Catering Convertible Design to Hedge Fund Demand

Transcription

1 Catering Convertible Design to Hedge Fund Demand Eric Duca and Marie Dutordoir Abstract Towards the end of the 20 th century the primary convertible bond market changed from one in which purchasers held only long posions in convertible bonds, to one dominated by convertible bond arbrageurs. In this paper, we provide new evidence about the types of firms that reacted to this increase in demand, and how they adapted the design of their offering. We find a significant amount of issuers that took advantage of the increased demand to be debt-like, rather than financially constrained firms. We also show that underpricing is less sensive to the risk associated wh the issuer. This is in line wh the hypothesis that arbrageurs are less concerned about issuer risk since they can hedge away. This allows firms to structure the design of the convertible to cater to hedge funds, obtaining favorable financing terms while doing so. Our paper contributes to a recent strand of lerature on the impact of buy-side characteristics on corporate finance transactions, and extends prior knowledge on the motivations for firms to issue convertible bonds. November 22, 2014 Keywords: Capal Supply, Convertible Debt, Hedge Funds * Eric Duca is from Colegio Universario de Estudios Financieros. Corresponding author: Eric Duca, Colegio Universario de Estudios Financieros (CUNEF), Serrana Angua 8, Madrid, Spain. tel ericduca@cunef.edu. Marie Dutordoir is from Manchester Business School, Manchester, M15 6PB, Uned Kingdom

2 1. Introduction Convertible bonds are hybrid securies that pay a fixed coupon, but also offer the possibily of conversion into stock, thus combining features of both straight debt and equy. Theoretical studies on convertible debt argue that convertibles are useful instruments for migating equy- and debt-related financing costs at the firm-sècific and aggregate level (Green, 1984; Brennan and Kraus, 1987; Brennan and Schwartz, 1988; Stein, 1992). 1 Empirical evidence in line wh this prediction is provided by Lewis, Rogalski, and Seward (1999, 2003), and Krishnaswami and Yaman (2008), among others. Until the end of the 20 th century, convertibles were purchased predominantly by investors such as insurance companies and pension funds, which required both a constant income stream and limed exposure to equies (see, e.g., Lummer and Riepe, 1993). At the turn of the century, however, a different type of buyer emerged: Convertible bond arbrage hedge funds consistently purchased around 80% of new offerings after the year 2000 (see, e.g., Brown, Grundy, Lewis, and Verwijmeren, 2012). These funds implement profable strategies by purchasing underpriced convertible bonds at issuance while hedging away other risks associated wh the issue (Calamos, 2003). 2 By hedging, arbrageurs transform the convertible into a secury wh significantly lower risk, thus reducing information asymmetries associated wh the valuation of the issue (Mchell, Pedersen, and Pulvino, 2007). This enables firms to quickly issue convertibles to this new type of buyer. There is anecdotal evidence that different types of 1 Between 1975 and 2007 companies issued convertibles worth on average 35% of their total assets, compared wh 49% raised from seasoned equy offerings and 11% raised from straight debt offerings. 2 Mchell, Pedersen, and Pulvino (2007) note that convertibles tend to be issued at prices below fundamental value (i.e., underpriced) in exchange for liquidy from hedge funds. 1

3 companies reacted to hedge fund demand for new issues. On hedge funds, Pilliam (2004) reports that these investors wouldn't need to meet the company or understand what does. On the other hand, Dinsmore, Dinsmore, and Finnican (2009) note that even better-rated issuers were tempted by attractive pricing that resulted from competion among hedge funds. Overall, the new demand from hedge funds presents a new opportuny for companies wishing to raise money, which is distinct from the tradional reasons for issuing convertibles that ce the migation of equy- or debt-related financing costs. In this paper we first examine how the shift in the investor base, from tradional long-only investors to convertible arbrage hedge funds, affects the types of firms issuing convertibles. One hypothesis is that issuers are riskier firms that take advantage of hedge fund demand to raise finance, since these funds are less sensive to issuer risk characteristics. The alternative hypothesis is that issuers are firms that can easily issue other types of securies, but issue convertibles to take advantage of better pricing that results from competion among hedge funds. Next, we examine whether the sensivy of convertible pricing to the risk characteristics of an issue changes wh the presence of convertible arbrage funds. Since arbrageurs are able to hedge away firm risk, we expect that pricing is less sensive to the tradional risk factors associated wh an issue. A corollary of this hypothesis is that the design of the convertible will change to appeal to hedge funds, rather than to migate equy- and debt-related financing costs. In particular, we expect convertibles to have a higher delta. Brown et al. (2012) and Calamos (2003) note that convertible arbrage 2

