Corporate Governance Strength and Cost of SEOs. Ali Sheikhbahaei 1. Balasingham Balachandran. Amalia Di Iorio. Huu Duong

Transcription

1 Corporate Governance Strength and Cost of SEOs Ali Sheikhbahaei 1 Department of Banking and Finance, Monash Business School Monash University, Australia Balasingham Balachandran Department of Economics and Finance, La Trobe Business School La Trobe University, Australia Amalia Di Iorio Department of Economics and Finance, La Trobe Business School La Trobe University, Australia Huu Duong Department of Banking and Finance, Monash Business School Monash University, Australia 1 Corresponding author - ali.sheikhbahaei@monash.edu

2 Abstract Although it is not directly observable, governance strength can give confidence to investors in the event of a seasoned equity offering. Prior studies find that corporate governance can be evaluated through proxies at firm level; however, these are subjected to endogeneity. Another strand of the literature employs a single takeover enactment to exogenously determine strength of governance. Empirical evidence shows that investors react positively if governance is not insulted by particular anti-takeover legal enactments. We employ the Hostile Takeover Index (HTI) as a middle ground proxy between firm-level and external governance mechanisms to gauge the strength of corporate governance. Overall, we find that investors react positively to issuing firms that experience a higher probability of a hypothetical hostile takeover. We also find that when the market is surprised by an announcement with a low price run-up, governance strength plays a significant role in reducing uncertainty among investors. 2

3 1. Introduction Theoretically, the strength of a firm s governance, as measured by growth opportunities, can reduce investors uncertainties about the misuse of equity offerings to raise capital (Jung et al., 1996). Empirical evidence shows that firms with stronger governance, as measured by the lack of particular external laws in takeover markets, experience a less negative market reaction around SEO announcement days (Kim & Purnanandam, 2013). However, this is not directly observable, and there is no consensus on which is the best proxy to measure the power of corporate governance. Another debate is over whether governance power should be determined endogenously, at firm level, or based on exogenous variables. The takeover market, with its legal changes in the past decades (Cain et al., 2017) and its firm-specific defences (Gompers et al., 2003; Bebchuk et al., 2008), is a vivid venue within which the strength of a firm s governance can be determined at both external and internal levels. Prior theoretical and empirical studies document that more external insults and interference from authorities in a takeover environment weaken governance. However, there is little evidence about how the market reacts to an issuing firm while its governance is facing external pressures from various takeover laws. In this study, we aim to use a measure newly-developed by Cain et al. (2017), the Hostile Takeover Index (HTI), as a proxy for strength of governance in order to identify investors reactions to the announcement of seasoned equity offerings (SEOs). Cain et al. (2017) construct the HTI using exogenous anti-takeover legal changes, and find that lower levels of hostility arising from higher takeover protection are related to lower firm values. Employing this index, we primarily hypothesize that issuing firms with higher HTIs will experience a less aggressive market reaction, since investors are less worried about the misuse of funds by firms with stronger governance. Since the HTI is constructed from variations in three different layers laws, macroeconomics, and firm-specific factors it can represent a middle ground between endogenous and exogenous governance proxies. Thus, the 3

4 index allows me to run cross-sectional analyses on different firm characteristics. Cain et al. (2017) argue that their index is not subjected to endogeneity as is the G-Index (Gompers et al., 2003), and does not suffer from a lack of power when a single law is imposed, such as Business Combination Statutes (BCS) enactment (Kim & Purnanandam, 2013). However, while specific studies have focused on endogenous governance measures or single legal shocks to governance, none has examined the effect of the full array of corporate legal changes on investors reactions to SEOs in a longitudinal time frame. To our knowledge, this is the first study to investigate the relationship between external shocks, adjusted by firm-specific factors, and governance and investors reactions to SEO announcements. Following Kim and Purnanandam (2010), we employ an exogenous shock to corporate governance as a proxy to evaluate the strength of a firm s internal governance. We examine changes in ex-post-capital expenditure as a channel through which governance can influence investors greatest concern, which is increases in capital expenditure. we also investigate how the market reacts to stronger governance, taking into account market timing as highlighted by DeAngelo et al. (2010) to be a prominent explanation for SEOs. In a co-determination approach, we consider both timing and governance when examining market reactions. We explicitly examine whether governance strengths can predict market reactions in the absence of a timing signal: that is, a so-called surprise. I use a data set of 2,415 SEO announcements that occurred in the US market between 1994 and 2014 to identify investors reactions to offered shares, considering their awareness of governance strength. We also co-determine timing theory as the dominant explanation for negative market reactions, along with the strength of the issuing firm s governance. We find that investors react less negatively when a firm has higher HTI, providing stronger governance. The coefficient of HTI in our base model is statistically significant at 1% and has an economically significant effect on the market reaction of 2.66% for the whole sample period, 4

