The performance of seasoned equity offerings

(1)

1

The Performance of Seasoned Equity Offerings

Bachelor Thesis in Finance and Organization University of Amsterdam

Faculty of Economics and Business

Author: Sjors van der Sluis Student number: 10784489 Date: January 13, 2017

(2)

2

Abstract

Previous literature proved that SEOs underperformed compared to their peers during the period 1975-1989. This thesis investigates the performance of SEOs during the period 2009-2015. Furthermore this thesis investigates if there is a difference in the performance of high- and low-growth SEOs. This research finds evidence supporting the underperformance of SEOs in a more recent time period, but does not find evidence supporting a difference between high- and low-growth SEOs.

Keywords: Seasoned equity offerings, underperformance

Statement of Originality

This document is written by Sjors van der Sluis who declares to take full responsibility for the contents of this document.

I declare that the text and the work presented in this document is original and that no sources other than those mentioned in the text and its references have been used in creating it. The Faculty of Economics and Business is responsible solely for the supervision of completion of the work, not for the contents.

(3)

3

1. Introduction

The financial market has developed to such a point that investors are constantly seeking for opportunities to make profits by investing in profitable stocks. This has led to research investigating stock markets. The short-term under-pricing of IPOs has been widely documented. However, Ritter (1991) found empirical evidence claiming that IPOs underperform in the long-run and Affleck-Graves & Spiess (1994) concluded that the

underperformance of IPOs is not an exception. Therefore SEOs underperform as well. There are several reasons explaining this phenomena. It could be explained by overvaluation of the firm or risk mismeasurement. Furthermore, Ross (1977) developed a theory called the incentive-signalling approach. Based on this approach, firms can send a credible signal by committing to large future payments by acquiring debt. In addition, the pecking order theory states that firms prefer debt over equity when externally generating funds (Myers 1984). Since theory states that firms prefer debt over equity and issuing debt sends a credible signal to the market, issuing equity will result in a negative signal towards investors. In addition,

McConnell & Servaes (1995) concluded that the corporate value for high-growth firms is negatively correlated with leverage, and the corporate value for low-growth firms is positively correlated with leverage. The research of Affleck-Graves & Spiess and McConnell & Servaes has led to the following research question: Is there a difference between the abnormal returns of high- and low-growth SEOs during the period 2009-2015?

This thesis will contribute to the existing literature by investigating the performance of SEOs in a recent time period. Furthermore, this thesis investigates if there is a difference between the performance of high- and low-growth SEOs. Firstly, the cumulative abnormal returns over a 3-year aftermarket period are computed. Secondly, a regression on the

cumulative abnormal returns is run to test if there is a difference between the performance of high- and low-growth SEOs.

Furthermore, the short-term under-pricing of IPOs is widely documented, and this anomaly has been studied for SEOs as well (Ghosh et al, 2000). Therefore it is more

interesting to investigate the overpricing anomaly. In addition, since the value of high-growth firms react different towards increasing debt than low-growth firms, it is interesting to

investigate whether high- and low-growth SEOs perform differently.

This thesis provides evidence supporting the underperformance of SEOs in a more recent time period. Using a sample that consists of 424 SEOs, this research has found an average cumulative abnormal return of -51.40% in the period 2009-2015. However, this thesis

(5)

5 did not find evidence supporting a difference in the abnormal returns of high- and low-growth SEOs.

At last, the thesis is organized as follows: The next section will discuss previous literature on the performance of SEOs. The third section will discuss the methodology and the sample that is used. The fourth section discusses the empirical results and the last section consists of a conclusion with recommendations for further research.

2. Literature review

This section of the thesis will elaborate on the existing literature on SEO performance. Firstly, the underperformance of IPOs and SEOs will be defined. Secondly, possible reasons for this underperformance will be discussed. Lastly, the expected outcome of the research will be discussed.

