Event risk bond covenants, agency costs of debt and equity, and stockholder wealth
Sung C. Bae
A recent innovation in the issuance of corporate debt is the inclusion in bond indentures of event risk protection language (or “poison puts”). These provisions typically allow for redemption of a bond issue in the event of a takeover, a merger, or any corporate anti-takeover restructuring that would drain a firm’s assets.
A number of recent empirical studies have examined the market valuation effect of event risk covenants (hereafter, ERCs) with respect to bond yield spreads on offering dates and in the secondary market (e.g., Bae, Levy, and Wannermacher (1991), Crabbe (1991), and Fields, Kidwell, and Klein (1991)) and the effect of event risk covenants on bondholder returns in leveraged buyouts (Asquith and Wizman (1990)). This paper investigates the effect of ERCs on agency costs of debt and equity and on stockholder wealth.
Agency problems typically center on two types of conflicts of interest in a firm. Conflicts between stockholders and managers arise because managers may pursue their own personal interest at the expense of stockholder wealth. Debt financing mitigates this problem by giving debtholders the option to force liquidation if the firm’s cash flows are poor (Harris and Raviv (1990)) or by reducing the free cash flow available to managers to engage in activities for their own benefit (Jensen (1986) and Stulz (1990)).(1)
Conflicts between debtholders and stockholders arise because bondholders face expropriation of their wealth through inappropriate investment decisions by the owners of the firm (Jensen and Meckling (1976) and Myers (1977)). Myers (1977) proposes two solutions to the agency problem between stockholders and bondholders – restrictive covenants and renegotiation provisions. ERCs in bond indentures are an example of the restrictive covenant solution.
Two different rationales may explain why a firm would include ERCs in the bond indenture. On the one hand, ERCs could be effective in protecting bondholders from wealth transfers that may result from debt-financed takeovers and recapitalizations, lowering agency costs of debt. Recent studies furnish evidence concerning the negative effects on bondholder wealth that occurred during the leveraged buyout craze of the 1980s (see Asquith and Wizman (1990) and Warga and Welch (1990)).
On the other hand, ERCs could act as a defacto deterrent for an otherwise more efficient transfer of corporate control by providing a mechanism for strengthening managerial resistance to hostile bids, increasing agency costs of equity. The vast majority of bond issues with ERCs impose limits restricting stock acquisitions, mergers, recapitalizations, and special dividends. Hence, in the presence of ERCs, management is less likely to engage in corporate restructuring activities that might otherwise enhance stockholder wealth.
These two motivations to offer ERCs imply that different stockholder wealth effects are likely to be associated with announcement of bond issues that include ERCs and announcement of issues without such covenants. To examine this issue, we compare the announcement-period stock returns of firms that issue protected bonds with the returns of firms that issue nonprotected bonds. We then attempt to identify any factors that may explain the different market response, if any, to the issuance of protected bonds.
We find that, on average, firms issuing protected bonds experience significantly more positive abnormal returns around the announcement of bond issuance than firms issuing nonprotected bonds. Further, regression analysis shows that 1) higher announcement-period abnormal returns are associated with firms that have a higher degree of agency costs of debt and 2) no strong relationship exists between the presence of ERCs and agency costs of equity of an issuing firm. These results support the notion that the issuance of protected bonds has a more positive impact on stockholder wealth than the issuance of nonprotected bonds primarily by lowering a firm’s agency costs of debt.
I. Related Studies and Development of Hypotheses
The agent/principal relationship and its associated problems are discussed by Jensen and Meckling (1976) in their seminal work on agency theory. The relationships between stockholders and bondholders and between stockholders and management give rise to agency costs of debt and agency costs of equity, respectively. The presence of ERCs may have an impact on agency costs of debt and equity and therefore on stockholder wealth.
A. Event Risk Covenants as a Tool for Stockholder Wealth Enhancement
ERCs would benefit stockholders and increase the value of a firm by lowering the firm’s agency costs of debt. The conflicting interests of bondholders and stockholders may cause two agency problems. First, a risk incentive (asset substitution) problem occurs if management/stockholders substitute riskier assets for assets of less risk, thereby increasing the volatility of a firm’s assets (Galai and Masulis (1976)). Several studies indicate that the increased use of debt may cause stockholders to invest in projects that increase the overall risk of the firm (see Barnea, Haugen, and Senbet (1980), Green and Talmor (1986), and Ramakrishnan and Thakor (1982)). If bondholders anticipate this action on the part of stockholders, they will pay less for the bonds at the time of issuance (or demand a higher interest rate to account for the greater risk), leading to a reduced value for the firm (Kahan and Klausner (1993)).
Second, an underinvestment problem occurs if the value of a firm’s investment opportunities is revealed to be less than the face value of the maturing debt plus the costs of undertaking the investment projects (Myers (1977)). As a result, managers will accept only projects whose value exceeds the face value of debt plus the initial investment and forgo some positive-net-present-value projects. Miller and Rock (1985), Myers (1984), and Myers and Majluf (1984) address the issues associated with signaling and the asymmetric information aspects of issuing debt versus equity and its potential impact on agency costs.
