The determinants of recent market reactions to tender offer share
repurchase announcements in North-America
Steffan Taanman 10819568
Economie en Bedrijfskunde Finance and Organization Supervisor: Evgenia Zhivotova 25-6-2018
2
Statement of Originality
This document is written by Steffan Taanman 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
In this thesis, the focus will be on the recent short-term market reaction to share repurchase tender offer announcements in the U.S. and Canada. Fixed price and Dutch auction tender offers will be considered. I find an average CAR with an event period of three days of 7.40%, which is significantly bigger than zero. My regression results indicate that the repurchase premium and the book to market ratio are both significant positive determinants of the CAR. These results are in line with the signaling theory and the undervaluation theory. I don’t find any evidence for the agency cost of free cash flow theory, neither a significant difference between the market reaction to fixed price tender offers and Dutch auction tender offers.
4 Section 1: Introduction
With the increase of share repurchase programs since 1980, it has become more important to look at the market reaction to such a program, as this is often directly observable when looking at the post-announcement returns. Namely, the returns tend to increase after a repurchase announcement is made. It is not strange that a repurchase announcement leads to a clear market reaction because a repurchase has a lot of important decisions in it; it’s an investment decision, a cash distribution decision, a change in the capital structure and a change in the ownership structure at the same time. I think it’s therefore interesting to examine what factors determine the share price increase.
In my thesis, I focus on the short-term market reaction. There are two repurchase strategies that lead to the strongest signal to the market; the fixed price tender offer and the Dutch auction tender offer. These strategies and the difference in market reactions will be examined further in this thesis. I look at observations from North-America; the buyback market is the biggest in this part of the world. Because most earlier research contains observations from 20th century, I want to compare my results from recent market reactions with those older findings. This in short leads to the following research question: What are the determinants of the short-term market reaction after a share repurchase tender offer announcement in the United States and Canada during 1996-2016 and which of the two tender offer strategies gives the strongest signal to the market?
I try to answer this question with the use of two methods. I use the Thomson One database to collect 54 observations. I do an event study where I first calculate the
cumulative abnormal return (CAR) of every observation. The CAR is calculated with the use of the market model, taking an estimation window of [-125,-7] and an event period of [-1, +1] to look at the short-term market reaction. Then I test if the CAAR (the average CAR) is significantly different from zero. This will be done with the use of a t-test. After this test, I will use the calculated CAR as the dependent variable in multiple regressions. The
independent variables are used to explain what factors determine the CAR after the repurchase announcement. These explanatory variables are related to the signaling theory and the agency cost of free cash flow theory. I also use a dummy variable to take the difference between fixed price and Dutch auction tender offers into account.
I find a CAAR of 7.40%, which is significantly different from zero with a p-value less than 0.001, indicating a strong result. The regression results indicate that the repurchase
5
premium and the book to market ratio significantly determine the short-term CAR of a firm in a positive way, with both a p-value less than 0.05. These results support the ideas of the signaling and the undervaluation theory. The agency cost of free cash flow theory is not supported by my regression results. The coefficients of the FCF-variable and the ROA-variable are both far from significant. I conclude that this theory doesn’t explain the market reaction to share repurchase announcements in my sample. My results do not indicate a significant difference between the fixed price and Dutch auction tender offers as well, which may be caused by the small sample size.
The rest of my thesis contains the following. In Section 2, I provide an overview of literature about share repurchase methods and the two theories considered in my thesis. Section 3 describes the event study, the regression model, my hypotheses and how they are tested. In Section 4, I present what databases are used to collect all the data of every
observation. It also contains some descriptive statistics. Section 5 describes the results of the t-test and the regression results. Finally, in Section 6, I provide a brief summary and
conclusion, and I discuss the biggest limitations of my research.
6 Section 2: Literature Review
2.1 Share Repurchase
A share repurchase is practically nothing more than a buyback of a firm’s own outstanding shares. Vermaelen (2005) says that a repurchase has got more in it than it seems. It actually captures four corporate governance mechanisms: It’s an investment decision, a payout mechanism, a capital structure decision and an ownership structure decision in one. A firm thus makes a couple of important decisions before it reaches the decision to make a share repurchase. First, a firm invests in the buyback of it’s own shares instead of making another investment. This could say something about the investment decisions of a firm. Second, a firm distributes their cash to shareholders through a repurchase instead of paying out a dividend. Third, a repurchase will change the capital structure of the firm; it decreases equity and therefore leads to an increase in the leverage ratio. Finally, a repurchase leads to a change in the ownership structure, because managers now own a higher percentage of the total shares outstanding (assuming that managers don’t tender their shares or sell less than other shareholders). For investors, these are quite important things to account for. This is why it is not strange that a repurchase announcement often leads to a reaction of the market. The determinants that have an impact on this reaction will be examined further in this thesis.
