The influence of industry on the announcement effect of convertible bond issues for US listed firms

(1)

The influence of industry on the announcement effect of convertible bond

issues for US listed firms

Author: Yves van den Berk Student number: 6063608 Supervisor: Mr. P. Trietsch

University of Amsterdam

BSc Economics and Business Track: Finance and Organization Amsterdam, 19 June 2015

(2)

1

1 Introduction

Convertible bonds are bonds that can be converted into shares. Convertible bonds are an important source of financing for companies. With $268 billion in convertible bonds outstanding, the convertible bond market is relatively small compared to the $9.6 trillion in U.S. corporate bonds outstanding. Nevertheless the convertible bond market is extensively examined in the finance literature. Announcements of convertible bonds issues are often associated with negative abnormal returns (Dann and Mikkelson (1984), Arshanapalli et al (2004), Rahim, Goodacre & Veld (2012).

The type of industry of a company can have an influence on the announcement effect. Suchard (2007) founds that industrial companies are associated with larger cumulative abnormal returns compared to other companies. During a financial crisis, markets become more volatile. It’s possible that during these years the convertible bond market behaves differently than before. This paper varies from existing literature because it will examine the effect of industry on the announcement effect of convertible bond issues for the US stock market before, during and after the crisis. The research question is as follows:

RQ: Does a firm’s industry influence the stock price reaction after the announcement of a convertible bond issue?

In order to answer the research question an event study will be used. The cumulative average abnormal returns of US listed firms will be determined using a market model. Furthermore a regression will be done in order to explain the possible average abnormal returns.

In the second chapter the different theories why firms issue convertible debt will be explained. Furthermore, the key drivers: industry, crisis and firm size will be discussed. In the third chapter the methodology to investigate the research question, the gathering of data and the expected results will be explained. In the fourth chapter the results will be presented and compared to the existing literature. The last chapter will give an overall conclusion.

(4)

3

2 Literature review

2.1 Convertible bond theories

A convertible bond is a form of debt that can be converted into shares. There are several reasons why firms issue convertible debt. One reason, first mentioned by Stein (1992), is that firms issue convertible debt as an indirect way to increase equity. This way to increase equity is called “equity through the back door”. It is used by companies in which information asymmetry plays an important role. Issuing convertible debt may decrease the adverse selection costs associated with the pure equity issues.

A second reason why firms issue convertible debt was first mentioned by Mayers (1998) and is called the sequential-financing theory. By using convertible bonds, companies lower the issuance costs of “sequential financing” and minimize the agency costs of

overinvestment. Convertible bonds provide sufficient internal funds for future investments and minimize the overinvestment costs.

A third theory why firms issue convertible bond is called the “risk-shifting” theory (Green, 1984). Shareholders of firms with high leverage will benefit from high risk projects while the debt-holders will suffer the losses. This may influence managers to take on high-risk projects, which can result in an overall cost to the firm. In order to prevent this from happening, firms can decide to issue hybrid debt.

2.2 Announcement Effect

Announcements of convertible bonds issues are generally associated with negative abnormal stock returns. Dann and Mikkelson (1984), Arshanapalli et al (2004) find negative cumulative abnormal returns in their research on convertible bond issues in the US stock market.

Research done by Christensen et al. (1996) and De Roon and Veld (1998) shows that it is also possible that positive abnormal returns occur.

A possible explanation for these differences could be corporate governance systems. Positive announcement effects are often found in network orientated countries like Japan. In Japan issues are guaranteed by a bank, this can explain the positive announcement effect. Negative announcement effects often occur in market orientated countries like the US in which issues are not guaranteed by a bank (Rahim, Goodacre & Veld 2012).

Another possible explanation for the differences in abnormal returns is called

(5)

4 their developed model. In this model managers have more information than shareholders. According to the model an equity issue is perceived as bad news because managers will try to sell overprized stock to maximize the wealth of the existing shareholders. The model predicts that equity issues will be associated with a more negative abnormal return than a debt issue. The differences in abnormal returns can occur because investors treat the convertible debt more as debt/equity which causes a smaller/bigger announcement effect.

