Funny you should ask....

Responding to Matt Bruce's sensible final parenthetical question in his Fancy Store-Bought Dirt weblog, I know just the place to look for an answer. In an article last year for the Journal of Human Resources, John Cawley asks whether obesity affects wages, while at the same time recognizing that wages may affect the risk of obesity. His summary of the various statistical approaches for tackling this type of chicken-and-egg problem is so lucid that I am using his article in my teaching. Here is his bottom line:

This paper finds that weight lowers wages for white females; OLS [regression] estimates indicate that a difference in weight of two standard deviations (roughly 65 pounds) is associated with a difference in wages of 9 percent. In absolute value, this is equivalent to the wage effect of roughly one and a half years of education or three years of work experience.

p.s. Bruce is right that correlation does not prove causation.