Debt covenant violation,competition and cost of new debt

Abstract

This article empirically shows that the cost of new debt is higher for firms that commit covenant violations. Using a proxy for product market competition to capture exogenous changes to a firm’s competitive environment, I find that the cost is systematically higher for firms that operate in competitive markets. Moreover, I identify channels through which violations can increase the cost of new debt, namely, the incidence, timing and frequency effects, and I document these effects to be more acute for competitive markets. Overall, the study finds that the market prices financial contracts by taking into account the information content of the violation and the risk arising from market competition.

JEL Classification: G12, G30

Keywords

Cost of debt debt covenant violation market competition

1. Introduction

Debt covenant violations, incurred when firms fail to meet the contractual requirements contained in debt agreements, increase the cost of both the debt and risk associated with a firm (e.g. Beneish and Press, 1995; Fargher et al., 2001). Recently, Gao et al. (2017) found higher bid-ask spreads and stock return volatility in firm-quarters that followed a violation, owing to information asymmetry and shareholder uncertainty. To avoid costs such as higher audit fees, a greater likelihood of receiving a going-concern opinion and experiencing an auditor resignation, firms engage in accrual-based and real earnings management (e.g. Bhaskar et al., 2016; Butt et al., 2016). Following a violation, creditors intervene and command remedial changes, though the concessions demanded by the creditors are often substantial and – in extreme cases – may lead to accelerated repayment of the violated debt (Dichev and Skinner, 2002). Earlier evidence on renegotiated loan contracts provides anecdotal evidence that violators pay an additional 39 basis points (Nini et al., 2012). However, these findings are limited to existing loans.

The information content of the violation is twofold. First, the firm was not able to maintain its debt agreement and is riskier than previously believed. Second, the firm has not been able to attain its objectives, and its short-term goals are not achievable. In this case, creditors would view the firm as riskier and hence would only extend future credit at a higher cost to the borrower. The effect will be acute for firms issuing bonds, considering that firms tend to avoid mitigation costs in private debt agreements and turn to bond financing (Bolton and Freixas, 2000). This raises an important question as to how debt covenant violations affect the pricing of new public debt.

Furthermore, most firms are constantly interacting with other firms while competing for customers and market share. This exogenous business environment fundamentally affects the firms’ operating decisions and the riskiness of their business environment (Valta, 2012). There are several potential reasons why the cost of covenant violations will be more profound for firms operating in a competitive environment when raising capital. One reason is that because competition reduces pledgeable income and increases cash flow risk, competition could also increase firms’ default risk, thereby increasing the cost of violation. Another reason is liquidation value, as higher liquidation value allows firms to obtain credit at lower cost (Benmelech et al., 2005), and the competitive nature of the product market could affect the number and the financial strength of potential buyers and hence the asset liquidity of an industry (Ortiz-Molina and Phillips, 2014). Therefore, competition could also affect the cost of new debt by changing the firm’s liquidation value, consequently magnifying the cost of the violation.

A more intriguing view of covenant violation is that it can exacerbate the difficulty of obtaining future credit. For example, bondholders are generally concerned with a firm’s ability to commit to contractual agreements, and failing to do so can dramatically affect the credit quality of the borrower. Notably, Denis and Mihov (2003) find that firms with the highest credit quality borrow from public sources, firms with medium credit quality borrow from banks and firms with the lowest credit quality borrow from non-bank private lenders. By measuring the effect of violation on the cost of new debt for firms with the highest credit quality (i.e. bond issues), this article presents a fuller and more detailed picture of the channels that affect the pricing of new debt.

To capture the effect of violation on the cost of new debt, I identify three different aspects of violation and separately measure their impact on the cost of new debt. First, I consider the incidence of covenant violation and measure the relative increase in the cost of new debt for firms that have violated a debt covenant. I label this the incidence effect because it increases the likelihood of both debt service default and bankruptcy, as well as the risk of the firm (Fargher et al., 2001; Wilkins, 1997). Second, I focus on the timing of the violation with respect to bond issuance. The covenant violation is reported to the Securities and Exchange Commission (SEC) at the end of the quarter in which the violation was committed. Although the financial performance of the firm deteriorates in the quarter leading up to the violation (as reported in section 3), the occurrence of violation is not known with certainty until the end of the quarter. Therefore, the issuance of bonds in the quarter of the violation will have different bearings on the cost of new debt than if the bonds were issued in the post-violation quarter. I label this the timing effect, as the information content of the reported violation preceding the bond issue affects the cost adversely. Third, I study the effect of multiple violations, in addition to a single violation, on the cost of new debt to determine whether frequent violators incur a higher cost. For completeness, I also look at multiple violators and the frequency of violations to determine the cost of each violation. I label this the frequency effect because it measures the direct effect of the number of violations on the cost of new debt. Finally, I document the magnitude of these effects in competitive firms and how it differs from firms operating in monopolistic markets.

Combining all these considerations, it is apparent that the effects of the occurrence, timing and frequency of violations on the cost of new debt can be quite substantial. To date, bondholders’ assessment and aggregation of the effects remain unclear, primarily because of the unavailability of violation data. How any of the hypothesized effects holds is ultimately an empirical question. Therefore, I conduct an empirical analysis of the relationship between violations and cost of new debt by employing three different specifications, namely, Offer Yield, Net Interest Cost and Spread to Treasury.

The incidence of violation measured as a binary variable reported as one for violation and zero otherwise. I model the three aspects of violations (incidence, timing and frequency) using an indicator variable approach used in prior research (e.g. Freudenberg et al., 2017; Qiu and Yu, 2009; Valta, 2012); this is referred to as the dummy variable approach.¹ The bond issue data consist of quarterly data for 722 US firms and 1494 public bond issues from 1996 to 2008. I include several well-documented measures of the cost of debt and number of control variables in the analysis. In addition, I use the Herfindahl–Hirschman index (HHI) as a proxy for market competition to measure the relation among market competition, cost of new debt and covenant violation.

The findings are briefly documented as follows: First, I find that the incidence of violation increases the cost of new debt by 79–90 basis points. This is both statistically and economically significant. The results, however, do not control for the differences between the borrower and bond characteristics. I further employ a regression model to measure a structured and controlled incidence effect and find that the price increase ranges between 37 and 61 basis points across the three specifications of the cost of new debt. This increase in the cost of new debt is both statistically and economically significant. To shed light on the timing effect, I create a dummy for the quarter of bond issue and extrapolate that the information content of the actual violation would have strong bearings on new debt issuance. I find that firms issuing bonds in the quarter of violation report an increased cost of 23–40 basis points, and I attribute this increase to the declining financial performance of the firm in the quarters leading up to the violation. The increase is not statistically significant. In contrast, I document a significant increase in the cost of new debt – between 87 and 121 basis points – for firms issuing bonds in the post-violation quarter. This result is consistent with Demiroglu and James’ (2010) findings on bank loan covenants. Finally, I detail the frequency effect and examine it using a two-pronged process. First, a dummy variable is created for firms that violated a covenant violation only once, and another dummy variable is created for firms that violated multiple covenants. I find mixed evidence of the cost of new debt increasing between 24 and 59 basis points for one-time violators, with two of the three specifications being statistically significant. The result is consistent with Roberts and Sufi’s (2009) findings on the cost of private debt covenant violation. The result for repeat violators is consistent with the hypothesis that multiple violations will lead to a higher cost of new debt. The results show strong evidence that repeat violators incur a higher cost of new debt, as evidenced by an increase of 44–52 basis points. This increase is statistically significant at the 1% level and is economically large. Finally, I include the number of actual violations by each firm and estimate the cost per violation to the borrower. The results show that the cost of new debt increases by 5–7 basis points for each violation, and these results are significant at the 1% level. The findings provide evidence that the reporting of the violation reduces information asymmetry and leads to a higher cost of new debt.

To conclude, the study finds that the intensity of the effects is systematically higher for competitive firms. First, I find that the incidence effect increases the cost of new debt by 43–71 basis points for competitive firms. Second, the cost of new debt does not increase for either classification of firms if the bond was issued in the same quarter as the violation. However, I find categorical evidence for the timing effect if the bond is issued in the post-violation quarter, with the cost of new debt being higher by 148–182 basis points for competitive firms. Third, I find mixed evidence for one-time violators, with two of the three specifications of the cost of new debt displaying a statistically significant increase of 51–83 basis points for competitive firms. The frequency effect is, however, more profound for repeat violators in competitive markets, with firms incurring a statistically significant cost of 49–73 basis points. Finally, the evidence for the actual number of violations is more profound, with competitive firms incurring an increased cost of 6–8 basis points. The results conclude that creditors price the incidence, timing and frequency effect more acutely for competitive firms.