5 controlling for important determinants of SEO cost in the literature. our findings are consistent with prior studies which suggest that when a firm s manoeuvrability faces less constraint by restrictions in takeover market, there is a positive effect on shareholders value (see for example Hackl & Testani, 1987; Karpoff & Malatesta, 1989). We also find that in the absence of a timing signal, where the market is surprised by offered shares that are not highly overpriced, the strength of a firm s governance plays a significant role in reducing investors concerns about the misuse of proceeds. I continue our analysis by examining the extent to which improvement in the takeover index influences investors reactions during our entire sample period. By measuring changes in the index for the year immediately prior to an SEO announcement, we find that investors react less negatively when there is an improvement in governance strength as measured by HTI. This result is only significant for the period , where the average HTI shows the minimum mean and standard deviations compared to prior decades in Cian et al. (2017) s index construction process. This result suggests that in the past decade, when more state-level protections have been imposed on takeover laws and as a result hostility in the corporate level market has reduced, any tiny change in the index has been significantly associated with investors reactions and their wealth. My supplementary test relates SEO announcement returns to HTI, while controlling for another measure of governance at the firm level. From a theoretical perspective, dedicated institutional investors try to use either their corporate voice or direct intervention (move) to make their desired changes. However, a recent survey by McCahery et al. (2016) shows that long-term institutional investors intervene in corporate governance more intensively. We argue that HTI is related to investors reactions, even when controlling for a major endogenous governance proxy percentage of dedicated institutional ownership in our regression analysis. 5

6 The results show that HTI is still significantly associated with market returns when a major governance proxy at firm level is in place. My results are not subjected to endogeneity, which most prior studies have struggled with. Manne (1965) theoretically shows that the takeover market is an external punitive mechanism for corporate governance. Prior studies have used single laws as external governance mechanisms to measure investors reactions to SEO announcements (Kim & Purnanandam, 2013). In contrast, HTI uses an array of anti-takeover laws over a period of four decades to provide a proxy to measure the hostility of the takeover market, which is positively related to firm value (Cain et al., 2017). My findings have important implications for corporate financing decisions, as they show that when raising external capital through the equity market, a firm s governance affects investors determinations. Although timing theory can mostly explain negative market reactions to SEO announcements, stronger governance as measured by external mechanisms can reduce investors uncertainties about the use of proceeds. Consistent with prior studies, we find evidence that capital expenditure is a channel through which strong governance can affect investors confidence. our robustness analysis also posits that if we exclude informed investors and rely solely on public announcement of SEOs, and control for timing and growth opportunities as major determinants from both firm and market perspectives, HTI is still significantly associated with a less negative reaction. The remainder of this paper is structured as follows: Part 2 presents a literature review and our hypothesis development; Part 3 discusses our data, sample characteristics, and research design; Part 4 presents our empirical findings and reports on robustness and supplementary tests; and Part 5 concludes the paper. 6

7 2. Literature Review Multiple theories have been offered to explain negative market reactions to announcements of SEOs. Brealey, Leland and Pyle (1977) explain them as negative signalling, when managers are too worried about moral hazards to be straightforward in transferring information about their projects to the market. The pecking-order model developed by Myers and Majluf (1984) argues that while managers try to maximize the wealth of existing shareholders, information asymmetry between new investors and managers can make equity issuance too costly. However, Jung et al. (1996) theoretically prove that if firms have a marked growth opportunity, they will bear the costs given that investors are aware of their growth opportunity and will react less negatively. The empirical implications of the pecking-order model are therefore problematic if we relax the assumption of information asymmetry. Consistent with trade-off theory and its variant extensions of tax and leverage cost models, firms should borrow through debt after an extraordinary share price increase rather than conducting equity issuance. Several empirical studies report extremely high price run-ups in the year prior to equity offerings (e.g. Asquith & Mullins, 1986; Korajczyk et al., 1989 Loughran & Ritter, 1995). Since SEOs are attached to high price run-ups, trade-off theory becomes empirically problematic. Myers and Majluf (1984) address the problem of price fall by adverse selection, explaining that firms that are performing well should raise their required capital through the debt market rather than through equity. The extent to which investors react to an SEO announcement therefore reflects their valuation of the issuing firm. Various studies use legal changes and laws imposed on the takeover market as exogenous proxies to determine a firm s governance strength. Empirical studies in this area are mostly supported by Manne s (1965) theoretical work, and use one specific external takeover defence as a proxy for ex-ante changes to governance power (see for example Karpoff & Malatesta, 1989; Bertrand & Mullainathan, 1999, 2003; Schwert, 2000). A recent study by 7

