When do firms issue exchangeable debt?
Exchangeable debt is bonds issued by one company that are convertible into common shares of a second company (hereafter the underlying firm) in which the issuing firm has a stake. Ghosh et al. (1990) and Barber (1993) propose that exchangeable debt is a way for the issuer to dispose of a block in the underlying firm. Several hypotheses have been put forth to explain what makes exchangeable debt a better alternative to a direct stock sale. Barber (1993) notes that exchangeable debt issuers often mention tax advantages (dividend tax arbitrage (1) and deferral of capital gains realization) as a motivating factor, but argues that tax incentives do not appear to drive the decision to issue this type of instrument. Both Barber (1993) and Gentry and Schizer (2002) examine whether firms issue exchangeable debt in part because it is less costly to do so than to conduct a secondary offering of a large block of stock, and find mixed support for this hypothesis. To shed more light on the growing use of exchangeables, this study investigates the relevance of information-based market timing in the issuance of exchangeable debt offerings.
Starting with the seminal study of Myers and Majluf (1984), there has been an ongoing stream of papers that suggest and provide evidence that asymmetric information can be exploited to market-time security issues. Myers and Majluf suggest that managers time security issues to take advantage of the information disparity between themselves and outside investors, and they predict that managers will be inclined to issue equity or convertible debt when common stock is overvalued. In an anonymous survey conducted by Graham and Harvey (2001), executives listed the market price of equity as the most important factor in the decision to issue equity and equity-linked debt. Baker and Wurgler (2002) suggest that corporate capital structure is a cumulative outcome of equity market timing by managers.
Recent studies question these findings. Korajczyk and Levy (2003) show that only financially unconstrained firms time security issues when macroeconomic conditions are favorable. Schultz (2003) suggests that apparent evidence on market timing is an artifact of another process and what appears to be market timing is what he calls pseudo market timing and an artifact of issuing equity when stock prices go up. Even if managers do not have any special market timing ability, it appears as if they do simply because more equity is issued when valuations are high. Butler, Grullon, and Weston (2005) argue that the proportion of equity in total new issues is the result of aggregate pseudo market timing.
In this paper we ask if managers can use asymmetric information in their decision to issue exchangeable debt. We hypothesize that issuers with unfavorable information about the future value of the underlying firm use exchangeable debt to profit from that information. Large (25.4%) initial stakes in underlying firms coupled with the history of stake formation suggest that the issuers have incentives and opportunities to be well-informed about the future prospects of underlying firms. (2)
The study makes three primary contributions. We re-examine price reaction at the announcement of exchangeable debt for a larger sample for both issuing firms and the underlying firms. We conduct a multiple regression analysis in order to examine the ability of managers to time exchangeable debt issues based on information they possess as large shareholders. Finally, we investigate whether a firm signals its intent to conduct an open market sale of the underlying shares by issuing exchangeable debt.
We build a dataset that includes 104 issues of exchangeable debt by corporate issuers between 1981 and 2001. In line with the previous literature, we find no abnormal issuer price reaction at the announcement of exchangeable debt. This could imply that the market expected the issuers to be well informed about the prospects of the underlying firms. For underlying firms, the abnormal returns at the announcement are significantly negative, corroborating previous studies. We find evidence in support of the market timing explanation for the occurrence of exchangeable debt announcements. The dollar amount of the exchangeable debt offer (scaled by total liabilities) is influenced by the underlying stock price run-up and the negative future earnings of the underlying firm even after holding constant the size and investment opportunities of the issuer. Furthermore, it is possible to predict an asset sale of the underlying stake in the open market subsequent to an exchangeable debt issue. The probability of the stock sale after an exchangeable debt issue increases with the dollar size of the issue, controlling for pre- and post-conversion stakes in the underlying firm and decreases with the level of institutional ownership.
This study is related to a recent study by Danielova et al. (2010), which studies the market timing ability of firms issuing exchangeable debt by recognizing that most exchangeable debt issuers have several stocks which could be chosen as the underlying asset for the offering and testing for differences between assets that are chosen and those that are not chosen as the underlying security. While reaching similar conclusions regarding the ability of insiders to time exchangeable debt offering based on proprietary information, this paper uses a different methodology. Whereas Danielova et al. track and compare the long-term operating and stock performance of corporate holdings chosen as the underlying stock and those that are not chosen, we look at the sample characteristics and circumstances surrounding the issue of exchangeable debt. We examine the market reaction of the issuer and underlying firms at the announcement of the exchangeable debt. In addition we shed more light on the tax advantage motivation behind exchangeable debt issues, as well as provide evidence on whether exchangeable debt issues signal the prospective open market disposition of the underlying shares by issuers.
Literature Review and Hypotheses Development
Several studies show that manager's time security issues and repurchases to take advantage of misvaluation in the market. Ritter (1991) and Loughran and Ritter (1995) argue that firms issue equity when they are overvalued. Conversely, Ikenberry, Lakonishok and Vermaelen (1995) argue that firms repurchase equity when they are undervalued. Spiess and Affleck-Graves (1999), Lee and Loughran (1998), and Dichev and Piotroski (1997) show that firms tend to issue convertible debt when they perceive their stock to be overvalued. (3) Exchangeable debt issuances provide an additional setting to test market timing. This study investigates whether managers can exploit temporary fluctuations in the value of the underlying stock when making an exchangeable debt offering.
When the issuer has inside adverse information about the true value of the underlying firm, this creates an incentive to unload the underlying shares expeditiously. The issuer's choices are to sell the underlying stock directly or to issue some form of exchangeable debt. The cost-saving benefits of delaying the capital gains tax, inherent in the exchangeable debt structure, could make exchangeable debt a better alternative to the direct sale of the underlying stake. These tax deferral benefits are more valuable if the underlying stock experienced a significant run-up in its price relative to the firm's acquisition price for the underlying shares. Better informed issuers time exchangeable debt offers to exploit a recent run-up in the price of the underlying stock before it reverses. The poorer the prospects of the underlying firm, the more shares the issuer wants to dispose of before this information is revealed to the market, and the larger is the stake that is monetized via an exchangeable debt offering. Thus, we obtain the first testable implication:
Hypothesis 1. The amount of exchangeable debt issued should be larger for issuers having superior information about the deteriorating underlying firm value and should increase to the extent of temporary overvaluation of the underlying stock.