2.1 The underperformance of IPOs and SEOs

Two anomalies in the pricing of IPOs have been widely documented. The short-run under-pricing phenomenon and the “hot issue” market phenomenon. Ritter (1991) proves that there is a third anomaly: in the long run, IPOs appear to be overpriced. The main difference is that short-term research claims that IPOs are under-priced, while long-term research claims that IPOs are overpriced. Ritter (1991) and Affleck-Graves & Spiess (1994) have found empirical evidence that IPOs and SEOs tend to be overpriced and therefore underperform. Both papers from Ritter (1991) and Affleck-Graves & Spiess (1994) have conducted an event study to investigate the performance of IPOs and SEOs respectively. In both articles the

underperformance of IPOs and SEOs is defined as long-term underperformance compared to their matched peers. Ritter (1991) matched the IPOs to stocks based on comparable size and industry. The average holding period return of IPOs in three years after going public is

34.47%, while the average holding period return of their matched counterpart is 61.86%. Even though investing in IPOs does not result in negative returns, one could better invest in their matched counterpart. Therefore IPOs underperform in the long-run.

Based on the results of Ritter, Affleck-Graves & Spiess (1994) have found empirical evidence claiming that SEOs underperform as well. The median five-year holding period return following a SEO is 10%, while the median five-year holding period return for non-issuing firms with a comparable size and similar industry is 42.3%. Furthermore Affleck-Graves & Spiess found the same results when controlling for size and book-to-market ratio, trading system, offer size and firm age as well.

(6)

6

2.2 Reasons for the underperformance

There are several reasons that could explain the underperformance of IPOs and SEOs. Firstly, managers could be able to use firm-specific information to issue equity when their firm is overvalued. Ritter (1991) found evidence claiming that firms choose to go public when investors are optimistic and Affleck-Graves & Spiess (1994) concluded that their findings are consistent with managers issuing equity when their firm is overvalued.

Secondly, the underperformance of SEOs and IPOs could reflect mismeasurement of risk. Ritter’s (1991) evidence does not support the risk mismeasurement, instead it supports the claim that firms go public when the market is overoptimistic. In addition, Affleck-Graves & Spiess concluded that the difference between the mean annualized risk premiums of SEO firms and matched firms was too large. Therefore their findings did not support the

mismeasurement of risk either.

At last, Ritter (1991) investigated the fact that the underperformance could be due to bad luck. However Ritter’s results did not support this claim.

2.3 Hypothesis

Firstly, due to previous literature negative cumulative abnormal returns are expected. This expectation is supported by the pecking order theory. The pecking order theory states that firms prefer internally generated funds over external funds and when external funds are needed, firms prefer debt over equity (Myers, 1984). Since firms prefer debt over equity, issuing equity has unfavourable influence on the market’s perception of the firms value which results in negative abnormal returns.

Secondly, low-growth firms tend to have stable cash flows. These stable cash flows provide opportunity for a high debt-to-equity ratio, because there is more certainty that debt repayments can be fulfilled (Berk & DeMarzo, 2014). Therefore debt is less expensive for low-growth firms. Since debt is less expensive, issuing equity could signal to the market that the firm does not have stable cash flows anymore. This negative signal will result in a drop in the stock price. However, past and current earnings are also used as a proxy for future

earnings and therefore influence the market’s perception of the firm’s value. Since high-growth firms have growing earnings, the signalling effect of equity might have a less large impact on the firm’s stock price. Hence, a difference between the returns of high- and low-growth SEOs is expected.

(7)

7 Thirdly, McConnell & Servaes (1995) concluded that the value of high-growth firms correlates differently with debt than the value of low-growth firms. Since the value of high-growth firms react different towards increasing debt than low-high-growth firms, this might also be for increasing equity. Therefore a difference in the performance of high- and low-growth SEOs is expected.

3. Methodology

This section of the thesis will describe how this research is conducted. First of all, what kind of research is performed, and what regressions and computations belong to this research. Additionally, the data and the criteria that were used to select the sample will be discussed.

3.1 Event study

Since seasoned equity offerings are not simultaneously issued at a specific time, an event study will be performed. An event study is performed to be able to compare the returns of SEOs in different periods of time. The rest of this section will describe what estimation- and event window is used. Moreover, the market model that is used will be described and the regression with corresponding computations will be discussed.