This article belongs to a group of recent studies that examine the various aspects of debt covenant violation. However, it is the first to provide explicit estimates of such infractions on the cost of future credit, and it compliments earlier literature (e.g. Bradley and Roberts, 2015) that details the changes in the structure of loan agreements following a covenant violation. It also advances research on earnings management. For example, Roychowdhury (2006) and Zang (2012) find evidence that managers are providing price discounts to increase sales temporarily, overproducing to report lower costs of goods sold and reducing discretionary expenses to avoid such violations. Similarly, DeFond and Jiambalvo (1994) and Sweeney (1994) find that firms go to great lengths to avoid technical defaults and engage in activities such as manipulation of accruals to avoid such events. Dyreng (2009) finds that as firms approach covenant violation, they engage in income-increasing earnings management, which increases their tax liability and associates an increase in the current tax liability by an amount equivalent to increasing the cost of debt financing by 12.92–22.72 basis points. In line with these studies, several papers have outlined the costs associated with such violations. For example, Beneish and Press (1993, 1995) report that common share prices respond negatively to reports of violations. Fargher et al. (2001) concluded that such breaches in firms’ loan agreements increase the likelihood of debt service default, bankruptcy and risk of the firm. Nini et al. (2012) provide anecdotal evidence of an increase in the renegotiated cost of existing private debt, which is incurred because of the violation. Ertan and Karolyi (2014) find that the information contained in changes in the probability of covenant violations is priced by the stock market, incremental to changes in firm fundamentals. Recently, Valta (2012) reported that the cost of bank debt is systematically higher for firms that operate in competitive product markets. This article furthers that research and provides explicit estimates of the cost of such violation on new debt. It also augments prior research by providing evidence that these effects are more severe for firms in competitive markets. It adds to the literature by articulating and then using empirical tests to identify the aspects of violation that affect the cost of new debt. This study outlines that violations are costly to the firm and contain important information content, which reduces information asymmetry for bondholders in line with Zhu and Gippel (2015). It is important to note that this article is concerned with the cost of new debt, which should be distinguished from the increase in the cost of existing debt as a result of contract renegotiation in the event of a covenant violation. The results provide the first documented evidence of incidence, timing and frequency effects for firms with the highest credit quality by outlining and examining the three channels through which the cost of new debt is affected.

The study makes several contributions to the literature. First, I use SEC filings of covenant violation to identify the relation between such violations and the cost of new debt issuance. I employ different measures of the cost of new debt to mitigate the ambiguity of results derived from a single source. While earlier papers focus on bankruptcy risk (Fargher et al., 2001), credit control rights (Nini et al., 2012), changes in structure of loan agreements (Bradley and Roberts, 2015) or earnings management (Bhaskar et al., 2016), this article focuses on the pricing of new debt. Taken as a whole, the effect of the violation seems to be substantial, and this article provides the first estimates of the costs of these violations.

Second, this study contributes to the literature by recognizing the various aspects of violation. Specifically, the costs of the incidence, timing and frequency of violation are measured, and all three are found to have important bearings on the cost of new debt.

Third, this study contributes to the literature analysing the impact of product market competition on the cost of debt (e.g. Qiu and Yu, 2009; Valta, 2012). Although previous studies have shed light on the relation between market competition and cost of debt, this article points to an important new dimension. Specifically, it documents the effect of the intensity of market competition on the pricing of new debt for violating firms. The results provide evidence that a firm’s competitive environment needs to be considered when assessing the increase in the firm’s cost of new debt financing as a result of a violation.

This study has some limitations and suggests some opportunities for further research. First, the severity of the violation would have an important impact on the cost of new debt. For example, a debt covenant violation of defaulting on debt payment would increase the risk of the firm, and thereby the associated cost, significantly more than violating a covenant that restricts the debt level to a certain ratio. The available data, however, do not differentiate the violation in this respect. Second, the covenant violation data only report whether violations occurred in a specific quarter and do not take into account the number of violations a firm committed in a quarter. This is, to some extent, covered in the incidence effect; however, it is acknowledged as a limitation because it is unobserved by the researcher. Third, another approach to measuring the incidence effect of violation would be to use a matched sample in which firms are matched on firm and bond characteristics to see if violation does increase the cost of new debt. This approach is not used because it requires bond issuance by violating and non-violating firms on two time dimensions: before and after the covenant violation. In the sample, this is unobservable to the researcher, as each firm issued two bonds on average. Because violations are evidenced to be cyclical in nature and matched firms can only be compared if the timing of the debt issuance is matched, the approach is not feasible for the sample. The bond issue time frame for the matched sample was extended to four quarters, but a matched sample could not be found.

The remainder of this article is structured as follows. The next section describes the sample selection process and the variables used in the analysis. Section 3 presents summary statistics and section 4 details the results. The final two sections of the report comprise robustness tests and the concluding remarks.

2. Data and methodology

This article uses three data sources to construct the sample used in the empirical analysis. First, I use the debt covenant violation reporting data constructed by Sufi. The data were constructed using the SEC Edgar website, which contains indices of every filing submitted to the Commission. The Commission made electronic filing mandatory for all SEC-registered firms in 1996, and the entire sample covers the period 1996–2008. The database reports a violation as one if a firm is in violation of a debt covenant in a quarter and zero otherwise.² I construct five different measurements of violation to capture incidence, timing and frequency effects: (1) VIOL is a binary variable that equals one for violating firms and zero for non-violating firms. (2) Viol is a binary variable that equals one if a debt covenant violation is reported in a quarter and zero otherwise. This measure is further used for Viol_t and Viol_t_− 1 to identify the timings of violation with respect to bond issuance. (3) Viol = 1 is a binary variable that equals one if a firm has reported exactly one violation in the entire data set. (4) Viol > 1 is a binary variable that equals one if a firm has ever reported more than one covenant violation, either in consecutive or intermittent quarters, and zero otherwise. (5) Viol = n is the total number of violations reported by the firm prior to issuing new bonds. VIOL is used to measure the incidence effect, Viol (Viol_t and Viol_t_− 1) is used to measure the timing effect and the last three (Viol = 1, Viol > 1 and Viol = n) are used to measure the frequency effect.

Second, I use the public bond issue data collected by SDC Platinum. The data set consists of new bond issues by corporations and contains specific bond issue characteristics such as offer yield,³ loan maturity and loan amount. I obtain the following variables from these data: (1) The cost of new debt. I include three different measures of the cost of new debt. First, I use Offer Yield, which is the yield offered to investors at the time of the bond issue. Earlier studies (e.g. Pástor et al., 2008) use the implied cost of capital (ICC) to capture time variations in expected stock returns. For bonds, Offer Yield is the measure that corresponds to ICC, and I use it as one of the measures of cost of capital. Second, I use Net Interest Cost, following the accounting literature on bonds (e.g. Apostolou et al., 2014). The Net Interest Cost is defined as the overall interest expense that is associated with the bond; it is based on the average coupon rate weighted to years of maturity and adjusted for any associated discounts or premiums. Third, I use Spread to Treasury, which is the difference in the bond yield and risk-free treasury security yield with similar maturity, to measure difference in differences following, for example, Qiu and Yu (2009). (2) Loan Size is the proceeds in dollars from bond issuance. (3) Log_Maturity is the log of maturity in months of issued bonds, and (4) S&P Ratings is the S&P rating for bonds and has been converted to a number⁴ following Qiu and Yu (2009). This yields a total of 722 US firms and 1494 public bond issues for the sample.

