Skip to main content

New data on non-compete contracts and what they mean for workers

Federal Reserve survey data open up new avenues for research

June 21, 2023


Tyler Boesch Data Scientist, Community Development and Engagement
Jacob Lockwood Board of Governors of the Federal Reserve System
Ryan Nunn Assistant Vice President, Community Development and Engagement
Mike Zabek Board of Governors of the Federal Reserve System
Inside a barber shop, a female barber in a red-checkered shirt and gray apron trims a male client's hair.
Thomas Barwick/Getty Images

Article Highlights

  • In new Fed data, about one in nine workers reports having a non-compete
  • Workers on West Coast less likely to have non-competes, while those in South Atlantic states more likely
  • Data allow researchers to see how non-competes relate to financial well-being, other outcomes
New data on non-compete contracts and what they mean for workers

Non-compete contracts, which limit the job options workers have when they leave their current employers, have been much in the news over the last few years. Policymakers at the federal and state levels have taken action to restrict the use of non-competes or their enforceability in court. However, our knowledge of who has these contracts has been limited, with relatively little survey evidence available. Fortunately, the Survey of Household Economics and Decisionmaking (SHED)—a key Federal Reserve survey conducted annually since 2013—newly includes a question on non-competes.1 We analyzed the latest release of SHED data, from 2022, and found that about one in nine adult workers currently has a non-compete, but this rate varies considerably by geographic region and worker age.

The SHED is not the first survey to ask about non-competes.2 However, the SHED data are valuable because they are broadly representative of the U.S. workforce and collected annually. The new data allow analysts to explore many topics, whether linked to a famous SHED question about financial resilience, questions about job search, or a host of other worker and household decisions. We explore some of these connections here, but note that others can also make use of SHED data to better understand non-competes and their effects on the labor market.

Who has non-competes

Consistent with original survey work by researchers Evan Starr, J.J. Prescott, and Norman Bishara (SPB) and a 2021 Federal Reserve Bank of Minneapolis analysis of U.S. Bureau of Labor Statistics (BLS) data, we find that non-competes can be found throughout the labor force, including for workers with less education and lower wages. The SHED data show that overall, 11.4 percent of adult workers currently have non-competes. However, the SHED data extend our understanding in key ways. For example, we find that workers on the West Coast are substantially less likely to have non-competes than workers in the South Atlantic, at rates of 9.0 percent and 13.3 percent, respectively. See Figure 1, which shows estimates for census regions rather than individual states.

Loading map 1...

Relatedly, we also find that workers are less likely to have non-competes in the three states that do not enforce them (California, North Dakota, and Oklahoma), where the overall rate is 7.0 percent, than in the other 47 states, where the overall rate is 12.0 percent. These patterns are somewhat different from earlier survey evidence showing similar rates of non-competes in states that do and do not enforce them (SPB 2021, page 68). Still, 7.0 percent is a significant share. The pattern suggests that, while some employers may avoid using non-competes in states where they are unenforceable, some employers use them regardless—perhaps because of limited understanding of how enforceability varies across states (Prescott and Starr 2021).

We also find that non-competes are much more common among mid-career workers (35- to 44-year-olds) than among younger and older workers. As shown in Figure 2, 13.2 percent of 35- to 44-year-olds report having non-competes, while only 7.3 percent of 65- to 74-year-olds have them. By contrast, the BLS data used in the 2021 Minneapolis Fed analysis included only workers aged 32–38 at the time, and the SPB survey indicated proportionally less variation across age groups. Both the BLS and SHED data indicate lower rates of overall non-compete holding than in the SPB survey.3

Loading chart 2...

The SHED data allow us to break out the incidence of non-competes by gender, race/ethnicity, educational attainment, industry, and income. We find that men are somewhat more likely to report having non-competes, as are workers with four-year college degrees. Industries vary widely in their use of non-competes: workers in professional services (19.2 percent) and finance (18.2 percent) are more likely to have non-competes than workers in construction (7.1 percent), education (7.8 percent), or public administration (4.7 percent). In line with previous analysis, we also find that workers with higher family incomes are more likely to have non-competes than those with lower incomes. These findings are shown in Figure 3, which enables users to select from a drop-down menu to explore various data cuts.4

Non-compete rates vary by worker characteristics
Loading chart 3...

Implications for workers

New data on non-competes in the SHED are also valuable because of other aspects of the survey that can help researchers understand how non-competes affect workers. In addition to questions on non-competes, the SHED contains questions about personal finances, income, employment, higher education, migration, and housing. The SHED also has a panel dimension that can allow researchers to see how outcomes change over time among workers with non-competes.

Relative to previous surveys that asked about non-competes, the SHED contains much more detail about personal finances among people earning lower incomes. For example, the SHED asks a) whether people have an emergency fund of savings built up in case of a job loss and b) if people would pay an unexpected $400 expense with cash or its equivalent. These questions about people’s liquid savings are relevant for understanding the possible effects of non-compete contracts that can restrict workers’ ability to accept new jobs. This connects with a burgeoning research literature that has found negative effects of non-competes (particularly non-competes that are stringently enforced) on wages of lower-paid workers (Balasubramanian et al. 2022, Lipsitz and Starr 2022) and increases in likelihood of career detours (Marx 2011; Marx, Singh, and Fleming 2015). And it is particularly relevant for those workers whose non-competes are enforceable even when they are fired without cause, as is the case in many states.

