Last week I read all the major research on the response of fertility to economic incentives.  There are actually two distinct literatures.  The first focuses on the effect of intentional “birth subsidies” on child-bearing.  The second focuses on the unintentional effect of welfare (and welfare reform) on child-bearing.

Striking contrast: In the “birth subsidy” literature, researchers usually find fairly large effects in the expected direction.  In the welfare literature, in contrast, most researchers find little or no evidence that welfare increases fertility – or that welfare reform reduces it.

The single best piece in both literatures put together: Kevin Milligan‘s “Subsidizing the Stork” (Review of Economics and Statistics 2006; ungated working paper here).  Milligan begins by describing a remarkable social experiment, Quebec’s Allowance for Newborn Children (ANC).  From 1988-1997, Quebec wrote checks to all households with new babies.   The payments gradually rose, plateaued from May 1992 through September 1997, and were then abolished.  At the plateau, the birth subsidy was: C$500 for the first child, two payments of C$500 for the second child, and twenty payments of C$400 for each additional child.

This program gives Milligan three distinct kinds of variation with which to estimate the effect of incentives on fertility. He can compare:

1. Quebec to the rest of Canada
2. Fertility in Quebec over time (since the birth subsidy repeatedly changed)
3. Fertility of families with zero, one, or more children

These comparisons allow Milligan to answer almost all complaints about the validity of his “natural experiment” assumption:

[E]ven if there were some change in unobservable determinants of fertility in Quebec contemporaneous with the introduction of the ANC, the triple-difference comparison of first births to higher order births would permit inferences about the effects of the policy. In other words, a social trend would have to have a differential impact on families of different sizes in order to hinder inferences.

Putting all this info together, he finds: “the responsiveness of fertility to a birth subsidy is estimated to be large–up to a 25% increase in fertility for families eligible for the full amount. A C$1,000 increase in first-year benefits is estimated to increase the probability of having a child by 16.9%.”

My favorite part of Milligan’s piece, though, is how he reconciles the birth subsidy and the welfare literatures.  To do so, he looks at the fertility response of young and/or unmarried women – the same populations that the welfare literature studies.  He finds that the fertility of this sub-sample is unusually unresponsive to incentives:

[W]omen who are single and women who are younger may be less responsive to the ANC than other women. These types of systematic differences between women likely to collect AFDC and all women may explain the different results found in AFDC studies and the ANC results presented above.

At first, this seems counter-intuitive.  But it is consistent with another surprising fact: Never-married moms in their thirties are about twice as likely to have one kid rather than two, while married moms are about twice as likely to have two kids rather than one. 

What’s the connection?  Well, both findings suggest that, as a rule, young unmarried women get pregnant by mistake.  They’re far from affirmatively wanting a baby, so modest financial incentives don’t change their plans.  Older, married women, in contrast, are already “in the baby business.”  They’re close enough to wanting another child that a little financial encouragement is often all it takes to tip them over the edge.  Does that make sense?