read article and answer question

Creditor Control Rights, Corporate Governance, and Firm Value

Greg Nini The Wharton School, University of Pennsylvania

David C. Smith McIntire School of Commerce, University of Virginia

Amir Sufi University of Chicago, Booth School of Business, and NBER

We provide evidence that creditors play an active role in the governance of corporations well outside of payment default states. By examining the Securities and Exchange Commission’s filings of all U.S. nonfinancial firms from 1996 through 2008, we document that, in any given year, between 10% and 20% of firms report being in violation of a financial covenant in a credit agreement. We show that violations are followed immediately by a decline in acquisitions and capital expenditures, a sharp reduction in leverage and shareholder payouts, and an increase in CEO turnover. The changes in the investment and financing behavior of violating firms coincide with amended credit agreements that contain stronger restrictions on firm decision-making; changes in the management of violating firms suggest that creditors also exert informal influence on corporate governance. Finally, we show that firm operating and stock price performance improve post-violation. We conclude that actions taken by creditors increase the value of the average violating firm. (JEL G21, G32, G34)

In their influential survey,Shleifer and Vishny(1997) argue that “corporate governance deals with the ways in which the suppliers of finance to corpora- tions assure themselves of getting a return on their investment.” In light of this definition, a natural question is:whichinvestors exert influence over managers

For helpful comments, we thank Sanjai Bhagat, Jian Cai, Mitchell Petersen, Carola Schenone, Michael Weisbach (editor), anonymous referees, the participants at the 2009FIRS Conference, the 2010WFA Conference, and the seminar audiences at Boston College, the University of Colorado, the University of Virginia, Wake Forest University, George Mason University, the Federal Reserve Board, and the Federal Reserve Bank of Philadelphia. We are indebted to Dirk Jenter for sharing his data set on CEO turnovers and Robin Greenwood for sharing his data set on activist hedge funds. For excellent research assistance, we thank Sneha Chiliveru, JiMin Lee, Waldo Ojeda, Ruoyu Wang, Le Yang, Susan Yoon, and Lin Zhu. G.N. is grateful for the support of the W.E. Upjohn Institute of Employment Research. D.C.S. is grateful for financial support from the Research Council of Norway’sFinansmarkedsfondand the McIntire Center for Financial Innovation. A.S. is grateful for financial support from the Center for Research in Security Prices at Chicago Booth. Send correspondence to Greg Nini, 2439 Steinberg Hall–Dietrich Hall, Philadelphia, PA 19104; telephone: (215) 898-7770. E-mail: [email protected].

c© The Author 2012. Published by Oxford University Press on behalf of The Society for Financial Studies. All rights reserved. For Permissions, please e-mail: [email protected]. doi:10.1093/rfs/hhs007 Advance Access publication March 13, 2012

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 1 Corporate governance: Which investors influence managerial decisions? Panel A shows a traditional view of corporate governance, in which creditor influence over managerial decisions is limited to payment default states. Panel B shows a more creditor-oriented view, in which both creditors and equityholders exert influence over managerial decisions as firm value deteriorates, but the firm is not yet in payment default.

to assure a good return? Panel A of Figure1 presents the traditional view. According to this view, corporate governance refers primarily to the ability of equityholders to influence managerial decision-making through the board of directors, either directly or indirectly. Corporate creditors are thought to remain passive bystanders until firms are in default, which is typically associated with failure to make a payment, as in the models ofTownsend(1979), Gale and Hellwig (1985), andHart and Moore(1998). The current corporate governance literature almost exclusively reflects the traditional view. Indeed,Shleifer and Vishny (1997) note that “although there has been a great deal of theoretical discussion of governance by large creditors, the empirical evidence of their role remains scarce” (p. 757).1

1 There are important exceptions, includingKaplan and Minton(1994), Kang and Shivdasani(1995), Ivashina et al.(2008),Santos and Rumble(2006), andWruck (1990).Gilson(1990) provides evidence on the strong role of creditors in bankruptcy.Lim, Minton, and Weisbach(2011) show that lenders that are also equityholders in a

The Review of Financial Studies / v 25 n 6 2012

1714

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

In Panel B of Figure1, we present an alternative view, one in which creditor influence over managerial decisions extends outside of payment default states. Creditors begin to play an active role in corporate governance when firm performance deteriorates, but well before bankruptcy. In the “mixed” region, the actions taken by creditors may be as important as the actions taken by equityholders. In other words,both creditors and equityholders play an important corporate governance role.

In this study, we present evidence consistent with this alternative view. Using a sample of more than 3,500 financial covenant violations reported by the universe of U.S. nonfinancial, public firms in quarterly filings with the Securities and Exchange Commission (SEC), we show that violations are associated with more conservative financial and investment policy and a sharp increase in CEO turnover. Furthermore, we provide evidence that firm oper- ating performance and equity valuation improve, on average, post-violation, suggesting that actions taken by creditors aid in company turnarounds.

There are several reasons why financial covenant violations are ideal events for studying the influence of creditors on corporate governance. First, covenant violations convey the same contractual rights to creditors as do payment defaults, including immediate repayment of principal and termination of further lending commitments. These rights provide creditors with a strong hand in negotiations post-violation. Second, violations are common, indeed much more common than are payment defaults. We find that between 10% and 20% of public firms were in violation of a covenant during any particular year of our sample period, and more than 40% of the firms were in violation at some point during the period. Finally, violations occur well before a firm is in danger of a payment default. The median firm in our sample that is a first-time covenant violator has a market-to-book ratio above one, positive operating cash flow, and enough liquidity to easily cover their current liabilities.

Whereas creditors have the right to demand immediate repayment, financial covenant violations rarely lead to liquidation or bankruptcy. Instead, creditors use the prospect of a waiver of the violation to renegotiate the credit agreement and impose stronger contractual restrictions on the borrower. We show that amended credit agreements following covenant violations provide less fund- ing, have a shorter maturity, and carry a higher interest rate spread compared with the contracts prior to the violation. Amended agreements are also more likely to require collateral and contain more restrictive covenants on the cash management and capital expenditures of violating firms.

Creditors can also apply noncontractual control over violating firms via behind-the-scenes advice on how best to “manage through” the performance problems that caused the violation. AsDaniels and Triantis(1995, p. 1,101) describe, “[t]he voice of a lender ranges from advice and exhortation to

borrower obtain loan terms that are more favorable compared with lenders that do not hold equity positions in their borrower.

Creditor Control Rights, Corporate Governance, and Firm Value

1715

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

exclusive control over the firm’s decisions.” Suggested fixes could be as simple as demanding better reporting and liquidity management. But for companies suffering from deeper structural problems, creditors can affect more substantial changes to the organization, including pushing for the hiring of a turnaround management firm and, if necessary, the replacement of top executives.

We formally identify the effect of violations using “quasi-discontinuity” regressions (as inRoberts and Sufi 2009), which employ a large sample of vio- lating and nonviolating firms in a dynamic model of firm outcomes. The regres- sions include linear and higher-order controls for performance metrics that are commonly used in financial covenants. By flexibly controlling for these vari- ables, the impact of a covenant violation is identified by the discontinuity that occurs at the level of the violation. Identifying the violation-related changes in firm behavior is fairly straightforward because we often observe short-term reversals in outcomes in the quarter immediately following the violation.

Following a covenant violation, we document a decline in the size of the asset base of violating firms and a move to a more conservative financial policy, beginning in the quarter immediately following the violation. New expendi- tures on property, plant, and equipment and acquisitions decline significantly, and violating firms begin to shed assets through sales and disposals. By four quarters post-violation, violating firms have shrunk their assets by more than 2%, as compared with otherwise similar firms. Meanwhile, issuance of new debt all but stops post-violation and balance sheet debt declines by almost 10%, resulting in a drop in leverage. Shareholder payouts fall sharply, and liquidity, measured by the ratio of cash to total assets, increases gradually post-violation.

Consistent with the notion that creditors use covenant violations to apply noncontractual control over the governance of firms, we also document a statistically and economically significant increase in CEO turnover following the announcement of a covenant violation. The impact on CEO turnover is particularly strong for those turnover events classified as forced. Holding other performance variables constant, the marginal likelihood of observing a forced CEO turnover is 60% higher during the quarter of a covenant violation. To put the magnitude of increase in perspective, the marginal impact of a covenant violation is about as large as the marginal impact of a two-standard-deviation decrease in operating cash flow or the market-to-book ratio.2

The interpretation of the consequences of creditor control crucially depends upon the subsequent impact of creditor intervention on borrower performance. We find evidence that both operating performance and equity-market valuation improve, on average, following a financial covenant violation. Both sets of results are particularly striking. After declining for five quarters prior to

2 In a concurrent study,Ozelge and Saunders(2009) document a similar connection between covenant violations and CEO firings.Ozelge and Saunders(2009) also show that the spike in CEO firings is strongest when loans are a relatively important source of financing for a borrower.

The Review of Financial Studies / v 25 n 6 2012

1716

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

the violation, operating cash flow experiences a sharp reversal directly after the violation and increases by 5% of lagged assets, on an annualized basis, in the three quarters post-violation. Most of the improvement in operating performance comes through a reduction in operating expenses, rather than increases in revenue, suggesting that changes made in response to the violation improve a violating firm’s operating efficiency. Likewise, firms that violate covenants experience large declines in their stock price in the months leading up to a violation but reverse their decline following the violation. Violating firms earn a statistically significant positive abnormal return of about 5% per year in the months following the violation, measured via traditional event study methods. The performance results suggest that creditor-led interventions are associated with a “turnaround” in company performance that improves firm value and eventually benefits the shareholders of violator firms.

The rest of the article is organized as follows. The next section places our article within the existing literature on corporate governance and creditor control. Section 2 provides more background on debt covenants and the consequences of covenant violations. Section 3 describes the data and presents some summary statistics on the frequency of violations. Sections 4, 5, and 6 present our main results. Section 7 concludes.

1. Related Literature

Our findings are related to the large body of research focusing on the corporate governance of firms by equityholders (seeShleifer and Vishny 1997; Hermalin and Weisbach 2003; Adams, Hermalin, and Weisbach 2010for surveys of this literature). Our contribution is to highlight that creditors, through the use of covenants and the control rights with which they are associated, also play an important role in the corporate governance of public firms well outside of bankruptcy or payment default states. Our results suggest that effective creditor interventions can boost, or even substitute for, equity-centered governance mechanisms. Indeed, our results could provide a partial explanation for why establishing a strong, causal relationship between equity-centered governance and performance is so difficult.3

Our article is also related to the growing body of literature on the effect of covenant violations on firm behavior (e.g.,Beneish and Press 1993, 1995a,b; Chen and Wei 1993; Sweeney 1994; Dichev and Skinner 2002; Chava and Roberts 2008;Roberts and Sufi 2009). We make the following contributions to this literature. First, we provide a much more comprehensive analysis of

3 For the debate over the association between corporate governance quality, board quality, and performance, see Gompers, Ishii, and Metrick(2003),Bebchuk, Cohen, and Farrell(2009),Bhagat and Black(2002), andBhagat and Bolton(2008); for arguments over the efficacy of shareholder control of board nominations, seeBebchuk (2007) and the subsequent replies to his article in theVirginia Law Review; for the relation between CEO pay and performance, seeJensen and Murphy(1990) andMurphy (1999); for studies examining CEO turnovers and performance, seeKang and Shivdasani(1995) andPerry and Shivdasani(2005). None of these studies consider the influence of creditor control on the overall quality of governance within the firm.

Creditor Control Rights, Corporate Governance, and Firm Value

1717

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

the impact of covenant violations on corporate behavior. More specifically, in addition to capital expenditures and net debt issuance, we also examine the effect of violations on asset growth, cash acquisitions, shareholder payouts, and cash holdings. Second, to the best of our knowledge, we are the first to study the extended impact of the violation on firm performance and firm value. Third, whereas the existing literature on covenants has emphasized contractual control following violations, our results on CEO turnover suggest that creditors also play an important role in advising management behind the scenes in ways that extend beyond contractual restrictions put in place at the time of the violation. Finally, to the best of our knowledge, we are the first to make data on financial covenant violations for the universe of publicly traded firms publicly available.

There are a number of studies that emphasize the prevalence of creditor control in restructurings following payment defaults or bankruptcy.4 More recent attention has focused on the role that nontraditional lenders, such as hedge funds, play in these restructurings. For instance,Lim (2011), Jiang, Li, and Wang (2011), andIvashina, Iverson, and Smith(2011) show that hedge fund investors frequently take positions in corporate debt to consolidate bargaining power during reorganization. Whereas such restructurings represent a relatively novel form of “creditor intervention,”Ivashina, Iverson, and Smith (2011) find that these investors typically enter the picture only when companies are severely distressed and face bankruptcy or some other default-related change in control. By contrast, our analysis focuses on bank-oriented creditor interventions among the much larger set of firms that violate a covenant well outside of financial distress.

Studies focusing on creditor control before payment default includeDaniels and Triantis(1995) andBaird and Rasmussen(2006), who provide anecdotal evidence of creditor influence and argue that this influence has been overlooked in the finance and legal literature. Similarly, in documenting the collapse of L.A. Gear, DeAngelo, DeAngelo, and Wruck(2002) make the point that covenants were the disciplining mechanism that constrained behavior at L.A. Gear, rather than interest-payment requirements. Relative to these articles, our contribution lies in the creation and analysis of novel data about covenant violations for a large number of firms. We view our study as providing a systematic, large sample of evidence that corroborates with the ideas previously presented byDaniels and Triantis(1995), Baird and Rasmussen (2006), andDeAngelo, DeAngelo, and Wruck(2002).

Finally, it is instructive to note the resemblance between the findings in our article and the evidence provided by studies of activist hedge fund

4 Early articles includeGilson(1989,1990),Gilson(1990),Wruck (1990),Gilson and Vetsuypens(1994),James (1995,1996),Hotchkiss and Mooradian(1997), andAndrade and Kaplan(1998).

The Review of Financial Studies / v 25 n 6 2012

1718

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

involvement in underperforming firms.5 As with covenant violations, targets of activist hedge funds experience an increase in operating performance, positive abnormal stock returns, and higher levels of CEO turnover.Brav, Jiang, and Kim(2009) summarize the existing literature as “These results are consistent with the view that hedge fund activism adds value through operational, financial, and governance remedies in the target firms” (p. 189). Of course, there are notable differences between activist hedge fund managers and creditors, who intervene following a covenant violation. Most obviously, hedge fund managers are typically equityholders, whose residual payoff differs notably from the senior and fixed payoffs to debtholders. This may explain why activist hedge funds tend to target firms with low leverage and dividends, which in turn subsequently increase following the intervention. However, the broad similarity of the results suggests that corporate creditors provide governance to firm managers in much the same way as do equityholders.6

2. Covenants in Corporate Credit Agreements: Background

Our central hypothesis is that creditors play an important role in corporate governance even outside of states of payment default or bankruptcy. Debt covenants play a crucial role in this process, since violation transfers significant contractual authority to creditors.

2.1 Covenants In practice, covenants are divided into three broad categories: affirmative covenants, negative covenants, and financial covenants.Affirmativecovenants require the borrower to take certain actions, such as meeting generally accepted accounting principles, timely submission of financial information to the lender, meeting all regulatory reporting demands, paying taxes, maintaining equipment, buying insurance, and remaining compliant with the law.Negative covenants prevent the borrower from taking certain actions, such as altering the fundamental nature of the business, changing control of the company (including through acquisition), disposing of assets, making excessive capital expenditures, and paying dividends.Financial covenants are accounting-based risk and performance limits. These covenants often consist of restrictions on a company’s leverage, interest coverage, total fixed charges (including, e.g.,

5 See, for example,Brav et al.(2008),Klein and Zur(2009),Clifford (2008),Greenwood and Schor(2009),Gillan and Starks(2007), andBrav, Jiang, and Kim(2009).

6 Activist hedge fund events tend to be relatively uncommon compared with covenant violations; for instance, first covenant violations are about five times more likely to occur than an activist hedge fund event, based on a comparison of our sample violations against the 786 investments tracked byGreenwood and Schor(2009) during the period 1993 to 2006. In unreported results, we show that activist events are positively correlated with covenant violations. However, the occurrence of an activist event does not explain any of our findings, mainly because activist events are so infrequent. Using the data collected inGreenwood and Schor(2009), we document that the likelihood of an activist event increases from 1.2 % to 1.7 % in the year post-violation. Most importantly, none of our results are affected if we examine the set of firms who violate a covenant and do not experience an activist event.

Creditor Control Rights, Corporate Governance, and Firm Value

1719

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

interest, rent, and capital expenditures), and net worth. Financial covenants almost always include a measure of periodic operating cash flow, such as earnings before interest, taxes, depreciation, and amortization (EBITDA).

