Do bondholders receive benefits from bank interventions?

Yili Lian (Pennsylvania State University, Worthington, Scranton, Pennsylvania, USA)

Review of Accounting and Finance

ISSN: 1475-7702

Publication date: 14 May 2018

Abstract

Purpose

The purpose of this study is to examine the effect of bank interventions on bond performance in relation to loan covenant violations.

Design/methodology/approach

This paper tests the following questions: do bondholders receive benefits from bank interventions? Is bond performance related to the probability of bank interventions? Is the turnover of a chief executive officer (CEO) associated with bank interventions and bond performance? Abnormal bond returns, the difference between bond returns and matched bond index returns are used to measure bond performance. An estimated outstanding loan balance is used to measure the probability of bank interventions. CEO turnover is identified from proxy statements and categorized into forced and voluntary CEO turnovers. Event studies and regression analysis were used to answer the above research questions.

Findings

This paper finds that both short-term and long-term bond returns increase after covenant violations, bond performance is positively related to the probability of bank interventions, forced CEO turnovers are positively associated with the probability of bank interventions and firms with forced CEO turnovers tend to have superior bond performance.

Originality/value

This paper is the first to explore the relation between bank interventions and bond performance.

Keywords

Citation

Lian, Y. (2018), "Do bondholders receive benefits from bank interventions?", Review of Accounting and Finance, Vol. 17 No. 2, pp. 177-197. https://doi.org/10.1108/RAF-09-2016-0148

Download as .RIS

Publisher

:

Emerald Publishing Limited

Copyright © 2018, Emerald Publishing Limited


1. Introduction

A growing number of literature in finance has studied the implication of covenant violations on corporate investment and capital structure decisions (Chava and Roberts, 2008; Roberts and Sufi, 2009), dividend policy (Bulan and Hull, 2013), chief executive officer (CEO) compensation (Francis et al., 2011), shareholder value (Nini et al., 2012) and corporate innovations (Gu et al., 2014). However, the implication of bank interventions on corporate bonds has not been sufficiently explored. In this paper, the effect of bank interventions on bond performance surrounding loan covenant violations is tested. As argued by Nini et al. (2012), banks are actively involved in the governance of corporations in technical default events and create shareholder value. This paper demonstrates that bondholders also receive benefits from bank interventions. As shown by Datta et al. (1999), bank cross-monitoring diminishes duplicative monitoring of bondholders and reduces bond yield. In addition to bank monitoring, this paper shows that bondholders free ride on bank interventions in firm governance.

Traditionally, corporate creditors are considered passive bystanders until firms default. Essentially, banks influence firms well before bankruptcy. Unlike public bonds, bank debt is tightly held, even for large syndicated loans. When a firm breaches a financial covenant, technical default is triggered and control rights shift to the bank. Therefore, banks can threaten to accelerate the existing bank loan and choose a desired course of action. As Shleifer and Vishny (1997, p. 757) argue:

Significant creditors, such as banks, are also large and potentially active investors. […] Their power comes in part because of a variety of control rights they receive when firms default or violate debt covenants. […] As a result of having a whole range of controls, large creditors combine substantial cash flow rights with the ability to interfere in the major decisions of the firm.

The traditional view on control rights shows that creditors can only influence firm decisions in states of payment defaults when the firm value is less than the creditors’ claim. However, creditors can influence firm decisions well before the state of payment default. As defined by Nini et al. (2012), the “mixed region” of corporate governance reflects deterioration in firm performance, though before payment defaults, such as when a firm violates a loan covenant. In a “mixed region,” both creditors and shareholders can influence firm decisions, and the actions of both these players are important to a firm’s corporate governance.

There are two reasons to believe that bank interventions create value for bondholders. First, banks discipline firm investments, dividend policies and capital structures to improve firm performance (Nini et al., 2009; Nini et al., 2012). Considering covenant violations, investments and dividends are cut, and leverage and operating costs are effectively controlled. Second, banks are involved in CEO hiring decisions. Both Ozelge and Saunders (2012) and Nini et al. (2012) show that banks exert influence over CEO turnover. An underperforming CEO is likely to exit a firm when a bank intervenes with the governance of that firm.

Alternatively, bank interventions could reduce the value of bondholders. As a large and active debtholder, a bank can renegotiate with the violator and obtain superior existing loan terms. For instance, Nini et al. (2012) compare loan terms for a sample of lines of credit that are renegotiated following a new covenant violation and show that banks are more likely to require collateral after renegotiation, which dilutes the claims of the existing bondholders. Therefore, whether bank interventions create value for bondholders remains an empirical issue.

This paper tests the above two hypotheses using event studies of bond returns. If bondholders benefit from bank interventions, the abnormal bond returns, which are deducted by matched bond index returns, should be significantly positive around covenant violations. The study finds that the value of bondholders increases after covenant violations. The average abnormal bond return is 1.71 per cent surrounding loan covenant violations, and the average cumulative abnormal bond returns are 9.10 and 17.36 per cent for 12 and 24 months after a covenant violation, respectively.

Further, this paper studies the relation between cross-section abnormal bond returns and the probability of bank interventions. When a firm violates loan covenants, banks can choose to waive the violation or renegotiate loan terms with the violator to gain more control rights, or even threaten the firm and push it to change its decisions. This study hypothesizes that if a bank is more likely to intervene, then the bondholders will gain more value if the bank intervention creates value for bondholders, and if the bank is less likely to intervene, then the bondholders will gain less value after covenant violations. However, we can only empirically observe whether a loan contract is renegotiated after covenant violations. Direct bank interventions on firm decisions are rarely available, though there are certain pieces of anecdotal evidence. Therefore, the probability of bank interventions ex ante by using the amount of outstanding loans is estimated. To test the validity of this proxy of bank interventions, this study shows that it has predictive power to explain the ex post probability of loan renegotiation after controlling for other firm characteristics. The study shows that abnormal bond returns are positively related to the probability of bank interventions.

Finally, this paper shows that forced CEO turnover is a possible mechanism through which banks intervene to improve bond performance, and that underperforming CEOs are likely to exit violating firms when the probability of bank interventions is high. It also finds that long-term bond returns are positively correlated with the probability of CEO turnover. In addition, long-term bond returns are higher for firms with forced CEO turnover in comparison to firms without a CEO turnover. Similar results were not found for firms with voluntary CEO turnover.

To better understand the relation between bank interventions and bond performance, consider the case of Magellan Health Services, Inc. (Figure 1), the nation’s largest managed behavioral health company at one point in time. On November 1, 2002, Magellan Health Services, Inc. obtained an interim waiver of its financial covenants through December 31, 2002, from bank lenders. Without the reprieve, bank lenders could have demanded an accelerated repayment schedule, potentially forcing the firm into bankruptcy. On the same day, Daniel S. Messina, CEO of the company, was forced to resign. According to Jones (2012):

Daniel S. Messina’s decision to step down came as Magellan’s banks granted a temporary reprieve, assuring that the company won’t go into technical default on its $300 million in bank debt at least until December 31 (2002).

From historical bond trading prices, the last average net trading price on September 12, 2002, which was before the news announcement, was 84.3 per cent of the face value, whereas the first average net trading price on December 31, 2002, which was after the news announcement, was 97.8 per cent. The price increased by approximately 16 per cent surrounding this event. Interestingly, Moody’s bond rating dropped from B3 to Caa2 during the same period.

