Share Repurchases in the Banking Industry:

Transcription

1 Share Repurchases in the Banking Industry: The Undervaluation Hypothesis Investigated Document: Author: Master Thesis Theresa M. Hoogendorp Administration Number: Program: Department: Supervisor: Chairperson: Master Finance TiSEM, Finance dr. F. Castiglionesi dr. M.R.R. van Bremen Date: September 17 th,

2 Acknowledgements This study focuses on the importance of the undervaluation hypothesis on the repurchase decision for financial institutions. This way, I was able to combine the knowledge I gathered during my two favorite courses in the master finance program. The repurchasing of shares attracted my attention during the lectures in the Advanced Corporate Finance course, while my main interest in the financial industry was stirred during the Global Banking lectures. Since not a lot of research has been done in this combined research field, I wanted to investigate whether the laws of corporate finance also hold in the field of financial institutions. This thesis marks the end of five years of studies at Tilburg University. During those five years I have grown to become an independent and academic woman; something I could not have done without the support of my family, close friends and the supervision and guidance of multiple professors at Tilburg University. I would like to thank dr. Fabio Castiglionesi in particular, not only for being my thesis supervisor but for making me enthusiastic about the financial industry as well. Also I would like to thank dr. Michel van Bremen for his time to read and review my thesis. I hope you enjoy reading my thesis. Theresa M. Hoogendorp 2

3 Table of Content 1. Introduction Literature Review and Hypotheses Developing Literature Review Hypotheses Developing Sample, Regression and Data Sample Regression Variables Results and Discussion Model Estimation Basic Regression Adding a Dummy for Financial Crisis Introducing Interaction Terms Determining the Impact of the Financial Crisis The Effect of the State of the Economy on the Repurchase Transactions The Effect of the State of the Economy on Undervaluation Conclusions References

4 List of Tables Table 1.1 Frequency Table of Public and Private Financial Institutions Page 10 Table 3.1 Overview of the Dependent Variable Page 20 Table 3.2 Descriptive Statistics of Market-to-Book Ratio for four Quartiles Page 22 Table 3.3 Correlation Matrix of Independent Variable and Control Variables Page 27 Table 4.1 Results of Tobit Regressions Page 31 Table 4.2 Results of the grouped t-test on REP Page 33 Table 4.3 Results of the grouped t-test on MKBK Page 34 4

5 Abstract The motives that companies have for making the share repurchase decision have already been investigated a lot. However, most studies exclude financial institutions from their sample since the financial industry would be too regulated and therefore the findings would be hard to compare or complement findings in other industries. The few papers that do focus on the repurchase decision made by financial institutions, concentrate on ex-post research including the market reaction after a repurchase announcement. In contrast, this thesis focuses on ex-ante market and firm characteristics regarding the actual repurchase transaction. Repurchase transactions performed by financial institutions from the second quarter of 2004 through the fourth quarter of 2013 will be investigated. This thesis directs its attention to the undervaluation hypothesis, which is investigated by using a tobit regression. The key finding here is that financial institutions that are more undervalued relative to their peers, engage in higher volume repurchase transactions. This implies that the undervaluation hypothesis also holds for financial institutions. Furthermore, the impact of the recent global financial crisis on the repurchase transaction is investigated. Although no real significant effect of the recent global financial crisis on the dollar volume of the repurchase transactions is found, the results show that financial institutions were more undervalued during the recent global financial crisis compared to the other years in the sample. 5

6 1. Introduction Since the 1980s, research has been done into motives for companies to buy back their own shares (e.g. (Vermaelen, 1981), (Dittmar, 2000) and (Comment & Jarrell, 1991)). However, most studies in this field exclude financial institutions from their research, due to the industry being too complex or too regulated. But financial institutions do have a huge impact on society. As we saw during the recent global financial crisis, choices made by banks and other financial institutions directly affect people by for example mortgages, pensions and life savings. Also, a lot of financial institutions were saved by governments, because of their importance to the people living in that country. (USA TODAY, 2013). It can be concluded that although financial institutions are in a regulated industry, investigating them could offer interesting insights because of their great importance and the effects of the choices they make. Compustat Bank does give detailed information about the amount of shares repurchased and the average repurchase price of financial institutions. Together with the importance of financial institutions, this data availability gives an interesting opportunity to test whether motives for financial institutions to repurchase stock are comparable to non-financial institutions. The undervaluation hypothesis is seen as one of the possible motives for firms to engage in a repurchase transaction. It states that if managers believe that the price of outstanding stock is too low compared to what the firm is really worth, there is no better investment than buying back their own stock. This redistribution of the firm s wealth is often compared with the firm s dividend payout policy and sometimes it is even argued that they are substitutes. Among others, Dittmar (2000) found evidence for the undervaluation hypothesis in a study conducted on a sample of non-financial institutions. Hirtle (1999) found positive stock market reactions after repurchase announcements in the financial industry and concluded this could be evidence for the undervaluation hypothesis being applicable for bank holding companies too. Laderman (1995) earlier found similar results, also conducting a study on stock market reactions after repurchase announcements. However, the aim of this thesis is not in the market reactions of financial institutions following a repurchase announcement. This study focuses on the firmspecific characteristics of a financial institution ex ante the repurchase transaction, in order to research whether the financial institution was really undervalued and if so, what the real reaction - rather than the reaction to the announcement - was to this undervaluation. Consequently, the purpose of this thesis is to see whether the undervaluation hypothesis is also applicable to financial institutions when 6

