An event study on the response of common equity holders of US firms
after the announcement of a preferred stock issuance.
Bachelor Thesis Economics & Business Specialization Economics and Finance
Kevin Lechner, 11040335 Supervisor: Jeroen Ligterink
1
Statement of OriginalityThis document is written by Student [Kevin Lechner] who declares to take full responsibility for the contents of this document. I declare that the text and the work presented in this document are original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.
2
AbstractThis thesis examines the response of common equity holders of US firms after the announcement of preferred stock issuance. To examine the response of common equity holders an event study is performed to calculate the cumulative abnormal common stock returns using data from 2010 until 2017. I found an average common equity effect of -1.38%, which confirms the main hypothesis of this thesis. The response of common equity holders is more negative when the event window is shorter and the estimation window to compute the normal returns is larger. Furthermore, to investigate the effect of other factors influencing the cumulative abnormal common stock return, a cross-sectional regression analysis is performed. The results are in line with existing empirical evidence found by Kallberg, Liu, Villupuram (2013) and Linn and Pinegar (1988). My results indicate that the response of common equity holders is weakened when the issued preferred stock is non-convertible, and the amount of information asymmetry between the firm and its common equity holders is low. I also introduced a new variable, Redeemable, to examine the effect of a call option attached to the issued preferred stock. The results showed that this new variable has a positive significant effect on the response of common equity holders.
3
Table of Contents1. Introduction ... 4
2. Literature review ... 6
2.1 The role and use of preferred stock in the capital structure... 6
2.2 Empirical evidence on security returns after a preferred stock issuance ... 9
3. Methodology ... 12
3.1. Research Design ... 12
3.2 Variables in the cross-sectional regression analysis ... 14
3.3 Sample Selection ... 17
4. Results Data Analysis ... 18
4.1 Descriptive Statistics ... 18
4.2 Cross-sectional Regression Analysis Results and Discussion ... 20
5. Conclusion... 23
Bibliography ... 25
4
1. IntroductionThe United States preferred stock market is approximated to be a 250-billion-dollar market according to a report published in October 2015 by S&P Global and has grown 400% compared to 1990 (Brzenk & Soe, 2015). However, the preferred industry is still considered miniscule compared to the 22.71 trillion-dollar US equity market (Brzenk & Soe, 2015). Besides, investors are often unacquainted with the preferred market. The preferred stock market is less clear compared to the much larger common stock market, as information is scarce, and the purpose of preferred stock can be difficult to interpret. Preferred stock is a hybrid security with both debt and equity features. Preferred stock as a way of external financing has an increasing influence in the capital structure of US publicly-traded firms. For example, preferred stock is more commonly used to diversify a firm’s capital structure.
However, research on preferred stock issues and the role of preferred stock in the capital structure is not fully explored. Most of the published papers are extensions on the foundation of the modern capital structure theory, the trade-off theory and pecking order theory. Academics mainly examine the decision making of firms on the choice of the optimal capital structure. In my thesis I examine if the cumulative abnormal returns of common stocks are significantly affected after the firm’s announcement of a preferred stock issuance. A significant effect in the cumulative abnormal returns would imply a positive or negative signal from common equity holders about the firm’s change in capital structure. The cumulative abnormal returns of common equity holders are measured by performing an event study. Thereafter, the measured cumulative abnormal returns are regressed onto multiple variables to determine whether the response of equity is strengthened or weakened by certain factors, such as the firm’s financial ratios and the relative size of the issuance. The input variables and the regression model are presented in the methodology section.
Preferred stock is a security with a variety of characteristics. Preferred stock generally has options attached, such as a cumulative dividend and convertibility (Crabbe, 1996). By purchasing a preferred stock, the investor often receives a periodic fixed preferred dividend. However, the firm is not obligated to pay a periodic fixed dividend. This gives the firm more flexibility compared to issuing debt. The firm may decide not to pay dividends, whereas they are obligated to pay interest on debt. Thus, receiving a periodic fixed dividend is comparable to a regular bondholder receiving a coupon, according to Crabbe (1996). But, the preferred investor takes on a larger risk compared to investing in bonds and receives a higher yield for taking on more risk (Bajaj, Mazumdar, & Sarin, 2002). Moreover, the issuance of preferred stock does not affect ownership dilution and is cheaper than a common stock offering (Brzenk & Soe, 2015). The study of Bajaj, Sarin and Mazumdar (2002) found that the average preferred issuance cost lies between the direct cost of equity and debt issuance. Additionally, the
5
preferred equity market is much more liquid compared to the bond market (Bessembinder, Kahle, Maxwell, & Xu, 2009).This way of raising funds offers a new perspective in the discussion on the optimal capital structure. Current research on the optimal capital structure is primarily centered around the optimal proportion of debt to equity, not extensively analyzing the role of preferred stock as a hybrid security. Moreover, each research on preferred stock is often centered about one characteristic of the hybrid security, such as convertibility. In my research I want to further contribute to the research of Kallberg, Liu and Villupuram (2013), who studied the short-term response of equity and bondholders of companies from different industries after the announcement of a preferred stock issuance. Although, they did research on the response of equity and bondholders of different companies from different industries, they did not specifically focus on the US market and the response of common equity investors after the 2008 financial crisis. In this paper, it will be further researched what the response of common equity holders is to a preferred stock issuance announcement of US firms using the data from 2010 until 2017.
When a preferred stock issue is interpreted as equity, rational equity investors assume a decrease in leverage. As a result, the issue affects the firm’s value. A decrease in leverage is seen as a negative signal by equity investors (Masulis & Korwar, 1986). Based on the results of Masulis and Korwar (1986), Linn and Pinegar (1988) and Kallberg, Liu and Villupuram (2013) I expect a negative abnormal return of common stocks after the announced preferred stock issuance. This is the main hypothesis of my thesis. However, the strength of the negative response is determined by the type of issued preferred stock. On average, Kallberg, Liu and Villupuram (2013) found a negative equity effect of -0,65%. But, If the issue is straight preferred stock, it is very likely to find an insignificant result based on the analysis of Kallberg, Liu and Villupuram (2013). But, if the preferred stock issue has certain options attached, the response of equity holders is assumed to be significantly strengthened or weakened.