Although covenants are common to all types of debt agreements, including bond and note indentures, they are typically more numerous, detailed, and tightly set in private loan agreements (Kahan and Tuckman 1993; Gilson and Warner 1998;Verde 1999; Sansone and Taylor 2007). Roberts and Sufi(2009) show that 96% of all private credit agreements contain at least one financial covenant, and leverage ratio and coverage ratio covenants are the most common. Financial covenants in private loan agreements aremaintenance- based, meaning that the borrower must be in compliance with the covenant on a regular basis, typically every fiscal quarter (Sansone and Taylor 2007). In contrast, financial covenants in bond indentures are usuallyincurrence- based, meaning that the borrower need be in compliance only at the time of a specific event, such as issuing new debt. The inability to avoid maintenance- based covenants makes private loan contracts much more restrictive.

2.2 Violations A violation of a covenant is considered an event of default, giving the creditor the right to demand immediate repayment of, oraccelerate, the entire loan balance.7 In practice, creditors rarely accelerate the loan, opting instead to use the acceleration right to initiate a renegotiation of the credit agreement. These renegotiations can lead to both changes in the terms of the loan and increases in monitoring by lenders.

The following description of a loan covenant violation, reported by Digital Generation Systems Inc. in a 10-Q disclosure filed on November 9, 2005, provides a typical example of how a violation is handled by the borrower and its lenders:

As of September 30, 2005, the Company was not in compliance with the covenant related to its leverage ratio. On November 9, 2005, the Company received a waiver from its lenders as of September 30, 2005. In connection with securing this waiver, certain other changes were made to the credit facility which, among other things, reduced the amount that can be borrowed under the Company’s revolving line of credit from $15.0 million to $4.5 million.

Beyond the $10.5 million reduction in the company’s line of credit, the “other things” required in connection with the waiver included a 100-basis- point increase in the interest rate spread charged on the loan, stronger restrictions on dividend and intercompany payments, a 50% reduction in

7 Creditors can also limit access to unused portions of any line of credit.

The Review of Financial Studies / v 25 n 6 2012

1720

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

allowed capital expenditures, and a requirement that the company comply with its capital expenditures restrictions on a quarterly, rather than annual, basis.8

We surmise that all of these contractually imposed restrictions on Digital Generations Systems serve as a form of governance, since they limit the control of management over company operations. Yet there is no mention of imminent payment default or bankruptcy in any of Digital Generations Systems’ SEC filings. In fact, Digital Generation Systems proceeded to have very strong cash flows in the following fiscal year.

Table 1 presents evidence that the Digital Generation Systems experience is common. Using a sample of loans from Reuters Loan Pricing Corpora- tion’s DealScan database, we show that loans renegotiated shortly following a covenant violation are significantly different from the loans that preceded the violation. Specifically, we compare a variety of loan terms made within six months preceding a covenant violation with the terms of a similar loan made to the same borrower following the violation, where the maturity of the original loan isafter the initiation of the renegotiated loan. Our assumption is that the loan made post-violation is a renegotiated version of the loan made pre-violation. In the table, we examine only covenant violations in which we have evidence that the firm was not in violation at the time the original loan was made.9

Table 1 shows that loans made following a covenant violation are smaller, carry higher interest rate spreads and fees, have a shorter maturity, and involve fewer lenders in the lending syndicate. The changes likely reflect an increase in the credit risk associated with the borrower and also a desire to facilitate additional monitoring by lenders, as evidenced by the shorter maturity and reduction in syndicate size. Consistent with this motivation, renegotiated loans also vary on a variety of nonprice loan terms. Renegotiated loans are significantly more likely to be secured with collateral and limit borrowing to a borrowing base, which is typically some fraction of a specific asset, such as inventory or accounts receivable. Renegotiated loans also are less likely to include a performance pricing provision, meaning that the new loans are less

8 This additional information is available from the actual credit agreements filed as attachments to the 10-Q. 9 We describe the data collection process in Section 3 and the Supplemental Data Appendix, which explains how

we identify covenant violations. Note that the sample of contracts in Table1 is smaller than our primary sample because we rely on information in DealScan to compare pre- and post-violation contracts. Many actual violations are cured with a waiver or loan amendment that is not picked up by DealScan because the data service tracks only amended loans that alter one of the major terms of the loan: pricing, maturity, or size. In an analysis of SEC filings,Roberts and Sufi(2009) show that more than one-third of all violations result in a change in the contractual terms of the loan, with the remainder of borrowers reporting receiving a waiver with no amendment to the contract. Whereas Table1 includes only those violations cured by changes to credit agreements that are significant enough to appear in DealScan, the changes are likely indicative of the types of changes in contracts we do not observe. We conclude that lenders frequently use the opportunity provided by a covenant violation to negotiate new loan terms.

Creditor Control Rights, Corporate Governance, and Firm Value

1721

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 1 Comparing loan contracts before and after a covenant violation

Panel A: Contract changes

N Before After Difference

Major loan terms Loan Size ($ M) 239 318 289 −29 Interest Rate Spread (bps) 239 174 213 39∗∗ Total Fees (bps) 239 22 32 11∗∗ Tenor (yr.) 239 3.7 3.1 −0.6∗∗ Syndicate Size (no.) 239 7.2 6.5 −0.7∗

Incidence of additional nonprice terms Secured 149 0.78 0.89 0.11∗∗ Performance Pricing 149 0.76 0.62 −0.14∗∗ Borrowing Base 149 0.31 0.45 0.14∗∗ Some Sweep Provision 65 0.77 0.95 0.18∗∗

Incidence of various covenants Max. CAPEX 239 0.22 0.32 0.10∗∗ Min. EBITDA 239 0.08 0.17 0.09∗∗ Max. Debt-to-EBITDA 239 0.50 0.36 −0.15∗∗ Min. Interest Coverage 239 0.32 0.23 −0.09∗∗

Level of various covenants Max. CAPEX ($ M) 28 44 30 −14∗ Min. EBITDA ($ M) 9 25 22 −3 Max. Debt-to-EBITDA 55 4.0 4.4 0.4∗ Min. Interest Coverage 37 2.7 2.3 −0.4∗∗

Panel A presents mean loan characteristics for a sample of lines of credit loans that are renegotiated following a new covenant violation. All loan characteristics are from Thomson Reuters LPC’s DealScan database. The sample includes all loans preceding a new covenant violation (“Before”) that can be matched to a loan of the same borrower after a new covenant violation (“After”), where the After loan initiates prior to the maturity of the Before loan. A new covenant violation is a financial covenant violation reported by a firm that has not experienced a violation for the previous four quarters. Interest Rate Spread is the contractual spread over a LIBOR base rate; Total Fees is the cost of all fees amortized over the life of the loan; Tenor is the maturity of the loan at initiation; Syndicate Size is the number of lenders at initiation; Secured is an indicator variable equal to one if the loan is secured with collateral; Performance Pricing indicates that the interest rate spread changes with some observable characteristic of the borrower; Borrowing Base indicates that borrowing is tied directly to some fraction of short term assets; Some Sweep Provision indicates that the loan contains at least one of the following directives for paying cash directly to the lender: an asset sales sweep, debt issuance sweep, equity issuance sweep, or insurance proceeds sweep.Incidence of various covenantstracks whether or not a given type of covenant is present in the contract.Level of various covenantsreports the level of a covenant when it is present in a contract. ** and * denote 1% and 5% levels of significance, respectively. M, millions; N, number of observations; yr, year; no., number; max., maximum; min., minimum; bps, basic points.

likely to adjust pricing automatically in response to changes in the borrower’s performance.

Table 1 also shows that covenants governing firm investment and fi- nancing decisions change significantly post-violation. Explicit restrictions on capital expenditures are more likely to be put in place for the first time post-violation. For loans that contain a restriction on capital expenditures prior to the violation, a covenant violation leads to a renegotiated contract with a tighter expenditure limit. Moreover, renegotiated loans also are more likely to dictate how borrowers can use cash revenues by requiring a “sweep” provision, which requires that cash flows from certain activities, such as asset sales or debt issuances, must be used to pay down outstanding balances on the loan.

The Review of Financial Studies / v 25 n 6 2012

1722

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Besides forcing changes through more restrictive contracts, there is substan- tial anecdotal evidence that creditors work behind the scenes to affect changes in the way that the company is managed. Although lender liability laws protect equityholders from creditors that directly interfere with the management of the firm, creditors can offer advice to management and the board, quid pro quo, and suggest actions the company can take to maximize the chance of receiving a covenant waiver.

Baird and Rasmussen(2006) cite the example of Krispy Kreme Doughnut Corporation, in which, following a covenant violation, concessions included replacing the CEO with a turnaround specialist.Baird and Rasmussen(2006) suggest that this type of activity may be widespread, writing that “lenders may need to do no more than make it understood that they will look more kindly on future waivers of loan covenants if a [chief restructuring officer] with whom they have worked before is in place and cleaning shop.”

The nature of behind-the-scenes negotiations makes it difficult to document the informal role of creditors on corporate governance. We use the occurrence of a covenant violation as a point in which we know that negotiations are taking place between the lenders and the borrower. Immediate changes in management that follow these negotiations can provide the large sample evidence of creditor influence on corporate governance suggested by theBaird and Rasmussen(2006) anecdote.

3. Data and Summary Statistics

We construct two data sets for the analysis that follows. First, we construct a firm-quarter level data set of financial information taken from Compustat. The broadest sample of Compustat observations used in this article consists of 7,661 nonfinancial U.S. firms and 181,704 firm-quarter observations from the second quarter of 1997 to the fourth quarter of 2008. For this data set, our primary variable of interest is an indicator for whether or not the firm reports a violation of a financial covenant during the corresponding quarter. We describe below how we construct this variable.

Second, we construct a sample of stock market returns for firms that report a covenant violation after having not violated a covenant in the previous four quarters. We use monthly stock returns from the Center for Research in Security Prices (CRSP) and include observations from 3,699 firms that reported a new covenant violation between September 1997 and June 2009.10 In the remainder of this section, we describe the sample construction and provide summary statistics. Full details of our data collection efforts are contained in the article’s Supplemental Data Appendix.

10 SEC filings typically are filed between two and four months after the end of the reporting period. In our stock return results, we examine performance following the date the SEC filing becomes public.

Creditor Control Rights, Corporate Governance, and Firm Value

1723

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

3.1 Data To construct our sample, we start with the universe of all nonfinancial U.S. firms in Compustat during the years 1996 to 2008.11 The sample begins in 1996 because we require electronic SEC filings to employ our text-search algorithm that finds covenant violations; the second calendar quarter of 1996 is the first quarter in which electronic filing became mandatory for all SEC-registered firms. We initially limit the sample to all firms with average book assets greater than $10 million in 2000 dollars and to firm-quarter observations with five available data items: total assets, total sales, common shares outstanding, closing share price, and the calendar quarter of the filing. Imposing these data restrictions leaves a sample of 8,284 firms and 205,863 firm-quarter observations from the second quarter of 1996 to the fourth quarter of 2008.

Next, we merge this data set with a firm-quarter data set of financial covenant violation indicators constructed from SEC filings. This data set is constructed in two steps. First, for every firm-quarter observation in the Compustat universe, we match the observation to its respective 10-Q or 10-K SEC filing that generates the Compustat data. Using these matches, we employ a text-search algorithm to search the actual filings for reports of violations. Our algorithm first locates the word “covenant” in the filing. Conditional on finding “covenant,” the algorithm then searches for the following five phrases within the seven lines surrounding the initial hit: “waiv,” “viol,” “in default,” “modif,” and “not in compliance.” After correcting for false positives, this algorithm captures 90% of actual violations in a random sample of 1,000 filings for which we manually read the entire filing.12 In less than 2% of firm-quarter observations, we are unable to successfully match a Compustat observation to the corresponding SEC filing. We drop these observations, since we cannot determine if the firm is in violation of a covenant.

Because we are interested in estimating the frequency of violations in the population of U.S. public corporations, it is important that our algorithm iden- tifies nearly all reported violations and has few false positives. As compared with prior studies of covenant violations (e.g.,Roberts and Sufi 2009; Nini, Smith, and Sufi 2009), we find roughly 40% more incidents of violations. The primary reason for the increase is that the text-search algorithm used in this analysis represents a significant improvement relative to the previous text- search algorithm used in these other articles. We refer interested readers to the Supplemental Data Appendix for more details on our search algorithm.

Our analysis focuses on initial or “new” financial covenant violations, which we define to be financial covenant violations for firms that have not violated

11 We include fiscal quarters through the fourth quarter of 2008, which may be filed in 2009 and become publicly available as late as June 2009.

12 For a comparison of this approach to the approach taken byRoberts and Sufi(2009) andNini, Smith, and Sufi (2009), see the Supplemental Data Appendix.

The Review of Financial Studies / v 25 n 6 2012

1724

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

a covenant in the previous four quarters.13 As benchmarks for financial and cash flow performance, financial covenants are by far the most common covenants to cause a violation. We exclude the rare occurrences of affirmative and other nonfinancial covenant violations, because they represent a deliberate breach by the borrower. We focus on initial violations, because they represent the first opportunity for creditor intervention and thus provide the cleanest identification of the effect of violations on corporate behavior. Firms often remain in violation of a covenant for several quarters, which reflects both the length of time necessary to cure a violation and increased monitoring of lenders subsequent to an initial violation. In our sample, 42% of firms remain in violation of a covenant in the quarter post-violation, and 32% remain in violation after one year. Even two years after the initial violation, roughly one- fifth of firms remain in violation of a covenant. Whereas there are potentially interesting questions regarding patterns of subsequent violations, we focus on the effect of the initial violation and leave these questions for future research.

Given the need to measure initial violations, an observation is only included in our sample if we have four previous quarters in which to measure whether a given violation is new. This restriction leaves us with the final sample of 7,661 firms and 181,704 firm-quarter observations from the second quarter of 1997 to the fourth quarter of 2008. Because of missing data on a variety of Compustat items, we often employ a smaller sample in our analysis.

We build our stock returns sample for the firms reporting a new covenant violation during the sample period. We use the date that the relevant SEC filing became public, as reported by Edgar, as the official report date of the violation. We merge these data with monthly stock returns from CRSP and capture stock returns between September 1997 and March 2010. We find sufficient stock return data for 3,699 initial violating firms.14

3.2 Summary statistics Figure2 reports the fraction of firms that violated a covenant in any given year from 1997 to 2008. The solid line shows that between 10% and 20% of firms were in violation of a covenant in a given year. The incidence of violations is cyclical, peaking during the 2001–2002 recession. There was also a sharp decline in the incidence of violations in the latter part of the sample—i.e., before the onset of the financial crisis and economic downturn of 2007 and 2008, at which time violations ticked up. The dotted line in Figure2 plots the fraction of firms that reported new financial covenant violations in each year. New violations follow the same cyclical pattern as total violations and reached a high of nearly 10% during 2001.

13 We use four quarters to ensure that we have a complete fiscal year that is free from any violations. 14 We lose a handful of firms due to insufficient return dataprior to the violation.

Creditor Control Rights, Corporate Governance, and Firm Value

1725

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 2 Covenant violations from 1997 to 2008 This figure presents the annual fraction of firms reporting a covenant violation in an SEC 10-K or 10-Q filing during the fiscal years 1997 to 2008 (covering reports through June 2009 in calendar time). A new covenant violation is a financial covenant violation by a firm that has not violated a covenant in the previous four quarters. The sample includes 7,661 firms.

Table 2 provides summary statistics on the incidence of violations. More than 40% of firms in our sample violated a financial covenant at some point during our 12-year sample period. Nearly 7% of firms were in violation in the average quarter, and 2% experienced a new violation in the average quarter. Table2 also shows that financial covenant violations are common across industries, although they have been more common in wholesale trade. Violations are common among both Standard & Poor’s (S&P’s) rated and nonrated firms. Although violations are negatively correlated with firm size, more than one-quarter of firms with over $5 billion in assets violated a financial covenant at some point in the sample. Although covenant violations are more common among small firms, we conclude that they are also quite common among the largest public firms in the economy.

