Explaining stock price underperformance of public Western European firms following seasoned e

(1)

Explaining stock price underperformance of public Western European firms

following seasoned equity offerings:

Examining the real growth options theory

Eva Louise Elisabeth Mynott (10737111)

Program: BSc. Economics & Business

Track: Finance and Organization

Specialization: Finance

Thesis supervisor: prof. dr. J.E. Ligterink

Thesis coordinator: prof. dr. P.J.P.M. Versijp

Faculty of Economics and Business,

University of Amsterdam

January 31, 2017

Abstract:

This paper examines the relationship between growth option exercise, SEO underperformance, and reductions in systematic risk. Explanations of SEO underperformance can be divided into two general market-‐related and risk-‐related categories (Koussis and Makrominas, 2015). This paper investigates the real growth options theory with a sample of Western European firms issuing equity between 2007 and 2014. This theory states that growth options are replaced by assets when firms issue equity, which decreases the issuing firm’s risk (Carlson et al., 2010). Betas, as a measure of systematic risk, are estimated using the Fama-‐French 3-‐factor model regression. Contrary to expectations, the results indicate an increase in average firm systematic risk, post-‐issue. In addition, for each SEO in the sample, relative offer size is regressed on the pre-‐ and post-‐issue risk

difference. On average, larger relative offer sizes are related to larger changes in risk. This result is significant at the five percent level, which indicates that equity issues indeed affect firm systematic risk. Finally, a fixed effects panel regression using quarterly observations was performed to estimate the effect of growth option changes, approximated by market-‐to-‐book ratio, on changes in firm risk. The results were not indicative of a significant effect of growth options on firm systematic risk changes after controlling for leverage, revenues, and the 2007-‐2008 financial crisis. The overall results might open a debate on the possibility of post-‐issue

systematic risk being largely dependent on the offer characteristics of the issuing firms in the sample, such as

(2)

Statement of Originality

This document is written by Student Eva Mynott, 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)

Declines in a firm’s stock price following seasoned equity offers (SEOs) have been documented for several decades (Koussis and Makrominas, 2015). However, no single, unambiguous explanation for this stock price underperformance has been found (Spiess and Affleck-‐Graves, 1995). Therefore, Loughran and Ritter (1995) documented the SEO underperformance phenomenon as ‘the new issues puzzle’. Although this was followed by extensive academic research into the explanation of this puzzle, there are still competing views on the cause of the post-‐issue stock price underperformance. Asymmetric information seems to explain only a part of the post-‐issue stock price underperformance on the short and long term. It is therefore useful to examine how firm characteristics change after equity issues, as this might give an additional insight in the cause of the empirically observed underperformance (DeAngelo et al., 2010).

Currently, market-‐related and risk-‐related explanations have been put forward (Koussis and

Makrominas, 2015). This paper focuses on the empirical implications of the rational, risk-‐based real growth options theory developed by Carlson et al. (2006; 2010). This model predicts a decrease in firm systematic risk, as less risky assets-‐in-‐place replace growth options when equity is issued. The basic assumption of this model is that firm value consists of cash flows generated by assets, and the future expected cash flows from growth options. The cash flows from unexercised growth options are highly uncertain, and are therefore contain more risk. When equity is issued, growth options are converted into assets, and the cash flows of the firm become less risky. Hence, overall firm

systematic risk decreases. Because of the reduction in risk, a lower return on the stock is required, which might explain the negative returns on issuing firms’ stocks (Carlson et al., 2006, 2010). The central question in this paper is: does the exercise of real growth options -‐ via a reduction of systematic risk – explain the negative stock returns of developed Western European firms after seasoned equity offerings?

This paper is unique in the sense that recent European SEO data are used, whereas all cited works in this paper have only analyzed SEOs from securities based in the United States, and listed on AMEX, NYSE, or NASDAQ. It is interesting to see whether U.S. results can be generalized to other markets, for example the European stock market, with many different individual stock exchanges. It is conceivable that the results might indicate a stronger relationship with the growth options theory due to the lower barrier to equity issuance confronting European firms when financing investments -‐ the (direct) costs of equity issuance in the European Union are almost 40% lower than in the United States (Krakstad and Molnár, 2014). Hence, the results of this paper could indicate whether the theoretical real growth option predictions are robust when using different geographical regions. Furthermore, in this paper a very recently proposed theory by Carlson et al. (2010) is investigated. At present, few scientific works have incorporated all three aspects: growth options, systematic risk,

