Actor Tom Chambers has caused a row by claiming that "many men's salaries aren't just for them, it's for their wife and children, too." As a justification for the gender pay gap, this is lousy. But there's a core of truth in what he says.

It's a lousy justification because if employers paid higher wages to people with dependents we'd expect to see women with children earn more than childless ones. But the opposite is the case. TUC research (pdf) has found that full-time working mothers aged 42 earn 7% less than otherwise similar but childless women. That might be because mothers take time off work to care for their babies and so accumulate less experience. But it also seems to be the case that single mothers earn less than married ones - which is inconsistent with Chambers' "breadwinner hypothesis."

However, he's got a point in another sense. Fathers and married men do, on average, earn far more than single men or those without children. The TUC has estimated that fathers earn more than 20% more than otherwise similar men without children. And Elena Bardasi and Mark Taylor have found (pdf) that married men on average earn 46% more than the never-married.

These are massive differences. The question is: why?

One reason is selection effects: some men only get married or have children if they can afford to do so.

Another big explanation is simply that married men are different from single ones, and so would get higher pay any way. The things that make men good marriage prospects also make them attractive to employers - such as good education or dependability. Bardasi and Taylor write:

A large proportion of the marriage premium is due to unobservable characteristics that are valued both by wives and by employers, such as motivation, loyalty, dependability and determination.

We should add to this list appearances. Handsome men earn more than munters (pdf) on average (though there are of course exceptions: Piers Morgan, for example, earns a fortune).

However, these things don't explain all of the married men's wage premium. Bardasi and Taylor say that married men earn more than singletons because marriage increases their productivity. They point out that the wage premium is higher for men whose wives do more household chores. Maybe, therefore, married men earn more simply because they're not as worn out as us singletons from having to do the housework.

The problem with this, though, is that it doesn't observe productivity directly. Naomi Feldman and Francesca Cornaglia address (pdf) this problem by studying professional baseball players, whose productivity can be measured directly. They find that married players on average aren't individually better than single ones, but they do earn more. One reason for this, they say, is that married players are more consistent, and this enhances team performance. Also, they say, married players are less exploited; wives help players to drive harder bargains (perhaps in part by reminding them of their team-mates pay).

For me, this chimes true. I would have thought that marriage and children would increase men's earnings by making them more dependable: married men are less likely to spend their evenings drinking and gambling and so are more likely to be fresh in the morning. As baseball manager Casey Stengel said, "Being with a woman all night never hurt no professional baseball player. It's staying up all night looking for a woman that does him in."

There is, though, another potential effect of marriage which I would expect to be important in some cases. I suspect that if I'd ever been married I would have earned more because I would have changed jobs. I'd have chosen higher-paid but riskier or less interesting jobs than the one I've had. I'd have chosen a bundle of compensating advantages in which money featured more highly. This, however, doesn't seem to be a significant factor in the research, which surprises me.

There are two points to all this. One is that, if we strip away the moralizing, Chambers has an empirical point. The other is that there are countless things that determine wages, as a classic paper (pdf) by Bowles Gintis and Osborne pointed out. One of these things is bargaining power.