The conflict of interest between stockholders and bondholders, and the consequent agency costs of debt, might be mitigated by inclusion of certain covenants in the bond indenture (Myers (1977)). Haugen and Senbet (1988) and Smith and Warner (1979) provide theoretical support for the idea that bond provisions can mitigate agency costs of debt. Along these lines, Thatcher (1985) examines the use of two-tiered call provisions as a form of a complex contract used to reduce agency costs of debt. She finds that firms with potentially higher agency costs of debt tend to use more complex contracts to protect bondholders from expropriation of their wealth. Allen, Lamy, and Thompson (1987) find that bond issues with call protection command lower yields, supporting the notion that agency costs of debt can be lowered through the granting of certain bond provisions.
ERCs afford additional protection to bondholders. Empirical studies by Asquith and Wizman (1990) and Warga and Welch (1990) examine the negative impacts of major capital restructurings on bondholder wealth. Although Asquith and Wizman examine ERCs different from those that we study, they provide evidence that bondholders with more protection have fared better in major restructurings than have bondholders with less protection. According to Bicksler and Chen (1991), this impact should be expected, as these covenants help align bondholder and stockholder interests and thus mitigate the asset substitution problem that might exist in the absence of these provisions.
If ERCs lead to a reduction in agency costs of debt, then stock returns for firms issuing bonds with ERCs in the indenture should be more positively affected than stock returns for firms issuing bonds without ERCs, ceteris paribus.
B. Event Risk Covenants as a Tool for Managerial Entrenchment
The presence of ERCs in a bond indenture, on the other hand, may lead to higher agency costs of equity and hence a lower firm value. A typical ERC requires the firm to compensate bondholders for losses associated with special events, such as a merger or takeover. Consequently, the cost of an acquisition, even when the acquisition is in the best interests of stockholders, will be higher than in the absence of the ERC (see Kahan and Klausner (1993)). This makes the firm a more costly, less attractive target for potential takeover or restructuring.
Because an issuance of bonds with ERCs makes a firm less likely to be taken over, management may use these covenants as a shield against the market forces associated with corporate control. Without ERCs, management would be under pressure to use the firm’s assets in the most productive, value-enhancing way possible. The presence of these covenants reduces or eliminates this pressure, leading to an entrenched management and a higher level of agency costs of equity. This view is consistent with the managerial entrenchment literature (Shleifer and Vishny (1989)) and is also in line with the rationalizations for poison pill changes in corporate charters (Malatesta and Walkling (1988) and Ryngaert (1988)).(2)
If ERCs lead to an increase in agency costs of equity, then stock returns for firms issuing bonds with ERCs should be more negatively affected than stock returns for firms issuing bonds without ERCs, ceteris paribus.
From Standard & Poor’s December 1990 issue of CreditWeek, we identified a preliminary test sample of 145 bond issues offered by 84 firms as including ERCs. These protected bonds were issued during the 1982-1990 period. In order to compare the announcement effect of firms issuing ERC-protected bonds with firms issuing straight bonds without such covenants, a control sample of bonds without ERCs is constructed from all bonds publicly offered during the same 1982-1990 period. The preliminary control sample,as provided by Investment Dealers’ Digest (IDD), consists of 746 corporate bond issues offered by 268 companies.(3)
A variety of selection criteria for inclusion in the final samples were applied to both the preliminary protected and nonprotected bond samples:
1) A bond must be nonconvertible and have a fixed coupon rate.
2) The announcement date of a bond issue must be identified. Since we focus on the first day when the information regarding the bond issue was released to the market, we use as the announcement date the earlier of the SEC registration filing date of each bond issue, as provided by IDD, or the day when the firm announced its bond issue plan, as reported in The Wall Street Journal.(4)
3) A firm must not appear in both protected and nonprotected samples. If a firm issues both a protected bond and a nonprotected bond during this time frame, only the protected bond issue is included in the study.
4) If a firm registers two or more bond issues with the SEC on the same date, only one bond issue is included in the final sample (such instances are relatively rare).
5) An issuing firm’s stock must be traded on either the New York Stock Exchange or the American Stock Exchange, and at least 121 daily stock returns (from 90 days preceding to 30 days following the announcement date) must be available from the Center for Research in Security Prices (CRSP) Daily Stock Return File.
6) Other sufficient data, such as bond rating and coupon rate, must be available.
Of 145 issues in the preliminary protected bond sample, 39 and 23 were excluded on the basis of criteria #2 and #5, respectively. This leaves 83 issues in the final protected bond sample, of which 82 were offered by industrial companies and 1 by a utility. Of the 746 issues in the preliminary control sample, 498 were excluded according to criterion #2, 11 by criterion #3, 13 by criterion #4, 13 by criterion #5, and 68 by criterion #6. This leaves 143 issues in the final nonprotected bond sample, of which 105 were offered by industrial companies, 12 by utilities, and 26 by financial companies.(5)
Table 1 reports the distribution of sample firms relative to several bond characteristics. About 55% of both protected and nonprotected bonds have a maturity of ten years or less, while the remaining bonds have maturities ranging from 11 to 30 years. According to S&P’s credit ratings, slightly less than 60% of both protected and nonprotected bonds were rated A or above when they were issued, with the largest number of bonds rated in the A category for both samples.
Separate from its traditional credit rating, S&P ranks ERCs on a scale from E-1 to E-5 (E-1 representing the strongest protection). Of the bonds issued with ERCs, only ten bonds (12%) offered strong or significant protection against event risk; 40 bonds (48%) offered moderate protection; and the remaining 33 bonds (40%) offered either weak or insignificant protection against event risk.(6) Shelf registration issues represent about 82% and 72% of the protected and nonprotected bond samples, respectively.
