.7 \. r04: Sky; 4 .334 uneven.» 13.x? 23.. .521 1; n a H. ,. r. Li... 4.. L . :3 .1 .. . i. 1.. 1 .13. I flnxulfianu. .\. . fie! 3:1... . .V 21.3. I. . 5 t! ES...) xitzndlnv. .. ‘ 3. r: ‘3.) 3C9...- ‘lv. .. I: 2.13.35. 3.314.? £1.31 .fi; b.3191... . is. 3...: S... .1 .51.. .23. .51.... 1...... IZ!AX.: . x 3!! {1.7.3}: 6.70.1. 4 sun , s . L 3 t. fa}! . 9:13;“ .. r :3! .L. . .a . 3,341.. ii... 12.. llllllllllllllllllllllllllllllllllllllllllllllllllllllllll 31293 010191918 This is to certify that the dissertation entitled COSTLY CONTRACTING: The Case of Event Risk Covenants presented by Claudia Sue Kocher has been accepted towards fulfillment of the requirements for Doctoral degree in Business Administration Major professor .’ I. . .' I I I Date October 7, 1993 MS U is an Affirmative Action/Equal Opportunity Institution 0-12771 LIBRARY Michigan State University PLACE II RETURN BOX to roman this checkout from your record. 1‘0 AVOID FINES return on or baton dd. duo. . DATE DUE DATE DUE DATE DUE MSU I: An Affirmative Action/Equal Oppommny Institution Wm: __ COSTLY CONTRACTING: THE CASE OF EVENT RISK COVENANTS BY Claudia Sue Kocher A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Finance and Insurance 1993 ABSTRACT COSTLY CONTRACTING: THE CASE OF EVENT RISK COVENANTS BY Claudia Sue Kocher The objective of this dissertation is to test the Costly Contracting Hypothesis, as described by Smith and Warner (1979), using the example of event risk covenants. Theory suggests that firms issuing bonds with event risk covenants have more severe agency problems than firms issuing bonds without event risk covenants. Theory also suggests that information asymmetry may be more of a problem for firms issuing bonds with event risk covenants. Empirical tests are performed to determine the relation, if any, between event risk covenant use and severity of agency problems. An alternative explanation for event risk covenant use is presented and tested. This explanation focuses on managerial entrenchment motives. Previous empirical studies find evidence that managers with small own-firm ownership stakes take actions which maximize their personal wealth instead of stockholders’ wealth. Empirical tests in this paper examine the relation between manager own-firm ownership stake and event risk covenant use. The sample includes all investment grade coupon bonds with maturities greater than five years which were issued by U.S. industrial firms during the period from January 1, 1989 through December 31, 1990. Moody's Bond Record is used to identify sample bonds. Logistic regression analysis is used to analyze the relation between agency problems, manager own-firm ownership stake and event risk covenant use. Independent variable coefficients are estimated using the method of maximum likelihood. . Results show support for the Costly Contracting Hypothesis. Event risk covenant use is associated with firm characteristics which indicate severe agency problems of free cash flow and debt. Empirical evidence does not find a systematic relation between manager own-firm ownership stake and event risk covenant use. Results of empirical tests of the relation between information asymmetry and event risk covenant use are inconclusive. DEDICATION I dedicate this dissertation to my husband, Shawn Rooney, my parents, Mary and Wilbur Kocher and my sisters, Brenda, Diane and Patricia. Their love for me and respect for my work have bolstered me during my long journey through the doctoral program. iv ACKNOWLEDGEMENTS I wish to thank my dissertation chairman, John Gilster, and the members of my dissertation committee, Joseph Anthony, Mary' Bange, and. Michael Mazzeo, for' their guidance and support. They were very generous with their ideas and time and provided me with excellent role models. I feel very fortunate to have had an opportunity to work with this talented group of people. I would also like to thank the other faculty members of the Department of Finance and Insurance at Michigan State for their encouragement and advice. A special thanks to Richard Simonds and Dale Domian for the guidance they offered in their roles as Ph.D. Program advisors and to Kirt Butler for sharing his computer programs to access the IBES data tapes. I appreciate the help and encouragement my friends and fellow doctoral students have provided and recognize that without them my graduate school experience would have been much less enjoyable and rewarding. Ming Shen Chen, Lise Graham, Dave Louton, Rick Osborne and Janet Todd were especially helpful to me over the course of my studies. Thank you to my husband, Shawn Rooney, who patiently helped me evaluate bond covenant descriptions and edit the final version of this dissertation. Thank you to Celeste Shoulders, Linda Driscoll and.Geri Pratt, in.the office of the Department of Finance and Insurance. 'They ‘were always available to offer' a cheerful greeting and. a practical solution to the latest crisis. Finally, I gratefully acknowledge the contribution of I/B/E/S Inc. for providing earnings per share forecast data, available through the Institutional Brokers Estimate System. vi TABLE OF CONTENTS LIST OF TABLES CHAPTER 1 INTRODUCTION 1.1 1.2 1.3 1.4 1.5 CHAPTER 2 Theoretical and Empirical Background Objective of this Research Event Risk and Event Risk Covenants Approach of this Research Organization of the Dissertation LITERATURE REVIEW Corporate Financial Restructuring Effects of Restructuring on Bondholder Wealth Agency Problems of Debt The Role of Indenture Covenants Summary of Chapter Two HYPOTHESIS DEVELOPMENT The Agency Problem of Free Cash Flow The Agency Problem of Debt Information Asymmetry Manager Entrenchment - An Alternative Explanation Summary vii 18 24 28 31 33 34 35 36 37 38 CHAPTER 4 DATA 4O 4.1 Sample Bonds and Firms 40 4.2 Financial Statement Data 43 4.3 Event Risk Protection 44 4.4 Insider Ownership Data 46 4.5 Earnings Forecast Variability as a Measure 47 of Information Asymmetry 4.6 Overall Covenant Sets for Sample Firms 48 i APPENDICES TO CHAPTER 4 50 4.1a List of Sample Firms 50 4.2a Sample Firm Insider Ownership Fractions 54 CHAPTER 5 DESCRIPTION OF VARIABLES WHICH PROXY FOR 58 AGENCY'PROBLEMS 5.1 Empirical Model 59 5.2 Independent Variables 60 5.21 Firm Size 60 5.22, Investment Opportunities 62 5.23 Free Cash Flow 64 5.24 Information Asymmetry 66 5.25 Leverage 68 5.26 Financial Subsidiary Indicator 69 5.27 Interaction Between Leverage and 69 Financial Subsidiaries 5.3 Empirical Methods 70 APPENDICES TO CHAPTER 5 71 5.1a Logit Regression 71 5.2a Information on Independent Variables 75 viii CHAPTER 6 DISCUSSION OF RESULTS 81 6.1 Empirical Results 84 6.2 "Substantial" Finance Subsidiaries Versus 92 All Finance Subsidiaries 6.3 Model Specification 95 6.4 Empirical Test of a Managerial Entrenchment 100 Motive for Event Risk Covenant Use 6.5 Summary of Results 103 APPENDIX TO CHAPTER 6 104 6.1a Robustness Tests for Independent Variables 104 CHAPTER 7 COVENANT SET EVALUATIONS 115 7.1 Background Information 115 7.2 Empirical Analysis of Covenant Sets 116 7.3 Discussion of Results 121 CHAPTER 8 CONCLUSION 122 8.1 Summary of Results 122 8.2 Implications of Results 123 8.3 Possible Extensions of this Research 126 BIBLIOGRAPHY 127 ix Table Table Table Table Table Table Table Table Table Table Table Table LIST OF TABLES Panel_b Description of Sample Bonds and Firms PaneLB Standard and Poor Corporation Event Risk Covenant Rankings for Sample Bonds with Event Risk Covenants Summary of Predictions of the Costly Contracting Hypothesis and the Manager Entrenchment Theory Logistic Regression Coefficients Base Model Logistic Regression Coefficients Information Asymmetry Variable Included Logistic Regression Coefficients Variations Related to Finance Subsidiaries Logistic Regression Coefficients Variations in Model Specification Logistic Regression Coefficients Insider Ownership Variable Included Logistic Regression Coefficients Tests of Size Variable Robustness Logistic Regression Coefficients Tests of Investment Opportunity Variable Robustness Logistic Regression Coefficients Tests of Cash Flow Variable Robustness Logistic Regression Coefficients Tests of Debt Variable Robustness Data on Firms Which Have Event Risk Protection in Bond Indentures During Sample Period (1989 - 1990) 42 82 85 90 93 96 101 104 106 108 110 119 Table 7.2 Data on Firms Which Do Not Have Event Risk 120 Protection in Bond Indentures During Sample Period (1989 - 1990) xi Chapter 1 Introduction Agency problems arise from conflicting interests among parties to the modern corporation. Agency problems between stockholders and managers and between stockholders and bondholders may affect financing and investing decisions of the firm and reduce firm value. (See Jensen and Meckling (1976), Barnea, Haugen and Senbet (1985), and Jensen (1986)) Thus the resolution of these problems is an important issue in corporate finance. 11]. W Smith and Warner (1979) present two competing hypotheses, the Irrelevance HypothesiS' and the Costly Contracting Hypothesis, to explain how’ agency' problems of debt are resolved. The Irrelevance Hypothesis states that market forces.are sufficient.to resolve(agency’problems‘through.price adjustments or claim restructuring; it is based on the work of Fama (1978) and Galai and Masulis (1976). The Costly Contracting Hypothesis states that market forces are insufficient to induce actions which maximize firm value. Under this hypothesis "there is a unique optimal set of financial contracts which maximize the value of the firm." (Smith and Warner (1979)) The Costly Contracting Hypothesis is based on the work of Jensen and Meckling (1976) and Myers (1977). M“ '7 2 Recent empirical studies have attempted to explain existing debt contracts by associating firm characteristics with the use of call provisions (Thatcher (1985)), dividend constraints, debt limitations and sinking funds (Malitz (1986)), and sinking funds (Kao and Wu (1990)). Results show support for the Costly Contracting Hypothesis. 1;; ijggtiyg of this Research The main objective of this research is to test the Costly Contracting and Irrelevance Hypotheses using the example of event risk covenants. Theory related to the Costly Contracting Hypothesis suggests that firms issuing bonds with event risk covenants have more severe agency problems than firms issuing bonds without event risk covenants. It also suggests that information asymmetry may be more of a problem for firms issuing bonds with event risk covenants. Theory related to the Irrelevance Hypothesis suggests that market forces can solve agency problems of debt and therefore firms issuing bonds with event risk covenants should have the same characteristics as firms issuing bonds without event risk covenants. Event risk covenants provide an interesting test of the Costly Contracting and Irrelevance Hypotheses. Existing empirical studies by Crabbe (1991) and Fields, Kidwell and Klein (1991) show that these covenants are priced by bondholders in the market. However, event risk covenant use did not persist. These covenants were heavily used in 1989 3 and 1990 and rarely used after 1990. Interesting questions that arise include: 1. Why did firms use event risk covenants? Why not adjust bond prices to compensate bondholders for event risk? Were they a fad or did some firms increase value by using them? 2. How' did event risk. covenants fit into overall covenant sets? Did they replace other covenants? Or were they used in addition to the usual covenants? What happened in 1991 - did other types of covenants replace event risk covenants? 3. Why did event risk covenant use decrease dramatically after the 1989-1990 period? Are the costs of event risk covenants greater than the benefits? Do macroeconomic factors, such as the availability of credit, affect the decision to use event risk covenants? The empirical results of this research provide insight into why firms use event risk covenants and. how event risk covenants fit into overall covenant sets. Examination of changes in financial market conditions provides insight into why event risk covenant use declined dramatically after 1990. 1.3 Eygnt Risk and Event Risk Covenants Event risk is defined as the risk of bondholder wealth loss due to a leverage-increasing event, such as a leveraged buyout, leverage-increasing’ takeover or leverage-financed share repurchase. Event risk covenants usually protect bondholders by allowing them to put bonds back to the issuing firm in exchange for par value if a pre-defined event occurs and bond ratings decrease to speculative grade. 