Creditor control and product-market competition

Accepted Manuscript

Creditor Control and Product-Market Competition Matthew T. Billett , Burcu Esmer , Miaomiao Yu PII: DOI: Reference:

S0378-4266(17)30149-8 10.1016/j.jbankfin.2017.06.016 JBF 5205

To appear in:

Journal of Banking and Finance

Received date: Revised date: Accepted date:

31 January 2017 22 May 2017 25 June 2017

Please cite this article as: Matthew T. Billett , Burcu Esmer , Miaomiao Yu , Creditor Control and Product-Market Competition, Journal of Banking and Finance (2017), doi: 10.1016/j.jbankfin.2017.06.016

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Creditor Control and Product-Market Competition Matthew T. Billett*, Burcu Esmer**, and Miaomiao Yu***

Abstract

CR IP T

May 2017

We explore how rival firms respond when firms in their industry violate debt covenants. We find that rival firms increase advertising expense, and that this increase is proportional to the

AN US

size of industry violators’ pre-existing market share. Rival firm product-market share also increases in the industry market share of violators, and this relation is more pronounced when products are more substitutable. Rival firm operating performance also increases in proportion

M

to the industry market share of violators. Overall, these findings suggest that the increased creditor control associated with covenant violations has a significant influence on rival firms

PT

ED

and product-market competition.

Keywords: Debt covenants, Covenant violations, Creditor control, Product market

CE

competition JEL Classification: G21, G30, M30

AC

We thank Redouane Elkamhi, Jon Garfinkel, Dave Mauer, Phuong-Anh Nguyen, Raunaq Pungaliya, David Smith, Xuan Tian, and seminar participants at Bilkent University, the University of Saskatchewan, and participants at the 2015 Northern Finance Association, 2014 Turkey Finance Workshop and 2013 Financial Management Association meetings for helpful comments and discussions. All errors remain our own. *Richard E. Jacobs Chair in Finance, Kelley School of Business, Indiana University, 1309 E 10th Street HH6100, Bloomington, IN 47405-1701. 812-855-3366 (Office). Email: [email protected] ; ** The Wharton School, University of Pennsylvania, 3620 Locust Walk, 2455 SHDH, Philadelphia, PA 19104. 215-898-1587 (Office). Email: [email protected] ; *** E.J. Ourso College of Business, Louisiana State University, Baton Rouge, LA 70803. 225-578-7676 (Office). Email: [email protected].

ACCEPTED MANUSCRIPT

1. Introduction We know from existing research that a firm’s capital structure decisions influence competition and product-market outcomes. Firms engage in price wars when their rivals are

CR IP T

constrained by high debt loads (see Chevalier 1995a and 1995b), and firm leverage plays a significant role in its overall competitive position (Brander and Lewis, 1986; Maksimovic, 1988; Philips, 1995; Zingales, 1998; Campello, 2003).1 One common link among these works is the notion that financial flexibility, or lack thereof, may significantly affect firm competitiveness.

AN US

Financial flexibility certainly depends on the quantity and price of leverage, as documented in the prior literature; however, financial flexibility may also depend on debt contract features, like restrictive covenants.

Debt covenants reduce a firm’s flexibility by restricting investment, financing and payout

M

policies directly via incurrence covenants and indirectly via maintenance covenants.2 These

ED

covenants are designed to alleviate agency costs by reducing adverse selection and moral hazard (see Rajan and Winton, 1995; Garleanu and Zwiebel, 2009; Gorton and Kahn, 2000). Prior studies

PT

document that bank loan covenant violations lead to increased creditor control over borrowing

CE

firms, resulting in significant changes in firm policies.3 Nini, Smith, and Sufi (2009, 2012) document that firms experience a decline in investment, sales growth and market share following a violation.

AC

We also know from the extant literature that covenant violations associate with

1 Related

work shows firms with excess cash gain product market share at the expense of their relatively cash poor rivals (Fresard 2010). 2 Incurrence covenants restrict specific actions while maintenance covenants require the firm to maintain financial ratios above or below specified thresholds. Begley (2015) shows that firm actions depend on the financial ratio thresholds commonly found in debt agreements. 3 See, for example, Beneish and Press (1993, 1995), Chen and Wei (1993), Sweeney (1994), Dichev and Skinner (2002), Chava and Roberts (2008), Roberts and Sufi (2009).

1

ACCEPTED MANUSCRIPT

significant decreases in trade credit (Zhang, 2016) and significant increases in asset sales (Ersahin, Irani and Le, 2016), all consistent with the firm contracting (rather than expanding) its operations. These changes may benefit the firm and its shareholders by constraining value-reducing managerial

CR IP T

behavior and limiting managerial inefficiency. In addition, firms decrease their net debt issuance and leverage following violations (Roberts and Sufi, 2009). These changes may strengthen violator firms’ position in the product market. However, it is also possible that some violations associate with lender self-interested policy changes at the expense of shareholders. In the incomplete

AN US

contracts setting covenants are written imperfectly and may trigger suboptimally:

“loan covenants can backfire if they are too inflexible or restrictive by slowing a borrower's growth and development. Borrowers may end up managing the loan covenants, rather than their business.”4 In such a circumstance, the violator’s lack of flexibility may provide a

M

competitive opportunity for rival firms. Regardless of whether covenant violations are optimal

ED

for all stakeholders of violating firms, the policy changes associated with violations and increased creditor control are likely to elicit predatory responses by rivals.

PT

Consistent with this notion, we find that rivals of firms that violate private debt covenants

CE

significantly increase their advertising expense. They experience significant market share gains and improvements in profitability. If these are indicative of strategic actions, then we would expect to

AC

see these actions and consequent effects increase with the potential size of the competitive opportunity to gain share – i.e., increase in the size of the violating firms’ market share. This is indeed the case. We find rivals’ advertising, market share gains, and profitability all increase with the pre-violation market share of the violating firms in their industry. We would

4 “What

you should know about loan covenants”, Spring 2005, Financial Lending Notes.

2

ACCEPTED MANUSCRIPT

also expect to see these relations vary with the degree to which rival products can substitute for violators’ products. Using product durability, R&D intensity, and product fluidity measures we find rival market share gains are more pronounced when violator firms sell less unique products.

CR IP T

We begin by examining changes in sales growth around covenant violations for the industry rivals of violating firms (violators) to capture changes in market share (following Fresard, 2010). Consistent with Nini et al. (2012), we find industry rivals gain significant market share from their covenant violating peers in the year following the violation. Specifically, we

AN US

find violators and matched5 rivals have similar sales growth patterns prior to the violation; however, rival firms experience an average (statistically insignificant) change in sales growth of -1.8% from pre- to post-violation, while the violators in the industry experience a -6.1% change, significant at the 1% level. The difference in these two changes (i.e., difference-in-differences,

M

DID) is 4.2%, significant at the 1% level. Given pre-violation sales growth averages 11% for

ED

matched rival firms, the sales growth difference-in-differences of 4.2% represents a 38.2% change for rivals relative to violators.

PT

We next explore how these changes in market share relate to the nature of the product

CE

market. First, we stratify the sample by the size of the violator firms’ pre-violation market share. We find market share gains for rivals is only significant (DID=4.8%) when violators have a relatively large

AC

market share. We also explore how product uniqueness affects product-market outcomes. If violator firms sell unique products, e.g. produced to a customer’s specification, then it may be costly for buyers to change suppliers. In this case, there is less opportunity for the industry rivals to take advantage of the covenant violation. We find this is indeed the case. When

5 We

match on industry, pre-violation sales growth, and a propensity score. The details are described in Section 4.1.

3

ACCEPTED MANUSCRIPT

violators produce nondurable goods, are less R&D intensive, and have high product fluidity, rival firms gain more market share (with DIDs of 4.6%, 3.2% and 6.8% respectively). We conduct multivariate tests to see how the change in sales growth of industry rivals

CR IP T

following the covenant violations relates to the pre-violation market share of violators. The regression analysis confirms the univariate observations: rivals’ market share gain increase in the violators’ market share.6 The relation is economically as well as statistically significant. A 10% previolation market share for covenant violators’ in a given industry associates with a 7.5% increase in

AN US

rivals’ market share in the year following the violation. Moreover, this relation between rival market share and violators’ market share is more pronounced when products are more similar. One way industry rivals may attempt to gain market share is by actively increasing advertising expense. We find that industry rivals increase their advertising budget following

M

violations, and these changes in advertising increase in proportion to the market share of the

ED

violators. Given that advertising is one way a firm may aggressively pursue rival market share, these results suggest the change in market share is not passive and simply due to a poor performing rival

PT

violator. We also explore rival firms’ operating performance following violations.

CE

We show significant increases in operating cash flow scaled by assets, ROA and ROE for the rival firms. Moreover, we find that all of these post-violation operating performance changes

AC

increase in magnitude with the market share of the violators. One concern for our study is that the rival reactions may be due to a common industry

shock that may have caused the violations. In this case the rivals may be responding to the shock,

6 In

multivariate tests, we aggregate the market share of all violators within an industry quarter to obtain violators’ market share.

4

ACCEPTED MANUSCRIPT

rather than the violators’ condition. To help separate this effect, we explore subsamples in three ways. First, we limit the sample to industries that have nonnegative returns (excluding the violators’ returns) during the last month of a violation quarter, given a negative shock would likely associate

CR IP T

with negative industry returns. Second, we limit the sample to those violations where violators experience negative idiosyncratic returns. Following Leary and Roberts (2014), we argue rivals should only respond to a violators’ idiosyncratic return for strategic reasons. Third, we explore the subsample that have both nonnegative industry returns and low violator idiosyncratic returns. For

AN US

all three subsamples we find that rival market share gains, changes in advertising expense, and changes in profitability increase in the violators’ market share. These results support the notion that rivals are strategically responding to violations.