8 Cain et al. (2017) reveals that out of 17 takeover laws, 12 prevalent legal statutes are related to hostile takeover rates, 2 with different significant levels of coefficients. Cain et al. (2017) suggest that the varying impacts of takeover laws are driven by their strength and whether they encourage or discourage takeover activities. Thus, to construct the HTI, they aggregate all laws and incorporate important determinants at firm level across four decades of changes in takeover laws. Cain et al. s (2017) main finding is that a firm s value is higher when governance is more powerful due to fewer insulting external takeover laws. We employ the index to argue that stronger governance arising from a higher index can reduce the magnitude of the negative market reaction to an SEO announcement. This forms Hypothesis 1: H1: Investors react less negatively to SEOs issued by firms with higher HTI. My first empirical hypothesis is a major statistical inference based on Manne s (1965) theoretical framework, and contributes to the existing body of literature by providing empirical evidence for the relationship between widely examined external shocks to governance and shareholders value. My second hypothesis arises from the notion that at the announcement of an SEO, investors are worried about improper use of proceeds caused by weak governance. According to the theoretical argument of Jung et al. (1996), investors react negatively due to concerns about the misuse of proceeds, and this is more pronounced for weakly-governed firms as measured by growth opportunity, given that they are more likely to waste the new capital. Kim and Purnanandam (2013) find that increases in capital expenditure are a channel through which 2 Related Statutes: Williams Act; Fair Price; Control Share Cash-Out; Mandatory Staggered Board; Disgorgement; Anti-Greenmail; Golden Parachute Restriction; Tin Parachute Blessing; Assumption of Labour Contracts; Revlon; Unocal; Blasius. Non-related statutes: Business Combination; Poison Pill; Control Share Acquisition; Expanded Constituency; 1st Generation Statutes. 8

9 external shocks to governance can affect investors concerns. We extend this empirical framework to hypothesize that an increase in capital expenditure funded by SEO proceeds comforts investors if a firm s governance is less insulted by external takeover legal changes. This is Hypothesis 2: H2: Investors react less negatively to SEOs issued by firms with higher HTI and higher changes to capital expenditure. In the absence of other theories that can reveal the primary reason for negative market reactions to SEOs, the market timing hypothesis by Loughran and Ritter (1995) and Baker and Wurgler (2002) can largely explain price falls, showing that managers attempt to sell overpriced shares. This becomes more convincing given Graham and Harvey s (2001) anonymous surveys, which report that overvaluation is an important consideration when issuing equity for two-thirds of the surveyed Chief Financial Officers (CFOs). Supportive empirical evidence by DeAngelo et al. (2010) also posits that the timing determinant is significant. Another explanation of the negative reaction, in Jung et al. (1996), is that investors react negatively when they realize that a firm s weak governance can lead to misuse of the capital raised. Empirical evidence provided by Kim and Purnanandam (2013) supports this notion in the US setting. However, the co-determination of timing and governance has received little attention. Our empirical question is therefore whether, if timing fails to predict an SEO, strong governance can mitigate the magnitude of investors negative reactions. The timing theory of Loughran and Ritter (1997) predicts the likelihood of an SEO when a firm is substantially overvalued, and of a repurchase when it is underpriced. Baker and Wurgler (2002) find that there is a strong relationship between a firm s current capital structure and the timing of its past attempts in the equity market. The empirical work of DeAngelo et al. (2010) shows that timing is a major consideration for firms conducting SEOs due to an urgent 9

10 need for cash, and has both statistical and economical significance and influence. The timing model seems to address the question of both why and when firms choose to issue equity, as well as how investors react to that choice; however, its underlying assumption is that the market fails to incorporate all the information conveyed by the issue. Kim and Purnanandam (2013) find that investors do not necessarily underreact to SEOs because they are worried about the misuse of proceeds, but instead welcome the increase in capital expenditure funded by SEOs if they are in the hands of strong governance. They use the passage of single BSC enactment as an external shock to a firm s governance body, and find that higher post-announcement capital expenditures by firms with stronger governance experience lower negative market reactions. We hypothesize that if the market is surprised by a low stock price run-up, stronger governance can reduce a negative reaction. This is Hypothesis 3: H3: Surprised investors react less negatively to SEOs issued by firms with higher HTI. According to Graham and Harvey (2001), one-third of surveyed CFOs do not consider overpricing to be a major consideration when issuing equity. Intuitively, we suggest that firms with low run-up issuing equity will surprise investors and market participants. If we find evidence supporting the above hypothesis, we will therefore be able to argue that in the event of SEO, strength of governance is an important consideration among surprised investors evaluating the issuing firm in the absence of high stock price run-up. 3. Data and Sample Statistics My sampling approach follows common methods used in the literature on equity and event study. SEO announcements are obtained from the Thompson Financial SDC database and include firm-commitment offerings, excluding pure secondary shares entirely offered by existing shareholders. Stock returns and firms accounting data are extracted from CRSP and 10