4 Event risk covenants were first included in bond indentures in 1986. These early event risk covenants, called "poison puts", were at least partially ineffective because they were triggered only if an event was "hostile". Many events which started out in a hostile manner ultimately were declared "friendly" by the target firm's board of directors. In late 1988, in the aftermath of the RJR Nabisco leveraged buyout, event risk covenants were strengthened by removal of the requirement that an event be ”hostile". The new event risk covenants, called "super poison puts", were popular with investors and issuers.1 Standard and Poor's Corporation responded to their popularity by developing criteria for ranking the strength of event risk covenant protection. 1.5 Approach of this Research Empirical tests are performed to determine the relation between event risk covenant use and severity of agency problems. .Agency'problems related to free cash flow, leverage level and asymmetric information are focused on because these problems are relevant to event risk; Results show support for 1The "super poison put" is the main type of event risk covenant used after late 1988. Two other types of event risk covenant are occasionally used, however. A coupon reset covenant calls for the coupon to be reset so that the bond trades at par if an event occurs and the bond is downgraded. The covenant may require that the rating decline be to speculative grade. A "credit sensitive note" calls for specified adjustment of the bond coupon for rating changes. This covenant protects against.bond rating declines.due to‘any cause. 5 the Costly Contracting Hypothesis. Event risk covenant use is systematically related to severity of agency problems of free cash flow and debt. The results regarding the relation between event risk covenant use and information asymmetry are inconclusive. Also, an alternative explanation for event risk covenant use is presented and tested. This explanation focuses on managerial entrenchment motives. Previous empirical studies (Amihud and Lev (1981) and Walkling and Long (1984)) find evidence that managers with small own-firm ownership stakes take actions which maximize their personal wealth instead of stockholders’ wealth. Empirical tests in this paper examine the relation between insider ownership and event risk covenant use. Results show no systematic relation between manager ownership fraction and event risk covenant use. .115 Organization of the Dissertation The dissertation is organized as follows: Chapter 2 presents a review of relevant literature. Chapter 3 develops a theoretical basis for the empirical tests of the Costly Contracting Hypothesis versus the Irrelevance Hypothesis. Chapter 4 describes sample selection and data collection. Chapter 5 explains how proxy variables for agency problems are calculated and discusses empirical methodology. Chapter 6 discusses empirical results. Chapter 7 discusses the analysis of overall covenant sets for long-term bonds issued between 1985 and 1991 by sample firms. Chapter 8 presents a summary .- MUT- 6 of results and concludes the dissertation. CHAPTER 2 Literature Review This chapter presents a survey of relevant theoretical and empirical research. Section 2.1 looks at research on why firms restructure. Motives for restructuring which are related to agency conflicts are emphasized. This research is helpful in identifying firm characteristics that are associated with event risk. Section 2.2 reviews empirical research which documents bondholder wealth changes due to financial restructuring. This research is relevant because it quantifies the effects of leverage-increasing "events" on bondholder wealth. Section 2.3 reviews theory related to agency problems of debt. The theory discussed in this section explains why agency problems related to firm restructuring may cause bondholder wealth losses. Section 2.4 reviews research which focuses on the role of bond contracts in reducing agency problems and maximizing firm value. 2.1 Copporate Financial Bestructuping There are numerous theories that offer explanations for firm restructurings. The theories that appear most relevant for event risk are those based on agency theory. Jensen’s free cash flow theory (1986) and Roll's hubris theory (1986) offer explanations for leverage-increasing restructurings. This review of the literature on corporate restructurings 7 8 focuses first on agency theory and then briefly discusses other theories on why corporate restructurings occur. Manne (1965) is the first to propose that corporate control is a valuable asset. He reasons that poorly managed firms have low share prices relative to similar firms with superior management. A low share price causes the firm to be attractive to outsiders who believe they have the ability to manage the firm more skillfully. Manne states that "only the takeover scheme provides some assurance of competitive efficiency among corporate managers and thereby affords strong protection to the interests of vast numbers of small, non-controlling shareholders" (p. 113). Manne's work looks at one aspect of agency problems related to corporate restructuring. Takeovers solve or minimize an agency problem. Jensen and Ruback (1983) define corporate control as '"the rights to determine the management of corporate resources -- that is, the rights to hire, fire and set the compensation of top-level managers" (p. 5). Their review of the extensive literature on the market for corporate control concludes that corporate takeovers generate positive gains, target firm shareholders gain wealth, and bidding firm shareholders do not lose wealth. Also, the gains from corporate takeovers do not appear to come from increased market power. Finally, with the exception of actions that eliminate bidders, such as targeted large block share repurchases and standstill agreements, they do not find that 9 managerial actions related to corporate control harm shareholders. Jensen (1986) examines conflicts of interest between shareholders and managers with regard to cash payouts and the role of debt in resolving these conflicts. He states that "the problem is how to motivate managers to disgorge the cash rather than investing it at below the cost of capital or wasting it on organization inefficiencies" (p. 323). Organizations which generate a large amount of free cash flow are likely to have severe conflicts of interest between shareholders and managers. Jensen suggests that these firms can issue new debt and repurchase shares with the proceeds from the debt issue. The interest payments due every period on the debt will impose discipline on managers and reduce the amount of cash over which they have control. Jensen states that "the control function of debt is more important in organizations that generate large cash flows but have low growth prospects and even more important in organizations that must shrink" (p. 324). Firms that go private through leveraged buyouts (LBO's) are likely to fit this description, according to the free cash flow theory. Free cash flow theory offers a two-tiered explanation for takeovers. Managers of firms with excess cash and unused borrowing power may attempt to acquire other firms rather than increase payouts to shareholders. These acquisitions are likely to have lower abnormal returns to 10 shareholders than acquisitions made by firms which have less free cash flow and higher growth prospects. On the other hand, hostile takeovers of firms which have severe agency problems of free cash flow may increase target firm shareholder wealth. Outside acquirers take over the firm, increasing its debt to a level where the firm cannot continue to operate inefficiently. In order to survive the firm must sell assets, increase the efficiency of operations, and reduce the size of its management staff. These activities are difficult to accomplish unless there is a sense of crisis in the firm. Hostile takeovers do not have to actually occur for free cash flow problems to be solved. The threat of a takeover may cause firms to engage in a large stock repurchase or an LBO. Jensen notes that "free cash flow theory predicts that many acquirers will tend to have exceptionally good performance prior to acquisition" (p. 329). He presents the oil industry as an example of an industry with severe free cash flow problems during the period of the late 1970's and the early 1980's. Oil industry firms obtained large amounts of cash flow through price increases in the mid- to late- 1970's and used this cash flow to fund additional exploration and drilling projects and to engage in conglomerate acquisitions. Returns from these projects were not favorable. The free cash flow theory implies that firms with a 11 large amount of free cash flow and few growth opportunities are takeover targets. They are more likely to be involved in a leveraged buyout or be the target of a leverage- increasing acquisition than the average firm. This has implications for bond investors. If leverage increases, the existing bonds will fall in value because the risk of bankruptcy increases. And if new bonds are issued by a firm with high cash flow and few growth opportunities, prospective investors should anticipate future leverage increases and adjust bond prices and contracts with this in mind. Lehn and Poulsen (1989) empirically investigate Jensen's free cash flow theory. They look at a measure of free cash flow scaled by the market value of the firm's equity, hereafter referred to as CF/EQ.1 This measure is calculated for a sample of 263 L80 firms and an equal number of similar size and industry control firms. Average CF/EQ is significantly larger for the LBO sample than for the control sample. Lehn and Poulsen also look at firm historical sales growth rates for the LBO and control firms. The LED sample "is characterized by systematically lower growth rates than 1CF/EQ = INC - TAX - INTEXP - PFDDIV - COMDIV where INC is operating income before depreciation, TAX is total income tax, INTEXP is gross interest expense on short- and long-term debt, PFDDIV is total amount of preferred dividend requirement on cumulative preferred stock and dividends paid on noncumulative preferred stock and COMDIV is total dollar amount of dividends declared on common stock. 12 the control group, significant for three of four measures" (p. 778). In order to evaluate whether going private transactions are influenced by a threat of a hostile takeover, Lehn and Poulsen create a variable, FOOTSTEPS. FOOTSTEPS takes a value of one if a Wall Street Journal search shows the firm received a takeover offer or was the subject of takeover F— speculation in the year preceding or following the going ! private transaction. FOOTSTEPS takes a value of one for i 42.6% of the going private sample and 15.1% of the control group. This result supports the idea that going private transactions are a response to the threat of a hostile takeover. Lang, Stulz and Walkling (1991) examine the relation between free cash flow and the quality of a firm's investment opportunities and bidder returns. Firm investment opportunities are measured by Tobin’s q, defined as the ratio of the market value of a firm’s assets to replacement cost.2 The authors note that "firms with substantial cash flow and a low Tobin's q" (Lang, Stulz and Walkling, 1991, p. 321) are expected to have the lowest abnormal bidder returns because these firms have the 2Lindenberg and Ross (1981) describe their procedure for calculating Tobin's q ratio~ They use data from SEC form 10-K on replacement costs. Reporting of this data was required starting in 1976. They note that the SEC gave firms "broad leeway" in reporting this data. Lang, Stulz and Walkling report using the procedure described by Lindenberg and Ross (with some modifications) to calculate Tobin's q. 13 greatest agency costs of free cash flow. The results show that firms with high cash flow measures and low q ratios have the lowest abnormal returns. Lang, Stulz and Walkling examine other bidder characteristics such as means of payment, bidder managerial ownership, the debt-equity_ratio of the bidder, and the logarithm of the size of the target in relation to the bidder. These characteristics do not affect bidder returns in their sample. Both Lehn and Poulsen and Lang, Stulz and Walkling provide empirical support for Jensen's free cash flow theory. The former provides evidence that firms engaging in LBO's often have a high level of free cash flow and low sales growth, as Jensen's theory predicts. The latter provides evidence in support of Jensen’s