Table 1 summarizes the research done on convertible bonds issues in different countries

Table 1: Previous research on convertible bond issues

Study Country Period Event

window

Number of observations

CAAR (%)

Dann and Mikkelson (1984)

USA 1970-1997 (-1,0) 132 -2.31***

Jayamaran, Shastri, and Tandon (1990)

USA 1977–1986 (-1,0) 54 -.064

Arshanapalli et al. (2004) USA 1993-2001 (-1,0) 85 -3.07*** Loncarski, Ter Horst, and

Veld (2008)

Canada 1991-2004 (-1,0) 86 −0.54*

Abhyankar and Dunning (1999) Suchard (2007) United Kingdom Australia 1986-1996 1980–2002 (0,1) (0,1) 112 58 -1.21*** −0.40 Christensen et al. (1996) Japan 1984–1991 (−1,0) 35 0.60 De Roon and Veld (1998) The Netherlands 1976–1996 (-1,0) 47 0.16

Rahim et. Al (2012) Malaysia 1996-2009 (-1,1) 105 -1,10**

It is clear from table 1 that convertible bonds issues in the US stock market are generally associated with negative cumulative abnormal returns. The only countries in table 1 that have

(6)

5 positive abnormal returns are the Netherlands and Japan. These results are insignificant though.

A possible explanation that announcements of convertible bond issues are more often associated with negative abnormal returns is general underpricing of all public offerings (Rock, 1986). Because of this underpricing, wealth will be transferred from stockholders to bondholders and that’s why negative abnormal returns occur.

2.3 Firm industry

Previous research has shown that a firm’s industry can have an influence on the

announcement effect of a hybrid debt issue. Research done by Smith (1986) shows that industrial companies have larger negative announcement effects than utility companies. A possible explanation for this effect is that utility companies tend to issue more external capital than other companies like industrial companies. As a result of this, the announcement effect is less negative.

Suchard (2007) compares resource firms with industrial firms in a research on convertible debt issues in the Australian stock market. It shows that the dummy industrial is significant negative. The possible explanation for this effect is that resource firms have more sequential-financing needs. When a next investment stage arrives a company can force the convertible bond into a share. Because of these sequential-financing needs the market reacts more positively (less negative) on a convertible bond announcement.

2.4 Crisis

In the convertible bond market the period (1984-1999) is called the Traditional Investor Period, the period (2000-september 2008) is called the Arbitrage Period (Duca et al, 2012). In the Traditional Investor Period the convertible bond market is characterized as market long- only investors. During the Arbitrage Period more convertible bond arbitrage funds appear which causes a more negative announcement effect. The arbitrage funds buy the convertible bonds and short the underlying stock, this causes a price pressure.

According to Masters (2009) during the financial crisis hedge funds play a much smaller role in the convertible bond market. Because of this it can be assumed that the financial crisis has a positive influence on the announcement effect of convertible bond issues.

(7)

6

2.5 Firm size

According to the signaling theory (Myers and Majluf 1984), firm size can have an influence on the announcement effect of convertible bond issues. Large (listed) firms have the

obligation to release certain information publicly. Small firms have more possibilities to keep certain information private. As a result of this the announcement effect of large firms is smaller than small firms. The more information a company releases, the smaller the market reaction to an announcement of an issue will be. Gebhard (2001), finds a positive abnormal return (+0.57%) for large German companies that issue convertible debt. This is lower than the effect for small companies (3.12%).

Brennan and Schwartz (1988) describe a model in which the firm’s choice of

financing instruments is described. Their model predicts that risky, high growth firms will be more likely to issue convertible debt. This implies that firm size has a negative influence on the announcement effect of convertible bond issues.

2.6 Firm Growth

A firm’s growth opportunities can be defined as the ratio of the market value of the firm to the book value of its assets. This ratio is called Tobin’s Q. According to the model of Brennan and Schwartz (1988) firms with a high Q will be associated with a more positive announcement effect than a firm with a low Q. Matured firms have limited growth

opportunities which causes a more negative announcement effect. Abhyankar and Dunning (1999) find a positive effect (+1.65) of Tobin’s Q in their research on the announcement effect of convertible bond issues in the UK stock market.