As motivated by credit risk models and subsequent validation by empirical research, a significant part of the changes in cost of debt can be explained by changes in firm characteristics (e.g. Collin-Dufresn et al., 2001; Ericsson et al., 2009). Following this literature, I obtain the following variables from the Compustat quarterly file: (1) The z-score is the Altman z-score used to predict corporate defaults. The z-score for manufacturing firms is computed following Hillegeist et al. (2004) and for non-manufacturing firms following Altman (2000). (2) Size is the natural log of total quarterly assets of the firm in year 2000 dollars. (3) Leverage is the ratio of the book value of long-term debt plus long-term debt in current liabilities to the book value of the total assets of the firm. (4) Coverage is the interest coverage ratio and is computed as the ratio of earnings before interest and taxes to the interest expense. (5) Tangibility is the ratio of property, plants and equipment to the total assets of the firm. (6) Market-to-Book is the ratio of the market value of assets, computed as the market value of equity plus the book value of debt, to the book value of assets. (7) Current Ratio is the ratio of current assets to current liabilities.

The main proxy to measure market competition is the HHI, which is widely used to measure competition. A lower HHI implies competitive markets and vice versa. The HHI combines Compustat data with Herfindahl data from the US Commerce Department and employee data from the Bureau of Labor Statistics. I follow earlier literature (e.g. Valta, 2012) and define firms in the lowest quartile of sample distribution for the HHI as competitive firms. For robustness, I define a dummy variable that is equal to one if the HHI is in the lowest quartile sample distribution and zero otherwise. The dummy variable, as opposed to an exact value of the HHI, should mitigate measurement problems, which are sometimes an issue with the HHI. Moreover, the interaction between the dummy variable and the violation specification allows for an intuitive economic interpretation of the coefficients while measuring the impact of violation on the cost of new debt.

3. Summary statistics

The sample is constructed by merging the dataset and winsorizing all variables at the 1% level. Table 1 summarizes the differences between violating and non-violating firms in terms of loan and firm-specific characteristics. The differences show that violating firms carry higher offer yield, net interest cost and spread to treasury, and they are smaller in denomination and have shorter maturities compared to non-violating firms. The differences are economically and statistically significant. Notably, non-violating firms are large organizations and have higher tangible assets, higher market-to-book ratios, higher z-scores (i.e. a lower chance of default), lower financial leverage, higher interest coverage ratios, higher current ratios and better S&P ratings compared to violating firms. The differences are significant at the 1% level. An important finding is that violating firms pay 79–90 basis points more than non-violating firms.

Table 1.

Comparing loan and firm characteristics for debt covenant non-violating and violating firms. Panel A presents mean loan characteristics for new bond issues by firms that did not violate a debt covenant (non-violating firms) and firms that violated at least one debt covenant (violating firms). Offer Yield is the yield offered to investors at the time of bond issue; Net Interest Cost is the overall interest expense that is associated with the bond and is based on the average coupon rate weighted to years of maturity adjusted for any associated discounts or premiums; Spread to Treasury is the difference in the bond yield and risk-free treasury security yield with similar maturity; Loan Size is the proceeds from bond issuance; Loan Maturity is the maturity in years of issued bonds. Panel B presents mean firm characteristics for new bond issues by firms that did not violate any debt covenant (non-violating firms) and firms that violated at least one debt covenant (violating firms). z-score is the Altman z-score used to predict corporate defaults and is computed differently for manufacturing and non-manufacturing firms. Assets are total assets of the firm; Leverage is the ratio of book value of long-term debt plus long-term debt in current liabilities to the book value of total assets of the firm; Coverage is the ratio of earnings before interest and taxes to the interest expense; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets; Current Ratio is the ratio of current assets to current liability; S&P Ratings is the S&P rating for bonds and has been converted to a number (see Supplementary Appendix).

	Non-violating firms	Violating firms	Difference
Panel A: loan characteristics
Offer Yield (%)	6.67	7.46	−0.79***
Net Interest Cost (%)	6.79	7.69	−0.90***
Spread to Treasury (%)	1.47	2.31	−0.84***
Loan Size ($ millions)	250.37	214.43	35.94***
Loan Maturity (months)	138.1	120.8	17.3***
Panel B: firm characteristics
z-score	2.19	1.58	0.61***
Assets ($ millions)	10,442	6103	4339***
Leverage (%)	0.36	0.42	−0.06***
Coverage Ratio	12.1	7.15	4.95***
Tangibility	0.423	0.399	0.024***
Market-to-Book	1.87	1.49	0.38***
Current Ratio	1.58	1.68	−0.1***
S&P Rating	14.7	12.43	2.27***

***

, ** and * denote 1%, 5% and 10% levels of significance, respectively.

The covenant violation data provide useful insight into the behaviour of violating firms. I report the fraction of violating firms in Figure 1 and find that 9%–17% of firms report at least one violation, with the incidence of violations peaking during the 2001–2002 recession and declining thereafter, indicating the cyclical nature of violations. Table 2 summarizes the incidence of violations, with nearly 38% of firms reporting a violation. Almost 9% of firms report one violation and 29% report multiple violations, with 6% of all firm-quarter observations reporting a violation. Small- and large-sized firms violate covenants less frequently than do mid-sized firms, and the same is evidenced for tangible assets.

Figure 1.

Covenant violation from 1996 to 2008.

Table 2.

Frequency of debt covenant violation. This table presents the percentage of firms that report a financial covenant violation at least once at any point between 1996 and 2008. The sample includes firm-quarter observations available from the Compustat universe that can be matched with the firm-quarter observation for the violation data. Assets are total quarterly assets of the firm; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets.

	Violator percentage (%)
Fraction of firms reporting a covenant violation	38.24
Fraction of firms reporting one covenant violation	9.40
Fraction of firms reporting more than one covenant violation	28.84
Firm-quarter observations with covenant violation	6.42
By Assets
25th percentile	35.38
Between 25th and 50th percentiles	46.75
Between 50th and 75th percentiles	45.11
Between 75th and 100th percentiles	36.02
Top 10%	30.53
Top 5%	25.42
By Tangibility
25th percentile	34.47
Between 25th and 50th percentiles	39.86
Between 50th and 75th percentiles	41.81
Between 75th and 100th percentiles	40.68
Top 10%	36.98
Top 5%	33.05
By Market-to-Book
25th percentile	48.75
Between 25th and 50th percentiles	45.05
Between 50th and 75th percentiles	40.26
Between 75th and 100th percentiles	32.86
Top 10%	26.88
Top 5%	22.70

Figure 2 summarizes violation reporting by size, tangibility and market-to-book. The incidence of violation decreases with an increase in size and tangibility and is the lowest for firms in the 95th percentile, while the middle quartiles have the highest incidence of violation. A possible explanation for this is that large firms can manipulate their accounts and avoid violation through earnings management (Butt et al., 2016). Small firms may also face a selection bias while acquiring financing, and only financially sound firms can secure debt financing. Violations are also found to be negatively correlated to market-to-book, with the incidence dropping from approximately 50% to approximately 22% for the change from the 25th to the 95th percentile.

Figure 2.

Covenant violation by firm characteristics.

Covenant violations reduce information asymmetry by outlining that the firm is going through a period of financial uncertainty; they do not mean that a firm is near default. Creditors, thereby, are concerned when the firm does not meet contractual obligations. Accordingly, I measure the pre- and post-violation data for violating firms. Figure 3 summarizes assets, tangibility and market-to-book for violators during the eight quarters leading up to, including and following a violation. Total assets and tangibility of the firm increase during the quarters leading up to the violation and decrease after the violation has occurred. Nini et al. (2012) attribute this trend to investment conservatism. In the quarters leading up to a violation, firms grow aggressively, with total assets and tangibility increasing, on average, more than 5% and 4%, respectively. Growth levels off in the quarter of the violation and decreases in the quarters following the violation, suggesting that violators engage in divestitures and investment conservatism. Although not reported, the stock price of the firm decreases in the quarters leading up to and after the violation quarter, with the decrease tapering off in the seventh quarter after the violation. The nearly 18% decline in market-to-book in the pre-violation quarters and the nearly 3% post-violation recovery suggest that the decrease in assets in the quarter following the violation does not entirely compensate for the decrease in the stock price. Figure 4 summarizes the z-score, leverage, coverage ratio, current ratio, S&P rating and number of new bond issues in eight pre- and post-violation quarters. The z-scores start, approximately, in the ‘grey zone’ and fall into the ‘distress zone’⁵ by the time of the violation. Financial leverage increases aggressively (from nearly 29% to nearly 36%) in the quarters leading up to the violation, levels off in the quarters immediately following the violation and decreases moderately following the fifth post-violation quarter. The coverage ratio declines sharply, from approximately 12 to 1, is negative in the quarter of the violation and increases immediately following creditor intervention. The current ratio declines by nearly 16% in the pre-violation quarters and begins to increase steadily afterwards. Firms do not face sharp liquidity shortages, as the lowest level (of 1.9) in the violation quarter is still adequate for short-term liquidity needs, although it may reflect high inventory levels. The S&P ratings of the bonds do not change drastically and stay in the range between ‘BBB+’ and ‘BB+’. The total number of new bonds issued decreases considerably following a violation, with nearly 75% of all bond issues occurring in the eight pre-violation quarters. Only 2.5% of bonds are issued in the quarter of the violation, and the rest are issued in the quarters following a violation. Overall, the financial condition of the firm deteriorates in the quarters leading up to a violation and only improves moderately after the intervention of creditors.