theory that firms with agency problems of free cash flow have increased incentive to use excess cash for low or negative return acquisitions rather than return it to shareholders. The hubris hypothesis of corporate takeovers (Roll, 1986) says that takeover bids are made because bidders erroneously believe their valuation estimates are correct. According to this hypothesis, bidders assess the value of a potential acquisition. If their assessment is below the current market price, they do not bid. If their assessment is above the current market price, they place a bid. If the current market price reflects the true value of the firm, bids above and below this price are random errors. The 14 hubris hypothesis predicts that the total combined wealth gains from a takeover are zero or negative. The increase in value to target shareholders is accompanied by a decrease in value to bidder shareholders. Roll examines previous studies on changes in total value associated with takeovers. He cites conflicting results from different studies and concludes that the results are uncertain. This is different from Jensen and Ruback’s (1983) conclusion that targets gain, bidders do not lose, and takeovers result in overall gains. Roll suggests that the hubris hypothesis can act as the null hypothesis of corporate takeovers because "it asserts that all markets are strong form efficient. He acknowledges that measurement problems make interpreting bidder returns difficult. For example, bidders usually increase leverage in an acquisition and leverage increases have been shown to result in excess positive returns to firms (Masulis,1980). Also, the bid may have been anticipated or may convey information about the bidder firm. Hubris may have implications for event risk. Roll notes the following: The entire sequence of returns for successful bidding firms is consistent with the hubris hypothesis. In the prebid period, excellent performance endows management with both hubris and cash. (p. 210) The hubris theory, like the free cash flow theory, may provide insight into which firms are likely to increase leverage in a corporate restructuring. Here the type of restructuring is an acquisition. As with free cash flow 15 theory, a large amount of cash is important. Jennings and Mazzeo (1991) look at whether management learns from security price changes which occur after the announcement of a proposed acquisition. Their results show no support for the view that bidder management believes share prices provide valuable information. This is consistent with the hubris hypothesis. Hubris suggests that managers believe their valuation estimates are more accurate than the market price. Other explanations of why corporations restructure can be categorized as follows:3 1. Efficiency Explanations 2. Information Explanations 3. Market Power Explanations 4. Tax Explanations Efficiency explanations state that strategically combining firms leads to operating economies of scale and replacing poor managers with skillful managers in a takeover increases the value of the acquired firm. It may be difficult to distinguish between efficiency explanations and agency theory explanations, note Copeland and Weston (1988), because it is difficult to determine if decisions which lead to poor results are due to manager opportunism or errors in judgement. Information theories related to mergers and acquisitions are examined by Bradley, Desai and Kim (1983). 3This categorization is taken from Copeland and Weston's text titled Financial Theogy aag Coppopapa Policy, 1988. 16 They examine two explanations for changes in firm value around merger announcements. The first explanation is the "kick in the pants" explanation. The threat of being acquired causes managers to reevaluate and improve their present strategy. The second explanation is the "sitting on a goldmine" explanation. The bidding process causes information about the firm to be released. This new information leads the market to believe the firm has been undervalued. Increased market power is a possible motivating force behind mergers. However, this force is opposed by the U.S. Department of Justice, on the grounds that monopolies are created and price fixing is facilitated. Tax factors appear to influence corporate restructuring actiVities. Several studies have found support for the theory that LBO gains may be partially attributed to tax gains. (See Marais, Schipper and Smith, 1989, and Kaplan, 1989) Mergers and acquisition decisions may be influenced by tax factors in situations where the firm being acquired has accumulated losses which may be used to reduce the taxable income of the acquirer. Managerial ownership stake appears to influence restructuring activities. Amihud and Lev (1981) examine motives for conglomerate mergers and hypothesize that managers may use conglomerate mergers to reduce employment risk. They find support for this hypothesis in that "manager-controlled firms were found to engage in more *’ p... 0“...- «he. 17 conglomerate acquisitions than owner-controlled firms." (p. 615) These results suggest that the classical assumption that firms maximize stockholder wealth may be violated when managerial ownership stake is small. Lewellen, Loderer and Rosenfeld (1985) investigate whether ”the impact of a merger on bidder firm stock returns is more likely to be negative when management's ownership of the firm's stock is small" (p. 209). A sample of 191 acquiring firms is divided into two subgroups. The first subgroup has positive cumulative prediction errors on stock returns during the period from five days prior to the merger offer announcement through merger resolution. The second subgroup has negative cumulative prediction errors on stock returns during this period. Average values for three variables are calculated for each of three senior executive categories.‘ Results show that managers and directors with large equity holdings in their firms are less likely than managers with small equity holdings to engage in acquisitions that decrease shareholder wealth. According to the results of this research, managers’ ‘The three variables are defined as follows: VALSHS/Pay = dollar ‘value of own-company stockholdings, divided by aggregate current remuneration: SHAREINC/Pay = expected annual income from own-company stockholdings, divided by aggregate current remuneration: .ALFA = number of shares held by management divided by the total number of shares outstanding: 18 and directors’ own-firm equity ownership stakes influence the acquisition decisions they make. There is no reason to believe acquisitions are the only decisions influenced by executive ownership stake. For example, firms with a low manager ownership stake may be more likely to attempt to discourage hostile takeovers than firms with a high manager ownership stake. A Walkling and Long (1984) look at a sample of cash tender offers and find that takeover bid resistance is related to manager and director personal wealth changes. If managers and directors stand to gain wealth in a takeover, they are less likely to resist. These results support the development of executive compensation plans which align owner and manager incentives. 2.2 Effagts of Resppacturing op Bopdhglga: Waaltn Part A examines possible reasons corporations enter into leverage-increasing restructurings. This section examines the effects of such restructurings on the wealth of preexisting bondholders. The risk of wealth loss due to a leverage-increasing corporate restructuring is often referred to as "event risk". Event risk is also discussed in terms of how it affects new issue bonds. Here event risk covenants are described and recent empirical studies which attempt to measure price effects of event risk covenants are summarized. Kim, McConnell and Greenwood (1977) examine how a 19 violation of a me-first rule affects the value of a corporation’s stock and bond securities. They define a me- first rule as a "prior arrangement to protect bondholders from uncompensated shifts of wealth from bondholders to stockholders through a change in the capital structure of the firm" (p. 789). A theoretical analysis of me-first rule violation is presented and then an empirical analysis of stockholder and bondholder returns in firms that have formed captive finance subsidiaries is performed. The theoretical analysis demonstrates that in perfect capital markets, bondholders will be worse off and stockholders will be better off if the firm is able to increase debt and violate the me-first rule protecting the original bondholders. In the portion of the analysis which considers corporate income tax, the authors show that the federal government shares the stockholders’ gains. Kim, McConnell and Greenwood present a situation in which a corporation forms a captive finance subsidiary as an example of a violation of a me-first rule in which the legal terms of the debt contract are not violated. The authors describe formation of the finance subsidiary as follows: firms ... organize the finance company which then issues debt in its own name, but which 'is guaranteed by the assets and earnings of the parent company. The proceeds of the debt issue are then used to purchase the parent company’s accounts receivable. Thereafter, the creditors of the subsidiary have first claim to the income produced by the sales contracts owned by the finance company. Only after the claims of the subsidiary’s creditors are met in full may any funds be 20 transferred from the wholly-owned subsidiary to the parent company to pay its creditors. This rearrangement of the asset and liability structure of the firm essentially creates a new class of security holders with claims that are superior to those of the old bondholders. (p. 797) The empirical analysis examines stockholder and bondholder abnormal returns for twenty-four firms which formed captive finance subsidiaries between 1940 and 1971. & Stockholder abnormal returns are calculated with a two- E factor model (as described by Black, Jensen, and Scholes (1972) and Fama and MacBeth (1973)). Bondholder abnormal returns are calculated using a paired-comparison procedure. The empirical results show that stockholders earned positive abnormal returns and bondholders experienced negative abnormal returns on formation of a captive finance subsidiary. A similar study on violation of me-first rules by Kim, Lewellen and McConnell (1978) looks at sale-leaseback transactions. This paper theoretically examines sale- leaseback transactions in a perfect market context. In a sale-leaseback arrangement, the lessor advances cash to the lessee in exchange for a series of promised lease payments and a priority claim to the residual value of the leased asset. The lessor should be concerned with the creditworthiness of the lessee and the value of the leased asset, which may be viewed as collateral for a loan. Kim, Lewellen and McConnell provide an analysis which shows how the market value of the firm’s bonds must decline 21 in a sale-leaseback transaction, as long as the lessee has a finite probability of becoming bankrupt. This is because "...the sale-leaseback diverts to the lessor a priority claim to a segment of the cash flow prospects which originally belonged to bondholders" (p. 875). Both the paper on sale-leasebacks by Kim, Lewellen, and McConnell and the paper on captive finance subsidiaries by Kim, McConnell and Greenwood provide examples of bondholder i wealth expropriation through rearrangement of firm capital structure. In other words, these papers address "event risk". The theoretical analyses presented in these papers helps clarify the security value changes that appear to occur in contemporary firm "events". Recent empirical studies have examined bondholder wealth changes after leverage-increasing financial restructurings. Lehn and Poulsen (1988) study bondholder returns after LBO’s and find that bond prices decrease by 2.46 percent over a 20-day period centered on the announcement date. Their sample is composed of only nine bonds, however. (106 LBO’s are in their original sample: bond price data is available on only 9 bonds.) Marais, Schipper and Smith (1989) study the effects of going-private transactions on the wealth of existing bondholders over the period 1974 through 1985. They find no evidence of abnormal returns to these bondholders, but do note that bondholders experienced rating downgrades as a consequence of the LBO’s. It does not seem correct that 22 bondholders would experience rating downgrades without losing wealth. Rating downgrades mean default risk is increased. The bond market will require yield increases to compensate for default risk increases. It seems probable that the results of Marais, Schipper and Smith are influenced by a lack of availability of bond price data. Asquith and Wizman (1990) study 214 bonds associated with 65 L80 targets over the 1980-1988 period. They find negative two percent abnormal returns for the entire sample over a period from two months before the announcement until two months after the LBO bid is either successfully completed or withdrawn. Covenant protection is important in explaining their results. Bonds with strong covenant protection receive average abnormal returns of positive 2.6 percent (over a period -2 months to +2 months). Bonds with weak (no) protection receive -0.7 percent (-5.2 percent) abnormal returns.5 Warga and Welch (1990) investigate bondholder wealth changes associated with LBO’s in the 1985-1989 period. They use trader-quoted prices and show that bondholders experience negative seven percent risk-adjusted returns over a period from two months prior to one month after the LBO 5Asquith and Wizman describe their method for classifying covenant protection. See Chapter 7 of this dissertation for a summary of their classification criteria. 