Before 1980, repurchase activity was very small in the U.S., and even
discouraged/illegal in other countries. Since then, the number of buybacks has increased in the last decades. For example, data from the U.S. shows that the total number of
repurchases has increased from 217 buybacks in 1985 to 1723 buybacks in 2003 (Vermaelen, 2005). A repurchase has thus become a more popular way to distribute cash to
shareholders.
The method that is mostly used is the open market repurchase strategy, which accounts for more than 80% of all the repurchases (Vermaelen, 2005). Tender offers are much less frequent strategies, with fixed price tender offers dominating the total amount of tender offers. The different repurchase methods will be discussed below.
7 2.2 Repurchase Methods
In this section, I will give a short overview of the most important repurchase methods with their characteristics.
2.2.1 Fixed Price Tender Offer
A fixed priced tender offer is the most traditional way to make a share repurchase. The company makes the offer to repurchase a specific amount of shares for a specific price, and announces a specific expiration date. This specific price is often higher than the market price, the repurchase premium, making it more attractive for shareholders to tender their shares. The repurchase premium is assumed to be an important factor in explaining the signaling theory, which will be discussed later in this section. The company commits itself to repurchase shares against this specific price announced. The shareholders should then decide whether they are willing to sell their share(s) for the indicated price. The company has to repurchase all shares tendered when the amount of shares tendered is less than the amount of shares sought. When there are more shares tendered than shares sought, the company can decide to buy back any number of shares within the target amount and the amount of shares tendered (Comment & Jarrell, 1991).
2.2 Dutch Auction Tender Offer
When a firm chooses for a Dutch auction tender offer, it should also indicate a specific number of shares sought, which is the same as with a fixed price tender offer. On the other hand, the indication of the repurchase price is something different when compared to the fixed price tender offer. A firm now indicates a specific price range instead of one certain repurchase price. The minimum of this range is often a few percentage points higher than the market price, the maximum of the range is quite similar to the repurchase price chosen when a fixed price tender offer is made. After the announcement of the offer is made, shareholders should decide if they are willing to tender their share(s), and they should also think of what is an acceptable price for them (within the specific price range) to sell their share. The shareholders inform the company about their decisions, and the firm then finds out what is the minimum price that will lead to the repurchase of the indicated amount of shares sought. This price will then be used (Persons, 1994).
8
The repurchase price used in a Dutch auction tender offer is typically lower than the repurchase price used in a fixed price tender offer (Comment and Jarrell, 1991). Because of this essential difference between the two tender offer strategies, it is interesting to look at the different reactions of the market to the tender offer announcements.
A firm may have different motives to choose for a Dutch auction tender offer instead of a fixed price strategy. First and most obvious, because of the lower repurchase price, it is less costly for the firm to use the Dutch auction strategy. If a firm just has the objective to get rid of pessimistic shareholders and lower the amount of shares outstanding, the Dutch auction is a relatively cheap way to achieve this because pessimistic shareholders are the most likely to tender their shares for the lowest price (Vermaelen, 2005). At the other hand, if the firm wants to give a strong signal to the market, a fixed price tender offer is a better strategy because now the insiders determine all the trade conditions and it’s therefore more informative for outsiders (Comment & Jarrell, 1991). Because the higher repurchase price, the fixed price strategy is also more costly for the firm, making it an even stronger signal.
2.3 Other Repurchase Strategies
While the focus of this research will be on the strategies discussed above, here is a short overview of the most imported other repurchase methods and the reasons why I won’t use these. First and most common, the open market repurchase. When looking at historical data, open market repurchase announcements accounted for more than 90% of all the repurchase programs in the period 1984-2000 in the United States (Grullon & Michaely, 2004). In an open market repurchase, the firm announces that it wants to repurchase a certain amount of shares for a certain repurchase price within a certain time period. The crucial difference here between the tender offers and the open market repurchase is that the announcement doesn’t mean that the firm commits itself to the repurchase. The firm just has the option to buy back the shares, it’s not an obligation (Vermaelen, 2005). This is the reason why the announcement of an open market repurchase gives a weaker signal to the market than a tender offer. According to Ikenberry et al. (1995), the average reaction of the market on such an announcement was only 3.5%, whereas Grullon and Michaely (2004) report that the average announcement return of a Dutch auction was 7.7% and of a fixed price tender offer was 11.9%. Because of this smaller observed return, only the market reaction to
9
Private or targeted share repurchases are used by firms when they want to take a number of shares from a big shareholder. This strategy is often used when firms are threatened by a hostile takeover. In that case, it’s usual that a repurchase premium is paid by the firm to get their shares back (Vermaelen, 2005). In other situations, repurchases could be made at a discount, for example when large investors just want to exit and take enough with a repurchase price that is lower than the market price. Because this strategy is often not used for giving a signal of undervaluation, these announcements will not be considered in this thesis.
