CAN AGENCY COSTS OF DEBT BE REDUCED WITHOUT EXPLICIT PROTECTIVE COVENANTS? THE CASE OF RESTRICTION ON THE SALE AND LEASE-BACK ARRANGEMENT

Transcription

1 CAN AGENCY COSTS OF DEBT BE REDUCED WITHOUT EXPLICIT PROTECTIVE COVENANTS? THE CASE OF RESTRICTION ON THE SALE AND LEASE-BACK ARRANGEMENT Jung, Minje University of Central Oklahoma Ellis, R. Barry University of Central Oklahoma ABSTRACT We explore the determination of debt contracting by testing how the probability of explicit contracting is related to amount to be borrowed and bond feature variables such as maturity, coupon, and bond ratings. Then, we examine the relationship between agency costs and contracting by comparing between two groups of bond: explicit restriction vs. implicit restriction on the sale of asset and/or sale and leaseback transaction. The probability of explicit restriction in place-is positively related to the amount of capital to be raised in support of the hypothesized relationship: the larger the amount of borrowing, the more likely an explicit restriction is to be included in debt contract. The negative coefficient for bond ratings is statistically significant in strong support of the hypothesized negative relationship between the probability of explicit contracting and bond ratings. The larger the agency costs of debt (that is, the lower the bond ratings), the more likely explicit restrictions are in place. Agency costs of debt (for which bond ratings are used as a proxy) are found to be positively related to the amount of issue and maturity, and negatively related to coupon rate and contracting variable. These positive relationships are interpreted as the size of capital to be raised by bond issue tends to be larger for the borrowing firms with lower agency costs of debt; and the longer the maturity, the higher the bond ratings, i.e. the lower agency costs of debt. INTRODUCTION Conflicts of interest are at the core of the agency problem between bondholders and stockholders of a firm. Stockholders can avoid debt financing while expropriating bondholders wealth through a sale and lease-back transaction. With this type of financial lease, a physical asset owned by a firm is sold, but leased back simultaneously. There is no interruption of use of the asset for regular business operations. Ownership of the asset, however, is transferred to the lessor. Bondholders wealth is jeopardized if stockholders invest the proceeds of the sale of the asset in higher risk-higher return projects. Stockholders would gain at the expense of bondholders. Stockholders will benefit if the new projects are successful. On the other hand, bondholders face a higher-than-originally-expected risk of default. Being wary of this possible loss in wealth as the value of their bonds decline, bondholders often require that a protective covenant be included in bond indenture explicitly restricting the sale and

2 lease-back transaction on a particular asset of the borrowing firm. The first role of a protective covenant is to prevent managers and shareholders from taking value-reducing actions that could be privately optimal because they expropriate bondholders (Tirole, 2006). The purpose of this research is twofold. First, we explore the determination of debt contracting by testing how the probability of explicit contracting is related to the amount to be borrowed and bond feature variables such as maturity, coupon, and bond ratings. Then, we examine the relationship between agency costs and contracting by comparing between two groups of bond: explicit restriction versus implicit restriction on the sale of asset and/or sale and leaseback transaction. The bond rating is used as a proxy for agency costs associated with bondholderstockholder conflict of interest. Two groups of sample bonds registered with the SEC in 2006 and 2007 are compared to test the hypothesis that there is no significant difference in agency costs between the explicit restriction group and the implicit restriction only group. LITERATURE REVIEW Conflicts of interest between stockholders and bondholders should exist in the system of corporation because of its unique contracting features. Bondholders bear default risk while stockholders have residual claims and limited liability in case of default. By making the asset and capital structure riskier, stockholders can reap most of the profit from high risk-high return projects if they are successful; yet stockholders can truncate loss in case of failure due to limited liability The higher risk of bankruptcy and increased financial distress due to stockholders limited liability characterize the unresolved stockholder-bondholder agency problem. Jensen and Meckling (1976) argue that bondholders are well aware of the possible expropriation of their wealth by stockholders, and discount the value of a bond at the time of issue for the expected losses that can result from stockholders effort to maximize their wealth at the expense of bondholders. Stockholders have an incentive to add a self-imposed protective covenant in the bond indenture designed to restrict this type of wealth transfer; however, the extent of protective covenants is limited by the contracting costs of monitoring and enforcing these covenants. In a nutshell, the stockholder-bondholder agency problem can be mitigated but not eliminated by contracting technology. Galai and Maulis (1976) analyze the stockholder-bondholder agency problem in an option pricing framework. The stockholders of a levered firm are equivalent to call option holders who can gain by increasing the risk of firm assets at the expense of bondholders who are equivalent to call writers. Bhojraj and Sengupta (2003) show corporate governance mechanisms are linked to higher bond ratings and lower bond yields. Using probit regression, they find that governance mechanisms can reduce default risk by mitigating agency costs and monitoring managerial performance and by reducing information asymmetry between the firm and the lenders. Bodie and Taggart (1978) as well as Thatcher (1985) show how inclusion of call provisions in bond indenture can reduce bondholder-stockholder agency costs. Low-risk profitable projects, despite their positive NPV, may be rejected by stockholders because these projects benefit primarily bondholders. A call provision, however, allows stockholders to limit the profit for bondholders to the call price, and the remaining profit is sufficient to induce shareholders to accept low-risk profitable projects. Also, a call provision makes stockholders less inclined to expropriate bondholders wealth by undertaking high-risk, negative NPV projects because the value of the call feature falls as firm value declines due to acceptance of high-risk, unprofitable projects. In sum, a call provision reduces bondholder-stockholder agency costs by rescuing low-