23 announcement date.“ Crabbe (1991) documents bondholder wealth losses for 56 industrial bonds that are downgraded after leveraged restructurings during the period between January 1983 and August 1989. Crabbe notes that "prices of the downgraded bonds fell an average of 11.83 percent when measured from the re-offering date to the downgrade date" (p. 694).7 Crabbe also measures abnormal returns six months and one year before the downgrade date and finds negative abnormal returns over these periods of 7.77 percent and 9.01 percent, respectively. The difference in findings between the earlier empirical works by Lehn and Poulsen (1988) and Marais, Schipper and Smith (1989) and the slightly more recent works by Asquith and Wizman (1990), Warga and Welch (1990) and Crabbe (1991) may be due to variations in sample periods and methodologies. Overall, the evidence appears to indicate that leveraged restructurings, especially LBO’s, cause existing bondholders to lose wealth. Section 2.3 discusses this wealth loss in terms of agency conflicts between bondholders and stockholders. 6Warga and Welch (1990) state that trader-quoted prices represent dealer offers or actual trades and reflect all available information in competitive markets. 7The re-offering date is the date the bonds were originally sold to the public through the underwriter. 24 2...}. WM Previous sections of this literature review have examined research related to why firms restructure and how restructuring affects bondholders. This section looks at research which attempts to explain why bondholders lose wealth when leverage suddenly and significantly increases. Jensen and Meckling (1976) demonstrate how the existence of agency costs leads to an internal optimal capital structure for the modern corporation. The first part of their analysis shows how a manager-owner of a firm has greater incentive to consume perquisites as his/her ownership fraction decreases. This occurs because the manager-owner bears only a portion of the cost of the perquisites consumed. The second part of the paper shows how agency costs of debt increase as the debt ratio increases, because stockholders have opportunities to expropriate bondholder wealth. Common mechanisms for bondholder wealth expropriation include firm cash flow variance increases and bondholder claim dilution. Agency costs arise when manager-owners have an opportunity to switch to high variance projects after issuing new debt. Suppose bonds are priced at B1 if project 1, a low variance project is accepted. The bonds are priced at 32 if project 2, a high variance project is accepted. (Assume here that project 1 and project 2 have the same expected total payoffs which occur at time T.) Manager- owners can maximize their own wealth, at the expense of the 25 bondholders, if they lead bondholders to believe they are going to choose project 1 and then, after debt is issued at 8,, switch to project 2. Galai and Masulis (1976) use the Black-Scholes (1973) option pricing model to explain why equity value increases when owner-managers switch to high variance projects. Stockholders of a levered firm can be thought of as holding a European call option on the value of the firm. This option has an exercise price equal to the maturity value of the firm’s risky debt. The option may be exercised at the maturity date of the debt. Merton (1973) shows how variance is positively related to option value. If the firm chooses project 2, the project with the higher variance, the value of the owner-managers’ option increases. The value of the debt must decrease as the value of the option increases, because B = V - S, where B is the value of the debt, V is the value of the firm, and S is the value of the equity. Bondholders know that owner-managers can promise to take project 1, issue risky debt at a price B1, and then switch to project 2, causing the risky debt to fall in value to B2. Bondholders anticipate this expropriation opportunity and only pay B2 for the firm’ 8 debt securities. In this case, there is no redistribution of wealth and no agency cost. If project 2 has a lower expected value than project 1, however, there is an agency cost. If owner-managers accept project 2, the high variance project, the value of the firm 26 will fall. The change in firm value can be expressed as follows: v1 - v2 = (s, - $2) + (B1- 132) In the above equation, owner-managers can gain while bondholders lose and the total value of the firm decreases. But if bondholders anticipate owner-manager opportunities to switch to high variance projects, no bondholder wealth loss occurs and the owner-managers incur the wealth loss. Jensen and Meckling note that bondholders can include covenants in debt agreements which prohibit managers from engaging in actions that expropriate bondholder wealth. Covenants are costly to use and for this reason are not written so as to protect bondholders from every managerial action that has potential to reduce bond value. Costs associated with covenants include "the costs involved in writing such provisions, the costs of enforcing them and the reduced profitability of the firm (induced because the covenants occasionally limit management’s ability to take optimal actions on certain issues)..." (Jensen and Meckling, 1976, p. 334). Jensen and Meckling also discuss bankruptcy and reorganization costs. They note that the expected value of bankruptcy and reorganization costs are of interest to purchasers of fixed claims because if these costs are incurred, there is less wealth available to satisfy existing fixed claims. Warner (1975) studies railroad bankruptcies and finds very small bankruptcy costs (as a fraction of the 27 value of the railroad three years before the bankruptcy occurred) for a sample of eleven railroad bankruptcies. Myers (1977) analyzes an agency problem of debt referred to as "the underinvestment problem". This problem occurs when stockholders forego low variance positive NPV projects because the benefits mostly accrue to bondholders. As leverage increases, this agency problem of debt becomes more severe. As with the other agency problems of debt, the agency costs are borne by the stockholders. Stiglitz and Weiss (1981) present a theoretical model of an equilibrium loan market characterized by credit rationing. In their model, banks are concerned with the interest rate they receive on a loan and the riskiness of the loan. Interest rates have screening properties, in that low risk borrowers will not pay high rates of interest. Assume that banks lend to two types of credit risks, good and poor credit risks. These two groups of borrowers are observationally identical. At low interest rates banks lend to both groups of credit risks. As interest rates are raised, however, the good credit risks drop out or switch to projects which have "lower probabilities of success but higher payoffs when successful" (p. 393). Banks offering loans at high interest rates will attract high risk borrowers. Therefore, each bank should have an optimal interest rate where the marginal income earned from making loans is equal to the marginal cost of lending to poor credit risks. Demand for credit may exceed supply at this 28 interest rate. But the bank will not raise the interest rate to bring demand down to supply. Instead credit rationing will occur. The Stiglitz and Weiss theory may help explain event risk and event risk covenant use. Some institutional investors, such as pension funds, do not want to invest in low-grade bonds. They also want to avoid investment grade bonds which may be downgraded due to a leveraged restructuring. A higher rate of return will not induce them to invest in low-grade bonds or investment grade bonds with high event risk because the risk of default is still present and this risk is incompatible with their risk preferences. These institutional investors may avoid corporate bonds altogether, or at least corporate bonds which have a significant probability of being downgraded in a restructuring, unless a provision, such as a "super poison put" provides them with a way to get rid of a downgraded bond without accepting the post-downgrade market price. 2.4 Ina Role of Indenture Covenants Smith and Warner (1979) describe competing hypotheses related to the use of bond covenants. The Irrelevance Hypothesis states that external markets have mechanisms for ensuring that stockholders maximize firm wealth rather than stockholder wealth. Therefore, stockholders should be indifferent between paying a higher rate of return on bonds and including a bond covenant in the bond indenture. Bond 29 covenants, according to this hypothesis, are neutral mutations. Fama (1978) supports the Irrelevance Hypothesis. He argues that if the firm does not follow a strategy to maximize total firm value, it will pay for outsiders to take over the firm and institute a value-maximizing strategy. The Costly Contracting Hypothesis states that covenants are costly and will not persist in bond contracts unless they are useful in reducing agency costs of debt. Jensen and Meckling (1976) support the Costly Contracting Hypothesis in their analysis of agency costs of debt. They refer to costs associated with bond covenants as monitoring costs and state that bondholders will use covenants up to the point where the "nominal" marginal cost of using covenants is equal to the marginal benefits. The term "nominal" is used because the cost of covenants is borne by stockholders, not the bondholders. This research hypothesizes that firms using event risk covenants have more severe agency problems of debt than firms not using these covenants. Empirical results which show a positive relation between magnitude of agency problems and the use of costly covenants support the Costly Contracting Hypothesis. Studies by Thatcher (1985) and Malitz (1986) investigate whether firm characteristics associated with agency problems of debt are related to the use of two-tiered call provisions (Thatcher) and to the use of dividend constraints, debt constraints, and sinking funds (Malitz). The results show a systematic relation between 30 firm characteristics which indicate severe agency problems of debt and complex contractual provisions. Kao and Wu (1990) investigate whether the probability of including a sinking fund provision in a bond contract is related to firm and bond characteristics. They show a systematic relation between firm characteristics which indicate severe agency problems and sinking fund use. This research focuses on the use of event risk covenants. Two previous studies examine event risk covenant use. Crabbe (1991) and Fields, Kidwell and Klein (1991) investigate whether event risk covenants are priced by the market.