7 focusing on the actual repurchase transaction, as in Dittmar (2000), rather than focusing on repurchase announcements, as in Hirtle (1999) and Laderman (1995). This research focuses on the actual repurchase transactions performed by financial institutions from the second quarter of 2004 through to the fourth quarter of The principle of undervaluation is really topical due to the recent global financial crisis. As a result, stock prices dropped heavily leading to low market equity value. In the financial industry, this led to bank runs, the burst of the housing bubble and the failure of some major financial institutions. Therefore this thesis is centered on the undervaluation hypothesis. The main question in this thesis is whether the undervaluation hypothesis, as posed for nonfinancial institutions, also holds for financial institutions. Although arguments upholding the undervaluation hypothesis for financial institutions can already be found, it must be remarked that these papers focus on the ex-post market reaction rather than the ex-ante firm-specific characteristics. Since financial institutions differ in many ways, for example having more financial assets relative to other commercial firms, one can assume the firm-specific characteristics leading to the repurchase transaction are also different or lead to the repurchase transaction in a different way. This thesis builds a bridge between the research done on ex-ante repurchase transactions in the non-financial industry as in Dittmar (2000) and the research done ex-post repurchase transactions in the financial industry as in Hirtle (1999) and Laderman (1995) Since the recent global financial crisis is part of the sample period used in this study, it seems interesting to zoom in on this period extensively. First, it will be investigated whether repurchasing financial institutions were more undervalued during the financial crisis of This can be expected due to the decreases in market equity value mentioned earlier. However, whether the book value of equity dropped evenly during this period is still to be investigated. Therefore, it is an open question whether financial institutions really were more undervalued during the global financial crisis. As a follow up, the repurchasing behavior of financial institutions can be compared during times of financial crisis and during economic stability. Although the latter is also investigated by Hirtle (2014), again the focus will be on the actual repurchase transaction rather that the stock market reaction to the repurchase announcement. Since the repurchase transaction is considered to be a redistribution of shareholders wealth, it can be assumed that financial institutions have less wealth to redistribute during the financial crisis. For this study, data on the repurchase transactions reported by financial institutions is downloaded from Compustat Bank Quarterly. Initially all financial institutions that reported a repurchase transaction from 7

8 the second quarter of 2004 through the fourth quarter of 2013 were included in the sample. Data on the market value and return of the financial institutions was downloaded from CRSP. Since CRSP did not report market data on all the financial institutions listed in Compustat Bank Annual, the initial sample was reduced to the number of financial institutions Compustat Bank Annual and CRSP both reported on. Since the sample is not representative for the whole population a tobit regression is used to estimate the basic model, as in Dittmar (2000). The sample is not representative since only repurchasing financial institutions are included, excluding al non-trading financial institutions. Whereas Dittmar (2000) focused to test multiple motives for repurchases simultaneously, this study only focuses on the undervaluation hypothesis. In the extended model, a dummy for crisis is included to check what the impact of the crisis is on the actual repurchase transaction. Also, the degree of undervaluation for financial institutions will be considered by comparing means of the undervaluation proxy. The measure for undervaluation will be the lagged market-to-book ratio, the other measures will be used as control variables. At first the basic model is estimated, using a tobit regression. When the regression is run only using the independent variables similar to Dittmar (2000), the market-to-book ratio enters the regression negative and significantly. Also firm size can be seen as a negative, significant influence on the repurchase transaction. Lastly, the return history enters the regression significantly. However, this is a positive correlation. It is remarkable that no significant relationship is found between the dividend payout policy of the financial institution and the repurchase transaction. In none of the regressions that were run, a significant relationship for dividend payout policies is found and therefore it is also excluded from the final model. To check whether the market-to-book ratio interacted with any of the control variables, regressions were run including the interaction terms. Significant evidence was found only for the interaction between the market-to-book ratio and firm size and the market-to-book ratio and the return history. It was shown both the size of financial institutions and the return history positively influences the effect of the market-to-book ratio on the repurchase transaction. In order to find whether the recent financial crisis is of influence on the repurchase transaction, a dummy variable for financial crisis was added to the basic model. However, the dummy did not enter any of the regressions significantly and therefore the relationship between the financial crisis and was further investigated by a two-group mean comparison test. This test confirmed the earlier findings, also in the mean comparison test there was no evidence for the relative dollar volume of repurchase 8

