Do Nonrefunding Provisions Constrain Corporate Behavior? Evidence from Calls.
I examine corporate call policy for nonrefundable debt. Nonrefundable bonds purportedly cannot be called with funds from lower-interest-cost debt. However, firms tend to call nonrefundable debt following interest rate decreases. The calling firms typically issue lower-interest-cost debt around the time of the call and do not identify the source of funds used to call the debt. These actions show the weakness of the constraints imposed by the nonrefunding provision on firms' call policies and may have led to the abandonment of the provision in the early 1990s and the adoption of potentially superior contracting mechanisms.
During the high-interest-rate periods of the 1970s and 1980s, many corporations issued debt with a form of call protection known as a nonrefunding provision. The offering circulars and bond indentures of nonrefundable debt issues usually specified that the bonds could not be called for early redemption if the source of call funds was the proceeds of borrowing at a similar or lower interest cost.
Although the nonrefunding provision appeared to protect investors against interest rate declines, the extent of that protection is not clear. For example, between February 1992 and February 1993, Kroger Company called $501 million of outstanding 12.785% and 13.125% nonrefundable debt issues, repurchased over $400 million of these issues in open-market transactions, and issued $1.6 billion of lower-interest-cost debt. Kroger announced that one of the purposes of the lower interest cost issues was "to refinance the company's higher cost debt" (Wall Street Journal, 1992). Kroger also issued $212.8 million of new equity following the calls of the nonrefundable issues. Although Kroger might not have violated the specifics of the nonrefunding provision, investors who purchased the nonrefundable debt to protect themselves against calls in periods of declining interest rates might have been disappointed with Kroger's actions.
Several court cases have addressed the legality of calls of nonrefundable debt by firms that are issuing lower-cost-debt. In 1978, the Franklin Life Insurance Company sued Commonwealth Edison Company for redeeming an issue of 9.44% nonrefundable preferred shares while at the same time borrowing funds at a lower rate. Commonwealth Edison identified the source of funds for the redemption as an equity issue. The court found in favor of Commonwealth Edison, because the nonrefunding provision requires an examination of only the source of the funds actually used to achieve redemption.
In 1983, Morgan Stanley & Company sued the Archer Daniels Midland Company (ADM) for calling a nonrefundable 16% issue while twice borrowing money at a lower interest rate. ADM identified two equity offerings as the source of the funds for the redemption. Morgan Stanley, a large holder of the nonrefundable issue, indicated that for ADM to identify the equity issues as the source was juggling funds. Morgan Stanley saw this action as aimed at circumventing the provision, since ADM had issued two lower-cost debt issues during the time between the issue of the nonrefundable debt and the call announcement. The court upheld the redemption, based on the source test precedent of the Franklin Life case.
The source test precedent seems to hold that a call of nonrefundable debt by a firm issuing lower-interest-cost debt is legal as long as the specific funds from the new issue are not used for the call. If a firm has sufficient cash from sources other than new debt, then the firm can sell a new issue and effectively achieve a refunding and still pass the scrutiny of the source test.
In another variation on redeeming nonrefundable debt, during the late 1980s and early 1990s, a number of firms used STACs (Simultaneous Tender and Call). With a STAC, the firm tenders for the debt and announces that all bonds not tendered will be called at a lower call price. Funds from new, lower-cost debt are used for the tender, and not the call. This approach led to civil litigation in several cases against firms that used STAGs (Dhillon, Noe, and Ramirez, 1998).
In practice, although some firms identify funds from common or preferred share issuance or asset sales for calls of nonrefundable debt, most announcements include very general descriptions of the sources of funds. The Wall Street Journal and press release announcements for the calls include descriptions of the sources of funds such as: "cash on hand," "excess cash currently available to the company," "general corporate funds," or "funds segregated and available for this redemption."
The way in which the nonrefunding provision functions as a financing choice depends on the effectiveness of the provision to constrain firms' ability to redeem bond issues. If the nonrefunding constraints hold, then bondholders should benefit in periods of declining interest rates. If the nonrefunding provision does not constrain at least some firms, then the provision can serve another purpose or persist as a "neutral mutation" that causes no harm. If the provision is a neutral mutation and future conditions show that the provision provides only limited protection, then the provision might disappear from use.  The use of nonrefunding provisions has declined dramatically since the peak in 1986. Only one corporate bond issued since 1994 (in 1996) has included the nonrefunding provision.
In this paper, I examine the practices of firms that call nonrefundable debt and consider the implications of these practices on the choice of the nonrefunding provision in the corporate debt contract. I find that although nonrefundable bonds are called at a lower rate than are bonds that do not have the provision, calls of nonrefundable debt are common and often occur in periods of declining interest rates. Firms that call during the nonrefunding period often do not report the source of funds used to call the bonds. Frequently, these calls occur at the same time that the company is issuing lower-interest-cost debt. Companies also use a variety of strategies to limit the effectiveness of the nonrefunding provision.
My evidence suggests limited protection against calls following interest rate decreases, particularly for investment-grade issues. These events are consistent with the abandonment in the early 1990s of the use of the nonrefunding provision, which identifies actions firms cannot take to call debt, and the adoption of prescriptive covenants, like make-whole and clawback provisions, which identify actions that firms must take to call debt.
In Section I, I discuss the background for this study and describe the data. In Section II, I discuss the reasons firms call debt. I develop variables to measure these incentives and the refunding constraints. In Section III, I develop hypotheses on the nature of the constraints imposed by the nonrefunding provision and describe my method for testing the hypotheses. In Section IV, I test the hypotheses. Section V summarizes and concludes.
I. Background and Data Description
Bond covenants are the terms on which bondholders and the firm contract. Because covenants are designed to restrict issuing firms from engaging in specific actions that might shift wealth from bondholders to stockholders (or other stakeholders) after the bonds are sold, covenants are intended to reduce the costs associated with the bondholder-stockholder conflict. The choice of which covenants are included in a particular debt contract should be those that most efficiently control the conflict.
The call provision is a covenant that is common in one form or another to all debt contracts. In the cases in which it allows the firm to call the debt, the call provision provides a schedule of prices for calling the debt and a description of the terms under which the debt can or cannot be called. 
Bonds with call provisions give the issuing firm the option to call the debt prior to maturity. Callable debt issues usually provide investors with call protection that lasts for several years, followed by a period during which the bonds are freely callable. This call protection can be either absolute, as with standard call protection, or conditional, as with nonrefunding protection. Standard call protection is more constraining than nonrefunding protection. Except for sinking fund calls, a bond with standard call protection cannot be called during the protection period.
In contrast, nonrefundable bonds are freely callable, but nonrefundable during the protection period. Some debt issues contain both call and refunding protection. Such issues are noncallable for a specified period. After that, they are callable but nonrefundable for an additional period. Thatcher (1985) calls these "two-tiered" bonds. Thus, it appears that nonrefundable debt affords investors some degree of protection against calls in response to interest rate declines, but not against calls for other purposes.
In the early 1990s, new types of call provisions with some similarities to nonrefunding provisions began to appear in corporate debt contracts. These provisions might have been developed in part as a substitute for the nonrefunding provision.
"Make-whole" call provisions began to appear in corporate debt contracts in 1991. These provisions provide a floating call price that is based on the current level of a benchmark interest rate. As the benchmark interest rate decreases, the call price on the bond increases. As interest rates fall, investors in called bonds must still reinvest the proceeds of the call at a lower interest rate, but they will have more money to invest at those lower rates.
Mann and Powers (2001) analyze 318 make-whole bonds issued from 1995 to 1999. They argue that the make-whole provision could provide firms with more flexibility than can a standard call or nonrefunding provision, but could still protect investors from reinvestment risk. The make-whole provision provides reinvestment risk protection similar to the intent of the nonrefunding provision, but without the source of funds constraint of the nonrefunding provision.
Mann and Powers (2001) surveyed the CFOs of 25 firms that issued make-whole debt. None of the CFOs identified the make-whole provision as a substitute for the nonrefunding provision. Only three CFOs identified refunding as a potential reason for calling make-whole debt.
"Clawback" provisions began to appear in high-yield issues in 1992. Clawbacks permit issuers to call specified portions of an issue only with the proceeds of a subsequent equity financing. Calls of clawback bonds are limited to specific portions of the issue, usually about 35%, at any one time.
The clawback provision is similar to the nonrefunding provision in that it restricts the source of funds that can be used to call outstanding debt. The callback provision is prescriptive, in that it identifies where the call funds must come from. The nonrefunding provision is restrictive in that it identifies only where the funds cannot come from. 
Table I shows the number of bonds issued by year and type of call protection, 1972 through 2000. Although this paper focuses on industrial debt, Table I provides data for all industrial, utility, and financial bonds issued over the period. The table lists each of the major call provision types for the period except for freely callable debt.
Firms made over 1500 offerings of new corporate nonconvertible industrial debt between 1982 and 1992. Over one third of the issues, representing one third of the dollar volume, included nonrefunding provisions. I examine the decision to call or not call for a comprehensive subset of 206 nonrefundable corporate bonds that were issued during the period 1982 through 1992 and which had information available on Compustat. Through 1996, firms called 84 of the 206 bonds prior to expiration of the nonrefunding provision.