If firms use exchangeable debt to take advantage of temporary price fluctuations, and may sell the underlying shares later, then exchangeable debt may signal the issuer's intent to sell the underlying stock. A greater degree of asymmetric information about the future prospects of the underlying stock thus increases the chances of selling the underlying stock. Furthermore, larger issue size and greater stake reduction via exchangeable debt suggest that the issuer has more unfavorable inside information and thus a greater incentive to sell the underlying asset. Therefore we test the following hypothesis:
Hypothesis 2. The likelihood of an open market sale of the underlying stock following an exchangeable debt issue increases with the degree of asymmetric information, increases as the market reaction of the underlying stock to the announcement becomes more negative, increases with the size of the exchangeable debt issue after controlling for stake reduction, and decreases with the run-up in the underlying stock prior to the announcement of the exchangeable debt (exchangeable debt becomes the preferred alternative due to possibility of tax deferral).
Data and Sample Characteristics
We were able to identify 104 transactions covering the period 1981 through 2001 that satisfy the following selection criteria: the exchangeable security payoff issued by the issuer depends on the stock of the underlying company; the underlying stock is publicly traded; the payoff can be made in the shares of underlying stock or in cash tied to the performance of underlying stock; the issuer has an ownership stake in the underlying stock prior to the announcement of the exchangeable issue; and the exchangeable debt is not used as payment to the underlying company in a merger.
We used several sources to identify exchangeable debt offerings. The original list of transactions came from the Security Data Company's Corporate New Issues database (SDC Platinum) for the period 1981 through December 2001. From this list we searched Lexis-Nexis, Dow Jones News Retrieval, and the Edgar database of SEC filings to retrieve news stories and prospectuses pertaining to these deals and to verify the details of each deal. The transactions that did not meet the above listed criteria were eliminated. This list was cross checked and augmented by the exchangeable transactions shared with us by Salomon Smith Barney and JP Morgan Convertible Research Groups. The rest of the sample came from the general search conducted using Lexis-Nexis, Dow Jones Interactive, SEC filings, and search engines on the Internet. The accuracy of data was verified by comparing them to SEC filings and public announcements of exchangeable offerings in the Wall Street Journal and news wires.
Further inclusion criteria required that accounting data and returns for the issuer and the underlying company be available on Compustat and CRSP. Applying these criteria resulted in a sample of 94 issuers and 97 underlying firms. There are 70 different firms represented in the issuers sample. Seventeen firms made two or more exchangeable debt announcements. The underlying sample has 78 different firms. Exchangeable debt was issued on fourteen of them more than once. Several issuers issued exchangeable debt on two (or in the case of Liberty Media, three) underlying firms. We based test statistics on the assumption that the observations are independent.
Table 1 contains the year-by-year distribution of the full sample of 104 announcements of exchangeable debt offerings spanning the 1981-2001 period. The 1990s feature prominently in the distribution of offerings. Exchangeable debt deals numbered in the single digits during the 1980s and the early 1990s, but numbered in the double digits in 1995, 1996, and 1999. Unprecedented spikes in the dollar amount of principal raised with exchangeable offerings are visible for 1997, 1998, and 1999.
In ninety-eight cases the issuer is a publicly traded firm. In six cases in the sample, however, an investment bank (Merrill Lynch in five cases and Salomon in one) offered its balance sheet for the transaction to a party that could not carry exchangeable debt on its balance sheet for regulatory reasons. (4,5)
Table 2 presents ex-ante characteristics of exchangeable debt offerings by corporate issuers during the period 1981-2001. The average amount raised by exchangeable debt is about $415 million. Average exchangeable debt issue has maturity of 13.9 years and 6.18 percent coupon. (6) Before their exchangeable debt announcements, issuers owned a 25.4% stake in their underlying firms on average; following conversion of the exchangeable debt those stakes would be reduced to an average of 9.39%.
Table 3 presents the origins of stake formation in the underlying firms. In 17 cases the issuing company retained some position in the underlying company after a public offering of its subsidiary stock in the equity carve-out. In 28 cases the issuers' stakes were formed as a result of a merger, aborted takeover attempt, or a process of exchange offers. Finally, in 19 cases stakes were the result of corporate investments. For the remaining 40 firms the origins of the stake in the underlying companies are unclear.
To investigate the possibility for information-based market timing, we collect additional data pertaining to exchangeable debt offers. In particular, we examine whether the issuer is required to hold the underlying shares to secure exchangeable debt and whether there are sales of the underlying stock by the issuers while the issue was outstanding. Not being obligated to hold the underlying shares in escrow allows companies to issue exchangeable debt based on the value of temporarily overpriced underlying shares and then dispose of them at their discretion. We also include cases when the issue was recalled in order to free shares from escrow to sell them.
Table 4 shows that about 45 percent of the issuers used open market operations to sell the underlying stock while the issue was still outstanding. This pattern confirms Barber's (1993) suggestion that firms issue exchangeable debt conditional on having made a decision to dispose of underlying shares. It seems that at the time of the exchangeable debt announcement, firms favor exchangeable debt over direct stock sales. This could be because issuers time exchangeable debt offerings while the underlying stock is overvalued, before disposing of it in a traditional way. Twenty-eight firms in the sample had to put underlying shares into escrow, while 35 did not have such a requirement. No information was available for the remaining 41 firms.
