> In those cases, saying something like this would mean 
  > finding another job.

Did you actually try it?

If wonder if companies with mandatory non-competes have lower salaries. They should.

You would expect them to have higher salaries, to compensate employees for the risk of being prohibited from working for N months after their employment terminates.

Higher initial salaries perhaps, but lower annual increases, since employees don't have as many jobs to switch to so there is less incentive for raises.

It appears that you are making the unstated assumptions that these employees have experience in exclusively non-transferable skills and that they had no knowledge of future changes in salary at the time they accepted the agreement. I find both of these claims to be dubious at best; if you have supporting evidence, please enlighten me.

Economic theory would suggest so. I suspect that, adjusting for everything else, it's a pretty small effect. Would be an interesting study though. I wonder if there's one there.

[EDIT: Duh. I misread the original post. Of course, the salary should be higher not lower to compensate for such a restriction--though I still suspect it would be small.]

What economic theory would suggest this? This non-compete appears to be an option, and would be priced as such.[1] I would guess that purchasing any additional assurance from an employee would cost the employer(s) more, perhaps in the form of salary, severance, or a signing bonus.

[1] http://en.wikipedia.org/wiki/Valuation_of_options

You're, of course, correct for the reason that you state.

In a good market, only more desperate candidates will go with non-compete positions. So lower salary argument makes sense

You're making the error of deducing market conditions from price (or price from market conditions).

Market price is determined (in a well-functioning market) by the intersection of supply and demand. So the salary could go other way: higher, if a non-compete makes it harder for the employer to find employees willing to sign the clause, lower, if result is fewer alternatives and/or more desperate employees interested in the position.

I'd actually argue the second: that noncompetes raise employees' switching costs, and reduce their alternatives, so that a state in which noncompetes are valid and widely applied would tend to have depressed labor rates.

No; in a good market the employer who wants a non-compete would need to pay me up front for the time period that I wouldn’t be able to work after leaving her firm, plus a premium for the expected difference between my salary at that point and my salary at hiring, which would make the wage for positions with a non-compete significantly higher than those without.