The literature that examines the use of the call provision in bonds also looks at the call provision as a contractual tool. These studies analyze the characteristics of the firms that include the different types of call provisions in debt contracts.  However, although the literature for calls of convertible debt is extensive (see, for example, Asquith, 1995), there are few empirical studies that examine calls of nonconvertible corporate debt in practice. Vu (1986) looks at call policy between 1962 and 1978 for 102 AMEX and NYSE corporate bonds, but he does not distinguish between callable and nonrefundable debt. King and Mauer (2000) look at call policy for 1,642 corporate bonds called during the period 1975 to 1994 and specifically eliminate nonrefundable debt from their sample. My paper might be the first paper to model the call decision by including in the data set all of the bonds that could have been called, rather than just the bonds that were called.
A. Issued and Called Bond Data Sets
I use the Securities Data Company (SDC) corporate new issues file to identify the full set of nonrefundable industrial bonds issued between 1982 and 1992. From this set, I identify which of these bonds were called, whether the bonds were called before or after the expiration of the nonrefunding period, and what portion of the outstanding issue was called if it was not a call for the complete issue.
I also use the Standard & Poor's Semi-Weekly Called Bond Record (CBR), for January 1983 through January 1996, to identify which nonrefundable bonds were called. Where available, I use Lexis and the Dow-Jones News Retrieval service to verify the call announcement date from the CBR and to identify new debt issues or significant financing events by the calling firm that occurred within six months of the call.
Panel C, Table II, gives summary information for the full set of called nonrefundable industrial bonds issued between 1982 and 1992, and the subset of issues that were called by 1996. There are 536 nonrefundable bonds issued by 397 different firms during the period.
Because of potential differences between the incentives for calling nonrefundable debt for firms issuing investment-grade debt and firms issuing speculative-grade debt, I distinguish between the two grades of debt based on Moody's bond ratings at the time of issue. Of the 536 issues, 402 include sinking funds, and all but 20 of the sinking funds begin after expiration of nonrefunding period. I do not include calls for sinking fund purposes in the sample. Prior to expiration of the nonrefunding period, 137 of the issues were called at least once, and 13 were called at least twice. I note that the nonrefunding provision does not limit the amount of the issue that can be called, so firms that make multiple calls of the same issue are constrained for some other reason from calling the full issue. Two of the investment grade issues and two of the speculative grade issues in the final sample were each called at least twice. Forty percent of the investment-grade issues and 11% of the speculative-grade issues were c alled prior to expiration of refunding protection.
For comparison purposes, Panels A and B of Table II show similar issue and call descriptions for bonds issued over the same period. Panel A shows bonds that had no call protection. Panel B shows bonds with absolute call protection that expires. Between 1982 and 1992, there were 31 bonds issued with no call protection and 947 bonds issued with an absolute call protection period. The bonds with no call protection and absolute call protection show a much shorter maturity than the nonrefundable bonds; a larger proportion of these bonds were called prior to 1996.
Panels A and B of Figure I show the proportions of freely callable bonds and nonrefundable bonds that were called by year for investment-grade and speculative-grade issues, respectively. In these figures, freely callable bonds include bonds with no call protection, bonds for which call protection has expired, and bonds for which refunding protection has expired. Nonrefundable bonds include those nonrefundable bonds for which refunding protection has not yet expired. The graphs in Figure I indicate that freely callable bonds are called at a greater rate than are nonrefundable bonds. The Wilcoxon signed rank test indicates that the mean proportion of called bonds is greater for the callable bonds than nonrefundable bonds at the 5% level for both investment and speculative grade issues.
Because interest rates continuously decreased over the 14-year period of the analysis, all of the calls identified in Table II and Figure I occur in periods where the market yield was less than at the time of issue. The average annual yield for Aaa corporate bonds in 1982 was 13.8%. This rate decreased during the period between 1982 and 1996 (except for a slight increase in 1984) to a level of 7.4% in 1996. This period is unique because of the consistent decrease in interest rate levels, and thus serves as a particularly useful period for examining the binding nature of the nonrefunding provision.
My empirical analysis focuses on the industrial corporate sector. I restrict my analysis to the firms that issue nonrefundable debt. Firms must have SIC codes between 2000 and 4899, 4940 and 5999, and be included in the Compustat files for the years 1982 through 1996. My sample excludes gas and eleciric utilities (SICs 4900 to 4939) and financial firms.
Panel D, Table II gives summary information for the subsample of the bonds that are issued by firms in the industrial sector, and which are included on Compustat. Panel D gives information for those nonrefundable issues that were called before the nonrefunding provision expired. The panel also shows firm size and leverage. I define firm Size as the sum of the book value of long-term debt plus the market value of equity. I measure firm Leverage as the ratio of the book value of long-term debt to the size measure for the fiscal year of issue. Investment-grade issues have a significantly longer maturity (mean of 28.5 years compared to 13.4 years) and length of refunding protection period (mean of 9.8 years compared to 5.1 years). Investment-grade issues represent a significantly smaller portion of firm value (mean of 3% compared to 23%) than do the speculative-grade issues. Firms that issue investment grade debt are significantly lower leveraged than are firms that issue speculative-grade debt (mean of 0.18 com pared to 0.33).
I collect Compustat information for the issuing firms for each fiscal year from the year each firm offers bonds through the earliest of the year that the firms either call the issue in full, the issue matures, or 1996. If an observation is for a year in which the bond is not callable, I eliminate the observation. I note that most of the nonrefundable bonds are issued prior to 1989 (only four bonds in the sample were issued after 1989). Absolute call protection for all of the bonds in the sample ends before 1996; therefore, no bond is included in the sample that could not have been called prior to 1996. Most of the nonrefundable bonds issued in the late 1980s did not have any period of absolute call protection, only refunding protection.
If an observation does not have information on Compustat for all of the variables included in the analysis, I eliminate it. I define a "bond year" for each year I can identify a complete set of variables. This process results in 1,545 bond-year observations and 110 calls.
Table III shows the distribution of observations and calls by issue grade and time period. The sample consists of 1,250 bond years and 89 calls for investment-grade issues and 295 bond years and 21 calls for speculative-grade issues. Twenty-five of the calls are made after the expiration of the nonrefunding provision. Because the nonrefunding constraint has expired, these are effectively calls of freely callable debt. I include these calls and the corresponding bond years after refunding protection expires in the data set to see if expiration of the nonrefunding constraint is an important determinant of calls of nonrefundable debt.
Bond-year information from Compustat is from the fiscal year during which I assume the call decision is made. I assume that the call decision is made on the anniversary date of the call for bonds that are called. Also, I assume that the call decision is made on the anniversary date of the issue for bonds that are not called.
In addition to identifying the market practice and firm characteristics for calls of nonrefundable debt, I identify new bond issues offered within six months of the call date for firm's calling nonrefundable debt. Of the 84 calls of nonrefundable debt, 60 of the calling firms issue new debt within six months of the call, and 42 of these issues are for an amount larger than the face value of the nonrefundable issue.
II. Reasons for Calling Debt
The literature offers several reasons for why firms to call debt prior to maturity. Each of the reasons indicates that a different set of variables is important to the call decision. However, if the nonrefunding provision imposes real constraints, then these sets do not completely explain the call decision.
A. Control Variables
I group the reasons for calls into six general classes. Call provisions provide firms with the opportunity to: 1) reduce the net interest cost of outstanding debt, 2) change capital structure, 3) transfer wealth between called securities and equity or other outstanding debt, 4) recontract with debtholders to address agency costs of debt, 5) maximize firm value by using the option characteristics of the call provision, or 6) eliminate debt with high levels of liquidity. 
1. Reduction of Net Interest Costs
Managers may call outstanding debt if doing so reduces net interest costs.  When market interest rates decline or the quality of the firm's debt increases after a bond issue, then the value of the bond will increase. Because the firm's opportunity cost of debt decreases, it is costly for the firm not to replace the bond with a lower-interest-cost bond. This incentive is consistent with the process of "refunding" outstanding debt.
To control for reduction in net interest costs due to market factors, I measure the Change in Interest Rates as the ratio of average annual yield on an index of corporate bonds from the year the bond was issued and the average annual yield on an index of corporate bonds for the year of the call decision. For bonds with a rating of Baa or better, I use an index of Aaa bonds. For bonds with a rating less than Baa, I use an index of Baa bonds. The annual yield from the issue year is my proxy for the coupon rate at which the bond is issued. The annual yield from the year of the call is my proxy for the rate at which the debt could be refinanced. To control for reduction in net interest cost due to firm-specific factors, I measure Change in Bond Rating as the difference in the S&P numerical bond rating (from Compustat) for the issuing firm for the fiscal year of the call minus the rating for the year of issue. An increased numerical rating indicates a decrease in quality. I predict that Change in Interest Rates w ill be lower and Change in Bond Rating will be higher for bond years without a call, rather than for bond years with a call.
2. Change in Capital Structure
Managers may call outstanding debt if doing so results in a value-increasing change in the capital structure of the firm. For example, firms might call outstanding debt to allow for the use of non-debt tax shields, or if the call is associated with leverage-increasing newly issued debt, the change could allow managers to use a higher level of debt-related tax shields or to signal higher firm quality (Ross, 1977). I measure the Change in Leverage as the fractional decrease from the previous year of the book value of long-term debt as a proportion of firm size for the fiscal year of the call decision. I measure the Change in Income Tax as the fractional decrease from the previous year of income taxes paid as a proportion of the book value of long-term debt for the fiscal year of the call decision. The predicted relation for Change in Leverage is uncertain, but I predict that Change in Income Tax will be lower for bond years without a call than bond years with a call.