4 hedge funds prefer issues that are more equy-like, since their strategies depend on bonds being sensive to stock price movements. We test our hypothesis by comparing issuance behavior in two periods, the prearbrage period ( ) and the arbrage period ( ), wh the latter being the era dominated by convertible arbrage hedge funds. Our analysis then proceeds in two steps: First we compare the types of companies that issue convertibles in the prearbrage period wh those in the arbrage period. We complement this preliminary analysis by estimating a secury choice model for the decision to issue convertibles rather than debt or equy, to examine how the influence of firm characteristics changes across periods. Overall, the shift in the investor base should influence the reasons for issuing convertibles, from those intended to minimize financing cost associated wh equy or straight debt issues, to intentions associated more wh opportunistic behavior. The evidence about the characteristics of issuers suggests that a significant portion of firms issuing convertibles could have otherwise issued straight debt. Hence, opportunistic behavior is not limed to riskier firms that would be financially constrained in normal condions. This is in spe of the fact that arbrageurs are less concerned wh firmspecific risk, since they can hedge away. The design of offerings also changes, wh firms increasing delta in line wh the idea that they cater to hedge fund preferences. In the second part of our analysis we examine how the sensivy of underpricing to the convertible bond design changes in accordance wh the change in investor base, and the new motives of issuers. We find that in the arbrage period, underpricing is less sensive to firm-specific and issue-specific characteristics that proxy for firm risk. The 3

5 most significant finding is that, whereas delta is posively related to underpricing in the pre-arbrage period, has a negative relationship in the period dominated by hedge funds. 3 At the same time, issuers increase delta, catering to hedge fund preferences for convertibles that are more sensive to stock price movements, and obtaining cheaper funding by doing so. Hence, issuers are able to structure their offerings in a way that caters to hedge funds because pricing has become less sensive to the convertible bond design. Convertible bond issuers in the hedge fund period can therefore quickly raise money whout sacrificing much in terms of pricing of the issue. Our results contribute to a recent strand of lerature that examines how changes in the supply of funds influence corporate finance decisions and asset prices. Baker (2009) provides a framework for analyzing these supply effects on corporate decisions. There is also an increasing amount of empirical evidence. Baker and Wurgler (2004) show that managers cater to investor preferences for dividend-paying firms. Manconi and Massa (2009) find that firms wh more fragmented ownership have more difficulties in catering to shareholder preferences for dividend payouts. Baker, Greenwood, and Wurgler (2009) obtain evidence that managers choose to maintain share prices at a low level in response to investor preferences for low-priced firms. Aghion and Stein (2008) provide a theory in which managers cater to the stock market s preferences for sales growth versus prof margins. Polk and Sapienza (2009) show that firms cater to investor preferences for the level of firm investment by altering their investment policy. Choi et al. (2010) and de Jong, Duca, and Dutordoir (2013) document that increases in convertible arbrage hedge fund flows have a posive impact on convertible debt issuance. 3 Delta measures whether a convertible is more equy-like or debt-like, and a larger delta suggests more uncertainty regarding the pricing of an issue. 4

6 Our results complement these studies by providing further evidence of firms reacting to the increased demand from hedge funds. In particular, we show that a significant part of the increased issuance is by companies that could have issued debt instead, rather than by companies that are financially constrained. We also find evidence of catering the design of convertible bonds to the specific requirements of hedge funds. More significantly, we find that the design is altered at the same time that becomes less sensive to pricing, allowing issuers to obtain cheaper funding when catering. These finding also have implications for tradional theories of issuance, which propose that firms design their convertible bonds to migate debt- and equy-related issuance costs. Our results suggest that firms give less importance to these contracting costs when there are opportunies to raise capal quickly. The remainder of the paper is structured as follows. In the next section we review the lerature, whereas in Section 3 we describe the data and summary statistics. In Section 4 we discuss our empirical results, and conclude the paper in Section Lerature Review In this section, we review the main reference lerature for our paper. A first strand of lerature analyzes the motivations for firms to issue convertible debt and describes how convertibles can be designed to migate financing costs. A second strand of related papers examines the factors influencing the design of convertible bonds. We then turn our attention to lerature that explores how the availabily of investor capal influences corporate financing decisions Motivations for convertible debt issuance 5

7 The theoretical lerature presents several viewpoints on why firms issue convertible debt. Mayers (1998) and Wang (2009) argue that convertibles are useful in reducing the issuance costs of sequential financing while at the same time migating overinvestment. Green (1984) demonstrates that, because of the convexy in their payoffs, convertible bonds are useful in reducing agency costs associated wh risk shifting behavior of shareholders. Brennan and Schwartz (1988) show that convertible bonds substute for straight debt if information asymmetry about the riskiness of the firm s assets is high, since convertibles are less sensive to risk as a result of their option component. Convertibles reduce this problem, because if the firm is riskier than the convertible bond holders originally thought, they will prof from the higher risk by means of the conversion option component of their bonds. In the Stein (1992) framework, firms wh high financial distress costs issue convertibles as an alternative to equy to alleviate equy-related adverse selection costs. In aggregate, these theories predict that convertibles are most useful for firms wh high costs of attracting equy or straight debt financing. Using a secury choice model that incorporates convertible debt, equy, and straight debt, Lewis, Rogalski, and Seward (1999, 2003) find empirical evidence consistent wh these hypotheses. Another expectation for convertible debt is that, irrespective of firm-specific characteristics, issuance should be more attractive during windows wh higher economywide costs for attracting equy or debt financing. A number of empirical studies have examined the impact of aggregate equy- and debt-related financing costs on the convertible debt issuance choice. Hoffmeister, Hays, and Kelley (1987), and Mann, Moore, and Ramanlal (1999) obtain evidence that aggregate convertible bond issuance is 6