The results of my research will further attribute to the literature on preferred stock issuance and the empirical evidence contributes to the short-term effect of a preferred stock issue on common stock returns. Besides, the research of Kallberg, Liu and Villupuram (2013) and Linn and Pinegar (1988), very little research has been published on preferred stock issues and its effects on other security returns. Moreover, no research has been done on the response after a preferred announcement of equity investors in the period following the financial crisis of 2008. Also, the effect of a call option attached the preferred stock has not been examined before. Furthermore, the primary focus of researchers studying equity issues is on the effects of seasoned equity offerings (SEO). There are many extensions on the paper of Masulis and Korwar (1986) investigating SEOs. The research on the effects of preferred stock issues is nihil. However, importance of preferred stock is growing in the capital structure
6
of firms as is shown in the percentage growth since 1990. The outcomes from the data analysis will show that equity markets are taking preferred issues seriously and the results will support the significant results found by Kallberg, Liu and Villupuram (2013), who used data from before the financial crisis. The dataset used for this research consists of data from 2010-2017 and will therefore add value to the current discussion on hybrid securities.In the next section, a literature review is presented in which the role and use of preferred stock and the empirical evidence on security returns after a preferred stock announcement is discussed. Following the literature review, the methodology and the data sample will be discussed. Thereafter, the outcomes will be presented. In the last section, conclusions will be made and suggestions for future research will be argued.
2. Literature review
2.1 The role and use of preferred stock in the capital structure
In order to go into detail on the role of preferred stock in the firm’s capital structure, a general overview of factors influencing the chosen way of financing is first discussed. The optimal capital structure is still the subject of ongoing research. According to the renowned paper and propositions of Modigliani and Miller (1953) debt and equity financing are perfect substitutes. Hence, firm value is unaffected by the choice of financing. However, the first proposition of Modigliani and Miller (1953) is built on the assumption of perfect markets. In practice, debt and equity financing are dissimilar as markets are imperfect. Firms have to account for market imperfections such as taxes, financial distress costs, transaction costs and agency costs. Considering these market imperfections, the trade-off theory was first introduced by Kraus and Litzenberger (1973). In their paper, Kraus and Litzenberger (1973) investigated the optimal relation between corporate tax benefits and financial distress costs. To build further on the trade-off theory of Kraus and Litzenberger (1973), Myers and Majluf (1984) introduced the pecking order theory. The pecking order theory argues that the way of financing is dependent on the amount of information asymmetry between the firm and its stakeholders. Myers and Majluf (1984) found that firms prefer internal financing over debt financing and debt financing over equity financing. The trade-off theory and pecking order theory are considered the foundation of the modern debate on the optimal capital structure of a firm.
Based on the conclusion drawn by Myers and Majluf (1984), firms are reluctant to issue equity. But why and when do firms issue equity when they generally see it as their last resort? This question is researched by Dittmar and Thakor (2007). Dittmar and Thakor (2007) proposed a new theory in which they argue that firms will either issue equity or debt based on
7
the alignment with the investor’s views. The chosen way of financing will according to Dittmar and Thakor (2007) maximize the degree of agreement between the firm and its investors. Their model predicts that when the stock price is high, and the degree of agreement is high, managers are more likely to issue equity rather than debt. Because, Dittmar and Thakor (2007) state that debt financing has two sides; the gain of the tax benefit, but the loss of autonomy for the manager. Dittmar and Thakor (2007) tested four hypotheses in their paper of which two are relevant to this thesis. Their first hypothesis is to test the common consensus of equity issuances: Firms will issue equity when their stock price is high. This statement was introduced by Myers and Majluf (1984) and proven by many other researchers such as Baker and Wurgler (2002). Baker and Wurgler (2002) studied the phenomenon of managers trying to time the equity market. Their theory argues that firms will issue equity when their market value is high and will issue debt when their market value is low. Therefore, the current capital structure of a firm is based on the historical market value of a firm. Their main finding is that highly-leveraged firms will issue equity when their market value is down relative to their book value. On the other hand, Baker and Wurgler (2002) found that low-leveraged firms have the incentive to issue equity when their market-to-book ratio is high. The results found by Dittmar and Thakor (2007) are also consistent with the general equity consensus. Moreover, Dittmar and Thakor (2007) found a significant difference in the market-to-book ratio between equity issuers and debt issuers. Besides, the market adjusted stock returns are significantly higher following the period after of the issuance for firms with a high market-to-book ratio. These results contradict the foundational theories on capital structure, the trade-off and pecking order theory (Dittmar & Thakor, 2007). The second hypothesis Dittmar and Thakor tested is: The issuance of equity is not dependent on the stock price when the degree of agreement is high. The results of Dittmar and Thakor (2007) show that more firms than expected will issue equity when the degree of agreement is high even though their stock price is low. They found the opposite result for debt issues.Thus, the stock price and the amount of agreement determines whether a firm will issue equity. But, preferred stock is a hybrid security consisting both of debt and equity features. The conclusions drawn by Dittmar and Thakor (2007) do therefore not fully apply to preferred stock. Linn and Pinegar (1988) state several potential motives why firms would issue preferred equity. Firstly, the firm wants to send a signal to the market about the firm’s prospects after changing the amount of leverage in the capital structure. A preferred issue could indicate financial distress or an additional constraint for the firm. This motive is studied in more detail in the papers of Smith (1986) and Blau and Fuller (2008). Also, distressed firms are able to issue preferred stock while issuing debt would trigger default (Blau & Fuller, 2008). And, preferred stock is not bound by rigid covenants, which is the case for bonds (Crabbe, 1996). Firms are allowed to omit a preferred fixed dividend without instantly triggering default of the
8
firm. This risk of not receiving a periodic fixed dividend is represented in the offered yield (Bajaj, Mazumdar, & Sarin, 2002). Secondly, Linn and Pinegar (1988) state that firms can quickly raise capital when they experience an unexpected shift in demand of funds. Brenzk and Soe (2015) argue that for instance banks can quickly increase their capital to meet regulatory requirements by issuing preferred stock. Thirdly, Linn and Pinegar (1988) argue that the type of preferred stock is dependent on the industry in which the firm operates. Where utility firms frequently issue straight preferred stock, financial firms are likely to issue preferred stock with an adjustable rate (Linn & Pinegar, 1988). The type of preferred stock signals the role of the hybrid security in the capital structure. Additionally, the accounting treatment accentuates this role in the capital structure. Firms decide if the preferred stock is treated as equity, debt or hybrid security (Kallberg, Liu, Villupuram, 2013). The announcement of the accounting treatment firm reveals signals to equity holders, Linn and Pinegar (1988) argue. Also, Miller and Rock (1985) argue that a large unexpected issue depicts a gap between actual and expected cash flows of a firm. Hence, a large unexpected issue evokes a negative signal by equity holders, because firm value is expected to decrease (Miller & Rock, 1985). Equity investors are skeptical of unexpected issues and take the accounting treatment of the issuance into consideration (Miller & Rock,1985). For example, an issue of straight preferred stock leads to an insignificant result, as equity investors interpret the issue as debt, Kallberg, Liu and Villupuram (2013) found. Linn and Pinegar (1988) found that the riskier the type of preferred stock issued, the more doubtful the equity or debtholder responds.The reason why firms issue preferred stock is further explored by Houston and Houston (1990). In 1988 they conducted a survey under 200 US firms asking the reason behind their preferred stock issue. The answers were put into two categories for which the proceeds were used: Restructuring the capital structure or acquiring assets. The proceeds of preferred stock issuances were mostly used for acquiring other assets, Houston and Houston (1990) found. An example of acquiring other assets is a financial take-over. For financial companies, preferred stock falls under the category of regulatory capital and is thus a substitute to common stock (Houston & Houston, 1990). Nowadays, bank regulators call this tier 1 capital (Brenzk & Soe, 2015). Issuing preferred stock does not thin the ownership of the firm and does not increase the leverage ratio, Brenzk and Soe (2015) state. Consequently, banks often favor to issue preferred stock over common stock in order to comply to regulations. Moreover, Houston and Houston (1990) argue that preferred stock is issued to continue corporate growth. Specifically, the proceeds are used for mergers & acquisitions. However, Houston and Houston (1990) conclude that the survey showed that the preferred stock proceeds are used for all kinds of corporate finance activities.