Figure 3.

Firm characteristics preceding and following a violation.

Figure 4.

Financial indicators preceding and following a violation.

Table 3 provides the summary statistics for the outcome and control variables used in the analysis. The first three variables represent the outcome variables and include the Offer Yield, Net Interest Cost and Spread to Treasury. The control variables for the cost of new debt include z-score, size, leverage, coverage ratio, tangibility, market-to-book, current ratio, S&P rating, log of the maturity of the bonds and log of the proceeds generated from the bond issue.

Table 3.

Summary statistics. This table presents loan characteristics for new bond issues by all firms before the restrictions of each data set are applied. Offer Yield is the yield offered to investors at the time of bond issue; Net Interest Cost is the overall interest expense that is associated with the bond and is based on the average coupon rate weighted to years of maturity adjusted for any associated discounts or premiums; Spread to Treasury is the difference in the bond yield and risk-free treasury security yield with similar maturity; Loan Size is the proceeds from bond issuance; Loan Maturity is the maturity in years of issued bonds; z-score is the Altman z-score used to predict corporate defaults and is computed differently for manufacturing and non-manufacturing firms (see Supplementary Appendix). Assets are total assets of the firm; Leverage is the ratio of the book value of long-term debt plus long-term loans in current liabilities to the book value of the total assets of the firm; Coverage is the ratio of earnings before interest and taxes to the interest expense; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets; Current Ratio is the ratio of current assets to current liabilities; S&P Rating is the S&P rating for bonds and has been converted to a number (see Supplementary Appendix); Log_Maturity is the log of the maturity of the bonds; Log_Amount is the log of the total value of the issued bonds.

Variable	Mean	Median	Standard deviation	25th percentile	75th percentile	N
Offer Yield (%)	6.840	6.849	1.556	6.000	7.625	1269
Net Interest Cost (%)	6.980	6.881	1.665	6.044	7.772	1152
Spread to Treasury (%)	1.660	1.300	1.241	0.800	2.100	1270
z-score	−9.323	2.044	67.076	−0.840	5.047	362,212
Size	5.021	5.111	2.691	3.277	6.825	512,908
Leverage	0.306	0.184	0.518	0.022	0.382	480,904
Coverage Ratio	7.410	4.692	105.145	1.006	11.653	306,416
Tangibility	0.260	0.160	0.265	0.042	0.415	496,767
Market-to-Book	3.315	1.414	7.657	1.057	2.456	446,445
Current Ratio	3.213	1.744	5.157	1.019	3.148	423,255
S&P Rating	15.171	15.000	3.040	14.000	17.000	1473
Log_Maturity	4.552	4.802	1.043	4.113	4.810	1483
Log_Amount	4.940	5.298	1.283	4.605	5.704	1486

4. Methodology and results

4.1. Empirical methodology

To empirically test the effect of violation on the three effects, I follow earlier literature that uses a regression model to capture the effect of violation (e.g. Freudenberg et al., 2017), and I use the description of violation as explained in section 2 to capture the incidence, timing and frequency effects. The regression model measures the effects of the three aspects of violations on the cost of new debt after controlling for firm and bond characteristics outlined in the previous section. In addition, one of the specifications of the cost of new debt, namely, Spread to Treasury, allows a difference-in-differences approach following Qiu and Yu (2009). This approach of employing three separate measures of the cost of new debt allows the results to be robust to the estimation methodology. I employ the following basic regression specification

y_{λ i j k} = α_{0} + γ V_{i λ} + θ X_{i} + δ K_{j} + φ Z_{k} + ε_{λ i j k}

(1)

where λ, i, j and k represent the violation specifications, firm, bond and year-quarter controls, respectively. Among the included variables, y denotes the cost of new debt, V represents the violation for the three aspects as explained in section 2,⁶ X firm-level control variables, K bond characteristics, Z other control variables and includes dummy variables for the year-quarter of the bond issue and the industry of the borrower, and ε represents the independently and identically distributed disturbances.

To gauge the relative importance of the incidence, timing and frequency effects for firms in competitive markets, I employ equation (1) for firms in competitive and concentrated markets. The regression is run separately on the two samples and I further run a test for the equality of the coefficients to see if the cost of new debt is different for the two classifications of firms. Furthermore, for robustness I augment equation (1) with the interaction between HHI and V dummy specifications following Valta (2012)⁷

y_{λ i j k} = α_{0} + γ (H H I \times V)_{i λ} + θ X_{i} + δ K_{j} + f Z_{k} + ε_{λ i j k}

(2)

This approach allows us to further condition the aforementioned effects for firms operating in competitive markets and to detail the coefficient for control variables for the complete sample.

4.2. Empirical results

The study uses the baseline regression (equation 1), which provides the estimate of the effects of the distinct outlined specifications of violation on the cost of new debt for the firms in the sample. The borrower and bond characteristics control for the feedback on the cost of new debt, and the results are reported in Tables 4 to 6 for incidence (VIOL), timing (Viol_t and Viol_t_− 1 jointly) and frequency (Viol = 1 and Viol > 1 jointly and Viol = n), respectively, across measures of the cost of new debt (Offer Yield, Net Interest Cost, Spread to Treasury). The battery of borrower and bond characteristics, and control variables into the regression equation, provides a spread of R-squared ranging from 0.59 to 0.68, suggesting that these variables help explain the variation in the cost of new debt. The sizes of the coefficients for debt covenant violation represent the answer to the empirical questions structured earlier.

Table 4.

Incidence effect. This table presents the regression results for equation (1) for the incidence effect. Offer Yield is the yield offered to investors at the time of bond issue. Net Interest Cost is the overall interest expense that is associated with the bond and is based on the average coupon rate weighted to years of maturity adjusted for any associated discounts or premiums. Spread to Treasury is the difference in the bond yield and risk-free treasury security yield with similar maturity. VIOL is a binary variable that equals one for violating firms and zero otherwise. z-score is the Altman z-score used to predict corporate defaults and is computed differently for manufacturing and non-manufacturing firms (see Supplementary Appendix). Size is the natural log of the total quarterly assets of the firm in year 2000 dollars; Leverage is the ratio of the book value of long-term debt plus long-term debt in current liabilities to the book value of total assets of the firm; Coverage is the ratio of earnings before interest and taxes to the interest expense; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets; Current Ratio is the ratio of current assets to current liabilities; S&P Rating is the S&P rating for bonds and has been converted to a number (see Supplementary Appendix); Log_Maturity is the natural log of the maturity of the loan in months. Log_Loan_Amount is the natural log of the proceeds from bond issuance. All regressions include year-quarter dummies. Figures in parentheses are robust clustered standard errors.

	Offer Yield	Net Interest Cost	Spread to Treasury
VIOL	0.475*** (0.124)	0.611*** (0.144)	0.369*** (0.105)
z-score	0.00302 (0.045)	0.0206 (−0.0493)	0.00004 (0.040)
Size_t _− 1	−0.158*** (0.045)	−0.123** (−0.0513)	−0.136*** (0.043)
Leverage_t _− 1	0.196 (0.353)	−0.0349 (0.393)	0.0999 (0.346)
Coverage_t _− 1	0.00191*** (0.001)	−0.000202 (0.001)	0.000397 (0.001)
Tangibility_t _− 1	−0.0466 (0.232)	−0.198 (0.268)	−0.362* (0.206)
Market-to-Book_t _− 1	−0.0775 (0.059)	−0.0275 (0.060)	−0.0764 (0.048)
Current Ratio_t _− 1	0.126* (0.068)	0.121 (0.076)	0.0712 (0.065)
S&P Rating	−0.226*** (0.028)	−0.274*** (0.034)	−0.224*** (0.020)
Log_Maturity	0.188*** (0.065)	0.320* (0.178)	−0.00291 (0.026)
Log_Loan_Amount	0.125*** (0.041)	0.109** (0.050)	0.0955** (0.040)
Constant	8.494***	8.464***	5.514***
Observations	732	660	724
R-squared	0.66	0.674	0.596

***

p < 0.01, **p < 0.05, *p < 0.1.