(8)

7

3 Methodology and data

3.1 Methodology

The research method used to answer the research question is called an event study. During an event study, the stock price reaction to a certain event will be determined. This will be done be comparing the actual stock prices with the stock prices that would have occurred without the influence of the event. In order to calculate the differences in stock prices, the abnormal returns will be determined.

To calculate the abnormal returns (AR) a standard event study is used (Mikkelson and Partch, 1986). The abnormal returns are calculated in two steps:

First the expected returns are calculated by using a market model. The alpha and beta are calculated for each firm individually by doing an OLS regression. The estimation period for this model is 171 days, starting 180 days before the announcement and continuing until 10 days before the announcement (-180, -10). This estimation period is based on earlier research done by Rahim (2012) on convertible bond issues.

The following formula will be used:

Market model: 𝑅_𝑖.𝑡 = α_i+ β_i𝑅_𝑟,𝑡+ ε_i,t

𝑅_𝑖,𝑡 = return of security in period t 𝑅_𝑚,𝑡 = return of the market

𝛼 = alpha of the security i

𝛽 = beta of the security i

ε_i,t= the error term with ε_i,t~ N(0, σ2)

Secondly, the abnormal returns for each firm in period t are calculated by subtracting the actual return minus the expected return.

(9)

8 To calculate the average rate of abnormal return (AAR) the following formula will be used:

𝐴𝐴𝑅𝑡 =

∑𝑛𝑖=1𝐴𝑅𝑖,𝑡 𝑛

in which n stands for the number of companies.

To calculate the cumulative abnormal returns (CAAR) of a period, the average abnormal returns of the corresponding days are cumulated:

𝐶𝐴𝐴𝑅𝑡 = ∑ 𝐴𝐴𝑅𝑡 −𝑛 𝑛

This research will use three different event windows: (-1,0), (-1,1) and (0,1). This is based on earlier research done on convertible bond issues. Dann and Mikkelson (1984), Arshanapalli et al. (2004) and Mohd Ashhari and Sin-Chun (2009) all use the event window (-1,0), this is the most common event window in the finance literature about hybrid debt. The event windows (-1,1) and (0,1) are used as well by Abhyankar and Dunning (1999), Suchard (2007) and Rahim (2012).

In order to test the significance of the average abnormal returns the t-statistic first presented by Brown and Warner (1985, pp. 7) will be used. The formula is as follows:

𝑇 − 𝑠𝑡𝑎𝑡𝑖𝑠𝑡𝑖𝑐 = 𝐴𝐴𝑅/𝑆𝐷(𝐴𝐴𝑅) in which AAR represent the average abnormal return and SD (AAR) represents the standard deviation for the average abnormal return calculated during the estimation period.

(10)

9

3.2 Regression

The following regression will be tested:

𝑐𝑎𝑎𝑟_𝑖,𝑡 = 𝛽₀+ 𝛽₁𝑖𝑛𝑑𝑢 + 𝛽₂𝑢𝑡𝑖𝑙 + 𝛽₃𝑐𝑟𝑖𝑠𝑖𝑠 + 𝛽₄ln(𝑖𝑠𝑠𝑢𝑒) + 𝛽₅ln(𝑎𝑠𝑠𝑒𝑡) + 𝛽₆Q + 𝜀_𝑖,𝑡

Table 2: Regression variables

Variable Type Definition Expected

Sign caar indu util crisis ln(issue) ln(asset) Q continuous dummy dummy dummy continuous continuous continuous

Cumulative average abnormal returns (%) 1 = industrial firm, 0 ≠ industrial firm* 1 = utility firm, 0 ≠ utility firm*