The paper contributes to literature on bank interventions. Prior literature suggests that bank interventions are associated with a decline in investment (Chava and Roberts, 2008; Nini et al., 2009), a reduction of financing activities (Roberts and Sufi, 2009) and an improvement of corporate governance, particularly CEO turnover (Nini et al., 2012; Ozelge and Saunders, 2012). In addition to the findings of prior studies, this paper shows that passive bondholders receive benefits from bank interventions, and cross-section bond returns are positively associated with the probability of bank interventions. This paper also shows that forced CEO turnover contributes to the positive effects of bank interventions on bond performance.

The paper also contributes to literature considering the relation between bond price and corporate governance mechanisms. Existing literature shows that bond yields are associated with managerial incentives (Ortiz-Molina, 2006), anti-takeover provisions and laws (Chen, 2011; Mansi et al., 2009), shareholder control (Cremers et al., 2007) and bank cross-monitoring (Datta et al., 1999). Besides the previously mentioned corporate governance mechanisms, this paper shows that bank interventions are positively related to bond performance. It also provides evidence that bonds with forced CEO turnover tend to have superior long-term performance.

The remainder of this paper is organized as follows. Section 2 discusses related literature. Section 3 describes data and methodology. Section 4 reports the empirical results, and Section 5 provides concluding remarks.

2. Related literature

This paper is related to the fast-growing literature on covenant violations and the implication on corporate decisions. As argued by Shleifer and Vishny (1997), banks have unique resources to influence their borrowers, and they could be very active in monitoring. Covenant violations provide a special setting where banks have the incentive and power to exert their influence on firms. Exiting literature provides significant evidence to prove this view.

Banks are active in monitoring corporate investment decisions. Chava and Roberts (2008) show that corporate investments sharply decline following financial covenant violations. Creditors use the threat of accelerating loan payments to influence firm decisions. Investment reduction is concentrated on firms with severe agency and information asymmetry problems, and creditor influence mitigates investment distortions. Using private credit agreements between banks and borrowers, Nini et al. (2009) show that capital expenditure restriction is common in such agreements, and that the restriction is sensitive to a borrower’s credit quality and contract terms and constrains firm investments. In addition to reducing inefficient investments, Mariano and Giné (2015) argue that creditors encourage efficient investments when firms have worthy investment opportunities.

Banks are active in monitoring corporate financial decisions. Roberts and Sufi (2009) find that net debt issuance declines sharply following covenant violations. The effect of creditor intervention on financing decisions is stronger when external financing is costly for firms. Using country-level creditor rights data, Brockman and Unlu (2009) show that creditors exert significant influence over corporate dividend policy. Creditors restrict dividend payout in weak creditor rights regimes.

Banks influence the corporate governance of borrowers. Nini et al. (2012) show that covenant violators experience a sharp decline in acquisitive and capital expenditures and leverage and dividend payouts. More importantly, they also experience a sharp increase in CEO turnover. The results suggest that banks exert informal influence on corporate governance of covenant violators. They also find that both firm operating performance and stock returns improve after bank interventions. Ozelge and Saunders (2012) find that high levels of outstanding bank loans are associated with high probabilities of forced CEO turnovers. This effect is significantly larger when underperforming firms violate loan covenants.

Prior literature shows that a bank’s involvement in corporate governance creates positive externality for bondholders. For instance, Datta et al. (1999) show that bank cross-monitoring reduces at-issue yield spreads. Using Japanese firm data, Cai et al. (1999) suggest that bank lending generates significant amounts of information and improves the contracting environment of public debt.

Prior literature provides evidence that bond prices are associated with other corporate governance mechanisms. Ortiz-Molina (2006) shows that managerial incentives are positively associated with bond yields. Bondholders obtain high bond yields by anticipating the future risk choices contained in managerial incentive structures. In corporate control markets, takeovers increase the leverage of target firms and expropriate the wealth of bondholders of those firms. Therefore, empirical findings show that bond yields are low for firms with anti-takeover provisions, such as classified boards (Chen, 2011), firms located in anti-takeover hostile states (Qiu and Yu, 2009; Francis et al., 2010) and firms located in states with extensive creditor protection, such as payout restriction laws (Mansi et al., 2009). In addition, Cremers et al. (2007) show that the impact of shareholder control on bond prices depends on the takeover vulnerability. Shareholder control is positively related to bond yields if a firm is exposed to takeovers, and vice versa.

This paper explores the governance role of banks after covenant violations and studies bond performance and forced CEO turnover surrounding covenant violations. This paper contributes to existing literature by showing that bank interventions after covenant violations improve corporate governance and provides evidence that forced CEO turnovers significantly increase after covenant violations and is positively associated with future bond performance. While existing literature finds that bank interventions are positively related to stock performance, this paper shows that bank interventions create positive externality for bondholders.

3. Data and methodology

3.1 Sample

A sample of non-financial firms from the USA with covenant violation information was obtained for 1996-2008 from Amir Sufi’s website[1]. Nini et al. (2012) collect the covenant violation information from annual (10-K) or quarterly (10-Q) Securities and Exchange Commission (SEC) filings. The construction of covenant violation data was initiated with data for all nonfinancial US firms in Compustat, with book assets greater than $10m in terms of the dollar value in the year 2000. Nini et al. (2012) use a text search algorithm that first locates the word “covenant” in the firm SEC filings, and conditional on finding this word, the authors search five terms: “waiv,” “viol,” “in default,” “modif” and “not in compliance.” The detailed sample selection process has been described by Nini et al. (2012) in the Appendix.

Because this paper analyzes the effect of bank interventions on bond performance, it requires that a firm has at least one bank loan and one bond outstanding. Bank loan data are obtained from the Loan Pricing Corporation (LPC) DealScan database (DealScan). Loan balance is constructed using the loan origination date, loan end date, loan type and loan size. The DealScan data set comprises firm filings and bank reports and has reasonable coverage of historical bank loans. According to Carey and Nini (2007), the DealScan data set has information on 50-70 per cent of all US commercial loan volumes up to the early 1990s, with coverage increasing to 80-90 per cent from 1992 to 2002.

Information of bond characteristics is collected from the Mergent Fixed Income Securities Database, and information on bond prices is collected from the Mergent National Association of Insurance Commissioners (NAIC) Bond Transaction Database. NAIC bond data include all purchases and sales of public fixed income securities by insurance companies that are required to report all their bond trades to the NAIC. The data cover only insurance company bond transactions. However, insurance companies hold approximately one-third of all public corporate bond issues and account for a quarter of all high-yield bond transactions (Campbell and Taksler, 2003; Hong and Warga, 2000). Considering that the violation sample starts from the year 1996, the NAIC data set is a reasonable data source for bond trading prices before the implementation of the Trade Reporting and Compliance Engine. Barclays Bond Indices are collected with different maturities and ratings from the DataStream database to calculate abnormal bond returns. To complement the results of the bond trading prices, bond quote data are obtained from DataStream to calculate bond returns for a three-day event window.

