Citation
Convertible bond issues and equity price impacts

Material Information

Title:
Convertible bond issues and equity price impacts
Creator:
Kuhlman, Bruce Robert, 1949-
Copyright Date:
1988
Language:
English

Subjects

Subjects / Keywords:
Arithmetic mean ( jstor )
Convertible bonds ( jstor )
Currency conversion rates ( jstor )
Debt ( jstor )
Equity ( jstor )
Financial securities ( jstor )
Market prices ( jstor )
Prices ( jstor )
Signals ( jstor )
Stock prices ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
AFM0874 ( ltuf )
20063449 ( oclc )
0024801400 ( ALEPH )

Downloads

This item has the following downloads:


Full Text

CONVERTIBLE BOND ISSUES AND
EQUITY PRICE IMPACTS













By

BRUCE ROBERT KUHLMAN


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN
PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA

1988


OF F LIBRARIES





To my wife, my parents, and Carolyn and Allyn





ACKNOWLEDGMENTS


Special thanks are due Professor Robert C. Radcliffe who chaired my dissertation. His guidance and support ensured the successful completion of this dissertation. I also thank Professor David T. Brown, and Professor James T. McClave for their assistance. Thanks are not enough for my wife, Jacalyn, for her loving support over these years.


iii





TABLE OF CONTENTS


Page


iii vi


ACKNOWLEDGMENTS . . . . . . . . . . . . . . . . . . .

ABSTRACT . . . . . . . . . . . . . . . . . . . . . . .


CHAPTER


ONE

TWO


INTRODUCTION . . . . . . . . . . . . . . . .

LITERATURE REVIEW . . . . . . . . . . . . . .

Common Stock and Straight Bonds . . . . .
Convertible Bonds . . . . . . . . * * * *
Theoretical Valuation of Convertible Bonds HYPOTHESES . . . . . . . . . . . . . . . . .

Announcement Date . . . . . . . . . . . .
Issue Date . . . . . . . . . . . . . . . .
Notes . . . . . . . . . . . . . . . . . .

DATA . . . . . . . . . . . . . . . . . . . .

Discussion of the Data . . . . . . . . . .
Notes . . . . . . . . . . . . . . . . . .


THREE


FOUR


FIVE


EMPIRICAL DESIGN . . . . . . . . . . . .

Event Study Methodology . . . . . . .
Announcement Date Tests of Hypotheses Issue Date Tests of Hypotheses . . . .

RESULTS . . . . . . . . . . . . . . . . .

Announcement Date Test Results . . . .
Issue Date Test Results . . . . . . .
Summary of Results . . . . . . . . . .
Additional Tests . . . . . . . . . . .
Single vs. Multi . . . . . . . . . .
Announcement Date Tests . . . . .
Issue Date Tests . . . . . . . . .
CP Ratios . . . . . . . . . . . . .
Notes . . . . . . . . . . . . . . . .


six


80 88 99
104 105 106 113 118 125





SEVEN SUMMARY, CONCLUSIONS,
AND FUTURE RESEARCH . . . . . . . . . . . 160

Summary and Conclusions . . . . . . . . . . 160
Announcing Convertible Bond Issues . . . 160 Selling Convertible Bond Issues . . . . . 163 Additional Tests . . . . . . . . . . . . 165
Future Research . . . . . . . . . . . . . . 167

REFERENCES . . . . . . . . . . . . . . . . . . . . . . 170

BIOGRAPHICAL SKETCH . . . . . . . . . . . . . . . . . 172





Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy



CONVERTIBLE BOND ISSUES AND EQUITY PRICE IMPACTS


By


Bruce Robert Kuhlman


December 1988


Chairman: Dr. Robert C. Radcliffe Major Department: Finance, Insurance, and Real Estate



That the announcement and sale of convertible bonds elicit negative equity price impacts is an established fact. The primary objective of this dissertation is to determine those factors, characteristics of the firm and the issue, which contribute to this impact.

Equity price impacts are studied at both the announcement and issue dates. It is hypothesized that the degree to which the market values the convertible issue as equity will determine the extent of the equity impact. As a proxy for the equity component of the convertible, the CP ratio is





calculated. This ratio, the conversion price divided by the current stock price, measures the sensitivity of the market value of the convertible to the underlying stock.

At both the announcement and issue dates, firms that set the lowest CP ratios experience the greatest (most negative) equity impacts. Although the results seem to support the null hypothesis, tests show that the CP ratio might not be capturing the equity component of the bonds.

To determine whether the CP ratio is actually masking the effects of other variables, several firm and issue characteristics are tested. When the firms are divided into

three strata by CP ratio, two variables, issue size and days between announcement and issue dates, are found to be significantly different across the strata.

A significant difference is observed between the issue date equity price impacts of firms selling their only issue of convertible bonds and those with outstanding previously issued convertible bonds. These are called single and multi firms, respectively. There is also evidence of a possible relationship between firm size and relative issue size and equity price impacts for multi firms.

The findings of this dissertation provide evidence that equity price impacts at the announcement and subsequent sale of convertible bonds may be related to characteristics of the convertible issue and/or the issuing firm.


vii





CHAPTER ONE
INTRODUCTION


This dissertation examines abnormal stock price behavior associated with the announcement and subsequent sale of

convertible bonds. Empirical research dealing with the impacts which new securities have on the price of existing

shares has documented three common findings:


(1) The announcement and sale of a straight debt security has little or no impact on firm equity value. For example, Smith (1986), in a survey of research published through 1986, reported an insignificant average two-day abnormal return of -0.26% at the announcement of an issue of straight debt. Barclay and Litzenberger (1986) report negative but insignificant abnormal returns during the announcement day. These examples are typical of the empirical research associated with straight debt securities. Studies have shown consistently that the announcement of straight debt causes negative but insignificant effects on outstanding share prices.

(2) The announcement and sale of common equity causes a relatively sizable negative impact on firm equity value. Smith (1986) reported an average two-day abnormal return of
-3.14%. This is confirmed by Barclay and Litzenberger who find that stock prices fall, on average, 1.5% in the first 15 minutes following an equity sale announcement and between 2.7% and 3.0% during the three hour period surrounding the announcement. The consistent results of studies
dealing with common equity are significant negative abnormal stock returns at both the announcement and issue dates.

(3) The announcement and sale of convertible bonds causes an impact between those of straight debt and common equity. Smith (1986) reports a statistically significant average two-day abnormal return of -2.07%.





2

Theories which have been developed to explain such price reactions can be broken into two general categories:

(1) leverage change, market inefficiency, and price pressure hypotheses based on symmetric information between management and investors, and (2) hypotheses based on asymmetric information.



Symmetric Information



Leverage Hypotheses

The various leverage hypotheses are restricted solely to changes in capital structure. Investment decisions are given and the offering conveys no information about profitability of existing or future assets. Empirical tests of leverage hypotheses, then, would be limited to security offerings which are classified as refunding of outstanding issues.

If a new security offering has the potential to change financial leverage and is as least partially unexpected by investors, its announcement can cause various effects on firm equity value depending upon the theory to which one subscribes. These theories can be classified as: (1) Tax Advantage of Debt, (2) Optimal Capital Structure, and (3) Redistribution of Wealth.





Tax advantage of debt

The tax advantage of debt hypothesis is based upon the analyses of Modigliani and Miller (1963). According to Modigliani and Miller, the effects of new debt or equity issues will depend upon the application of the proceeds. An issue which increases total interest tax shield will increase share prices, while the opposite is true for a tax shield decreasing issue. For example, an issue of common equity sold to retire straight debt has the effect of reducing the firm's interest tax shield. Refunding outstanding debt by selling a comparable issue of debt has no impact on tax shields. And, repurchasing equity shares with funds acquired through the sale of bonds increases tax shields.

The effect of selling convertible bonds, then, should also depend upon the net effect on interest payments. Since convertible bonds carry reduced coupons and are rarely used as refinancing tools, their announcement date impact is difficult to predict according to the Tax Advantage of Debt Hypothesis. However, Mikkelson (1981 and 1985) has examined stock returns associated with the call of an issue of convertible bonds. Since the motivation for calling convertible bonds is most often to force their conversion to equity, the action will usually decrease the net interest tax shield. Although the direction of stock price movements was as would be predicted, Mikkelson was unable to attribute the negative call date equity impact to loss of tax shield.





optimal capital structure

This hypothesis suggests that all unanticipated security offerings should have a positive effect on share

prices. If (1) an optimal capital structure exists, (2) management always acts in the interest of shareholders, and

(3) there is symmetry of information between managers and shareholders, any new security offering could be viewed as movement towards the optimal structure. Since sale of the new issue is assumed to contain no information? this hypothesis would imply that only issue size, not type, would affect share prices.



Redistribution of wealth

The redistribution hypothesis is based upon option pricing theory and suggests that new equity issues reduce default risk and transfer a portion of firm asset value from stockholders to debtholders. Issuing debt increases default risk and shifts wealth from bondholders to stockholders. It is implied, then, that new equity issues will decrease stock prices, while new debt issues will have the opposite effect.

Since, by definition, convertible bonds are (usually) convertible into common equity, they have characteristics of both debt and equity. If the bonds increase (decrease) the firm's debt-equity ratio, their sale will be met with





5

increased (decreased) share prices. Without the ability to accurately measure the firm's debt ratio before and after the sale, the exact effect of selling convertible bonds is very difficult to predict.



Testing the application of the leverage hypotheses to the sale of convertible bonds is very difficult. First, they are rarely used solely for realigning capital structure, so both investment and capital structure impacts might be present at the announcement of a proposed issue of convertible bonds. Also, convertible issues have both immediate and future impacts on firm debt structure which can not be predicted at the announcement or sale of the issue. For example, the future date at which the issue will be converted and financing decisions at that time are unknown.



Market Inefficiency

One view of the negative abnormal returns associated with common equity and convertible issues is that investors consider a new issue as a statement by management that the security is overvalued. By issuing securities at inflated values, management effectively transfers wealth from new shareholders to existing shareholders.

There are two reasons why investors might believe that management thinks the security is overvalued: (1) management has privileged information and, thus, knows better





6

what the security is worth, and (2) management and shareholders have symmetric information, but management is somehow better able to value the firm's debt and equity. It is the latter that is considered market inefficiency, but it is difficult to accept that managers can consistently assess firm values more accurately than the market when they hold the same body of information.



Price Pressure

The sale of new securities is often said to have a negative price impact due to "price pressure" associated with the issue. Price pressure can arise from either a downward sloping demand curve or from transactions costs.

If a security has a downward sloping demand curve, additional securities can only clear at increasingly lower prices. A downward sloping demand curve can be the result of differences in information among investors in a complete market or an incomplete market with symmetry of information. Regardless of the cause, if perfect substitutes for a firm's securities do not exist, the issuing firm will face a downward sloping demand curve for its securities.

High grade (straight) debt securities are generally thought to have the greatest number of substitutes and, thus, the least price pressure, while common stock is said to have fewer substitutes and more price pressure. convertible bonds are difficult to assess, but the relatively few





convertibles outstanding and the high costs of replicating convertible payoffs with the firm's common stock and debt suggest the number of substitutes is small. This would suggest that price pressure associated with convertible bonds is considerable. The convertible bond is a hybrid security, however, and might add to the "completeness" of the market. This characteristic might be viewed as attractive by investors who would be willing to pay a premium for the convertible. The degree to which this would offset the price pressure associated with the issue makes it difficult to assess the net effect of selling convertible bonds using the downward sloping demand curve hypothesis.

Price pressure can also arise from price inducements which might be given to initial purchasers. These inducements are considered transactions costs associated with the sale of securities and are commonly thought of as an offset to the costs which buyers incur in adjusting their portfolios to optimal positions.

The distinguishing difference between these two forms of price pressure is that negative equity impacts caused by downward sloping demand curves are permanent while price inducements cause only temporary price declines. It is important to note that price pressure impacts should affect only the security being sold. Convertible bonds, with their

debt and equity components, are difficult to assess using a price pressure argument.





Asymmetric Information



The leverage, market inefficiency, and price pressure hypotheses for equity price impacts of new security offerings are all based on forces other than asymmetric information between investors and firm managers. That is, no new information is conveyed by the unanticipated sale of new securities. Recent theoretical work, however, has investigated the possible role of management signalling in security offerings. This literature is not yet well developed, but three information hypotheses have become popular: (1) cash flow signalling, (2) asset value signalling, and (3) goodpoor firm signaling.


Cash Flow Signalling

The cash flow hypothesis was first suggested by Miller and Rock (1985). They assume symmetric information about the level of the firm's planned investments and the firm's value (conditional on the firm's internal cash flow), but asymmetric information about the firm's cash flows. Their analysis suggests that any unanticipated security offering conveys unfavorable information about the firm's internal cash flow and will result in decreased equity prices. Unanticipated announcements of new security sales signal that internal cash flows are insufficient to finance a known





investment budget. Thus, announcements of debt, equity, and convertible securities all provide negative information and cause negative stock price reactions. An implication is that the larger the issue size, regardless of type, the greater (more negative) the equity impact.



Asset Value Siqnalling

This hypothesis is based on the work of Myers and Majluf (1984). They suggest that management and investors have asymmetric information about the value of the firm's assets in place and potential new investments. A principal result of their model is that management might not to accept all positive NPV projects if new securities need to be issued to do so. This can occur if benefits which existing stockholders would receive from investing in the projects are insufficient to offset losses in stock values caused by negative information released when a new security offering is disclosed to the public. An implication of their model, however, is that equity sales would cause a larger negative equity impact than a straight debt issue. This would imply that the larger the equity component inherent in the convertible bond, the greater the negative stock price reaction.



Good-Poor Firm Signalling

In a recent working paper Narayanan and Mishra (1987)





present a different manner in which information asymmetries might affect stock prices. This will be referred to as the Good-Poor Firm signalling hypothesis. Their model is addressed directly to the issue of convertible securities and is driven by the option which management has to decide the call date of the security.

They assume that all convertibles are issued with a call provision. (Although not strictly true, virtually all convertible bonds are callable at the firm's option with very little or no call protection.) Managers are assumed to operate either good or bad firms. A good firm is one in which managers have relatively good knowledge about the potential outcomes from current and future investments. A poor firm, on the other hand, is one in which managers have more diffuse knowledge about investment outcomes. As a result, managers of poor firms prefer financing alternatives which provide them with a device by which they can signal information to investors in the future. In their model, a call conveys negative information and a decision not to call conveys either no information or good information. The critical aspect of their paper is their suggestion that the issue of a convertible is in itself a signal that management desires the ability to provide future signals via the call feature. That is, the sale of a convertible is a signal that the firm is a "poor firm" and the stock price will fall at the time of issue.





11

The intent of this study is to examine the empirical validity of the above information asymmetry hypotheses. Both the Asset Value Signalling and the Good-Poor Firm Signalling hypotheses would suggest that abnormal returns will be positively related to the size of the equity component in a new convertible bond issue. In addition, these hypotheses would predict larger (more negative) returns for larger issues. In contrast, the Cash Flow Signalling hypothesis would predict only a relationship between negative abnormal returns and issue size. The relative size of the equity component is irrelevant.

The measurement of the equity and debt components inherent in a typical convertible bond is a theoretically and empirically difficult problem. Conceptually the investor purchases a portfolio of the following securities: (1) a straight debt security with fixed coupons and maturity date, plus (2) an option to purchase shares by tendering the straight debt security, less (3) a call option sold back to the firm which allows management to determine the exercise date of the conversion option.

The straight debt value would be the present value of the coupons expected between the issue date and expected call date and the present value of the expected call price. The discount rate would reflect yields on bonds of similar default risk having a maturity equal to the expected call date. Of course, determining the market's





12

perception of the expected call date is an empirically difficult task.

In short, an explicit valuation for the debt and equity components of a convertible bond is presently not possible with any degree of accuracy. As a proxy for the relative amount of equity inherent in the convertible bond, the conversion price is divided by the market price of the issuing firm's common equity. This will be referred to as the CP

ratio throughout the dissertation. The ratio is basically a measure of how far out of the money the convertible's call option is at the announcement or issue date. As such, it is an approximation of how sensitive the convertible's price is to the underlying stock. All else equal, the higher the CP ratio, the less sensitive the call's value to changes in stock prices.

The CP ratio can be interpreted in a slightly different manner in the case of the Good-Poor Firm Signalling hypothesis. In that case, the lower the CP ratio, the earlier management's ability to use the information value of a call/no call decision. As such, the lower the CP ratio,

the more likely the firm is to be a Poor firm. With either of the interpretations, the suggestion is that the lower the CP ratio, the greater (more negative) the equity price impact.

For each convertible in the respective sample, a CP ratio will be calculated at both the announcement and issue





13
dates. In each case, the closing stock price two days preceding the event and the conversion price at issue are used. It is anticipated that those firms setting low CP ratios will experience the greatest equity price impact.





CHAPTER TWO
LITERATURE REVIEW


This study deals with issuing convertible bonds and the accompanying impact on the issuing firm's stock price. To understand this impact one must study not only the literature dealing with convertible bonds, but also the literature dealing with other forms of financing. This chapter is divided into a brief review of the literature dealing with the impacts of issuing common stock and straight debt and a somewhat more detailed review of the literature dealing with the impacts of issuing convertible bonds. These sections are followed by a theoretical discussion of the valuation of convertible bonds.



Common Stock and Straight Bonds


The reaction of equity prices to the issuance of new securities is an area of Finance that has received considerable attention. Masulis (1980) looked at the equity

price reactions to security exchanges, the issuance of one security to retire another. Of particular interest to this study are his findings dealing with debt for equity











and equity for debt exchanges. Masulis found that when debt was issued to retire equity, stockholders benefited with positive residuals of over 10.5%. However, where equity was exchanged for debt, the resulting residuals were negative and significant at around -7.5%. Masulis interpreted the results as support for the theory that equity value is a positive function of leverage. Studies of leverage increasing transactions which lend support to Masulis' 1978 findings are studies by Dann (1981), and Vermaelen (1981) who found residuals of 15.41%, and 14.14%, respectively, for stock repurchases, and Dann and IMikkelson (1984) and Eckbo (1986) who find insignificant residuals of

-.37% and -.06%, respectively, with the issuance of (straight) debt. Korwar (1982), Asquith and Mullins (1983), and Asquith and Mullins (1986) find negative residuals of about -2.5%, -3%, and -3.14%, respectively, with the sale of equity which is a leverage decreasing transaction. This abundance of evidence would seem to indicate that Masulis (1983) was correct in suggesting that residuals experienced by the issuing firm's stockholders are a positive function of the change in leverage.

Several studies, including some of the above, offer results that seem to dispute the conclusion that equity impacts are due to changes in leverage. Eckbo (1986) looks at offering size, tax shield change, rating, method of issuance, and possible abnormal earnings experienced before





16

and after an issue of debt is sold in an attempt to explain price behavior witnessed with debt offerings, but finds none significant. He interprets this as a direct contradiction to Masulis (1983). Mikkelson and Partch (1986) test issue rating and size, type of security issued, and relative amount of new offering for refunding and find only security type to be significant. Rogers and Owers (1985) study equity for debt exchanges and five possible causes for the negative response noted in equity returns. After ruling out wealth transfer, tax shield loss, etc., they suggest that the only remaining possible cause is information contained in the transaction. Vermaelen

(1981), as noted before, found positive residuals with the repurchase of equity. However, since the firm usually tendered for shares in the market at a premium, and a large percentage of the premium remained in the stock price long after the tender expired, he concluded that the tender offer actually signalled to the market that the firm's shares were undervalued. This signalling of "inside" information was offered as reason for the positive residuals with the repurchase. This is in agreement with the suggestions of Myers and Majluf (1984) who suggest that firms will offer securities for sale when management feels the securities are overvalued in the market and purchase the securities if they are undervalued.

According to Miller and Rock (1985), as the issuance of securities becomes more regular, the reaction in the market





will become less severe. That is, if financing is unanticipated, equity prices will drop with its announcement. Even if the security issue is anticipated but the actual amount issued is greater than expected, equity prices will drop. In essence Miller and Rock suggest that a larger than expected issue signals lower than expected earnings. The findings of Eckbo (1986) and Mikkelson and Partch (1986) that type of security offered, not offering size, seems to affect the reaction in equity returns would, however, tend to contradict Miller and Rock.

From the above studies it would seem that capital structure changes, size of issue, loss of tax shield, method of issue, quality rating of issue, and transactions costs all fall short of explaining the reaction in equity returns when the firm announces the need for external funds through the issuance of new securities. only the type of security which is issued remains a significant variable. And, it has been demonstrated repeatedly that issuing equity causes significant negative residuals while the residuals at the issuance of debt are not statistically differently from zero.



Convertible Bonds


Whether the negative reaction in equity returns from the issue of common stock is caused by leverage change or









18

signalling, the fact remains that the issuance of new equity is usually met with negative equity price reactions. Issuing straight debt, as many studies have shown, seems not to be an economically important event. It would seem reasonable that convertible bonds, seen as some combination of debt and equity, could cause a reaction somewhere in between those of debt and equity, and studies have shown this to be the case.

Dann and Mikkelson (1984) study the impact on equity returns for 132 convertible debt issues between 1970 and 1979. Treating the convertible bonds as 100% debt and using market value for equity and book value for debt, they measure the two-day abnormal returns around announcement date and issue date to find the correlation between leverage change caused by the convertible debt issue and the sign of any residuals. They reason that in every case the convertible must have increased leverage, but in every case the residuals are negative. The average two-day announcement date residual for convertible debt issues was -2.31% and significant at the 0.01 level. Of their sample of 132, 29 had positive residuals, and the remaining 103 had negative residuals. With the straight debt offerings from 150 firms during the same time period, they found insignificant residuals of -0.37% at announcement date. At the issue date they find two-day abnormal returns of -1.54% for convertible debt, and, again, reject the null hypothesis at





19

the 0.01 significance level. The abnormal returns for straight debt at issue date are insignificant. Also, for the issues of convertible debt, 32 had positive abnormal returns and 97 had negative, while the distribution for straight debt was about half and half.

Dann and Mikkelson (1984) tested three possible explanations for the response to convertible debt issuance: (1) change in leverage, defined as book value of long term debt divided by market value of equity, (2) new, unfavorable information, defined as the degree to which the proceeds are used for new financing rather than refunding existing debt, and (3) underpricing of the issue. Their finding of negative residuals, even though the convertible bonds increased leverage, was interpreted as contrary to the change in leverage hypothesis that would predict a positive relationship between the direction of the leverage change and the sign of the residual. Recall, however, that, in measuring debt ratios, the authors treated the convertibles as 100% long term debt. They argue that "the equity

component of the new convertible would have to comprise almost two-thirds of the new financing for the average firm in order for the issuance to reduce leverage." Whether or not convertible bonds increase the firm's financial leverage may not be the appropriate question, however.

The term "leverage" can take on two meanings: (1) the tax subsidy connected with the deductibility of interest





20

payments, and (2) the signalling connotation of Myers and Majluf (1984). If the authors are referring to the firm's use of tax shields, the assumption is questionable. It is obvious that convertible bonds, even with their reduced coupons, have the potential to increase the firm's fixed obligations. However, since the proceeds are often partially used to repay other short and long term borrowings with, no doubt, higher interest costs, the net change in tax shield is uncertain without close scrutiny. It is unclear whether the authors actually measured interest costs before and after the convertible issue to determine any changes in tax shield.

If the authors use debt-equity ratios to measure leverage in the Myers and Majluf (1984) signalling context, again their conclusion is questionable. They ignore the potential for complete conversion to equity at a future date and the uncertainty about actions the firm might take at that time if funds are needed. Also, the proceeds of the convertibles send mixed signals of refunding existing long or short term debt, financing capital expansion or working capital needs, or redeeming common or preferred stock. Perhaps it would be better to measure the relative sizes of the debt and equity components of the convertible, the issue's debt-equity ratio, than to look at that of the issuing firm.

To test whether the convertible bond issues signalled new, unfavorable information, they broke their sample into





21

those classified as debt refunding and those representing new financing. They argue that debt refunding send no signal, while using convertibles to finance new projects, etc., represents the unexpected need for external financing; a negative signal in the Miller and Rock (1985) context. When the two samples experienced almost identical announcement period average prediction errors! the authors interpreted it as evidence that new information was not necessarily responsible for the entire reaction. It might be incorrect to assume that issuing convertible bonds to refund debt is not a negative signal, however. since convertible bonds are recognized as containing an equity component, this action must be seen as issuing both debt and equity to replace the outstanding debt.

The signal contained in issuing convertible bonds to refund existing debt, then, is mixed but probably negative as Myers and Majluf (1984) would predict. That the authors found no significant difference between the impacts from issuing convertibles for new financing or to refund existing debt is not surprising.

To test the underpricing hypothesis, Dann and Mikkelson tested for equity price reactions to the issuance of convertible bonds sold through rights offerings. Their

reasoning was that underpricing of new debt securities unnecessarily strips wealth from the stockholders and is accompanied by negative residuals, while an issue through a





22

rights offering should not shift wealth even if the (debt) security is underpriced. Their findings do not support underpricing as the cause for the equity reactions, however. Further, they calculated that convertible debt would have to be underpriced by an implausible 15% for underpricing to account for the entire reaction in equity returns.

Mikkelson and Partch (1986) took a random sample of 360 firms from the Center for Research in Security Prices Daily Returns File (CRSP) for 1972. They studied equity price reactions to announcements by these firms of issues of common stock, straight debt, convertible debt, and preferred stock over the period from 1972 to 1982. They found significant residuals for common stock and convertible debt of

-3.56% and -1.97%, respectively, and insignificant residuals for straight debt and preferred stock. As with the results of Dann and Mikkelson (1984), about three-fourths of the observations for both common stock and convertible debt were negative, while about half of the straight debt observations were negative. The authors apply the results to the hypotheses of Miller and Rock (1985) and Myers and Majluf (1984). Miller and Rock (1985) suggest that unanticipated financing conveys unfavorable future earnings potential, and, hence, is followed by a negative reaction in equity returns. A necessary implication of this hypothesis is that the size of the issue will affect the size of the equity impact. Miller and Rock do not distinguish between





23

types of securities, however. The model of Myers and

Majluf (1984), on the other hand, suggests that new financing conveys information to the market about the firm's assets in place. Specifically, the issuance of equity conveys less favorable information than does the issuance of debt. The findings of Mikkelson and Partch (1986) do not support Miller and Rock in that offering size and net amount of new financing are insignificant in explaining the abnormal returns experienced at the announcement of a convertible bond issue. on the other hand, they find security type to be significant, as Myers and Majluf would predict. Mikkelson and Partch conclude by stating that their evidence is consistent with the prediction of Myers and Majluf (1984) in that issues of common stock and convertible debt convey less than favorable information to the market about the firm's assets in place. Market participants infer

equity and convertible debt to be overpriced whenever the firm issues them.

Vu (1986) studied the effects of the announcement of issues of straight (non-convertible) and convertible debt. He ran cross-sectional regressions to test for the effects of size and riskiness of the issue, tax shield increase, and use of the funds. Again, the author's results agree with the predictions of Myers and Majluf (1984) and disagree with Miller and Rock (1985). While he found no significant relationship between any of the variables and





announcement date abnormal returns, he did find a significant difference in the reactions to straight debt and convertible debt. Although straight debt caused no significant reaction in equity returns, convertible debt caused a significant abnormal return of -1.25%. Since Vu found no relationship between increase in tax shield and equity impacts, he interprets this as evidence that leverage-increasing issues do not necessarily increase stockholder wealth.

It has been demonstrated in many of the above tests that the equity price impact associated with the announcement of a security issue is not a function of increased or decreased tax-shield. In each case the only significant

variable in determining announcement date equity price impacts is the type of security offered. And, the impact from announcing convertible bonds consistently falls between those from announcing common equity or straight debt.




Theoretical Valuation of Convertible Bonds


That convertible bonds contain components of both debt and equity is widely accepted, and the above empirical results would tend to support that belief. Further, it is reasonable to assume that it is this mixture of debt and





25

equity that causes the announcement date reaction for convertible bonds to be less than that for common equity but more than the reaction to announcing straight debt.

However, this is only conjecture without a theoretical basis for the assumption that convertible bonds are indeed valued as part debt and part equity. Ingersoll (1977) presented a thorough model for the valuation of convertible bonds and demonstrated the source of their debt and equity components. He began by stating the value of a straight, non-callable bond to be a function of firm value, V, maturity, t, coupon, c, and final payment, B. Letting F denote the value of a straight, non-callable bond, then



(4) F = F(V, t; B, c).



If we let G be a non-callable bond equivalent to F in all respects except that it is convertible into fraction, a, of the firm's post-conversion equity, Ingersoll showed that G can be represented as



(5) G = G(V, t; B, c, a).



Note that if G is convertible into a fraction a = 0 of the firm's equity, G is identical to F.

If we now allow the same bond to be callable, and let H denote the value of this callable convertible, the relation ship can be expressed as follows:





(6) H = H(V, t; k(t), B, a)



where k(t) is the call price at time t. If k(t > 0) is infinite, H( is equivalent to G( ). In words, the value

of a callable convertible with infinite call price prior to maturity is the same as a non-callable convertible with otherwise identical characteristics. In fact, the difference in value for callable and non-callable debt arises purely out the firm's right to call the issue. The difference in the value of straight, non-callable debt and that of callable, convertible debt is then due to the combination of the call feature and the conversion feature.

Ingersoll shows that the value of convertible debt must be a non-decreasing function of a, the fraction of equity into which the bond is convertible, and k(t), the price at which the convertible is callable. Therefore, a convertible issue, G( ), must be at least as valuable as a nonconvertible issue, F( ). And, since a non-callable issue can be thought of as a callable with infinite call price, a callable convertible cannot be more valuable than a noncallable one. That is, G( ) > H( ).

Ingersoll continues by showing the value of a callable, convertible discount bond to be





27

(7) H(V, t) F( + W( ) + [F F2( H

where


F( The value of a non-callable, nonconvertible discount bond of maturity, t, face value, B, and issued by firm with value V. (See equation 4 for the equivalent coupon bond.)

W( The value of a warrant issued on
the same firm convertible into a portion, a, of the firm's value. (Ingersoll considers the firm to have only common stock and one issue of convertible debt outstanding.)


Fl( and F2( represent the reduction in value

due to the call feature. This call feature reduction in value implicitly considers the reduction in the value of the bond, F( as well as reduction in the value of the

warrant, W( In other words, the value of the noncallable straight debt portion of the convertible is reduced by the addition of the call feature, and the value of the warrant, the investor's option to convert to common equity, is reduced by the firm's ability to determine its maturity through calling the issue. That is, there are two components to the security which bondholders "sell back" to management which allow management to determine both the maturity of the bond and that of the option to convert.

Ingersoll shows that the value of the equity portion of the convertible is a function of the post-conversion value





28

of the percentage of firm held if the bonds are converted and the maturity of the "attached" warrant. The value of this warrant is derived through the Black-Scholes model (1973), and is shown to be a positive function of stock price and maturity, and a negative function of exercise price (conversion price).

Although Ingersoll does not present a closed form model for pricing the various components of a convertible bond, he has demonstrated the theoretical basis for valuing them as both debt and equity. The exact valuation of the components of the convertible bond is a topic for future research.

It is the purpose of this study to demonstrate that convertible bonds vary in the extent to which they are valued as equity and, further, that announcement date equity price impacts are a function of this equity valuation. It is hypothesized that those convertibles perceived to have a relatively large equity component will impact much the same as an issue of common equity. If the market feels the conversion terms will never be attractive enough to warrant conversion, from the standpoint of the firm or the investor, the convertible will be perceived as mostly debt and will impact much the same as an issue of straight debt.





CHAPTER THREE
HYPOTHESES


The hypotheses tested in this study are divided into two categories: (1) those dealing with equity price

impacts at the announcement of a firm's intent to issue convertible bonds, and (2) those dealing with equity price impacts at the date of issue. As such this chapter is

divided into two sections dealing with the respective hypotheses.



Announcement Date

The Announcement Date Event

Investors continually gather and analyze information about publicly held firms, and, at any point in time, some consensus expectation will exist as to the probability that a given firm will issue convertible bonds. Whether changes in this probability have an effect on equity share prices is the subject of this dissertation.

The announcement by management of the intent to issue convertible bonds will have no effect on equity prices if the announcement provides no information of value to market participants. In an efficiently priced market this has two





30

possible explanations. First, the issuance of convertible bonds might be an economically important event, but the announcement of such is fully anticipated by the market. That is, knowledge of a convertible bond issue is valuable, but the public announcement by management is not. Second, the issue announcement might be unexpected but economically unimportant.

While it is theoretically conceivable that a convertible bond announcement could be accurately predicted by the market, as a practical matter it is doubtful. Thus, we will assume that the announcement does change the public's expectations that the firm will issue convertibles. Given this assumption, the nature of equity price movements at the date of announcement can be used to infer the type of information contained in the announcement. For example, if share prices remain unchanged at the date of public announcement, we can infer that the announcement contains no economically important information. However, if on average prices fall by a statistically significant amount, the announcement contains negative information.

Investors become aware of the announcement at the actual announcement date through various wire services, computer networks, news services, etc. However, the announcement might be made during or after market trading hoursso any response to the announcement can take place on that day or on the following day when the announcement appears in





31

the Wall Street Journal or other financial news media. To capture any equity price impact on either of these days, this study utilizes a two-day window consisting of the date the issue is announced in the Wall Street Journal and the preceding day.



The Announcement

The type of information disclosed at the announcement date is important since this could have implications about possible price impacts at the date of issue or withdrawal. As part of this study over 200 convertible bond announcements were read in the Wall Street Journal. Virtually all of these announcements contained only the following type of information: (1) that the firm intended to sell convertible bonds, (2) the approximate size of the issue, (3) the name of the underwriter(s), and (4) a period during which the issue might be sold. The proposed use of the proceeds was occasionally mentioned, but only in very broad terms such as "general corporate needs."

Certain potentially important information is not revealed in a convertible bond announcement, however. In particular, the issue's conversion terms, coupon rate, maturity, and call features are not available until the issue date. Expectations about such terms can be formed at the announcement date given the market's knowledge of past convertible offerings, current capital market conditions, the financial





32

status of the issuing firm, etc., They will not be known with certainty, however, until the issue is sold.


Announcement Date Impact

One purpose of this study is to investigate why convertible bond sales have a negative impact on the stock price of the issuing firm. However, as noted above, certain information is not disclosed until the actual issue date. Thusl any price impacts observed at the announcement date might be due to the fact that a firm intends to issue a generic convertible bond. Evidence pointing to the true

source of the negative returns, however, might not be observed until the actual issue date when all terms are publicly disclosed for the first time.

For example, if a larger equity component inherent in a convertible bond leads to a larger negative equity return, we might not be able to observe such a relationship until the issue date . the date when the public is told the terms of the issue. Announcement date equity price impacts might be due to the equity component inherent in any convertible bond issue but be unrelated to the, as yet, undisclosed terms of the specific issue in question.

Even the probability that the firm will actually sell the convertible is not 1.0 at the announcement date since the firm retains an option to withdraw the issue. Due to the research design of this study, precise statistics on





33

announced convertible issues that are subsequently withdrawn from sale are not available. However, security

offerings are frequently withdrawn. Mikkelson and Partch (1986), in a randomly drawn sample of security offerings from 1972 to 1982, find 23% (18 of 80) common stock offerings cancelled, 20% (35 of 172) straight debt offerings cancelled, and 24% (8 of 23) convertible bond offerings cancelled. Therefore, investors must not only estimate any potential economic impact from the firm's proposed issue of convertible bonds, they must estimate the probability that the issue will actually be sold.

In short, equity price impacts might be observed at both the announcement and issue dates, but the nature of such impacts might be different due to different information available at each date.





Announcement Date Hypotheses

Hol: The announcement of an issue of convertible bonds has no impact on the announcing firm's equity value.

This test will examine whether the announcement of an issue of convertible bonds is an economically important event. If the announcement is fully anticipated by the market and the market efficiently interprets the information, there should be no effect on the firm's stock price. If, on the other hand, the announcement of convertible bonds





34

imparts new information to the market, the effect on the firm's equity value should be reflected in the firm's price per share.

It is anticipated that the announcement of an issue of convertible bonds cannot be accurately predicted by the market, and that it is an economically important event as Myers and Majluf (1984) or Miller and Rock (1985) would predict. As such, a significant negative average equity price response is expected at the announcement date.



Ho2: The announcement of an issue of convertible

bonds causes no increase in the variance of daily stock returns.

On any given day excess returns for a sample of firms will be expected to have a zero mean and some non-zero variance due to insistently movements in stock prices. That is, on any day the movement of a firm's stock may result in a positive or negative excess return, but over a large sample of firms the average excess return on that day should be zero. The release of information, however, could affect the distribution of excess returns, and the type of information can be a factor in the effect. For example, there might be a consensus that on average the type of information released is a negative signal about true equity values. Then, most firms announcing the same type of information will experience a negative equity price impact on





35

that date, and the average excess return for a sample of firms making the same type of announcement will be nonzero.

If the announcement made by each firm does not contain exactly the same information, however, there might be varying equity price impacts. For example, consider the following which demonstrates that the return for any given firm on any given day depends on some normal flow of information as well as any information that might be contained in an unexpected convertible bond announcement:



R = Rn + Rc

where

R = the total return.

Rn= the return due to some "normal" flow of information.

Rc= the return due to information contained in an
unexpected convertible bond announcement. and
(1) (2) (3) (4)

where VAR(R) = VAR(Rn) + VAR(RC) + 2COV(RnRc)

(1) = the variance of the return on any given day.

(2) = the normal variance caused by the usual flow
of information.

(3) = the variance caused by information contained
in the convertible bond announcement.

(4) = the covariance of the returns due to normal
information flow with returns caused by any information in the convertible bond
announcement.









36

On a typical non-event day, terms (3) and (4) are zero, and the variance of the firm's return is simply VAR(Rn). However, with the release of the convertible bond announcement , term (2) might become non-zero, and the value of term (3) depends upon the correlation, r, of Rn and Rc. That is, if managers only make convertible bond announcements on "good" days, rRn,Rc = -1.0. If managers pay no attention to the normal information flow when making these "bad" public announcements, rRn,Rc is more or less zero, and if convertible bond announcements are released only on "bad" days, rRn,Rc = +1.0.

VAR(R) can be summed up as follows:



if rRn,Rc 2 0, VAR(R) > VAR(Rn)

if rRn,Rc < 0, VAR(Rn) < [VAR(Rn) + VAR(Rc)]


In either case, as long as VAR(Rc) > 0, the variance of the daily returns will probably increase with the announcement of a convertible bond issue. If rRn,Rc is negative, as a practical matter it is probably not -1.0. And, even if the correlation of "normal" returns with these "abnormal" returns was perfectly negative, term (4) will not necessarily make VAR(R) < VAR(Rn).

Firms announcing convertible bonds come from a variety of different industries, vary greatly in size, and have varying capital structures and degrees of financial soundness.