8 influenced both by equy market and bond market condions. Krishnaswami and Yaman (2008) find that companies are more likely to substute convertible debt for straight debt during periods wh high economy-wide debt-related financing costs Convertible bond structure Convertibles can be structured to be more equy- or debt-like, allowing companies wh high costs of attracting common equy or straight debt to access the capal markets. A common component used in the lerature to capture the design is a variation of the Black-Scholes (1973) delta, which measures the sensivy of a convertible bond to the underlying stock price. 4 Lewis, Rogalski, and Seward (2003) argue that convertibles are more equy-like when they substute for equy, thus reducing financing costs arising from asymmetric information, as in the theory of Stein (1992). Convertibles that are structured hedge-like are more suable to deal wh the asymmetric information problem that is suggested by Brennan and Schwartz (1988). Finally, convertibles that are structured more debt-like are used to reduce debt-related financing costs associated wh the risk-shifting hypothesis of Green (1984). Krishnaswami and Yaman (2008) examine the firm-specific characteristics affecting delta and conclude that convertible bonds are more equy-like when bankruptcy costs and adverse selection costs are higher and more debt-like when information asymmetry is larger. Overall, the lerature suggests that delta is more likely to be larger when issuers suffer from valuation problems that are more typical of equy issues. Regarding the more specific elements of design, two important determinants of the equy-likeness are the conversion premium and matury. Kim (1990) proposes a model 4 Delta is corrected for continuous dividend payments as suggested by Merton (1973). 7

9 in which a larger conversion premium is seen as a posive signal of future firm value. A high conversion premium implies that the share price must rise by more before the convertible option becomes in the money, signaling that managers are optimistic about future prospects. On the other hand, a longer matury is usually a bad signal and detrimental to the convertible value, since gives more opportuny for risk-shifting in the future (see, e.g., Brennana and Shwarz, 1988, and Lewis, Rogalski, and Seward, 1998). The aforementioned lerature shows that the design is adapted to minimize costs of information asymmetry, which have a negative influence on the pricing of convertible bonds. However, convertible arbrage funds can hedge firm-specific risks, and are less sensive to the bond design. As noted by Brown et al. (2012) and Calamos (2003), hedge funds prefer convertibles wh a larger delta, since allows for more trading profs from market-making while dynamically hedging. 2.3 Impact of capal availabily on corporate finance decisions In the neoclassical framework, markets are efficient and the supply of funds is perfectly elastic at a price that reflects fundamental value. A recent strand of lerature, however, contends that changes in the supply of funds do have an effect on corporate finance decisions and asset prices. Baker (2009) argues that changes in the supply of capal can be caused by two factors, i.e., (i) changes in investor tastes or preferences, and (ii) changes in the funds available to investors. In this paper, we examine the impact of the second factor funds available to convertible hedge fund arbrageurs on convertible bond issuers and the sensivy of underpricing to design. 8

10 Several previous papers provide empirical evidence that corporate finance decisions and secury prices are indeed influenced by investor tastes. Baker and Wurgler (2004) show that managers cater to investor preferences for dividend-paying firms. Manconi and Massa (2009) find that firms wh more fragmented ownership have more difficulties in catering to shareholder preferences for dividend payouts. Baker, Greenwood, and Wurgler (2009) obtain evidence that managers choose to maintain share prices at a low level in response to investor preferences for low-priced firms. Aghion and Stein (2008) provide a theory in which managers cater to the stock market s preferences for sales growth versus prof margins. Polk and Sapienza (2009) show that firms cater to investor preferences for the level of firm investment by altering their investment policy. 3. Sample selection and descriptive statistics 3.1. Sample Selection We obtain data for U.S. convertible debt, seasoned equy, and straight debt offerings made between January 1992 and December 2007 from the Securies Data Company s New Issues database (henceforth SDC). 5 We exclude utilies (SIC codes ) and financial firms (SIC codes ), and consolidate multiple tranches of convertibles issued by the same firm on the same date. We only include plain vanilla convertible bonds (no exchangeable bonds, mandatory convertible bonds, or convertible preferred stock). To be included in the sample of SEOs, firms must also have data available in both CRSP and Compustat. We thus obtain a data set of 954 convertible debt issues. 5 We start our period in 1992 because pricing data are unavailable before this date. In addion, our sample ends in 2007 to avoid the confounding effects associated wh the collapse of Lehman Brothers in 2008, as documented in Duca, Dutordoir, Veld, and Verwijmeren (2012). 9

11 Our analysis involves the comparison of two regimes: the era wh long-only buyers (pre-arbrage period) and the era dominated by convertible bond arbrageurs (arbrage period). The precise date when convertible bond arbrage hedge funds surpassed longonly investors as the main purchasers of convertible bond issues is not clear-cut. Duca et al. (2012), however, find evidence that these funds dominated the market after the year We follow this paper and use the year 2000 as the starting point for the period dominated by hedge funds. Therefore the pre-arbrage period spans and accounts for convertible bond 319 issues, whereas the arbrage period spans , and comprises 635 issues. We obtain balance sheet and income statement data from the Compustat Fundamentals Annual database, stock price-related data (i.e., prices, shares outstanding, and trading volumes) from the Center for Research in Secury Prices (CRSP), issuespecific information from SDC, and macroeconomic data from Datastream. The detailed definion of all the variables used in this study is given in Appendix A Descriptive statistics Table 1 provides detailed descriptive statistics for characteristics of convertible debt issues over the entire sample period, as well as sub-periods. A comparison of the firm characteristics across sub-periods suggests that in the arbrage period firms are more debt-like, being bigger (LogAssets), and have more leverage. In addion, a larger proportion of firms had issued debt in the three years prior to the convertible offering. Firms in the arbrage period, however, also have larger information asymmetries, captured by the stock volatily. 10