To calculate the control variables, firms’ accounting information is collected from the Compustat database, stock returns data from the CRSP database, institutional ownership data from the Thomson’s CDA/Spectrum database and CEO and director information from SEC proxy filings. The covenant violation information is collected from 10-Q or 10-K filings, and therefore, it coincides with earnings announcements. According to Datta and Dhillion (1993) and Easton et al. (2009), the bond market reacts to earnings announcements. Therefore, earnings surprise is calculated from the I/B/E/S database as a control variable. The Compustat and DealScan data are matched using the Compustat-DealScan link made publicly available by Michael Roberts and Wharton Research and Data Services (Chava and Roberts, 2008). To determine whether CEO turnover is voluntary, corporate news surrounding CEOs’ turnover date is collected from the Factiva database.

3.2 Variable construction

3.2.1 Measuring abnormal bond returns.

As shown by Cai et al. (2007) and Bessembinder et al. (2009), bonds are traded in different sizes and at different prices during a day. Therefore, the daily volume-weighted bond prices and accrued interests were used to calculate bond returns:

(1) Bond return=Pt+1Pt+Accrued interestsPt

For short-term bond returns, the last trading price before a violation announcement and the first trading price after a violation announcement are used to calculate the price changes. The last trading price before an announcement is required to be within a three-month pre-announcement window, and the first trading price after an announcement is required to be within a three-month post-announcement window. Kedia and Zhou (2014) use a similar rule to calculate bond returns surrounding acquisitions. As bonds are traded less frequently in comparison to stocks, the average time lag between the two trading prices is 52 days. Bond trading data are merged with the covenant violations data. The final sample comprises 471 bonds in event studies and 456 bonds in a two-year event window. Using daily bond quote data, three-day abnormal bond returns were obtained surrounding violation dates to complement the abnormal bond returns calculated by trading data. After merging bond quote data and covenant violations data, the final sample comprises 96 bonds. As the coverage of bond quote data is restrictive, the sample using bond quote data is smaller than the sample using bond trading data. Considering long-term bond returns, the last trading price close to the 12-month end is used to calculate the 12-month bond return, and the last trading price close to the 24-month end is used to calculate the 24-month bond return. Figure 2 shows the timeline of the construction of bond returns.

Further, abnormal bond returns are estimated by calculating the difference between bond returns and returns of matched portfolios, which are proxied by the Barclays Capital Bond Indices[2]. Default risk and time-to-maturity are controlled by matching credit ratings and maturities of bonds with covenant violations and Barclays Capital Bond Indices. Cai et al. (2007), Bessembinder et al. (2009) and Kedia and Zhou (2014) use this approach to calculate abnormal bond returns. As Barclays Capital Bond Indices comprehensively cover bonds, most of the bonds in the indices are issued by firms that do not violate loan covenants. Abnormal bond returns are the difference between bond returns of firms with covenant violations and bond index returns.

3.2.2 Proxy of probability of bank interventions.

When a firm violates a loan covenant, creditors can choose to waive the violation (no intervention), choose to renegotiate the terms of the loan with the violator or further intervene in the firm’s decisions. As argued by Nini et al. (2012), firm performance is influenced by creditors through changes in investment and financing policies. Therefore, the likelihood of creditors’ influence plays an important role in determining the performance of bonds as well. To estimate the probability and variation of bank interventions in violations, loan balance is used to measure the variation of bank interventions on violating firms. There are two reasons why loan balance is related to the probability of bank interventions. First, when the outstanding loan balance of a firm is high, more effort of the lender is required to improve firm performance and avoid greater loss. Outstanding loan balance is a natural variable to measure bank influence. Second, an outstanding loan balance can successfully predict the probability of future loan amendments, which is a form of bank intervention. As shown in Table I, outstanding loan balance is positively associated with future loan amendment six months after a covenant violation, supporting the conjecture that a loan balance is a valid proxy of the probability of bank interventions.

As loan balances are not available in Compustat, the approach of Allen et al. (2012) is adopted to estimate loan balances in the following manner. For term loans, linear interpolation is used to obtain the loan balances over time. For lines of credit, a 50 per cent utilization of the credit line is assumed. For example, if a firm borrows $5m with a five-year term loan and $2m with a three-year line of credit in 2000, the principle of the term loan reduces 20 per cent every year. Thus, the term loan balance in 2002 is $3m. Assuming a 50 per cent utilization of the lines of credit, the balance of the line of credit in 2002 is $1m and the loan balance for the firm for the same year is $4m. The main results of the paper are robust to different assumptions on utilization of lines of credit and interpolation of the term loan principle[3].

3.2.3 Forced chief executive officer turnover.

Data concerning CEO age, name and departure date are collected from the SEC proxy filings, and whether a departure is voluntary is determined by searching corporate news reports from the Factiva database. Following Parrino (1997), turnover is classified into forced and voluntary, based on the following criteria. All departures that the press reports as firings, forced departures, retirements or resignations due to policy differences or pressures are classified as forced. If the CEOs who exit are below 60 years of age and the press did not report their departures as death, poor health or the acceptance of another position or report a CEO’s retirement within six months of succession, the turnover is classified as forced. All remaining turnovers are considered voluntary.

Each firm is traced for two years following violation announcement. Therefore, it is possible that some firms subsequently became bankrupt. For CEO turnover as a result of bankruptcy, the method adopted by Lehn and Zhao (2006) is followed. If a firm fails to emerge from bankruptcy or if it emerges from bankruptcy but the CEO is replaced, the CEO turnover is considered as forced. The sample has 22 firms that became bankrupt within two years of a covenant violation.

3.3 Summary statistics

The final sample contains 471 bonds and 297 firms in the event windows, which is approximately 1.6 bonds per firm. Table II shows the bond, firm and board characteristics. The event[4] and 24-month abnormal bond returns are 1.71 and 17.36 per cent, respectively. The average bond maturity is 102 months, which is approximately eight years. The average offering amount is $360m.

The average leverage ratio is 50 per cent, market-to-book is 1.26 and return on assets (ROA) is 8.6 per cent[5]. The firm characteristics of the sample are weaker in comparison to those of the whole Compustat data set as the sample focuses on covenant violations. The median standardized earnings surprise, which is defined as the difference between the actual earnings per share (EPS) and analysts’ expected EPS scaled by stock price, is 0 per cent, and the mean standardized earnings surprise is −1.3 per cent. The average blockholder ownership is 24.84 per cent with 2.7 blockholders on average for a firm.

The average insider ownership is 12.8 per cent, which is the aggregate ownership of executives and directors on the board from proxy filings. The average board size is 9.2. The directors are categorized into insiders, gray directors and outsiders. The average number of outsiders on a board is 6.3. The percentage of outsiders is a measurement of board independence. On average, 15.5 per cent of firms have at least one bank director on the board. A total of 62.7 per cent firms have CEOs with the title of a “Chairman.” The average age of a CEO is 54 years.

The overall CEO turnover for two years following the covenant violation is 45.5 per cent, forced CEO turnover is 28.6 per cent and voluntary CEO turnover is 16.9 per cent. CEO turnovers are also classified in the table. Forced CEO turnover is 11.4, 9.8 and 7.4 per cent for covenant violation announcement year, one year and two years after the announcement year, respectively. Voluntary CEO turnover is 5.4, 6.4 and 4.7 per cent for covenant violation announcement year, one year and two years after the announcement year, respectively. Jenter and Kanaan (2015) study CEO turnover for all firms in the S&P ExecuComp database from 1993 to 2009 and show that the overall CEO turnover is 10.25 per cent, forced CEO turnover is 2.77 per cent and voluntary CEO turnover is 7.85 per cent. In comparison to CEO turnovers in Jenter and Kanaan (2015), forced CEO turnover in the sample of this study is higher and voluntary CEO turnover is lower as the sample of this study contains firms with covenant violations and banks are more likely to be involved in corporate governance.