Summary statistics for the sample bond issues are reported in Table 2. The average time to maturity from the announcement year is slightly less than 16 years for both protected and nonprotected bond issues. A typical protected or nonprotected bond is issued in the same year as the announcement, as evidenced by the time interval between announcement and issuance of less than a year. The average offering size of protected bonds is slightly above $150 million, which is significantly smaller than that of nonprotected bonds. A typical event risk-protected bond in the test sample has an S&P event risk ranking between E-3 and E-4, equivalent to moderate or weak protection quality, while the average bond in both samples has an S&P bond rating close to A-. Average asset size of firms issuing protected bonds is significantly smaller than for firms issuing nonprotected bonds.
Table 1. Distribution of Sample Nonprotected and Protected Bond
Issues Offered during the Period January 1982-December 1990
The samples consist of bond issues offered during the period January
1982-December 1990. Years to Maturity is the years from the
registration year to maturity year. Standard & Poor’s Bond Rating is
the bond rating as of the offering date. n.a. indicates “not
Nonprotected Bonds Protected Bonds
Classification (n = 143) (n = 83)
Years to Maturity
1-5 26 7
6-10 54 39
11-15 9 8
16-20 9 7
21-30 45 22
Standard & Poor’s Bond Rating
AAA 1 0
AA 31 8
A 52 39
BBB 44 32
BB 10 3
B 5 1
Event Risk Protection Ranking
E-1 n.a. 6
E-2 n.a. 4
E-3 n.a. 40
E-4 n.a. 12
E-5 n.a. 21
Shelf 103 68
Nonshelf 40 15
III. Analysis of Stock Return Response to Announcement of Bond Issuance
We test two competing hypotheses, the stockholder wealth enhancement hypothesis and the management entrenchment hypothesis, by comparing stock return effects associated with the announcement of protected bonds and the announcement of nonprotected bonds. We use the traditional event study methodology, estimating a set of specified market model parameters over a 120-day period, from 150 days to 31 days before the announcement date of event risk-protected and nonprotected bonds. Using these parameters in the market model, we estimate. the abnormal [TABULAR DATA FOR TABLE 2 OMITTED] returns (ARs) over a 61-day period, from 30 days before to 30 days after the announcement date.(7)
The announcement date of the bond issue in our sample represents the date the public first has knowledge of the issue (day t = 0). Because it is possible that the information regarding a bond issue may not be reflected in the market until the day following the announcement date, we examine the ARs over a two-day announcement period (day t = 0 and day t = +1).
Daily stock returns for the firms that issue bonds in this study are assumed to be generated by the model:[R.sub.it] = [[Alpha].sub.i] + [[Beta].sub.i] [R.sub.mt] + [[Epsilon].sub.it] (1)
where [R.sub.it] is the daily return for stock i on day t, and [R.sub.mt] is the daily return associated with the market portfolio. The market index employed here is the equally weighted index of stock returns (with dividends included) on the CRSP tape. Ex post ARs are generated by:
[Mathematical Expression Omitted]
where A[R.sub.it] represents the abnormal returns for stock i on day t.
The ARs are averaged across the securities in each sample for each day t to arrive at an average AR for day t, A[R.sub.t]. In addition, average cumulative abnormal returns (CARs) for each day from day t = -30 to day t = +30 are computed to study the trend in daily ARs. Both the traditional T-test and the nonparametric Wilcoxon signed-rank test are performed to test the significance of these measures.
The mean two-day announcement-period abnormal return (APAR) for the 226 bonds in the combined sample (not reported in a table) is -0.321% (the median = -0.178%), which is not statistically significant at the 0. 10 level, with a t-statistic of -1.56. These findings are consistent with those of previous studies that document that the issuance of nonconvertible debt has little, if any, effect on the common stock of the firm. Dann and Mikkelson (1984) find an insignificant negative return of -0.37% associated with the announcement of nonconvertible bond issues. Eckbo (1986) and Mikkelson and Partch (1986) provide similar evidence.
Table 3 shows the mean abnormal returns around the announcement date for protected and nonprotected bonds. While the average two-day APAR for nonprotected bonds is a negative 0.588%, the average APAR for protected bonds is a positive 0.138%. The difference in the average APARs between the two samples is significantly different from zero at the 0.05 level, with a t-statistic of -2.134. As further support, the nonparametric Wilcoxon signed-rank test also indicates a significant difference in the APARs between the two samples at the 0.05 level, with a Z-statistic of 2.028 (median two-day APAR of -0.487% for nonprotected and 0.104% for protected bonds, not reported in Table 3).(8)
An examination of daily average ARs during the announcement period shows that the ARs for nonprotected bonds at days t = 0 and t = +1 are negative, with the AR at day t = +1 statistically significant at the 0.01 level. In addition, 88 of the 143 firms in the sample (61.5%) register negative daily ARs at day t = +1.
For protected bonds, a different set of results are obtained. Although the average AR at day t = 0 is negative, the average AR at day t = +1 is significantly positive and more than makes up for the negative average AR at day t = 0. For this sample, only 34 of the 83 firms (41%) register negative daily ARs at day t = +1.
An overview of the trends in the CARs for the two samples appears in Figure 1. While nonprotected bonds show a noticeable negative drift in CARs from day t = -10 through day t = +10, protected bonds show a noticeable positive drift during the same period.