(5)

and SEOs -‐ aside from Carlson et al. (2010). For example, Koussis and Makrominas (2015) perform a similar study in which they regress systematic risk against firm fundamental characteristics, which include R&D expense as a proxy for growth opportunities. However, their sample does not consist specifically of firms that have issued seasoned equity offerings. Hence, it does not answer the question of how systematic risk, through growth option exercise, changes due to equity issues in particular. Moreover, in this paper market-‐to-‐book ratio is used as a proxy for growth options. Market-‐to-‐book ratio can capture immediate changes in firm value related to the presence of growth options, whereas R&D expense is usually only stated once a year. Furthermore, different control variables than Koussis and Markominas (2015) are used based on the literature. Finally, Carlson et al. (2010) regress the change in pre-‐ and post-‐issue beta against change in investment, including some control variables. This is a cross-‐sectional regression that observes one before and after difference for each individual equity issue. However, in this paper, changes in firm betas are regressed against changes in growth option exercise during eight time periods using a panel regression.

Knowledge of the factors driving this post-‐stock price underperformance is valuable for

several reasons. Firstly, from an academic perspective this insight challenges some fundamental concepts such as the efficient market hypothesis and the rationality of expectations. Secondly, an understanding of SEO dynamics is beneficial within the context of corporate practices, as this could influence capital structure decisions, stock price predictions, and methods of funding profitable investment opportunities. Thirdly, investors could be interested in SEO return patterns, for example when they intend to trade in stocks that may announce, or have announced, equity issuances (Spiess & Affleck-‐Graves, 1995).

The question examined in this paper is addressed by estimating three different regression

equations. Firstly, the Fama-‐French three-‐factor model is used to calculate pre and post-‐issue betas for each individual issuing firm. These are then compared to assess whether there is a reduction in risk, i.e. whether the average pre-‐issue beta is higher than the average post-‐issue beta. Secondly, a simple regression is carried out with the difference in pre and post-‐issue betas as dependent variable against the relative equity offer size as explanatory variable. A panel regression, in conclusion, is carried out to test whether changes in growth opportunities (proxied by market-‐to-‐book ratio changes) explain the differences in quarterly betas. The control variables include leverage, asset turnover, a dummy for the financial crisis and several interaction variables.

The results are not indicative of a reduction in risk. On the contrary, on average systematic risk increases after equity issues. This difference in systematic risk is, however, and in line with expectations, significantly explained by the relative offer size. Nevertheless, the results from the panel regression do not exhibit a significant relationship between changes in growth options, which is measured by changes in market-‐to-‐book ratio, and changes in firm systematic risk of the firms in

(6)

this sample. These findings might be due to the use of offer proceeds for purposes other than firm expansion. It is also possible that market-‐to-‐book ratio is not a sufficient proxy for the existence of growth options. The findings could indicate that the use of offer proceeds largely determines whether the firm systematic risk decreases or increases.

The layout of this paper is as follows: Section II contains a brief review of the divergent explanations of SEO underperformance, followed by the real growth options theory and an outline of the accompanying empirical studies. The Section concludes with the formulation of a hypothesis relating to the literature and the question to be addressed in this paper. Section III discusses the data selection process and the sample characteristics. Section IV introduces the three different regression equations used to estimate the relationship between systematic risk, growth option exercise and seasoned equity offers. Section V presents the results on the analysis of systematic risk and issuing firm characteristics. The last section, Section VI, completes the paper with the conclusions and a discussion of the findings.

II. Literature review

Seasoned equity offerings and subsequent stock underperformance is widely covered in academic literature. This review begins with a presentation of the different theories proposed as an

explanation of the seasoned equity underperformance phenomenon. This is followed by a more detailed discussion of the real growth options theory and the associated empirical studies. The section concludes with the formation of three hypotheses based on the current literature.