Another concern is that rivals may be responding to changes in the violator’s condition

M

that led to the covenant violation rather than the violation itself. To address this concern, we

ED

use a regression discontinuity design to identify the violated covenants on loans for which we know the covenant thresholds following Chava and Roberts (2008). We also follow the “quasi-

PT

discontinuity” regression approach of Roberts and Sufi (2009) and Nini, Roberts and Sufi (2012)

CE

and find similar results. Last, we re-estimate our baseline regressions excluding financially distressed violators. For this sample of relatively healthy covenant violators, we continue to find

AC

rivals’ market share increases in that of violators. This study contributes to the literature on the link between firms’ financing decisions and

product-market behavior. Theoretical papers argue that leverage may soften (Fudenberg and Tirole, 1986; Bolton and Scharfstein, 1990; Phillips, 1995; Chevalier and Scharfstein, 1996) or strengthen (Brander and Lewis, 1986; Maksimovic, 1998; Rotemberg and Scharfstein 1990) the 5

ACCEPTED MANUSCRIPT

product market competition. Several empirical papers test these predictions by studying how debt influences the product prices, the market shares gained or lost, or the probability of exit or entry into the market. 7

CR IP T

Although prior studies show how the level of debt in firms’ capital structure affects their product-market strategies, it has been less clear how different features of debt contacts influence these decisions, such as restrictive covenants. Restrictive covenants define the circumstances under which creditors may intervene in management (Aghion and Bolton, 1992;

AN US

Hart and Moore, 1995). Building on this view, several papers investigate whether firm behavior changes once control rights are transferred to creditors by examining bank loan covenant violations.8 These studies document that financial covenant violations lead to increased creditor control over borrowing firms that significantly changes borrowing firms’ investment, financing

M

and payout policies. To our knowledge, our paper is the first to document how industry rivals

ED

respond to their peers’ violations and the product-market outcomes of such actions. 9 Our paper is related to the literature exploring the effect of bankruptcy and financial

PT

distress on product-market competition (Lang and Stulz, 1992; Hertzel, Zhi, Officer, and Rodgers,

CE

2008; Hotchkiss, 1995; Eberhart, Altman and Aggarwal, 1999; Zhang, 2010). These studies examine changes in bankrupt firms’ or their rivals’ stock price and operating performance. Relative to these

AC

studies, we focus on financial covenant violations that occur well outside of

Phillips (1995), Kovenock and Philips (1997), Chevalier (1995a), Chevalier (1995b), Zingales (1998), Opler and Titman (1994), Chevalier and Scharfstein (1996), Khanna and Tice (2000), and Campello (2003). 8 Beneish and Press (1993, 1995), Chen and Wei (1993), Sweeney (1994), Dichev and Skinner (2002), Chava and Roberts (2008), Roberts and Sufi (2009). 9 A recent paper by Nordlund (2016) confirms our result that peer firms increase their sales growth when rivals violate a covenant using a double-randomization identification technique. He concludes that covenant violations enable rivals to strategically prey on violating firms. 7

6

ACCEPTED MANUSCRIPT

financial distress and rarely lead to default (Gopalakrishnan and Parkash, 1995). 10 Moreover, financial covenant violations are very common.11 Thus, the potential impact covenant violations have on product-market strategies is not limited to a small fraction of firms facing extreme circumstances. In fact, even if rivals were responding to changes in violator condition rather

CR IP T

than the violation itself, our study demonstrates that relatively small changes in rival conditions, identified by covenant violations, elicit significant competitive responses.

By studying financial covenant violations, we are able to document the changes in

AN US

product-market outcomes, and how the industry rivals respond when creditors have the opportunity to exert control over corporate behavior in solvent firms on a frequent basis.

2. Related literature and hypothesis development

M

Numerous papers examine the link between firms’ capital structure and product-market

ED

behavior, focusing specifically on the link between the level of debt and product market competition. Theoretical papers have mixed predictions about how leverage affects competition.

PT

Fudenberg and Tirole (1986), Bolton and Scharfstein (1990), Phillips (1995), and Chevalier and

CE

Scharfstein (1996) argue that competition becomes softer when leverage increases. However, other theoretical papers such as Brander and Lewis (1986), Maksimovic (1998), and Rotemberg and

AC

Scharfstein (1990) predict that firms have an incentive to increase debt to gain a strategic advantage over their rivals. The link between financing and product-market outcomes has also

10 Firms

which violate financial covenants may have performance deterioration leading to the violation. However, the median firm in violation of a financial covenant has comparable liquidity and valuation measures compared to the median firm in the full sample. 11 Nini et al. (2012) show that between 10% and 20% of public firms were in violation of a covenant during any particular quarter and more than 40% of the firms were in violation at some point during the 1996- 2007 period.

7

ACCEPTED MANUSCRIPT

been investigated empirically. Phillips (1995) and Kovenock and Philips (1997) focus on increased leverage in manufacturing industries. Chevalier (1995a, 1995b) analyzes the prices at local supermarkets after LBOs, Zingales (1998) analyzes the deregulation in the trucking industry. In each

CR IP T

of these studies, debt is shown to influence the product prices, the market shares gained or lost, or the probability of exit or entry into the market. Complementing these studies, Opler and Titman (1994), Chevalier and Scharfstein (1996), Khanna and Tice (2000), and Campello (2003) examine how debt influence firms’ responses to shocks to competitive environments.

AN US

The extant literature on capital structure and product-market competition has mostly focused on how the level of leverage impacts product-market outcomes. However, there are other important features of leverage that may have similar effects on product-market outcomes, one of which is covenant restrictions in debt agreements. The existence of covenants is motivated and

M

rationalized by their ability to mitigate agency conflicts by reducing adverse selection and moral

ED

hazard through better screening of borrowers and lowering monitoring costs (Smith and Warner, 1979; Myers, 1977; Smith, 1993; Hart, 1995; Tirole, 2006). The optimal contracting literature

PT

examines security design models where covenants define the circumstances under which creditors

CE

are permitted to intervene in management (e.g. Aghion and Bolton 1992; Dewatripont and Tirole 1994). Building on this, recent studies document that financial covenant violations lead to increased

AC

creditor control over borrowing firms, resulting in significant changes in firms’ financial, investment, and payout policies. Covenant violations are followed by a decrease in capital and R&D investments (Chava and Roberts, 2008; Nini et al. 2009, Gu, Mao and Tian, 2013; Chava, Nanda and Xiao, 2016), a decrease in net debt issuing activity, leverage (Roberts and Sufi, 2009) and shareholder payouts (Nini et al., 2012), a decline in sales 8

ACCEPTED MANUSCRIPT

growth and market share and an increase in CEO turnover (Nini et al., 2012), an increase in employment risk (Falato and Liang, 2016), improvements within-firm resource allocation (Ersahin et al. , 2016), an increase in risk and risk-taking activities (Esmer, 2016; Gao, Khan and

Lobo, 2014), and a decrease in trade credit (Zhang, 2016).

CR IP T

Tan, 2017), higher accounting and real earnings management activities (Franz, HassabElnaby,

These lender induced policy changes may be designed to curb prior overinvestment or induce underinvestment (to the benefit of lenders). Nini et al. (2012) argue that shareholders

AN US

benefit from violator policy changes, on average, and Ersahin et al. (2016) show that violators sell or close inefficient plants. In addition, Roberts and Sufi (2009) show that borrowing firms adopt more conservative financing policies following violations. These suggest that the lenderinduced policy changes may be value maximizing and may enhance borrowing firm’s

M

competitiveness in the product market.

ED

It is also possible that these creditor-induced, creditor-favorable policy changes in borrowing firms reduce managers’ ability to pursue first-best outcomes (Grossman and Hart,

PT

1986; Hart and Moore, 1988 and 1990; Hart, 1995; Tirole, 1999) at the expense of shareholders.

CE

In this case, the violator’s lack of flexibility may permit rivals to strategically prey on the firm. These two countervailing arguments form the basis of our main hypothesis:

AC

Hypothesis 1: Covenant violations enhance firms’ competiveness by improving

investment and financing policies, strengthen their product market position, and result in rivals having market share losses.

9

ACCEPTED MANUSCRIPT

Alternative Hypothesis 2: Covenant violations hinder firms’ competiveness by reducing their flexibility through restrictive investment and financing policies, and result in rivals having market share gains.

CR IP T

3. Data

Financial covenant violation data is obtained from Amir Sufi’s website.12 The sample construction below follows Nini et al. (2012) using the period 1997 through 2008. To be included in

AN US

the sample, we require firms to have data available on the Compustat database, and have average book assets greater than $10 million in 2000 dollars. We exclude financial firms and firms with missing information on total assets, sales, common shares outstanding, and closing share price. Imposing these restrictions leaves a sample of 8,199 firms and 180,335 firm-quarter observations.

M

We follow Nini et al. (2012) and focus our analysis on new financial covenant violations, which are

ED

defined as financial covenant violations for firms that have not violated a covenant in the previous four quarters.13 The reason to focus on the new covenant violations is that they help pinpoint the

PT

initiation of creditor intervention. Therefore, how industry rivals respond when another firm in the

CE

industry is the subject of creditor control will be more clearly identified. In our sample, we have

AC

3,604 new violations between 1997 and 2008.14

This data is available at http://faculty.chicagobooth.edu/amir.sufi/data.htm. Nini, Smith, and Sufi extract information from every 10-Q and 10-K filing on SEC Edgar website. Using a text-searching algorithm, they determine whether a firm is in violation of a covenant. Then they match this information to COMPUSTAT file. For more information on the data, please see Appendix of “Creditor Control Rights, Corporate Governance and Firm Value” by Nini et al. (2012). 13 Based on the violation sample downloaded from Sufi’s website, we assume that a firm is not in violation of a covenant during the quarter if there is missing violation variable in the dataset after the starting quarter of the collection for each firm. 12

10

ACCEPTED MANUSCRIPT

4. Results 4.1. Univariate Results Nini et al. (2012) find that firms that violate covenants experience negative industry

CR IP T

adjusted sales growth following the violation. Using a difference-in-differences framework, we start by corroborating their results by analyzing sales growth of the rival firms when their peers in the same industry violate a financial covenant. One concern is whether the differences in sales growth patterns are truly due to creditor control threats and not simply due to fundamental differences

AN US

between the rivals and covenant violators. Violations may be preceded by deteriorating performance, suggesting they are not random events. If sales growth is serially correlated then previolation performance may be driving the decline in sales growth for violators relative to the rivals and not the creditor intervention. To see if this is the case, we need to compare violators’ sales

M

growth with that of nonviolator rival firms with similar pre-violation performance. To address this

ED

issue, we conduct difference-in-differences tests which control for both time-invariant firm-level effects that may be different between rival firms and the violators as well as the expected changes

PT

in sales growth following deterioration in violators’ performance.

CE

We match peer firms using industry-propensity score matching. We first estimate a probit regression of the probability (i.e. the “propensity score”) of a firm violating a covenant. We include

AC

a large set of observable characteristics consisting of our full set of control variables in equation (1), higher order covenant controls, the four quarter lags of covenant controls, calendar quarter-year, fiscal quarter-year and industry fixed effects. We then take all potential matches

14 See

Appendix A.2 and Table A.1 to see the descriptive statistics of variables for firms in new covenant violations and non-violator rival firms.

11

ACCEPTED MANUSCRIPT

that are in the same industry (three-digit SIC code) and that have pre-violation average sales growth within one-half of a standard deviation of the violator firm’s pre-violation average sales , , ,

growth and choose the one with the closest propensity score to that of the violator firm. 15 16 17 18

CR IP T

We also verify that the matched sample of peers satisfy the parallel trend assumption which requires a similar pattern of sales growth between the matched samples prior to the violation. Testing this assumption is particularly important for our study given the possibility that financially strong firms (rivals) may prey upon financially weak firms (violators) and cause

AN US

weak firms to violate covenants. As shown in Table 1, both rivals and violating firms have similar sales growth patterns in the quarters leading up to the violation. This is inconsistent with the notion that rival actions did cause the violation and suggests our sample satisfies the parallel trends assumption.