Evidence on Stock Price Effects of Exchangeable Debt Offers Announcements
Ghosh, Varma, and Woolridge (1990) find no abnormal returns at the announcement for firms issuing exchangeable debt, whereas they find significantly negative (-1.11%) announcement-period abnormal returns for underlying firms in 36 exchangeable offers between 1969 and 1987. Barber (1993) analyzes 37 exchangeable debt offerings over the period 1970 through 1987, and obtains similar results. Gentry and Schizer (2002) document an abnormal announcement return of -1.03% in the underlying stock and an abnormal return of -2.80% in the underlying stock on the issue date for exchangeable debt transactions offered between 1992 and 2000. Neither announcement nor execution of exchangeable debt transactions has a statistically significant effect on the issuer's stock price. Because these results were found using smaller sample sizes and different time periods, we begin by confirming these results using a larger sample and longer time horizon.
We define the announcement date to be the earlier of the first report of the exchangeable debt offer in the Wall Street Journal or major news wire, or the date the issue was registered with the Securities and Exchange Commission. We use a 4-day event window due to the fact that for a considerable share of the sample, the SEC filing date was used as the date of the announcement. (7) For each firm on each day of the event window, we estimate abnormal returns by calculating the daily difference between the actual return of the event window and an expected "normal" return. To better control for time-varying risk premium, expected returns are defined as the forecasts from a Fama-French three-factor market model conditional on the contemporaneous return on the market index, a high-minus-low (HML) market-to-book ratio factor, and a small-minus-big (SMB) market capitalization factor. Parameters of the model are estimated for each firm over a 200-day estimation window, ending 61 days prior to the event. The CRSP NYSE/AMEX/NASDAQ equal and value-weighted market indexes are used as market proxies. We report results based on both market proxies. As a robustness check we also conduct an event study using Fama-French calendar time portfolio regressions. A security is assigned to a calendar-date portfolio if the calendar date falls in a desired window relative to the security's literal event date. (8)
Table 5.1 shows the average stock price responses surrounding announcements of exchangeable debt. The average issuer's stock price response at the announcement is not significantly different from zero. In contrast, the underlying stock demonstrates a highly significant price decline of 2.69% (2.58%) at the announcement of the exchangeable debt issue using an equal-weighted (value-weighted) market index. The underlying firms' price response is -2.74% (-2.65%) using the Fama-French calendar time approach in Table 5.2. These results are consistent with the results previously documented in the literature. We also document a significant run-up in price for the preannouncement period and a deep, about 31%, decline in stock value, relative to "normal" expected return (defined as the forecasts from a Fama-French three-factor market model), for the year after the announcement for the underlying stock. This result substantiates the prediction of the market-timing hypothesis.
To summarize, our market-timing hypothesis has direct implications for the stock price reaction and pre- and post-announcement stock price patterns. In particular, it suggests that issuers time the exchangeable debt offerings when they believe that the underlying firm is overvalued. Consistent with that we find significantly negative abnormal returns for the underlying firms. We also document significant post-announcement price declines in the value of the underlying stock. The next section provides cross sectional evidence on the market-timing hypothesis.
Cross Sectional Determinants of Market Timing of Exchangeable Debt Issues
In this section we test whether the extent of adverse information about the prospects of the underlying firm and the temporary overvaluation in the underlying stock price are related to the level of proceeds raised by the exchangeable debt issues.
The main dependent variable is the size of the exchangeable debt issue in dollar terms. We use two deflators for the dependent variable. In the first set of regressions, we express the size of the exchangeable debt issue as a percentage of the firm's total debt at the fiscal year end prior to the announcement. This allows us to compare the importance of the funds raised through the new exchangeable debt issue to the issuer's total previous debt. In the second regression set we look at the ratio of exchangeable debt to the sum of exchangeable debt raised and the total debt prior to the announcement. This measure reflects the post-issue contribution of the exchangeable debt to the total liabilities. For an average issuer, the prospective exchangeable debt issue accounts for 74% of total debt prior the issue, and for about 21% of total liabilities after the issue is made. The amount of exchangeable debt raised equals 14% of the market cap of the issuer.
To proceed we need to identify variables that proxy for the asymmetric information and temporary overvaluation of the underlying stock. The financial literature does not offer a universally accepted proxy for asymmetric information. Our hypotheses assume that issuers have better information about the quality of the underlying firm than do investors. To measure the quality of the underlying firm empirically, we follow Barclay and Smith (1995) and use the firm's abnormal future earnings. We assume that high quality underlying firms, undervalued at the time of the exchangeable debt announcement, have positive future abnormal earnings, and low quality underlying firms, overvalued at the announcement, have negative future abnormal earnings. We define abnormal earnings at year t+2 as earnings per share in year t+2, excluding extraordinary items (Compustat annual data item 58), minus earnings per share in year t, both divided by the share price in year t (Compustat annual data item 199). For robustness we also use a run-down measure, calculated as a cumulative return on the underlying stock for two years starting one year after the announcement. We also control for the level of institutional ownership in the underlying firm, since some studies have found it to be related to the degree of information asymmetry. For example, O'Brien and Bhushan (1990) suggest that the level of institutional ownership and analysts' coverage are closely associated with each other. Firms with greater analysts' coverage undergo greater scrutiny. A higher level of institutional holdings implies a lower degree of asymmetric information about the underlying company, all else equal. Since larger firms are more likely to be followed by analysts, (9) we also control for the underlying firm size.
To assess the extent of recent overvaluation of the underlying stock we include a measure of the appreciation in the underlying firm's stock price. We define it as a cumulative return on the underlying stock for the two years ending three months before the announcement. We use a two-year cumulative return to better reflect the fact that issuers have been holding stakes in the underlying firms for several years. In addition to being a measure of temporary overvaluation, this price run-up measure also captures the magnitude of prospective capital gains taxes. The grater the run up in the price of the underlying shares, the larger the capital gains tax to be paid if underlying shares are sold. The increase in the present value of delayed tax savings offered by exchangeable debt makes it a better alternative than a direct stock sale.