3.1.1 Estimation- and Event window

The existing literature has studied the aftermarket performance of IPOs and SEOs for a three- and five-year period. Affleck-Graves & Spiess (1994) studied the aftermarket performance of SEOs for a three- and five-year period and their paper will be used as a guideline for this thesis. Therefore, the three-year aftermarket performance will be studied in this thesis. To define an event window, the amount of days that is contained in a year has to be defined. In this thesis trading days will be used and therefore a month contains 21 trading days. This results in 252 trading days a year. Using this information, the event window can be defined as (0, +756) where t = 0 is the issue date obtained from the Thomson One database and t = 756 is 3 years after the issue date.

An estimation window is defined to predict the normal returns in the event period. These normal returns are the expected returns to be observed if there had been no event. Affleck-Graves & Spiess (1994) have obtained normal returns by using the stock returns of matching non-issuing firms based on size and comparable industry, but also book-to-market

(8)

8 ratio, trading system, offer size and firm age. Since previous literature used a matching

technique to compute abnormal returns, their estimation window is identical to the event window. Considering the matching technique to be too time consuming, in this thesis normal returns are predicted using a market model. Therefore an estimation window that is different from the event window is used. The estimation window is defined as (-777, -22) which represents a period of 3 years. Since estimation windows are typically chosen prior to the event period, in this thesis an estimation window prior to the event is chosen as well (Peterson, 1989). The estimation window is chosen to end 1 trading month before the issue date, to be sure that the predicted betas are not influenced by announcements prior to the issue. Table 3.1.1 provides a timeline representing the estimation- and event window.

Table 3.1.1

3.1.2 Market model

To predict the normal returns for the event period a market model is used. This thesis uses a market model that is a basic form of the CAP-model (Fama & French 2004):

𝐸(𝑅𝑖) = 𝛼 + 𝛽𝑖∗ 𝐸(𝑅𝑚) 𝑖 = 1, . . . , 𝑁.

E(Ri) is the expected return of firm i that will be predicted by this model. Rm is the return of

the market portfolio. In this thesis the return of the equally weighted S&P 500 index will be used as a market proxy (Rm), because the performance of SEOs in the United States is

examined. The alphas and betas will be estimated over the estimation window. Using the estimated alphas and betas, the normal returns for the event window will be predicted.

(9)

9

3.1.3 Regression

To examine if there is a difference in the returns of low- and high-growth stocks after a seasoned equity offering, a regression on the cumulative abnormal returns has to be run. First of all, the cumulative abnormal returns are computed as follows:

𝐶𝐴𝑅_𝑖 = ∑(𝑅_{𝑠𝑒𝑜, 𝑖𝑡}

𝑛

𝑡=1

− 𝑅_{𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑, 𝑖𝑡})

Where Rseo, it is the return of stock i at event time t and Rpredicted, it is the predicted (normal)

return of stock i at event time t. These abnormal returns are summed up for n = 756 trading days resulting in the cumulative abnormal returns per SEO.

In addition, the main independent variable is the book-to-market ratio, which is computed as: 𝐵𝐸 𝑀𝐸𝑖 = (1 𝑛𝑡 ) ∑(𝐵𝐸 𝑀𝐸𝑖𝑡 ) 𝑛 𝑡=1

Where the book-to-market ratio for every firm i is the average book-to-market ratio over a period of n = 7 years where t = 1 is the fiscal year 2009. All accounting data computations are based on the computations of Fama & French (1992). BE is the Common Equity minus Deferred Taxes (Balance Sheet) obtained at the end of the fiscal year for every firm. ME is the market capitalization which is computed by multiplying the amount of shares outstanding by the closing price at the end of every fiscal year. The book-to-market ratio is the main independent variable, because Fama & French (1993) found that low BE/ME is characteristic for high-growth stocks, and low-growth stocks tend to have a high BE/ME ratio. Therefore the book-to-market ratio is used to distinguish high- and low-growth stocks.

Besides the main independent variable, there are 3 control variables added to this regression. These control variables are chosen, because prior literature has proven that size, leverage and E/P help to explain the cross-section of average returns (Fama & French, 1992).