Overall, the stock price responses indicate that the issuance of bonds with ERCs in a bond indenture has, when compared to the issuance of nonprotected bonds, a more positive impact on stockholder returns. Consequently, the results provide evidence that supports the stockholder wealth enhancement hypothesis but does not support the management entrenchment hypothesis.(9)
IV. Analysis of the Impact of Agency Costs of Debt and Equity on Stockholder Wealth
Our results support the notion that the issuance of bonds with ERCs allows stockholders to realize a net benefit when the reduction of agency costs of debt is greater than an increase in agency costs of equity. The effect of ERCs, however, may not be equal across all firms because of differences in the magnitude of the agency costs of debt and equity and other firm and bond characteristics.
A. Measures of Agency Costs of Debt
Two measures of the agency costs of debt are discussed here: bond callability and a firm’s cumulative profitability.(10)
As previous studies indicate, noncallable debt creates externalities for stockholders, particularly for a firm that is in financial distress (see Barnea, Haugen, and Senbet (1980), Bodie and Taggart (1978), and Thatcher (1985)). Under these circumstances, bondholders will realize a portion of the residual value in the firm’s assets, given a change in the company’s fortunes. If stockholders know that some of the value from their investments will flow to bondholders, stockholders will choose to invest suboptimally (Myers (1977)). With a call provision, however, stockholders can renegotiate the old debt with new debt that fully reflects the value of the new investment opportunity; thus, they will be more inclined to invest. The presence of a call provision, therefore, can lower the agency costs of debt by mitigating the suboptimal investment problem associated with debt, leading to greater firm value.
The reduction in agency costs of debt should be even greater when an ERC is included along with a call provision. Accordingly, it is hypothesized that the stock response will [TABULAR DATA FOR TABLE 3 OMITTED] be more positive upon issue of a bond with both an ERC and a call provision.
The second measure of the agency costs of debt is a firm’s cumulative profitability, CP, measured as retained earnings divided by total assets. Cumulative profitability has been widely used in discriminant analysis studies to identify the default risk associated with a firm (see Altman, Haldeman, and Narayanan (1977)). Thatcher (1985) points out that a firm with lower CP has higher bankruptcy risk. Accordingly, the issuance of debt by a firm with low CP would increase the agency costs of debt and, therefore, lower the value of the firm. Because inclusion of an ERC in the debt contract would help lower agency costs of debt, a firm with low CP would benefit more by issuing debt with an ERC.
B. Measures of Agency Costs of Equity
For agency costs of equity, we use two measures frequently used in the finance literature: free cash flow and insider ownership. We measure free cash flow for the year immediately preceding the announcement year following Lehn and Poulsen (1989). This variable is defined as
FCF = INC – TAX – INTEXP – PFDDIV – COMDIV (3)
INC = operating income before depreciation (COMPUSTAT item #13)
TAX = total income taxes (COMPUSTAT item #6) minus change in deferred taxes from the previous to the current year (Change in COMPUSTAT item #35)
INTEXP = gross interest expense on short- and long-term debt (COMPUSTAT item #15)
PFDDIV = total amount of preferred dividend requirement on cumulative preferred stock and dividends paid on noncumulative preferred stock (COMPUSTAT item #19)
COMDIV = total dollar amount of dividends declared on common stock (COMPUSTAT item #21)
FCF measures post-tax cash flow that was not distributed to security holders as either interest or dividend payments. In our analysis, we express FCF as a percentage of the value of a firm’s total assets to remove any size effects. Jensen (1986) hypothesizes that firms with greater amounts of free cash flow have potentially greater agency costs of equity, as managers can use the free cash flow in ways that could reduce stockholder wealth. Jensen argues further that debt can be used to reduce the free cash flow and, hence, agency costs of equity, since debt requires the disbursement of cash flows in the form of principal and interest. We thus expect a firm with a higher level of free cash flow to benefit more by issuing debt than a firm with less free cash flow. Hence, a positive relationship between APAR and free cash flow is expected.
The presence of ERCs would, however, increase agency costs of equity (Bicksler and Chen (1991), Kahan and Klausner (1993), Smith and Warner (1979), and Thatcher (1985)). Therefore, a firm that issues protected bonds is expected to benefit less than a firm that issues nonprotected bonds. Consequently, we expect a greater negative impact on stockholder return if a firm with a higher level of free cash flow issues a protected bond.
The second measure of the agency costs of equity is the proportion of insider ownership (INS). This variable is measured by the number of shares owned by managers and directors divided by the total number of shares outstanding. For this variable, we obtain the information on insider ownership from various issues of Spectrum VI during the period examined in our study. Previous studies have shown that the lower the proportion of insider ownership, the greater the agency costs of equity (see Haugen and Senbet (1981), Jensen and Meckling (1976), and Lehn and Poulsen (1989)). Since the inclusion of ERCs may increase the agency costs of equity further, a firm with lower INS will benefit less by issuing protected bonds than by issuing nonprotected bonds.