Loughran and Ritter (1995) showed that stock price underperformance is not exclusive to initial public offerings, but also occurs with seasoned equity offerings. This finding is consistent for both the short run (up to one year), and the longer run (three to five years). Similarly, Spiess and Affleck-‐ Graves (1995) find evidence of seasoned equity offering (SEO) underperformance from their calculations of holding-‐period abnormal returns in which they match issuing firms with non-‐issuing firms with comparable firm characteristics, such as firm size, book-‐to-‐market ratio, and industry. Brav et al. (2000), in line with Spiess and Affleck-‐Graves (1995), conclude that underperformance is more pronounced for issuing firms of smaller size, and those with higher market-‐to-‐book ratio. Finally, Jegadeesh (2000) shows that this underperformance of stocks issuing equity is robust. He, using different benchmarks, including equal-‐ and value-‐weighted indexes, matching on the basis of firm characteristics, and factor-‐model benchmarks, finds that compounded returns are significantly lower for issuing firms. However, there is no agreement on one explanation for this underperformance.

(7)

II.A. Market-‐ and Risk-‐Related Explanations

A first explanation proposed for post-‐issue stock price underperformance is the market-‐related explanation of price pressure. According to market supply and demand dynamics, the issue of additional shares results in an upward shift in supply. This decreases market liquidity (Aggarwal and Zhao, 2008). Assuming that the demand curve is downward sloping, the share price would adjust downward to correct for the rise in supply. However, this price decrease is expected to occur immediately after the offer announcement. Thus, whether price pressure and market liquidity serve as an explanation for actual post-‐issue negative returns is not unambiguous (Corwin, 2003).

A second potential explanation introduced by Myers and Maljuf (1984) is the pecking order

theory explanation, a more of a behavioral approach to stock price underperformance that relies on information asymmetries. This proposes that managers will prefer debt financing to equity financing. Debt financing is a positive signal, as it shows that a company’s financial position is sufficient to handle a larger amount of debt. However, equity financing may signal that the company cannot issue more debt, and therefore has to resort to issuing equity. The financing decision is new information to the public, so managers possess superior information about the company. Upon SEO announcement, the public incorporates the new, unfavorable information, and the stock price declines. The ongoing stock price underperformance, a year at minimum, is explained by market imperfections: the market is assumed not to adjust immediately.

However, following this reasoning, it is not clear why there is a substantial decline in stock prices upon equity issuance, instead of only upon equity announcement (Loughran and Ritter, 1995). Loughran and Ritter (1995) and Spies and Affleck-‐Graves (1995) both opt for the market timing theory as an explanation for post-‐issue negative returns. This theory suggests that managers are knowledgeable about firm value, and therefore know when the company is overvalued. Since equity is sold at the market price, this represents a gain on every share sold. Managers therefore, postpone equity issues until it is apparent that the stock price is overvalued, at some point in the future. The public then realizes that the share price was too high, and the share price declines. However,

DeAngelo et al. (2010) perform a logit regression to examine what motivates firms to issue seasoned equity: they find that 81.1% of the issuing firms would have run out of cash without the offer. Hence, one of the main motivations to issue is the need for cash. If this is urgent, it is in general not possible to wait until the market conditions are optimal. They conclude that market timing can explain a part of the decision to issue, but not all. Thus, it is necessary to look at firm-‐specific characteristics as well.

In addition to market-‐related explanations, other non-‐behavioral approaches have been

developed that suggest that SEO underperformance is related to the risk characteristics of the issuing firms. Firstly, Brav et al. (2009) adopt the yield on institutional private loans as a measure of risk. They find a significant reduction in the loan yield around the time of the equity issue, consistent with

(8)

the notion that firm systematic risk declines after equity offers. They argue that the stock price underperformance after equity issues might not be due to mispricing, but could instead have a rational risk-‐based explanation (Brav et al., 2009).

Similarly, changes in the systematic risk characteristics of an issuing firm can also be examined by estimating the firm’s beta. Healy and Palepu (1990) have done so and find that

systematic risk increases after issuing equity, which is in contrast to the findings of Brav et al. (2009). Healy and Palepu (1990) argue that a firm's capital structure is determined by its managers'

expectations of the level and riskiness of its future earnings. They state that managers are more likely to sell new equity when they expect the cash flows generated by existing assets to be lower or riskier. Hence, an equity issue will indicate that managers believe that earnings will become more volatile or will decline more than previously anticipated. The market then responds by adjusting the stock price downward. The increased risk is reflected in a higher asset beta. They estimate that the mean asset beta increases by 19 percent in the first year after the offer.