3. Wealth Transfer
Managers, acting in the interest of equityholders, may call outstanding debt if doing so transfers wealth from bondholders to equityholders. If the market value of outstanding callable debt is significantly higher than the current call price, a call will transfer wealth from the debtholders to the other security holders of the firm. (However, if the market expects the debt to be called, then the bond price will show negative convexity and will not increase much above the current call price.) If the called issue is the firm's only outstanding debt, then the call will result in a wealth transfer to or from equity. With a single outstanding debt issue, minimizing the value of the debt will maximize the value of equity. Therefore, the optimal policy will be to call the debt whenever the market price equals the call price.  If managers wish to transfer wealth to equityholders, then there might be a positive reaction in equity prices on the call announcement date. In my analysis, I do not define a proxy for the wealth transfer motivation. However, I do examine wealth transfer by using standard event-study methodology.
Managers may call outstanding debt to modify the maturity and duration of the firm's debt. This option to recontract can be valuable for immunizing equity value from interest rate changes, addressing problems of information asymmetry, or reducing the agency costs of debt. Kraus (1983) suggests that call provisions could reduce the sensitivity of equity values to unexpected changes in interest rates. I define Interest Immunization as net working capital divided by the sum of net working capital and long-term assets. This measure shows the fraction of net assets that are short term.
The call provision also can address problems associated with information asymmetry. The provision does so by providing for debt to be called when the market determines the quality of the firm is higher than originally perceived (Flannery, 1986; and Robbins and Schatzberg, 1986). It addresses the underinvestment problem (Myers, 1977) by providing for the removal of debt prior to managers making the investment decision. Firms that have problems with information asymmetry or underinvestment, particularly those firms that have identified new growth opportunities, will show a relatively higher market-to-book ratio. I define the ratio of the market value of the firm's assets to the book value from the fiscal year of the call decision, Market-to-Book, as a proxy for the growth opportunities available to the firm.
I predict that Interest Immunization and Market-to-Book will be lower for bond years without a call than for bond years with a call. 
5. Call Provisions as Options
The option nature of the call provision could have some impact on firms' decisions to call nonrefundable debt.  The call prices for debt are usually the face value of the debt plus a call premium that declines by approximately $C/n per year on the anniversary date of the bond issue. (C is the annual dollar coupon payment and n is the length in years until the call premium falls to zero.) For bonds with a maturity of 20 years or more, the call premium typically falls to zero 20 years after the call protection ends. For bonds with maturities less than 20 years, the call premium falls to zero about 2/3 through the life of the bond. This decreasing schedule, in what is essentially the exercise price of the option, can motivate firms to wait to call the issue. Since the call premium declines with time, postponing the call will increase the value of the option.
However, the call premium also depends on the time-to-maturity of the bond, which could also affect firms' decisions to call nonrefundable debt. I measure the Call Premium as the approximate percentage of bond face value that would have to be paid by the issuer (in addition to the face value) to call the bond.
The value of the asset underlying the call option depends on the interest rate at which the fixed cash flows are discounted. If the bond can be called at a price less than the current market value, then a decrease in the interest rates will render the option "in the money." The Change in Interest Rates measure controls for interest-rate changes due to market-wide factors. The Change in Bond Rating controls for interest-rate changes due to firm-specific factors.
The slope of the yield curve can also affect the call decision. The implied forward rates of an upward-sloping yield curve might indicate that the market predicts that the value of outstanding debt and call option will decline. However, Livingston (1987) shows that because of possible future increases in the value of the call option, a firm might find it better to wait until the market price is well above the call price to call the bond.
Barclay and Smith (1995a) measure the term structure as the difference in month-end yields between a ten-year government bond and a six-month government bond. Similarly, I measure the Term Structure of interest rates as the difference in the month-end yield on ten-year government bonds and the month-end yield on three-month government bonds. I match this yield spread with the month and year in which I assume the firm makes the call decision.
The effect of the volatility of returns of the underlying asset on the value of the option is not clear. It is reasonable to expect that interest rates have a lower bound below which they will not drop. This translates to an upper bound in the market value of the underlying noncallable bond, which would limit the benefit of increased volatility.
The predicted relation for Call Premium is uncertain. However, I predict that Change in Bond Rating will be higher and that Change in Interest Rates and Term Structure will be lower for bond years without a call, rather than for bond years with a call.
King and Mauer (2000) find that firms with outstanding nonconvertible debt are more likely to call this debt if they have higher levels of liquidity. I define Cash as the sum in dollars of the accounting variables for cash and short-term securities; sales of common and preferred stock; sales of property, plant, and equipment; and sales of investments for the fiscal year of the call decision. The Cash variable uses only Compustat information from the fiscal year of the call. I do not include funds from new security issues or asset sales that occurred around the time of the call in these numbers if the issues or sales occurred in a different fiscal year.
I scale Cash by the total assets of the firm to yield the Cash/Total Assets control variable. I predict that Cash/Total Assets will be lower for bond years without a call, rather than for bond years with a call.
B. Refunding Constraint Variables
I propose three constraints that can be imposed by the nonrefunding provision. First, I use Cash/Total Call Price to examine funds available at the time of the call. Total Call Price is defined as the funds necessary to call all of the outstanding nonrefundable issues at the current call price. The Cash/Total Call Price measure is similar to the Cash/Total Assets control variable. However, if the constraint is binding, then the liquidity measures should be more important for calls of nonrefundable debt. The Cash/Total Call Price variable in the regressions below is included only as a variable that I interact with the dummy Bind. I define Bind as one for bond years before the nonrefunding constraint expires, and zero for bond years after that.
Second, I control for the expiration of refunding protection by using Years to Refunding Protection Expiration. If the nonrefunding provision is binding, then constrained firms with important incentives to call cannot call unless the firms are ready to face the costs of litigation from the debtholders or loss of some reputation in the capital markets. However, if the legal interpretation of the provision appears to limit the constraints imposed, then firms may be willing to bear the potential costs. If the nonrefunding provision does not expire for a number of years, then firms may have the incentive to call the debt now, rather than waiting until the provision expires. If the provision expires soon, then firms may postpone the call until after expiration. I code Years to Refunding Protection Expiration variable as a positive integer for the years before the expiration of refunding protection and zero for the years after expiration.
Third, I measure the incentive to call as the product of the Years to Refunding Protection Expiration and the Change in Interest Rates. The Years to Refunding Protection Expiration might be an incomplete measure of actual incentives. The more "in the money" the call option is, the more valuable it will be for the firms to call the nonrefundable bond, and the more willing firms may be to be exposed to potential litigation or capital market costs. The incentive to call might be influenced by the interaction between the time to expiration of refunding protection and a measure of the value of the call option.
The predicted relation for Years to Refunding Protection Expiration* Change in Interest Rates is uncertain. I predict that Cash/Total Call Price will be lower and that Years to Refunding Protection Expiration will be higher for bond years without a call than bond years with a call.
C. Comparisons of Means for Control and Refunding Constraint Variables
Table IV shows the comparisons of means for the control variables and the refunding constraint variables for firms that call and do not call nonrefundable issues for both the full sample and the two subsamples of investment- and speculative-grade bonds. The table compares the means for those bond years in which there was not a call to the means for those bond years in which there was a call.
The signs of the differences in means in Table 1V are consistent with the predicted signs. The results indicate that Change in Interest Rates and Term Structure are important parameters that affect calls of nonrefundable issues for both the full sample and the two subsamples. The differences in the Cash/Total Assets and Cash/Total Call Price measures of liquidity are not significant in the full sample or either subsample. The other refunding constraint variables for the full sample and subsamples are statistically significant and of the predicted sign.
There are two exceptions to the predicted signs. The Change in Bond Rating variable for the investment-grade subsample indicates that the average called investment-grade issue has a lower quality rating than do the bonds that were not called, and a lower quality rating than when the bond was issued. Consistent with the predictions, the average called speculative-grade issue has a higher quality rating than the bonds that were not called and a higher quality rating than when the bond was issued.
I propose two hypotheses--the enforceable contract hypothesis and the unenforceable contract hypothesis--to explain the constraints imposed by the nonrefunding provision.
A. The Enforceable Contract Hypothesis
The enforceable contract hypothesis holds that the nonrefunding provision precludes firms from using new, lower-interest-cost debt as a funding source for calling a nonrefundable issue. There are different implications for this hypothesis, depending on the constraint imposed by the provision.
For at least some firms with limited liquidity, the nonrefunding provision provides a binding constraint. Thatcher (1985) argues that some firms choose the nonrefunding provision to give themselves the option to call an outstanding issue for reasons other than a decrease in interest rates. At the same time, the provision protects the investor from calls for refunding purposes. Such firms are willing to forgo the option to refund debt if they believe that interest rates will not decrease enough for refunding to be valuable, or that the future prospects of the firm will not allow the firm to take advantage of a refunding opportunity anyway. These firms issue speculative-grade debt, have limited sources of funds available for redeeming debt, and limited growth prospects for generating resources to call outstanding nonrefundable debt in the future. Expiration of the refunding protection period will remove the constraint. Therefore, when interest rates decline, these firms will call nonrefundable debt at a higher rate after expiration of the provision than before.
For another set of firms with sufficient liquidity, the nonrefunding provision does not provide a binding constraint. These firms will forgo the opportunity to refund debt if they perceive that this constraint will not be binding because the firm will have sufficient funds from sources other than new debt issues to call the outstanding nonrefundable debt. These firms issue investment-grade debt, have several sources of funds available for redeeming debt, and have future growth prospects. These firms also have sources of funds other than new debt issues to call nonrefundable debt during the refunding protection period. Expiration of refunding protection will not affect the rate at which these issues are called.