Table III reports the bond ratings and bond provisions. The sample contains 128 investment grade bonds and 343 junk bonds. The number of bonds with a callable provision is 405, credit enhancement is 91 and putable provision is 11. The number of bonds placed under Rule 144a is 45. These bond characteristics are controlled in the multivariate analysis.

3.4 Empirical strategy

This paper uses an event study approach to analyze short-term and long-term bond performance surrounding covenant violation announcements. Further, a multivariate analysis is undertaken to explore the relation between bank interventions and abnormal bond returns. If bank interventions create positive values for bondholders, the coefficient of probability of bank interventions should be positive and significant in regressions. The relation between abnormal bond returns and the probability of bank interventions is tested by the following ordinary least squares (OLS) regression for bond i of firm j at time t:

(2) BondCARi,t=BankInterventionj,t+EarningsSurprisej,t+BondControli,t+FirmControlj,t+Rating+ShareholderMonitoring+εi,j,t

As independent variables in the regression model control various dimensions of firm and bond characteristics, and the dependent variable, bond CAR, is a continuous variable, the OLS regression method is used in the model. Similarly, Kedia and Zhou (2014) use the OLS regression method to study the information content of pre-announcement abnormal bond returns surrounding acquisitions. For a firm with multiple outstanding bonds, standard errors are clustered at the firm level, as the standard errors of bonds from the same firm are correlated. The dependent variable is the cumulative abnormal bond returns. The key independent variable is the probability of bank interventions, measured by the loan balance. Bank interventions are associated with corporate governance improvement, thereby enhancing bond performance. The coefficient of the key variable is expected to be positive and significant in regressions. As the announcement of covenant violations coincide with earnings announcements, standardized earnings surprise is used to control the unexpected surprise in the capital market, following existing literature. To disentangle the effect of bank interventions from the effect of shareholder monitoring, blockholder ownership and the number of blockholders in each firm in the sample are controlled. Shareholder monitoring reduces agency problems and is expected to be positively associated with bond performance. It is possible that bonds with different credit ratings react differently to bank interventions. Thus, credit ratings are controlled in regressions. Firm and bond characteristics are included in the regression as well. For instance, small firms are relatively easier to be intervened in comparison to large firms; thus, it is expected that firm size is negatively related to bond performance. Leverage is a proxy of a firm’s financial leverage and is expected to be negatively correlated with bond performance. ROA is a proxy of a firm’s profitability. On one hand, profitability is positively associated with bond performance because of good financial health conditions, and on the other hand, bank interventions are more likely to occur in low profitability firms. Thus, there is a possibility that ROA is negatively associated with bond performance. Bond characteristics, such as maturity, security and seniority, are included in regressions and are associated with the risk profile of bonds.

Further, this paper studies the effect of bank interventions on CEO turnover and their implications on bond returns. As the dependent variable is an indicator that takes a value of 1 if a CEO turnover occurs and 0 otherwise, the following probit regression model is used to study the relation between bank intervention and CEO turnover for firm j at time t:

(3) CEOTurnoverj,t=BankInterventionj,t+CorporateGovernanceVariables+FirmControlj,t+ShareholderMonitoring+εj,t

CEO turnover indicator is the dependent variable. Forced and voluntary CEO turnovers are tested separately in the regressions. The key independent variable is the probability of bank interventions, measured by the loan balance. As argued by existing literature, bank interventions improve corporate governance. Therefore, the coefficient of bank interventions is expected to be positive and significant for forced CEO turnover regressions and not significant for voluntary CEO turnover that is not driven by bank interventions. To separate the effect of bank interventions from the effects of shareholder monitoring and other corporate governance mechanisms, blockholder ownership and number of blockholders are included to control for shareholder monitoring, and a series of variables from proxy filings are included to control for other corporate governance channels, such as percentage of outside directors, board size and insider ownership. It is expected that independent board and shareholder monitoring are positively related to forced CEO turnover, and that the CEO-chairman duality and insider ownership are negatively related to forced CEO turnover. Firm characteristics such as firm size, leverage and profitability are included in the regressions. CEOs are more likely to be forced out of firms with high financial leverage and low profitability because of poor performance. Holding other factors constant, external corporate governance interventions are more likely to occur in small firms. Thus, CEOs of small firms are more likely to be forced out in comparison to those of large firms.

4. Empirical findings

4.1 Event studies

Panel A of Table IV reports the abnormal bond returns in short-term and long-term event windows. A covenant violation by itself represents negative news in the market as firms violate covenants by deteriorating their performance. However, considering the positive effect of bank interventions, it is shown that reactions to covenant violations could be positive in the bond market. Figure 3 illustrates the average abnormal bond returns for each violator from one year before a covenant violation to two years after a covenant violation. Considering that most firms have several bonds outstanding as shown in Table II, the abnormal bond returns are value-weighted if a firm has more than one bond in the sample. Time zero in Figure 3 represents the year of a covenant violation. The graph shows that abnormal bond returns increase dramatically after loan covenant violations.

In Panel A of Table IV, the average abnormal bond return surrounding the announcements of covenant violations is 1.71 per cent, and the 12-month and 24-month average abnormal bond returns following the announcements are 9.1 and 17.36 per cent, respectively[6]. Both the mean and median of abnormal bond returns are significantly different from zero. Firm-level abnormal bond returns are also measured and are also statistically different from zero. As argued previously in this paper, most firms have more than one bond outstanding. Firm-level bond returns are value-weighted. Considering that corporate bonds are not traded as frequently as stocks, the event window of bonds is 52 days on average, which is longer than a typical stock event window. To complement the results, the bond’s cumulative abnormal returns (CAR) are calculated from daily bond quote data from the DataStream database. Owing to the limited coverage of bond quote data, only 96 bonds were found with covenant violations. In Panel B of Table IV, three-day bond CAR [0, 2] are 0.376 per cent and CAR [−1, 1] are 0.596 per cent. Both are significantly above zero. Similar results are found in the sample using bond quote data.

4.2 Bank interventions and bond performance

Table V reports the regression results of bank interventions on abnormal bond returns from short-term and long-term event windows. The number of bonds in the cross-section regression is marginally smaller than the number in the event studies as the observations with missing firm characteristics are eliminated in the final sample. In cross-section, the probability of bank interventions positively correlates with short-term and long-term abnormal bond returns. The coefficients of the loan balance are robust in both event and long-term analysis after controlling for the monitoring effect of institutional investors. From Specifications (3) and (6), one standard deviation that increases in the logarithm of loan balance (1.3) corresponds to a 1.78 per cent increase in short-term abnormal bond returns (1.3 × 1.37 per cent) and to a 6.94 per cent increase in long-term abnormal bond returns (1.3 × 5.34 per cent). The results show that bond returns are positively related to the probability of bank interventions.

4.3 Probability of bank interventions, forced chief executive officer turnover and long-term bond performance

Table VI shows that forced CEO turnover is positively related to the probability of bank intervention. From Specification (1), 1 per cent increase of an average logarithm of loan balance is related to a 7.02 per cent increase of the probability of forced CEO turnover (marginal effects of loan balance at means in the probit regression). From Specifications (1) and (2), the probability of bank interventions is significantly related to forced CEO turnover, but not to voluntary CEO turnover. The results show that banks are influential in board decisions with respect to forced CEO turnovers. Specification (3) shows that the main results still hold when CEO turnover in Year 2 is excluded.