4. Financial Covenant Violations, Payment Default, and Firm Exit

It is well established that creditors play an important role in bankruptcy and in the aftermath of a payment default, but our maintained hypothesis is that creditors play an important role, even when bankruptcy or payment default is not imminent. In this section, we examine this hypothesis by exploring the

The Review of Financial Studies / v 25 n 6 2012

1726

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 2 Frequency of financial covenant violations

Violator Percentage

Fraction of firms ever reporting covenant violation 40.5 Fraction of firm-quarter observations with covenant violation 6.9 Fraction of firm-quarter observations with new covenant violation 2.1 By industry

Agriculture, minerals, construction 41.1 Manufacturing 40.2 Transportation, communication, and utilities 38.3 Trade—wholesale 53.3 Trade—retail 42.1 Services 39.1

By size (book assets) Less than $100 M 43.7 $100 M to $250 M 43.1 $250 M to $500 M 41.7 $500 M to $1,000 M 39.3 $1,000 M to $2,500 M 33.2 $2,500 M to $5,000 M 26.7 Greater than $5,000 M 25.3

Borrower does not have credit rating 40.6 Borrower has credit rating 40.2

This table presents the percentage of firms that report a financial covenant violation in a 10-K or 10-Q SEC filing at some point between 1997 and 2008. The sample includes firm-year quarters for which we have available information on violations in SEC filings for the previous four quarters. A new covenant violation is a financial covenant violation by a firm that has not experienced a violation for the previous four quarters. The sample includes 7,661 firms and 181,704 firm-quarter observations. M, millions.

financial condition of new violators and compare them with nonviolators, in accordance with standard measures of solvency.

4.1 Financial condition at the time of violation Figure 3 produces a series of six panels that summarize the performance of new violators during the eight quarters leading up to and including the violation. The message across all of the panels is similar and not surprising: firm performance declines in the quarters leading up to the violation, which is typically most severe in the year prior to the violation. Operating cash flows fall, market valuations deteriorate, liquidity declines, and leverage increases. Since we are plotting six ratios that are commonly used in financial covenants, it is likely that the violation was caused by deterioration in one or more of these ratios. However, it is important to note that the deterioration does not happen exclusively in the quarter of the violation but rather happens over several quarters leading up to the violation. This fact gives us more confidence in attributing the turnaround in firm performance, which we show later, to the covenant violation, as opposed to other factors that might cause mean- reversion in performance metrics.

Despite the deterioration shown in Figure3, at the time of the violation, the typical violator is not on the verge of payment default or bankruptcy.

Creditor Control Rights, Corporate Governance, and Firm Value

1727

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 3 Firm performance in quarters preceding New Financial Covenant Violation This figure presents sample medians for various firm performance measures leading up to a new financial covenant violation. A new violation is a violation by a firm that has not violated in the previous four quarters. In each respective figure, the sample is limited to firms that have the variable available for the seven quarters leading up to the violation.

Panel A of Table3 provides percentiles of the distribution of various liquidity and solvency measures for firms in violation of a financial covenant. Panel B provides the percentiles of the rank of each violating firm within the broader sample of violating and nonviolating firms from the same industry during the same quarter.15 For example, within the sample of violators, the firm at the twenty-fifth percentile of the asset distribution has $43 million in assets. That firm falls at the twenty-second percentile of the broader distribution of

15 Throughout the analysis, we use the 38 industry classification of SIC codes provided by Ken French; seeFama and French(1997).

The Review of Financial Studies / v 25 n 6 2012

1728

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 3 Which firms violate covenants?

Panel A: Distribution of variables for new violators

All Firms N 10th 25th 50th 75th 90th 50th

Assets ($ M) 3, 755 20 43 131 480 1,700 206 Change in Ln(Sales) 3, 255 −0.406 −0.142 0.028 0.177 0.401 0.064 Operating cash flow / average

assets 3, 713 −0.289 −0.081 0.035 0.105 0.170 0.104

Sales / average assets 3, 748 0.336 0.660 1.054 1.587 2.299 0.999 Operating costs / average assets3, 713 0.358 0.682 1.090 1.639 2.382 0.934 Net worth / assets 3, 755 0.065 0.238 0.410 0.576 0.720 0.498 Leverage ratio 3, 666 0.029 0.140 0.296 0.461 0.636 0.197 Current ratio 3, 667 0.652 1.033 1.519 2.330 3.505 1.950 S&P issuer credit rating 748 B− B BB− BB BBB BB+ Market-to-book ratio 3, 755 0.767 0.920 1.158 1.616 2.456 1.442 Payout yield 3, 251 0.001 0.008 0.029 0.080 0.186 0.034 Dividend payer 3, 266 0.000 0.000 0.000 1.000 1.000 0.000

Panel B: Within industry-quarter rank of new violators

N 10th 25th 50th 75th 90th

Assets ($ M) 3, 755 0.10 0.22 0.42 0.67 0.84 Change in Ln(Sales) 3, 255 0.08 0.19 0.44 0.72 0.90 Operating cash flow / average

assets 3, 713 0.06 0.15 0.31 0.53 0.75

Sales / average assets 3, 748 0.12 0.29 0.53 0.77 0.92 Operating costs / average assets3, 713 0.15 0.33 0.59 0.82 0.93 Net worth / assets 3, 755 0.08 0.18 0.37 0.62 0.81 Leverage ratio 3, 666 0.24 0.44 0.67 0.85 0.95 Current ratio 3, 667 0.06 0.16 0.35 0.62 0.81 S&P issuer credit rating 748 0.00 0.08 0.24 0.47 0.67 Market-to-book ratio 3, 755 0.08 0.17 0.36 0.61 0.83 Payout yield 3, 251 0.10 0.25 0.49 0.73 0.89 Dividend payer 3, 266 0.09 0.21 0.42 0.68 0.88

Panel A shows the distribution of variables for new covenant violators at the time of violation. The column “All firms” shows the median for the entire sample of firms, including violators and nonviolators. Panel B shows the within industry-quarter rank of new violators within the full sample of violators and nonviolators. A new covenant violation is a financial covenant violation for a firm that has not experienced a financial covenant violation in the previous four quarters. All change variables are one-year changes. N, number of observations; M, millions.

violating and nonviolating firms. Panel B provides a sense of the relative performance of violating firms compared with nonviolating firms in the same industry measured at the same time.

By the measures reported in Table3, violators are not extremely leveraged. The median net worth scaled by assets is 0.41, which puts the median violator at the thirty-seventh percentile of the broader distribution of firms. As a point of comparison, Campbell, Hilscher, and Szilagyi (2008) report that firms that subsequently “fail”—defined as firms that file for bankruptcy, delist for financial reasons, or receive a D rating from S&P—are much more highly leveraged shortly before failing than are our sample of violators. The median firm that subsequently fails has a ratio of liabilities to total assets of 0.82, suggesting a net worth ratio less than 0.20, which is below the bottom

Creditor Control Rights, Corporate Governance, and Firm Value

1729

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

quartile of covenant violators. Similarly, the median leverage ratio of violators is 0.30, which is in the sixty-seventh percentile of the broader distribution. Even firms at the extreme percentiles of the violator distribution do not appear more extreme than do nonviolators. In terms of market valuations, the median violator has a reasonably high market valuation relative to book assets (1.158), which falls at the thirty-sixth percentile of the broader distribution. By comparison, the median market-to-book ratio for Campbell, Hilscher, and Szilagyi’s (2008) firms is 0.751, which is in the bottom 10% of violators. Perhaps the best summary measure of credit risk in Table3 is the firm’s credit rating, which is largely between B- and BBB for our sample of violating firms. This is only slightly more risky than the distribution for nonviolators, who have a larger fraction of firms above BBB. As a point of reference, BB-rated firms have tended to experience a one-year payment default rate of less than 1%.

Violators also do not appear to suffer from sharp liquidity shortages. The median current ratio for violators in Table3 is 1.519, which puts the median violator at the 35th percentile of the broader set of firm-quarter observations. The median violator has operating cash flow scaled by lagged assets of 0.035 on an annualized basis, which puts the median violator at the 31st percentile of the broader distribution. Certainly a significant fraction of violators have negative operating cash flow, but more than half have positive cash flow, and the share with negative cash flow is not much larger than the share of nonviolators with negative income. Campbell, Hilscher, and Szilagyi (2008) report that the median firm that subsequently fails has markedly negative net income. Overall, financial covenant violations appear to serve more as an indicator of achangein performance, rather than as an indicator of a lowlevel of performance.

4.2 Exit rates post-violation An alternative perspective on the proximity to bankruptcy or payment default comes from examining the frequency with which firms exit our sample. Table4 provides some evidence on the frequency and cause of firm exits from our sample.

Panel A of Table4 shows the frequency with which firms exit from the Compustat sample within one year for both violators and nonviolators. We count a firm as exiting if the firm ceases to have available data for total assets, total sales, common shares outstanding, and the closing share price. We use a combination of Compustat and CRSP data to determine the reason for exit.16 To account for delayed filings, any firm that survives through the second quarter of 2008 is counted as a survivor.

16 We use the delisting code from CRSP and the reason the firm moved to the historical file from Compustat. We confirm the classification of distress and non-distress reasons for exit by examining firm operating performance and market valuation preceding the exit. Being acquired or going private is not correlated with ex ante declines in performance or valuation, whereas the other exit reasons are.

The Review of Financial Studies / v 25 n 6 2012

1730

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 4 The effect of financial covenant violations on firm exit

Panel A: Frequency of exit from sample within four quarters

New Violators Nonviolators

Any Exit 0.123 0.084 Non-distress-related exit 0.059 0.058

Acquired 0.057 0.057 Become private 0.003 0.002

Distress-related exit 0.064 0.026 Bankrupt and/or liquidated 0.011 0.004 Dropped from exchange 0.010 0.007 Stop filing with SEC for other reason 0.043 0.015 Missing data 0.000 0.001

Panel B: Marginal effects on sample exit from proportional hazardmodel

(1) (2) (3) (4) Any exit Non distressed exit Distressed exit Distressed exit

New financial covenant violation 0.273∗∗ 0.102 0.412 0.325 (0.089) (0.121) (0.215) (0.198)

New financial covenant violationt−1 0.097 −0.254 0.263 0.363 (0.130) (0.183) (0.281) (0.254)

New financial covenant violationt−2 0.328∗∗ 0.075 0.208 0.173 (0.094) (0.205) (0.442) (0.419)

New financial covenant violationt−3 −0.040 −0.059 0.012 −0.278 (0.149) (0.207) (0.376) (0.384)

New financial covenant violationt−4 0.080 0.100 0.214 0.117 (0.073) (0.118) (0.181) (0.202)

Ln(assets) −0.127∗∗ −0.044∗ −0.375∗∗ −0.299∗∗ (0.021) (0.018) (0.058) (0.065)

Operating cash flow / average assets−0.493∗∗ 0.130 −1.266∗∗ −3.334∗∗ (0.127) (0.110) (0.185) (0.486)

Leverage ratio −0.431∗∗ −0.326 −0.356 −3.842∗∗ (0.128) (0.181) (0.228) (1.246)

Interest expense / average assets 2.418∗∗ −2.988 5.436∗∗ 44.645∗∗ (0.669) (1.584) (1.943) (17.145)

Net worth / assets −0.860∗∗ −0.406∗∗ −1.847∗∗ −2.191∗∗ (0.090) (0.110) (0.165) (0.294)

Current ratio 0.000 0.000 0.018 −0.279 (0.012) (0.014) (0.033) (0.164)

Market-to-book ratio −0.179∗∗ −0.124∗∗ −0.471∗∗ −1.707∗∗ (0.031) (0.029) (0.048) (0.216)

Higher-order covenant controls No No No Yes Lagged first-difference controls No No No Yes Lagged covenant controls No No No Yes No. of observations 136,699 136,699 136,699 130,947

Panel A presents the sample exit frequencies of new covenant violators, measured over the four quarters following the violation, compared to “nonviolators,” defined as firms that do not violate any covenant for the current or subsequent three quarters. Reasons for an exit are obtained fromCompustatandCRSP. To take into account delayed filings at the end of the sample, any firm that survives until the second quarter of 2008 is assumed to be surviving even if there are missing observations for 2008Q3 or 2008Q4. Panel B shows estimates of the marginal effect of a covenant violation on the firm-exit hazard rate using a Cox proportional hazard specification. Covenant controls include the following six variables: operating cash flow scaled by average assets; the leverage ratio; the ratio of interest expense to average assets; the ratio of net worth to total assets; the current ratio; and the market-to-book ratio. Specification (4) also includes the second and third power of the levels of the covenant control variables, two lags of the first differences of the covenant control variables, and three- and four-quarter lags of the levels of the covenant control variables. All specifications include industry, calendar-quarter, and fiscal-quarter fixed effects. Standard errors are robust to arbitrary heteroskedasticity and clustered by calendar quarter. ** and * denote 1% and 5% levels of significance, respectively. No., number.

Creditor Control Rights, Corporate Governance, and Firm Value

1731

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

The unconditional probability of exit is higher for violators, and the difference with nonviolators is notable, but the frequency of exit is quite low for violators. On average, a new violator is four percentage points more likely to exit the sample within a year. Although the difference in exit frequency is driven by distress-related exits, only 6.4% of violators exit the sample for a distress-related reason. Liquidation, bankruptcy, and delisting are far from the primary outcomes for firms that violate covenants.

Since covenant violations follow on the heels of deteriorating firm per- formance, a fair comparison of exit probabilities between violators and nonviolators should control for performance declines prior to exit. Panel B of Table4 presents estimated marginal effects from a Cox proportional hazard model, relating the hazard rate of firm exit to a sequence of covenant violation indicators and control variables. We use a quarterly data set and model the baseline hazard rate as a function of the firm’s age, measured in quarters from first appearance in Compustat. We control for right-censoring of observations beginning with the third quarter of 2008. All specifications include calendar quarter, industry, and fiscal quarter indicator variables. Specification (1) ex- amines any exits, specification (2) examines only non-distressed-related exits, and specifications (3) and (4) examine distress-related exits. As compared with specification (3), specification (4) includes additional control variables.

When controlling for firm performance, the estimated impact of violations on firm exit falls significantly. The unconditional hazard of firm exit for any reason is 3.2% per quarter. Based on the estimated coefficients in specification (1), this rises to 4.1% in the quarter of a violation. Based on specification (2), there is no impact of violations on nondistressed exits, suggesting that the impact on total exits arises through the impact on distressed exits.

Specifications (3) and (4) confirm this. The unconditional hazard of a distressed exit is 1.6% per quarter. The point estimates on the new covenant violation indicators in specification (4) suggest that this rises by 0.5 per- centage points and 0.6 percentage points in the two quarters immediately following the violation, although the estimated coefficients are not signif- icantly different from zero at conventional levels. In the subsequent three quarters, there is no evidence that violations lead to an increase in the hazard rate. Without controlling for firm performance, the estimated hazard coefficients suggest about a tripling of the hazard rate in the four quarters post- violation.17

The combined evidence from Tables3 and 4 demonstrates that covenant violations are quite common, even for firms that are far from insolvent and not on the verge of payment default or bankruptcy. Contractual covenant levels are clearly set to be tripped well before severe financial distress, and violations are not typically resolved by means of bankruptcy, liquidation, or merger.

17 This is based on untabulated regressions but largely reflects the summary statistics reported in Panel A of Table4.

The Review of Financial Studies / v 25 n 6 2012

1732

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

5. The Corporate Response to Financial Covenant Violations

In this section, we explore the consequences of creditor intervention by examining changes in firm investment policy, financial policy, and man- agement around a covenant violation. Given the results of the previous section—showing that violating firms are not on the verge of insolvency and are in relatively good health—the evidence provided here reflects the extensive influence of creditors in “normal” states outside of payment default.

5.1 Methodology In the following subsections, we explore the effect of a new financial covenant violation on several firm outcomes, including asset growth, capital expendi- tures, firm payout policy, and CEO turnover. For each outcome, we first plot the mean and the median of the outcome for violators from four quarters pre-violation through four quarters post-violation. Given that we focus on new violations, which are defined to be violations in which the firm has not violated a financial covenant in the previous four quarters, we know that the pre-period is one in which the firm is not in violation of any covenant. For each outcome, we isolate the sample to firms that have available data for four quarters both before and after the violation. These plots are intended to explore unconditional changes in outcomes around the violation and highlight the exact timing of the changes.

We also estimate regressions designed to test whether the changes in outcomes observed post-violation are statistically significant and robust to the inclusion of control variables. We focus on four-quarter changes post-violation in order to capture the cumulative effects over a relatively long period. For most outcome variables, our broadest specification is of the form

yi ,t+4 − yi ,t = β ∗ V i olati oni ,t + θ1 ∗ C onvenant C ontr olsi ,t +θ2 ∗

( H i gher Or der C onvenant C ontr oli ,t

)

+ θ3 ∗ ( C onvenant C ontr olsi ,t−4

) + I ndustr yi

+ Quar t eri + F i scal Quar t eri ,t + εi ,t , (1)

where Violation is an indicator variable that equals 1 for a new financial covenant violation;Industry represents 38 SIC-based industry indicator vari- ables;Quarterrepresents calendar quarter indicator variables; andFiscalQuar- ter represents fiscal quarter indicator variables. The latter indicator variables are included because firm outcomes may have seasonal patterns related to fiscal quarters and financial covenant violations are more common in 10-K filings than in 10-Q filings. In some specifications, we also include the four-quarter lagged value ofViolation, which picks up the effect of a violation one year prior.

