Want to help kids? Build skills instead of pointing fingers.

Economist Kevin Milligan responds to critics of his controversial new paper on Quebec’s low-cost childcare program


Child care policy is a key policy battleground in the current federal election. Into this fray comes a new research paper evaluating the long-term impact of Quebec’s low-cost child care program on things such as the health of teenagers and even how much they fall afoul of the law.

This research comes from the National Bureau of Economic Research in Cambridge, Mass., which published a new working paper this week by me, Michael Baker from the University of Toronto, and Jonathan Gruber from MIT.  What economists call “working papers” are not peer-reviewed, but are posted online so that others can offer feedback on the completed drafts. In today’s fast-paced, Internet-based world, controversial working papers sometimes capture immediate attention, and this new child care paper did just that.

The study found: “a lasting negative impact on the non-cognitive skills of exposed children, but no consistent impact on their cognitive skills. At older ages, program exposure is associated with worsened health and life satisfaction, and increased rates of criminal activity. Increases in aggression and hyperactivity are concentrated in boys, as is the rise in the crime rates.”

In 1960s Michigan, a famous randomized experiment took 123 preschool kids from disadvantaged backgrounds and randomly assigned around half of them to a high-quality child care program, with the rest of the kids getting nothing. The Perry Preschool program was the “Cadillac” of child care: a new learning-based curriculum with four experienced elementary-school teachers for 20 to 25 preschool kids, and also featured weekly home visits to each family to involve the parents directly. The Perry Preschool experiment has been followed now for nearly 50 years, spawning a large number of long-run evaluations comparing the 58 kids in the program to the 65 kids who were randomized to receive nothing.

The research on Perry Preschool uncovered important long-run improvements in the life of kids who were randomized into the program. Earnings are higher, females are more likely to attend college, and males are less likely to get in trouble with the law. Most important, recent findings from Nobel Prize-winner James Heckman identify non-cognitive skills (such as not disrupting the classroom or being aggressive or other forms of externalizing behaviour) as being the driving factor behind the Perry Preschool success story. Better non-cognitive skills led to better life outcomes for the Perry Preschool kids.

Now we can fast-forward to Quebec in the 1990s. Quebec introduced a $5/day universal child care program in 1997, phasing it in fully by 2000. In the next decade, three independent research teams evaluated this Quebec program by comparing Quebecers before and after the program was introduced to residents of other provinces.

In peer-reviewed studies, all three research teams reached similar conclusions. My own team (with Baker and Gruber) found that the subsidized care in Quebec led to a big boost in women working, but had a negative impact on many non-cognitive aspects, such as aggression. Steve Lehrer and Mike Kottelenberg found similar results, and showed that the results were driven by children who took up the child care in response to the policy. Finally, a Quebec-based team also found that the work-boosting impact of the program persisted through time.

Now for a quick methodology break, since that’s where our critics have focused their comments in recent days. When you’re trying to figure out the effect of a program, you can’t just compare the “takers” to the “non-takers.” Why? Because you mix up the true impact of the program with any background differences between the people who are “takers” and “non-takers.” We social scientists call that a self-selected sample, and we try to avoid it as much as we can.

One way to avoid self-selection problems is by taking the whole population of one region that experienced a policy change and comparing them to other similar regions. This so-called “natural experiment” approach has become an anchor of program evaluation in the social sciences. This method is akin to a full randomized experiment such as Perry Preschool, but on a larger scale. All three research teams studying the Quebec child care program used this method extensively, and published the findings in reputable academic journals.

The stage is now set for our new study. If better non-cognitive skills in Perry Preschool led to better life outcomes, what effect will the worsened non-cognitive skills in Quebec have for kids in the long run? The new research tries to find out. We show that the change in non-cognitive skills uncovered in the first wave of studies persisted to school age. This is just like the result of the Perry Preschool program, but in the opposite direction. Then, looking at kids in their teens, we find indicators of health and life satisfaction got worse, along with teens being in more trouble with the law.

The implications of these findings for policy? If the evidence is correct, the case is even stronger for building the best possible environments for developing early-life skills for kids. When the environments are strong, Perry Preschool tells us at-risk kids get a big boost. When those skills are weakened, the new research suggests the kids are doing worse. Rather than finger-pointing, I’d much rather the debate move toward how best to improve the quality of kids’ early-life environments.

As with any study, there are conditions and caveats. In particular, this working paper has not undergone peer review and some problems may be uncovered as scholars kick and prod the results to see if they hold up. Also, we may just be seeing a transitional impact as the program went through its growth, or the negative impact may disappear at still-older ages when we have more data. The new research is not the first and, we hope, not the last word on this important topic; more research always helps to refine what we know.

But, the evidence is mounting that a renewed focus on making sure kids’ early-life environments build skills is fruitful for the long run. Which exact aspect of the early-life environment will achieve that goal? I don’t know. But the new research reinforces the idea that the absence or presence of stronger early-life skills is an important determinant of later-life success. Myself, I think more kids are helped by investigating whether the Quebec program is improving early-life skills rather than just assuming it does.

Kevin Milligan has contributed to policy discussions with officials from all three major parties, most recently as an outside economic adviser to Liberal Leader Justin Trudeau. These discussion did not include childcare policy. A full disclosure statement is here.

Filed under:

Sign in to comment.