ABSTRACT The annual proportion of U.S. public firms that reported a financial covenant violation fell roughly 70% between 1997 and 2019. To understand this trend, we develop an estimable model of covenant design that depends on the ability to differentiate between distressed and nondistressed borrowers and the relative costs associated with screening incorrectly. We find that the drop in violations is best explained by an increased willingness to forgo early detection of distressed borrowers in exchange for fewer inconsequential violations, which we attribute largely to a shift in the composition of public borrowers and partly to heightened investor sentiment during the 2010s.