9 transactions being lower during the financial crisis compared to the other years included in the sample. Furthermore, a mean comparison test was performed to check whether the financial crisis was of influence on the market-to-book ratio. This resulted in a significant difference, stating that the average market-to-book ratio was lower for financial institutions during the financial crisis compared to the other years included in the sample. However, when an interaction term between the dummy for crisis and the market-to-book ratio was added to the basic model, no significant relationship was found. In her research on motives for firms to engage in a repurchase transaction, Dittmar (2000) studied five possible motivations simultaneously. However, financial institutions were excluded from this study. One of the motivations she found evidence for, was the undervaluation hypothesis. This hypothesis is also tested by the tobit regression on the basic model in this study. Now the sample only includes financial institutions. The first thing to be noticed is that the average relative dollar volume per repurchase transaction for financial institutions is lower than reported in the Dittmar (2000). Further, Dittmar (2000) found support for the undervaluation hypothesis shown by a negative, significant relationship between the lagged market-to-book ratio and the repurchase transaction. The same relationship, albeit smaller, is found in this study on financial institutions. Vermaelen (1981) argued that smaller firms should have more incentive to engage in a repurchase transaction due to the fact smaller firms should experience more information asymmetries. However, this view is not broadly supported. For example, Dittmar (2000) and Ikenberry, Lakonishok and Vermaelen (1995) later argued that firm size is positively related to the relative dollar volume of repurchase transactions since also large firms want to take advantage of mispricing. Although the study of Dittmar (2000) is of key interest; her findings on the relationship between firm size and the repurchase transaction do not hold in this study. In the sample of financial institutions a significant negative relationship between firm size and the repurchase transaction is found, contradicting the argument of Dittmar (2000) and Ikenberry, Lakonishok and Vermaelen (1995). Dittmar (2000) included the dividend payout policy in her regressions, to check the tax motive for repurchases. However, more researchers (Jagannathan, Stephens and Weisbach (2000), Lintner (1956), Julio and Ikenberry (2004) and Skinner (2008)) discussed the relationship between dividend payouts and repurchase transactions and made it clear a real effect does exist, although the opinions whether this effect is positive or negative differs. Due to the very occupant relationship, the model in this study is controlled for dividend payout policy also. However, whereas Dittmar (2000) found positive significant relationship between dividend payouts and the repurchase transaction, the study performed on 9

10 financial institutions did not find evidence for any significant relationship. Further, Dittmar (2000) did not find conclusive results about the relationship between the firm s return history and the repurchase transaction, while a positive relationship was found in this study for financial institutions. Hirtle (2014) found that the occurrence of the recent global financial crisis was of negative influence on the repurchase transactions of bank holding companies. However, this effect is not established in this study. An explanation for this effect not being found in this study could be the fact that this study focuses on traded individual banks and not on bank holding companies. During the sample period, Compustat reports data on 1098 different banks, only 765 of which are traded on the financial market as projected in table 1.1. Since bank holding companies, as investigated by Hirtle (2014), include both individual public and private banks, this difference could be an implication for the differences in established effects in her paper, compared to this thesis. Since private banks involve a whole different facet of risk taking decisions and therefore bring along different firm-specific characteristics relative to public financial institutions, this could be an explanation for the difference in findings in Hirtle (2014) compared to this thesis. Table 1.1: Frequency Table of Public and Private Financial Institutions Divides the financial institutions listed in Compustat Bank in traded and non-traded. Traded Frequency Percentage No Yes Total 1, This thesis found arguments for the undervaluation hypothesis to be applicable to financial institutions, empirically supported by the results of the tobit regression. No conclusive proof was found for any effect of the recent global recent financial crisis on the relative dollar volume of the repurchase transaction. However, evidence was found for financial institutions reporting lower market-to-book ratios during the recent financial crisis. This thesis was limited by the focus being solely on the undervaluation hypothesis. It could be the case that the repurchase decision depends on multiple criteria rather than just one and this is also indicated by the highly significant constants the regressions reported. The relationship between the undervaluation hypothesis and other hypotheses is ignored. This leaves room for further research on 10

11 the repurchase decision in the banking industry, for which this thesis can be used as a complement or as a reference. The rest of this thesis is organized as follows. Chapter two discusses the relevant existing papers on share repurchases, motives for share repurchases with the undervaluation hypothesis in particular, share repurchases in the financial industry and a short discussion on legislation. Chapter three discusses the way the sample was selected, the model that was used and the variables that were included into the model. Chapter four discusses the results from the statistical analysis. Lastly, chapter five consists of some concluding remarks and a brief explanation of the limitations that should be taken into account for further research. 11

