As Brian notes, non-competes protect the investments that employers make in their employees. This is a specific instance of the more general–and commonsense–principle that investment is more likely when it’s protected. On-the-job training (a form of investment), which raises the marginal productivity and wages of workers, becomes more likely under a regime of free contracting. The employee voluntarily “ties his hands” in order to receive other benefits he deems more valuable than a low cost exit option. If he preferred the ability to swiftly switch to a rival employer, he could take lower wages in return for this perk.
Check out chapter twenty-four of Alchian and Allen’s recently published Universal Economics. A&A describe and defend one of the most (in)famous examples of a non-compete: the baseball reserve clause. In their own words:
How could the reserve clause be defended in baseball? The defense rests on the necessity of expensive training and testing of athletes during their early careers. A team owner makes exploratory, developmental investments in many rookies, hoping that a few players will ultimately be worth the expense. The reserve clause restriction protects the team owner’s investments. Absent the initial investments by the team owners, the aspiring players would have had to bear more of their own investment costs during their early careers, possibly playing with no salary in the minor leagues or not trying at all.
The same sort of reasoning applies to the seemingly “lopsided” contracts signed by no-name actors and musical artists. But good luck explaining this reasoning to your average Hollywood leftist.
Incidentally, Harold Demsetz wondered if the “reserve clause” would have any allocational effects at all, given the Coase Theorem. However, to me, this seems like the classic high-transaction-costs-case that Coase cared most about.