II.B. Growth Options and SEO Underperformance

Another concept that can explain SEO underperformance is the growth options concept. Lee (1997) concludes that growth firms -‐ firms that possess more growth options prior to the equity issue -‐ experience a more significant negative post-‐issue stock performance, which is measured in terms of EBIT, operating income before depreciation, revenue, and total assets. Underperformance is robust for each of the performance measures used. Cooper and Priestley (2011), building on the work carried out by Lee (1997), relate growth options to the firm’s systematic risk. They argue that a firm’s systematic risk is the average of the systematic risks of its assets in place. When a growth firm invests uncertainty is resolved and the share risk premium decreases, which explains the lower stock prices. With this decline in risk, the growth options theory predicts a decline in beta upon equity issuance, in contrast to the findings of Healy and Palepu (1990). Koussis and Makrominas (2015) also find

evidence of a relationship between growth options and firms’ systematic risk. They find that firms with higher R&D expenses, as a proxy for growth options, experience larger decreases in beta after exercising their growth options.

Carlson et al. (2006) develop a theoretical model that incorporates all three aspects: growth

options, SEOs, and systematic risk changes. The main assumptions of this real growth options framework are that equity offers are related to firm expansion, and that managers have superior information about the firm’s growth opportunities. Furthermore, similar to Cooper and Priestley (2011), they argue that firm value consists of assets-‐in-‐place and risky growth options. Hence, firm systematic risk equals the riskiness of the firm’s combination of assets. They also assume that changes in asset risk are caused only by changes in endogenous factors, such as a firm’s cash flows

(9)

changes due to investment. Finally, when firms issue equity they exercise their growth options which then become assets-‐in-‐place. Even though an all-‐equity firm is modeled, they argue that substituting assets for a growth option always reduces firm risk, irrespective of whether the firm is leveraged.

In a later study, Carlson et al. (2010) focus specifically on risk dynamics around seasoned

equity offerings. They also address a limitation of their previous model. An immediate decline in the stock price and beta of a firm would be expected when the option is exercised. However, an

examination of the long-‐term underperformance and gradual post-‐issue beta decline of issuing firms indicates that this is not realistic (Loughran and Ritter, 1995). Carlson et al. (2010) account for this by including ‘commitment-‐to-‐invest’: firms first need to make a lumpy immediate investment financed by the SEO, and then invest at a fixed rate over a fixed period of time. Risk then continues to fall with investment. Empirically, they show that beta does actually increase before the equity issue, and then gradually decreases. Comparing this result to non-‐issuing matched firms, they find that this pattern is unique to issuing firms.

The studies most closely related to this paper are those of Carlson et al. (2010) and Koussis and Makrominas (2015). This paper, in line with Koussis and Makrominas (2015), regresses firms’ beta against firm fundamental variables, including a proxy for growth option exercise. However, their sample does not focus on equity issuing firms. This paper also extends the study carried out by Carlson et al. (2010), who perform a cross-‐sectional regression of changes in beta pre-‐ and post-‐issue against other variables which include change in investment: my panel regression includes

observations of firm beta changes over more time-‐periods. I also use a different method to estimate betas. Finally, the results from this paper could indicate whether the theoretical real growth option

predictions are robust when using other estimation methods and other geographical regions.

II. C. Hypotheses

From the literature review, three hypotheses can be developed for the research question in this paper: does the exercise of real growth options – via a reduction of systematic risk – explain the negative stock returns of developed Western European firms after seasoned equity offerings? The first expectation, from the real growth options theory perspective, is that systematic risk declines after the offering. A decline in post-‐issue systematic risk is expected because, from a theoretical and empirical perspective, uncertainty is resolved, risky growth options have been turned into less risky assets, and leverage has decreased (Carlson et al. 2006, 2010; Koussis and Makrominas, 2015; Loughran and Ritter, 1995; Cooper and Priestley, 2011). Secondly, assuming that this difference in risk is related to the SEO and is proportional to the offer size, it can be expected that the change in pre and post-‐issue risk is significantly explained by offer size. The expectation is that a potential change in systematic risk post-‐issue is solely due to the equity issue and potential SEO-‐induced

(10)

changes in firm characteristics. Hence it is not due to other firm-‐specific news or events, or general market changes (Aggarwal and Zhao 2008; Cooper and Priestley, 2011). Finally, it is expected that growth option exercise significantly accounts for the post-‐issue difference in systematic risk. In line with the real growth options theory and related empirical findings, growth option exercise -‐ in this paper approximated by market-‐to-‐book ratio changes -‐ is expected to explain the change in risk following the equity issue (Carlson et al., 2006, 2010; Lee, 1997; Koussis and Makrominas, 2015).