Table 5.

Timing effect. This table presents the regression results for equation (1) for the timing effect. Offer Yield is the yield offered to investors at the time of bond issuance. Net Interest Cost is the overall interest expense that is associated with the bond and is based on the average coupon rate weighted to years of maturity adjusted for any associated discounts or premiums. Spread to Treasury is the difference in the bond yield and risk-free treasury security yield with similar maturity. Viol_t is a binary variable that equals one if the firm reported a debt covenant violation in the quarter of bond issuance and zero otherwise. Viol_t_− 1 is a binary variable that equals one if the firm reported a debt covenant violation in the quarter preceding the bond issue and zero otherwise. z-score is the Altman z-score used to predict corporate defaults and is computed differently for manufacturing and non-manufacturing firms (see Supplementary Appendix). Size is the natural log of the total quarterly assets of the firm in year 2000 dollars; Leverage is the ratio of book value of long-term debt plus long-term debt in current liabilities to the book value of total assets of the firm; Coverage is the ratio of earnings before interest and taxes to the interest expense; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets; Current Ratio is the ratio of current assets to current liabilities; S&P Rating is the S&P rating for bonds and has been converted to a number (see Supplementary Appendix); Log_Maturity is the natural log of the maturity of the loan in months. Log_Loan_Amount is the natural log of the proceeds from bond issuance. All regressions include year and quarter dummies. Figures in parentheses are robust clustered standard errors.

	Offer Yield	Net Interest Cost	Spread to Treasury
Viol_t	0.359 (0.363)	0.227 (0.407)	0.402 (0.445)
Viol_t _− 1	1.048* (0.579)	1.214* (0.688)	0.874 (0.585)
z-score	−0.0138 (0.044)	−0.0284 (0.048)	−0.00154 (0.043)
Size_t _− 1	−0.178*** (0.044)	−0.159*** (0.048)	−0.134*** (0.042)
Leverage_t _− 1	0.0834 (0.356)	−0.0297 (0.371)	0.129 (0.350)
Coverage_t _− 1	0.00146* (0.001)	0.000127 (0.001)	0.000515 (0.001)
Tangibility_t _− 1	0.0271 (0.230)	−0.163 (0.251)	−0.316 (0.210)
Market-to-Book_t _− 1	−0.0994 (0.062)	−0.0819 (0.061)	−0.114** (0.050)
Current Ratio_t _− 1	0.145** (0.065)	0.152** (0.069)	0.0749 (0.061)
S&P Rating	−0.230*** (0.028)	−0.257*** (0.025)	−0.229*** (0.020)
Log_Maturity	0.145** (0.068)	0.262 (0.190)	0.00132 (0.033)
Log_Loan_Amount	0.0886** (0.038)	0.0855** (0.043)	0.0816*** (0.030)
Constant	9.272***	11.10***	5.617***
Observations	696	630	693
R-squared	0.634	0.676	0.592

***

p < 0.01, **p < 0.05, *p < 0.1.

Table 6.

Frequency effect. This table presents the regression results for equation (1) for the frequency effect. Offer Yield is the yield offered to investors at the time of bond issuance. Net Interest Cost is the overall interest expense that is associated with the bond and is based on the average coupon rate weighted to years of maturity adjusted for any associated discounts or premiums. Spread to Treasury is the difference in the bond yield and risk-free treasury security yield with similar maturity. Viol = 1 is a binary variable that equals one if a firm has reported one debt covenant violation in the period from 1996 to 2008 and zero otherwise. Viol > 1 is a binary variable that equals one if a firm has reported more than one covenant violation either in consecutive or in intermittent quarters and zero otherwise. Viol = 1 and Viol > 1 are measured jointly in equation (1). The results are reported in the first three columns of the table. Viol = n is the total number of violations reported by the firm and is measured separately in equation (1). The results are in the last three columns of the table. z-score is the Altman z-score used to predict corporate defaults and is computed differently for manufacturing and non-manufacturing firms (see Supplementary Appendix). Size is the natural log of the total quarterly assets of the firm in year 2000 dollars; Leverage is the ratio of the book value of long-term debt plus long-term debt in current liabilities to the book value of the total assets of the firm; Coverage is the ratio of earnings before interest and taxes to the interest expense; Tangibility is the ratio of property, plants and equipment to the total assets of the firm; Market-to-Book is the ratio of the market value of assets to the book value of assets; Current Ratio is the ratio of current assets to current liabilities; S&P Rating is the S&P rating for bonds and has been converted to a number (see Supplementary Appendix); Log_Maturity is the natural log of the maturity of the loan in months. Log_Loan_Amount is the natural log of the proceeds from bond issuance. All regressions include year and quarter dummies. Figures in parentheses are robust clustered standard errors.

	Offer Yield	Net Interest Cost	Spread to Treasury	Offer Yield	Net Interest Cost	Spread to Treasury
Viol = 1	0.498*** (0.180)	0.593** (0.234)	0.236 (0.152)	−	−	−
Viol > 1	0.486*** (0.140)	0.522*** (0.149)	0.443*** (0.128)	−	−	−
Viol = n	−	−	−	0.0577*** (0.021)	0.0700*** (0.025)	0.0485*** (0.019)
z-score	0.00156 (0.043)	0.0209 (−0.047)	0.00231 (0.040)	−0.014 (0.041)	0.0204 (0.052)	−0.00642 (0.040)
Size_t _− 1	−0.164*** (0.044)	−0.152*** (0.051)	−0.143*** (0.044)	−0.186*** (0.042)	−0.145*** (0.047)	−0.156*** (0.040)
Leverage_t _− 1	0.026 (0.344)	0.0738 (0.366)	0.163 (0.337)	0.295 (0.339)	−0.0866 (0.419)	0.0272 (0.348)
Coverage_t _− 1	0.00141* (0.001)	−0.000157 (0.001)	0.000465 (0.001)	0.00230*** (0.001)	−0.000151 (0.001)	0.000411 (0.001)
Tangibility_t _− 1	−0.0259 (0.213)	−0.167 (0.244)	−0.29 (0.197)	−0.15 (0.196)	−0.162 (0.261)	−0.337* (0.194)
Market-to-Book_t _− 1	−0.0551 (0.059)	−0.0253 (0.061)	−0.0743 (0.048)	−0.138** (0.057)	−0.064 (0.064)	−0.0909* (0.048)
Current Ratio_t _− 1	0.133* (0.069)	0.108 (0.072)	0.0655 (0.066)	0.117* (0.062)	0.123 (0.075)	0.0697 (0.063)
S&P Rating	−0.225*** (0.028)	−0.271*** (0.033)	−0.222*** (0.021)	−0.219*** (0.022)	−0.284*** (0.033)	−0.232*** (0.020)
Log_Maturity	0.193*** (0.068)	0.306* (0.173)	0.0145 (0.028)	0.159*** (0.060)	0.312 (0.194)	−0.00253 (0.027)
Log_Loan_Amount	0.121*** (0.043)	0.154*** (0.049)	0.103*** (0.038)	0.120*** (0.032)	0.0865* (0.046)	0.0923** (0.039)
Constant	8.378***	8.344***	5.385***	8.918***	9.153***	5.889***
Observations	727	658	723	732	660	724
R-squared	0.654	0.672	0.595	0.68	0.66	0.592

***

p < 0.01, **p < 0.05, *p < 0.1.

Among the explanatory variables, most of the estimated coefficients are consistent with earlier studies. z-score and leverage are not significant and explain that the firms are not necessarily at risk of default (Nini et al., 2012) or highly leveraged (Binsebergen et al., 2010). Size, tangibility, market-to-book and S&P ratings have negative coefficients and reduce the cost of debt (Anderson et al., 2004). Coverage does not display a consistent relationship across the differing specifications of the cost of new debt and appears to be an insignificant determinant in the sample. Current Ratio, Log_Maturity and Log_Loan_Amount generally display a positive relationship across the specifications of the cost of new debt. While the coefficient for current ratio is unexpected, the results for Log_Maturity and Log_Loan_Amount are consistent with it being a measure of illiquidity (Sarig and Warga, 1989).

