales or

(15)

15 | P a g e  A/ME (Assets divided by market equity)

 A/BE (Assets divided by book equity)

These research variables are based in the models of Fama and French, 1992, and Solarz, 2010. The model of Solarz is technically an updated model of the Fama and French model. Leverage has been used in multiple research models, which were all based on the models of Fama and French(1992), most of these models included Tobin´s Q, as well as return on equity and profit margin (Solarz, 2010). The variables A/ME and A/BE were used in the Fama and French model. Fama and French used A/ME and A/BE as a form of leverage indication, whereas A/ME is the market leverage and A/BE is the book leverage. In their research, higher market leverage was associated with higher average returns, and higher book leverage was associated with lower average returns. Most models also use market capitalization, but since the target firms are from all over the world, and the market cap is mostly given in local currencies, this would not have been a viable variable. Therefore, the variables A/ME and A/BE were included.

For the second hypothesis, the dataset is split into two sub samples. This in order to make a distinction between low leverage target firms and high leverage target firms. The first sub sample is the group with high leverage, the second group being the low leverage sub sample group. In order to split the sample, the average debt to equity ratio was taken for the complete sample and all the target firms with an above average ratio were classified as high leverage firms and the target firms with a lower than average ratio were classified as low leverage firms.

(16)

16 | P a g e

Short term performance

This paper will be using the initial share filing announcements in order to calculate the abnormal returns. According to literature, Cumulative Abnormal Returns (CAR) can be best used in order to analyze these share filings. According to Brown and Warner (1980), and Brown and Warner (1985), abnormal returns of a security can be calculated by looking at the difference between the normal return and the actual return during the same period. In their research, Brown and Warner used the Market Adjusted Model to measure the abnormal returns, which will also be used in this paper. A ten year period will be used in the paper to find as many share filing announcement events as possible, 1999 till present will be the time period.

The stock exchange indices from the different countries where the target firms are listed in will be used in order to calculate the abnormal returns. The assumption will be made that there is a normal distribution for the underlying securities of the market indices.

This paper will use a 5-day period around the announcement date to the calculate announcement returns of the initial filings of hedge funds. The announcement returns are the cumulative abnormal returns (CARs) of the target firms from 2 days prior and 2 days after the announcement of the initial filing (-2, +2). In previous research on similar matters, the same method was used, however, the window event in most studies are different. Brav, et. al. (2008) uses a (-20, +20) window, whereas Boyson and Mooradian (2009) choose to use a (-25, +25) window, but also (0, 10) (Boyson and Mooradian, 2007). Klein and Zur (2009), used an even bigger window (-30, +30). The (-2, +2) event window was chosen for this paper, which was also used in the research papers of Doukas, Gonenc and Plantinga (2008) and Clifford (2008).

(17)

17 | P a g e Gonenc and Plantinga, 2008). Boyson and Mooradian (2009) also used a similar model to calculate CAR. Therefore, in this paper the same model will be used to calculate CAR.

In order to find the average cumulative abnormal returns for all target firms, the number of CARs will be divided by the number of events. This will give the average CAR for the entire event and will determine the outcome of hypothesis 1. This will result in the following formula:

With H0: = 0 H1: ≠ 0

For the second hypothesis, CAR will also be used as the measurement of performance, but the research variables will also be added to the mix. The formula used in order to measure the influence of the independent variables and to test hypothesis 2 is:

(

(18)

18 | P a g e

Long term performance

For hypothesis 3, the performance will be measured for the long term. Brav et al. (2008) and Solarz (2010) found that it is hard to measure holding periods for hedge funds, therefore an assumption will be made that the holding period in this paper will be at least one year. The long term change in Tobin’s Q will be calculated for a time span of one year before the event date till one year after the event date in this paper. The model in order to calculate this will look as follow:

(

(19)

19 | P a g e

Results

In this section, the target firm characteristics will be analyzed. This is done by looking at the correlations between the characteristics. As was done in the research of Boyson and Mooradian (2007), a Wilcoxon signed rank test will be performed on the data sets to test for differences in medians, as well as a t-test for the difference in means.

It was expected that the variables A/ME and A/BE are highly correlated with each other, since they both use assets and the difference between the book and market value. The variables ROE and Net Profit also had these expectations. The results of the correlations for the variables are in table 1, the same variables in this model were used by Fama and French (1992), and Solarz (2010). As can be seen in the results, the highest correlation is Tobin´s Q and Net profit. Other high correlations are Tobin´s Q and A/ME. As expected, ROE, Net Profit, A/ME and A/BE also have a high correlation. According to Landis and Koch (1997), it is advised to remove variables with correlations above the .5. As shown in table 1, there is no correlation above the .5 level. According to previous research, when the levels are above .5, the correlation is too high and the variable should not be used. Since none of the variables used in this model have a correlation higher than 0.5, the model will be used the way it was.