III. Data

The data is selected in three steps. The first step is to collect data on seasoned equity offerings. This data includes the date of the offering, the firm’s name, the offer size and its proceeds. The second step is to match daily stock prices to the issuing firms. The third and last step is to collect quarterly firm fundamental data, including data from balance sheets and income statements, such as total assets, stockholders’ equity, and revenue.

III.A. Databases

Firstly, the firms that have issued seasoned equity need to be identified. The Thomson ONE database on new issues is used for this purpose. The sample contains firms of Western European origin that issued seasoned equity between 01/01/2007 and 31/12/2014, and which originate from 16 countries, namely Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, the Netherlands, Norway, Portugal, Spain, Sweden, Switzerland and the United Kingdom. These countries are selected on the basis of the definition of ‘developed Europe’ formulated by Kenneth French1_{.
In} later stages of this examination French’s database is used to estimates the betas, for which reason the countries selected for the size and book-‐to-‐market portfolios in this database need to coincide with the countries used in the sample.

Secondly, stocks prices between 01/01/2006 and 31/12/2015 are collected from the

CompuStat Global Securities Daily database. The countries that issued equity between 2007 and

2014 are matched to these stock price data on the basis of their international security identification number (ISIN), a unique code for every security. This then provides for an analysis of stock returns one year before and one year after the issue. These stock prices are, together with Kenneth French’s data, then used to estimate betas for each individual security.

Thirdly, firm fundamental data from balance sheets and income statements is obtained from

the CompuStat Global Quarterly Fundamentals database on Wharton Research Data Services

1

_{http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data_library.html}

(11)

(WRDS). The VLOOKUP function in Excel and other tools are used to match the issuing firms’ fundamentals with their respective equity offering and stock price data, using ISINs.

III.B. Data Selection Procedure

Using the Thomson ONE database, a total of 7,367 follow-‐on offerings are identified. However, this is not the final sample, but rather a first rough selection. The data selection procedure consists of six consecutive steps. The criteria for inclusion in the sample are:

(1) The firm does not have the ‘financial’ or ‘utilities’ industry classification . As these industries are heavily regulated, the characteristics and effects of equity offers of firms in these industries may differ substantially from offers from firms in other industries (Cooper and Priestley, 2011; Loughran and Ritter, 1995).

(2) The equity issue is an offering of new, common shares. Consequently, rights issues and warrants are excluded.

(3) The equity issue is a primary offering of shares: secondary offerings are excluded. The rationale behind this is that secondary offerings include the sale of shares by one or more major stockholders in the firm, selling all or a large portion of their shares, whereby the total amount of company shares does not change (DeAngelo et al., 2010; Carlson et al., 2010). Equity issues that are a combination of primary and secondary offerings are included.

(4) The firm that has issued equity is included in the CompuStat Global Fundamentals database and sufficient information is available: the firm must be listed in the CompuStat database and the specific quarters under analysis must be available in the database.

(5) The firm that has issued equity exists in the CompuStat Global Daily Stock Prices database and sufficient information is available: daily stock prices are available at least one year before, and one year after the specific equity issue. Furthermore, there needs to be trade volume, so that the price varies from day to day.

(6) The firm has not issued equity one year before and one year after the equity issuance under observation. For example, an equity offer observation on January 31, 2008, will be included in the sample only when: 1. The preceding equity issue was on January 31, 2006 or earlier, and 2. The subsequent equity issue on January 31, 2009 or later.

Many of the 7,367 follow-‐on offerings identified during the 2007-‐2014 period were rights issues and not pure common stock offerings. Once criteria 1 to 3 had been assessed, only 683 seasoned equity offers remained. Of these, 419 equity issues met criterion 4. The 419 issues were checked manually to determine whether they met criteria 5 and 6. Some firms, mostly from the United Kingdom, had

(12)

issued equity every year, sometimes as often as three or four times a year. A total of 94 individual seasoned equity issuances ultimately met all criteria. Appendix A contains a detailed list of the sample. A summary of the sample characteristics is presented below.

III.C. Data Summary and Characteristics

This section presents summary statistics of the Thomson One data on equity offerings and of the Compustat data on quarterly firm fundamentals. Table I summarizes the equity offering

characteristics of SEOs in Western Europe between 2007 and 2014.