Creditor Control Rights, Corporate Governance, and Firm Value

1733

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Given that we estimate (1) using four-quarter differences in outcomes on a firm-quarter data set, we have overlapping observations that induce a mechanical serial correlation in our dependent variable. To control for this in estimating standard errors, we cluster our standard errors by firm and quarter, as in Petersen(2009).18 All of our inferences are robust to choosing a single quarter to estimate (1), which avoids mechanical serial correlation but ignores three-quarters of the sample.

The set of variables labeledCovenantControlsare included to account for variables that may have an independent effect on the outcome of interest. These variables are the ratio of operating cash flow to lagged assets, the leverage ratio (debt-to-assets), the ratio of interest expense to lagged assets, the ratio of net worth to assets, the current ratio (current assets/current liabilities), and the market-to-book ratio. The first five of these variables capture the most common ratios included in financial covenants (seeRoberts and Sufi 2009). We also include the market-to-book ratio because it is a powerful predictor of many firm outcomes. We include these variables linearly, squared, and to the third power, as indicated byHigherOrderCovenantControls.We also include the four-quarter lag of these variables.

The primary challenge we hope to address with this specification is to identify separately the effect of violations from expected changes in outcomes related to differences in the underlying fundamentals of violators and nonvi- olators. Our approach mimics the “quasi-discontinuity” approach inRoberts and Sufi(2009), in which identification is based on comparing firms just above and just below the contractually written covenant threshold. The idea is to control flexibly for continuous functions of the underlying variables, on which covenants are written, and exploit the discontinuity created at the point of vi- olation. By using a first-difference specification, we control for time-invariant, firm-level effects that may be different between violators and nonviolators. By flexibly controlling for the current and lagged level of a variety of variables known to affect outcomes, we hope to control for the expected time-series path of outcomes following deterioration in firm performance. The upshot is that we identify the effect of a violation based on differences in outcomes for violators relative to differences in outcomes for nonviolators with a similar pre- violation pattern in performance.19 Additionally, in all of our reported results, we show various specifications that include an increasing number of control variables. In most cases, the estimated effect shrinks as we add more controls but converges to a level different than zero.

18 We thank Mitchell Petersen for making his code available. 19 Since we do not observe the contractual level of the covenant, we cannot use a standard regression-discontinuity

(RD) design. However, it is worthwhile noting that our strategy works identically to a standard RD if all firms have covenants written at the same level, since the level of the covenant control variables will perfectly determine a violation. By including lags of the covenant control variables, we proxy for each firm’s condition at about the time the loan contract was originated, which is a strong proxy for the unobserved level of the covenant.

The Review of Financial Studies / v 25 n 6 2012

1734

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 5 Summary statistics

N Mean SD 10th Median 90th

Operating cash flow / average assets 179,3500.057 0.237 −0.200 0.104 0.260 Leverage ratio 176,291 0.242 0.242 0.000 0.197 0.549 Interest expense / average assets 172,1900.021 0.028 0.000 0.013 0.050 Net worth / assets 181,698 0.475 0.306 0.145 0.498 0.830 Current ratio 177,749 2.787 2.783 0.818 1.950 5.482 Market-to-book ratio 181,698 2.027 1.775 0.850 1.442 3.764 Change in Ln(assets) 165,968 0.057 0.380 −0.258 0.041 0.381 Change in Ln(PPE) 164,990 0.038 0.484 −0.299 0.023 0.418 CAPX / average assets 177,014 0.059 0.078 0.005 0.034 0.138 Acquisitions / average assets 173,681 0.025 0.106 0.000 0.000 0.032 Net debt issuance / average assets 174,6810.031 0.289 −0.137 0.000 0.204 Change in Ln(total debt) 129,399 0.038 1.017 −0.637 −0.005 0.785 Cash / assets 181,652 0.182 0.220 0.006 0.082 0.537 Ln(1 + shareholder payout) 158,339 1.279 2.036 0.000 0.000 4.615 Change in Ln(sales) 164,789 0.062 0.536 −0.310 0.064 0.438 Change in Ln(operating costs) 162,195 0.053 0.396 −0.266 0.061 0.378 Forced CEO turnover 64,630 0.009 0.096 0.000 0.000 0.000 Unforced CEO turnover 64,630 0.020 0.141 0.000 0.000 0.000 Total CEO turnover 64,630 0.030 0.170 0.000 0.000 0.000

This table presents summary statistics for the sample of 7,661 firms and 181,704 firm-quarter observations from fiscal quarters between 1997 and 2008. The sample includes firm-year quarters for which we have available information on violations in SEC filings for the previous four quarters. All flow variables (operating cash flow, interest expense, capital expenditures, cash acquisitions, and net debt issuance) are annualized with the exception of shareholder payouts. Shareholder payouts include cash dividends and share repurchases. N, number of observations; SD, standard deviation.

Table 5 provides summary statistics for the outcome and control variables used in our analysis. All of these variables are defined in the Supplemental Data Appendix. The first five variables in Table5 are the covenant control variables; the distributions of the variables are in line with data used in previous studies. Our relevant outcome variables include measures of fixed investment (total assets, property, plant, and equipment (PPE), capital expenditures scaled by assets, and cash acquisitions scaled by lagged assets) and measures of financing activity (net debt issuance scaled by assets, total debt, cash scaled by total assets, and total shareholder payouts). The final outcome variable of interest is CEO turnover. We start with Jenter and Kanaan’s (2010) CEO turnover data, which are available through 2001, and extend the sample through 2007, using an identical procedure for classifying CEO turnover as forced or voluntary. The CEO turnover data are limited to firms in the S&P 1500, which is why the CEO sample is much smaller than our original sample.20

5.2 Investment conservatism We begin by exploring the investment and financial consequences of covenant violations in the next two subsections. Figure4 plots mean and median values

20 We are grateful to Dirk Jenter for sharing these data. Please seeJenter and Kanaan(2010) for more details on forced and unforced CEO turnover.

Creditor Control Rights, Corporate Governance, and Firm Value

1735

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

F ig

u re

4 F

in a

n ci

a lc

o ve

n a

n t vi

o la

tio n

s a

n d

in ve

st m

e n

t co

n se

rv a

tis m

T hi

s fig

ur e

pr es

en ts

m ea

ns an

d m

ed ia

ns fo

rm ea

su re

s of

in ve

st m

en ta

ro un

d a

ne w

fin an

ci al

co ve

na nt

vi ol

at io

n. A

ne w

vi ol

at io

n is

a vi

ol at

io n

by a

fir m

th at

ha s

no tv

io la

te d

in th

e pr

ev io

us fo

ur qu

ar te

rs .T

he vi

ol at

io n

is fir

st re

po rt

ed at

qu ar

te r0

,w hi

ch im

pl ie

s th

at it

to ok

pl ac

e at

so m

e po

in tb

et w

ee n

qu ar

te r

− 1

an d

0.

The Review of Financial Studies / v 25 n 6 2012

1736

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

of various measures of firm investment policy in the window extending from four quarters before the violation to four quarters after the violation. The graphs reveal that the financial covenant violations are followed by decreases in capital expenditures and cash acquisitions and sharp reductions in the growth rate of total assets and PPE. Violators’ growth is fairly aggressive pre- violation, with total assets increasing by an average of 10% in the year before the violation. Growth levels off in the quarter of the violation and reverses in the quarters immediately after the violation. Growth in PPE exhibits a similar pattern. The nearly 10% decline in PPE over the year following the violation suggests that violators engage in asset sales and divestitures post-violation.

Consistent with the evidence inChava and Roberts(2008) andNini, Smith, and Sufi(2009), capital expenditures also drop in the quarter after a violation. Capital expenditures decline prior to the violation, but a downward kink in capital expenditures occurs in the period after the violation. Cash acquisitions are fairly stable pre-violation but fall markedly in the quarter immediately following the violation. As discussed inNini, Smith, and Sufi(2009), corporate credit agreements often contain an explicit restriction on capital expenditures and acquisitions, which provides a contractual mechanism for creditors to limit investment following a covenant violation.

Table 6 presents estimates of Equation (1) for these four measures of investment. In all four cases, the estimated effect of a covenant violation remains statistically significant and economically important. The inclusion of control variables tends to reduce the estimated impact, but even in our strictest specifications, covenant violations are estimated to have an important effect on real investment outcomes. Based on the fourth specification in each panel, changes in investment policy begin to wane one year after the violation.

5.3 Financial conservatism Figure5 investigates changes in financial policy after covenant violations. We explore changes in net debt issuance, total debt outstanding, the ratio of cash and liquid securities to total assets, and total shareholder payouts (including dividends and share repurchases). The evidence suggests a clear increase in financial conservatism following covenant violations, consistent with creditors imposing more constraints on the financial policy of firms.

Similar to the results inRoberts and Sufi(2009), net debt issuance falls immediately following the covenant violation and stays low up to a year afterward. This translates into a reduction in the stock of outstanding debt for violators; however, the reduction only partially reverses the large run-up in pre-violation debt. The liquidity of violators, proxied by the ratio of cash- to-assets, declines sharply in the quarters before the violation but levels off in the quarter of the violation and begins to increase thereafter. At least part of the buildup in cash can be attributed to a reduction in payouts to shareholders,

Creditor Control Rights, Corporate Governance, and Firm Value

1737

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

T a

b le

6 F

in a

n ci

a lc

o ve

n a

n t vi

o la

tio n

s a

n d

in ve

st m

e n

t co

n se

rv a

tis m

P a

n e

lA

C h

a n

g e

in L

n (a

ss e

ts )

C h

a n

g e

in L

n (P

P E

)

(1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t −

0. 04

1∗ ∗

− 0.

02 7∗

∗ −

0. 02

5∗ ∗

− 0.

02 2∗

∗ −

0. 04

2∗ ∗

− 0.

03 1∗

∗ −

0. 03

3∗ ∗

− 0.

03 3∗

∗

(0 .0

07 )

(0 .0

06 )

(0 .0

06 )

(0 .0

05 )

(0 .0

11 )

(0 .0

10 )

(0 .0

11 )

(0 .0

12 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t− 4

− 0.

02 9∗

∗ −

0. 02

6∗ ∗

(0 .0

06 )

( 0 .0

08 )

O pe

ra tin

g ca

sh flo

w /a

ve ra

ge as

se ts

0. 30

9∗ ∗

0. 51

2∗ ∗

0. 46

6∗ ∗

0. 45

5∗ ∗

0. 31

2∗ ∗

0. 45

9∗ ∗

0. 36

2∗ ∗

0. 35

7∗ ∗

(0 .0

30 )

(0 .0

36 )

(0 .0

30 )

(0 .0

32 )

(0 .0

22 )

(0 .0

29 )

(0 .0

29 )

(0 .0

33 )

Le ve

ra ge

ra tio

− 0.

09 5∗

∗ −

0. 17

1∗ ∗

− 0.

14 9∗

− 0.

16 5∗

− 0.

02 4

− 0.

26 8∗

∗ −

0. 31

4∗ ∗

− 0.

33 5∗

∗

(0 .0

23 )

(0 .0

65 )

(0 .0

66 )

(0 .0

69 )

(0 .0

29 )

(0 .0

95 )

(0 .1

07 )

(0 .1

15 )

In te

re st

ex pe

ns e

/a ve

ra ge

as se

ts 0.

60 4∗

∗ 4.

06 3∗

∗ 3.

33 9∗

∗ 3.

56 6∗

∗ 0.

78 5∗

∗ 5.

31 2∗

∗ 4.

17 9∗

∗ 3.

91 6∗

∗

(0 .2

08 )

(0 .5

86 )

(0 .6

10 )

(0 .6

67 )

(0 .2

40 )

(0 .9

51 )

(0 .9

37 )

(1 .0

51 )

N et

w or

th /a

ss et

s 0.

03 1

0. 14

1∗ ∗

0. 16

6∗ ∗

0. 15

1∗ ∗

0. 11

0∗ ∗

0. 21

6∗ ∗

0. 23

5∗ ∗

0. 22

4∗ ∗

(0 .0

23 )

(0 .0

31 )

(0 .0

37 )

(0 .0

40 )

(0 .0

23 )

(0 .0

34 )

(0 .0

44 )

(0 .0

47 )

C ur

re nt

ra tio

− 0.

00 5∗

∗ 0.

02 5∗

∗ 0.

02 3∗

∗ 0.

02 5∗

∗ 0.

00 6∗

∗ 0.

02 1∗

∗ 0.

01 8∗

0. 02

1∗ ∗

(0 .0

01 )

(0 .0

05 )

(0 .0

04 )

(0 .0

05 )

(0 .0

01 )

(0 .0

07 )

(0 .0

07 )

(0 .0

07 )

M ar

ke t-

to -b

oo k

ra tio

0. 06

3∗ ∗

0. 18

4∗ ∗

0. 21

0∗ ∗

0. 20

0∗ ∗

0. 04

7∗ ∗

0. 16

6∗ ∗

0. 18

5∗ ∗

0. 16

2∗ ∗

(0 .0

03 )

(0 .0

12 )

(0 .0

13 )

(0 .0

13 )

(0 .0

02 )

(0 .0

12 )

(0 .0

13 )

(0 .0

12 )

H ig

he r-

or de

rc ov

en an

tc on

tr ol

s N

o Ye

s Ye

s Ye

s N

o Ye

s Ye

s Ye

s La

gg ed

co ve

na nt

co nt

ro ls

N o

N o

Ye s

Ye s

N o

N o

Ye s

Ye s

N o.

of ob

se rv

at io

ns 15

0, 47

4 15

0, 47

4 14

3, 53

2 11

9, 95

1 15

0, 01

9 15

0, 01

9 14

3, 11

9 11

9, 60

8 R

2 0.

17 0.

19 0.

20 0.

19 0.

14 0.

14 0.

15 0.

14

(c o

n ti n

u e

d )

The Review of Financial Studies / v 25 n 6 2012

1738

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

T a

b le

6 (C

o n

tin u

e d

)

P a

n e

l B

C h

a n

g e

in C

A P

X / a ve

ra g

e a

ss e

ts C

h a

n g

e in

a cq

u is

iti o

n s

/ a ve

ra g

ea ss

e ts

(9 )

(1 0)

(1 1)

(1 2)

(1 3)

(1 4)

(1 5)

(1 6)

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t −

0. 00

5∗ −

0. 00

5∗ −

0. 00

4∗ −

0. 00

3 −

0. 00

9∗ ∗

− 0.

00 8∗

∗ −

0. 00

5∗ −

0. 00

5∗ (0

.0 02

) (0

.0 02

) (0

.0 02

) (0

.0 02

) (0

.0 02

) (0

.0 02

) (0

.0 02

) (0

.0 02

) N

ew fin

an ci

al co

ve na

nt vi

ol at

io n t−

4 0.

00 1

0. 00

5∗ (0

.0 01

) (0

.0 02

) O

pe ra

tin g

ca sh

flo w

/a ve

ra ge

as se

ts 0.

01 1∗

∗ 0.

01 7∗

∗ 0.

02 6∗

∗ 0.

02 6∗

∗ 0.

02 5∗

∗ 0.

02 7∗

∗ 0.

03 7∗

∗ 0.

03 6∗

∗

(0 .0

02 )

(0 .0

02 )

(0 .0

03 )

(0 .0

03 )

(0 .0

02 )

(0 .0

03 )

(0 .0

03 )

(0 .0

03 )

Le ve

ra ge

ra tio

− 0.

00 7∗

− 0.

02 9∗

∗ −

0. 02

8∗ ∗

− 0.

02 2∗

− 0.

05 4∗

∗ −

0. 11

8∗ ∗

− 0.

13 4∗

∗ −

0. 13

7∗ ∗

(0 .0

03 )

(0 .0

08 )

(0 .0

09 )

(0 .0

09 )

(0 .0

05 )

(0 .0

14 )

(0 .0

16 )

(0 .0

17 )

In te

re st

ex pe

ns e

/a ve

ra ge

as se

ts 0.

12 2∗

∗ 0.

62 1∗

∗ 0.

54 3∗

∗ 0.

47 2∗

∗ 0.

18 3∗

∗ 1.

21 7∗

∗ 0.

85 4∗

∗ 0.

87 3∗

∗

(0 .0

21 )

(0 .0

80 )

(0 .0

85 )

(0 .0

82 )

(0 .0

26 )

(0 .1

47 )

(0 .1

48 )

(0 .1

57 )

N et

w or

th /a

ss et

s 0.

00 1

− 0.

00 2

− 0.

00 2

0. 00

3 −

0. 01

1∗ ∗

− 0.

00 8∗

0. 01

5∗ 0.

01 7∗

∗

(0 .0

01 )

(0 .0

02 )

(0 .0

04 )

(0 .0

04 )

(0 .0

03 )

(0 .0

04 )

(0 .0

06 )

(0 .0

05 )

C ur

re nt

ra tio

0. 00

1∗ ∗

0. 00

1 0.