The SHED’s question about emergency savings is especially relevant: “Have you set aside emergency or rainy day funds that would cover your expenses for 3 months in case of sickness, job loss, economic downturn, or other emergencies?” A rainy-day fund is particularly important for someone with a non-compete because the non-compete makes it more difficult for them to find a new job.

Looking strictly at the association between non-competes and having an emergency fund, we find that workers with non-competes are 10.8 percent more likely to have an emergency fund. However, the association is complicated by the fact that, as shown in Figure 3, non-competes are more common among mid-career, highly educated workers who tend to have more savings. We therefore present unadjusted estimates as well as estimates adjusted for differences in worker characteristics.

When we adjust for those differences in Figure 4, we find much smaller and statistically insignificant associations between non-competes and savings. While workers with non-competes are more likely to have emergency funds than are workers in general, they appear to have emergency funds at similar rates to workers with similar backgrounds and jobs.

Loading chart 4...

Another important dimension of personal finances is how easily someone could handle a relatively modest expense. The SHED’s “$400 question” asks how respondents would cover an unanticipated $400 expense; we distinguish those who would pay the expense using cash (or a credit card they would pay off in full at the next statement) from those who would pay it in some other way, including with a loan or sale of property.

Overall, people with non-competes are more likely to handle a $400 expense with cash or its equivalent, despite a substantial share still reporting that they would use something else. However, the gap closes and even reverses when we adjust for differences in worker and job characteristics. After adjusting for differences in age, education, and gender, that gap is eliminated. After further adjustments for rural location, occupation, industry, and state—in addition to age, education, and gender—those with non-competes are actually 4.4 percentage points less likely to say they would use cash or its equivalent to meet the emergency expense.

Loading chart 5...

We also conduct the same exercise with several questions about job search and negotiations to show some of the possibilities the survey opens for researchers interested in career outcomes. We find that workers with non-competes are 10 percentage points more likely to ask for a raise or promotion and 7 percentage points more likely to apply for new jobs. These differences persist, in large part, after adjusting for the worker characteristics described above.5 The results are somewhat in contrast to findings that non-competes (and/or their stringent enforcement) tend to reduce workers’ job-search activity (Prescott and Starr 2021), wages (Lipsitz and Starr 2019), and mobility (Balasubramanian et al. 2022). As before, however, we do not have reason to believe these estimates reflect a causal effect of non-competes, but they suggest avenues for deeper investigation.

Informing the policy discussion

The recent explosion of public discussion about non-competes has made clear the need for better and more systematic data collection. The BLS and now the Federal Reserve have invested in this effort through the introduction of questions in their long-running survey initiatives. These investments are all the more timely because of the numerous state and federal policy actions now underway—actions whose effects will be difficult to measure without ongoing data collection.

We encourage other researchers and policy analysts to explore the SHED data, which offer new avenues for investigating non-compete contracts and their implications for workers. Particular strengths of the SHED include its focuses on personal finances, job search behavior, and a number of other topics relevant for people earning low incomes. Non-competes matter for reasons that go beyond what the SHED and other worker surveys can speak to, but the surveys do provide an important factual basis for the decisions policymakers are grappling with.

We thank Matt Marx and Evan Starr for insightful feedback on an earlier draft. Any errors remain the authors’ own.


1 The 2022 SHED yielded a final sample of 11,667 respondents. See Board of Governors of the Federal Reserve System (2023) for more details.

2 Notably, researchers Evan Starr, J.J. Prescott, and Norman Bishara conducted their own groundbreaking survey in 2014 on non-compete contracts. Later, in its long-running study of Americans born in the early 1980s, the U.S. Bureau of Labor Statistics followed up with questions about non-competes, which researchers from the Federal Reserve Bank of Minneapolis analyzed in a 2021 article.

3 Our overall estimate, from the SHED, is 11.4 percent, by contrast to 18 percent overall in the SPB survey. Differences between them may be due in part to differences in handling of “Don’t know” responses; see SPB (2021) for details of their imputation procedure. In the 2021 Minneapolis Fed analysis and here, these responses are omitted. (However, in the appendix of the 2022 SHED report, “Don’t know” responses are not omitted, leading to a slightly lower estimate.) In the sample used in this article, 9.5 percent of respondents were not sure whether they currently have a non-compete. Another difference between the SHED and the SPB survey is the time they were conducted; in the years between the surveys, considerable policy action and public attention have focused on non-compete contracts.

4 Because the sample of American Indian or Alaska Native respondents is small and the estimate is correspondingly imprecise, Figure 3 does not show an estimate for the group. The share of American Indian or Alaska Native workers with a non-compete is not statistically significantly different than the overall share.

5 Relatedly, Rothstein and Starr (2022) find a positive association between having a non-compete and being likely to bargain. However, their data included task-level controls, the inclusion of which nearly eliminated the association. In other words, when comparing workers who are assigned similar tasks, the difference disappeared.

Tyler Boesch
Data Scientist, Community Development and Engagement
Tyler Boesch analyzes data, develops visualizations, and creates statistical models to help the Community Development and Engagement team understand issues affecting low- and moderate-income communities. Before joining the Bank, he was a graduate research assistant with the University of Minnesota Center for Urban and Regional Affairs.
Ryan Nunn
Assistant Vice President, Community Development and Engagement
Ryan Nunn is an assistant vice president in the Minneapolis Fed’s Community Development and Engagement Department. Leading the Bank’s applied research function, Ryan works to improve outcomes for low- and moderate-income communities with the help of better evidence and analysis.