12 2. Literature Review and Hypotheses Developing 2.1 Literature Review A share repurchase is the transaction in which a company, being the issuer, buys back its own stock. This transaction can have internal and external effects. It reduces the number of outstanding shares (Laderman, 1995) and it also alters the leverage ratio or capital-to-assets ratio, when debt or cash is used to complete the transaction. Another internal affect is the change in ownership structure, since the percentage of ownership increases for the remaining shareholders after the repurchase transaction. What is more, when the share repurchase is financed with debt, the transaction could increase the company s debt tax shield and therefore lower the company s cost of capital. However, most openmarket repurchases are paid for with excess cash (Rau & Vermaelen, 2002). As an external effect the share repurchase transaction distributes cash to shareholders selling their shares to the company and, in case of equal equity value, the value of the shares held by the remaining shareholders increases. This external effect is part of the substitution hypothesis. A firm redistributing money through repurchasing shares often does so instead of paying dividends (Grullon & Michaely, 2002) or alongside paying dividends as a complement (Jagannathan, Stephens, & Weisbach, 2000). However, dividends and repurchases have another effect on the company s structure than share repurchases. Whereas dividend payouts reduce a firm s retained earnings and therefore could reduce the value of the firm s common equity, share repurchases only reduce the amount of common equity outstanding while not altering the firm s value. However, a repurchase announcement could increase the stock value of the firm. Evidence is found for abnormal stock returns among bank holding companies directly after the repurchase announcement (Hirtle, 2014). Much research into the motives to repurchase shares has already been done. Back in the 1980s, Vermaelen found that the stock price for firms repurchasing their shares increased. He found evidence for his information hypothesis together with his signaling hypothesis, meaning that firms are willing to repurchase shares in a tender offer for a premium when future growth is expected, and that dividends are a signal for future growth. However, the results from his research were not conclusive for repurchases through open market transactions (Vermaelen, 1981). Around the same time, Dann (1981) also investigated tender offers and found conclusive evidence for Vermaelen s signaling and information motives. However, he was not able to answer the question why managers would want to disclose the information via a tender offer or repurchase announcement and also he could not identify the nature of 12

13 the information which led to value changes. Dann further found positive returns for common stockholders after a repurchase on the short term (Dann, 1981). Fama later used the findings of Vermaelen, Dann and others to explain his arguments against the Efficient Market Hypothesis by concluding that the market underreacts to the signal about future performance in share repurchases. In other words, according to Fama, abnormal returns after a share repurchase are the result of the market failing to incorporate the expectations about future performance into the stock prices (Fama, 1998). Later, Dittmar used the hypotheses of Vermaelen (1981) to see which motive is the most important in the repurchase decision. She found that the main reason for firms repurchasing their own stock is the potential undervaluation of their stock. It could be the case that when managers find their stock price is too low, they feel like there is no better other investment than in their own company. A remarkable conclusion from Dittmar is that larger firms are more likely to repurchase, while it is expected that most information is present about large firms and therefore it is highly unlikely that larger firms are the ones to be valued incorrectly. It is important to remark that Dittmar excluded financial institutions from her sample, because she concluded that due to regulations the motives for those companies to repurchase stock could be different from other non-regulated companies (Dittmar, 2000). However, could it not be that all motives for engaging in a stock repurchase are related and could be traced back to the same motive? If there is information asymmetry between a firm s managers and, albeit potential, shareholders, managers may try to send a signal to the market, capturing the information hypothesis and the signaling hypothesis. When the signal is undervaluation, managers may want to signal the market by announcing a share repurchasing program. Portfolio managers stick to the buy low and sell high principle and therefore the best thing a well educated manager could do is buying back the shares of his own company (Ikenberry, Lakonishok, & Vermaelen, 1995), when he believes the price of the shares is now lower than the real enterprise value. These firms are undervalued, meaning their market value is below the firm value (Barth & Kasznik, 1999). However, the undervaluation hypothesis is in contrast with the signaling hypothesis, since the undervaluation hypothesis assumes the market is not working efficiently by not incorporating information in the market value, while the signaling hypothesis assumes the market to work well, since the repurchase announcement will cause the market value adjust to the signal (Rau & Vermaelen, 2002) (Ikenberry, Lakonishok, & Vermaelen, 1995). Incorrect valuation can occur because of two independent conditions. The first one focuses on the market, which is not able to interpret the public information available correctly. This way, analysts could 13