The first control variable is Size: 𝑆𝑖𝑧𝑒_𝑖 = (1 𝑛_𝑡) ∑(𝑆ℎ𝑎𝑟𝑒𝑠 𝑜𝑢𝑡𝑠𝑡𝑎𝑛𝑑𝑖𝑛𝑔𝑖𝑡 𝑛 𝑡=1 ∗ 𝐶𝑙𝑜𝑠𝑖𝑛𝑔 𝑝𝑟𝑖𝑐𝑒_𝑖𝑡)

(10)

10 Size is the average market capitalization for every firm i, which is computed by calculating the average of the ME over 7 years. The second control variable is earnings-to-price ratio which is computed as:

𝐸 𝑃𝑖 = (1 𝑛_𝑡) ∑( 𝐸 𝑃𝑖𝑡 ) 𝑛 𝑡=1

Where E is earnings per share and P is the closing price at the end of the fiscal year. Earnings are computed as:

𝐸𝑎𝑟𝑛𝑖𝑛𝑔𝑠 = 𝐼𝑛𝑐𝑜𝑚𝑒 𝐵𝑒𝑓𝑜𝑟𝑒 𝐸𝑥𝑡𝑟𝑎𝑜𝑟𝑑𝑖𝑛𝑎𝑟𝑦 𝐼𝑡𝑒𝑚𝑠 + 𝐼𝑛𝑐𝑜𝑚𝑒 𝑇𝑎𝑥𝑒𝑠 𝐷𝑒𝑓𝑒𝑟𝑟𝑒𝑑 − 𝐷𝑖𝑣𝑖𝑑𝑒𝑛𝑑𝑠 𝑃𝑟𝑒𝑓𝑒𝑟𝑟𝑒𝑑

The price ratio belonging to every firm i is also the average of the earnings-to-price ratio over 7 years. At last, the leverage ratio is used as a control variable, which is computed as: 𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖 = (1 𝑛_𝑡) ∑( 𝐴𝑠𝑠𝑒𝑡𝑠 𝐵𝐸 𝑖𝑡 ) 𝑛 𝑡=1

Where leverage for every firm i is the average leverage ratio over a period of 7 years. Assets are defined as total assets at the end of the fiscal year.

A combination of the variables results in the following regression formula:

𝐶𝐴𝑅_𝑖 = 𝛽₀+ 𝛽₁∗ 𝑙𝑜𝑔 (𝐵𝐸 𝑀𝐸𝑖 ) + 𝛽₂∗ 𝑙𝑜𝑔(𝑆𝑖𝑧𝑒_𝑖) + 𝛽3∗ 𝐸 𝑃𝑖 + 𝛽₄∗ 𝑙𝑜𝑔(𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖) (1)

Without using a log function the coefficient of the size variable became very small. Since Fama & French (1992) used logs for book-to-market, size and leverage, and taking into account the very small coefficient of size, logs for the independent variables except E/P are used.

First of all, to conclude that there is an effect of book-to-market ratio on the CAR’s the 𝛽1 has to be significant. The corresponding statistical hypotheses are:

𝐻0 : 𝛽1= 0

(11)

11 Secondly, to be able to answer the research question a second regression has to be run:

𝐶𝐴𝑅_𝑖 = 𝛽₀+ 𝛽₁∗ 𝐷𝑢𝑚𝑚𝑦(𝐻𝑖𝑔ℎ 𝐵𝑇𝑀) + 𝛽₂∗ 𝐷𝑢𝑚𝑚𝑦 (𝐿𝑜𝑤 𝐵𝑇𝑀) + 𝛽₃∗ 𝑙𝑜𝑔(𝑆𝑖𝑧𝑒_𝑖) + 𝛽₄∗𝐸

𝑃𝑖

+ 𝛽₅∗ 𝑙𝑜𝑔(𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒_𝑖) (2)

In this second regression two dummy variables are added. The book-to-market ratios are split into 3 groups: high, low and medium. If the average book-to-market ratio belongs to the top 33%, then the high dummy equals one and if the ratio belongs to the bottom 33% than the low dummy is one. In order to avoid multicollinearity only 2 dummies are used.

In order to test if there is a difference in the return of high- and low-growth stocks this statistical test has to be done:

𝐻0 : 𝛽1 = 𝛽2 = 0

𝐻₁ : 𝛽₁ ≠ 0 𝑜𝑟 𝛽₂ ≠ 0

If 𝛽1 , 𝛽2 or both are significantly different from zero there can be concluded that there is a

difference between the returns of high- and low-growth stocks.