B. The Unenforceable Contract Hypothesis
The unenforceable contract hypothesis holds that the nonrefunding provision does not provide a binding constraint because the intent of the provision is not enforceable. Firms might include the provision in their debt contracts, because investors think that it imposes constraints, but the firms recognize that it does not.
The nonrefunding provision provides a binding constraint only if the courts are willing to enforce the provision in a way that affects some firms' decisions. If no legal precedent has established minimum obstacles to circumvent the constraints of the provision, then the provision cannot bind the future actions of a firm and, thus, is not a constraint.
IV. Method and Results
If the nonrefunding provision provides a binding constraint, then the factors affecting the decision to call nonrefundable debt will be different from those for freely callable debt. I use pooled time-series and cross-sectional logistic (logit) regression analysis to examine the constraint imposed by the nonrefunding provision. First, I examine the call decision without considering the constraint imposed by the nonrefunding provision. This set of regressions identifies the important control variables as if the debt were freely callable. My second set of regressions includes the refunding constraint variables that are proxies for the constraints imposed by the nonrefunding provision.
For both the pooled and cross-sectional regressions, I use all bond-year observations, including those after the nonrefunding provision expires. If the nonrefunding provision is not binding, there should be no difference in the variables that affect the call decision before or after the provision's expiration. If the nonrefunding provision is binding, then the model that includes variables that discriminate between bonds called before and after expiration of refunding protection should provide superior predictive power.
In the pooled time series regressions, I treat each year after the bond is issued and could be called as an independent observation (1,545 bond-year observations and 110 calls). For the dependent variable, I create a binary variable that has a value of one for the fiscal year of a call and zero otherwise. If there are multiple calls in one fiscal year, I define the amount called as the total amount called during the whole year.
Because the assumption of independent errors is unlikely to be satisfied, the p-values in the pooled regressions may be underestimated. Most firms have several bond-years of observations, and these observations could be autocorrelated across individual firms. I use cross-sectional logit regression analysis to account for the potential error-dependence problem. For the cross-sectional logits, I use the time-series mean of each variable by bond issue and type of bond year. I separate the bond years for each issue into two types, years without a call and years with a call. I calculate the mean of each variable for each bond for both types (306 mean observations and 102 call observations). The single cross-section eliminates the problem of serially correlated errors and preserves the dispersion across the bonds, but does not exploit most of the time-series variation in the observations.
The cross-sectional logit regression also resolves any problem with the pooled regression that arises because the population of bond years without calls is much greater than the population of bond years with calls. With the pooled sample, just over 7% of the 1,545 observations are for bond years in which a nonrefundable bond is called. Because of the unbalanced sample, a naive model that predicts that no bonds will be called has predictive power comparable to those of more complex models. However, with the cross-sectional samples all of the bond years in which a bond is not called are represented by a single observation that is the mean of the variables from these years. All of the bond years in which a bond is called (most bonds are called only once) also are represented by a single observation that is the mean of the variables from these years. All of the bonds that are called have two observations in the cross-sectional data set--one for the years in which the bond was not called and one for the years in which the bond was called. All of the bonds that are not called will have one observation in the cross-sectional data set for the years in which the bond was not called. In the cross-sectional data set, just fewer than 34% of the 306 observations are for bond years in which a nonrefundable bond is called.
I also use event study methodology to examine the changes in equity values around call announcements for nonrefundable debt. Doing so provides insight into the market response to calls of nonrefundable debt. If the market is surprised by the call, then, as wealth is transferred between debt and equity, there might be a change in equity values around the call announcement. This change suggests that the market perceives that the nonrefunding provision provides a binding constraint for at least some firms. If there is no important change in equity values around the call announcement, then there are two possible explanations. One explanation is that the market perceives that the constraint is not binding--the market can predict which firms will not be constrained by the provision and will have already adjusted the prices to the likelihood of the call. The other explanation is that the effect is too small to detect by this approach.
B. Logit Regression Results
I examine six different explanations for calls of nonrefundable debt: reduction of net interest cost (controlled with Change in Interest Rates and Change in Bond Rating), change in capital structure (Change in Leverage, Change in Income Tax), wealth transfer, recontracting (Interest Immunization, Market-to-Book), call provisions as options (Call Premium, Change in Interest Rates, Change in Bond Rating, Term Structure), and liquidity (Cash/Total Assets). If the nonrefunding provision is binding, then refunding constraints will also be important ((Cash/Total Call Price) *Bind, Years to Refunding Protection Expiration, and Years to Refunding Protection Expiration * Change in Interest Rates).
Table V shows the pooled and cross-sectional logit regressions examining the calls of nonrefundable debt. Because there may be different incentives for using nonrefunding provisions for issuers of investment- and speculative-grade debt, I include dummy variables to distinguish between the two grades of debt. The dummy is one for speculative-grade issues, and zero otherwise. In addition, I include the speculative-grade dummy by itself. I do not include interacted variables for the Market-to-Book, Change in Interest Rates, and Term Structure variables. I also do not show likelihood ratio statistics comparing logit regressions that do and do not include the variables interacted with the speculative-grade dummy variable. These statistics indicate that these variables have explanatory power that is significant at the 5% level for the pooled and cross-sectional regressions.
Panel A in Table V includes those control variables that are appropriate under the assumption that the nonrefunding provision is not binding. Panel B in Table V includes the same control variables as in the regressions in Panel A, and also the refunding constraint variables that are important if the nonrefunding provision is binding. The panels include both the parameter estimate and the calculated p-value for the estimate. The bottom of each panel lists -2 times the log likelihood for the independent variables and the pseudo-[R.sup.2]. These statistics are measures of the goodness of fit for the logistic models. The pseudo-[R.sup.2] uses the likelihood function maximized for all of the parameters and the likelihood function maximized for only the intercept. It is analogous to the [R.sup.2] in multiple regressions. (See Maddala, 1983, for a more complete description of the goodness of fit measures.)
Panel C of Table V shows the likelihood ratios comparing the models with and without refunding constraints, the p-values for these likelihood ratios, and the number of observations and calls in the data set for which the parameters are estimated. The difference between the -2 times the log likelihood measures for different models that use the same data is a likelihood ratio statistic that is distributed approximately chisquare for large samples. I calculate the likelihood ratio statistic for comparing models that do and do not include the refunding constraint variables by taking the difference between the -2 times the log likelihood measures in Panels A and B. This statistic tests the hypothesis that the coefficients on the refunding constraint variables are jointly equal to zero.
1. Control Variables
Panel A in Table V indicates that Change in Interest Rates, Change in Bond Rating for speculative-grade issues, and Term Structure are strongly associated with calls of outstanding nonrefundable debt. If there are no refunding constraints, then these variables are significant predictors of calls of nonrefundable debt for both the pooled and cross-sectional logit regressions. Cash/Total Assets is marginally significant for only the pooled regression.
My results are consistent with two reasons for calls--the reduction in net interest cost and the call provisions as options. Change in capital structure to achieve a leverage change or tax reduction is not associated with calls. Similarly, recontracting for interest immunization or because of superior growth opportunities are not associated with call announcements.
2. Refunding Constraints
The regressions shown in Panel B of Table V are the same regressions as those in Panel A, except that these include the refunding constraint variables. I include two types of refunding constraints in the models. The first type, represented by (Cash/Total Call Price) *Bind, measures firms' ability to generate sources of cash other than new, lower-cost debt to call nonrefundable debt. If the firm has other sources of cash, then it can call the nonrefundable debt without abrogating the nonrefunding provision. The second type, represented by Years to Refunding Protection Expiration and the interacted variable that is the product of Years to Refunding Protection Expiration and the Change in Interest Rates, measures the effects of the time to expiration of the provision. If the provision is binding, then expiration of the provision should affect the call decision.
The empirical evidence in Table V indicates that (Cash/Total Call Price) *Bind does not provide significant explanatory power for calls of nonrefundable debt. The results suggest that firms are more likely to call debt when liquidity levels are higher, but expiration of the refunding provision does not alter this effect.
The Years to Refunding Protection Expiration variable is significant and inversely related to the likelihood of calls of nonrefundable debt. This suggests that firms are less likely to call nonrefundable debt when the nonrefunding provision is in place than they are if the provision has expired or will expire soon. The inverse relation is much stronger for speculative- than investment-grade debt for both the pooled and cross-sectional regressions, but the differences are not significant. These results are consistent with the nonrefunding provision constraining at least some firms. The results also show that firms that issue speculative-grade debt are more likely to be constrained than are firms that issue investment-grade debt.
The interacted variable of Years to Refunding Protection Expiration * Change in Interest Rates is significant and positively related to the likelihood of calls of nonrefundable debt. This result suggests that firms are more likely to call nonrefundable debt when the option to call is in-the-money and the nonrefunding protection will not expire for several years. If the bond cannot be called for several years, then firms might be willing to call nonrefundable debt early and expose themselves to potential litigation. The positive relation is stronger for speculative-grade debt, but the difference is not significant.
The likelihood ratio statistics for comparing the models with and without the refunding constraints indicate that the refunding constraints provide significant explanatory power to the model. The chi-square p-value for the likelihood ratio is significant at the 1% level for the pooled and cross-sectional regressions. The empirical results for the refunding constraints indicate that the nonrefunding provision restricts at least some of those firms that include it in the debenture indenture.
If the refunding constraints are important, then the regressions shown in Panel A of Table V are essentially misspecified because I omit relevant independent variables. The coefficients in Panel A will be unbiased only if the omitted variables are orthogonal to the included variables.