4.3. Incidence effect

The preliminary findings in the summary statistics document that violating firms incur a higher cost of new debt. As shown by Demiroglu and James (2010), covenant violations provide associated information regarding the performance of the firm. First, the covenant threshold conveys information to other market participants about expectations regarding the prospects and riskiness of the borrower (e.g. Diamond, 1991; Rajan, 1992). Second, contract design models (e.g. Gârleanu and Zwiebel, 2009) and collateral requirement models (e.g. Besanko and Thakor, 1987) imply that contract terms require borrowers to convey credible private information regarding a firm’s future. In the framework of these models, information asymmetry between borrowers and lenders regarding the borrower’s credit quality and risk-shifting opportunities determines the tightness of the covenant design. To the lender, the information content of a violation of such debt restrictions is straightforward: The borrower has not been able to meet its targets or stay on course for future projections, and hence that borrower is riskier. A further implication is that the violation results in lower firm value (e.g. Qiu and Yu, 2009), thereby increasing the cost of debt. Therefore, the incidence effect is certainly consistent with the baseline results reported in Table 4. The Offer Yield, Net Interest Cost and Spread to Treasury for a violating firm are higher by approximately 48, 61 and 37 basis points, respectively. The coefficients are significant at the 1% level. To summarize, the results find evidence that violating a debt covenant significantly increases the price of debt for firms.

4.4. Timing effect

A valid concern with respect to incidence effect is whether the cost of new debt is higher because of the violation or because the firm is fundamentally weak. I examine the potentially dynamic effect of the timing of the violation and the issuance of the bond. In other words, I am interested in the effect of violation on the cost of new debt, and I measure the information content of the reported violation, which results in reducing information asymmetry between the firm and the bondholder. As evidenced earlier, firm-specific indicators deteriorate in the quarters leading up to a violation; however, bondholders do not learn of the actual violation until the end of the quarter when the violation is reported to the SEC. Therefore, I measure the cost of debt for bonds issued in the quarter of the violation and dynamically set that cost for comparison with firms that issued bonds in the quarter following the violation. I follow the model of information between borrowers and lenders developed by Gârleanu and Zwiebel (2009), who contend that covenants are designed to reduce information asymmetry between borrowers and lenders, and the reporting of a covenant violation would make the violating firm riskier for the lender. The results are listed in Table 5 and provide evidence for bond issuance in the violation quarter (Viol_t) and bond issuance in the post-violation quarter (Viol_t_− 1). The Offer Yield, Net Interest Cost and Spread to Treasury for a firm reporting a violation in the bond issue quarter, though higher by approximately 36, 23 and 40 basis points, respectively, are not statistically significant. The results provide evidence that while bondholders are restrained owing to deteriorating firm characteristics, the cost of new debt does not increase, as the violation is not known with certainty in the bond issue quarter. Subsequently, I find an increase in the cost of new debt when the bond is issued in the post-violation quarter, and I identify two important characteristics. First, I find that the Offer Yield, Net Interest Cost and Spread to Treasury for a firm reporting a violation in the quarter preceding the bond issue are higher by approximately 105, 121 and 87 basis points, respectively. Second, the increase in the cost of borrowing is higher than the increase in cases where the violation occurs in the bond issue quarter. The difference is approximately 69 basis points for Offer Yield, 98 basis points for Net Interest Cost and 47 basis points for Spread to Treasury. The coefficients are significant and economically substantial for Offer Yield and Net Interest Cost but not for Spread to Treasury. To summarize, I find that the actual reporting of violation reduces information asymmetry and acts as a signal regarding the financial health of the firm. Bondholders therefore demand a higher return once they know of the violation with certainty.

4.5. Frequency effect

So far, this study has identified an unconditionally positive relationship between violation and the cost of new debt issuance. Further analysis finds evidence for the incidence and timing effects, both of which articulate the relative importance of violation on the cost of new debt. The collective analyses suggest that these two independent effects dominate how bondholders view the wealth implications of violation. The frequency effect, however, is consistent with Beneish and Press (1995), who suggest that investors do not immediately incorporate covenant violation information into their assessments of the financial health of the firm. Nini et al. (2012) follow up on their argument and reason that firms violating a restriction and correcting it in the following quarter should benefit from creditors’ discretionary right to waive the penalties of covenant violation because the violation is corrected before the information can be incorporated into investors’ analyses. Violations, however, are costly as firms engage in accrual-based and real earnings management to avoid such events. Therefore, this study needs to be especially careful about the robustness of the frequency effect and cannot merely estimate it for firms that violate only once. Following the earlier approach, I examine not only one-time violators but also firms with multiple violations. For completeness, I develop the approach such that it considers the effect of multiple violations and additionally measures the cost of new debt for each violation.

I measure single and multiple violators jointly and report the results in the first three columns of Table 6. The Offer Yield, Net Interest Cost and Spread to Treasury for one-time violators are higher by approximately 50, 59 and 24 basis points, respectively. The estimated coefficient is not significant for Spread to Treasury, but the results suggest that one-time violations increase the cost of new debt issue for the firm. The follow-up for multiple violations in the sample outlines two important characteristics of the frequency effect. First, the Offer Yield, Net Interest Cost and Spread to Treasury for repeat offenders are higher by approximately 49, 52 and 44 basis points, respectively. Second, for repeat offenders, the cost of new debt issuance is significant at the 1% level across all specifications of the cost of new debt. To lend robustness to frequency effect, I also estimate the increase in the cost of new debt issue for each violation and find that the Offer Yield, Net Interest Cost and Spread to Treasury increase by approximately 6, 7 and 5 basis points, respectively, significant at the 1% level. This stepwise approach to the number of violations provides unequivocal evidence for the frequency effect.

4.6. HHI

Firms often face competition in the product market, which affects their operating decisions, the risk of their business environment and their liquidation value (Valta, 2012), which means that the competitive nature of the business in which a firm operates could also be a determinant of the severity of the incidence, frequency and timing effects on the cost of new debt. There are several reasons why competition can affect the cost of new debt through a firm’s default risk, consequently magnifying the aforementioned effects. First, competition reduces market power, profits and pledgeable income while increasing cash flow risk, making it more difficult for borrowers to raise funds (e.g. Tirole, 2006). This argument implies that the cost of debt should increase with the intensity of competition, thereby amplifying the violation effects. Second, firms are constantly facing competition and the threat of rapacity by rival firms. For instance, the business risk of the incumbent firm, having limited access to external finance, can increase significantly when financially strong firms adopt aggressive pricing strategies (e.g. Fre’sard, 2010). Accordingly, underinvestment and missed opportunities have implications for financing and predation risk (e.g. Haushalter et al., 2007). In addition, the liquidation value of assets is of central importance if a firm defaults. In particular, higher liquidation value lowers the cost of liquidation, increases firms’ debt capacity and reduces the promised debt yield for a given debt level (e.g. Benmelech and Bergman, 2009). In this case, competition could significantly affect the financial strength of the buyers and hence the asset liquidity of the industry (e.g. Ortiz-Molina and Phillips, 2014), thereby amplifying the incidence, timing and frequency effects on the cost of new debt for firms operating under competitive settings. To test this, I employ the HHI, defined as the sum of squared market share at the industry level, as a proxy for the level of industry competition. In a perfectly competitive industry with atomistic firms, the HHI is equal to zero; in a monopolistic industry, the Herfindahl index is equal to one. As a first measure, I segregate the data according to competitive and concentrated firms and assign the lowest quartile firms to competitive markets (e.g. Valta, 2012). The data are then tested for incidence, timing and frequency effects by employing equation (1).

The results are listed in Table 7 and only report the coefficients for the specifications of violation described in section 2. The estimates provide evidence for the intensity of the incidence effect in competitive markets with Offer Yield, Net Interest Cost and Yield to Maturity being higher by 59, 71 and 43 basis points, respectively. The results are statistically significant at the 1% level and are economically larger than the measure for the entire sample. Alternatively, the measures of the cost of new debt are largely not significant for firms operating in monopolistic markets. I also test for the difference in the estimated coefficients across the two classifications of firms and find that the coefficients are significantly different at the 1% level. The results provide evidence that market competition magnifies the incidence effect on the cost of new debt for violating firms and provide robustness for my earlier analysis.