1 = during crisis, 0 = before/ after crisis natural logarithm of convertible bond issue size natural logarithm of market value

a firms Tobin’s Q - - + + ? +/- +

*Other firms are mostly financial firms

In the next section the expected signs of the coefficients will be predicted. The dummy industry is the main explanatory variable. The expected sign for industrial firms is negative. This is based on earlier research done by Janjigian (1987) and Suchard (2007) who find a negative coefficient of - 0.08 and -.07. In order to categorize the different firms the SIC codes will be used

The expected sign for utility firms, the second dummy, is expected to be positive. This is based on research done by Janjigian (1987) and Suchard (2007). Firms which are not categorized as industrial firms or utility firms are mostly financial firms and healthcare firms. The dummy crisis is expected to be positive, this is based on research done by Duca et al (2012). Because less convertible bond arbitrage fund are active during the financial crisis the dummy is expected to be positive. The fall of Lehman Brothers is taken as a start of the

(11)

10 financial crisis and June 2009 as the end of the crisis.

The first control variable that is included in the regression is issue size of the convertible bond issue. Based on earlier research there is nog indication that issue size will influence the announcement effect. But since there is a large variation in issues size (2.1 million euros smallest value, 6900 euros biggest value) this control variable is added. The logarithmic issue size is used because it is more normally distributed than the discrete issue size (Strong, 1992).

The second control variable that is included in the regression is market value of assets. Based on earlier research done by Myers and Maljuch (1984) and Gebhard (2001) the expected sign is positive. Smaller firms have more possibilities to keep information private, this creates a larger information asymmetry between investors and management. This causes a bigger announcement effect. Based on earlier research done by Brennan and Schwartz (1988) it is also possible that the sign of market value of assets is negative. Firms that typically issue convertible bonds are small high growth firms. Large firms have limited growth opportunities which causes a bigger announcement effect.

Tobin’s Q is the third control variable that is included in the regression. The expected sign is positive, firms with limited growth opportunities cause a larger announcement effect. This is based on earlier research done by Abhyankar and Dunning (1999).

3.3 Hypothesis

Based on earlier research done by Dann and Mikkelson (1984), Jayamaran, Shastri, and Tandon (1990) and Arshanapalli et al. (2004) on the announcement effect of convertible bond issues in the US stock market, it can be assumed that there will be a negative announcement effect. This implies that CAAR will be negative.

To explain the possible negative announcement effect, the most important hypothesis that will be tested is:

A firms industry doesn’t have any influence on the stock price reaction of an announcement of a convertible bond issue.

𝛽₁ = 0

A firm’s industry does have an influence on the stock price reaction of an announcement of a convertible bond issue.

(12)

11

3.4 Data

This study investigates convertible bond issues between 2004 and 2014 in the US stock market. This period is chosen because it includes a sufficient number of issues before and after the crisis. The following search criteria are used in Bloomberg to collect the sample: the announcement date is between 2004 and 2014, the issue size is at least 1 million dollar, the company is a listed US company. Using these criteria 73 issues are found.

The daily stock returns of each corresponding company is found be using Datastream. The daily stock price returns of the benchmark, the S&P 500, will be gathered from

Datastream as well. The benchmark S&P 500 is chosen because all companies in the sample are US listed companies and the S&P500 contains the 500 biggest US listed companies. For each company individually the following data is obtained from CRSP: type of company (using the SIC code) and market value of assets and leverage.

Using these three different databases, 63 announcements of convertible bonds issues remain. Table 3 shows the distribution of the different convertible bond issues in years.

