Infant Interventions

My oped in the AFR this month is on early childhood intervention. Full text over the fold. I had the benefit of comments from Nicholas Gruen and Justin Wolfers, though the usual disclaimer applies.

Intervention: Better Earlier, Australian Financial Review, 12 July 2007

Mentioning disadvantage in an economic boom is sometimes regarded as a faux pas, like asking the host at a fashionable party why she didn’t invite her badly-dressed friends. But given that today’s golden age isn’t glittering for everyone, the right response is to work out how to improve the life chances for the poorest Australians.

The trouble is, the cycle of disadvantage has proven agonisingly hard to break. We know relatively little about patterns of poverty, but we do know that a son whose father is out of work for six months has triple the odds of being long-term unemployed when he grows up. And a boy who witnesses parental violence in childhood is six times as likely to hit his spouse in later life. Correlation isn’t causation, but early experiences seem to matter.

One promising solution is high-impact early childhood intervention programs. In the US, careful economic evaluations – based on randomised trials from the Abecedarian, Perry Preschool, and Early Training Projects – have shown that providing intensive assistance to disadvantaged children and their parents isn’t just morally right – it can be wildly cost-effective too.

These programs admitted children into preschool at an early age (sometimes as young as 4 months), and focused on developing cognitive, language, and social skills. The target population was extremely disadvantaged. From a young age, their IQ scores were below the US average. In the Perry Preschool program, two-thirds of girls in the control group had fallen pregnant in their teens, while more than half the boys had been arrested.

When researchers followed both the treatment and control groups, they found that those who received early childhood interventions were doing better on most measures than those in the control group. The programs cost A$15,000-50,000 per child, yet they easily paid for themselves in reduced welfare spending, higher tax revenues, and less crime.

Set against other anti-poverty programs, early childhood interventions look even better. Put to the test, many programs designed to help low-income adults have only minimal results. Job training programs for unskilled workers are rarely very effective. Prison education programs often fail to prevent recidivism. A steady diet of welfare payments does make people less poor, but has its own negative consequences. Until we find better solutions, it would be rash to defund these other programs. But when a promising solution for breaking the poverty cycle comes along, it’s time to grab it with both hands.

You would think that the lessons from the United States’ early childhood intervention programs were obvious: high-impact programs are effective, and randomised trials are the best way to work out whether a program makes a difference. Yet Australian policymakers act like they read the headline, but skipped the story. They have largely eschewed targeted, high-impact interventions in favour of universal, low-impact programs. Offering more publicly provided childcare to the middle class may have a high electoral impact, but it is not going to transform the life chances of the poorest.

So far as evaluations are concerned, a handful of randomised trials of small early intervention programs have been run by researchers at the University of Queensland. But for the most part, Australian early childhood programs are not rigorously evaluated in this way. Ironically, the effect of this will be to ensure that the next generation of early childhood researchers will be as disadvantaged by lack of evidence as we are today.

Unlike the United States, where Struggle Street and Main Street intersect in the CBD, low-income Australians tend to live in regional areas, or on the outskirts of cities. Clontarf and Campsie may both be Sydney suburbs, but it’s a fair bet that most residents of the former have never visited the latter. While the media spotlight has lately highlighted run-down Indigenous communities, anyone who has walked through some of the rougher housing estates on our urban fringes knows that you don’t have to be Indigenous to be part of the underclass.

The fact that you can’t see disadvantage from the city shouldn’t blind us to action. If equality of opportunity means anything, doesn’t it mean that governments should allocate more money to poor children than rich ones? Unlike ‘tax and churn’ programs like the Baby Bonus, early childhood intervention allocates resources where they are needed the most. Devoting a few extra dollars to early childhood intervention programs today sure beats spending it on jails tomorrow.

Dr Andrew Leigh is an economist in the Research School of Social Sciences at the Australian National University.


  • The intergenerational unemployment calculations are my own, based on the HILDA dataset.
  • The intergenerational domestic violence estimates are from this paper.
  • MBS held an event on this topic a couple of weeks ago, and the papers are here.
  • In 2006, James Heckman visited Australia and spoke about this topic, and I wrote up his talk.
  • Two excellent review papers on the US randomised trials are by AIFS and Michael Anderson.
  • One of the main Australian researchers doing randomised trials of early intervention programs is Matthew Sanders (UQ), and I’m grateful to Ross Homel for drawing my attention to his work.
This entry was posted in Economics of Education. Bookmark the permalink.