It is also possible for a firm to make a repurchase using derivatives. This is called a synthetic repurchase. This strategy could be beneficial when a firm believes that its share is undervalued but does not have a lot of excess cash to buy back those shares. At the other hand, this kind of strategy is a commitment to the firm in contrast to the more flexible open market share repurchase strategy (Vermaelen, 2005). In this thesis, these announcements will not be used too.
2.3 Theories
There are two theories which are the underlying reasons for doing this research. They say something about the expected market reaction to a share repurchase announcement. These theories will be discussed below.
2.3.1 Signaling Theory
The signaling theory says that the insiders of a firm give an informative signal to the
outsiders by announcing a share repurchase. First, managers may want to let the outsiders know that they think their share is undervalued. A share is undervalued when the market underestimates the expected cash flows or overestimates the risks a firm faces or will face (Vermaelen, 2005). If the outsiders believe the management, it will probably lead to a share price increase.
The crucial assumption of the signaling theory states that the signal has to be costly for a firm to be a credible signal to the market. With a repurchase, a firm increases its leverage and this has higher bankruptcy costs as a consequence. This could lead to serious damage for firms that are already in some financial trouble, but this would not be very dangerous for stronger firms because the probability that they will go bankrupt is not very
10
high (Vermaelen, 2005). Making the repurchase more costly (or increasing leverage) should therefore lead to a stronger signal. If a repurchase is costly for a firm depends on a couple of things. First, a high repurchase premium leads obviously to a more costly repurchase for the firm. The firm signals that it wants to and is capable of paying more for a share than the market states the share is worth, which is also a signal of strength and trust in the future. Second, buying back a huge amount of shares is also more costly than buying back a small amount of shares. Third, if the management has no reason to believe that the share is undervalued (in other words, they are lying), it leads to a false signal and the share price will not increase. The costs will then be higher for the management when they own a big stake in the company, which also leads to a stronger signal (Vermaelen, 2005). Unfortunately, the information about managerial ownership at the moment of the announcement is not
available and this third assumption is therefore not discussed further in this thesis.
There has been found evidence for the signaling theory in many studies. Comment and Jarrel (1991) for example find that Dutch auctions give a weaker signal of undervaluation because of the lower repurchase premium and the less shares sought, which both are important factors in determining the expensiveness of a repurchase. Vermaelen (1981) tries to test the signaling theory with comparing tender offers with open market repurchases. He finds that the CAR [-1,0] of firms using a tender offer is substantially higher (14.14%) than firms using the open market repurchase (3.37%). This is consistent with the theory; the signal of an open market repurchase is expected to be weaker because of the lower repurchase premium and because it is an option instead of a commitment for the firm after the announcement.
2.3.2 Agency Cost of Free Cash Flow Theory
This theory is about possible agency costs arising from excess cash held by a firm. If a
manager has the opportunity to expend excess cash, he can do it in a way that isn’t good for the firm, for example by investing in empire building or making negative NPV investments because of bad investment opportunities. It would be better for shareholders that the manager distributes the amount of cash to his shareholders. This can be done with paying out a dividend, but this isn’t a very strong statement because the manager can reduce the dividend in the future. With taking on debt for the repurchase of shares, the manager promises to pay out future cash flows so that he can’t waste it with bad investments. In this
11
way, a share repurchase is an attractive way of cash distribution for shareholders (Jensen, 1986). Because the risk of wasting the excess cash by managers is bigger for firms with a large free cash flow and a small return on assets (which implies few good investment opportunities or low profitability), the hypothesis of this theory states that these firms are more likely to make a share repurchase (Grullon & Michaely, 2004). Dittmar (1999) found evidence that firms with bigger free cash flows make a share repurchase more often than firms with smaller free cash flows.