Table 1

Offering Characteristics for Western EU Seasoned Equity Offers, 2007-‐2014

Offer proceeds is the dollar amount raised by the sale of new common shares. Relative offer size is calculated as the number of shares offered in the equity issuance divided by the number of shares outstanding before the offer. Market value is the share price multiplied by the number of shares outstanding. Shares outstanding is the number of shares that firms had outstanding during the period of -‐252 days to +252 days before and after the equity issue. Data source: Thomson Reuters SDC New Issues Database.

Variable Median Mean St. Dev. Min. Max.

Offer proceeds (mil.) 8.23 30.92 118.60 .0093 1116.67

Relative offer size (%) 10.00 19.50 21.36 0.09 119.09

Market value (mil.) 144.10 507.68 1028.30 2.09 5851.03

Shares outstanding (mil.) 53.01 157.88 309.35 1.35 1800.04

In the sample, the mean offer proceeds is 30.92 million dollars and the median is 8.23 million dollars. There is a substantial variation in size, as the standard deviation is high compared to the mean, and the minimum offer proceeds is only 9,300 dollars, whereas the maximum equals 1.12 billion.

Similarly, although the mean offer size is 19.50 percent, the standard deviation of 21.36

percent indicates that the spread is quite large. In the sample, the most common offer size is about ten percent. The smallest offer size is 0.09%. However, the largest equity offer of 119.09%, results in the firm more than doubling its shares outstanding. This also explains the substantial minimum and maximum differences in offer proceeds. Nevertheless, more than doubling the number of shares outstanding is not common, as 62% of the relative offer sizes in the sample are of between 5% and 20%. This indicates that the mean is raised by a few, more extreme observations.

The median of market value is 144 million, which is not so large as compared to the largest firm with a market value of 5.9 billion. The mean reflects this by being almost four times higher. The same holds for shares outstanding, of which the mean is about three times higher than the median.

(13)

Overall, the sample reflects a combination of issuing firms of a wide range of sizes as measured by shares outstanding and by market capitalization.

Furthermore, although two firms in the sample had issued equity twice, the interval between

the issuances was sufficient to analyze the effects of the two issues. Finally, the stated purposes for the equity offer proceeds in the sample consist of general corporate purposes, working capital, increasing cash balances, reducing indebtedness, conducting investment, and acquisition finance. SEOs intended to finance acquisitions may have different effects on the stock price than SEOs with other stated purposes. For example, Porrini (2015) finds that firms’ beta decreases after acquisitions, where beta is estimated by running a regression of the stock's return on the return on the market index, and using a time period of 254 business days. Hence, equity issues for which the proceeds will be used for acquisition purposes, may reinforce the expected decrease in post-‐issue beta. In the total sample of 94 Western European SEOs, 15 equity issues stated ‘acquisition finance’ or ‘future acquisitions’ as their issue purpose. To account for the possible deviant effect of acquisition-‐ motivated SEOs, a dummy variable for M&A activity is added to the second regression. In this regression, offer size is regressed on the difference in pre-‐ and post-‐issue risk.

In addition to the equity offer characteristics, an overview is presented on the number of SEOs conducted in each individual year in the sample period. Graph 1 shows the results. Of the total of 94 equity offerings, most companies in the sample issued equity in 2014. Furthermore, there seems to be no increase in equity offers during the global financial crisis of 2007-‐2008 compared to the other years in the sample. Hence, it does not seem to be the case that SEO activity peaked during the financial crisis and dropped afterwards, in 2009 and 2010. In contrast, SEO activity in the first few years after the financial crisis is higher than the average of 11.75. The reason for this observation is

(14)

hard to determine, but one explanation could be that after the global financial crisis of 2007-‐2008 European firms were more in need of increasing cash balances, or reducing indebtedness (Dissanaike et al., 2014).

Finally, of interest is whether average firm characteristics of the pre-‐issue period significantly

differ from the post-‐issue period. Table 2 summarizes the mean differences in firm fundamentals before and after the equity issue. The variables included are the variables used for the panel regression in section V, except for market capitalization.

Table 2

Differences in means for the variables used in the panel regression

Market capitalization is calculated as the price per share multiplied by the number of shares outstanding. Leverage is defined here as net debt, which is calculated as long-‐term debt outstanding minus cash and short-‐ term investments. Asset turnover is calculated as revenue divided by total assets. Finally, the market-‐to-‐book ratio is defined as market capitalization divided by total shareholders’ equity. The means for these variables are divided into two subsample groups: before the equity issue and after the equity issue. The p-‐values are

represented by stars, where p<0.10 = *, p<0.05= **, and p<0.01= ***. Data source: Compustat Global Quarterly Fundamentals.