3.2 Data

The sample consists of seasoned equity offerings during the period 2009-2012, obtained from the Thomson One database. This period is chosen because it is a more recent period that has not been studied yet. The stock price and accounting data is obtained from the

CRSP/Compustat database. In order to be able to compute a 3-year cumulative abnormal return, the last date of issuance could be 31/12/2012, because the CRSP database provides data until the end of 2015. To determine which equity issues are eligible for this research, the issues have to meet the following criteria: (1) the equity offering should be labelled a follow-on offering in the Thomsfollow-on One database, (2) the offering should be a commfollow-on stock

offering, (3) the gross proceeds over all markets must exceed $1 million, (4) and the company’s nation should be the United States.

This results in a sample of 3640 SEOs. In order to be able to compute unbiased

cumulative abnormal returns, an issue has to be independent. An independent issue is defined as an issue that has no consecutive issue within the 3-year aftermarket period. Removing non-independent issues results in a sample of 1270 non-independent issues. To be eligible for

(12)

12 computing the abnormal returns, the stock price data of 3 years after and 3 years plus a month before the issue date has to be available. This results in 621 issues that had matching CRSP data available. Furthermore, there has to be accounting data available from 2009 until the end of 2015 for the issues to be eligible for the regression. Merging the accounting data with the issues results in 424 issues with removing 2 definite outliers (Tickers: EGLE, CVGI) and removing issues that had matching negative leverage and book-to-market ratios. The descriptive statistics belonging to this sample are given in table 3.2.1.

Table 3.2.1 Without any winsorizing

(1) (2) (3) (4) (5)

VARIABLES N mean sd min max

CAR 424 -0.514 2.116 -27.99 10.25

E/P 424 -0.0154 0.182 -2.169 0.263

Log(Size) 424 7.166 1.870 2.611 12.23

Log(BTM) 424 -0.562 0.679 -3.126 1.265

(13)

13

4. Results

In this section the empirical results of the regressions will be presented and discussed. Firstly, the regression without dummies will be discussed, and after that the regression with dummies will be discussed.

4.1 Significance test for the cumulative abnormal returns

The first regression that has been run in this thesis is the regression without dummy variables. This regression has been run to investigate if the book-to-market ratio has a significant effect on the cumulative abnormal returns. First of all, to be sure that the CAR’s are significant a simple regression on CAR is run.

Table 4.1

After winsorizing CAR 5% percentile

(1)

VARIABLES Significance test CAR

Constant -0.514***

(0.0475)

Observations 424

R-squared 0.000

Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1

The coefficient of the constant is significantly different from zero, which indicates that the CAR’s are significant (Table 4.1). The mean CAR is negative which supports the theory that SEOs underperform, because a negative abnormal return indicates that the stock has

underperformed the predicted normal return for the stock. While Affleck-Graves & Spiess (1994) found an average cumulative abnormal return of -30.99% in the period 1975-1989, this research has found an average cumulative abnormal return of -51.40% in the period 2009-2015. This evidence suggests that SEOs underperform even more in a recent time period.

4.2 Regression without dummy variables

In this section the regression results and variations in the regression will be discussed. First of all, table 4.1.1 provides the coefficients for the independent variables by using regression formula 1. Table 4.1.1 presents three different regressions. The coefficient of the constant

(14)

14 represents the underperformance of the SEOs. As issues are removed the coefficient of the constant increases up to -1.750 which suggest that SEOs underperform their predicted values by a 175%. The coefficient of our main independent variable is 0.00902 without removing any issues. This means that when the book-to-market ratio increases by 1%, the value of the CAR increases by 0.00902 * 0.01, which is 0.0000902. This increase is so small that it is almost neglectable. In addition, the coefficients of the book-to-market ratio are never

