Why do Companies Issue Convertible Bond Loans

Document Sample
Why do Companies Issue Convertible Bond Loans Powered By Docstoc
					Canadian Journal of Administrative Sciences Revue canadienne des sciences de l’administration 25: 214–236 (2008) Published online 21 August 2008 in Wiley Interscience (www.interscience.wiley.com). DOI: 10.1002/CJAS.64

Why do Companies Issue Convertible Bond Loans? An Empirical Analysis for the Canadian Market
Igor Loncarski*
University of Ljubljana

Jenke ter Horst
Tilburg University

Chris Veld
University of Stirling and Simon Fraser University

Abstract To identify issuer motives, we study the determinants of announcement effects of convertible debt issues in the Canadian market. Classified into equity- and debt-like, wealth effects are significantly more negative for equitylike convertible bond issuers. Equity-like convertibles are significantly negatively affected by agency costs of equity. However, agency costs of debt have no significant effect on debt-like convertibles. Consistent with Stein (1992), this suggests convertibles in particular represent a substitute for equity. Moreover, convertible debt offers announced by income trusts experience significantly less negative wealth effects than offers by nontrusts—a finding explained by a more debt-like convertible design, very low agency costs of equity in case of income trusts, or both. Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd. Keywords: security issuance decision, convertible bonds, wealth effects, agency costs, income trusts JEL classifcations: G30, G32

Résumé Le présent article examine les intentions de l’émetteur par l’étude des déterminants des effets d’annonce d’émissions de dette convertible dans le marché canadien. Si les coûts de délégation des actions ordinaires influencent très négativement les effets de la richesse (classés en capital actions ordinaires et en capital assimilable à un titre d’emprunt), les coûts de délégation des dettes par contre n’ont qu’un effet limité sur les actions convertibles assimilables à des dettes. Comme l’a montré Stein (1992), les actions convertibles peuvent remplacer les actions ordinaires. Par ailleurs, les offres de dettes convertibles annoncées par les fiducies de revenus connaissent des effets de richesse nettement moins négatifs que celles annoncées par les compagnies traditionnelles. Ce résultat s’explique par une conception des obligations convertibles de type dette et/ou par les coûts très faibles de délégation des actions ordinaires dans le cas des fiducies de revenu. Copyright  2008 ASAC. Published by John Wiley & Sons, Ltd. Mots-clés : décision d’offre de valeurs mobilières, obligations convertibles, effet de richesse, coûts de délégation, fiducies de revenu

Chris Veld gratefully recognizes the financial support of the Social Sciences and Humanities Research Council of Canada. We thank Yakov Amihud, Enrique Arzac, Abe de Jong, Frans de Roon, John Doukas, Phil Molyneux, Luc Renneboog, Anil Shivdasani, Mola Simona, Yulia Veld-Merkoulova, and participants at the European Financial Management Association Conference and Doctoral Tutorial in Milan (June 2005) for helpful comments and suggestions. Special thanks go to two anonymous referees and to Jason Wei (the Associate Editor) for their very useful and insightful comments. In addition, the excellent research assistance of Yuriy Zabolotnyuk and editing assistance of Betty Chung is gratefully acknowledged. *Please address correspondence to: Igor Loncarski, University of Ljubljana, Faculty of Economics, Kardeljeva ploscad 17, 1000 Ljubljana, Slovenia. Email: igor.loncarski@ef.uni-lj.si

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

214

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Why companies issue convertible debt is receiving considerable attention in both the theoretical and empirical corporate finance literature. While practitioners put forward notions such as delayed equity, lower coupon rate and “sweetening” of deals that are otherwise hard to sell,1 academics have proposed theories that relate the use of convertible debt to informational asymmetries (Brennan & Kraus, 1987; Kim, 1990; Stein, 1992), agency issues (Green, 1984), and tax motives (Jalan & Barone-Adesi, 1995). These theories in general suggest that companies that face high debt- and/or equity-related agency costs could benefit from issuing convertible debt as opposed to other “straight” means of financing. Prime candidates for issuing convertible debt are companies for which straight debt or equity does not provide the most efficient way of financing. These include companies to which the following problems apply: difficulty in estimating risk, possession of ample growth opportunities, high costs of financial distress, financial constraints, and/ or high agency costs. A convertible bond, from now on to be referred to as a convertible, is a bond that can be exchanged for a predetermined fixed number of “new” shares of the issuing company within a predetermined period of time. In essence, a convertible is a package consisting of a straight bond and warrants written on the issuing company stock.2 Empirically, it is well documented that different security types induce different wealth effects at the time of their announcements. For example, seasoned equity offerings induce the strongest negative wealth effects (see Eckbo & Masulis, 1995) of between −2.5 and −4.5 percent, while straight debt issues induce only slightly (often insignificant) negative wealth effects (see Eckbo & Masulis). Given the hybrid character of convertibles, we expected the size of the wealth effects associated with the announcements of convertible security offerings to be between those for straight debt and equity. Previous studies on stock market reactions to the announcements of convertible debt issues in the US market have documented significant negative effects of convertible debt announcements in the range of −1 to −3 percent.3 Other studies such as Magennis, Watts, and Wright (1998) and Abhyankar and Dunning (1999) found significantly negative effects for the Australian and the UK markets respectively. Outside the US and the UK market empirical evidence has been somewhat less conclusive. Burlacu (2000), Dutordoir and Van de Gucht (2007), and Ammann, Fehr, and Seiz (2006) found similar effects for markets in France, Western Europe, and Germany and Switzerland respectively. However, results for other markets show the contrary. More specifically, Christensen, Faria, Kwok, and Bremer (1996) found positive effects for the Japanese market; Chang, Chen,

and Liu (2004) positive (insignificant) effects for the Taiwanese market; and De Roon and Veld (1998) significantly positive effects for the Dutch market. The hybrid nature of convertibles and the institutional and regulatory differences among countries and markets seem to be the driving force of the divergence. This makes the analysis an interesting field in empirical corporate finance since convertible debt can be structured to be either more debtor equity-like so as to mitigate some of the risks and deficiencies associated with each of “plain” securities. The first objective of our study was to provide further evidence on market reactions to convertible debt offerings by studying the Canadian market. We therefore examined the influence of several issuer characteristics on announcement reactions in the Canadian market for the period between 1991 and 2004. To the best of our knowledge, this is the first study to examine wealth effects associated with convertible debt issues in the Canadian market. The second objective was to examine the nature and determinants of the size of wealth effect with respect to issuer characteristics and to relate these findings to theories about motives for the use of convertible debt. Although the Canadian market shares many of its design features with the US, closer scrutiny reveals some differences, which make the analysis of the Canadian market particularly interesting. First of all, the Canadian market is much smaller than that of the US.4 This potentially implies that the issuers in the Canadian market differ from their US counterparts, thus having a different motivation for the use of convertible bond loans. Secondly, investment (income) trusts have become a very popular means of business organization in Canada after 2000 due to a very favourable tax treatment.5 The convertible issues by income trusts in Canada are different from those issued by “standard” corporations in the sense that income trusts’ convertibles are more debt-like, while standard corporations issue more equity-like convertibles. This empirical observation can be explained by the characteristics of income trusts and their obligation to distribute most of the free cash flow through very high dividend payments compared to standard corporations, in particular. Moreover, as argued by Jog and Wang (2004), income trusts reduce agency costs of equity and therefore provide an interesting setting to further explore motives for the issuance of convertibles. Our empirical findings are mostly in line with the seminal work of Myers and Majluf (1984) on external financing and the role of informational asymmetry. As in the US, the event study analysis showed that wealth effects associated with the announcements of Canadian convertible offerings yielded significantly negative abnormal returns of about −2.7%. The analysis showed

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

215

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

this is to be attributed to the more equity-like nature of most of the convertibles issued in the Canadian market in the period under consideration, in particular before 2000. After 2000 many issues are made by income trusts and are very much debt-like in nature. The result is robust using different convertible bond design measures. With respect to firm-specific determinants of announcement price reactions, we found that the abnormal returns are driven by factors related to equity-like features of convertible debt. Debt-related costs, as proxied by interest coverage and leverage, do not have a significant effect on the market valuation. On the other hand, equity-related costs, proxied by the stock price run-up (overvaluation issue) in particular, have a negative market valuation effect. Firms with higher dividend yields are mostly found to have higher cumulative average abnormal returns related to the announcement of the convertible offerings, as the dividend payout serves as a disciplining device that lowers equity-related agency costs. However, we found that this result is driven by convertible issues made by income trusts (debt-like issues), which typically pay high dividends. Our results appear to be robust across different specifications, that is when we control for the stated use of the proceeds (acquisitions, capital expenditures, or refinancing) and the design of convertibles (debtversus equity-like). These findings are consistent with theories that relate the use of convertible debt to mitigate different aspects of informational asymmetries, in particular those related to the agency costs of equity (Stein, 1992). We did not find support for tax arguments for the use of convertible debt. Shareholder Reactions to Convertible Debt Announcements Wealth Effects Associated with the Announcements of Convertible Debt Offerings A general explanation as to why investors react negatively to security offerings follows from the informational asymmetry between managers and the market with respect to value of assets in place and/or future growth opportunities. As such, security offerings are viewed as special examples of the “lemons problem” presented by Akerlof (1970). The models of Myers and Majluf (1984) and Miller and Rock (1985) can be viewed as specific applications of the lemons problem. According to these models, when a company issues risky securities, investors will demand a discount on the security price as compensation for potential overvaluation of the firm. Therefore, the announcements of convertible issues are predicted to have a negative impact on the issuer’s stock price.

Based on results of previous studies, abnormal returns may be driven by the type of financial system. Market-oriented systems, including those in the US, Canada, and the UK have well-developed financial markets and open corporations with widely dispersed share ownership. On the other hand, network-oriented systems, including those in Japan, Germany, Switzerland, and the Netherlands have strong banks with large share ownership and a greater role in monitoring. In market-based systems it is expected that managers are more likely to act in the interest of existing shareholders, and informational asymmetries may be larger. It follows from Myers and Majluf’s (1984) adverse selection model that in these systems market reaction to convertible debt issues may be less favourable. In network-oriented systems, where managers are more likely to be entrenched given their institutional settings, the Myers and Majluf model may not hold. A second explanation for negative stock returns at the announcement of convertible debt issues attributes these returns at least in part to systematic underpricing of public offerings. If public offerings are underpriced, then wealth is transferred from the firm’s current stockholders to the purchasers of the underpriced securities. Evidence of underpricing for convertibles at the issue date is reported by Loncarski, ter Horst, and Veld (2007) for the Canadian market and Kang and Lee (1996) for the US market. Moreover, Loncarski et al. showed that more equity-like convertibles are more underpriced at the issuance date than more debt-like convertibles. This implies that a more negative market response can be expected for announcements of the issuance of equitylike convertibles compared to debt-like convertibles. Given the adverse selection model of Myers and Majluf (1984) and the underpricing of convertibles, we tested the following hypothesis regarding the wealth effects associated with the announcements of convertible debt offerings.
H1: The market valuation effect is more negative for equity-like convertibles than for debt-like convertibles.