Table 1: Research variables correlations.

This table shows the correlations of the variables used in the research models. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=128 Tobin´s Q A/ME A/BE ROE Net Profit Leverage

(20)

20 | P a g e For hypothesis 2, two sub samples were made. One consisting out of high leverage target firms and the other out of low leverage target firms. In table 2, the Wilcoxon signed rank test shows that two out of the five variables are significant. The significance for these variables are with a P-value at the 5% level.

The results show that only 2 of the 5 variables are significant, CAR and A/BE. The findings about CAR are in contrast with previous research, low leverage firms have higher CARs. On the other hand, high leverage firms have a significant higher asset to book equity ratio.

Table 2: Wilcoxon Signed Rank Test, t-test and sample Means and Medians for hypothesis 2.

This table shows the characteristics of target firms one year before the event for hypothesis 2, with a high and low leverage sample. CAR is the cumulative abnormal returns of the 5 day event window. CAR is calculated as the individual stock returns minus the market return per day. The derived daily abnormal returns are then cumulated to calculate CAR. A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N = 128 High Lev (N=64) Low Lev (N=64) t-test (diff.) Z

(21)

21 | P a g e The third hypothesis also made use of two sub samples. One consisting out of target firms with a large change in leverage and the other out of target firms with a small change in leverage.

Table 3: Wilcoxon Signed Rank Test, t-test and sample Means and Medians for hypothesis 3.

This table shows the characteristics of target firms one year before the event, for firms with a large and small change in leverage. A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N = 128 Large Change

(N=93)

Small Change (N=35)

t-test (diff.) Z

Lev Change Mean 0.0820 -0.0003 _.139

Lev Change Median -0.1130 0.0000 _-1.245

A/ME Mean 2.2287 2.8397 _-.302 A/ME Median 1.5301 2.0404 .723 A/BE Mean 3.1631 4.2160 _-.252 A/BE Median 2.6176 2.7181 _.213 ROE Mean 9.5652 4.2197 _-.320 ROE Median 9.6900 6.7700 _.862 NPM Mean 5.4253 2.2797 _.493 NPM Median 3.7000 3.0000 _.769 Leverage Mean 0.2575 0.2847 _.147 Leverage Median 0.2192 0.3219 _-4.820***

(22)

22 | P a g e

Table 4: Logistic Regression Results full sample.

This table shows the importance of firm characteristics. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=128 B

Tobin´s Q -.046

A/ME -.007

A/BE 1.654

ROE -1.876*

Net Profit Margin 2.717***

Leverage -1.891*

The table above (4) shows the characteristics of the target firms that are important. The findings show that 3 out of the 6 variables are significant. Return on Equity, Net Profit margin and leverage. Return on equity is found not to be important as well as leverage. The Net Profit Margin on the other hand is of great importance. This indicates that the target firms tend to be very efficient, an increase in efficiency normally leads to an increase in profit margin.

(23)

23 | P a g e

Short term performance

In this section, the short term stock performances are shown for the target firms.

The stock performance is measured as cumulative abnormal returns (CAR) around the event date, the window size for the returns is 5 days (-2, +2) (Clifford, 2008). Higher abnormal returns are found significant for this event. With cumulative abnormal returns of 3%, the outcome of this test is in line with previous research. Research states that hedge funds cause abnormal returns for target firms stock surrounding the announcement date. When comparing with other research that also used the (-2, +2) window, the results of Clifford (2008) can be used. In his research, a 1.7% CAR was reported.

Table 5: Short term stock performance hypothesis 1.

This table shows the short term stock performance by mean and median comparison. CAR is the cumulative abnormal returns of the 5 day event window. CAR is calculated as the individual stock returns minus the market return per day. The derived daily abnormal returns are then cumulated to calculate CAR. A T-test is performed to the significance of the CARs. All data is winsorized at the 10% level. *, ** and *** refer to statistical

significance at the 10%, 5% and the 1% levels.

N = 239 Target

CAR Mean 0,029515***

CAR Median 0,018217

The results of table 5 are in line with research done by Brav, Jiang Partnoy and Thomas (2008), Clifford (2008), Klein and Zur (2009), Solarz (2010), Boyson and Mooradian (2007) and Boyson and Mooradian (2009) and was found to be significant at the 1% level.