12 << Please include Fig. 1. here >> << Please insert Table 1 here >> Turning to the issue specific characteristics, an important observation is that underpricing decreases from an average of 21.5% in the pre-arbrage period, to 15.5% in the arbrage period. This represents a significant reduction in the cost of issuing convertible bonds. The annual change in underpricing can also be seen in Figure 1. The delta is also higher, as firms structured their issues to be more equy-like. Both this feature and the increased volatily of issuers are particularly appealing to hedge funds. The term to matury and the conversion premium are also larger, whereas the amount raised is less. These design features, on the other hand, are more beneficial to issuers, since a longer matury secures financing longer into the future, whereas a larger conversion premium leads to less dilution if the convertible bond is converted into equy. Announcement returns also became more negative, in line wh the short selling pressure documented by Duca et al. (2012). Finally, we observe an increase in unrated issues in the arbrage period Sub-sample analysis In Table 2 we shed further light on the increase in unrated issues. This increase is partly driven by higher demand by hedge funds, which are less concerned by firmspecific risk. The table suggests that unrated issuers in the arbrage period are more debt- 6 Table 1 shows only a slight increase in percentage hedge fund flows in the arbrage period. In dollar terms, however, the change is striking: Monthly flows into convertible arbrage hedge funds are on average $28 million till 1999, rising to an average of $278.4million in the arbrage period. Since hedge funds use leverage these differences would be more dramatic in practice. 11

13 like, having lower market-to-book-value, and larger size and leverage. These companies are also more likely to have issued debt in the three years prior to a convertible offering. The volatily of unrated issuers is similar across periods, although the delta is higher for unrated issuer sin the arbrage period. Both of these characteristics are more appealing to hedge funds. These features might explain why underpricing for unrated issues is lower in the arbrage period. An even more sharp decrease in underpricing can be observed for rated issuers. In the arbrage period, the characteristics of these issuers suggest that they are debt-like, but they also have significantly higher volatily and market-to-book-value than in the prearbrage period. Issues are made when flows into hedge funds are particularly large, suggesting that they benef most from the demand effect on underpricing. << Please insert Table 2 here >> << Please insert Table 3 here >> Table 3 presents sub-sample characteristics based on whether underpricing is above or below the median in each period. The most striking observation is that the largest decrease in underpricing occurs in those firms that are below the median level. Underpricing for this sub-group falls from 14.1% in the pre-arbrage period, to 4.8% in the arbrage period. These companies are more debt-like, being larger and having more leverage, but are also riskier, wh a higher MB and stock volatily. The drop in underpricing might be related to their larger delta, which makes them more attractive to hedge funds. This hypothesis is explored deeper in the regression analysis in the next section. 12

14 << Please insert Table 4 here >> In Table 4 we separate issuers according to whether they are predicted to have issued debt or equy, had they not chosen convertible debt financing. Firms are predicted to have issued equy if the Probabily issue equy dummy (defined in Appendix A6) is one, and debt otherwise. The sub-sample analysis indicates that the increase in debtlikeness in the arbrage period is not constrained to those issuers that are predicted to have issued debt (Debt-prob.). For instance, size and leverage is higher in the arbrage period for both sub-groups. On the other hand, the increase in stock volatily is most prominent for the Debt-prob group, suggesting that debt-like issuers swching to convertibles took advantage of their larger volatily to appeal to hedge funds. << Please insert Table 5 here >> In Table 5 we examine the characteristics of issuers according to whether they issue convertibles wh a high or low delta. 7 Tradional explanations for the design of convertibles suggest that a high delta is more likely to be chosen by companies that need to migate financing costs associated wh equy issues. Similarly, convertibles wh a low delta are more typical of debt-like companies, which swch to convertibles to alleviate debt-related issuance costs. These predictions are clearly apparent in both periods. Low delta issuers tend to have significantly lower stock price run-up, market-to- 7 High (low) delta is defined as being above (below) Brown et al. (2012) and Lewis et al. (2003) also classify issuers by delta. 13

15 book-value, and volatily, but are larger and have more leverage. In addion, they are more likely to have issued debt in the three years prior to the convertible issue, especially in the pre-arbrage period, where 93.8% of firms had a debt issue prior to the convertible offering. While the underpricing for both high and low delta issuers is lower in the arbrage period, the drop is more marked for the high delta group, wh underpricing decreasing from 24.2% to 17.4%. This supports the hypothesis that the preference of arbrageurs for a larger delta reduces underpricing, as funds compete for these issues. 4. Regression analysis The summary statistics in the previous section suggest that issuers taking advantage of increased hedge fund demand were more debt-like than equy-like. In this section we test the hypotheses about the shift in convertible bond buyers from long-only to arbrage hedge funds by adopting a two-stage approach. In the first step we use a secury choice model to examine how this shift in demand alters the influence of tradional determinants of the choice between convertible issuance and straight debt or equy. After that, we document the effect of the change in investor base on underpricing of convertible bond offerings Choice between convertible debt, straight debt, and equy issues In the first step of our analyses we use a secury choice model to examine how the increasing importance of convertible bond arbrageurs has influences the choice of companies to issue convertible bonds. We first estimate a logistic regression model for 14