M

Table 1 reports the average sales growth for the violators and their rivals for the four

ED

quarters preceding the violation quarter and the four quarters following the violation. Both the rivals and violating firms have similar sales growth pattern before the violation. Pre-violation

PT

average sales growth is around 11% both for the rivals and violating firms and the difference is

standard deviation of pre-violation average sales growth is about 74%. As robustness, we identify the rival firm as the firm within the same 3-digit SIC code industry which has a pre-violation average sales growth within one-quarter of the standard deviation above and below the violator firm’s pre-violation average sales growth and choose the one with the closest propensity score to that of the violator firm. The results do not change, therefore not reported. 16 We also match rivals and violators on industry and size. In this case, to identify the matched rival firm, we use a non-violator rival firm with the same three-digit SIC code and has a pre-violation average sales growth within onehalf (one-quarter) of the standard deviation above and below the violator firm’s pre-violation average sales growth that has book value of assets closest to that of the violator firm. The results are robust to using these matching criteria. 17 We also use four-digit SIC code as the industry definition for robustness. The results are quantitatively similar, therefore not reported. 18 Table A.2 reports the mean and median statistics of firm characteristics for the violators and their rivals.

AC

15 The

CE

insignificantly different from zero. The rival firms experience an average change in sales growth

12

ACCEPTED MANUSCRIPT

of -1.8% from pre- to post-violation, while their covenant violating peers experience a 6.1% decrease, significant at the 1% level. Moreover, the difference in these two changes (i.e., difference-in-differences, DID) is 4.2%, significant at the 1% level. Given sales growth averages 11% for rivals, the sales growth difference-in-differences of 4.2% represents a 38.2% change in

CR IP T

sales growth for the rivals relative to their peer violating firms. 19

In panels B through E, we explore changes in sales growth for subsamples of firms. In Panel B, we divide the sample into small and big violators based on their industry market share.

AN US

Small (Big) violators are defined as new violators having the ratio of sales to total industry sales lower (greater) than the median ratio for the sample of new violators. Specifically, the ratio of sales to total industry sales is equal to the sum of the past four quarters’ sales (Sales t-1+Salest2+Salest-3+Salest-4)

of violators, divided by the sum of the past four quarters’ sales of that industry

M

(Salest-1+Salest-2+Salest-3+Salest-4). These results show that the DID is negative and significant for

ED

the big violator sample but insignificant for the small violator sample. One interpretation is that the rivals have little market share to gain from small violators.

PT

We next explore how different characteristics of violator firms’ products affect the

CE

changes in rival firm market share. If the violating firm is selling more unique products then it may be costly for buyers to change suppliers, making it difficult for the rival firms to take

AC

advantage of the violation. To test this argument, we define product uniqueness using three different measures: research and development (R&D) intensity, product durability, and product fluidity. Titman and Wessels (1988) argue that R&D intensity measures product uniqueness because firms selling products with close substitutes spend less on R&D due to easy duplication

19 These

results are consistent with Nini et al. (2012).

13

ACCEPTED MANUSCRIPT

of their technology. They also argue that successful R&D projects lead to new products that differ from those existing in the market. Moreover, firms spending more on R&D are more likely to produce more specialized products (Hertzel et al., 2008). If a violator firm’s non-missing annual R&D

CR IP T

spending (scaled by average assets) is more than the median ratio for the new violators in the year preceding the violation, that firm is considered high R&D intensive.20 Our second measure is product durability. Titman and Wessels (1988) argue that firms manufacturing machines and equipment require specialized servicing and spare parts. In this case, it may be more costly for the

AN US

customers to switch suppliers when their suppliers violate covenants. Durable goods industries are defined as those with an SIC industry code between 3400 and 4000 (firms producing machines and equipment). Our last measure of product uniqueness is product fluidity. Hoberg, Phillips and Prabhala (2014) use a text-based algorithm to define product fluidity as a

M

“measure of the competitive threats faced by a firm in its product market, which captures changes

ED

in rival firms' products relative to the firm’s products.” Product fluidity measures the speed with which firms introduce new products. If product fluidity is high, then the firm is considered to be

PT

producing less unique products. We use Hoberg et al. (2014)’s classifications to measure product

CE

fluidity. If a firm’s product fluidity is more (less) than the median level for the sample of new violators prior to the violation year, the firm is considered as a firm with High

AC

(Low) product fluidity.21

Firms are not required to report R&D expenses separately from sales and general administrative (SG&A) expenses if they are less than 10% of SG&A (SEC Regulation 5-03.2). If a firm’s annual R&D spending prior to the violation year is missing, the firm is also considered as a low R&D intensive firm, on the assumption that these firms have low R&D spending (Gentry and Shen, 2013). 21 We use violators’ R&D expenses and product fluidity since we are interested in the product uniqueness of rivals compared to violators. 20

14

ACCEPTED MANUSCRIPT

We create subsamples based on these three measures of product uniqueness and report the results in panels C, D, and E. In Panel C, we see the sales growth DID is -3.2%, significant at the 5% level, for the low R&D intensity sample. We also find the DID for the high

CR IP T

R&D intensity sample is negative and marginally significant. Thus R&D intensity provides weak support for the notion that product similarity leads to greater rival market share gains. In Panel D, we divide the sample based on nondurable and durable goods industries. We see that rival firms’ market share gains are only significant for the nondurable subsample where the DID is -

AN US

4.6%, significant at the 5% level. Lastly, Panel E stratifies the sample based on product fluidity. We see insignificant rival firm gains for the low fluidity group; however, rival firms gain significant market share in the high fluidity group where we find the DID is -6.8%, significant at the 5% level. Overall the results in Panels B through E suggest the market share gains of rival

M

firms are more pronounced when the violators have a greater share to lose and when their

ED

products less unique. Next we explore these relations in a multivariate setting. 4.2. Multivariate Results

PT

We estimate various outcomes for the rivals of firms that violate a covenant using the

CE

following regression:

outcome variable = β1* Industry market share of violators + Θ1*Covenant Controlsi,t (1)

AC

+ Θ2*(Covenant Controlsi,t-4) + Calendar Quarter-Yeari,t + Fiscal Quarter-Yeari,t + εi,t

Industry market share of violators measures the aggregate market share of all violators in a given industry for a given quarter. If gains by rivals come at the expense of violators then the magnitude of rival gains should relate to the size of the violators’ market share. We compute Industry market share of violators as annualized total sales of new violators within the three-digit SIC code 15

ACCEPTED MANUSCRIPT

industry prior to the violation quarter divided by the annualized industry sales prior to the violation quarter.22 More specifically, it is equal to the sum of past four quarters’ sales (Sales t1+Salest-2+Salest-3+Salest-4)

of peer violators, divided by the sum of past four quarters’ sales

CR IP T

(Salest-1+Salest-2+Salest-3+Salest-4) of the industry, where subscript t refers to new covenant violation quarter. In the regressions, we limit the sample to rival firms, i.e, firms that are not in a covenant violation.23

We include calendar quarter-year and fiscal quarter-year indicator variables because

AN US

seasonal patterns may exist and financial covenant variables are more common in 10-K filings than 10-Q filings. Following Nini et al. (2012), we include the most common ratios used in debt agreements in the analysis as Covenant Controls. These ratios include: operating cash flow to average assets, leverage (debt-to-assets), interest expense to average assets, net worth to assets,

M

and the current ratio (current assets/current liabilities). We also include market-to-book ratio. In

ED

some specifications, we include four-quarter lagged value of these variables. Since we have overlapping observations that induce a mechanical serial correlation in the dependent variable, ,

PT

we cluster our standard errors by firm.24 25

CE

Table 2 presents the results for nonviolating rival firms. The dependent variable is Market Share, annualized sales growth minus the industry average, where the average excludes the

AC

firm itself, or equivalently, is a proxy for market share growth (following Fresard 2010). Columns (1) to (3) show the regression results using the set of variables in equation (1) as well as

22 We

also use four-digit SIC code as the industry definition for robustness. The results are quantitatively similar. To be included in the sample, we require firms to have available covenant control variable items. We have 76,624 firm-quarter observations in our sample. 24 Please see Peterson (2009) for detailed information on clustered standard errors. 25 To address clustering of violations within industries, we cluster standard errors by industry and firm as a robustness check. The results are similar. 23

16

ACCEPTED MANUSCRIPT

the level and first differences of ln(assets) and the level and first differences of net fixed assets scaled by total assets. We restrict our sample to industry-quarters which have at least one firm in a new covenant violation.26 We find the rival firm’s market share increases in the market share of violators in their

CR IP T

industry. In each of the first three specifications we find a positive and significant coefficient on Industry market share of violators. In Column (3) of Table 2, the coefficient of 0.753 on Industry market share of violators indicates that the rival firms’ market share grows by 0.753% when the

AN US

combined industry market share of violators is 1%.

We also analyze the change in rival firms’ sales growth when the industry market share of violators is above (below) the median for the sample of rival firms. Columns (4) to (6) of Table 2 show that rival firms’ market share economically and significantly increases following

M

the violation if the industry market share of the violators is above the median. If the market

ED

share of the violators is small, however, the increase in market share of rival firms is not statistically significant at a reasonable level. These results are consistent with the notion that

PT

the rivals see their peers’ covenant violations as providing an opportunity to increase their

CE

competitive position in the product market.27 In untabulated robustness analysis, we conduct “quasi-discontinuity” regressions

AC

following Roberts and Sufi (2009) and Nini, Roberts and Sufi (2012) where we include linear and

26 The

results are robust if we include industry-quarters without any new covenant violations.

27 If

the whole industry experiences a decline in sales due to a demand or technology shock, all firms in the industry would be affected. In this case, financially weak firms (covenant violators) may have lower sales compared to their rivals. To address this, we include variables such as operating cash flow/average assets, leverage ratio, interest expense/average assets, and add industry-fixed effets to the regression. This way, we hope to control for financial differences between rivals and violating firms as well as industry-specific shocks. When we sub-sample the data, we show rival firms’ market share economically and significantly increases only when the industry market share of the

17

ACCEPTED MANUSCRIPT

higher order transformations of the covenant control variables. We continue to find that rivals market share increases in the market share of violators. While prior work demonstrates covenant violations are common place and frequently occur well outside the realm of financial

CR IP T

distress, our results could be driven by distressed violators. To see if this is the case we estimate Altman’s z-score for all violators and reconstruct the violator market-share variable omitting those violators defined as financially distressed (z-score < 1.81). Regressions of rival market share on based on this sample result in a significant coefficient on Industry market share of

AN US

violators ranging of 0.586, significant at the 1% level. Taken together, these results suggest that the rival reactions are not driven by financially distressed violators.