The marginal tax rate is another independent variable included in the model. Unlike individuals who face lower tax rates on capital gains income than on ordinary income, U.S. corporations do not receive preferential tax rates on realized capital gains. Net realized capital gains are added to ordinary income in computing the firm's taxable income (Desai and Gentry, 2003). The key variation in effective tax rates arises due to the rules related to operating losses. Due to the impossibility of inferring precisely the tax position of a firm from public financial documents, (10) similar to Desai and Gentry we employ two marginal tax measures devised by Graham (1996) that identify probabilities of having net operating losses: the first is based on income before interest expense, and the second is based on income after interest expense, (11) The tax system encourages firms facing higher effective marginal tax rates to issue more exchangeable debt that pays tax-deductible interest and to minimize disposition of investments. Using Graham's (1996) proxies for the marginal tax rate and controlling for firm characteristics and time-varying investment opportunities, Desai and Gentry found that the sales of investments are less likely and considerably smaller in high-tax years.
As a robustness check we also use two simpler tax rate calculation methods as in Welch (2002). The first is the ratio of income taxes paid (Compustat annual item 16 or 317) to total assets (Compustat annual item 6). The second is the ratio of total income taxes (16 or 317), divided by earnings plus total income taxes (*+).
To isolate the effects of market timing on exchangeable debt levels we employ two control measures to guard against the possibility that levels of exchangeable debt are related to issuers' investment and financing activities. Although firm size is one of an organization's endogenous choices, we believe that it is important to control for the issuer's size in our regression specifications because it is a function of past investment opportunities and may also affect corporate capital structure choices. Because large firms should have a comparative advantage in issuing securities publicly, this implies that, all else equal, larger firms would have less exchangeable debt in their capital structures. Firm size is measured as the natural logarithm of the dollar market value of the firm.
Smith and Watts (1992) find that book-to-market is significantly related to the firms' investment policy choices. Myers (1977) suggests that firms with substantial growth and investment opportunities have the most to lose because of costly financial distress, and thus should have less debt in their capital structure. Thus, book-to-market of the issuer is included as a control variable in the model. The book-to-market ratio is calculated as the book value of assets divided by the estimated market value of assets. We define the market value of the firm's assets as the sum of market value of equity (*), book value of long-term debt (9), book value of current debt (34), and book value of preferred stock (130). The book-to-market measure for issuers and the levels of exchangeable debt are expected to be positively associated.
To separate out mispricing and investment opportunities perceived by issuers to exist in the underlying firm, we control for current cross-sectional variation in the level of book-to-market of the underlying firm. Chan and Chen (1991) propose that low stock prices and high book-to-market ratios are signals for firms that the market judges as having poor prospects. Fama and French (1992) suggest that book-to-market captures cross-sectional variation in average returns that is related to relative distress. Companies with higher levels of book-to-market could be judged by the market as having a poor current investment outlook.
The market price of exchangeable debt can be influenced by the volatility of the underlying stock because the variance of returns is likely to affect the value of the underlying option embedded in the exchangeable debt structure. Thus, we include the variance of underlying stock as a control, computing it as the simple variance of daily returns over the twelve months preceding the announcements of exchangeable debt issues and ending three months before the announcements, using CRSP data. In addition, the volatility of stock returns controls for the differences in risk among the underlying firms.
The regressions in Table 6, Panels A and B, are ordinary least squares cross-sectional regressions with robust standard errors clustered at the year level. For all of the regression specifications, the [R.sup.2] values are quite high, ranging from 54% to 61%.
Consistent with the first hypothesis, there is a positive relationship between the magnitude of the underlying stock price appreciation and the level of exchangeable debt issued. This relationship is significant across all regression specifications. Marginal tax rates, (12) on the other hand, are statistically insignificant across all regression specifications. The significantly positive price run-up measure coefficient is consistent with both an overvaluation hypothesis and a tax advantage hypothesis. However, the insignificance of the marginal tax rate coefficient diminishes the support for the tax advantage hypothesis. Thus, our results are consistent with Barber's (1993), suggestion that capital gains tax is not a driving force for the issuance of exchangeable debt. In addition, the insignificance of the marginal tax coefficient based on income after interest expense, suggests that dividend tax arbitrage does not hold up under closer scrutiny, similar to Barber.
Consistent with the first hypothesis, there is a strong significant relationship between the private information issuers have about the negative prospects of the underlying firm and the level of exchangeable debt. The abnormal earnings variable, which proxes for the quality of the underlying firm, has a negative sign in all regression specifications in Panel A. The results using the run-down variable in Panel B are qualitatively similar and statistically more significant. The more negative the private information issuers have, the more cash they raise with exchangeable debt issues. This relationship is significant across all regression specifications. This is an important finding because it contributes to the existing disagreement in the literature about whether the issuers have negative information regarding the future prospects of their underlying firms. Our results provide an empirical test of Barber's (1993) suggestion that firms issue exchangeable debt when they have negative information about underlying firms. Contrary to our results, Ghosh et al. (1990) find no support for the negative information explanation.
A positive relationship between the magnitude of the underlying stock price appreciation and the level of exchangeable debt issued coupled with a negative relationship between two proxies for the quality of the underlying firm and the amount of exchangeable debt provides support for the market-timing hypothesis (that issuers time exchangeable debt issues to profit from their information about the underlying firm) instead of pseudo market timing (issuing more exchangeable debt when underlying stock prices appreciated). Our results are also in line with Danielova et al. (2010) finding that exchangeable debt issuers choose underlying firms based on their superior past performance, but that performance deteriorates after the issue.
Several control variables are significant with coefficients of the expected sign. Size of the issuer has a negative effect on the magnitude of exchangeable debt issues. The issuer's book-to-market has a positive effect. Size, book-to-market, and volatility of the underlying firm lack significance in most regression specifications.