significant. Moreover, after removing issues before 2011 an increase in the size of the firm by 1% would result in an increase in the value of the CAR of 0.0897 * 0.01, which is 0.000897. This effect is small but significant. In addition, the coefficient of E/P can be interpreted as follows. Increasing the E/P ratio by 100% increases the cumulative abnormal return by 91.5%. This positive coefficient seems logical because current earnings are used as a proxy for future earnings. When earnings increase, the E/P ratio increases, but also the expectation of the stocks value increases. Therefore an increase in the E/P ratio has a positive effect on the cumulative abnormal returns. Furthermore, the coefficient of leverage is 0.206 when no issues have been removed. Increasing the leverage ratio by 1% results in an increase in the value of CAR by 0.206*0.01, which is 0.00206. This effect is positive and the reasoning for this will be explained later on. After winsorizing the cumulative abnormal returns, the significance level of leverage and E/P increased to 1%. In none of the 3 regressions the book-to-market coefficient is significant. Therefore there can be concluded that there is no significant effect of the book-to-market ratio. Fama & French (1992) concluded that size and book-to-market ratio seem to absorb the roles of leverage and E/P in explaining the average stock returns during the period 1963-1990. In the period of 2009-2015 the roles tend to be the other way around. Based on table 4.1.2 there can be concluded that leverage absorbs the explanatory power of the book-to-market ratio. Adding size and E/P increases the t-value of the book-to-market coefficient slightly. However, when leverage is added to the regression the t-value of BE/ME drops from 1.221 to 0.114. This supports the claim that leverage absorbs the explanatory power of the book-to-market ratio, which is in contrast with the findings of Fama & French (1992).

This contrast could be explained by the incentive-signalling approach of Ross (1977). The capital structure of a firm may reveal information regarding the firm’s future prospects. Ross (1997) modified the Modigliani-Miller theorem on the irrelevancy of financial structure by taking into account for features of the real world. The irrelevancy of financial structure would suggest that a firm should be financed by debt only. Ross designed an incentive-signalling model for the determination of the financial structure of a firm that dropped the

(15)

15 assumptions of the Modigliani-Miller theorem. Ross (1997) concluded that the values of firms will rise with leverage, since it sends a credible sign to the market about the firm’s positive future prospects. This can be elaborated by giving an example. When a firm issues debt, it commits to future debt repayments. If a firm has unpromising future prospects, a manager will know that the firm will not be able to fulfil these debt repayments. Therefore issuing debt sends a credible sign to the market that the firm will be able to fulfil these debt payments and this increases the market’s perception of the firm’s value. Since the coefficient of leverage is positive, increasing the leverage ratio has a positive effect on the CAR’s. This supports the theory that acquiring debt sends a credible signal to the investors that the firm is able to commit to large future payments. Therefore this signals that the firm has good future prospects.

For the issues that took place in 2009 and 2010, the estimation window overlaps with the fall of the Lehman Brothers in 2008. This might cause a bias in the results of the first regression. Removing the issues before 2011 results in 152 CAR’s which average is -72.0%. This suggests that after removing the issues that might be influenced by the fall of the Lehman Brothers, SEOs still underperformed. In fact, SEOs performed worse. Furthermore, there is still no significant effect for the book-to-market ratio. After removing the issues before 2011, the effect of size becomes significant for 5% and the explanatory power of the model increased from 0.062 to 0.221 (Table 4.1.1).

Another data manipulation that has been done is to decrease the estimation- and event window to 2 years. This results in 439 observations with a mean CAR of -53.5%. The mean CAR is only 2.1% lower than when using an estimation- and event window of 3 years. After winsorizing the 5% percentile for CAR, the only significant coefficient is the leverage coefficient with a significance of 5%. Why the coefficient of E/P is not significant after decreasing the estimation- and event window is unclear.

In addition, when increasing the estimation window to 5 years and remaining the event window 3 years, the results present significant coefficients for leverage (1%), E/P (1%) and size (5%). However, the book-to-market ratio coefficient remains non-significant.

At last, another manipulation of the data has been done. Instead of removing negative book-to-market and leverage ratios, these negative ratios are winsorized into positive ratios. The results for the estimation- and event window of 3 years are roughly the same. The significant coefficients are roughly the same level in comparison to the initial regression. However, without removing any issues the coefficients of leverage and E/P are significant at a level of 5% instead of 1% compared to the initial regression. In addition, the E/P coefficient is

(16)

16 not significant after removing issues before 2011, while this coefficient is significant (5%) after running the initial regression. Finally, the book-to-market coefficients remain non-significant.