Furthermore, Houston and Houston (1990) examine the traditional view on preferred stock issues. Preferred stock is perceived as having a tax disadvantage to debt and is less
9
flexible to common stock. Also, preferred stock has no tax advantage over common stock, Houston and Houston (1990) argue. But why would a firm then issue preferred stock, they asked. According to Fooladi and Roberts (1986), this perspective on preferred stock versus debt is too conservative. Firms issuing preferred stock have special tax incentives and therefore issue preferred stock instead of debt, they argue. Fooladi and Roberts (1986) state that firms with a below-average tax rates are likely to issue preferred stock and investors with an above-average tax rate are more likely to buy preferred stock. Fooladi and Roberts (1986) state that this is because firms can benefit from a so-called ‘’pass through’’ tax benefit mechanism. However, the research is done using data from the 1980’s. Tax regulations change over time and therefore the tax hypothesis touched upon by Fooladi and Roberts (1986) and supported by the results of Houston and Houston(1990) is unlikely to hold anymore.2.2 Empirical evidence on security returns after a preferred stock issuance
A preferred stock issue changes the firm’s capital structure. As a result, this announced change evokes a response of common equity holders. Before analyzing the effect on security returns after a preferred stock issue, the general effect of a capital structure change on security returns is first discussed. The paper of Masulis (1980) studies the effect of a capital structure change on outstanding securities of the firm. A change in capital structure of a firm, changes the probability of bankruptcy. In effect, the revision of probability affects the firm’s value (Masulis, 1980). The market values must therefore be equal to the market values of the firm’s securities, Masulis (1980) states. Masulis (1980) argues that a firm’s investment decisions differ when there is a change in capital structure. Consequently, the rate of return variance is affected (Masulis, 1980). When the variance of return decreases, debt and preferred stock increase in value, while common stock is negatively affected (Masulis, 1980). He defines this phenomenon as a secondary redistribution effect.
Masulis (1980) found in his results that when debt is issued, the common stock price tends to increase. Of his sample, 80% of the common stocks had a positive return following the announcement. When debt was announced to decrease, common stocks responded with a negative return according to Masulis (1980). Both the positive and negative returns are significant. According to Masulis (1980) the results he found where both caused by a tax shield effect and a redistribution effect. This effect was present across debt, preferred stock and common stock assets. Also, Masulis (1980) states that the change in returns is larger when the tax shield effect and wealth effect are reinforced by each other. A decrease in leverage is interpreted as a negative signal to stockholders, because stockholder wealth is not maximized (Masulis, 1980). Furthermore, He analyzed convertibles in his paper and the relevance to the
10
redistribution of wealth in the firm. According to Masulis (1980), preferred stockholders are worse off when common stock can be converted into debt. On the grounds that debt is senior to preferred stocks. Additionally, the paper Masulis (1980) only studied two situations: Recapitalizations and exchange offers within the firm. The effect on security returns after an issuance is not the focus of his paper.However, the effects of a preferred stock issue on common stockholder and existing preferred stockholder wealth is investigated in the paper of Linn and Pinegar (1988). Their motivation behind their research is similar to Kallberg, Liu and Villupuram (2013). Linn and Pinegar (1988) are of the opinion that there is little attention given to the effects and possibilities of preferred stock as a way of external financing. In their study, Linn and Pinegar (1988) compare different industries and make a distinction between the different types of preferred stock using issuance data from 1963 to 1984. Linn and Pinegar (1988) found that each industry favors a certain type of preferred stock. For example, more than fifty percent of the preferred stock issues of financial firms are preferred stock with an adjustable dividend rate. To add, the preferred stock issue has an insignificant effect on the common stock return of financial firms, but a positive significant effect on banking firms (Linn & Pinegar, 1988). Furthermore, they conducted a sectional analysis on the abnormal returns. Their cross-sectional analysis suggests that the relative issue size combined with the ratings announcement, reveal information to equity and debt holders about the firm’s value. This strengthens the theory of Myers and Majluf (1984) about the role information asymmetry. The amount of information asymmetry determines the strength of the response.
Linn and Pinegar (1988) found that the average abnormal return in the preannouncement period is insignificant. The abnormal return within the announcement period is significant, but small according to Linn and Pinegar (1988). Similar outcomes are also found in the paper of Kallberg, Liu and Villupuram (2013). However, the sign of the abnormal return is largely dependent on the chosen industry and the type of preferred stock. For example, a preferred issue in the utility sector yields a significant positive return, while the return in the industrial industry is negative (Linn & Pinegar, 1998). Because, the utility sector is different compared to other industries in terms of regulatory environment. Overall, they conclude that the equity market responds differently to preferred stock issuances. Linn and Pinegar (1988) also examined the effects on wealth-redistribution and price-pressure but did not find any support for these hypotheses.
More recently, Kallberg, Liu and Villupuram (2013) analyzed the response of equity and bondholders to the announcement of a preferred stock issuance. Kallberg, Liu and Villupuram (2013) used abnormal common stock returns and the Credit Default Swap (CDS) spread to examine the response of the equity and bondholders. Since, preferred stock has both characteristics of debt and equity, they argue that the effect on firm value is ambiguous.