The study also shows that long-term abnormal bond returns are positively related to forced CEO turnovers. Forced and overall CEO turnover dummies are included in the long-term abnormal bond returns regression [equation (2)]. Table VII shows that forced CEO turnover is positively related to abnormal bond returns, whereas the overall CEO turnover, including voluntary CEO departure, is not significantly related to long-term abnormal bond returns. The evidence shows that bonds react positively to firms that discipline poor performing CEOs. Adams and Mansi (2009) show that CEO turnover events are associated with low bondholder values. The results of this study are different from those of Adams and Mansi (2009), as the latter study the general CEO turnover that is driven by shareholders, and it harms bondholders by increasing the uncertainty of future operating performance, such as asset restructuring. In this paper, CEO turnover is driven by creditors. As shown by Nini et al. (2012), creditor involvement is associated with improvement of operating performance, which in turn increases bondholder values.

4.4 Discussion

A possible reason to explain the positive bond returns surrounding covenant violations is fire selling. Bondholders may tend to fire sell bonds before the announcement date by expecting a possible default. Bond prices overreact to fire selling and bounce back after the covenant violation date. Wei and Zhou (2011) show that informed bond trading is concentrated around 10 days before an earnings announcement. However, the event window used in Panel A of Table IV is 52 days long on average. In addition, Wei and Zhou (2011) show that informed trading is mainly due to negative earnings news, while in the sample of this study, the median of earnings surprise is zero. Therefore, the results of this study are less likely to be driven by informed fire selling of corporate bonds.

Existing evidence shows that short-term abnormal stock returns are negative (Beneish and Press, 1995; Nini et al., 2012). Nini et al. (2012) argue that negative abnormal stock returns suggest that investors do not immediately incorporate future performance improvements into stock prices owing to arbitrage limits or sell-side pressures. The event studies of this paper show that short-term abnormal bond returns are positive. A possible explanation for the coexistence of negative stock returns and positive bond returns is the transfer of wealth by banks from shareholders to bondholders. If the wealth transfer story is true, more wealth will be transferred from shareholders to bondholders if the probability of bank interventions is higher. Therefore, cross-section stock returns are negatively associated with the probability of bank influence. Table VIII shows that cumulative abnormal stock returns surrounding covenant violations are positively related to the probability of bank interventions, which contradicts the wealth transfer story. When the probability of bank interventions is high, the stock returns are also high. In conclusion, short-term abnormal stock returns are negative because of the limits of arbitrage or sell-side pressures and are less affected by these factors when the probability of bank interventions is high.

5. Conclusion

This paper shows that bondholders receive benefits from bank interventions and provides evidence that both short-term and long-term abnormal bond returns increase significantly after covenant violations. Outstanding loan balance is used as a proxy of probability of bank interventions. The paper shows that the cross-section abnormal bond returns are positively related to the probability of bank interventions. The higher is the probability of bank interventions, the greater is the benefit to the bondholders. Additionally, forced CEO turnover is associated with the probability of bank interventions, and long-term bond returns are positively related to the probability of bank interventions and forced CEO turnovers. Finally, the paper analyzes the relation between abnormal stock returns and the probability of bank interventions and concludes that bank interventions have a positive effect on value for both shareholders and bondholders. Policy makers should encourage creditors to take actions in corporate governance practices. As a result of the positive interventions on covenant violators, operating efficiency of firms will increase and the value of shareholders and creditors will improve.

The findings of this paper suggest that banks play an active role in corporate governance. Creditor governance is significantly different from traditional corporate governance mechanisms as banks have unique resources and incentives to monitor and influence borrowers. Bank interventions on corporate governance create positive externality for bondholders. Consequently, passive bondholders benefit from bank’s influence on corporate governance. This paper sheds some light on this topic. One limitation of the paper is that it only studies the covenant violation events in the USA. It would be interesting to implement the research topic at an international level as the levels of creditor protection vary among countries. Different levels of creditor protection and culture provide a unique setting to study creditor governance, particularly bank interventions surrounding covenant violations. For instance, La Porta et al. (2000) argue that differences in laws and the effectiveness of law enforcement in different countries explain the differences in ownership concentration, breadth and depth of capital market, dividend policies and access to external finance. Bank interventions, as one of the corporate governance channels to influence firm activities, are affected by laws and their enforcement. Cross-country studies will provide a unique setting where creditors, shareholders and the legal environment interact with each other.

Figures

Magellan Health Service, Inc. bond trading prices

Figure 1.

Magellan Health Service, Inc. bond trading prices

Time line for calculating bond returns

Figure 2.

Time line for calculating bond returns

Cumulative abnormal bond returns

Figure 3.

Cumulative abnormal bond returns

Outstanding loan balance and the probability of loan amendments

(1) Amendment (2) Amendment
Loan 0.316*** (2.79) 0.350*** (2.83)
Logat −0.309*** (−2.89) −0.328*** (−2.66)
Lev −0.458 (−1.21) −0.569 (−1.52)
MB 0.167 (0.90) 0.219 (1.16)
ROA −0.956 (−0.87) −1.257 (−1.11)
Chairman 0.222 (0.93)
Insider ownership 0.191 (1.52)
Board size −0.366 (−0.76)
Outside 0.775 (0.75)
CEO age 0.986 (1.25)
d_bank 0.314 (1.00)
Intercept −0.552 (−0.72) −4.810 (−1.48)
N 249 247
Pseudo R2 0.041 0.069
Notes:

This table presents the probit regression of the probability of loan amendments on the outstanding loan balance. Loan amendment is identified when a loan is amended within six months of covenant violations. Loan variable is the natural logarithm of the outstanding loan balance. Definitions of other variables are available in the Appendix. Z-statistics are heteroskedasticity-robust. Significance at the 10%, 5% and 1 levels are indicated by

*

;

**

; and

***

, respectively

Summary statistics

Variable N Mean Median SD p25 p75
Panel A: Bond characteristics
Abnormal return (event) 471 0.0171 0.0122 0.127 −0.0176 0.0488
Abnormal return (24 months) 456 0.1736 0.1809 0.3551 0.0701 0.3267
Maturity (months) 471 102.03 79 86.33 52 111
Offering amount (mm) 471 360.22 275 313.26 200 400
Panel B: firm and board characteristics
Firm characteristics:
Leverage 282 0.499 0.479 0.225 0.343 0.619
Outstanding loan balance (mm) 282 1,015 458 1563 239 1,083
Loan balance (log) 282 6.20 6.13 1.303 5.479 6.988
Assets (mm) 282 5,481 1,961 9,217 832 5,024
ROA 250 0.086 0.092 0.089 0.056 0.132
Market to book 281 1.26 1.12 0.59 0.95 1.36
Standardized earnings surprise 197 −0.0134 0 0.0569 −0.004 0.0031
Stock return (past 12 months) 258 0.063 −0.051 0.8856 −0.484 0.365
Blockholder 238 0.2484 0.206 0.302 0.125 0.308
Number of blockholders 238 2.71 2 1.71 1 3
Board characteristics:
Board size 295 9.23 9 2.17 8 11
Outside 294 6.29 6 2.42 5 8
d_bank 297 0.155 0 0.362 0 0
Chairman 295 0.627 1 0.484 0 1
CEO age 295 54 54 7.49 49 59
Insider ownership 295 0.128 0.061 0.156 0.0282 0.1603
Panel C: CEO turnover
All turnovers 297 0.455 0.499 0 1
Year 0 297 0.168 0.375 0 1
Year 1 297 0.162 0.369 0 1
Year 2 297 0.121 0.327 0 1
Forced turnovers 297 0.286 0.453 0 1
Year 0 297 0.114 0.319 0 1
Year 1 297 0.098 0.297 0 1
Year 2 297 0.074 0.262 0 1
Voluntary turnovers 297 0.169 0.363 0 1
Year 0 297 0.054 0.182 0 1
Year 1 297 0.064 0.253 0 1
Year 2 297 0.047 0.226 0 1
Notes:

In this table, Panel A presents bond characteristics of firms with covenant violations. Panel B shows the firm-level statistics of firms with covenant violations in the sample period of 1996-2008. Panel C presents CEO turnover of firms with covenant violations following covenant violations

Distribution of bond ratings and bond provisions

Priority of claims N
Panel A. Payment hierarchy
Secured 37
Senior 333
Subordinate 1
Sub subordinate 97
Non 3
Total 471
Panel B: Credit ratings
Rating N
AAA 1
AA 4
A 40
BAA 83
BA 107
B 153
CAA 51
CA, C, D 21
NR 11
Total 471
Panel C: Bond provisions
Bond provisions N
Credit enhancement 91
Rule 144A 55
Putable provision 11
Callable provision 405
Notes:

In this table, Panel A presents payment hierarchy of bonds with covenant violations. Panels B and C show the distribution of credit ratings and bond provisions of those bonds, respectively

Event study

Mean Annualized
Mean
Z-statistics Wilcoxon
z stat
Positive (%) Obs
Panel A. Abnormal bond returns using trading data
Abnormal return (event) 0.0171 0.1146 2.92 4.92 60.50 471
Abnormal return (event, firm) 0.0228 0.1486 2.72 3.77 61.40 254
Abnormal return (12 months) 0.0910 0.0910 7.40 11.50 79.62 530
Abnormal return (12 months, firm) 0.1190 0.1190 1.92 6.58 75.10 257
Abnormal return (24 months) 0.1736 0.0868 10.44 13.532 86.40 456
Abnormal return (24 months, firm) 0.1383 0.0692 5.34 8.26 82.50 223
Panel B. Abnormal bond returns using quote data
Event window Bond CAR (%) t-statistic Observations
[−1,1] 0.596 2.26 96
[0,2] 0.376 1.70 96
Notes:

In this table, Panel A presents abnormal bond returns surrounding covenant violation announcements 12 months and 24 months following covenant violation. Event abnormal bond returns measure buy-and-hold abnormal bond returns surrounding violation announcements, while long-term abnormal returns measure buy-and-hold abnormal bond returns 12-months and 24-months after covenant violation announcements. Firm-level abnormal bond returns are bond value-weighted returns for each firm. Panel B presents abnormal bond returns surrounding violation announcements, using daily bond quote data. Abnormal bond returns are calculated in [−1, 1] and [0, 2] three-day windows

Abnormal bond returns and bank interventions

(1) CAR (event) (2) CAR (event) (3) CAR (event) (4) CAR (2 years) (5) CAR (2 years) (6) CAR (2 years)
Loan 0.0110* (1.80) 0.0144* (1.75) 0.0137* (1.96) 0.0511*** (2.74) 0.0393** (2.02) 0.0534*** (2.77)
SUE −0.146 (−1.37) −0.0852 (−1.43) −0.123 (−0.95) −0.0670 (−0.50)
Lev −0.0452 (−0.80) −0.117* (−1.95) −0.137*** (−2.99) −0.340 (−1.19) −0.449** (−2.29) −0.324 (−1.56)
ROA −0.000544 (−0.00) −0.182* (−1.84) −0.0579 (−0.50) −0.817** (−2.05) −1.027*** (−2.88) −0.802* (−1.79)
Logat −0.0184*** (−2.67) −0.0136 (−1.60) −0.0149* (−1.75) −0.00563 (−0.42) −0.0490* (−1.85) −0.0654** (−2.19)
MB −0.0285 (−1.34) −0.00174 (−0.19) −0.0282 (−1.64) 0.0656 (1.19) 0.000878 (0.01) 0.0180 (0.32)
Logm −0.00717 (−1.29) −0.00486 (−0.86) 0.0270 (1.28) 0.0273 (1.55)
Logs −0.0157 (−0.92) 0.000969 (0.07) 0.105** (2.07) 0.128** (2.52)
A and above 0.0306* (1.71) 0.0238 (1.29) −0.0614 (−1.09) −0.0505 (−0.96)
Baa 0.0120 (0.75) 0.0133 (0.78) −0.0653 (−1.42) −0.0525 (−1.43)
B 0.0154 (0.72) 0.0159 (0.90) −0.00328 (−0.06) −0.0836* (−1.67)
Caa 0.0450 (1.39) 0.0375 (1.32) 0.111 (1.32) 0.00464 (0.04)
C and below 0.186 (1.08) 0.256 (1.15) 0.185  (1.81) 0.0322 (0.49)
No rating −0.0538 (−1.51) −0.0638* (−1.72) −0.287** (−2.07) −0.271* (−1.68)
Secured or senior −0.0273 (−0.99) −0.0591** (−2.34) 0.0944 (1.57) 0.0671 (1.14)
Enhancement −0.0299 (−1.08) −0.0371** (−2.08) −0.0865 (−1.27) −0.0521 (−0.84)
Rule144a 0.0140 (0.77) 0.00369 (0.22) 0.0232 (0.26) 0.0150 (0.16)
Callable 0.0249 (1.10) −0.0106 (−0.58) −0.0564 (−0.82) −0.0672 (−1.13)
Putable −0.0164 (−0.52) −0.0326 (−0.87) −0.183 (−1.62) −0.223* (−1.95)
Block −0.00221 (−0.03) 0.107 (0.32)
Number of block −0.00872 (−0.48) 0.0761 (1.10)
Intercept 0.174** (2.22) 0.253*** (3.42) 0.271*** (3.44) 0.112 (0.48) 0.0754 (0.24) −0.176 (−0.59)
N 294 343 294 289 343 288
Adj. R2 0.056 0.054 0.152 0.119 0.176 0.218
Notes:

This table represents the regression results on the effect of bank interventions on abnormal bond returns. The dependent variables are event and 24-month abnormal bond returns. The loan variable represents the natural logarithm of outstanding loan balance, which measures the probability of bank interventions. The details of variable definitions and measurements of all other variables are reported in the Table A1. Coefficients of the credit ratings should be interpreted as incremental effects with respect to the Ba bonds. T-statistics are heteroskedasticity-robust and clustered by firm. Significance at the 10%, 5% and 1% levels are indicated by:

*

;