Predictability of Stock Sale Following the Exchangeable Debt Announcement
If firms issue exchangeable debt to take advantage of temporary price fluctuations when they possess private adverse information about the underlying firm's value, then exchangeable debt may signal the issuer's intent to sell the underlying stock later on. In Table 7 we report logistic regression results demonstrating what factors influence the likelihood of an open market sale of the underlying stock following an exchangeable debt issue. The dependent variable is 1 if the issuer sells at least some part of its stake in the underlying company after the exchangeable debt issue was announced and while the issue has been outstanding, or if the issue was retired in order to release underlying shares from escrow in order to sell them. It takes on a value of 0 if no direct market sale was made while the issue was outstanding.
Consistent with Hypothesis 2, the coefficient for the size of the exchangeable debt issue is significant and has the predicted positive sign. The positive and significant coefficient of the post-conversion stake in the underlying firm suggests that the issuer is more inclined to shed further its stake in the underlying firm if the exchangeable debt was issued on a smaller portion of the initial stake in the underlying firm. Firms also have a greater tendency to dispose of underlying stock when underlying firms have lower levels of institutional ownership.
The likelihood of an underlying stock sale is negatively associated with the private information that issuers have about the prospects of the underlying firm. Greater negative abnormal earnings increase the likelihood of stock disposition. The coefficient for run-up in the underlying stock price prior to the exchangeable debt announcement is negative, but not significant.
The two specifications of the Logit model produce close to 87 percent of the association between predicted probabilities and observed responses, suggesting that it does a fairly good job of revealing issuers' intentions to conduct an open market sale after the exchangeable debt offer.
In summary, by issuing exchangeable debt a firm signals its intent to dispose of the underlying shares on the open market. An open-market stock sale subsequent to the exchangeable debt issue is more likely for larger dollar-volume issues, a smaller pre-conversion stake in the underlying firm, a larger post-conversion stake in the underlying firm, and, if the underlying firm has greater level of asymmetric information, smaller percentage of institutional investors.
The current evidence on the reasons for issuing exchangeable debt is scant and at best inconclusive. Although several hypotheses have been advanced, none have been thoroughly investigated. This article provides an initial empirical investigation of the different explanations to use an exchangeable debt. We believe that exchangeable debt issuers take advantage of transitory windows of opportunity when they perceive the underlying stock to be overvalued.
Our results show that the driving force behind the market-timing behavior of issuers is their perception of temporary mispricing of the associated underlying stock based on information they possess as large shareholders. The reported evidence offers consistent support for the hypothesis that exchangeable debt is a resourceful method of fully exploiting temporary fluctuations in the value of underlying stock by the issuers with unfavorable information about the value of the underlying firm. In particular, the results of this paper indicate that there is a strong relationship between the relative amount of exchangeable debt raised and the measures of asymmetric information and temporary overvaluation in the underlying stock. Finally, we document the conditions under which the exchangeable debt issue signals the prospective open market sale of the underlying shares.
I would like to thank anonymous referees, John Boquist, Robert Jennings, Richard Rosen, Charles Trzcinka, Greg Udell, Andrey Ukhov, Xiaoyun Yu, the seminar participants at Indiana University, McMaster University, University of Calgary, University of Manitoba BI, Norwegian School of Management, University of Technology Sydney, McMaster CIBC Chair Symposium on Capital Markets, and the Financial Management Association meeting for their comments and suggestions. A special expression of gratitude goes to Scott Smart for his ongoing support and helpful insights, I thank Charles Trzcinka for sharing the institutional ownership data with me. All errors are mine. I appreciate funding provided by the Social Sciences & Humanities Research Council (SSHRC) of Canada and the Arts Research Board of McMaster University.
Baker, M., and J. Wurgler, "Market Timing and Capital Structure," The Journal of Finance, 57, No. 1 (February 2002), pp. 1-32.
Barber, B., "Exchangeable Debt," Financial Management, 22, No. 1 (Summer 1993), pp. 48-60.
Barclay, M., and C. Smith, Jr., The Priority Structure of Corporate Liabilities," The Journal of Finance, 50, No. 3 (July 1995), pp. 899-917.
Bhushan, R., "Firm Characteristics and Analyst Following," Journal of Accounting and Economics, 11, No. 2-3 (July 1989), pp. 255-274.
Butler, A.W., G. Grullon, and J.P. Weston, "Can Managers Forecast Aggregate Market Returns?" The Journal of Finance, 60, No. 2 (April 2005), pp. 963-986.
Chan, K.C., and N. Chen, "Structural and Return Characteristics of Small and Large Firms," The Journal of Finance, 46, No. 4 (September 1991), pp. 1467-1484.
Collins, D.W., S.P. Kothari, and J.D. Rayburn, "Firm Size and Information Content of Prices with Respect to Earnings," Journal of Accounting and Economics, 9, No. 2 (July 1987), pp. 111-138.
Danielova, A.N., S.B. Smart, and J. Boquist, "What Motivates Exchangeable Debt Offerings?" Journal of Corporate Finance, 16 (2010), pp. 159-169.
Desai, M.A., and W. M. Gentry, "The Character and Determinants of Corporate Capital Gains," NBER, Working Paper 10153, (2003).
Dichev, I., and J. Piotroski, "The Performance of Long-Run Stock Returns Following Issues of Public and Private Debt," University of Michigan, Working Paper, (1997).
Fama, E., "Market Efficiency, Long-term Returns, and Behavioral Finance," The Journal of Financial Economics, 49, No. 3 (September 1998), pp. 283-306.
Fama, E., and K.R. French, "The Cross-Section of Expected Stock Returns," The Journal of Finance, 47, No. 2 (June 1992), pp. 427-465.