C. Regression Model
Using regression analysis, we examine whether the increase in stockholder wealth associated with the issuance of protected bonds is attributable to a reduction in agency costs of debt, given measures of agency costs of equity and other firm and bond characteristics. The variables measuring firm and bond characteristics are well documented (e.g., Eckbo (1986) and Mikkelson and Partch (1986)) and are only briefly discussed here. A firm’s APAR is assumed to be a function of eight variables:
APA[R.sub.i] = [[Beta].sub.0] + [[Beta].sub.1]ln[(SIZE).sub.i] + [[Beta].sub.2][DEBT.sub.i]
+ [[Beta].sub.3][RANKING.sub.i] (or COV) + [[Beta].sub.4][IND.sub.i]
+ [[Beta].sub.5][ACEQ.sub.i] + [[Beta].sub.6][ACEQ.sub.i] x COV + [[Beta].sub.7][ACDEBT.sub.i]
+ [[Beta].sub.8][ACDEBT.sub.i] x COV + [e.sub.i] (4)
APA[R.sub.i] represents the two-day announcement-period abnormal return for firm i. Combining the two samples results in a total of 226 observations.
SIZE represents the dollar size of the bond issue in natural log form. All else held equal, a larger issue will result in larger cash outflows from the firm as payments to bondholders. Therefore, a negative relationship between SIZE and APAR is expected.(11)
DEBT represents the degree of a firm’s leverage and is measured by the ratio of total debt to total assets for the year prior to the announcement date. It is expected that the greater the degree of leverage, the more risk to stockholders. Thus, a negative relationship between APAR and DEBT is expected. It is also plausible that a change in the value of a firm may be induced by a change in the firm’s capital structure, regardless of the presence of ERCs in the bond indenture. In this regard, DEBT will effectively control the effect of such a capital structure change on the value of the firm.
IND is a zero-one dummy variable representing the industry classification of issuing firms. This variable is included in the regression to control for different industry characteristics. Since financial firms and utilities are perceived to be less exposed to event risk because of regulatory restrictions (e.g., on leverage), the announcement-period returns may be influenced by different industry characteristics. IND has a value of 1 for industrial firms and 0 for financial and utility firms.(12)
Test Variables for ERCs
COV is a zero-one dummy variable that accounts for the presence of ERCs, with one for protected bonds and zero for nonprotected bonds. Although we have found the presence of ERCs to be positively related to APARs, it is worthwhile to examine this relationship in the regression context. As the issuance of bonds with ERCs increases stockholder wealth relative to bonds without ERCs, we expect a positive relationship between COV and APAR.
RANKING represents the quality of ERCs as evaluated by S&P. Values of 1 through 5 are assigned to the S&P E-1 through E-5 rankings. An additional category with a value of 6 is included to represent bonds without ERCs. Since ERCs with stronger protection are expected to have a more positive impact on firm value, a negative relationship between APAR and RANKING is hypothesized.
Since COV and RANKING are highly correlated with other variables in the regression (particularly with the interactive variables discussed below), these two control variables are dropped from multiple regressions and run individually in a simple regression.
Test Variables for Agency Costs of Equity
ACEQ measures the agency costs of equity present within a firm at the time of the bond issue announcement. We expect a positive relationship between free cash flow to total assets, FCF, and APAR and a negative relationship between degree of insider ownership, INS, and APAR.
An interactive variable, ACEQ x COV (FCF x COV and INS x COV, respectively), is included in the regression to test whether a firm with greater agency costs of equity will benefit to a smaller degree by issuing protected bonds than by issuing nonprotected bonds. Since the variables FCF and INS may not measure the relationship between the agency costs of equity and APAR associated with the issuance of debt with or without ERCs, the interactive variable must be included in the regressions. As discussed previously, firms with higher FCF (lower INS) that issue protected bonds are expected to benefit less than firms that issue nonprotected bonds. Hence, we expect a negative relationship between FCF x COV and APAR and a positive relationship between INS x COV and APAR.
Test Variables for Agency Costs of Debt
ACDEBT measures the agency costs of debt present within a firm at the time of the bond issue announcement. Two measures of the agency costs of debt, bond callability, CALL, and cumulative profitability, CP, are used.(13) A positive relationship between CALL and APAR and between CP and APAR is expected.
As in the analysis for the agency costs of equity, an interactive variable of ACDEBT x COV (CALL x COV and CP x COV, respectively) is included to examine whether a firm with greater agency costs of debt will benefit more by issuing protected bonds than by issuing nonprotected bonds. Firms issuing debt with a call provision or firms with a lower CP that issue protected bonds are expected to benefit more than firms that issue nonprotected bonds. Therefore, a positive relationship is expected between CALL x COV and APAR, and a negative relationship is expected between CP x COV and APAR.
A summary of definitions and measures of variables and their expected relationships to APAR is presented in Table 4.
[TABULAR DATA FOR TABLE 4 OMITTED]
D. Empirical Results
The regression results are presented in Table 5. Six separate regressions were performed. Regressions 1 and 2 take COV and RANKING as the sole independent variable to measure the effect of ERCs on APAR. Regressions 3 and 4 take FCF and INS as the measure of agency costs of equity. Since FCF and INS are proxies for the same measure of the agency costs of equity, each of these two measures enters Regressions 3 and 4, respectively, with two measures of agency costs of debt (CALL and CP) and other control variables. Finally, Regressions 5 and 6 take CALL and CP separately in the regressions with FCF and other control variables.(14) We perform separate Regressions 3 through 6 in order to avoid any confounding influences among the measures of the agency costs of debt and equity.