As an additional robustness check, we conduct a regression discontinuity regression analysis following Chava and Roberts (2008). We identify the firms which have the financial

M

covenant information (net worth, tangible net worth and current ratio) on Loan Pricing

ED

Corporation’s (LPC) DealScan database. We label “barely violators” as those firm-quarter observations where the firm’s financial ratio falls below the threshold, but is no further than 20%

PT

from the threshold. We label “barely nonviolators” as those firm-quarter observations when the

CE

firm’s financial ratio is within 20% of the threshold but above the threshold. We then compute the industry market share of “barely violators” and “barely nonviolators” by summing the firm-quarter

AC

observations by industry. The untabulated results show that a firm’s market share increases in the market share of “barely violator” peer firms (the coefficient is 0.55, significant at 5% level) but does not increase in that for “barely nonviolators”. Moreover, the coefficients on

violators is above the median. We also address this by using subsamples based on industry returns and violator idiosyncratic returns. We discuss this in detail in Section 4.6.

18

ACCEPTED MANUSCRIPT

the two variables differ significantly at the 10% level. This result suggests that rivals react to the violation, rather than the changes in the financial condition of their peers, before the violation. Last, we explore whether our results are affected when the violator and rival firms share a

CR IP T

common lender. If the same bank is lending both to the violator and rival firms, we may observe a different effect of the violation on the product market competition. If the rival firm and its peer violator have the same lead lender for any loans originated during the prior 12 quarters, we define these two firms as having a common lender. We then sum the market share of the industry

AN US

violators for those violators that share a common lender with the rival nonviolator, and we sum the market share of violators without a common lender. We then include these two variables in our main market-share regressions, in place of Industry market share of violators. The untabulated results show that the coefficient of Industry market share of violators with common lender is 0.84,

M

significant at 1% level. Similarly, the coefficient of Industry market share of violators without a

ED

common lender is 0.75, significant at 1%. We conduct tests on whether the coefficients differ and

PT

find no significant difference between the coefficients. 28

CE

4.3. Changes in market share and product similarity The previous section shows that the rival firms’ market share significantly increases in

AC

proportion to the industry market share of violators following a financial covenant violation. In this section, we explore how product uniqueness influences the relation between changes in rival firms’ market share and the industry market share of violators. The results are presented in Table 3. In the first three columns, the Product Uniqueness Dummy is one for durable goods industries

28 The

results are similar for advertising expense.

19

ACCEPTED MANUSCRIPT

and zero otherwise. In columns 4-6, Product Uniqueness Dummy is defined using R&D intensity. If a firm’s annual R&D spending is more than the median level for the full sample in the year preceding the violation, that firm is considered R&D intensive. If more than half of the violator firms in the industry are R&D intensive, product uniqueness dummy is defined as one, zero

CR IP T

otherwise.29 In columns 7-8, Product Uniqueness Dummy is defined using product fluidity. Using Hoberg et al. (2014)’s classifications to measure product fluidity, we categorize a firm’s products as having low fluidity when it is below the median level for the full sample in the year

AN US

preceding the violation. If more than half of the violator firms in the industry have low product fluidity then the Product Uniqueness Dummy equals one.

We include both the Product Uniqueness Dummy and the interaction of Product Uniqueness Dummy with Industry market share of violators in the regressions. We conjecture

M

that the interaction term will have a negative coefficient, consistent with the notion that rivals

ED

will not be able to take as much market share from violators when violators’ products are more unique. This is indeed what we find. In all nine specifications, we find negative coefficients on

PT

the interaction term that are significant at the 1% level. This result supports our conjecture that

CE

if peer violators are producing more unique and customer specific products, there is less room for rival firms to take advantage of the violation. We next explore how rival firms’ advertising

AC

expense relates to violators’ industry market share. If rivals are indeed actively attempting to capture share then we would expect to see advertising expense to rise as a function of the violators’ share.

29 The

results are qualitatively similar when we classify firms into high and low R&D based on industry adjusted measures.

20

ACCEPTED MANUSCRIPT

4.4. Advertising expense and industry market share of violators In Table 4, we examine the growth in the annual advertising expense, adjusted annual advertising expense (advertising expense/lagged sales), and the growth in the adjusted annual

CR IP T

advertising expense as a function of industry market share of violators.30 Our specifications are similar to those in Table 2 but with our advertising expense measures as the dependent variables and with annual values of covenant controls. In the first two specifications where the dependent variable is change in natural log of advertising expense, we find the coefficient on Industry market

AN US

share of violators is 0.575 and 0.522, both significant at the 1% level. The later coefficient suggests that for each 1% of market share held by industry violators’, rival firms grow their advertising budget by 5.22%. When we explore the adjusted advertising expense and the growth in adjusted advertising expense, the results are similar. The coefficient of Industry market share of violators is

M

positive and statistically significant in all specifications.31 We also find similar results for advertising

ED

expense when we estimate “quasi-discontinuity” regressions following Roberts and Sufi (2009) and Nini, Roberts and Sufi (2012). Overall these findings suggest that increases in advertising expense of

PT

the rivals may explain the increase in sales growth with respect to the peer violators, and may be

CE

indicative of strategic actions of the rivals. These results suggest that covenants play an important

AC

role in industry competitiveness and product-market outcomes.

4.5. Operating performance and industry market share of violators The results above suggest that creditor control threats over borrowing firms following

violations have positive effects on rival firms. To provide evidence on the valuation consequences

21

ACCEPTED MANUSCRIPT

of the increased market share for the rivals following covenant violations, we examine rival firms’ performance following violations. We use three different variables as dependent variables to measure operating performance

CR IP T

in equation (1). We explore, change in operating cash flow scaled by average assets, where operating cash flow is defined as operating income before depreciation and amortization, change in return on assets (ROA), where ROA is defined as operating income before depreciation minus depreciation and amortization divided by average assets, and change in return on equity (ROE),

AN US

where ROE is defined as net income divided by average common equity.

Table 5 reports the results. We see in all nine specifications, using three measures of profitability, the relation between rival firm profits and industry market share of violators is positive and significant. In other words, not only do the rival firms experience significant

M

improvements in these profitability measures relative to the violators in their industry, they

ED

also experience these increases in proportion to the violators’ industry market share. 32

PT

4.6. Industry returns, violator idiosyncratic returns, and rival outcomes

CE

One concern may be that industry shocks drive both violations and actions by rivals. If an

AC

industry shock hits the weakest firms the hardest then these firms will be more likely to violate.

30 To

explore the annual change in the advertising expense, we restrict the sample to covenant violations that are reported in the fourth quarter of each fiscal year. We use the fourth quarter since financial covenant violation announcements are more common in 10-K filings than in 10-Q filings and advertising expense is reported annually. 31 We explore the relation between advertising expense and product uniqueness; however, there are countervailing factors that drive the predictions. On the one hand, we would expect advertising to increase more when products are less unique given the greater product substitutability – leading to a negative relation between advertising and product uniqueness. On the other hand, firms may need to spend more on advertising to persuade rival customers when products are more unique – leading to a positive relation. In general we find the relation is statistically insignificant. 32 We find similar result using quasi-discontinuity regressions.

22

ACCEPTED MANUSCRIPT

The less weak firms will be less likely to violate a covenant, and the relative change in sales growth could simply reflect the degree to which the shock affected the cross section of firms in the industry. To minimize the impact of such channel we re-estimate the regressions using

CR IP T

subsamples where industry shocks are less likely to be a driving force using two approaches. Following Leary and Roberts (2014), we decompose violators’ stock returns into systematic, industry, and idiosyncratic components. We regress the total return of the firm on the excess market return (CRSP value weighted return minus the risk-free rate) and on excess

AN US

industry return based on equally weighted three-digit SIC code industry portfolio (excluding the firm’s return). We estimate parameters for year t using monthly returns, on a rolling annual basis, over a five year window (year t, t-1, t-2, t-3, t-4), requiring at least 24 months of nonmissing returns. We use parameters estimated at year t and realized monthly returns at

M

year t+1 to compute the idiosyncratic component for each month in year t+1. After obtaining

ED

the idiosyncratic component, we merge the sample of idiosyncratic returns with the covenant violation sample on the last month of the violation quarter.

PT

First, given negative industry shocks likely result in negative industry returns, we limit the

CE

sample to industries that experience non-negative returns in the violation quarter. We compute the excess industry return, omitting violator firms, as the equal weighted average of the returns to the

AC

industry (based on three-digit SIC code) net of the risk-free rate. We then estimate the regressions from Table 2, 4, and 5 for this subsample. The results are contained in the first column of Table 6. Panel A reports the results for the regressions of rival firm sales growth (changes in market share), the coefficient on Industry market share of violators is 0.787 and significant at the 1% level. Panel B (Panel C) reports the results where advertising expense (operating 23

ACCEPTED MANUSCRIPT

performance) is the left hand side variable. In all three panels, we see results that are very similar to our full sample results. The next column of Table 6 reports the results using the complement sample (industries with non-positive industry average excess return in the violation quarter). For

CR IP T

this sample where industry returns are negative, we still find a positive and significant coefficient in Panel A, but none of the coefficients in panels B and C are statistically significant. Taken together these results suggest that industry shocks are not the likely cause of our findings.

In our second set of tests, we focus the sample on violators where the violator’s shock is

AN US

likely to be firm specific. Leary and Roberts (2014) argue that a firm’s response to the peer’s idiosyncratic returns indicates strategic action. We limit the sample based on whether violators experience negative idiosyncratic returns in the last month of the violation quarter using the following procedure. We compute the fraction of Industry market share of violators that comes

M

from violators with negative idiosyncratic returns only including industry quarters which have 33

The results from regressions

ED

more than 67% of violators with negative idiosyncratic returns.

based on this sample are in the next two columns of Table 6. In Panels A, B, and C, we again find

PT

results similar to those of our full sample.

CE

While the results focusing on violators with low idiosyncratic returns support a strategic channel, low idiosyncratic returns could coincide with low industry returns. To rule this out, we

AC

further limit the sample to observations where 67% or more of Industry market share of violators comes from violators and where the excess industry return is nonnegative. This sample helps ensure that not only the likely cause of the violation is firm specific, but also that it occurs during

33 As

robustness, we (1) require this ratio to be 75% or (2) relax this assumption and only limit the sample to only some violators in the quarter with low idiosyncratic return. Both tests give similar results, therefore not reported.

24

ACCEPTED MANUSCRIPT

a time when the industry is not experiencing any negative shock. These results are in the last two columns of Table 6, where we again see results comparable to that for the full sample. Overall, we conclude from this section that our results on rival actions outcomes are not likely

CR IP T

to be driven by common omitted factors such as industry shocks.

5. Conclusion

There is a large body of work which examines how firms’ capital structure affects

AN US

product-market behavior. While previous studies focus on the level of leverage, there are other dimensions of leverage that may also affect product-market outcomes. In this study, we investigate how creditor control threats impact product-market competition. More specifically, we explore whether bank loan covenant violations result in commensurate changes in product-

M

market outcomes.