Determinants of the Size of Wealth Effects The effect of issuer characteristics on the size of the wealth effect associated with announcements of convertible debt offerings can, in general, be separated according to the dominating nature of the convertible issue (debt- versus equity-likeness) and related to the motives for issuing such security. Convertible debt is a particularly useful financing instrument in cases where

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

216

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

informational asymmetries and market imperfections make the use of straight debt or equity more costly or even impossible. Agency costs of debt. Brennan and Kraus (1987) have shown that convertible debt mitigates problems associated with the estimation of returns of assets in place. According to this explanation, convertible debt represents an alternative to straight debt, which is very costly or difficult to issue. Green (1984) also considered convertible debt as a resolution to the agency conflict between bondholders and shareholders where shareholders may be inclined to expropriate debt-holders by substituting less risky investment policies for riskier ones due to their limited liability in a standard debt contract. Since convertible debt can be turned into equity at the discretion of bondholders, it alleviates the risk shifting problem and can therefore be viewed as a substitute for straight debt. When treated as a substitute for straight debt, the information signalling model of Ross (1977) suggests that the issuance of debt securities conveys favourable information to the market. A manager of a successful firm may choose to increase leverage to positively signal to the market future performance of the firm;6 unsuccessful firms cannot mimic these signals because they have insufficient earnings to meet the debt payments. On the other hand, Myers (1977) has demonstrated the opposite–firms with a higher share of growth opportunities with respect to the current value of the firm issue less debt. In the spirit of Myers, increases in leverage can be interpreted as negatively signalling future growth opportunities of the company. From the debt perspective, the effect of convertible debt issuance on leverage is not obvious since it has both debt-like and equity-like components if the entire sample of convertible issues is analyzed. However, for more debt-like convertibles, the level of debt-related costs at the firm level should have a negative impact on the price response. This leads to Hypothesis 2:
H2: Agency costs of debt have a negative effect on the market valuation for the more debt-like convertibles and a nonnegative effect for the more equitylike convertibles.

H2a: Higher financial leverage negatively affects the market valuation, in particular for more debt-like convertibles. H2b: Interest coverage positively affects the market valuation.

Agency costs of equity. From the equity component perspective, Kim (1990) and Stein (1992) argued that convertibles are delayed equity and are used to signal the quality of the firm in the framework of informational asymmetry. This is consistent with the adverse selection model of Myers and Majluf (1984), where conventional equity issues are unattractive due to high issue costs and dilution. Kim demonstrated that the conversion ratio serves as a credible signal of a company’s future earnings. Stein argued that good quality firms issue debt while medium quality firms differentiate themselves from bad quality firms by issuing convertibles. If the nature of a convertible issue is more equity-like, the equity-related adverse selection costs should negatively affect the price reaction to convertible debt offerings. We therefore postulate the following hypothesis:
H3: Agency costs of equity have a negative effect on the market valuation for the more equity-like convertibles and a nonnegative effect for the more debt-like convertibles.

If the nature of a convertible issue is more equitylike, the equity-related adverse selection costs should negatively affect the price reaction to convertible debt offerings. Lucas and McDonald (1990) showed why equity issues on average are preceded by positive abnormal returns. However, in line with the pecking order theory of Myers and Majluf (1984), costs associated with issuing equity should be higher for companies with larger stock price run-ups, since they are more likely to be overvalued.
H3a: A period of positive abnormal returns preceding the announcement date negatively affects the market valuation.

Firms are expected to face high debt-related costs when their financial leverage is high and earnings are not sufficiently adequate to service the interest payments, as these factors increase risk of financial distress and bankruptcy threat. In terms of debt-related costs, where leverage and the Times-Interest-Earned (interest coverage) ratio are considered as proxy measures for the level of debt and the risk of financial distress, we tested the following two subhypotheses:

Another aspect of the issuer’s characteristics is related to the equity-like nature of convertibles – the free cash flow. Jensen (1986) examined the adverse effect of free cash flow on the value for shareholders and in particular, for low growth firms. He proposed debt as a better control or bonding device for managers than payout policy as a company’s future payouts can be changed, while debt has to be repaid. Nevertheless, it has been documented that reductions in dividends are associated with negative wealth effects for shareholders and that managers try to avoid negative changes in payout

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

217

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

policy. This is especially the case if compensation schemes are related to shareholder value creation. Therefore, payout policy has a disciplining function for managers to act in shareholders’ best interests. We therefore tested the following two subhypotheses with respect to the agency cost of free cash flow (agency cost of equity):
H3b: Higher free cash flow negatively affects the market valuation. H3c: Dividend payments positively affect the market valuation.

Tax motivation. Jalan and Barone-Adesi (1995) considered convertibles as delayed equity financing, and motivated their use with the different tax treatment of coupon interest and dividend payments in a setting with market frictions and incompleteness. In such a setting, issuing convertibles increases the residual equity value of the firm, since the firm benefits from the tax shield as opposed to up-front equity financing. The cooperative game, and the fact that firms have repeated need for the financial markets, assures that both firms and investors have an incentive to use convertibles and share their benefits. Compared to straight debt, convertibles offer much less trade-off between interest tax shields and cost of financial distress. In cases of straight bonds, higher interest tax shields are only achievable through higher indebtedness, which increases the probability of financial distress. On the other hand, convertibles offer the benefit of interest tax shields, although they give a smaller probability of financial distress.7 We expected a positive effect of the tax burden (marginal tax rate) on the size of abnormal returns, especially in cases of more equity-like convertibles, implying some evidence on the tax motive argument. Accordingly,
H4: Income taxes positively affect the market valuation, in particular for more equity-like convertibles.

Income trusts. Income trusts or publicly-traded flow-through entities (FTEs) are specially designed financing vehicles where the trust is positioned as an immediate full owner of a typically mature business. The cash flows from the ultimate operating company, which the trust owns, are usually fully distributed to the trust (typically in the form of interest payments) and then passed on to the unit holders (owners of the income trust) as dividends. Since the trust accrues no tax payments, investors then (depending on the tax status of their investment) either pay no or at least somewhat reduced taxes than they would have otherwise. The main benefit of the income trust is therefore tax driven. Aggarwal and Mintz

(2004) argued that creation of income trusts is in part a response to high corporate taxes and double taxation. Businesses therefore benefit from a lower cost of capital. Typically, income trusts were designed for businesses with stable earnings, such as investments in real estate portfolios (REITs) and natural resources (energy trusts). However, income trusts have become very popular in the Canadian market in the last few years. Jog and Wang (2004) reported that the number of income trust IPOs grew from 9 in 1998 to 64 and 36 in 2002 and 2003 respectively, with the highest increase in the number of business trusts. Moreover, while the market capitalization of trusts was less than 20 billion CAD in 2000, it grew to 120 billion CAD by 2004 with approximately half of the capitalization attributed to business trusts. Aggarwal and Mintz claimed that income trust financing is distorting capital markets and favours slower-growth companies since businesses with lower economic performance (typically those with a steady income stream) seem to benefit the most. On the other hand Cleary and MacKinnon (2007) have shown that in terms of riskadjusted returns, income trusts have outperformed both equities and bonds. Jog and Wang (2004) argued that income trusts reduce agency costs of equity because they distribute most of the free cash flow through very high dividend payments as compared to standard companies. This claim has an important implication for motives of issuance of convertible debt by income trusts. Income trust convertible bond issues should not be “plagued” with agency costs of equity. Therefore, announcements of convertible bond issues placed by income trusts should induce a less negative market valuation. However, this potential result can also stem from the design of a convertible bond (debt-like convertible bond issues are less susceptible to agency costs of equity) or both the relative unimportance of agency costs of equity and the design. In relation to income trusts we therefore investigated two issues: 1. Whether convertible bond issues placed by income trusts induce a market valuation that is different from the convertibles placed by standard companies. 2. Whether the reason behind this potential difference is low (or even nonexistent) agency costs of equity or convertible debt design. In 2007, the Canadian government changed the taxation on distributions from income trusts. Existing income trusts were given a four year transition period, while the distributions from new income trusts (and also those that announced the conversion at the time) would be taxed in the same way as dividend distributions from corporations. This implies that nothing changes in terms of taxation for taxable Canadian residents, while tax-exempt

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

218

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Canadian residents and taxable foreign investors pay higher taxes on distributions from income trusts. Other observations. From the reasoning so far it follows that price reactions to convertible debt announcements should be negatively influenced by both debt- and equity-related agency costs since convertible debt encompasses both debt-like and equity-like components. We considered three additional factors that influence both debt- and equity-related costs. First, both debt-related costs (e.g., risk uncertainty and financial distress costs) and equity-related adverse selection costs should be lower for larger companies. Larger firms tend to be more familiar to the market, lowering their respective issuing costs because less information-search and -processing costs are required. On the other hand, the size of the company increases the complexity and analysis, so that the larger company might actually be more opaque. Size, therefore, does not necessarily translate into a smaller adverse selection problem. We therefore used the size of the firm as a control variable, but did not have any a priori expectation about the direction of the effect. The size of the company captures complex interactions between different issuer characteristics. Secondly, De Jong and Veld (2001) argued that the problem of perceived overvaluation is worse for firms with sufficient slack in the form of liquid assets. The reason for this is that slack provides an alternative source for financing of new projects and thus enhances the potential agency problem (overinvestment) between managers and shareholders. This negative impact should be more pronounced for equity-like convertibles and is not likely to be detected in the overall sample of convertibles since its role should be less strong for more debtlike convertibles. However, there is also the opposite potential impact of slack. It can be viewed as a build-up of internally generated and needed funds for increased capital expenditures, when the external sources of financing are very costly (i.e., along the lines of the peckingorder theory of capital structure). In particular, this is the case for companies with higher risk and larger growth opportunities (more equity-like issuers). We therefore included slack in our cross-sectional analysis without hypothesizing its overall effect on the valuation as it has not only a negative effect of increased agency cost of equity, but also a positive effect of internal (less expensive) build-up of funds. In addition, slack can also be viewed as collateral, in which case it should have a positive effect on valuation in case of debt-like convertibles, where it mitigates agency costs of debt. Thirdly, a firm with good growth opportunities should face reduced debt- and equity-related agency costs. De Jong and Veld (2001) have argued that expecta-

tions in the market regarding the profitability of the firm’s projects reduce the potential for both the asset substitution problem and adverse selection problem described earlier.

Method Sample The sample consists of convertibles issued between January 1991 and December 2004 by Canadian companies that were listed on the Toronto Stock Exchange (TSE). During that period, there were 207 convertible bond issues in total. We excluded issues made by financial companies (SIC division H – Finance, Insurance and Real Estate), and were left with 149 issues by nonfinancial companies. Data on announcement dates and other features of the convertible bond issues were obtained from the SDC database and checked against press releases in Lexis-Nexis, Canadian newswires, company web sites, and the SEDAR8 database. For 26 issues in our final sample, we found discrepancies in the announcement dates. In those cases we used the earliest announcement date that we could find. The criteria for an issue to be included in our sample were: (a) The announcement date had to be verifiable through a source other than SDC; (b) The issuing firm’s stock price data had to be available in DataStream; (c) The issuing firm’s accounting data had to be available in DataStream; (d) The announcement did not confound other corporate announcements; (e) The conversion option related to the equity of the issuing company; and (f) The issues of the same issuer had to be at least 120 trading days apart in order for the estimation and event periods for different issuers not to overlap. Given the criteria, the initial 149 issues by nonfinancial companies first shrunk to 129 due to stock price data availability, and further down to 107 issues due to accounting data availability. Of those 107 issues, we could not verify the announcement date for 10; 4 were convertible in the equity of another company; 3 were too close together with the previous issues of the same issuer causing an overlap of event periods; and 4 were joined together with the convertible bond issues (by the same issuer) announced on the same or the previous day. Accordingly, our final sample consisted of 86 bond issues offered by 77 different companies.

Event Study Approach We used the standard event study method to estimate the wealth effects associated with the

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

219

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

announcement of convertible debt offerings. To estimate the parameters of the market model, we used the Standard & Poor’s TSX (Toronto Stock Exchange) valueweighted total return index to compute the market return. This is widely considered as the benchmark for Canadian equities as it accounts for more than 200 stocks listed on the TSX or about 70% of the total market capitalization. The estimation period ranged from day −120 to day −20 relative to the announcement date of the offering. As there was no significant clustering of the announcement dates of the offerings, residuals in the market model regressions for individual issuers were not likely to be correlated.