11 COMPUSTAT. Institutional ownership and classification data are obtained from Thomson Reuters Institutional (13F) Holdings database through the WRDS database. The takeover index is available at Stephen McKeon s webpage through the University of Oregon. After merging the lead announcement with the above-mentioned databases, we drop observations with missing data. We also drop penny stock and issues following the first issue within one year. Units, warrants and financial firms with SIC coded between 6000 and 6999 are also excluded. My final SEO sample consists of 2,415 announcements spanning the period Following Kim and Purnanandam (2013), we set the SDC s filing date as the first public announcement of a firm s intention to issue equity, fulfilling the requirement for an accurate event day. The 20-year sample enables me to examine samples from both before and after the year 2000, to keep consistency with the subsets of HTI in Cain et al. (2017). The average index falls during the passage of the years as the takeover market becomes more restricted and hostile thanks to more state-level protections. Therefore, dividing our sample into pre-2000 and post subsamples enables me to capture the greatest variation in HTI, which is already adjusted for time series and cross-sectional variations. Panel A of Table 1 shows the frequency distribution of our sample across the sample period. The sample is almost evenly distributed into pre-2000 and post-2000 subsamples (41.78% from the period and 58.22% from the period ). Panel B shows the sample in a firm cluster setting, where 63.32% are single firms that made announcements only once, followed by 23% and 8.2% which are firms that issued a second and third time respectively. The smallest cluster is made up of three firms that issued eight times during the sample period. [Table 1 here] 11

12 Table 2 presents the mean, standard deviation, minimum, and maximum of key variables used in our sample of SEOs during the entire period. Consistent with prior empirical studies (see for example Masulis & Korwar, 1986; Gao & Ritter, 2010), the average cumulative abnormal return for event windows of 3 and 2 days are -2.43% and -2.31% respectively. The takeover index shows a mean of 8.61%, indicating the probability of a hypothetical hostile takeover (HTI) in the year prior to the SEO announcement date. The mean market-to-book ratio of 5.72 is a relatively high figure, but not surprising for issuing firms that show a mean return of % in the previous year (pastret). [Table 2 here] 4. Results In this section, we provide model descriptions and results for our proposed empirical hypotheses. However, before drawing any inferences, we need to ensure that no specific variables at either firm or market level are driving the results. Any strong correlation between the main independent variable of HTI and other controlling variables can lead to a biased estimation of our main dependant variable. Table 3 shows all correlation coefficients between the variables of interest in this study. We find no strong correlations between the variables employed in this study, suggesting that our further estimations of market reactions are not driven by any particular endogenous factor. [Table 3 here] 12

13 First, we start our main analysis by estimating the market reaction CAR (-1, +1), regressed on HTI as our main independent variable. We classify our analysis into 3 different models with samples drawn from the periods , and We follow the same classification as in this base model for the rest of the analysis. Second, we estimate the market reaction, using the interactive effect of changes in capital expenditure and HTI. This allows me to identify a channel through which governance affects market reaction. Third, we draw two different sets of analysis, first set as estimating market reaction crosssectional on market timing and second set as estimation of market reaction cross-sectional on growth opportunity, to find that the predominant explanations of SEOs are investors joint considerations of governance and timing. Fourth, we examine the effect of improvement in HTI on investors reactions. As a supplementary test, we employ a governance measure at firm level percentage of dedicated institutional ownership to estimate market reaction, and finally we check the robustness of our base model using a shorter window of time: (0,+1). 4.1 Base Model, HTI and Market Reaction In this section, we primarily examine our first hypothesis: that strong governance has a positive relationship with investors reactions to SEO. We employ HTI as a measure of governance power in a cross-sectional approach. Consistent with Cain et al. (2017), we only include year fixed effects in all models, as other fixed effects such as firm or industry may cancel out variations in slow-moving HTI. In Table 4 Model 1, we estimate CAR (-1, +1) using the single variable of HTI, which reveals a coefficient (t-stat) of 0.095(4.29). In Table 2 Model 2, we control for the important controlling variables that prior studies find important in explaining SEOs including market to book, log of sale, cash to total asset, leverage ratio, standard errors of residuals in market model, stock price run-up in the last year prior to SEO, indicator of mixed offered stock. The following model describes the estimation: 13

14 CAR(-1,+1)= α + βhti + γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (i) When controlling for other important variables, HTI still shows a coefficient (t-stat) of 0.065(2.66), suggesting that stronger governance due to fewer external takeover laws makes investors more confident. The economic magnitude of the benefits associated with HTI on investors reactions is meaningful. Our estimation in Model 2 shows that the median firms gain 2.66% (0.0648/0.0244) more positive market reaction due to stronger governance achieved through higher levels of hostility in the takeover environment. As described earlier, HTI shows the lowest mean in the past two decades out of all four studied decades (Cain et al., 2017). Therefore, in order to clearly identify the effect of its variations during each decade in our sample, we classify the sample into the two subcategories of pre-2000 and post Models 4 and 6 represent these subcategories respectively. For the pre-2000 sample, we find a statistically significant coefficient for HTI at a significance level of 5%; however, we find only a marginally significant coefficient for the post-2000 sample. A test of difference between the two coefficients reveals a p-value of 0.79 for the equality of the two slopes. Therefore, we can conclude that HTI is an important determinant for investors to gain confidence in a firm s value, and that its importance is more pronounced for issues announced prior to One explanation for this, according to the specifications of the passing laws, would be that these firms have experienced more state-level protections and consequently a lower rate of hostility in the takeover market. 3 This affects the HTI variation rate in the post-2000 sample, in comparison to the pre-2000 sample. [Table 4 here] 3 For example, the following takeover laws became effective after 2000 on the state level: Mandatory Staggered Board (states: IN, IA and OK); Revlon (states: CA, IL, KS, MD, NH and NC). 14

