THE non-compete clause has been causing trouble for over 600 years. In 1414 an English court heard the case of John Dyer, an apprentice whose master had stopped him from plying his trade for six months. The judge was having none of it. “The contract is contrary to common law,” he ruled. Individuals should be free to pursue the livelihood of their choice.

That principle has been diluted in the intervening centuries—most countries give businesses some leeway to use non-compete clauses, whereby workers promise not to start or join firms that go head-to-head with their ex-employer. But their prevalence in America is striking (see article). According to a study by the Treasury in 2016, almost 20% of American workers are bound by a non-compete agreement, and almost 40% have been subject to one at some point. Efforts to rein them in are intensifying. Rightly so.

Incumbency we trust

Defenders of these agreements put forward several arguments. One is that non-competes encourage innovation by stopping rivals waltzing off with trade secrets; there is some evidence that levels of investment are higher at firms where they are used. Another argument is that firms are less likely to train workers if newly skilled employees are able to up sticks and take what they have learned with them to a rival. Again, research backs up this claim. A third argument is that firms and employees should be free to contract as they wish.

The counter-arguments are stronger. The prevalence of non-compete agreements is clear evidence that they are being used indiscriminately. Roughly 15% of American employees without a college degree, and a similar share of those earning less than $40,000 a year, are bound by them. Burger-flippers and care-home workers do not have trade secrets to hawk.

The gains in investment and training must be set against the wider costs. In one study, in Michigan, researchers found that workers’ job mobility fell by 8% when non-competes were allowed. When people cannot work for another employer who would value their skills, wage growth suffers, too, because people typically achieve the biggest bumps in their salary when they move firm. Non-competes are also associated with a decline in enterprise. One study found that the rate of entry of new firms into knowledge-intensive industries fell by 18% when non-compete clauses could more easily be enforced. It is no accident that California, a notably innovative state, is one of only three to make the clauses unenforceable except in special circumstances, such as the sale of a business.

The costs spill over to all workers—even those who are not subject to non-competes. Young firms are disproportionately important for job growth, for example; if fewer firms are formed, it will affect everyone in the labour market. And non-competes can have a chilling effect even in places that do not recognise them. One study found that 40% of employees who turn down job offers from competitors cite the clauses as a reason, whether they are in enforcing or non-enforcing states.

The drawbacks of non-compete clauses are all the more worrying because of today’s business climate. The incentive to invest and train counts for less when, as now, the American economy suffers from a lack of competition. Clamping down on such agreements would not solve the problem—just look at the tech giants that call California home—but it would help.

Non-competes are also more worrying when the balance of power between companies and employees is already skewed. The spread of mandatory-arbitration clauses in employment contracts and the decline of trade unions are both signs of that imbalance.

The bar to making all non-compete clauses illegal is high. But the circumstances in which they can be enforced should be narrow, as they are in California. They should be negotiated before employees accept a job offer and they should apply for short times. If a company takes an ex-employee to court, it should be required to demonstrate genuine harm to its business. Non-compete agreements were a bad idea in the 15th century. They still are.