Measurement of the Convertible Debt Design The most difficult variable to proxy is the measurement of the equity component of convertible debt. As previously shown and used in the literature (e.g., Burlacu, 2000; Lewis, Rogalski, & Seward, 1999), different approaches can determine the size of the equity component embedded in a convertible bond design. Following Lewis et al. (1999), Burlacu (2000), and Dutordoir and Van de Gucht (2004, 2007), we estimated the structure of the convertible debt design (i.e., how debt- or equitylike it is) by employing the delta measure. The delta was derived from the option pricing model of Black and Scholes, and adjusted for continuous dividend payments in the way suggested by Merton (e.g., Hull, 2006, page 317–318):

Proxies The variables used in the analysis are related to the hypotheses described earlier. Leverage (LEV) was measured as the ratio between total debt and total assets. The Times-Interest-Earned ratio (TIE) is a measure often employed in practice, in particular in restrictive covenants that govern typical debt contracts. It is defined as EBIT (Earnings Before Income and Taxes) over interest expense. SLACK was measured as the ratio of cash and equivalents over total assets.. Free cash flow (FCFA) was measured as the ratio of free cash flow (net income plus depreciation minus capital expenditures) over total assets. The dividend yield (DY) was measured as the average dividend yield between (−15, −5) days relative to the announcement date. The tax burden (TAXA) was measured as a ratio of income taxes over total assets. Growth opportunities were measured using Tobin’s Q (Q). The nominator of this ratio was computed as the sum of market value of equity (measured as the average between (−15, −5) days relative to the announcement date), long term, and short term debt. The denominator was calculated as the book value of total assets. We added a control variable for firm size, which is the natural logarithm of firm size (LNTA). We defined the size of the issue proceeds relatively to the book value of equity (RISSEQ). The variable captured the dilution effect, but it also proxied for the agency cost of equity. Higher relative proceeds indicated higher agency costs of equity. Based on the stated use of proceeds, we defined three dummy variables in the analysis: M&ADUM was given a value of 1 if the issuer intended to use the proceeds for acquisitions and 0 otherwise; REFDUM was assigned a value of 1 if the issuer intended to use the proceeds for refinancing and 0 otherwise; and CGXDUM is a dummy variable with a value of 1 if the issuer intended to use the proceeds for capital and general expenditures and 0 otherwise.

σ2    S  ln +  r − δ +  ⋅ T   K  2 ∆ = e−δ T ⋅ N  , (1) σ⋅ T       where S is the current price of the underlying stock, K is the conversion price, d is the continuously compounded dividend yield, r is the continuously compounded yield on a selected “risk-free” bond, s is the annualized stock return volatility, T is the maturity of the bond and N(.) is the cumulative standard normal probability distribution. The delta measure always takes a value between 0 and 1. Values closer to 1 indicate a high sensitivity of the convertible bond value to the underlying equity (stock) value, implying a high probability of conversion. As a proxy for the risk-free rate we used the yield of a Canadian government benchmark bond of the closest matching maturity rounded upwards. For the stock price volatility measure we used the annualized volatility of stock returns as estimated over the period (−120, −20) relative to the announcement date of the offering. In order to differentiate between equity- and debt-like convertibles we used a delta cut-off value of 0.5. The subsample is denoted with a delta smaller than 0.5 as more debt-like, while the subsample with a delta greater than (or equal to) 0.5 is indicated as the more equity-like sample. For comparison, Burlacu denoted convertibles with delta values below 0.33 as debt-like and those with delta values above 0.66 as equity-like. Lewis, Rogalski, and Seward (2003) used delta cut-off values of 0.4 and 0.6 for classifying bonds as either debt or equity-like. The delta measure only captures the equity component of a convertible. It does not take into account the credit quality and other important characteristics of convertibles such as callability, putability, and early conversion. Similar to other studies, the delta measure acts as our main measure of the convertible debt design. We

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

220

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

used two alternative measures for the size of the equity component relative to the size of the debt component in convertible debt for robustness tests. The first is the ratio of equity to straight debt component value of the convertible bond (ED measure); this was estimated using the convertible debt valuation approach proposed by Tsiveriotis and Fernandes (1998). This ratio accounts for different credit qualities among issuers and other important features of convertibles.9 We estimated the model price of the convertible bond at the issue where the price is the sum of equity (value of the conversion right) and straight debt component. We used values of ED greater than 1 as the reference for more equity-like convertibles, and values of ED lower than 1 as the reference for the more debt-like convertibles. The second alternative measure is the ratio of conversion value to the par value of the convertible bond (M measure)—“moneyness.” This is defined as the conversion value at the announcement of the bond issue (CR·S0, where CR represents the conversion ratio, and S0 represents the stock price at the announcement of the bond issue) over the par value of the convertible bond. Note that this simple measure ignores a number of factors that affect the convertible debt design such as maturity, volatility, and dividend yield. For classification purposes, we treated issues with the value of M > 1 as equity-like and those with the value M < 1 as debt-like.

Results Distribution of the Sample and the Convertible Bond Design Table 1 presents the distribution of the issues over the years and across organizational structure and convertible debt design. The data in Panel A of the table shows that around 48% of the issues in our sample were offered from the beginning of 2001. Offerings exhibited some bunching, with hot periods being 1993–1994 and the end of the 1990s onwards. We classified convertible debt into more equity and more debt-like using three different measures. Although delta does not take into account credit quality and other important characteristics of convertibles such as callability, putability, and early conversion, it has been extensively used in previous studies. In order to be able to compare the design of convertible debt in the Canadian market with other markets (other studies) we also used the delta measure. In contrast to studies for other markets (Dutordoir & Van de Gucht, 2004, for example, found convertible bonds in the Western European markets to be

more debt-like than in the US market; Lewis et al. 1999 on the other hand reported an almost normal distribution of the delta for the US market) we found a bimodal distribution of the delta of the convertible bond issues. We found that prior to 2000, convertible debt issues in the Canadian market were mostly equity-like (value of the delta > 0.5). From 2001 more of the convertible bond issues became debt-like with very low values of the delta. This change coincides with and seems to be driven by the increasing popularity of conversion of businesses into income trusts. Compared to the findings based on the delta measure, we found that the ED and M measures were more normally distributed with asymmetry towards high values (more equity-like convertible debt design). This confirms the result that most of the issues were more equity-like. More importantly, classification of issues into more debt and more equity-like base on ED and M measures in Panel A of Table 1 confirms the results based on the delta measure–prior to 2000 most of the convertible bond issues were equity-like (ED < 1 and M < 1), while after 2000 more issues became debt-like.10 The numbers in Panel A of Table 1 also show that convertible debt offerings made by income trusts were virtually nonexistent up to 2002, which also coincides with the increased proportion of debt-like convertible bond issues compared to equity-like issues—as measured by the delta or any of the two alternative measures of the convertible debt design. This is further demonstrated by the data in Panel B of Table 1, where most issues made by income trusts were debt-like and the vast majority of issues made by other corporations were equity-like.

Wealth Effects Associated with the Announcement Dates of Convertible Debt Offerings Table 2 presents the Cumulative Average Abnormal Returns (CAAR) as well as tests for Hypothesis 1 regarding the wealth effects associated with the announcements of the convertible debt offerings. In Panel A of the table the results for the total sample and the subsamples split according to the delta measure of the convertible design are presented. From the results based on the total sample, we observed that CAARs were significantly negative over different event windows. In particular, the effect for the event window (−1, 1) was significantly negative at −2.7%. These results are in line with the results from previous studies, in particular those for the US market. Panel A of Table 2 further reports the CAAR for the subsamples with a value of the delta measure above 0.5 and below 0.5 respectively. The first interesting result is the comparison of wealth effects for

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

221

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Table 1 Breakdown of Convertible Debt Issues According to the Year of the Issue, Organizational Structure, and the Convertible Debt Design Panel A: Distribution of the Issuers According to the Year of the Issue and the Convertible Debt Design
Year Number of issues 10 13 12 6 3 7 4 4 5 2 8 5 3 4 86 Percent Number of income trusts 10 4 4 0 0 0 2 0 0 0 0 0 0 0 20 ∆ < 0.5 ∆ > 0.5 ED < 1 ED > 1 M<1 M>1

2004 2003 2002 2001 2000 1999 1998 1997 1996 1995 1994 1993 1992 1991 Total

11.6 15.1 14.0 7.0 3.5 8.1 4.7 4.7 5.8 2.3 9.3 5.8 3.5 4.7 100.0

9 5 4 0 0 1 1 0 0 0 1 0 2 1 24

1 8 8 6 3 6 3 4 5 2 7 5 1 3 62

8 7 7 2 1 4 0 3 0 0 0 0 0 0 32

2 6 6 4 2 3 4 1 5 1 6 2 2 3 47

10 10 9 3 2 4 3 3 3 0 2 0 1 1 51

0 3 3 3 1 3 1 1 2 2 6 5 2 3 35

Panel B: Distribution of the Issuers According to DELTA and Organizational Structure
∆ < 0.5 Standard corporation Income trust Total 6 18 24 ∆ > 0.5 60 2 62 Total 66 20 86

Notes: Distribution of nonfinancial Canadian companies that announced a convertible bond loan in the period from January 1991 to December 2004 by announcement year. The announcements are identified from the SDC database. Announcements were eliminated for the following reasons (1) no stock and/or accounting data available; (2) nonverifiable announcement dates; (3) bonds convertible in the equity of another company; (4) issuance dates overlap or are very close to announcement dates of other securities. ∆ is the measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. It is calculated using the option pricing model of Black and Scholes corrected for continuous dividend payments (see Equation 1). ED is the equity-to-debt component ratio based on the Tsiveriotis and Fernandes convertible bond valuation model. M is the “moneyness” measure defined as a ratio between conversion value at the announcement of the convertible bond issue and the par value of the bond.

the subsamples in the event window (−10, −2), where the CAAR of 2.24% was significantly positive for more equity-like convertibles (∆ > 0.5) and significantly negative (−0.77%) for the more debt-like convertibles (∆ < 0.5). The difference between the two values was also significant, which implies that prior to the announcement of the issue, more equity-like issuers experience a significant stock price run-up. This suggests that issuers try to time their announcements after periods of favourable stock price movements. It also suggests that the market is more likely to perceive the more equity-like issuers as overvalued at the announcement dates of the convertible debt offerings in our sample given the prior streak of positive abnormal returns. Therefore, they react more

negatively to the announcement. The most negative CAAR for the more equity-like issuers was in the event window (0, 20) with a significantly negative −6.32%, while more debt-like issuers did not experience significant wealth effects during that period. Based on the results in Panel A of Table 2 we can conclude that wealth effects associated with the announcements of convertible bond offerings are significantly more negative for the more equity-like convertible issues than for the more debt-like issues.11 This confirms Hypothesis 1—the market valuation effect is more negative for equity-like convertibles than for debt-like convertibles. Next to investigating the wealth effects associated with the announcements of convertible debt offerings

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

222

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Table 2 Cumulative Average Abnormal Returns (CAAR) for Different Event Windows Panel A: Subsamples Split Based on the Convertible Debt Design
CAAR (T1, T2) Total sample n = 86 CAAR −10, −2 −5, −2 −2, 0 −1, 0 −1, 1 −1, 2 −1, 5 0 0, 1 0, 2 0, 5 0, 20 1.43% 0.68% −0.60% −0.54% −2.70% −2.87% −2.87% −1.35% −3.52% −3.69% −3.68% −4.62% J1 stat. 3.92 *** 2.00 ** −1.25 −1.45 * −6.53 *** −7.48 *** −8.60 *** −3.74 *** −11.58 *** −12.21 *** −12.91 *** −17.17 *** CAAR 2.24% 0.99% −1.15% −0.86% −3.67% −4.00% −4.00% −1.98% −4.79% −5.12% −5.12% −6.32% ∆ < 0.5 n = 62 J1 stat. 4.62 *** 2.20 ** −1.80 ** −1.75 ** −6.63 *** −7.80 *** −8.99 *** −4.37 *** −11.84 *** −12.82 *** −13.52 *** −17.75 *** CAAR −0.77% −0.16% 0.89% 0.33% −0.10% 0.15% 0.17% 0.35% −0.08% 0.17% 0.19% −0.04% ∆ > 0.5 n = 24 J1 stat. −2.42 *** −0.47 2.16 ** 0.93 −0.29 0.43 0.58 0.67 −0.29 0.56 0.74 −0.14 3.01% ** 1.15% −2.04% ** −1.19% −3.57% *** −4.14% *** −4.17% *** −2.33% *** −4.72% *** −5.28% *** −5.32% *** −6.29% *** Difference in means