3 risk, positive NPV projects from stockholders rejection and by deterring stockholders from accepting high-risk, negative NPV projects just to maximize their wealth, not the total firm value. DATA & VARIABLES All data is compiled directly from the Securities and Exchange Commissions EDGAR database which posts the prospectus of all recent and past long term debt issues from each firm studied. Bond indentures of 204 sampled bond issues, registered with SEC in 2006 and 2007 by the Dow Jones thirty blue chip industrial firms, are examined for the trustee-stipulated explicit prohibition on the disposition of assets through the sale of assets and/or a sale and leaseback agreement. Other bond features such as amount raised, maturity, coupon rate, and bond rating are collected for empirical hypothesis testing. Table 1: Descriptive statistics of variables (all bonds) statistics amount($) maturity(years) coupon ratings Mean Median Mode Range Minimum Maximum # of bonds The contractual feature of whether or not the bond indenture has an explicit restriction is quantified by using an indicator variable: 1 for explicit restriction and 0 for implicit (i.e. no) restriction. Compared to 35.6 percent of sample bonds in the study of Smith and Warner (1976), 41.7 percent (85 out of 204) of bonds in our study are found to have explicit constraint on the firm s disposition of assets through either the sale of asset or a sale and lease-back agreement. Bond ratings are proxies for agency costs that are highly correlated with default risk (Crabbe and Helwege, 1988). Our basic conjecture is that bond ratings are inversely related to the agency costs of debt, which, in turn, are directly related to the probability of including restrictive covenants to protect bondholders interest. It is conjectured that the implicit subsample has low enough default risk that bondholders have sufficient trust in issuers to behave prudently to not require explicit restrictions. Thus, it is hypothesized that our dependent variable (probability of including restriction on sale and lease-back) is negatively related to bond ratings. All the sample bonds are investment-grade at the time of issue. For bond ratings, the following conversion scale is used to assign a numerical value to each rating: AAA+ = 12, AAA = 11, AAA- = 10, AA+ = 9, AA = 8, AA- = 7, A+ = 6, A = 5, A- = 4, BBB+ = 3, BBB = 2, BBB- = 1. The sampled bonds in our study have an average bond rating of 6.28, approximately equivalent to A+. Descriptive statistics for the two subsamples are given in Tables 2 and 3. The explicit group has an average of 6.06 and the implicit group has an average of This comparison is consistent with the conjecture that the lower default risk of the implicit group contributes to higher bond ratings than that of explicit group of bonds. The total amount of capital to be raised through a bond issue should be a significant contributor to the agency costs of debt. The bond issue will increase the total amount of debt, and with more

4 debt in capital structure, Jensen and Meckling (1976) argue that shareholders have more incentive to alter investment projects so that they can expropriate bondholders wealth. Explicit restrictions on asset disposition reduce the default risk of issuing firms. Thus, it is hypothesized that the larger the amount borrowed, the more likely an explicit restriction is to be included in debt contract. For the sampled bonds, the average amount of capital to be borrowed by issuing bonds is $679 million dollars ($303 million for implicit contracting bonds and $1,210 million for explicit contracting group.) As posited, on average, bond issues with explicit restriction are associated with a larger amount of capital to be raised than bond issues with an implicit only restriction. Table 2: Descriptive statistics of variables (implicit group) statistics amount($) maturity(years) coupon ratings Mean 3.03E Median Mode 7.5E Range 3E Minimum Maximum 3E # of bonds Table 3: Descriptive statistics of variables (explicit group) statistics amount($) maturity(years) coupon ratings Mean 1.21E Median 9E Mode 7.5E Range 4.73E Minimum Maximum 4.75E # of bonds Maturity should be a key variable explaining the inclusion of explicit restrictive covenants in the bond indenture. The model of Barnea et al. (1980) suggests that long-term securities provide less discipline on borrowing firms. Hence, long-term bonds should carry higher agency costs. It is predicted, therefore, that the probability of including explicit restriction on asset disposition increases as the maturity of bonds increases. The average maturity of sampled bonds is years. However the difference in average maturities the two subsamples is opposite to our prediction (14.72 years for the explicit group and years for the implicit group). Finally, the coupon rate should help explain the inclusion of restrictive covenants. A high coupon rate should be negatively related to the probability of explicit restriction due to the accelerated cash flow to bondholders. Then, it is expected that the higher the coupon rate, the lower the probability of including explicit restrictions on disposition of assets. Consistent with this conjecture, the average coupon rate for the total sample is 5.85 percent, 5.71 percent for the explicit group and 5.94 percent for the implicit group.