11
Moreover, they studied the short-term reaction of equity and bondholders. Kallberg, Liu and Villupuram (2013) used multiple event windows, but only found significant abnormal returns in the (-1,1) window. Kallberg, Liu and Villupuram (2013) argue that the interpretation of the preferred stock affects the response of equity holders. When equity holders perceive the issuance as equity, the response is either negative or insignificant, they found. The response depicted in the abnormal stock return is insignificant when the firm issues straight preferred stock. On other hand, equity holders respond negatively to options such as convertibility (Kallberg, Liu, Villupuram, 2013). In their data analysis, Kallberg, Liu and Villupuram (2013) found a negative average equity effect of -0,65%. However, the response of equity holders is positively influenced by the transparency of the firm (Kallberg, Liu, & Villupuram, 2013). Convertibility and the accounting treatment (debt, equity or hybrid) of the hybrid security negatively influence the response of the equity holders. The overall effect of the preferred stock announcement leads to an increase in the firm’s value (Kallberg, Liu, & Villupuram, 2013). But, the increase in the firm’s value is not driven by financial innovations or regulatory decisions, Kallberg, Liu and Villupuram (2013) conclude.Newly issued preferred stock can be different type then the existing preferred stock in a firm. Therefore, a possibility arises to replace existing preferred stock for a new type of preferred stock. The replacement of preferred stock for a new type of preferred stock is examined in the paper of Irvine and Rosenfeld (2000). Irvine and Rosenfeld (2000) studied the effect of Monthly Income Preferred Stock (MIPS) on the common stock price of the firm. The issuing firm used the MIPS in order to retire existing preferred stock. This new innovation introduced by Goldman Sachs is debt repackaged as preferred stock according to Irvine and Rosenfeld (2000). Issuing MIPS has specific tax benefits as the issuer may deduct preferred dividends as interest. However, a complicated structure is necessary, making use of a special-purpose vehicle (Irvine & Rosenfeld, 2000). The focus of the research of Irvine and Rosenfeld is primarily on the relationship between the firm’s value and the redemption of straight preferred stock by issuing MIPS. According to Irvine and Rosenfeld (2000), the weighted average cost of capital should decrease and in return the firm’s value should rise. As a result, the abnormal return of common stock adapts positively. However, Irvine and Rosenfeld (2000) state that the response of the equity holders primarily depends on the announced use of the proceeds. When the proceeds are used to repay debts, the common stock price is negatively affected (Irvine & Rosenfeld, 2000). This is also concluded in the research of Masulis (1980), because a leverage decrease is perceived as a negative signal to stockholders. Also, Irvine and Rosenfeld (2000) argue that the underlying argument behind this negative response is the increased flexibility for the firm. It allows the firm to potentially omit a dividend payment (Irvine & Rosenfeld, 2000). However, this decision to issue preferred stock gives off a different signal to investors, who might expect a possible bankruptcy. If the firms use the proceeds for
12
corporate purposes, the effect is ambiguous (Irvine & Rosenfeld, 2000). Also, they found a negative common stock return occurs when the proceeds are used to repay long-term debt. Irvine and Rosenfeld (2000) found that MIPS issuers are highly-leveraged firms with the motivation of gaining more financial flexibility. In addition, the issuance of MIPS exposes equity holders to higher agency costs, they conclude. This should negatively impact equity holder. Moreover, Irvine and Rosenfeld (2000) found that the negative abnormal common stock returns of firms are strengthened when they have a lower credit rating.3. Methodology
3.1. Research Design
To measure the effect of a preferred stock issuance announcement, the cumulative abnormal return (CAR) is used. The way of computing the cumulative abnormal return is done performing an event study. The sample of US firms will be discussed in the next section. The event date is determined by the filling date of the preferred stock issuance. In this research all preferred stock issues are assumed to be accounted as equity issues. An event window to compute the cumulative abnormal return differs per research. Kallberg, Villupuram and Liu (2013) used a 3-day (-1,1), 6-day (2,3) and 7-day (2,4) event window. The event window includes the announcement date, day 0. However, only the 3-day event window (-1,1) resulted in a significant t-statistic. Also, Linn and Pinegar (1988) only found a significant event window by including the announcement in the event window. Hence, I decide to use a 3-day event window (-1,1) and a 5-day event window (-2,2). In order to compare the cumulative abnormal returns, the normal or expected returns are required. The normal returns can be calculated using a market model. Mackinlay (1997) argues that the market model parameters can be estimated over the 120 days before the event date but can be extended. Based on the example of Mackinlay (1997) an estimation window of 250 trading days (-280, -30) is used. To compare if there is a significant difference in the estimation period, another estim ation window of 120 trading days is used (-130, -10). The 3-factor market model of Fama and French (1993) will be used to compute the normal returns. The 3-factor market model to calculate the normal return is:
(1) 𝐸(𝑅𝑖) = 𝛼 + 𝛽1(𝑅𝑚 − 𝑅𝑓) + 𝛽2𝐻𝑀𝐿 + 𝛽3𝑆𝑀𝐵 + 𝜀
Where E(Ri) is the expected ‘normal’ return, Rf is the risk-free rate return, Rm-Rf is the market risk premium, HML is the high minus low risk factor and SMB is the small minus big risk factor.
13
The small minus big risk factor is used to copy the returns relative to size (Fama & French, 1993). The difference in returns between small stocks and big stocks with the same weighted-average book-to-market equity is portrayed in the SMB risk factor as suggested by Fama and French (1993). The high minus low risk factor resembles the returns relative to the book-to-market equity (Fama & French, 1993). The difference in returns between firms with a high book-to-market equity and firms with a low book-to-market equity with the same weighted-average size is depicted in the HML risk factor as suggested by Fama and French (1993).The calculated normal returns are necessary to compute the cumulative abnormal returns. The abnormal return is the difference between the realized return and the expected return. This is formulated in the following way:
(2) 𝐴𝑅𝑖,𝑡= 𝑅𝑖,𝑡 − 𝐸(𝑅𝑖,𝑡 | 𝑋𝑡)
Where Xt is the conditioning information derived from the chosen market model (Mackinlay, 1997). Since, the Fama French 3-factor market model is used, Xt is the market return of the stock. Ri,t is the realized return of a firm (i) on day (t) during the event window. E(Ri,t I Xt) is
the expected return of a firm (i) on day (t) during the event window calculated using the Fama French 3-factor market model. Combining these two calculation leads to the following cumulative abnormal returns:
(3) 𝐶𝐴𝑅 = ∑ 𝐴𝑅𝑖,𝑡 1 −1 𝐶𝐴𝑅 = ∑ 𝐴𝑅𝑖,𝑡 2 −2
Using the WRDS Event Study tool, the cumulative abnormal stock returns are calculated. Nextly, the CARs are used in a cross-sectional regression analysis to investigate which other variables affect the returns besides the Fama French 3-factor market model. The cross-sectional regression will be run in Stata. An Ordinary Least Regression (OLS) with non-robust standard errors will be performed. This is tested using a Breusch-Pagan (1979) / Cook-Weisberg (1983) test for heteroskedasticity. Also, the normality of the residuals is tested using a Jarque-Bera (1980) test, because the sample size is small. There will be two event windows and two estimation windows, which in total leads to four regressions in Stata. The outcomes will be discussed in the results section.