Although a share repurchase is a promise to distribute the excess cash, the debt that has been taken on also has other consequences. Debt can encourage managers to do their best because of the threat caused by the possible failure of paying back. This motivating device can make an organization more efficient, and could contribute to the positive reaction of the market to a share repurchase announcement (Jensen, 1986).
Nohel and Tarhan (1998) did some research on tender offers to test the free cash flow hypothesis. They found that firms with a low market to book ratio, which implies low growth opportunities, face a higher short term CAR than firms with a higher market to book ratio. This indicates that the post-announcement returns are higher when investors think that the risk of agency costs of a firm is high, which supports the theory. Grullon and Michaely (2004) also found evidence for the free cash flow theory. In their research about open market repurchases, they found that the reaction of the market was stronger to announcements of firms that are more likely to overinvest (firms with a larger free cash flow and low growth opportunities).
In this research, I will also focus on these hypotheses. I want to look if my findings about recent reactions correspond with prior research.
12 Section 3: Hypotheses and Methodology
This research contains two parts. The first part is an event study; there will be considered if there exists a significant market reaction to a share repurchase announcement. The second part focuses on the factors that cause this reaction. The theories discussed in section 2 will be tested with different variables in a multiple regression.
3.1 Event study
The event study will be used to determine if there exists a significant reaction of the market to the share repurchase announcements. First, the cumulative abnormal return (CAR) has to be calculated to do this kind of test. To get to the CAR, a couple of steps have to be taken. The CAR is nothing more than the accumulation of the abnormal returns during the event period. The abnormal return (AR) is defined as the return during the event period (R) minus the normal return (NR):
(1) ARit = Rit – NRit,
where i stands for the observation and t stands for time.
The returns during the event period are simply collected with the use of the databases. The normal returns serve as a benchmark to compare with the return during the event period. The expected returns are calculated with the market model. First, the market model is estimated:
(2) Rit = αi + βiRmt + εi, with t =[-125, -7]
I chose to use an estimation period of about half a year, which comes down to the period [-125, -7]. I think half a year is enough to give an accurate view of the connection between the return of the company’s share and the return on the market index. After this regression, the normal returns for the event period are calculated:
(3) NRit = 𝛼̂i+ 𝛽̂iRmt, with t = [-1, +1]
13
The event period [-1, +1] is used because I will only look at the short-term
post-announcement reaction of the market. This period is also used in most literature about this topic.
After collecting Rit and NRit, the calculation of ARit can be done with formula 1. Accumulating the abnormal returns gives us the CAR:
(4) CARi =∑ 𝐴𝑅𝑖𝑡𝑡1𝑡2
Because there will be looked at different observations, the mean of the CAR (CAAR) is calculated to test if this significantly deviates from zero:
(5) CAARi = 𝑁1 ∑𝑁𝑖=1𝐶𝐴𝑅𝑖,
where N stands for the number of observations.
The CAAR will be used in the following t-test to look if this significantly deviates from zero:
(6) T-value =
√𝑁
𝐶𝐴𝐴𝑅 𝑠≈ 𝑁(0,1)
where s = √ 1 𝑁−1∑ (𝐶𝐴𝑅𝑖 − 𝐶𝐴𝐴𝑅) 2 𝑁 𝑖=13.2 Regression Formula and Hypotheses
In the previous section, I discussed how the CARs of the different observations will be
obtained. The CARs will be used as dependent variable in a multiple regression model. In this way, we can look at the different factors that influence the post-announcement returns. I will discuss each variable and the expectations about the regression in this section. The regression formula is:
CAR[-1, +1] = α + β1Premium + β2SharesSought + β3Book/Market + β4FCF + β5ROA + β6FirmSize + β7DummyFixedPrice + ε
14
3.2.1 Signaling Theory
The first three variables in the regression formula will be used to test whether the signaling theory predicts the reaction to a repurchase announcement in the used sample. The most important assumption of the signaling theory is that the strength of the signal depends on how costly it is for a firm to make this signal. The variables ‘Premium’ and ‘Shares Sought’ will test this assumption. The variable ‘Premium’ obviously stands for the repurchase premium that’s offered by the firm. It is calculated as a percentage of the closing market price at the trading day prior to the announcement. The hypothesis is that a higher repurchase premium will lead to a stronger signal and therefore leads to a more positive reaction of the market; in other words, that β1 is bigger than zero.