Concluding, for every variation in the regression, winsorizing the 5% percentile of CAR increases the explanatory power of the model. Furthermore, leverage seems to absorb the explanatory power for every variation. Summing up, there can be concluded that the book-to-market ratio has no significant effect on the cumulative abnormal returns.

4.3 Regression with dummy variables

The previous section provides evidence claiming that the book-to-market ratio has no

significant effect on the cumulative abnormal returns. Even though there is a great chance that the coefficients of the dummies will not be significant, the regression with the dummy

variables is still run. Table 4.3.1 provides the results from the regression using formula 2. When using an estimation- and event window of 3 years the results are almost identical. As expected after running the initial regression, the coefficients for high and low book-to-market ratio are not significant.

When decreasing the estimation- and event window to 2 years the coefficients of the high and low dummies are still not significant. However, the significance level and the value of the coefficient of leverage decrease. Furthermore, after decreasing the estimation- and event window the significance level of E/P diminishes, which happened using regression formula 1 as well.

Concluding, after running a regression using formula 2 there is still no evidence claiming that there is a difference between the performance of high- and low-growth SEOs.

5. Summary and conclusion

Previous literature found evidence claiming that SEOs underperform during the period 1975-1989. This thesis investigates if SEOs underperform in a more recent time period and if there is a difference in the abnormal returns of high- and low-growth SEOs. The reasoning behind the difference in the abnormal returns of high- and low-growth SEOs is mainly based on the incentive-signalling theory of Ross (1977). This theory suggests that issuing debt sends a credible signal to the market about the future prospects. Therefore issuing equity would have a negative effect on the market’s perception of the value of the firm. Since an investor’s perception of the firm’s value is also affected by historical and current earnings, the impact of

(17)

17 issuing equity is expected to be less large for high-growth firms.

This thesis found evidence supporting the underperformance of SEOs in a more recent time period. Using a sample that consists of 424 SEOs that had matching stock and

accounting data, this thesis found an average cumulative abnormal return of -51.40% in the period 2009-2015. Affleck-Graves & Spiess (1994) found an average cumulative abnormal return of -30.99% in the period 1975-1989. Thus, SEOs perform worse in a more recent time period. However, no evidence for the difference between the abnormal returns of high- and low-growth stocks is found.

Although this research has not found evidence supporting the research question, some recommendations for further investigation are named. Firstly, a broader time period can be used to increase sample size. Secondly, it would be interesting to use a different way of categorizing high- and low-growth stocks in order to possibly finding a difference. Thirdly, using the matching technique used by Ritter and Affleck-Graves & Spiess could improve this research.

(18)

18

6. References

Affleck-Graves, J. J., & Spiess, K. D. (1995). Underperformance in long-run stock returns following seasoned equity offerings. Journal of Financial Economics(38), 243-267. Berk, J., & DeMarzo, P. (2014). Corporate Finance. Harlow: Person Education Limited. Fama, E. F., & French, K. R. (1992). The Cross-Section of Expected Stock Returns. The

Journal of Finance(47), 427-465.

Fama, E. F., & French, K. R. (1993). Common risk factors in the returns of stocks and bonds.

Journal of Financial Economics(33), 3-56.

Fama, E. F., & French, K. R. (1995). Size and Book-to-Market Factors in Earnings and Returns. The Journal of Finance(50), 131-155.

Fama, E. F., & French, K. R. (2004). The Capital Asset Pricing Model: Theory and Evidence.

Journal of Economic Perspectives(18), 25-46.

Ghosh, C., Nag, R., & Sirmans, C. F. (2000). The Pricing of Seasoned Equity Offerings: Evidence from REITs. Real Estate Economics(28), 363-384.

Masulis, R. W., & Korwar, A. N. (1986). Seasoned equity offerings: An empirical investigation. Journal of Financial Economics(15), 91-118.

McConnell, J. J., & Servaes, H. (1995). Equity ownership and the two faces of debt. Journal

of Financial Economics(39), 131-157.