Table VI shows the Pearson correlation coefficients for the variables included in the regressions, but does not include the variables that are interacted with the speculative-grade dummy. Of the 66 reported, 30 are significantly different from zero at the 5% level. One of the refunding constraint variables is significantly correlated with each of the control variables except for the Change in Leverage and Change in Income Tax.
Another way to evaluate the inferences from the pooled and cross-sectional regressions in Table V is to examine the classification success of the predicted structural models. However, any interpretation of the results from the pooled regressions must address the unbalanced sample between bond years in which there is or is not a call (110 calls out of 1,545 observations). Panels A and B of Table VII show the classification success for the pooled and cross-sectional models shown in Panels A and B, respectively, of Table V and compares the models to a naive model predicts no bonds will be called. The classification success for the pooled regression indicates that the naive model has superior predictive power to both the models with and without the refunding constraints (86.2% correctly predicted by the naive model compared to 84.9% and 85.1% correctly predicted for the logit models without and with refunding constraints, respectively).
Classification success for the cross-sectional model supports the inferences of the pooled model. The cross-sectional model has more closely matched sample sizes (102 calls out of 306 observations) and significantly superior predictive power over the naive model (66.9% and 72.1% correctly predicted for the model without and with refunding constraints, respectively, compared to 36% correctly predicted by the naive model). These results show that my variables have explanatory value for calls of nonrefundable debt. The cross-sectional results also indicate that the logit model that includes the refunding constraint variables provides superior predictive power to the model that does not include them.
C. Equity Value Changes Around Call Announcements
The logit regressions do not address the wealth transfer and signaling incentives for calls of nonrefundable debt. If these incentives are important, I predict they will have direct implications for the value of firms' equity.
To examine these possibilities, I measure the average cumulative abnormal returns associated with the call announcement for which sufficient data are available.  Of the 84 nonrefundable issues called before the expiration of the nonrefunding provision, 67 have call announcements identified from either the press release newswire, or The Wall Street Journal. I evaluate the six months prior to the call date for announcements to identify the call or sources of funds to be used to finance a call. I eliminate the calls without clear announcement dates, calls that occur within 220 calendar days after a call by the same firm, and calls by firms for which data is not available on CRSP. The resulting sample of 51 calls comprises 47 investment-grade calls and four speculative-grade calls. I also examine sub samples of the called-bond data set by grade and resulting change in leverage.
Event studies for call announcement dates for nonrefundable debt issues indicate that there are no statistically significant changes in equity value for the full sample or any of the subsamples around the call announcement date. These results are consistent with the results in King and Mauer (2000), who find no significant change in equity values around call announcements for 255 non-utility calls for the period of 1975 to 1994. King and Mauer find statistically significant equity value changes around call announcements for leverage increasing and leverage-decreasing subsamples of non-utility calls. These results are consistent with optimizing the use of debt tax shields. I find no significant change in equity values for a similar partitioning for calls of nonrefundable debt.
The results for the nonrefundable calls suggest that neither wealth transfer nor capital structure signaling are important incentives for calls. The market either perceives that the constraint is not binding, it can predict which firms will not be constrained by the provision and adjusts prices to the likelihood of the call, or the amount of the wealth transfer is not material. The event study methodology cannot distinguish among these possibilities.
V. Summary and Conclusion
Bond covenants persist because they represent a contractual solution that is efficient for both the issuer and the investor. This study provides insight into one aspect of the total package of covenants that are often included in an indenture. I examine what has been, over the last four decades, one of the most common types of provisions--the nonrefunding provision. The nonrefunding provision is intended to protect investors from calls of outstanding debt following interest-rate decreases. My empirical examination of the nonrefunding provision indicates that it constrains at least some firms that include it in the debt contract.
I use logit models to examine the decision to call nonrefundable debt. I include variables to measure those incentives to call outstanding debt that are identified in the literature. I also include variables to control for the nonrefunding constraints.
My empirical results indicate that calls of outstanding debt are strongly associated with a decrease in interest rates, an upward-sloping yield curve, and an increase in bond quality. The models that include the nonrefunding provision constraints provide a significantly better explanation of the calls of nonrefundable debt than can those that do not include the refunding constraints. These results indicate: i) that firms with non-debt sources of funds are more likely to call nonrefundable debt; ii) that the expiration of the refunding protection provision affects the call decision of at least some firms; iii) and, that firms call nonrefundable debt prior to expiration of the provision if the option to call is in the money.
Calls of investment- and speculative-grade debt are consistent with the enforceable contract hypothesis that holds the nonrefunding provision precludes firms from using new, lower-interest-cost debt as a source of funds for the call of a nonrefundable issue. However, since 1983, there have been over 192 nonrefundable industrial bonds called prior to expiration of refunding protection. The source of the call funds can be precisely identified from public information for just over 50 of these calls. Flow-of-funds information indicates that the median firm issues at least three times more debt in the year of the call than is outstanding in the called nonrefundable issue. Examination of individual calls shows that at least 115 of the industrial calls are made by firms issuing contemporaneous debt. The majority of these contemporaneous issues are at a lower coupon rate, and for at least 96 of these calls, the amount of debt issued is larger in face value than the original nonrefundable issue.
The analysis in this study suggests that the nonrefunding provision might not provide an efficient contracting tool if investors cannot discriminate between which firms are or are not constrained by the provision. This deficiency in the provision may not have been evident when firms had little incentive to call outstanding debt. However, the dramatic decrease in interest rates from the early 1980s to the early 1990s provided managers with significant incentives to circumvent the nonrefunding provision and call outstanding nonrefundable debt.
If investors expected that the nonrefunding provision protected the bonds from being called when interest rates declined, then the calls would have affected their perception of the constraint imposed by the provision. Legal precedence, establishing only minor obstacles to circumventing the constraints of the provision, further weakened the usefulness of the provision as a contracting tool from investors' point of view. Thus, the nonrefunding provision might have been a "neutral mutation" that persisted for several decades, but did no harm. The survival value of the provision was lost with the change in interest rate environment.
The use of the nonrefunding provision was effectively eliminated as a contracting choice from the beginning of 1994. At the same time, firms began to use the make-whole provision and the clawback provision. These provisions might provide a superior contracting solution and have perhaps replaced the nonrefunding provision as a contract choice.
I thank Jeff Coles, Ranjan D'Mello, Michael Joehnk, Kenneth Lehn, Michael P. Smith and particularly Richard L. Smith, as well as seminar participants at the University of Otago, and the 1998 Financial Management Association International Meetings for helpful comments. I also thank the reviewers and the Editors for numerous suggestions that helped me improve this study.
(*.) Francis J. Kerins, Jr. is an Assistant Professor at Washington State University.
(1.) See Miller(1977) for a discussion of financial heuristics.
(2.) A firm can always purchase outstanding debt on the secondary market at the market price or through a tender offer.
(3.) See Goyal, Gollapudi, and Ogden (1998) for a complete description of the clawback provision. Goyal et al. analyze 85 high-yield equity clawback bonds issued from 1992 to 1994. They argue that the provision is used to mitigate wealth transfers from equity to debt when new equity is issued.
(4.) See, for example, Crabbe and Helwege (1994), Mitchell (1991), and the literature cited therein.
(5.) Smith and Warner (1979), Vu (1986), and King and Mauer (2000) suggest that removal of restrictive covenants could also be an important incentive for calls of outstanding debt. This motivation is not considered here.
(6.) See, for example, Bowlin (1966), Jen and Wert (1967), Kidwell (1976), and Pye (1966).
(7.) Longstaff and Tuckman (1994) show that the wealth transfer effects are less clear if a firm has multiple outstanding debt issues and the call changes the capital structure of the firm. Fischer, Heinkel, and Zechner (1989) and Mauer (1993) note that transaction costs may result in it being optimal to wait until the market price significantly exceeds the call price.
(8.) See Guedes and Opler (1996) and Stohs and Mauer (1996) for a discussion of asset maturity. Also, see Barclay and Smith (1995a, 1995b), Gayer and Gayer (1993), Smith and Watts (1992), and Stohs and Mauer (1996) for discussions of growth opportunities and the market-to-book measure. This variable does not address potential problems arising from comparing ratios in increasing markets with ratios in decreasing markets.
(9.) See, for example, Barnea, Haugen, and Senbet (1980), Brennan and Schwartz (1977), Kraus (1983), Livingston (1987), and Yawitz and Anderson (1977).
(10.) I obtain returns from CRSP; I use conventional event study methodology to estimate abnormal returns. See Brown and Warner (1985).
Asquith, P., 1995, "Convertible Bonds Are Not Called Late," Journal of Finance 50, 1275-1289.
Barclay, M.J. and C.W. Smith, 1995a, "The Maturity Structure of Corporate Debt," Journal of Finance 50, 609-631.
Barclay, M.J. and C.W. Smith, 1995b, "The Priority Structure of Corporate Liabilities," Journal of Finance 50, 899-917.
Barnea, A., R.A. Haugen, and L.W. Senbet, 1980, "A Rationale for Debt Maturity Structure and Call Provisions in the Agency Theoretic Framework," Journal of Finance 35, 1223-1234.
Bowlin, O.D., 1966, "The Refunding Decision: Another Special Case in Capital Budgeting," Journal of Finance 21, 55-68.
Brennan, M. and E. Schwartz, 1977, "Convertible Bonds: Valuation and Optimal Strategies for Call and Conversion," Journal of Finance 32, 1699-1715.
Brown, S. and J. Warner, 1985, "Using Daily Stock Returns: The Case of Event Studies," Journal of Financial Economics 14, 3-32.
Crabbe, L. and J. Helwege, 1994, "Alternative Tests of Agency Theories of Callable Corporate Bonds," Financial Management 23, 3-20.