16 the choice between convertible debt and straight debt using firm characteristics and aggregate financing costs as explanatory variables. We then repeat the exercise for the choice between convertible debt and equy issues. We estimate the model over the whole period ( ), as well as the pre-arbrage period ( ) and the arbrage period ( ). The logistic regression model for the choice between convertible bond and straight debt or equy issues is specified as: Pr(Conv 5 9 1) StockRunup LogAssets TermSpread t Leverage 10 MktRunup t 7 Slack StockVolat 11 2 MktVolat t 3 8 Tax Rate 12 t 4 MB Sentiment υ (1) Where Conv is a binary variable that takes a value of one if a convertible bond is issued in quarter t by firm i, and a value of zero if straight debt is issued (Table 6, Panel A), or if equy is issued (Panel B). 8 Data on debt and equy issues are obtained from SDC. The explanatory variables are defined in appendixes A1 and A3. << Please insert Table 6 here >> Column (1) of Table 6 Panel A show estimated coefficients of the logistic regression model for the choice between convertible debt and straight debt, for the entire period. Wh regards to the firm-specific characteristics, we find a significantly posive coefficient for Stock Run-up and Slack. As expected, convertible issuers also have less to gain from tax benefs. The coefficients of Size and Leverage are also significantly negative. Finally, in line wh previous studies, we find that convertible issuers are riskier 8 We treat multiple secury issues by the same firm at different points in a quarter as separate observations. 15

17 than straight debt issuers (significantly higher StockVolat). In Column (2) we add the variable Hedge Fund Flows, and find that companies are more likely to issue convertible bonds if hedge fund flows are larger. Comparing the pre-arbrage period (Column 3) wh the arbrage period (Column 4), we find that the coefficients are mostly stable. The most notable difference is that stock volatily is around four times as important for the choice between convertibles and straight debt in the arbrage period than is in the previous period. This is in line wh the hypothesis that riskier firms take advantage of hedge fund demand for volatile stocks and issue convertibles rather than debt. In Panel B of Table 6, we show that the estimated coefficients of the logistic regression model for the choice between convertible debt and seasoned equy. We find that, across all periods, convertible issuers have a significantly lower Stock Run-up and Leverage, and significantly higher Size and StockVolat than equy issuers. These findings are largely consistent wh those of previous studies (see, e.g., Lewis, Rogalski and Seward (1999, 2003) and Krishnaswami and Yaman (2008)). In Column 2, we introduce hedge fund flows, which has a posive influence on the decision to issue convertibles. In Column 3 and 4 we see that the coefficients for firm characteristics are relatively stable across sub-periods, apart from Tax, which is only significant in the pre-arbrage period, and Leverage, which is significant in the Arbrage period Underpricing effects of convertible bond issues In the final part of our analysis we examine the effect of the change in investor base on underpricing of convertible bond offerings. Several studies report that the large majory of convertibles are underpriced at issuance (Ammann, kind, and Wilde, 2003; 16

18 Chan and Chen, 2007; Loncarski, ter Horst, and Veld, 2009; de Jong, Dutordoir, and Verwijmeren, 2011; Mchell, et al., 2007). Convertibles are sold at a discount for reasons such as their relative illiquidy, and the complexies associated wh the valuation of hybrid securies (Lhabant, 2002). Lerature on underpricing of equy issues also ces information asymmetry as a main reason for the offering discount (see, e.g., Eckbo Masulis, and Norli, 2007 for a review). Hence, we expect issue characteristics that proxy for information asymmetry to have a posive influence on underpricing. We estimate the following regressions of firm-level underpricing on several proxies for financing costs: UP StockRunup TermSpread LogAssets Matury t t 15 6 Leverage 10 ConvPremium 2 Slack MktRunup t t 7 Tax StockVolat MktVolat Proceeds t t 4 8 MB Rate t Sentiment Sentiment t t υ 13 Delta t (2) Underpricing is measured at the issue date of new convertible bond offerings, and is calculated as shown in Appendix A4. We include the same set of firm-specific and economy-wide control variables as in the logistic regression analysis, to control for the financing costs associated wh an issue. We also include issue-specific variables to examine the influence of the convertible bond design on underpricing. We use Delta to capture the overall equy-likeness of the issue. More specific design features are captured by the matury of the bond, and the conversion premium, obtained from SDC. Ammann, et al. (2003) find that longer-dated bonds tend to be more underpriced, reflecting the larger uncertainty associated wh their longer matury. The other aspect of design is the conversion premium, wh a larger premium signaling better prospects, as in 17

