Corporate syndicated loan contracts frequently include financial covenants, namely provisions that require a borrowing corporation to periodically achieve specified accounting-based performance objectives or maintain specified accounting-based capital requirements. Practitioners, academics, and policymakers are interested in these provisions because they mitigate costly agency frictions that arise between borrowers and lenders, thereby enhancing the flow of capital in the economy.
Financial covenants reduce agency costs by creating a mechanism whereby rights of control shift from borrowers to lenders when performance deteriorates. For example, a borrower with a high amount of debt relative to its earnings might be in jeopardy of missing future debt payments, thereby losing control of the corporation. As prospects become increasingly bleak, the borrower might be tempted to make riskier and riskier investments. If those investments are successful, the borrower can fulfill debt payment obligations and retain control of the corporation. If those investments are instead unsuccessful, the borrower loses control of the corporation, something that might have happened anyway.[1]
Financial covenants reduce the incentive to make risky investments by providing early signals to the lender that the borrower’s performance is deteriorating, but before the borrower makes suboptimal investments. If the borrower fails to maintain a required level of performance, a covenant violation occurs. This violation triggers a technical default, which usually grants lenders the right to call the loan or take other legal action against the borrower. These rights generally force lender and borrower back to the negotiating table, and tend to tilt negotiating power in favor of the lender, allowing the lender to protect itself against the borrower’s natural tendency for risky investment.
Loan contracts can include different types of financial covenants, as illustrated by Ruby Tuesday’s revolving loan contract dated February 28, 2007. The contract binds the company to maintain three financial covenants on a quarterly basis – (1) a minimum fixed charge ratio of 2.0; (2) a maximum debt-to-earnings before interest, taxes, depreciation, and amortization (EBITDA) ratio of 3.25; and (3) a minimum net worth equal to $300 million. If Ruby Tuesday fails to maintain any of these three covenants, their syndicated lenders (which includes Bank of America and Citibank) could take legal action against them.
Because financial covenants are nearly ubiquitous in corporate syndicate loans and play an integral role in alleviating agency frictions between corporate borrowers and lenders, academicians have dedicated considerable effort to understanding both the determinants and consequences of covenant violations. A critical challenge faced by researchers is precisely measuring how close a corporate borrower is to a covenant violation (often referred to as covenant slack) and whether a covenant violation has occurred. Although this might seem like a simple exercise, it is challenging for at least two reasons.
First, syndicated loan contracts typically include non-standard financial definitions, which makes comparing loan contracts difficult. For example, the definition of debt and earnings numbers in Ruby Tuesdays’ debt-to-EBITDA covenant differ from the way debt and earnings numbers are defined in a syndicated loan granted to Cracker Barrell.
Second, syndicate loan contracts are frequently amended, including changes to the covenant definitions, the required covenant thresholds, or a combination of the two. While these events are common, data available to academics does not reflect these contractual amendments.
Despite these known challenges, researchers have often estimated covenant violations and/or a borrower’s distance from a violation using standardized covenant definitions, which does not account for the heterogeneous way in which financial covenants are defined and measured, and initial contractual terms, which fail to account for the occurrence of contractual amendments.
In our recent paper, we seek to understand the extent to which true measures regarding a borrower’s distance to a covenant violation differ from estimated measures currently used by academics. To do this, we begin by extracting information about financial covenants from corporate borrowers’ periodic filings (e.g., 10-Q and 10-K) that are publicly available on the Securities and Exchange Commission’s (SECs) website. For example, for its fiscal period ending on December 2, 2008, Ruby Tuesday reports its required covenant threshold for its debt-to-EBITDA covenant is 4.50 compared to its actual debt-to-EBITDA ratio of 4.22. Because the debt-to-EBITDA covenant imposes a maximum threshold, we determine that Ruby Tuesday’s true distance to a covenant violation (or covenant slack) for its debt-to-EBITDA covenant is 0.28 which is the 4.50 required threshold less the 4.22 actual ratio.
We next compare our true measures of distance to covenant violation with estimated measures used in prior academic research, which are based on standardized covenant definitions and initial contract terms. We document that the latter approach is likely to induce error. For example, these metrics indicate that Ruby Tuesday’s debt-to-EBITDA covenant requirement for the fiscal period ending December 2, 2008 amounts to 3.25, while the actual debt-to-EBITDA ratio is 5.96. Because the actual debt-to-EBITDA ratio of 5.96 exceeds the maximum allowable estimated threshold, the researcher would identify Ruby Tuesday as being in technical default, i.e., as having violated their debt-to-EBITDA covenant, when they were not.
In our study, we show that the Ruby Tuesday’s example is not an exception: commonly used methods to estimate the distance to a covenant violation result in an overestimation of the frequency of violations by as much as 2,500% relative to using true measures, with the magnitude of the error, depending on the covenant type, included in the contract. We also show that this measurement error is not random, as overestimation of violations are much more common than underestimation of violations – that is, failure to estimate a covenant violation when a true violation has occurred. Overall, these results suggest that prior approaches to identify covenant violations can induce severe measurement error, which cast doubts on the validity of some results published in prior literature.
We next seek to determine whether the primary source of the measurement error problem associated with estimated measures of covenant slack is due to measurement error in the required covenant threshold or the realized covenant ratio. To clarify, in Ruby Tuesday’s case overestimating a covenant violation associated with its debt-to-EBITDA covenant could arise because the estimated covenant threshold differs from the true required threshold (3.25 vs. 4.50), or because the estimated debt-to-EBITDA ratio differs from the actual debt-to-EBITDA ratio (5.96 vs. 4.22), or a combination of the two. Our analysis suggests that the primary source of measurement error comes from covenant ratio realizations, not covenant thresholds. More specifically, using true covenant ratios in combination with estimated thresholds reduces the overestimation of covenant violations by 79%, whereas using true required thresholds in combination with estimated covenant ratios only reduces the overestimation of covenant violations by 2%.
Given that extracting covenant information from borrower’s periodic filings is time-intensive and possibly unfeasible for large-sample studies, we suggest adjustments to standardized covenant definitions that researchers can rely upon to mitigate measurement error. We show that such adjustments – including adjustments for share-based compensation and pension expense – in the calculation of EBITDA can reduce the overestimation of covenant violations for EBITDA-based covenants by as much as 25%.
Collectively, our findings suggest that estimated measures of covenant violations grossly overstate the frequency of covenant violations, which may inhibit academics’ and policymakers’ ability to draw meaningful inferences from analyses that rely upon these estimated measures. To overcome these challenges, we make several recommendations to adjust standardized covenant definitions if researchers choose to use these measures. Finally, we offer our hand-collected data that is devoid of measurement error as a source for future research.
Scott Dyreng is a Professor of Accounting at the Fuqua School of Business at Duke University.
Elia Ferracuti is an Assistant Professor of Accounting at the Fuqua School of Business at Duke University.
Robert Hills is an Assistant Professor of Accounting at Penn State University.
Matt Kubic is an Assistant Professor of Accounting at the University of Texas at Austin.
This post is adapted from their paper, “Measurement Error when Estimating Covenant Violations” available on SSRN.
[1] A famous example of this behavior is the Fred Smith, founder of FedEx who, during the early days of the company, took the company’s last $5,000 to Las Vegas and won $27,000 gambling on blackjack to cover the company’s $24,000 fuel bill.
