Startups are an important source of job creation in the U.S., generating over 4 million jobs since the early 1980s. Although the majority of startups fail within the first few years, those that survive can have a disproportionate impact on job generation, as evidenced by companies such as Microsoft, Google, and Amazon.¹ The development and advancement of startups is often propelled by workers leaving their current jobs to create new businesses. However, employers generally want to avoid competing with former employees and thus frequently require their workers to sign noncompete agreements. The agreements typically bar former employees from starting or moving to a similar business for a set time period and within a specific geographic area.

In his paper, “Do Non-Compete Covenants Influence State Startup Activity? Evidence from the Michigan Experiment,” Philadelphia Fed emeritus economist Gerald Carlino analyzes whether noncompete agreements interfere with entrepreneurial activity. Noncompete agreements are used across industries throughout the U.S., but, as Carlino explains, the states, not the federal government, are responsible for enforcing these agreements, and the states differ with respect to the degree and manner of enforcement. To what extent does enforcing noncompete covenants hinder the information flows beneficial for startups?

According to recent literature, enforcement of noncompete agreements can affect startups in opposing ways. For example, some researchers have argued that enforcement of noncompete contracts can help startups protect their investments, thereby stimulating innovation and employment.² However, other researchers have argued that startups, many of which lack the resources to train employees and instead look to hire workers previously employed by their competitors, may choose to locate in states that don't enforce noncompetes.³ And some research indicates that knowledge workers migrate to nonenforcement states, triggering “brain drain” in the enforcement states. This research also indicates that, in localities where workers are covered by the agreements, knowledge spillovers are reduced because of limits on job-hopping and the sharing of information.⁴

To address these differing results in the literature, Carlino asks, To what extent did judicial enforcement of noncompete agreements affect Michigan’s startup entry rate and job creation rate? The state is an interesting case study because it did not begin enforcing noncompete agreements until the 1985 Michigan Antitrust Reform Act (MARA).⁵ Carlino employs an approach (difference-in-difference analysis) that compares a panel of data on startup activity in Michigan with similar data in various control groups, including the group of states sharing a land border with Michigan, and a group of 10 nonenforcement states that did not change their laws affecting the enforcement of noncompetes between 1977 and 2013.⁶ By comparing the state’s startup activity before and after the MARA, Carlino was able to identify the impact of noncompetes on entrepreneurial activity.

The authors found that the change in laws affecting noncompetes in Michigan had little or no impact on the entry rate of startups in Michigan when compared to any of the control groups. However, Carlino did find that enforcement of noncompete agreements in Michigan had a positive impact on startup job creation when compared to the control group of nonenforcement states — a doubling of enforcement led to about an 8 percent increase in startup job creation in the state.

Because the number of patents issued provides a useful indicator of entrepreneurial activity, Carlino also analyzed the granting of patents pre- and post-MARA. He found that enforcement of noncompete agreements positively affected the total number of patents issued to inventors in Michigan. Although the impact varied among six patent classifications, it was particularly strong in the mechanical category, the most important patent classification in the state.⁷

Next, Carlino examined Michigan’s high-tech sector, which is disproportionally covered by noncompete agreements. This sector drives U.S. economic growth and is a focus of public policy. (For example, numerous city leaders nationwide offered subsidies to attract Amazon’s second headquarters.) Carlino found that enforcement of noncompete covenants had a positive impact on Michigan’s high-tech sector’s startup entry rate and job creation rate when compared to nonenforcement states. Carlino’s new state-level research adds to the important literature on this topic. Some researchers have questioned, for instance, whether Massachusetts, which has long enforced noncompete agreements, may have inadvertently suppressed its high-tech sector (notably in the Boston area’s Route 128 belt) compared to California (particularly in Silicon Valley), which has not actively enforced noncompetes.⁸

In summary, Carlino, in his case study of Michigan, finds little evidence confirming the widely held view in the literature that enforcement of noncompete agreements negatively impacts startup activity and job creation. Such findings are important for city and state policymakers who seek to protect the interests of firms and workers and promote overall productivity and growth.