C&B Notes

EBITDA Definition Danger

EBITDA calculations vary widely across businesses and industries, and loosey-goosey “adjusted” EBITDA definitions exacerbate risk for lenders who are already offering covenant-lite terms.  As this recent study reveals, 5.0x debt-to-EBITDA ratios are not all created equal when the denominator is calculated differently.  This issue is particularly acute in private equity.

We show that the degree to which covenants vary in complexity and tightness has been substantially underestimated in the literature. Many of the major covenants in debt contracts today depend in some way on the borrower’s EBITDA, including in “covenant-lite” deals that do not include financial maintenance covenants. Yet this financial measure has no uniform definition across debt contracts. Originally intended as a measure of the borrower’s cash flows that is unaffected by financing choices, EBTIDA as defined in credit agreements has evolved to a highly individualized metric that may bear little relation to the borrower’s actual cash flow. In many cases, the definition grants to the borrower so much discretion as to what is included that the borrower has substantial leeway to avoid a covenant violation.

In this study we extract EBITDA definitions from 4,170 credit agreements and combine this text with characteristics of the deal and information about the borrowers, lead banks, private equity sponsors, and law firms that are parties to the deal. Using text analysis we are able to show the sources that these parties draw from when they craft this contractual language. We are also able to categorize how permissive these definitions are. To do so, we hand-code a subsample of roughly ten percent of the definitions and use this hand-coded sample to train a model to make out-of-sample predictions about the permissiveness of every definition in the sample. We use this information to develop five primary findings.

First, we find very wide variation among the EBITDA definitions in the data sample, contrary to the implicit assumptions underlying related empirical studies. Second, we find that these differences in language matter: there is significant variation in permissiveness as well, and the degree of permissiveness has increased overall over most of the sample period. Third, we identify factors that are associated with more permissive EBITDA definitions. Private equity sponsored deals are significantly more permissive than non-sponsored deals, all else equal. However, non-sponsored deals have been catching up to the private equity deals both in terms of covenant formulation and permissiveness, particularly in the wake of the 2013 Interagency Guidance on Leveraged Lending, which effectively barred banks from underwriting loans with leverage greater than six times EBITDA. Fourth, and relatedly, we show that there is a negative relationship between the permissiveness of EBITDA definitions and the amount of covenant slack in loan agreements, which suggests that the tightness of covenant levels and EBITDA definitions are substitutes. Finally, borrowers appear to value covenant uniformity across their own loan agreements through time more than standardization across similar borrowers and credits: a borrower’s past deals are highly predictive of the EBITDA definition in the current deal. Private equity sponsors also appear to predict content, to a lesser degree.

Referenced In This Post