Measurement Error when Estimating Covenant Violations

Abstract

We hand-collect true covenant realizations and thresholds from SEC filings and show that estimating covenant slack using realizations and thresholds from commercial databases frequently overestimates, yet rarely underestimates violations. We show this asymmetric measurement error, which is largely driven by differences between true and estimated realizations, can meaningfully affect inferences. For example, we document that measurement error is particularly concerning for research that relies on precisely identifying covenant violations (e.g., lender forbearance), and that true, but not estimated, violations are associated with negative market reactions and renegotiation outcomes. Finally, we propose ways to reduce measurement error.

Keywords: Debt covenants, Covenant Slack, Covenant Violation

JEL Classification: G30, M40, M41