Panel B: Subsamples Split Based on the Organizational Design of the Issuer
CAAR (T1, T2) Total sample n = 86 CAAR −10, −2 −5, −2 −2, 0 −1, 0 −1, 1 −1, 2 −1, 5 0 0, 1 0, 2 0, 5 0, 20 1.43% 0.68% −0.60% −0.54% −2.70% −2.87% −2.87% −1.35% −3.52% −3.69% −3.68% −4.62% J1 stat. 3.92 *** 2.00 ** −1.25 −1.45 * −6.53 *** −7.48 *** −8.60 *** −3.74 *** −11.58 *** −12.21 *** −12.91 *** −17.17 *** CAAR 4.50% 1.90% −1.82% −1.70% −5.36% −5.47% −5.50% −1.88% −5.55% −5.65% −5.68% −5.92% ITRUST = 0 n = 66 J1 stat. 8.41 *** 3.32 *** −3.18 *** −3.67 *** −7.96 *** −8.63 *** −9.63 *** −4.13 *** −9.16 *** −9.28 *** −10.07 *** −12.27 *** CAAR −0.52% 0.26% 0.73% 0.11% −0.25% 0.06% −0.10% 0.09% −0.27% 0.04% −0.11% −0.51% ITRUST = 1 n = 20 J1 stat. −1.46 * 0.81 1.64 ** 0.27 −0.68 0.17 −0.32 0.20 −0.86 0.14 −0.43 −1.88 ** 5.01% 1.64% −2.54% * −1.81% −5.12% *** −5.52% *** −5.40% *** −1.97% *** −5.28% *** −5.69% *** −5.57% *** −5.41% *** Difference in means

Notes: Cumulative average abnormal returns for the sample of 86 convertible bond announcements by Canadian companies from January 1991 to December 2004. The convertible bond announcements are identified from the SDC database. Abnormal returns are based on the market model, estimated over a 100-day period for each company (from day −120 to day −20). ∆ is the measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. It is calculated using the option pricing model of Black and Scholes corrected for continuous dividend payments (see Equation 1). ITRUST is a dummy variable that takes a value of 1 if the company is organized as an income trust and a value of 0 otherwise. Under the null hypothesis CAAR equals 0. CA A R (T1 ,T2 ) J1 = 1 [V A R [CA A R (T1,T2 )]]2 * p < .10,** p < .05,*** p < .01.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

223

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Figure 1. Cumulative average abnormal returns (CAAR)
Subsamples Split According to the Convertible Debt Design

Subsamples Split According to the Organizational Design of the Issuer

Notes: Cumulative average abnormal returns (CAAR) for the total and two subsamples split according to the value of the ∆ (DELTA) and the organizational design over the event window (−20, 20).

announcements of the convertible debt issues by standard corporate issuers induce significantly more negative wealth effects than issues announced by income trusts. Figure 1 shows the evolution of the CAARs over the event window (−20, 20) for the total sample as well as for the subsamples split according to the convertible debt design (as measured by the delta) and organizational design of the issuer (income trusts versus “standard” corporations). A striking result is that the wealth effect continued to grow negatively after the announcement date. For the total sample, we found a CAAR of −1.35% at the announcement of the issue, while over the event window (−1, 2) the CAAR dropped to −2.87% and continued to fall to −4.62% over the event window (0, 20). From the analysis of the two subsamples it appears that the more debt-like convertible issues (∆ < 0.5) experienced negative abnormal returns somewhat prior to the announcement, that is −0.77% over the event window (−10, −2), and this rebounded after the announcement of the offering to around 0. Conversely, the more equity-like convertible issues (∆ > 0.5) exhibited a significantly positive abnormal return reaction prior to the issue announcement (2.24% in the event window −10, −2), but this became significantly negative after the announcement by decreasing to around −4% over the event window (−1, 2) and even further to −6.32% over the event window (0, 20). Loncarski et al. (2007) analyzed convertible arbitrage in the Canadian market and showed that short positions in stock of equity-like convertible bond issuers increase significantly more immediately following the announcement of the issue than short positions in stock of debt-like convertible bond issuers. They demonstrated that the increase in the short positions is negatively correlated with the abnormal returns between the announcement and the issue date of the bond.

Issue and Issuer Characteristics with respect to the convertible design, we looked at the differences in market valuations when issuers were income trusts (in comparison to the issues placed by standard corporations). Based on the results presented in Panel B of Table 2, we found that the wealth effect in the event window (−1, 1) for the subsample of income trusts was −0.25%, but the result was not significantly different from 0. On the other hand, standard corporate issuers experienced a significantly negative wealth effect of −5.36% in the same event window. The difference of 5.12% points between the two was statistically significant (p < .01). We therefore can conclude that the In order to explore the characteristics of the issues and the issuing companies, we examined some sample statistics for several subsamples based on different measures of the convertible debt design and the issuer’s business structure (income trusts versus standard corporations). Panel A of Table 3 shows the differences between debt- and equity-like convertibles based on the delta measure. From the differences it appears that the more debt-like convertible issues had significantly lower conversion premiums (ratio between conversion price and stock price at the announcement date of the issue) and

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

224

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Table 3 Sample Statistics for Issue and Issuer Characteristics Panel A: Delta Subsamples Split
Statistics Variable mean min max median CV n mean min max ∆ > 0.5 0.610 2.244 −8.184 1.790 3.993 0.371 0.405 0.067 2.554 0.613 0.892 0.313 0.290 0.818 3.401 0.449 21 0.254 0.000 0.697 20 3.654 −35.920 101.798 24 0.022 −0.548 0.311 21 0.085 0.000 0.390 21 −0.069 −0.540 0.149 24 0.006 0.000 0.045 21 1.385 0.250 6.454 21 13.779 10.583 17.957 21 0.011 0.000 0.073 24 9.896 2.003 30.041 21 0.484 0.049 3.326 24 1.233 0.484 3.134 24 0.484 0.188 1.466 24 0.782 0.509 0.991 21 4.455 0.574 83.846 23 1.011 0.548 1.949 0.178 1.173 0.022 0.066 −0.017 0.000 0.992 13.803 0.002 7.134 0.258 1.188 0.418 0.811 1.468 0.960 0.921 5.500 5.388 0.989 −2.071 1.723 0.933 0.118 1.499 0.653 1.281 0.281 0.512 0.165 2.969 0.277 59 58 62 60 57 62 59 60 58 62 60 62 62 62 54 57 median CV n difference in means 0.035 −0.798 0.028 * 0.061 *** −0.085 *** −0.116 *** 0.081 0.538 ** 0.005 * 3.541 *** 0.090 0.143 ** 0.279 *** 0.622 *** 1.544 −0.030

∆ < 0.5

LEV 0.218 0.000 0.490 0.199 TIE 4.452 −2.223 41.621 2.004 SPRUN −0.006 −0.104 0.116 −0.005 SLACK 0.024 0.000 0.165 0.000 FCFA 0.016 −0.143 0.121 0.023 DY 0.122 0.038 0.205 0.120 Q 1.304 0.506 3.049 1.168 LNTA 13.240 11.435 15.475 13.333 TAXA 0.006 0.000 0.064 0.000 MATURITY 6.355 0.161 20.989 5.416 RISSEQ 0.394 0.089 1.250 0.232 CPREM 1.090 0.316 2.509 1.078 VOLAT 0.205 0.119 0.350 0.192 DELTA 0.159 0.008 0.495 0.143 ED 2.912 0.000 45.970 0.602 M 1.041 0.867 3.176 0.937

Panel B: ED Subsamples Split
Statistics Variable LEV TIE SPRUN SLACK FCFA DY Q LNTA TAXA MATURITY RISSEQ CPREM VOLAT DELTA ED M 0.264 3.091 0.023 0.044 −0.032 0.081 1.242 13.480 0.005 5.464 0.297 1.180 0.348 0.371 0.597 0.880 mean min max median CV n mean min max ED > 1 0.911 2.876 3.961 1.763 −3.973 0.832 0.501 0.072 2.527 0.465 0.909 0.181 0.649 0.805 0.487 0.131 29 0.213 28 6.073 32 0.014 30 0.083 29 −0.058 30 0.014 29 1.486 30 13.748 29 0.013 32 10.076 28 0.507 32 1.168 32 0.459 30 0.755 32 6.095 32 1.097 0.000 −35.920 −0.548 0.000 −0.540 0.000 0.250 10.583 0.000 2.751 0.049 0.316 0.169 0.033 1.007 0.558 0.697 0.171 101.798 2.510 0.311 0.022 0.384 0.066 0.104 −0.002 0.152 0.000 6.454 1.127 17.957 13.526 0.056 0.007 30.041 7.214 3.326 0.309 3.134 1.169 1.466 0.406 0.991 0.811 83.846 1.901 3.176 0.976 0.861 3.689 8.747 0.980 −2.422 2.365 0.928 0.128 1.273 0.620 1.240 0.323 0.561 0.261 2.506 0.377 44 42 47 44 41 45 44 44 43 47 45 47 47 45 47 47 median CV n difference in means −0.051 2.982 −0.009 0.039 ** −0.026 −0.067 *** 0.244 0.268 0.008 ** 4.612 *** 0.210 ** −0.012 0.110 ** 0.385 *** 5.498 *** 0.218 ***

ED < 1 0.000 0.551 0.193 −3.519 41.621 0.168 −0.104 0.304 −0.002 0.000 0.390 0.018 −0.456 0.121 0.002 0.000 0.205 0.086 0.329 3.049 1.142 11.435 15.292 13.550 0.000 0.064 0.000 0.115 10.025 5.256 0.060 1.250 0.203 0.981 1.901 1.103 0.119 1.064 0.266 0.008 0.882 0.268 0.000 0.982 0.657 0.548 1.043 0.916

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

225

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Table 3 Panel C: “Moneyness” Subsamples Split
Statistics Variable LEV 0.231 TIE 4.477 SPRUN 0.026 SLACK 0.053 FCFA −0.060 DY 0.053 Q 1.300 LNTA 13.459 TAXA 0.008 MATURITY 7.665 RISSEQ 0.316 CPREM 1.238 VOLAT 0.359 DELTA 0.512 ED 1.133 M 0.858 mean min max M<1 0.000 0.606 0.182 −35.920 101.798 1.632 −0.121 0.311 0.014 0.000 0.390 0.022 −0.540 0.121 0.000 0.000 0.182 0.019 0.329 3.411 1.163 10.910 16.413 13.406 0.000 0.064 0.000 0.162 25.003 5.499 −0.246 1.250 0.213 0.976 3.134 1.147 0.119 1.064 0.299 0.008 0.991 0.622 0.000 4.436 0.963 0.548 0.991 0.881 0.901 4.152 3.363 1.356 −2.439 1.197 0.581 0.094 1.926 0.606 0.859 0.263 0.592 0.631 0.803 0.125 47 47 51 48 46 51 47 48 46 51 48 51 51 51 50 51 0.263 0.000 2.921 −35.078 −0.003 −0.548 0.092 0.000 −0.027 −0.417 0.016 0.000 1.455 0.250 13.901 10.583 0.012 0.000 10.719 0.252 0.649 0.049 1.128 0.316 0.475 0.129 0.747 0.139 9.805 0.000 1.303 1.008 median CV n mean min max M>1 0.697 69.054 0.273 0.384 0.149 0.205 6.454 17.957 0.073 30.041 3.326 2.509 1.466 0.980 83.846 3.176 0.198 1.441 0.007 0.068 0.003 0.000 0.992 13.735 0.003 9.970 0.297 1.123 0.406 0.819 2.881 1.171 0.829 5.926 −47.675 0.933 −4.077 2.382 1.058 0.125 1.440 0.687 1.212 0.333 0.586 0.301 2.080 0.324 33 31 35 33 32 35 33 33 33 35 33 35 35 35 25 29 median CV n difference in means 0.033 −1.556 −0.029 0.039 ** 0.033 −0.037 *** 0.155 *** 0.443 0.004 3.054 ** 0.332 * −0.110 *** 0.116 ** 0.235 *** 8.672 *** 0.445 ***