‘ Crabbe documents savings of 24 basis points for firms that include event risk covenants in new issue bond indentures. He also examines the secondary market and documents savings of 32 basis points for bonds with event risk covenants at the end of 1989. Fields, Kidwell and Klein relate the value of event risk covenants to the environment for corporate control. They find that before the RJR/Nabisco buyout, bonds with event risk covenants "sold for penalty yields compared with similar bonds without poison put provisions" (Fields, Kidwell and Klein, 1991, p. 19). After the RJR/Nabisco buyout, they find that new issue yields increased by 26.4 basis points and bonds with event risk covenants sold for 12.3 basis points less than similar bonds without poison put provisions. They also find that bonds with low ratings value event risk covenants more and that event risk 31 covenants with the highest S&P event risk covenant rankings save issuers the most basis points. Warga and Welch (1990) analyze bond maturity and bond rating in terms of their relation to the magnitude of bondholder wealth losses in LBO’s. They find that bonds with long maturities lose more wealth than bonds with short maturities. This is consistent with findings by Crabbe and Fields, Kidwell and Klein that bonds with long maturities include event risk covenants more often than bonds with shorter maturities. Warga and Welch also find that bonds with high ratings lose more wealth than bonds with lower ratings. This is inconsistent with the finding of Crabbe and Fields, Kidwell and Klein. Crabbe and Fields, Kidwell and Klein find that bonds with ratings below AA most frequently include event risk covenants. Fields, Kidwell and Klein also find that as rating decreases toward speculative grade, the market values event risk covenants more . 2.5 Summazy of Chappe; Two A major finding of studies reviewed in section 2.1 is that firms may increase leverage in a restructuring event in order to decrease agency problems between managers and stockholders. As agency problems between managers and stockholders decrease, however, agency problems between stockholders and bondholders increase. Empirical research reviewed in section 2.2 provides evidence that bondholder 32 wealth is reduced by leverage-increasing restructuring events. Section 2.3 reviews research which offers explanations of how bondholder wealth is reduced by an increase in firm leverage. The explanations focus on claim dilution, a risk incentive problem, and an underinvestment problem. Finally, section 2.4 looks at the role of bond indenture covenants in resolving agency problems of debt. Several empirical studies show a positive relation between severe agency problems of debt and the use of call provisions, sinking funds, dividend constraints and debt constraints. This dissertation seeks to provide further insight into the role of bond covenants in reducing agency problems. More specifically, it empirically analyzes the relation between firm characteristics which proxy for agency problems and event risk covenant use. Chapter three develops a theoretical basis for the empirical analysis that follows. Chapter 3 Hypothesis Development Smith and Warner (1979) present two competing hypotheses about how firm value is influenced by debt contracts. "The Irrelevance Hypothesis is that the manner of controlling the bondholder-stockholder conflict does not change the value of the firm." (Smith and Warner, 1979, p.120) "The Costly Contracting Hypothesis is that control of the bondholder-stockholder conflict through financial contracts can increase the value of the firm." (Smith and Warner, 1979,p.121) Empirical evidence which shows a systematic relation between covenant use and firm characteristics which proxy for agency problems is more consistent with the Costly Contracting Hypothesis. This research focuses mainly on the relation between covenant use and agency problems related to event risk. Event risk is an interesting problem to investigate because it is affected by manager/stockholder conflicts as well as stockholder/bondholder conflicts. The research addresses interdependencies that naturally exist between parties to the firm and examines how bond contracts are affected by these interdependencies. Theory is used to identify firm characteristics which proxy for agency problems related to event risk. The paragraphs that follow discuss relevant theory and identify firm characteristics which proxy for agency problems in the 33 34 empirical tests. The agency problem of free cash flow, agency problem of information asymmetry and agency problem of debt are considered relevant to event risk. _l3- WW Jensen’s (1986) free cash flow theory states that firms with few growth opportunities and excess free cash flow have incentives to increase leverage. Leverage can be increased by a leveraged buyout, a debt-financed share repurchase, or a debt-financed takeover. The discipline imposed by the additional debt burden reduces managers’ opportunities to waste cash flow. Prospective bondholders of firms with a free cash flow problem recognize the potential for a dramatic increase in leverage. A dramatic increase in leverage causes claim dilution and a more severe risk incentive problem. It is rational for prospective bondholders to demand compensation for event risk or require that an event risk covenant be present in the bond contract. Empirical findings which show that firms with excess free cash flow and few investment opportunities are likely to use event risk covenants provide support for the Costly Contracting Hypothesis. Findings which show no such relation provide support for the Irrelevance Hypothesis. 35 MW Jensen and Meckling (1976) show how agency problems of debt become more severe as debt ratio increases. The risk of bankruptcy rises and the risk incentive problem becomes more severe as leverage increases. Event risk is about future possible leverage increases. This research postulates that the debt ratio at the time a bond is issued influences event risk. Firms that already have a large amount of debt have less ability to remain solvent when an additional increment of debt is issued. They are, in a sense, ’close to the edge’. Stiglitz and Weiss (1981) present a model which shows how markets can fail when agency problems of debt are severe. The analysis assumes that good credit risks refuse to pay high interest rates. Some lenders are better off providing capital to the average borrowers at low rates than lending to high risk borrowers at high rates. Thus credit rationing occurs in equilibrium. The Stiglitz and Weiss theoretical framework offers an explanation for event risk covenant use. Institutional bond investors, such as pension funds, often prefer investment- grade debt. Riskier debt is not compatible with their investment objectives. Some of these investors face restrictions on the amount of speculative-grade debt they may carry in their portfolios. Thus, investment-grade bonds issued by firms which are likely to be involved in a leverage-increasing event are undesirable unless event risk 36 covenants are present. The closer a bond is to a speculative rating before an event, the more likely it is to become speculative-grade after an event. Empirical results which show a positive relation between debt ratio and event risk covenant use support the Costly Contracting Hypothesis. Results which show no significant relation between debt ratio and event risk covenant use support the Irrelevance Hypothesis. MW The agency problem of information asymmetry may lead to a situation in which firm value is maximized by including an event risk covenant in the bond indenture. (Barnea, Haugen and Senbet, 1985) Suppose a firm decides to issue bonds to fund a new project. Firm insiders (managers) believe the project is worth V;. Outsiders (prospective bondholders) believe the project is worth Vb, where Vb is less than V.’ The market price of the bonds will be VL‘unless insiders can send an unambiguous signal about the value of the project. The difference between V. and Vb is the agency cost of information asymmetry. High event risk firms are in a situation similar to the one described above. Prospective bondholders anticipate a leverage-increasing event and price bonds accordingly. The firm’s managers may believe prospective bondholders are overly pessimistic and are unfairly pricing the bonds. They can use an event risk covenant to signal that they believe 37 an event is unlikely. This is an effective signal because it is costly to mimic for firms in which an event is likely. Event risk covenants provide prospective bondholders with an opportunity to cash in their bonds for par value if both a leverage-increasing event and a bond rating downgrade occur. The agency problem of information asymmetry cannot be solved by market forces. Therefore, results which show a systematic relation between event risk covenant use and firm characteristics which proxy for information asymmetry provide insight into the role of information asymmetry in determining bond contracts without differentiating between the Irrelevance Hypothesis and the Costly Contracting Hypothesis. 3.4 Maaaga; Entrenchment - An Alternative Explanation Manager Ownership fraction appears to influence restructuring activity. Amihud and Lev (1981) and Lewellen, Loderer and Rosenfeld (1985) provide empirical evidence that firms with low manager ownership stakes may be more likely to attempt to discourage hostile takeovers than firms with high manager ownership stakes. Morck, Shleifer and Vishny (1988) hypothesize that mangers respond to two opposing forces: 1. Manager personal interests (which may conflict with shareholder interests). 2. Manager interests in the value of the firm’s equity. The relation between ownership and value depends on which of 38 the above forces dominate. Morck, Shleifer and Vishny (1990) find empirical evidence that managers’ personal objectives influence acquisition activities. Managers may use event risk covenants to make takeovers more difficult. If outstanding bonds have event risk covenants which are triggered in a takeover, the new management must either refund those bonds or raise the coupon rate (depending on the specific provisions of the covenant). According to theory, the managers most likely to use event risk covenants to discourage a takeover are those with a small own-firm ownership stake. Empirical tests which show a negative relation between manager ownership stake and event risk covenant use provide support for this explanation of event risk covenant use. élé Summarx This research relates event risk covenant use to severity of agency problems in an attempt to find support for either the Irrelevance Hypothesis or the Costly Contracting Hypothesis. Firm characteristics which proxy for agency problems are identified using existing theory. Results which show no systematic relation between proxy variables and event risk covenant use support the Irrelevance Hypothesis. 2": i ‘I r. W. \ l .. 1 39 Results which show the following relations provide support for the Costly Contracting Hypothesis: 1. a negative relation between firm size and event risk covenant use: a negative relation between future investment opportunities and event risk covenant use; a positive relation between debt level and event risk covenant use: a positive relation between information asymmetry and event risk covenant use; Managerial Entrenchment is examined as another explanation for event risk covenant use. Results which show a negative relation between manager own-firm ownership stake and event risk covenant use provide support for the managerial entrenchment explanation. Chapter 4 Data This chapter presents a description of the data used in this research. Section 4.1 describes how sample bonds and firms are identified. Section 4.2 discusses financial statement data used in the-calculation of proxy variables for agency problems. Section 4.3 discusses how information on specific event risk covenants is obtained. Data used in the calculation of the information asymmetry proxy and manager ownership proxy are discussed in sections 4.4 and 4.5, respectively. Data on overall covenant sets for sample bonds is described in section 4.6. 4.1 Sappla Bonds aag Firms To examine the relation between agency problems and event risk covenant use, all U.S. industrial firms issuing publicly traded investment grade bonds between January 1, 1989 and December 31, 1990 are identified. This study focuses on industrial firms because other industry groups often have regulatory constraints which may influence the likelihood of a leverage-increasing event. Investment grade bonds are selected because speculative grade bonds rarely have event risk covenants. The study period is chosen because this is when event risk covenants were most heavily 40 41 used.1 Among the identified firms, only those issuing at least one bond with a face value of greater than or equal to $25 million and a maturity of greater than or equal to five years are included in the sample. This criteria is established to facilitate comparison with other studies on event risk covenant use. Firms issuing only zero coupon bonds are also excluded from the sample.2 'This is because _ 1..