Table 3: distribution of years of convertible debt issues

Year of announcement Number of announcements in the final sample 2004 7 2005 10 2006 3 2007 5 2008 8 2009 1 2010 1 2011 2 2012 7 2013 7 2014 12 2004-2014 63

(13)

12 Table 4 and table 5 present descriptive statistics about issue size, market value, Tobin’s Q and the dummy variables.

Table 4: Descriptive statistics

Average Minimum Maximum

Issue Size (million) 545.4 2.1 6900

Market Value (million) 30748 30 183124

Tobin’s Q (%) 52.9 7.16 412

A large difference in market value of the different companies and issue size can be identified

Table 5: Descriptive statistics of dummy variables

Variable name N Value Percentage

Crisis 56

7

0= before/ after crisis 1= during crisis

88.9% 11.1%

Industrial 49

14

0= not industrial company 1= industrial company

77.8% 22.2%

Utility 44

19

0= not utility company 1= utility company

69.8% 30.2%

(14)

13

4 Results

This section discusses the results for the cumulative abnormal returns of the 63 US listed firms. First the average abnormal returns and the cumulative abnormal returns will be discussed. Secondly the regression with CAAR as a dependent variable and industry and crisis as independent variables will be discussed. The last part will be a discussion of the results.

4.1 Results

Table 6 shows the abnormal returns for three different event days (-1,0,1). It is clear from this table that biggest announcement effect is at T=-1 (-1.18), this effect is significant (1%). The smallest announcement effect is at T=1 (0.03), this result is not significant.

Table 6: Average abnormal returns

Event Day AAR (%) T-test Minimum Maximum

-1 0 1 -1.18 -0.69 0.03 -2.70*** -1.37 0.10 -19.86 -24.17 9.02 6.35 10.92 9.29 CAAR (-1,0) CAAR (-1,1) CAAR (0,1) -1.87 -1.15 -0.66 2.61** 2.15** 0.98 -24.17 -19.86 -24.17 10.92 9.29 10.92

Significance levels at 1%, 5% and 10% presented by ***, **, *, respectively

The largest cumulative abnormal returns can be found in the event window (-1,0). This is similar to earlier research done on convertible bond issues in the US stock market by Dann and Mikkelson (1984) and Arshanapalli et al. (2004).

Table 7 shows the results of the regression with CAAR (-1,0) as a depended variable and the dummy’s industrial, utility, crisis and the control variables as independent variables.

(15)

14

Table 7: Regression on cumulative average abnormal returns for the event window (-1,0)

CAAR (-1,0) CAAR (-1,0) CAAR (-1,0) CAAR (-1,0)

1 2 3 4 indu util crisis ln(issue) ln(market) Tobins Q constant -.0585*** (.0173) -.0150 (.0160) -.0015 (.0097) -.0581*** (.0173) -.0168 (.0160) -.0252 (.0217) .0016 (.0100) -.05401*** (.0206) -.0129 (.0193) -.0234 (.0230) -.0016 (.0045) .0007 (.0030) -.0041 (.0496) -.0526*** (.0203) -.0023 (.0204) -.0239 (.0227) -.0015 (.0044) -.0003 (.0031) -.0188 (.0125) .0196 (.0516) R-squared Adjusted R2 Number of obs 0.1591 0.1311 63 0.1780 0.1362 63 0.1802 0.1083 63 0.2120 0.1275 63

Significance levels at 1%, 5% and 10% presented by ***, **, *, respectively Standard errors are between brackets

Model 1 in table 7 shows that when only the dummy’s industrial, utility and a constant are included in the regression there seems to be a negative influence of industrial and utility firms on the cumulative average abnormal returns . Industrial firms have a 5.85 % smaller CAAR, significant at the 1% significance level. Utility firms have a 1.5% smaller CAAR: result is not significant though. When the R-squared is taken into account, only 15.9% of the variation is explained by this model.

In model 2 crisis is added to the regression. The dummy industrial is a little less than the previous model (5.81%) but remains significant. The dummy utility increases to 1.68% but remains not significant. The dummy variable crisis is negative (2.52%) but not

(16)

15 significant. The R-squared has increased to 17.8%.

In model 3 the control variables issues size and market value are included in the regression. The dummy industrial remains significant. The control variable issues size is negative but not significant and the control variable market value is positive but not significant.

In the last model all control variables are included in the regression. In this model 21.2% of the variation can be explained by the model. The dummy industrial is 5.26% and significant. No significant influence of the dummy utility or crisis can be identified. No significance influence of the control variables can be identified.