37

In that case, even if firms issued identical convertible bonds, the information contained in the convertible bond announcement might be valued differently from firm to firm. And, the equity price impacts might vary across the firms, even if they are mostly negative, causing VAR(Rc) to be firm-specific.

When a firm announces a convertible bond issue, it could be signalling information about its assets in place, Myers and Majluf (1984), or its projected cash flows, Miller and Rock (1985). Investors must analyze each firm and issue to determine the type and value of any signal. On average the announcement date equity impact could be negative. But, since the signal can vary greatly across a large sample of convertible bond announcements, it is expected that the announcement of convertible bond issues will be met with not only a negative shift in the distribution of daily excess returns but also an increase in the variance of those returns, indicating that the announcement of an issue of convertible bonds has a significantly different impact across firms.


Ho3: Variation in the relative values of the debt and equity components of the convertible bond has no effect on any impact on the announcing firm's equity value.

Using the reasoning of Myers and Majluf (1984) the type of security issued signals something about the condition of





38

the issuing firm. Specifically, issuing equity signals weakness. The empirical results of studies discussed in the literature survey support Myers and Majluf. In all cases the announcement of an equity issue is met with significant negative abnormal returns. The announcement date equity price impacts for firms selling straight debt are insignificant.

The straight debt component of convertible bonds, fixed coupons and maturity, should cause no announcement date equity price impact. However, convertible bonds may be converted into the common equity of the issuing firm at the discretion of the holder or may be forced into equity by the issuing firm.2 Therefore, they must be considered part equity, and this equity component may affect the value of the firm's outstanding common equity. In that case the announcement date equity price impact would be directly related to the size of the equity component.

If market participants can reasonably predict the equity component inherent in a given announcement, then the announcement date equity price impact will be directly related to the relative size of the underlying equity component. Hence, those convertibles perceived as having a relatively large equity component will cause a greater (more negative) equity price reaction than issues with relatively smaller equity components.





39

H04: Issue size has no effect on any equity price impact at the announcement date.

This hypothesis tests whether issue size, measured relative to the market value of the issuing firm's equity, is important in determining announcement date equity price impacts. Miller and Rock (1985) maintain that, by going to the financial markets for external funds, the firm signals an unexpected short fall in cash flows. Any announced need for external financing will be interpreted as negative information by the market and should result in a negative reaction in stock price. Myers and Majluf (1984) agree but state further that weaker firms issue equity.

Convertible bonds contain characteristics of both debt and equity. If we assume that the size of the equity component is not a function of issue size, a sample of large convertible bond issues would be expected to have a larger average equity valuation than a sample of small issues. On average, then, we would expect relatively larger issues of convertibles to cause a greater announcement date equity price impact.



Ho5: Firm size has no effect on any equity price impact at the announcement date.

According to Miller and Rock (1985), any unexpected need for external funds signals a short fall in internally generated funds. Since this will be read as a negative





40

signal, the firm's equity will be analyzed and reduced in value accordingly. Myers and Majluf (1984) maintain that a firm's need for external financing signals that its assets are over valued By definition, the announcement is unexpected and should cause reevaluation of the firm's equity.

If we assume that the unexpected component of any release of information is the cause of stock price adjustments, then the size of the announcing firm could influence the equity price impact at the convertible bond announcement date. For example, information on a larger firm may

be more readily available than that on a smaller firm, especially if the larger firm is older and/or has gone to the markets in the past. Or the larger firm may be listed on one of the larger exchanges. In any case, expectations would be that actions of larger firms should be more easily monitored.

It has already been established that the announcement of an issue of convertible bonds is an economically important event. By definition, then, the announcements are unexpected signals of weakness. By the above argument, information on larger firms should be more complete, and a signal of weakness might be more of a surprise when issued by a larger firm. If a relationship exists between firm size and announcement date equity price impacts, it would be expected that the largest firms would experience the greatest (most negative) impacts.













Issue Date


The Issue Date Event

In an efficient capital market, any impact from issuing securities should be experienced when the issue is first announced. Any economically important information contained in the announcement should be analyzed immediately by market participants, and the firm's stock price should reflect any necessary adjustments instantaneously. The date the issue is actually sold should be of no economic importance, since selling the issue is simply the transaction announced previously and has already been priced by the market. However, this makes two important assumptions: (1) all necessary information is contained in the announcement, and (2) there is no uncertainty that the announced issue will be sold.

Violation of either of these assumptions can cause an issue date equity price impact. As noted earlier, the

announcement of a convertible bond issue might not contain all the information necessary to price the impact of the sale at the announcement date. The typical announcement contains information only that the firm plans to sell convertibles, the underwriter(s), an estimate of when the sale may take place, and some probable uses of the proceeds.









42

A potentially crucial item, the conversion ratio, is only released immediately prior to the sale and, therefore, can only be estimated at the announcement date. 3 If the conversion terms have economic importance, the equity price impact can only be estimated at the announcement date. Then, release of the terms at the issue date may constitute new information and cause a reevaluation of the issue and an adjustment in stock prices.

The second implied assumption is that investors know with probability one that the announced issue will not be withdrawn. If any doubt remains that the issue will be sold, the full equity impact might not be felt at the announcement date. The market will continually revise estimates of if and when the issue will be sold, but the probability of issue can never be one until the sale actually takes place.


Issue Date Impacts

At the sale or cancellation date, then, all uncertainty is resolved. If the issue is to be sold, the terms of the issue are made public, and the market can properly assess the economic importance of the particular convertible. The firm's stock price will then be adjusted accordingly. If the issue is withdrawn, however, the announcing firm should experience a positive stock price reaction as found by Mikkelson and Partch (1986).









43

Although not tested empirically, it seems reasonable to assume that the average issue date impact for a mixed sample of firms selling and firms withdrawing convertible bond issues might be zero. Since this study includes no issues that were withdrawn, the average issue date impact will only be some combination of the economic impact of releasing the issue terms and resolution of uncertainty about whether the issue would be sold. Since there will be no firms experiencing the positive impact from withdrawing the announced issue, it is expected that the average issue date impact will be negative.



Issue Date Hypotheses

Hol: The sale of an issue of convertible bonds has

no impact on the issuing firm's equity value.

Any issue date equity price impact will depend upon whether or not economically important information is released when convertible bonds are sold. Anything that has the capacity to affect the market's estimate of the firm's equity value can be construed as information. For example, many of the convertible bond's terms, such as the conversion rights, coupon, and maturity, are not released until the issue is actually sold. if these terms are critical in determining the equity price impact of issuing the convertible, and if the market is not capable of precisely estimating them at the announcement date, it may be reasonable to expect their release to constitute information.









44

Even if the market can accurately predict the terms of the issue and the issue's impact on the market value of the firm's equity, there remains uncertainty that the issue will be sold. In that casel only a portion of the total

equity impact will be felt at the announcement date as the market waits to see the final disposition of the issue. If the issue is withdrawn, the firm might recoup all or part of its announcement date loss as documented by Mikkelson and Partch (1986). If, however, the issue is sold, and the exact sale date has not been predicted by the market, the firm could experience a further reduction in equity value.

Whether it is the release of the terms of the issue or simply the resolution of uncertainty about the sale or withdrawal of the issue, it is expected that economically important information is released at the issue date, and the average issue date equity price impact will be negative.



Ho2: The sale of an issue of convertible bonds

causes no increase in the variance of daily stock returns.

That firms on average experience a negative equity price impact on the date convertible bonds are sold has been established. If the noted equity impact represents

only a shift in the mean issue date excess return, we would say that it is due to resolution of uncertainty. That is, if firms sell convertible bonds, there is a more or less constant equity cost that will be assessed. If, however,





45

along with a shift in the mean comes an increase in the cross-sectional variance of excess returns, we would say that issue date equity price impacts vary significantly across firms. selling an announced issue of convertible

bonds conveys different information about different firms.

There are four scenarios that might explain the negative average equity price impact at the sale of convertible bonds: (1) There is a common equity price impact from issuing convertible bonds, and the probability of issue is considered about equal across firms. The sale of the issue only constitutes resolution of uncertainty about whether the issue will be sold, and any issue date impact is more or less equal across firms. The issue date reaction is a shift in the distribution of daily excess returns. (2) The probability of sale or withdrawal is considered constant across firms. The equity price impact from issuing convertible bonds is firm specific but accurately estimated at the announcement date. Any impact at the issue date is the probability of withdrawal times the true equity impact, and, therefore, it can vary across firms. The issue date reaction is a shift in mean and increased cross-sectional variance. (3) The probability of selling an announced issue of convertible bonds is considered constant across firms, but firm specific equity price impacts cannot be accurately predicted at the announcement date. If the

terms of a convertible bond issue are necessary in





46

calculating equity price impacts, and they can not be accurately estimated, their release at the issue date is economically important. Each firm will experience an issue date equity price impact according to how management sets the terms. Then, since terms vary from issue to issue, it is reasonable to expect the issue date equity impacts to vary in size even though they are mostly negative. The issue date reaction is an increase in cross-sectional variance of daily excess returns and a negative shift in the average daily excess return. (4) The probability of selling an announced issue of convertible bonds varies across firms, and the firm specific equity price impacts cannot be accurately predicted at the announcement date. There is much uncertainty surrounding the signal contained in a convertible bond announcement. The market tends to penalize

firms about the same amount at the announcement date and wait for the release of the terms at the sale date to determine how much, if any, further adjustment in equity value is needed. Much of the equity valuation is left for the issue date, and the effect on the equity value of the selling firms can vary. The result is a shift in the average daily excess return and an increase in cross-sectional variance.



Ho3: Variation in the relative values of the debt and equity components of the convertible bond has no effect





on any issue date impact on the issuing firm's equity value.

The probability of conversion is a potentially important variable in the market's assessment of the signal contained in a convertible bond announcement, and where management sets the conversion terms could have an effect on this probability. For example, if the conversion price is set very close to the firm's current stock price, conversion in the near future might be highly probable. In that case, the market might consider the convertibles no more than a delayed equity issue. Setting the conversion price very high with respect to the current stock price could cause the market to view conversion as improbable. The issue could be considered non-convertible from a practical standpoint.

Since the terms of a convertible bond issue are not released until the issue date, it is reasonable to expect investors to estimate announcement date equity price impacts based on some average conversion terms or CP ratio. For this study, the CP ratio is defined as the ratio of conversion price to current stock price. The CP ratio could be thought of as a measure of the value of the issue's call option to the firm. A low CP ratio would indicate that the call option is "near the money." That is, management might not have to wait long before the stock price is greater than the conversion price and they can call the bonds and





force conversion to equity. A higher CP ratio for the same firm could indicate less probability of forced conversion in the near future.

If, at the issue date, the terms of the issue vary from the perceived average, there could be a reaction by market participants to further reevaluate the issuing firm's equity. It is expected that issues with lower than average CP ratios will be viewed more as equity than the announcement date impact would indicate. The issue date equity price impact for those firms should more negative than for firms setting terms at or above the average.


Ho4: Issue size has no effect on any equity price impact at the issue date.

Any issue date equity price impact must be a result of the release of information or the resolution of uncertainty that the announced issue would be sold. Rejecting the null hypothesis, then, suggests a relationship between issue size, measured as the size of the convertible bond issue relative to the market value of the firm's equity, and the market's ability to accurately predict equity price impacts and/or the probability that the announced issue will be sold.

The arguments of Miller and Rock (1985) and Myers and Majluf (1984) suggest a reaction to the announcement of a proposed issue of convertible bonds. Both arguments





49

suggest that the size of the issue may affect the market's announcement date reaction. But even if the size of the

issue has an effect on announcement date equity price impacts, there is no theoretical argument which suggests that issue size can cause systematic under or overestimation of equity price impacts or probability of issue. There should be no recognizable relationship between relative size and any issue date equity price impact.



Ho5: Firm size has no effect on any equity price

impact at the issue date.

Issue date equity price impacts must be the result of the release of economically important information. As discussed above, this information can take the form of above or below average conversion terms or other information contained in the issue terms, or it can be simply the resolution of uncertainty that the issue would be sold.

There is no theory that would support the hypothesis that any of these would be more or less predictable according to the size of the issuing firm. That is, there should be no systematic over or underestimation of the terms of the issue associated with the size of the issuing firm, and the probability of sale has not been shown empirically to be related to the size of the announcing firm. There might be an increase in the cross sectional variance of daily excess returns and even a shift in the average excess return.





However, there should be no discernible relationship between any observed issue date impacts and the size of the

issuing firm.



Notes


1. For announcements to be classified as information, they
must be unexpected or at least not fully anticipated.

2. If the conversion price is below the market price of
the underlying stock, the issuing firm can force bondholders to convert to equity by calling the issue.

3. The conversion ratio is the number of shares into which
each bond is convertible. The lowest conversion ratio in the sample is 6.58 and the highest is 142.86. Although this might seem quite variable, it is somewhat misleading. To better compare conversion ratios, conversion prices are computed ($1000/conversion ratio) and compared to the firm's current stock price. This is called the CP ratio in the study and ranged from 1.00 to 1.91 with the majority falling between 1.13 and 2.00. Looked at from this angle, there is not a great
difference in the terms of most convertible bonds.





CHAPTER FOUR
DATA


An initial sample of 558 outstanding convertible bond issues was obtained from Moody's Bond Guide.1 As stock price information is critical to the study, the issuing firm must be listed on the Center for Research in Security Prices (CRSP) tapes. 259 issues were deleted from the sample because they were (1) issued prior to July 2, 1962, (2) the issuing firm was never listed on the CRSP tapes, or (3) the firm was listed after issuing the convertibles. 2

The date of sale or issue date was identified in Moody's Industrial Manual. Twenty convertibles that were issued along with other securities were removed from the issue date sample. Of the 20, 15 were issued with common stock, four with straight debt, and one with preferred stock.

Another 25 were deleted from the sample for other reasons. Six of the issues were convertible into the common stock of a parent company, two were convertible into preferred stock or preferred plus common, and one issue was convertible into straight debt. Seven issues were deleted





because they were issued outside the United States, and four were deleted because detailed information was unavailable. One issue was deleted because the exact issue date was unclear. This leaves a total of 298 issues in the

issue date sample.3

original announcements of the proposed convertible bond issues were located in the Wall Street Journal.4 Of the 298 convertibles in the issue date sample, fourteen issues were deleted from announcement date testing for release of confounding information along with the announcement of the proposed issue of convertible bonds. This confounding

information was the announcement of earnings (9 issues), a dividend change (1 issue), a change in bond ratings (1 issue), or any other information with potential for economic impact on the firm's stock price (3 issues).

Another 53 issues were deleted from announcement date testing because no information was released in the Wall Street Journal prior to the issue date.5 This left a

final announcement date sample of 232 convertible bond issues. Table 1 contains a list of the data gathered, when available, for each issue in both samples.

The convertibles in the final sample were issued over the years 1963 to 1986. Table 4-2 shows the frequency of issue per year along with the maturities of the issues. The typical convertible bond is issued with a maturity of 20 or 25 years. Only seven have maturities less than 20





53

years, and only two of those are less than 15 years. This is not unexpected since virtually all convertible bonds are callable, and the value of the call option to the firm is a positive function of the length of maturity of the option.

A survey by Brigham (1966) revealed that managers might look at convertible bonds, with their generally reduced coupons, as "cheap" debt. Table 4-2 shows average yields on long term (> 10 years) treasury issues and AAA rated corporate bonds for each year in the test sample. If managers are simply trying to avoid high borrowing costs, convertible bonds should be particularly attractive during periods of high or climbing interest rates. Over 61%, 184 of 300, of the bonds in the sample were issued between 1965 and 1971.6 During this time, interest rates on AAA corporates climbed steadily from 4.57% to 7.28%, and those on government issues also rose. It appears that, during the earlier years in the sample, cheap borrowing might have played a part in convertible bond sales. However, about 30% of the issues in the sample were sold from 1980 to 1986, and interest rates were generally falling over this period. Also, during the eight years beginning with 1972, very few convertible bonds were issued, and interest rates were rising over this period. This would seem to indicate that now there are motives other than reduced coupons for selling convertible bonds.

Tables 4-3, 4-4, and 4-5 show the distribution of issue, firm, and relative sizes, respectively. The average





54

convertible bond issue in the sample is about $51.9 million. Ninety percent of the issues are $100 million or less. The smallest issue sold was $5.5 million, and the largest was just over $250 million. The smallest firm selling convertible bonds had total equity value of just over $6.6 million. That for the largest was almost $7 billion, and the average firm was valued at about $550 million. Since not all firms in the issue date sample are included in the announcement date sample, Tables 4-4 and 4-5 show the figures for the announcement date firms at both the announcement date and the issue date, and the total issue date sample of firms. Comparing the issue date distribution of the firms in the announcement date sample and the distribution of "All Firms Issue Date" shows the distribution of the issue date firms that were not included in the announcement date tests.

Table 4-5 shows that the average issue of convertible bonds is about 19% of the value of the issuing firm's equity. over 30% of the issues are 10% or less of the value of the firm's equity, and over 89% are less than 35%.

Table 4-6 gives the distributions of CP and conversion ratios. The conversion ratio is the number of shares of common stock into which an individual bond is convertible. Dividing the face value of the convertible by the conversion ratio yields the conversion price, which, divided by the current stock price, gives the CP ratio. The





55

conversion ratio and CP ratio are not necessarily related, however. Much depends on the firm's current stock price. For example, consider the following:







Firm A

Conversion Ratio 40.00 30.00 20.00

Conversion Price $25.00 $33.33 $50.00

Current Stock Price $25.00 $25.00 $25.00 CP Ratio 1.00 1.33 2.00



Firm B

Conversion Ratio 40.00 30.00 20.00

conversion Price $25.00 $33.33 $50.00

Current Stock Price $15.00 $15.00 $15.00 CP Ratio 1.67 2.22 3.33



The average conversion ratio was approximately 28.4, indicating that the average bond in the sample was convertible into just over 28 shares of the issuing firm's common stock. Most of the bonds were convertible into between 10 and 40 shares of common.

The call option on a convertible bond allows the firm the opportunity to force conversion of the bonds to equity once the firm's stock price is above the conversion price.





The CP ratio, therefore, measures how far "out of the money" the option to force conversion is. The minimum CP ratio was 1.0, and the maximum was 1.91. The average CP ratio was 1.17, with over 95% falling between 1.00 and 1.3. Further, almost 60% of the CP ratios were between 1.1 and 1.2.

Table 4-7 shows the distribution of 2-day excess returns at both the announcement and issue dates. The average announcement date return is -1.43%. The minimum and maximum, respectively, were -10.55% and 10.72%. Approximately 73% of the firms, 169 of 232, experienced negative announcement date 2-day excess returns. The average issue date 2-day excess return was -0.61%, and, again, about 73%, 219 of 298, of the returns were negative. The minimum 2day issue date excess return was -12.97%, and the maximum was 15.13%.




NOTES


1. Issues were found outstanding in at least one of the
Moody's Bond Guides dated 1968, 1970, 1972, 1975, 1977, 1980, 1982, 1984, or 1986. Therefore, any issue called before early 1968 or issued in the latter part of 1986
is not in the sample.

2. The CRSP tapes began recording stock price and return
information on July 2, 1962.

3. None of the 298 issues in the final sample was issued
to retire outstanding convertible bonds.

4. The typical announcement for a convertible bond issue
(as it appears in the Wall Street Journal) contains











information on the approximate size of the proposed issue, a period of time during which the issue will occur, and the underwriter(s). In most cases the proposed use of the funds is stated but only in very broad terms such as general corporate needs, refund debt, plant expansion, working capital needs, or some combination of any or all the above. Only in a very small number of cases was a specific use detailed in the announcement, such as retirement of a specific issue of
straight debt or preferred stock.

5. In some cases announcement of the issue came the day
the issue was sold or only a day or two prior to the issue date. Usually the announcement and issue dates
were separated by three to four weeks.

6. None of the issues in the final sample was sold during
1973, and very few were sold during the period 1972 to 1979. To test whether there were different motives for issuing convertibles and, hence, potentially different market reactions, the sample was divided into two periods: those issues sold from 1963 through 1972, and those sold from 1974 through 1986. The average announcement and issue date two-day equity price impacts for the two samples were virtually identical.
The announcement date two-day abnormal returns for 1963-1974 and 1974-1986 were -1.41 and -1.47, respectively. The corresponding issue date returns
were -0.56 and -0.67.

To test whether equity valuation affected equity price impacts, each sample was divided into three approximately equal strata by CP ratio. The three strata for each sample are designated High, Mid, and Low. It was assumed that, although the relative values of CP ratios may vary across comparable strata (ie. high vs. high), each stratum was indicative of a "high", "medium", or
"low" CP ratio for its time of issuance.

The issue date two-day residuals for the high, mid, and low CP strata for the pre and post 1973 issues were 0.12, -0.82, -1.82 and 0.44, -1.15, -2.14, respectively. Although the high and low strata for each sample were significantly different, comparison of the respective strata for the two samples showed no significance. Hence, it will be assumed that the impact of the relative equity valuation of convertibles
does not vary across time.









TABLE 4-1
Data gathered for each issue in the final sample.


ISSUE DATE: The date the issue was sold.

ANNOUNCEMENT DATE: The first time the proposed issue is
mentioned in the Wall Street Journal.

MATURITY DATE: The date on which the issue matures.

COUPON: The annual coupon paid to holders of the bond. This is usually paid in two semi-annual payments, the dates for which were also gathered.

ISSUE SIZE: The original dollar amount registered with the securities and Exchange Commission.

QUALITY RATING: The original rating given by Moody's Bond Rating Service. This was obtained from Moody's Weekly News Reports in the week following the issue of the bonds.

CONVERSION RATIO: The original number of shares into which each bond is convertible. It is used to compute conversion price, the price at which a bond may be converted into the issuing firm's common. It was obtained from Moody's Weekly News Reports in the week following the issue of the bonds.

CLOSING STOCK PRICE: The closing stock price two days
prior to both the announcement and issue dates. These were obtained from the CRSP Daily Stock Master tape.

SHARES OUTSTANDING: The number of shares outstanding two
days prior to both the announcement and issue dates. These were obtained from the CRSP Daily Stock Master tape.

ANNOUNCED USE OF FUNDS: The use of the proceeds from the issue as announced at the time the bonds were sold.

ACTUAL CALL DATE: The date on which the issue was called
for redemption by the issuing firm. This was obtained from Standard and Poor's Bond Guide.

ACTUAL CALL PRICE: The dollar amount offered to each
bondholder for redemption of a single bond. This was
obtained from Standard and Poor's Bond Guide.

CONVERSION PRICE AT CALL: The conversion price on the date the issue was called. It was usually adjusted over the
life of the bonds for stock dividends and/or splits. This and the date on which the conversion privilege expired were obtained from Standard and Poor's Bond Guide.


Unless otherwise noted, data were gathered from Moody's Industrial Manual.





Table 4-2
Frequency of issue and maturity of convertible bonds in the sample and average yields on long term U.S. government and AAA corporate bonds.


Convertible Bonds


Maturity


Ave. Yield *


Number Year Issues


25 Yrs


2 0 Yrs


< 2 0 Yrs


US Gov I t


AAA Corp.


1963
1964 1965 1966 1967 1968 1969 1970 1971 1972 1973
1974 1975 1976 1977 1978 1979 1980 1981 1982 1983
1984 1985 1986


Total


4.07 4.14 4.32 4.65 5.11
5.46 6.47 6.28 5.62
5.54 6.23 6.78 7.06 6.58 7.62 8.61 9.67 11.75 13.30 11.61
11.49 11.94 10.29 7.93


4.30 4.46 4.57 5.33 6.00 6.55 8.03 8.02 7.28 7.17 7.68 8.77 8.99 8.25 8.12 9.00 10.15 12.86 15.36 13.91 12.30
12.04 11.45 9.24


300


204


k Source: Moody's Industrial Manual


















" is " is " is " is " is " is " is " is " is " is " is " is " is " is " is " is " is " is

is


Table 4-3
Convertible Bond Issue Sizes ($ Million)


Issue S


ize Issues




* 250 7

* 200 2

* 175 11

* 150 1

* 140 0

* 130 6

* 120 2

* 110 25

* 100 0

* 90 3

* 80 16

* 70 20

* 60 36

* 50 is

* 40 43

* 30 46

* 20 47

* 10 16
. AVE. IS 51.9


250

200 175 150

140 130

120 110 100 90 80 70 60 50

40 30

20 10


















Firm Equity Value
($Million)


61

Table 4-4 Distribution of Firm Sizes, Announcement and Issue Date.

Announcement Date Sample


All Firms Issue Date


Announ.
Date


Issue Date


5000
4000 3000
2000 1000 900 800 700 600 500
400 300
200 100 90 80 70 60 50
40 30
20 10


FS
FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS < FS <


5000
4000 3000
2000 1000 900 800 700 600 500
400 300
200 100 90 so 70 60 50
40 30
20 10




















Announ. Issue Date Date


2 3

2 1

0 0

3 3

1 0

0 0

5 4

6 7

13 15

16 is

23 20

35 38

47 40

44 46

16 16


Table 4-5
Relative Sizes of Convertible Bond Issues,
Announcement Date and Issue Date


Announcement Date Sample


Relative
Size ($Million)


All Firms Issue Date


4 1 0

4 2 2 6 9 16

17 28

44 49 61

is


1.00 0.90 0.80 0.70 0.60 0.50

0.40 0.35 0.30 0.25

0.20 0.15 0.10 0.05


1.00 0.90 0.80 0.70 0.60 0.50

0.40 0.35 0.30 0.25

0.20 0.15 0.10 0.05








63

Table 4-6 Distribution of Conversion and Issue Date CP Ratios


Conversion Ratios


Ratio


CP Ratios


Issues


CR 4

CR < 100 0

CR < 90 1

CR < 80 2

CR < 70 5

CR < 60 7

CR < 55 8

CR < 50 7

CR < 45 9

CR < 40 20

CR < 35 24

CR < 30 48

CR < 25 55

CR < 20 49

CR < 15 33

CR < 10 5

CR < 5 0

AVE. CR = 28.4


Ratio


1.50 1.40 1.35 1.30 1.25 1.20 1.15 1.10 1.05 1.00


Issues


< CP < 2.00 4

< CP < 1.50 3

< CP < 1.40 2

< CP < 1.35 3

< CP < 1.30 15

< CP < 1.25 34

< CP < 1.20 72

< CP < 1.15 77

< CP < 1.10 37

< CP < 1.05 10

AVE. CP = 1.17


100 < 90 < 80 < 70 < 60 <

55 < 50 < 45 < 40 < 35 < 30 < 25 <

20 < 15 < 10 < 5 <








64

Table 4-7
Distribution of Two-day Excess Returns


Excess Returns (%)


10.00 8.00 6.00

4.00 2.00 1.00 0.00

-1.00

-2.00

-4.00

-6.00

-8.00

-10.00 AVE. 2AR


Announ. Date


1 2 2 9 14 11

24 35 38

46 36 11

2 1

-1.43


Issue Date


2 1 4 11

29 32 36 41 55 52 27

6 0 2

-0.61


10.00 8.00

6.00 4.00 2.00 1.00 0.00

-1.00

-2.00

-4.00

-6.00

-8.00

-10.00


< 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR < 2AR

2AR





CHAPTER FIVE
EMPIRICAL DESIGN


This is a study of the impact on firm equity value on the date convertible bonds are announced and the date they are actually sold. As such an event study methodology will be employed extensively. Following a discussion of the

event study methodology used, the discussion of the empirical design is divided into two sections: (1) a discussion of the announcement date tests, and (2) a discussion of the issue date tests.



Event Study Methodology



Abnormal returns are obtained from the Center for Research in Security Prices (CRSP) Daily Excess Returns Tape. To estimate excess returns, CRSP first estimates each security's beta. Beta is computed for a given year only if the stock traded on at least half of the trading days in that year. The individual beta is calculated as

follows:



Bi = (COVim)/(VARM)









66

COVim = Z (reti,tmret3t) - (1/n)( mret3t)
t t

VARm = Y (mrettmret3t)
t 1l/n) ( I mrett) ( 2: mret3t)
t t

where

reti,t = log (1 + return for security i on day t) mrett = log (1 + value weighted market return on day t) mret3t = mrett_1 + mret + mrett+1 (a moving average market window)

n = number of observations for the year


Next, the firms are ranked and divided into 10 portfolios according to beta. Betas used in the ranking are

either from the preceding year or, if not available, from the current year. If neither is available, the firm is

placed in a portfolio of firms for which no excess returns are calculated. Average daily returns are then calculated over the year for each portfolio.



ARit =1/m : (ARj,t) j = 1 to m
t
where

ARit = the average return for portfolio i on day t

ARj,t = the CRSP daily return for firm j on day t

m = the number of firms in portfolio i


Daily excess returns for firm j are then the difference between the firm's actual return on day t and the average day t return for the portfolio containing firm j.














Ej't = Rj't - Ai't

where

ERjt = the excess return for firm j on day t

Rjit = the actual return for firm j on day t

Ai't = the average return for portfolio i on day t


For each firm in a given sample, 121 daily excess returns are gathered, 60 days before to 60 days following the day of the proposed event. The event will be either the announcement of a convertible bond issue or the date the issue is sold. These daily excess returns are then combined to form 60 two-day excess returns for each firm, the thirtieth, t = 30, of which contains the event date and the preceding day. The two-day excess returns are then averaged at each time t, t = 1 to 60, across the firms in the sample.



AERt = 1/n ERj't


j =l1to n t = 1to 60

where

AERt = the average 2-day excess return for the n firms in the sample at time t ERj't = the 2-day excess return for firm j at time t
n = the number of firms in the sample









68

To test the significance of each average two-day excess return a Student's T-test will be performed. The standard deviation of the 2-day abnormal returns is estimated from the 59 2-day average excess returns from trading days -60 to -2 and +1 to +60. Calculating the standard deviation in this manner avoids any potential effects of abnormal variance on the event date.

In most of the empirical tests the sample will be stratified in some fashion to identify any equity price impact related to the variable by which the firms have been stratified. To test for a significantly different event date impact between two strata, a new variable will be created which amounts to the difference between the 2-day average excess returns for the two strata. For example,



where AERdt = AERa t - AERbt t = 1 to 60

AERd t= the difference in the average 2-day excess returns for strata a and b at time
t

AERa,t= the average 2-day excess return at time t for the firms in stratum a

AERb t= the average 2-day excess return at time t for the firms in stratum b



Once the new variable is calculated, it is a simple procedure to perform a T-test to determine significance. As before, the standard deviation of the 2-day excess returns (this time using AERdt) is calculated after









69

omitting AERd,3O, the difference between event date average excess returns. If the T-test determines that AERd,3O is significantly different from zero, this would indicate a difference in the event date average 2-day excess returns for the two strata. That is, the variable along which the strata have been formed is a factor in event date equity price impacts.





Announcement Date Tests of Hypothesis



H01: The announcement of an issue of convertible bonds has no impact on the announcing firm's equity value.

This first test looks at the overall effects of announcing a convertible bond issue. No stratification or classification of any kind is utilized. For all 232 firms in the announcement date sample, 60 2-day excess returns are gathered centering on the announcement date. A T-test is performed as described above to determine whether the announcement date excess return is significantly different from zero. If the return is significant and positive (negative), this suggests that the announcement contains good (bad) information.



0o2: The announcement of an issue of convertible bonds causes no increase in the variance of daily stock returns.









70

To test whether a convertible bond announcement constitutes the release of firm specific information, the distribution of the announcement date 2-day excess returns will be analyzed. on any given (non-event) day a firm may experience a non-zero excess return. Over many firms on that same day or for the same firm over many non-event days, however, the average excess return will have an expected value of zero with some positive variance. on an event day the mean excess return for a portfolio of firms might be nonzero, and increased uncertainty about the valuation of the firms could cause an increase in cross-sectional stock return variance. The following test was utilized to compare the cross-sectional variance at the announcement date with that of a typical non-event period.

The standard deviation of the twenty-nine 2-day excess returns preceding the announcement date is calculated for each firm as follows:



Sj,a = (1/29 (ERj,t)2 )1/2
t

t = 1 to 29
where
Si.a= the standard deviation of prei~a announcement date excess returns for firm j

ERj It= excess return for firm j at time t


An F-test is performed for each firm that compares announcement date variance with pre-announcement date variance:













Vj= (ERj,30/Sj,a) 2


where


ERj,3 = exesrtr o imthe mean-adjusted announcement date


Sj,a = the pre-announcement date standard deviation of excess returns for firm j

Vi = the ratio of the variance of the announcement date excess return to that of the pre-announcement date period and is
distributed as F (1,30)


The expected Therefore, we let


value of Vi is n/(n - 2) =29/27. U= (Vj * 27/29), and


E (Uj) =1. 0

VAR(Uj) = 2(29 -3)/(29 - 6)


and


This is the standard variance formula for an F with

Vi 1 and V2 = 29 degrees of freedom after the

adjustment factor, 27/29, has been applied. See Warner, et al (1987) for more on this procedure.

Dealing with a large sample of independent observations allows the use of the Central Limit Theorem in testing whether the mean of the U. is significantly different from one. Specifically,


z (U - 1) (2~


n


j = 1 to n





72

As noted above, the announcement date excess return for each firm is mean-adjusted before performing the F-test. Subtracting the average announcement date effect from each firm removes the mean effect from the individual excess returns. This allows testing of the variance alone, and shows whether, aside from some average impact affecting all firms, each firm has a specific equity price impact. In other words, there are two potential impacts at the announcement of economically important information: (1) a mean effect representing a shift in excess return distribution due to some average impact each firm will experience, and (2) an increase in variance of equity returns signalling a differing impact across firms beyond the average impact. That is,



ERj't = AERt + bj't

where

ERjt = the excess return for firm j at time t

AERt = the average excess return at time t for the n firms in the sample

bjt = the firm specific equity impact, either positive or negative.


A significant Z, then, indicates that the signal contained in a convertible bond announcement is valued according to the firm making the announcement. The typical bjt term is large enough to significantly affect the variance of the announcement date excess returns.





73

H03: Variation in the relative values of the debt and equity components of the convertible bond has no affect on any impact on the announcing firm's equity value.

Before this hypothesis can be tested, a proxy for the debt and equity components has to be obtained. It should be noted that since virtually all convertible bond issues are sold at par, if a proxy for either the debt or equity component can be obtained, the other value is automatic.

As a proxy for the equity component inherent in each convertible bond, a ratio of the conversion price to current price per share is used and is referred to as the CP ratio. This ratio measures how far the conversion option is "out of the money" or how likely the issue is to be converted either voluntarily or through a forced conversion. A large CP ratio indicates that the option is far out of the money. The issue is less likely to be converted and

should impact more as an issue of straight debt. A small CP ratio would indicate that the issue has a relatively higher probability of being converted and should impact more like common equity. In other words, those issues with small CP ratios are considered to have a relatively larger equity component than an issue with a larger CP ratio.

The conversion price for each issue is obtained from Moody's Weekly News Reports during the week following the issue date. Obtaining the conversion price very early in

the bond's life is important as it is released immediately





prior to the sale of the bonds and is subject to adjustment for stock dividends or splits during the life of the bond. The conversion price is then divided by the closing stock price two days prior to the announcement date. This stock price was utilized to help avoid any effects of the announcement on equity value.

The minimum and maximum CP ratios for the announcement date sample were 0.75 and 2.03, respectively, with a mean of 1.15. In option pricing jargon the 0.75 ratio is "in

the money", and the 2.03 ratio could be considered "far out of the money." "In the money" would indicate that the firm could immediately call the issue and force its conversion to equity. It is important to remember, however, that

these CP ratios are based on conversion terms not available until the issue date. Therefore, these announcement date figures can only be considered expectations.

Once (expected) CP ratios are calculated for each firm, the firms are divided into three approximately equal strata according to CP ratio. T-tests are performed to test for significance within as well as between strata. Significance between strata indicates that the announcement date equity price impact is a function of the (expected) relative values of the debt and equity components.


Ho4: Issue size has no affect on any equity price impact at the announcement date.





75

To test whether the dollar size of the convertible issue is a factor in the announcement date impact, the firms will be stratified by the relative size of the issue. This is defined as the dollar value of the announced issue divided by the market value of the firm's equity. The dollar size of the issue is obtained from Moody's Industrial Manual, and the market value of the firm's equity is calculated as the market price per share times the number of shares outstanding. Both the price per share and the number of shares are obtained from the CRSP Daily Stock Master tape, and, to avoid any possible effects of the announcement on the share price, the closing price two days before the announcement is used.

Once relative size is calculated for each firm for which the data are available, the firms are divided into three approximately equal strata. A significant difference between strata indicates a difference in announcement date equity price impacts based on relative size.


Ho5: Firm size has no affect on any equity price impact at the announcement date.

To test whether the size of the issuing firm is a factor in the announcement date impact, the firms are stratified by total market value of equity. As in the last test, the market value of the firm is defined as the closing stock price two days before the announcement times the





76

number of shares on that date. Significance between strata indicates that the equity price reaction to the announcement is a function of the size of the issuing firm.


Issue Date Tests of Hypothesis



Hol: Sale of an issue of convertible bonds has no impact on the issuing firm's equity value.

This is a general test of the stock price affects of issuing convertible bonds on the date the bonds are actually sold. For the 298 securities in the issue date sample, the sale date is identified in Moody's Industrial Manual. Significant issue date excess returns indicate that a firm issuing convertible bonds will experience an equity price impact at not only the announcement of the issue but also at the sale date.


Ho2: The sale of an issue of convertible bonds

causes no increase in the variance of daily stock returns.

As with the announcement date tests, the cross-sectional variance of excess returns on the issue date will be compared to that of a typical non-event period. For this test, however, the standard deviation of post issue date excess returns will be compared to issue date excess return variance. The remainder of the test is exactly as its announcement date counterpart. A significant Z value will





77

indicate that firm specific information is released when convertible bonds are actually sold.



Ho3: Variation in the relative values of the debt and equity components of the convertible bond has no affect on any issue date impact on the issuing firm's equity value.

Again the CP ratio is used as a proxy for the equity component of the individual convertible bond. To calculate the CP ratio the conversion price as stated in Moody's Weekly News Reports the week after the issue sold is divided by the closing stock price for the issuing firm two days before the issue date. The minimum and maximum CP ratios at the issue date were 1.0 and 1.91, with an average of

1.17.

After the CP ratio is figured for the 257 firms for which the data are available, the firms are divided into three approximately equal strata. Significance between

strata indicates that the relative values of the debt and equity components inherent in the convertibles are a factor in the issue date equity impact.


Ho4: Issue size has no affect on any equity price impacts at the issue date.

To test whether the value of the convertible relative to the total market value of the issuing firm's equity is a





78

factor in any issue date equity price impact, the issuing firms are divided into three approximately equal strata by relative issue size. Relative size is defined as the dollar size of the issue divided by the market value of the issuing firm's common equity. The dollar amount of convertibles sold is obtained from Moodv1s Industrial Manual. The value of the firm's equity is the closing price per share two days preceding the issue date times the number of shares on that date. Both of these figures are obtained

from the CRSP Daily Stock Master tape. Closing price is obtained from two days before the event to avoid any effects of the release of the terms of the issue either on the sale date or the day before.