For all the regressions, the estimated coefficients for the two control variables – ln(SIZE) and DEBT – have the expected signs, although not significant at the 0.05 level. In addition, the estimated coefficient for IND is not significant at the 0.05 level. As shown in Regressions 1 and 2, both COV and RANKING have the expected signs and estimated [TABULAR DATA FOR TABLE 5 OMITTED] coefficients significant at the 0.05 level. The magnitude (0.7%) and t-statistic (2.16) for the estimated coefficient for COV are consistent with the results associated with the two-day APAR reported in Table 3. This further supports the notion that the presence of ERCs in debt contracts enhances stockholder wealth. RANKING measures the effect of the strength of ERCs on APAR. The significantly negative regression coefficient of -0.003 (t-statistic = -2.33) for RANKING indicates that stronger protection (that is, lower ERC ranking) is associated with greater APAR.
For the test variables, the regression results generally support the positive impact of ERCs on stockholder wealth through a reduction in the agency costs of debt. The estimated coefficients for the two noninteractive variables proxing agency costs of debt, CALL and CP, show inconsistent results; only CP has the expected positive signs in all regressions. However, none of these regression coefficients for CALL and CP is significant at the 0.05 level.
More important, the estimated coefficients for the interactive variables, CALL x COV and CP x COV, are of the expected signs: positive for CALL x COV and negative for CP x COV. Furthermore, the coefficient for CALL x COV is significantly different from zero at the 0.05 level in all regressions. Since these two interactive variables measure the differential effect of the agency costs of debt on stockholder wealth for protected bonds over nonprotected bonds, the significantly positive relationship between CALL x COV and APAR strongly suggests that the presence of ERCs provides a net positive benefit to stockholders by further lowering the agency costs of debt. Thus, a firm with higher agency costs of debt will benefit more by issuing protected bonds.
The regression results associated with the agency costs of equity provide further evidence that issuing protected bonds does not increase the agency costs of equity and, hence, does not affect stockholder wealth.
As shown in Regressions 3 through 6, the estimated regression coefficients for FCF have the positive signs as expected and are significantly different from zero, at least at the 0.10 level. The estimated coefficient of INS, however, is not significant. This result indicates that the higher the FCF, the more positive the impact on stockholder wealth. The significant positive relationship between FCF and APAR provides evidence that issuance of debt by a firm will lead to a reduction in the agency costs of equity. This finding is consistent with Jensen’s (1986) free cash flow hypothesis.
The estimated coefficients for the interactive variables for the agency costs of equity, FCF x COV and INS x COV, have mixed signs and are not significant at the 0.10 level. These results indicate that there is no strong relationship between the presence of ERCs and the agency costs of equity of an issuing firm. Thus, the issuance of protected bonds has neither cost nor benefit for stockholder wealth of a firm with higher agency costs of equity.
The overall results of the regression analysis confirm a strong positive relationship between the presence of ERCs and stockholder wealth. The results further indicate that the increase in stockholder wealth is primarily attributable to a significant reduction in the agency costs of debt by the issuance of protected bonds.(15)
V. Summary and Conclusions
Two different rationales may explain why a firm includes event risk covenants in the bond indenture. On the one hand, these covenants could offer an effective mechanism to protect bondholders from wealth transfers that may result from debt-financed takeovers and recapitalizations; they would then reduce the agency costs of debt associated mainly with the risk incentive problem between stockholders and bondholders. On the other hand, event risk covenants could act as a de facto deterrent for an otherwise more efficient transfer of corporate control, thereby increasing the agency costs of equity. We test these two competing rationales and examine stockholder wealth effects associated with announcements of bond issues that include ERCs.
We find that, on average, firms issuing protected bonds experience significantly more positive abnormal returns around the announcement of bond issuance than firms issuing nonprotected bonds. This finding provides evidence that the issuance of protected bonds has a more positive impact on stockholder returns than the issuance of nonprotected bonds does.
We also examine the relative impact of the agency costs of debt and equity on stockholder wealth. Regression results confirm evidence of a strong positive relationship between the presence of ERCs and stockholder wealth. They further indicate that higher announcement-period abnormal returns are associated with firms characterized by greater agency costs of debt that issue bonds with event risk covenants. These results support the notion that the increase in stockholder wealth is primarily attributable to a significant reduction in the agency costs of debt resulting from the issuance of protected bonds.
The results reported in this paper suggest that not all firms would benefit from issuing protected bonds. First, the magnitude of a firm’s agency costs of debt and equity appears to govern the issuance of protected or nonprotected bonds. That is, firms with relatively greater agency costs of debt stand to benefit from issuing protected bonds, while firms with relatively greater agency costs of equity have no incentive to issue such bonds. Second, firms issuing protected bonds are significantly smaller than firms issuing nonprotected bonds. Hence, if size proxies for a firm’s reputation in the market and for degree of financial distress, relatively smaller firms may be required to provide additional protection in the form of ERCs that would not be required for relatively larger firms.
An interesting avenue for future research would be examination of bond price reactions to the announcement of protected and nonprotected bonds. The results would shed additional empirical light on the impact of event risk covenants on the value of the firm.
1 Several studies suggest that agency problems can be reduced or eliminated through the use of managerial incentive schemes and/or more complicated financial securities, such as convertible bonds and warrants (see Barnea, Haugen, and Senbet (1985), and Green (1984)). See also Emery and Finnerty (1992) for a review of research on corporate debt provisions.
2 Note that there is great variation among ERCs. Some ERCs are triggered when a takeover event is accompanied by a rating decline. Such covenants afford some protection to bondholders and thus may reduce the agency costs of debt. At the same time, these covenants may also entrench management, as takeover events may very well accompany rating declines without triggering ERCs (e.g., a proxy contest initiated because of poor performance).