Panel D: Income Trust Subsamples Split
Statistics Variable mean min max median CV n mean min max median CV n difference in means −0.067 ** −2.445 −0.016 −0.071 *** 0.058 *** 0.113 *** 0.113 −0.772 *** −0.007 ** −4.462 *** −0.160 * −0.165 *** −0.242 *** −0.581 *** −3.839 ** −0.047 *

ITRUST = 0 0.748 4.153 6.490 0.963 −2.411 1.917 0.915 0.116 1.407 0.681 1.215 0.306 0.546 0.238 2.939 0.263 66 0.183 0.000 64 1.997 −35.920 66 0.001 −0.092 66 0.017 0.000 64 0.004 −0.144 63 0.123 0.000 66 1.433 0.590 66 12.983 11.435 65 0.005 0.000 66 5.303 0.252 66 0.346 0.089 63 1.082 0.992 66 0.219 0.119 60 0.168 0.008 53 0.703 0.000 57 0.934 0.833

ITRUST = 1 0.490 0.182 41.621 0.168 0.116 −0.003 0.132 0.000 0.121 0.002 0.205 0.129 3.194 1.269 13.945 12.966 0.064 0.000 7.216 5.277 1.250 0.240 1.235 1.088 0.426 0.199 0.758 0.121 2.208 0.602 1.060 0.938 0.686 7.091 41.324 2.018 19.013 0.461 0.465 0.055 3.154 0.259 0.851 0.051 0.360 1.266 0.800 0.061 20 18 20 20 19 19 20 20 20 20 20 20 20 19 19 20

LEV 0.249 0.000 0.697 0.201 TIE 4.442 −35.078 101.798 1.713 SPRUN 0.018 −0.548 0.311 0.019 SLACK 0.088 0.000 0.390 0.069 FCFA −0.054 −0.540 0.149 −0.004 DY 0.010 0.000 0.106 0.000 Q 1.320 0.250 6.454 1.000 LNTA 13.754 10.583 17.957 13.694 TAXA 0.012 0.000 0.073 0.003 MATURITY 9.765 0.115 30.041 7.133 RISSEQ 0.506 0.049 3.326 0.253 CPREM 1.247 0.484 3.134 1.183 VOLAT 0.461 0.129 1.466 0.410 DELTA 0.749 0.239 0.991 0.788 ED 4.542 0.000 83.846 1.501 M 0.981 0.526 1.786 1.000

Notes: Sample statistics for subsamples of income and nonincome trust issuers and debt- and equity-like convertible bond announcements by Canadian companies from January 1991 to December 2004. The convertible bond announcements are identified from the SDC database. The subsamples are divided according to the ∆ of the conversion rights, equity-to-debt component ratio (ED), simple moneyness of the conversion option (M) and whether the company is organized as an income trust (ITRUST = 1) or not (ITRUST = 0). ∆ (DELTA) is the measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. It is calculated using the option pricing model of Black and Scholes corrected for continuous dividend payments (see Equation 1). ED is the equity-to-debt component ratio based on the Tsiveriotis and Fernandes convertible bond valuation model. M is the “moneyness” measure defined as a ratio between conversion value at the announcement of the convertible bond issue and the par value of the bond. LEV is computed as the ratio between total debt and total assets. TIE is the Times-Interest-Earned ratio. This is defined as EBIT (Earnings Before Income and Taxes) over interest expense. SPRUN is the cumulative average abnormal stock return measured over the window (−10,−2) relative to the announcement date. SLACK is the ratio of cash and equivalents over total assets. FCFA is the ratio of free cash flow (net income + depreciation – capital expenditures) over total assets. DY is the dividend yield. Q is Tobin’s Q-ratio measured as the sum of the market value of equity measured as the average between (−15,−5) days relative to the announcement date and the book value of long and short term debt over the book value of total assets. LNTA is the natural logarithm of total assets. TAXA is the ratio of income taxes over total assets. MATURITY is the maturity of the convertible bond defined as the difference in time between the maturity date of the bond and the issue date. RISSEQ is the ratio of the issue proceeds over the book value of equity. CPREM is a conversion premium, defined as the ratio between conversion price and the stock price. VOLAT is the annualized stock returns volatility, measured during the period (−120, −20) relative to the announcement date of the issue. * p < .10, ** − p < .05, *** p < .01.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

226

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

shorter maturities than more equity-like convertibles— that is a conversion premium of 1.090 versus 1.233 and a maturity of 6.4 years versus 9.9 years. A significantly lower conversion premium for the more debt-like convertibles was surprising. Typically, the conversion premium for the more debt-like convertibles should be higher than for the equity-like convertible since the probability of conversion should be lower. This was correctly reflected in a significantly lower maturity and also in the lower volatility (0.205 for more debt-like convertibles versus 0.484 for more equity-like). This result can be explained in terms of time varying elements (conversion price, maturity, volatility, dividend yield) that affect the value of the delta measure. Most of the debt-like issues in our sample occurred towards the end of our sample period, while the opposite is true for the more equity-like issues. We have shown in a previous section that issuers of the more equity-like convertibles experienced significantly positive abnormal returns prior to the announcement of the issue while those of more debt-like convertibles experienced significantly negative CAARs. The same conclusion can be inferred from Table 3, as the stock price run-up over the period (−10, −2) days prior to the announcement was significantly larger by 3 percentage points for the more equity-like issuers. Both types of issuers seemed to have similar leverage on average (0.254 for the more equity-like versus 0.218 for the more debt-like). The difference between interest coverage capacities was not significant. There was no statistically significant difference (at p < .05) between the Q-ratios of the equity-like and the debt-like issues. The equity-like issues appeared to be accompanied by more risk, as indicated by a higher volatility of respectively 48% versus 21% (annually). Note that issuers of equity-like convertibles were characterized as those that might have wanted to issue equity, but due to adverse selection and agency problems, this would have been too costly or impossible. The level of slack was significantly higher for the equity-like convertibles (8.5% of the total assets versus 2.4% of the total assets for debt-like issuers). The dividend yield was also significantly different between the issuers of the more equity-like and those of more debtlike convertibles. While the issuers of debt-like convertibles had an average dividend yield of around 12%, the issuers of equity-like convertibles had a significantly lower dividend yield of 0.6%. More equity-like issuers also had, on average, negative free cash flow relative to the total assets (−6.9% of the total assets), while the free cash flow for the more debt-like issuers was, on average, positive (1.6% of the total assets). This implies that, given the costly external finance, more equity-like issuers

seem to be more financially constrained than more debtlike convertible bond issuers. Panel B of Table 3 explores the differences between more debt- and more equity-like convertibles where we split the convertibles based on the alternative measure of convertible bond design—the equity-to-debt component ratio (ED). The results are similar to those in Panel A except the statistical significance for some of the differences changes. The findings here suggest that equity-like issuers (ED > 1) have a significantly lower dividend yield, have more slack, pay more taxes, have larger relative issue size, issue bonds with longer maturities, and are riskier. In Panel A only the delta measure significantly differed (by construction of the two subsamples) between the two subsamples. However, in Panel B all three measures of the convertible bond design—the delta (∆i), the ED measure, and the simple moneyness (M measure)— significantly differed between debt- and equity-like convertibles. The delta was, as expected: significantly higher for the equity-like convertibles (0.755) than for debt-like convertibles (0.371). Similar in the case of the ED measure (6.095 versus 0.597). The simple ratio of conversion value to par value of the bond (M) shows that equity-like convertibles were in the money at the issue (1.097), while the debt-like convertibles were not (0.880). This suggests that the ED measure may be somewhat better at capturing the design of a convertible bond, but on the other hand, the delta does provide similar results. Panel C of Table 3 uses the third measure of the convertible debt design—moneyness (M). As mentioned earlier, this measure ignores a number of factors that affect the convertible debt design, such as maturity, volatility, and dividend yield. We classified the bonds into more debt- and more equity-like by treating issues with the value of M > 1 as equity-like and those with the value M < 1 as debt-like. The results were very similar to those in Panels A and B—more equity-like convertible issues had significantly more slack, had better growth opportunities, had a longer maturity, represented a larger proportion of the book value of issuer’s equity, and had more volatile stock returns. Overall, the more equity-like convertible issuers had more slack and lower free cash flow. Lower free cash flow implies that equity-like convertible bond issuers are more financially constrained than the issuers of debt-like convertibles. However, issuers of the equity-like convertibles had significantly higher slack than the issuers of debt-like convertibles, which mitigates the problem of the financial constraint. The issuers of equity-like convertibles also had a lower dividend yield and were riskier than the more debt-like convertible issuers. This is in line with many previous findings (e.g., Lewis et al., 1999;

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

227

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Jen, Choi, & Lee, 1997) on the characteristics of convertible debt issuers. Finally, Panel D of Table 3 splits the sample according to the organizational structure of the issuer—into income trusts and standard corporations. Results show that income trusts had significantly lower leverage (0.183 compared to 0.249 for standard corporations). They also had significantly less slack (0.017 compared to 0.088), significantly more free cash flow (0.004 compared to −0.054) and paid significantly higher dividends (0.123 compared to 0.010). These results confirm the notion that income trusts face very low if not nonexistent agency costs of equity (note that free cash flow was not significantly different from 0 in case of the income trust subsample). In addition, income trusts were faced with a significantly lower tax burden than standard corporations (0.005 versus 0.012). In addition, income trusts had less volatile stock returns (lower risk) and were smaller in size. When it came to the convertible debt design, they issued bonds with lower maturity and lower conversion premiums. Although lower conversion premiums imply a more equity-like nature of convertible bond design, high dividend yields (high payout), low volatility, and shorter maturity resulted in a significantly lower value of the delta (0.168 versus 0.749). This suggests that income trusts issue convertible bonds that are more debt-like. This finding was also confirmed when we looked at the alternative measures of convertible debt design—the mean value of ED was significantly lower in the case of income trusts (0.703 compared to 4.542 for standard corporations) and the same holds for the mean value of the M measure (0.934 versus 0.981). Our results show that income trusts were significantly different from other issuers (standard corporations) of convertible bonds. First of all, they were faced with very low or almost nonexistent agency costs of equity compared to standard corporations. Secondly, they issued convertible bonds that were more debt-like. A significantly less negative wealth effect in the event window (−1, 1) for the issues announced by income trusts can therefore be explained by both lower agency costs of equity and the convertible debt design.