A T-test is used to test for significance within each stratum as well as between the strata. Significance between strata indicates that the issue date equity price impact is a function of relative size.



Ho5: Firm size has no affect on any equity price impact at the issue date.

The firms are divided into three approximately equal strata according to market value of equity. Market value of equity is defined as the closing price per share of common two days before the sale date times the number of shares outstanding on that date. The source for these data is the CRSP Daily Stock Master. Significance between the









79

strata indicates that firm size is a factor in the equity price impact at the date convertible bonds are sold.





CHAPTER SIX
RESULTS


This chapter is divided into four sections: (1) a discussion of the results of the announcement date tests, (2) a discussion of the results of the issue date tests, (3) a summary of the results of the announcement and issue date tests, and (4) a discussion of additional tests performed.



Announcement Date Test Results



Hol: The announcement of an issue of convertible bonds has no affect on the announcing firm's equity.

Table 6-la shows the average two-day excess returns for the 121 day period surrounding the announcement of 232 issues of convertible bonds. observation 30 contains the announcement date, as identified in the Wall Street Journal, and the preceding day. The average announcement date equity impact on the 232 firms is -1.43% and is significant at the 1% level. Of the 232 two-day excess returns, 72% (168) were negative.

The results indicate that the announcement of an upcoming issue of convertibles is received as unfavorable





information by the market and has an economically significant impact on the market value of the average announcing firm's equity. The null hypothesis that the announcement

of an issue of convertible bonds will have no effect on the announcing firm's equity is rejected at the 1% level.

Table 6-lb shows average daily excess returns and average cumulative excess returns for 121 days surrounding the announcement date. The average equity impacts for the announcement day and the preceding day are -0.727% and

-0.697%, respectively, and both are significant at the 1% level. The cumulative excess returns are positive and statistically significant for most of the sixty days preceding the announcement. The average cumulative excess return

reaches 4.25% on day -5 and is significant at the 1% level. The negative returns for the five days preceding and including the announcement date bring down the cumulative excess return to about 2.4% which is only significant at the 5% level. By the first day following the

announcement, the average cumulative return is approaching 2.0% and only significant at the 10% level. This cumulative return and significance is maintained for the ten days following the announcement date, but by day +11 any significance is lost. This seems to indicate some validity in

the theory that management tries to time security





announcements. According to Myers and Majluf (1984) management will issue a security it deems to be overpriced. The accumulation of over 4.0% in excess returns for the sixty days preceding the announcement and the subsequent loss of this excess return would support Myers and Majluf. The announcement of the impending convertible bond issue signals that the firm's equity is not worth as much as its current stock price would indicate.

This test has determined that the announcement of an issue of convertible bonds is a statistically as well as an economically significant event. It will be the goal of the following tests to help determine those characteristics of the issue and/or the firm that affect the market's reaction to the announcement of a convertible bond issue.



Ho2: The announcement of an issue of convertible

bonds causes no increase in the variance of daily stock returns.

If the signal contained in an announcement of convertible bonds is valued according to the announcing firm, the equity impact could vary greatly from firm to firm. The resulting movement of stock prices in differing amounts, both positive and negative, will cause an increase in the crosssectional variance of stock returns. If the information available at the announcement date is not complete enough to form estimates of equity values, however, the market





83

might simply assess some more or less constant amount from each announcing firm's equity. If this is the casel the distribution of excess returns at the announcement date would simply shift to the left with no increase in variance.

The Z-score from testing all 232 firms in the announcement date sample was only 0.624, and the null hypothesis cannot be rejected. since there is no significant increase in the cross-sectional variance of equity values at the announcement date, it would appear that firms are simply charged a flat amount for announcing convertible bonds. This could have two explanations: (1) there is no attempt by the market to estimate the terms and other information not contained in the announcement, or (2) the information omitted from the announcement is of no economic importance. It only matters that the firm intends to issue convertible bonds.



Ho3: Variation in the relative values of the debt and equity components of the convertible bond has no effect on any impact on the announcing firm's equity value.

According to Myers and Majluf (1984) managers prefer to issue debt, and issuing equity or convertible bonds is considered a negative signal. It follows from their reasoning that the impact from announcing convertible bonds might be due to the equity component inherent in the issue. This





84

test divides the issues into three strata according to equity value. The equity component of each bond in the sample is measured by its CP ratio, the ratio of conversion price to stock price at issue date. Table 6-2 shows the results of this stratification.

The third of the firms with the highest CP ratios experienced an announcement date 2-day excess return of -1.04% which is significant at the 1% level. The middle and lowest strata experienced 2-day excess returns of -1.63% and

-1.84%, respectively. Both are significant at the 1% level. Sixty-nine percent (49 of 71) of the high strata 2-day excess returns, seventy-three percent (52 of 71) of the middle strata 2-day excess returns, and seventy-eight percent (55 of 70) of the low strata 2-day excess returns were negative. T-tests revealed no significant difference between announcement date reactions in any of the strata. No statistical relationship is established between the announcement date impact and the equity component of the convertible bond. The T-test comparing the announcement date equity impacts for the high and low strata produced a t-value of 1.43, and, therefore, the null hypothesis cannot be rejected at the 10% level. However, a pattern in announcement date impacts does seem to be present. The absolute magnitude of the announcement appears to be a negative function of CP ratio. That is, as the CP ratio increases, the equity impact decreases. This would seem to





indicate that even though the conversion terms are not released until the issue date, the market is capable of predicting where management will set the terms or at least makes an attempt to do so.



Ho4: Issue size has no affect on any equity price impact at the announcement date.

Miller and Rock (1985) claim that any unexpected announcement of external financing signals a short-fall in operating funds. A necessary implication of their theory is that the impact on the announcing firm's equity should be a positive function of the size of the issue of securities sold. If this is the case there should be a larger impact on the equity of the firm selling a relatively larger issue of convertible bonds. Table 6-3 presents the results when the firms are stratified according to the relative size of the convertible bond issue defined as the dollar value of the issue divided by the market value of the issuing firm's equity.

The firms were divided approximately into thirds. The 70 firms with the relatively largest issues experienced an impact of -1.75% which is significant at the 1% level. Seventy six percent (53 of 70) had negative 2-day excess returns. The next stratum experienced an average 2-day excess return of -1.21%, again significant at the 1% level. Sixty-eight percent (48 of 71) of the 2-day excess returns









86

returns were negative for this group. The 71 firms in the stratum with the relatively smallest issues experienced an average 2-day excess return of -1.12%. This was significant at the 1% level and 73% (52 of 71) of the excess returns were negative.

To test for a significantly different impact across the strata, a T-test was performed. The null hypothesis that the impacts were equal across the three strata could not be rejected at the 10% level. This would indicate that the relative size of the issue of convertible bonds is not a factor in the announcement date impact on the issuing firm's equity, and it does not support the theory of Miller and Rock (1985).


Ho5: Firm size has no affect on any equity price impact at the announcement date.

According to Myers and Majluf (1984) management will issue securities when they are overpriced. Further, management would prefer to issue debt. The announcement of the sale of convertible bonds would be a negative signal to the market and would be met with a negative impact on stock prices. Miller and Rock (1985) maintain that any unexpected announcement of external funding is a negative signal and would be met with a negative reaction in stock prices. An implication of both is that the predictability of any announcement will affect the reaction by the market.





87

That is, if the market is able to predict the announcement of the need for external funding and the type of security to be issued, there should be no perceptible impact in conjunction with the announcement of the security offering. The availability of information is then critical in determining the impact on the announcing firm's equity.

If the availability of information varies from firm to firm, market reactions could be firm specific. If, for instance, information on large firms is more readily accessible than that for smaller firms, the market would be better able to predict the actions of larger firms. For this test the firms were divided into three strata according to the market value of equity. A T-test was performed to determine significance within as well as among the strata.

Table 6-4 shows the results of this stratification. For the 71 firms in the stratum of the largest firms, the average 2-day announcement date excess return was -1.48% and was significant at the 1% level. Seventy-six percent (54 of 71) of the 2-day excess returns were negative. The middle strata experienced a significant (1%) reaction of

-1.06% while that for the smallest firms was -1.55%, also significant at the 1% level. The percentages of negative 2-day excess returns were sixty-nine and seventy-three, respectively, for the middle and smallest strata. As with relative size, there was no statistical difference between the reactions of any two strata. This indicates that the





size of the issuing firm is not a factor in determining the announcement date impact, and the null hypothesis cannot be rejected.



Table 6-5 presents a summary of the results for the announcement date tests. These tests have determined that announcing an issue of convertible bonds is an economically important event for the announcing firm, and there is some evidence that management attempts to time the announcement to take advantage of abnormal pricing. The average firm experiences an announcement date equity price impact of nearly -1.5%. Factors such as firm size and relative size of the issue have been shown generally not to affect announcement date equity impacts. Even though the size of the equity component inherent in convertible bonds was not demonstrated to have a statistically significant affect on announcement date equity price impacts, the observed pattern of excess returns may indicate that the market attempts to estimate the terms of the issue before they are announced.





Issue Date Test Results



Hol: Sale of an issue of convertible bonds has no impact on the issuing firm's equity value.





89

Table 6-6a presents 60 two-day residuals around the date of sale for 298 issues of convertible bonds. Observation thirty contains the issue date, as identified in Moody's Industrial Manual, and the preceding day. The average issue date equity price impact is -0.61% and is significant at the 1% level. Of the 298 firms issuing convertibles, 72% (168) experienced a negative issue date equity impact. This indicates that the date convertible bonds are sold is an important event for the selling firm, and the null hypothesis is rejected at the 1% level.

Table 6-6b shows average daily and cumulative average abnormal returns for the same period. The average daily abnormal returns for days -40, -14, and -10 show statistical significance. The average abnormal return on the day the bonds were sold is -0.48% and is significant at the 1% level. Besides the issue date, however, there are no significant daily excess returns between day -10 and day +60, and none of the average cumulative abnormal returns from day -12 to day +60 is significant.

Although the issue date itself has been shown to be a statistically significant event for the average convertible bond issuing firm, there are no patterns in the cumulative returns before or after the event which may be of help in explaining either the firm's motive for selling vs. withdrawing the issue or even why the issue date is an event. It is important to make note at this time that the





90

convertible bond issues used in this study were located in Moody's Bond Record and, hence, were already outstanding. This, by definition, excludes any issues that were announced and subsequently withdrawn by the announcing firm. Including canceled issues and treating cancellation dates as issue dates might alter the results of the test. As the results of Mikkelson and Partch (1986) imply, the cancellation of convertible bonds may be viewed as a positive signal and cause positive average cancellation date returns. In that case, including cancelled offerings could push the average issue date abnormal return to zero. Also, at the date the cancelled issues are announced, the market may accurately predict that the issues will be withdrawn, and the announcing firms will experience much smaller announcement date equity price impacts. This could greatly affect the overall average announcement date equity price impacts for firms announcing convertible bonds. This is only conjecture at this point, however, but it warrants future testing.

The next test of hypothesis tests the nature of the issue date equity price impact. Specifically, whether the issue date impact is a result of the release of new information or simply the result of resolution of the uncertainty surrounding sale of the issue.



Ho2: The sale of an issue of convertible bonds

causes no increase in daily stock return variance.





91

As discussed in the announcement date tests, an increase in the cross-sectional variance of issue date excess returns will indicate the presence of a firm

specific equity impact in addition to the average impact felt by all firms. Comparing the cross-sectional variance at the issue date with the average cross-sectional variance of the 30 post-issue date 2-day excess returns yielded a Zscore of 1.10, and the null hypothesis cannot be rejected at the 10% level. There is no evidence to support the alternative hypothesis that each firm has a potentially different issue date equity price impact. The results seem to indicate that firms that sell previously announced convertible bonds are assessed a more or less constant amount.



Ho3: Variation in the relative values of the debt and equity components of the convertible bond has no affect on any issue date impact on the issuing firm's equity value.

Table 6-7 shows the results from stratifying the firms according to CP ratio, the ratio of conversion price to current stock price. The stratum of firms with the large CP ratios experienced an insignificant average issue date impact of +0.19%. Approximately 44% of the firms in this

stratum, 37 of 84, had negative issue date excess returns. A test of whether the number of negative returns was significantly different from 50% yielded an insignificant Z_





92

of 1.09. This indicates that the probability of a negative issue date equity impact for the firms in this stratum is only 50%.

The firms in the medium stratum experienced an average equity impact of -0.59%, significant at the 5% level. over 65% of these firms, 56 of 86, experienced negative returns. The firms in the small stratum experienced an average issue date equity price impact of -1.40% which is significant at the 1% level. Almost 78% of these firms, 67 of 86, had negative issue date 2-day excess returns. The hypothesis that the probability of a negative return is greater than 50% is rejected for both strata at the 1% level.

Testing between strata yielded significant results. The hypothesis that the issue date 2-day excess return for the stratum with the small CP ratios is equal to that of the medium stratum is rejected at the 10% level. That the small stratum equity price impact is equal to that of the the large stratum was rejected at the 1% level. This is strong statistical evidence of a relationship between CP ratio and issue date equity price impact. Firms in the stratum that set low CP ratios experience the greatest equity impact followed by the firms in the stratum of middle CP ratios. The firms that set their CP ratios

relatively high had no significant impact on their equity value at the issue date.





93

This test would indicate that the equity impact for firms selling convertible bonds is related to where management sets the CP ratio. There are two possible interpretations of these results: (1) The equity valuation of the convertibles directly affects the issue date equity price impact, the CP ratio captures the equity valuation, and investors are not able to predict the issue's terms at the announcement date. (2) the equity valuation of the convertible is unrelated to the issue date equity price impact, and the CP ratio is actually masking the effects of another variable related to the firm or the issue.

The goal of the remaining tests in this section is to determine whether there are variables other than the CP ratio which affect the issue date equity price impact. In each test of hypothesis the firms will be stratified according to the variable in question, and the average issue date equity impacts will be compared across the strata. Relationships between these and other variables and the CP ratio will be tested in the last section, "Additional Tests."



Ho4: Issue size has no affect on any equity price impacts at the issue date.

Table 6-8 shows the results from stratifying the firms according to relative size of the issue of convertible bonds. The stratum of firms selling the relatively largest




Full Text

PAGE 1

CONVERTIBLE BOND ISSUES AND EQUITY PRICE IMPACTS By BRUCE ROBERT KUHLMAN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1988 F liBRARfES

PAGE 2

To my wife, my parents, and Carolyn and Allyn

PAGE 3

ACKNOWLEDGMENTS Special thanks are due Professor Robert c. Radcliffe who chaired my dissertation. His guidance and support ensured the successful completion of this dissertation. I also thank Professor David T. Brown, and Professor James T. Mcclave for their assistance. Thanks are not enough for my wife, Jacalyn, for her loving support over these years. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS ABSTRACT CHAPTER . . . . . . . . . . . . . . . . . . . . . . iii vi ONE TWO THREE FOUR FIVE SIX INTRODUCTION LITERATURE REVIEW. . . . . . . . . . . . . . . . . . . 1 14 Common Stock and Straight Bonds . . . 14 Convertible Bonds . . . . . . . . . . . . 17 Theoretical Valuation of Convertible Bonds 24 HYPOTHESES . . . . . . . . . . . . . . . . . . . . . . Announcement Date Issue Date . . . . . . . . . . . . Notes ..... . . . . . . . DATA . . . . . . . . . . . . . . . . . . Discussion of the Data . . . . . . . . Notes . . . . . . . . . . EMPIRICAL DESIGN . . . . . . . . . . . . Event Study Methodology ...... . Announcement Date Tests of Hypotheses Issue Date Tests of Hypotheses .... RESULTS . . . 29 29 41 50 51 51 56 65 65 69 76 80 Announcement Date Test Results . . . . 80 Issue Date Test Results . . . . . . . . . 88 Summary of Results . . . . . . . . . . . . 99 Additional Tests ............. 104 Single vs. Multi ............ 105 Announcement Date Tests ... 106 Issue Date Tests . . . . . . . 113 CP Ratios . . . . . . . .... 118 Notes . . . . . . . . 125 iv

PAGE 5

SEVEN SUMMARY, CONCLUSIONS, AND FUTURE RESEARCH . . 160 Summary and Conclusions .......... 160 Announcing Convertible Bond Issues ... 160 Selling Convertible Bond Issues ... 163 Additional Tests . . . . . . . .. 165 Future Research. . . . . . 167 REFERENCES . . . . . . . . 170 172 BIOGRAPHICAL SKETCH . . . . . . . . . . . . . V

PAGE 6

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy CONVERTIBLE BOND ISSUES AND EQUITY PRICE IMPACTS By Bruce Robert Kuhlman December 1988 Chairman: Dr. Robert c. Radcliffe Major Department: Finance, Insurance, and Real Estate That the announcement and sale of convertible bonds elicit negative equity price impacts is an established fact. The primary objective of this dissertation is to determine those factors, characteristics of the firm and the issue, which contribute to this impact. Equity price impacts are studied at both the announce ment and issue dates. It is hypothesized that the degree to which the market values the convertible issue as equity will determine the extent of the equity impact. As a proxy for the equity component of the convertible, the CP ratio is vi

PAGE 7

calculated. This ratio, the conversion price divided by the current stock price, measures the sensitivity of the market value of the convertible to the underlying stock. At both the announcement and issue dates, firms that set the lowest CP ratios experience the greatest (most negative) equity impacts. Although the results seem to support the null hypothesis, tests show that the CP ratio might not be capturing the equity component of the bonds. To determine whether the CP ratio is actually masking the effects of other variables, several firm and issue char acteristics are tested. When the firms are divided into three strata by CP ratio, two variables, issue size and days between announcement and issue dates, are found to be sig nificantly different across the strata. A significant difference is observed between the issue date equity price impacts of firms selling their only issue of convertible bonds and those with outstanding previously issued convertible bonds. These are called single and multi firms, respectively. There is also evidence of a possible relationship between firm size and relative issue size and equity price impacts for multi firms. The findings of this dissertation provide evidence that equity price impacts at the announcement and subsequent sale of convertible bonds may be related to characteristics of the convertible issue and/or the issuing firm. vii

PAGE 8

-------------------CHAPTER ONE INTRODUCTION This dissertation examines abnormal stock price behav ior associated with the announcement and subsequent sale of convertible bonds. Empirical research dealing with the im pacts which new securities have on the price of existing shares has documented three common findings: (1) The announcement and sale of a straight debt securi ty has little or no impact on firm equity value. For exam ple, Smith (1986), in a survey of research published through 1986, reported an insignificant average two-day ab normal return of -0.26% at the announcement of an issue of straight debt. Barclay and Litzenberger (1986) report nega tive but insignificant abnormal returns during the announce ment day. These examples are typical of the empirical re search associated with straight debt securities. Studies have shown consistently that the announcement of straight debt causes negative but insignificant effects on outstand ing share prices. (2) The announcement and sale of common equity causes a relatively sizable negative impact on firm equity value. Smith (1986) reported an average two-day abnormal return of -3.14%. This is confirmed by Barclay and Litzenberger who find that stock prices fall, on average, 1.5% in the first 15 minutes following an equity sale announcement and be tween 2.7% and 3.0% during the three hour period surround ing the announcement. The consistent results of studies dealing with common equity are significant negative abnor mal stock returns at both the announcement and issue dates. ( 3) causes equity. average The announcement and sale of convertible bonds an impact between those of straight debt and common Smith (1986) reports a statistically significant two-day abnormal return of -2.07%. 1

PAGE 9

2 Theories which have been developed to explain such price reactions can be broken into two general categories: (1) leverage change, market inefficiency, and price pres sure hypotheses based on symmetric information between man agement and investors, and (2) hypotheses based on asymmet ric information. Symmetric Information Leverage Hypotheses The various leverage hypotheses are restricted solely to changes in capital structure. Investment decisions are given and the offering conveys no information about profit ability of existing or future assets. Empirical tests of leverage hypotheses, then, would be limited to security of ferings which are classified as refundings of outstanding issues. If a new security offering has the potential to change financial leverage and is as least partially unexpected by investors, its announcement can cause various effects on firm equity value depending upon the theory to which one subscribes. These theories can be classified as: (1) Tax Advantage of Debt, (2) Optimal Capital Structure, and (3) Redistribution of Wealth.

PAGE 10

,-------------------------3 Tax advantage of debt The tax advantage of debt hypothesis is based upon the analyses of Modigliani and Miller (1963). According to Modigliani and Miller, the effects of new debt or equity is sues will depend upon the application of the proceeds. An issue which increases total interest tax shield will in crease share prices, while the opposite is true for a tax shield decreasing issue. For example, an issue of common equity sold to retire straight debt has the effect of reduc ing the firm's interest tax shield. Refunding outstanding debt by selling a comparable issue of debt has no impact on tax shields. And, repurchasing equity shares with funds acquired through the sale of bonds increases tax shields. The effect of selling convertible bonds, then, should also depend upon the net effect on interest payments. since convertible bonds carry reduced coupons and are rare ly used as refinancing tools, their announcement date im pact is difficult to predict according to the Tax Advantage of Debt Hypothesis. However, Mikkelson (1981 and 1985) has examined stock returns associated with the call of an issue of convertible bonds. Since the motivation for calling con vertible bonds is most often to force their conversion to equity, the action will usually decrease the net interest tax shield. Although the direction of stock price move ments was as would be predicted, Mikkelson was unable to attribute the negative call date equity impact to loss of tax shield.

PAGE 11

4 Optimal capital structure This hypothesis suggests that all unanticipated securi ty offerings should have a positive effect on share prices. If (1) an optimal capital structure exists, (2) management always acts in the interest of shareholders, and (3) there is symmetry of information between managers and shareholders, any new security offering could be viewed as movement towards the optimal structure. Since sale of the new issue is assumed to contain no information, this hypoth esis would imply that only issue size, not type, would af fect share prices. Redistribution of wealth The redistribution hypothesis is based upon option pric ing theory and suggests that new equity issues reduce de fault risk and transfer a portion of firm asset value from stockholders to debtholders. Issuing debt increases de fault risk and shifts wealth from bondholders to stockhold ers. It is implied, then, that new equity issues will de crease stock prices, while new debt issues will have the op posite effect. Since, by definition, convertible into common convertible bonds are (usually) equity, they have characteristics of both debt and equity. If the bonds increase (decrease) the firm's debt-equity ratio, their sale will be met with

PAGE 12

5 increased (decreased) share prices. Without the ability to accurately measure the firm's debt ratio before and after the sale, the exact effect of selling convertible bonds is very difficult to predict. Testing the application of the leverage hypotheses to the they sale of convertible bonds is very difficult. First, are rarely used solely for realigning capital struc ture, so both investment and capital structure impacts might be present at the announcement of a proposed issue of convertible bonds. Also, convertible issues have both immediate and future impacts on firm debt structure which can not be predicted at the announcement or sale of the issue. For example, the future date at which the issue will be con verted and financing decisions at that time are unknown. Market Inefficiency One view of the negative abnormal returns associated with common equity and convertible issues is that investors consider a new issue as a statement by management that the security is overvalued. By issuing securities at inflated values, management effectively transfers wealth from new shareholders to existing shareholders. There are two reasons why investors might believe that management thinks the security is overvalued: (1) manage ment has privileged information and, thus, knows better

PAGE 13

6 what the security is worth, and (2) management and share holders have symmetric information, but management is some how better able to value the firm's debt and equity. It is the latter that is considered market inefficiency, but it is difficult to accept that managers can consistently as sess firm values more accurately than the market when they hold the same body of information. Price Pressure The sale of new securities is often said to have a neg ative price impact due to "price pressure" associated with the issue. Price pressure can arise from either a downward sloping demand curve or from transactions costs. If a security has a downward sloping demand curve, ad ditional securities can only clear at increasingly lower prices. A downward sloping demand curve can be the result of differences in information among investors in a complete market or an incomplete market with symmetry of informa tion. Regardless of the cause, if perfect substitutes for a firm's securities do not exist, the issuing firm will face a downward sloping demand curve for its securities. High grade (straight) debt securities are generally thought to have the greatest number of substitutes and, thus, the least price pressure, while common stock is said to have fewer substitutes and more price pressure . Convert ible bonds are difficult to assess, but the relatively few

PAGE 14

7 convertibles outstanding and the high costs of replicating convertible payoffs with the firm's common stock and debt suggest the number of substitutes is small. This would sug gest that price pressure associated with convertible bonds is considerable. The convertible bond is a hybrid security, however, and might add to the "completeness" of the mar ket. This characteristic might be viewed as attractive by investors who would be willing to pay a premium for the con vertible. The degree to which this would offset the price pressure associated with the issue makes it difficult to as sess the net effect of selling convertible bonds using the downward sloping demand curve hypothesis. Price pressure can also arise from price inducements which might be given to initial purchasers. These induce ments are considered transactions costs associated with the sale of securities and are commonly thought of as an offset to the costs which buyers incur in adjusting their portfo lios to optimal positions. The distinguishing difference between these two forms of price pressure is that negative equity impacts caused by downward sloping demand curves are permanent while price in ducements cause only temporary price declines. It is impor tant to note that price pressure impacts should affect only the security being sold. Convertible bonds, with their debt and equity components, are difficult to assess using a price pressure argument.

PAGE 15

8 Asymmetric Information The leverage, market inefficiency, and price pressure hypotheses for equity price impacts of new security offer ings are all based on forces other than asymmetric informa tion between investors and firm managers. That is, no new information is conveyed by the unanticipated sale of new se curities. Recent theoretical work, however, has investigat ed the possible role of management signalling in security offerings. This literature is not yet well developed, but three information hypotheses have become popular: (1) cash flow signalling, (2) asset value signalling, and (3) good poor firm signaling. Cash Flow Signalling The cash flow hypothesis was first suggested by Miller and Rock (1985). They assume symmetric information about the level of the firm•s planned investments and the firm's value (conditional on the firm's internal cash flow), but asymmetric information about the firm's cash flows. Their analysis suggests that any unanticipated security offering conveys unfavorable information about the firm's internal cash flow and will result in decreased equity prices. Un anticipated announcements of new security sales signal that internal cash flows are insufficient to finance a known

PAGE 16

9 investment budget. Thus, announcements of debt, equity, and convertible securities all provide negative information and cause negative stock price reactions. An implication is that the larger the issue size, regardless of type, the greater (more negative) the equity impact. Asset Value Signalling This hypothesis is based on the work of Myers and Majluf (1984). They suggest that management and investors have asymmetric information about the value of the firm's assets in place and potential new investments. A principal result of their model is that management might not to ac cept all positive NPV projects if new securities need to be issued to do so. This can occur if benefits which existing stockholders would receive from investing in the projects are insufficient to offset losses in stock values caused by negative information released when a new security offering is disclosed to the public . An implication of their model, however, is that equity sales would cause a larger negative equity impact than a straight debt issue. This would imply that the larger the equity component inherent in the con vertible bond, the greater the negative stock price reac tion. Good-Poor Firm Signalling In a recent working paper Narayanan and Mishra (1987)

PAGE 17

i 10 present a different manner in which information asymmetries might affect stock prices. Good-Poor Firm signalling This will be referred to as the hypothesis. Their model is addressed directly to the issue of convertible securities and is driven by the option which management has to decide the call date of the security. They assume that all convertibles are issued with a call provision. (Although not strictly true, virtually all convertible bonds are callable at the firm's option with very little or no call protection.) Managers are assumed to operate either good or bad firms. A good firm is one in which managers have relatively good knowledge about the po tential outcomes from current and future investments. A poor firm, on the other hand, is one in which managers have more diffuse knowledge about investment outcomes. As a result, which managers provide of poor firms prefer financing alternatives them with a device by which they can signal information to investors in the future. In their model, a call call The conveys negative information and a decision not to conveys either no information or good information. critical aspect of their paper is their suggestion that the issue of a convertible is in itself a signal that man agement desires the ability to provide future signals via the call feature. That is, the sale of a convertible is a signal that the firm is a "poor firm" and the stock price will fall at the time of issue.

PAGE 18

11 The intent of this study is to examine the empirical validity of the above information asymmetry hypotheses. Both the Asset Value Signalling and the Good-Poor Firm Signalling hypotheses would suggest that abnormal returns will be positively related to the size of the equity compo nent in a new convertible bond issue. In addition, these hypotheses would predict larger (more negative) returns for larger pothesis issues. In contrast, the Cash Flow Signalling hywould predict only a relationship between negative abnormal returns and issue size. The relative size of the equity component is irrelevant. The measurement of the equity and debt components inher ent in a typical convertible bond is a theoretically and em pirically difficult problem. Conceptually the investor pur chases a portfolio of the following securities: (1) a straight debt security with fixed coupons and maturity date, plus (2) an option to purchase shares by tendering the straight debt security, less (3) a call option sold back to the firm which allows management to determine the exercise date of the conversion option. the call The straight debt value would be the present value of coupons expected between the issue date and expected date and the present value of the expected call price. similar The discount rate would reflect yields on bonds of default risk having a maturity equal to the expected call date. Of course, determining the market's

PAGE 19

12 perception of the expected call date is an empirically dif ficult task. In short, an explicit valuation for the debt and equity components of a convertible bond is presently not possible with any degree of accuracy. As a proxy for the relative amount of equity inherent in the convertible bond, the con version price is divided by the market price of the issuing firm's common equity. This will be referred to as the CP ratio throughout a measure of how the dissertation. The ratio is basically far out of the money the convertible's call option is at the announcement or issue date. As such, it is an approximation of how sensitive the convertible's price is to the underlying stock. All else equal, the high er the CP ratio, the less sensitive the call's value to changes in stock prices. The CP ratio can be interpreted in a slightly different manner in the case of the Good-Poor Firm Signalling hypoth esis. In that case, the lower the CP ratio, the earlier management's ability to use the information value of a call/no call decision. As such, the lower the CP ratio, the more likely the firm is to be a Poor firm. With either of the interpretations, the suggestion is that the lower the CP ratio, the greater (more negative) the equity price impact. For each convertible in the respective sample, a CP ratio will be calculated at both the announcement and issue

PAGE 20

13 dates. In each case, the closing stock price two days pre ceding the event and the conversion price at issue are used. It is anticipated that those firms setting low CP ratios will experience the greatest equity price impact.

PAGE 21

CHAPTER TWO LITERATURE REVIEW This study deals with issuing convertible bonds and the accompanying impact on the issuing firm's stock price. To understand this impact one must study not only the litera ture dealing with convertible bonds, but also the litera ture dealing with other forms of financing. This chapter is divided into a brief review of the literature dealing with the impacts of issuing common stock and straight debt and a somewhat more detailed review of the literature deal ing with the impacts of issuing convertible bonds. These sections are followed by a theoretical discussion of the valuation of convertible bonds. Common Stock and Straight Bonds The reaction of equity prices to the issuance of new securities is an area of Finance that has received considerable attention. Masulis (1980) looked at the equity price reactions to security exchanges, the issuance of one security to retire another. Of particular interest to this study are his findings dealing with debt for equity 14

PAGE 22

15 and equity for debt exchanges. Masulis found that when debt was issued to retire equity, stockholders benefited with positive residuals of over 10.5%. However, where equity was exchanged for debt, the resulting residuals were negative and significant at around -7.5%. Masulis inter preted the results as support for the theory that equity value is a positive function of leverage. Studies of lev erage increasing transactions which lend support to Masulis' 1978 findings are studies by Dann (1981), and Vermaelen (1981) who found residuals of 15.41%, and 14.14%, respectively, for stock repurchases, and Dann and Mikkelson (1984) and Eckbo (1986) who find insignificant residuals of -.37% and -.06%, respectively, with the issuance of (straight) debt. Korwar (1982), Asquith and Mullins (1983), and Asquith and Mullins (1986) find negative residuals with the of about -2.5%, sale of equity -3%, and -3.14%, respectively, which is a leverage decreasing transaction. This abundance of evidence would seem to indicate that Masulis (1983) was correct in suggesting that residuals experienced by the issuing firm's stockholders are a positive function of the change in leverage. Several studies, including some of the above, offer results that seem to dispute the conclusion that equity impacts are due to changes in leverage. Eckbo (1986) looks at offering size, tax shield change, rating, method of issu ance, and possible abnormal earnings experienced before

PAGE 23

16 and after an issue of debt is sold in an attempt to explain price behavior witnessed with debt offerings, but finds none significant. diction to Masulis test issue rating relative amount of only security type (1985) study equity He interprets this as a direct contra (1983). Mikkelson and Partch (1986) and size, type of security issued, and new offering for refundings and find to be significant. Rogers and Owers for debt exchanges and five possible causes for After ruling they suggest information the negative response noted in equity returns. out wealth transfer, tax shield loss, etc., that the only remaining possible cause is (1981), as repurchase dered for percentage after the contained in the transaction. Vermaelen noted before, found positive residuals with the of equity. However, since the firm usually ten shares in the market at a premium, and a large of the premium remained in the stock price long tender expired, he concluded offer actually signalled to the market that the tender that the firm's shares were undervalued. This signalling of "inside" infor mation was offered as reason for the positive residuals with the repurchase. tions of Myers and This is in agreement with the sugges Majluf (1984) who suggest that firms will offer securities for sale when management feels the securities are overvalued in the market and purchase the securities if they are undervalued. According to Miller and Rock (1985), as the issuance of securities becomes more regular, the reaction in the market

PAGE 24

17 will become less severe. That is, if financing is unantic ipated, equity prices will drop with its announcement. Even if the security issue is anticipated but the actual amount issued is greater than expected, equity prices will drop. In essence Miller and Rock suggest that a larger than expected issue signals lower than expected earnings. The findings of Eckbo {1986) and Mikkelson and Partch (1986) that type of security offered, not offering size, seems to affect the reaction in equity returns would, however, tend to contradict Miller and Rock. From the above studies it would seem that capital structure changes, issue, quality fall short of size of issue, loss of tax shield, method of rating of issue, and transactions costs all explaining the reaction in equity returns when the firm announces the need for external funds through the issuance of new securities. Only the type of security which is issued remains a significant variable. And, it has been demonstrated repeatedly that issuing equity causes significant negative residuals while the residuals at the issuance of debt are not statistically differently from zero. Convertible Bonds Whether the negative reaction in equity returns from the issue of common stock is caused by leverage change or

PAGE 25

18 signalling, the fact remains that the issuance of new equity is usually met with negative equity price reac tions. Issuing straight debt, as many studies have shown, seems not to be an economically important event. It would seem reasonable that convertible bonds, seen as some combi nation of debt and equity, could cause a reaction somewhere in between those of debt and equity, and studies have shown this to be the case. Dann and Mikkelson (1984) study the impact on equity returns for 132 convertible debt issues between 1970 and 1979. Treating the convertible bonds as 100% debt and using market value for equity and book value for debt, they measure the two-day abnormal returns around announcement date and issue date to find the correlation between lever age change caused by the convertible debt issue and the sign of any residuals. They reason that in every case the convertible must have increased leverage, but in every case the residuals are negative. The average two-day announce ment date residual for convertible debt issues was -2.31% and significant at the 0.01 level. Of their sample of 132, 29 had positive residuals, and the remaining 103 had nega tive residuals. With the straight debt offerings from 150 firms during the same time period, they found insignificant residuals of -0.37% at announcement date. At the issue date they find two-day abnormal returns of -1.54% for con vertible debt, and, again, reject the null hypothesis at

PAGE 26

19 the 0.01 significance level. The abnormal returns for straight debt at issue date are insignificant. Also, for the issues of convertible debt, 32 had positive abnormal returns and 97 had negative, while the distribution for straight debt was about half and half. Dann and Mikkelson (1984) tested three possible expla nations for the response to convertible debt issuance: (1) change in leverage, defined as book value of long term debt divided by market value of equity, (2) new, unfavorable information, defined as the degree to which the proceeds are used for new financing rather than refunding existing debt, and (3) underpricing of the issue. Their finding of negative residuals, even though the convertible bonds increased leverage, was interpreted as contrary to the change in leverage hypothesis that would predict a positive relationship between the direction of the leverage change and the sign of the residual. Recall, however, that, in measuring debt ratios, the authors treated the convertibles as 100% long term debt. They argue that "the equity component of the new convertible would have to comprise almost two-thirds of the new financing for the average firm in order for the issuance to reduce leverage." Whether or not convertible bonds increase the firm's financial leverage may not be the appropriate question, however. The term "leverage" can take on two meanings: (1) the tax subsidy connected with the deductibility of interest

PAGE 27

20 payments, and (2) the signalling connotation of Myers and Majluf (1984). If the authors are referring to the firm's use of tax shields, the assumption is questionable. It is obvious that convertible bonds, even with their reduced cou pons, have the potential to increase the firm's fixed obli gations. However, since the proceeds are often partially used to repay other short and long term borrowings with, no doubt, higher interest costs, the net change in tax shield is uncertain without close scrutiny. It is unclear whether the authors actually measured interest costs before and after the convertible issue to determine any changes in tax shield. If the authors use debt-equity ratios to measure lever age in the Myers and Majluf (1984) signalling context, again their conclusion is questionable. They ignore the potential for complete conversion to equity at a future date and the uncertainty about actions the firm might take at that time if funds are needed. Also, the proceeds of the convertibles send mixed signals of refunding existing long or short term debt, financing capital expansion or working capital needs, or redeeming common or preferred stock. Perhaps it would be better to measure the relative sizes of the debt and equity components of the convertible, the issue's debt-equity ratio, than to look at that of the issuing firm. To test whether the convertible bond issues signalled new, unfavorable information, they broke their sample into

PAGE 28

21 those classified as debt refundings and those representing new financing. They argue that debt refundings send no sig nal, while using convertibles to finance new projects, etc., represents the unexpected need for external financ ing; a negative signal in the Miller and Rock (1985) context. When the two samples experienced almost identical announcement period average prediction errors, the authors interpreted it as evidence that new information was not nec essarily responsible for the entire reaction. It might be incorrect to assume that issuing convertible bonds to refund debt is not a negative signal, however. Since con vertible bonds are recognized as containing an equity com ponent, this action must be seen as issuing both debt and equity to replace the outstanding debt. The signal contained in issuing convertible bonds to refund existing debt, then, is mixed but probably negative as Myers and Majluf (1984) would predict. That the authors found no significant difference between the impacts from issuing convertibles for new financing or to refund exist ing debt is not surprising. To test the underpricing hypothesis, Dann and Mikkelson tested for equity price reactions to the issuance of con vertible bonds sold through rights offerings. Their reasoning was that underpricing of new debt securities unnecessarily strips wealth from the stockholders and is accompanied by negative residuals, while an issue through a