3 The original IDD sample consists of 19,000 U.S. corporate public debt issues, brought to market during the January 1982 – December 1990 period. We excluded 10,086 mortgage-related issues and 7,983 issues without identifiable CUSIPs. Also excluded are 185 issues having a maturity date before July 13, 1991.
4 Norton and Pettengill (1991) use the announcement of event risk ranking by S&P as the relevant event day. In fact, information regarding bond issue characteristics, especially the inclusion of ERCs in the bond indenture, is first available to the market on the day the firm announces the bond offering, not on the day S&P announces the ERC ranking.
5 Regulated companies, such as financial companies and utilities, are perceived to be less exposed to event risk because of regulatory restrictions on leverage. Inclusion of such firms in higher proportion in the control sample may bias the estimated abnormal returns associated with the announcement period. Analysis that excludes these regulated companies from both samples produces results qualitatively similar to those reported using the full sample of firms.
6 By S&P ERC rankings, for a bond with an E-3 ranking, “protection may not be provided against some anticipated events, or the effectiveness of protective provisions is questionable, or the benefits of protection are modest.” (S&P’s CreditWeek, December 24, 1990, p. 29)
7 We also estimate the abnormal returns using both the market model with a value-weighted market index and the mean adjusted return model. The results are qualitatively the same as those reported here.
8 ERC-protected bonds are very rare prior to 1986. Consequently, it is possible that the stock price response results reported here may be affected by the timing distribution of bonds in the two samples, In particular, 30% of the nonprotected bonds (43 of 143) were issued in the 1982-1985 period, while only 4% of the protected bonds (3 of 83) were issued in this period. In order to test the robustness of our empirical findings, stock price responses associated with the announcements of bond issues are re-examined using only the sample bonds for the 1986-1990 period, with qualitatively similar results to those reported here.
9 The inclusion of ERCs conceivably may signal a probability that a particular firm is a takeover target. Hence, the more positive abnormal returns associated with the announcement of ERCs may be a result of sending such a signal to the market. The more positive abnormal returns may also be due to the relative change in capital structure between the two samples. We compared the debt ratios of the two samples for the year before and after the debt announcement assuming that the new debt is used either to acquire additional assets (scenario 1) or to buy back equity (scenario 2). The results showed that the debt ratio increases from 46.3% to 48.4% or 50.7% for the control sample, and from 47.0% to 49.9% or 53.1% for the test sample under scenarios 1 and 2, respectively. The differences in debt ratios between the two samples are not significant either for the pre-announcement period (t-statistic = -0.30) or for the post-announcement period (t-statistics = -0.86 and -1.17 under each scenario). Hence, the observed significant difference in announcement-period abnormal returns does not appear to be associated with changes in capital structure for the firms in the two samples.
10 We also use the bond rating as a proxy for agency costs of debt in the regression analysis. We do not report the results here because of the high correlation of bond rating and other variables in the regression. The estimated regression coefficients for the bond rating variable and its interactive term with COV are insignificant, and inclusion of these variables does not materially affect the estimated coefficients of other variables.
11 Krasker (1986) demonstrates that an adverse selection problem is associated with the issuance of risky securities, debt or equity, suggesting that larger issues are viewed more unfavorably than smaller issues. His model predicts a negative relationship between the stock price reaction to the issuance of risky securities and issue size.
12 We also ran regressions using two zero-one industry dummy variables accounting for three industry classifications of industrial, financial, and utility companies. The overall regression results remain the same with insignificant regression coefficients for both industry dummy variables.
13 The proportion of bonds with a call provision is 59% (84 out of 143) and 35% (29 out of 83) for nonprotected and protected bonds, respectively. The average call protection period is 8.2 years and 11.3 years for nonprotected and protected bonds, respectively.
14 Substituting INS for FCF in Regressions 5 and 6 as a variable for the agency costs of equity results in qualitatively similar results for estimated regression coefficients, with slightly lower adjusted [R.sup.2] and F-values.
15 We also performed weighted least squares (WLS) regressions using the standard error of the estimation period residuals as the weighting factor to remove any possible heteroscedasticity associated with APAR. This analysis yields signs and significance levels for the variables similar to those reported in Table 5.
Allen, D.S., R.E. Lamy, and G.R. Thompson, 1987, “Agency Costs and Alternative Call Provisions: An Empirical Investigation,” Financial Management (Winter), 37-44.
Altman, I., R.G. Haldeman, and P. Narayanan, 1977, “Zeta Analysis: A New Model to Identify Bankruptcy Risk of Corporations,” Journal of Banking and Finance (June), 29-51.
Asquith, P. and T.A. Wizman, 1990, “Event Risk, Covenants, and Bondholder Returns in Leveraged Buyouts,” Journal of Financial Economics (July), 195-213.
Bae, S.C., H. Levy, and P.J. Wannermacher, 1991, “An Empirical Analysis of Event Risk Protection Provisions on Corporate Bonds: Some Early Evidence,” University of Florida Unpublished Manuscript.
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 (December), 1223-1234.
Barnea, A., R.A. Haugen, and L.W. Senbet, 1985, Agency Problems and Financial Contracting, Englewood Cliffs, NJ, Prentice-Hall.
Bicksler, J.L. and A.H. Chen, 1991, “The Valuation of Risky Debt with Event Risk Provisions,” Pacific-Basin Markets Research (March), 21-32.
