An Empirical Analysis of Noncompetition Clauses and Other Restrictive Postemployment Covenants

Abstract

Employment contracts for most employees are not publicly available, leaving researchers to speculate about whether they contain postemployment restrictions on employee mobility, and if so, what those provisions look like. Using a large sample of publicly available CEO employment contracts, we are able to examine these noncompetition covenants, including postemployment covenants not to compete (CNCs or noncompetes), nonsolicitation agreements (NSAs), and nondisclosure agreements (NDAs). What we found confirms some long-held assumptions about restrictive covenants but also uncovers some surprises.We begin by discussing why employers use restrictive covenants and examining how the courts have treated them. We then analyze an extensive sample of CEO employment contracts drawn from a large random sample of 500 S&P 1500 companies. We find that 80% of these employment contracts contain CNCs, often with a broad geographic scope, and that these generally last only one to two years. Similarly, we find that NSAs routinely appear in these contracts, barring solicitation of the firm's employees and customers or clients. We demonstrate that NDAs are prevalent and prohibit the CEOs from disclosing unspecified In addition, we note that there is a strong California effect, whereby firms from that state are less likely to put CNCs in employment contracts.Our research also uncovers several previously undocumented trends. First, increasingly more and more restrictive covenants are appearing over time, and they are appearing with increasingly expansive enforcement rights for the firm.. Second, there is clear path dependence for these clauses, with a prior CNC being a strong predictor of CNC use in future employment contracts. Third, long-term contracts are more likely to have CNC clauses than short-term contracts, probably because firms have more confidence in making investments in CEOs that are committed to staying for longer periods. We argue that this shows that for some firms the risk of harm from a departing executive may simply be more acute than for other firms.I. IntroductionA departing key employee can seriously harm a firm by leaving to work for the firm's competitors or by providing these competitors with the firm's confidential business information. Firms, therefore, have strong incentives to seek to impose broad covenants not to compete and other postemployment restrictive covenants on these employees. However, most employment contracts are not publicly available, leaving researchers to speculate on the prevalence of these restrictions and their contents. Fortunately, public companies must disclose their CEOs' employment contracts. This mandatory disclosure permits us to examine them to determine how frequently, and how broadly, employers use covenants not to compete (CNCs or noncompetes), nonsolicitation agreements (NSAs), and nondisclosure agreements (NDAs).1In this paper, we first discuss what is currently known about the use and purpose of these restrictive covenants and why they remain controversial. We then analyze 874 CEO employment contracts initiated between 1996 and 2010 from a random sample of 500 S&P 1500 companies to develop the first comprehensive study of the prevalence and details of restrictive covenants in public company employment contracts. We find that 80% of these employment contracts contain CNCs that commonly last one to two years. Although 89% of the CNCs forbid CEOs from working for a competitor during the term of the CNC, only 25% prohibit CEOs from financing competitors. These covenants often have a broad geographic scope: for example, 38% bar the CEO from working in any jurisdiction where the company has operations.We also find that NSAs and NDAs frequently show up in CEO employment contracts. NSAs routinely appear in CEO contracts with CNCs and typically restrict CEOs' behavior for a period of one to two years: 75. …

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 52 citations exist upstream.