ED

Recent papers show that following bank loan covenant violations, creditors increase their control over violating firms in ways that significantly affect firms’ corporate policies. These lender-

PT

induced changes in violating firms may improve their investment and financing policies and

CE

strengthen their position in the product market. Alternatively, covenant violations may reduce firms’ flexibility, providing competitive opportunities for their rivals. Regardless of whether

AC

covenant violations benefit shareholders, the influence of covenants on product-market actions may be an important channel that affects overall industry competitiveness. The rival firms

25

ACCEPTED MANUSCRIPT

may consider their peers’ exposure to control threats when developing product-market strategies. Our paper is the first to thoroughly examine the product-market outcomes following

CR IP T

financial covenant violations. We show that the rival firms gain market share when a firm in their industry violates a financial covenant, this gain increases with the industry market share of violators. The rival firms gain less market share if peer violators sell more unique goods. We also show that rival firms increase advertising spending significantly after covenant violations.

AN US

ROA, ROE, and operating cash flow significantly increase for rival firms once their peers violate a financial covenant. These findings indicate that the rivals follow more aggressive strategies to improve their positions in the product market when another firm in their industry violates a covenant.

M

We find evidence that the rival firm actions (i.e., changes in market share and

ED

advertising spending) are indicative of strategic actions. Namely we see their effects increase with the potential size of the competitive opportunity – i.e., the size of the violating firms’

PT

market share and the nature of the product violating firms sell. Taken together, results suggest

CE

that creditor control threats can indeed have dramatic effects on product-market outcomes. Rival firms may be considering their peers’ covenant violations when making competitive

AC

decisions in the product market.

26

ACCEPTED MANUSCRIPT

References Aghion, P., Bolton, P., 1992. An incomplete contracts approach to financial contracting. Rev. Econ. Stud. 59, 473–494. Begley, T., 2015. The real costs of corporate credit ratings. SSRN working paper 2404290.

CR IP T

Beneish, M., Press, E., 1993. Costs of technical violation of accounting– based debt covenants. Account. Rev. 68, 233–57. Beneish, M., Press, E., 1995. The resolution of technical default. Account. Rev. 70, 337–353.

Bolton, P., Scharfstein, D., 1990. A theory of predation based on agency problems in financial contracting. Am. Econ. Rev. 80, 93–106.

AN US

Brander, J., Lewis, T., 1986. Oligopoly and Financial Structure: The Limited Liability Effect. Am. Econ. Rev. 76, 956–970. Campello, M., 2003. Capital structure and product market interactions: Evidence from business cycles. J. Financ. Econ. 68, 353–378. Chava, S., Nanda, V., Xiao, S., 2016. Lending to innovative firms. Working paper, Georgia Institute of Technology.

M

Chava, S., Roberts, M., 2008. How does financing impact investment? The role of debt covenants. J. Finance 63, 2085–2121.

ED

Chen, K., Wei, J., 1993. Creditors' decisions to waive violations of accounting–based debt covenants, Account. Rev. 68, 218–232.

PT

Chevalier, J.A., 1995a. Do LBO supermarkets charge more?: An empirical analysis of the effects of LBOs on supermarket pricing. J. Finance 50, 1095−1112.

CE

Chevalier, J.A., 1995b. Capital structure and product–market competition: empirical evidence from the supermarket industry. Am. Econ. Rev. 85, 415−35.

AC

Chevalier, J.A., Scharfstein, D.S., 1996. Capital–market imperfections and countercyclical markups: theory and evidence. Am. Econ. Rev. 86, 703−725. Dichev, I., Skinner, D., 2002. Large sample evidence on the debt covenant hypothesis. J. Account. Res. 40, 1091–1123. Eberhart, A.C., Altman, E.I., Aggarwal, R., 1999. The equity performance of firms emerging from bankruptcy. J. Finance 54, 1855–1868. Ersahin, N., Irani, R., and Le, H., 2015. Creditor control rights and resource allocation within firms. Working paper. Esmer, B., 2016. Creditor control rights and managerial risk shifting. Working paper. Indiana University.

27

ACCEPTED MANUSCRIPT

Falato, A., Liang, N., 2016. Do creditor rights increase employment risk? Evidence from loan covenants. J. of Finance 71, 2545–2590. Franz D., HassabElnaby, H. R., Lobo. G.J., 2014, Impact of proximity to debt covenant violation on earnings management. Rev. of Account. Stud. 19,473–505. Fresard, L., 2010. Financial strength and product market behavior: The real effects of corporate cash holdings. J. Finance 65, 1097–1122.

CR IP T

Fudenberg, D., Tirole, J., 1986. A 'signal–jamming' theory of predation. Rand J. Econ. 17, 366–376.

Garleanu, N., Zwiebel, J., 2009. Design and renegotiation of debt covenants. Rev. Financ. Stud. 22, 749– 781. Gao, Y., Khan, M., Tan. L., 2017. Further evidence on consequences of debt covenant violations. Con. Account. Research. Forthcoming.

AN US

Gentry, R., Shen, W., 2013. The impacts of performance relative to analyst forecasts and analyst coverage on firm R&D intensity. Strategic Manage. J. 34, 121–130.

Gopalakrishnan, V., Parkash, M., 1995. Borrower and lender perceptions of accounting information in corporate lending agreements. Account. Horizons 9, 13–26. Gorton, G., Kahn, J., 2000. The design of bank loan contracts, Rev. Financ. Stud. 13, 331–364.

M

Grossman, S., Hart, O., 1986. The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy 94: 691–719.

ED

Gu, Y., Mao, C. X., Tian, X., 2013. Creditor interventions and firm innovation: Evidence from debt covenant violations. Working paper.

PT

Hart, O., Moore, J., 1995. Debt and seniority: An analysis of the role of hard claims in constraining management. Am. Econ. Rev. 85, 567–585.

CE

Hertzel, M., Li, Z., Officer, M., and Rodgers, K., 2008. Interfirm linkages and the wealth effects of financial distress along the supply chain. J. Financ. Econ. 87, 374–387.

AC

Hoberg, G., Phillips, G., and Prabhala, N., 2014. Product market threats, payouts, and financial flexibility. J. Finance 69, 293–324. Hotchkiss, E.S., 1995. The post–emergence performance of firms emerging from Chapter 11. J. Finance 50, 3–21. Khanna, N., Tice, S., 2005. Pricing, exit, and location decisions of firms: Evidence on the role of debt and operating efficiency. J. Financ. Econ. 75, 397–427. Khanna, N., Tice, S., 2000. Strategic responses of incumbents to new entry: the effect of ownership structure, capital structure, and focus. Rev. Financ. Stud. 13, 749–779.

28

ACCEPTED MANUSCRIPT

Kovenock D., Phillips, G.M., 1997. Capital structure and product market behavior: An examination of plant exit and investment decisions. Rev. Financ. Stud. 10, 767–803. Lang, L., Stulz, R., 1992. Contagion and competitive intra–industry effects of bankruptcy announcements. J. Financ. Econ. 8, 45–60. Leary, M., Roberts, M., 2014. Do peer firms affect corporate financial policy? J. Finance 69, 139–178.

CR IP T

Maksimovic, V., 1988. Capital structure in repeated oligopolies. Rand J. Econ. 19, 389–407. Nini, G., Smith, D., Sufi, A., 2009a. Creditor control rights and firm investment policy. J. Financ. Econ. 92, 400–420. Nini, G., Smith, D., Sufi, A., 2012. Creditor control rights, corporate governance, and firm value. Rev. Financ. Stud. 25, 1713–1761.

AN US

Nordlund, J., 2016. Strategic responses to peer–firm financial flexibility: Evidence from indirect treatment effects. Working paper.

Opler, T., Titman, S., 1994. Financial distress and corporate performance. J. Finance 49, 1015–1040. Petersen, M., 2009. Estimating standard errors in finance panel data sets: Comparing approaches. Rev. Financ. Stud. 22, 435–480.

M

Phillips, G., 1995. Increased debt and product market competition: An empiric analysis. J. Financ. Econ. 37, 189–238.

ED

Rajan, R.G., Winton, A., 1995. Covenants and collateral as incentives to monitor. J. of Finance 50, 1113– 1146.

PT

Roberts, M., Sufi, A., 2009. Control rights and capital structure: An empirical investigation. J. Finance 64, 1657–1695.

CE

Rotemberg, J., Scharfstein, D., 1990. Shareholder–value maximization and product market competition. Rev. Financ. Stud. 3, 367–391. Sweeney, A., 1994. Debt covenant violations and managers’ accounting responses. J. Account. Econ. 17, 281–308.

AC

Tirole, J., 1999. Incomplete contracts: where do we stand? Econometrica 67: 741–781. Titman, S., 1984. The effect of capital structure on a firm's liquidation decision. J. Financ. Econ. 13, 137– 151. Titman, S., Wessels, R., 1988. The determinants of capital structure choice. J. Finance 43, 1–19. Zhang, G., 2010. Emerging from Chapter 11 bankruptcy: is it good news or bad news for industry competitors? Financ. Manage. 39, 1719–1742.

29

ACCEPTED MANUSCRIPT

Zhang, Z., 2016. Bank interventions and trade credit: evidence from debt covenant violations. Working paper.

AC

CE

PT

ED

M

AN US

CR IP T

Zingales, L., 1998. Survival of the fittest or fattest? Exit and financing in the trucking industry. J. Finance 53, 905–938.

30

ACCEPTED MANUSCRIPT

A.1 Variable definitions All cash flow statement variables are first disaggregated into quarterly flows. Sales growth = (Salest – Sales t–4)/Sales t–4 where Sales = saleq

CR IP T

Industry market share of violators = (Salest–1+Salest–2+Salest–3+Salest–4) of violators / (Salest– 1+Salest–2+Salest–3 +Salest–4) of the three–digit SIC code industry. Change in Market Share= ((Salest+4–Salest)/Salest) adjusted by the three–digit SIC code industry average sales growth (excluding the firm itself) Average assets = (Total assets + lagged total assets) / 2 where total assets = atq

AN US

Market–to–book ratio = Market value / total assets where market value = Market value of equity – book value of equity + total assets

Market value of equity = Price close quarterly * common shares outstanding (prccq*cshoq) Book value of equity = Total assets – total liabilities (ltq) + deferred taxes and investment tax credit (txditcq)

M

Total debt = Debt in current liabilities (dlcq) + total long–term debt (dlttq) Leverage ratio = Total debt / total assets

ED

Net worth / assets = Total shareholders' equity (seqq) / total assets Current ratio = Current assets (actq) / current liabilities (lctq)

PT

Interest expense / average assets = Interest and related expense (xintq) / average assets If interest expense is missing, we calculate the interest expense using this equation:

AC

CE

Operating income before depreciation (oibdpq) – depreciation and amortization (dpq) + non– operating income (nopiq) – pretax income (piq) +impairment of goodwill pretax (gdwlipq) +settlement pretax (setpq) + writedowns pretx (wdpq) +other special items pretax (spiopq) + restructuring cost pretax (rcpq)+ gain/loss pretax (glpq) If non–operating income (nopiq), pretax income (piq), impairment of goodwill pretax (gdwlipq), settlement pretax (setpq), writedowns pretx (wdpq), other special items pretax (spiopq), restructuring cost pretax (rcpq), gain / loss pretax (glpq) are missing, we set them equal to zero. Operating cash flow / average assets = Operating income before depreciation quarterly (oibdpq) / average assets 31

ACCEPTED MANUSCRIPT

Net fixed assets / total assets = Plant, property and equipment (Ppentq) / total assets (atq) Advertising Expense / lagged sales= Annual advertising Expense (xad) / annual sales (sale) Return on assets (ROA) = Operating income after depreciation quarterly (oibdpq) / average assets where operating income after depreciation= Operating income before depreciation quarterly (oibdpq)–depreciation and amortization (dpq)

AC

CE

PT

ED

M

AN US

CR IP T

Return on equity (ROE) = Net income (niq) / average common equity where average common equity= (Common equity (ceqq) + lagged common equity) / 2

32

ACCEPTED MANUSCRIPT

A.2 Descriptive Statistics

Since we use the data of Amir Sufi and follow the data construction of Nini et al. (2012), we refer the reader to Nini et al. (2012) for the data characteristics. As Nini et al. (2012) show in

CR IP T

Table 2 of their paper, violations are common events, 40 percent of firms violate a financial covenant in 1997–2008 period, although small firms are more likely to violate a covenant, violations are also common among large firms (25 percent of firms with greater than $5 billion in assets are in violation in the sample). We find that violations are more common among

AN US

manufacturing firms. As Nini et al. (2012) show in Figure 1 of their paper, violations are more likely to occur in bad times (e.g. 2001 crises), the incidence of violations declines in the later part of the sample.

M

Table A.1 displays the distribution of performance measures for new violator firms and the non–violating rival firms. A new covenant violation is a financial covenant violation for a firm that

ED

has not experienced a financial covenant violation in the previous four quarters. We confirm Table 3

PT

of Nini et al. (2012) showing that although violator firms have high leverage and interest expense scaled by average assets compared to rivals, they are not highly leveraged. The median violator firm

CE

has debt ratio of 0.34, compared to 0.21 of the median non–violator rival firm. Operating cash flow/average assets are lower for the median violator firm compared to the median rival firm;

AC

however, violators are not experiencing deep liquidity problems. The median violator has a reasonably high market–to–book ratio, 1.5, relative to the median non–violator rival firm’s market– to–book ratio of 2.3. Operating performance of the median violator is lower than the median rival firm. As Nini et al. (2012) point out “Overall, financial covenant violations

33

ACCEPTED MANUSCRIPT

appear to serve more as an indicator of a change in performance, rather than as an indicator of a low level of performance” (p. 1730). Table A.1 also shows that the median new violator firm’s change in market share in the year preceding the violation and advertising expense/lagged sales are similar to that of the

AC

CE

PT

ED

M

AN US

CR IP T

median non–violator rival firm.

34

CR IP T

ACCEPTED MANUSCRIPT

Table 1 Differences-in-differences (DID) tests of sales growth of rival firms and covenant violators

Panel A

AC

CE

PT

Pre-violation average sales growth (Quarters -4,-3,-2, -1) Post-violation average sales growth (Quarters +1,+2,+3,+4) Post Violation - Pre Violation

ED

M

AN US

This table presents changes in sales growth of rival firms and violators around new financial covenant violations. 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. Sales growth is defined as sales growth with respect to same quarter last year i.e. (Salest-Salest-4)/Salest-4, where subscript t refers to the new covenant violation quarter. A rival firm is identified as a non-violator firm with the same three-digit SIC code, a pre-violation average sales growth within one-half of the standard deviation above and below the violator firm’s pre-violation average sales growth, and a propensity score of violating a covenant closest to that of the violator firm. In Panel B, Small (Big) violators are defined as new violators having the ratio of sales to total industry sales lower (greater) than the median ratio for the sample of new violators. Specifically, the ratio of sales to total industry sales is equal to the sum of the past four quarters’ sales (Salest-1+Salest-2+Salest-3+Salest-4) of violators, divided by the sum of the past four quarters’ sales of that industry (Salest-1+Salest-2+Salest-3+Salest-4). From Panel C to Panel E, product uniqueness is defined by three measures: Research and Development (R&D) Intensity, Non-durable/Durable Goods Industry and Product Fluidity. If a violator firm’s non-missing annual R&D spending (scaled by average assets) is less than the median ratio for the new violators in the year preceding the violation, the firm is considered as a Low (High) R&D intensity firm. If a firm’s annual R&D spending prior to the violation year is missing, the firm is also considered as a Low R&D intensity firm. If a firm’s SIC industry code is between 3400 and 4000 (firms producing machines and equipment), then it is defined as a durable goods firm; otherwise it is defined as a non-durable goods firm. If a firm’s product fluidity is more (less) than the median level for the sample of new violators prior to the violation year, this firm is considered as a High (Low) product fluidity firm. In each panel, first rows show the average sales growth of firms in the four quarters prior to the violation. Second rows show the average sales growth of firms in the four quarters following the violation. Last rows show the difference between average sales growth for the post-violation and pre-violation periods both for the rivals and their peer violator firms. The difference-in-differences estimator is the difference between the average difference of the rivals and peer violator firms. ***, ** and * denote 1%, 5% and 10 % levels of significance, respectively.

N 1896

New Violators 0.116***

Rival Firms 0.110***

1896

0.055***

0.091***

1896

-0.061***

-0.018

DID 0.006 -0.036** -0.042***

35

Panel B

N 933 933 933

0.104*** -0.047*

Panel C

N 1364

0.045***

1364

-0.054***

1299

CE AC

1299

0.078***

-0.021*

Non-Durable Goods New Rival Violators Firms 0.128*** 0.128***

ED N 1299

PT

Pre-violation average sales growth (Quarters -4,-3,-2, -1) Post-violation average sales growth (Quarters +1,+2,+3,+4) Post Violation - Pre Violation

Low R&D intensity New Rival Violators Firms 0.099*** 0.100***

1364

Panel D

-0.010

M

Pre-violation average sales growth (Quarters -4,-3,-2, -1) Post-violation average sales growth (Quarters +1,+2,+3,+4) Post Violation - Pre Violation

0.128***

0.064***

-0.064***

DID 0.012** -0.024

N 963 963

AN US

Pre-violation average sales growth (Quarters -4,-3,-2, -1) Post-violation average sales growth (Quarters +1,+2,+3,+4) Post Violation - Pre Violation

Small Violators New Rival Violators Firms 0.150*** 0.138***

CR IP T

ACCEPTED MANUSCRIPT

0.110*** -0.018

-0.036

963

DID -0.001

N 532

-0.033**

532

-0.032**

532

DID 0.000

N 597

Big Violators New Rival Violators Firms 0.083*** 0.082*** 0.009

-0.074***

0.056***

-0.026**

High R&D intensity New Rival Violators Firms 0.160*** 0.136*** 0.081*** -0.079**

0.125*** -0.010

Durable Goods New Rival Violators Firms 0.091*** 0.070***

-0.046**

597

0.037**

-0.046**

597

-0.054**

0.051*** -0.018

DID 0.001 -0.048*** -0.048***

DID 0.025*** -0.044 -0.069*

DID 0.021*** -0.014 -0.035

36

Panel E

932

0.027***

932

-0.042***

0.051***

DID 0.000 -0.023

N 860 860

AN US

N 932

-0.019

-0.023

860

High Fluidity New Rival Violators Firms 0.169*** 0.156*** 0.084***

-0.085***

0.139***

-0.017

DID 0.013** -0.055* -0.068**

AC

CE

PT

ED

M

Pre-violation average sales growth (Quarters -4,-3,-2, -1) Post-violation average sales growth (Quarters +1,+2,+3,+4) Post Violation - Pre Violation

Low Fluidity New Rival Violators Firms 0.069*** 0.070***

CR IP T

ACCEPTED MANUSCRIPT

37

Table 2 Rival firm sales growth and the industry market share of covenant violators

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN US

This table presents changes in sales growth of rival firms when firms in their industry violate a financial covenant. The dependent variable is Change in Market Share defined as the sales growth in the year following the violation ((Salest+4-Salest)/Salest) adjusted by the three-digit SIC code industry average sales growth, where the average excludes the firm itself. Industry market share of violators is the total sales prior to the new covenant violation quarter of violators within the industry (three-digit SIC code) divided by the total corresponding industry sales. Specifically, it is equal to the sum of the past four quarters’ sales (Salest-1+Salest-2+Salest-3 +Salest-4) of violators, divided by the sum of the past four quarters’ sales of that industry (Salest-1+Salest-2+Salest-3+Salest-4), where subscript t refers to the new covenant violation quarter. We limit the sample to rival firms, i.e, firms that are not in a covenant violation. We include industry-quarters which have at least one firm in a new covenant violation. Covenant controls include the following: operating cash flow scaled by average assets, leverage ratio, interest expense scaled by average assets, net worth scaled by assets, current ratio and market-to-book ratio. All specifications include calendar quarter-year fixed effects, fiscal quarter-year fixed effects, the level and first difference of Ln(assets), and the level and first difference of the ratio of net fixed assets to total assets. The first difference is computed as the value in quarter t minus the value in quarter t-4. In some specifications, we include four-quarter lag of covenant controls (i.e. Covenant Controlt-4). Columns (4) to (6) show the results for the sample where Industry market share of violators is above the sample median for rival firms. Columns (7) to (9) show the results for the sample where Industry market share of violators is below the sample median for rival firms. Definitions of the control variables are described in the appendix. Standard errors are clustered by firm. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

38

Whole Sample 1 2 3 0.769*** 0.754*** 0.753*** (10.31) (10.09) (10.10) -2.569*** -2.605*** -2.608*** (-15.54) (-13.59) (-13.56) 0.213*** 0.163** 0.164** (5.35) (2.45) (2.45) -0.849 -0.819 -0.818 (-1.02) (-0.98) (-0.98) 0.130*** 0.206*** 0.206*** (4.06) (3.96) (3.98) 0.017*** 0.010** 0.010** (5.15) (2.17) (2.14) 0.025*** 0.036*** 0.036*** (5.24) (7.08) (7.08) 0.001 (0.09)

Industry market share of violators Operating cash flow / average assets

Industry market share of violators is above the median 4 5 6 0.588*** 0.573*** 0.559*** (7.78) (7.56) (7.39) -2.018*** -2.054*** -2.105*** (-10.47) (-8.87) (-8.88) 0.212*** 0.191*** 0.196*** (5.21) (2.94) (3.03) 0.299 0.382 0.369 (0.23) (0.29) (0.28) 0.140*** 0.204*** 0.210*** (3.89) (3.63) (3.72) 0.018*** 0.019*** 0.020*** (4.45) (2.95) (3.06) 0.024*** 0.035*** 0.035*** (4.30) (5.21) (5.21) 0.017 (1.26)

Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio Change in Market Share t–t-4

No Yes 76,624 0.055

Yes Yes 76,624 0.059

Yes Yes 76,624 0.059

No Yes 38,180 0.056

Yes Yes 38,180 0.059

Yes Yes 38,180 0.059

Industry market share of violators is below the median 7 8 9 1.813 4.704 4.719 (0.59) (1.55) (1.55) -3.008*** -3.023*** -2.960*** (-13.97) (-11.60) (-11.37) 0.196*** 0.115 0.111 (3.34) (1.14) (1.10) -1.462 -1.502 -1.524 (-1.56) (-1.59) (-1.62) 0.125*** 0.197*** 0.188** (2.93) (2.65) (2.57) 0.017*** 0.005 0.004 (4.25) (0.84) (0.65) 0.025*** 0.037*** 0.038*** (4.36) (6.12) (6.15) -0.014 (-1.28) No Yes 38,444 0.063

Yes Yes 38,444 0.068

Yes Yes 38,444 0.068

AC

CE

PT

ED

Four-quarter lag of covenant controls Calendar quarter-year and fiscal quarter-year fixed effects Number of observations 2 Adjusted R

M

AN US

Leverage ratio

CR IP T

ACCEPTED MANUSCRIPT

39

CR IP T

ACCEPTED MANUSCRIPT

Table 3 Rival firm sales growth, product uniqueness, and the industry market share of covenant violators

AC

CE

PT

ED

M

AN US

This table presents changes in sales growth of rival firms when firms in their industry violate a financial covenant based on different characteristics of products peer violators sell. The dependent variable is Change in Market Share defined as the sales growth in the year following the violation ((Salest+4-Salest)/Salest) adjusted by the three-digit SIC code industry average sales growth, where the average excludes the firm itself. Industry market share of violators is the total sales prior to the new covenant violation quarter of violators within the industry (three-digit SIC code) divided by the total corresponding industry sales. Specifically, it is equal to the sum of the past four quarters’ sales (Salest1+Salest-2+Salest-3+Salest-4) of violators, divided by the sum of the past four quarters’ sales of that industry (Salest-1+Salest-2+Salest-3 +Salest-4), where subscript t refers to the new covenant violation quarter. We limit the sample to rival firms, i.e, firms that are not in a covenant violation. We include industry-quarters which have at least one firm in a new covenant violation. We use three measures for product uniqueness. In columns (1) to (3), we define product uniqueness by the nature of the industry. The product uniqueness dummy equals one for firms with SIC industry code between 3400 and 4000 (firms producing machines and equipment) and zero otherwise. In columns (4) to (6), we define product uniqueness by Research and Development (R&D) Intensity. If a firm’s annual R&D spending is more than the median level for the full sample in the year preceding the violation, that firm is considered as R&D intensive. If more than half of the violator firms in the industry are R&D intensive, product uniqueness dummy is defined as one, zero otherwise. In columns (7) to (9), we define product uniqueness by product fluidity. If a firm’s product fluidity is less than the median level in the full sample in the year preceding the violation, the firm’s product fluidity is defined as low. If more than half of the violator firms in the industry have low product fluidity, the product uniqueness dummy is defined as one, zero otherwise. Covenant controls include the following: operating cash flow scaled by average assets, leverage ratio, interest expense scaled by average assets, net worth scaled by assets, current ratio and market-to-book ratio. All specifications include calendar quarter-year fixed effects, fiscal quarter-year fixed effects, the level and first difference of Ln(assets), and the level and first difference of the ratio of net fixed assets to total assets. The first difference is computed as the value in quarter t minus the value in quarter t-4. In some specifications, we include four-quarter lag of covenant controls (i.e. Covenant Controlt-4). Definitions of the control variables are described in the appendix. Standard errors are clustered by firm. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

40

Industry market share of violators *Product Uniqueness Dummy Product Uniqueness Dummy Operating cash flow/ average assets Leverage ratio Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio Change in Market Share t–t-4

No Yes 76,624

AC

CE

0.057

R&D Intensity 5 0.615*** (7.75)

6 0.616*** (7.78)

7 0.882*** (6.78)

-0.697*** (-3.41) -0.124*** (-11.21) -2.613*** (-13.59) 0.158** (2.36) -0.759 (-0.91) 0.213*** (4.07) 0.010** (2.08) 0.038*** (7.33) -0.002 (-0.19)

-0.480*** (-3.29) 0.161*** (17.42) -2.725*** (-15.38) 0.200*** (4.83) -0.527 (-0.57) 0.137*** (4.00) 0.015*** (4.56) 0.028*** (5.65)

-0.959*** (-5.64) 0.086*** (7.05) -2.636*** (-13.69) 0.160** (2.42) -0.802 (-0.96) 0.197*** (3.84) 0.009** (1.98) 0.037*** (7.20) -0.001 (-0.08)

-0.726*** (-3.53) -0.130*** (-11.79) -2.618*** (-15.74) 0.199*** (4.97) -0.765 (-0.92) 0.145*** (4.39) 0.017*** (5.18) 0.027*** (5.70)

-0.697*** (-3.40) -0.123*** (-11.24) -2.619*** (-13.67) 0.159** (2.37) -0.758 (-0.91) 0.214*** (4.07) 0.010** (2.14) 0.038*** (7.32)

Yes

Yes

No

Yes

Yes

Yes 76,624

Yes 76,624

Yes 76,624

Yes 76,624

0.061

0.061

0.060

0.063

PT

Four-quarter lag of covenant controls Calendar quarter-year and fiscal quarter-year fixed effects Number of observations 2 Adjusted R

-0.958*** (-5.63) 0.086*** (7.09) -2.638*** (-13.75) 0.160** (2.42) -0.802 (-0.96) 0.197*** (3.84) 0.009** (2.03) 0.037*** (7.20)

ED

-0.984*** (-5.77) 0.088*** (7.27) -2.621*** (-15.84) 0.208*** (5.25) -0.819 (-0.98) 0.120*** (3.82) 0.016*** (4.94) 0.025*** (5.42)

4 0.619*** (7.85)

Product Fluidity 8 9 0.874*** 0.880*** (6.73) (6.80)

-0.484*** (-3.32) 0.157*** (16.77) -2.653*** (-12.93) 0.170** (2.35) -0.584 (-0.63) 0.226*** (3.97) 0.010** (1.97) 0.038*** (7.08)

-0.486*** (-3.34) 0.157*** (16.68) -2.635*** (-12.81) 0.169** (2.34) -0.589 (-0.63) 0.223*** (3.95) 0.009* (1.88) 0.038*** (7.09) -0.005 (-0.52)

No

Yes

Yes

Yes 76,624

Yes 67,781

Yes 67,781

Yes 67,781

0.063

0.064

0.067

0.067

AN US

Industry market share of violators

Durable Goods 2 3 0.991*** 0.992*** (11.00) (11.01)

M

1 1.012*** (11.25)

CR IP T

ACCEPTED MANUSCRIPT

41

Table 4 Rival firm advertising expense and the industry market share of covenant violators

CR IP T

ACCEPTED MANUSCRIPT

AC

CE

PT

ED

M

AN US

This table presents changes in advertising expense of rival firms when firms in their industry violate a financial covenant. The dependent variable is the change in natural logarithm of annual advertising expense in the year following the violation, the change in annual advertising expense divided by last year’s sales in the year following the violation, and the change in natural logarithm of the annual advertising expense divided by last year’s sales in the year following the violation. Industry market share of violators is the total sales of violators within the industry (three-digit SIC code) prior to the new covenant violation year divided by the total corresponding industry sales. We limit the sample to rival firms, i.e, firms that are not in a covenant violation. We include industry-years which have at least one firm in a new covenant violation. In this regression, covenant controls are computed using annual data. Covenant controls include the following: operating cash flow scaled by average assets, leverage ratio, interest expense scaled by average assets, net worth scaled by assets, current ratio and market-to-book ratio. All specifications include year fixed effects, the level and first difference of Ln(assets), and the level and first difference of the ratio of net fixed assets to total assets. The first difference is computed as the value in year t minus the value in year t-1, where t is the violation year. In some specifications, we include one-year lag of covenant controls (i.e. Covenant Controlt-1). Standard errors are clustered by firm. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

42

Change in Ln(Advertising Expense) 1 2 0.575*** 0.522*** (3.35) (3.07) 0.594*** 0.454*** (8.90) (4.27) -0.029 -0.095 (-0.44) (-1.02) 0.045 0.161 (0.11) (0.37) -0.027 -0.031 (-0.63) (-0.52) -0.000 0.003 (-0.04) (0.59) 0.029*** 0.045*** (6.00) (7.28) -0.047*** -0.045*** (-6.88) (-6.72) -0.094*** -0.093*** (-4.33) (-4.23) -0.018 -0.005 (-0.96) (-0.25)

Industry market share of violators Operating cash flow / average assets

Change in (Advertising Expense/Lagged Sales) 3 4 0.091*** 0.087*** (2.71) (2.83) -0.139*** -0.147* (-2.67) (-1.69) 0.008 -0.021 (0.51) (-1.28) 0.044 0.053 (0.31) (0.35) 0.013 0.008 (0.94) (0.52) -0.001 -0.001 (-0.90) (-0.65) 0.012** 0.013** (2.53) (2.32) -0.875*** -0.874*** (-34.62) (-34.04) -0.004 -0.006 (-0.50) (-0.75) -0.032*** -0.031*** (-3.92) (-4.27)

AN US

Leverage ratio

CR IP T

ACCEPTED MANUSCRIPT

Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio

M

Level of the dependent variable First difference of the dependent variable

ED

Change in Market Share t–t-1

No Yes 7,054

Yes Yes 7,054

No Yes 7,015

Yes Yes 7,015

No Yes 7,015

Yes Yes 7,015

0.131

0.140

0.817

0.817

0.215

0.233

AC

CE

PT

One-year lag of covenant controls Year fixed effects Number of observations 2 Adjusted R

Change in Ln(Advertising Expense/Lagged Sales) 5 6 0.619*** 0.525*** (3.62) (3.14) 0.363*** -0.069 (4.59) (-0.52) -0.028 -0.046 (-0.39) (-0.46) -0.047 -0.223 (-0.10) (-0.44) -0.029 0.066 (-0.67) (1.11) 0.007 0.020*** (1.60) (2.95) 0.018*** 0.037*** (3.12) (5.16) -0.086*** -0.075*** (-11.99) (-10.76) 0.003 -0.024 (0.14) (-1.19) -0.319*** -0.290*** (-13.72) (-12.64)

43

CR IP T

ACCEPTED MANUSCRIPT

Table 5 Rival firm operating performance and the industry market share of covenant violators

AC

CE

PT

ED

M

AN US

This table presents changes in operating performance of rival firms when firms in their industry violate a financial covenant. Columns (1) - (3) show operating cash flow (operating income before depreciation and amortization) scaled by average assets. Columns (4) to (6) show return on assets (ROA), calculated by operating income before depreciation minus depreciation and amortization divided by average assets. Columns (7) to (9) show return on equity (ROE), calculated by net income divided by average common equity. Industry market share of violators is the total sales prior to the new covenant violation quarter of new violators within the industry (three-digit SIC code) divided by the total corresponding industry sales. Specifically, it is equal to the sum of the past four quarters’ sales (Salest-1+Salest-2+Salest-3+Salest-4) of violators, divided by the sum of the past four quarters’ sales of that industry (Salest-1+Salest-2+Salest-3+Salest-4), where subscript t refers to the new covenant violation quarter. We limit the sample to rival firms, i.e, firms that are not in a covenant violation. We include industry-quarters which have at least one firm in a new covenant violation. Covenant controls include the following: operating cash flow scaled by average assets, leverage ratio, interest expense scaled by average assets, net worth scaled by assets, current ratio and market-to-book ratio. All specifications include calendar quarter-year fixed effects, fiscal quarter-year fixed effects, the level and first difference of Ln(assets), the level and first difference of the ratio of net fixed assets to total assets, the level and first difference of the corresponding dependent variable. The first difference is computed as the value in quarter t minus the value in quarter t-4. Definitions of the control variables are described in the appendix. In some specifications, we include four-quarter lag of covenant controls (i.e. Covenant Controlt-4). Standard errors are clustered by firm. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively.

