Thursday, February 09, 2006

Economics of human skills

The University of Chicago's James Heckman gave a talk, 'The Economics of Human Skills: Evidence and Policy Implications', at the University of Melbourne today. The slides accompanying the talk are here.

Heckman's argument centres on the idea that wages are the source of most people's income with wage inequalities reflecting differences in abilities and skills, broadly conceived as intelligence skills such as IQ and socio-emotional skills ('non-cognitive skills'). Like Plato, Heckman believes major sources of inequality are based on 'ability gaps' created in families. Ability gaps open up even before formal schooling begins and are produced by differences in family environments. His view is that early experiences can modify the biochemistry and architecture of neural circuits and that these early effects have major implications for the way we devise education policy. We need to reduce these inequalities early.

Targeting disadvantaged kids improves equity and has no bad efficiency effects provided the money gets targeted towards the kids. This contrasts with policies directed towards later age groups. Indeed the returns to giving teachers more money and reducing class size are small as are more specific programs that reduce tuition charges for the disadvantaged and job training programs. Basically the damage has already been done at earlier ages.

The idea is to target disadvantaged very young kids and their mothers where rates of return on 'educational investment' (broadly conceived) are very high. 'Skills-beget-skills' so that the emphasis of educational policy should be on the dynamics of skill formation.The talk is based on a book with Alan Kreuger and a productivity argument, presented here.

The difficulty is that early family environments have deteriorated around the world over the past 40 years. Increasing numbers of children live (and/or were born) into single-parent homes and live (and/or were born) into poverty. This has led to a decline in the quality of the labour force which will reduce productivity in the years to come. Educational performance is declining (for example numbers completing high schools are falling) and crime rates are increasing because of a strong negative relation between crime and education. It is cheaper to deal with crime by using education than by employing more police. There is also evidence that early interventions reduce crime.

Children cannot buy good parents - to Heckman this is the key credit constraint. A young Mozart cannot buy himself a good environment to be creative in. On the other hand credit constraints and low family incomes are not major determinants of college enrollments. Indeed, the observed relation between incomes and university participation is spurious because incomes then are a good predictor of the circumstances by which children are brought up at young ages. Uneducated single mothers with low cognitive ability have children early and provide low cognitive and emotional stimulation. Early interventions such as pre-school programs have strong long-term effects in increasing educational attainment, income and in reducing crime.

Moreover the relation between current and future investments in education was 'approximately Leontief' - there was little substitution between these alternative forms of investment. If you don't get to educate through stimulation kids when they are young it is expensive to do it later and there are then big equity/efficiency tradeoffs.

It was a fascinating talk by a major University of Chicago economist. My major criticism of the talk is that the early interventions were not accurately identified. Heckman talked about taking single mothers into daycare groups where both where education focused on stimulating children by reading to them (very important!), taking them to the zoo etc. He called this 'enriched child-care'. He argued that single mothers tended to have poor cognitive and non-cognitive skills and didn't create stimulating environments for their children, and that this could be addressed.

I also think there are political and practical constraints in 'interfering in families' that limit the practicality of Heckman's views. That the effects of early interventions are strong does not, in itself, drive a strong case for public policy based on them.

I was interested that Heckman believes the evidence he provided refutes the R.J. Herrnstein and Charles Murray thesis, in The Bell Curve , that only cognitive skills matter and that these are genetically-determined. Heckman disagrees with this (he believes that IQ itself is manipulable) but did claim the ability gaps, though not genetic, were determined early by the family environment. And it is true that abilities, broadly conceived, don't change much after age 8.

Fascinating stuff.

Update: See the comments on this by Joshua Gans with more hyperlinks to Heckman's work here.


dearieme said...

"a fascinating talk by a major University of Chicago economist": and yet it seemed to be about physiology/psychology. What's up: is Economics so well understood that Discipline Imperialism is necessary, or are you economist chappies a bit like the physicists i.e. stuck?

hc said...


Economists are interested in psychology and they are interested in neuroscience. Heckman is an economist - he won the Nobel Prize in economics in 2000.

Your comment is interesting because, in fact, a lot of Heckman's talk did rely on these ideas. But like a true economist he put the thing in dollar terms. The economic component of the talk was to compare the effectiveness of interventions at different phases of an individual's life cycle by looking at rates of return.