14
3.2 Variables in the cross-sectional regression analysis
To examine whether the cumulative abnormal common stock return (CAR) is affected by other factors than the Fama French 3-factor market model, a cross-sectional regression analysis is performed. The cross-sectional regression variables are based on the research of Kallberg, Villupuram and Liu (2013) and Linn and Pinegar (1988). Also, I introduce a new dummy variable, Redeemable, to examine whether a call option has a significant effect on the response of equity investors. The CAR is the dependent variable in the model. In total eight explanatory variables are implemented in the cross-sectional regression model. The explanatory variables will be discussed individually in this section.
In order to measure the importance and influence of information asymmetry, three variables are added in the regression model. The first variable is the Earnings to Price ratio of the previous year. This is the long-term expected earnings per share divided by the market price per share at the end of the year before the event year. The variable 12-month forward Earnings per Share from DataStream is divided by the stock price of the firm at the end of the year before the event year. According to Kallberg, Villupuram and Liu (2013) the higher the earnings to price ratio, the higher future expected earnings and the lower the information asymmetry. Based on the research of Dittmar and Thakor (2007), the higher the degree of agreement, thus the lower information asymmetry, the better. Consequently, a preferred stock announcement should be positively affected by this variable for the CAR of common stocks.
Secondly, to measure information asymmetry, the dividend to free cash flow ratio is used as a variable. Firms have less information asymmetry when they pay out dividends compared to firms who are not paying dividends, according to Booth and Chang (2011). The difference between firms is represented in the dividend to free cash flow ratio. The dividend to free cash flow ratio at the end of the previous year is used. The positive size of the ratio should positively affect the announcement of a preferred stock issue for the CAR of common stocks, as a higher ratio indicates lower information asymmetry. This theory is reinforced by the model of Miller and Rock (1985), where they found a smaller change in returns for dividend paying firms than non-dividend paying firms. The data on the ratio of the US firms derives from DataStream.
Thirdly, the market capitalization of a firm is expected to influence the response of equity and bondholders. Market capitalization indicates information asymmetry, Huang and Tompkins (2010) conclude. They argue that financial analysts focus predominantly on larger firms. Consequently, Larger firms try to decrease their amount of information asymmetry. Kallberg, Villupuram and Liu (2013) included the variable market capitalization as an addition to the market model in terms of measuring risk and liquidity. The market capitalization of a firm (i) is the value at the end of the fiscal year prior to the event year (Huang & Tompkins, 2010).
15
I expect that the size of the market capitalization of a firm should positively affect the response of common equity investors. The data on the market capitalization is retrieved from DataStream.To build further on the market capitalization of a firm and the effect on the cumulative abnormal stock return, it is also important to investigate the effect of the size of the issue. Therefore, a control variable size is included in the regression model. Size is the total value of the preferred stock issue relative to the market capitalization of a firm before the issue. A firm issues stock when it thinks its shares are overvalued, according to Huang and Tompkins (2010). However, this depends on the accounting treatment of the issued preferred stock (Kallberg, Liu, Villupuram, 2013). The larger the issue size relative to the market value of a firm then investors assume that the difference between actual and expected cash flows is large (Linn & Pinegar, 1988). Hence, if a firm labels preferred stock as equity, issuing preferred stock should evoke a negative response by equity holders. I expect a response of equity holders which is unequal to zero, because Linn and Pinegar (1988) found mixed results. The data on the issue size is generated from the Thomson One database.
Another control variable in the regression model is the interest coverage ratio, ICR. The interest coverage is the earnings before interest and taxes (EBIT) divided the total interest expense. A high ICR indicates that the probability of a wealth transfer from stockholders to bondholders is low (Kallberg, Liu, Villupuram, 2013). Consequently, a high ICR should have a positive effect on the cumulative abnormal stock return after a preferred stock announcement (Kallberg, Liu, Villupuram, 2013). The data on the ICR ratio is retrieved from DataStream.
The response of investors is also dependent on the industry in which the firm is active. Hence, three dummy variables are included in the regression model. The first dummy variable, is a bank dummy. If the firm issuing preferred stock is a bank institution, the dummy variable takes on a value of 1. If the firm is not a bank, the dummy variable takes on a value of 0. The cross-sectional regression of Kallberg, Villupuram and Liu (2013) found a significant effect for this variable. The introduction of the bank dummy variable is based on the research of Kim and Stock (2012). Kim and Stock (2012) investigated the effect of the TARP program on bonds, common stock and preferred stock of banks. The TARP program was introduced by the US government during the 2008 credit crises to raise bank capital. Kim and Stock (2012) and Linn and Pinegar (1988) both found a positive response by equity holder after the announcement if the issuer is a bank. Hence, the dummy variable bank is expected to positively affect the cumulative abnormal return of common stocks.
Not only does the response of equity holders depend on the chosen macro industry, but the options attached to the hybrid security also determine the strength of the response. The two dummy variables, convertible and redeemable, are introduced in the regression model. The dummy variables take on the value 1 if the option is attached and 0 if the option is
16
not attached to the preferred stock issue. The options attached to the issues are found in the filed prospectuses available on the SEC website. Based on the paper of Linn and Pinegar (1988) convertible preferred stock issues have a negative effect on the cumulative abnormal return of bonds and common stock. Because, it allows preferred holders to convert their shares to common shares, which dilutes ownership. Hence, I expect a negative effect for equity abnormal returns if the preferred stock is convertible. Also, this result was found in the cross-sectional analysis of Kallberg, Liu and Villupuram (2013). Kallberg, Liu and Villupuram (2013) nor Linn and Pinegar (1988) investigated the effect of a call option on preferred stock. A redemption option allows the firm to recall the outstanding preferred stock after a determined period, but before the maturity date. Consequently, it allows the firm to reinvest against a more favorable rate in future. As a result, I expect equity holders and bondholders to respond positively to this option. Hence, I expect the beta of both dummy variables to be unequal to zero.The cross-sectional regression model is the following: (5)
𝐶𝐴𝑅 = 𝛽0 + 𝛽1𝑃𝑟𝑖𝑐𝑒 + 𝛽2 𝐷𝑖𝑣
𝐹𝐶𝐹+ 𝛽3𝑚𝑘𝑡𝑐𝑎𝑝 + 𝛽4𝑆𝑖𝑧𝑒 + 𝛽5𝐼𝐶𝑅 + 𝛽6𝐵𝑎𝑛𝑘 + 𝛽7𝐶𝑜𝑛𝑣𝑒𝑟𝑡𝑖𝑏𝑙𝑒 + 𝛽8𝑅𝑒𝑑𝑒𝑒𝑚𝑎𝑏𝑙𝑒 + 𝜀
Overall, I expect a negative response of equity holders on the announcement of a preferred stock announcement:
𝐻0: 𝐶𝐴𝑅 = 0 , 𝐻1: 𝐶𝐴𝑅 < 0
Tabulating the hypotheses of the variables in the cross-sectional regression analysis leads to the following:
Price Div/FCF Mktcap Size ICR Bank Convertible Redeemable
17
3.3 Sample Selection
The data on the preferred stock issues is generated using the Thomson One (SDC Platinum) database. Applying the screening and analysis tool, we are able to search for preferred stock issues. Following up, I filtered on preferred stock issue data between 01/01/2010 and 31/12/2017. 466 worldwide preferred stock issues are found. Of these 466 observations, 112 took place in the United States. However, the issue date is unequal to the announcement date, which is needed to determine the event window. Following up, a filter on the filing date is applied. As a result, 99 issues worldwide filings are found. 64 of those filings took place in the United States.