Table 7.

Regression results: Herfindahl–Hirschman index. This table provides only the coefficient for incidence (VIOL), timing (Viol_t and Viol_t_− 1 jointly) and timing (Viol = 1 and Viol > 1 jointly and Viol = n separately) effects for Offer Yield, Net Interest Cost and Spread to Treasury for competitive and concentrated firms under the Herfindahl–Hirschman index. VIOL is a binary variable that equals one for violating firms and zero otherwise. Viol_t is a binary variable that equals one if the firm reported a debt covenant violation in the quarter of bond issue and zero otherwise. Viol_t_− 1 is a binary variable that equals one if the firm reported a debt covenant violation in the quarter preceding the bond issue and zero otherwise. Viol_t and Viol_t_− 1 are measured jointly in the nested model using equation 1. Viol = 1 is a binary variable that equals one if a firm has reported one debt covenant violation in the period from 1996 to 2008 and zero otherwise. Viol > 1 is a binary variable that equals one if a firm has reported more than one covenant violation either in consecutive or in intermittent quarters and zero otherwise. Viol = 1 and Viol > 1 are measured jointly in the nested model using equation (1). Viol = n is the total number of violations reported by the firm. Figures in parentheses are robust clustered standard errors.

	Offer Yield		Net Interest Cost		Spread to Treasury
	Competitive	Monopolistic	Competitive	Monopolistic	Competitive	Monopolistic
Incidence effect
VIOL	0.594*** (0.130)	0.18 (0.250)	0.706*** (0.150)	0.39 (0.270)	0.431*** (0.120)	0.34* (0.200)
Timing effect (nested model)
Viol_t	0.589 (0.465)	−0.453 (0.462)	0.756 (0.545)	−0.422 (0.446)	1.065 (0.749)	−0.219 (0.394)
Viol_t_− 1	1.48** (0.690)	0.91* (0.535)	1.821*** (0.888)	0.884 (0.586)	1.603** (0.684)	0.3 (0.586)
Frequency effect (nested model)
Viol = 1	0.513** (0.245)	0.356 (0.259)	0.833*** (0.295)	0.125 (0.348)	0.255 (0.183)	0.176 (0.228)
Viol > 1	0.586*** (0.139)	0.335 (0.362)	0.725*** (0.157)	0.569 (0.382)	0.492** (0.139)	0.404 (0.307)
Frequency effect
Viol = n	0.0699*** (0.020)	0.03 (0.040)	0.08*** (0.020)	0.08 (0.050)	0.06*** (0.020)	0.05 (0.040)

***

p < 0.01, **p < 0.05, *p < 0.1.

The results of the timing effect are also consistent with the discussion above. The cost of new debt does not increase for either classification of the firms if the debt was issued in the violation quarter. As expected, I find contrary results for timing effect if the bond was issued in the post-violation quarter. For firms in competitive markets, Offer Yield, Net Interest Cost and Yield to Maturity are higher by 142, 182 and 160 basis points, respectively. The results are statistically significant across the specifications and are economically larger than those for the entire sample. In contrast, I find that the timing effect is largely non-existent for firms operating in monopolistic markets, with the effect only being significant for Offer Yield at the 10% level; the coefficient is also statistically and economically smaller than that for competitive firms. The test for the difference of coefficients also reveals that the coefficients are significantly different for competitive and monopolistic firms at the 1% level. Overall, this article provides strong evidence that competition amplifies the timing effect on the cost of new debt.

Consistent with the discussion of the frequency effect, this article finds evidence of a higher cost of new debt for competitive markets. I find statistical evidence for the frequency effect for one-time violators, with Offer Yield and Net Interest Cost being statistically significant for firms operating in competitive markets. The effect is non-existent for the firms operating in concentrated markets. For multiple violators, the frequency effect is profound for firms in competitive markets, with Offer Yield, Net Interest Cost and Yield to Maturity being higher by 59, 72 and 49 basis points, respectively. The results are statistically significant and economically larger than those for the entire sample. For firms in concentrated industries, however, the frequency effect is non-existent and identifies the market power of comparatively monopolistic markets. I also test for the difference in coefficients and find that the coefficients are significantly different across competitive and monopolistic firms at the 1% level. For completeness, I also estimate the frequency effect per violation and find statistically significant results for competitive firms with Offer Yield, Net Interest Cost and Yield to Maturity increasing by 7, 8 and 6 basis points, respectively. In contrast, I find that the frequency effect is non-existent for firms operating in concentrated markets. The results highlight the impact of this aspect of violation on the cost of new debt in competitive markets.

To summarize, I have attempted to identify three different effects of violation on the cost of new debt and have embodied the effects under distinct market competition. All three predict an increase in the cost of new debt after the violation and provide insight into the different ways through which violations affect the cost of new debt. This article identifies positive and significant overall results and extends the literature on covenant violation. The effects are more profound for firms operating in competitive markets, with firms operating under comparatively monopolistic markets having either a non-existent or a negligible increase in the cost of new debt.

5. Robustness

In this section, I consider the robustness of the results under alternative econometric specifications by employing equation (2). The aim is to identify and measure the effects of the violation specifications for firms operating in competitive markets and to provide robust evidence for the incidence, timing and frequency effects.

For the incidence effect, the results reveal a significant increase of 37–68 basis points in the cost of new debt among the three specifications for competitive firms. Second, the results highlight that the intensity of the incidence effect is primarily greater for competitive firms when comparing the results with those of the entire sample highlighted in section 4.3. These results are coherent with the earlier discussion and findings.

The study further employs this approach to measure the intensity of the timing effect for competitive firms. First, I find that the cost of new debt does not increase if the bond was issued in the violation quarter. This is in line with the argument that covenant violation filing has important debt-pricing implications. Second, I find that for competitive firms the cost of new debt increases by 123, 159 and 150 basis points for Offer Yield, Net Interest Cost and Spread to Treasury, respectively, if the bond was issued in the post-violation quarter. The results are consistent with the findings in section 4.6. Third, I find evidence for the intensity of the timing effect to be greater for firms operating in competitive markets when collating the findings in section 4.4. Overall, the results are in line with the assertion that the timing effect has stronger bearings on firms in competitive markets.

Finally, I address the intensity of the frequency effect for competitive firms by employing equation (2). First, I find that the frequency effect for one-time violators increases the cost of new debt by 25–67 basis points, with two of the three specifications being statistically significant. This is in line with the outcome in section 4.5 and highlights that the cost is comparatively higher for one-time violators in competitive industries. For multiple violators, I find that the cost of new debt increases by 54, 62 and 42 basis points for Offer Yield, Net Interest Cost and Spread to Treasury, respectively. The frequency effect appears to have stronger implications for multiple violators in competitive markets when compared with the results in section 4.5. Finally, the findings demonstrate that the increase in the cost of new debt for competitive firms per violation is higher by 5–8 basis points. As a whole, the results find the frequency effect to be acutely aligned to the competitive nature of markets.

To summarize, this approach lends robustness to the discussion of the cost of new debt, covenant violation and market competition. It provides unequivocal evidence that the incidence, timing and frequency effects are prevalent and more acute for firms operating in competitive markets.

6. Conclusion

The violation of contractual agreements has often been considered as an indicator of financial distress. Such events reduce information asymmetry between bondholders and creditors regarding the risk and financial health of the firm. Recent empirical research finds that periods of economic turbulence incite managers to manipulate earnings (Ahmad-Zaluki et al., 2011). However, our understanding of the costs of such violations remains unclear. The goal of this article is to provide empirical estimates of the cost of such violations and to extend the literature by quantifying the penalties on future credit.

The important outcome of this study is that it documents the cyclical nature of violations, with 38% of firms reporting a violation in at least one quarter, and firms in the middle quartiles of size, tangibility and market-to-book reporting a higher frequency of violation. In addition, the financial health of a firm deteriorates in the quarters leading up to the quarter of violation and improves thereafter because of creditor intervention (Nini et al., 2012).