00 2∗

0. 00

2∗ 0.

00 2∗

∗ 0.

00 7∗

∗ 0.

00 9∗

∗ 0.

01 0∗

∗

(0 .0

00 )

(0 .0

01 )

(0 .0

01 )

(0 .0

01 )

(0 .0

00 )

(0 .0

01 )

(0 .0

01 )

(0 .0

01 )

M ar

ke t-

to -b

oo k

ra tio

− 0.

00 1∗

∗ −

0. 00

2∗ 0.

00 0

0. 00

0 0.

00 3∗

∗ 0.

00 8∗

∗ 0.

00 7∗

∗ 0.

00 8∗

∗

(0 .0

00 )

(0 .0

01 )

(0 .0

01 )

(0 .0

01 )

(0 .0

00 )

(0 .0

01 )

(0 .0

01 )

(0 .0

02 )

H ig

he r-

or de

rc ov

en an

tc on

tr ol

s N

o Ye

s Ye

s Ye

s N

o Ye

s Ye

s Ye

s La

gg ed

co ve

na nt

co nt

ro ls

N o

N o

Ye s

Ye s

N o

N o

Ye s

Ye s

N o.

of ob

se rv

at io

ns 14

5, 44

2 14

5, 44

2 13

8, 69

4 11

6, 25

3 14

1, 19

8 14

1, 19

8 13

4, 66

4 11

2, 81

5 R

2 0.

06 0.

06 0.

06 0.

06 0.

02 0.

02 0.

03 0.

03

T hi

s ta

bl e

pr es

en ts

fir st

-d iff

er en

ce es

tim at

es of

th e

m ar

gi na

le ffe

ct of

a ne

w co

ve na

nt vi

ol at

io n

on th

e in

ve st

m en

td ec

is io

ns of

fir m

s fr

om th

e qu

ar te

ro ft

he vi

ol at

io n

to fo

ur qu

ar te

rs af

te rt

he vi

ol at

io n.

P an

el A

ex am

in es

ch an

ge s

in to

ta la

ss et

s an

d P

P E

;P an

el B

ex am

in es

C A

P X

an d

ac qu

is iti

on s,

bo th

sc al

ed by

av er

ag e

as se

ts .S

ub sc

rip t

t re

fe rs

to a

vi ol

at io

n in

th e

co nt

em po

ra ne

ou s

qu ar

te r,

an d

(t −

4 ) re

fe rs

to a

vi ol

at io

n fo

ur qu

ar te

rs pr

io r.

C ov

en an

tc on

tr ol

s ar

e th

e si

x va

ria bl

es :o

pe ra

tin g

ca sh

flo w

sc al

ed by

av er

ag e

as se

ts ;t

he le

ve ra

ge ra

tio ;t

he ra

tio of

in te

re st

ex pe

ns e

to av

er ag

e as

se ts

; th

e ra

tio of

ne t

w or

th to

to ta

la ss

et s;

th e

cu rr

en t

ra tio

; an

d th

e m

ar ke

t- to

-b oo

k ra

tio .

H ig

he r-

or de

r co

ve na

nt co

nt ro

ls re

fe rs

to th

e se

co nd

an d

th ird

po w

er of

th e

co ve

na nt

co nt

ro lv

ar ia

bl es

;l ag

ge d

co ve

na nt

co nt

ro ls

re fe

rs to

th e

fo ur

-q ua

rt er

la gg

ed le

ve lo

ft he

co ve

na nt

co nt

ro lv

ar ia

bl es

.A ll

sp ec

ifi ca

tio ns

in cl

ud e

in du

st ry

,c al

en da

r- qu

ar te

r, an

d fis

ca l-q

ua rt

er fix

ed ef

fe ct

s an

d th

e fo

llo w

in g

co nt

ro lv

ar ia

bl es

:t he

le ve

la nd

fir st

di ffe

re nc

e of

Ln (a

ss et

s) ;t

he le

ve la

nd fir

st di

ffe re

nc e

of (P

P E

/a ss

et s)

.S ta

nd ar

d er

ro rs

ar e

cl us

te re

d by

fir m

an d

ca le

nd ar

qu ar

te r.

** an

d *

de no

te 1%

an d

5% le

ve ls

of si

gn ifi

ca nc

e, re

sp ec

tiv el

y. N

o. ,n

um be

r.

Creditor Control Rights, Corporate Governance, and Firm Value

1739

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

F ig

u re

5 F

in a

n ci

a lc

o ve

n a

n t vi

o la

tio n

s a

n d

fin a

n ci

a lc

o n

se rv

a tis

m T

hi s

fig ur

e pr

es en

ts m

ea ns

an d

m ed

ia ns

fo rm

ea su

re s

of fin

an ci

al po

lic y

ar ou

nd a

ne w

fin an

ci al

co ve

na nt

vi ol

at io

n. A

ne w

vi ol

at io

n is

a vi

ol at

io n

by a

fir m

th at

ha s

no tv

io la

te d

in th

e pr

ev io

us fo

ur qu

ar te

rs .T

he vi

ol at

io n

is fir

st re

po rt

ed at

qu ar

te r0

,w hi

ch im

pl ie

s th

at it

to ok

pl ac

e at

so m

e po

in tb

et w

ee n

qu ar

te r

− 1

an d

0.

The Review of Financial Studies / v 25 n 6 2012

1740

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

as total shareholder payouts decline sharply in the quarter after the violation and stay low for at least a year post-violation.

Table 7 presents estimates of Equation (1) for these four measures of firm financial policy. The inclusion of control variables only slightly lowers the estimated impact of covenant violations and leaves the qualitative conclusion unchanged. Covenant violations are associated with nearly a 10% reduction in total debt and a 4% reduction in shareholder payouts, although the latter is estimated with substantial error. Net debt issuance also falls, and the cash-to- assets ratio increases by about 0.5 percentage points. This effect is relatively small but reflects a sharp reversal from the trend during the year before the violation, when violators burn through about one-quarter of the cash on their balance sheets. During the second year after the violation, there is little evidence of continued changes in financial policy, except that total debt continues to decrease.

5.4 CEO turnover We examine CEO turnover in Figure6 and Table8. We focus on total CEO turnover and forced turnover, defined byJenter and Kanaan(2010) to be observed CEO turnovers that most likely represent a CEO firing, forced resignation, or forced retirement. The solid line in Figure6 shows the incidence of total CEO turnover around a violation, and the dashed line shows the incidence of forced turnover. Despite violators experiencing at least four quarters of declines in operating performance and market valuation before a violation (as shown in Figure3), the frequency of CEO turnover is about constant during the four quarters before a violation. In contrast, during the quarter of the violation (between -1 and 0) and the quarter immediately following the violation (between 0 and +1), the incidence of forced CEO turnovers increases sharply. The frequency of forced turnover rises from 1.5% per quarter pre-violation to 2.5% in the quarter of the violation and then peaks in the post-violation quarter at above 3%. Over the two quarters following the violation, the likelihood of observing a CEO being fired increases to roughly 5%, a sharp increase from the year before the violation.

An obvious concern with the interpretation that violations lead to CEO turnover is that violating firms experience consecutive declines in performance prior to the violation, which likely increases the probability of a forced CEO turnover, even in the absence of the violation itself. In addition to again highlighting that the turnover is concentrated in the quarters immediately following the violation, we also estimate regressions to control for firm performance, as shown in Table8. We estimate Cox proportional hazard models relating the one-quarter likelihood of CEO turnover to a variety of control variables. We specify the baseline hazard rate as a function of the CEO’s tenure and focus on the pre-violation quarter through the fourth quarter following the violation. The set of control variables is similar to Equation (1),

Creditor Control Rights, Corporate Governance, and Firm Value

1741

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

T a

b le

7 F

in a

n ci

a lc

o ve

n a

n t vi

o la

tio n

s a

n d

fin a

n ci

a lc

o n

se rv

a tis

m

P a

n e

l A

C h

a n

g e

in N

D I / a ve

ra g

e a

ss e

ts C

h a

n g

e in

L n

(t o

ta l d

e b

t)

(1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t −

0. 03

4∗ ∗

− 0.

03 4∗

∗ −

0. 02

0∗ ∗

− 0.

01 9∗

∗ −

0. 10

6∗ ∗

− 0.

08 0∗

∗ −

0. 09

7∗ ∗

− 0.

09 8∗

∗

(0 .0

08 )

(0 .0

07 )

(0 .0

07 )

(0 .0

08 )

(0 .0

17 )

(0 .0

17 )

(0 .0

17 )

(0 .0

17 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t− 4

− 0.

00 1

− 0.

04 2∗

∗

(0 .0

08 )

(0 .0

21 )

O pe

ra tin

g ca

sh flo

w /a

ve ra

ge as

se ts

0. 09

2∗ ∗

0. 07

5∗ ∗

0. 06

4∗ ∗

0. 06

3∗ ∗

0. 06

0 0.

21 6∗

∗ 0.

00 0

− 0.

06 6

(0 .0

08 )

(0 .0

12 )

(0 .0

11 )

(0 .0

13 )

(0 .0

34 )

(0 .0

49 )

(0 .0

57 )

(0 .0

56 )

Le ve

ra ge

ra tio

− 0.

48 5∗

∗ −

1. 08

8∗ ∗

− 1.

50 2∗

∗ −

1. 48

2∗ ∗

− 0.

86 2∗

∗ −

4. 82

2∗ ∗

− 4.

92 4∗

∗ −

4. 85

1∗ ∗

(0 .0

24 )

(0 .0

53 )

(0 .0

54 )

(0 .0

59 )

(0 .0

60 )

(0 .2

43 )

(0 .2

45 )

(0 .2

56 )

In te

re st

ex pe

ns e

/a ve

ra ge

as se

ts 2.

22 1∗

∗ 12

.8 56

∗ ∗

10 .8

78 ∗ ∗

10 .2

49 ∗ ∗

1. 49

0∗ ∗

14 .9

11 ∗ ∗

13 .1

49 ∗ ∗

12 .8

47 ∗ ∗

(0 .1

68 )

(0 .5

40 )

(0 .5

41 )

(0 .5

58 )

(0 .3

86 )

(1 .4

50 )

(1 .3

82 )

(1 .4

75 )

N et

w or

th /a

ss et

s −

0. 03

3∗ ∗

− 0.

08 8∗

∗ −

0. 02

7 −

0. 02

9 −

0. 10

7∗ 0.

09 2

0. 04

6 0.

03 2

(0 .0

07 )

(0 .0

17 )

(0 .0

15 )

(0 .0

16 )

(0 .0

42 )

(0 .0

64 )

(0 .0

86 )

(0 .0

88 )

C ur

re nt

ra tio

0. 00

1 0.

00 5

0. 00

3 0.

00 3

0. 01

4∗ ∗

0. 01

2 0.

00 9

0. 00

1 (0

.0 01

) (0

.0 03

) (0

.0 03

) (0

.0 03

) (0

.0 04

) (0

.0 14

) (0

.0 14

) (0

.0 17

) M

ar ke

t- to

-b oo

k ra

tio 0.

00 8∗

∗ 0.

02 6∗

∗ 0.

02 7∗

∗ 0.

02 7∗

∗ 0.

03 8∗

∗ 0.

15 0∗

∗ 0.

15 6∗

∗ 0.

12 4∗

∗

(0 .0

01 )

(0 .0

05 )

(0 .0

05 )

(0 .0

04 )

(0 .0

06 )

(0 .0

25 )

(0 .0

25 )

(0 .0

25 )

H ig

he r-

or de

rc ov

en an

tc on

tr ol

s N

o Ye

s Ye

s Ye

s N

o Ye

s Ye

s Ye

s La

gg ed

co ve

na nt

co nt

ro ls

N o

N o

Ye s

Ye s

N o

N o

Ye s

Ye s

N o.

of ob

se rv

at io

ns 14

8, 03

6 14

8, 03

6 14

1, 42

9 11

8, 23

8 11

9, 88

2 11

9, 88

2 11

4, 11

0 95

,2 01

R 2

0. 06

0. 08

0. 12

0. 11

0. 04

0. 06

0. 06

0. 06

(c o

n ti n

u e

d )

The Review of Financial Studies / v 25 n 6 2012

1742

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

T a

b le

7 C

o n

tin u

e d

P a

n e

lB

C h

a n

g e

in ca

sh / to

ta la

ss e

ts C

h a

n g

e in

L n

(s h

a re

h o

ld e

r p

a yo

u ts

)

(9 )

(1 0)

(1 1)

(1 2)

(1 3)

(1 4)

(1 5)

(1 6)

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t 0.

00 8∗

∗ 0.

00 6∗

∗ 0.

00 56

∗ 0.

00 5∗

− 0.

08 1∗

∗ −

0. 07

4∗ ∗

− 0.

04 7∗

− 0.

04 0

(0 .0

02 )

(0 .0

02 )

(0 .0

02 )

(0 .0

02 )

(0 .0

23 )

(0 .0

23 )

(0 .0

23 )

(0 .0

23 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t− 4

0. 00

3 0.

03 4

(0 .0

02 )

(0 .0

25 )

O pe

ra tin

g ca

sh flo

w /a

ve ra

ge as

se ts

0. 01

8∗ ∗

0. 02

1∗ ∗

0. 03

0∗ ∗

0. 02

9∗ ∗

0. 09

8∗ ∗

0. 13

2∗ ∗

0. 13

7∗ ∗

0. 15

3∗ ∗

(0 .0

04 )

(0 .0

05 )

(0 .0

06 )

(0 .0

06 )

(0 .0

23 )

(0 .0

34 )

(0 .0

43 )

(0 .0

48 )

Le ve

ra ge

ra tio

− 0.

00 9

0. 10

6∗ ∗

0. 09

5∗ ∗

0. 08

6∗ ∗

− 0.

18 8∗

∗ −

0. 37

7∗ ∗

− 0.

60 1∗

∗ −

0. 66

2∗ ∗

(0 .0

06 )

(0 .0

19 )

(0 .0

20 )

(0 .0

21 )

(0 .0

42 )

(0 .1

23 )

(0 .1

30 )

(0 .1

39 )

In te

re st

ex pe

ns e

/a ve

ra ge

as se

ts 0.

08 9

− 0.

16 2

− 0.

04 0

0. 06

6 2.

04 4∗

∗ 7.

45 6∗

∗ 5.

43 5∗

∗ 5.

61 3∗

∗

(0 .0

47 )

(0 .1

84 )

(0 .1

80 )

(0 .1

96 )

(0 .2

39 )

(1 .1

91 )

(1 .1

28 )

(1 .1

59 )

N et

w or

th /a

ss et

s 0.

00 0

− 0.

02 3∗

∗ −

0. 04

3∗ ∗

− 0.

04 1∗

∗ 0.

14 2∗

∗ 0.

15 3∗

∗ 0.

27 3∗

∗ 0.

29 4∗

∗

(0 .0

04 )

(0 .0

06 )

(0 .0

08 )

(0 .0

08 )

(0 .0

28 )

(0 .0

47 )

(0 .0

63 )

(0 .0

69 )

C ur

re nt

ra tio

− 0.

00 7∗

∗ −

0. 02

0∗ ∗

− 0.

02 0∗

∗ −

0. 01

8∗ ∗

0. 00

0 −

0. 00

3 −

0. 00

4 0.

00 1

(0 .0

00 )

(0 .0

01 )

(0 .0

01 )

(0 .0

01 )

(0 .0

01 )

(0 .0

10 )

(0 .0

10 )

(0 .0

11 )

M ar

ke t-

to -b

oo k

ra tio

0. 00

0 −

0. 01

0∗ ∗

− 0.

01 1∗

∗ −

0. 00

7∗ 0.

03 0∗

∗ 0.

10 4∗

∗ 0.

14 6∗

∗ 0.

15 1∗

∗

(0 .0

01 )

(0 .0

02 )

(0 .0

03 )

(0 .0

03 )

(0 .0

03 )

(0 .0

17 )

(0 .0

18 )

(0 .0

21 )

H ig

he r-

or de

rc ov

en an

tc on

tr ol

s N

o Ye

s Ye

s Ye

s N

o Ye

s Ye

s Ye

s La

gg ed

co ve

na nt

co nt

ro ls

N o

N o

Ye s

Ye s

N o

N o

Ye s

Ye s

N o.

of ob

se rv

at io

ns 15

0, 44

7 15

0, 44

7 14

3, 50

9 11

9, 93

4 12

4, 97

2 12

4, 97

2 11

9, 41

2 10

0, 60

0 R

2 0.

06 0.

07 0.

07 0.

06 0.

03 0.

03 0.

03 0.