PAGE 29

22 rights offering should not shift wealth even if the (debt) security is underpriced. Their findings do not support underpricing as the cause for the equity reactions, how ever. Further, they calculated that convertible debt would have to be underpriced by an implausible 15% for underpric ing to account for the entire reaction in equity returns. Mikkelson and Partch (1986) took a random sample of 360 firms from the Center for Research in Security Prices Daily Returns File (CRSP) for 1972. They studied equity price reactions to announcements by these firms of issues of com mon stock, straight debt, convertible debt, and preferred stock over the period from 1972 to 1982. They found signif icant residuals for common stock and convertible debt of -3.56% uals and -1.97%, for straight respectively, and insignificant resid debt and preferred stock. As with the and Mikkelson (1984), about three-fourths results of Dann of the observations for both common stock and convertible debt were negative, while about half of the straight debt observations were negative. The authors apply the results to the hypotheses of Miller and Rock (1985) and Myers and Majluf (1984). Miller and Rock (1985) suggest that unan ticipated financing conveys unfavorable future earnings potential, equity is that and, hence, is followed by a negative reaction in returns. A necessary implication of this hypothesis the size of the issue will affect the size of the equity impact. Miller and Rock do not distinguish between

PAGE 30

23 types of securities, however. The model of Myers and Majluf (1984), on the other hand, suggests that new financ ing conveys information to the market about the firm's assets in place. Specifically, the issuance of equity con veys less favorable information than does the issuance of debt. The findings of Mikkelson and Partch (1986) do not support Miller and Rock in that offering size and net amount of new financing are insignificant in explaining the abnormal returns experienced at the announcement of a con vertible bond issue. On the other hand, they find security type to be significant, as Myers and Majluf would predict. Mikkelson and Partch conclude by stating that their evi dence is consistent with the prediction of Myers and Majluf (1984) in that issues of common stock and convertible debt convey less than favorable information to the market about the firm's assets in place. Market participants infer equity and convertible debt to be overpriced whenever the firm issues them. Vu (1986) studied the effects of the announcement of issues of straight (non-convertible) and convertible debt. He ran cross-sectional regressions to test for the effects of size and riskiness of the issue, tax shield increase, and use of the funds. Again, the author's results agree with the predictions of Myers and Majluf (1984) and dis agree with Miller and Rock (1985). While he found no sig nificant relationship between any of the variables and

PAGE 31

24 announcement date abnormal returns, he did find a signif icant difference in the reactions to straight debt and convertible debt. Although straight debt caused no sig nificant reaction in equity returns, convertible debt caused a significant abnormal return of -1.25%. Since Vu found no relationship between increase in tax shield and equity impacts, he interprets this as evidence that lever age-increasing issues do not necessarily increase stock holder wealth. It has been demonstrated in many of the above tests that the equity price impact associated with the announce ment of a security issue is not a function of increased or decreased tax-shield. In each case the only significant variable in determining announcement date equity price impacts is the type of security offered. And, the impact from announcing convertible bonds consistently falls between those from announcing common equity or straight debt. Theoretical Valuation of Convertible Bonds That convertible bonds contain components of both debt and equity is widely accepted, and the above empirical results would tend to support that belief. Further, it is reasonable to assume that it is this mixture of debt and

PAGE 32

25 equity that causes the announcement date reaction for con vertible bonds to be less than that for common equity but more than the reaction to announcing straight debt. However, this is only conjecture without a theoretical basis for the assumption that convertible bonds are indeed valued as part debt and part equity. Ingersoll (1977) pre sented a thorough model for the valuation of convertible bonds and demonstrated the source of their debt and equity components. He began by stating the value of a straight, non-callable bond to be a function of firm value, V, matur ity, t, coupon, c, and final payment, B. Letting F denote the value of a straight, non-callable bond, then (4) F = F(V, t; B, c). If we let G be a non-callable bond equivalent to Fin all respects except that it is convertible into fraction, a, of the firm's post-conversion equity, Ingersoll showed that G can be represented as (5) G = G(V, t; B, c, a). Note that if G is convertible into a fraction a= O of the firm's equity, G is identical to F. If we now allow the same bond to be callable, and let H denote the value of this callable convertible, the relation ship can be expressed as follows:

PAGE 33

26 (6) H = H(V, t; k(t), B, a) where k(t) is the call price at time t. If k(t > O) is infinite, H( ) is equivalent to G( ) . In words, the value of a callable convertible with infinite call price prior to maturity is the same as a non-callable convertible with otherwise identical characteristics. In fact, the difference in value purely out the ference in the for callable and non-callable debt arises firm's right to call the issue. The dif value of straight, non-callable debt and that of callable, convertible debt is then due to the com bination of the call feature and the conversion feature. Ingersoll shows that the value of convertible debt must be a non-decreasing function of a, the fraction of equity into which the bond is convertible, and k(t), the price at which the convertible is callable. Therefore, a convert ible issue, G( ) , must be at least as valuable as a noncon vertible issue, F( ) . And, since a non-callable issue can be thought of as a callable with infinite call price, a callable convertible cannot be more valuable than a noncall able one. That is, G( ) H( ) . Ingersoll continues by showing the value of a callable, convertible discount bond to be

PAGE 34

(7) where 27 H(V, t) = F( ) + W( ) + [F1( ) F2( ) ] F ( ) = W( ) = The value of a non-callable, non convertible discount bond of matur ity, t, face value, B, and issued by firm with value V. (See equa tion 4 for the equivalent coupon bond.) The value of a warrant issued on the same firm convertible into a portion, a, of the firm's value. (Ingersoll considers the firm to have only common stock and one issue of convertible debt outstand ing.) F 1 ( ) and F 2 ( ) represent the reduction in value due to the call feature. This call feature reduction in value implicitly considers the reduction in the value of the bond, F( ), as well as reduction in the value of the warrant, W( ) . In other words, the value of the noncallable straight debt portion of the convertible is reduced by the addition of the call feature, and the value of the war rant, the investor's option to convert to common equity, is reduced by the firm's ability to determine its maturity through calling the issue. That is, there are two compo nents to the security which bondholders "sell back" to man agement which allow management to determine both the matu rity of the bond and that of the option to convert. Ingersoll shows that the value of the equity portion of the convertible is a function of the post-conversion value

PAGE 35

28 of the percentage of firm held if the bonds are converted and the maturity of the "attached" warrant. The value of this warrant is derived through the Black-Scholes model (1973), and is shown to be a positive function of stock price and maturity, and a negative function of exercise price (conversion price). Although Ingersoll does not present a closed form model for pricing the various components of a convertible bond, he has demonstrated the theoretical basis for valuing them as both debt and equity. The exact valuation of the compo nents of the convertible bond is a topic for future re search. It is the purpose of this study to demonstrate that con vertible bonds vary in the extent to which they are valued as equity and, further, that announcement date equity price impacts are a function of this equity valuation. It is hy pothesized that those convertibles perceived to have a rela tively large equity component will impact much the same as an issue of common equity. If the market feels the conver sion terms will never be attractive enough to warrant con version, from the standpoint of the firm or the investor, the convertible will be perceived as mostly debt and will impact much the same as an issue of straight debt.

PAGE 36

CHAPTER THREE HYPOTHESES The hypotheses tested in this study are divided into two categories: (1) those dealing with equity price impacts at the announcement of a firm's intent to issue con vertible bonds, and (2) those dealing with equity price impacts at the date of issue. As such this chapter is divided into two sections dealing with the respective hypotheses. Announcement Date The Announcement Date Event Investors continually gather and analyze information about publicly held firms, and, at any point in time, some consen sus expectation will exist as to the probability that a giv en firm will issue convertible bonds. Whether changes in this probability have an effect on equity share prices is the subject of this dissertation. The announcement by management of the intent to issue convertible bonds will have no effect on equity prices if the announcement provides no information of value to market participants. In an efficiently priced market this has two 29

PAGE 37

30 possible explanations. First, the issuance of convertible bonds might be an economically important event, but the announcement of such is fully anticipated by the market. That is, knowledge of a convertible bond issue is valuable, but the public announcement by management is not. Second, the issue announcement might be unexpected but economically unimportant. While it is theoretically conceivable that a convert ible bond announcement could be accurately predicted by the market, as a practical matter it is doubtful. Thus, we will assume that the announcement does change the public's expectations that the firm will issue convertibles. Given this assumption, the nature of equity price movements at the date of announcement can be used to infer the type of information contained in the announcement. For example, if share prices remain unchanged at the date of public announcement, we can infer that the announcement contains no economically important information. However, if on average prices fall by a statistically significant amount, the announcement contains negative information. Investors become aware of the announcement at the actu al announcement date through various wire services, compu ter networks, news services, etc. However, the announce ment might be made during or after market trading hours,so any response to the announcement can take place on that day or on the following day when the announcement appears in

PAGE 38

31 the Wall Street Journal or other financial news media. To capture any equity price impact on either of these days, this study utilizes a two-day window consisting of the date the issue is announced in the Wall Street Journal and the preceding day. The Announcement The type of information disclosed at the announcement date is important since this could have implications about possible price impacts at the date of issue or withdrawal. As part of this study over 200 convertible bond announce ments were read in the Wall Street Journal. Virtually all of these announcements contained only the following type of information: (1) that the firm intended to sell convertible bonds, (2) the approximate size of the issue, (3) the name of the underwriter(s), and (4) a period during which the issue might be sold. The proposed use of the proceeds was occasionally mentioned, but only in very broad terms such as "general corporate needs." Certain potentially important information is not reveal ed in a convertible bond announcement, however. In partic ular, the issue's conversion terms, coupon rate, maturity, and call features are not available until the issue date. Expectations about such terms can be formed at the announce ment date given the market's knowledge of past convertible offerings, current capital market conditions, the financial

PAGE 39

32 status of the issuing firm, etc., They will not be known with certainty, however, until the issue is sold. Announcement Date Impact One purpose of this study is to investigate why convert ible bond sales have a negative impact on the stock price of the issuing firm. However, as noted above, certain in formation is not disclosed until the actual issue date. Thus, might any be price due to impacts observed at the announcement date the fact that a firm intends to issue a generic convertible bond. Evidence pointing to the true source of the negative returns, however, might not be ob served until the actual issue date when all terms are pub licly disclosed for the first time. For example, if a larger equity component inherent in a convertible bond leads to a larger negative equity return, we might not be able to observe such a relationship until the issue date ... the date when the public is told the terms of the issue. Announcement date equity price impacts might be due to the equity component inherent in any convertible bond issue but be unrelated to the, as yet, undisclosed terms of the specific issue in question. Even the probability that the firm will actually sell the convertible is not 1.0 at the announcement date since the firm retains an option to withdraw the issue. Due to the research design of this study, precise statistics on

PAGE 40

33 announced convertible issues that are subsequently withdrawn from offerings (1986), in available. However, security sale are not are frequently withdrawn. Mikkelson and Partch drawn sample of security offerings a randomly from 1972 to 1982, find 23% (18 of 80) common stock offer ings cancelled, 20% (35 of 172) straight debt offerings can celled, and 24% (8 of 23) convertible bond offerings can celled. Therefore, investors must not only estimate any potential economic impact from the firm's proposed issue of convertible bonds, they must estimate the probability that the issue will actually be sold. In short, equity price impacts might be observed at both the announcement and issue dates, but the nature of such impacts might be different due to different infor mation available at each date. Announcement Date Hypotheses H 01 : The announcement of an issue of convertible bonds has no impact on the announcing firm's equity value. This test will examine whether the announcement of an issue of convertible bonds is an economically important event. If the announcement is fully anticipated by the mar ket and the market efficiently interprets the information, there should be no effect on the firm's stock price. If, on the other hand, the announcement of convertible bonds

PAGE 41

34 imparts new information to the market, the effect on the firm's equity value should be reflected in the firm's price per share. It is anticipated that the announcement of an issue of convertible bonds cannot be accurately predicted by the mar ket, and that it is an economically important event as Myers and Majluf (1984) or Miller and Rock (1985) would predict. As such, a significant negative average equity price response is expected at the announcement date. The announcement of an issue of convertible bonds causes no increase in the variance of daily stock returns. On any given day excess returns for a sample of firms will be expected to have a zero mean and some non-zero vari ance due to insistently movements in stock prices. That is, on any day the movement of a firm's stock may result in a positive or negative excess return, but over a large sam ple of firms the average excess return on that day should be zero. The release of information, however, could affect the distribution of excess returns, and the type of infor mation can be a factor in the effect. For example, there might be a consensus that on average the type of informa tion released is a negative signal about true equity values. Then, most firms announcing the same type of infor mation will experience a negative equity price impact on

PAGE 42

35 that date, and the average excess return for a sample of firms making the same type of announcement will be nonzero. If the announcement made by each firm does not contain exactly the same information, however, there might be vary ing equity price impacts. For example, consider the follow ing which demonstrates that the return for any given firm on any given day depends on some normal flow of information as well as any information that might be contained in an unexpected convertible bond announcement: where and where R = the total return. R = n R = C the return due to some "normal" flow of infor mation. the return due to information contained in an unexpected convertible bond announcement. (1) VAR(R) = (2) (3) (4) VAR(Rn) + VAR(Rc) + 2COV(Rn,Rc) (1) = the variance of the return on any given day. (2) = the normal variance caused by the usual flow of information. (3) = the variance caused by information contained in the convertible bond announcement. (4) = the covariance of information flow information in announcement. the returns due to normal with returns caused by any the convertible bond

PAGE 43

36 On a typical non-event day, terms (3) and (4) are zero, and the variance of the firm's return is simply VAR(Rn). However, with the release of the convertible bond announcement term (2) might become non-zero, and the value of term (3) depends upon the correlation, r, of Rn and Re. That is, if managers only make convertible bond announcements on "good" days, rRn,Rc = -1. o. If managers pay no attention to the normal information flow when making these "bad" public announcements, rRn Re is more or less , zero, and if convertible bond announcements are released only on "bad" days, rRn,Rc = +1.0. VAR(R) can be summed up as follows: if rRn,Rc O, VAR(R) > VAR(Rn) if rRn,Rc < o, VAR(Rn) < [VAR(Rn) + VAR(Rc)J In either case, as long as VAR(Rc) > o, the variance of the daily returns will probably increase with the an nouncement of a convertible bond issue. If rRn Re is neg, ative, as a practical matter it is probably not -1.0. And, even if the correlation of "normal" returns with these "abnormal" returns was perfectly negative, term (4) will not necessarily make VAR(R) < VAR(Rn). Firms announcing convertible bonds come from a variety of different industries, vary greatly in size, and have var ying capital structures and degrees of financial soundness.

PAGE 44

------------37 In that case, even if firms issued identical convertible bonds, the information contained in the convertible bond announcement might be valued differently from firm to firm. And, the equity price impacts might vary across the firms, even if they are mostly negative, causing VAR(Rc) to be firm-specific. When a firm announces a convertible bond issue, it could be signalling information about its assets in place, Myers and Majluf (1984), or its projected cash flows, Miller and Rock (1985). Investors must analyze each firm and issue to determine the type and value of any signal. On average the announcement date equity impact could be negative. But, since the signal can vary greatly across a large sample of pected that the convertible bond announcements, it is ex announcement of convertible bond issues will be met with not only a negative shift in the distri bution of daily excess returns but also an increase in the variance of those returns, indicating that the announcement of an issue of convertible bonds has a significantly differ ent impact across firms. H 03 : Variation in the relative values of the debt and equity components of the convertible bond has no effect on any impact on the announcing firm's equity value. Using the reasoning of Myers and Majluf (1984) the type of security issued signals something about the condition of

PAGE 45

38 the issuing firm. Specifically, issuing equity signals weakness. The the literature empirical results of studies discussed in survey support Myers and Majluf. In all cases the announcement of an equity issue is met with sig nificant negative abnormal returns. The announcement date equity price impacts for firms selling straight debt are insignificant. The straight debt component of convertible bonds, fixed coupons and maturity, should cause no announcement date equity price impact. However, convertible bonds may be con verted into the common equity of the issuing firm at the discretion of the holder or may be forced into equity by the issuing firm. 2 Therefore, they must be considered part equity, and this equity component may affect the value of the firm's outstanding common equity. In that case the announcement date equity price impact would be directly related to the size of the equity component. If market participants can reasonably predict the equity component inherent in a given announcement, then the announcement date equity price impact will be directly re lated to the relative size of the underlying equity compon ent. Hence, those convertibles perceived as having a rel atively large equity component will cause a greater (more negative) equity price reaction than issues with relatively smaller equity components.

PAGE 46

39 H 04 : Issue size has no effect on any equity price impact at the announcement date. This hypothesis tests whether issue size, measured rela tive to the market value of the issuing firm's equity, is important in determining announcement date equity price impacts. Miller and Rock (1985) maintain that, by going to the financial markets for external funds, the firm signals an unexpected short fall in cash flows. Any announced need for external financing will be interpreted as negative information by the market and should result in a negative reaction in stock price. Myers and Majluf (1984) agree but state further that weaker firms issue equity. Convertible bonds contain characteristics of both debt and equity. If we assume that the size of the equity component is not a function of issue size, a sample of large convertible bond issues would be expected to have a larger average equity valuation than a sample of small issues. On average, then, we would expect relatively larger issues of convertibles to cause a greater announcement date equity price impact. H 05 : Firm size has no effect on any equity price impact at the announcement date. According to Miller and Rock (1985), any unexpected need for external funds signals a short fall in internally generated funds. Since this will be read as a negative

PAGE 47

40 signal, the firm's equity will be analyzed and reduced in value accordingly. Myers and Majluf (1984) maintain that a firm's need for external financing signals that its assets are over valued By definition, the announcement is unex pected and should cause reevaluation of the firm's equity. If we assume that the unexpected component of any re lease of information is the cause of stock price adjust ments, then the size of the announcing firm could influence the equity price impact at the convertible bond announce ment date. For example, information on a larger firm may be more readily available than that on a smaller firm, especially if the larger firm is older and/or has gone to the markets in the past. Or the larger firm may be listed on one of the larger exchanges. In any case, expectations would be that actions of larger firms should be more easily monitored. It has already been established that the announcement of an issue of convertible bonds is an economically impertant event. By definition, then, the announcements are unexpected signals of weakness. By the above argument, in formation on larger firms should be more complete, and a signal of weakness might be more of a surprise when issued by a larger firm. If a relationship exists between firm size and announcement date equity price impacts, it would be expected that the largest firms would experience the greatest (most negative) impacts.

PAGE 48

41 Issue Date The Issue Date Event In an efficient capital market, any impact from issuing securities should be experienced when the issue is first announced. Any economically important information con tained in the announcement should be analyzed immediately by market participants, and the firm's stock price should reflect any necessary adjustments instantaneously. The date the issue is actually sold should be of no economic importance, since selling the issue is simply the trans action announced previously and has already been priced by the market. However, this makes two important assump tions: (1) all necessary information is contained in the announcement, and (2) there is no uncertainty that the announced issue will be sold. Violation of either of these assumptions can cause an issue date equity price impact. As noted earlier, the announcement of a convertible bond issue might not contain all the information necessary to price the impact of the sale at the announcement date. The typical announcement contains information only that the firm plans to sell con vertibles, the underwriter(s), an estimate of when the sale may take place, and some probable uses of the proceeds.

PAGE 49

42 A potentially crucial item, the conversion ratio, is only released immediately prior to the sale and, therefore, can only be estimated at the announcement date. 3 If the con version terms have economic importance, the equity price impact can only be estimated at the announcement date. Then, release of the terms at the issue date may constitute new information and cause a reevaluation of the issue and an adjustment in stock prices. The second implied assumption is that investors know with probability one that the announced issue will not be withdrawn. If any doubt remains that the issue will be sold, the full equity impact might not be felt at the an nouncement date. The market will continually revise esti mates of if and when the issue will be sold, but the proba bility of issue can never be one until the sale actually takes place. Issue Date Impacts At the sale or cancellation date, then, all uncertainty is resolved. If the issue is to be sold, the terms of the issue are made public, and the market can properly assess the economic importance of the particular convertible. The firm's stock price will then be adjusted accordingly. If the issue is withdrawn, however, the announcing firm should experience a positive stock price reaction as found by Mikkelson and Partch (1986).

PAGE 50

43 Although not tested empirically, it seems reasonable to assume that the average issue date impact for a mixed sam ple of firms selling and firms withdrawing convertible bond issues might be zero. Since this study includes no issues that were withdrawn, the average issue date impact will only be some combination of the economic impact of releas ing the issue terms and resolution of uncertainty about whether the issue would be sold. Since there will be no firms experiencing the positive impact from withdrawing the announced issue, it is expected that the average issue date impact will be negative. Issue Date Hypotheses H 01 : The sale of an issue of convertible bonds has no impact on the issuing firm's equity value. Any issue date equity price impact will depend upon whether or not economically important information is releas ed when convertible bonds are sold. Anything that has the capacity to affect the market's estimate of the firm's equi ty value can be construed as information. For example, many of the convertible bond's terms, such as the conver sion rights, coupon, and maturity, are not released until the issue is actually sold. If these terms are critical in determining the equity price impact of issuing the convert ible, and if the market is not capable of precisely estimat ing them at the announcement date, it may be reasonable to expect their release to constitute information.

PAGE 51

44 Even if the market can accurately predict the terms of the issue and the issue's impact on the market value of the firm's equity, there remains uncertainty that the issue will be sold. In that case, only a portion of the total equity impact will be felt at the announcement date as the market waits to see the final disposition of the issue. If the issue is withdrawn, the firm might recoup all or part of its announcement date loss as documented by Mikkelson and Partch (1986). If, however, the issue is sold, and the exact sale date has not been predicted by the market, the firm could experience a further reduction in equity value. Whether it is the release of the terms of the issue or simply the resolution of uncertainty about the sale or with drawal of the issue, it is expected that economically impor tant information is released at the issue date, and the average issue date equity price impact will be negative. The sale of an issue of convertible bonds causes no increase in the variance of daily stock returns. That firms on average experience a negative equity price impact on the date convertible bonds are sold has been established. If the noted equity impact represents only a shift in the mean issue date excess return, we would say that it is due to resolution of uncertainty. That is, if firms sell convertible bonds, there is a more or less constant equity cost that will be assessed. If, however,

PAGE 52

45 along with a shift in the mean comes an increase in the cross-sectional variance of excess returns, we would say that issue date equity price impacts vary significantly across firms. Selling an announced issue of convertible bonds conveys different information about different firms. There are four scenarios that might explain the nega tive average equity price impact at the sale of convertible bonds: (1) There is a common equity price impact from issu ing convertible bonds, and the probability of issue is con sidered about equal across firms. The sale of the issue only constitutes resolution of uncertainty about whether the issue will be sold, and any issue date impact is more or less equal across firms. The issue date reaction is a shift in the distribution of daily excess returns. (2) The probability of sale or withdrawal is considered constant across firms. The equity price impact from issuing convert ible bonds is firm specific but accurately estimated at the announcement date. Any impact at the issue date is the probability of and, therefore, reaction is a withdrawal times the true equity impact, it can vary across firms. The issue date shift in mean and increased cross-sectional variance. issue of firms, but accurately terms of (3) The probability of selling an announced convertible bonds is considered constant across firm specific equity price impacts cannot be predicted at the announcement date. If the a convertible bond issue are necessary in

PAGE 53

46 calculating equity price impacts, and they can not be ac curately estimated, their release at the issue date is ec onomically important. Each firm will experience an issue date equity price impact according to how management sets the terms. Then, since terms vary from issue to issue, it is reasonable to expect the issue date equity impacts to vary in size even though they are mostly negative. The is sue date reaction is an increase in cross-sectional vari ance of daily excess returns and a negative shift in the av erage daily excess return. (4) The probability of selling an announced issue of convertible bonds varies across firms, and the firm specific equity price impacts cannot be accurately predicted at the announcement date. There is much uncertainty surrounding the signal contained in a con vertible bond announcement. The market tends to penalize firms about the same amount at the announcement date and wait for the release of the terms at the sale date to de termine how much, if any, further adjustment in equity val ue is needed. Much of the equity valuation is left for the issue date, and the effect on the equity value of the selling firms can vary. daily excess return variance. The result is a shift in the average and an increase in cross-sectional Variation in the relative values of the debt and equity components of the convertible bond has no effect

PAGE 54

47 on any issue date impact on the issuing firm's equity value. The probability of conversion is a potentially impor tant variable in the market's assessment of the signal con tained in a convertible bond announcement, and where manage ment sets the conversion terms could have an effect on this probability. For example, if the conversion price is set very close to the firm's current stock price, conversion in the near future might be highly probable. In that case, the market might consider the convertibles no more than a delayed equity issue. Setting the conversion price very high the with market respect to to view the current stock price could cause conversion as improbable. The issue could be considered non-convertible from a practical stand point. Since the terms of a convertible bond issue are not re leased until the issue date, it is reasonable to expect in vestors to estimate announcement date equity price impacts based on some this study, version price the average conversion terms or CP ratio. For CP ratio is defined as the ratio of conto current stock price. The CP ratio could be thought of as a measure of the value of the issue's call option to the firm. A low CP ratio would indicate that the call option is "near the money." That is, management might not have to wait long before the stock price is greater than the conversion price and they can call the bonds and

PAGE 55

48 force conversion to equity. A higher CP ratio for the same firm could indicate less probability of forced conversion in the near future. If, at the issue date, the terms of the issue vary from the perceived average, there could be a reaction by market participants to further reevaluate the issuing firm's equi ty. It is expected that issues with lower than average CP ratios will be viewed more as equity than the announcement would indicate. The issue date equity price date impact impact for those firms should more negative than for firms setting terms at or above the average. H 04 : Issue size has no effect on any equity price impact at the issue date. Any issue date equity price impact must be a result of the release of information or the resolution of uncertainty that the announced issue would be sold. Rejecting the null hypothesis, then, suggests a relationship between issue size, measured as the size of the convertible bond issue relative to the market value of the firm's equity, and the market's ability to accurately predict equity price impacts and/or the probability that the announced issue will be sold. The arguments of Miller and Rock (1985) and Myers and Majluf (1984) suggest a reaction to the announcement of a proposed issue of convertible bonds. Both arguments

PAGE 56

49 suggest that the size of the issue may affect the market's announcement date reaction. But even if the size of the issue has an effect on announcement date equity price impacts, there is no theoretical argument which suggests that issue size can cause systematic under or overesti mation of equity price impacts or probability of issue. There should be no recognizable relationship between rel ative size and any issue date equity price impact. Firm size has no effect on any equity price impact at the issue date. Issue date equity price impacts must be the result of the release of economically important information. As dis cussed above, this information can take the form of above or below average conversion terms or other information con tained in the issue terms, or it can be simply the resolu tion of uncertainty that the issue would be sold. that ing be the the be be cess There is no theory that would support the hypothesis any of these would be more or less predictable accordto the size of the issuing firm. That is, there should no systematic over or underestimation of the terms of issue associated with the size of the issuing firm, and probability of sale has not been shown empirically to related to the size of the announcing firm. There might an increase in the cross sectional variance of daily exreturns and even a shift in the average excess return.

PAGE 57

50 However, there should be no discernible relationship be tween any observed issue date impacts and the size of the issuing firm. Notes 1. For announcements to be classified as information, they must be unexpected or at least not fully anticipated. 2. If the conversion price is below the market price of the underlying stock, the issuing firm can force bond holders to convert to equity by calling the issue. 3. The conversion ratio is the number of shares into which each bond is convertible. The lowest conversion ratio in the sample is 6.58 and the highest is 142.86. Al though this might seem quite variable, it is somewhat misleading. To better compare conversion ratios, con version prices are computed ($1000/conversion ratio) and compared to the firm's current stock price. This is called the CP ratio in the study and ranged from 1.00 to 1.91 with the majority falling between 1.13 and 2.00. Looked at from this angle, there is not a great difference in the terms of most convertible bonds.

PAGE 58

CHAPTER FOUR DATA An initial sample of 558 outstanding convertible bond issues was obtained from Moody's Bond Guide. 1 As stock price information is critical to the study, the issuing firm must be listed on the Center for Research in Security Prices (CRSP) tapes. 259 issues were deleted from the sam ple because they were (1) issued prior to July 2, 1962, (2) the issuing firm was never listed on the CRSP tapes, or (3) the firm was listed after issuing the convertibles . 2 The date of sale or issue date was identified in Moody's Industrial Manual. Twenty convertibles that were issued along with other securities were removed from the issue date sample. Of the 20, 15 were issued with common stock, four with straight debt, and one with preferred stock. Another 25 were deleted from the sample for other reasons. Six of the issues were convertible into the common stock of a parent company, two were convertible into pre ferred stock or preferred plus common, and one issue was convertible into straight debt. Seven issues were deleted 51

PAGE 59

52 because they were issued outside the United States, and four were deleted because detailed information was unavailable. One issue was deleted because the exact issue date was unclear. This leaves a total of 298 issues in the issue date sample. 3 Original announcements of the proposed convertible bond issues were located in the Wall Street Journal. 4 Of the 298 convertibles in the issue date sample, fourteen issues were deleted from announcement date testing for release of confounding information along with the announcement of the proposed issue of convertible bonds. This confounding information was the announcement of earnings (9 issues), a dividend issue), change or any (1 issue), a change in bond ratings (1 other information with potential for economic impact on the firm's stock price (3 issues). Another 53 issues were deleted from announcement date testing because no information was released in the Wall Street Journal prior to the issue date. 5 This left a final announcement date sample of 232 convertible bond issues. Table 1 contains a list of the data gathered, when available, for each issue in both samples. The convertibles in the final sample were issued over the years 1963 to 1986. Table 4-2 shows the frequency of issue per The typical 20 or 25 year along with the maturities of the issues. convertible bond is issued with a maturity of years. Only seven have maturities less than 20

PAGE 60

53 years, and only two of those are less than 15 years. This is not unexpected since virtually all convertible bonds are callable, and the value of the call option to the firm is a positive function of the length of maturity of the option. A survey by Brigham (1966) revealed that managers might look at convertible bonds, with their generally reduced cou pons, as "cheap" debt. Table 4-2 shows average yields on long term (> 10 years) treasury issues and AAA rated cor porate bonds for each year in the test sample. If managers are simply trying to avoid high borrowing costs, convert ible bonds should be particularly attractive during periods of high or climbing interest rates. over 61%, 184 of 300, of the bonds in the sample were issued between 1965 and 1971. 6 During this time, interest rates on AAA corpor ates climbed steadily from 4.57% to 7.28%, and those on government issues also rose. It appears that, during the earlier played 30% of years in a part in the issues the sample, cheap borrowing might have convertible bond sales. However, about in the sample were sold from 1980 to 1986, and interest rates were generally falling over this period. Also, during the eight years beginning with 1972, very few convertible bonds were issued, and interest rates were rising over this period. This would seem to indicate that now there are motives other than reduced coupons for selling convertible bonds. Tables 4-3, 4-4, and 4-5 show the distribution of issue, firm, and relative sizes, respectively. The average

PAGE 61

54 convertible bond issue in the sample is about $51.9 million. Ninety percent of the issues are $100 million or less. The smallest issue sold was $5.5 million, and the largest was just over $250 million. The smallest firm sell ing convertible bonds had total equity value of just over $6.6 million. That for the largest was almost $7 billion, and the average firm was valued at about $550 million. Since not all firms in the issue date sample are included in the announcement date sample, Tables 4-4 and 4-5 show the figures for the announcement date firms at both the announcement date and the issue date, and the total issue date sample of firms. Comparing the issue date distri bution of the firms in the announcement date sample and the distribution of "All Firms Issue Date" shows the distribu tion of the issue date firms that were not included in the announcement date tests. Table bonds is equity. 4-5 shows that the average issue of convertible about 19% of the value of the issuing firm's Over 30% of the issues are 10% or less of the value of the firm's equity, and over 89% are less than 35%. Table 4-6 gives the distributions of CP and conversion ratios. The conversion ratio is the number of shares of common stock Dividing the sion ratio the current into which an individual bond is convertible. face value of the convertible by the conver yields the conversion price, which, divided by stock price, gives the CP ratio. The

PAGE 62

55 conversion ratio and CP ratio are not necessarily related, however. Much depends on the firm's current stock price. For example, consider the following: Firm Conversion Ratio 40.00 Conversion Price $25.00 Current Stock Price $25.00 CP Ratio 1.00 Firm Conversion Ratio 40.00 Conversion Price $25.00 Current Stock Price $15.00 CP Ratio 1.67 A 30.00 $33.33 $25.00 1.33 B 30.00 $33.33 $15.00 2.22 20.00 $50.00 $25.00 2.00 20.00 $50.00 $15.00 3.33 The average conversion ratio was approximately 28.4, indicating that the average bond in the sample was convert ible into just over 28 shares of the issuing firm's common stock. Most of the bonds were convertible into between 10 and 40 shares of common. The call option on a convertible bond allows the firm the opportunity to force conversion of the bonds to equity once the firm's stock price is above the conversion price.

PAGE 63

56 The CP ratio, therefore, measures how far "out of the mon ey" the option to force conversion is. The minimum CP ratio was 1.0, and the maximum was 1.91. The average CP ratio was 1.17, with over 95% falling between 1.00 and 1.3. Further, almost 60% of the CP ratios were between 1.1 and 1.2. Table 4-7 shows the distribution of 2-day excess re turns at both the announcement and issue dates. The aver age announcement date return is -1.43%. The minimum and maximum, respectively, were -10.55% and 10.72%. Approx imately 73% of the firms, 169 of 232, experienced negative announcement date 2-day excess returns. The average issue date 2-day excess return was -0.61%, and, again, about 73%, 219 of 298, of the returns were negative. The minimum 2day issue date excess return was -12.97%, and the maximum was 15.13%. NOTES 1. Issues were found outstanding in at least one of the Moodv's Bond Guides dated 1968, 1970, 1972, 1975, 1977, 1980, 1982, 1984, or 1986. Therefore, any issue called before early 1968 or issued in the latter part of 1986 is not in the sample. 2. The CRSP tapes began recording stock price and return information on July 2, 1962. 3. None of the 298 issues in the final sample was issued to retire outstanding convertible bonds. 4. The (as typical announcement it appears in the for a convertible bond issue Wall street Journal) contains

PAGE 64

57 information on the approximate size of the proposed issue, a period of time during which the issue will occur, and the underwriter(s). In most cases the pro posed use of the funds is stated but only in very broad terms such as general corporate needs, refund debt, plant expansion, working capital needs, or some com bination of any or all the above. Only in a very small number of cases was a specific use detailed in the announcement, such as retirement of a specific issue of straight debt or preferred stock. 5. In some cases announcement of the issue came the day the issue was sold or only a day or two prior to the issue date. Usually the announcement and issue dates were separated by three to four weeks. 6. None of the issues in the final sample was sold during 1973, and very few were sold during the period 1972 to 1979. To test whether there were different motives for issuing convertibles and, hence, potentially different market reactions, the sample was divided into two periods: those issues sold from 1963 through 1972, and those sold from 1974 through 1986. The average announcement and issue date two-day equity price impacts for the two samples were virtually identical. The announcement date two-day abnormal returns for 1963-1974 and 1974-1986 were -1.41 and -1.47, respectively. The corresponding issue date returns were -0.56 and -0.67. To test whether equity valuation affected equity price impacts, each sample was divided into three approxi mately equal strata by CP ratio. The three strata for each sample are designated High, Mid, and Low. It was assumed that, although the relative values of CP ratios may vary across comparable strata (ie. high vs. high), each stratum was indicative of a "high", "medium", or "low" CP ratio for its time of issuance. The issue date two-day residuals for the high, mid, and low CP strata for the pre and post 1973 issues were 0.12, -0.82, -1.82 and 0.44, -1.15, -2.14, respectively. Although the high and low strata for each sample were significantly different, comparison of the respective strata for the two samples showed no significance. Hence, it will be assumed that the impact of the relative equity valuation of convertibles does not vary across time.

PAGE 65

58 TABLE 4-1 Data gathered for each issue in the final sample.* ISSUE DATE: The date the issue was sold. ANNOUNCEMENT DATE: The first time the proposed issue is mentioned in the Wall Street Journal. MATURITY DATE: The date on which the issue matures. COUPON: The annual coupon paid to holders of the bond. This is usually paid in two semi-annual payments, the dates for which were also gathered. ISSUE SIZE: The original dollar amount registered with the Securities and Exchange Commission. QUALITY RATING: The original rating given by Moody's Bond Rating Service. This was obtained from Moody's Weekly News Reports in the week following the issue of the bonds. CONVERSION RATIO: The original number of shares into which each bond is convertible. It is used to compute conversion price, the price at which a bond may be converted into the issuing firm's common. It was obtained from Moody's Weekly News Reports in the week following the issue of the bonds. CLOSING STOCK PRICE: The closing stock price two days prior to both the announcement and issue dates. These were obtained from the CRSP Daily Stock Master tape. SHARES OUTSTANDING: The number of shares outstanding two days prior to both the announcement and issue dates. These were obtained from the CRSP Daily Stock Master tape. ANNOUNCED USE OF FUNDS: The use of the proceeds from the issue as announced at the time the bonds were sold. ACTUAL CALL DATE: The date on which the issue was called for redemption by the issuing firm. This was obtained from Standard and Poor's Bond Guide. ACTUAL CALL PRICE: The dollar amount offered to each bondholder for redemption of a single bond. This was obtained from Standard and Poor's Bond Guide. CONVERSION PRICE AT CALL: The conversion price on the date the issue was called. It was usually adjusted over the life of the bonds for stock dividends and/or splits. This and the date on which the conversion privilege expired were obtained from Standard and Poor's Bond Guide. * Unless otherwise noted, data were gathered from Moody's Industrial Manual.