14 value stocks of a particular firm too low, which leads to undervaluation (Doukas, Kim, & Pantzalis, 2005). The second condition focuses on insider information. When managers have inside information that is not publicly available, they could interpret the company s stock price to be mispriced. When managers have information about the firm s positive future growth opportunities, this could lead to undervaluation (O'Brien, Schmid Klein, & Hilliard, 2007). Most of the research on motives for repurchasing stock excluded financial institutions from the sample. The exclusion was mainly caused by financial institutions being more severely regulated and also, these companies are in a more complex industry (Dittmar, 2000) (Grullon & Michaely, 2002). The question here is what makes the financial industry different from other industries. When focusing on financial institutions, one special feature is the high percentage of financial assets. Also, these types of companies are more leveraged and hold relatively less equity compared to non-financial firms. The motives for share repurchases in the banking industry are examined in only a few papers. Srivastav, Armitage and Hagendorff (2014) focused their research on risk-shifting through bank payout policies. They argue that any payout, no matter being dividends or share repurchases will be decreasing liquid assets and retained earnings, while the more illiquid assets would remain constant. This indicated that debt holders now have to incur more risk than the paid out equity holders, since there is less capital left to absorb potential losses. Equity holders therefore benefit from risk-taking more, because their payoffs are expected to rise with increased risk, while the debt holders payoff does not increase and is more at risk in case of the risky behavior of the bank will lead to losses (Srivastav, Armitage, & Hagendorff, 2014). Laderman (1995) considers this risk-shifting in a different way by looking at the alteration of the capitalto-asset ratio. However, due to regulations as the deposit insurance for banks, they are made to hold less equity than the banks would prefer. The deposit insurance makes the investment for debt holders less risky compared to a case with no deposit insurance. However, Laderman (1995) claims this would not eliminate the risk-shifting effect of share repurchases for banks completely. Laderman further found three possible explanations for bank holding companies to engage in share repurchases. The first one, as described above focuses on the risk-shifting from equity holders to debt holders by decreasing the capital-to-asset ratio. The second motive by Laderman argues that the repurchasing behavior could be a result of the firm repurchasing its own shares. He argues that this is investment possible, since in some circumstances the repurchasing can lead to an increase in equity value. Lastly, Laderman also found evidence for the undervaluation hypothesis. However, the research 14

15 conducted by Laderman is based on repurchase announcements rather than actual repurchase transactions (Laderman, 1995). Although Hirtle found evidence for positive stock price reactions to repurchase announcements among bank holding companies (Hirtle, 1999), these results are not broadly supported. For example, Billingsley et al. (1989) did not find any stock market reaction to the repurchase announcement made by bank holding companies. Therefore, the signaling hypothesis is not considered a plausible motive for share repurchases in the banking industry. Another different facet for the repurchasing in the banking industry compared to other non-financial industries is the way a firm s management gets compensated. Due to legislation, the managers compensation in the banking industry is composed differently. In the banking industry, a much smaller part of the managers compensation depends on the stock price behavior compared to other industries. Therefore, it is unlikely that managers in the banking industry have the same tax incentives to prefer share repurchases over dividends, as they would in a non-financial industry (Hirtle, 2001). Prior to 2004, firms that disclosed the intention to engage in a repurchase program, were not obliged in any way to actually do so. However, from March 2004 on, a change in Rule 10b-18 of the Securities Exchange Act obliged firms to disclose all open market repurchases in their annual and quarterly statements. From 2004 onwards, firms must disclose the total number of shares repurchased, the average price paid per share, the total number of shares purchased as part of publicly announced plans or programs and the maximum number of shares that may yet be purchased under the plans or programs (Securities and Exchange Commission, 2003). A direct implication of the change mentioned above is increased transparency, since now firms that have announced to engage in a repurchase program, can be checked if they really did. This way, share repurchase announcements should have more real effects. Firms do not want to lose their credibility by not completing a program and therefore the announcement itself will not be just a signal anymore, but a commitment to complete the program (Bonaimé, 2012). Ivashina and Scharstein (2008) have shown that during the fiscal year of 2008, the first year of the global financial crisis, lending for restructuring purposes has decreased from $150 billion to $50 billion in the United States. Restructuring loans are those for M&A, LBO and stock repurchases. 15