Cross-sectional Analysis of Determinants of the Size of the Wealth Effect Analysis of the total sample, convertible debt design, and the effect of income trust business design. In order to examine the impact on the size of the wealth effect due to the implicit design of convertibles (e.g., delta and the alternative measures of the convertible debt structure) and the issuer characteristics associated with debt- and

equity-related agency costs, we performed a number of cross sectional regressions. The dependent variable was the cumulative average abnormal return (CAAR) in the event window (−1, 1) in all regression specifications we considered. Table 4 presents the results based on the total sample. In the first specification we tested our hypotheses regarding the effects of debt-related agency costs (Hypothesis 2), equity-related agency costs (Hypothesis 3), the effect of the tax burden (Hypothesis 4), and the effect of some other determinants for the total sample of convertible debt issues. Based on the results of the first regression specification in Table 4,12 we did not find support for Hypotheses 2a and 2b that agency costs of debt have a negative effect on the market valuation. Leverage (LEV) had no significant effect on the market valuation in any of the specifications. The same holds for the proxy for TimesInterest-Earned (TIE). Hypothesis 3a is supported because the market valuation was worse after a period of significant stock price run-up (a significantly negative coefficient of −0.3339). This implies that an increase in cumulative abnormal stock returns in the event window (−10, −2) of 3% decreased the announcement related CAAR by 1 percentage point. This supports the hypothesis that investors are more concerned with overvaluation when the announcement of the issue is preceded by a streak of positive abnormal stock returns. Next, we found that the level of slack significantly positively affects CAAR. The coefficient of 0.1812 implies that an increase in slack of 10 percentage points increased the CAAR related to the announcement of the issue by 1.8%. As mentioned earlier, this result suggests that slack can be viewed as a build-up of internally generated funds. These are particularly important when the external sources of financing are very costly. Judging by our results, this effect dominates the effect of the high agency costs of slack capital. We expected the market valuation effect to be less favourable when the issuing firm had more free cash flow (Hypothesis 3b). However, even though we found the expected negative sign for the coefficient, it was not significant. Therefore we cannot confirm Hypothesis 3b that higher free cash flow negatively affects the market valuation. The coefficient for the payout proxy variable was a significantly positive 0.3059. This implies that an increase in dividend yield by 1 percentage point has, on average, a positive effect on the CAAR at the announcement of convertible debt offering of around 0.3%, all else being equal. This yields support for Hypothesis 3c that dividend payments positively affect the market valuation and is in line with the disciplining role of the payout policy. On the other hand, it could also account for the fact that dividend-paying companies are usually mature and less risky companies. The more direct

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

228

Can J Adm Sci 25(3), 214–236

Table 4 OLS Regressions of the Cumulative Average Abnormal Returns in the Event Window (−1,−1) on the Issue and Issuer Characteristics
2 bj t bj bj bj bj bj t t t t t bj 3 4 5 6 7 8 t

1

Variable

bj

t

−0.0039 −0.0001 0.1812 −0.3339 −0.0556 0.3059 −0.0077 0.0012 0.0001 0.1484 −0.3480 −0.0480 0.1410 −0.0087 −0.5471 0.0058 0.0286 −0.0008 0.0785 −0.0333 79 0.327 *** 76 0.314 *** 65 0.333 *** 70 0.373 *** −0.48 −0.0332 −0.46 −0.0937 −0.89 −0.0719 −0.78 0.89 0.0189 −0.0479 0.85 −0.62 76 0.311 *** 0.0233 −0.1169

−0.16 −0.63 2.06** −2.34** −1.23 2.74*** −1.32

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd. −0.0073 0.0000 0.1847 −0.3291 −0.0494 0.2951 −0.0079 −0.2849 −0.0005 0.0157 −0.0008 −1.98* −0.29 −0.11 2.08** −2.30** −1.06 2.65*** −1.36 −0.55 −0.11 1.34 −0.0051 0.0000 0.1732 −0.3327 −0.0465 0.3849 −0.0085 −0.2576 −0.0016 0.0140 0.0242 −0.20 −0.09 2.00** −2.38** −0.96 1.61 −1.42 −0.49 −0.31 1.16 0.49 −0.0009 0.0001 0.1405 −0.3525 −0.0478 0.2678 −0.0076 −0.5375 0.0042 0.0250 −0.03 0.29 1.30 −2.33** −0.80 1.83* −0.94 −0.76 0.60 1.38 −0.0158 0.0001 0.1204 −0.2925 −0.0933 0.3142 −0.0081 −0.4714 −0.0027 0.0055 −0.65 0.49 1.37 −2.12** −0.99 2.46** −1.28 −0.76 −0.42 0.43 −0.0038 0.0000 0.1824 −0.3297 −0.0450 0.3068 −0.0095 −0.2526 −0.0009 0.0159 0.0296 −0.14 0.08 1.99** −2.36*** −0.90 1.34 −1.53 −0.47 −0.18 1.25 0.58 0.51 0.04 1.34 −2.27** −0.77 0.92 −1.07 −0.77 0.77 1.53 −2.10** −0.0143 0.0002 0.1281 −0.2889 −0.0930 0.2173 −0.0089 −0.4681 −0.0018 0.0076 −0.57 0.65 1.42 −2.07** −0.96 1.63 −1.40 −0.75 −0.26 0.56 1.02 −1.02 65 0.327 *** 0.0784 0.0181 −0.0865 0.87 0.88 −0.87 70 0.366 ***

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

−0.0010 0.0155

−0.22 1.35

229

LEV TIE SLACK SPRUN FCFA DY Q TAXA LNTA RISSEQ DELTA ED M ITRUST intercept

−0.0306

−0.45

n Adj. R2 F-test

79 0.333 ***

Notes: Dependent variable is the cumulative average abnormal return in the event window (−1,−1) around the convertible debt offering announcement. Cumulative average abnormal returns are for the sample of 86 convertible bond announcements by Canadian companies from January 1991 to December 2004. The convertible bond announcements are identified from the SDC database. Abnormal returns are based on the market model, estimated over a 100-day period for each company (from day −120 to day −20). LEV is computed as the ratio between total debt and total assets. TIE is the Times-Interest-Earned ratio. This is defined as EBIT (Earnings Before Income and Taxes) over interest expense. SLACK is the ratio of cash and equivalents over total assets. SPRUN is the cumulative average abnormal stock return measured over the window (−10,−2) relative to the announcement date. FCFA is the ratio of free cash flow (net income + depreciation – capital expenditures) over the total assets. DY is the dividend yield. Q is the Tobin’s Q-ratio measured as the sum of the market value of equity measured as the average between (−15,−5) days relative to the announcement date and the book value of long and short term debt over the book value of total assets. TAXA is the ratio of income taxes over total assets. LNTA is the natural logarithm of total assets. RISSEQ is the ratio of the issue proceeds over the book value of equity. ∆ (DELTA) is a measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. This measure is calculated using the option pricing model of Black and Scholes corrected for continuous dividend payments (see Equation 1). ED is the equityto-debt component ratio based on the Tsiveriotis and Fernandes convertible bond valuation model. M is the “moneyness” measure defined as a ratio between conversion value at the announcement of the convertible bond issue and the par value of the bond. ITRUST is a dummy variable that takes a value of 1 if the company is organized as an income trust and a value of 0 otherwise. All the standard errors are White heteroskedasticity corrected. Under the null of the F-test all the bj are equal to 0. * p < .10, ** p < .05, p < .01.

LONCARSKI ET AL.

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

effect of the disciplining role of dividend payments needs to be explored on the subset of more equity-like convertible debt issuers, where the agency costs of equity are assumed to be more important. The coefficients for growth opportunities (Q) and relative issue size (RISSEQ) did not significantly affect the market valuation. The second specification in Table 4 additionally included a taxation proxy in the cross sectional regression to test for the effect of income taxes on the wealth effect. While other coefficients remained practically unchanged, we found no significant effect of taxes on the wealth effect associated with the announcement of the convertible debt offering. We therefore cannot confirm Hypothesis 4, which states that income taxes positively affect the market valuation. Note that an analysis for the total sample was not the most appropriate, since the design of the convertible had to be taken into account as we argued earlier. We therefore also estimated the third specification in Table 4, where we included a control variable for implicit issue characteristics by adding the delta measure as an explanatory variable. The delta measure reflects how debt- or equity-like the convertible issue was, and therefore it captured the issue characteristics comprehensively. Since a value of delta closer to 1 indicates a more equity-like convertible issue, we expected to find a negative relationship between the size of the wealth effect and the value of delta. The results of the third specification in Table 4 are very similar to those in specifications 1 and 2. The effect of delta on CAAR was not significant. Specification 4 in Table 4 includes the ED variable to account for the design of a convertible bond (instead of the delta measure). We observed that ED had an expected negative effect (coefficient of −0.0008); more equity-like convertible issuers experienced a more negative CAAR. However, the effect was only marginally significant, both in statistical and economical terms. Additionally, in specification 5 in Table 4, we used the second alternative measure of the convertible debt design—the M measure. The results remain comparable to specifications 3 and 4. The results of specifications 3, 4, and 5 suggest that issuer characteristics and convertible bond design seem to be closely related, as there was no significant additional information in the measures of convertible debt design that affected market valuation beyond the issuers characteristics. Given the specific role and organizational design of income trusts, we investigated the impact of such business design on the size of the wealth effect controlling for the convertible debt design, measured with delta and both alternative measures—the ED and the M measures. The results are presented in specifications 6, 7, and 8 of

Table 4. As can be observed, there is no difference in specification 6, where delta was used to measure the convertible debt design. On the other hand, in specifications 7 and 8, we observed that dividend yield no longer had a significant effect on the market valuation when we controlled for the organizational design. These results can be attributed to the fact that delta measure captured the effect of dividend payments, while the two alternative measures (ED and M) did not. The effect of dividend payments disappeared when we controlled for the organizational design of the issuer, since income trusts had significantly higher dividend yields and experienced significantly less negative wealth effects in the event window (−1, 1). The overall results suggest that perceived overvaluation (stock price run-up), slack, and dividend yield significantly affected the size of the wealth effect. The overvaluation is the most robust result, as it persisted even when controlling for a convertible bond design and organizational structure of the issuer. Agency costs of debt and equity versus convertible debt design and organizational structure of the issuers. In order to test hypotheses related to impact of debtrelated and equity-related costs on the size of wealth effects, we estimated the regressions separately for subsamples split according to implicit issue characteristics or the design of the issue (as measured by the delta, the ED measure and/or the M measure). Convertible issues with a value of delta below 0.5 are denoted as more debt-like, while those with a value of delta above 0.5 are identified as more equity-like. We expected debt-related costs to have a market valuation impact for more debt-like convertibles, and equity-related costs to have a valuation effect for more equity-like convertibles. As the results of the first specification (∆ < 0.5) in Table 5 show, we did not find evidence that leverage has a significant effect on the excess returns. This means that, just like in Table 4, we did not find any evidence to support Hypothesis 2a that higher financial leverage negatively affects the market valuation. The interest coverage had a significantly positive effect on CAAR of more debt-like convertible bond issuers. This can be interpreted as an evidence to support Hypothesis 2b that interest coverage positively affects the market valuation. The economic significance of the effect of interest coverage seems to be rather small, but given the very high variation in the TIE ratio (between around −2 to around 40 for debt-like convertible issues), the coefficient of 0.0013 suggests that an increase in TIE of 8 led to the increase in CAAR of around 1%, all else being equal. We found a positive, but not significant effect of free cash flow. Thus we were not able to confirm Hypothesis 3c that dividend payments positively affect the market

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

230

Can J Adm Sci 25(3), 214–236

Table 5 OLS Regressions of Abnormal Returns at the Announcement Date of Convertible Debt Offering on Issue and Issuer Characteristics for Various Subsamples
∆ > 0.5 bj 0.0121 0.0001 0.1215 −0.5215 −0.1519 −0.0139 −0.0066 −0.4296 −0.0105 −0.0199 0.1478 25 0.478 ** 0.15 0.08 0.69 −3.38*** −1.53 −0.07 −0.56 −0.34 −0.87 −1.29 0.76 t bj bj bj bj bj t t t t t ED < 1 ED > 1 M<1 M>1 ITRUST = 0 ITRUST = 1 bj −0.0441 0.0010 −0.1214 0.1431 0.1563 0.0006 −0.0288 0.0527 0.0017 0.3804 60 0.277 *** t −0.46 1.45 −0.50 0.65** 1.57 0.01 −0.85 0.19 0.02 0.83 18 0.110

