Casual empiricism suggests that celebrities engage in more anti-social and other socially unapproved behavior than non-celebrities. I consider a number of reasons for this stylized fact, including one new theory, in which workers who are less substitutable in production are enabled to engage in greater levels of misbehavior because their employers cannot substitute away from them. Looking empirically at a particular class of celebrities – NBA basketball players – I find that misbehavior on the court is due to several factors, including prominently this substitutability effect, though income effects and youthful immaturity also may be important.
Elsewhere, here is a Kaushik Basu micro piece on the law and economics of sexual harassment. And a more recent piece from the sociology literature. The practice increases quits and separations, with some of the costs borne by harassment victims and not firms; given imperfect transparency, recruitment incentives may not internalize this externality. On other issues, here is a relevant AER article. And this piece applies an insider-outsider model. Here is Posner (1999), perhaps he has changed his mind. Here is work by Elizabeth Walls, from Stanford.
I see negative externalities to sexual harassment across firms and sectors, and so, contra Posner (1999) and Walls, the most just and also efficient outcome is to tolerate one explicit and transparent form of the practice in the sector of formal prostitution and otherwise to keep it away from normal business activity. I believe such a ban boosts womens’ human capital investment, investment in firm-specific skills, aids the optimal production of status, and limits one particular kind of uninsurable risk, with all of those benefits correspondingly higher in an O-Ring or Garett Jones model of productivity.