15 4.2 Interactive Effect of Capital Expenditures and HTI Negative market reactions to SEO announcements can be explained by investors concerns about the misuse of SEO proceeds (Jung et al., 1996). Kim and Purnanandam (2103) empirically examine the effects the passage of BSC law as an exogenous governance measure has had on investors concerns about misuse of capital. We extend their study by examining capital expenditure as a channel through which HTI can affect investors reactions. Following DeAngelo et al. (2010), we measure the level of capital expenditure and changes in acquisition expenditures by subtracting the respective disclosed figures in the year of an SEO from those of the prior fiscal year, which is used as the benchmark. The following equation describes the cross-sectional changes in capital expenditures (acquisitions) around the SEO year and associates them with investors reactions: CAR(-1,+1)= α + βhti + ωhti*δcapex (acq)+ ʊδcapex(acq))+ γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (ii) Where Δcapex is changes in dollar capital expenditures and Δacq is changes in the dollar value of disclosed acquisitions performed by each firm. Consistent with our base model, we include year fixed effects with White heteroscedasticity-consistent robust standard errors for all models. I am particularly interested in the interaction term of HTI*Δcapex (acq). If ω is significantly different from zero, then we can show the effect of changes in capital expenditures made by firms with stronger governance. Here, we examine our second hypothesis to see whether strong governance can mitigate investors uncertainties about the misuse of SEO proceeds. Kim and Purnanandam (2013) find evidence that when a firm s governance is 15

16 weakened by the adoption of BCS in its state of incorporation and there is more capital expenditure in the year of SEO, this is interpreted as bad news and the market reacts negatively. In Table 5, Models 1 and 2 show no significant coefficients for our variable of interest, the interactions of HTI and Δcapex and of HTI and Δacq. However, when we isolate the pre and post-2000 subsamples, we find surprising results for the post-2000 group. In Models 5 and 6, both interactive terms show significant and positive coefficients, which is consistent with prior findings by Kim and Purnanandam (2013). It seems intuitive that higher HTI means governance is less affected by external shocks, and that investors favour the increase in capital expenditure and acquisitions accomplished by firms with strong governance. Notwithstanding, the past year return (run-up) is still a significant variable, which shows that the timing factor is a predominant explanation of SEO in all models. [Table 5 here] 4.3 Growth Opportunity, Market Timing, and HTI One important question is whether high-growth firms spend wisely or carelessly. A theoretical model created by Jung et al. (1996) suggests that investors react negatively to SEOs because they are worried about the misuse of proceeds. However, they suggest that high growth firms tend to spend wisely and that investors are aware of that, resulting in a less negative market reaction to their offered shares. They use market-to-book ratio as measure of growth opportunity. We argue that high-growth firms give investors confidence if their high growth is accompanied by stronger governance. As in Part 4.2, the variable of interest in our argument is the interaction between governance and growth opportunity: HTI* mtb. The following 16

17 equation describes the cross-sectional changes in market-to-book and governance strength around the SEO year and associates them with investors reactions: CAR(-1,+1)= α + βhti + ωhti*mtb+ γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (iii) Table 6 shows the empirical results of the above estimation equation in Models 1, 2, and 3. We find a positive and significant coefficient for the interaction variable, indicating that when high-growth firms are accompanied by stronger governance, investors react positively to an SEO announcement. This result is still pronounced for firms issuing prior to 2000, which again suggests that the takeover market is more restricted post-2000, and governance is more insulted than in the 1990s. [Table 6 here] Next, with the same justifications, we replace growth (mtb) with stock price run-up (pastret) as a major determinant of negative market reactions in order to identify the impact of market timing as a negative signal to the market considering governance power. Here our variable of interest is the interaction of governance and share price run-up: HTI*pastret. The following equation shows the empirical model for our estimation of market reaction: CAR(-1,+1)= α + βhti + ωhti*pastret+ γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (iv) In Table 6, Models 4, 5 and 6 present the empirical evidence. Not surprisingly, the interaction term shows no significant impact on the dependant variable. This is most likely due to the strong explanatory power of pastret, as this washes out the variation effect of the interaction term, as well as the slow movement of HTI as explained by Cain et al. (2017). 17