16 The findings of Ivashina and Scharfstein are in line with those of Hirtle (2014), who found that share repurchases done by bank holding companies dropped heavily in 2008 and hit the lowest level since 2005 mid It is remarkable that, during the same time, dividend payouts by bank holding companies remained constant while share repurchases dropped down to zero. This difference is explained by the fact that repurchases of shares are instruments used in periods of high income, while dividend payouts always are rather constant in the banking industry, since managers are reluctant to announce decreases in dividend payouts due to the negative stock price reaction this announcement is likely to cause. While the focus in the previously mentioned studies is solely on bank holding companies, this study focuses on all financial institutions available. Hirtle (2014) further found that bank holding companies engaging in repurchase transactions before the recent global financial crisis could postpone a decrease in dividend payout. Matching these findings to the earlier describes motives for share repurchases, an implication is that for banks the substitution hypothesis of Grullon & Michaely (2002) only holds in financially bad times, as we see banks paying out dividends during the financial crisis, while they are not engaging in any repurchasing activity. The statement that repurchases can be done alongside paying out dividends of Jagannathan et al. (2002) is confirmed by Hirtle s results from before the recent global financial crisis. 2.2 Hypotheses Developing Since evidence is found on positive market reactions to repurchase announcements in the financial industry as in the non-financial industry, one should expect a similar relationship between both industries looking at the actual repurchase transaction rather than the repurchase announcement. Because the ex-post research already found arguments for the undervaluation hypothesis in the financial industry, similar arguments are expected to be found in this ex-ante research as stated in hypothesis 1. Hypothesis 1: The actual repurchase transaction reported by financial institutions is dependent on the degree of undervaluation of the particular financial institution. As lending for restructuring purposes dropped during the recent financial crisis, one should expect less shares to be repurchased as a consequence of the fact that less capital is available to execute the transaction. Therefore, the occurrence of the recent global financial crisis should lead to a decrease in repurchase transactions. This should not only be true for bank holding companies as Hirtle (2014) reported, but also for the individual banks which are investigated in this thesis, as stated in hypothesis 2. 16

17 Hypothesis 2: The amount of shares repurchased by financial institutions depends on whether the repurchase transaction is executed during times of financial crisis or not. Since the undervaluation hypothesis is of key interest here, it might be possible to link the financial crisis to undervaluation directly. That is, during the recent financial crisis stock prices dropped heavily. As a consequence, the market value of equity should decrease as well and keeping everything else constant, the degree of undervaluation of the particular financial institution should increase. Therefore it should be expected that financial institutions are more undervalued during times of a financial crisis, as stated in hypothesis 3. Hypothesis 3: The degree of undervaluation of financial institutions depends on the presence of a financial crisis. 17

18 3. Sample, Regression and Data 3.1 Sample In March 2004, Rule 10b-18 of the Securities Exchange Act changed the disclosure of repurchases in annual and quarterly statements into an obligation for firms (Securities and Exchange Commission, 2003). Therefore in the sample only repurchase transactions from the second quarter of 2004 through the fourth quarter of 2013 are included into. This measure prevents the results from being blurred, since more repurchase transactions are expected to be reported due to the new obligation of reporting. Given that the data may or may not be complete before the second quarter of 2004, it doesn t seem reasonable to also include data from before the implementation of Rule 10b-18 in the sample. From Compustat Bank Quarterly, quarterly data from 1090 different banks was downloaded, resulting in lines of data. From this initial dataset the quarters for which the banks did not engage in a repurchase transaction, defined by the bank reporting the amount of repurchased shares per quarter and the average repurchase price per share, were eliminated. This action resulted in a dataset of 7694 quarters in which 710 unique banks engaged in a repurchase transaction. In order to prepare the sample to perform statistical analysis, the data obtained from Compustat Bank Quarterly was supplemented with market and return data from CRSP. However, not all banks or lagged quarters were available in CRSP. As a consequence only the quarters wherein a bank engaged in a repurchase transaction for which lagged market data were available, made it to the final sample. As a result the final sample consists of 6114 repurchase transactions, performed by 541 unique banks from the second quarter of 2004 to the fourth quarter of Regression In order to find out what the effect of undervaluation on repurchase transactions in the banking industry is, a part of the model of Dittmar (2000) is complemented with other variables. The model considers different factors that are expected to have an impact on the repurchase transaction. REP i,t = β 0 + β 1 MKBK i,t 1 + β X X i,t + β I IT i,t + ε i,t (1) Expression (1) describes the model that will be used, where β 0 denotes a constant. Further, i denotes the bank and t or t-1 denotes time. MKBK is the main variables of interest, since it is used as a proxy for undervaluation. This variable is taken from the part of the regression Dittmar (2000) uses to find evidence for the undervaluation hypothesis and is further specified in section X is a vector of 18