The variable ‘Shares Sought’ doesn’t need a big explanation as well. It accounts for the number of shares offered to repurchase as a percentage of the total shares outstanding on the announcement day. The more shares sought, the more costly it is for a firm to make a repurchase, so the hypothesis is that this variable is also positively related with the CAR (β2 is bigger than zero).
As discussed in Section 2.3.1, a firm often announces a share repurchase because the management thinks that its share is undervalued. If this isn’t a false signal, it should lead to a positive reaction of the market. This is part of the signaling theory and is also known as the undervaluation theory (Vermaelen, 2005). The variable ‘Book/Market’ refers to the book to market ratio, which is the book value of the firm divided by the market value of the firm. This measures how much a share is undervalued; the higher this ratio, the more a share is undervalued. The assumption is therefore that the book to market ratio is positively related with the CAR (β3 is bigger than zero).
3.2.2 Agency Cost of Free Cash Flow Theory
The fourth and the fifth variable are used to test the agency cost of free cash flow theory. As discussed in section 2.3.2, the theory says that firms with a lot of excess cash and few good investment opportunities are more likely to face agency costs, for example because of the risk of making negative NPV investments. Announcing a tender offer repurchase is the same as promising the distribution of cash to shareholders. Shareholders and other people
involved will be satisfied with such a decision, which will lead to a positive reaction of the market according to this theory. The variable ‘FCF’ stands for the amount of excess cash a
15
company possesses according to the financial information of the fiscal year prior to the announcement year. It is calculated by subtracting capital expenditures from the cash flow from operating activities, and then dividing it by the total assets for scaling purposes. The hypothesis about this variable is that the bigger the free cash flow of a firm, the more positive the market reaction to a share repurchase announcement.
The variable ‘ROA’ stands for the return on assets, which is calculated by dividing the net income by the total assets, both according to financial information of the prior fiscal year. The ROA measures the profitability of a firm with the assets taken into account and works in that way as a proxy for agency problems. A low ROA means low profitability, which in turn could lead to more agency problems. The bigger the possible agency problems, the better it would be for stakeholders that the excess cash is distributed to shareholders (by making a repurchase), so the more satisfied these stakeholders will be. This lead to the assumption that the CAR is negatively related to the ROA (β5 is smaller than zero).
The variable ‘Firm Size’ acts as a control variable in this regression model. This is measured as ln(Total assets). The natural logarithm is used to account for the big differences between observations. Prior evidence showed that firm size also has a relation with the post-announcement returns; the smaller the firm, the higher the post-announcement returns (Vermaelen, 2005). This is because a bigger company is followed more carefully and by more people than a small firm. This leads to the existence of more information
asymmetry between the insiders and the outsiders of a small firm, which creates a bigger reaction to a share repurchase announcement. The expectation is therefore that firm size is negatively related with the CAR, which means β6 should be smaller than zero. Because firm size is expected to be related to other explanatory variables as well, I take firm size as a control variable to prevent an omitted variable bias.
3.3.3 Differences between the tender offer strategies
Finally, the variable ‘DummyFixedPrice’ is used to account for the differences between the fixed price tender offers and the Dutch auction tender offers. The fixed price tender offer is expected to give a stronger signal to the market, because the insiders give more information to the outsiders with a specified price. This price is often higher than the price paid in Dutch auctions. Because I could only collect the dataset of 17 fixed price tender offers (out of 54 observations in total), this comparison is limited.
16 Section 4: Data Description
I’ve used a couple of databases to get to the sample of 54 observations in total. First, I used Thomson One to collect the different tender offer announcements between 1996 and 2016 in the United States and Canada. Information about the repurchase price was missing for a lot of announcements, so it gave me about 100 observations. Unfortunately, many of them were odd-lot self-tender offers or tender offers in the financial sector, which I had to remove from my sample.
The next step was to collect the information about the different variables for the regression. I‘ve collected these data with the use of Compustat and Datastream. Because not all information of every observation was available, I searched in Annual Reports for the missing info.
After that, the data was matched with data about the returns, which is collected with the CRSP-database. Here I used the S&P 500 index returns as a benchmark to calculate the normal returns and the CAR at the end (I used the S&P/TSX index collected from
https://www.investing.com/ to act as a benchmark for the Canadian observations). Again,
this database didn’t contain all company’s return-information, so I had to remove some more observations. This led to only 54 observations with the data about all variables, 51 announcements from the U.S. and 3 from Canada. This is a limitation to my research, because 54 observations is really close to the minimum sample size.