Mean Pre-‐issue Post-‐issue Difference (t-‐statistic)

Market Capitalization (mil.) 486 555 -‐15.23***

Leverage (mil.) 136 170 -‐14.06***

Asset Turnover 0.137 0.148 -‐9.45***

Market-‐to-‐book ratio 2.296 3.765 -‐8.06***

Differences between the means are calculated using a paired t-‐test to account for the dependence

between the pre and post-‐issue subgroups (Wiedermann and Von Eye, 2013).

All pre and post-‐issue differences are significant at the one percent level. Market

capitalization is higher in the year after the equity issue than before the issue. As market

capitalization is calculated as shares outstanding multiplied by the stock price, this higher market value may be due to a higher stock price or a larger number of shares outstanding, or both. As the firms have issued equity, the average number of their shares outstanding has increased, whilst the predicted decline in stock price decline is gradual and may well have yet to offset the increase in shares, thereby increasing the overall market capitalization.

Moreover, the mean of leverage increases after the equity issue. The reason for this is not known, although it could, for example, indicate that firms decide to issue equity to balance out the effects of issuing new debt, or to use the offer proceeds as collateral for a loan. The higher amount of

(15)

leverage would, theoretically, predict an increase in risk and firm beta (Koussis and Makrominas, 2015).

Furthermore, average asset-‐turnover increases, which means that revenue as a fraction of

total assets increases. This indicates an efficiency gain, as more sales can be generated from the same total amount of assets. This higher efficiency level is expected to lower the probability of firm failure, thereby decreasing systematic risk (Logue and Merville, 1972).

Finally, market-‐to-‐book ratio increases. This is contrary to expectations. As market-‐to-‐book

ratio will be used as a proxy for growth options, it is hypothesized that firms which have exercised their growth options have lower market-‐to-‐book ratios (Carlson et al., 2006). Growth option exercise does not appear to be present in the sample. This is investigated in detail in the third regression. Leverage and market-‐to-‐book ratio would appear to indicate that overall systematic risk increases post-‐issue, which would be in line with expectations based on theory. Whether this is actually the case is examined in the first regression.

IV. Methodology

The main question of this paper is addressed by dividing the research question into several sub-‐ questions. The questions are as follows: (1) does firm systematic risk decrease after the SEO? (2) Does equity offer size explain the post-‐issue change in firm systematic risk? (3) Do growth option changes account for changes in firm systematic risk? Various types of regressions are required to test the hypotheses for each sub-‐question. The first regression tests whether beta after the equity issue is significantly lower than beta before the issue. The second, a simple regression, tests whether the difference in beta for each equity issue is explained by the offer size. The third, a panel regression, tests whether changes in growth opportunities explain the differences in quarterly betas. Growth option changes are approximated by market-‐to-‐book ratio changes, and control variables on possible risk-‐affecting firm and market characteristics are included, such as leverage, revenue, the global financial crisis, and several interaction variables.

IV.A. Regression 1 – Difference between Pre and Post-‐Issue Beta

The first sub-‐question to be answered is whether a firm’s beta decreases after a seasoned equity offering. This requires an estimation of beta before and after the equity offering. These pre and post betas can then be compared to test for a significant difference.

First, each period is assigned a number. The first seasoned equity offering in the sample, for example SEO1, is assigned number ‘1’ for the 252 trading days before the equity offering and number ‘2’ for the 252 days after the equity offering. The second equity issue in the sample, for example SEO2, is then assigned ‘3’ and ‘4’. This continues until the last seasoned equity offering, SEO94. This

(16)

then results in 188 groups, where all odd group numbers correspond to the 252 days before the offering.