19 other variables which can influence the repurchase transactions and are therefore used as control variables. The IT vector is constructed of the interaction terms of the main variable of interest and the control variables. The interaction terms are used to test whether the control variables are in any way precursors for repurchase transaction to be related to undervaluation (Drichoutis, 2011). 3.3 Variables Dependent Variable The dependent variable REP i,t is the term denoting the repurchase transaction of bank i reported during quarter t. Whereas previous studies focused mainly on the repurchase announcement (Rau & Vermaelen, 2002) (Hirtle, 2014), now the interest in the actual repurchase transaction as reported in the quarterly statements of the banks. This variable is comparable but slightly different from the dependent variable used in Dittmar (2000) and Stephens and Weisbach (1998), since it is the relative dollar volume of the repurchase transactions during quarter t, divided by the market value of equity of quarter t-1, as stated in expression (2). However, since Dittmar (2000) used the Compustat item Purchase of Stock, she had to deal with difficulties in readjusting the variable to only obtain the repurchase transactions. These readjustments were not necessary in this analysis since the necessary Compustat items were at hand, as explained in more detail below. REP i,t = (AMREPSHARES i,t PRICEREPSHARES i,t ) MEV i,t 1 (2) The relative dollar volume of the repurchase transaction is calculated by multiplying the amount of repurchased shares with the average repurchase price per share. The amount of the repurchased shares and the average repurchase price per share are both downloaded from Compustat Bank Quarterly, which offers these items directly so no further readjustment was needed. The market value of equity is calculated by multiplying the number of outstanding shares by the month s end closing prices. When closing prices were not available, the bid-ask average was used. By averaging the market values of the three months of the prior quarter, the market value of equity of quarter t-1 was obtained. The monthly share prices and the number of shares outstanding were downloaded from the CRSP database. Looking at the distribution of the dependent variablerep i,t, summarized in Table 3.1, a strong decline in the frequency of transactions can be found in the years 2008 and However, a growth or decline in frequency not always goes with a similar change in the mean of the transactions. It is also remarkable 19

20 that the total column which displays the mean of the transactions is below one percent of the market value. Whereas Dittmar (2000) deleted all observations lower than one per cent out of the lagged market value of equity it doesn t seem advisable to do so in this study since the mean of the transaction is already below one per cent. This difference in volume could be an indicator of a difference in repurchase transactions of financial institutions comparing to the volume of repurchase transactions in nonfinancial industries. The distribution and also the mean of the repurchase transaction will be further examined with the use of the independent and control variables. Table 3.1: Overview of the Dependent Variable Table 3.1 summarizes the dependent variable, being the dollar volume of the actual transaction divided by the lagged market value of equity. N denotes the number of transactions in a particular year and Mean shows the mean of the depended variable in a certain year. In the Total column the total number of transactions is shown and the mean of all transactions done is displayed. Year Total N Mean Independent Variables and Control Variables Market-to-Book Ratio Firms with a low market-to-book ratio generally suffer more from financial distress than firms with a higher market-to-book ratio. Also the market-to-book ratio is positively related to return on stock. Furthermore, it is shown the market-to-book ratio for firms is persistent, meaning that firms will find difficulties to alter their market-to-book ratio over time (Fama & French, 1995). Stocks that are underpriced are said to have low market-to-book ratios relative to shares that are overpriced (Dittmar, 2000) (Zhang, 2005) 1. Therefore, if the repurchase transaction of a firm depends on the undervaluation of the company s stock, it can be expected that firms with a lower market-to-book ratio will be engaging in larger repurchase transactions. However, when looking at the propensity to announce a repurchase transaction, Ikenberry et. al. (1995) do not find a significant difference for low and high market-to-book ratios. Their reasoning is based on that not all firms with low market-to-book 1 Low in this context means a market-to-book ratio lower than one, indicating that the market value of equity is lower than the book value of equity. 20

21 ratios are really undervalued and therefore the propensity to announce a repurchase program is not solely dependent on the degree of undervaluation. MKBK i,t 1 = Market value of Equity i,t 1 Book value of Equity i,t 1 (3) The lagged market-to-book ratio is measured by dividing the market value of equity at the quarter prior to the repurchase transaction by the book value of equity at the quarter prior to the repurchase transaction, as denoted in expression 3. The market value of equity is downloaded from CRSP, multiplying the monthly amount of outstanding shares and the month s end closing price per firm 2. Those monthly results were averaged to the quarter level. The book value of equity was downloaded using Compustat Bank Quarterly. Table 3.2 gives an overview of the independent variable when sorted and divided into quartiles. All quartiles are constructed per year, which explains the overlap of some interquartile ranges in the totaling row. When looking at the means, it is striking that only the first quartile displays a mean marketto-book ratio below 1 as a total. However, the first three quartiles all include market-to-book ratios below one. Since the average means are rather high in the second, third and fourth quartile it might be the case most of the financial institutions included in the sample are actually overvalued instead of undervalued. Still, the relationship between the market-to-book ratio and the repurchase transaction can be studied in terms whether or not a negative relationship exists between the two variables. As in Dittmar (2000), a negative relationship is expected concerning repurchase transaction and the market-to-book ratio. The coefficient being negative would implicate that the lower the market-to-book ratio and therefore the more undervalued a bank is, the more likely it is to engage in a larger repurchase transaction. 2 Following Ho, Liu and Ramanan (1997). 21