Freeman, R., "The Association between Accounting Earnings and Security Returns for Large and Small Firms," Journal of Accounting and Economics, 9, No. 2 (July 1987), pp. 195-228.
Gentry, W., and D. Schizer, "Frictions and Tax-Motivated Hedging: An Empirical Exploration of Publicly-Traded Exchangeable Debt," Columbia University, Working Paper, (2002).
Ghosh, C., R. Varma, and J.R. Woolridge, "An Analysis of Exchangeable Debt Offers," The Journal of Financial Economics, 28, No. 1-2 (November-December 1990), pp. 251-263.
Graham, J.R., "Debt and the Marginal Tax Rate," The Journal of Financial Economics, 41, No. 1 (May 1996), pp. 41-73.
Graham, J., and C. Harvey, "The Theory and Practice of Corporate Finance: Evidence from the Field," The Journal of Financial Economics, 60, No. 2-3 (May 2001), pp. 187-243.
Ikenberry, D., J. Lakonishok, and T. Vermaelen, Market Underreaction to Open Market Share Repurchases," The Journal of Financial Economics, 39, No. 2-3 (October-November 1995), pp. 181-208.
Jones, E.P., and S. Mason, "Equity-Linked Debt," Midland Corporate Finance Journal, (Winter 1986), 47-58.
Korajczyk, R.A., and A. Levy, Capital Structure Choice: Macroeconomic Conditions and Financial Constraints," The Journal of Financial Economics, 68, No. 1 (April 2003), pp. 75-109.
Lee, I., and T. Loughran, "Performance Following Convertible Debt Issuance," Journal of Corporate Finance, 4 (1998), pp. 185-207.
Loughran, T., and J. Ritter, "The New Issues Puzzle," The Journal of Finance, 50, No. 1 (March 1995), pp. 23-51.
Myers, S., Determinants of Corporate Borrowing," The Journal of Financial Economics, 5 (1977), pp. 147-175.
Myers, S., and N. Majluf, "Corporate Financing and Investment Decisions When Firms have Information Investors Do Not Have," The Journal of Financial Economics, 13 (1984), pp. 187-221.
O'Brien, P.C., and R. Bhushan, "Analyst Following and Institutional Ownership," Journal of Accounting Research, 28 (January 1990), pp. 56-76.
Ritter, J.R., "The Long-Run Performance of Initial Public Offerings," The Journal of Finance, 46, No. 1 (March 1991), pp. 3-27.
Schultz, P., "Pseudo Market Tinting and the Long-Run Underperformance of IPOs," The Journal of Finance, 58, No. 2 (April 2003), pp. 483-517.
Shores, R., "The Association between Interim Information and Security Returns Surrounding Earnings Announcements," Journal of Accounting Research, 28, No. 1 (April 1990), pp. 164-181.
Smith, C., and R. Watts, "The Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies," The Journal of Financial Economics, 32, No. 3 (December 1992), pp. 263-292.
Spiess, D.K., and J. Affleck-Graves, "The Long-Run Performance of Stock Returns following Debt Offerings," The Journal of Financial Economics, 54, No. 1 (October 1999), pp. 45-73.
Welch, I., "Columbus' Egg: The Real Determinant of Capital Structure," NBER. Working Paper 8782, (2002).
Anna N. Danielova
(1) The coupon payments to the bondholders are deductible as interest expense while the dividends received on the underlying stock receive a corporate dividend tax exclusion. Jones and Mason (1986) argue that this differing tax treatment effectively lowers the coupon paid by the issuer on exchangeable debt. The differing tax treatments become more pronounced at higher corporate tax rates.
(2) There are four primary avenues for the creation of the large block holding in the underlying firm: the issuer could retain a majority stake in the carved-out former subsidiary; a stake could be formed as part of an exchange offer or former strategic alliance; a stake could be a result of an aborted takeover or inherited with a merger with a third company; finally, a stake could be created as a corporate investment.
(3) Fama (1998) advises that long-term return studies are sensitive to the methodology used, and thus the results of these studies could be controversial.
(4) Individuals and mutual insurance companies are prohibited from issuing public securities. To get around this, Salomon Brothers provided their own balance sheet to Western-Southern Life and issued exchangeable debt in November of 1996. At the same time, the investment bank entered into private swap-like contracts with the insurance company. In another example, Merrill Lynch rented its balance sheet to Bobby G. Stevenson, the chairman and chief executive officer of CIBER, Inc in January of 1998.
(5) To make our sample consistent, in these 6 cases we replaced the investment bank information with the relevant data for the corresponding customer, the actual issuer behind Salomon or Merrill Lynch.
(6) A few exchangeable debt issues with lower coupons also passed the underlying firm's dividend onto exchangeable debt holders.
(7) We have also performed tests using a 3-day event window. Results are similar.
(8) For example, suppose parameters of the model are estimated for each firm over a 189-trading-day estimation window, ending 63 trading days prior to the event. Firms A and B have event dates on January 1, 2007 and January 4, 2007, respectively. Then, firm A's estimation period is January 1, 2006 through September 30, 2006 and B's is January 4, 2006 through October 4, 2006. If we test window (-2, +2), a security belongs to a given calendar day portfolio if the day falls within two days before to two days after the security's day 0. Thus, both A and B belong to the January 2, 2007 and January 3, 2007 calendar time portfolios.
(9) There are numerous studies that suggest that more information is generated for larger firms. Among many are Collins, Kothari, and Rayburn (1987), Freeman (1987). Bhushan (1989) and Shores (1990) note that large firms generally have a higher analyst following.
(10) In addition, financial accounting differs from tax accounting so the measured gain or loss from selling assets differs across accounting systems.
(11) We thank John Graham for making his tax rate variables available via his website at http://www.duke.edu/~graham/. A further discussion of the methodology underlying his tax rate measure is available in Graham (1996).
(12) Regressions with simple tax rate proxies provide similar results and are omitted to save space.