Below are some descriptive statistics.
Table1
Descriptive Statistics of both tender offer strategies, 54 observations.
Mean Median Std. Dev. Min. Max.
CAR[-1, +1] 0.073951 0.058808 0,077605 -0.16222 0.26359 Rep. Premium (%) 0.14255 0.10497 0.13301 0 0.57746 Shares Sought (%) 0.16821 0.12915 0.15977 0.0308 1 Book/Market ratio 0.72 0.6632 0.39810 0.00021 1.8172
17 FCF/total assets 0.04647 0.0306 0.10143 -0.18728 0.52131 ROA 0.05340 0.04715 0.08117 -0.24701 0.25895 Firm Size (ln(total assets)) 19.70162 19.42802 1.73500 16.40583 23.96445 Dummy Fixed Price 0.31482 0.06380 0.46880 0 1
Below are the descriptive statistics for each tender offer strategy separately to compare their characteristics.
Table 2
Descriptive statistics of the fixed price tender offers, 17 observations.
Mean Median Std. Dev. Min. Max.
CAR[-1, +1] 0.09156 0.10920 0.11472 -0.16222 0.26359 Rep. Premium (%) 0.139422 0.12846 0.13347 0 0.40351 Shares Sought (%) 0.18195 0.1 0.23247 0.0401 1 Book/Market ratio 0.74565 0.58756 0.46797 0.14414 1.8172 FCF/total assets 0.02486 0.01795 0.10315 -0.10229 0.25192 ROA 0.03938 0.03371 0.09908 -0.24701 0.25895 Firm Size (ln(total assets)) 19.21155 18.98865 1.55507 16.67129 21.62308
18
Table 3
Descriptive statistics of the Dutch auction tender offers, 37 observations.
Mean Median Std. Dev. Min. Max.
CAR[-1, +1] 0.06586 0.05875 0.05295 -0.06074 0.18846 Rep. Premium (%) 0.14355 0.10313 0.13462 0 0.57746 Shares Sought (%) 0.16189 0.1335 0.11589 0.0308 0.603 Book/Market ratio 0.71067 0.67181 0.36823 0.00021 1.7058 FCF/total assets 0.0564 0.03593 0.10047 -0.18728 0.52131 ROA 0.05984 0.05142 0.07212 -0.14152 0.25895 Firm Size (ln(total assets)) 19.92678 19.522 1.78646 16.40583 23.96445
The CAR of the fixed price offers is higher than the CAR of the Dutch auctions, but the difference is smaller than expected; it’s only a difference of 2.57%. The repurchase premiums are approximately the same; this is because I used the maximum premium offered in Dutch auctions. The other variables don’t differ very much between the two strategies.
19 Section 5: Results
In this section, I will provide the most important results from my research. First, the
outcome of the t-test will be discussed. After that, the regression analysis will be made and I will zoom in on every variable separately with reference to the signaling theory and the agency cost of free cash flow theory.
5.1 The CAAR and its significance
First, let’s look at the average abnormal returns for the days in the event period [-1, +1].
Table 4
The average abnormal returns of the event period [-1, +1].
AAR [-1] AAR [0] AAR [+1] CAAR [-1, +1]
-0,001466 0,05773 0,01769 0,07395
The average abnormal return at the announcement date is clearly the highest with 5.773 %. The announcement of a share repurchase seems to lead to a direct market reaction. The average abnormal return on the day after the announcement is also positive, but this is only 1.769%. When the AAR [-1] had shown an obvious positive return, it could be caused by leaked information about a share repurchase program. This relation is not indicated by the AAR [-1]; it’s almost zero.
As shown in table 4, the sum of the average abnormal returns is 7.395%. The t-test is used to test this result and look if there exists a significant reaction of the market to a share repurchase announcement.
Table 5
Results from the t-test of the CAAR [-1, +1].
CAAR [-1, +1] Std. Dev. T-value Critical t-value p-value
0,07395 0,07761 7,00244 >|2.009| <0.001
The T-value is bigger than the critical value, so the null-hypothesis that the CAAR is zero is rejected. The CAAR thus significantly deviates from zero with a p-value smaller than 0.001, which is a strong result. Comparing with early research, it is not very surprising that the
20
CAAR is significantly bigger than zero. Comment and Jarrell (1991) for example found an even higher CAAR; 7.7% for Dutch auctions and 11.9% for fixed price tender offers.