18 However, pastret still shows a significant negative relationship to market reaction, which again posits the importance of the timing factor and its negative signal to the market, consistent with Loughran and Ritter s (1997) timing theory and other empirical evidence in this line. In order to tackle this issue and clearly identify the cross-sectional effect of HTI in the presence of timing as a key variable, we divide our sub-samples into two different groups of high and low growth opportunity and stock price run-up. This classification allows me to narrow our investigation and better capture the variations in slow-moving HTI. For both the pre-2000 and the post-2000 subsample, we categorize firms as having low (high) market-tobook and timing if the mtb and pastret of the respective firm are lower than (above) the median. We then investigate the effect of governance power on investors reactions for the subsamples of firms with low and high growth opportunity and timing. We report the results in Table 7, with the same model description as our base model (equation i). [Table 7 here] Models 1, 2, 5 and 6 present the results of the effect of governance power on investors reactions for low (high) growth firms. We highlight that HTI has significant implications only for firms with high growth opportunities. The effect is more pronounced for issues that occurred prior to These findings suggest that in the absence of timing as one-third of the CEOs surveyed by Graham and Harvey (2001) do not declare timing to be an initiative for SEOs investors are surprised and cannot react purely negatively due to stock not being overpriced. Therefore, good governance can give them confidence that the offered share is valuable. 18

19 My main inference in this section arises from Models 3, 4, 7 and 8, where we isolate the two groups of high run-up (High pastret) and low run-up (Low pastret), then run the base regression model separately for both. As stated in our third hypothesis, we expect to find evidence of a positive relationship between HTI and investors reactions when investors are surprised by an SEO for which they see no signal of overpriced stock. In this situation, strong governance will become a mechanism for spreading certainty across the market about the firm s value. Masulis and Korwar (1986) find that a larger price run-up is associated with a larger price drop. They also report that a large fraction of their sample does not make a stock offering following a run-up. According to timing theory, investors expect to see an equity offering after observing a large price run-up, and they have sufficient time to scrutinize the firm s quality. Thus, the effect of governance strength on investors reactions can be more pronounced when the market is surprised by an SEO and has no time in which to assess the firm s quality. I argue that if there is a high likelihood of a hypothetical hostile takeover effort, the market translates it to a firm having strong governance in place, and finally into investors having more confidence in an SEO announcement. We continue to run our regression based on the pre-2000 and post-2000 classifications. Models 3 and 7 confirm that there is a positive and significant association between HTI and CAR (-1, +1) for both pre- and post-2000 year-wise classification across the low pastret group, representing surprised investors. However, we find no evidence of a relationship between HTI and HTI and CAR (-1, +1) within the high pastret group that has a price run-up above the median range. This result suggests that HTI can reflect an insight into information asymmetry, which is not as of our focus in this study. We therefore call for further research on the relationship between HTI and information asymmetry in order to shed more light on this newly-developed measure. 19

20 4.4 Robustness Analysis In this section, we perform two separate robustness tests on the impact of HTI on investors reactions. First, while investors react positively to SEOs by firms with higher HTI, they may react similarly if there is an improvement in HTI. We define improvement as a percentage change between HTI in the year of an SEO and HTI one year prior to that. If the change is positive, it is considered to be an improvement in HTI, and if it is negative it is considered to be a diminishment. we substitute HTI with ΔHTI as a representative of percentage change. The estimating equation is as follows: CAR(-1,+1)= α + β ΔHTI + γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (v) Table 8 represents the results of the above estimation. We find positive and significant coefficient for ΔHTI only for the post-2000 subsample. This indicates that an improvement in HTI from the year prior to an SEO to the year of the SEO gives investors confidence in the issuing firm and its governance strength. The insignificant coefficient for the pre-2000 subsample may be due to the large movement in HTI prior to 2000, which was imposed by fewer legal changes and changes in laws compared to the post-2000 subsample, which can be also explained by Cain s et al. (2017) index construction process. [Table 8 here] Following Kim and Purnanandam (2013), we use a 3-day window of (-1, +1) to capture the full effect in the main analysis; however, the full effect may also include informed investors reactions. A shorter window can capture the immediate market reaction through a public announcement and eliminate these informed reactions. As a second robustness test, we 20

21 therefore use a shorter event window of (0, +1) to comply with prior studies on signalling and adverse selection hypothesis in SEOs, particularly Brealey et al. (1977) and Myers and Majluf (1984). I substitute CAR(-1,+1) with CAR(0,+1) in our base model and run 4 different regressions as follows: Model 1 tests the sole effect of HTI on CAR(0,+1), without considering other controlling variables; Model 2 incorporates a growth factor (mtb) into our regression; Model 3 captures the effect of HTI along with a timing factor (pastret); and finally, Model 4 includes HTI and all controlling variables as previously employed in our base model. This allows me to include a horse race as an additional robustness test and identify which prominent component or components can explain un-informed investors reactions through a shorter event window. Table 9 represents the results of our second robustness test and the final outcome of the horse race between timing and growth factors. First, the result of Model 1 shows that governance is still significantly associated with investors reactions when informed investors reactions are removed by narrowing the event window. Second, the results of Model 2 and 3 provide empirical evidence for a significant association between HTI and un-informed investors reactions, while the two prominent factors of timing (pastret) and growth (mtb) can largely explain the negative reaction. And finally, Model 4 shows that HTI has a significant positive relation to CAR(0,+1) when controlling for all important variables. These results show that our main finding that there is a significant positive relationship between governance strength as indicated by fewer external legal enforcements and investors reactions to SEOs is robust and can be extended to both informed and uninformed investors. [Table 9 here] 21