--..”1 firm characteristics which influence the choice of zero coupon bonds versus coupon bonds may also influence decisions related to event risk covenant use. Moody’s Bond Survey is used to identify new issue bonds. In 1989, 62 firms issuing 114 bonds meet sample criteria. In 1990, 64 firms issuing 117 bonds meet sample criteria. Panel A of Table 1 presents a brief description of sample bonds. Appendix 4.1 lists sample firms, the dollar face value of the debt they issue during the sample period, and whether or not the debt had event risk protection. ‘According to Fields, Kidwell and Klein (1991), one bond with an event risk covenant was issued between October 21, 1988 and December 31, 1988. 2Three firms are excluded from the sample in 1989 because they issue only zero coupon bonds. Seven firms are excluded in 1990 because they issue only zero coupon bonds. 42 TABLE 4.1 21321.3 lumber of Firms and Honda That Meet Sample Criteria During Period from January 1, 1989 to December 31, 1990 Calendar Year 1989 1990 Number of Bonds 114 117 Number of Firms , 64 62 Number of Bonds with 25 17 Event Risk Covenants Number of Convertible 5 1* Bonds The sample includes all industrial firms issuing investment- grade coupon bonds which have a face value greater than $25 million and a maturity of greater than 5 years during the sample period. 23221.9 number of Bonds in Bach standard and Poor Corporation Event Risk Ranking Category 1989 Bonds with Event Risk Covenants Ranking = E1 1 Ranking = E2 0 Ranking = E3 21 Ranking = E4 3 Ranking = E5 0 1990 Bonds with Event Risk Covenants Ranking = E1 4 Ranking = E2 0 Ranking = E3 9 Ranking = E4 3 Ranking = E5 1 The levels of protection as ranked by Standard and Poor Corporation are listed below: E1 Strong protection E4 Weak protection E2 Significant protection E5 Insignificant or E3 Some protection no protection * Three zero coupon convertible bonds were issued in 1990. Zero coupon bonds are not included in this sample. w" puuujathl 43 MW Financial statement data for sample firms is obtained from Compustat and is used to calculate proxy variables for firm size, investment opportunities, free cash flow, and leverage level.3 Compustat data is obtained for the fiscal years ending prior to the year in which sample bonds are issued.‘ A sample firm that issues several bonds in a fiscal year is observed only once for that year in empirical tests.5 There are two reasons for this practice. First, variables which measure firm-specific characteristics such as size and cash flow are identical for several observations if each bond is considered an observation. Second, some firms issue several small face value bonds and others issue one large face value bond. Thus, the number of bonds issued 3Two sample firms have very little Compustat data available. Lyondell Petrochemical was spun off from Arco in 1988. It appears on the Annual Compustat tape but many data items are unavailable. Conagra acquired Beatrice foods in 1988. It appears on the Annual Compustat tape but many 1989 data items are unavailable. ‘For sample firms which issue one bond in 1989 and/or one bond in 1990, the issue date of the bond(s) is used to determine which fiscal years financial statement data comes from. For example, if a bond is issued in November 1989 by a firm with a September fiscal year end, it is issued in the 1990 fiscal year. Financial statement data from fiscal 1989 (and fiscal years prior to 1989 for variables which require more than one year worth of data) is used to calculate proxy variables. 5A firm with a fiscal year-end month other than December may issue several bonds in a specified calendar year that are not in the same fiscal year; This does not occur here. If it did occur, the firm would be observed once for each fiscal year in which a bond is issued. DIP a 44 is not a useful measure. For firms which issue more than one bond in a fiscal year, the bond with the largest face value is chosen as the sample bond.‘ The sample bond is used to determine the firm’s event risk protection status and the fiscal years for which financial statement data is needed. MW Standard and Poor’s Creditweek provides information on whether or not a new issue bond has event risk protection. Standard and Poor’s Corporation (hereafter S&P) began ranking new issue bonds in terms of their event risk protection in July 1989. S&P’s ranking system assesses the strength of covenant protection against a sudden and dramatic decrease in credit quality. This ranking system does not assess the likelihood of an "event". S&P’s ranking system has five categories: E 1 Strong Protection E 2 Significant Protection E-3 Some Protection E 4 Weak Protection E 5 Insignificant or No Protection Bonds issued prior to July 1989 which have event risk covenants were assigned rankings in the months following the initiation of this service. New issues with event risk covenants were ranked throughout the rest of 1989 and 1990. “If the two largest bonds have the same face value, then the bond with the longest maturity is chosen. If the maturities are identical, then the bond issued first is chosen. 45 In 1991, event risk covenant rankings appeared infrequently as event risk covenant use became infrequent. Sample period issues of Creditweek provide information on which firms use event risk covenants, the ranking of the covenants, and a brief description of the covenant details. Sample bonds are coded "1” if a covenant ranked E-1, E-2, E-3, or E-4 is present. Sample bonds are coded "0" if a covenant ranked E-5 or no covenant is present.7 Panel 2 of Table 1 presents data on the number of sample bonds in each event risk covenant ranking category.l Twenty two firms issued 26 bonds with event risk protection in 1989. Eleven firms issued 16 bonds with event risk covenants in 1990. In terms of dollar face value, 19% of the coupon debt issued by U.S. industrial firms in 1989 has event risk protection and 10% of the coupon debt issued by U.S. industrial firms in 1990 has event risk protection. As mentioned previously, some firms issue more than one bond in a fiscal year. For these firms the event risk ranking of the bond with the largest face value is used to determine the firm’s event risk protection status. Most sample firms which issue more than one bond in a fiscal year either have event risk protection on all of their bonds or have no event risk protection on all of their bonds. Three firms, however, issue bonds which differ in event risk protection during the sample period. For all three of these 7Only one bond issued after January 1, 1989 had an E-5 covenant. 46 firms, the bonds with event risk protection have larger face values and longer maturities than the bonds without event risk protection.8 Most sample firms which issue bonds in both 1989 and 1990 either include event risk covenants in both years or do not use event risk covenants in either year. Four sample firms which issue bonds in both years include event risk covenants in only 1989 or only 1990, however. In three of the four firms it appears that bond maturity may influence the decision to include an event risk covenant. 4.4 Inside; Ownership Data Data on the fraction of equity owned by people who are considered insiders at a corporation is provided by Value Line Investment Survey. Value Line publishes quarterly company evaluations for investors. For each sample bond, data on insider ownership is collected from the Value Line analysis which is within 1.5 months of the bond issue date. The objective is to use insider ownership information which would have been considered current by managers engaged in approving new borrowing agreements. For most sample firms, ownership fraction is stable from quarter to quarter. For firms which 8An alternative rule is to classify firms as having event risk protection if the face value amount of bonds with event risk protection is greater than the face value amount of bonds without event risk protection. This rule results in the same event risk status for sample firms as the rule which chooses the bond with the largest face value. 47 issue more than one bond in a fiscal year, the ownership fraction associated with the bond which has the largest face value is chosen as the insider ownership fraction for the firm. If the two largest bonds have the same face value, the bond with the longest time to maturity is chosen. If the two largest bonds have the same face value and time to maturity, the bond with the earliest issue data is chosen. Appendix 4.1 of this chapter lists sample firms by insider ownership fraction. This appendix also lists "other significant ownership" fraction as reported in Value Line. For twelve firm observations, the information in Value Line does not provide a clear picture of insider ownership. For example, the Value Line information on insider ownership of Ford Motor Company states that "Ford family, officers and directors own 9% of stock, have 40% of voting power”.9 Insider ownership fraction was coded as a missing value for these firms. 4.5 a s Forecast V riabi it s a Meas e o Infgpmation Asymmetpy IBES analysts’ earnings forecasts are used in a measure of information asymmetry. Earnings forecasts for the fiscal year end preceding the sample bond(s) issue date(s) are collected from the I/B/E/S U.S. detail tapes. Only new or newly updated forecasts made six months before the fiscal 9This information is from Value Line Investment Survey, December 23, 1988. 48 year end are used. The decision to use forecasts made six months before the fiscal year end is arbitrary. There is no theoretical reason for choosing the mid—year forecast over beginning or end of the year forecasts. It is important that forecasts from the same month of the fiscal year be chosen for all sample firms. Otherwise, differences in variability of forecasts across firms might be due to the fact that some firms are near the end of their fiscal years while others are at the start or midpoint. As a firm approaches the end of its fiscal year, much uncertainty regarding year-end earnings is resolved. gag ve a l Covenant Sets 0 e ' s Covenant sets of sample firms are evaluated by examining bond indenture descriptions in Moody’s Industrial Manual (1992). Covenant sets are examined during three time periods: 1. January 1985 - December 1988 2. January 1989 - December 1990 3. January 1991 - June 1992 These three time periods are chosen to provide a summary of how covenant use varied before, during and after the time period during which event risk covenants were popular. 49 Data on individual bonds issued during the specified time periods is collected and recorded. The following information is included: 1. 2. 3. 4. Bond maturity, face value, coupon rate and rating: Type of bond (such as subordinated debenture, convertible bond, or medium term note); Purpose of the bond issue: Underwriter(s):. Also, note is made of the presence of the following indenture covenants or provisions: a. b. c. d. e. f. g. h. i. j. call provision event risk covenant sale/leaseback covenant security provision limit on additional secured debt limit on dividends and other cash payouts limit on total funded debt postmerger net worth restriction put provision sinking fund APPENDIX 4 . 1a List of Sample Firms 1989 Sample Firms Firm Name AAR Corp Alcan Aluminum American Brands Anadarko Petroleum Anheuser-Busch Arco Chemical Arkla, Inc. Ashland Oil Baxter International Becton Dickinson Boise Cascade Borden Bowater Caterpillar Chrysler Coastal Corporation Deere & Co. Dillard Dept. Stores Dow Chemical Dupont de Nemours Eastman Kodak Eaton Corporation Exxon Fleming Companies Ford Motor Co. General Electric General Motors Grumann Corporation Home Depot IBM ITT Corporation Knight-Ridder Limited (The) Lockheed Corporation Loews Corporation Lyondell Petrochemical Maytag McDonalds Monsanto Occidental Petroleum Oryx Energy Penn Central Pennzoil 50 APPENDIX 4.18 Face Value 4 ($ millions) 65 150 100 100 842 125 200 200 250 100 250 150 300 300 250 200 350 100 150 300 650 100 400 150 850 700 $1550 s 200 $ 225 s 750 $1400 3 200 s 100 s 300 $1075 $ 300 s 175 $ 200 $ 100 $1506 5 175 $ 200 $ 550 mmmmmmmmmmmmmmmmmmmmmmmmmm Event Riak Covenant Yes No No No Yes * No Yes No No Yes Yes * Yes Yes No No Yes No No No No No Yes No No No No No Yes No No No Yes No Yes No Yes No No Yes No No No No 5 1 APPENDIX 4.1a (cont’d) Firm Name Face Value Event Riak (S milliona) Covenant Philip Morris $1250 No Pitney Bowes $ 250 No Potlatch Corporation $ 100 Yes Procter & Gamble $ 150 No Ralston Purina $ 200 No Rockwell International S 300 No Rohm and Haas $ 100 Yes Safety-Kleen $ 100 Yes Sears Roebuck $ 600 No Sequa Corporation $ 150 Yes Sonat, Inc. $ 100 No Super Valu Stores $ 45 Yes Tenneco $ 350 No Texaco $ 600 No Texas Instruments $ 150 No Times Mirror Co. $ 100 No Union Camp 5 100 Yes United Technologies $ 400 Yes * VF Corporation $ 200 Yes Westvaco $ 200 No Xerox $ 900 No # This column provides the $ face value of all fixed rate coupon bonds issued in 1989. i A portion of the new issue bonds have event risk protection. In terms of $ face value, 71% of Anheuser- Busch’s new issue bonds, issue bonds and 75% of United Technologies’ new issue bonds have event risk protection. 