03

T hi

s ta

bl e

pr es

en ts

fir st

-d iff

er en

ce es

tim at

es of

th e

m ar

gi na

le ffe

ct of

a ne

w co

ve na

nt vi

ol at

io n

on th

e fin

an ci

ng de

ci si

on s

of fir

m s

fr om

th e

qu ar

te ro

ft he

vi ol

at io

n to

fo ur

qu ar

te rs

af te

rt he

vi ol

at io

n. P

an el

A ex

am in

es ne

td eb

ti ss

ua nc

e an

d gr

ow th

in to

ta ld

eb t;

P an

el B

ex am

in es

ca sh

ho ld

in gs

an d

ch an

ge s

in sh

ar eh

ol de

r pa

yo ut

s, in

cl ud

in g

bo th

di vi

de nd

pa ym

en ts

an d

sh ar

e re

pu rc

ha se

s. S

ub sc

rip t

t re

fe rs

to a

vi ol

at io

n in

th e

co nt

em po

ra ne

ou s

qu ar

te r,

an d

( t −

4 ) re

fe rs

to a

vi ol

at io

n fo

ur qu

ar te

rs pr

io r.

C ov

en an

tc on

tr ol

s ar

e th

e si

x va

ria bl

es :o

pe ra

tin g

ca sh

flo w

sc al

ed by

av er

ag e

as se

ts ;t

he le

ve ra

ge ra

tio ;t

he ra

tio of

in te

re st

ex pe

ns e

to av

er ag

e as

se ts

;t he

ra tio

of ne

tw or

th to

to ta

la ss

et s;

th e

cu rr

en tr

at io

;a nd

th e

m ar

ke t-

to -b

oo k

ra tio

.H ig

he r-

or de

r co

ve na

nt co

nt ro

ls re

fe rs

to th

e se

co nd

an d

th ird

po w

er of

th e

co ve

na nt

co nt

ro lv

ar ia

bl es

;l ag

ge d

co ve

na nt

co nt

ro ls

re fe

rs to

th e

fo ur

-q ua

rt er

la gg

ed le

ve lo

ft he

co ve

na nt

co nt

ro lv

ar ia

bl es

.A ll

sp ec

ifi ca

tio ns

in cl

ud e

in du

st ry

,c al

en da

r- qu

ar te

r, an

d fis

ca l-q

ua rt

er fix

ed ef

fe ct

s an

d th

e fo

llo w

in g

co nt

ro lv

ar ia

bl es

:t he

le ve

la nd

fir st

di ffe

re nc

e of

Ln (a

ss et

s) ;t

he le

ve la

nd fir

st di

ffe re

nc e

of (P

P E

/a ss

et s)

.S ta

nd ar

d er

ro rs

ar e

cl us

te re

d by

fir m

an d

ca le

nd ar

qu ar

te r.

** an

d *

de no

te 1%

an d

5% le

ve ls

of si

gn ifi

ca nc

e, re

sp ec

tiv el

y. N

o. ,n

um be

r.

Creditor Control Rights, Corporate Governance, and Firm Value

1743

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 6 Financial covenant violations and CEO turnover This figure presents the fraction of new violators that experience CEO turnover in the quarters around a new covenant violation. A new violation is a violation by a firm that has not violated in the previous four quarters. The violation is first reported at quarter 0, which implies that it took place at some point between quarter−1 and 0. Total turnover is shown by the blue solid line and plotted against the left axis; forced turnover is shown by the red dashed line and plotted against the right axis.

except that the dependent variable is the hazard of CEO turnover during a quarter.

Column (1) of Table8 shows that the likelihood of any CEO turnover increases in the quarter of the violation and stays high for the subsequent two quarters, although the only statistically significant effect is in the first post-violation quarter. The unconditional hazard rate is 3% per quarter, so the coefficient estimate of 0.372 suggests that the probability of turnover increases by 1.1 percentage points following the violation. The lack of significant coefficients in column (2)’s specification suggests that the increase in turnover is driven by forced turnovers, which is confirmed in columns (3) and (4).

The specifications (3) and (4) focus exclusively on forced turnover, and the estimated coefficients confirm that violations are followed by a sharp increase in turnover that persists for at least a year and is most dramatic in the two quarters following the violation. The estimates in column (4) suggest that the cumulative impact in the two quarters post-violation is more than a doubling of the rate of forced turnover. The unconditional hazard of a forced turnover is about 1% per quarter, so the estimates imply that that rate increases to over 2% in the six months following the violation. Using a different technique for

The Review of Financial Studies / v 25 n 6 2012

1744

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 8 Financial covenant violations and CEO turnover

(1) (2) (3) (4)

Any Unforced Forced Forced turnover turnover turnover turnover

New financial covenant violationt+1 0.204 0.151 0.307 0.357 (0.203) (0.247) (0.321) (0.309)

New financial covenant violationt 0.235 −0.211 0.724∗ 0.616∗ (0.182) (0.323) (0.283) (0.301)

New financial covenant violationt−1 0.372∗ 0.177 0.710∗∗ 0.529∗ (0.161) (0.228) (0.245) (0.264)

New financial covenant violationt−2 0.319 0.171 0.564 0.611 (0.244) (0.295) (0.346) (0.316)

New financial covenant violationt−3 0.048 −0.188 0.378 0.311 (0.216) (0.316) (0.306) (0.350)

New financial covenant violationt−4 0.226 0.019 0.555∗ 0.642∗ (0.188) (0.321) (0.280) (0.255)

Ln(assets) 0.053∗ 0.064∗∗ 0.037 0.060 (0.023) (0.030) (0.040) (0.041)

Operating cash flow / average assets −1.187∗∗ −0.303 −2.283∗∗ −3.153∗∗ (0.205) (0.253) (0.283) (0.826)

Leverage ratio −0.045 0.605 −0.673 −8.103∗∗ (0.351) (0.433) (0.533) (1.967)

Interest expense / average assets −4.590 −13.043∗∗ 3.570 31.226 (3.116) (4.241) (3.759) (22.199)

Net worth / assets −0.199 −0.192 −0.261 −1.343 (0.219) (0.298) (0.337) (1.052)

Current ratio −0.037∗ −0.047∗∗ −0.004 0.009 (0.016) (0.018) (0.031) (0.153)

Market-to-book ratio −0.030 −0.008 −0.144∗∗ −0.637∗ (0.016) (0.018) (0.048) (0.258)

Higher-order covenant controls No No No Yes Lagged first-difference controls No No No Yes Lagged covenant controls No No No Yes No. of observations 47,523 47,523 47,523 44,413

This table presents estimates of marginal effects of a financial covenant violation on the hazard rate of CEO turnover in the quarters around the violation. The baseline hazard is specified as a function of the CEO’s tenure. Subscript (t + 1) refers to a violation in the subsequent quarter,t refers to a violation in the contemporaneous quarter, (t − 1) refers to a violation in the prior quarter, etc. Covenant controls are the six variables: operating cash flow scaled by average assets; the leverage ratio; the ratio of interest expense to average assets; the ratio of net worth to total assets; the current ratio; and the market-to-book ratio. Specification (4) also includes the second and third power of the levels of the covenant control variables, two lags of the first differences of the covenant control variables, and three- and four-quarter lags of the levels of the covenant control variables. All specifications include industry, calendar-quarter, and fiscal-quarter fixed effects. Standard errors are robust to arbitrary heteroskedasticity and clustered by calendar quarter. ** and * denote 1% and 5% levels of significance, respectively.

classifying covenant violations,Ozelge and Saunders(2009) produce similar estimates of the impact of violations on forced CEO turnover.

The impact of a covenant violation on forced CEO turnover is economically meaningful, comparable to dismissal rates following large declines in market- to-book values or operating cash flows. For example, a two-standard-deviation decrease in the market-to-book ratio leads to a 0.5-percentage-point increase in the probability of forced CEO turnover, and a two-standard-deviation decrease in operating cash flow scaled by lagged assets leads to a 1.1-percentage-

Creditor Control Rights, Corporate Governance, and Firm Value

1745

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

point increase in the probability of a forced CEO turnover.21 It takes about a two-standard-deviation change in these standard measures of performance to generate the same change in CEO turnover, as does a covenant violation. In assessing the economic importance of violations for CEO turnover, we again note that the sample used in Table8 is limited to firms in the S&P 1500. Although it is an empirical hypothesis that could be answered with expanded data, we conjecture that the effect would be at least as strong in smaller firms that rely more heavily on bank debt.

The specification shown in column (4) adds a large set of additional control variables, which only slightly reduces the marginal impact of a violation. Thus, even with the inclusion of a rigorous set of controls for variables that affect the probability of forced CEO turnover, we still find a very large effect of covenant violations.

We conclude this section by highlighting the difficulty in identifying the mechanism through which financial covenant violations lead to forced CEO turnover. The difficulty arises because legal concerns give lenders strong incentives to avoid the appearance of having excessive control over borrower management.22 However, we can rely on the legal scholarship ofDaniels and Triantis (1995) to help understand the potential mechanisms. They argue that the lender first and foremost uses the threat of exit to affect board decision- making, which is closely in line with the Krispy Kreme anecdote inBaird and Rasmussen(2006). Additionally, creditors can prompt action by other stakeholders, as noted byDaniels and Triantis(1995): “. . . the monitoring bank detects evidence of slack or misbehavior when it does occur. If the bank exits or takes new security in the assets of the borrower, it provides a signal to the other stakeholders that prompts intervention” (p. 1103). As we emphasize in the conclusion, more evidence is needed on the exact mechanism through which violations lead to CEO turnover.

5.5 Robustness As a check on the robustness of our results, we estimate the impact of covenant violations on all of our outcome variables using propensity-score-matching methods. That is, we use a first-stage probit regression to estimate the likeli- hood that a firm violates a covenant and then match violators to nonviolators based on their fitted violation probability, i.e., “propensity score.” The second stage measures differences in outcomes between covenant violators and their propensity-score-matched nonviolators. As discussed inRoberts and Whited

21 These values are computed using the estimated coefficients from specification (3) that does not use high-order terms and the standard deviations from Table5. The unconditional hazard rate of forced CEO turnover is 1 % per quarter.

22 The bankruptcy doctrine ofequitable subordination“authorizes the court to subordinate a lender’s claim if the lender obtains an advantage at the expense of other creditors as a result of its control over the borrower’s management” (Daniels and Triantis 1995, p. 1,097). As a result, lenders will often work in unobservable ways to influence the management team.

The Review of Financial Studies / v 25 n 6 2012

1746

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

(2011), such matching estimators allow us to control for a variety of observable traits between the two samples in a highly nonlinear and flexible fashion.

Supplemental Data Appendix shows that the propensity-score estimates only strengthen our inferences regarding the impact of covenant violations on firm behavior. For every outcome variable, the propensity-score estimates are larger in absolute magnitude than are the OLS estimates.

6. The Value Implications of Creditor Intervention

The previous sections show that creditors influence the behavior and man- agement of companies that violate financial covenants, even when these companies are relatively healthy and unlikely to default on debt payments. In this section, we study how creditor intervention impacts firm value by examining changes in operating and stock price performance in the period before and after the violation of a covenant.

The theoretical prediction of the effects of credit interference on firm value is ambiguous. Creditor interference can exacerbate conflicts of interest between equity- and debtholders, as inJensen and Meckling(1976) andGorton and Kahn (2000). Excessive conservatism by creditors can lead them to thwart positive net present value projects that create too much risk. Creditors may also “hold up” financing following a covenant violation to extract surplus from the borrower via amendment fees and increased interest charges. On the other hand, a number of theories suggest that creditor interventions may increase firm value that would benefit shareholders by constraining value- reducing managerial behavior and affecting a turnaround following poor performance. These theories view agency conflicts arising not between equity- and debtholders but between external investors (equity- and debtholders) and management.23 By limiting managerial inefficiency, creditor interventions may jointly benefit creditors and equityholders.

6.1 Accounting-based evidence We estimate the impact of the covenant violation on firm value using both accounting-based measures of performance and stock market returns. For accounting-based measures, we examine operating cash flow scaled by assets, sales growth, and cost growth, where operating costs are defined as the difference between sales and operating cash flow.

Figure 7 plots the three accounting-based performance measures in the quarters surrounding a covenant violation. The operating cash flow panel reproduces the swift pre-violation decline in operating cash flow from Figure3 but also includes the mean and median operating performance of covenant

23 Relevant articles includeJensen(1986),DeAngelo, DeAngelo, and Wruck(2002),Aghion and Bolton(1992), Dewatripont and Tirole(1994), andSmith (1993).

Creditor Control Rights, Corporate Governance, and Firm Value

1747

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 7 Financial covenant violations and operating performance This figure presents unconditional means and medians for measures of firm performance around a new financial covenant violation. A new violation is a violation by a firm that has not violated in the previous four quarters. The violation is first reported at quarter 0, which implies that it took place at some point between quarter−1 and 0.

The Review of Financial Studies / v 25 n 6 2012

1748

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

violators following the violation. The results are striking; both the mean and median violating firm undergo a turnaround in operating cash flows in the quarter immediately following the violation. For instance, within one year of the violation, median operating cash flow recovers to nearly 7% of assets, compared with a median of about 4% in the quarter of the violation. The bottom two panels show that changes in costs are the primary source of improvement in operating cash flow. In the quarters immediately following the violation, operating costs fall by 5% to 10% relative to the quarter of the violation and remain low before drifting up several quarters post-violation. The level of sales falls immediately following the violation but begins to increase a few quarters out.

Table9 extends and confirms the conclusions taken from Figure7. Table9 reports estimates of Equation (1) using the three accounting-based measures of performance as the dependent variable. We also include the lagged level and the one-year first difference of the dependent variable in all specifications.24 The new financial covenant violation dummy enters the operating cash flow and operating cost regressions with a statistically significant estimate that suggests an improvement in performance. For instance, Column (3) of Table9 indicates that the operating cash flow/average assets of firms violating a covenant are 1.2 percentage points higher in the year following the violation, compared with the set of control firms that have experienced similar drops in operating performance. This represents about a 34% increase in income as compared with the median level of operating cash flow from the quarter of the violation (3.5%) and an increase equivalent to 11% of the cash flow of the median Compustat firm (10.4 %). The increase in operating cash flow is not short-lived and increases again during the second year post-violation. The estimated coefficient in specification (4) shows that operating cash flow is 1.5 percentage points higher two years post-violation.

Based on the regression results in Table9, the short-run improvement in operating cash flow is derived entirely from a decrease in operating costs. Specification (7) shows that sales growth decreases following a covenant violation, and specification (11) shows that operating cost growth falls by a statistically significant 4.4%. Two years after a violation, there is evidence that improvements in operating cash flow follow from both an increase in revenue and a decrease in costs (column (8) and column (12)). The pattern suggests that the initial actions taken post-violation work to first reduce costs and later increase revenue growth.

24 This is especially important here becauseBarber and Lyon(1996) show that cash-flow-based measures of operating performance exhibit strong mean reversion, so we would bias our estimates of the impact of a violation upward by not properly controlling for performance prior to the violation.

Creditor Control Rights, Corporate Governance, and Firm Value

1749

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

T a

b le

9 F

in a

n ci

a lc

o ve

n a

n t vi

o la

tio n

s a

n d

o p

e ra

tin g

p e

rf o

rm a

n ce

P a

n e

l A

C h

a n

g e

in o

p e

ra tin

g ca

sh flo

w / a ve

ra g

e a

ss e

ts C

h a

n g

e in

L n

(s a

le s)

(1 )

(2 )

(3 )

(4 )

(5 )

(6 )

(7 )

(8 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t 0.

02 0∗

∗ 0.

02 3∗

∗ 0.

01 2∗

∗ 0.

01 1∗

∗ −

0. 03

3∗ ∗

− 0.

02 1∗

∗ −

0. 02

9∗ ∗

− 0.

02 7∗

∗

(0 .0

04 )

(0 .0

04 )

(0 .0

03 )

(0 .0

04 )

(0 .0

09 )

(0 .0

08 )

(0 .0

08 )

(0 .0

09 )

N ew

fin an

ci al

co ve

na nt

vi ol

at io

n t− 4

0. 01

5∗ ∗

0. 01

1 (0

.0 03

) (0

.0 07

) O

pe ra

tin g

ca sh

flo w

/a ve

ra ge

as se

ts −

0. 31

9∗ ∗

− 0.

23 2∗

∗ −

0. 37

4∗ ∗

− 0.

37 6∗

∗ 0.

06 2∗

∗ 0.

23 6∗

∗ 0.

08 6∗

∗ 0.

08 5∗

∗

(0 .0

12 )

(0 .0

13 )

(0 .0

15 )

(0 .0

17 )

(0 .0

23 )

(0 .0

35 )

(0 .0

37 )

(0 .0

41 )∗

∗

Le ve

ra ge

ra tio

0. 01

5 0.

05 3∗

− 0.

02 6

− 0.

03 0

0. 02

1 0.