PAGE 66

59 Table 4-2 Frequency of issue and maturity of convertible bonds in the sample and average yields on long term U.S. government and AAA corporate bonds. Convertible Bonds Maturity Ave. Yield* Number Year Issues 25 Yrs 20 Yrs < 20 Yrs us Gov't AAA Corp. 1963 1 0 1 0 4.07 4.30 1964 3 0 3 0 4.14 4.46 1965 10 8 1 1 4.32 4.57 1966 14 9 5 0 4.65 5.33 1967 55 35 17 3 5.11 6.00 1968 37 19 18 0 5.46 6.55 1969 37 29 8 0 6.47 8.03 1970 13 5 7 1 6.28 8.02 1971 18 12 5 1 5.62 7.28 1972 7 6 1 0 5.54 7.17 1973 0 0 0 0 6.23 7.68 1974 4 4 0 0 6.78 8.77 1975 3 3 0 0 7.06 8.99 1976 3 3 0 0 6.58 8.25 1977 0 0 0 0 7.62 8.12 1978 2 0 2 0 8.61 9.00 1979 2 1 1 0 9.67 10.15 1980 20 16 4 0 11.75 12.86 1981 9 8 1 0 13.30 15.36 1982 4 3 1 0 11.61 13.91 1983 18 12 6 0 11.49 12.30 1984 10 9 1 0 11.94 12.04 1985 16 10 6 0 10.29 11.45 1986 14 12 1 1 7.93 9.24 Total 300 204 89 7 Source: Moody's Industrial Manual

PAGE 67

60 Table 4-3 Convertible Bond Issue Sizes ($ Million) Issue Size Issues 250 < IS 1 200 < IS < 250 7 175 < IS < 200 2 150 < IS < 175 11 140 < IS < 150 1 130 < IS < 140 0 120 < IS < 130 6 110 < IS < 120 2 100 < IS < 110 25 90 < IS < 100 0 80 < IS < 90 3 70 < IS < 80 16 60 < IS < 70 20 50 < IS < 60 36 40 < IS < 50 18 30 < IS < 40 43 20 < IS < 30 46 10 < IS < 20 47 IS < 10 16 . AVE. IS = 51.9

PAGE 68

61 Table 4-4 Distribution of Firm Sizes, Announcement and Issue Date. Announcement Date Sample Firm Equity Value Announ. Issue All Firms ($Million) Date Date Issue Date 5000 < FS 1 1 1 4000 < FS < 5000 2 1 1 3000 < FS < 4000 6 7 8 2000 < FS < 3000 3 3 5 1000 < FS < 2000 14 18 20 900 < FS < 1000 7 4 4 800 < FS < 900 5 4 6 700 < FS < 800 7 7 10 600 < FS < 700 11 11 13 500 < FS < 600 4 8 10 400 < FS < 500 12 9 12 300 < FS < 400 18 21 22 200 < FS < 300 25 19 28 100 < FS < 200 44 48 62 90 < FS < 100 7 10 12 80 < FS < 90 7 2 3 70 < FS < 80 5 9 11 60 < FS < 70 8 4 4 50 < FS < 60 5 8 9 40 < FS < 50 6 2 3 30 < FS < 40 7 6 7 20 < FS < 30 5 5 7 10 < FS < 20 3 3 4 FS < 10 l 1 3

PAGE 69

62 Table 4-5 Relative Sizes of Convertible Bond Issues, Announcement Date and Issue Date Announcement Date Sample Relative Announ. Issue All Firms Size ($Million) Date Date Issue Date 1.00 < RS 2 3 4 0.90 < RS < 1.00 2 1 1 0.80 < RS < 0.90 0 0 0 0.70 < RS < 0.80 3 3 4 0.60 < RS < 0.70 1 0 2 0.50 < RS < 0.60 0 0 2 0.40 < RS < 0.50 5 4 6 0.35 < RS < 0.40 6 7 9 0.30 < RS < 0.35 13 15 16 0.25 < RS < 0.30 16 15 17 0.20 < RS < 0.25 23 20 28 0.15 < RS < 0.20 35 38 44 0.10 < RS < 0.15 47 40 49 0.05 < RS < 0.10 44 46 61 RS < 0.05 16 16 18

PAGE 70

100 90 80 70 60 55 50 45 40 35 30 25 20 15 10 5 63 Table 4-6 Distribution of Conversion and Issue Date CP Ratios Conversion Ratios CP Ratios Ratio Issues Ratio < CR 4 1.50 < CP < 2.00 < CR < 100 0 1.40 < CP < 1.50 < CR < 90 1 1.35 < CP < 1.40 < CR < 80 2 1.30 < CP < 1.35 < CR < 70 5 1.25 < CP < 1. 30 < CR < 60 7 1.20 < CP < 1.25 < CR < 55 8 1.15 < CP < 1.20 < CR < 50 7 1.10 < CP < 1.15 < CR < 45 9 1.05 < CP < 1.10 < CR < 40 20 1.00 < CP < 1.05 < CR < 35 24 AVE. CP < CR < 30 48 < CR < 25 55 < CR < 20 49 < CR < 15 33 < CR < 10 5 CR < 5 0 AVE. CR = 28.4 Issues 4 3 2 3 15 34 72 77 37 10 = 1.17

PAGE 71

64 Table 4-7 Distribution of Two-day Excess Returns Announ. Issue Excess Returns (%) Date Date 10.00 < 2AR 1 2 8.00 < 2AR < 10.00 2 1 6.00 < 2AR < 8.00 2 4 4.00 < 2AR < 6.00 9 11 2.00 < 2AR < 4.00 14 29 1.00 < 2AR < 2.00 11 32 0.00 < 2AR < 1.00 24 36 -1.00 < 2AR < o.oo 35 41 -2.00 < 2AR < -1.00 38 55 -4.00 < 2AR < -2.00 46 52 -6.00 < 2AR < -4.00 36 27 -8.00 < 2AR < -6.00 11 6 -10.00 < 2AR < -8.00 2 0 2AR < -10.00 1 2 AVE. 2AR -1.43 -0.61

PAGE 72

CHAPTER FIVE EMPIRICAL DESIGN This is a study of the impact on firm equity value on the date convertible bonds are announced and the date they are actually sold. As such an event study methodology will be employed extensively. Following a discussion of the event study methodology used, the discussion of the empiri cal design is divided into two sections: (1) a discussion of the announcement date tests, and (2) a discussion of the issue date tests. Event Study Methodology Abnormal returns are obtained from the Center for Research in Security Prices (CRSP) Daily Excess Returns Tape. To estimate excess returns, CRSP first estimates each security's beta. Beta is computed for a given year only if the stock traded on at least half of the trading days in that year. The individual beta is calculated as follows: 65

PAGE 73

covi,m = VA~= where reti,t mrett mret3t n 1 t 1: t 66 (reti,tmret3t) (1/n) ( 1: t (mrettmret3t) ( 1/n) ( 1: t mrett) ( L t = log (1 + return for security ion day t) = log (1 + value weighted market return on day t) = mrett-l + mrett + mrett+l (a moving average market window) = number of observations for the year Next, the firms are ranked and divided into 10 portfolios according to beta. Betas used in the ranking are either from the preceding year or, if not available, from the current year. If neither is available, the firm is placed in a portfolio of firms for which no excess returns are calculated. Average daily returns are then calculated over the year for each portfolio. where AR t 1, ARi, t =1/m 1: t = the average return j = l tom for portfolio AR t = the CRSP daily return for firm j J ' m = the number of firms in portfolio i on day t on day t i Daily excess returns for firm j are then the difference between the firm's actual return on day t and the average day t return for the portfolio containing firm j.

PAGE 74

where ER t = the J ' R t = the J ' AR t = the l. ' 67 ER t J ' = Rj,t ARi,t excess return for firm j on actual return for firm j on average return for portfolio day t day t i on day t For each firm in a given sample, 121 daily excess returns are gathered, 60 days before to 60 days following the day of the proposed event. The event will be either the announcement of a convertible bond issue or the date the issue is sold. These daily excess returns are then com bined to form 60 two-day excess returns for each firm, the thirtieth, t = 30, of which contains the event date and the preceding day. The two-day excess returns are then averaged at each time t, t = 1 to 60, across the firms in the sample. where AERt = ER t = J ' AERt = 1/n I j ER t J ' j = 1 ton t = 1 to 60 the average 2-day excess return for the n firms in the sample at time t the 2-day excess return for firm j at time t n = the number of firms in the sample

PAGE 75

68 To test the significance of each average two-day excess return a Student's T-test will be performed. The standard deviation of the 2-day abnormal returns is estimated from the 59 2-day average excess returns from trading days -60 to -2 and +l to +60. Calculating the standard deviation in this manner avoids any potential effects of abnormal vari ance on the event date. In most of the empirical tests the sample will be strat ified in some fashion to identify any equity price impact related to the variable by which the firms have been stratified. To test for a significantly different event date impact between two strata, a new variable will be created which amounts to the difference between the 2-day average excess returns for the two strata. For example, where AERd,t = AERa,t AERb,t t = 1 to 60 the difference in the average 2-day ex cess returns for strata a and bat time t the average 2-day excess return at time t for the firms in stratum a the average 2-day excess return at time t for the firms in stratum b Once the new variable is calculated, it is a simple procedure to perform a T-test to determine significance. As before, the standard deviation of the 2-day excess returns (this time using AERd t) is calculated after '

PAGE 76

69 omitting AERd,JO' the difference between event date average excess returns. If the T-test determines that AERd,JO is significantly different from zero, this would indicate a difference in the event date average 2-day ex cess returns for the two strata. That is, the variable along which the strata have been formed is a factor in event date equity price impacts. Announcement Date Tests of Hypothesis The announcement of an issue of convertible bonds has no impact on the announcing firm's equity value. This first test looks at the overall effects of an nouncing a convertible bond issue. No stratification or classification of any kind is utilized. For all 232 firms in the announcement date sample, 60 2-day excess returns are gathered centering on the announcement date. AT-test is performed as described above to determine whether the announcement date excess return is significantly different from zero. If the return is significant and positive (neg ative), this suggests that the announcement contains good (bad) information. The announcement of an issue of convertible bonds causes no increase in the variance of daily stock returns.

PAGE 77

70 To test whether a convertible bond announcement consti tutes the release of firm specific information, the distri bution of the announcement date 2-day excess returns will be analyzed. On any given (non-event) day a firm may experience a non-zero excess return. Over many firms on that same day or for the same firm over many non-event days, how ever, the average excess return will have an expected value of zero with some positive variance. On an event day the mean excess return for a portfolio of firms might be non zero, and increased uncertainty about the valuation of the firms could cause an increase in cross-sectional stock return variance. The following test was utilized to compare the cross-sectional variance at the announcement date with that of a typical non-event period. The standard deviation of the twenty-nine 2-day excess returns preceding the announcement date is calculated for each firm as follows: sj,a = (1/29 I t where S = 1,a ER t= J ' t = 1 to 29 the standard deviation of pre announcement date excess returns for firm j excess return for firm j at time t An F-test is performed for each firm that compares announcement variance: date variance with pre-announcement date

PAGE 78

where The ERj,30 = V• = J 71 the mean-adjusted announcement excess return for firm j date the pre-announcement date standard dev iation of excess returns for firm j the ratio of the variance of the an nouncement date excess return to that of the pre-announcement date period and is distributed as F (1,30) expected value of V J is n/(n 2) = 29/27. Therefore, we let Uj = (Vj * 27/29), and E (Uj) = 1. 0 and VAR(Uj) = 2(29 3)/(29 6) This is the standard variance formula for an F with = 1 and v 2 = 29 degrees of freedom after the adjustment factor, 27/29, has been applied. See Warner, et al (1987) for more on this procedure. Dealing with a large sample of independent observations allows the use of the Central Limit Theorem in testing whether the mean of the U• J is from one. Specifically, z = ( I (Uj 1))/(2 j j = 1 ton significantly different I (26/23)) 1 / 2 n

PAGE 79

72 As noted above, the announcement date excess return for each firm is mean-adjusted before performing the F-test. Subtracting the average announcement date effect from each firm removes the mean effect from the individual excess returns. This allows testing of the variance alone, and shows whether, aside from some average impact affecting all firms, each firm has a specific equity price impact. In other words, there are two potential impacts at the announcement of economically important information: ( 1) a mean effect representing a shift in excess return distri bution due to some average impact each firm will exper ience, and (2) an increase in variance of equity returns signalling a differing impact across firms beyond the aver age impact. That is, where ER t = J ' AERt = ER t J ' the excess return for firm j at time t the average excess return at time t for then firms in the sample the firm specific equity impact, either positive or negative. A significant z, then, indicates that the signal contained in a convertible bond announcement is valued according to the firm making the announcement. The typical bj,t term is large enough to significantly affect the variance of the announcement date excess returns.

PAGE 80

73 H 03 : Variation in the relative values of the debt and equity components of the convertible bond has no affect on any impact on the announcing firm's equity value. Before this hypothesis can be tested, a proxy for the debt and equity components has to be obtained. It should be noted that since virtually all convertible bond issues are sold at par, if a proxy for either the debt or equity component can be obtained, the other value is automatic. As a proxy for the equity component inherent in each convertible bond, a ratio of the conversion price to cur rent price per share is used and is referred to as the CP ratio. This ratio measures how far the conversion option is "out of the money" or how likely the issue is to be con verted either voluntarily or through a forced conversion. A large CP ratio indicates that the option is far out of the money. The issue is less likely to be converted and should impact more as an issue of straight debt. A small CP ratio would indicate that the issue has a relatively higher probability of being converted and should impact more like common equity. In other words, those issues with small CP ratios are considered to have a relatively larger equity component than an issue with a larger CP ratio. The conversion price for each issue is obtained from Moody's Weekly News Reports during the week following the issue date. Obtaining the conversion price very early in the bond's life is important as it is released immediately

PAGE 81

74 prior to the sale of the bonds and is subject to adjustment for stock dividends or splits during the life of the bond. The conversion price is then divided by the closing stock price two days prior to the announcement date. This stock price was utilized to help avoid any effects of the an nouncement on equity value. The minimum and maximum CP ratios for the announcement date sample were 0.75 and 2.03, respectively, with a mean of 1.15. In option pricing jargon the 0.75 ratio is "in the money", and the 2.03 ratio could be considered "far out of the money." "In the money" would indicate that the firm could immediately call the issue and force its conversion to equity. It is important to remember, however, that these CP ratios are based on conversion terms not available until the issue date. Therefore, these announcement date figures can only be considered expectations. Once (expected) CP ratios are calculated for each firm, the firms are divided into three approximately equal strata according to CP ratio. T-tests are performed to test for significance within as well as between strata. Signifi cance between strata indicates that the announcement date equity price impact is a function of the (expected) rel ative values of the debt and equity components. H 04 : Issue size has no affect on any equity price impact at the announcement date.

PAGE 82

75 To test whether the dollar size of the convertible issue is a factor in the announcement date impact, the firms will be stratified by the relative size of the issue. This is defined as the dollar value of the announc ed issue divided by the market value of the firm's equity. The dollar size of the issue is obtained from Moody's Industrial Manual, and the market value of the firm's quity is calculated as the market price per share times the number of shares outstanding. Both the price per share and the number of shares are obtained from the CRSP Daily stock Master tape, and, to avoid any possible effects of the an nouncement on the share price, the closing price two days before the announcement is used. Once relative size is calculated for each firm for which the data are available, the firms are divided into three approximately equal strata. A significant difference between strata indicates a difference in announcement date equity price impacts based on relative size. H 05 : Firm size has no affect on any equity price impact at the announcement date. To test whether the size of the issuing firm is a fac tor in the announcement date impact, the firms are strat ified by total market value of equity. As in the last test, the market value of the firm is defined as the clos ing stock price two days before the announcement times the

PAGE 83

76 number of shares on that date. significance between strata indicates that the equity price reaction to the announce ment is a function of the size of the issuing firm. Issue Date Tests of Hypothesis Sale of an issue of convertible bonds has no impact on the issuing firm's equity value. This is a general test of the stock price affects of issuing convertible bonds on the date the bonds are actually sold . For the 298 securities in the issue date sample, the sale date is identified in Moody's Industrial Manual. Significant issue date excess returns indicate that a firm issuing convertible bonds will experience an quity price impact at not only the announcement of the is sue but also at the sale date. The sale of an issue of convertible bonds causes no increase in the variance of daily stock returns. As with the announcement date tests, the cross-section al variance of excess returns on the issue date will be com pared to that of a typical non-event period. For this test, however, the standard deviation of post issue date ex cess returns will be compared to issue date excess return variance. The remainder of the test is exactly as its an nouncement date counterpart. A significant z value will

PAGE 84

77 indicate that firm specific information is released when convertible bonds are actually sold. H 03 : Variation in the relative values of the debt and equity components of the convertible bond has no affect on any issue date impact on the issuing firm's equity val ue. Again the CP ratio is used as a proxy for the equity component of the individual convertible bond. To calculate the CP ratio the conversion price as stated in Moody's Weekly News Reports the week after the issue sold is divid ed by the closing stock price for the issuing firm two days before the issue date. The minimum and maximum CP ratios at the issue date were 1.0 and 1.91, with an average of 1.17. After the CP ratio is figured for the 257 firms for which the data are available, the firms are divided into three approximately equal strata. Significance between strata indicates that the relative values of the debt and equity components inherent in the convertibles are a factor in the issue date equity impact. H 04 : Issue size has no affect on any equity price impacts at the issue date. To test whether the value of the convertible relative to the total market value of the issuing firm's equity is a

PAGE 85

78 factor in any issue date equity price impact, the issuing firms are divided into three approximately equal strata by relative issue size. Relative size is defined as the dol lar size of the issue divided by the market value of the is suing firm's common equity. The dollar amount of convert ibles sold is obtained from Moody's Industrial Manual. The value of the firm's equity is the closing price per share two days preceding the issue date times the number of shares on that date. Both of these figures are obtained from the obtained fects of CRSP Daily Stock Master tape. Closing price is from two days before the event to avoid any ef the release of the terms of the issue either on the sale date or the day before. A T-test is used to test for significance within each stratum as well as between the strata. Significance be tween strata indicates that the issue date equity price im pact is a function of relative size. H 05 : Firm size has no affect on any equity price impact at the issue date. The firms are divided into three approximately equal strata according to market value of equity. Market value of equity is defined as the closing price per share of com mon two days before the sale date times the number of shares outstanding on that date. The source for these data is the CRSP Daily Stock Master. Significance between the

PAGE 86

79 strata indicates that firm size is a factor in the equity price impact at the date convertible bonds are sold.

PAGE 87

CHAPTER SIX RESULTS This chapter is divided into four sections: (1) a discussion of the results of the announcement date tests, (2) a discussion of the results of the issue date tests, (3) a summary of the results of the announcement and issue date tests, and (4) a discussion of additional tests performed. Announcement Date Test Results The announcement of an issue of convertible bonds has no affect on the announcing firm's equity. Table 6-la shows the average two-day excess returns for the 121 day period surrounding the announcement of 232 is sues of convertible bonds. Observation 30 contains the an nouncement date, as identified in the Wall Street Journal, and the preceding day. The average announcement date eq uity impact on the 232 firms is -1.43% and is significant at the 1% level. (168) were negative. Of the 232 two-day excess returns, 72% The results indicate that the announcement of an up coming issue of convertibles is received as unfavorable 80

PAGE 88

81 information by the market and has an economically signif icant impact on the market value of the average announcing firm's equity. The null hypothesis that the announcement of an issue of convertible bonds will have no effect on the announcing firm's equity is rejected at the 1% level. Table 6-lb shows average daily excess returns and aver age cumulative excess returns for 121 days surrounding the announcement date. The average equity impacts for the an nouncement day and the preceding day are -0.727% and -0.697%, respectively, and both are significant at the 1% level. The cumulative excess returns are positive and sta tistically significant for most of the sixty days preceding the announcement. The average cumulative excess return reaches 4.25% on day -5 and is significant at the 1% level. The negative returns for the five days preceding and including the announcement date bring down the cum ulative excess return to about 2.4% which is only signif icant at the 5% level. By the first day following the announcement, the average cumulative return is approaching 2.0% and only significant at the 10% level. This cumula tive return and significance is maintained for the ten days following the announcement date, but by day +11 any signif icance is lost. This seems to indicate some validity in the theory that management tries to time security

PAGE 89

82 announcements. According to Myers and Majluf (1984) management will issue a security it deems to be overpriced. The accumulation of over 4.0% in excess returns for the six ty days preceding the announcement and the subsequent loss of this excess return would support Myers and Majluf. The announcement of the impending convertible bond issue sig nals that the firm's equity is not worth as much as its cur rent stock price would indicate. This test has determined that the announcement of an is sue of convertible bonds is a statistically as well as an economically significant event. It will be the goal of the following tests to help determine those characteristics of the issue and/or the firm that affect the market's reaction to the announcement of a convertible bond issue. The announcement of an issue of convertible bonds causes no increase in the variance of daily stock re turns. If the signal contained in an announcement of convert ible bonds is valued according to the announcing firm, the equity impact could vary greatly from firm to firm. The re sulting movement of stock prices in differing amounts, both positive and negative, will cause an increase in the cross sectional variance of stock returns. If the information available at the announcement date is not complete enough to form estimates of equity values, however, the market

PAGE 90

83 might simply assess some more or less constant amount from each announcing firm's distribution of excess would simply shift to ance. The z-score from equity. If this is the case, the returns at the announcement date the left with no increase in varitesting all 232 firms in the announcement date sample was only 0.624, and the null hypothesis cannot be rejected. Since there is no significant increase in the cross-sectional variance of equity values at the announcement date, it would appear that firms are simply charged a flat amount for announcing convertible bonds. This could have two explanations: (1) there is no attempt by the market to estimate the terms and other information not contained in the announcement, or (2) the information omitted from the announcement is of no economic importance. It only matters that the firm intends to issue convertible bonds. Variation in the relative values of the debt and equity components of the convertible bond has no effect on any impact on the announcing firm's equity value. According to Myers and Majluf {1984) managers prefer to issue debt, and issuing equity or convertible bonds is con sidered a negative signal. It follows from their reasoning that the impact from announcing convertible bonds might be due to the equity component inherent in the issue. This

PAGE 91

84 test divides the issues into three strata according to eq uity value. The equity component of each bond in the sam ple is measured by its CP ratio, the ratio of conversion price to stock price at issue date. Table 6-2 shows the re sults of this stratification. The third of the firms with the highest CP ratios exper ienced an announcement date 2-day excess return of -1.04% which is significant at the 1% level. The middle and low est strata experienced 2-day excess returns of -1.63% and -1.84%, respectively. Both are significant at the 1% lev el. Sixty-nine percent (49 of 71) of the high strata 2-day excess returns, seventy-three percent (52 of 71) of the mid dle strata 2-day excess returns, and seventy-eight percent (55 of 70) of the low strata 2-day excess returns were neg ative. T-tests revealed no significant difference between announcement date reactions in any of the strata. No sta tistical relationship is established between the announce ment date impact and the equity component of the convert ible bond. The T-test comparing the announcement date eq uity impacts for the high and low strata produced at-value of 1.43, jected at nouncement and, therefore, the null hypothesis cannot be rethe 10% level. However, a pattern in andate impacts does seem to be present. The ab of the announcement appears to be a negsolute magnitude ative function of creases, the equity CP ratio. That is, as the CP ratio inimpact decreases. This would seem to

PAGE 92

85 indicate that even though the conversion terms are not released until predicting where the issue date, the market is capable of management will set the terms or at least makes an attempt to do so. H 04 : Issue size has no affect on any equity price impact at the announcement date. Miller and Rock (1985) claim that any unexpected an nouncement of external financing signals a short-fall in op erating funds. A necessary implication of their theory is that the impact on the announcing firm's equity should be a positive function of the size of the issue of securities sold. If this is the case there should be a larger impact on the equity of the firm selling a relatively larger issue of convertible bonds. Table 6-3 presents the results when the firms are stratified according to the relative size of the convertible bond issue defined as the dollar value of the issue divided by the market value of the issuing firm's equity. The firms were divided approximately into thirds. The 70 firms impact of with the relatively largest issues experienced an -1.75% which is significant at the 1% level. six percent (53 of 70) had negative 2-day excess Seventy returns. The next stratum experienced an average 2-day ex cess return of -1.21%, again significant at the 1% level. Sixty-eight percent (48 of 71) of the 2-day excess returns

PAGE 93

86 returns were negative for this group. The 71 firms in the stratum average icant with the relatively smallest issues experienced an 2-day excess return of -1.12%. This was signif at the 1% level and 73% (52 of 71) of the excess returns were negative. To test for a significantly different impact across the strata, a T-test was performed. The null hypothesis that the impacts were equal across the three strata could not be rejected at the 10% level. This would indicate that the relative size of the issue of convertible bonds is not a factor in the announcement date impact on the issuing firm's equity, and it does not support the theory of Miller and Rock (1985). H 05 : Firm size has no affect on any equity price impact at the announcement date. According to Myers and Majluf (1984) management will is sue securities when they are overpriced. Further, manage ment would prefer to issue debt. The announcement of the sale of convertible bonds would be a negative signal to the market and would be met with a negative impact on stock prices. Miller and Rock (1985) maintain that any unex pected announcement of external funding is a negative sig nal and would be met with a negative reaction in stock prices. An implication of both is that the predictability of any announcement will affect the reaction by the market.

PAGE 94

87 That is, if the market is able to predict the announcement of the need for external funding and the type of security to be issued, there should be no perceptible impact in con junction with the announcement of the security offering. The availability of information is then critical in deter mining the impact on the announcing firm's equity. If the availability of information varies from firm to firm, market reactions could be firm specific. If, for in stance, information on large firms is more readily access ible than that for smaller firms, the market would be bet ter able to predict the actions of larger firms. For this test the firms were divided into three strata according to the market value of equity. AT-test was performed to de termine significance within as well as among the strata. Table 6-4 shows the results of this stratification. For the 71 firms in the stratum of the largest firms, the average 2-day announcement date excess return was -1.48% and was significant at the 1% level. Seventy-six percent (54 of 71) of the 2-day excess returns were negative. The middle strata experienced a significant (1%) reaction of -1.06% while that for the smallest firms was -1.55%, also significant at the 1% level. The percentages of negative 2-day excess returns were sixty-nine and seventy-three, respectively, for the middle and smallest strata. As with relative size, there was no statistical difference between the reactions of any two strata. This indicates that the

PAGE 95

88 size of the issuing firm is not a factor in determining the announcement date impact, and the null hypothesis cannot be rejected. Table 6-5 presents a summary of the results for the an nouncement date tests. These tests have determined that an nouncing an issue of convertible bonds is an economically important event for the announcing firm, and there is some evidence that management attempts to time the announcement to take advantage of abnormal pricing. The average firm ex periences an announcement date equity price impact of nearly -1.5%. Factors such as firm size and relative size of the issue have been shown generally not to affect announce ment date equity impacts. Even though the size of the eq uity component inherent in convertible bonds was not demon strated to have a statistically significant affect on an nouncement date equity price impacts, the observed pattern of excess returns may indicate that the market attempts to estimate the terms of the issue before they are announced. Issue Date Test Results H 01 : Sale of an issue of convertible bonds has no impact on the issuing firm's equity value.

PAGE 96

89 Table 6-6a presents 60 two-day residuals around the date vation of sale thirty for 298 issues of convertible bonds. Obsercontains the issue date, as identified in and the preceding day. The price impact is -0.61% and is Of the 298 firms issuing conMoody's Industrial Manual, average issue date equity significant at the 1% level. vertibles, 72% equity impact. (168) experienced a negative issue date This indicates that the date convertible bonds are sold is an important event for the selling firm, and the null hypothesis is rejected at the 1% level. Table 6-6b shows average daily and cumulative average abnormal returns for the same period. The average daily abnormal returns for days -40, -14, and -10 show statis tical significance. The average abnormal return on the day the bonds were sold is -0.48% and is significant at the 1% level. Besides the issue date, however, there are no signi ficant daily excess returns between day -10 and day +60, and none of the average cumulative abnormal returns from day -12 to day +60 is significant. Although the issue date itself has been shown to be a statistically significant event for the average convertible bond issuing firm, returns before or explaining either drawing the issue It is important there are no patterns in the cumulative after the event which may be of help in the firm's motive for selling vs. with or even why the issue date is an event. to make note at this time that the

PAGE 97

90 convertible bond issues used in this study were located in Moody's Bond Record and, hence, were already outstanding. This, by definition, excludes any issues that were announc ed and subsequently withdrawn by the announcing firm. Ineluding issue results canceled dates of might issues and treating cancellation dates as alter the results of the test. As the cellation Mikkelson and Partch (1986) imply, the can of convertible bonds may be viewed as a positive signal and cause positive average cancellation date re turns. In that case, including cancelled offerings could push the average issue date abnormal return to zero. Also, at the date the cancelled issues are announced, the market may accurately predict that the issues will be withdrawn, and the announcing firms will experience much smaller an nouncement date equity price impacts. This could greatly affect the overall impacts for firms only conjecture at future testing. average announcement date equity price announcing convertible bonds. This is this point, however, but it warrants The next test of hypothesis tests the nature of the issue date equity price impact. Specifically, whether the issue date impact is a result of the release of new infor mation or simply the result of resolution of the uncer tainty surrounding sale of the issue. The sale of an issue of convertible bonds causes no increase in daily stock return variance.

PAGE 98

91 As discussed in the announcement date tests, an increase in the cross-sectional variance of issue date excess returns will indicate the presence of a firm specific equity impact in addition to the average impact felt by all firms. Comparing the cross-sectional variance at the issue date with the average cross-sectional variance of the 30 post-issue date 2-day excess returns yielded a score of 1.10, and the null hypothesis cannot be rejected at the 10% level. There is no evidence to support the al ternative hypothesis that each firm has a potentially dif ferent issue date equity price impact. The results seem to indicate that firms that sell previously announced convert ible bonds are assessed a more or less constant amount. H 03 : Variation in the relative values of the debt and equity components of the convertible bond has no affect on any issue date impact on the issuing firm's equity value. Table 6-7 shows the results from stratifying the firms according to CP ratio, the ratio of conversion price to cur rent stock price. The stratum of firms with the large CP ratios experienced an insignificant average issue date im pact of +0.19%. Approximately 44% of the firms in this stratum, 37 of 84, had negative issue date excess returns. A test of whether the number of negative returns was significantly different from 50% yielded an insignificant z

PAGE 99

92 of 1.09. This indicates that the probability of a negative issue date equity impact for the firms in this stratum is only 50%. The firms in the medium stratum experienced an average equity impact of -0.59%, significant at the 5% level. Over 65% of these firms, 56 of 86, experienced negative returns. The firms in the small stratum experienced an average issue date equity price impact of -1.40% which is significant at the 1% level. Almost 78% of these firms, 67 of 86, had negative issue date 2-day excess returns. The hypothesis that the probability of a negative return is greater than 50% is rejected for both strata at the 1% level. Testing between strata yielded significant results. The hypothesis that the issue date 2-day excess return for the stratum with the small CP ratios is equal to that of the medium stratum is rejected at the 10% level. That the small stratum equity the large stratum strong statistical was price impact is equal to that of the rejected at the 1% level. This is evidence of a relationship between CP ratio stratum equity middle and issue that set date low impact followed equity price impact. Firms in the CP ratios experience the greatest by the firms in the stratum of CP ratios. The firms that set their CP ratios relatively high had no significant impact on their equity value at the issue date.

PAGE 100

93 This test would indicate that the equity impact for firms selling convertible bonds is related to where man agement sets the CP ratio. There are two possible inter pretations of these results: (1) The equity valuation of the convertibles directly affects the issue date equity price impact, the CP ratio captures the equity valuation, and investors are not able to predict the issue's terms at the announcement date. (2) the equity valuation of the con vertible is unrelated to the issue date equity price im pact, and the CP ratio is actually masking the effects of another variable related to the firm or the issue. The goal of the remaining tests in this section is to determine whether there are variables other than the CP ratio which affect the issue date equity price impact. In each test of hypothesis the firms will be stratified accord ing to the variable in question, and the average issue date equity impacts will be compared across the strata. Rela tionships between these and other variables and the CP ratio will be tested in the last section, "Additional Tests." H 04 : Issue size has no affect on any equity price impacts at the issue date. Table 6-8 shows the results from stratifying the firms according to relative size of the issue of convertible bonds. The stratum of firms selling the relatively largest

PAGE 101

94 issues experienced an average 2-day equity price reaction of -0.57%. Although over 65% of these firms, 57 of 87, had negative issue date equity impacts, the average issue date impact for this stratum is insignificant at the 10% level. The results for the medium stratum are very similar. These firms had a nearly identical average 2-day equity impact of -0.59%. Even though a smaller percentage of these firms had negative issue date equity impacts, about 63%, the average impact apparently sue date is statistically significant at the 5% level, due to less variability among the individual isreactions. The stratum of firms with the relatively smallest issues had an insignificant average equity impact of -0.41%, and about 56% of these firms had negative issue date reactions. As expected, t-tests revealed no significant difference between the average issue date 2-day excess returns for any of the strata. From these results, there is no indication that the issue date equity price impact is related to the size of the issue of convertible bonds. H 05 : Firm size has no affect on any equity price impact at the issue date. The results of stratifying the firms by market value are presented in Table 6-9. The smallest firms in the sample -0.19%, experienced an insignificant average impact of even though almost 64% of these firms, 56 of 88,

PAGE 102

95 had negative issue date equity reactions. The medium-sized firms had an average issue date equity impact of -0.46%, again insignificant at the 10% level. About 58% of these firms, 51 of 88, had negative excess returns at the issue date. The largest firms in the sample had a highly signif icant issue date impact of -0.85%. Over 60% of these firms, 53 of 88, had negative issue date 2-day excess re turns. It would seem that the larger the issuing firm, the greater the issue date equity price impact. The smallest firms in the sample had the smallest issue date impact, fol lowed by the middle sized firms, and neither of the average impacts of these groups of firms was significant. The larg est firms in the sample experienced the largest and only significant issue date equity price impact. Although, T-Tests showed no statistical significance between any of the strata, the relationship between firm size and issue date impact deserves further attention due to the issue date reaction of large firms. Whether the probability of a negative issue date reac tion is 50% was tested next. The zvalue for the smallest firms was highly significant at 2.56, indicating that the null hypothesis can be rejected at the 1% level. Even though the probability of a negative response among the smallest firms is the greatest, 64%, the average impact for this group of firms is insignificant. Apparently, most

PAGE 103

96 small firms experience negative issue date equity price impacts, but the impacts are very small. The middle stratum's zwas 1.5, and the null hypothesis that the probability of a negative reaction is 50% cannot be rejected at the 10% level. The probability of a negative response for a medium-sized firm is more or less random. The stratum with the largest firms had a zvalue of 1.93. The null hypothesis that the probability of a negative equity response is 50% can only be rejected with 10% sig nificance, and these firms had the greatest issue date impact. Table 6-10, below, shows the average sizes of firms and issues in the three strata. It is clear from the figures that relative size does not remain constant across firms. Table 6-10: Average firm size, issue size, and relative size for the three strata by firm size. Stratum by Firm Size Small Medium Large Ave. Firm Size ($ Million) 74.2 244.3 1271.9 Ave. Issue Size ($ Million) 18.8 47.4 91.8 Ave. Relative Size 0.35 0.21 0.09 In relative terms, the issues sold by the smallest firms were almost four times the size of those sold by the

PAGE 104

97 largest firms. It seems obvious that, in the previous test, the stratum of firms selling the relatively smallest issues would contain many of the largest firms. However, since relative size has been demonstrated not to affect issue date equity impacts, it must be another factor deal ing with firm size that might cause the largest firms to experience the greatest issue date equity price impacts. Table 6-11 is a summary of the issue date results. It is clear that the date convertible bonds are actually sold is a statistically significant event. The average firm suf fered an issue date equity price impact of -0.61%, signifi cant at the 1% level. Over 61% of the firms in the issue date sample, 183 of 298, had negative issue date reactions. The null hypothesis that the probability of a negative issue date equity impact is 50% is rejected at the 1% level. Whether the -0.61% reaction represents an econom ically significant impact is unclear. An investor attempt ing to capitalize on this seeming market inefficiency, would have to short the stock (or some other similar stra tegy) of every convertible announcing firm. Then, since this would entail numerous transactions over a long period, broker fees might eliminate any potential profits. Also, as mentioned earlier, the issues in the study were found already outstanding. The sample, then, does not contain any issues that were announced and subsequently withdrawn. It is within reason that including issues that were

PAGE 105

98 withdrawn would change significantly the issue/cancellation date results. When the firms were stratified by CP ratio, the equity price impact was dramatically different across the strata. Firms that set the conversion price closest to the current stock price, the lowest CP ratios, experienced the greatest (negative) issue date equity price impact. The middle strata experienced a much less severe impact, and that of the highest strata was even slightly positive, albeit insig nificant. If we accept that the CP ratio captures the eq uity valuation of the convertible, we must conclude that the issue date equity price impact is a function of the equity component of the convertible bond issue, and invest ors are unable to estimate the issue's terms before the issue date. If the CP ratio is actually masking the effects of another variable, however, the conclusion is not so obvious. Therefore, the relationship of the CP ratio to other potentially important variables was tested further. Also, if the issue date equity impact is related to the CP ratio, its release at the date of sale must represent the release of economically important information. Testing for increased cross-sectional variance at the issue date, however, could explored further. not support The results this. This point was also of stratifying according to relative size showed no relationship between the size of the convertible issue and the equity price impact, but a

PAGE 106

99 possible relationship was discovered between size of the issuing firm and the impact on stock prices. The largest firms experienced the greatest average issue date equity impact. Since relative size was ruled out as an explan ation for issue date impacts, the fact that larger firms tend to sell relatively smaller convertible issues does not explain this phenomenon. Again, this will be tested fur ther. Summary of Results Announcement Date Tests It has been shown that the information contained in a convertible bond announcement is of economic importance to the market. The average convertible bond announcing firm is assessed an announcement date equity price impact of almost -1.5%. Also, the average cumulative abnormal return before the issue is announced is large and statistically significant. Miller and Rock (1985) hypothesize that sel ling any type of security signals a shortfall in operating funds, a negative signal to the market. Myers and Majluf (1984) argue that managers will sell a security if it is overpriced in the market. The negative announcement date impact supports both arguments, and the positive performance preceding the announcement adds further support for Myers and Majluf (1984).

PAGE 107

100 The arguments of Miller and Rock (1985) and Myers and Majluf (1984) are built on the assumption of asymmetric information between managers and investors, and any stock price impacts must be the result of the release of "inside" information. The tests in this study are designed to help determine the nature of the information contained in a con vertible bond announcement. For example, does the announce ment contain firm specific information to be valued differ ently according to the announcing firm and/or terms of the issue, or are firms simply assessed some more or less flat fee for selling a generic convertible bond? If the impact varies from firm to firm, what is the cause of this variance? First, the nature of the information was tested. If the announcement is to be classified as information, and if the value of the information varies across firms, the re sult would be an increase in the cross-sectional variance of stock returns in conjunction with the announcement. An F-test was performed and showed that there was no statistically significant increase in the cross-sectional variance. In essence, the distribution of excess returns on any non-event date is simply shifted to the left with the announcement of a convertible bond issue. This result would indicate that, rather than analyzing each announce ment to determine the impact on the equity of the announc ing firm, the market simply charges some flat fee for announcing convertible bonds.