This article uses SEC filings of covenant violations reported between 1996 and 2008 and analyses their impact on the cost of new debt issuance. It augments earlier literature by measuring the impact of violations in competitive markets. The study employs three specifications of the cost of new debt and finds the results to be robust to such configurations. Using an estimation approach, it finds that the cost of new debt for firms that commit violations increases, on average, by 37–61 basis points. The incidence effect is more profound for firms in competitive markets. This article also presents evidence of the timing effect, in which the cost of new debt increases only once the violation is known with certainty. Finally, the evidence is consistent with creditors’ discretionary right to forgo the penalties of the violation for one-time violators. The cost of new debt is substantially higher, both significantly and economically, for repeat offenders. The timing effect is found to be more acute for firms in competitive markets. The increase in the cost of new debt per violation is estimated to be between 5 and 7 basis points. Empirical evidence compliments the acuteness of the frequency effect for competitive markets.

The results for the incidence, timing and frequency effects are robust to three different measures of the cost of new debt. The alternative econometric specification (equation (2)) finds robustness for the results for competitive markets. Building on earlier literature, these results are the first to provide explicit estimates of the cost of covenant violations. A more precise qualification of the incidence, timing and frequency effects, however, must be conditioned on the firm’s economic environment. This analysis details the cost of such effects in competitive markets and shows that market competition is an important conditioning variable.

Supplemental Material

AJM-17-0269.R2_Main_Doc_with_AE-AD – Supplemental material for Debt covenant violation, competition and cost of new debt

Supplemental material, AJM-17-0269.R2_Main_Doc_with_AE-AD for Debt covenant violation, competition and cost of new debt by Umar Butt in Australian Journal of Management

Footnotes

Acknowledgements

I would like to thank Amir Sufi, Professor of Finance, Chicago Booth, Chicago, IL, for providing access to the debt covenant violation data. I would also like to thank Trevor Chamberlain, Anna Danielova, Sudipto Sarkar and Jiaping Qiu for their valuable comments and suggestions. The remaining errors are the sole responsibility of the author.

Final transcript accepted 10 September 2018 by Sue Wright (AE Accounting).

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

Notes

References

Ahmad-Zaluki

Campbell

Goodacre

(2011) Earnings management in Malaysian IPOs: The East Asian crisis, ownership control, and post-IPO performance. The International Journal of Accounting 46: 111–137.

Altman

(2000) Predicting financial distress of companies: Revisiting the Z-Score and ZETA models. Working Paper, New York University, New York, July.

Anderson

Mansi

Reeb

(2004) Board characteristics, accounting report integrity, and the cost of debt. Journal of Accounting and Economics 37: 315–342.

Apostolou

Dorminey

(2014) The association of departures from spending rate equilibrium to municipal borrowing cost. Advances in Accounting 30: 1–8.

Beneish

Press

(1993) Costs of technical violations of accounting-based debt covenants. The Accounting Review 68: 233–257.

Beneish

Press

(1995) Interrelation among events of default. Contemporary Accounting Research 12: 57–84.

Benmelech

Bergman

(2009) Collateral pricing. Journal of Financial Economics 91: 339–360.

Benmelech

Garmaise

Moskowitz

(2005) Do liquidation values affect financial contracts? Evidence from commercial loan contracts and zoning regulation. Quarterly Journal of Economics 120: 1121–1154.

Besanko

Thakor

(1987) Collateral and rationing: Sorting equilibria in monopolistic and competitive credit markets. International Economic Review 28: 671–689.

10.

Bhaskar

Gopal

(2016) Debt covenant violations, firm financial distress, and auditor actions. Contemporary Accounting Research 34: 186–215.

11.

Binsebergen

JHV

Graham

Yang

(2010) The cost of debt. The Journal of Finance 65: 2089–2136.

12.

Bolton

Freixas

(2000) Equity, bonds, and bank debt: Capital structure and financial market equilibrium under asymmetric information. Journal of Political Economy 108: 324–351.

13.

Bradley

Roberts

(2015) The structure and pricing of corporate debt covenants. Quarterly Journal of Finance 5: 1–37.

14.

Butt

Chamberlain

Sarkar

(2016) Accruals and real earnings management: Testing the debt covenant hypothesis. International Journal of Accounting and Finance 6: 167–196.

15.

Collin-Dufresn

Goldstein

Martin

(2001) The determinants of credit spread changes. The Journal of Finance 56: 2177–2207.

16.

DeFond

Jiambalvo

(1994) Debt covenant violation and manipulation of accruals. Journal of Accounting and Economics 17: 145–176.

17.

Demiroglu

James

(2010) The information content of bank loan covenants. Review of Financial Studies 23: 3700–3737.

18.

Denis

Mihov

(2003) The choice among bank debt, non-bank private debt, and public debt: Evidence from new corporate borrowings. Journal of Financial Economics 70: 3–28.

19.

Diamond

(1991) Monitoring and reputation: The choice between bank loans and directly placed debt. Journal of Political Economy 99: 689–721.

20.

Dichev

Skinner

(2002) Large-sample evidence on the debt covenant hypothesis. Journal of Accounting Research 40: 1091–1122.

21.

Dyreng

(2009) The cost of private debt covenant violation. Working Paper, Duke University, Durham, NC, September.

22.

Ericsson

Jacobs

Oviedo

(2009) The determinants of credit default swap premia. The Journal of Financial and Quantitative Analysis 44: 109–132.

23.

Ertan

Karolyi

(2014) Debt covenants, expected default costs, and the implicit cost of financial reporting manipulation. Working Paper, London Business School, London, October.

24.

Fargher

Wilkins

Holder-Webb

(2001) Initial technical violations of debt covenants and changes in firm risk. Journal of Business Finance & Accounting 28: 465–480.

25.

Freudenberg

Imbierowicz

Saunders

et al . (2017) Covenant violations and dynamic loan contracting. Journal of Corporate Finance 45: 540–565.

26.

Fre’sard

(2010) Financial strength and product market behavior: The real effects of corporate cash holdings. The Journal of Finance 65: 1097–1122.

27.

Gao

Khan

Tan

(2017) Further evidence on consequences of debt covenant violations. Contemporary Accounting Research 34: 1489–1521.

28.

Gârleanu

Zwiebel

(2009) Design and renegotiation of debt covenant. Review of Financial Studies 22: 749–781.

29.

Haushalter

Klasa

Maxwell

(2007) The influence of product market dynamics on a firm’s cash holding and hedging behavior. Journal of Financial Economics 84: 797–825.

30.

Hillegeist

Elizabeth

Donald

et al . (2004) Assessing the probability of bankruptcy. Review of Accounting Studies 9: 5–34.

31.

Nini

Smith

Sufi

(2012) Creditor control rights, corporate governance, and firm value. Review of Financial Studies 25: 1713–1761.

32.

Ortiz-Molina

Phillips

(2014) Real asset illiquidity and the cost of capital. The Journal of Financial and Quantitative Analysis 49: 1–32.

33.

Pástor

Sinha

Swaminathan

(2008) Estimating the intertemporal risk–return tradeoff using the implied cost of capital. The Journal of Finance 63: 2859–2897.

34.

Qiu

(2009) The market for corporate control and the cost of debt. Journal of Financial Economics 93: 505–524.

35.

Rajan

(1992) Insiders and outsiders: The choice between informed and arm’s-length debt. The Journal of Finance 47: 1367–1400.

36.

Roberts

Sufi

(2009) Control rights and capital structure: An empirical investigation. The Journal of Finance 64: 1657–1695.

37.

Roychowdhury

(2006) Earnings management through real activities manipulation. Journal of Accounting and Economics 42: 335–370.

38.

Sarig

Warga

(1989) Some empirical estimates of the risk structure of interest rates. The Journal of Finance 44: 1351–1360.

39.

Sweeney

(1994) Debt covenant violations and managers’ accounting responses. Journal of Accounting and Economics 17: 281–308.

40.

Tirole

(2006) The Theory of Corporate Finance. Princeton, NJ: Princeton University Press.

41.

Valta

(2012) Competition and the cost of debt. Journal of Financial Economics 105: 661–682.

42.

Wilkins

(1997) Technical default, auditors’ decisions and future financial distress. Accounting Horizons 11: 40–48.

43.

Zang

(2012) Evidence on the trade-off between real activities manipulation and accrual-based earnings management. The Accounting Review 87: 675–703.

44.

Zhu

Gippel

(2015) Informed trading on debt covenant violations: The long and short of it. Accounting and Finance 57: 289–315.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.41 MB