Table 1--Distribution and Proceeds of 104 Announcements of Exchangeable Debt Offerings by Corporate Issuers during the Period 1981-2001 Total Year Number Gross Principal Amount Raised ($ millions) 1981 4 436.00 1982 2 200.00 1983 6 660.00 1984 2 280.00 1985 6 547.00 1986 7 760.00 1987 2 280.00 1988 3 3,100.00 1989 3 399.70 1990 2 2,375.00 1991 3 420.88 1992 5 1,728.73 1993 5 2,144.20 1994 4 1,380.47 1995 12 2,399.40 1996 11 2,158.84 1997 5 5,251.60 1998 4 4,914.52 1999 11 8,211.84 2000 4 3,131.10 2001 3 2,388.10 Total 104 43,167.38 Table 2--Ex ante Characteristics of Exchangeable Debt Offerings by Corporate Issuers during the Period 1981-2001 Number Average Average Average Average Average Size mm$ Stake in Stake in Term Coupon Underlying Underlying of ED on ED Company in Company in % before ED if ED Converted 104 415.08 25.40 9.39 13.90 6.18 Table 3--Origins of the Stake in the Underlying Company for the Sample of 104 Issuers of Exchangeable Debt during the Period 1981-2001 Stake Origin Number Equity carve out 17 Corporate investment 19 Exchange offer or strategic alliance 21 Aborted takeover or part of merger agreement 7 Unclear 40 Table 4--Ex post Characteristics of Exchangeable Debt Offerings by Corporate Issuers during the Period 1981-2001 as of January 2003 Total 104 Stock Sale 46 Shares pledged 28 Shares not pledged 35 No information shared 41 Table 5.1--Abnormal Returns for Issuers and Underlying Firms Surrounding the Announcement of Exchangeable Debt Offers, 1981-2001 For each firm on each day of the event window we estimate abnormal returns by calculating the daily difference between the actual return of the event window and an expected "normal" return. Expected returns are defined as the forecasts from a Fama-French three-factor market model conditional on the contemporaneous return on the market index, a high-minus-low (HML) market-to-book ratio factor, and a small-minus- big (SMB) market capitalization factor. Parameters of the market model are estimated for each firm over a 200-day estimation window, ending 61 days prior to the event. The CRSP NYSE-AMEX-NASDAQ equal/ value-weighted market index is used as a market proxy. t-stats with crude dependence adjustment are given in the second row of each grouping. Statistical significance is indicated for 10% (*), 5% (**), and 1% (***) levels. Panel A Fama-French Model: Equally-Weighted Market Index -60, -2 -22, -2 -1, +2 Issuer 1.61 1.39 -0.14 94 0.967 1.404 -0.328 Underlying -1.23 0.02 -2.69 *** 97 -0.606 0.018 -5.075 +3, +23 +3, +45 +46, +252 Issuer 0.51 0.51 2.63 94 0.512 0.359 0.844 Underlying -0.67 -3.41 ** -31.23 *** 97 -0.556 -1.968 -8.203 Panel B Fama-French Model: Value-Weighted Market Index -60, -2 -22, -2 -1, +2 Issuer 1.76 1.45 -0.07 94 1.069 1.482 -0.168 Underlying -1.38 -0.01 -2.58 *** 97 -0.673 -0.008 -4.838 +3, +23 +3, +45 +46, +252 Issuer 0.50 0.30 0.11 94 0.511 0.214 0.037 Underlying -0.99 -3.62 ** -34.63 *** 97 -0.811 -2.070 -9.014 Table 5.2--Abnormal Returns for Issuers and Underlying Firm Surrounding the Announcement of Exchangeable Debt Offers Using Fama- French Calendar Time Portfolio Regression Equally-and value-weighted (EW, VW) Fama-French intercepts are estimated. Securities formed into portfolios by event date. The estimated intercept comes from regression of portfolio excess returns on 3 Fama-French factors. The intercepts represent the mean daily abnormal return in the event period. Under each parameter are t- statistics (2-tail test). Statistical significance indicated as 10% (*), 5% (**), and 1% (***) levels. Panel A Fama-French Calendar Time Portfolio Regression: EW Market Index -60, -2 -22, -2 -1, +2 Issuer 1.64 1.41 -0.12 94 0.734 0.965 0.009 Underlying -0.96 0.18 -2.74 *** 97 -0.699 0.495 -4.914 +3, +23 +3, +45 +46, +252 Issuer 0.53 0.51 2.63 94 0.547 0.920 1.366 Underlying -0.78 -3.56 -31.16 *** 97 0.064 -1.555 -31.16 Panel B Fama-French Calendar Time Portfolio Regression: VW Market Index -60, -2 -22, -2 -1, +2 Issuer 1.77 1.47 -0.05 94 0.807 1.009 0.172 Underlying -1.14 0.15 -2.65 *** 97 -0.832 0.464 -4.765 +3, +23 +3, +45 +46, +252 Issuer 0.52 0.38 0.66 94 0.470 0.773 0.818 Underlying -1.10 -3.77 * -34.71 *** 97 -0.175 -1.847 -4.092 Table 6--Results from Regression of Exchangeable Debt Proceeds on Run- up, Information Asymmetry Proxies, and Controls Dependent variable is exchangeable debt scaled by total debt and sum of total debt and exchangeable debt. All accounting data used are reported as fiscal year end prior to the announcement of exchangeable debt issues. Abnormal earning are defined as earning per share in year t+2 (excluding extraordinary items) minus earning per share in year t, divided by the share price in year t. Institutional ownership is a proportion of institutional ownership in the underlying company. Run- up is defined as a cumulative return in the underlying stock for two years before the announcement ending three months before the announcement. Issue size is equal to dollar value of