∆ < 0.5

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd. −0.0045 −0.0001 0.2011 −0.3407 −0.0836 0.6603 −0.0075 −0.0782 −0.0002 0.0160 −0.0454 0.0014 0.0010 0.1030 −0.0959 0.0416 0.2908 −0.0177 −0.2806 0.0074 0.0616 −0.1406 26 0.188 39 0.306 ** 45 0.180 * 0.03 1.23 0.84 −0.54 0.44 1.60 −0.81 −0.49 0.46 0.90 −0.63 0.0461 −0.0001 0.2575 −0.4156 −0.1448 0.6200 −0.0016 −0.1014 0.0006 0.0066 −0.0798 0.78 −0.01 1.49 −2.27** −1.54 1.97** −0.16 −0.08 0.07 0.33 −0.56 56 0.276 ** −0.15 −0.39 1.91* −2.22** −1.35 0.56 −1.12 −0.10 −0.04 1.16 −0.53 −0.0352 0.0000 0.0259 −0.1136 −0.0632 0.4573 −0.0004 −0.4588 0.0073 0.0379 −0.1623 −0.98 −0.07 0.22 −0.85 −0.92 3.10*** −0.03 −0.56 0.75 0.87 −1.17 −0.0045 0.0000 0.2026 −0.3375 −0.0785 0.4865 −0.0079 −0.1747 0.0004 0.0158 −0.0544 −0.11 −0.02 1.87* −2.13** −1.23 1.47 −0.99 −0.23 0.09 1.12 −0.68

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

Variable

bj

t

LEV TIE SLACK SPRUN FCFA DY Q TAXA LNTA RISSEQ intercept

−0.0559 0.0013 0.0970 −0.0271 0.1455 −0.0516 −0.0364 0.0445 −0.0360 0.0139 0.5283

−0.92 2.46** 0.69 −0.19 1.26 −0.44 −2.39** 0.12 −3.96*** 0.46 4.00***

231

n Adj. R2 F-test

20 0.537 **

Notes: Dependent variable is the cumulative average abnormal return in the event window (−1,−1) around the convertible debt offering announcement. Cumulative average abnormal returns are for the sample of 86 convertible bond announcements by Canadian companies from January 1991 to December 2004. The convertible bond announcements are identified from the SDC database. Abnormal returns are based on the market model, estimated over a 100-day period for each company (from day −120 to day −20). LEV is computed as the ratio between total debt and total assets. TIE is the Times-Interest-Earned ratio. This is defined as EBIT (Earnings Before Income and Taxes) over interest expense. SLACK is the ratio of cash and equivalents over total assets. SPRUN is the cumulative average abnormal stock return measured over the window (−10,−2) relative to the announcement date. FCFA is the ratio of free cash flow (net income + depreciation capital expenditures) over the total assets. DY is the dividend yield. Q is the Tobin’s Q-ratio measured as the sum of the market value of equity measured as the average between (−15,−5) days relative to the announcement date and the book value of long and short term debt over the book value of total assets. TAXA is the ratio of income taxes over total assets. LNTA is the natural logarithm of total assets. RISSEQ is the ratio of the issue proceeds over the book value of equity. ∆ (DELTA) is a measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. This measure is calculated using the option pricing model of Black and Scholes corrected for continuous dividend payments (see Equation 1). ED is the equity-to-debt component ratio based on the Tsiveriotis and Fernandes convertible bond valuation model. M is the “moneyness” measure defined as a ratio between conversion value at the announcement of the convertible bond issue and the par value of the bond. ITRUST is a dummy variable that takes a value of 1 if the company is organized as income trust and a value of 0 otherwise. All the standard errors are White heteroskedasticity corrected. Under the null of the F-test all the bj are equal to 0. * p < .10, ** p < .05, *** p < .01.

LONCARSKI ET AL.

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

valuation. Growth opportunities had a significantly negative effect on CAAR (coefficient of −0.0364), where an increase in the Q-ratio of 0.3 led to an approximate 1% decrease in CAAR. Company size also had a significantly negative effect on CAAR. Although the negative effect of growth opportunities seemed somewhat surprising, it is possible to think of the Q-ratio as a proxy for the risk of the company. This might lead to difficulties in risk estimation such that larger issuers with higher growth opportunities (in the universe of more debt-like convertible issues) are perceived to be riskier. A somewhat striking result is the significant negative effect of the size of the issuing company on CAAR (coefficient of −0.0360). As stated earlier, firm size may not only mitigate adverse selection and agency problems, but could actually make them more acute since the opaqueness can also increase with the size. To sum up, we found support for one of the hypotheses relating to the effect of debt-related agency costs on the wealth effects associated with the announcement of convertible debt offerings for the subsample of debt-like convertibles, while equity-related agency costs did not adversely affect the valuation in this case. We did not find any significant effect of tax burden on the valuation (Hypothesis 4). The second specification (∆ > 0.5) in Table 5 relates to the subsample of more equity-like convertible bond issues. Here, we found that proxies relating to the agency cost of equity significantly affected the wealth effects at the announcement of convertible debt issues. More specifically, the prior stock price run-up negatively affected the valuation, as there is more concern about the potential overvaluation of the equity. As in Table 4, this result can be interpreted as a confirmation of Hypothesis 3a. The coefficient for SPRUN of −0.3407 suggests that a 5% positive cumulative average abnormal return in a ten day period prior to the announcement of the issue leads to a negative −1.7% CAAR following the announcement. Again, as in Table 4, we found a marginally significant positive effect of slack (coefficient of 0.2011) on the valuation. As discussed previously, this confirms that the overall effect of slack is positive, or, put differently, the flexibility benefits of such “buffer” funds in the case of costly external financing outweigh the agency cost of slack. With respect to the taxation proxy, we did not find a significant effect on the valuation in the case of more equity-like convertible issues, again leading to the conclusion that a positive effect of income taxes on the market valuation cannot be confirmed. The specifications 3–6 in Table 5 revisit the subsample analysis for more debt-like versus more equitylike convertibles using the alternative measures of convertible debt design. The results for the more equity-

like convertibles (ED > 1) in specification 4 are partly similar in terms of statistical and economic significance as in the case of subsample analysis based on the delta measure in specification 2. One difference is the effect of slack, which was not significant in this case. In addition, the effect of dividend yield on the wealth effect here was significantly positive (a coefficient of 0.6200), giving support to Hypothesis 3c. This suggests that in the case of equity-like convertibles, an increase in dividend yield of 0.01 led to an increase in CAAR of around 0.62 percentage points, all else being equal. Contrary to the results in specification 1, the results in specifications 3 and 5 for the subsample of the more debt-like convertibles, as defined with the value of ED below the value of 1 and the value of M < 1, yield no support to Hypothesis 2 that agency costs of debt had a negative effect on the market valuation in the case of more debt-like convertibles and a nonnegative effect for the more equity-like convertibles. The results for the more equity-like convertibles in specification 6 (M > 1) were similar in terms of statistical and economic significance as in the case of the subsample analysis based on the ED measure in specification 4 (ED > 1)—the results in both specifications support Hypothesis 3a that stock price run-up prior to the announcement of the issue negatively affects market valuation. This suggests that in the case of equitylike convertibles, a positive CAAR of 2% in a ten day period prior to the announcement of the issue leads to a negative CAAR of around 1% following the announcement. In specifications 7 and 8 of Table 5 we more closely investigated the role of organizational design, which is very particular to Canadian market—an income trust. We split the sample into issuers that were organized as standard corporations (ITRUST = 0) and those that were organized as income trusts (ITRUST = 1). The results show that investors react more negatively to the announcements of convertible debt issues by standard corporations when the issuers experience stock price run-up prior to the issue (a coefficient of −0.3375). There is also a positive effect of slack on the valuation. The economic significance of these findings is of the same order of magnitude as in the case of the more equity-like convertibles in specification 2 (∆ > 0.5). This is not surprising, as we showed in Panel B of Table 1 that most of the issues made by standard corporations are equity-like in nature. For the income trusts we have a very small subsample of issues and cannot make reliable estimates. Results based on the same set of variables as in specification 7 (ITRUST = 0) suffer from strong multicolinearity between proxies for size (LNTA) and growth opportunities (Q). We therefore estimated a model specification 8 where we excluded one of the two

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

232

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

highly correlated variables (Q) and found no significant effects on valuation by any of the proxies. This is in line with previous findings that agency costs of equity are very low or nonexistent for the income trusts, as well as the fact that convertible bonds issued by income trusts are almost exclusively debt-like. This implies that due to convertible debt design and characteristics of income trusts, agency costs of equity have no significant effect on the market valuation.

debt-like convertibles (−0.10%). Equity-like convertibles were significantly affected by agency costs of equity, but this was not the case for debt-like convertibles.

Contributions to Scholarship We contribute to the literature on wealth effects associated with convertible bond announcements. Previous studies found significantly negative results for AngloSaxon countries (US, UK, Australia), but mixed results for other countries. In line with previous studies for Anglo-Saxon countries, we found significantly negative wealth effects. However, a deeper analysis shows that the wealth effects are significantly more negative for equitylike than for debt-like convertibles. These results suggest that convertibles are used to mitigate information asymmetries (see Stein, 1992). Special attention is paid to income trusts, which are a special feature of the Canadian market. Convertible bonds issued by income trusts are associated with insignificantly positive abnormal returns. This result can be explained by a more debt-like convertible design and/or very low agency costs in case of income trusts.

Relationship between the Results and the Motives for the Use of Convertible Debt Table 6 presents the summary of hypotheses and the results. Although convertibles are hybrid securities that share characteristics of both equity and straight debt, they can nevertheless be classified according to their specific debt- or equity-like nature. Results for the total sample suggest that the agency costs of debt do not have any significant effect on the CAAR around the announcement of the convertible bond issue (H2 is not confirmed). On the other hand, agency costs of equity seem to have a negative effect (H3a always confirmed and H3c mostly confirmed), especially for the degree of potential overvaluation (stock price run-up prior to the announcement). We found similar results when we took into account the hybrid nature of convertibles and split the sample into more debt- and more equity-like convertibles. Results based on such classification suggest that the agency costs of debt have no effect on wealth effects of both debt-like (H2a not confirmed and H2b mostly not confirmed) and equitylike convertible issuers. The opposite holds for the agency costs of equity (H3a confirmed and H3c mostly confirmed). This lends support to the motives for the use of convertible debt as proposed by Stein (1992). We found no evidence for the tax hypothesis relating to the benefits of the use of convertible debt as opposed to the use of equity (Jalan & Barone-Adesi, 1995).

Applied Implications Issuers of equity-like convertibles are better off issuing convertible debt than equity. The reason for this is that the wealth effects associated with the announcement of the convertible debt are negatively affected by the agency costs of equity. This implies that in such a setting, convertible debt, which is not 100% equity (lower agency costs of equity), can be seen as a good substitute for equity. It creates a less negative response from the market as the debt-like feature acts either as a commitment or as a controlling device reducing the impact of the agency costs of equity.

Limitations and Future Research Directions Conclusion Summary We studied the announcement effects and the determinants of convertible bonds issued in Canada between 1991 and 2004 in order to identify issuer motives. The average wealth effect for the event window of day −1 to day +1 was −2.7%. This effect was significantly negative for equity-like issues (−3.67%) and insignificant for A remaining question of interest is that survey studies have generally found that managers don’t mention concepts related to agency costs of equity as their motivation to issue convertible bonds. Ideally a follow-up study would combine different research techniques, such as surveys and event study methodology. The answers to survey questions could then be related to the wealth effects. Such an approach could possible bridge the gap between theory and practice.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

233

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Table 6 Overview of Hypotheses and the Results of the Tests of the Hypotheses
Hypothesis Proxy Table with the result Result Result (∆ specific subsample) Result (ED specific subsample) Result (M specific subsample)

H1: The market valuation effect will be more negative for equity-like convertibles than for debtlike convertibles.