Dhillon, U., T. Noe, and G Ramirez, 1998, "Bond Calls, Credible Commitment, and Equity Dilution: A Theoretical and Clinical Analysis of Simultaneous Tender and Call (STAC) Offers," Binghamton University Working Paper.
Fischer, E.O., R. Heinkel, and J. Zechner, 1989, "Dynamic Recapitalization Policies and the Role of Call Premia and Issue Discounts," Journal of Financial and Quantitative Analysis 24, 427-446.
Flannery, M.J., 1986, "Asymmetric Information and Risky Debt Maturity Choice," Journal of Finance 41, 19-37.
Gaver, J. and K. Gaver, 1993, "Additional Evidence on the Association Between the Investment Opportunity Set and Corporate Financing, Dividend and Compensation Policies," Journal of Accounting and Economics 16, 125-160.
Goyal, V., N. Gollapudi, and J. Ogden, 1998, "A Corporate Bond Innovation of the 90s: The Clawback Provision in High-Yield Debt," Journal of Corporate Finance 4, 301-320.
Guedes, J. and T. Opler, 1996, "The Determinants of the Maturity of Corporate Debt Issues," Journal of Finance 51, 1809-1833.
Jen, F.C. and J.E. Wert, 1967, "The Effect of Call Risk on Corporate Bond Yields," Journal of Finance 22, 637-651.
Kidwell, D.S., 1976, "The Inclusion of Call Provisions by State and Local Governments," Journal of Money, Credit, and Banking 8, 391-398.
King, D. and D.C. Mauer, 2000, "Corporate Call Policy for Nonconvertible Bonds," Journal of Business 73, 403-444.
Kraus, A., 1983, "An Analysis of Call Provisions and the Corporate Refunding Decision," Midland Corporate Finance Journal 1, 46-60.
Livingston, M., 1987, "Measuring the Benefit of a Bond Refunding: The Problem of Nonmarketable Call Options," Financial Management 16, 38-40.
Longstaff, F.A. and B. Tuckman, 1994, "Calling Nonconvertible Debt and the Problem of Related Wealth Transfer Effects," Financial Management 23, 21-27.
Maddala, G.S., 1983, Limited-Dependent and Qualitative Variables in Econometrics, Cambridge, UK, Cambridge University Press.
Mann, S. and E. Powers, 2001, "Indexing a Bond's Call Price: An Analysis of Make-Whole Call Provisions," University of South Carolina Working Paper.
Mauer, D.C., 1993, "Optimal Bond Call Policies Under Transaction Costs," Journal of Financial Research 16 ,23-37.
Miller, M. 1977, "Debt and Taxes," Journal of Finance 32, 261-275.
Mitchell, K., 1991, "The Call, Sinking Fund, and Term-to-Maturity Features of Corporate Bonds: An Empirical Investigation," Journal of Financial and Quantitative Analysis 26, 201-222.
Myers, S.C., 1977, "Determinants of Corporate Borrowing," Journal of Financial Economics 5, 147-175.
Pye, G., 1966, "The Value of the Call Option on a Bond," Journal of Political Economy 74, 200-205.
Robbins, H.R. and J.D. Schatzberg, 1986, "Callable Bonds: A Risk-Signaling Mechanism," Journal of Finance 41, 935-949.
Ross, S., 1977, "The Determination of Financial Structure: The Incentive-Signaling Approach," Bell Journal of Economics 8, 23-40.
Smith, C.F. and J. Warner, 1979, "On Financial Contracting: An Analysis of Bond Covenants," Journal of Financial Economics 7, 117-161.
Smith, C.F. and R.L. Watts, 1992, "The Investment Opportunity Set and Corporate Financing, Dividend, and Compensation Policies," Journal of Financial Economics 32, 263-292.
Stohs, M. H. and D. C. Mauer, 1996, "The Determinants of Corporate Debt Maturity Structure," Journal of Business 69, 279-312.
Thatcher, J.S., 1985, "The Choice of Call Provision Terms: Evidence of the Existence of Agency Costs of Debt," Journal of Finance 40, 549-561.
Vu, J.D., 1986, "An Empirical Investigation of Calls of Non-Convertible Bonds," Journal of Financial Economics 16, 235-265.
Wall Street Journal, 1992, "Financing Business: Kroger Co.," (November 23).
Yawitz, J.B. and J.A. Anderson, 1977, "The Effect of Bond Refunding on Shareholder Wealth," Journal of Finance 32, 1738-1746.
Numbers of Bonds Issued by Year and Type of Call Protection
This table is a list of the number of bonds issued by type of call protection from 1972 through August 4, 2000. Data include industrial, financial, and utility bond issues. Data source is Securities Data Company. The table does not include data for freely callable issues or equity callbacks. The use of equity clawback provisions in high-yield debt contracts began in 1992, but I was unable to find a reliable source for numbers of issues that include this provision.
Call Provision Type Noncallable/ Year Nonrefundable [a] Nonrefundable [b] Noncallable [c] 1972 138 0 28 1973 108 0 20 1974 125 0 35 1975 167 0 59 1976 178 0 140 1977 175 0 126 1978 161 0 88 1979 142 0 107 1980 169 1 162 1981 137 0 120 1982 146 4 152 1983 165 4 98 1984 100 11 153 1985 149 21 284 1986 263 40 403 1987 125 32 299 1988 85 25 323 1989 44 13 281 1990 17 2 163 1991 23 1 227 1992 18 9 562 1993 16 4 930 1994 0 0 1126 1995 0 0 2286 1996 1 0 2988 1997 0 0 3353 1998 0 0 5605 1999 0 0 5309 2000 0 0 2003 Noncallable Make Year for Life [d] Whole [e] 1972 255 - 1973 148 - 1974 223 - 1975 297 - 1976 138 - 1977 83 - 1978 54 - 1979 31 - 1980 67 - 1981 62 - 1982 143 - 1983 144 - 1984 184 - 1985 209 - 1986 382 - 1987 419 - 1988 300 - 1989 433 - 1990 530 - 1991 1352 3 1992 1533 1 1993 2285 1 1994 2324 0 1995 3044 25 1996 3985 100 1997 6515 187 1998 6765 316 1999 6649 256 2000 5169 113
(a.)Nonrefundable issues restrict refunding for a period following issuance and then are freely callable.
(b.)Noncallable/nonrefundable issues restrict calls for a period following issuance, restrict refunding for an additional period, and then are freely callable.
(c.)Noncallable issues restrict calls for a period following issuance and then are freely callable.
(d.)Noncallable for life issues restrict calls for the life of the bond.
(e.)Call prices increase with decreases in index interest rates for make-whole issues.
Summary Information for Nonconvertible Industrial Bonds Issued 1982-1992 and the Subset Called Prior to January, 1996, by Issue Grade
Panel A shows information for freely callable industrial bonds. Panel B shows noncallable industrial bonds. Panel C shows nonrefundable bonds issued during the period and the subset called prior to 1996, including bonds called after nonrefunding protection expired. Panel D shows bonds from Panel C that are issued by firms with information on Compustat from 1982 to 1995 that had SIC codes from 2000 to 4899, 4940 to 5999, and the subset of bonds called prior to expiration of refunding protection.