06 7

0. 02

3 −

0. 01

8∗ ∗

(0 .0

08 )

(0 .0

23 )

(0 .0

25 )

(0 .0

25 )

(0 .0

33 )

(0 .0

81 )

(0 .0

84 )

(0 .0

88 )∗

∗

In te

re st

ex pe

ns e

/a ve

ra ge

as se

ts −

0. 24

4∗ ∗

− 0.

79 6∗

∗ −

0. 41

6∗ ∗

− 0.

51 1∗

∗ −

0. 59

8∗ ∗

0. 49

2 0.

19 2

0. 45

4∗ ∗

(0 .0

54 )

(0 .2

14 )

(0 .2

02 )

(0 .2

04 )

(0 .2

60 )

(0 .7

95 )

(0 .8

38 )

(0 .8

82 )∗

∗

N et

w or

th /a

ss et

s −

0. 00

7 −

0. 02

3∗ ∗

− 0.

07 6∗

∗ −

0. 07

4∗ ∗

0. 00

5 0.

05 7

0. 06

4 0.

05 2∗

∗

(0 .0

06 )

(0 .0

10 )

(0 .0

12 )

(0 .0

12 )

(0 .0

22 )

(0 .0

34 )

(0 .0

43 )

(0 .0

45 )∗

∗

C ur

re nt

ra tio

− 0.

00 4∗

∗ 0.

00 0

− 0.

00 1

− 0.

00 1

− 0.

01 2∗

∗ 0.

01 9∗

∗ 0.

01 4∗

∗ 0.

01 6∗

∗

(0 .0

00 )

(0 .0

02 )

(0 .0

02 )

(0 .0

02 )

(0 .0

02 )

(0 .0

06 )

(0 .0

06 )

(0 .0

06 )∗

∗

M ar

ke t-

to -b

oo k

ra tio

− 0.

00 1

0. 00

1 0.

00 3

0. 00

7 0.

03 2∗

∗ 0.

10 9∗

∗ 0.

13 5∗

∗ 0.

12 4∗

∗

(0 .0

01 )

(0 .0

04 )

(0 .0

04 )

(0 .0

04 )

(0 .0

03 )

(0 .0

11 )

(0 .0

12 )

(0 .0

13 )∗

∗

H ig

he r-

or de

rc ov

en an

tc on

tr ol

s N

o Ye

s Ye

s Ye

s N

o Ye

s Ye

s Ye

s La

gg ed

co ve

na nt

co nt

ro ls

N o

N o

Ye s

Ye s

N o

N o

Ye s

Ye s

N o.

of ob

se rv

at io

ns 14

8, 83

5 14

8, 83

5 14

2, 04

3 11

8, 77

7 14

9, 49

9 14

9, 49

9 14

2, 64

3 11

9, 29

3 R

2 0.

17 0.

18 0.

21 0.

21 0.

09 0.

10 0.

10 0.

10

(c o

n ti n

u e

d )

The Review of Financial Studies / v 25 n 6 2012

1750

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 9 Continued

PanelB

Change in operating cash flow / averageassets

(9) (10) (11) (12)

New financial covenant violationt −0.052∗∗ −0.052∗∗ −0.044∗∗ −0.043∗∗ (0.009) (0.009) (0.009) (0.011)∗∗

New financial covenant violationt−4 −0.014∗ (0.007)

Operating cash flow / average assets 0.282∗∗ 0.296∗∗ 0.362∗∗ 0.370∗∗ (0.027) (0.030) (0.030) (0.033)∗∗

Leverage ratio 0.056∗ 0.172∗∗ 0.221∗∗ 0.183∗∗ (0.026) (0.060) (0.065) (0.070)∗∗

Interest expense / average assets −0.032 0.752 −0.237 0.240∗∗ (0.189) (0.563) (0.580) (0.590)∗∗

Net worth / assets 0.072∗∗ 0.148∗∗ 0.207∗∗ 0.185∗∗ (0.021) (0.030) (0.038) (0.039)∗∗

Current ratio 0.003∗∗ 0.013∗∗ 0.012 0.012∗ (0.001) (0.004) (0.005) (0.005)∗∗

Market-to-book ratio 0.032∗∗ 0.118∗∗ 0.135∗∗ 0.118∗∗ (0.002) (0.008) (0.009) (0.008)∗∗

Higher-order covenant controls No Yes Yes Yes Lagged covenant controls No No Yes Yes Number of observations 148,945 148,945 142,150 118,866 R2 0.12 0.13 0.14 0.14

This table presents first-difference estimates of the marginal effect of a new covenant violation on the operating performance of firms from the quarter of the violation to four quarters after the violation. Panel A examines changes in operating cash flow scaled by assets and sales growth; Panel B examines growth in operating costs. Subscriptt refers to a violation in the contemporaneous quarter, and (t − 4) refers to a violation four quarters prior. Covenant controls are the six variables: operating cash flow scaled by average assets; the leverage ratio; the ratio of interest expense to average assets; the ratio of net worth to total assets; the current ratio; and the market-to-book ratio. Higher-order covenant controls refers to the second and third power of the covenant control variables; lagged covenant controls refers to the four-quarter lagged level of the covenant control variables. All specifications include industry, calendar-quarter, and fiscal-quarter fixed effects and the following control variables: the level and first difference of Ln(assets), the level and first difference of (PPE / assets), and the level and first difference of the corresponding dependent variable. Standard errors are clustered by firm and calendar quarter. ** and * denote 1% and 5% levels of significance, respectively. No., number.

6.2 Stock return evidence We explore the impact of violations on stock returns in Figures8 and 9 and Tables10 and11. Figure8 and Table10 present standard event-time estimates of long-run abnormal stock returns to firms following a covenant violation. Figure 9 and Table11 summarize the performance of a “covenant violator portfolio” that holds an equally weighted portfolio of covenant violating firms, adding firms in the month after they report a violation and dropping them after a fixed holding period (Figure9 assumes a two-year holding period). As later discussed in more detail, we judge performance by comparing the stock return performance of covenant-violating firms to a benchmark portfolio with similar risk characteristics.

Figure 8 plots event-time abnormal returns around the report of a new covenant violation and shows substantial abnormal gains in the returns to violating firms following a covenant violation. In a pattern similar to the

Creditor Control Rights, Corporate Governance, and Firm Value

1751

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 8 Financial covenant violations and stock price performance (event time) This figure displays results from a stock price event study around the occurrence of a new covenant violation by estimating the event-study monthly abnormal returns of stocks following the report of a new loan covenant violation in an SEC 10-K or 10-Q filing. A new violation is a violation by a firm that has not violated in the previous four quarters. The estimates are for event months September 1997 through June 2009 and include 3,699 observations. The violation occurs at month 0, and the figure shows monthly cumulative average abnormal return estimates beginning one year before the violation. Abnormal returns measured against a four-factor return model, measured on a monthly basis, are (1) the excess return on the NYSE/AMEX market return; (2) the difference between the returns on small and big stocks; (3) the return performance of value stocks relative to growth stocks; and (4) the return performance of high momentum stocks relative to low momentum stocks.

turnaround in operating performance shown in Figure7, violating firms experience a large drop in their stock price prior to the violation—on the order of 20% in the 12 months leading up to the violation—and a recovery in the stock price post-violation. Whereas abnormal returns remain relatively flat in the first three months following the violation report, returns increase thereafter at a rate of about 5% per year above the risk-adjusted benchmark.

For a more detailed look at the event-time returns of covenant violators, we report in Table9 estimates of abnormal performance over a variety of event windows. We construct the abnormal return estimates using the regression format developed in Thompson (1985) and Sefcik and Thompson (1986):

ri = ai + X ′ Bi + 1

′ iΓi + εi , i = 1, 2, ....., N. (2)

In Equation (2), ri is a T x 1 vector of monthly stock returns in excess of the one-month Treasury bill rate for each covenant violatori . We regress these returns on an intercept, aT x m set of benchmark monthly return factorsX (m = 1 or m = 4) and aT x (k0 + k1 + 1) matrix of dummy variables that identify thek0 months prior to the event, the event month, and thek1 months after the event. We define the event month to be the month in which a firm reports a new covenant violation to the SEC. The coefficientsBi are a set of loadings on them factors, and the coefficientsΓ i are thek0 + k1 + 1

The Review of Financial Studies / v 25 n 6 2012

1752

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Figure 9 Financial covenant violations and stock price performance (calendar time) This figure displays cumulative stock returns for a covenant violator portfolio and the implied return for a portfolio of all stocks (including nonviolators) of similar risk as measured by a four-factor benchmark model. The figure covers the period September 1997 through March 2010, and includes violations reported between September 2007 and June 2009. The covenant violator portfolio is formed by purchasing stocks of firms that report a new covenant violation and holding the stocks for two years. A new violation is a violation by a firm that has not violated in the previous four quarters. The stocks are purchased on the first trading day of the month following the reported violation, and the portfolio is equally weighted. The implied returns are constructed using a four-factor benchmark portfolio based on (1) the excess return on the NYSE/AMEX market return; (2) the difference between the returns on small and big stocks; (3) the return performance of value stocks relative to growth stocks; and (4) the return performance of high momentum stocks relative to low momentum stocks.

monthly abnormal returns around firmi ’s report of a violation. The intercept ai measures firmi ’s abnormal return in the nonevent period.

We include four risk factors in our benchmark model, including the three factors fromFama and French(1993), as well as a momentum factor. The factor returns are (1) the monthly return on the equally weighted average index of NYSE/AMEX and NASDAQ stocks, measured in excess of the one-month Treasury bill rate; (2) the average return on a small capitalization portfolio minus the average return on a large capitalization portfolio; (3) the average return on a value (high book-to-market) minus the return on a growth (low book-to-market) portfolio; and (4) the difference in the monthly return of stocks with high returns over the trailing 11 months and stocks with low returns over the trailing 11 months. All four monthly series are downloaded from Kenneth French’s Web-based data library.25

For each violating firm, we estimate the parameters of Equation (2) using all monthly return observations between September 1997 and March 2010. We require that a firm have at least one usable return during the relevant event

25 Available athttp://mba.tuck.dartmouth.edu/pages/faculty/ken.french/data library.html.

Creditor Control Rights, Corporate Governance, and Firm Value

1753

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 10 Event-study estimates of stock price performance following a covenant violation

Panel A: Factor loading estimates

Excess market Small minus High minus High minus low Intercept return big stocks low growth momentum

Mean factor loading estimates

0.000 1.000 0.922 0.047 −0.335∗∗

(0.001) (0.014)∗∗ (0.016)∗∗ (0.024)∗∗ (0.014)∗∗

Panel B: CAR estimates

Event mo. Event mo. Event mo. Event mo. Event mo. Event mo. [0] [+1, +3] [+1,+6] [+1, +12] [+1, +24] [+1,+60]

Mean CAR −0.0221 −0.0008 0.0025 0.0032 0.0046 0.0010 Standard error

estimates (0.0039)∗∗ (0.0023) (0.0016) (0.0012)∗ (0.0009)∗∗ (0.0006) [0.0065]∗∗ [0.0040] [0.0027] [0.0019] [0.0012]∗∗ [0.0008]

This table reports event-time estimates of stock price performance of firms violating a loan covenant by estimating the event-study monthly abnormal returns of stocks following the report of a new loan covenant violation in an SEC 10-K or 10-Q filing. A new violation is a violation by a firm that has not violated in the previous four quarters. The estimates are for event months September 1997 through June 2009 and include 3,699 observations. Abnormal returns measured against a four-factor return model, measured on a monthly basis, are (1) the excess return on the NYSE/AMEX market return; (2) the difference between the returns on small and big stocks; (3) the return performance of value stocks relative to growth stocks; and (4) the return performance of high momentum stocks relative to low momentum stocks. Standard error estimates based on cross-sectional averages of firm-level standard are in parentheses, and time-series clustering-robust standard errors are in brackets. ** and * denote 1% and 5% levels of significance, respectively. CAR, cumulative abnormal return; mo., month.

window and enough return observations to reliably estimate the parameters in Equation (2). Importantly, we account for firms that delist over the period by using the CRSP delisting amount (dlamt) in the month of the delisting to compute a delisting return.26 The delisting amount is the value of the delisting share following an exchange or merger offer or, if no such information exists, the price of the stock on its last trading day.

We judge the statistical significance of the post-event CARs using both cross-sectional averages of individual standard error estimates and time-series standard errors that are robust to clustering within a month.27 To calculate the clustering-robust standard errors, we first group the covenant violator-level CAR estimates by month and calculate the mean CAR within the month. We then compute the standard error of the monthly CARs, weighted by the number of covenant violators within the month. This method provides a conservative adjustment for clustering since it excludes any information obtainable from within-month variation in CARs across covenant violators.

26 This is important because delisting firms tend to have negative returns. The average stock of a firm that violates a covenant and later delists because of a merger earns 2.0 % in its last month on the exchange. The average stock of a firm that violates a covenant and later delists because of financial distress experiences an average return in the last month of trading of−28 %.

27 SeeCampbell, Lo, and MacKinlay(1997, pp. 160–61).

The Review of Financial Studies / v 25 n 6 2012

1754

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Table 11 Calendar–time estimates of stock price performance following a covenant violation

Holding period

12 mo. 12 mo. 24 mo. 24 mo. 60 mo. 60mo.

Abnormal performance (α) 0.0028 0.0019 0.0049 0.0039 0.0046 0.0037 (0.0049) (0.0037) (0.0046) (0.0035) (0.0043) (0.0033)

Excess market return 1.408∗∗ 1.002∗∗ 1.365∗∗ 0.986∗∗ 1.314∗∗ 0.952∗∗ (0.099) (0.098) (0.091) (0.009) (0.082) (0.087)

Small minus big stocks 1.129∗∗ 1.076∗∗ 1.027∗∗ (0.127) (0.118) (0.116)

High minus low growth 0.104 0.096 0.042 (0.145) (0.134) (0.125)

High minus low momentum −0.440∗ −0.399∗ −0.358∗ (0.116) (0.108) (0.110)

Observations (mo.) 151 151 151 151 151 151

This table reports calendar-time estimates of the stock price performance of aCovenant Violator Portfolio measured against a four-factor return model over the period September 1997 through March 2010, for violations reported between September 2007 and June 2009. Stocks are added to the portfolio in the month following the first report of the covenant violation in an SEC 10-K or 10-Q filing, and are held in the portfolio for k months, or until they are delisted, whichever comes first. Delisting firms are assumed to earn the CRSP delisting amount in the month following the delisting and are zero afterward. The portfolio is rebalanced monthly. The abnormal performance estimateα is the intercept from a regression of the monthly return on the covenant violator portfolio on four-factor returns measured at a monthly frequency, all downloaded from /mba.tuck.dartmouth.edu/pages/faculty/ken.french/: (1) the excess return on the NYSE/AMEX market return; (2) the difference between the returns on small and big stocks; (3) the return performance of value stocks relative to growth stocks; and (4) the return performance of high momentum stocks relative to low momentum stocks. White’s (1980) heteroskedasticity consistent standard errors are in parentheses. ** and * denote 1% and 5% levels of significance, respectively. mo., month.

The first result to note in Table10 is that the AR estimate from the event month is negative and statistically significant. This is not surprising because the event-month AR incorporates information about the events leading up to the covenant violation. Moreover, evidence from Beneish and Press (1995)— who document a decline in stock prices in the days around the announcement of a covenant violation—suggests that investors do not immediately impound future performance improvements into the stock price of a violator once a violation becomes public.28

Following the event month and in the first three months following the violation, CAR estimates are positive. Within 12 months of the violation report, violating firms earn an average CAR of 0.32% per month (3.84% per year); these estimates are statistically significant at the 5% level using the cross-sectional standard errors and marginally insignificant using the pooled time-series standard errors. By 24 months post-violation, the violators earn an average abnormal return of 0.46% per month (5.5% per year) that is

28 Our performance and valuation results stand in contrast to existing accounting studies of the consequences of loan covenant violations.Sweeney(1994), DeFond and Jiambalvo(1994), andBeneish and Press(1993, 1995a,b) find that violations are associated with greater subsequent accounting manipulation, poorer loan terms for the borrower, increased financial distress, and declines in borrower value. These articles emphasize some of the costs of covenant violations, but do not account for the possibility that the violations also force actions that improve borrower performance. Additionally, the investigations of the impact of the violation on stock returns focus only on the few days around the announcement.

Creditor Control Rights, Corporate Governance, and Firm Value

1755

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

statistically significant at the 1% level using either standard error estimate. Abnormal returns begin to level off beyond two years following the violation; the cumulative abnormal return at month 60, though positive, is largely indistinguishable from zero. This reflects the inclusion of a large number of months with near-zero abnormal returns, suggesting that the impact of the violation does not persist past two years.