5 EMPIRICAL MODELS AND TEST RESULTS 1. Determination of contracting Inclusion of covenants in the debt contract is rationalized (Tirole, 2006) by its ability to the reduce the agency problem (Jensen and Meckling 1976; Smith and Warner 1979). The dependent variable of this study, presence of restriction on sale of assets or sale and lease-back agreement, depends upon the following independent variables: amount raised, maturity, coupon rate, and degree of agency problems as indicated by bond ratings. In our paper, the dependent variable has two possible outcomes, and therefore can be represented by an indicator variable taking on values 0 and 1. Sampled bonds are classified dichotomously into two groups. The first group consists of bonds that have an explicit restriction on the sale of assets and/or sale and lease-back agreement in bond indenture (coded 1). The second group is comprised of bonds without explicit (i.e., implicit) restrictions (coded 0), respectively. This binary dependent variable (Y) is regressed on the above independent variables in order to estimate the mean response (E(Y)), i.e., the probability that explicit restrictions are present in bond indenture, to the levels of the independent variables: amount of issue (X 1 ), maturity (X 2 ), coupon rate (X 3 ) and bond ratings (X 4 ). The linear regression model is Y i = α + β 1 X 1i + β 2 X 2i + β 3 X 3i + β 4 X 4i + ε i where the responses Y i = 0, 1 As explained by Neter et al. (1985), the expected response E(Y i ) = α + β 1 X 1i + β 2 X 2i + β 3 X 3i + β 4 X 4i is simply the probability that a bond has explicit restriction on sale and lease-back agreement when the levels of independent variables are X i s. The ordinary least square fit to our data will show how significantly each explanatory variable is related to the probability of explicit contracting. The ordinary least square fit to the data leads to the results shown in Table 4.

6 Table 4: Regression of contracting dummy variable on independent variables SUMMARY OUTPUT Regression Statistics Multiple R R Square Adjusted R Standard E Observatio 204 ANOVA df SS MS F Significance F Regression E-15 Residual Total Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept amount E E E E E-10 maturity coupon ratings As shown in Table 4, the dependent variable (the probability of explicit restriction in place) is positively related to the amount of capital to be raised. This positive relationship is statistically significant with its p-value close to zero supporting the hypothesized relationship: the larger the amount of borrowing, the more likely an explicit restriction is to be included in debt contract. The coefficients for maturity and coupon rate are not statistically significant at 5 percent. The negative coefficient for bond ratings is statistically significant with a p-value of , in strong support of the hypothesized negative relationship between the probability of explicit contracting and bond ratings. The larger the agency costs of debt (that is, the lower the bond ratings), the more likely explicit restrictions are in place. The empirical model for determination of contracting (either explicit restriction or implicit restriction) by the four independent variables is robust with an F-statistic and a highly significant p-value close to zero. The R-square of implies that about 30 percent of variation in the dependent variable (the probability of explicit contracting) is explained by the four independent variables: amount of issue, maturity, coupon rate, and bond ratings. 2. Two group comparison: explicit restriction vs. implicit restriction This part first examines how the agency costs of debt are dependent upon the contracting indicator variable (1 for explicit, and 0 for implicit restriction). Testing for statistical significance of the contracting indicator variable will reveal the effect of contracting on agency costs of debt. The linear regression model is Y i = α + β 1 X 1i + β 2 X 2i + β 3 X 3i + β 4 X 4i + ε i where:

7 Y i = Bond ratings (proxy for agency costs of debt), X 1i = amount of issue, X 2i = maturity, X 3i = coupon rate and contracting variable X 4i = 1 for explicit restriction and 0 for implicit restriction. The linear regression will exhibit how the agency costs of debt (proxied by bond ratings) is related to each explanatory variable. Specifically, we focus on the contracting indicator variable, X 4i. Statistical significance of the coefficient for the contracting variable will determine whether or not the variable has an effect on agency costs of debt. The ordinary least square fit to the data leads to the results shown in Table 5. Table 5: Regression of bond ratings on bond feature variables and contracting variable SUMMARY OUTPUT Regression Statistics Multiple R R Square Adjusted R Standard E Observatio 204 ANOVA df SS MS F Significance F Regression Residual Total Coefficients Standard Error t Stat P-value Lower 95% Upper 95% Intercept E amount 3.195E E E E-10 maturity coupon contracting As shown in Table 5, the dependent variable (bond ratings, as a proxy of agency costs of debt) is positively related to the amount of issue and maturity, and is negatively related to coupon rate and contracting variable. Coefficients of all independent variables are statistically significant at 5 percent. Interpretation of this result is as follows: The larger the amount of issue, the higher bond ratings, i.e., the lower the agency costs of debt. This result indicates that the size of capital to be raised by bond issue tends to be larger for the borrowing firms with lower agency costs of debt. The maturity coefficient, with a p-value of 0.01, is positively related to agency costs of debt such that the longer the maturity, the higher the bond ratings, i.e., the lower agency costs of debt. This result contradicts the common notion that a long maturity is highly correlated with high default risk that would lower the bond ratings. The statistically significant positive coefficient indicates that long term bonds are issued by firms that have low agency costs of debt. The coupon rate is found to be negatively related to the agency costs of debt. The coefficient for the coupon rate is negative and statistically significant with p-value close to zero and indicates

8 that the bonds that have a low coupon rate are rated more highly by the rating agencies. It can be interpreted that the firms with low agency costs of debt (i.e., high bond ratings) can raise capital in bond markets at low coupon rates. Most noteworthy, the binary contracting variable (i.e., 1 for explicit restriction and 0 for implicit restriction) is negatively related to the agency costs of debt, and its coefficient is statistically significant with p-value This result indicates that the explicit restriction reduces bond ratings by This negative effect of explicit restriction on the agency costs of debt can be expressed in the 95 percent confidence interval for β 4, the coefficient of contracting variable, X 4i : β Thus, with 95 percent confidence, it is concluded that bonds under explicit restriction tend to have ratings that are by somewhere between 0.16 and 1.16 lower, on the average, than those only under implicit restriction. The regression results show that the amount of issue and maturity are positively related to bond ratings, indicating that the firms with low agency costs of debt tend to issue bonds with long maturity for a large amount of debt capital. The coupon rate is negatively related to bond ratings and it indicates that firms with low agency costs of debt can raise debt capital at low coupon rates. Lastly, the contracting variable, that is found to be negatively related to bond ratings, indicates that firms with low agency costs of debt are positioned to raise debt capital relatively free of explicit restriction placed in bond indenture. Evidently, bonds issued under implicit restriction appear to perform better than bonds under explicit restriction in mitigating the conflict of interest between bondholder and shareholder as indicated by bond ratings differential between 0.16 and In a nutshell, the bond ratings as a proxy for agency costs of debt are negatively related to explicit contracting, and bonds issued under implicit restriction are perceived to be less risky in terms of potential risk of expropriation of bondholders wealth. The regression analysis is now followed by a t-test to compare the two subsamples. This is a test of the null hypothesis that there is no difference in bond ratings (i.e., agency costs of debt) between the explicit and implicit restriction groups of bonds. H 0 : Ratings exp Ratings imp H a : Ratings exp > Ratings imp where Ratings exp is the mean bond rating for the explicit restriction group and Ratings imp is for implicit restriction group. As specified by Tirole (2006), the first role of a protective covenant (such as explicitly stated restriction of the sale of asset and/or sale and lease-back transaction) is to prevent managers and shareholders from expropriating bondholders wealth. Thus, it is predicted that bonds with explicit restriction have lower agency costs of debt (i.e., higher bond ratings) than bonds with an implicit only restriction. If the null hypothesis is not rejected, it should be concluded that the implicit contracting group has been more effective than or at least as effective as the explicit contracting group in mitigating the agency costs of debt. Table 6 shows that the difference in bond ratings between the two groups of bonds (i.e., bonds issued under explicit restriction and bonds issued under implicit only restriction) is statistically significant at 5 percent. The coefficient for the bond rating difference (Ratings exp - Ratings imp ) is with p-value It indicates that bonds issued under implicit restriction are rated higher by rating agencies than bonds issued under explicit restriction. The result of this t-test is consistent with the above regression result as far as the effect of contracting variable on agency costs of debt. It is noteworthy that bonds issued under implicit restriction are rated higher than the bonds with explicit restriction. Table 6: t-test for bond rating difference between