(24)

24 | P a g e When adding leverage to the equation, test results show that target firms in the high leverage group have lower cumulative abnormal returns than their low leverage counterparts, which is shown in table 6. For high leverage firms CAR is 1.1%, whereas CAR for low leverage target firms is 2.7%. These findings are in contrast with previous research. According to text book models high leverage firms should have higher returns. More debt increases risk, which should increase return. George and Hwang (2010), state that this is actually not the case and low leverage firms have higher returns due to them being more risky to begin with.

Table 6: Short term stock performance hypothesis 2.

This table shows the short term stock performance by mean and median comparison. CAR is the cumulative abnormal returns of the 5 day event window. CAR is calculated as the individual stock returns minus the market return per day. The derived daily abnormal returns are then cumulated to calculate CAR. A T-test is performed to the significance of the CARs. All data is winsorized at the 10% level. *, ** and *** refer to statistical

significance at the 10%, 5% and the 1% levels.

N = 128 High Lev (N=64) Low Lev (N=4) t-test (diff.) Z

CAR Mean 0.0111 0.0266 _-.272

(25)

25 | P a g e Table 7 shows that return on equity has a significant negative relation with the cumulative abnormal returns, and net profit has a significant positive relation with CAR.

Table 7: Regression Results CAR.

This table shows relationship between the dependent variables and CAR. CAR is the cumulative abnormal returns of the 5 day event window. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=128 CAR Tobin´s Q -.001 A/ME .015 A/BE .118 ROE -.176* Net Profit .262** Adjusted R Square .033 F 1.860

(26)

26 | P a g e Table 8 shows the relation of the variables including leverage. As indicated before, return on equity still has a significant negative relation with CAR, and Net Profit Margin has a significant positive relation. This test shows that leverage has also a significant relation, which is as previous research stated, a negative relation. This shows that a higher leverage means a lower CAR.

Table 8: Regression Results CAR with Leverage.

This table shows the short term performance of target firms only and coefficient of the independent variables used in the model. CAR is the cumulative abnormal returns of the 5 day event window. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=128 CAR Tobin´s Q -.046 A/ME -.001 A/BE .163 ROE -.180* Net Profit .286*** Leverage -.172* Adjusted R Square .053 F 2.179**

(27)

27 | P a g e When we just look at the sub sample group with high leverage and test this group, another variable has a strong positive relation with CAR, A/BE. The relation of CAR with ROE has become even stronger negative in this case, and the relation with net profit margin has also become stronger.

Table 9: Regression Results CAR High leverage.

This table shows the short term performance of target firms only and coefficient of the independent variables used in the model with the high leverage sample. CAR is the cumulative abnormal returns of the 5 day event window. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=64 CAR Tobin´s Q -.090 A/ME -.013 A/BE .392*** ROE -.349** Net Profit .419*** Leverage -.160 Adjusted R Square .142 F 2.734**

(28)

28 | P a g e For the sub sample group with low leverage, there are no variables with a significant level. When looking at the variables, it is still interesting to see that A/ME is positively related to CAR, and in table 9 it was negatively related. Which is in contrast with previous research by Fama and French (1992). In line with this research, A/BE is negatively related with CAR and in table 9 it was positively related. Another note can be made about Tobin´s Q, the low leverage group has a positive relation with Tobin´s Q, whereas the high leverage group has a negative relation with Tobin´s Q.

Table 10: Regression Results CAR Low leverage.

This table shows the short term performance of target firms only and coefficient of the independent variables used in the model with the low leverage sample. CAR is the cumulative abnormal returns of the 5 day event window. Tobin´s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=64 CAR Tobin´s Q .093 A/ME .110 A/BE -.051 ROE -.187 Net Profit .185 Leverage .096 Adjusted R Square -.032 F .675

(29)

29 | P a g e

Long term performance

In this section, the post even performance of the target firms will be analyzed. The tests will conclude whether or not a change in leverage has influence on target a firm’s performance. As shown in table 11, target firms with a large change in leverage, the change in the Tobin's Q Mean is -0.015%. The target firms who only had a small change or no change in the debt-to-asset ratio have a 0.067% change in their Tobin's Q Mean. This indicated that target firms with a small change in leverage are less undervalued after this period whereas target firms with larger leverage changes are more undervalued.

Table 11: Post event performance.

