Saturday, August 25, 2012

Remember: family size is endogenous

This is a great post from Jared Bernstein on income inequality that I wish you'd hear more. He writes:

"This sounds, and is, a bit weedy/wonky, but it’s a point that needs to be taken more seriously among economists, journalists, and policy makers who focus on trends in family income: it’s not necessarily correct to adjust incomes for family size, because family size is endogenous to economic conditions."

You would think this would also be of interest to people who talk about inequality a lot - like Don Boudreaux and Steve Horwitz - who also (I assume) agree with Bryan Caplan's enthusiasm for children and population growth.

When the earning potential of a family declines the size of that family is going to decline too, in response to the reduced earning potential. That endogenous response is going to understate shifts in earning potential if you adjust for family size, because families are accommodating to their income stream.

You don't just have to imagine the poorest families to see this, although it's a group we often focus on. Kate and I probably would have had kids three years ago or so if I didn't have plans to go for a PhD. We're not going to wait until I'm all the way out, but we definitely didn't want a toddler around when our income stream took a temporary nosedive*.

I've adjsted family income for family size in analyses before, but it was typically because I wanted to control for the resources of a given family. Even that, of course, has endogeneity problems but since neither family size nor family income was an outcome I was interested in it wasn't quite as big of a deal. But if you're looking at trends over time this could be very important.

* - in my earlier post on ability to pay, a lot of people talked about poor people accessing credit. Kate and I aren't even poor - what do you think the bank would say if I just went up to them and asked for a loan because we're having a kid and we really just need to make up the tens of thousands of dollars I lost by leaving the Urban Institute. I don't think I'd get it. They'll give me a loan for an actual underlying asset, like my house, but not for that. Credit rationing and credit constraints are real, and I was surprised how many people just appealed to accessing credit markets in the comment section of that post.

3 comments:

  1. The only problem I see with this is that the point isn't exactly about "family size"... it's about *new* children. Whether or not a family has another child may track with the economic situation, but every new child is going to have something like an 18 year stay in the household.* A family with two children born from 2003-2006 is not going to reflect anything about current economic conditions, even if their family is the same size as mine (with kiddos born in 2008 and 2010). It seems like you'd need to control the family count very tightly for only very young children to pull out the endogenous factor, and maybe after that apply a much lighter general adjustment for all household dependents, regardless of age.


    *Or, in this economy, even longer, as 20-somethings without jobs move back in and are potentially still dependents for some of that time.

    ReplyDelete
  2. This notion of endogenous family size seems to ignore any effects from different levels of education. Certainly on a global scale, there is research to the effect that poorer, less-educated women/families tend to have more children. As incomes and education rise, women/families tend to have less children. It seems the argument here implies that the US, individually, operates under the reverse of this global trend.

    ReplyDelete
  3. The point of the argument was that there is still substantial median income growth. That is still true, regardless of what this says.

    You would need some strangely shaped demand curves to show that the median individual is both richer today but is having fewer children in response to that.

    ReplyDelete

All anonymous comments will be deleted. Consistent pseudonyms are fine.