PAGE 108

101 If each firm that announces a convertible bond issue is charged to any To test according value of by the a flat fee, the equity impact should be unrelated characteristics if this is to three of the firm or the issue of bonds. the case, the firms were stratified variables: (1) the estimated relative the equity component of the convertible, measured CP ratio, (2) the relative equity value of the issue, measured by the CP ratio, and (3) the size of the an nouncing firm. No relationship was detected between the announcement date equity impact and the size of the firm or the issue. component ment date However, there is an indication that the equity of the convertible may enter into the announce impact for the individual firm. The stratum of firms which set the conversion price closest to the current stock price experienced the greatest (most negative) aver age equity price impact of -1.84%. The stratum of firms that set the conversion price farthest from the current stock price experienced the least average impact, -1.04%. The middle stratum of firms experienced an average impact of -1.63%. All were significant at the 1% level, but none was shown to be significantly different from the others. So, even though there is no statistical basis upon which to draw a conclusion, there does seem to be a pattern relating CP ratio and equity price impact. Increasing the number of strata from the arbitrary choice of three might help to uncover a subtle relationship. It should not be

PAGE 109

102 ruled out at this point that investors attempt to estimate the conversion terms of the issue before it is sold. The results of the announcement date tests can be summed up as follows: (1) The announcement of a convertible bond issue is an economically significant event. (2) It appears that management might be trying to time the announcement of the issue to capture over pricing of the firm's stock. (3) There is no statistical evidence specific announcement date impact either firm or issue size. of a firm related to (4) There is evidence, although statistically insig nificant, that the announcement date equity impact might be affected by the expected terms of the con vertible bond issue. Issue Date Results Any additional equity impact felt by the firm at the issue date can have two causes: (1) information released at the issue date, and (2) resolution of uncertainty about whether the issue would be sold or withdrawn. The average issue date equity price impact was a sta tistically significant -0.61%, even though the accumulated excess returns preceding and following the issue date were never significant. To test the nature of the issue date equity impact, i.e. whether it represents information release or resolution of uncertainty about whether the issue would be sold, the cross-sectional

PAGE 110

103 variance of excess returns at the issue date was compared to the typical non-event variance. The zvalue of 1.1 did not support the hypothesis that the issue date impact represents the release of firm specific information. This would indicate that firms are simply charged an additional flat fee at the issue date, and this fee is simply a result of resolving the uncertainty about whether the issue would be sold. To further were compared value of the test this result, the issue date impacts to three variables: (1) the relative equity issue, using the CP ratio, (2) the relative size of the issue, and (3) the size of the issuing firm. No issue relationship and the issue was detected between the size of the date equity impact, but there might exist a relationship between the size of the selling firm firm and the impact. Using three strata, the difference between the impacts for the smallest and largest firms was not shown to be significant. However, since the difference in impacts was fairly large, -0.85% for the large firms and -0.19% for the small firms, this point deserves further investigation. The relationship between equity valuation and issue date impact was highly significant. Firms setting the lowest CP ratios experienced an average issue date impact of -1.4%, significant at the 1% level, while that for firms setting the highest CP ratios was an insignificant +0.19%.

PAGE 111

104 This indicates that: (1) the equity valuation of the convertible is important in determining issue date equity price impacts, and (2) investors are either incapable of estimating the issue's terms or make no attempt to do so prior to the issue date. Since this result seems to contradict the finding of no increased cross-sectional variance, it warrants further testing. In the following chapter, the firms will be strat ified by CP ratio, and characteristics of the firms and issues will be checked to rule out the possibility that the CP ratio is actually masking the effects of another var iable. The issue date results can be summarized as follows: (1) There is a uity price are sold. statistically significant average eq impact on the date convertible bonds (2) There is no evidence of above or below average per formance around the issue date by firms that an nounce and then sell convertible bonds. (3) There is no statistical evidence that issue date equity impacts are related to the size of the issue, but there is a possible relationship be tween firm size and the issue date impact. (4) Where management sets the conversion terms affects the issue date impact. Additional Tests The section is divided into two main parts: ( 1) a discussion of the results of testing for differing equity

PAGE 112

price impacts vertible bond 105 according to the number of outstanding con issues, and (2) an examination of the characteristics of the CP Ratio and its relationship to other firm and issue characteristics. Single vs. Multi The central question in this study is the impact on the equity value of a firm that announces and issues convert ible bonds. Another, related, question is the impacts on outstanding issues of convertible bonds with the announce ment by the same firm of a new issue. The preceding tests have shown that the terms of a convertible bond issue can have an economic impact on firm equity values. In turn, however, movements in stock prices can affect the conversion terms of outstanding issues. For example, consider the CP ratio used above to measure the relative equity value of the convertible issues. The drop in stock price at both the announcement and issue dates for the firm that sets a low CP ratio can affect the CP ratio of outstanding issues and potentially their relative equity values. Any change in the equity value of outstanding issues has the potential of affecting the firm's equity val ue. The potential for this confounding information raises the question of whether firms that have previously issued convertibles outstanding at the announcement of a new issue

PAGE 113

experience different price impacts than ibles. 106 announcement and/or issue date equity firms with no other issues of convertThe firms were divided into two samples according to whether they had outstanding convertible bonds at the date the issue in the sample was announced. If the issue in the sample is the firm's only issue, the firm is classified as a "single." A firm with other issues outstanding is class ified as a "multi." Announcement date tests There are 171 firms classified as single and 61 class ified as multi in the announcement date tests that follow. In each test the firms are divided into three strata before testing, and firms each. the strata for multi firms are as small 18 Any results when the firms are stratified should, sample therefore, be size has the considered preliminary as small potential to affect the results for 6-12a shows the announcement date multi firms. Table results when the firms are divided into single and multi. Both announcement date 2-day excess returns are highly significant with t-values in excess of 4.0. The announce ment date reaction for single firms was -1.37% with 71%, 121 of 171, of the 2-day excess returns negative. The reaction for multi firms was -1.59%. Seventy-eight per cent, 48 of 61, of the multi firms experienced negative

PAGE 114

107 reactions. The single or multi equity impacts level, however. null will cannot hypothesis that classification of not affect affect announcement date be rejected (t = 0.5) at the 10% At the announcement of a convertible bond issue, whether the firm has other issues of convertibles outstanding apparently does not affect the equity reaction. Tables 6-12b and 6-12c show average daily and cumula tive average excess returns for single and multi firms, 121 days around the announcement of a respectively, for convertible bond issue. Both single and multi firms experience significant negative excess returns on the announcement date and the preceding day, and, as stated above, these reactions are found to be approximately equal. Single firms, however, enjoy rather substantial cumulative abnormal returns prior to the announcement. Up to day -5 single firms average +4.65% which is significant at the 1% level. Multi firms experience only marginally significant abnormal returns prior to the announcement, and by day -5 they are insignificant. Even at the announcement date and the day before, the cumulative abnormal return for single firms is significant. By day -1 this cumulative return has fallen to 3.2%, significant return for significant hand, do at the single at the not 5% level. At day zero the cumulative firms is down to 2.5%, but it is only 10% level. Multi firms, on the other experience significant positive

PAGE 115

108 cumulative excess returns after day -6. For days -30 to -6 multi firms maintain a cumulative abnormal return around 3.4%, but it is never significant above the 10 % level. Both single and multi firms experience no significant cumulative date. abnormal returns following the announcement Although the announcement date equity reactions for single and multi firms are not found to be significantly different, the performance in equity returns preceding the announcement date calls for further testing. There is an obvious difference in pre-announcement returns that might indicate different motives for selling convertible bonds. The positive performance by single firms follows the Myers and Majluf (1984) reasoning that management will sell securities when they are overpriced. Witness that the positive performance disappears following the announce ment. Multi firms do not experience as significant a positive performance preceding the announcement, but still may be attempting to capture this marginal performance. The remaining tests will repeat the tests performed earlier to determine if issue or firm size or equity valuation affects announcement date reactions for single vs. multi firms. Table 6-12d shows the results for single firms divided into three divided by strata by relative size, defined as issue size market value of the firm's equity. The results

PAGE 116

109 are fairly identical. The large, medium, and small strata experience average announcement date 2-day excess returns of -1.33%, -1.25% and -1.26%, respectively, with 73%, 38 of 52, 66%, 35 of 53, and 73%, 39 of 53, of the 2-day excess returns negative. Each is significant at the 1% level, but none of the results is found to be significantly different from the others at the 10% level. Table 6-12e is the same stratification for multi firms. In this case the firms issuing the relatively largest issues experience an average announcement date 2-day excess return of over -3.5%, and it is significant at the 1% level. The other strata experience much smaller and insignificant 2-day excess returns. Eighty-nine percent, 16 of 18, of the firms with the largest relative size experienced negative of the middle reactions. Sixty-seven percent, 12 of 18, stratum had negative reactions, as did 78%, 14 of 18, of the firms with the smallest relative size. Preliminary results would seem to indicate that rel ative issue size only matters for multi firms. The T-test comparing the stratum with the relatively largest issues with the middle stratum yielded a T-value of -3.22. That for the test comparing the largest stratum with the small est stratum was -2.68. Both are significant at the 1% level, but as test are very mentioned earlier the sample sizes in this small. Care must be taken in interpreting these results as they might be driven by sample size.

PAGE 117

110 The next test is to determine the affects of firm size on the announcement date impact for single vs. multi firms. Table 6-12f shows the results from dividing single firms into three strata by firm equity value. The 2-day average excess return for the smallest firms -1.53%, and it is significant at the 1% level. seventy-four percent, 39 of 53, of the smallest firms experienced negative announce ment date reactions. The middle stratum of firms experi enced an average 2-day excess return -0.85%. This was sig nificant at the 5% level, and 67%, 35 of 52, experienced negative 2-day returns. The average impact for the stratum of largest firms was -1.45% and was significant at the 1% level. Seventy-two percent, 38 of 53, of these firms experienced a negative announcement date impact. None of the strata was found to experience an impact significantly different from the other two strata at the 10% level. This indicates that firm size is not a factor in equity price impacts for firms announcing their only issue of convertible bonds. Table 6-12g gives the results for the multi firms divid ed into three strata by firm size. The smallest firms ex perienced an average 2-day excess return of -1.45%. this was only significant at the 10% level and 72% (13 of 18) of the firms experienced negative excess returns. The stratum with the middle third of the firms experienced a -1.84% 2-day excess return, significant at the 1% level. Seventy

PAGE 118

111 eight percent of these firms (14 negative announcement date impact. of 18) experienced a The largest firms experienced a 2-day -1.56%, significant of the 18 announcement date abnormal return of at the firms of 1% level, and eighty-three this stratum experienced a percent negative reactions reaction to the announcement. was significantly different Again, none of the from the others at the 10% level. These tests have demonstrated that firm size is not a (statistically significant) factor in the announcement date impact on convertible bond issuing firms. However, there appears to be what might be interpreted as a pattern be tween firm size and announcement date equity price impacts. For multi firms, the largest firms experienced the greatest impacts. The middle-sized firms experienced the next greatest impacts, and those for the smallest firms were statistically insignificant. Once again, however, the interpretation of any patterns must be viewed as prelimin ary due to small sample sizes. The next test looks at the effects of equity valuation on the impacts on single and multi firms. Table 6-12h shows the results for single firms divided into three strata by equity valuation of the issue of convertible bonds. It should be noted that those issues with the lowest CP ratios would be considered to have the

PAGE 119

112 greatest equity component. The three strata from lowest to highest CP ratio experienced average announcement 2-day excess returns of -1.72%, -1.49%, and -1.04%, all signif icant at at least the 5% level. Sixty-seven percent (35 of 52) of the firms in the stratum with the highest CP ratios experienced middle and 52) and 75% negative 2-day excess returns. Figures for the smallest strata, respectively, were 71% (37 of (39 of 52). None of the strata experienced a 2-day excess return which was significantly different from the others at a level of 10%. There is no statistical rela tionship between perceived equity valuation and announce ment date equity price impacts. However, the reactions are such that there would appear to be an attempt by investors to estimate the terms of the issue before they are releas ed. The results for single firms, although not significantly different from each other, follow a pattern suggesting a relationship between relative equity value and announcement date equity price impacts. Table 6-12i contains the results for the multi firms. The stratum of issues with the lowest CP ratios experienced an average 2-day excess return of -2.24% which was signif icant at the 1% level. Eighty-nine percent (16 of 18) were negative. The middle stratum had an average 2-day excess return of -1.47%, significant at the 5% level, and that for the remaining stratum was -1.44%, significant at the 10% level. These strata experienced a 74% (14 of 19) and 74%

PAGE 120

113 (14 of 19) negative 2-day excess returns, respectively. Again, none of the strata experienced an average 2-day ex cess return determined to be significantly different from the others, indicating that the equity valuation of the convertible bond is not a factor in the announcement date impact for firms with previously issued convertible bonds outstanding. Once again, however, firms which set the lowest CP ratios experienced the greatest announcement date equity price impact. Table 6-12j is a summary of the results of announcement date tests performed for single and multi firms: (1) There is no statistical relationship between classification as single or multi and announcement date equity price impacts. (2) Although no statistical relationship was detected which related firm and issue size to announcement date equity price impacts for either single or multi firms, in each case either the largest firms or the largest issues caused the greatest equity impacts. This is a point that deserves further testing. (2) Again, no statistical relationship was detected which related equity valuation with announcement date equity price impacts. However, once again, a pattern emerges for both single and multi firms. In each case, the issues with the lowest CP ratios cause the greatest impacts. It appears that investors attempt to estimate issue terms before they are released. Issue date tests There are 298 firms in the issue date tests, 225 classified as single and 73 classified as multi. Table

PAGE 121

114 6-13a shows average 2-day excess returns for single and multi firms for the 121 days surrounding the sale of a con vertible bond issue. There is nothing significant to re port either before or after the sale, but sale date average excess returns are significantly different for the two strata. Multi firms experienced an average excess return of -1.28% significant at the 1% level. Over 68%, 50 of 73, of the multi firms experienced negative issue date returns. The average impact for single firms was only -0.39% significant at the 10% level, and about 60%, 133 of 225, had negative issue date returns. AT-test (t = 2.07) revealed that the impacts for single and multi firms are different at the 5% level. Tables 6-13b and 6-13c excess returns for the 121 which show average cumulative days surrounding the sale of convertible bonds for single and multi firms, respectively, indicate nothing remarkable that would help explain this result. There are several potentially contradicting hypotheses which might explain why issue date equity impacts are dif ferent for single and multi firms, but none will be explored at this time. The causes for issue date equity price impacts differing according to whether the issuing firm has other outstanding convertibles is a topic for future research.

PAGE 122

115 Tables 6-13d and 6-13e show the results from strat ifying single and multi firms, respectively, by issue size. Again, issue size is measured as relative size, the ratio of issue size to firm equity value. There is nothing remarkable to report for either single or multi firms. The large, medium, and small single strata experienced average 2-day issue date excess returns of -0.34%, -0.34%, and -0.25%, respectively. Sixty-three percent, 41 of 65, of the single firms selling the relatively largest issues had negative issue date and small strata are respectively. None returns. The figures for the medium 60%, 39 of 65, and 56%, 35 of 63, of the strata was found statistically different from the others. The large, medium, and small multi strata experienced average issue date 2-day excess returns of -1.28%, -0.96%, and -1.19%. Seventy-four percent, 17 of 23, of the multi firms selling the relatively largest issues experienced negative issue date returns. Sixty-five percent, 15 of 23, of the medium stratum had negative issue date returns. Sixty-four percent, 14 of 22, of the small stratum issue date excess returns were negative. None of the strata was found significantly different from the others. Table 6-13f shows the issue date results for single firms divided into three approximately equal strata by firm size. The large, medium, and small single firms experi enced average issue date excess returns of -0.52%, -0.51%,

PAGE 123

116 and +0.15%, respectively, with 62%, 58%, and 57% of the firms experiencing negative issue date returns. None of the average issue date returns was statistically different from zero or significantly different from the others. Table 6-13g shows the same stratification for multi firms. The large and medium strata had average issue date 2-day excess returns of -1.88% and -1.98%, respectively, with 61% of the largest firms and 70% of the medium firms experiencing negative issue date returns. The stratum with the smallest multi firms had an insignificant average issue date excess return of -1.3%, and 70% of these firms had negative issue date issue date returns from the others. returns. None of the average 2-day is shown to be significantly different The results for firm size lend weak firm size and issue however, it appears single and multi firms stratified by support for the relationship between date equity price impacts. Again, that the largest firms experience the greatest issue date equity impacts. When the firms are stratified according to CP ratio, the results for both the single and the multi firms show statistical significance. Table 6-13h gives the results for the single firms stratified by CP ratio. As in the previous issue date ratios experience the For single firms in tests, firms that set the lowest CP greatest issue date equity impacts. the low stratum, the average 2-day

PAGE 124

117 issue date excess return is -1.03, and it is significant at the 1% level. Seventy-one percent of these firms, 44 of 62, had negative issue date returns. The medium stratum averaged an insignificant -0.44% with 63% of the observa tions negative. The high stratum average 2-day excess return is +0.30%. The average excess return is insignif icant and only 47% of these firms had negative issue date returns. When the stratum of low CP ratios is compared to the stratum of high CP ratios, the T-value is -2.46 indicating a significant (5% level) difference between reactions. Comparing the low stratum with the medium stratum yielded an insignificant T-value of -0.98. The indication for the multi firms is even more pro nounced. Multi firms in the stratum with low CP ratios had a highly significant average 2-day issue date excess return of -2.43% with over 91% of the firms experiencing negative issue date returns. The medium stratum had an average re turn of -1.16%, this time significant at the 5% level. About 70%, 16 of 23, experienced negative issue date re turns. The stratum of high CP ratios experienced an average issue date return of o.oo, and less than 45% of these firms had negative issue date returns. For both the single and multi firms, then, the percentage of negative responses is not shown to be significantly different from 50% for firms that set high CP ratios.

PAGE 125

118 T-tests comparing the low CP stratum with the others revealed that CP ratio is a factor in issue date returns for multi firms. The equity impact for the low stratum is different from the high stratum at the 1% level (t = 2.96) and different from the medium stratum at the 5% level (t = 1. 96). Table 6-13j gives a summary of the issue date test results for single and multi firms: (1) Issue date equity impacts are statistically different for single and multi firms. (2) Although not shown statistically, there appears to be a relationship between issue size and issue date equity impacts. The relatively largest issues elicit the greatest equity impacts for both single and multi firms. (3) Although not shown statistically, there appears to be a relationship between firm size and issue date equity impacts. (4) There is strong statistical evidence that issue date equity impacts vary inversely with CP ratios for both single and multi firms. As mentioned earlier, no explanation is offered at this time for differing impacts across single and multi firms. This matter is left for future study. CP ratios The CP ratio has been shown in the preceding tests to be inversely related to issue date equity price impacts. For all firms taken together as well as for single and

PAGE 126

119 multi firms as subsets, it appears that the lower the CP ratio, the ratio of conversion price to current stock, the greater (more negative) the equity price impact. Although insignificant, the relationship appears to be present at the announcement date, also. The degree to which convertible bond returns are cor related with the equity returns for the issuing firm should be determined by the relative equity valuation of the con vertible. That is, if the CP ratio captures equity valua tion, the returns for issues with low CP ratios should be correlated to a greater degree with the issuing firm's com mon stock returns than issues with high CP ratios. Returns for issues with high CP ratios, seen as mostly debt, should be correlated to a greater degree with interest rate changes. When available, the returns over the first 30 trading days following the issue date were gathered for the con vertible, the issuing firm's common stock, and long term U. S. Treasury bonds. The issues were divided into three approximately equal strata by CP ratio. If the CP ratio captures equity valuation as assumed, the strata with the lowest CP ratios should show the highest correlation with common stock returns and the lowest correlation with treasury issues. The following regression was run first for 130 convertible bonds for which data were available: 1

PAGE 127

where CBOND•= l. STK = l. TBOND= 120 the 30-day post issue date return for the convertible bond issued by firm i. the 30-day post issue date return for the common stock issued by firm i. the 30-day post issue date return for long term government bonds. The regression was then run for the 130 firms divided into three approximately equal strata by CP ratio. Table 6-14, below, shows the results. (t-values are in paren theses below coefficient estimates.) The results for all Table 6-14 Results from regressing convertible bond returns against common stock and government bond returns. Stratum Unstratified 0.04 0.49 0.30 62 (8.21) (13.91) (2.25) High CP 0.03 0.49 0.32 .65 (4.09) (8.53) (1. 77) Mid CP 0.04 0.49 0.19 .59 (5.06) (6.97) (0.67) Low CP 0.04 0.51 0.38 .62 (4.56) (7.75) (1.37)

PAGE 128

121 four regressions are almost identical. The returns of con vertible bonds seem to be correlated to about the same degree with the issuing firm's common stock returns regard less of CP ratios. b 1 across the three strata is approx imately equal, and only the strata with high CP ratios has a b 2 that is even marginally significant. The R 2 for the regressions are about equal and reasonably high indi cating that convertible bond returns are explained fairly well by the returns on the firm's common stock and by move ments in interest rates. The regressions do not seem to support the assumption that the CP ratio captures equity value. However, the relationship between CP ratio and equity price impacts has been shown clearly in the other tests. Therefore, if the CP ratio is not capturing equity valuation, it must be masking the affects of some other variable(s). As a means for testing whether the CP ratio is actually masking the affects of one or more other variables, the issue date firms are once again divided into three strata by CP ratio. found: For each of the strata the following were (1) Average CP ratio (2) Average firm size (3) Average issue Size (4) Average relative issue size (5) Average conversion ratio (6) Average number of days between announcement and issue dates (7) Frequency of issue per year (8) Frequency of issue quality ratings

PAGE 129

122 The results of the stratification are contained in Table 6-15a. The average CP ratios for the high, mid, and low CP strata are 1.26, 1.15, and 1.09, respectively, and they are significantly different at the 1% level. The average firm sizes for the high, mid, and low CP strata are 541.74, 609.43, and 505.61 million dollars, respectively. None of the strata average firm sizes was significantly different from the others, so firm size is ruled out as a possible explanation for the relationship between CP ratio and equity price impacts. A significant difference was found between average issue sizes and relative issue sizes across the strata. The average issue are 59.43, 50.72, the high CP strata average issue size sizes for the high, mid, and low strata and 45.79. The average issue size for is significantly different from the for the low CP strata at the 5% level. Average relative issue sizes for the high, mid, and low strata are 0.17, 0.20, and 0.21, respectively. The average relative size for the high strata is significantly dif ferent from that of the low strata at the 10% level. Aver age issue size and relative issue size appear to be invers ely related. The largest issues are in the high CP strata, but the greatest relative issue sizes are in the low CP strata. Since firm size is not significantly different for the strata, this is a curious result. However, since the low CP strata experienced the greatest equity price impact,

PAGE 130

123 this would seem to indicate that the small issues and those with the greatest relative size cause the greatest issue date equity price impacts. Relative size, however, has been previously ruled out as a factor in issue date equity price impacts. If issue size, as contrasted with relative size, is a factor in issue date equity impacts, this would seem to indicate that price responses are simply a function of transactions costs. That is, small issues cost more to have underwritten since there are economies of scale in issuing securities. This is only speculation, of course, but if relative issue size is not a factor and issue size is, this would tend to remove firm characteristics as possible causes of equity impacts. Conversion ratio, the number of shares into which an individual bond is convertible, has been shown not to be related to equity impacts. The average conversion ratios for the high, mid, and low CP strata are 28.25, 26.8, and 30.21. None is significantly different from the others. When the average length of time between announcement date and issue date is found for the three strata, strong significance appears. The average number of calendar days from announcement to issue, shown as AD-> ID, for the high, mid, and low CP strata are 20.97, 28.03, and 32.27. AD -> ID for the high strata is significantly different from that of the mid strata at the 5% level and different from the low strata at the 1% level.

PAGE 131

124 The issue date equity price reaction seems to be directly related to how long the firm waits to sell an an nounced issue. The greatest issue date equity price impact was experienced by the firms which set the lowest CP ratios, these are also the firms that waited the longest to sell their announced issues. AD-> ID cannot be ruled out as a possible explanation for the relationship between CP ratio and equity price impacts. The frequency of issue per year was also found. If there is any pattern to be discerned from the data, poss ibly issues with low CP ratios tended to be sold toward the beginning of the period over which the sample was obtained. Newer issues seem to have higher CP ratios, but no glaring patterns emerge. Also, since the frequency of issuance is not compared to any macroeconomic variables such as GNP or the CPI, etc., care must be taken in arriving at conclusions. Clear from Table 6-15a is that very few issues were sold in the 1975-1978 period. In fact, no issues in this sample were sold during 1973 or 1977. different economic factors were at divided into two subsamples, 1963 To determine whether work, the sample was to 1972 and 1974 to 1986. Tests showed that there were no differences in equity impacts between the page 57 contains a summary these two samples. two samples. Footnote six on of the results from testing

PAGE 132

125 In conclusion, regression estimates indicate that the use of CP ratios to estimate convertible bond equity values might not be appropriate. Since a definite relationship exists between CP ratios and the equity price impacts, it is possible that the CP ratio is actually masking the affects of some other variable(s). To test this possibility, seven characteristics of the firms and the issues were examined. Firm size, relative size, conversion ratio, issues per year, and quality rating were ruled out. Only issue size and the length of time between announcement and issuance remain as potential explanations for the relationship between CP ratios and equity price reactions. These conclusions are only preliminary at this time, however. Further empirical research is warranted. There is some variable or group of variables that determine the impact on the equity value of firms that sell convertible bonds. These tests have shown that such variables exist. It must be the goal of future research to determine what those variables are. Notes 1. The correlation of the two independent variables, return on the underlying stock and return on longterm government bonds, is only +0.09. Thus, results of the regression are only minimally affected by multi collinearity.

PAGE 133

*** ** * 126 Table 6-la Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Full Sample n = 232 Obs. 2AR t 1 0.23 % 0.92 5 0.15 0.60 10 0.37 1.48 15 -0.29 -1.16 16 0.18 0.72 17 0.18 0.72 18 0.14 0.56 19 -0.21 -0.84 20 0.07 0.28 21 0.42 1.68 * 22 -0.08 -0.32 23 0.02 0.08 24 0.03 0.12 25 -0.03 -0.12 26 0.31 1.24 27 0.19 0.76 28 0.11 0.44 29 -0.37 -1.48 30 a -1.43 -5.72 *** 31 -0.42 -1.68 * 32 0.13 0.52 33 0.13 0.52 34 0.04 0.16 35 -0.15 -0.60 36 -0.41 -1.64 * 37 -0.19 -0.76 38 -0.16 -0.64 39 o.os 0.20 40 -0.18 -0.72 41 -0.12 -0.48 42 0.66 2.64 *** 43 0.17 0.68 44 0.10 0.40 45 -0.06 -0.24 50 0.16 0.64 55 -0.02 -0.08 60 -0.31 -1.24 Significant at the 1% level Significant at the 5% level Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 134

127 Table 6-lb Daily Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) For 121 Days Around the Announcement of Convertible Bond Issues: Full Sample n = 232 Trading Day AR t CAR t -60 0.06 % 0.38 0.06 % 0.38 -50 0.05 0.35 1.15 2.30 *** -40 0.11 0.72 2.13 3.07 *** -30 -0.20 -1.30 2.91 3.46 *** -20 -0.06 -0.37 3.27 3.38 *** -15 0.15 0.98 3.75 3.66 *** -14 -0.13 -0.83 3.63 3.50 *** -13 -0.08 -0.52 3.55 3.39 *** -12 0.11 0.71 3.66 3.45 *** -11 0.03 0.18 3.68 3.45 *** -10 -0.06 -0.43 3.62 3.35 *** -9 0.07 0.49 3.69 3.39 *** -8 0.24 1.60 3.94 3.58 *** -7 0.17 1.10 4.10 3.69 *** -6 0.02 0.15 4.12 3.68 *** -5 0.13 0.86 4.25 3.76 *** -4 -0.02 -0.10 4.24 3.71 *** -3 -0.18 -1.20 4.06 3.52 *** -2 -0.19 -1.26 3.87 3.33 *** -1 -0.70 -4.61 *** 3.17 2.71 *** 0 a -0.73 -4.81 *** 2.44 2.07 ** 1 -0.21 -1.40 2.23 1.87 * 2 -0.21 -1.40 2.02 1.68 * 3 0.11 0.74 2.13 1.76 * 4 0.02 0.11 2.15 1.76 * 5 0.06 0.40 2.21 1.80 * 6 0.07 0.46 2.28 1.84 * 7 -0.06 -0.41 2.21 1.78 * 8 0.10 0.68 2.32 1.85 * 9 -0.08 -0.51 2.24 1.77 * 10 -0.07 -0.45 2.17 1.71 * 11 -0.34 -2.24 ** 1.83 1.43 12 -0.07 -0.48 1. 76 1. 36 13 0.11 0.75 1.87 1.44 14 -0.30 -1.96 ** 1.58 1.21 15 -0.21 -1.39 1.37 1.04 20 -0.04 -0.29 1.29 0.95 30 -0.14 -0.93 2.02 1.40 40 0.16 1.07 1.25 0.82 50 -0.09 -0.58 0.60 0.38 60 -0.18 -1.19 -0.30 -0.18 *** Significant at 1% Level ** Significant at 5% Level * Significant at 10% Level a The announcement date as identified in the WSJ

PAGE 135

128 Table 6-2 Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by CP Ratio High n = 71 Middle n = 71 Low n = 70 Obs. 2AR t 2AR t 2AR t 1 0.62 % 1.82 * -0.25 % -0.76 0.29 % 0.62 5 0.52 1.53 0.05 0.15 0.09 0.19 10 0.29 0.85 -0.02 -0.06 0.47 1.00 15 -0.30 -0.88 -0.30 -0.91 -0.37 -0.78 16 0.10 0.29 -0.13 -0.39 0.32 0.68 17 0.33 0.97 0.33 1.00 0.18 0.38 18 0.18 0.53 -0.14 -0.42 0.09 0.19 19 -0.61 -1.79 * 0.32 0.97 -0.36 -0.77 20 -0.08 -0.24 0.20 0.61 0.05 0.11 21 0.12 0.35 0.54 1.64 0.26 0.55 22 -0.12 -0.35 -0.19 -0.58 -0.19 -0.40 23 0.30 0.88 0.08 0.24 -0.39 -0.83 24 0.65 1.91 * -0.31 -0.94 0.05 0.11 25 0.01 0.03 -0.03 -0.09 0.11 0.23 26 0.09 0.26 0.50 1.55 0.17 0.36 27 -0.07 -0.21 0.24 0.73 0.39 0.83 28 0.07 0.21 -0.07 -0.21 0.56 1.19 29 -0.48 -1.41 -0.16 -0.48 -0.17 -0.36 30 a -1.04 -3.06 *** -1.63 -4.94 *** -1.84 -3.91 *** 31 0.08 0.24 -0.33 0.10 -0.70 -1.51 32 0.49 1.44 0.16 0.48 -0.23 -0.49 33 0.41 1.21 -0.27 -0.82 0.35 0.74 34 0.49 1.44 0.08 0.24 -0.54 -1.15 35 0.33 0.97 0.06 0.18 -1.06 -2.25 ** 36 -0.24 -0.71 -0.40 -1.21 -0.51 -1.08 37 0.10 0.29 -0.56 -1.70 * -0.19 -0.40 38 0.17 a.so -0.15 -0.45 -0.25 -0.53 39 0.29 0.85 0.30 0.91 -0.61 -1.30 40 -0.35 -1.02 0.40 1.21 -0.34 -0.72 41 -0.08 -0.23 -0.17 -0.51 -0.14 -0.30 42 0.20 0.59 0.73 2.21 ** 0.71 1.51 43 0.65 1.91 * 0.05 0.15 -0.01 -0.02 44 0.17 0.50 0.02 0.06 0.15 0.32 45 -0.35 -1.03 -0.08 -0.24 -0.09 -0.19 50 0.20 0.59 -0.13 -0.39 0.22 0.47 55 0.07 0.21 -0.36 -1.09 0.52 1.11 60 0.55 -1.62 0.04 0.12 -0.42 -0.89 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 136

129 Table 6-3 Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by Relative Size Large n = 70 Medium n = 71 Small n = 71 Obs. 2AR t 2AR t 2AR t 1 0.69 % 1.64 -0.09 % -0.26 -0.14 % -0.38 5 -0.22 -0.52 0.28 0.82 0.58 1.57 10 0.40 0.95 0.16 0.47 0.06 0.16 15 -0.31 -0.74 -0.15 -0.44 -0.03 -0.08 16 0.10 0.24 0.55 1.62 -0.23 -0.62 17 -0.44 -1.05 0.47 1.38 0.33 0.89 18 0.26 0.62 0.16 0.47 -0.12 -0.32 19 -0.05 -0.12 -0.40 -1.18 0.15 0.41 20 -0.02 -0.05 0.04 0.12 -0.01 -0.03 21 0.35 0.83 0.20 0.59 0.68 1.84 22 -0.18 -0.43 0.20 0.59 -0.80 -2.16 ** 23 -0.17 -0.40 -0.04 -0.12 0.07 0.19 24 -0.50 -1.19 o.oo 0.00 0.54 1.46 25 -0.35 -0.83 0.32 0.94 -0.13 -0.35 26 0.01 0.02 0.28 0.82 0.39 1.05 27 0.23 0.55 0.27 0.79 -0.25 -0.68 28 0.49 1.17 0.12 0.35 -0.10 -0.27 29 -0.61 -1.45 -0.65 -1.91 * 0.23 0.62 30 a -1.75 -4.17 *** -1.21 -3.56 *** -1.12 -3.03 *** 31 -0.30 -0.71 -0.70 -2.42 ** -0.09 -0.24 32 -0.15 -0.36 o.oo 0.00 0.48 1.30 33 0.11 0.26 -0.18 -0.53 0.23 0.62 34 0.28 0.67 -0.24 -0.71 0.12 0.32 35 0.06 0.14 -0.28 -0.82 -0.10 -0.27 36 0.05 0.12 -0.43 -1.26 -0.60 -1.62 37 -0.43 -1.02 -0.35 -1.03 -0.06 -0.16 38 -0.26 -0.62 0.01 0.03 -0.18 -0.49 39 0.28 0.67 0.29 0.85 -0.02 -0.05 40 -1.01 -2.40 ** 0.30 0.88 0.40 1.08 41 -0.22 -0.52 0.08 0.24 -0.40 -1.08 42 1.11 2.64 *** 0.13 0.38 0.45 1.22 43 -0.07 -0.17 0.03 0.09 0.68 1.84 44 -0.64 -1. 52 0.24 0.71 0.56 1.51 45 0.55 1.31 -0.20 -0.59 -0.25 -0.68 50 0.21 0.50 0.02 0.06 0.16 0.43 55 -0.07 -0.17 0.13 0.38 0.32 0.86 60 0.07 0.17 -0.56 -1.65 * -0.61 -1.65 * *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 137

130 Table 6-4 Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by Firm Size Large n = 71 Medium n = 71 Small n = 70 Obs. 2AR t 2AR t 2AR t 1 0.25 % 0.81 -0.14 % -0.47 0.36 % 0.75 5 0.33 1.06 0.46 1.53 -0.14 -0.29 10 0.26 0.84 0.21 0.70 0.16 0.33 15 -0.09 -0.29 -0.57 -1.90 * 0.16 0.33 16 0.12 0.39 0.49 1.63 -0.16 -0.33 17 0.37 1.19 0.67 1.23 -0.69 -1.43 18 0.27 0.87 -0.10 -0.33 0.13 0.27 19 -0.24 -0.77 -0.14 -0.47 0.06 0.13 20 -0.15 -0.48 -0.24 -0.80 0.39 0.81 21 0.17 0.55 0.22 0.73 0.82 1.71 * 22 -0.21 -0.68 -0.48 -1.60 -0.10 -0.21 23 0.28 0.90 -0.02 -0.07 -0.39 -0.81 24 0.42 1.35 -0.20 -0.67 -0.18 -0.38 25 -0.05 -0.16 -0.01 -0.03 -0.11 -0.23 26 0.60 1.94 * 0.18 0.60 -0.11 -0.23 27 -0.29 -0.94 0.52 1.73 * 0.02 0.04 28 0.03 0.10 -0.13 -0.43 0.62 1.29 29 -0.02 -0.06 -0.10 -0.33 -0.92 -1.92 * 30 a -1.48 -4.77 *** -1.06 -3.53 *** -1.55 -3.23 *** 31 -0.32 -1.03 -0.52 -1.73 * -0.29 -0.60 32 0.34 1.10 -0.04 -0.13 0.03 0.06 33 0.22 0.71 0.26 0.87 -0.31 -0.65 34 -0.19 -0.61 0.21 0.70 0.14 0.29 35 -0.01 -0.03 -0.20 -0.67 -0.11 -0.23 36 -0.54 -1. 74 * -0.20 -0.67 -0.25 -0.53 37 -0.09 -0.29 -0.65 -2.17 ** -0.11 -0.23 38 -0.45 -1.45 0.01 0.03 -0.01 0.03 39 0.36 1.16 -0.29 -0.97 0.47 0.98 40 -0.18 -0.58 0.10 0.33 -0.22 -0.46 41 -0.40 -1.29 -0.09 -0.30 -0.06 -0.13 42 0.27 0.87 -0.27 -0.90 1.69 3.52 *** 43 0.44 1.42 0.21 0.70 -0.01 -0.02 44 0.10 0.32 0.24 0.80 -0.18 -0.38 45 -0.10 -0.32 -0.25 -0.83 0.45 0.94 50 0.06 0.19 -0.44 -1.47 0.77 1.60 55 0.23 0.74 0.30 1.00 -0.13 -0.27 60 -0.36 -1.16 -0.35 -1.17 -0.39 -0.81 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 138

131 Table 6-5 summary Results for Announcement Date Tests Test N 2AR z No. Neg z Issue Date 232 -1.43% -5.72 *** 168 6.83 CP Ratio Low 70 -1.84 -3.91 *** 55 4.78 Middle 71 -1.63 -4.94 *** 52 3.92 High 71 -1.04 -3.06 *** 49 3.20 Relative Size Small 71 -1.12 -3.03 *** 52 3.92 Medium 71 -1.21 -3.56 *** 48 2.97 Large 70 -1.75 -4.17 *** 53 4.30 Firm Size Small 70 -1.55 -3.23 *** 51 3.82 Medium 71 -1.06 -3.53 *** 49 3.20 Large 71 -1.48 -4.77 *** 54 4.39 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level *** *** *** *** *** *** *** *** *** ***

PAGE 139

132 Table 6-6a Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bonds: Full Sample n = 298 Obs. 2AR 1 0.03 % 5 0.04 10 -0.44 15 -0.10 16 0.18 17 -0.35 18 0.11 19 -0.02 20 -0.10 21 0.23 22 0.05 23 0.18 24 -0.11 25 -0.28 26 -0.25 27 -0.04 28 o.oo 29 -0.22 30 a -0.61 31 -0.27 32 -0.04 33 0.09 34 0.05 35 -0.08 36 0.25 37 -0.13 38 0.06 39 -0.25 40 -0.20 41 -0.19 42 -0.26 43 -0.01 44 0.04 45 0.19 50 0.06 55 -0.09 60 -0.18 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level t 0.16 0.21 -2.32 -0.52 0.95 -1.84 0.58 -0.11 -0.53 1. 21 0.26 0.95 -0.58 -1.47 -1.32 -0.21 o.oo -1.16 -3.21 -1.42 -0.21 0.47 0. 26 -0.42 1.32 -0.68 0.32 -1.32 -1.05 -1.00 -1.37 -0.05 0.21 1.00 0.32 -0.47 -0.95 ** * *** a Two-day excess return consisting of the sale date and the day before.