**

; and

***

, respectively

CEO Turnovers and bank interventions

(1) forced (2) voluntary (3) forced
Loan 0.228* (1.95) −0.072 (−0.98) 0.212* (1.91)
Lev 0.869** (2.12) −0.137 (−0.32) 1.172*** (2.68)
ROA −2.941*** (−2.67) −0.110 (−0.07) −2.719* (−1.67)
Logat −0.302*** (−2.76) −0.083 (−0.79) −0.319** (−2.25)
MB 0.0804 (0.49) −0.052 (−0.35) −0.306 (−0.76)
Chairman −0.0411 (−0.19) 0.341 (1.30) −0.194 (−0.71)
Outside 1.037 (1.25) −1.132 (−1.30) 0.313 (0.30)
d_bank 0.272 (0.95) 0.149 (0.46) −0.117 (−0.36)
CEO age −1.624** (−2.16) 2.914*** (3.32) −0.910 (−1.04)
Board size 0.532 (1.03) 0.391 (0.65) 1.056 (1.55)
Insider ownership 0.0574 (0.08) −2.980*** (−2.47) −1.050 (−0.94)
Block −0.958 (−0.79) −2.972* (−1.73) −1.464 (−0.96)
Number of block −0.0690 (−0.25) 0.472 (1.49) 0.218 (0.64)
Intercept 4.528 (1.52) −11.17*** (−3.38) 1.374 (0.38)
N 209 209 209
Pseudo R2 0.133 0.159 0.142
Notes:

This table shows the effect of bank interventions on forced CEO turnovers. The dependent variable for specifications (1) and (2) are forced and voluntary CEO turnovers, respectively. The dependent variable in specification (3) is forced CEO turnover, excluding Year 2’s data. The loan variable is the natural logarithm of the outstanding loan balance, which measures the probability of bank interventions. The details of variable definitions and measurements of all other variables are reported in the Appendix. Heteroskedasticity-robust z-statistics for probit regression are reported in parentheses. Significance at the 10%, 5% and 1% levels are indicated by

*

;

**

; and

***

, respectively

Bank interventions, CEO turnover and long-term bond performance

(1) CAR (two years) (2) CAR (two years)
Loan 0.0436** (2.52) 0.0493*** (2.63)
Forced turnover 0.219*** (2.66)
All turnover 0.110 (1.60)
Block 0.200 (0.66) 0.217 (0.71)
Number of block 0.0569 (0.93) 0.0546 (0.85)
SUE −0.00471 (−0.04) −0.0192 (−0.14)
Lev −0.319* (−1.88) −0.345* (−1.78)
ROA −0.561 (−1.34) −0.759* (−1.75)
Logat −0.0433 (−1.59) −0.0631** (−2.09)
MB −0.0108 (−0.22) 0.00667 (0.13)
Logm 0.0266 (1.58) 0.0297* (1.75)
Logs 0.104** (2.46) 0.112** (2.41)
A and above −0.122* (−1.82) −0.131 (−1.43)
Baa −0.0428 (−1.13) −0.0437 (−1.09)
B −0.0616 (−1.10) −0.0667 (−1.20)
Caa 0.0106 (0.11) 0.0149 (0.15)
C and below −0.0115 (−0.16) 0.0477 (0.66)
No rating −0.294** (−2.08) −0.275* (−1.72)
Secured or senior 0.0733 (1.27) 0.0601 (0.99)
Enhancement −0.0220 (−0.36) −0.0546 (−0.86)
Rule144a −0.0251 (−0.31) −0.0104 (−0.12)
Callable −0.0555 (−1.09) −0.0480 (−0.90)
Putable −0.207* (−1.91) −0.244** (−2.20)
Intercept −0.203 (−0.81) −0.104 (−0.35)
N 288 288
Adj. R2 0.276 0.233
Notes:

This table represents the regression results on the effect of bank interventions and CEO turnover on long-term bond performance. The dependent variable is the 24-month abnormal bond return. The loan variable is the natural logarithm of outstanding loan balance, which measures the probability of bank interventions. The details of variable definitions and measurements of all other variables are reported in the Table A1. Coefficients of the credit ratings should be interpreted as incremental effects with respect to the Ba bonds. T-statistics are heteroskedasticity-robust and clustered by firm. Significance at the 10%, 5% and 1% levels are indicated by

*

;

**

; and

***

, respectively

Bank interventions and abnormal stock returns

CAR0 CAR (0,1) CAR(−1,1)
Loan 0.00611*** (2.82) 0.00768*** (3.00) 0.00683** (2.11)
MB −0.0108* (−1.67) −0.0176 (−1.61) −0.0200* (−1.67)
Lev −0.0394** (−2.10) −0.0646** (−2.30) −0.0404 (−1.34)
Logat −0.0000478 (−0.02) −0.000733 (−0.20) −0.00124 (−0.31)
ROA 0.0864* (1.83) 0.145 (1.25) 0.201* (1.73)
SUE −0.0282 (−0.88) −0.0689 (−1.50) −0.101* (−1.92)
Block −0.0437 (−0.73) 0.0186 (0.25) 0.0473 (0.52)
Number of block 0.0105 (0.92) −0.00595 (−0.43) −0.00607 (−0.36)
Intercept −0.0822* (−1.97) −0.0848 (−1.48) −0.0914 (−1.39)
N 162 162 162
Adj. R2 0.040 0.054 0.054
Notes:

The table shows the relation between bank interventions and cross-section abnormal stock returns. Abnormal stock returns are estimated by the four-factor model, including three factors from Fama and French (1993) and a momentum factor. CAR0, CAR(0,1) and CAR(-1,1) represent one-day, two-day and three-day abnormal stock returns, respectively. The loan variable is the natural logarithm of outstanding loan balance, which measures the probability of bank interventions. The details of variable definitions and measurements of all other variables are reported in the Appendix. Heteroskedasticity-robust z-statistics are reported in parentheses. Significance at the 10%, 5% and 1% levels are indicated by

*

;

**

and

***

, respectively

Definition of variables

Variables Definition
Bank influence
loan Natural logarithm of the outstanding loan balance
Firm characteristics
logat Natural logarithm of the total assets of a firm
lev Book leverage
MB (Market value of equity + book value of debt)/total assets
ROA Previous four quarters’ EBIT/total assets
SUE Standardized earnings surprise
block Blockholder ownership
number of block Logarithm of number of blockholders
chairman CEO is also the chairman on a board
outside Percentage of outside directors on a board
CEO age Logarithm of CEO age
board size Logarithm of number of directors
d_bank A dummy equals 1 if bank directors are on a board
insider ownership Shareholding of executives and directors
forced CEO turnover All CEO departures that the press reports as firings, forced departures, retirements or resignations owing to policy differences or pressures are classified as forced. If the CEOs exiting are below 60 years and the press did not report their departures as death, poor health or the acceptance of another position or report a CEO’s retirement within six months of the succession, the turnover is classified as forced
Bond characteristics
logm Logarithm of bond maturity, measured by month. Logs: logarithm of bond offering amount
A and up A dummy equals 1 if the credit rating is equal or above A. BAA: a dummy equals 1 if the credit rating is BAA
BA A dummy equals 1 if the credit rating is BA
B A dummy equals 1 if the credit rating is B
CAA A dummy equals 1 if the credit rating is CAA
CA and down A dummy equals 1 if the credit rating is equal or below CA. No rating: a dummy equals 1 if no credit rating is available
secured or senior A dummy equals 1 if a bond is senior or senior and secured
enhancement A dummy equals 1 if a bond has credit enhancement characteristics. Rule144A: a dummy equals 1 if a bond is issued under rule 144A
callable A dummy equals 1 if a bond has a callable provision
putable A dummy equals 1 if a bond has a change of control provision

Notes

2.