5.2 Regression results
The regression formula that I used, as described in section 3.2, is:
CAR[-1, +1] = α + β1Premium + β2SharesSought + β3Book/Market + β4FCF + β5ROA + β6FirmSize + β7DummyFixedPrice + ε
To look for the significance of the influence of the different independent variables on the CAR[-1, +1], I use different regressions. First, I look at the significance of the coefficients when I regress each independent variable separately on the dependent variable.
Table 6
The independent variables individually regressed on the CAR[-1, +1]. The t-values are the numbers between the parentheses. Every model contains 54 observations.
Model Variable 1 2 3 4 5 6 7 Rep. Premium (%) 0.32582*** (4.07) Shares Sought (%) .05779 (1.13) Book/Market ratio 0.06361** (2.33) FCF/total assets -0.00110 (-0.01) ROA -0.09389 (-0.58) Firm Size (ln(total assets)) -0.00794 (-1.18)
21 Dummy Fixed Price 0.02570 (0.89) Constant 0.02760* (1.90) 0.06423*** (5.56) 0.02804 (1.34) 0.07400*** (6.35) 0.07896*** (6.07) 0.23045 (-1.66) 0.06586*** (7.53) R-squared 0.3118 0.0142 0.1065 0.0000 0.0096 0.0315 0.0241 * p<0.10, ** p<0.05, *** p<0.01
The most obvious and strongest result from this table is that the repurchase premium is significant with a p-value that’s less than 0.01. This is not really a surprise; prior evidence, Comment & Jarrell (1991) for example, also found a significant positive relation between the repurchase premium and the CAR. The result is also in line with the most important
assumption of the signaling theory and confirms the hypothesis described in section 3; that a more costly repurchase leads to a stronger reaction of the market.
The only coefficient that is significant except the coefficient of the repurchase
premium is the book to market ratio, with a p-value less than 0.05. This result gives strength to the undervaluation theory and confirms our hypothesis.
Surprisingly, the other coefficients aren’t significant, not even when we look at a p-value < 0.10. Table 1 does not give any support for the agency cost of free cash flow theory.
Table 2 shows result from multiple regression models. Model 2.1 is used to primarily look at the signaling theory, model 2.2 is used for the agency cost of free cash flow theory. Model 2.3 contains all the independent variables.
Table 7
Multiple regressions with CAR[-1, +1] as dependent variable. The t-values are the numbers between the parentheses. Every model contains 54 observations.
Model Variable 2.1 2.2 2.3 Rep. Premium (%) 0.31288*** (3.94) 0.33243*** (4.08) Shares Sought (%) -0.03171 (-0.64) -0.03567 (-0.64)
22 Book/Market ratio 0.05103** (2.34) 0.06135** (2.12) FCF/total assets 0.02516 (0.22) -0.04945 (-0.60) ROA - 0.08153 (-0.50) 0.13726 (0.86) Firm Size (ln(total assets)) 0.00034 (0.06) 0.00158 (0.26) Dummy Fixed Price 0.02609 (1.23) 0.03036 (0.82) 0.02802 (1.23) Constant -0.01687 (-0.15) 0.06932 (5.31) -0.05650 (-0.41) R-squared 0.4003 0.0316 0.4182 * p<0.10, ** p<0.05, *** p<0.01
As we compare table 2 with table 1, there aren’t any differences between the significance of the coefficients. The repurchase premium is still highly significant with a coefficient higher than 0.30, both in model 2.1 as in model 2.3. Also book to market ratio remains significant at a 5% p-value level; it has a coefficient of more than 0.05 in both models. These results support the signaling theory and are again in line with the expectations.
Staying with the signaling theory, it is remarkable that the amount of shares sought isn’t significant in both tables. In contrast with the hypothesis, the coefficient of ‘shares sought’ is negative in models 2.1 and 2.3. The assumption that a signal should be costly to be credible isn’t supported with these results.
Model 2.2 focuses on the agency cost of free cash flow theory. The coefficients of the variables that are used to test this theory, the FCF and the ROA, are not significant when we look at the results. When these variables are added to model 2.1 in model 2.3, we can see that the R-squared value doesn’t change very much. This indicates that these variables don’t add a lot of value in explaining the variance in the CAR, which is in contrast to the
23
The coefficient of the control variable firm size isn’t significant as well. The expectation was that firm size would be negatively related with the CAR because of a difference in information asymmetry between small and bigger firms. The results do not indicate support for this expectation.