However, not all filings can be used for the research. The filings of preferred stock on which there is no credit rating or offered price present, are removed from the dataset. Kallberg, Liu and Villupuram (2013) removed all observations on which there was no CRSP data match. A similar method is applied to the found observations. Consequently, 35 observations remained in the sample.
The sample can be divided up into different macro industries: Real estate, Financials, Energy and Power, Healthcare and Consumer Staples. 15 observations of the sample (±40%) are issues of financial firms. Of those 15 financial issuances, 10 are bank issues. Banks have a different capital structure due to the heavy regulatory environment. 7 issues of the sample derive from Energy and Power companies, 8 issues of the sample derive from the Real Estate industry, 2 healthcare industry issues, 1 technology industry issue and 2 from the Consumer Staple industry. All preferred stock issues are from US firms listed on either the New York Stock Exchange or the Nasdaq.
18
4. Results Data Analysis4.1 Descriptive Statistics
There were 64 preferred stock issues filed in the period between 1st of January 2010 and 31st
of December 2017. However, only 35 observations of those 64 are usable for the regression analysis. In the table below, the descriptive statistics on the Cumulative Abnormal Returns and the other explanatory variables are presented. The kurtosis is measured as the deviation from the normal distribution. The table shows that the cumulative abnormal returns have a high kurtosis. This indicates that data has a sharper peak compared to the normal distribution. Hence, the tails of the normal distribution are flatter, meaning there are almost no tail events in the data. The negative skewness of the CARs indicate that the CARs are approximately negatively distributed. This indicates that the probability density of a negative effect is greater than of a positive effect. As a result, the mean is skewed to the left. This is visible in the negative mean of the CARs in the table on below.
The descriptive statistics table above shows an average equity response of -1,38% which is higher in the 3-day event window than the 5-day event window. Also, the response is more negative when the estimation window is longer. The results on the CAR confirm the main hypothesis of the thesis: Common equity investors respond negatively to a preferred stock issuance. As a result, the cumulative abnormal common stock return is negative. To test the significance of the abnormal returns, I used a Patell (1976) Z-test. The results can be found in the appendix and show significant abnormal returns on the day before the announcement and the announcement date. The standard deviation of the cumulative abnormal returns is nearly identical for the 4 periods. The gap between the minimum and maximum abnormal shows the strong difference in responses of equity holders to different preferred stock issues, which is in line with the statement of Linn and Pinegar (1988). The remaining control variables statistics are mixed.
19
The normality of the residuals in the OLS regression is examined performing a Jarque-Bera test (Jarque & Jarque-Bera, 1980). I have done this test, because the outcomes of the results above indicate a negative skewness and positive kurtosis deviating from the expectations of a normal distribution. The results can be found in the appendix and conclude that the error terms are normally distributed.A correlation matrix of the used variables is presented below. From the matrix, it can be concluded that there are some variables that are highly correlated, such as earnings to price ratio with the CAR. Therefore, the correlation of the variables is further investigated. To examine the high correlation, I looked at Variance inflation Factors (VIF). The result can be found in the appendix, but indicate a moderate multicollinearity, which does not significantly affect the regression models.
Table 2: Correlation Matrix
The variables Earnings to Price ratio and CAR, and Redeemable and CAR, have the highest correlation. This is not surprising because nearly all issued preferred stock in the sample have a clause, which allows the firm to repurchase the preferred stock after a predetermined period. Because, this allows the firm to repurchase preferred stock and reissue securities against a more favorable rate in the future, which is a strong positive signal to common equity holders. As a result, the call option is strongly positively related to the abnormal return of common stocks. Moreover, the variables Earnings to Price ratio and the CAR are strongly correlated. Also, this is not surprising because the higher the future earnings per share to the current stock price, the lower information asymmetry (Kallberg, Liu, Villupuram, 2013). The amount of information asymmetry strongly determines the response of the equity investors. Of the three variables used to measure the influence of information asymmetry on the response of common equity holders, Earnings to Price ratio shows the highest correlation. The other two variables, Div/FCF and Market Cap, show a slight positive correlation with the CAR.
To determine whether robust standard errors are required to correct for heteroskedasticity of the residuals, I performed a Breusch-Pagan (1979) /Cook-Weisberg (1983) test for heteroskedasticity. The results are found in the appendix but show that non-robust standards errors are allowed.
20
4.2 Cross-sectional Regression Analysis Results and Discussion
In sections 3.1 the research set-up is discussed. I used two event windows in combination with two estimation windows. As a result, four regressions are performed to obtain the regression results. A difference occurs between the two event windows, but there is a less significant difference between using different estimation windows. The significant variables remain significant, while almost all insignificant variables stay insignificant. However, the two event windows lead to significant differences between the models. In the shorter event window, more variables show significance. On the contrary, the disparity between the two event windows did not lead to sign changes of the coefficients. Of each event window, two models are presented. One model is the combination of the event window with the first estimation window and the second model is the combination of the event window with the second estimation window.
In table 3, the regression models with dependent variable CAR (-1,1) are presented. The coefficients derived from the regression all have the hypothesized sign, regardless of the estimation window. However, not all coefficients are significant. The most significant variable is the earnings to price variable, which is significant for both estimation windows with a p-value smaller than 1%. The earnings to price ratio significantly positively affects the response of equity holders. This result is in line with the results found by Kallberg, Liu and Villupuram (2013), who argued that a higher earnings to price ratio indicates a good future prospect of the firm, hence the equity investors responds less negative to a preferred stock issue announcement. The response is less negative because overall equity investors respond negatively to a firm announcing a preferred stock issuance.