This table shows the long term performance of the target firms by mean and median comparison. ΔTobin's Q is the change in Tobin's Q for year t-1 till t+1. Whereas Tobin’s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A T-test is performed to the significance of the CARs. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels. N = 128 Large Change (N=93) Small Change (N=35) t-test (diff.) Z ΔTobin's Q Mean -0.0115 0.0671 _-.063 ΔTobin's Q Median -0.1141 -0.0268 _-1.199

(30)

30 | P a g e When we take a look at the drivers on the longer term, the test indicate that A/ME has a significant negative relation with the change in Tobin's Q. This means that a high market leverage has a negative effect on Tobin's Q, whereas according to literature a high market leverage should have a positive effect on average returns.

Table 12: Post event Regression Results.

This table shows the long term performance of target firms only and coefficient of the independent variables used in the model. ΔTobin's Q is the change in Tobin's Q for year t-1 till t+1. Whereas Tobin’s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

N=128 ΔTobin's Q A/ME -.195** A/BE .115 ROE .010 Net Profit -.101 Leverage -070 Adjusted R Square .002 F 1.054

(31)

31 | P a g e When we also include the change in leverage, it can be stated that still only A/ME has a significant negative relation with the change in Tobin's Q. The same statements can be made as for table 12; the results indicate a negative long term performance.

Table 13: Post event Regression Results including leverage changes.

This table shows the long term performance of target firms only and coefficient of the independent variables used in the model. ΔTobin's Q is the change in Tobin's Q for year t-1 till t+1. Whereas Tobin’s Q is calculated as the (long term debt plus market value equity divided by (long term debt plus book value equity). A/ME is the total assets divided by market equity. A/BE is the total assets divided by book equity. ROE is the return on equity, calculated as Earnings per Share for the most recent fiscal year divided by the previous year’s book value per share. Net profit is the net profit margin, calculated as net Income before preferred dividends / net sales or revenues * 100. Leverage is the debt to asset ratio. Leverage is the debt to asset ratio. All data is winsorized at the 10% level. *, ** and *** refer to statistical significance at the 10%, 5% and the 1% levels.

(32)

32 | P a g e When only analyzing the group that has the small leverage change, the same conclusions can be drawn as for the whole group. The difference is that the variable A/ME has an even stronger negative relation with the change in Tobin's Q.

For the other variables, leverage and the ΔTobin's Q are highly negatively related. This means that firms with lower leverage have smaller changes. As stated before, the other variables lead to an indication of a negative long term performance.

Table 14: Post event Regression Results with the no change in leverage sample.

(33)

33 | P a g e For the group with large leverage changes, no used variable has a significant relation with the changes in Tobin's Q.

The variables do however have the same indication as in table 14, but on a non-significant level. There is only one variable that is different, leverage. Which is in contrast with table 14, were low leverage firms have smaller leverage changes, whereas higher leverage firms have larger changes according to table 15.

Table 15: Post event Regression Results with the change in leverage sample.

N=93 ΔTobin's Q A/ME -.115 A/BE .128 ROE .027 Net Profit -.067 Leverage .029 Leverage Change -.057 Adjusted R Square -.038 F .395

(34)

34 | P a g e

Conclusion

Previous research has given the expectations that hedge fund targets have positive short term cumulative abnormal returns (CAR) for the announcement event window. According to previous research, the long term performance for the same target firms should also be positive. This paper’s findings are in accordance with previous research for the short term; however, for the longer term the results contradict this same research. For the short term, a 5 day (-2,+2) window was used around the announcement event, for the long term this was a 2 year (-1,+1) window.

As the results show in this research, hedge funds initial share purchase announcements have a positive relation with CAR in the short-term, which concludes the acceptance of the first hypothesis. Target firms do achieve higher abnormal returns during the event window of the hedge fund share purchase announcement. Thus, my findings are in line with previous research.

In the short-term, target firms with low leverage have higher abnormal returns than target firms with high leverage. Both target firms with high and low leverage have abnormal returns, but for low leverage firms it is higher. This leads to the acceptance of the second hypothesis, a difference in leverage does affect CAR. When looking at CAR, the net profit margin has the biggest significant positive relation.

(35)

35 | P a g e Implications of this research are fairly clear for hedge funds, when they invest in a company, the short term returns will be most likely positive. When a hedge fund is an active investor it will result in even higher returns. When looking for suitable investments, they will have to keep in mind that a company with relatively low leverage and a high profit margin will also result in even higher short term returns. In the long term, however, this paper finds no evidence of a positive performance for target companies with hedge fund capital structure activism involvement.