19 the model of Kim (1990). We also control for the ratio of offering proceeds to the firm s total market value (obtained from Compustat and measured at fiscal year-end preceding the issuance month), since previous studies on seasoned equy offerings indicate that secury underpricing is higher for offerings wh a larger size (Altinkilic and Hansen, 2003; Intintoli and Kahle, 2010). << Please insert Table 7 here >> The results of the underpricing analysis are shown in Table 7. t-statistics are based on standard errors adjusted for clustering of observations whin each year. In Column 1 we analyze the whole sample period. Market-to-book-value has a negative influence on underpricing, suggesting that investors view this variable as a proxy for growth opportunies rather than overvalued equy. Higher growth opportunies are expected to be associated wh lower underpricing, since they increase the option value of a convertible. LogAssets has a negative influence on underpricing, reflecting the higher information asymmetry associated wh smaller firms. Stock volatily also captures information asymmetry, having a posive relationship wh underpricing, in line wh King (1986) and Kang and Lee (1996). Leverage, which can capture costs of financial distress, has a marginally significant posive influence on underpricing. We also find a significant posive coefficient for Rate, and significant negative coefficients for TermSpread. In the second column we include the proxy for flows into hedge funds, which has a negative influence on underpricing. This finding is in line wh de Jong et al. (2013), who 18

20 show that increased demand for convertible bonds by hedge funds reduces underpricing. In Column 3 and 4, we examine the pre-arbrage and arbrage periods separately. The regression coefficients for the firm-specific characteristics are mostly stable. MB, however, is only significant in the latter period, whereas leverage becomes insignificant. Next we examine the influence of design on underpricing by introducing delta as an explanatory factor in the pre-arbrage period (Column 5) and the arbrage period (Column 6). Delta has a posive influence on underpricing in the first period, in line wh the hypothesis that more equy-like issues are more difficult to value. In the period dominated by arbrageurs, however, delta has a negative effect on underpricing. This striking finding supports the hypothesis that convertible arbrageurs prefer equy-like securies and are willing to accept lower underpricing for issuers that structure them in this way. In economic terms, a one-standard deviation increase in delta in the arbrage period leads to a decrease in underpricing of 1.3 percentage points. An increase in delta in the pre-arbrage period, on the other hand, leads to an increase in underpricing of 2.1 percentage points. This finding is an important factor in explaining the lower average underpricing in the arbrage period. In Columns 7 and 8 we introduce specific design factors. 9 The most prominent finding is that matury has a posive influence on underpricing in the pre-arbrage period, which then becomes negative in the arbrage period. This evidence supports the idea that hedge funds are less concerned wh proxies for issue risk, given that they can hedge away. In fact, Brown et al. (2012) and Huang and Ramirez (2010) note that hedge funds most likely do not hold the convertible for more than three months after purchase. As expected, a larger conversion premium, which signals better future stock 9 We exclude log assets from these regressions because this variable is highly correlated wh proceeds. 19

21 prospects, leads to less underpricing. The offering size, on the other hand, only has the predicted posive influence in the pre-arbrage period. This finding suggests that hedge funds are less concerned about the negative signal associated wh a larger offering size. Another interesting finding is that market-to-book-value has a negative influence on underpricing in the arbrage period. Whereas this finding suggests that MB captures growth opportunies, could also be partly explained by the fact that hedge funds prefer issues that are harder to value. Supportive evidence comes the from smaller coefficient of volatily in the arbrage period, which can be seen as evidence that hedge funds prefer issues that have the potential for more price movements. Altogether, the evidence in columns 5 8 supports the notion that hedge funds are less concerned about the valuation uncertainties surrounding riskier convertible issues. This can partly explain why issuers in the arbrage period are riskier and adapt the design of their convertible issues to the new type of buyer. Conclusions Convertible bonds have tradionally been used as an alternative to equy or debt financing by firms wh high equy- or debt-related financing costs (Green, 1984; Brennan and Kraus, 1987; Brennan and Schwartz, 1988; Stein, 1992). Investors in these securies typically purchase convertibles as part of a balanced portfolio. At the turn of the century, however, there was a shift in the investor base, wh convertible arbrage hedge funds purchasing most of the new issues. These arbrageurs adopt a variety of strategies that target specific aspects of convertible bonds, while hedging away other 20

22 risks, thus reducing information asymmetries and enabling firms to quickly issue the convertible. In this paper, we examine how the characteristics of issuers change in reaction to the new source of demand from convertible bond arbrageurs. We show that the shift in the investor base, from tradionally long-only investors to convertible arbrage hedge funds presented opportunies for a wider variety of issuer to tap the primary market for convertibles. The characteristics of issuers suggest that a significant portion of firms issuing convertibles could have otherwise issued straight debt. Hence, opportunistic behavior is not limed to riskier firms that would be financially constrained in normal condions. This is in spe of the fact that arbrageurs are less concerned wh firmspecific risk, since they can hedge away. The design of offerings also changes, wh firms increasing delta, in line wh the idea that they cater to hedge fund preferences. We also show that the design is altered at the same time that is less sensive to pricing, allowing issuers to obtain cheaper funding when catering. Our results contribute to a recent strand of lerature that examines how changes in the supply of funds influence corporate finance decisions and asset prices (see, e.g. Baker, 2009). 21