Bodie, Z. and R.A. Taggart, Jr., 1978, “Future Investment Opportunities and the Value of the Call Provision on a Bond,” Journal of Finance (September), 1187-1200.
Crabbe, L., 1991, “Event Risk: An Analysis of Losses to Bondholders and Super Poison Put Bond Covenants,” Journal of Finance (June), 689-706.
Dann, L.Y. and W.H. Mikkelson, 1984, “Convertible Debt Issuance, Capital Structure Change and Financing-Related Information: Some New Evidence,” Journal of Financial Economics (Jane), 157-186.
Eckbo, B.E., 1986, “Valuation Effects of Corporate Debt Offerings,” Journal of Financial Economics (January/February), 119-151.
Emery, D.R. and J.D. Finnerty, 1992, “A Review of Recent Research Concerning Corporate Debt Provisions,” Financial Markets, Institutions & Instruments (December), 23-39.
Fields, J.A., D.S. Kidwell, and L.S. Klein, 1991, “RJR/Nabisco and Event Risk Protection,” University of Connecticut Unpublished Manuscript.
Galai, D. and R.W. Masulis, 1976, “The Option Pricing Model and the Risk Factor of Stock,” Journal of Financial Economics (January/March), 53-81.
Green, R.C., 1984, “Investment Incentives, Debt, and Warrants,” Journal of Financial Economics (March), 115-136.
Green, R.C. and E. Talmor, 1986, “Asset Substitution and the Agency Costs of Debt Financing,” Journal of Banking and Finance (October), 391-400.
Harris, M. and A. Raviv, 1990, “Capital Structure and the Informational Role of Debt,” Journal of Finance (June), 321-349.
Haugen, R.A. and L.W. Senbet, 1981, “Resolving the Agency Problems of External Capital Through Options,” Journal of Finance (June), 629-647.
Haugen, R.A. and L.W. Senbet, 1988, “Bankruptcy and Agency Costs: Their Significance to the Theory of Optimal Capital Structure,” Journal of Financial and Quantitative Analysis (March), 27-38.
Jensen, M.C., 1986, “Agency Costs of Free Cash Flow, Corporate Finance, and the Market for Takeovers,” American Economic Review (May), 323-329.
Jensen, M.C. and W. Meekling, 1976, “Theory of the Firm: Managerial Behavior, Agency Costs, and Capital Structure,” Journal of Financial Economics (October), 305-360.
Kahan, M. and M. Klausner, 1993, “Antitakeover Provisions in Bonds: Bondholder Protection or Management Entrenchment?” UCLA Law Review (April), 931-982.
Krasker, W.S., 1986, “Stock Price Movements in Response to Stock Issues Under Asymmetric Information,” Journal of Finance (March), 93-105.
Lehn, K. and J. Poulsen, 1989, “Free Cash Flow and Stockholder Gains in Going Private Transactions,” Journal of Finance (July), 771-788.
Malatesta, P. and R. Walkling, 1988, “Poison Pill Securities: Stockholder Wealth, Profitability, and Ownership Structure,” Journal of Financial Economics (January/March), 347-376.
Mikkelson, W.H. and M.M. Partch, 1986, “Valuation Effects of Security Offerings and the Issuance Process,” Journal of Financial Economics (January/February), 31-60.
Miller, M.H. and K. Rock, 1985, “Dividend Policy Under Asymmetric Information,” Journal of Finance (September), 1031-1051.
Myers, S.C., 1977, “Determinants of Corporate Borrowing,” Journal of Financial Economics (March), 147-175.
Myers, S.C., 1984, “The Capital Structure Puzzle,” Journal of Finance (July), 575-592.
Myers, S.C. and N.S. Majluf, 1984, “Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have,” Journal of Financial Economics (June), 187-221.
Norton, E. and G. Pettengill, 1991, “The Impact of Event Risk Covenant Ranking Announcements on Shareholder Wealth,” Fairleigh Dickinson University Unpublished Manuscript.
Ramakrishnan, R.T.S. and A.V. Thakor, 1982, “Moral Hazard, Agency Costs, and Asset Prices in a Competitive Equilibrium,” Journal of Financial and Quantitative Analysis (November), 503-532.
Ryngaert, M.D., 1988, “The Effect of Poison Pill Securities on Shareholder Wealth,” Journal of Financial Economics (January/March), 377-417.
Shleifer, A. and R.W. Vishny, 1989, “Management Entrenchment: The Case of Manager-Specific Investments,” Journal of Financial Economics (June), 123-140.
Smith, C.W., Jr. and J.B. Warner, 1979, “On Financial Contracting: An Analysis of Bond Covenants,” Journal of Financial Economics (June), 117-162.
Stulz, R., 1990, “Managerial Discretion and Optimal Financing Policies,” Journal of Financial Economics (July), 3-28.
Thatcher, J.S., 1985, “The Choice of Call Provision Terms: Evidence of the Existence of Agency Costs of Debt,” Journal of Finance (June), 549-561.
Warga, A. and I. Welch, 1990, “Bondholder Losses in Leveraged Buyouts,” University of California Unpublished Manuscript.
Sung C. Bae, Daniel P. Klein, and Raj Padmaraj are all Associate Professors of Finance at Bowling Green State University, Bowling Green, Ohio.
COPYRIGHT 1994 Financial Management Association
COPYRIGHT 2004 Gale Group