Investors and target companies can also take these same results and use it for their specific needs. Investors should keep an eye on companies with hedge fund interests, and buy their stock as soon as the initial share filing announcement has been made, or before they expect a hedge fund to buy a certain stock. Again, companies with low leverage and a high net profit margin are expected to have a higher stock price increase during the event window. For companies it can be interesting to see that hedge fund involvement in their company will lead to a higher value in the short term, but are likely to result in a decrease in performance over time. Companies with low leverage and a high profit margin will be more interesting for hedge funds, this information can be useful when developing a strategy to either attract or repel hedge funds.

(36)

(37)

37 | P a g e

References

Bhandari, L., (1988), “Debt/Equity Ratio and Expected Stock Returns: Empirical Evidence”,

Journal of Finance, 43, p 507-528.

Boyson, N. M., Mooradian, R. M., (2007), “Hedge Funds as Shareholder Activists from 1994–2005”, Financial market trends, 92, p 87-104.

Boyson, N. M., Mooradian, R. M., (2009), “Intense Hedge Fund Activists”,

http://ssrn.com/abstract=1492641.

Brav, A., Jiang, W., Partnoy, F., and Thomas, R., (2008) “Hedge Fund Activism, Corporate Governance, and Firm Performance”, Journal of Finance, 63, p 1729- 1775.

Brown, S.J., and Warner, J.B., (1980), “Measuring security price performance”, Journal of

Financial Economics, 6, p 54-67.

Brown, S.J., Warner, J.B., (1985), “Using daily stocks returns”, Journal of Financial

Economics, 14, p 3-31.

Clifford, C.P., (2008), “Value creation or destruction? Hedge funds as shareholder activists”,

Journal of Corporate Finance, 14, p 323-336.

Coval, J.D., Moskowitz, T.J., (1999), “Home bias at home: Local equity preference in domestic portfolios”, Journal of Finance, 54, p 2045-2073.

Doukas, J., Gonenc, H. and Plantinga, A., (2008), The Gains from Private Acquisition Deals, working paper, University of Groningen.

Fama, E. F., French, K. R., (1992), “The Cross-Section of Expected Stock Returns, Journal of

Finance, 47, p 3427-3465.

George, T., J., Hwang, C., Y., (2010), “A Resolution of the Distress Risk and Leverage Puzzles in the Cross Section of Stock Returns”, Journal of Financial Economics, 96, p 56-79.

Gomes, J., F., Schmid, L., (2007), “Levered Returns”, http://ssrn.com/abstract=966076. Johnson, T.C., (2004), “Forecast Dispersion and the Cross Section of Expected Returns”, The

Journal of Finance, 59, p 1957-78.

Jong, A., Verbeek, M., Verwijmeren, P., (2011), “Firms’ debt-equity decisions when the static tradeoff theory and the pecking order theory disagree”, Journal of Banking and Finance, 35, p 1303-1314.

Kyle, A.S, Vila, ., J.L., (1991), “Noise trading and takeovers”, The RAND Journal of

(38)

38 | P a g e Landis, J.R., and Koch, G.G., (1977), “The Measurement of Observer Agreement for

Catergorical Data”, Biometrics, 33, p 159–174.

Lewis, K.K., (1999), “Trying to explain home bias in equities and consumption”, Journal

of Economic Literature, 37, p 571-607.

Livdan, D., Sapriza, H., Zhang, L., (2009), “Financially Constrained Stock Returns”, Journal of

Finance, 64, p 1827-1861.

MacKinlay, C. (1997) "Event Studies in Economics and Finance", Journal of Economic

Literature, 35, p 13-39.

Modigliani, F., Miller, M., (1958), “The Cost of Capital, Corporation Finance, and the Theory of Investment”, American Economic Review, 48, p 261-297.

Partnoy, F., Thomas, R., (2006), “Gap Filing, Hedge Funds, and Financial Innovation”,

Financial Analyst Journal, 64, p 45-61.

Pearson, R., Altman, K., (2005), “Hedge Funds and Shareholder Activism”, The Corporate

Governance Advisor, 14, p 25-27.

Penman, S., H., Richardson, S., A., Tuna, I., (2007), “The Book-to-Price Effect in Stock Returns: Accounting for Leverage”, Journal of Accounting Research, 45, p 427-467.

Solarz, S., (2010), “Stock Picking in Disguise? New Evidence that Hedge Fund Activism Adds Value”, The Michigan Journal of Business, 3, p 101-160.

Teo, M., (2009), “The geography of hedge funds”, The Review of Financial Studies, 22, p 3531-3562.

Tesar, L. L., Werner, I.M., (1995),” Home bias and high turnover”, Journal of International

Money and Finance, 14, p 467-493.

The influence of hedge funds on performance through leverage