Table 8: Expected signs and results

Variable Definition Expected

Sign Result caar indu util crisis ln(issue) ln(asset) Q

Cumulative average abnormal returns (%) 1 = industrial firm, 0 ≠ industrial firm* 1 = utility firm, 0 ≠ utility firm*

1 = during crisis, 0 = before/ after crisis natural logarithm of convertible bond issue size natural logarithm of market value

a firms Tobin’s Q - - + + ? +/- + - - - - - +- -

When using robust errors the variables industry and utility are significant (5 %) in model 1 and 2. In model 3 and 4 nog significant results can be found (appendix: table 9). A possible explanation for this is the number of observations. To check for possible multicollinearity, a correlation matrix is used (appendix: table 10). There doesn’t seem to be any

(17)

16

4.2 Discussion

In this section the results of the regression will be compared to the literature.

For all the models presented in table 7 the dummy industrial is significant and the value is between -5.26% and -5.85%. The results are in line with the findings of Janjigian (1987) and Suchard (2007) who found a negative effect of industrial firms of -8.0 and -7.0 %. The dummy utility was expected to be less than the dummy industrial. The value of the regression is between -.23% and -1.68%. This is not in line with the literature, the expected sign was positive.

The coefficient for the dummy crisis is negative in all models, but without statistical significance. This contradicts the existing research done by Ducae et.al (2012). A possible explanation for this negative effect is the high market volatility and very high offering underpricing that are generally associated with a financial crisis. This causes a larger announcement effect.

No significant influence of the control variables can be found. The dummy market value of assets is both negative and positive. The dummy Tobin’s Q is negative, this

contradicts earlier research done by Abhyankar and Dunning (1999). A possible explanation for the insignificance of the control variables is the number of observations.

(18)

17

5 Conclusion

In this thesis the influence of a firm’s industry on the stock price reaction after the

announcement of a convertible bond is studied. In total 63 US listed firms are included in the model. In order to answer the research question an even study is used. The cumulative average abnormal returns are determined using a market model. Furthermore a regression is used in order to explain the cumulative average abnormal returns.

We find support for the testable implications of research done by Janjigian (1987) and Suchard (2007). A firm that is classified as an industrial company has a negative effect of 5.26% on the cumulative average abnormal returns when the control variables are included in the regression. This result is significant at the 1% level. Therefore 𝐻₀which stats that 𝛽₁ = 0 can be rejected.

Furthermore the influence of the financial crisis is negative on the announcement effect of hybrid debt (-2.39%). This result is insignificant and contradicts with earlier research done by Duca (2012). A possible explanation for this effect is that during a crisis there is a higher market volatility and very high offering underpricing, this causes a larger announcement effect.

The convertible bond market is changing rapidly. Future research can focus more on the different types of convertibles bonds that are used. For example the contingent

convertible bonds. These are bonds that are automatically converted when a firm or bank reaches a specific level of financial distress.

(19)

18

References

- Abhyankar, A. and A. Dunning (1999): “Wealth effects of convertible bond and convertible preference share issues: An empirical analysis of the UK market” Journal of Banking and

Finance

- Arshanapalli, F. Fabozzi, L. Switzer, G. Gosselin (2004) New evidence on the market impact of convertible bond issues in the US. Finance Letters, 3 (2005), pp. 1–6

- Brennan, M. and Schwartz, E., (1988). The case of convertibles. Journal of Applied Corporate Finance 1, pp. 55-64.

- Chang, S.-C., Chen, S.-S., Liu, Y. (2004) Why firms use convertibles: A further test of the sequential-financing hypothesis. Journal of Banking and Finance

- Christensen, D.G., Faria, H.J., Kwok, C.C.Y. and Bremer M., (1996). Does the Japanese

stock market react differently to public security offering announcements than the US stock market. Japan and the World Economy 8, pp. 99-119.