60% of Boise Cascade’s new 52 APPENDIX 4.1a (cont’d) 1990 Sample Firms Firm Name Face Value #4 Event Risk ($ millions) Covenant Alcan Aluminum $ 500 No Anheuser-Busch $ 100 No Arco Chemical $ 625 No Arvin Industries $ 50 No Boise Cascade $ 325 Yes Burlington Resources $ 300 No Campbell Soup $ 100 No Capital Cities/ABC $ 250 No Caterpillar $ 300 No Champion International $ 350 No Chesapeake Corporation $ 55 No Coastal Corporation $ 250 No Coca Cola $ 250 No Comdisco $ 200 No Conagra $ 400 No Cyprus Minerals 5 150 Yes Dayton Hudson $ 650 No Dillard Dept. Stores $ 50 No Dow Chemical $ 200 No Dupont de Nemours $ 600 No Eastman Kodak $ 750 No Exxon $ 250 No Ford Motor $4202 No General Electric $1550 NO General Motors $2250 No Georgia Pacific $ 600 Yes Home Depot $ 200 No International Paper $ 400 Yes ITT Corporation $ 350 No Johnson and Johnson $ 250 No Kimberly Clark $ 100 No May Dept. Stores $ 325 No Maytag $ 200 Yes McDonalds $ 100 No McGraw Hill $ 250 No Morrison Knudsen $ 500 No Morton International $ 200 Yes Nynex $ 450 No Occidental Petroleum $ 150 No Penn Central $ 150 Yes Penny (J.C.) $ 500 No Pennzoil $ 250 No Philip Morris $1600 No Phillips Petroleum 5 300 No Premark International S 100 No Firm Name Rohm and Haas Scott Paper Sears Roebuck Tenneco Texaco Times Mirror TRW, Inc. Union Camp Union Pacific Unisys Unocal Wal Mart Westinghouse Westvaco Weyerhauser Whirlpool Xerox APPENDIX 4.1a (cont’d) Face Value (8 millions) mmmmmmmmmmmmwmmmm 53 250 550 300 325 400 100 100 100 100 300 500 500 300 200 200 200 400 Event Risk Covenant Yes No No No No No No No No Yes No No No No Yes No No ## This column provides the $ face value of all fixed rate coupon bonds issued in 1990. APPENDIX 4 . 2a 54 APPENDIX 4.2a Sample Firm Insider Ownership Fractions 1989 Sample Firms Firm Name Insider Other Signif. Ownership Ownership AAR Corp _.O66 .000 Alcan Aluminum .003 .000 American Brands .005 .000 Anadarko Petroleum .020 .120 Anheuser-Busch .130 .110 Arco Chemical .005 .834 Arkla, Inc. .015 .000 Ashland Oil .020 .000 Baxter International .022 .000 Becton Dickinson .023 .000 Boise Cascade .045 .000 Borden .005 .000 Bowater .005 .000 Caterpillar .006 .000 Chrysler .010 .000 Coastal Corporation .054 .150 Deere & CO. .006 .101 Dillard Dept. Stores n.a. n.a. Dow Chemical .014 .000 Dupont de Nemours .220 .230 Eastman Kodak .005 .000 Eaton Corporation .010 .000 Exxon .005 .000 Fleming Companies .050 .000 Ford Motor Co. n.a. .000 General Electric .005 .000 General Motors .010 .000 Grumann Corporation .020 .426 Home Depot .140 .000 IBM .006 .000 ITT Corporation .005 .110 Knight-Ridder .400 .000 Limited (The) .360 .000 Lockheed Corporation .017 .221 Loews Corporation .240 .000 Lyondell Petrochemical .005 .500 Maytag .015 .000 McDonalds .080 .000 Monsanto .005 .000 Occidental Petroleum n.a. n.a. Oryx Energy .005 .260 Penn Central .341 .000 Pennzoil .030 .000 55 APPENDIX 4.23 (cont’d) Firm Name Insider Other Signif. Ownership Ownership Philip Morris .005 .000 Pitney Bowes .005 .000 Potlatch Corporation .140 .000 Procter & Gamble .122 .000 Ralston Purina ..070 .000 Rockwell International .017 .000 Rohm and Haas .470 .000 Safety-Kleen .066 .123 Sears Roebuck .005 .150 Sequa Corporation .380 .000 Sonat, Inc. .007 .114 Super Valu Stores .014 .088 Tenneco .010 .000 Texaco .005 .000 Texas Instruments .140 .000 Times Mirror Co. n.a. .000 Union Camp .005 .000 United Technologies .005 .080 VF Corporation .005 .231 Westvaco .050 .140 Xerox .010 .000 -4 -'-ll“a- 56 APPENDIX 4.2a (cont’d) 1990 Sample Firms Firm Name Insider Other Signif. Ownership Alcan Aluminum .002 .000 Anheuser-Busch .130 .110 Arco Chemical .005 .834 Arvin Industries .170 .000 Boise Cascade .045 .000 Burlington Resources n.a. n.a. Campbell Soup .580 .000 Capital Cities/ABC .216 .000 Caterpillar .006 .000 Champion International .005 .000 Chesapeake Corporation .182 .000 Coastal Corporation n.a. n.a. Coca Cola .050 .068 Comdisco .300 .000 Conagra .070 .000 Cyprus Minerals .005 .280 Dayton Hudson .005 .000 Dillard Dept. Stores n.a. n.a. Dow Chemical .014 .000 Dupont de Nemours .220 .230 Eastman Kodak .005 .000 Exxon .005 .000 Ford Motor n.a. n.a. General Electric .005 .000 General Motors .010 .000 Georgia Pacific .020 .000 Home Depot .140 .000 International Paper .005 .000 ITT Corporation .005 .110 Johnson and Johnson .016 .086 Kimberly Clark n.a. n.a. May Dept. Stores n.a. n.a. Maytag .080 .000 McDonalds .010 .070 McGraw Hill .060 .000 Morrison Knudsen .030 .000 Morton International .014 .000 Nynex n.a. n.a. Occidental Petroleum n.a. n.a. Penn Central .349 .000 Penny (J.C.) n.a. n.a. Pennzoil .030 .000 Philip Morris .005 .000 Phillips Petroleum .190 .000 57 APPENDIX 4.2a (cont’d) Firm Name Insider Other Signif. Ownership Premark International .025 .094 Rohm and Haas .470 .000 Scott Paper .014 .000 Sears Roebuck .005 .160 Tenneco .010 .000 Texaco .005 .000 Times Mirror n.a. .000 TRW Inc. .012 .000 Union Camp .005 .000 Union Pacific .010 .000 Unisys .005 .000 Unocal .090 .000 Wal Mart .420 .000 Westinghouse .005 .000 Westvaco .050 .140 Weyerhauser .077 .000 Whirlpool .020 .000 Xerox .010 .000 n.a. Either no information is available in Value Line regarding insider ownership fraction or the information provided is not comparable with the information provided for the other sample firms. CHAPTER 5 Description of Variables which Proxy for Agency Problems This chapter presents the empirical model to be tested and describes how the independent variables used in this model are calculated. It also discusses the empirical methodology used in this research. Chapter 3 examines theories and arguments which are useful in identifying firm characteristics which proxy for agency problems. The amount of free cash flow and future investment opportunities are identified as characteristics which proxy for agency problems of free cash flow. Leverage level is identified as a proxy for agency problems of debt. Firm size and the variability of analysts’ earnings forecasts are identified as proxies for information asymmetry. Manager own-firm ownership stake is identified as a proxy for an agency problem related to manager entrenchment. This chapter describes how these characteristics are quantified in the independent variables. 58 59 £11 EmEiIi£§l_EQQ§l A cross-sectional logit regression is used to analyze the relation between agency problems and event risk covenant use. Equation (1) presents the logit model. (1) ln(P,)/(1-Pi) = a0 + <11SIZEi + aZOPPORi + a3CASHi + a‘INFOR‘ + (stETDEBTi + c16YEARi + “WINS“: + “3‘30““: where Pi = the probability that COVPROTj equals "1". COVPROT‘ is a b1nary variable which is coded "1" if firmi has issued one or more bonds with event risk protection and "0" otherwise. SIZEi = the size of firm. OPPORi = future investment opportunities or growth prospects of firmi. CASHi = the amount of free cash flow available to managers of firm INFORi = the severity of information asymmetry that characterizes firm. NETDEBTi = the leverage level of firmi. YEARi = a binary variable to indicate whether an observation is from 1989 or 1990. YEARi is coded "1" if an observation is from 1989 and "0" if it is from 1990. FINSUBi = a binary variable to indicate whether a significant portion of the firm’s assets support finance and/or insurance activities. FINSUB is coded "1" if the firm has significant finance or insurance activities. COMBINi = an interaction variable, NETDEBTi * FINSUBi. The model shown in equation (1) is a logit transformation of the logistic model. When the response variable is binary, several assumptions of ordinary least squares (OLS) regression are violated. Logistic regression 60 gets around the problems encountered with OLS. Appendix 5.1 discusses the problems encountered with OLS and how logistic regression overcomes them. In the present analysis of event risk covenant use, a bond is categorized as having event risk protection if Standard and Poor’s Corporation gives it a ranking of E-l, E-2, E-3 or E-4. A bond with no ranking or a ranking of E-5 is categorized as having no event risk protection. As stated previously, the levels of protection, as ranked by Standard and Poor’s, are E-l Strong event risk protection: E-2 Significant event risk protection: E-3 Some event risk protection; E-4 Weak event risk protection: E-5 Insignificant event risk protection. 3W 5‘21 21pm Size Firm size is measured by the natural logarithm of the book value of firm total assets (Compustat item #6) at the fiscal year end prior to sample bond issuance. Natural logarithm of book value, rather than unadjusted book value of firm assets is used in order to minimize the effects of outlier firms on logit regression results. Panel A of Appendix 5.2 provides statistics on the firm size variable. A list of the five smallest and five largest observations is included. Other empirical research uses a variety of proxies for 61 firm size. For example, Banz (1981) and Fama and French (1992) use market value of equity as a size proxy. These two studies relate firm size to the risk/return characteristics of equity. Barclay and Smith (1993) use the natural logarithm of the market value of firm total assets as a size proxy.1 IMalitz_(1986) proxies for size with the natural logarithm of the book value of total assets in her research relating firm characteristics to the use of several bond covenants. Book value is preferred to market value of assets in this research because it results in a size proxy which does not overlap with the variable which measures future investment opportunities of the firm. Market value of assets overlaps because it includes the expected value of future projects. Total asset book value, instead of equity book value, is preferred as a size proxy in order to prevent leverage level from influencing the size measure. Leverage is measured with another variable. As discussed in Chapter 3, results which provide support for the Costly Contracting Hypothesis show a negative relation between firm size and event risk covenant use. Small firms are likely to have relatively severe information asymmetry which is reduced by event risk 1Barclay and Smith calculate the market value of total assets by subtracting book value of equity from book value of total assets and adding’ market value of equity to the resulting difference. 62 covenants. LEW Quality of future investment opportunities is proxied for with the ratio of market value of common equity to book value of common equity (hereafter ME/BE). The numerator of this ratio takes into account the market consensus present value of expected future projects. It is calculated as closing stock price (Compustat item #24) multiplied by the number of common shares outstanding at year end (Compustat item #25). The denominator is total common equity (Compustat item #60). Data for the fiscal year end preceding sample bond issuance is used. Panel B of Appendix 5.2 provides statistics on the investment opportunity variable. A list of the five smallest and five largest observations is included. ME/BE (or its reciprocal BE/ME) is used in other research as a measure of quality of investment opportunities. Fama and French (1992) state that "firms the market judges to have poor prospects, signaled here by low stock prices and high rates of book to market equity, have higher expected returns than firms with strong prospects." They do note, however, that BE/ME may just capture "the unraveling (regression toward the mean) of irrational market whims about the prospects of firms." Barclay and Smith (1993) use a ratio of market value of total assets to book value of total assets (MA/BA) as a 63 proxy for investment opportunities or "growth options". This research looks at MA/BA in tests of robustness of the investment opportunity proxy variable. There is not a strong theoretical argument for favoring the use of assets over equity for this proxy. Barclay and Smith (1993) also use earnings price ratio (E/P) and annual research and development expense divided by firm value as alternative proxies for future investment opportunities. They state that they expect firms with relatively few growth options to have high E/P and low research and development expenses. Lehn and Poulsen (1989) use recent historical sales growth rates as proxies for future investment opportunities. They reason that if sales have been growing rapidly in the past 2-year or 4-year period, then it is probable that they will grow rapidly in the near future. Lang, Stulz and Walkling (1991) use Tobin’s q ratio to distinguish between firms that have positive net present value investment opportunities and those that do not. Tobin’s q ratio is defined as the market value of a firm’s assets to their replacement cost. They state that high q ratio firms are likely to have positive prospects. Tobin’s q ratio is a theoretically superior measure of future investment opportunities because replacement cost provides more information to decision-makers than book value. However, replacement cost is very difficult to measure and accounting data related to replacement cost is 64 not widely available for firms after 1987. Thus, ME/BE instead of Tobin’s q ratio is used in this research. Results which show that firms with poor investment prospects are more likely to use event risk covenants than firms with good investment prospects support the Costly Contracting Hypothesis. 