Was Research — on Physicians and Noncompete Agreements — Before Its Time?

Years after a paper goes unpublished, it’s fodder for a major Federal Trade Commission proposal

A research team was interested in the economic effects of noncompete agreements, and especially in how they might affect access to health care. With independent medical practices rapidly being sold to investors or corporations, more doctors are signing contracts that bar them from quitting to start their own local practices or to join existing ones nearby. (Typically, the bans last a year or more.) Are these restrictions, the researchers wondered, resulting in more physician choices for patients or fewer?  

Fewer, they found. Looking at states that tightened or loosened noncompete restrictions between 1976 to 2010, the researchers saw that the number of medical practices rose, and practice closures fell, when employers had little or no control over where a doctor could work after leaving their employment.

The study never went very far. With publishers uninterested, the co-authors — Syracuse’s Natarajan Balasubramanian, UCLA Anderson’s Mariko Sakakibara, University of Maryland’s Evan Starr and State University New York’s Seethalakshmi Ramanathan — turned to other projects, both as a team and individually. 

Their notable teamwork includes a paper titled, “Screening Spinouts,” published in Management Science in 2017. The study suggests noncompetes reduce entrepreneurship that naturally flows from workers leaving corporate jobs to start their own businesses in the same field. Another, titled, “Locked In,” published in the Journal of Human Resources in 2020, found that tech workers under noncompetes average lower pay throughout their careers than unrestrained counterparts. The team was joined by University of Michigan’s Jin Woo Chang and Jagadeesh Sivadasan in that collaboration.

A Ban Coming in 2024?

Fast forward to early 2023, when the Federal Trade Commission publishes a notice of proposed rulemaking regarding noncompetes. Calling their use “a widespread and often exploitative practice that suppresses wages, hampers innovation and blocks entrepreneurs from starting new businesses,” the agency proposes invalidating some 30 million existing noncompete agreements and banning, with few exceptions, their use in the future. About 1 in 5 U.S. workers currently works under noncompete restrictions, according to FTC estimates.

FTC officials repeatedly cite findings from “Locked In” and “Screening Spinouts” as support for the proposal, in the official notice as well as numerous speeches and presentations. A reference to the research team’s physician access study, which has not been officially published, shows up in a director’s slide presentation. Organizations that file comments for or against the ban — it’s a hot topic, attracting tens of thousands of comments to the FTC so far — also frequently refer to studies by these authors.

The FTC is expected to vote on the proposed ban in April 2024. Meanwhile, the question of how noncompete agreements affect health care has become a lot more popular. And the data needed to pose credible arguments for and against barring noncompetes has piled up. 

Hospitals vs. Doctors

Traditionally, noncompetes were limited to high-level workers that had access to proprietary information or apprentices who received extensive training for a specific skill. Today hairdressers, security guards and fast food workers — some 14% of workers earning less than $40,000 a year — are subjected to the same career restrictions as executives, tech workers and corporate sales reps who likely can better afford sabbaticals or moves to escape the restrictions. 

The FTC estimates that eliminating all of these contracts would raise worker pay between $250 billion and $296 billion per year nationally. 

Ahead of the FTC proposal, the restraints on doctors have notably spurred limited legislation restricting noncompetes so far. Only California, North Dakota and Oklahoma ban enforcement of noncompetes for all employees. Eleven states have some limitations on enforcement of noncompetes for lower income workers, with widely varying definitions of lower income. Four other states have laws that specifically prohibit enforcement against physicians. (New York is considering its own ban.)

Notably, attorneys do not practice under noncompetes in any state, because the American Bar Association prohibits them from signing the contracts. The rule is meant to protect clients’ ability to maintain their preferred representation, even if their lawyer’s place of employment changes.

The American Medical Association has not granted the same coverage to its constituency, despite research that shows patients suffer when they lose access to their doctors. For example, a study covering about 16 million recipients finds that losing a longtime primary care physician increased patient mortality by 4%, emergency department visits by 4% and hospital admissions by 3%. The study, by Stanford’s Adrienne Sabety, is published in the Journal of Public Economics. 

Traditionally, U.S. physicians were self-employed, outside the realm of noncompetes. A vast consolidation of health care has changed that. Between 2020 and mid-2023 alone, more than 7,000 medical practices sold. Today, the  majority of physicians work for large employers. According to figures by consultants Levin Associates, some 40% of these practices are now owned by private equity firms, an industry especially fond of noncompetes. 