22 4.5 Additional Analyses In their recent study, McCahery et al. (2016) give a novel insight into the role of institutional investors in corporate governance by surveying them. However, prior theoretical and empirical papers show little direct or analogical evidence on the influential moves and voices of institutional investors on corporate governance (see for example Shleifer & Vishny, 1986; Huddart, 1993; Faure-Grimaud & Gromb, 2004; Admati & Pfleiderer, 2009; Edmans, 2009; Edmans & Manso, 2010). McCahery et al. (2016) find that long-term investors intervene more intensively as a complementary governance mechanism. In Table 10, we examine the relationship between the HTI and investors reactions to SEOs, controlling for the percentage of shares held by dedicated institutional investors, as a proxy to measure the extent to which an additional governance mechanism is in place. Dedicated institutions are characterized as having large stakes and low turnovers in their holdings (Bushee & Noe, 2000) and often have better private information about their portfolio firms (Porter, 1992). we identify a dedicated type of investors among the three possible types described in Bushee (1998, 2001) (the other two are quasi-indexers and transient ). we argue that a higher value in HTI represents stronger corporate governance arising from an external shock, and thus is positively related to investors reactions, and we expect that the relationship will persist when controlling for a complementary governance mechanism at firm level. [Table 10 here] 22

23 Table 10 reports the results of our estimation of investors reactions to HTI, controlling for dedicated institutional investors (IO_ded) and all other controlling variables so far used in this study, as shown in the following equation: CAR(-1,+1)= α + β HTI + ψio_ded + γmtb + δlogsale + θcash/ta + λleverage + ζresstd + ηpastret+ φmixed + ε (vi) In all models, HTI carries a positive and significant coefficient except for the final column, in which the sample includes SEOs made after 2000 with a positive but marginally significant level of 10%. Notwithstanding, all results are consistent with our prior findings and again posit that there is a significant relationship between external shocks strengthening governance and investors confidence in a firm s value. Overall, we find evidence for the importance of exogenous factors contributing to governance strength in investors determinations about SEOs and firms value in general, while controlling for an endogenous governance mechanism at firm level. A deep investigation of the relationship between firm-level governance mechanisms and exogenous shocks to governance is outside the scope of this study, but further examination of the equilibrium between the HTI and any firm-level governance protection can be investigated by future researchers. 5. Conclusion The hostility of a takeover market exerts a statistically and economically significant influence on investors reactions to the announcement of SEOs. A positive relationship between the two is more pronounced for equity offerings made prior to the year 2000 due to less regulation being imposed on the takeover market compared to the years post The evidence suggests that investors are less worried about the misuse of capital if a firm s governance is strong. 23

24 Although timing theory can generally explain why the market reacts negatively to SEOs, there is remarkable portion of firms that can initiate equity offerings without a drastic increase in their stock price prior to their intention. In these so-called surprised market circumstances, we find that governance strength as measured through the HTI mitigates the magnitude of negative investor reactions. We also find that while high-growth firms are more subject to improper use of the raised capital, higher HTI is a channel through in which investors can gain more confidence that capital will be correctly used. My cross-sectional analyses are not subjected to endogeneity, as HTI serves as a middle-ground proxy between external and internal corporate governance mechanisms. our robustness test confirms that there is a significant and positive association, as using an alternative event window plus price run-up and growth opportunities can still exclusively explain negative market reactions to SEOs. Finally, the significant and positive relationship between HTI and investors reactions still holds while controlling for an endogenous corporate governance measure at firm level. Overall, our evidence on SEO announcements and the importance of the hostile takeover index and timing effects support theories which argue that corporate governance and stock overpricing can drive the magnitude of investors negative reactions. Tellingly, a situation in which there are fewer external protective takeover laws empowers corporate governance and extends more certainty about the proper use of proceed to the market. 24

25 Table 1. Frequency distribution Panel A shows number of SEOs, percentage, and cumulative percentage by year. Panel A Year Freq. Percent Cum Total 2,415 Panel B shows distributions across firm and year clusters. Panel B Freq. Percent Cum Number of issues by single firm N 2,415 1,