5.232225351111211 The free cash flow proxy variable used in this research is identical to that used by Lehn and Poulsen (1989). It is measured using data from the fiscal year end prior to sample bond issuance. Lehn and Poulsen’s cash flow equation is shown in (2). (2) CF = INC - TAX - INTEXP - PFDDIV - COMDIV where INC = operating income before depreciation, (Compustat item #13): TAX = total income taxes, (Compustat item #16), minus change in deferred taxes from the previous year to the current year (change in Compustat item #35); gross interest expense on short- and long-term debt (Compustat item #15): total amount of preferred dividend requirement on cumulative preferred stock and dividends paid on noncumulative preferred stock (Compustat item #19): total dollar amount of dividends declared on common stock (Compustat item #21). INTEXP PFDDIV COMDIV Lehn and Poulsen scale CF by market value of equity. Panel C of Appendix 5.2 provides statistics on the free cash flow variable. 65 Lang, Stulz and Walkling (1991) use Lehn and Poulsen’s CF measure as a proxy for free cash flow. They scale CF by book value of total assets because "...depending on the stochastic process followed by cash flow, an increase in cash flow can increase, decrease or leave unchanged the ratio of cash flow to market value of equity." (P.319) Based on this reasoning, CF is scaled by book value of total assets in this paper. Lehn and Poulsen’s cash flow measure is appropriate for this research because it provides cash flow information after all payments to suppliers of capital have been made. The free cash flow problem refers to the propensity of managers to waste the residual or "free" cash flow rather than pay it out to shareholders. Lehn and Poulsen’s measure ignores cash flow distortions due to accrual accounting however. A measure suggested by Compustat appears to account for some effects of accrual accounting. This measure is calculated as income before extraordinary items plus depreciation and amortization and is scaled by book value of total assets. As discussed in Chapter 3, according to Jensen’s free cash flow theory, firms with relatively large amounts of free cash flow have severe agency problems between managers and owners and may lever up to decrease these agency problems. Results which show a systematic positive relation between the cash flow measure and event risk covenant use provide support for the Costly Contracting Hypothesis. They 66 also provide support for Jensen’s free cash flow theory. MW The independent variable INFOR proxies for information asymmetry between firm insiders, such as managers, and firm outsiders, such as prospective bondholders. A cross- sectional standard deviation of IBES analysts’ earnings forecasts is calculated for sample firms. The fiscal year for which data is obtained is the year which precedes sample firm bond issuance. The earnings forecasts used are those made six months before fiscal year end. Only new and newly- verified forecasts are included in the standard deviation calculation. For example, Anadarko Corporation issued a bond in 1989. Its fiscal year end occurs on December 31. The standard deviation of forecasts of 1988 year-end earnings which were made in June 1988 are calculated. If an Anadarko analyst does not submit a new forecast in June 1988 but verifies to IBES that his/her previous forecast still holds, the verified forecast is included in the standard deviation calculation. The selection of forecasts made six months prior to fiscal year end as opposed to forecasts made earlier or later in the year is arbitrary. The objective of specifying a particular month from which forecasts are taken is to control for the resolution of uncertainty that occurs as an earnings announcement date approaches. For example, if Firm A’s fiscal year ends July 31 and Firm B’s fiscal year ends 67 December 31, and June forecasts are used for all sample firms, then Firm A’s analysts would face less uncertainty than Firm B’s analysts regarding the profitability of the current fiscal year. The standard deviation of analysts’ forecasts is scaled by fiscal year-end stock price. The mean earnings forecast is not appropriate as a scale factor because it may be equal to zero or a negative number or may vary systematically by industry, leading to a distortion of the measure of earnings variability. Panel E of Appendix 5.2 provides statistics on the information asymmetry variable. The use of this proxy variable assumes that variability of forecasts can be attributed to information asymmetry. To the extent that other factors contribute to the variability of forecasts, this proxy variable will be of limited usefulness. Malitz (1986) uses firm size, measured as total asset book value, as a proxy for information asymmetry. As stated previously, size is used in this research as a proxy for information asymmetry as well as a measure of risk. It is expected that firms with severe problems of information asymmetry use event risk covenants more frequently than firms with less severe problems of information asymmetry. Thus, INFOR is expected to be positively related to the use of event risk covenants. - m .1": i. ‘ 68 EIZQ LQYQIEQB Firm debt level is measured by Bruner’s (1988) net debt ratio in which debt is reduced by the cash and near-cash assets of the firm. Consideration of cash and near-cash assets provides a clearer picture of the firm’s bankruptcy risk. The net debt ratio is calculated as follows: Net Debt Ratio Net Debt/(CS + PS + Net Debt) - ixA‘i - a" where Net Debt book value of long-term debt (Compustat item #9) plus book value of short-term debt (Compustat item #34) less cash and short-term investments (Compustat item #1); C8 = common stock market value (Compustat item #24 multiplied by item #25): PS = preferred stock book value (Compustat item #130). Panel D of Appendix 5.2 provides statistics on the net debt ratio. Bruner (1988) also looks at a traditional debt ratio in order to test the robustness of his findings. The debt ratio is calculated as follows: Debt Ratio = Debt/(CS + PS + Debt) where Debt = book value of long-term debt plus book value of short-term debt (as calculated above): CS = as above: PS = as above. This research also uses the traditional debt ratio in robustness tests. 69 Theory tells us that firms with high debt ratios have relatively severe agency problems of debt. Thus it is expected that there is a positive relation between event risk covenant use and debt ratio. LEW This binary variable indicates whether a sample firm has a substantial financial subsidiary. The line of business description in Moody’s Industrial Manual is examined to determine if financing activities are one of the firm’s main activities. If financing activities are one of the firm’s main activities, then FINSUB equals one. Otherwise it equals zero. The presence of substantial finance operations at some firms may decrease comparability among sample firms. FINSUB is included in the model to capture systematic influences due to differences in industrial and financial operations. Many firms have finance subsidiaries which are not a major part of their business. A variation of the logit model sets FINSUB equal to one if the firm has a finance subsidiary, regardless if it is a major part of the firm’s business or not. Finance subsidiaries are identified using lists of subsidiaries found in Moody’s Industrial Manual. 5. 7 Inpepaction between Levarage and Financial Snpsidianias COMBIN is an interaction variable between FINSUB and DEBT. It is used to determine if there is a systematic 70 relation between debt ratio and event risk covenant use for firms which have substantial finance subsidiaries. Finance companies are characterized by higher debt ratios than non-finance companies. The presence of these firms in the sample could obscure the relation between debt ratio and event risk covenant use in nonfinancial companies. LAW The logistic model in equation (1) is estimated using the method of maximum likelihood. The principle of maximum likelihood estimation (MLE) is to choose a set of coefficient estimates that "imply the highest probability or likelihood of having obtained the observed sample Y." (Aldrich and Nelson, p. 51, 1984) Appendix 5.1a shows how maximum likelihood estimates are calculated. An Iteratively Reweighted Least Squares (IRLS) algorithm is used to solve the log likelihood equation. Wald chi square statistics are calculated and used in tests of the null hypotheses that the individual coefficients equal zero. The Wald chi square statistic is calculated as the square of the coefficient estimate divided by its variance estimate. Coefficient estimates reflect the effect of a change in an independent variable on the logarithm of the odds ratio, log[Pi/(1-Pi]. The effect of a change in an independent variable on P3 is not constant since the relation between a given variable and event risk covenant use is not linear. APPENDIX 5 . 1a w-*yr . ..— 71 APPENDIX 5 . 1a LOGIT REGRESBION A; , 0 class .28 So a ‘: -;.,-1 _s _1-o.-_o-_a : The empirical analysis in this research is concerned with factors affecting the decision to use or not use an event risk covenant. The dependent variable is qualitative with two possible outcomes. Ordinary least squares regression (hereafter OLS) is inadequate for estimating parameters when the dependent variable is qualitative because the error terms have undesirable properties. Specifically, if we assume that E(y1-) =PI[Y1=1] =x,’p then the probability that.y3 equals one is unbounded and may be outside the unit interval. Also, since y} can only take two values then ei can only take two values. with probability and 91 = -11,” with probability 72 1 '81,“ when yi = 0 With this probability structure e” is heteroskedastic. var(e1)_=E[yi] (1-EIY11) Thus, OLS produces unbiased but not best coefficient estimates. Hypotheses tests of the coefficients will be invalid. 51W In this research we are modeling choice behavior of managers/firms in determining whether or not to use event risk covenants. Each dependent variable observation has a Bernoulli distribution. flyi) = (pi)”1(1-p,)1"’1 A particular probability distribution for ei must be chosen. The most common choices are the normal and logistic distributions. The logistic distribution is chosen here.2 The logistic CDF is 2The choice of the logistic distribution is arbitrary. Aldrich and Nelson (1984) note that "the logistic and normal curves are so similar as to yield essentially identical results." (p.34) 73 1 1+exp(-t) F(t)= The logistic random variable variance is The logistic distribution is symmetric with zero mean. Choice of the logistic distribution for ei results in the logit statistical model. Logit leads to probabilities that are confined to the 0,1 interval. g; Eapaneter Estination The object of estimation is a vector of unknown parameters, 8. Maximum likelihood estimation is used to estimate 8 in cases where there is one decision per individual decision maker. With T independent observations, the joint probability or likelihood function is T L= fl f(y,)= II Pf‘u-Ppl'“ (i=1) (i=1) F(xfp)’1[1-F