PAGE 140

133 Table 6-6b Daily Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) For 121 Days Around the Sale of Convertible Bonds: Full Sample n = 298 Trading Day AR t CAR t --60 0.06 0.48 0.06 0.48 -50 0.14 1.13 1.06 2.58 -40 -0.26 -2.10 0.87 1.53 -30 0.03 0.24 1.19 1.73 -20 -0.03 -0.24 1.01 1.27 -15 -0.04 -0.32 1.25 1.49 -14 0.21 1.70 1.46 1.72 -13 0.04 0.32 1.50 1.75 -12 -0.15 -1.21 1.35 1.56 -11 -0.02 -0.16 1.33 1.52 -10 -0.26 -2.10 1.07 1.21 -9 -0.12 -0.97 0.95 1.06 -8 -0.13 -1.05 0.82 0.91 -7 -0.04 -0.32 0.78 0.86 -6 o.oo 0.00 0.78 0.85 -5 -0.05 -0.40 0.73 0.79 -4 0.05 0.40 0.78 0.83 -3 -0.06 -0.48 0.72 0.76 -2 -0.16 -1.29 0.56 0.59 -1 -0.13 -1.05 0.43 0.45 0 -0.48 -3.88 -0.05 -0.05 1 -0.09 -0.73 -0.14 -0.14 2 -0.18 -1.45 -0.32 -0.33 3 -0.05 -0.40 -0.37 -0.37 4 0.01 0.08 -0.36 -0.36 5 0.17 1.37 -0.19 -0.19 6 -0.08 -0.65 -0.27 -0.27 7 0.04 0.32 -0.23 -0.23 8 0.01 0.08 -0.22 -0.21 9 0.01 0.08 -0.21 -0.20 10 -0.09 -0.73 -0.30 -0.29 11 0.20 1.62 -0.10 -0.10 12 0.04 0.32 -0.06 -0.06 13 -0.19 -1.54 -0.25 -0.23 14 0.06 0.48 -0.19 -0.18 15 0.05 0.40 -0.14 -0.13 20 -0.10 -0.81 -0.58 -0.52 30 -0.02 -0.16 -0.81 -0.69 40 -0.06 -0.48 -1.02 -0.82 50 -0.07 -0.57 -1.55 -1.19 60 -0.15 -1.21 -2.25 -1.65

PAGE 141

134 Table 6-7 Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bonds: Three Strata by CP Ratio Large n = 84 Medium n = 86 Small n = 86 Obs. 2AR t 2AR t 2AR t 1 0.13 % 0.38 0.04 % 0.12 -0.11 % -o. 31 5 0.01 0.03 0.11 0.32 -0.24 -0.69 10 -0.72 -2.12 ** -0.50 -1.47 -0.35 -1.00 15 -0.06 -0.18 -0.02 -0.06 -0.23 -0.66 16 0.07 0.21 a.so 1.47 0.27 0.77 17 0.33 0.97 -0.58 -1.71 * -0.77 -2.20 ** 18 -0.14 -0.41 -0.36 -1.06 0.51 1.46 19 -0.07 -0.21 0.12 0.35 -0.15 -0.43 20 0.10 0.29 -0.36 -1.06 0.04 0.11 21 0.49 1.44 0.11 0.32 -0.09 -0.26 22 0.48 1.41 -0.14 -0.41 -0.34 -0.97 23 0.09 0.26 0.31 0.91 -0.09 -0.26 24 -0.18 -0.53 -0.15 -0.44 0.09 0.26 25 0.05 0.15 -0.65 -1.91 * -0.39 -1.11 26 -0.03 -0.09 -0.14 -0.41 -0.28 -0.80 27 -0.01 -0.03 -0.31 -0.91 0.04 0.11 28 -0.04 -0.12 0.32 0.94 -0.52 -1.48 29 -0.60 -1.76 * 0.24 0.71 -0.33 -0.94 30 a 0.19 0.55 -0.59 -1.73 * -1.40 -4.00 *** 31 0.31 0.91 -0.35 -1.03 -0.38 -1.09 32 0.39 1.15 -0.25 -0.74 -0.42 -1.20 33 -0.32 -0.94 0.25 0.74 -0.03 -0.09 34 0.23 0.68 -0.52 -1.53 0.33 0.94 35 -0.33 -0.97 -0.10 -0.29 0.60 1.71 * 36 0.28 0.82 -0.11 -0.32 0.10 0.29 37 -0.36 -1.06 -0.07 -0.21 0.05 0.14 38 0.36 1.06 -0.23 -0.68 -0.08 -0.23 39 -0.35 -1.03 -0.11 -0.32 -0.28 -0.80 40 0.18 0.53 -0.04 -0.12 -0.89 -2.54 ** 41 -0.14 -0.41 -0.07 -0.21 -0.42 -1.20 42 -0.57 -1.68 * 0.03 0.09 -0.40 -1.14 43 0.33 0.97 0.27 0.79 -0.50 -1.43 44 -0.65 -1.91 * 0.53 1.56 0.10 0.29 45 0.00 0.00 0.32 0.94 0.39 1.11 50 0.19 0.56 0.41 1.21 -0.43 -1.23 55 -0.12 -0.35 -0.38 -1.12 0.24 0.69 60 0.29 0.85 -0.34 -1.00 -0.59 -1.69 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 142

135 Table 6-8 Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bonds: Three strata by Relative Size Large n = 87 Medium n = 87 Small n = 87 Obs. 2AR t 2AR t 2AR t l -0.34 % -0.87 0.30 % 1.15 0.17 % 0.52 5 0.26 0.67 . 0.15 0.58 -0.08 -0.24 10 -0.64 -1.64 -0.46 -1.77 * -0.98 -2.97 *** 15 -0.41 -1.05 0.09 0.35 0.38 1.15 16 0.10 0.26 0.19 0.73 0.27 0.82 17 -1.11 -2.84 *** -0.42 -1.62 0.34 1.03 18 0.44 1.13 -0.28 -1.08 o.oo 0.00 19 0.37 0.95 -0.15 -0.58 0.35 1.06 20 0.02 0.05 -0.36 -1.38 -0.24 -0.73 21 0.01 0.03 0.46 1.77 * 0.44 1.33 22 0.12 0.31 -0.12 -0.46 0.25 0.76 23 -0.21 -0.54 0.53 2.04 ** -0.10 -0.30 24 -0.27 -0.69 -0.33 -1.27 0.17 0.52 25 -0.23 -0.59 -0.35 -1.35 -0.18 -0.55 26 -o. 31 -0.79 -0.25 -0.96 0.06 0.18 27 -0.25 -0.64 -0.22 -0.85 0.20 0.61 28 0.02 0.05 -0.15 -0.58 0.30 0.91 29 0.00 -0.87 -0.14 -0.54 -0.35 -1.06 30 a -0.57 -1.46 -0.59 -2.27 ** -0.41 -1.24 31 -1.18 -3.03 *** 0.12 0.46 0.58 1.76 * 32 -0.09 -0.23 0.35 1.35 -0.13 -0.39 33 0.15 0.38 -0.19 -0.74 0.14 0.42 34 -0.06 -0.15 0.24 0.92 -o. 20 -0.61 35 0.30 0.77 -0.06 -0.23 -0.10 -0.30 36 0.40 1.03 -0.01 -0.04 0.05 0.15 37 -0.41 -1.05 0.07 0.27 -0.37 -1.12 38 0.01 0.03 0.05 0.19 -0.18 -0.55 39 -0.56 -1.44 -0.34 -1.31 -0.02 -0.06 40 -0.29 -0.74 -0.22 -0.85 -0.06 -0.18 41 -0.12 -0.31 -0.27 -1.04 -0.20 -0.69 42 -0.43 -1.10 0.00 0.00 -0.47 -1.42 43 0.28 0.72 -0.01 -0.04 -0.11 -0.33 44 0.11 0.28 -0.08 -0.31 0.07 0.21 45 0.64 1.64 0.27 1.04 0.04 0.12 50 -0.15 -0.38 -0.08 -0.31 0.54 1.64 55 0.50 1.28 0.03 -0.57 -0.57 -1.73 * 60 -0.19 -0.49 -0.23 -0.88 0.07 0.21 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the Issue date and the day before.

PAGE 143

136 Table 6-9 Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bonds: Three Strata by Firm Size Large n = 88 Medium n = 88 Small n = 88 Obs. 2AR t 2AR t 2AR t 1 0.18 % 0.62 0.10 % 0.29 -0.17 % -0.44 5 -0.04 -0.14 0.17 0.50 0.18 0.46 10 -0.79 -2.72 *** -0.70 -2.06 ** -0.57 -1.46 15 0.05 0.17 -0.47 -1.38 0.58 1.49 16 0.45 1.55 0.31 0.91 -0.28 -0.72 17 0.50 1.72 * -0.22 -0.65 -1.48 -3.79 *** 18 -0.19 -0.66 0.00 o.oo 0.28 0.72 19 0.15 0.52 -0.32 -0.94 0.63 1.62 20 -0.18 -0.62 -0.14 -0.41 -0.22 -0.56 21 0.34 1.17 0.16 0.47 0.42 1.08 22 0.45 1.55 -0.58 -1.71 * 0.35 0.90 23 -0.40 -1.38 0.69 2.03 ** -0.15 -0.38 24 0.07 0.24 -0.80 -2.35 ** 0.38 0.97 25 -0.74 -2.55 ** 0.51 1.50 -0.50 -1.28 26 0.33 1.14 -0.44 -1.29 -0.37 -0.95 27 0.09 0.31 -0.17 -a.so -0.27 -0.69 28 0.38 1.31 -0.13 -0.38 -0.06 -0.15 29 -0.23 -0.79 -0.15 -0.44 -0.40 -1.02 30 a -0.85 -2.93 *** -0.46 -1.35 -0.19 -0.49 31 0.37 1.28 -0.21 -0.62 -0.61 -1.56 32 0.24 0.83 -0.26 -0.76 0.04 0.10 33 0.23 0.79 -0.29 -0.85 0.21 0.54 34 -0.42 -1.45 -0.09 -0.26 0.59 1.51 35 -0.23 -0.79 0.16 0.47 0.17 0.44 35 0.09 0.31 -0.05 -0.15 0.35 0.90 37 -0.16 -0.55 -0.43 -1.26 -0.06 -0.15 38 -0.14 -0.48 0.31 -0.91 -0.16 -0.41 39 -0.13 -0.45 -0.29 -0.85 -0.39 -1.00 40 -0.05 -0.17 -0.09 -0.26 -0.39 -1.00 41 -0.26 -0.90 -0.09 -0.27 -0.33 -0.85 42 -0.36 -1.24 -0.61 -1.79 * 0.05 0.13 43 -0.09 -o. 31 0.14 0. 41 -0.01 -0.03 44 0.15 0.52 -0.01 -0.03 -0.06 -0.15 45 0.10 0.34 0.30 0.88 0.53 1.36 50 0.23 0.79 -0.01 -0.03 0.10 0.26 55 -0.27 -0.93 -0.02 -0.06 0.30 0.77 60 0.04 0.14 -0.62 -1.82 * 0.26 0.67 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 144

137 Table 6-11 Summary Results for Issue Date Tests Test N 2AR z No. Neg Issue Date 298 -0.61% -3.21 *** 183 CP Ratio Low 86 -1.40 -4.00 *** 67 Middle 86 -0.59 -1.73 * 56 High 84 0.19 0.55 37 Relative Size Small 87 -0.41 -1.24 49 Medium 87 -0.59 -2.27 ** 55 Large 87 -0.57 -1.46 57 Firm Size Small 88 -0.19 -0.49 56 Medium 88 -0.46 -1.35 51 Large 88 -0.85 -2.93 *** 53 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level z 3.94 *** 5.18 *** 2.80 *** -1.09 1.18 2.47 ** 2.89 *** 2.56 *** 1.49 1.92 *

PAGE 145

138 Table 6-12a Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues Single Firms n = 171 Obs. 1 5 10 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 a 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 50 55 60 2AR 0.26 % 0.15 0.24 -0.36 0.22 0.10 0.30 -0.30 0.21 0.40 -0.26 0.04 0.02 0.01 0.33 0.27 0.22 -0.26 -1.37 -0.51 -0.15 0.20 -0.03 -0.20 -0.61 0.06 0.05 0.24 -0.17 -0.11 0.84 0.19 0.18 -0.15 0.08 -0.12 -0.31 t 0.96 0.56 0.89 -1.33 0.81 0.37 1.11 -1.11 0.78 1.48 -0.96 0.15 0.07 0.04 1.22 1.00 0.81 -0.96 -5.07 *** -1.89 * -0.56 0.74 -0.11 -0.74 -2.26 ** 0.22 0.19 0.89 -0.63 -0.41 3.11 *** 0.70 0.67 -0.56 0.30 -0.44 -1.15 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level Multi Firms n = 61 2AR 0.11 % 0.16 0.76 -0.11 0.03 0.39 -0.29 0.04 -0.34 0.47 0.43 -0.04 0.05 -0.18 0.27 -0.05 -0.20 -0.69 -1.59 -0.19 0.90 -0.08 0.24 0.03 0.16 -0.84 -0.73 -0.47 -0.20 -0.19 0.14 0.11 -0.14 0.22 0.40 0.29 -0.31 t 0.28 0.41 1.95 * -0.28 0.08 1.00 -0.74 0.10 -0.87 1.21 1.10 -0.10 0.13 -0.46 0.69 -0.13 -0.51 -1.77 * -4.08 *** -0.49 2.31 ** -0.21 0.62 0.08 0.41 -2.15 ** -1.87 * -1.21 -0.51 -0.49 0.36 0.28 -0.36 0.56 1.03 0.74 -0.79 a Two-day excess return consisting of the announcement date and the day before.

PAGE 146

139 Table 6-12b ily Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) or 121 Days Around the Announcement of Convertible Bond Issues: Single Firms n = 171 Trading Day -60 -so -40 -30 -20 -15 -14 -13 -12 -11 -10 -9 -8 -7 -6 -5 -4 -3 -2 -1 o a 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 20 30 40 50 60 AR 0.10 % 0.00 0.10 -0.13 -0.01 0.15 -0.11 -0.17 0.19 0.02 -0.01 0.04 0.29 0.23 0.04 0.27 -0.05 -0.20 -0.58 -0.64 -0.73 -0.25 -0.26 -0.05 -0.10 0.14 0.06 -0.19 0.16 0.01 -0.21 -0.42 -0.19 0.23 -0.17 -0.08 -0.04 -0.17 0.09 -0.16 -0.06 t 0.58 0.02 0.54 -0.73 -0.04 0.87 -0.63 -0.95 1.07 0.11 -0.04 0.24 1.66 * 1.33 0.23 1.54 -0.26 -1.13 -3.27 *** -3.62 *** -4.12 *** -1.40 -1.46 -o. 30 -0.54 0.80 0.35 -1.09 0.92 0.03 -1.21 -2.40 ** -1.09 1.28 -0.99 -0.48 -0.25 -0.98 0.49 -0.92 -0.32 *** Significant at 1% Level ** Significant at 5% Level * Significant at 10% Level CAR 0.10 % 1.26 2.16 3.01 3.55 3.84 3.73 3.56 3.75 3.77 3.77 3.81 4.10 4.33 4.37 4.65 4.60 4.40 3.82 3.18 2.46 2.21 1.95 1.90 1.80 1.94 2.01 1.81 1.98 1.98 1.77 1.35 1.15 1.38 1.21 1.12 1.31 2.26 1.78 1.28 0.24 t 0.58 2.14 ** 2.67 *** 3.06 *** 3.14 *** 3.21 *** 3.08 *** 2.91 *** 3.04 *** 3.02 *** 2.99 *** 2.99 *** 3.19 *** 3.34 *** 3.34 *** 3.52 *** 3.45 *** 3.27 *** 2.82 *** 2.33 ** 1.78 * 1.59 1. 39 1.34 1.27 1.36 1.39 1.25 1.35 1.34 1.19 0.90 0.77 0.91 0.79 0.73 0.82 1.34 1.00 0.69 0.12 a The announcement date as identified in the WSJ

PAGE 147

140 Table 6-12c Daily Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) For 121 Days Around the Announcement of Convertible Bond Issues: Multi Firms n = 61 Trading Day AR t CAR t -60 -0.07 % -0.25 -0.07 % -0.25 -50 0.19 0.70 0.86 0.95 -40 0.15 0.54 2.04 1.63 -30 -0.38 -1.40 2.64 1.73 * -20 -0.19 -0.71 2.48 1.42 -15 0.13 0.48 3.50 1.89 * -14 -0.17 -0.62 3.34 1.78 * -13 0.17 0.64 3.51 1.85 * -12 -0.12 -0.45 3.38 1.77 * -11 0.05 0.17 3.43 1.78 * -10 -0.23 -0.82 3.21 1.64 * -9 0.17 0.61 3.37 1.71 * -8 0.10 0.36 3.47 1.75 * -7 -0.02 -0.08 3.45 1.72 * -6 -0.03 -0.09 3.42 1.69 * -5 -0.27 -0.98 3.15 1.54 -4 0.07 0.26 3.22 1.56 -3 -0.13 -0.47 3.10 1.49 -2 -0.56 -2.06 ** 2.53 1.21 -1 -0.86 -3.15 *** 1.67 0.79 0 a -0.73 -2.66 *** 0.95 0.44 1 -0.11 -0.40 0.84 0.39 2 -0.08 -0.30 0.75 0.35 3 0.57 2.10 ** 1.33 0.61 4 0.33 1.20 1.66 0.75 5 -0.17 -0.61 1.49 0.67 6 0.09 0.33 1.58 0.71 7 0.30 1.10 1.88 0.83 8 -0.06 -0.23 1.82 a.so 9 -0.31 -1.13 1.51 0.66 10 0.34 1.24 1.84 0.80 11 -0.10 -0.37 1.74 0.75 12 0.26 0.95 2.00 0.86 13 -0.20 -0.74 1.80 0.77 14 -0.64 -2.33 *** 1.16 0.49 15 -0.56 -2.06 ** 0.60 0.25 20 -0.04 -0.16 -0.23 -0.10 30 -0.05 -0.19 -0.09 -0.03 40 0.38 1.38 -1.68 -0.61 50 0.12 0.45 -2.76 -0.96 60 -0.52 -1.92 * -3.25 -1.08 *** Significant at 1% Level ** Significant at 5% Level * Significant at 10% Level a The announcement date as identified in the WSJ

PAGE 148

141 Table 6-12d Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of convertible Bond Issues: Three Strata by Relative size Single Firms Large n = 52 Medium n = 53 Small n = 53 Obs. 2AR t 2AR t 2AR t 1 0.59 % 1.23 -0.12 % -0.31 0.06 % 0.14 5 -0.37 -0.77 0.39 1.00 0.37 0.88 10 0.29 0.60 -0.05 -0.13 0.22 0.52 15 -0.52 -1.08 -0.29 -0.74 -0.03 -0.07 16 0.49 1.02 0.33 0.85 -0.27 -0.64 17 -0.58 -1.21 0.23 0.59 0.49 1.17 18 0.62 1.29 0.40 1.03 -0.15 -0.36 19 -0.25 -0.52 -0.45 -1.15 0.12 0.29 20 0.22 0.46 -0.47 -1. 21 0.53 1.26 21 0.47 0.98 -0.13 -0.33 0.58 1.38 22 -0.45 -0.94 0.07 0.18 -0.98 -2.33 ** 23 -0.05 -0.10 0.16 0.41 o.oo o.oo 24 -0.74 -1.54 0.17 0.44 0.65 1.55 25 -0.33 -0.69 0.11 0.28 0.19 0.45 26 -0.04 -0.08 0.20 0.51 0.50 1.19 27 0.27 0.56 0.30 0.77 -0.28 -0.67 28 0.85 1.77 * 0.04 0.10 0.21 0.50 29 -0.60 -1.25 -0.66 -1.69 * 0.39 0.93 30 a -1.33 -2.77 *** -1.25 -3.21 *** -1.26 -3.00 *** 31 -0.34 -0.71 -0.65 -1.67 * -0.34 -0.81 32 -0.31 -0.65 -0.30 -0.77 0.31 0.74 33 -0.01 -0.02 -0.09 -0.23 0.38 0.90 34 0.35 0.73 -0.27 -0.69 0.06 0.14 35 0.05 0.10 -0.16 -0.41 -0.20 -0.48 36 -0.14 -0.29 -0.83 -2.12 ** -0.84 -2.00 ** 37 -0.13 -0.27 -0.32 -0.82 0.42 1.00 38 0.06 0.13 0.01 0.03 0.09 0.21 39 0.66 1.38 0.34 0.87 0.34 0.81 40 -0.90 -1.88 * 0.01 0.03 0.26 0.62 41 -0.21 -0.44 -0.10 -0.26 -0.24 -0.57 42 1.25 2.60 *** 0.50 1.28 0.31 0.74 43 0.04 0.08 -0.12 -0.31 0.69 1.64 44 -0.34 -0.71 0.33 0.85 0.52 1.24 45 0.53 1.10 -0.41 -1.05 -0.22 -0.52 50 0.34 0.71 -0.14 -0.36 0.07 0.17 55 -0.44 -0.92 -0.02 -0.05 0.48 1.14 60 0.22 0.46 -0.52 -1.33 -0.60 -1.42 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 149

142 Table 6-12e Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by Relative Size Multi Firms Large n = 18 Medium n = 18 Small n = 18 Obs. 2AR t 2AR t 2AR t 1 1.17 % 1.48 -0.65 % -1.06 -0.26 % -0.40 5 0.10 0.13 0.51 0.84 0.75 1.17 10 0.87 1.10 0.67 1.10 -0.41 -0.65 15 0.56 0.71 0.23 0.38 -0.26 -0.41 16 -1.04 -1.31 1.40 2.30 ** -0.25 -0.39 17 0.13 0.16 0.85 1.39 0.01 0.02 18 -1.12 -1.42 0.02 0.03 -o. 31 -0.48 19 0.71 0.90 -0.56 -0.92 0. 30 0.47 20 -0.73 -0.92 1.18 1.93 * -1.21 -1.89 * 21 0.07 0.09 1.12 1.84 * 0.96 1.50 22 0.53 0.67 0.62 1.02 -0.28 -0.44 23 -0.73 -0.92 -0.49 -0.80 0.37 0.58 24 0.09 0.11 -0.45 -0.74 0.23 0.36 25 -0.41 -0.52 0.32 0.52 -0.48 -0.75 26 0.01 0.13 0.06 0.08 0.66 1.03 27 0.03 0.04 1.21 1.98 * -1.08 -1.68 28 -0.62 -0.78 -0.28 -0.46 -0.33 -0.52 29 -0.35 -0.44 -0.93 -1.52 -0.20 -0.31 30 a -3.54 -4.48 *** -0.25 -0.41 -1.06 -1.65 31 -0.06 -0.08 -0.65 -1.06 0.20 0.31 32 0.44 0.56 0.74 1.21 1.05 1.64 33 0.32 0.41 -0.02 -0.03 -0.48 -0.75 34 -0.15 -0.19 -0.13 -0.21 0.52 0.81 35 -0.12 -0.15 -0.06 -0.10 -0.20 -0.31 36 1.00 1.27 0.18 0.30 0.19 0.30 37 -1.30 -1.64 -1.08 -1.77 * -0.85 -1.33 38 -1.29 -1.63 -0.25 -0.41 -0.59 -0.92 39 -0.85 -1.07 0.12 0.19 -1.01 -1.58 40 -1.44 -1.82 * 1.29 2.11 ** 0.79 1.23 41 -0.25 -0.32 0.35 0.57 -0.67 -1.04 42 1.10 1.39 -1.29 -2.11 ** 0.78 1.22 43 -0.41 -0.52 0.57 0.93 0.54 0.84 44 -1.97 -2.49 ** 0.23 0.38 0.90 1.41 45 1.02 1.29 -0.14 -0.23 -0.20 -0.31 50 0.00 0.00 0.24 0.39 0.51 0.80 55 1.24 1.57 0.12 0.20 0.11 0.17 60 -0.49 -0.62 -0.56 -0.92 -0.59 -0.92 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date . and the day before.

PAGE 150

143 Table 6-12f Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by Firm Size Single Firms Large n = 53 Medium n = 52 Small n = 53 Obs. 2AR t 2AR t 2AR t 1 0.26 % 0.76 0.22 % 0.65 0.06 % 0.11 5 0.51 1.50 -0.28 -0.82 0.16 0.28 10 0.14 0.41 0.18 0.53 0.13 0.23 15 -0.15 -0.44 -0 . 74 -2.17 ** 0.03 0.05 16 0. 21 0.62 0.39 1.15 -0.04 -0.07 17 0.20 0.59 0.83 2.44 ** -0.87 -1.52 18 0.04 0.12 0.14 0.41 0.70 1.23 19 -0.43 -1.26 -0.29 -0.85 0.14 0.46 20 0.02 0.06 -0.17 -0.50 0.42 0.74 21 0.09 0.26 0.09 0.26 0.74 1.29 22 -0.57 -1.67 * -0.31 -0.91 -0.47 -0.82 23 0.22 0.65 0.16 0.47 -0.27 -0.47 24 0.53 1.74 * -0.49 -1.44 -0.03 -0.05 25 0.02 0.06 -0.07 -0.21 0.01 0.02 26 0.44 1.29 0.21 0.62 0.01 0.02 27 -0.30 -0.88 0.46 1. 35 0.13 0.23 28 0.05 0.15 0.16 0.47 0.90 1.58 29 0.10 0.29 0.21 0.62 -1.19 -2.08 ** 30 a -1.45 -4.26 *** -0.85 -2.50 ** -1.53 -2.68 *** 31 -0.33 -0.97 -0.89 -2.61 *** -0.14 -0.25 32 0.09 0.26 0.00 o.oo -0.42 -0.74 33 0.29 0.85 0.25 0.74 -0.26 -0.46 34 0.31 0.91 -0.16 -0.47 -0.01 -0.02 35 -0.13 -0.28 0.07 0.21 -0.24 -0.42 36 -0.94 -2.76 *** -0.69 -2.03 ** -0.18 -0.32 37 0.16 0.47 -0.25 -0.74 0.05 0.09 38 -0.39 -1.15 0.36 1.06 0.19 0.33 39 0.57 1.68 * -0.22 -0.65 0.98 1.72 40 -0.11 -0.32 -0.10 -0.29 -0.42 -0.74 41 -0.23 -0.68 -0.17 -0.50 -0.16 -0.28 42 0.56 1. 65 * -0.24 -0.71 1.73 3.03 *** 43 0.84 2.47 ** -0.35 1.03 0.11 0.19 44 0.37 1.09 0.47 1.38 -0.34 -0.60 45 -0.33 -0.97 -0.33 -0.97 0.55 0.96 50 -0.26 -0.76 -0.30 -0.88 0.83 1.45 55 0.27 0.79 0.22 0.65 -0.46 -0.81 60 -0.35 -1.03 -0.01 -0.03 -0.53 -0.93 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 151

144 Table 6-12g Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by Firm Size Multi Firms Large n = 18 Medium n = 18 Small n = 18 Obs. 2AR t 2AR t 2AR t 1 -0.44 % -0.86 0.14 % 0.20 0.53 % 0.71 5 0.27 0.53 2.09 2.94 *** -1.01 -1. 31 10 0.79 1.55 0.01 0.01 0.32 0.42 15 0.34 0.67 -0.32 -0.45 0.51 0.66 16 0.27 0.53 0.37 0.52 -0.51 -0.66 17 0.67 1.31 0.45 0.63 -0.13 -0.17 18 0.69 1.35 -0.36 -0.51 -1.74 -2.26 ** 19 0.06 0.12 0.25 0.35 0.14 0.18 20 -0.31 -0.61 -0.27 -0.38 -0.19 -0.25 21 0.28 0.55 0.71 1.00 1.15 1.49 22 0.22 0.43 -0.56 -0.79 1.21 1.57 23 0.40 0.78 -0.49 -0.69 -0.75 -0.97 24 -0.07 -0.14 0.80 1.13 -0.84 -1.09 25 0.08 0.16 0.06 0.08 -0.71 -0.92 26 0.19 0.37 0.85 1.20 -0.30 -0.39 27 0.01 0.02 0.02 0.03 0.13 0.17 28 -0.90 -1.76 * -0.05 -0.07 -0.29 -0.38 29 -0.60 -1.18 -0.58 -0.82 -0.30 -0.89 30 a -1.56 -3.06 *** -1.84 -2.59 *** -1.45 -1.88 * 31 -0.18 -0.35 -0.26 -0.36 -0.06 -0.08 32 1.06 2.08 * -0.29 -0.41 1.45 1.88 * 33 0.12 0.23 0.24 0.34 -0.55 -0.71 34 -0.70 -1.37 0.51 0.72 0.44 0.57 35 0.02 0.04 -0.37 -0.52 -0.01 -0.01 36 -0.04 -0.08 1.80 2.53 ** -0.37 -0.48 37 -1.00 -1.96 * -1.45 -2.04 * -0.77 -1.00 38 -1.01 -1.98 * -0.33 -0.46 -0.79 -1.03 39 -0.28 -0.55 -0.69 -0.97 -0.78 -1.01 40 0.09 0.18 -0.19 -0.27 0.74 0.96 41 -0.37 -0.73 -0.74 -1.04 0.55 0.71 42 0.04 0.08 -0.34 -0.48 0.90 1.17 43 -0.31 -0.61 0.68 0.96 0.32 0.42 44 -0.29 -0.57 -0.95 -1.34 0.41 0.53 45 -0.38 -0.74 0.99 1.39 0.06 0.08 50 0.73 1.43 -0.60 -0.85 0.62 0.81 55 -0.26 -0.51 1.15 1.62 0.59 0.77 60 -0.31 -0.61 -1.01 -1.43 -0.32 -0.42 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 152

145 Table 6-12h Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by CP Ratio Single Firms High n = 52 Medium n = 52 Low n = 52 Obs. 2AR t 2AR t 2AR t 1 0.39 % 0.95 -0.06 % -0.15 0.32 % 0.59 5 0.27 0.66 0.07 0.17 0.16 0.30 10 0.36 0.88 -0.10 -0.24 0.51 0.94 15 -0.09 -0.22 -0.77 -1.88 * -0.44 -0.81 16 0.09 0.22 0.32 0.78 0.02 0.04 17 0.23 0.56 0.48 1.17 -0.04 -0.07 18 0.25 0.61 0.06 0.15 0.23 0.43 19 -0.78 -1.90 * 0.29 0.71 -0.60 -1.11 20 0.46 1.12 0.21 0.51 -0.04 -0.07 21 0.00 0.00 0.42 1.02 0.22 0.41 22 -0.48 -1.17 -0.04 -1.00 -0.48 -0.89 23 0.03 0.07 0.33 0.80 -0.19 -0.35 24 0.59 1.44 0.05 0.12 -0.22 -0.41 25 0.14 0.34 -0.03 -0.07 0.09 0.17 26 0.23 0.56 0.65 1.58 0.16 0.30 27 -0.22 -0.54 0.50 1.22 0.35 0.65 28 0.58 1.41 -0.06 -0.15 0.60 1.11 29 -0.35 -0.85 -0.12 -0.29 -0.12 -0.22 30 a -1.04 -2.54 ** -1.49 -3.63 *** -1.72 -3.19 *** 31 -0.05 -0.12 -0.43 -1.05 -0.73 -1.35 32 0.61 1.48 -0.32 -0.78 -0.47 -0.87 33 0.24 0.59 -0.30 -0.73 0.63 1.17 34 0.35 0.85 0.41 1.00 -0.66 -1.22 35 0.14 0.34 -0.05 -0.12 -0.93 -1.72 * 36 -0.45 -1.10 -0.68 -1.66 -0.81 -1.50 37 0.33 0.80 -0.09 -0.22 0.05 0.09 38 0.29 0.71 -0.03 -0.07 0.06 0.11 39 0.97 1.37 0.45 1.10 -0.78 -1.44 40 -0.33 -0.81 0.21 0.51 -0.52 -0.96 41 0.08 0.20 -0.12 -0.29 -0.37 -0.69 42 0.47 1.15 0.54 1.32 1.06 1.96 ** 43 0.39 0.95 0.16 0.39 0.28 0.52 44 0.45 1.10 0.01 0.02 0.18 0.33 45 -0.57 -1.39 -0.15 -0.37 -0.15 -0.28 50 0.22 0.54 -0.17 -0.41 0.05 0.09 55 0.04 0.10 -0.42 -1.02 0.30 0.56 60 -0.51 -1.24 0.01 0.02 -0.29 -0.54 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 153

--146 Table 6-12i Two-Day Abnormal Returns (2AR) For 121 Days Around the Announcement of Convertible Bond Issues: Three Strata by CP Ratio Multi Firms High n = 19 Medium n = 19 Low n = 18 Obs. 2AR t 2AR t 2AR t l 1.52 % 2.03 ** -0.86 % -1.26 0.08 % 0.13 5 1.19 1.59 0.27 0.40 -0.32 .-0.52 10 0.19 0.25 0.05 0.07 0.43 0.69 15 -1.23 -1.64 1.32 1.94 * -0.23 -0.37 16 -0.50 -0.67 -0.76 -1.12 1.11 1.79 * 17 0.20 0.27 0.39 0.57 0.75 1.21 18 -0.14 -0.19 -0.66 -0.97 -0.20 -0.32 19 0.17 0.23 0.21 0.31 0.24 0.39 20 -1.47 -1.96 * 0.27 0.40 0.01 0.02 21 0.54 0.72 0.91 1.34 0.28 0.45 22 0.16 0.21 -0.06 -0.09 0.76 1.23 23 1.18 1.57 -0.67 -1.00 -0.90 -1.45 24 0.41 0.55 -1.22 -1.79 * 1.14 1.84 * 25 -0.78 -1.04 0.13 0.19 0.38 0.61 26 -0.38 -0.51 0.36 0.53 0.01 0.02 27 0.67 0.89 -0.75 -1.10 0.43 0.69 28 -0.74 -1.00 -0.50 -0.74 0.26 0.42 29 -0.26 -0.35 -0.84 -1.24 -0.26 -0.42 30 a -1.44 -1.92 * -1. 47 -2.16 ** -2.24 -3.61 *** 31 0.32 0.43 0.12 0.18 -0.71 -1.15 32 0.10 0.13 1.54 2.26 ** 0.46 0.74 33 0.92 1.23 -0.08 -0.12 -0.54 -0.87 34 0.92 1.23 -0.81 -1.19 -0.20 -0.32 35 0.62 0.83 0.60 0.88 -1. 35 -2.20 ** 36 0.76 1.01 0.06 0.09 0.24 0.39 37 -0.28 -0.37 -2.11 -3.10 *** -0.83 -1.34 38 -0.16 -0.21 -0.33 -0.48 -1.19 -1.92 * 39 -1.00 -1.33 -0.63 -0.93 -0.17 -0.27 40 -0.25 -0.33 0.76 1.12 0.14 0.21 41 -0.45 -0.60 -0.29 -0.43 0.41 0.66 42 -0.81 -1.08 1. 47 2.16 ** -0.29 -0.47 43 1.51 2.01 ** -0.47 -0.69 -0.67 -1.08 44 -0.69 -0.92 0.00 0.00 0.14 0.23 45 0.13 0.17 0.34 0.50 0.00 o.oo 50 -0.42 -0.56 0.42 0.62 0.82 1.32 55 -0.01 -0.01 0.07 0.10 1.02 1.64 60 -0.49 -0.65 -0.13 -0.19 -0.71 -1.14 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the announcement date and the day before.