A beta is then estimated for each of the 188 groups, i.e. 𝛽!, 𝛽!, …, 𝛽!"". This requires the programming of local macros and loops in Stata. These Stata commands are enclosed in Appendix C.1. Each beta with an odd number corresponds to a pre-‐issue beta and each beta with an even number corresponds to a post-‐issue beta. In total, there are 94 pre-‐issue betas and 94 post-‐issue betas. The time-‐series regression equation used to estimate each beta is the 3 factor model of Fama

and French (1993). This model is as follows: 𝑟!− 𝑟𝑓! = 𝑎!+ 𝛽 𝑟𝑚!− 𝑟𝑓! + 𝑎!𝑆𝑀𝐵!+ 𝑎!𝐻𝑀𝐿!+

𝜀!. The ordinary least squares (OLS) regression method is used. The dependent variable is the excess return of an individual stock over the risk-‐free return, at a specific date 𝑡. The constant in the model is denoted by 𝑎!. The beta is the coefficient of interest, as this is the measure for firm systematic risk in this model. The variable 𝑆𝑀𝐵! is the return on a portfolio constructed to mimic risk factors related to size at time 𝑡. 𝐻𝑀𝐿! is the return on a portfolio for book-‐to-‐market equity at time 𝑡 and, 𝜀! is the error term.

This asset-‐pricing model tends to explain more about asset-‐return variations than, for

example, the capital asset pricing model that only examines the slope of a stock’s return on a market return. In the three-‐factor model, size and book-‐to-‐market equity can serve as proxies for common risk factors in stock returns (Fama and French, 1993). The three-‐factor model has very recently been extended to a five-‐factor model that incorporates profitability and investment portfolios (Fama and French, 2016). Few checks of the robustness of the model and its superiority over the three-‐factor model have been made to date, and for this reason the three-‐factor model has been adopted for this paper.

Finally, the hypothesis is tested. As stated in Section II, it is expected that systematic risk declines after the offering. The 94 pre and the 94 post-‐issue betas are averaged, and it is then estimated whether the average pre-‐issue beta is significantly larger than the average post-‐issue beta. In statistical terms, the hypothesis can be formulated as:

𝐻!: 𝛽!"#$%"− 𝛽!"#$% = 0 vs. 𝐻!: 𝛽!"#$%"− 𝛽!"#$%> 0. This comparison is made using a paired t-‐ test to account for the interdependence of the two groups. If the normality assumption is violated, the paired t-‐test has been shown to be robust with respect to type 1 errors (Wiedermann and Von Eye, 2013).

(17)

IV.B. Regression 2 – The Difference in Beta in Relation to the Equity Offer Size

The second sub-‐question to be answered is whether the estimated difference in beta is explained by the seasoned equity offer size.

Firstly, the regression estimates from the first regression are used to create a new variable.

In the first regression, 94 pre-‐issue betas and 94 post-‐issue betas were estimated with the Fama-‐ French three factor model. The new variable is the difference in beta for each equity offering, which is denoted by 𝛽!!. The regression equation is the simple regression model: 𝛽!! = 𝑎!+ 𝑎!𝑂𝐹𝑆!+ 𝜀!. This regression model is estimated using ordinary least squares. In this model, 𝑎! represents the constant, 𝑂𝐹𝑆! the relative size of the seasoned equity offering for each individual offering,

calculated as the number of common shares offered to the market divided by the number of shares

outstanding before the offering, and 𝜀! is the error term. Furthermore, an extended regression

equation is performed in which a dummy variable on merger and acquisition activity is included. As mentioned in section III.C., SEOs intended for (future) acquisitions may either enhance or decrease the predicted post-‐issue change firm systematic risk, as compared to equity issues for non-‐

acquisition purposes (Porrini, 2015). The dummy variable on M&A activity equals one if the issuing company states ‘acquisition finance’ or ‘future acquisitions’ as the equity offer purpose, and zero for other purposes. The extended equation is as follows: 𝛽!!= 𝑎!+ 𝑎!𝑂𝐹𝑆!+ 𝑎!𝐷_!&!!+ 𝜀!.

The hypothesis with respect to offer size and the difference in beta can then be tested. According to the real growth options theory, it is expected that the seasoned equity issue induces a risk reduction and, as a result, a decrease in beta (Carlson et al., 2010). The difference between the pre and post-‐issue betas tends to increase with relative offer size (Corwin, 2003; Aggarwal and Zhao, 2008). For this reason, the test needs to examine whether a significant fraction of the difference in the beta is related to the relative offer size of the equity issue. The second statistical hypothesis is then whether the coefficient on the variable offer size is significant, which can be written as: 𝐻!: 𝑎!= 0 vs. 𝐻!: 𝑎!≠ 0 .

Explaining stock price underperformance of public Western European firms following seasoned e