Table 11 reports calendar-time estimates of abnormal performance of a covenant violator portfolio over various holding periods. The covenant violator portfolio is created by buying all stocks that report a new covenant violation at the beginning of the month following the report of the violation and holding the stocks in an equally weighted portfolio forn months before selling the stocks, wheren is a pre-specified holding period. We examine holding periods of 12, 24, and 60 months. The portfolio is created starting in September 1997 and rolls forward by adding new violators each month and dropping violators at the end of their holding period. The performance of the portfolio is judged relative to a factor model specified using time-series regressions of monthly portfolio excess returns on monthly estimates of the factors and a “Jensen’s alpha” intercept. We report estimates from four- and one-factor market models. Under each specification and assumed holding period, the estimate of alpha provides our measure of average abnormal performance.

Figure 9 plots the time-series returns of the violator portfolio and the implied returns on the benchmark portfolio for a 24-month holding period. The figure shows that the covenant violator portfolio begins to outperform the benchmark portfolio in late 1999, approximately two years following the start of the portfolio. Returns on the violator portfolio outperform the benchmark through mid-2007, stay well above the benchmark through mid-2008, and then experience a large decline for three months starting in September 2008, coinciding with the worsening of the financial crisis following the bankruptcy filing of Lehman Brothers. By early 2009, the covenant violator portfolio recovers, so by March 2010, the cumulative return on the violator portfolio over the entire period is 184%, more than double the benchmark cumulative return of 76% over the same timespan.

Table11 provides point estimates and standard errors for the intercept from the time-series regressions, along with factor loadings. Most notably, the time- series regressions produce point estimates that are similar in magnitude to the event-time estimates in Table10. Average abnormal returns are positive across all holding periods and peak in the 24-month portfolio at about 0.5% per month (6% per year) before trailing off in the five-year holding period portfolio. The difference between the results in Tables10 and 11 is that our calendar-time abnormal performance measures are estimated with less precision. None of the abnormal return estimates in Table11 are statistically significant at conventional levels. The larger standard errors likely reflect the low power of tests of long-run abnormal returns using the simple Jensen’s alpha framework (Gur-Gershgoren, Hughson, and Zender 2008). Because our

The Review of Financial Studies / v 25 n 6 2012

1756

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

goal is to document the general pattern of abnormal returns in the violator portfolio—rather than to devise the most powerful test for the significance of those results—we stop with the basic specification presented in Table11.

Taken together, the operating and stock return results imply that violating firms, on average, increase in value following a covenant violation. Holding the value of the creditors’ claims constant, the event-study methodology measures the abnormal performanceof the firm, since equityholders are the residual claimants on firm cash flows. Of course, observed changes in shareholder wealth are a function of both the value of the firm and the value of debtholder claims. Because it is unlikely that creditors will act in a way that reduces the value of their claims post-violation, concerns of confounding changes in shareholder value with changes in firm value would only arise if equity returns fall in response to the violation. We show that the value of equity claims actually rises post-violation.

A potential concern with our performance results is that it may reflect sample selection, rather than creditor intervention. It could be that creditors quickly liquidate very poor-performing firms and let continue those with a brighter outlook. However, as discussed in Section 4, a covenant violation is associated with only a 4-percentage-point increase in the probability of exit within a year, and this effect is reduced to 1 percentage point when we include basic controls for ex ante performance. Almost all violators survive for at least one year post-violation, and it is over this year that we see improvements in average performance. Moreover, our stock return results are not subject to any bias created by bankruptcy or liquidation since we account for delisting returns that accrue to stockholders.

A natural question arising from the stock return results is: why does it take so long for stock prices to reflect the influence of violations? Forward-looking investors, who understand that a covenant violation leads, on average, to a turnaround in performance, should quickly incorporate this information into stock prices, thereby “flattening” any effect of a covenant violation on stock returns. Our preferred explanation is that investors would have had difficulty knowing which firms were in violation of a covenant and would not have likely known the value implications. Our data collection procedure suggests that extracting the necessary information from footnotes in SEC filings is not a trivial procedure. Moreover, the Edgar system that facilitates our data collection has only been around since 1996, which until recently has not provided much history with which to evaluate the performance implications. Of course, it is difficult for us to rule out the alternative explanation in which observed returns are compensation for some risk that is not in our benchmark model. However, combined with our operating results, we believe that stock returns provide supportive evidence that firm valuations increase, and certainly do not decrease, following covenant violations.

In sum, our performance estimates suggest that creditor interventions following covenant violations are associated with improvements in firm value.

Creditor Control Rights, Corporate Governance, and Firm Value

1757

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

These findings are consistent with the idea that increases in creditor influence on the governance of the firm represent an optimal shift in control rights to the party that has the most incentive to monitor and influence the firm when it is performing poorly. This control shift has a positive knock-on effect that benefits equityholders, even as the creditors move to protect their own claims.

7. Conclusions

We offer evidence that firms in violation of a covenant in a private debt agreement change senior management, become more conservative in their financial and investment policy, and thus improve performance. Given the well-documented set of control rights given to creditors following a covenant violation, we interpret the evidence as suggesting that creditors serve a corporate governance role that helps increase the value of the firm. These changes occur despite the fact that violators are not on the verge of bankruptcy or payment default; creditors play an important corporate governance role, even outside of payment default states. Taken together, our results provide a unique look into an aspect of corporate governance that has been largely overlooked by the traditional corporate governance literature.

We strengthen the extant evidence ofRoberts and Sufi(2009) andNini, Smith, and Sufi(2009), who find that contract terms can become more restrictive following a covenant violation and the new restrictions influence firm behavior. Creditors can also influence firm behavior through exerting behind-the-scenes pressure on managers and force CEO turnover. We also demonstrate that creditor influence extends beyond affecting debt issuance and capital expenditures; violations of financial covenants lead to important changes in virtually every dimension of firms’ investment and financing. These results are consistent with the extensive body of literature showing that financial intermediaries are valuable as delegated monitors, especially when there are unresolved conflicts of interest between managers and the providers of finance in public companies.

In terms of future research, we have not provided evidence on the precise channel through which creditors affect CEO turnover. Do creditors use re- structuring or turnaround firms to force the CEO out? Do board members use the violation as a signal to fire the CEO? These are interesting questions that should be addressed. Regardless of the exact mechanism, our findings support the view that models of corporate governance should recognize the control rights available to creditors in the optimal governance structure.

References Adams, R., B. Hermalin, and M. Weisbach. 2010. The Role of Boards of Directors in Corporate Governance: A Conceptual Framework and Survey.Journal of Economic Literature48:58–107.

Aghion, P., and P. Bolton. 1992. An Incomplete Contracts Approach to Financial Contracting.Review of Economic Studies59:473–94.

The Review of Financial Studies / v 25 n 6 2012

1758

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Andrade, G., and S. Kaplan. 1998. How Costly Is Financial (Not Economic) Distress? Evidence from Highly Leveraged Transactions That Become Distressed.Journal of Finance53:1443–93.

Baird, D., and R. Rasmussen. 2006. Private Debt and the Missing Lever of Corporate Governance.University of Pennsylvania Law Review154:1209–51.

Barber, B., and J. Lyon. 1996. Detecting Abnormal Operating Performance: The Empirical Power and Specification of Test Statistics.Journal of Financial Economics41:359–400.

Beneish, M., and E. Press. 1993. Costs of Technical Default.Accounting Review68:233–57.

. 1995a. The Resolution of Technical Default.Accounting Review70:337–53.

. 1995b. Interrelation Among Events of Default.Contemporary Accounting Research12:57–84.

Bebchuk, L. 2007. The Myth of the Shareholder Franchise.Virginia Law Review93:675–732.

Bebchuk, L., A. Cohen, and A. Ferrell. 2009. What Matters in Corporate Governance?Review of Financial Studies22:783–827.

Bhagat, S., and B. S. Black. 2002. The Non-correlation Between Board Independence and Long-term Firm Performance.Journal of Corporation Law27:231–73.

Bhagat, S., and B. Bolton. 2008. Corporate Governance and Firm Performance.Journal of Corporate Finance 14:257–73.

Brav, A., W. Jiang, F. Partnoy, and R. Thomas. 2008. Hedge Fund Activism, Corporate Governance, and Firm Performance.Journal of Finance63:1729–75.

Brav, A., W. Jiang, and H. Kim. 2009. Hedge Fund Activism: A Review.Foundations and Trends in Finance 4:1–66.

Campbell, J. Y., A. W. Lo, and A. C. MacKinlay. 1997.The Econometrics of Financial Markets. Princeton, NJ: Princeton University Press.

Chava, S., and M. Roberts. 2008. How Does Financing Impact Investment? The Role of Debt Covenants.Journal of Finance63:2085–121.

Chen, K., and J.Wei. 1993. Creditors’ Decisions to Waive Violations of Accounting-based Debt Covenants. Accounting Review68:218–32.

Clifford, C. 2008. Value Creation or Destruction? Hedge Funds as Shareholder Activists.Journal of Corporate Finance14:323–36.

Daniels, R., and G. Triantis. 1995. The Role of Debt in Interactive Corporate Governance.California Law Review 83:1073–113.

DeAngelo, H., L. DeAngelo, and K. Wruck. 2002. Asset Liquidity, Debt Covenants, and Managerial Discretion: The Collapse of L.A. Gear.Journal of Financial Economics64:3–34.

DeFond, M., and J. Jiambalvo. 1994. Debt Covenant Violation and Manipulation of Accruals.Journal of Accounting and Economics17:145–76.

Dewatripont, M., and J. Tirole. 1994. A Theory of Debt and Equity: Diversity of Securities and Manager- shareholder Congruence.Quarterly Journal of Economics109:1027–54.

Dichev, I., and D. Skinner. 2002. Large Sample Evidence on the Debt Covenant Hypothesis.Journal of Accounting Research40:1091–123.

Fama, E., and K. French. 1993. Common Risk Factors in the Returns on Stocks and Bonds.Journal of Financial Economics33:3–56.

. 1997. Industry Costs of Equity.Journal of Financial Economics47:427–65.

Gale, D., and M. Hellwig, 1985. Incentive-compatible Debt Contracts: The One-period Problem.Review of Economic Studies52:647–63.

Creditor Control Rights, Corporate Governance, and Firm Value

1759

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Gillan, S., and L. Starks. 2007. The Evolution of Shareholder Activism in the United States.Journal of Applied Corporate Finance19:55–73.

Gilson, S. 1989. Management Turnover and Financial Distress.Journal of Financial Economics25:241–62.

. 1990. Bankruptcy, Boards, Banks, and Blockholders: Evidence on Changes in Corporate Ownership and Control When Firms Default.Journal of Financial Economics27:355–87.

Gilson, S., K. John, and L. Lang. 1990. Troubled Debt Restructurings: An Empirical Study of Private Reorganization of Firms in Default.Journal of Financial Economics2:315–53.

Gilson, S., and M. Vetsuypens. 1994. Creditor Control in Financially Distressed Firms: The Empirical Evidence. Washington University Law Quarterly72:1005–25.

Gilson, S., and J. Warner. 1998. Private versus Public Debt: Evidence from Firms That Replace Bank Loans with Junk Bonds. Working Paper, SSRN abstract 140093.

Gompers, P., J. Ishii, and A. Metrick. 2003. Corporate Governance and Equity Prices.Quarterly Journal of Economics118:107–55.

Gorton, G., and J. Kahn. 2000. The Design of Bank Loan Contracts.Review of Financial Studies13:331–64.

Greenwood, R., and M. Schor. 2009. Hedge Fund Investor Activism and Takeovers.Journal of Financial Economics92:362–75.

Gur-Gershgoren, G., E. Hughson, and J. Zender. 2008. A Simple-but-powerful Test for Long-run Event Studies. Working Paper, SSRN abstract 847791.

Hart, O., and J. Moore. 1998. Default and Renegotiation: A Dynamic Model of Debt.Quarterly Journal of Economics113:1–41.

Hermalin, B., and M. Weisbach. 2003. Boards of Directors as an Endogenously Determined Institution: A Survey of the Economic Literature.Economic Policy Review9:7–26.

Hotchkiss, E., and R. Mooradian. 1997. Vulture Investors and the Market for Control of Distressed Firms. Journal of Financial Economics3:401–32.

Ivashina, V., V. Nair, A. Saunders, N. Massoud, and R. Stover. 2008. Bank Debt and Corporate Governance. Review of Financial Studies22:41–77.

Ivashina, V., B. Iverson, and D. C. Smith. 2011. The Ownership and Trading of Debt Claims in Chapter 11 Restructurings. Working Paper, University of Virginia.

James, C. 1995. When Do Banks Take Equity in Debt Restructurings?Review of Financial Studies8:1209–34.

. 1996. Bank Debt Restructurings and the Composition of Exchange Offers in Financial Distress.Journal of Finance51:711–27.

Jensen, M. 1986. Agency Costs of Free Cash Flow.American Economic Review76:323–29.

Jensen, M., and W. Meckling. 1976. Theory of the Firm: Managerial Behavior, Agency Costs, and Capital Structure.Journal of Financial Economics3:305–60.

Jensen, M., and K. Murphy. 1990. Performance Pay and Top-management Incentives.Journal of Political Economy98:225–63.

Jenter, D., and F. Kanaan. Forthcoming. CEO Turnover and Relative Performance Evaluation.Journal of Finance.

Jiang, W., K. Li, and W. Wang. Forthcoming. Hedge Funds and Chapter 11.Journal of Finance.

Kahan, M., and B. Tuckman. 1993. Private vs. Public Lending: Evidence from Covenants. Working Paper, Anderson Graduate School of Management, UCLA.

Kang, J.-K., and A. Shivdasani. 1995. Firm Performance, Corporate Governance, and Top Executive Turnover in Japan.Journal of Financial Economics38:29–58.

The Review of Financial Studies / v 25 n 6 2012

1760

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from

Kaplan, S., and B. Minton. 1994. Appointments of Outsiders to Japanese Boards: Determinants and Implications for Managers.Journal of Financial Economics36:225–58.

Klein, A., and E. Zur. 2009. Entrepreneurial Shareholder Activism: Hedge Funds and Other Private Investors. Journal of Finance64:187–229.

Lim, J. 2011. The Role of Activist Hedge Funds in Distressed Firms. Working Paper, Ohio State University.

Lim, J., B. A. Minton, and M. S. Weisbach. 2011. Do Equity Holders Receive Better Terms When They Lend to Their Firms? Working Paper, Ohio State.

Murphy, K. J. 1999. Executive Compensation. Working Paper, SSRN abstract 163914.

Nini, G., D. Smith, and A. Sufi. 2009. Creditor Control Rights and Firm Investment Policy.Journal of Financial Economics92:400–420.

Ozelge, S., and A. Saunders. Forthcoming. The Role of Lending Banks in Forced CEO Turnovers.Journal of Money, Credit, and Banking.

Perry, T., and A. Shivdasani. 2005. Do Boards Affect Performance? Evidence from Corporate Restructuring. Journal of Business78:1403–31.

Petersen, M. 2009. Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches.Review of Financial Studies22:435–80.

Roberts, M., and A. Sufi. 2009. Control Rights and Capital Structure: An Empirical Investigation.Journal of Finance64:1657–95.

Roberts, M., and T. Whited. 2011. Endogeneity in Empirical Corporate Finance. In George Constantinides, Milton Harris, and Rene Stulz (eds.),Handbook of the Economics of Finance, vol. 2. Amsterdam: Elsevier.

Sansone, A., and A. Taylor. 2007.The Handbook of Loan Syndications and Trading. New York: McGraw-Hill.

Santos, J., and A. Rumble. 2006. The American Keiretsu and Universal Banks: Investing, Voting, and Sitting on Nonfinancials’ Corporate Boards. Journal of Financial Economics 80:419–54.

SEC. 2003. Interpretation: Commission Guidance Regarding Management’s Discussion and Analysis of Financial Condition and Results of Operations. Available at http://www.sec.gov/rules/interp/33-8350.htm.

Shleifer, A., and R. Vishny. 1997. A Survey of Corporate Governance.Journal of Finance52:737–83.

Smith, C. 1993. A Perspective on Accounting-based Debt Covenant Violations.Accounting Review68:289–303.

Sweeney, A. 1994. Debt Covenant Violations and Managers’ Accounting Responses.Journal of Accounting and Economics17:281–308.

Townsend, R. 1979. Optimal Contracts and Competitive Markets with Costly State Verification.Journal of Economic Theory20:265–93.

Verde, M. 1999. Loan Preserve: The Value of Covenants. FitchIBCA Loan Products Special Report. Available at www.fitchibca.com.

Wruck, K. 1990. Financial Distress, Reorganization, and Organizational Efficiency.Journal of Financial Economics27:419–44.

Creditor Control Rights, Corporate Governance, and Firm Value

1761

at D rexel H

ahnem ann L

ibrary on A pril 20, 2013

http://rfs.oxfordjournals.org/ D

ow nloaded from