23 References Aghion, P., Stein. J., Growth vs. Margins: Destabilizing Consequences of Giving the Stock Market What Wants. Journal of Finance, 63, Ammann, M., A. Kind, and C. Wilde, 2003, Are convertible bonds underpriced? An analysis of the French market, Journal of Banking and Finance 27, Baker, M., Capal Market-Driven Corporate Finance. Annual Review of Financial Economics 1, Baker, M., Greenwood, R., Wurgler, J., Catering through Nominal Share Prices. Journal of Finance 64, Baker, M., Wurgler, J., A Catering Theory of Dividends. Journal of Finance 59, Black, F., Scholes, M., 1973, The pricing of options and corporate liabilies. The Journal of Polical Economy 81, Brennan, M., Kraus, A., Efficient financing under asymmetric information. The Journal of Finance 42, Brennan, M., Schwartz, E., The case for convertibles. Journal of Applied Corporate Finance 1, Brown, S., Grundy, B., Lewis, C., Verwijmeren, P., Convertibles and hedge funds as distributors of equy exposure. Review of Financial Studies 25, Brown, S.J., Warner, J.B., 1985, Using daily stock returns: The case of event studies, Journal of Financial Economics 14, Burlacu, R., New evidence on the pecking order hypothesis: The case of French convertible bonds. Journal of Multinational Financial Management 10, Calamos, N., 2003, Convertible Aribtrage: Insights and Techniques for Successful Hedging. John Wiley & Sons, Hoboken, New Jersey. Choe, H., Masulis, R., Nanda, V., Common stock offerings across the business cycle: theory and evidence. Journal of Empirical Finance 1, Choi, D., Getmansky, M., Henderson, B., Tookes, H., Convertible bond arbrageurs as suppliers of capal. Working Paper (Yale International Center for Finance). 22

24 Choi, D., Getmansky, M., Tookes, H., Convertible bond arbrage, liquidy externalies and stock prices. Journal of Financial Economics, 2, De Jong, A., Duca, E., Dutordoir, M., Do Convertible Bond Issuers Cater to Investor Demand?, Financial Management 42, De Jong, A., Dutordoir, M., Verwijmeren, P., Why do convertible issuers simultaneously repurchase stock? An arbrage-based explanation. Journal of Financial Economics 100, Dinsmore, T., Dinsmore, J., Finnican, P., A review of the convertible securies market. Dinsmore Capal Management. Duca, E., Dutordoir, M., Veld, C., Verwijmeren P., Why Are Convertible Bond Announcements Associated wh Increasingly Negative Abnormal Stock Returns? An Arbrage-Based Explanation, Journal of Banking and Finance 36, Dutordoir, M., Van de Gucht, L., Are there windows of opportuny for convertible debt issuance? Evidence for Western Europe. Journal of Banking and Finance 31, Eckbo, B., Masulis, R., Seasoned Equy Offerings: A Survey. In R. Jarrow, V. Maksimovic, and B. Ziemba (Eds.), Finance: Series of Handbooks in Operations Research and Management Science. North-Holland, Amsterdam, Green, R., Investment incentives, debt, and warrants. Journal of Financial Economics 13, Hoffmeister, J., Hays, P., Kelley, G., Condions affecting the timing of convertible bond sales. Journal of Business Research 15, Huang, R., Ramirez, G.G., Speed of issuance, lender specialization, and the rise of the 144A debt market. Financial Management 39, Hughes, J., An appete for convertibles. Financial Times, June 18. Kang, J. K., Lee, Y. W The pricing of convertible debt offerings. Journal of Financial Economics 41, King, R Convertible Bond Valuation: An Empirical Test. The Journal of Financial Research, 9, pp Kim, Y. O., Informative Conversion Ratios: A Signalling Approach." Journal of Financial and Quantative Analysis, 25 (June 1990),

25 Kim, Y. O., Stulz, R., Is there a global market for convertible bonds? Journal of Business 65, Korajczyk, R., Levy, A., Capal structure choice: macroeconomic condions and financial constraints. Journal of Financial Economics 68, Krishnaswami, S., Yaman, D., The role of convertible bonds in alleviating contracting costs. Quarterly Review of Economics and Finance 48, Lewis, C.M., Rogalski, R.J., Seward, J.K Is convertible debt a substute for straight debt or for common equy? Financial Management 28, Lewis, C.M., Rogalski, R.J., Seward, J.K, Industry condions, growth opportunies and market reactions to convertible debt financing decisions. Journal of Banking and Finance 27, Lewis, C.M., R.J. Rogalski, and J.K. Seward, 1998, Agency Problems, Information Asymmetries, and Convertible Debt Secury Design, Journal of Financial Intermediation 7, Liu. J., Swzer, L.N., Liquidy risk, firm risk, and issue risk premium effects on the abnormal returns to new issues of convertible bonds. Working Paper. Concordia Universy. Loncarski, I., ter Horst, J., Veld. C., Why do companies issue convertible bonds? An empirical analysis for the Canadian market. Canadian Journal of Administrative Sciences 25, Loncarski, I., Ter Horst, J., Veld, C., The rise and demise of the convertible arbrage strategy. Financial Analysts Journal, 65(5), Lowry, M., Why does IPO volume fluctuate so much? Journal of Financial Economics 67, Lucas, D., McDonald. R., Equy issues and stock price dynamics. Journal of Finance 45, Lummer, S., Riepe M., Convertible bonds as an asset class; Journal of Fixed Income 3, Manconi, A., Massa, M., A Servant to Many Masters: Competing Shareholder Preferences and Lims to Catering. Insead Working Paper. Mann, S., Moore, W., Ramanlal, P., Timing of convertible debt issues. Journal of Business Research 45,

