Pre-school investments

James Heckman publishes new papers faster than I can read papers. And so I can't pretend to be on top of all of the work he's been doing on early childhood education. But there has been a lot of it over the past several years.

I was pretty excited about the whole line of work when he started it. Investments in early childhood education looked to yield huge benefits, especially for children from disadvantaged backgrounds. NPR provides a decent summary of that work in an interview with Heckman in December:

If you got 13 percent back on your investments every year, you'd be pretty happy, right? Remember, the S&P 500, historically, has averaged about 7 percent when adjusted for inflation.

What if the investment is in children, and the return on investment not only makes economic sense but results in richer, fuller, healthier lives for the entire family?

That's the crux of a new paper out Monday, The Life-Cycle Benefits of an Influential Early Childhood Program, co-authored by Nobel laureate James Heckman, a professor of economics at the University of Chicago and the director of the Center for the Economics of Human Development.

The new paper looks again at the Abecedarian Project, and finds returns that last through to the long-term. It sounds like the kind of thing that would make sense under the New Zealand government's investment approach. But I have a couple of worries.

Every time we see great results like the ones Heckman demonstrates, they're from really small-scale, really resource intensive programmes like Abecedarian or Perry.

And the couple of studies I've seen looking at what happens with larger roll-outs don't suggest that these things scale well.

Michael Baker, Jonathan Gruber and Kevin Milligan found that a universal childcare programme in Quebec yielded a huge increase in use of childcare, but worsening of outcomes for kids. NBER provides a summary here.

And an evaluation of Tennessee's pre-K programme, rolled out to 18,000 lower income kids across the state, showed temporary gains followed by worse outcomes than those for kids who didn't participate. This one was focused on providing high quality care to a large group of deprived kids, rather than being a universal roll-out like Quebec's. The programme only included 4 year olds, and so could be critiqued as not having the very early intervention that the Heckman-style interventions recommend, but it should give pause that the effect wound up being negative.

I don't know if these things don't scale well because you quickly run out of teachers that match the quality of those involved in Abecedarian and Perry, or if larger scale programmes in general disappoint. But it does suggest we need to be careful in monitoring roll-outs of things that looked good in smaller scale trials.

And I worry too about what happens if it turns out that the thing really can only work in really small and focused interventions. You then wind up setting eligibility criteria to focus on the groups in most need, and really pissing off the people who are just outside the eligibility threshold. Like, "You mean if I'd had more CYFs notifications my kids would be eligible for this fantastic pre-school programme but because I was a better parent than my crazy neighbour she gets the programme and I don't?!" The usual answer on the left to that kind of concern is to demand that it be universal, but look again at the results in Quebec.

I don't pretend to have any good answers here, just worries. I wonder if anybody's used IDI data to see what happened after New Zealand shifted to 20-hours free daycare.

FWIW, both our kids were in pre-school from about 3 months old, when Susan returned to work. So our revealed preference was to rely on professional childcare.