PAGE 154

147 Table 6-12j Summary Results for Announcement Date Tests: Single and Multi Firms Single Firms 171 -1.37 -5.07 *** 121 5.43 Multi Firms 61 -1.59 -4.08 *** 48 4.48 CP Ratio Single Firms Small 52 -1.72 -3.19 *** 39 3.61 Medium 52 -1.49 -3.63 *** 37 3.05 Large 52 -1.04 -2.54 ** 35 2.50 Multi Firms Small 18 -2.24 -3.61 *** 16 3.30 Medium 19 -1.47 -2.16 ** 14 2.06 Large 19 -1.44 -1.92 * 15 2.52 Relative Size Single Firms Small 53 -1.26 -3.00 *** 39 3.43 Medium 53 -1.25 -3.21 *** 35 2.34 Large 52 -1.33 Multi Firms -2.77 *** 38 3.33 Small 18 -1.06 -1.65 14 2.36 Medium 18 -0.25 -0.41 12 1.41 Large 18 -3.54 -4.48 *** 16 3.30 Firm Size Single Firms Small 53 -1.53 -2.68 *** 39 3.43 Medium 52 -0.85 -2.50 ** 35 2.50 Large 53 -1.45 Multi Firms -4.26 *** 38 3.16 Small 18 -1.45 -1.88 * 13 1.89 Medium 18 -1.84 -2.59 *** 14 2.36 Large 18 -1.56 -3.06 *** 15 2.83 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level

PAGE 155

148 Table 6-13a Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bonds Multi Firms n = 73 Single Firms n = 225 Obs. 2AR t 2AR t 1 0.49 1.36 -0.12 -0.55 5 -0.25 -0.69 0.13 0.59 10 -0.71 -1.97 ** -0.34 -1.55 15 -0.19 -0.53 -0.07 -0.32 16 0.44 1.22 0.10 0.45 17 0.28 0.78 -0.56 -2.55 ** 18 0.45 1.25 0.00 0.00 19 0.25 0.69 -0.11 -0.50 20 0.06 0.17 -0.16 -0.72 21 0.10 0.28 0.26 1.18 22 -0.53 -1.47 0.23 1.05 23 0.11 0.31 0.19 0.86 24 -0.02 -0.06 -0.14 -0.64 25 -0.54 -1.50 -0.20 -0.91 26 0.51 1.42 -0.51 -2.32 ** 27 -0.31 -0.86 0.06 0.27 28 -0.29 -0.81 0.10 0.45 29 -0.24 -0.67 -0.22 -1.00 30 a -1.28 -3.56 *** -0.39 -1.77 * 31 -0.16 -0.44 -0.31 -1.41 32 -0.14 -0.39 -0.01 -0.05 33 0.05 0.14 0.10 0.45 34 -0.24 -0.67 0.15 0.68 35 o.oo 0.00 -0.12 -0.55 36 0.22 0.61 0.25 1.13 37 -0.29 -0.81 -0.07 -0.32 38 0.54 1.50 -0.10 -0.45 39 -0.22 -0.61 -0.26 -1.18 40 -0.64 -1.78 * -0.06 -0.27 41 -0.12 -0.33 -0.20 -0.91 42 -0.41 -1.14 -0.21 -0.95 43 0.05 0.14 -0.03 -0.14 44 -0.61 -1.69 * 0.26 1.18 45 0.08 0.22 0.23 1.05 50 0.07 0.19 0.05 0.23 55 0.38 1.06 -0.24 -1.09 60 -0.21 -0.58 -0.17 -0.77 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 156

-~ -----149 Table 6-13b Daily Abnormal Returns (AR) and Cumulative Abnormal Returns (CAR) For 121 Days Around the Sale of Convertible Bonds: Single Firms n = 225 Trading Day AR t CAR t -60 -0.05 -0.35 -0.05 -0.35 -50 0.17 1.20 0.79 1.68 -40 -0.24 -1.69 0.83 1.28 -30 0.04 0.28 1.33 1.68 -20 -0.02 -0.14 0.60 0.66 -15 0.03 0.21 1.12 1.16 -14 0.16 1.13 1.28 1.32 -13 0.05 0.35 1.33 1.35 -12 -0.19 -1.34 1.14 1.15 -11 -0.01 -0.07 1.13 1.13 -10 -0.19 -1.34 0.94 0.93 -9 -0.25 -1.76 0.69 0.67 -a -0.26 -1.83 0.43 0.42 -7 -0.06 -0.42 0.37 0.35 -6 0.11 0.78 0.48 0.46 -5 0.06 0.42 0.54 0.51 -4 0.04 0.28 0.58 0.54 -3 -0.07 -0.49 0.51 0.47 -2 -0.15 -1.06 0.36 0.33 -1 0.08 0.56 0.44 0.40 0 -0.47 -3.31 -0.03 -0.03 l -0.07 -0.49 -0.10 -0.09 2 -0.24 -1.69 -0.34 -0.30 3 -0.03 -0.21 -0.37 -0.33 4 0.02 0.14 -0.35 -0.31 5 0.19 l. 34 -0.16 -0.14 6 -0.09 -0.63 -0.25 -0.22 7 0.05 0.35 -0.20 -0.17 8 0.10 0.70 -0.10 -0.08 9 -0.03 -0.21 -0.13 -0.11 10 -0.09 -0.63 -0.22 -0.18 11 0.15 1.06 -0.07 -0.06 12 0.10 0.70 0.03 0.02 13 -0.14 -0.99 -0.11 -0.09 14 0.07 0.49 -0.04 -0.03 15 -0.07 -0.49 -0.11 -0.09 20 -0.04 -0.28 -0.46 -0.36 30 -0.02 -0.14 -0.42 -0.31 40 o.oo 0.00 -0.80 -0.56 50 -0.20 -1.41 -1.18 -0.79 60 -0.18 -1.27 -1.92 -1.23 __ J

PAGE 157

-~-~ --------150 Table 6-13c Daily Abnormal Returns {AR) and Cumulative Abnormal Returns (CAR) For 121 Days Around the Sale of Convertible Bonds: Multi Firms n = 73 Trading Day AR t CAR t -60 0.41 1.69 0.41 1.69 -50 0.02 0.08 1.88 2.34 -40 -0.31 -1.28 0.99 0.89 -30 o.oo 0.00 0.76 0.56 -20 -0.07 -0.29 2.23 1.44 -15 -0.27 -1.12 1.53 0.93 -14 0.38 1.57 1.91 1.15 -13 o.oo o.oo 1.91 1.14 -12 -0.02 -0.08 1.89 1.12 -11 -0.06 -0.25 1.83 1.07 -10 -0.48 -1.98 1.35 0.78 -9 0.27 1.12 1.62 0.93 -8 0.24 0.99 1.86 1.06 -7 0.03 0.12 1.89 1.06 -6 -0.34 -1.40 1.55 0.86 -5 -0.40 -1.65 1.15 0.63 -4 0.11 0.45 1.26 0.69 -3 -0.05 -0.21 1.21 0.66 -2 -0.19 -0.78 1.02 0.55 -1 -0.76 -3.14 0.26 0.14 0 -0.52 -2.15 -0.26 -0.14 1 -0.16 -0.66 -0.42 -0.22 2 0.00 0.00 -0.42 -0.22 3 -0.12 -0.50 -0.54 -0.28 4 -0.02 -0.08 -0.56 -0.29 5 0.10 0.41 -0.46 -0.23 6 -0.05 -0.21 -0.51 -0.26 7 0.02 0.08 -0.49 -0.25 8 -0.26 -1.07 -0.75 -0.37 9 0.11 0.45 -0.64 -0.32 10 -0.11 -0.45 -0.75 -0.37 11 0.38 1.57 -0.37 -0.18 12 -0.16 -0.66 -0.53 -0.26 13 -0.34 -1.40 -0.87 -0.42 14 0 . 05 0.21 -0.82 -0.39 15 0.43 1.78 -0.39 -0.18 20 -0.28 -1.16 -1.14 -0.52 30 -0.04 -0.17 -2.15 -0.93 40 -0.25 -1.03 -1.82 -0.75 50 0.34 1.40 -2.87 -1.13 60 -0.06 -0.25 -3.46 -1.30

PAGE 158

151 Table 6-13d Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by Relative Size Single Firms Large n = 65 Medium n = 65 Small n = 63 Obs. 2AR t 2AR t 2AR t 1 -0.53 % -1.06 0.34 % 1.06 -0.12 % -0.33 5 0.07 0.14 0.03 0.09 0.33 0.92 10 -0.62 -1.24 -0.46 -1.44 -0.92 -2.56 ** 15 -0.32 -0.64 0.01 0.03 0.51 1.42 16 0.26 0.52 -0.11 -0.34 0.08 0.22 17 -1.42 -2.84 *** a.so -1.56 -0.09 -0.25 18 0.66 1.32 -0.41 -1.28 -0.11 -0.31 19 0.30 0.60 -0.26 -0.81 0.32 0.89 20 -0.18 -0.36 -0.32 -1.00 -0.39 -1.08 21 -0.02 -0.04 0.41 1.28 0.64 1.78 * 22 0.62 1.24 -0.26 -0.81 0.47 1.31 23 -0.34 -0.68 0.24 0.75 0.10 0.28 24 -0.45 -0.90 -0.09 -0.28 0.09 0.25 25 -0.08 -0.16 -0.22 -0.69 -0.12 -0.33 26 -0.79 -1.58 -0.33 -1.03 -0.15 -0.42 27 0.32 0.64 -0.34 -1.06 0.37 1.03 28 0.08 0.16 -0.26 -0.81 0.59 1.64 29 -0.68 -1.36 -0.09 -0.28 -0.20 -0.56 30 a -0.34 -0.68 -0.34 -1.06 -0.25 -0.69 31 -1.28 -2.56 ** 0.20 0.63 0.59 1.64 32 0.11 0.22 0.11 0.34 -0.15 -0.42 33 0.46 0.92 -0.46 -1.44 0.05 0.14 34 0.07 0.14 0.47 1.47 -0.14 -0.39 35 0.36 0.72 -0.24 -0.75 -0.13 -0.36 36 0.58 1.16 -0.07 -0.19 0.02 0.06 37 -0.57 -1.14 0.24 O. 75 -0.03 -0.75 38 -0.49 -0.98 0.01 0.03 -0.14 -0.39 39 -0.51 -1.02 -0.52 -1.63 0.26 0.72 40 0.10 0.20 -0.35 -1.09 0.17 0.47 41 -0.17 -0.34 -0.38 -1.19 -0.28 -0.78 42 -0.40 -0.80 -0.03 -0.09 -0.23 -0.64 43 0.40 0.80 0.14 0.44 -0.35 -0.97 44 0.86 1.76 * -0.12 -0.38 0.35 0.97 45 0.61 1.22 0.17 0.53 0.38 1.06 50 -0.12 -0.24 -0.08 -0.25 0.42 1.17 55 0.60 1.20 -0.20 -0.63 -0.92 -2.56 ** 60 -0.16 -0.32 -0.46 -1.44 0. 26 0.72 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 159

152 Table 6-13e Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by Relative Size Multi Firms Large n = 23 Medium n = 23 Small n = 22 Obs. 2AR t 2AR t 2AR t 1 0.20 % 0.29 0.03 % 0.05 1.21 % 1.92 * 5 0.77 1.12 0.41 0.75 -1.15 -1.83 10 -0.69 -1.00 -0.36 -0.65 -1.32 -2.09 ** 15 -0.50 -0.72 -0.03 -0.05 0.25 0.40 16 -0.34 -0.49 1.10 2.00 ** 0.78 1.24 17 -0.22 -0.32 0.79 1.44 0.60 0.95 18 -0.15 -0.22 0.23 0.42 0.11 0.17 19 0.51 0.74 0.15 0.27 0.58 0.92 20 0.48 0.70 -0.81 -1.47 0.63 1.00 21 0.24 0.35 0.81 1.47 -0.42 -0.67 22 -1.32 -1.91 * 0.33 0.60 -0.37 -0.59 23 0.16 0.23 1.16 2.11 ** -0.52 -0.83 24 0.25 0.36 -0.85 -1.55 0.26 0.41 25 -0.78 -1.13 -0.74 -1.35 -0.21 -0.33 26 1.16 1.68 0.11 0.20 0.36 0.57 27 -1.88 -2.72 *** -0.13 -0.24 0.01 0.02 28 -0.80 -0.12 0.21 0.38 -0.65 -1.03 29 0.66 0.96 -0.58 -1.05 -0.49 -0.78 30 a -1.28 -1.86 * -0.96 -1.75 * -1.19 -1.89 * 31 -0.84 -1.22 0.41 0.75 0.08 0.13 32 -0.69 -1.00 1.34 2.43 ** -0.37 -0.59 33 -0.64 -0.93 0.27 0.49 0.66 1.05 34 -0.31 -0.45 -0.87 -1.58 0.01 0.02 35 0.05 0.07 0.37 0.67 0.13 0.21 36 -0.09 -0.13 0.43 0.78 -0.16 -0.25 37 -0.08 -0.12 -0.86 -1.56 -0.07 -0.11 38 1.53 2.22 ** -0.48 -0.87 0.29 0.46 39 -0.59 -0.86 0.05 0.09 -0.79 -1.25 40 -1.41 -2.04 ** -0.02 -0.04 -0.45 -0.71 41 0.14 0.20 -0.27 -0.49 0.26 0.41 42 -0.51 -0.74 -0.19 -0.35 -0.90 -1.43 43 -0.11 -0.16 -0.36 -0.65 0.58 0.92 44 -2.07 -3.00 *** -0.21 -0.38 -0.39 -0.62 45 0.69 1.00 -0.07 -0.13 -0.26 -0.41 50 -0.27 -0.39 0.08 0.15 0.80 1.27 55 0.27 0.39 0.57 1.04 0.44 0.70 60 -0.33 -0.48 0.52 0.95 -0.47 -0.75 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 160

153 Table 6-13f Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by Firm Size Single Firms Large n = 65 Medium n = 65 Small n = 65 Obs. 2AR t 2AR t 2AR t 1 -0.07 % -0.22 0.19 % 0.51 -0.50 % -1.04 5 -0.04 -0.13 0.46 1.24 -0.05 -0.10 10 -0.90 -2.81 *** -0.61 -1.65 * -0.36 -0.75 15 0.15 0.47 -0.46 -1.24 0.50 1.08 16 0.27 0.84 0.13 0.35 -0.24 -0.50 17 0.24 0.75 -0.64 -1.73 * -1.64 -3.41 *** 18 -0.19 -0.59 -0.07 -0.19 0.39 0.81 19 -0.36 -1.13 -0.21 -0.57 0.93 1.94 * 20 -0.25 -0.78 -0.30 -0.81 -0.26 -0.54 21 0.48 1.50 0.25 0.68 0.25 0.52 22 0.38 1.19 -0.33 -0.89 0.71 1.48 23 -0.13 -0.41 0.66 1.78 * -0.50 -1.04 24 0.13 0.41 -0.77 -2.08 ** 0.15 0.31 25 -0.68 -2.13 ** 0.63 1.70 * -0.33 -0.69 26 0.12 0.38 -0.55 -1.48 -0.79 -1.65 * 27 0.26 0.81 -0.06 -0.16 0.09 0.19 28 0.39 1.22 -0.14 -0.38 0.15 0. 31 29 -0.06 -0.19 -0.39 -1.05 -0.53 -1.10 30 a -0.52 -l.63 -0.51 -1.38 0.15 0. 31 31 0.24 0.75 0.03 0.08 -0.76 -1.58 32 0.32 1.00 -0.59 -1.59 0.19 0.40 33 0.08 0.25 -0.06 -0.16 0.00 0.17 34 -0.08 -0.25 -0.05 -0.14 0.58 1.21 35 -0.54 -1.69 * 0.18 0.49 0.37 0.77 36 0.25 0.78 -0.27 -0.73 0.52 1.08 37 -0.08 -0.25 -0.41 -1.11 -0.04 -0.08 38 -0.22 -0.69 -0.03 -0.08 -0.28 -0.58 39 -0.14 -0.44 -0.30 -0.81 -0.37 -0.77 40 0.08 0.25 -0.04 -0.11 -0.13 -0.27 41 -0.28 -0.88 -0.23 -0.62 -0.33 -0.69 42 -0.19 -0.59 -0.25 -0.68 -0.21 -0.44 43 -0.12 -0.38 0.01 0.03 0.22 0.46 44 0.48 1.50 0.06 0.16 0.51 1.06 45 0.28 0.88 0.29 0.78 0.61 1.27 50 -o. 36 -1.13 a.as 0.14 0.50 1.04 55 -0.71 -2.22 ** -0.11 -0.30 0.31 0.65 60 0.15 0.47 -0.54 -1.46 o.oo 0.00 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 161

154 Table 6-13g Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by Firm Size Multi Firms Large n = 23 Medium n = 23 Small n = 23 Obs. 2AR t 2AR t 2AR t 1 0.89 % 1.54 0.05 % 0.08 0.53 % 0.75 5 -0.25 -0.43 -0.24 -0.37 0.52 0.73 10 -1.08 -1.86 * -0.21 -0.32 -1.05 -1.48 15 -0.17 -0.29 -0.68 -1.05 0.57 0.80 16 1.09 1.88 0.59 0.91 -0.14 -0.20 17 1.23 2.12 ** 0.76 1.17 -0.95 -1.34 18 -0.17 -0.29 0.11 0.17 0.26 0.36 19 0.96 1.65 0.48 0.74 -0.29 -0.41 20 -0.54 -0.93 0.74 1.14 0.04 0.06 21 0.18 0.31 -0.46 -0.71 0.86 1.21 22 -0.50 -0.86 0.07 0.11 -1.04 -1.46 23 -1.12 -1.93 * 0.20 0. 31 1.54 2.17 ** 24 0.17 0.29 -1.32 -2.03 ** 0.93 1.31 25 -0.68 -1.17 0.22 0.34 -1.25 -1.76 26 0.38 0.66 0.63 0.97 0.52 0.73 27 -0.14 -0.24 -0.07 -0.11 -1.81 -2.55 ** 28 0.08 0.14 -0.32 -0.49 -0.23 -0.32 29 -0.56 -0.97 0.44 0.68 -0.38 -0.54 30 a -1.09 -1.88 * -1.29 -1.98 ** -0.92 -1.30 31 0.59 1.02 -0.68 -1.05 -0.32 -0.45 32 0.26 0.45 0.30 0.46 -0.40 -0.56 33 0.15 0.26 -0.67 -1.04 0.78 1.10 34 -1.05 -1.81 * 0.51 -0.78 0.40 0.56 35 0.32 0.55 -0.14 -0.21 0.24 0.34 36 -0.28 -0.48 0.66 1.02 -0.15 -0.21 37 -0.23 -0.40 -0.87 -1.34 o.oo o.oo 38 -0.57 -0.98 1.48 2.27 ** 0.63 0.89 39 -0.02 -0.03 -0.13 -0.20 -1.03 -1.45 40 -0.53 -0.91 -0.44 -0.68 -1.10 -1.55 41 -0.13 -0.22 -0.11 -0.17 0.41 0.58 42 -0.81 -1.40 -1.69 -2.60 *** 0.85 1.20 43 -0.19 -0.33 0.10 0.15 0.10 0.14 44 -0.42 -0.72 -1.01 -1.55 -1.21 -1.70 45 -0.49 -0.85 0.58 0.89 0.19 0.27 50 1.10 1.90 * 0.55 0.85 -0.97 -1.37 55 0.60 1.03 0.51 0.78 0.20 0.28 60 0.25 0.43 -0.61 -0.94 0.12 0.17 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 162

155 Table 6-13h Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by CP Ratio Single Firms Large n = 62 Medium n = 63 Small n = 62 Obs. 2AR t 2AR t 2AR t 1 -0.01 % -0.03 -0.02 % -0.05 -0.47 % -1.09 5 0.02 0.06 0.35 0.85 -0.32 -0.74 10 -0.48 -1.37 -0.37 -0.90 -0.46 -1.07 15 -0.21 -0.60 0.23 0.56 -0.15 -0.35 16 0.28 0.80 0.09 0.22 0.27 0.63 17 0.06 0.17 -0.74 -1.80 * -1.07 -2.49 ** 18 -0.15 -0.43 -0.51 -1.24 0.50 1.16 19 -0.25 -0.71 0.20 0.49 -0.33 -0.77 20 -0.26 -0.74 -0.23 -0.56 0.13 0.30 21 0.42 1.20 0.05 0.12 0.01 0.02 22 0.21 0.60 0.31 0.76 -0.07 -0.16 23 -0.14 -0.40 0.24 0.59 0.12 0.28 24 -0.21 -0.60 -0.24 -0.59 0.04 0.09 25 0.21 0.60 -0.65 -1.59 -0.23 -0.53 26 -0.11 -0.31 -0.38 -0.93 -o. 64 -1.48 27 0.37 1.06 -0.26 -0.63 0.27 0.63 28 0.19 0.54 0.36 0.88 -0.58 -1.35 29 -0.65 -1.85 * 0.19 0.46 -0.38 -0.88 30 a 0.30 0.86 -0.44 -1.07 -1.03 -2.39 ** 31 0.49 1.40 -0.25 -0.61 -0.70 -1.63 32 0.32 0.91 -0.22 -0.54 -0.43 -1.00 33 -0.25 -0.71 0.37 0.90 -0.20 -0.47 34 0.56 1.60 -0.45 -1.10 0.49 1.14 35 -0.09 -0.26 -0.34 -0.83 0.49 1.14 36 0.15 0.43 -0.10 -0.24 0.17 0.40 37 -0.05 -0.14 -0.55 -1.34 0.32 0.74 38 0.11 0.31 -0.27 -0.66 -0.37 -0.86 39 -0.01 -0.03 -0.28 -0.68 -0.38 -0.88 40 0.17 0.49 0.24 0.59 -0.63 -1.46 41 -0.24 -0.69 -0.16 -0.39 -0.52 -1.21 42 -0.48 -1.37 0.22 0.54 -0.46 -1.07 43 0.22 0.63 0.60 1.46 -0.66 -1.53 44 0.22 0.63 0.34 0.83 0.34 0.79 45 0.12 0.34 0.38 0.93 0.28 0.65 50 0.40 1.14 0.30 0.73 -0.45 -1.05 55 -0.32 -0.91 -0.74 -1.80 * 0.16 0.37 60 0.64 1.83 * -0.63 -1.54 -0.81 -1.88 * *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before. _j

PAGE 163

156 Table 6-13i Two-Day Abnormal Returns (2AR) For 121 Days Around the Sale of Convertible Bond Issues: Three Strata by CP Ratio Multi Firms Large n = 23 Medium n = 23 Small n = 23 Obs. 2AR t 2AR t 2AR t 1 0.30 % 0.44 0.33 % 0.62 0.91 % 1.71 * 5 -0.07 -0.10 -0.40 -0.75 -0.15 -0.28 10 -0.83 -1.22 -1.39 -2.62 ** 0.00 0.00 15 -0.43 -0.63 0.06 0.11 -0.49 -0.92 16 -0.42 -0.62 1.49 2.81 *** 0.28 0.53 17 1.02 1.50 -0.14 -0.26 o.oo o.oo 18 0.22 0.32 -0.29 -0.55 0.55 1.04 19 -0.01 -0.10 0.38 0.72 0.27 0.51 20 0.74 1.09 -0.48 -0.91 -0.13 -0.24 21 0.40 0.59 0.53 1.00 -0.34 -0.64 22 0.76 1.12 -0.97 -1.83 * -1.08 -2.03 ** 23 0.30 0.44 0.83 1.57 -0.56 -1.06 24 -0.19 -0.28 0.17 0.32 0.25 0.47 25 -0.16 -0.23 -0.87 -1.64 -0.83 -1.57 26 0.05 0.07 0.77 1.45 0.58 1.09 27 -0.89 -1.31 -0.61 -1.15 -0.52 -0.98 28 -0.15 -0.22 -0.29 -0.55 -0.40 -0.75 29 -0.37 -0.54 0.35 0.66 -0.30 -0.57 30 a o.oo o.oo -1.16 -2.19 ** -2.43 -4.58 *** 31 -0.49 -0.72 -0.04 -0.76 0.19 0.36 32 0.28 0.41 0.05 0.09 -0.54 -1.02 33 -0.84 -1.24 0.35 0.66 0.35 0.66 34 -0.70 -1.03 -0.65 -1.23 -0.11 -0.21 35 -0.82 -1.21 0.49 0.93 0.88 1.66 36 0.68 1.00 -0.14 -0.26 -0.10 -0.19 37 -0.98 -1.44 1.09 2.06 ** -0.68 -1.28 38 1.19 1.75 * -0.10 -0.19 0.53 1.00 39 -1.29 -1.89 * 0.50 0.94 -0.15 -0.28 40 -0.42 -0.62 -0.15 -0.28 -1.67 -3.15 *** 41 0.36 0.53 -0.01 -0.02 -0.20 -0.38 42 -0.67 -1.00 -0.48 -0.91 -0.34 -0.64 43 0.38 0.56 -0.53 -1.00 0.04 0.08 44 -2.54 -3.73 *** 0.28 0.53 -0.17 -0.32 45 -0.39 -0.57 0.40 0.75 0.57 1.08 50 -0.41 -0.60 0.77 1.45 -0.47 -0.89 55 0.74 1.09 0.27 0.51 0.46 0.87 60 -0.70 -1.03 0.48 0.91 0.01 0.02 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level a Two-day excess return consisting of the issue date and the day before.

PAGE 164

157 Table 6-13j Summary Results for Issue Date Tests Single and Multi Firms Single Firms 225 -0.39 -1.77 * 133 Multi Firms 73 -1.28 -3.56 *** 50 CP Ratio Single Firms Small 62 -1.03 -2.39 ** 44 Medium 63 -0.44 -1.07 40 Large 62 0.30 0.86 29 Multi Firms Small 23 -2.43 -4.58 *** 21 Medium 23 -1.16 -2.19 ** 16 Large 23 o.oo o.oo 10 Relative Size Single Firms Small 63 -0.25 -0.69 35 Medium 65 -0.34 -1.06 39 Large 65 -0.34 -0.68 41 Multi Firms Small 22 -1.19 -1.89 * 14 Medium 23 -0.96 -1.75 * 15 Large 23 -1.28 -1.86 * 17 Firm Size Single Firms Small 65 0.15 0.31 37 Medium 65 -0.51 -1.38 38 Large 65 -0.52 -1.63 40 Multi Firms Small 23 -0.92 -1.30 16 Medium 23 -1.29 -1.98 * 16 Large 23 -1.09 -1.88 * 14 *** Significant at the 1% level ** Significant at the 5% level * Significant at the 10% level 59.11% 68.49% 70.97% 63.49% 46.77% 91.30% 69.57% 43.48% 55.56% 60.00% 63.08% 63.64% 65.22% 73.91% 56.92% 58.46% 61.54% 69.57% 69.57% 60.87%

PAGE 165

158 Table 6-15 Summary Statistics Issue Date Sample Divided into 3 Strata by CP Ratio HIGH CP MID CP LOW CP ------------------------CP RATIO (N = 84) (N = 86) (N = 87) --------AVG 1.26 1.15 1.09 STD DEV 0.12 0.01 0.03 MIN 1.18 1.13 1.00 MAX 1.91 1.18 1.13 FIRM SIZE --------AVG 541.74 609.43 505.61 STD DEV 624.99 907.18 1026.94 MIN 22.78 16.77 6.61 MAX 3305.99 4726.38 6785.98 ISSUE SIZE --------AVG 59.43 50.72 45.79 STD DEV 48.06 42.39 41.19 MIN 5.00 7.50 5.50 MAX 250.00 200.00 200.00 REL SIZE --------AVG 0.17 0.20 0.21 STD DEV 0.13 0.39 0.17 MIN 0.03 0.04 0.02 MAX 0.74 3.55 1.06 CONV RATIO --------AVG 28.25 26.80 30.21 STD DEV 17.33 15.27 20.11 MIN 8.00 11.11 6.58 MAX 125.00 111.11 142.86 AD -> ID --------AVG 20.97 28.03 32.27 STD DEV 18.47 18.31 19.62 MIN o.oo 1.00 1.00 MAX 123.00 98.00 101.00

PAGE 166

159 Table 6-15 continued HIGH CP MID CP LOW CP ------------------------ISSUE DATES (N = 84) (N = 86) (N = 87) ----------1963-1966 10 10 6 ( 2 6) 1967-1970 21 41 56 (118) 1971-1974 5 9 13 (27) 1975-1978 0 6 2 (8) 1979-1982 18 7 7 (32) 1983-1986 30 . 13 3 ( 46) ------------------------84 86 87 (257) RATINGS ------HIGH A 0 0 0 (0) MID A 4 2 0 (6) LOW A 6 8 4 (18) HIGH B 17 18 14 (49) MID B 44 40 46 (130) LOW B 13 18 23 (54) HIGH C 0 0 0 ( 0) MID C 0 0 0 ( 0) LOW C 0 0 0 ( 0) ------------------------84 86 87 (257) CP RATIO is the ratio of conversion price at issue to the firm's stock price two days prior to the issue date. FIRM SIZE is the total market value of the issuing firm's equity in millions of dollars. ISSUE SIZE is the size of the issue in millions of dollars. REL SIZE is the ratio of ISSUE SIZE to FIRM SIZE. CONV RATIO is the original number of shares into which an indi vidual bond is convertible. AD-> ID is the number of calendar days from announcement date to issue date.

PAGE 167

CHAPTER SEVEN SUMMARY, CONCLUSIONS, AND FUTURE RESEARCH Summary and Conclusions Announcing Convertible Bond Issues Announcing a proposed issue of convertible bonds has been shown to be an economically as well as a statistically significant event. The average firm experienced a negative equity price impact of -1.43% when news of the issue is released to the public. Even though 73% of the firms experienced negative announcement day price reactions, the impacts ranged from -10.55% to+ 10.72%. To help determine the cause of this phenomenon, various characteristics of the convertible bonds and the issuing firms were studied. The average firm experiences highly significant cumula tive abnormal returns over the 60 days preceding the public announcement indicating that management might be timing the announcement to coincide with otherwise positive price per formance. The announcement might be signalling overpriced shares as Myers and Majluf (1984) would suggest, since this performance disappears almost immediately. 160

PAGE 168

161 To determine whether announcement date impacts are due to the release of information or simply some "flat charge", the cross-sectional variance of abnormal returns at the announcement date was compared to that during a nonwoven period. impacts An increase in variance would indicate differing according to the informational content of the announcement. A z-test could not reject the null specific hypothesis that there is no increase in cross-sectional variance, however. This seems to indicate that firms are simply charged a fee for announcing an issue of convertible bonds. Additional tests were performed to further test this point. The size of the firm and the issue size were both found insignificant in explaining the announcement date impact, but a possible relationship was detected with the issue CP ratio. This ratio, conversion price divided by current stock price, measures how far out of the money the firm's call option is. That is, how soon the firm can call the issue and force investors to convert to equity. Even though no statistical significance is shown, it appears that firms that set the lowest CP ratios experience the greatest (most negative) equity impacts. But, the con version terms are not released until the issue date. There fore, either investors are estimating the terms at the

PAGE 169

162 announcement date, or there is another variable related to the CP ratio that is available when the issue is announced. The firms were then divided into two groups according to whether they had outstanding convertible bond issues when the bond in the sample was announced, multi firms, or not, single firms. Although no difference in announcement date equity impacts was found for single versus multi firms, the performance preceding the announcement is quite different for the two classes of firms. Single firms experience highly the previous multi firms disappears by significant cumulative abnormal returns over 60 days, while the abnormal performance for is only marginally significant and completely day -5. This could be interpreted as evidence of different motives for issuing convertible bonds. Further tests showed issue size to be a potential factor in announcement date equity impacts. Multi firms selling the relatively largest issues experienced the great est announcement date impact. Also, it appears that announcement date equity impacts for multi firms are pos itively related to firm size. These conclusions must be considered preliminary, however, due to small sample sizes. In conclusion, the tests have shown that announcing an issue of convertible bonds is an important event and there is some evidence that management tries to time the

PAGE 170

163 announcement to take advantage of abnormal pricing. Fac tors such as firm and issue size have been shown, general ly, not to affect announcement date impacts. However, when the firms are divided into single and multi, there appears to be a relationship between firm and issue size for multi firms. Also, even though no statistical relationship is deter mined, there would appear to be a relationship between CP ratio and announcement date equity impacts. This means either investors attempt to estimate issue terms before they are of another date. released, or the CP ratio is masking the effects variable which is available at the announcement Selling Convertible Bond Issues The average firm that sells a previously announced con vertible bond experiences an issue date equity price impact of -0.61%. Unlike the announcement date event, however, there is no indication of abnormal performance either before or after the issue date. It should be noted that only issues that were ultimately sold were included in this study. It is possible that firms which withdraw announced issues have different performance patterns preceding the cancellation announcement. This is a matter for future research, however, as it was not pursued in this study.

PAGE 171

164 Again, there was no evidence that the cross-sectional variance of abnormal returns increased at the issue date. This would indicate that firms are also charged a flat fee for selling a previously announced issue of convertible bonds. To further test this point, however, various firm and issue characteristics were again tested. When the firms were stratified according to CP ratio, a significant relationship was observed. Firms which set the lowest CP ratios experienced the greatest (most negative) equity impacts. The issue date impact for firms which set the highest ratios was slightly positive but insignifi cant. This is additional evidence that conversion terms are important factors in convertible bond equity impacts. No relationship was detected between issue size and the issue date impact, but there is evidence that firm size might be a factor. Again, even though it appears that the impact for largest firms was the greatest, no statistical significance is found. Once again, the firms were divided into single and multi. This time, the two impacts were significantly different. Multi firms experienced a significant issue date impact of -1.28%. Single firms, on the other hand, only experienced an average impact of -0.39% which was statist ically insignificant. A T-test revealed statistical

PAGE 172

165 significance between the two, but cumulative abnormal returns preceding the issue date showed no remarkable difference. There are several contradicting hypotheses which might explain this effect. However, they will not be explored at this time. Again, it appears that the largest multi firms and multi firms selling the relative largest issues experience the greatest impacts, but, again, there is no statistical significance to support this finding. Additional Tests In both the announcement date and issue date tests, a relationship was observed between CP ratio and equity price impacts. In both cases, the firms which set the lowest ratios experienced the greatest impacts. It was assumed that the CP ratio captured the relative equity component of the issue, and the lower ratios indicated greater equity components. To test whether the CP ratio was capturing the equity component, the first one month returns for the con vertible bonds were regressed against the return for the underlying stock and long term government bonds after divid ing the firms into three strata by CP ratio. If the convertible has a greater equity component, it should be more sensitive to price movements in the underlying stock.

PAGE 173

166 The results were almost identical for the three regres sions, so it would appear that the CP ratio is actually masking some other variable. To test whether the CP ratio is actually masking another variable, the firms were once again divided into three strata by CP ratio. Since a statistical relationship between CP ratio and equity impact was only found at the issue date, the following issue date figures were found: (1) Average CP ratio ( 2) Average firm size ( 3) Average issue size (4) Average relative issue size (5) Average conversion ratio (6) Average number of days between announcement and issue dates (7) Frequency of issue per year (8) Frequency of quality ratings As would be expected, the average CP ratios were signif icantly different among the strata. There is no difference in average firm size, but a significant difference is observed across the strata for average issue size and average relative issue size. It appears that small issues and those with the greatest relative size cause the greatest issue date equity impact since the low CP strata has a smaller average issue size and larger average relative issue size than the other two strata. Relative issue size, however, has been ruled out as a factor by previous tests. If issue size, as contrasted with relative

PAGE 174

167 issue size, is a factor, it would tend to eliminate firm characteristics as factors in issue date equity impacts. It would further tend to indicate that issue date impacts are solely a function of transactions costs. That is, smaller issues cost proportionally more to sell than larger issues. The only other variable that proved significantly different across the three strata was AD-> ID, the number of days between announcing and selling the issue. The greatest equity impact was experienced by firms in the low CP strata, and this strata also waited the longest to sell the announced issue of convertibles. No explanation is offered at this time for this finding. It will be left for future research to determine why waiting to sell an announced issue of convertible bonds will subject the issuing firm to greater issue date equity impacts. Future Research Often in the course of attempting to answer an empiri cal question, many other, unanswered, questions surface. This research has determined that there is indeed a vari able or set of variables that determines the amount of equity impact experienced by a firm that announces and then sells a convertible bond issue.

PAGE 175

168 For instance, it has been determined that multi firms experience greater issue date equity price impacts than single firms. It must be determined exactly what classi fies a firm as "multi." That is, are the reactions different if the announces be less according to the number of issues. For example, firm has one other issue outstanding when it an issue of convertibles, will its equity impact than if it had two or three other issues outstanding. Also, it must be determined if announcing a convertible bond issue has any impact on the market value of the firm's other convertible bonds. In short, the difference in equity impacts between single and multi firms deserves further study. The CP ratio, conversion price divided by current stock has been shown highly related to issue date equity price, impacts. tion that A regression attempting to validate the assump the CP ratio is capturing the relative equity valuation of the convertible showed nothing conclusive. It appears that the CP ratio by itself does not capture the degree equity. to which the market values the convertible as It apparently is masking the effects of some other variable(s). Two possible candidates for these other variables are issue size and days between announcement date and issue date. This is a point that deserves further testing.

PAGE 176

169 In the course of this study, various characteristics were assumed constant across the firms in the samples. One of these variables is firm growth. It is possible that the regressions testing for a relationship between CP ratio and equity valuation showed nothing remarkable because it is actually an interaction between CP ratio and firm growth rate that affects equity price impacts. Future research much allow for this interactive term. Another characteristic held constant was that every firm in the sample sold the issue it had previously announced. No issues that were announced and then subsequently can be withdrawn were formed as to included. Before any conclusions the causes of announcement or issue date equity reactions, a representative sample of withdrawn issues must be included in the tested sample. Further empirical research is warranted. There is some variable or group of variables that determines the impact on the equity value of firms that sell convertible bonds. It must be the goal of future research to determine what those variables are.

PAGE 177

REFERENCES Asquith, P. and D. Offerings Dilution." 1986, pp. 61-89. W. Mullins. "Equity Issues and Journal of Financial Economics, 15, Barclay, M. J. and R. H. Effects of New Equity Issues Data." Working Paper, The Pennsylvania. Litzenberger. "Announcement and the Use of Intraday Price Wharton School, University of Billingsly, R., Lamy, R., and G. Thompson. "Valuation of Primary Issue Convertible Bonds." Journal of Financial Research, Vol. IX, No. 3, Fall, 1986, pp. 251-259. Dann, L. Y. and w. H. Mikkelson. "Convertible Debt Issuance, Capital Structure Change and Financing-Related Information." Journal of Financial Economics, 13, 1984, pp. 157-186. Eckbo, B. Offerings." 119-151. E. "Valuation Effects of Corporate Debt Journal of Financial Economics, 1986, 15, pp. Ingersoll, J. E. "A Convertible Securities." 4, 1977, pp. 289-322. Contingent-Claims Valuation of Journal of Financial Economics, Marsh, P. Empirical 121-134. "The Study." Choice Between Debt and Equity: Journal of Finance, 32, 1982, Masulis, Analysis Finance, R. W. "Stock Repurchase by Tender Offer: of Common Stock Price Changes." Journal 35, 1980, pp. 305-319. An pp. An of Masulis, R. W. "The Impact of Capital Structure Change on Firm Value: Some Estimates." Journal of Finance, 38, 1983, pp. 107-126. Mikkelson, W. "Convertible Calls and Security Returns." Journal of Financial Economics, Vol. 9, 1981, pp. 237-264. 170

PAGE 178

171 Mikkelson, W. "Convertible Calls and Stock Price Declines." Financial Analysts Journal, January-February, 1985, pp. 63-69. Mikkelson, W. and M. Partch. "Valuation effects of Security Offerings and the Issuance Process." Journal of Financial Economics, 15, 1986, pp. 31-61. Miller, M. and K. Rock. "Dividend Policy Under Asymmetric Information." Journal of Finance, 40, 1985, 1031-1051. Mishra, B. and M. P. Callable Convertibles?" Florida, December, 1986. Narayanan. Working "Why Paper, Do Firms Issue University of Modigliani, F. and M. Miller. "The Cost of Capital, Corporation Finance and the Theory of Investments." American Economic Review, 48, 1963, pp. 261-297. Myers, s. c. and N. s. Majluf. "Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have." Journal of Financial Economics, 13, 1984, pp. 187-221. Rogers, R. c. and J.E. Owers. "Equity for Debt Exchanges and Stockholder Wealth." Financial Management, Vol. 14, No. 3, Autumn, 1985, pp. 18-26. Smith, C. W., Jr. Acquisition Process." 1986, pp. 3-29. "Investment Banking and the Capital Journal of Financial Economics, 15, Vermaealen, Signalling: T. "Common Stock Repurchases and Market Economics, 9, An Empirical Study." Journal of Financial 1981, pp. 139-183. Vu, J. D. "An Empirical Investigation of Calls of Non-Convertible Bonds.'' Journal of Financial Economics, 16, 1986, 235-265. Warner, J., Prediction, Management Rochester. R. Watts, and K. Wruck. "Stock Prices, Event and Event Studies: An Examination of Top Changes." Working Paper. University of May, 1987.

PAGE 179

BIOGRAPHICAL SKETCH Bruce Kuhlman, the son of Anita and Henry Kuhlman, was born in st. Petersburg, Florida, on September 14, 1949. He earned a from the University Doctor of in 1988. bachelor's degree in Food and Resource Economics University of Florida in 1980 and an MBA from the of Florida in 1982. He expects to receive a Philosophy degree from the University of Florida Bruce is now an Assistant Professor of Finance at the University of Toledo in Toledo, Ohio. 172

PAGE 180

I certify that opinion it conforms presentation and is as a dissertation for I certify that opinion it conforms presentation and is as a dissertation for I certify that opinion it conforms presentation and is as a dissertation for I have read this study and that in my to acceptable standards of scholarly fully adequate, in scope and quality, the degree of Doctor of Philosophy. Robert c. Radcliffe, rman Associate Professor of Finance, Insurance and Real Estate I have read this study and that in my to acceptable standards of scholarly fully adequate, in scope and quality, the degree of Doctor of Philosophy. David T. Brown Assistant Professor of Finance, Insurance and Real Estate I have read this study and that in my to acceptable standards of scholarly fully adequate, in scope and quality, the degree of Doctor of P 'losophy. Jams . Mcclave Asso a e Professor of Deci n and Information Sciences This dissertation was submitted to the Graduate Faculty of the Department of Finance, Insurance and Real Estate in the College of Business Administration and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December, 1988 Dean, Graduate School

PAGE 181

UNIVERSITY O F FL O RIDA \\ \ \\ \\\\\\ \\\ \\\ \\\\\ 1\\1\ \\ \\\\\I \\I\\\ \\\1\1 \\ \\\I \1\1\\ \II\\ . 3 1262 08553 4781