As in other industries, in health care employers love noncompetes and employees despise them. Responding to the FTC’s request for comments, the American Hospital Association says banning noncompetes would exacerbate serious staff shortages at a time when many health care workers are expected to exit the profession. (The organization also questioned the FTC’s authority to establish a ban on several levels.)

In its opposition letter, the association says that struggling rural hospitals in particular need noncompete contracts, as well as perks for recruits such as signing bonuses and moving costs, to both attract and retain physicians and other professionals “when they are inevitably presented with other professional opportunities.” The implications are that these recruits would quickly leave these positions  if allowed to quit to work nearby at will, exacerbating hospital staff shortages while allowing other employers to benefit from their recruiting expenses.

Representing medical doctors, the American Medical Association pledged to support, with some exceptions, an end to noncompetes in the profession. Several specialist groups also responded with wholehearted support of a ban, including American College of Emergency Physicians. Its letter included quotes from emergency room physicians particularly galled at their career limitations, considering they do not have patients they would take with them in a move. The practice, one doctor wrote, “is quite frankly ridiculous. No one chooses an Emergency Room in an emergency based on … whether a certain doctor is working.”

The Data Gap

When Balasubramanian, Sakakibara and Starr first began to collaborate, the study of noncompetes lacked some basic empirical evidence, such as the extent of their use. Starr and University of Michigan’s Norman Bishara, who have become oft-cited experts in the field, published a catalog of the missing data in a 2016 paper in Lewis and Clark Law Review. 

A wave of academic interest in the topic since has helped fill the research gap. The FTC’s proposal cites more than 40 studies on noncompetes, most of which came out in the past six years. Among them were some credible estimates about how many and what kinds of workers sign them: about 62% of chief executives, according to a study by Georgia State’s Omesh Kini, University of Arizona’s Ryan Williams and Miami University’s Sirui Yin; 30% of hair stylists, according to research by Duke’s Matthew S. Johnson and the FTC’s Michael Lipsitz; and 45% of primary care physicians, in findings by Ohio State’s Kurt Lavetti, The Levin Group’s Carol Simon and Cornell’s William D. White.

In 2021, Starr and Bishara, along with University of Michigan’s J.J. Prescott, published results of a nationwide survey they conducted about noncompete use. They found that 18% of 11,500 respondents worked under noncompetes, and 38% had at some point in their lives. Some 53% of surveyed workers covered by a noncompete were paid by the hour. The Federal Reserve called the study “groundbreaking” and now includes questions about noncompete agreements in its own annual household surveys. 

In a surprise finding, the survey also uncovered some suspected, but previously untested, information about noncompete use. Laws prohibiting enforcement of noncompetes did not, in the least, dissuade employers from demanding workers sign them, according to the results. Ostensibly, the tactic scares employees into compliance to avoid becoming the subject of potentially expensive lawsuits, baseless or not. It appears to work. The authors’ later collaboration found employees under even legally unenforceable noncompetes commonly cited fear of being sued for breach as reason for declining job offers from competitors. 

The studies that the FTC cites overwhelmingly conclude that noncompetes reduce worker mobility, or turnover, and increase employers’ investment in training. Several find that worker pay increases when the contracts are banned, but there is ongoing debate over the methodologies used to collect this data. 

The American Hospital Association, one of the most powerful opponents of a ban,  notes that the Lavetti study found that physicians who worked under noncompetes earned significantly more money than those without the contracts. It contends that the finding should exempt physicians from a noncompete ban, because raising worker pay is one of the FTC’s key stated goals. “What is clear is that the only actual study regarding physician pay supports the use of noncompetes,” according to the association’s formal objection to the proposal. 

In an email exchange, Sakakibara notes that, “Lavetti’s paper is a correlational study,” not asserting a causal relationship between noncompetes and pay, (which the hospital association seems to imply). Lavetti’s results perhaps reflects selection, meaning better, or more sought-after, physicians sign noncompetes and get higher wages, Sakakibara adds.

Meanwhile, the specific question of whether noncompetes give patients fewer doctor choices in today’s overburdened healthcare markets remains empirically unsettled. But state actions targeting physician noncompetes suggest that lawmakers have serious concerns that the contracts are contributing to the problem, not helping it.

For example, Florida, a state that supports noncompete agreements more relentlessly than most, in 2019 carved out a narrow ban on them for physician specialists in counties where one entity controls the market for their services. According to the legislative consensus, allowing more doctors to compete freely would be good for patients in those areas.