Industrial Bonds Issued 1982-1992 Investment Grade Med. Mean Freely Callable Bonds Number of Issues 12 Number of Firms 11 Amount Issued (Million $) 140.3 160.4 Offer Yield (%) 14.2 12.4 Maturity (Years) 17.0 16.3 Bonds with a Call Protection Period Number of Issues 441 Number of Firms 237 Amount Issued (Million $) 100.0 144.7 Offer Yield (%) 9.8 10.1 Maturity (Years) 10.0 13.6 Length of Call Protection (Years) 7.0 7.1 Speculative Grade Med. Mean Freely Callable Bonds Number of Issues 19 Number of Firms 16 Amount Issued (Million $) 10.0 34.2 Offer Yield (%) 15.0 14.5 Maturity (Years) 5.0 8.0 Bonds with a Call Protection Period Number of Issues 506 Number of Firms 387 Amount Issued (Million $) 105.0 154.8 Offer Yield (%) 12.8 12.7 Maturity (Years) 10.0 10.8 Length of Call Protection (Years) 5.0 4.9 Industrial Bonds Called 1982-1995 Investment Grade Med. Mean Freely Callable Bonds Number of Issues 1 Number of Firms 1 Amount Issued (Million $) 58.3 58.3 Offer Yield (%) 12.1 12.1 Maturity (Years) 5.0 5.0 Bonds with a Call Protection Period Number of Issues 135 Number of Firms 106 Amount Issued (Million $) 100.0 131.2 Offer Yield (%) 10.6 10.7 Maturity (Years) 10.0 9.0 Length of Call Protection (Years) 7.0 6.1 Speculative Grade Med. Mean Freely Callable Bonds Number of Issues 1 Number of Firms 1 Amount Issued (Million $) 37.1 37.1 Offer Yield (%) 13.7 13.7 Maturity (Years) 3.0 3.0 Bonds with a Call Protection Period Number of Issues 73 Number of Firms 62 Amount Issued (Million $) 100.0 150.9 Offer Yield (%) 13.1 13.2 Maturity (Years) 10.0 11.2 Length of Call Protection (Years) 5.0 4.9 Bonds with a Refunding Protection Period Number of Issues 274 Number of Firms 173 Amount Issued (Million $) 100.0 147.5 150.0 Offer Yield (%) 10.5 10.7 10.8 Maturity (Years) 30.0 27.0 30.0 Length of Ref. Protection (Years) 10.0 9.4 10.0 Bonds with Refunding Protection and Included on Compustat Number of Issues 159 Number of Firms 91 Amount Issued (Million $) [a] 100.0 155.0 120.0 Offer Yield (%) [b] 10.3 10.5 13.1 Maturity (Years) [b] 30.0 28.5 10.0 Length of Ref. Protection (Years) [b] 10.0 9.8 5.0 Size (Billion $) [c, d] 5.5 9.7 1.0 Leverage [c, e] 0.17 0.18 0.31 Amount Issued/Size [a, f] 0.02 0.03 0.12 Bonds with a Refunding Protection Period Number of Issues 262 134 Number of Firms 224 93 Amount Issued (Million $) 171.0 150.0 Offer Yield (%) 10.9 10.8 Maturity (Years) 29.0 30.0 Length of Ref. Protection (Years) 10.0 10.0 Bonds with Refunding Protection and Included on Compustat Number of Issues 47 78 Number of Firms 39 52 Amount Issued (Million $) [a] 151.50 150.0 Offer Yield (%) [b] 15.0 10.8 Maturity (Years) [b] 13.4 30.0 Length of Ref. Protection (Years) [b] 5.1 10.0 Size (Billion $) [c, d] 1.8 6.3 Leverage [c, e] 0.33 0.19 Amount Issued/Size [a, f] 0.23 0.02 Bonds with a Refunding Protection Period Number of Issues 79 Number of Firms 65 Amount Issued (Million $) 171.0 100.0 128.6 Offer Yield (%) 10.9 14.1 14.1 Maturity (Years) 29.0 12.0 12.8 Length of Ref. Protection (Years) 10.0 5.0 5.0 Bonds with Refunding Protection and Included on Compustat Number of Issues 6 Number of Firms 4 Amount Issued (Million $) [a] 170.8 50.0 275.0 Offer Yield (%) [b] 10.8 13.1 13.0 Maturity (Years) [b] 29.1 10.0 11.4 Length of Ref. Protection (Years) [b] 9.9 5.0 5.0 Size (Billion $) [c, d] 11.3 0.6 3.7 Leverage [c, e] 0.19 0.32 0.35 Amount Issued/Size [a, f] 0.03 0.07 0.09
(a.)Differences in means between investment- and speculative-grade issues are significant at the .05 level for only the called bonds subset.
(b.)Differences in means between investment- and speculative-grade issues are significant at the .05 level for both the issued bonds and called bonds subset.
(c.)Differences in means between investment- and speculative-grade issues are significant at the .05 level for only the issued bonds.
(d.)Size is defined as the sum of the book value of debt plus the market value of equity for the fiscal year of issue.
(e.)Leverage is defined as the ratio of the book value of long-term debt to the size measure for the fiscal year of issue.
(f.)Amount issued/size is defined as the issue amount divided by the size measure for the fiscal year of issue.
Distribution of Bond-Year Observations and Calls
The full sample contains 1,545 bond-year observations for the years 1982 through 1995. The bonds include all nonrefundable debt issued from 1982 through 1992 by firms on Compustat that have SIC codes from 2000 to 4899 and 4940 to 5999. A bond-year observation develops for each year after absolute call protection expires and before the bond is either called or matures. The resulting observations include bond-years before and after refunding protection expires. The calls include those bonds in the observations that are called prior to 1996. These observations also include calls that occurred before and after refunding protection expired.
Investment Speculative Grade Grade Total Observations: During Refunding Protection Period 1,154 196 1,350 After Refunding Protection Period 96 99 195 Total 1,250 295 1,545 Calls: During Refunding Protection Period 79 6 85 After Refunding Protection Period 10 15 25 Total 89 21 110
Differences Between Variables for Bond Years During Which Nonrefundable Bonds Are Called Versus Bond Years Where Nonrefundable Bonds Are Not Called
This table shows the predicted and realized relations between the means of the control and refunding constraint variables for bond years during which firms do not call nonrefundable bonds compared to years during which firms call nonrefundable bonds. The table shows means for all bond years and the investment- and speculative-grade subsamples. In the predicted relations column, NC represents years during which nonrefundable bonds are not called and C represents years during which nonrefundable bonds are called.
All Bond Years Means Predicted Relation NC Control Variables Change in Interest Rates NC[less than]C 1.14 Change in Leverage - -0.17 Change in Income Tax NC[less than]C 1.06 Interest Immunization NC[less than]C 0.20 Market-to-Book NC[less than]C 1.53 Call Premium - 8.15 Change in Bond Rating NC[greater than]C 0.86 Term Structure NC[less than]C 1.91 Cash/Total Assets NC[less than]C 0.02 Refunding Constraint Variables Cash/Total Call Price NC[less than]C 0.78 Years to Ref. Prot. Exp. NC[greater than]C 5.14 Years to Ref. Prot. Exp [*] - 5.47 Change in Interest Rates N= 1,435 Differences in Means C NC-C t-Stat Control Variables Change in Interest Rates 1.31 -0.17 -8.75 Change in Leverage 0.01 -0.18 -1.12 Change in Income Tax 1.79 -0.73 -0.27 Interest Immunization 0.18 0.01 0.20 Market-to-Book 1.53 0.00 -0.76 Call Premium 7.44 0.71 2.52 Change in Bond Rating 0.58 0.27 1.17 Term Structure 2.44 -0.53 -5.10 Cash/Total Assets 0.08 -0.06 -1.39 Refunding Constraint Variables Cash/Total Call Price 0.78 0.00 0.02 Years to Ref. Prot. Exp. 3.29 1.85 5.78 Years to Ref. Prot. Exp [*] 4.24 1.23 3.70 Change in Interest Rates N= 110 Investment Grade Bond Years Differences Means in Means NC C NC-C Control Variables Change in Interest Rates 1.13 1.31 -0.19 Change in Leverage -0.10 0.00 -0.10 Change in Income Tax 1.03 1.46 -0.43 Interest Immunization 0.20 0.17 0.02 Market-to-Book 1.52 1.52 -0.01 Call Premium 8.15 7.62 0.53 Change in Bond Rating 0.59 1.03 -0.45 Term Structure 1.92 2.38 -0.47 Cash/Total Assets 0.06 0.08 -0.02 Refunding Constraint Variables Cash/Total Call Price 0.57 0.66 -0.09 Years to Ref. Prot. Exp. 5.75 3.98 1.78 Years to Ref. Prot. Exp [*] 6.10 5.13 0.97 Change in Interest Rates N= 1,161 89 Speculative Grade Bond Years Means t-Stat NC C Control Variables Change in Interest Rates -8.83 1.20 1.31 Change in Leverage -0.85 -0.46 0.05 Change in Income Tax -1.28 1.18 3.16 Interest Immunization 1.12 0.20 0.22 Market-to-Book -0.16 1.38 1.55 Call Premium 1.95 8.18 6.68 Change in Bond Rating -2.47 2.00 -1.33 Term Structure -4.04 1.89 2.68 Cash/Total Assets -1.53 0.14 0.15 Refunding Constraint Variables Cash/Total Call Price 0.88 1.69 1.30 Years to Ref. Prot. Exp. 5.20 2.55 0.38 Years to Ref. Prot. Exp [*] 2.77 2.79 0.47 Change in Interest Rates N= 274 21 Differences in Means NC-C t-Stat Control Variables Change in Interest Rates -0.11 -2.33 Change in Leverage -0.52 -0.75 Change in Income Tax -1.98 -0.14 Interest Immunization -0.02 -0.07 Market-to-Book -0.16 -1.46 Call Premium 1.50 1.59 Change in Bond Rating 3.33 7.99 Term Structure -0.79 -3.36 Cash/Total Assets -0.01 -0.41 Refunding Constraint Variables Cash/Total Call Price 0.39 0.57 Years to Ref. Prot. Exp. 2.17 4.02 Years to Ref. Prot. Exp [*] 2.33 4.03 Change in Interest Rates N=
Pooled and Cross-Sectional LOGIT Regressions Examining the Predictability of the Calls of Nonrefundable Debt
The table reports on calls of 206 nonrefundable bonds (159 investment grade and 47 speculative grade issues) issued between 1982 and 1992 and called before 1996. These observations include calls after the nonrefunding provision expired. Spec. Dummy is zero for investment grade issues and one for speculative grade issues. Bind is one for bond years before refunding protection expires and zero for years after refunding protection expires.
Panel A shows the LOGIT regressions assuming the firms are not constrained by the nonrefunding provision. Panel B shows the LOGIT regressions that include the variables that are proxies for the refunding constraint. The likelihood ratios comparing the models in Panel A with the models in Panel B are shown in Panel C.