11 Responses to Infant Interventions

  1. Daniel says:

    Thanks Andrew.

    Keep pushing this until they hear you. It is really important. That said, I’d be interested in your thoughts on Labor’s very liberal use of Heckman’s research to support their early childhood policy of accessible kindergarten for all 4 year olds.

    I’d argue that the COAG human capital document…while still not having effective Australian data…understands the real impact is made with specifically-targetted, well-resourced programs rather than broad hit and miss universal programs.


  2. Andrew Leigh says:

    Daniel, here’s my answer on that one.

  3. conrad says:

    I love that teenage pregnancy measure. You’d think that people having children early in life were aliens, despite the rest of human history.

  4. Of course the opening premise that a time of economic boom is not the time to talk of disadvantaged is utterly wrong. When else do the middle class have the confidence to start interfering in the lives of their inferiors?

  5. Bruce Bradbury says:

    You imply that low-impact programs are ineffective, but I wonder if this simply reflects the inability of research to identify such effects. A very large sample size is required to statistically identify a small effect, and I suspect that few evaluations of weak interventions are able to reject the hypothesis that a weak effect exists.

    Nonetheless, a program with a weak but beneficial effect can be substantively important if it is applied across a large enough population (eg all children).

  6. Andrew Leigh says:

    Bruce, what I meant to say (and I clearly didn’t express myself very well), was that low-impact programs don’t have the kind of super-impressive benefit/cost ratios of the high-impact programs. It’s true that some lite programs have small effects (see Matthew Sanders’ evaluations, for example), but what frustrates me is politicians who cite Perry as evidence for implementing universal preschool.

  7. backroom girl says:


    Is it the case that most if not all of the benefits from the Perry preschool project arise from the reduced incidence of teenage pregnancy and male incarceration (associated presumably with increased labour force participation and reduced welfare receipt), rather than from substantially higher earnings by the participants?

    If so, I find that a fairly fascinating result from a program that was actually aimed at increasing human capital. (Not an argument against doing it, of course, but just a reinforcement of the message that social programs often have unintended consequences, both good and bad.)

    The other thing that I’ve always wondered about the Perry results is why they seem to be so much better than the long-term outcomes of many apparently similar intervention experiments. Did Perry involve parents in some way that was different to the other experiments? If it did, and therefore changed things at the family as well as the individual level, it might help explain how the effects on teenage pregnancy and male incarceration were so large.

  8. backroom girl says:

    “You’d think that people having children early in life were aliens, despite the rest of human history.”

    Surely you jest Conrad. The point is that we are no longer in the rest of human history, when having children was the only thing that most women did with their lives (thank goodness).

    Do you really dispute that having a child as a teenager is not associated with poorer life outcomes in general, including for the children of the teenage mother? If intervening intensively with young children at risk of such outcomes (and perhaps just as importantly with their parents) helps some young women to take a different life course to dropping out of school and raising another generation of kids on welfare, then I’m all for it.

    I agree with Andrew, though, that Australians have a fairly inexplicable aversion to programs targeted clearly to populations at risk, perhaps because we think it is stigmatising or somehow inegalitarian. I find it difficult to believe that if teenage mothers living on a Housing Commission estate were offered a special preschool program (or any other ‘special’ program) for their kids, that they would feel stigmatised.

  9. Andrew Leigh says:

    BG, the Perry results aren’t that different from other programs that involve thousands of hours per child (eg. Elmira, Early Training, Abecedarian). They’re definitely bigger than most interventions that are trialled, however. And one possible reason for this is that your typical intervention involves <10 hours in the program.

  10. backroom girl says:

    Well, the amazing thing is that anyone could actually imagine that 10 hours or less of intervention would work. Even universal preschooling amounts to some hundreds of hours in total.

    On the preschooling issue, does anyone know if there is any evidence of preschool take-up problems for lower socio-economic groups? Universal provision of preschool will be less effective unless there are strategies to ensure high participation by ‘at risk’ children – otherwise it will just be another case of most of the benefits of universal programs going to the middle classes.

  11. Pingback: Andrew Leigh » Blog Archive » Poor kids first, please

Comments are closed.