26 Table 2. Descriptive statistics of main variables This table provides the descriptive statistics of SEOs in our sample. We provide the mean, standard deviation, min and max for key variables in our entire sample. CAR(-1,+1) is the cumulative abnormal return as benchmarked by the value-weighted market index over a 3-day window around the filing date. GS is the gross spread paid to the lead underwriter as measured by the percentage amount of gross proceeds. HTI is the hostile takeover index as measured in Cain et al. (2017). Mtb is the MB ratio measured by the market value of firm s equity plus value of assets minus book value of assets, all scaled by the book value of assets. logsale is the natural logarithm of sales of the last fiscal year in millions. Cash/ta measures cash and marketable securities as a percentage of book value of assets. Leverage measures the ratio of total long- and short-term debt scaled by total assets. Resstd is the log of standard deviation of residuals from the market model regression using the previous year s daily returns. Pastret is the log of 1 plus return of firm over the year before 1 month prior to filing. Variable Mean Std. Dev. Min Max CAR(-1,+1) CAR(0,+1) HTI Mtb logsale Cash/ta Leverage Resstd pastret

27 Table 3. Correlation This table provides the correlation coefficients between key variables in this study. CAR(-1,+1) is the cumulative abnormal return as benchmarked by the valueweighted market index over a 3-day window around the filing date. GS is the gross spread paid to the lead underwriter as measured by the percentage amount of gross proceeds. HTI is the hostile takeover index as measured in Cain et al. (2017). Mtb is the MB ratio measured by the market value of firm s equity plus value of assets minus book value of assets, all scaled by the book value of assets. logsale is the natural logarithm of sales of the last fiscal year in millions. Cash/ta measures cash and marketable securities as a percentage of book value of assets. Leverage measures the ratio of total long- and short-term debt scaled by total assets. Resstd is the log of standard deviation of residuals from the market model regression using the previous year s daily returns. Pastret is the log of 1 plus return of firm over the year before 1 month prior to filing. CAR(-1,+1) 1 CAR(-1,+1) HTI mixed mtb mixed logsale Cash/ta Leverage Resstd pastret HTI mixed mtb logsale Cash/ta Leverage Resstd pastret

28 Table 4. Base model, Takeover index and market reaction This table presents OLS estimation results of CAR (-1,+1) as dependant variable with White heteroscedasticity robustness of standard errors. Year fixed effects are included in all models. We also separate models into all, filings occurs prior to year 2000, and after CAR(-1,+1) is the cumulative abnormal return as benchmarked by the value-weighted market index over a 3-day window around the filing date. HTI is the hostile takeover index as measured in Cain et al. (2017). Mtb is the MB ratio measured by the market value of firm s equity plus value of assets minus book value of assets, all scaled by the book value of assets. logsale is the natural logarithm of sales of the last fiscal year in millions. Cash/ta measures cash and marketable securities as a percentage of book value of assets. Leverage measures the ratio of total long- and short-term debt scaled by total assets. Resstd is the log of standard deviation of residuals from the market model regression using the previous year s daily returns. Pastret is the log of 1 plus return of firm over the year before 1 month prior to filing.. t-stats are reported are reported in brackets with ***,**,and * representing statistical significance of coefficients at 1%, 5% and 10% level, respectively. 28

29 Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 All All HTI (4.29)*** (2.66)*** (3.04)*** (2.09)** (3.05)*** (1.76)* mtb (-0.42) (-1.05) (0.59) logsale (-2.18)** (-2.69)*** (-0.79) Cash/ta (-1.64) (-1.36) (-1.01) Leverage (-1.23) (-1.71)* (-0.33) Resstd (-4.46)*** (-3.37)*** (-3.04)*** pastret (-4.89)*** (-4.58)*** (-2.79)*** mixed (-2.12)** (-0.91) (-2.11)** Constant (-4.63)*** (-0.63) (3.07)*** (-3.33)*** -0.9 Fixed Effects Year Year Year Year Year Year R N

30 Table 5. Interactive effect of takeover index and capital expenditures This table presents OLS estimation results of CAR(-1,+1) as dependant variable with White heteroscedasticity robustness of standard errors. Year fixed effects are included in all models. We also separate models into all, filings occurs prior to year 2000, and after CAR(-1,+1) is the cumulative abnormal return as benchmarked by the value-weighted market index over a 3-day window around the filing date.. Δacq measures changes in the dollar amount of acquisitions in the year of an SEO minus the respective amount for the prior fiscal year, scaled by total assets at the beginning of the prior fiscal year. Similar to Δacq, we measure Δcapex using capital expenditures and acquisition included. HTI is the hostile takeover index as measured in Cain et al. (2017). Mtb is the MB ratio measured by the market value of firm s equity plus value of assets minus book value of assets, all scaled by the book value of assets. logsale is the natural logarithm of sales of the last fiscal year in millions. Cash/ta measures cash and marketable securities as a percentage of book value of assets. Leverage measures the ratio of total long- and short-term debt scaled by total assets. Resstd is the log of standard deviation of residuals from the market model regression using the previous year s daily returns. Pastret is the log of 1 plus return of firm over the year before 1 month prior to filing. t-stats are reported are reported in brackets with ***,**,and * representing statistical significance of coefficients at 1%, 5% and 10% level, respectively. 30