Modigliani, F., & Miller, M. H. (1963). Corporate income taxes and the cost of capital: A correction. American Economic Review(53), 433-443.

Myers, S. C. (1984). The capital structure puzzle. Journal of Finance(39), 575-592. Peterson, P. P. (1989). Event Studies: A Review of Issues and Methodology. Quaterly

Journal of Business and Economics(28), 36-66.

Ritter, J. R. (1991). The long-run performance of initial public offerings. Journal of

(19)

19 Ross, S. (1977). The determination of financial structure: The incentive-signalling approach.

(20)

20

7. Appendix

Table 4.1.1

Regression using the formula:

𝐶𝐴𝑅𝑖= 𝛽0+ 𝛽1∗ 𝑙𝑜𝑔 ( 𝐵𝐸 𝑀𝐸𝑖) + 𝛽2∗ log(𝑆𝑖𝑧𝑒𝑖) + 𝛽3∗ 𝐸 𝑃𝑖+ 𝛽4∗ 𝑙𝑜𝑔(𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖 ) Estimation- and event window of 3 years

CAR 5% percentile winsorized

(1) (2) (3)

VARIABL ES

Without removing issues

Issues before 2010 have been removed

Issues before 2011 have been removed Log(BTM) 0.00902 0.0458 0.0653 (0.0793) (0.102) (0.0897) Log(Size) -0.0346 -0.00579 0.0897** (0.0287) (0.0402) (0.0413) E/P 0.915*** 0.899 1.378** (0.343) (0.607) (0.669) Log(Lever age) 0.206*** 0.309*** 0.321*** (0.0605) (0.0858) (0.0998) Constant -0.539** -0.955*** -1.750*** (0.227) (0.335) (0.366) Observatio ns 424 277 152 R-squared 0.062 0.081 0.221

(21)

21

Table 4.1.2

4 different regressions by adding control variables Estimation- and event window is 3 years

Without removing any issues

(1) (2) (3) (4)

VARIABLES CAR CAR CAR CAR

Log(BTM) 0.0754 0.0839 0.0919 0.00902 (1.022) (1.095) (1.221) (0.114) Log(Size) 0.0137 -0.0169 -0.0346 (0.473) (-0.589) (-1.204) E/P 1.001*** 0.915*** (3.176) (2.669) Log(Leverage) 0.206*** (3.412) Constant -0.471*** -0.564** -0.325 -0.539** (-6.956) (-2.571) (-1.538) (-2.380) Observations 424 424 424 424 R-squared 0.003 0.003 0.035 0.062

Robust t-statistics in parentheses *** p<0.01, ** p<0.05, * p<0.1

(22)

22 Table 4.3.1 𝐶𝐴𝑅𝑖= 𝛽0+ 𝛽1∗ 𝐷𝑢𝑚𝑚𝑦(𝐻𝑖𝑔ℎ 𝐵𝑇𝑀) + 𝛽2∗ 𝐷𝑢𝑚𝑚𝑦 (𝐿𝑜𝑤 𝐵𝑇𝑀) + 𝛽3∗ 𝑙𝑜𝑔(𝑆𝑖𝑧𝑒𝑖) + 𝛽4∗ 𝐸 𝑃𝑖 + 𝛽5 ∗ 𝑙𝑜𝑔(𝐿𝑒𝑣𝑒𝑟𝑎𝑔𝑒𝑖)

Estimation- & event window is 3 years CAR 5% percentile winsorized

(1) (2) (3)

VARIABL ES

Without removing issues

Issues before 2010 have been removed

Issues before 2011 have been removed High 0.0395 0.166 0.187 (0.122) (0.190) (0.231) Low 0.0267 0.136 0.0508 (0.110) (0.148) (0.173) Log(Size) -0.0347 -0.00965 0.0898** (0.0288) (0.0404) (0.0416) E/P 0.920*** 0.909 1.402** (0.344) (0.605) (0.676) Log(Lever age) 0.205*** 0.302*** 0.306*** (0.0623) (0.0903) (0.102) Constant -0.562** -1.046*** -1.852*** (0.242) (0.342) (0.378) Observatio ns 424 277 152 R-squared 0.062 0.084 0.223

The performance of seasoned equity offerings