Difference in CAARs between equity-like (∆ > 0.5) and debt-like (∆ < 0.5) convertibles.

Table 2

C

Agency Cost of Debt H2: Agency costs of debt will have a negative effect on the market valuation for the more debt-like convertibles and a nonnegative effect for the more equity-like convertibles. H2a: Higher financial leverage negatively affects the market valuation, in particular for more debtlike convertibles. H2b: Interest coverage positively affects the market valuation.

Leverage (LEV): ratio between total debt and total assets.

Tables 4 and 5

NC

NC

NC

NC

Times-interest-earned ratio (TIE): Earnings Before Income and Taxes over interest expense on debt.

Tables 4 and 5

NC

C

NC

NC

Agency Cost of Equity H3: Agency costs of equity will have a negative effect on the market valuation for the more equity-like convertibles and a nonnegative effect for the more equitylike convertibles. H3a: A period of positive abnormal returns preceding the announcement date negatively affects the market valuation. H3b: Higher free cash flow negatively affects the market valuation.

CAAR over the window (−10,− 2) relative to the announcement date. Free cash flow (FCFA): ratio of free cash flow (net income plus depreciation less capital expenditures) over total assets. DY: the dividend yield.

Tables 4 and 5

C

C

C

C

Tables 4 and 5

NC

NC

NC

NC

H3c: Dividend payments positively affect the market valuation.

Tables 4 and 5 Tax Hypothesis

C

NC

C

NC

H4: Income taxes positively affect the market valuation, in particular for more equity-like convertibles.

TAXA: ratio of income taxes over total assets.

Tables 4 and 5

NC

NC

NC

NC

Notes: This table gives an overview of the hypotheses that we test in this paper as well as of the results of these tests. The delta-specific, EDspecific, and M specific subsample results are from specifications in 1, 3, and 5 in Table 5 for hypotheses 2a and 2b, from specifications 2, 4, and 6 in Table 5 for hypotheses 3a to 3c, and hypothesis (4) as well as from columns (2) to (8) of Table 4 for hypothesis (4). ∆ (DELTA) refers to a measure of the sensitivity of the value of convertible bond with respect to the value of the underlying equity. ED is the equity-to-debt component ratio based on the Tsiveriotis and Fernandes convertible bond valuation model. M is the “moneyness” measure defined as a ratio between conversion value at the announcement of the convertible bond issue and the par value of the bond. C denotes “Confirmed” and NC denotes “Not Confirmed”.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

234

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Notes
See, for example, surveys of managers by Billingsley and Smith (1996) for the US market and Bancel and Mittoo (2004) for the European markets. 2 Given that the exercise price is “paid” by redeeming the bonds, convertible bonds are in fact warrants with a variable exercise price. 3 See Table 12.4 of Loncarski, ter Horst, and Veld (2006) for an overview of US studies on wealth effects associated with convertible debt issue announcements. 4 Loncarski, ter Horst, and Veld (2006) report that in the period 1990–2003, 30% of all the convertible bond issues were placed in the US market, compared to less than 4% that were placed in the Canadian market. Moreover, the average convertible bond issue size in the US market in the same period was approximately 237 million USD, compared to the average issue of roughly 123 million USD in the Canadian market. 5 Note that the Canadian government already sidestepped this regulation due to heavy tax losses. Although income trusts have been around for more than 20 years, they were initially designed for mature industries with steady cash flows. However, lately the conversions of all sorts of business into income trusts have grown out of proportion (see for example The Economist, June 22, 2005: Canada’s income trusts—Cold shower.) 6 It is assumed that manager’s compensation policy includes a penalty (loss of position) in cases of bankruptcy, as is usually the case in reality. This makes the signal costly for the sender (manager). 7 A direct test of this tax motivated argument for the issue of convertible debt is also related to calls of convertibles, which we will not address in this paper. 8 SEDAR stands for “System for Electronic Document Analysis and Retrieval” and is a service of CSA (Canadian Securities Administration) providing public securities filings. (http://www.sedar.com/) 9 See Tsiveriotis and Fernandes (1998) for details on the valuation approach. 10 Note that the number of observations is less than 86 (initial sample) due to missing values of the ED measure for some issues. 11 The findings in both panels of Table 2 are confirmed using nonparametric test results as well. The Wilcoxon signed rank test, which tests the difference in sums of ranks of the mean adjusted CAAR above and below medium, gives significant differences. These differences were statistically significant for different event windows (up to 20 trading days) following the announcement date. A similar result, using the difference in means between subsamples in both Panels in Table 2 was obtained using the nonparametric Kruskall-Wallis test for the equality of subpopulations. These results are available on request from the authors. 12 Note that the number of observations is less than 86 (initial sample) due to either missing accounting items or the values of the convertible bond design measures for some issues. 1

References
Abhyankar, A., & Dunning, A. (1999). Wealth Effects of Convertible Bond and Convertible Preference Share Issues:

An Empirical Analysis of the UK Market. Journal of Banking and Finance, 23(7), 1043–1065. Aggarwal, L., & Mintz, J. (2004). Income Trusts and Shareholder Taxation: Getting It Right. Canadian Tax Journal, 52(3), 792–818. Akerlof, G. A. (1970). The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism. Quarterly Journal of Economics, 84(3), 488–500. Ammann, M., Fehr, M., & Seiz, R. (2006). New Evidence on the Announcement Effect of Convertible and Exchangeable Bonds. Journal of Multinational Financial Management, 16(1), 43–63. Bancel, F., & Mittoo, U. R. (2004). Why Do European Firms Issue Convertible Debt? European Financial Management, 10(2), 339–373. Billingsley, R. S., & Smith, D. M. (1996). Why do Firms Issue Convertible Debt? Financial Management, 25(2), 93–99. Brennan, M. J., & Kraus, A. (1987). Efficient Financing Under Asymmetric Information. The Journal of Finance, 42(5), 1225–1243. Burlacu, R. (2000). New Evidence on the Pecking Order Hypothesis: the Case of French Convertible Bonds. Journal of Multinational Financial Management, 10(3–4), 439–459. Chang, S.-C., Chen, S.-S., & Liu, Y. (2004). Why Firms Use Convertibles: A Further Test of the Sequential-Financing Hypothesis. Journal of Banking & Finance, 28(5), 1163–1183. Christensen, D. G., Faria, H. J., Kwok, C. C. Y., & 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(1), 99–119. Cleary, S., & MacKinnon, G. (2007). The Investment Nature of Income Trusts and their Role in Diversified Portfolios. Canadian Journal of Administrative Sciences, 24(4), 314–325. De Jong, A., & Veld, C. (2001). An Empirical Analysis of Incremental Capital Structure Decisions Under Managerial Entrenchment. Journal of Banking & Finance, 25(10), 1857–1895. De Roon, F., & Veld, C. (1998). Announcement Effects of Convertible Bond Loans and Warrant-Bond Loans: An Empirical Analysis for the Dutch Market. Journal of Banking & Finance, 22(12), 1481–1506. Dutordoir, M., & Van de Gucht, L. (2004). Are European Convertibles More Debt-Like than the US Issues? An Empirical Analysis. Tijdschrift voor Economie en Management, 49(4), 533–568. Dutordoir, M., & Van de Gucht, L. (2007). Are There Windows of Opportunity for Convertible Debt Issuance? Evidence for Western Europe. Journal of Banking & Finance, 31(9), 2828–2846. Eckbo, E.B., & Masulis, R.W. (1995). Seasoned equity offerings: A survey. In R. Jarrow, V. Maksimovic, & W. Ziemba, (Eds.), Finance (Series of Handbooks in Operations Research and Management Science), (pp. 1017–1072). Amsterdam: Elsevier-North Holland.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

235

Can J Adm Sci 25(3), 214–236

WHY DO COMPANIES ISSUE CONVERTIBLE BOND LOANS?

LONCARSKI ET AL.

Green, R. C. (1984). Investment Incentives, Debt, and Warrants. Journal of Financial Economics, 13(1), 115–136. Hull, J. (2006). Options, Futures, and Other Derivatives (6th ed.). New Jersey, Upper Saddle River: Pearson-Prentice Hall. Jalan, P., & Barone-Adesi, G. (1995). Equity Financing and Corporate Convertible Bond Policy. Journal of Banking & Finance, 19(2), 187–206. Jen, F. C., Choi, D., & Lee, S.-H. (1997). Some New Evidence on Why Companies Use Convertible Bonds. Journal of Applied Corporate Finance, 10(1), 44–53. Jensen, M. C. (1986). Agency Costs of Free Cash Flow, Corporate Finance, and Takeovers. The American Economic Review, 76(2), 323–329. Jog, V., & Wang, L. (2004). The Growth of Income Trusts in Canada and the Economic Consequences. Canadian Tax Journal, 52(3), 853–880. Kang, J. K., & Lee, Y. W. (1996). The Pricing of Convertible Debt Offerings. Journal of Financial Economics, 41(2), 231–248. Kim, Y. O. (1990). Informative Conversion Ratios: A Signalling Approach. Journal of Financial & Quantitative Analysis, 25(2), 229–243. Lewis, C. M., Rogalski, R. J., & Seward, J. K. (2003). Industry Conditions, Growth Opportunities and Market Reactions to Convertible Debt Financing Decisions. Journal of Banking & Finance, 27(1), 153–181. Lewis, C. M., Rogalski, R. J., & Seward, J. K. (1999). Is Convertible Debt a Substitute for Straight Debt or for Common Equity? Financial Management, 28(3), 5–27. Loncarski, I., ter Horst, J., & Veld, C. (2007). The Rise and Demise of the Convertible Arbitrage Strategy. Working

paper, Tilburg University, Simon Fraser University and University of Stirling. Loncarski, I., ter Horst, J., & Veld, C. (2006). Why do Companies Issue Convertible Bonds? A Review of Theory and Empirical Evidence. In L. Renneboog, (Ed.), Advances in Corporate Finance and Asset Pricing, (pp. 311–339). Amsterdam: Elsevier-North Holland. Lucas, D. J., & McDonald, R. L. (1990). Equity Issues and Stock Price Dynamics. The Journal of Finance, 45(4), 1019–1043. Magennis, D., Watts, E., & Wright, S. (1998). Convertible Notes: the Debt Versus Equity Classification Problem. Journal of Multinational Financial Management, 8(2–3), 303–315. Miller, M. H., & Rock, K. (1985). Dividend Policy Under Asymmetric Information. The Journal of Finance, 40(4), 1031–1051. Myers, S. C. (1977). Determinants of Corporate Borrowing. Journal of Financial Economics, 5(2), 147–175. Myers, S. C., & Majluf, N. S. (1984). Corporate Financing and Investment Decisions When Firms Have Information that Investors do Not Have. Journal of Financial Economics, 13(2), 187–221. Ross, S. A. (1977). The Determination of Financial Structure: The Incentive-Signalling Approach. The Bell Journal of Economics, 8(1), 23–40. Stein, J. C. (1992). Convertible Bonds as Backdoor Equity Financing. Journal of Financial Economics, 32(1), 3– 21. Tsiveriotis, K., & Fernandes, C. (1998). Valuing Convertible Bonds with Credit Risk. The Journal of Fixed Income, 8(2), 95–102.

Copyright © 2008 ASAC. Published by John Wiley & Sons, Ltd.

236

Can J Adm Sci 25(3), 214–236


				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:213
posted:6/24/2009
language:English
pages:23