9 two groups of bond Ratings exp Ratings imp Mean Variance Observations Pooled Variance Hypothesized Mean Difference 0 df 202 t Stat P(T<=t) one-tail t Critical one-tail CONCLUSION Rational debt contracting is to stipulate protective covenants intended to protect bondholders from possible wealth expropriation by managers and shareholders. This paper explores the relationship between contracting scheme (either explicit or implicit) and bond feature variables. The probability that the debt contract contains explicit restriction on disposition of assets is found to be positively related to the amount of capital to be borrowed by issue of bonds and maturity of bonds issued. The results are consistent with hypothesized relationships: (i) the larger the amount of borrowing, the more likely an explicit restriction is to be included in debt contract; (ii) the probability of including an explicit restriction on asset disposition increases as the maturity of the bond increases; (iii) the coupon rate and bond ratings are negatively related to the probability of the debt contract being explicit on the restriction of sale and leaseback transaction: (iv) the higher the coupon rate is, the lower the default risk, thus lower the probability of including explicit restriction on disposition of assets; and (v) the larger the agency costs of debt (that is, the lower the bond ratings), the more likely explicit restriction to be in place. Agency costs of debt (for which bond ratings are used as a proxy) are found to be positively related to the amount of issue and maturity, and negatively related to coupon rate and contracting variable. These positive relationships are interpreted as the size of capital to be raised by bond issue tends to be larger for the borrowing firms with lower agency costs of debt; and the longer the maturity, the higher the bond ratings, i.e. the lower agency costs of debt. The coupon rate is negatively related to the agency costs of debt and it is interpreted in the following way: the firms with low agency costs of debt (i.e. high bond ratings) can raise capital in bond market at low coupon rates. The binary contracting variable (i.e., 1 for explicit restriction and 0 for implicit restriction) is negatively related to the bond rating, which may be understood as the agency costs of debt are negatively related to explicit contracting. That is, bonds issued under implicit restriction are perceived to be less risky in terms of potential risk of expropriation of bondholders wealth. The t-test results indicate that bonds issued under implicit restriction are rated even higher than the other group of bonds. That is, the bonds issued without explicit restriction on the sale and leaseback transaction are rated higher than bonds issued with explicit restriction stipulated as a protective covenant in debt contract.

10 There are two possible interpretations for the results concerning the contracting variable. First, firms with low agency costs of debt are simply able to issue debt without explicit restrictions on the disposition of assets and choose to do so. The second (and more surprising if true) interpretation is that implicit restriction reduces agency costs more efficiently than do explicit restrictions. Under this interpretation, the mere issuance of bonds without explicit restrictions in the debt contract will result in the firm being viewed as behaving prudently by following their own self-imposed restriction and hence higher bond ratings. This second interpretation contradicts the common notion that the agency costs of debt would be reduced by the inclusion of explicit restrictions on the disposition of assets in the debt contract. Further empirical work is needed to determine which competing interpretation better reflects reality. REFERENCES Barnea, A., Haugen, R.A., and Senbet, L.W. (1980). A Rationale for Debt Maturity Structure and Call Provisions in the Agency Theoretic Framework. Journal of Finance (December), Bodie, Z., Taggart, R. (1978). Future Investment Opportunities and the Value of the Call Provision on a Bond. Journal of Finance, vol. 33, n. 4. Bhojraj, S., Sengupta, P. (2003). Effect of Corporate Governance on Bond Ratings and Yields: The Role of institutional Investors and Outside Directors. Journal of Business, vol. 76, no.3. Crabbe, L., Helwege, J. (1994). Alternative Tests of Agency Theories of Callable Bonds.Financial Management, vol. 23, n 4. Galai, D., Masulis, R.(1976). The Option Pricing Model and the Risk Factor of Common Stock. Journal of Financial Economics, vol. 3, n. 3. Jensen, M.C., Meckling, W.H. (1976). Theory of the Firm: Managerial Behavior, Agency Costs, and Capital Structure. Journal of Financial Economics 3, Neter, J, Wasserman, W., and Kutner, M.H. (1985). Applied Linear Statistical Models, 2 nd edition, Irwin. Smith, C., Warner, J. (1979). On Financial Contracting: An Analysis of Bond Covenants. Journal of Financial Economics, 7, Thatcher, J.S. (1985). The Choice of Call Provision Terms: Evidence of the Existence of Agency Costs of Debt. Journal of Finance (June), Tirole, J. (2006). The Theory of Corporate Finance. Princeton University Press, Princeton.