Lehman Brothers Corporate Indices were renamed to Barclays Capital Bond Indices after the bankruptcy of Lehman Brothers in 2008.

3.

The main results are consistent under assumptions of 30 or 70 per cent utilization of lines of credit and the assumption of no interpolation or linear interpolation of term loans.

4.

The event abnormal bond returns refer to short-term bond returns surrounding covenant violations.

5.

Quarterly ROAs are converted into annual ROAs.

6.

The number of bonds in the short-term event window is smaller than the number of bonds in the 12-month window as the first trading prices observed more than three months after a violation are discarded.

Appendix

Table AI

References

Adams, J.C. and Mansi, S.A. (2009), “CEO turnover and bondholder wealth”, Journal of Banking & Finance, Vol. 33 No. 3, pp. 522-533.

Allen, L., Gottesman, A., Saunders, A. and Tang, Y. (2012), “The role of banks in dividend policy”, Financial Management, Vol. 41 No. 3, pp. 591-613.

Beneish, M.D. and Press, E. (1995), “The resolution of technical default”, Accounting Review, Vol. 70 No. 2, pp. 337-353.

Bessembinder, H., Kahle, K.M., Maxwell, W.M. and Xu, D. (2009), “Measuring abnormal bond performance”, Review of Financial Studies, Vol. 22 No. 10, pp. 4219-4258.

Brockman, P. and Unlu, E. (2009), “Dividend policy, creditor rights, and the agency costs of debt”, Journal of Financial Economics, Vol. 92 No. 2, pp. 276-299.

Bulan, L. and Hull, T. (2013), “The impact of technical defaults on dividend policy”, Journal of Banking & Finance, Vol. 37 No. 3, pp. 814-823.

Cai, J., Cheung, Y.L. and Goyal, V.K. (1999), “Bank monitoring and the maturity structure of Japanese corporate debt issues”, Pacific-Basin Finance Journal, Vol. 7 Nos 3/4, pp. 229-250.

Cai, N., Helwege, J. and Warga, A. (2007), “Underpricing in the corporate bond market”, Review of Financial Studies, Vol. 20 No. 6, pp. 2021-2046.

Campbell, J. and Taksler, G. (2003), “Equity volatility and corporate bond yields”, Journal of Finance, Vol. 58 No. 6, pp. 2321-2349.

Carey, M. and Nini, G. (2007), “Is the corporate loan market globally integrated? A pricing puzzle”, The Journal of Finance, Vol. 62 No. 6, pp. 2969-3007.

Chava, S. and Roberts, M.R. (2008), “How does financing impact investment? The role of debt covenants”, The Journal of Finance, Vol. 63 No. 5, pp. 2085-2121.

Chen, D. (2011), “Classified boards, the cost of debt, and firm performance”, available at: https://ssrn.com/abstract=1729472

Cremers, K.M., Nair, V.B. and Wei, C. (2007), “Governance mechanisms and bond prices”, Review of Financial Studies, Vol. 20 No. 5, pp. 1359-1388.

Datta, S. and Dhillion, U.S. (1993), “Bond and stock market response to unexpected earnings announcements”, Journal of Financial and Quantitative Analysis, Vol. 28 No. 4, pp. 565-577.

Datta, S., Iskandar-Datta, M. and Patel, A. (1999), “Bank monitoring and the pricing of corporate public debt”, Journal of Financial Economics, Vol. 51 No. 3, pp. 435-449.

Easton, P., Monahan, S. and Vasvari, F. (2009), “Initial evidence on the role of accounting earnings in the bond market”, Journal of Accounting Research, Vol. 47 No. 3, pp. 721-766.

Fama, E.F. and French, K.R. (1993), “Common risk factors in the returns on stocks and bonds”, Journal of Financial Economics, Vol. 33 No. 1, pp. 3-56.

Francis, B.B., Hasan, I. and Sun, X. (2011), “CEO compensation and covenant violations”, available at: https://ssrn.com/abstract=1786590

Francis, B.B., Hasan, I., John, K. and Waisman, M. (2010), “The effect of state antitakeover laws on the firm’s bondholders”, Journal of Financial Economics, Vol. 96 No. 1, pp. 127-154.

Gu, Y., Mao, C.X. and Tian, X. (2014), “Bank interventions and firm innovation: evidence from debt covenant violations”, Research Paper, Kelley School of Business.

Hong, G. and Warga, A. (2000), “An empirical study of corporate bond market transactions”, Financial Analysts Journal, Vol. 56 No. 2, pp. 32-46.

Jenter, D. and Kanaan, F. (2015), “CEO turnover and relative performance evaluation”, The Journal of Finance, Vol. 70 No. 5, pp. 2155-2184.

Jones, S. (2012), “Magellan health CEO quits amid overhaul”, Washington Post, 2 November.

Kedia, S. and Zhou, X. (2014), “Informed trading around acquisitions: evidence from corporate bonds”, Journal of Financial Markets, Vol. 18, pp. 182-205.

La Porta, R., Lopez-de-Silanes, F., Shleifer, A. and Vishny, R. (2000), “Investor protection and corporate governance”, Journal of Financial Economics, Vol. 58 Nos 1/2, pp. 3-27.

Lehn, K.M. and Zhao, M. (2006), “CEO turnover after acquisitions: are bad bidders fired?”, The Journal of Finance, Vol. 61 No. 4, pp. 1759-1811.

Mansi, S.A., Maxwell, W.F. and Wald, J.K. (2009), “Creditor protection laws and the cost of debt”, Journal of Law and Economics, Vol. 52 No. 4, pp. 701-726.

Mariano, B. and Giné, J.A.T. (2015), “Creditor intervention, investment, and growth opportunities”, Journal of Financial Services Research, Vol. 47 No. 2, pp. 203-228.

Nini, G., Smith, D.C. and Sufi, A. (2009), “Creditor control rights and firm investment policy”, Journal of Financial Economics, Vol. 92 No. 3, pp. 400-420.

Nini, G., Smith, D.C. and Sufi, A. (2012), “Creditor control rights, corporate governance, and firm value”, Review of Financial Studies, Vol. 25 No. 6, pp. 1713-1761.

Ortiz-Molina, H. (2006), “Top management incentives and the pricing of corporate public debt”, Journal of Financial and Quantitative Analysis, Vol. 41 No. 2, pp. 317-340.

Ozelge, S. and Saunders, A. (2012), “The role of lending banks in forced CEO turnovers”, Journal of Money, Credit and Banking, Vol. 44 No. 4, pp. 631-659.

Parrino, R. (1997), “CEO turnover and outside succession: a cross-sectional analysis”, Journal of Financial Economics, Vol. 46 No. 2, pp. 165-197.

Qiu, J. and Yu, F. (2009), “The market for corporate control and the cost of debt”, Journal of Financial Economics, Vol. 93 No. 3, pp. 505-524.

Roberts, M.R. and Sufi, A. (2009), “Control rights and capital structure: an empirical investigation”, The Journal of Finance, Vol. 64 No. 4, pp. 1657-1695.

Shleifer, A. and Vishny, R. (1997), “A survey of corporate governance”, Journal of Finance, Vol. 52 No. 2, pp. 737-783.

Wei, J. and Zhou, X. (2011), “Informed trading in corporate bonds prior to earnings announcements”, available at: https://ssrn.com/abstract=2023494

Corresponding author

Yili Lian can be contacted at: yxl50@psu.edu