the offering. Run-down is defined as a cumulative return on the underlying stock for two years starting one year after the announcement. Issue size is equal to dollar value of the offering. Two marginal tax measures provided by Graham (1996): the first is based on income before interest expense, and the second is based on the income after interest expense. Volatility is the simple variance of daily returns over the twelve months preceding the announcements of exchangeable debt issues and ending three month before the announcements. Size of issuer and underlying firm is the natural logarithm of the dollar market value of the firm. The market-to-book ratio is calculated as the estimated market value of assets, defined as sum of market value of equity, book value of long-term debt, book value of current debt, book value of preferred stock, divided by the book value of assets. Statistical significance is indicated for 10% (*), 5% (**), and 1% (***) levels and is based on robust standard errors clustered at the year level. p- values are in parentheses. Panel A Dependent Variable Dependent Variable is ED/TD is ED/(ED+TD) 1 2 3 4 Intercept 1.2133 *** 0.9802 *** 0.6801 *** 0.5472 *** (0.0100) (0.0020) (0.0008) (<0.0001) MTRb -0.0041 -0.0021 (0.1269) (0.1111) MTRa -0.0015 -0.0007 (0.2871) (0.3014) Run-up 0.0003 *** 0.0003 *** 0.0001 *** 0.0001 *** (0.0003) (0.0004) (0.0001) (0.0003) Abn. Earnings -0.7567 * -0.6675 * -0.3316 * -0.2782 * (0.0795) (0.0825) (0.0713) (0.0905) Inst. Owner 0.0162 0.0259 0.0338 0.0415 (0.9363) (0.9063) (0.7090) (0.6775) Size Issuer -0.0977 *** -0.0929 *** -0.0595 *** -0.0572 *** (<.0001) (<.0001) <0.0001) (<0.0001) B/M Issuer 0.1440 *** 0.1453 *** 0.0384 ** 0.0396 ** (0.0090) (0.0063) (0.0301) (0.0163) Size Under -0.0304 -0.0222 -0.0027 0.0026 (0.2719) (0.3195) (0.8243) (0.7761) B/M Under -0.0359 -0.0280 -0.0226 *** -0.0163 ** (0.2051) (0.2568) (0.0100) (0.0482) Volatility 0.0070 0.0089 0.0017 0.0027 Under (0.2850) (0.1696) (0.5469) (0.3387) [R.sup.2] 0.6082 0.6008 0.5614 0.5426 Panel B Dependent variable Deltendent variable is ED/TD is ED/(ED+TD) 1 2 3 4 Intercept 1.0014 *** 0.9049 *** 0.5930 *** 0.5513 *** (0.0043) (0.0004) (<.0001) (<.0001) MTRb -0.0015 -0.0005 (0.4949) (0.5995) MTRa -0.0004 -0.0002 (0.7487) (0.7579) Run-up 0.0003 *** 0.0003 *** 0.0001 *** 0.0001 *** (0.0025) (0.0018) (0.0008) (0.0010) Run-down -0.0961 *** -0.0975 ** -0.0501 *** -0.0515 *** (0.0061) (0.0110) (0.0002) (0.0009) Inst. Owner -0.0547 -0.0548 -0.0228 -0.0229 (0.7540) (0.7523) (0.7625) (0.7600) Size Issuer -0.0868 *** -0.0851 *** -0.0543 *** -0.0541 *** (0.0001) (<.0001) (<.0001) (<.0001) B/M Issuer 0.1559 *** 0.1559 *** 0.0427 *** 0.0435 *** (0.0021) (0.0016) (0.0049) (0.0021) Size Under -0.0255 -0.0215 -0.0022 0.0003 (0.2434) (0.2335) (0.8135) (0.9641) B/M Under -0.0100 -0.0039 -0.0088 -0.0026 (0.7534) (0.8829) (0.3749) (0.7206) Volatility 0.0051 0.0058 0.0007 0.0009 Under 0.3486 (0.3035) (0.7808) (0.7387) [R.sup.2] 0.6105 0.6092 0.5782 0.5727 Table 7--Results from Logistic Regression of Underlying Stock Sale on Proxies for the Market Assessment of Unfavorable Information and Controls The table presents logistic regression analysis of the underlying equity direct market sale. The dependent variable is 1 if the issuer sells at least some part of its stake in the underlying company after the exchangeable debt issue was announced and while the issue has been outstanding or was retired to be able to sell the stake, and 0 if no direct market sale was made while the issue was outstanding. All accounting data used are reported as fiscal year end prior to the announcement of exchangeable debt issues. Abnormal earnings are defined as earnings per share in year t+2 (excluding extraordinary items) minus earning per share in year t, divided by the share price in year t. Institutional ownership is a proportion of institutional ownership in the underlying company. Run-up is defined as a cumulative return in the underlying stock for two years before the announcement ending three month before the announcement. Issue size is equal to dollar value of the offering. Stake before is the pre-offer issuer's ownership in underlying firm. Stake after is the post conversion stake in the underlying firm by issuer. 1 2 Intercept 4.0757 ** 4.0059 ** (0.0123) (0.0115) EW 0.0219 (0.7638) VW 0.0184 (0.8240) Stake Before -0.0558 * -0.0548 * (0.0644) (0.0646) Stake After 0.0871 * 0.0860 * (0.0840) (0.0853) Issue size 0.00221 * 0.0022 * (0.0752) (0.0781) Abn. Earnings -5.4660 -5.4662 (0.2085) (0.2152) Inst. Ownership -8.3812 *** -8.2738 *** (0.0051) (0.0048) Run-up -0.0003 -0.0002 (0.8378) (0.8713) Concordant 86.9 86.7
|Printer friendly Cite/link Email Feedback|
|Author:||Danielova, Anna N.|
|Publication:||Quarterly Journal of Finance and Accounting|
|Date:||Mar 22, 2011|
|Previous Article:||The information risk drivers: a long-term analysis to support a risk premia modeling.|
|Next Article:||To be or not to be public: the impact of SOX.|