Panel A. Assuming Refunding Constraints Are Not Binding on Any Firms Pooled Regression Independent Variable Estimate p-value Intercept -8.91 0.00 Spec. Dummy 0.30 0.65 Change in Interest Rates 3.61 0.00 Change in Leverage 0.11 0.52 Change in Leverage * Spec. Dummy 0.21 0.71 Change in Income Tax 0.03 0.26 Change in Income Tax * Spec. Dummy -0.03 0.31 Interest Immunization -0.48 0.49 Interest Immunization * Spec. Dummy 0.35 0.63 Market-to-Book 0.29 0.18 Call Premium 0.03 0.49 Call Premium * Spec. Dummy -0.05 0.55 Change in Bond Rating 0.05 0.39 Change in Bond Rating* Spec. Dummy -0.39 0.00 Term Structure 0.52 0.00 Cash/Total Assets 1.95 0.07 (Cash/Total Assets)*Spec. Dummy -1.54 0.43 -2 Log Likelihood for Model 677.29 pseudo-[R.sup.2] 0.12 Cross-Sectional Regression Independent Variable Estimate Intercept -14.47 Spec. Dummy -1.41 Change in Interest Rates 9.88 Change in Leverage 0.26 Change in Leverage * Spec. Dummy 2.31 Change in Income Tax 0.03 Change in Income Tax * Spec. Dummy 0.23 Interest Immunization -0.43 Interest Immunization * Spec. Dummy 0.04 Market-to-Book 0.55 Call Premium -0.17 Call Premium * Spec. Dummy 0.03 Change in Bond Rating 0.33 Change in Bond Rating* Spec. Dummy -0.88 Term Structure 1.08 Cash/Total Assets 3.14 (Cash/Total Assets)*Spec. Dummy -0.32 -2 Log Likelihood for Model 220.46 pseudo-[R.sup.2] 0.29 Independent Variable p-value Intercept 0.00 Spec. Dummy 0.34 Change in Interest Rates 0.00 Change in Leverage 0.47 Change in Leverage * Spec. Dummy 0.15 Change in Income Tax 0.68 Change in Income Tax * Spec. Dummy 0.10 Interest Immunization 0.70 Interest Immunization * Spec. Dummy 0.97 Market-to-Book 0.16 Call Premium 0.05 Call Premium * Spec. Dummy 0.87 Change in Bond Rating 0.04 Change in Bond Rating* Spec. Dummy 0.00 Term Structure 0.00 Cash/Total Assets 0.22 (Cash/Total Assets)*Spec. Dummy 0.94 -2 Log Likelihood for Model pseudo-[R.sup.2] Panel B. Assuming Refunding Constraints Are Binding on at Least Some Firms Pooled Regression Independent Variable Estimate p-value Intercept -2.73 0.05 Spec. Dummy -0.09 0.90 Change in Interest Rates -0.55 0.54 Change in Leverage 0.07 0.64 Change in Leverage * Spec. Dummy 0.08 0.89 Change in Income Tax 0.02 0.53 Change in Income Tax * Spec. Dummy -0.02 0.59 Interest Immunization -0.73 0.33 Interest Immunization * Spec. Dummy 0.60 0.43 Market-to-Book 0.07 0.76 Call Premium -0.03 0.65 Call Premium * Spec. Dummy 0.08 0.45 Change in Bond Rating 0.04 0.56 Change in Bond Rating* Spec. Dummy -0.34 0.01 Term Structure 0.60 0.00 Cash/Total Assets 3.68 0.04 (Cash/Total Assets)*Spec. Dummy -3.05 0.23 (Cash/Total Call Price)*Bind -0.16 0.44 (Cash/Total Call Price)*Bind*Spec. Dummy 0.17 0.55 Years to Ref. Prot. Expiration -1.89 0.00 Years to Ref. Prot. Exp * Spec. Dummy -1.60 0.48 Years to Ref. Prot. Exp* Ch. in Int. Rates 1.52 0.00 Years to Ref. Prot. Expiration * 0.72 0.69 Change in Int. Rates * Spec. Dummy -2 Log Likelihood for Model 611.48 pseudo-[R.sup.2] 0.19 Cross-Sectional Regression Independent Variable Estimate p-value Intercept 6.69 0.08 Spec. Dummy -2.12 0.22 Change in Interest Rates -5.80 0.02 Change in Leverage 0.52 0.25 Change in Leverage * Spec. Dummy 3.31 0.23 Change in Income Tax -0.05 0.52 Change in Income Tax * Spec. Dummy 0.48 0.03 Interest Immunization -3.89 0.03 Interest Immunization * Spec. Dummy 3.05 0.11 Market-to-Book 0.83 0.23 Call Premium 0.03 0.89 Call Premium * Spec. Dummy 0.17 0.59 Change in Bond Rating 0.41 0.09 Change in Bond Rating* Spec. Dummy -0.61 0.14 Term Structure 0.99 0.00 Cash/Total Assets 2.25 0.68 (Cash/Total Assets)*Spec. Dummy -2.54 0.73 (Cash/Total Call Price)*Bind 0.61 0.22 (Cash/Total Call Price)*Bind*Spec. Dummy 1.99 0.19 Years to Ref. Prot. Expiration -6.32 0.00 Years to Ref. Prot. Exp * Spec. Dummy -20.03 0.27 Years to Ref. Prot. Exp* Ch. in Int. Rates 4.82 0.00 Years to Ref. Prot. Expiration * 11.45 0.42 Change in Int. Rates * Spec. Dummy -2 Log Likelihood for Model 117.57 pseudo-[R.sup.2] 0.56 Panel C. Model Comparison Number of Observations 1,545 306 Number of Calls 110 102 Likelihood Ratio for Comparison 65.81 102.89 Chi-Square p-value for LR 0.00 0.00
Pearson Correlation Coefficients for the Control Variables and Refunding Constraint Variables
The table shows 66 correlation coefficients for pairs of control and refunding constraint variables used to evaluate the predictability of calls of nonrefundable debt. Table does not include variables interacted with the investment-grade dummy. Correlation coefficients with absolute values larger than 0.04 (0.05) are statistically significant at a 0.10 (0.05) confidence level.
Change in Change in Interest Change in Income Interest Rates Leverage Tax Immun. Change in Leverage -0.040 Change in Income Tax 0.026 0.047 Interest Immunization -0.003 -0.028 0.036 Market-to-Book -0.051 0.034 0.016 -0.010 Call Premium -0.330 -0.070 -0.001 -0.010 Term Structure 0.058 0.003 0.011 -0.011 Change in Bond Rating 0.192 -0.078 0.051 -0.069 Cash/Total Assets 0.007 -0.675 0.012 0.092 Cash/Total Call Price 0.033 -0.016 0.014 0.098 Years to Ref. Prot. Exp. -0.612 -0.006 -0.015 0.021 (Years to Ref. Prot. Exp) -0.464 0.000 -0.039 0.020 Change in Market-to- Call Term Bond Book Premium Structure Rating Change in Leverage Change in Income Tax Interest Immunization Market-to-Book Call Premium -0.017 Term Structure 0.004 0.007 Change in Bond Rating -0.022 -0.116 -0.005 Cash/Total Assets -0.074 0.128 0.004 -0.105 Cash/Total Call Price -0.086 0.027 0.015 -0.062 Years to Ref. Prot. Exp. -0.019 0.571 -0.079 -0.111 (Years to Ref. Prot. Exp) -0.032 0.593 -0.074 -0.080 Cash/ Cash/ Years to Total Total Call Ref. Prot. Assets Price. Exp. Change in Leverage Change in Income Tax Interest Immunization Market-to-Book Call Premium Term Structure Change in Bond Rating Cash/Total Assets Cash/Total Call Price 0.567 Years to Ref. Prot. Exp. -0.055 -0.130 (Years to Ref. Prot. Exp) -0.058 -0.133 0.970 (*.)(Ch. in Interest Rates)
Naive and Logit Model Classification Comparisons
The table shows classification comparisons between naive models that assume mat no bonds are called, logit models without the refunding constraint variables, and logit models including the refunding constraint variables. Panels A and B show the classification success of the pooled and cross-sectional regressions, respectively, from Panels A and B of Table V.
Classification Success Naive Model Pooled Regressions Expected/Actual # % Call/Call 0 0.0% No Call/No Call 1,435 93.3% Correct 1,435 93.3% Call/No Call 110 7.2% No Call/Call 0 0.0% Incorrect 110 7.2% Correct-Incorrect 1,325 86.2% Total 1,545 Observations: Cross-Sectional Regressions Expected/Actual Call/Call 0 0.0% No Call/No Call 204 72.1% Correct 204 72.1% Call/No Call 102 36.0% No Call/Call 0 0.0% Incorrect 102 36.0% Correct-Incorrect 102 36.0% Total 306 Observations: Logit Model Logit Model without Refunding with Refunding Constraints Constraints Pooled Regressions Expected/Actual # % # Call/Call 1 0.1% 4 No Call/No Call 1,424 92.6% 1,423 Correct 1,425 92.7% 1,427 Call/No Call 11 0.7% 12 No Call/Call 109 7.1% 106 Incorrect 120 7.8% 118 Correct-Incorrect 1,305 84.9% 1,309 Total 1,545 1,545 Observations: Cross-Sectional Regressions Expected/Actual Call/Call 71 23.1% 80 No Call/No Call 185 60.1% 184 Correct 256 83.1% 264 Call/No Call 19 6.2% 20 No Call/Call 31 10.1% 22 Incorrect 50 16.2% 42 Correct-Incorrect 206 66.9% 222 Total 306 306 Observations: Pooled Regressions Expected/Actual % Call/Call 0.3% No Call/No Call 92.5% Correct 92.8% Call/No Call 0.8% No Call/Call 6.9% Incorrect 7.7% Correct-Incorrect 85.1% Total Observations: Cross-Sectional Regressions Expected/Actual Call/Call 26.0% No Call/No Call 59.7% Correct 85.7% Call/No Call 6.5% No Call/Call 7.1% Incorrect 13.6% Correct-Incorrect 72.1% Total Observations:
|Printer friendly Cite/link Email Feedback|
|Author:||Kerins Jr., Francis J.|
|Date:||Mar 22, 2001|
|Previous Article:||The Initial Listing Decisions of Firms That Go Public.|
|Next Article:||The Debt Maturity Structure of Small Firms.|