26 Mayers, D., Why firms issue convertible bonds: the matching of financial and real investment options. Journal of Financial Economics 47, Merton, R., Theory of rational option pricing. Bell Journal of Economics and Management Science 4, Mchell, M., Pedersen, L. Pulvino, T., Slow moving capal. American Economic Review 97, Monnelly, M., The rise and fall of convertible arbs. Euromoney. November Myers, S.C., Majluf, N.S., Corporate financing and investment decisions when firms have information that investors do not have. Journal of Financial Economics 13, Pernambuco, A., Consider convertible arbrage to balance performance, risk. Financial Executive. May Polk, C., Sapienza, P., The Stock Market and Corporate Investment: A Test of Catering Theory. Review of Financial Studies 22, Pulliam, S., Mixed blessing: How hedge-fund trading sent a company s stock on wild ride. Wall Street Journal December 28, A1. Ranaldo, A., Eckmann A., Convertible bonds: characteristics of an asset class. Working Paper (UBS). Stein, J., Convertible bonds as backdoor equy financing. Journal of Financial Economics 32, Wang, S Convertibles in sequential financing. Review of Finance 13,

27 Appendix A: Definion of firm-specific, issue-specific, and macroeconomic variables A.1. Firm-specific variables We include the following standard firm-specific financing costs proxies (see, e.g., Lewis, Rogalski, and Seward, 1999, 2003; Krishnaswami and Yaman, 2008). Stock price data are obtained from CRSP. Trading days are measured relative to the convertible bond announcement date. Firm characteristics are retrieved from the Compustat Fundamentals Annual database, and measured at the end of the fiscal year prior to the offering, unless mentioned otherwise. The symbol # denotes a Compustat data em: Stock Run-up: The stock return over the window ( 60, 2) relative to the issue date. As shown by Lucas and McDonald (1990), a pre-announcement stock run-up can reflect lower equy-related adverse selection costs. Slack: Cash and short-term investments (# 1) divided by total assets (# 6). Financial slack acts as a measure for adverse selection costs, since firms wh higher slack could engage in wasteful use of resources. Tax: Captures the tax liabilies benef associated wh issuing debt and is computed as income tax (# 16) divided by total assets. MB: Market to book value, calculated as the number of shares (# 24) multiplied by the share price (# 25), divided by the book value of common equy (# 60). This variable can proxy for growth opportunies, but may also capture asymmetric information and/or underinvestment problems. LogAssets: Calculated as the natural logarhm of total assets (# 6). Larger firms are assumed to face smaller information asymmetries regarding their value and risk. 26

28 Leverage: Long-term debt (# 9) divided by total assets. Higher leverage is assumed to be associated wh higher financial distress and risk-shifting costs. Stock Volat: The volatily in the stock price prior to issue, calculated over days -240 to -40, and expressed in annualized terms. Issued SD previously: A dummy variable equal to one if the firm issued straight debt at least once over the past three years prior to the convertible bond issue. Issued EQ previously: A dummy variable equal to one if the firm issued equy at least once over the past three years prior to the convertible bond issue. Probabily issue equy: A dummy variable that takes the value of one if the fted probabily of equy (defined in Appendix A6) is greater than 0.5. A.2. Issue-specific variables Underpricing: The underpricing of the convertible bond, as defined in Appendix A4 Delta: The delta of the convertible bond, as defined in Appendix A5 Matury: The years to matury of the convertible, obtained fro SDC ConvPremium: The conversion premium of the convertible, defined as the conversion price minus the issue price, expressed relative to the issue price. Proceeds: The relative size of the convertible debt offering, calculated as the issue proceeds (obtained from SDC) divided by the total assets of the firm (# 6). In line wh Lewis, Rogalski, and Seward (1999, 2003), and Dutordoir and Van de Gucht, 2007, we expect larger offerings to induce more negative announcement returns due to higher external financing costs. 27

29 Announce. Return: The percentage abnormal stock returns measured over the window (-1, 0) relative to the announcement date using standard event study methodology. Rating: The bond rating of each issue as assigned by Moody s, or the equivalent Standard and Poor s ratings converted to a Moody s rating (both obtained from SDC). We assign a rating of BAA2 to unrated convertibles, as in Loncarski, ter Horst, and Veld (2009). Not rated: A dummy variable equal to one if the bond is not rated. A.3. Macroeconomic Variables 10 Next to the proxies for investor demand, we include several widely-used macroeconomic variables to control for intertemporal variations in general debt- and equy-related contracting costs. All variables are obtained from Datastream. The real interest rate (Rate) serves as a proxy for bankruptcy risk, as in Krishnaswami and Yaman (2008). This variable is calculated as the difference between yields on 10-year U.S. Treasury Bonds and the inflation rate, defined as the continuously-compounded annual change in the U.S. consumer price index. Following Korajczyk and Levy (2003), we control for business condions and expected investment opportunies by including the term spread (Term Spread), defined as the difference between yields on 10-year U.S. Treasury Bonds and 3-month U.S. Treasury Bills. Both the real interest rate and the term spread are averaged over the 3 months prior to the convertible bond offering. The market run-up (Mkt Run-up), calculated as the return on the S&P 500 index over the quarter preceding the issue, is included to control for general market condions as in Korajczyk 10 The variable Hedge fund flows is described in Appendix A7. 28