The dummy variable doesn’t indicate a significant relation with the dependent variable as well. This is also against the hypothesis; the abnormal returns in the event period of fixed price tender offers were expected to be higher than the abnormal returns of the Dutch auctions.
24 Section 6: Conclusion
The purpose of this thesis was to examine whether the short-term reaction of the market to recent share repurchase announcements in North-America could be explained by two theories; the signaling theory and the agency cost of free cash flow theory. Furthermore, I wanted to look whether there was a difference between the reactions to fixed price tender offers and Dutch auctions, and how these differences could be explained. I collected data from 17 fixed price and 37 Dutch auction tender offer announcements in the U.S. and Canada between 1996 and 2016 and I did a t-test on the CAAR to examine if this average three-day return was significant. After that, I did a regression with the CAR as the dependent variable to look which factors had an impact and whether these factors could be related to the two theories.
My results indicate that there exists a significant positive reaction of the market to a repurchase announcement; the three-day CAAR is 7.40%, which is big enough to be
statistically significant with a p-value less than 0.001. This result is in line with the
expectation that a repurchase announcement leads to a share price increase because of the information given by insiders about their ideas and expectations of the firm and its value. The result is actually a bit low when comparing it with the overview given by Vermaelen, who summarizes results of several earlier studies in his paper from 2005. This difference can be explained by time differences, by the Dutch auctions dominating my sample, or by the smaller sample size.
The regression results only support the signaling theory. The coefficient of
repurchase premium is significant with a p-value less than 0.01, indicating a strong relation with the CAR as dependent variable. This result supports the assumption of the signaling theory that a signal is more credible when it’s more expensive, and is in line with results from for example Comment and Jarrell (1991). On the other hand, the coefficient of the ‘shares sought’ variable isn’t significant when looking at the individual and the multiple regressions, indicating the opposite. The coefficient of the book to market ratio is also significant with a p-value less than 0.05, supporting the ideas of the undervaluation theory.
The regression results don’t give any sign of evidence for the agency cost of free cash flow theory. The coefficients of the FCF and the ROA variable are both far from significant. I tried several options with the use of different combinations of variables, for example the use of number of employees, sales, and total assets (the one that is finally used in this thesis).
25
These attempts didn’t lead to significant results as well. It could be that the determinants of the signaling theory are more important to outsiders than the possible determinants of the agency theory and that the market reaction therefore is only caused by these variables (repurchase premium and book to market ratio) in my sample . I conclude that the CAR is not determined by the ideas of the agency cost of free cash flow theory in my sample.
The biggest limitation of my work is that the sample is not very large (54
observations). Unfortunately, it wasn’t possible to get all the data to extend my sample with more announcements. To make conclusions with more certainty, it would be better to test the theories with a broader sample in future researches. My small sample could be the reason for the insignificant results of the coefficients testing the agency cost of free cash flow theory. It also leads to a limited conclusion about the difference between the fixed price and Dutch auction tender offers. I don’t find evidence for the hypothesis that the fixed price offer gives a stronger signal to the market, but this can be caused by the small fixed price sample of only 17 observations.
Because of the small sample size, my results can’t be fully generalized to other populations. Besides that, I’ve focused on firms in North-America. Findings about market reactions with regard to firms from other parts of the world could therefore differ with the findings of this research.
26 References
Comment, R., & Jarrell, G. A. (1991). The relative signalling power of Dutch‐auction and fixed‐price self‐tender offers and open‐market share repurchases. The Journal of
Finance, 46(4), 1243-1271.
Dann, L. Y. (1981). Common stock repurchases: An analysis of returns to bondholders and stockholders. Journal of Financial Economics, 9(2), 113-138.
Dittmar, A. K. (2000). Why do firms repurchase stock. The Journal of Business, 73(3), 331-355.
Grullon, G., & Michaely, R. (2004). The information content of share repurchase programs. The Journal of Finance, 59(2), 651-680.
Jensen, M. C. (1986). Agency costs of free cash flow, corporate finance, and takeovers. The
American economic review, 76(2), 323-329.
Nohel, T., & Tarhan, V. (1998). Share repurchases and firm performance:: new evidence on the agency costs of free cash flow1. Journal of Financial Economics, 49(2), 187-222.
Persons, J. C. (1994). Signaling and takeover deterrence with stock repurchases: Dutch auctions versus fixed price tender offers. The journal of finance, 49(4), 1373-1402. Vermaelen, T. (1981). Common stock repurchases and market signalling: An empirical study. Journal of financial economics, 9(2), 139-183.