Secondly, the coefficients of the dummy variables are in line with previous researches. Kim and Stock (2012) and Linn and Pinegar (1988) found that preferred stock issues by banks are interpreted as a positive signal by both debt and equity investors. This result is apparent in the positive coefficient but is small and insignificant. Therefore, the hypothesis on the banking dummy cannot be rejected based on this data analysis but is in line with the expectations made in section 3.2.
The dummy variable convertibility is negative and significant. Convertibility strengthens the negative response of equity holders, which was already found by Linn & Pinegar (1988) and Kallberg, Liu and Villupuram (2013). The convertible option allows preferred stockholders to convert their preferred shares into common stock. As a result, ownership is diluted which negatively affects existing common stockholders. Therefore, common equity holders respond more negative to this particular option. However, the significance decreases when the estimation window is longer. The redeemable option is significant in model 2, but insignificant
21
in model 1. This confirms the prediction on the newly introduced variable Redeemable, which was not examined in previous researches. The call option allows firms to repurchase outstanding preferred stock after a pre-determined period, but before the maturity date. Consequently, the firm can refinance against a favorable rate, which is positive for the future of the firm and inherently for common equity holders.Thirdly, the effect of the control variables Div/FCF, MarketCap and ICR is positive, but insignificantly small. The positive effect of the dividends to cash flow found by Miller and Rock (1985) cannot be confirmed with my regression models. The same holds for MarketCap and ICR. In my regression model the market capitalization of the firms has a very little positive insignificant effect on the response of common equity holders. Huang and Tompkins (2010) found a negative effect between market capitalization and a SEO, but this result was insignificant. Although, a preferred stock issuance is perceived differently than a SEO, both outcomes are insignificant. Kallberg, Liu and Villupuram (2013) found a significant positive effect of the ICR on the response of equity holders, when the preferred stock is treated as equity. In my results I found a positive but insignificant effect. Hence, the regression outcome cannot support the results by Kallberg, Liu and Villupuram (2013) which show that equity investors respond significantly less negative when the ICR of a firm is high. Therefore, the hypotheses of the three control variables cannot be confirmed with the found outcomes. Increasing the sample size or time period might potentially lead to a significant effect of these three control variables.
Lastly, the variable Size is positively small but significant at a 10% significance level. The prediction on the variable Size was that the coefficient is unequal to zero. Therefore, the hypothesis is only rejected with a significance level of 10%. This positive effect on the cumulative abnormal return of common stocks is in contrast to the foundings of Miller and Rock (1985). They argued that a large issue is perceived as a negative signal, because the actual cash flow differs from the expected cash flow. As a result, firm value is expected to decrease. This leads to a negative response of equity investors. Linn & Pinegar (1988) found a small positive coefficient for the variable size for the total sample, but this outcome was also insignificant.
22
Table 3: Regression results with dependent variable CAR derived from event window (-1,1) & (-2,2). CAR (-1,1) is combined with estimations windows 250 trading days (Model 1) and 120 trading days (Model 2). Model 3 is the CAR (-2,2) combined with estimation window 250 trading days and the estimation window 120 trading days is used in Model 4.23
In table 3, the regression models with dependent variable CAR (-2,2) are also presented. The signs of the coefficients are similar to the regression models with dependent variable CAR ( -1,1). However, less variables show significance compared to the regression models with CAR (-1,1) as dependent variable. This could indicate that the effect of the preferred stock announcement is only short-lived. In both models only the earnings to price, redeemable and the regression intercept are significant. This could indicate that equity investors primarily focus on the long-term changes for the firm. Both the earnings to price ratio and the redeemable variable are of importance for the future prospects of the firm.Overall, the event window (-1,1) combined with both estimation windows yield the higher R-squared and adjusted R-squared. Model 2 is the model with the best goodness-of-fit compared to the other three models. However, all four regression models have a significant F-statistic, which indicates that the four regressions all fit the data.
5. Conclusion
In conclusion, I examined the effect of a preferred stock issuance on the abnormal returns of a common stock of US firms between 2010 and 2017. I found that on average, the cumulative abnormal return of common stock after the firm announces a preferred stock issuance is -1,38%. The negative response of equity investors is stronger when the event window is narrower as can be seen in the descriptive statistics table in section 4.1. I used the event study tool of WRDS to compute the Cumulative Abnormal Returns (CAR) for common stocks. To compute the normal returns, the Fama-French 3-factor model is used. The other explanatory variables are derived from the DataStream database and are based on the accounting data from the year before the event year.
Furthermore, the response of equity investors is significantly strengthened when the issued preferred stock is convertible, while a call option leads to a significantly weakened response by common equity holders. The dummy variable Bank shows a positive but insignificant effect on the CAR. The same holds for the three control variables: Div/FCF, ICR and Market Cap. Size is the only control variable which shows significance in the 3-day event window but only at a 10% significance level.
In contrast to the research of Kallberg, Liu and Villupuram (2013), no separation is made between accounting treatments of preferred stock. In my analysis I assumed all preferred stock issues to be treated as equity. Therefore, the variable Market Capitalization potentially yields an insignificant result. However, in comparison with previous research done by Linn and Pinegar (1988) and Kallberg, Liu and Villupuram (2013), similar results are found in my data analyses. For instance, the signs of the coefficients found in the regression
24
analyses are in line with coefficients found by Kallberg, Liu and Villupuram (2013) and my hypotheses formulated in section 3.2. However, not all explanatory variables are significant.Moreover, the analyses did not incorporate the whole spectrum of factors influencing the response of equity holders of a firm announcing a preferred stock issuance. Where Kallberg, Liu and Villupuram (2013) did take into account the accounting treatment of the issuance, but they did not take into consideration the relative size of the issue and the potential effect of a call option attached to the preferred stock. In future research, it would be interesting to see the influence of specific options attached to preferred stock, such as perpetual preferred stock and adjustable dividend rate preferred stock, and how these options affect the response of common equity holders. Also, the significance of the variables used in my regression models can be further explored by increasing the sample size. For this research only 35 observations were used, which could explain the insignificance of some input variables. By increasing the sample size, the support for the found results in this research could emphasize the growing importance of preferred stock in the capital structure and the serious response of common equity holders.