22 Table 3.2: Descriptive Statistics of Market-to-Book Ratio for four Quartiles The variable market-to-book ratio is summarized into four quartiles per year. The last row includes the total for each year. IQR describes the interquartile range, defined by range of the minimum value to the maximum value of the market-to-book ratio in that quartile. Mean describes the mean value of the market-to-book ratio within a particular quartile, specified per year. N defines the number of observations per quartile per year. YEAR LEGENDA First Quartile Second Quartile Third Quartile Fourth Quartile 2004 IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N IQR Mean N Total IQR Mean N

23 Firm Size One of the first papers investigating the relationship between a firm s size and the repurchase transactions was written by Vermaelen (1981). He argued that there is a positive relationship between undervaluation and repurchases, which is fueled by the idea that larger firms have less information asymmetries and therefore are less likely to be undervalued and in turn are also less likely to engage in a repurchase transaction. Vermaelen (1981) further argued that larger firms are less likely to have information asymmetries compared to smaller firms, due to the fact larger firms are more often covered by financial analysts. Renneboog, Simons and Wright (2007) agree with the view of Vermaelen (1981) by stating that smaller firms may have trouble with attracting potential investors to fund expansion. In order to resolve this lack of interest, small firms should engage in transactions to attract attention 3. However, the view of Vermaelen (1981) that small undervalued firms are more likely to engage in a repurchase transaction is not broadly supported in all its hypotheses. Two key papers of interest (Dittmar, 2000) (Ikenberry, Lakonishok, & Vermaelen, 1995) for this thesis report that firm size is positively related to the relative dollar volume of the repurchase transaction, since also large firms want to take advantage of mispricing. SIZE i,t 1 = LN(Total Assets i,t 1 ) (4) Following Dittmar (2000) 4, size is measured as the natural log of the total assets of the bank at time t-1, as denoted in expression 4. Total assets are downloaded from Compustat Bank Quarterly and lagged per company at the quarter level. Firm size is used as a control variable since it also has a proven influence on repurchase transactions. However, since the relationship between undervaluation and firm size is not widely supported, firm size is not used as a proxy for undervaluation. A positive effect of size on the repurchase transaction is expected as Dittmar (2000) found that large firms also try to exploit mispricing via a buy back History of Returns A history of low returns could indicate an increase in undervaluation due to information asymmetry (Stephens & Weisbach, 1998). However, Dittmar (2000) did not find exclusive evidence concerning a negative relationship between prior performance as a proxy for undervaluation defined by the return history and repurchase transactions. 3 Although the Renneboog, Simons and Wright (2007) paper is about the decision for UK firms to go private, they link undervaluation to firm size and therefore is used here as a complementary argument. 4 Dittmar (2000) is followed here since this paper of key interest for this thesis. 23

24 However, when focusing on repurchase announcements, Råsbrant (2013) did find a negative relationship between the pre-announcement cumulative abnormal return and the reaction of the market to the repurchase announcement 5. This indicates that there is a significant relationship between the return history of the company and the repurchase transaction and therefore this variable is included in the model as a control variable. RETURN i,t 1 = Value Weighted Return i,t 1 (5) Following Stephens and Weisbach (1998) return is defined as the return prior to the quarter the repurchase transaction is reported in, as denoted in expression 5. For this variable, monthly returns were downloaded using CRSP and are averaged on a quarterly basis. As in Dittmar (2000), the valueweighted return is used 6. A negative relationship between prior returns and the actual repurchase transaction is expected since Stephens and Weisbach (1998) found evidence for undervaluation being one of the results of poor performance Dividends In their paper about financial flexibility, Jagannathan, Stephens and Weisbach (2000) state that a firm declaring dividends engages in an ongoing commitment of redistributing a company s wealth to its shareholders, while an engagement to a repurchase program does not implicitly commits a firm to do so in the future. Therefore, performing buy backs implicate more financial flexibility for a firm than paying out dividends do. Lintner (1956) represented a model which argued that larger firms would have a higher propensity to pay dividends than smaller firms and those dividend-paying firms should have stable and higher cash flows. One factor influencing the firm s choice between dividends and repurchases is tax. Dividends and repurchase transactions are taxed the same at corporate level, however on the personal level dividend payouts are taxed more heavily than income from a repurchase transaction (Jagannathan, Stephens, & Weisbach, 2000). The personal tax level depends on margins and is not available in public databases. 5 Ikenberry, Lakonishok, & Vermaelen (1995) did found a similar relationship, as did Comment and Jarrell (Comment & Jarrell, 1991). 6 Dividends are excluded from the return measure since dividends are already controlled for by another independent, control variable. 7 Dittmar (2000) and Stephens and Weisbach (1998) do not find similar results concerning the relationship between prior returns and repurchases. Dittmar (2000) deserted this dissimilarity to the difference in measuring the returns. Since Stephens and Weisbach (1998) follow a quarterly measurement, their findings are followed here. 24