- Dann, L.Y. and W. H. Mikkelson (1984): „Convertible debt issuance, capital structure change and financing-related information: Some new evidence,” Journal of Financial

Economics

-Duca, E. Dutordoir, M., Veld, C., and Verwijmeren, P. (2012) Why are convertible bond announcements associated with increasingly negative issuer stock returns? An arbitrage-based explanation. Journal of Banking and Finance

- De Roon, F., Veld, C. (1998). Announcement effects of convertible bond loans: An empirical analysis for the Dutch market. Journal of Banking and Finance

- Gebhardt, G. (2001) .Announcement effects of financing decisions by German companies: Synthesis of an empirical research programme. Working paper, Johann Wolfgang Goethe University Frankfurt, Germany

- Green, C., (1984), Investment incentives, debt, and warrants. Journal of Financial Economics 13: 115-136.

- Jayamaran, N., Shastri, K. and Tandon, K., (1990). Valuation effects of warrants in new security issues. Working Paper, Salomon Center, New York University.

- Lewis, C. M., Rogalski, R. J., Seward, J. K. (2003) Industry conditions, growth opportunities and market reactions to convertible debt financing decisions. Journal of

(20)

19

- Loncarski, I., Ter Horst, J. and Veld, C., (2008). Why do companies issue convertible

bonds? An empirical analysis for Canadian market. Canadian Journal of Administrative

Sciences 25, pp. 214-236.

- Mayers, D. (1998) Why firms issue convertible bonds: the matching of financial and real investment options. Journal of Financial Economics 47 (1998) 83-102

-Masters, B., (2009) Companies Return to Convertible Bond Market. Financial Times, May 11.

- Mikkelson, W.H. and M.M. Partch (1986): “Valuation effects of security offerings and the issuance process” Journal of Financial Economics, 15 , pp. 30–60

- Myers, S. C. and N. S. Majluf (1984): “ Corporate financing and investment decisions when firms have information that investors do not have,” Journal of Financial Economics

- Rahim, A., A. Goodacre & C. Veld (2012): “Wealth effects of convertible-bond and warrant-bond offerings: a meta-analysis” The European Journal of Finance

-Rock,, (1986) Why new issues are underpriced, Journal of Financial Economics, 15 (1986), pp. 187–212

- Stein, C. (1992) Convertible bonds as backdoor equity financing. Journal of Financial Economics 32: 3-21.

- Strong, N., (1992). Modelling abnormal returns: a review article. Journal of Business

Finance and Accounting 19, pp. 533-553.

- Suchard, J.A., (2007). The impact of right issued of convertible debt in Australian markets. Journal of Multinational Financial Management 17, pp. 187-202.

(21)

20

Appendix

Table 9: Regression on cumulative average abnormal returns with robust errors

CAAR (-1,0) CAAR (-1,0) CAAR (-1,0) CAAR (-1,0)

1 2 3 4 indu util crisis ln(issue) ln(market) Tobins Q constant -.0585** (.0279) -.0150* (.0075) -.0015 (.0054) -.0581** (.0272) -.0168** (.0077) -.0252 (.0361) .0016 (.0062) -.05401 (.0316) -.0129 (.0122) -.0234 (.0375) -.0016 (.0035) .0007 (.0022) -.0041 (.0410) -.0526 (.0304) -.0023 (.0120) -.0239 (.0362) -.0015 (.0034) -.0003 (.0024) -.0188 (.0148) .0196 (.0452) R-squared Number of obs 0.1591 63 0.1780 63 0.1802 63 0.2120 63

Significance levels at 1%, 5% and 10% presented by ***, **, *, respectively Standard errors are between brackets

Table 10: Correlation matrix

CRISIS INDUSTRIAL UTILITY TOBIN’S Q MARKET VALUE ISSUE SIZE CRISIS 1.000 INDUSTRIAL 0.0540 1.000 UTILITY -0.1118 -0.3381 1.000 TOBIN’S Q -0.0735 -0.0143 0.3942 1.000 MARKET VALUE 0.0832 -0.3203 0.0406 -0.2347 1.000 ISSUE SIZE 0.2041 0.1886 0.3868 0.1562 0.0950 1.000

The influence of industry on the announcement effect of convertible bond issues for US listed firms