44

Operating cash flow / average assets Leverage ratio Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio Level of the dependent variable

M

First difference of the dependent variable Change in Market Share t–t-4

-0.000 (-0.42)

Yes

Yes

No

Yes 72,628

Yes 72,628

Yes 72,628

Yes 67,745

0.108

0.132

0.132

0.140

ED

No

Yes

Change in Return on Equity 7 8 9 0.067*** 0.062*** 0.063*** (3.07) (2.89) (2.90) 1.147*** 1.076*** 1.078*** (22.33) (16.81) (16.61) -0.125*** -0.135*** -0.135*** (-8.51) (-4.84) (-4.85) -0.291 -0.254 -0.254 (-1.48) (-1.29) (-1.29) -0.020* -0.001 -0.001 (-1.69) (-0.05) (-0.06) 0.002*** 0.003*** 0.003*** (5.42) (4.80) (4.79) -0.001* 0.000 0.000 (-1.79) (0.48) (0.48) -0.688*** -0.704*** -0.704*** (-29.17) (-28.68) (-28.67) -0.104*** -0.093*** -0.093*** (-8.55) (-6.51) (-6.51) -0.000 (-0.16)

Yes

No

Yes

Yes

Yes 67,745

Yes 67,745

Yes 66,850

Yes 66,850

Yes 66,850

0.141

0.141

0.211

0.212

0.212

AC

CE

PT

Four-quarter lag of covenant controls Calendar quarter-year and fiscal quarter-year fixed effects Number of observations 2 Adjusted R

Change in Return on Assets 4 5 6 0.018*** 0.017*** 0.017*** (4.62) (4.56) (4.64) 0.267*** 0.256*** 0.258*** (5.53) (4.73) (4.77) -0.006* -0.014*** -0.014*** (-1.81) (-2.71) (-2.73) 0.006 0.022 0.022 (0.12) (0.47) (0.46) -0.006** -0.013*** -0.013*** (-2.20) (-2.67) (-2.70) -0.001*** 0.000** 0.000** (-4.88) (-2.44) (-2.55) 0.000** -0.000 -0.000 (-2.09) (-0.78) (-0.76) -0.464*** -0.489*** -0.488*** (-9.91) (-10.40) (-10.39) -0.161*** -0.129*** -0.129*** (-15.04) (-4.29) (-4.30) -0.000 (-0.81)

AN US

Industry market share of violators

Change in Operating cash flow/average assets 1 2 3 0.015*** 0.012*** 0.012*** (4.11) (3.33) (3.37) -0.261*** -0.377*** -0.376*** (-23.55) (-28.50) (-27.92) -0.002 -0.011** -0.011** (-0.59) (-2.22) (-2.23) -0.017 0.033 0.032 (-0.34) (0.70) (0.69) -0.002 -0.010** -0.010** (-0.61) (-2.36) (-2.37) -0.001*** -0.001*** -0.001*** (-7.59) (-4.48) (-4.51) -0.001*** 0.000* 0.000* (-3.26) (-1.72) (-1.72)

CR IP T

ACCEPTED MANUSCRIPT

45

CR IP T

ACCEPTED MANUSCRIPT

Table 6 Coefficient estimates of İndustry market share of violators based on industry returns and violator idiosyncratic returns

AN US

Following Leary and Roberts (2014), we decompose each firm’s stock return into systematic, industry, and idiosyncratic components. We regress the total return of the firm on the excess market return (CRSP VWRETD minus the risk-free rate) and on excess industry returns based on equally weighted three-digit SIC code industry portfolio (excluding the firm’s return). We estimate parameters for year t using monthly returns, on a rolling annual basis, over a five year window (year t, t-1, t-2, t-3, t-4), requiring at least 24 months of nonmissing returns. We use parameters estimated at year t and realized monthly returns at year t+1 to compute the idiosyncratic component for each month in year t+1. After obtaining the idiosyncratic component, we merge the sample of idiosyncratic returns with the covenant violation sample on the last month of the violation quarter. First, we run the subsample tests based on whether the industry equally weighted average return in excess of risk free rate (excluding the violator's returns) is positive or non-positive. Second, we run tests for a subsample of industry quarters which have more than two thirds of violators with negative idiosyncratic returns. Finally, we run the tests for a subsample of industry quarters which have positive industry equally weighted average excess return and also have more than two thirds of violators with negative idiosyncratic returns. Panels A, B, and C contain the coefficients and t-statistics on the variable Industry market share of violators from estimations of the regression specifications from Tables 2, 4, and 5, respectively, based on the sample described in the column heading.

M

Industry market share of violators

ED

Non-Positive industry average excess return Coefficient N (t) 27,161 0.680*** (7.44)

Higher than two thirds violators with negative idiosyncratic return Coefficient N (t) 59,054 0.593*** (7.89)

AC

CE

PT

Panel A: Market share (1) Change in market share

Positive industry average excess return Coefficient N (t) 49,463 0.787*** (7.57)

Positive industry average excess return & Higher than two thirds violators with negative idiosyncratic return Coefficient N (t) 38,200 0.644*** (6.11)

46

Panel B: Advertising expense

CR IP T

ACCEPTED MANUSCRIPT

Change in Ln(advertising expense)

3,420

0.777*** (2.99)

3,634

0.270 (1.23)

4,508

0.367** (2.01)

2,160

0.499** (2.01)

(2)

Change in (advertising expense/lagged sales)

3,416

0.024* (1.81)

3,599

-0.003 (-0.17)

4,484

0.009 (0.69)

2,159

0.022 (0.90)

(3)

Change in Ln(advertising expense/lagged sales)

3,416

0.682*** (2.69)

3,599

0.338 (1.50)

4,484

0.426** (2.33)

2,159

0.455* (1.75)

0.014*** (3.16)

25,799

0.005 (0.90)

56,007

0.012*** (3.33)

36,160

0.013*** (3.05)

0.019*** (4.14)

23,976

0.009 (1.56)

52,278

0.016*** (4.32)

33,806

0.017*** (3.72)

0.064** (2.43)

24,317

0.056 (1.60)

51,450

0.043* (1.93)

32,801

0.042 (1.55)

46,829

(2)

Change in return on assets

43,769

(3)

Change in return on equity

42,533

AC

CE

PT

ED

M

Panel C: Operating performance (1) Change in operating cash flow/average assets

AN US

(1)

47

CR IP T

ACCEPTED MANUSCRIPT

Table A.1 Descriptive Statistics

ED

M

New Violators Mean -0.004 0.335 0.009 0.382 1.942 1.507 5.024 0.285 0.181 -0.021 -0.150 0.013 0.055 -0.197

Median 0.009 0.302 0.006 0.411 1.508 1.173 4.840 0.210 -0.015 -0.004 -0.033 0.012 0.017 -0.193

N 76624 76624 76624 76624 76624 76624 76624 76624 76624 74213 72661 7054 7015 76624 76624

Rival Firms Mean 0.011 0.212 0.005 0.525 3.220 2.257 5.229 0.248 0.262 -0.004 -0.041 0.031 0.062 -0.175 0.015

Median 0.024 0.141 0.002 0.565 2.188 1.599 4.970 0.162 0.066 0.012 0.012 0.035 0.018 -0.224 0.004

AC

CE

PT

Operating cash flow /average assets Leverage ratio Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio Log(assets) Log(PPE/assets) Sales growth ROA ROE Change in Ln(advertising expense) Advertising expense/Lagged sales Change in Market Share Industry market share of violators

N 3563 3526 3522 3604 3511 3604 3604 3599 3561 3413 3333 453 511 3218

AN US

This table shows the descriptive statistics of variables for new covenant violation firms and non-violator rival firms. 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. Rival firms are firms that are not in a covenant violation. Please see the appendix for a detailed description of the variables.

48

Table A.2 Firm characteristics of industry-propensity score matched rival firms and covenant violators

CR IP T

ACCEPTED MANUSCRIPT

M

New Violators Mean Median 0.002 0.012 0.318 0.281 0.008 0.005 0.405 0.431 2.017 1.587 1.460 1.134 5.158 4.981 0.292 0.218 0.053 -0.024 -0.014 -0.001 -0.097 -0.025

Industry-propensity score matched Rival Firms Mean Median 0.009 0.019 0.303 0.285 0.007 0.005 0.428 0.430 2.085 1.704 1.440 1.185 5.417 5.216 0.295 0.225 0.055 0.004 -0.006 0.006 -0.071 0.002

Difference Mean Median -0.007*** -0.007*** 0.016** -0.005 0.001*** 0.004** -0.023*** 0.001 -0.068 -0.118** 0.020 -0.051*** -0.260*** -0.235*** -0.003 -0.008 -0.002 -0.027** -0.008*** -0.007*** -0.027*** -0.027***

AC

CE

PT

ED

Operating cash flow /average assets Leverage ratio Interest expense / average assets Net worth / assets Current ratio Market-to-book ratio Log(assets) Log(PPE/assets) Sales growth ROA ROE

N 1896 1896 1896 1896 1896 1896 1896 1896 1896 1785 1676

AN US

This table reports mean and median statistics for industry-propensity score matched rival firms and new covenant violators. A rival firm is identified as a nonviolator firm with the same three-digit SIC code, a pre-violation average sales growth within one-half of the standard deviation above and below the violator firm’s pre-violation average sales growth, and a propensity score of violating a covenant closest to that of the violator firm. The last two columns report the differences in means and medians. Definitions of the control variables are described in the appendix. ***, ** and * denote 1%, 5% and 10 % levels of significance, respectively.

49