Not only common equity investors are affected by a change in the capital structure of a firm. Debtholders also respond to a firm announcing a preferred stock issuance. The response of debtholders is visible in the Credit Default Swap spread of firms or in the bond returns. Kallberg, Liu and Villupuram (2013) examined the Credit Default Swap spreads but did not examine the bond returns. It would be interesting for further research to examine the effect of a preferred stock issuance on bond returns, which has not been examined before. Moreover, Kallberg, Liu and Villupuram (2013) used data from before the financial crisis of 2008, while I used data from after the financial crisis. For future research, a comparable analysis can be performed on the difference between responses before and after the financial crisis by equity and debt investors on the response of a preferred stock announcement. This could be either examined using worldwide data or solely with US firms. Also, the comparative analysis could account for different industries or worldwide regions and the accounting treatment of the hybrid security.
25
BibliographyBajaj, M., Mazumdar, S., & Sarin, A. (2002). The Costs of Issuing Preferred Stock. Journal
of Financial Research, 25(4), 577-592.
Baker, M., & Wurgler, J. (2002). Market Timing and Capital Structure. Journal of
Finance,57(1), 1-32.
Bessembinder, H., Kahle, K., Maxwell, W., & Xu, D. (2009). Measuring Abnormal Bond Performance. The Review of Financial Studies, 22(10), 4219-4258.
Blau, B.M., & Fuller, P. (2008). Flexibility and dividends. Journal of Corporate Finance, 14(2), 133-152.
Breusch, T., & Pagan, A.R. (1979). A simple test for heteroscedasticity and random
coefficient variation. Econometrica : Journal of the Econometric Society, an Internat.
Society for the Advancement of Economic Theory in Its Relation to Statistics and Mathematics, 47(5), 1287-1294.
Brzenk, P., & Soe, A.M. (2015). Digging Deeper into U.S. Preferred Market. Retrieved from:
http://us.spindices.com/documents/research/research-digging-deeper-into-the-us- preferred-market.pdf
Cook, R., & Weisberg, S. (1983). Diagnostics for Heteroscedasticity in Regression.
Biometrika, 70(1), 1-10.
Crabbe, L. (1996). Estimating the Credit-Risk Yield Premium for Preferred Stock. Financial
Analysts Journal, 52(5), 45-56.
Dittmar, A., & Thakor, A. (2007). Why Do Firms Issue Equity? Journal of Finance,62(1), 1-54.
Fama, E.F., & French, R. (1993). Common risk factors in the returns on stocks and bonds.
Journal of Financial Economics, 33(1), 3-56.
Fooladi, I., & Roberts, G. (1986). On Preferred Stock. Journal of Financial Research, 9(4), 319-324.
Houston, A., & Houston, C.O. (1990). Financing with preferred stock. Financial
Management, 19(3), 42-54.
Huang, R., & Tompkins, J.G. (2010). Corporate Governance and Investor Reactions to Seasoned Equity Offerings. Managerial Finance. 36(7), 603-628.
Irvine, P., & Rosenfeld, J. (2000). Raising Capital Using Monthly Income Preferred Stock: Market Reaction and Implications for Capital Structure Theory. Financial
Management, 29(2), 5-20.
Jarque, C.M., & Bera, A.K.(1980). Efficient tests for normality, homoscedasticity and serial independence of regression residuals. Economics Letters, 6(3), 255-259.
26
Kallberg, J., Liu, C.H, & Villupuram, S. (2013). Preferred stock some insights into capitalstructure. The Journal of Corporate Finance : Contracting, Governance and
Organization, 77-86.
Kim, D.H., & Stock, D. (2012). Impact of the TARP financing choice on existing preferred stock. Journal of Corporate Finance, 18(5), 1121-1142.
Kraus, A., & Litzenberger, R.H. (1973). A State-Preference Model of Optimal Financial Leverage. The Journal of Finance. 28(4), 911-922.
Linn, S.C., & Pinegar, M.J. (1988). The effect of issuing preferred stock on common and preferred stockholder wealth. Journal of Financial Economics, 22(1), 155-184. MacKinlay, A. (1997). Event studies in economics and finance. Journal of Economic
Literature, 35(1), 13-39.
Masulis, R. (1980). The effects of capital structure change on security prices: A study of exchange offers. Journal of Financial Economics, 8(2), 139-178.
Masulis, R.W., & Korwar, A.N. (1986). Seasoned equity offerings: An empirical investigation. Journal of Financial Economics, 15, 91.
Miller, M. H., & Modigliani, F. (1958). The Cost of Capital, Corporation Finance, and The Theory of Investment. The American Economic Review. 49(4), 655-669.
Miller, M.H., & Rock, K. (1985). Dividend Policy Under Asymmetric Information. The
Journal of Finance. 40(4), 1031-1051.
Miles, J., & Shevlin, M. (2001). Applying regression & correlation : A guide for students and
researchers. London.
Myers, S. C, & Majluf, N. C. (1984). Corporate Financing and Investment Decisions when Firms have Information that Investors do not have. Journal of Financial Economics.
13(2), 187-221.
Patell, J. (1976). Corporate forecasts of earnings per share and stock price behavior empirical tests. Journal of Accounting Research, 14(2), 246-276.
SEC Company filings database. Information retrieved from
http://www.sec.gov.com/edgar/searchedgar/companysearch.html
Smith, C. (1986). Raising capital: Theory and evidence. Midland Corporate Finance Journal, 4(1), pp.6-22.
27
Appendix1. Patell Z-test for abnormal returns
The results of the Patell Z-test (Patell, 1976)are tabulated below. The results show that the day leading up to the announcement and the date of the announcement only show significance, with an outlier of day 2 with estimation window 250 trading days. *,**,*** show the level of significance, which is respectively 10%,5%,1%.
Table 4: Results Patell Z-test
2. Jarque-Bera test for normality
The Jarque-Bera (Jarque & Bera, 1980) test is performed on the residuals derived from the 4 regressions. The hypothesis of the Jarque-Bera test is the following: 𝐻0: Normality 𝐻1: Non-normal. The p-values calculated with the Jarque-Bera test will be presented below. From the p-values results found using the Jarque-Bera test, it is concludable that the error terms are normally distributed.
28
3. Inspection multicollinearity
To determine the amount of multicollinearity between the variables. The average VIF of the variables is 2.02. Miles and Shevlin (2001) argue that a VIF of 1 indicates no multicollinearity, a VIF value between 1 and 5 indicates moderate multicollinearity and a VIF above 5 equals strong multicollinearity. They argue that multicollinearity only strongly biases the results when the VIF is above 5, which is not the case for our variables.
Table 6: Results VIF
4. Breusch-Pagan / Cook-Weisberg test for heteroskedasticity
The hypothesis for this test is the following: 𝐻0: Constant variance, 𝐻1: Non-constant Variance. Failing to reject the null hypothesis indicates that the residuals are homoscedastic and robust standard errors are not required. The p-values of the 4 four regression I ran are tabulated below. From the p-values we can conclude that error terms are homoscedastic and robust standards errors are not necessary.