There is nothing inevitably wrong with one discipline borrowing ideas - perhaps heavily - from another.

Anonymous said...

I always thought that the a solution to terrorism will be found via education, and by increasing the marketability of the poor who become the suicide bombers of tomorrow.

Robert Merkel said...

Harry, if I recall my developmental psych textbook the social engineers have an idea called "two-generation intervention" - that helping impoverished children may not pay off all that well directly, but the next generation will receive much better early parenting and thus *that* generation will gain the skills needed to haul themselves out of poverty.

Dunno whether the economists have analysed this or not, though.

hc said...

I don't know about such 'two-generation intervention theories, Robert, but I will follow up on this and find out. The weakness of Heckman's argument to me was the lack of specificity about the form of the interventions for children at young ages.

In the illicit drugs area we know that problems of addiction are likewise often linked to family triggers such as sexual abuse but that does not make it easy to address problems of this type by targeting families. It is just too broad a target to address with policy.

I have problems believing that the huge numbers of skill-poor females, having children out of wedlock, will be that easy to assist. And it was clear that Heckman was targeting parent(s) as well as the young children.

You might be right, interventions might not have an effect on the current generation but may make the daughters of skill-poor single females better parents themselves. This might be true even if the daughters themselves are again single-mothers.

But to be clear this isn't what Heckman believes. He thinks substantive improves in cognitive skills and non-cognitive skills (resiliance, planning, emotional skills) could be made to work over a few years.

Anonymous said...

Interesting post.

I have a clarifying question:

"I have problems believing that the huge numbers of skill-poor females, having children out of wedlock, will be that easy to assist"

Is it the 'out of wedlock' that matters or is it 'skill-poor parents'? By parents I mean mothers and fathers.

The idea that there are 'skill poor mothers' without corresponding 'skill poor fathers' seems to suggest to me that the 'skilled fathers' are bloody irresponsible because they should know better.

Providing publicly financed kindergarten facilities would seem to be one way to provide assistance, irrespective of the answer to the foregoing question.

Removing incentives for relatively wealthy parents to send their children to private schools would seem to me to be a practical way to provide an opportunity for children from 'less skilled parents' to learn about 'opportunities' by hearing what other childrens' parents do.

hc said...

Anonymous if you go to the slide show that I refer to by Heckman you will find evidence cited that over 25% of US children live with a single parent. Generally this will be the mother. About 40% of those children live in families where the mother has not completed high school.

Basically uneducated teenage mothers provide low levels of cognitive stimulation to children -- about half that of educated mothers.

I wasn't making moral points about fathers. I just think that often they are not part of the picture.

Heckman's argument is that teenage mothers as well as their children need to be educated. On simple things such as the dangers to children of cigarette smoking, need to read to kids, take them on excursions etc.

Investing in pre-schools sounds like a good idea to me but I think your last para is not sensible. This might have redistributive effects but won't help poor kids.

Anonymous said...

harry clarke,

Yes,there are the statistics provided by Heckman.

Yes, you do not make a moral point about the fathers. But it is the absence of this point which caused me to ask a question.

I find your statement "I just think that often they (fathers)are not part of the picture" somewhat intriguing. How many immaculate conceptions do take place in the U.S.A. per year?

You consider my last paragraph not sensible because it has redistributive effects without helping the poor kids.

My last paragraph was:
"Removing incentives for relatively wealthy parents to send their children to private schools would seem to me to be a practical way to provide an opportunity for children from 'less skilled parents' to learn about 'opportunities' by hearing what other childrens' parents do. "

As for redistributive effects:

a) Without redistribution (of wealth) there seems to be little point in talking about the poor kids. I am not saying that one should not try to get the 'best value' per dollar of redistribution. I am saying, wealth redistribution is one policy variable.

b) As you know, the concept of Pareto efficiency is insensitive to wealth redistribution (as long as the minimum wealth constraint is met). So, there is no problem from the perspective of 'allocative efficiency'.

c) I can imagine that it might be psychologically tough on 'poor children' to be exposed to direct experience of how the other half lives. However, since you do not see a problem with that for pre-school, this cannot be the justification for your conclusion regarding later years.

d) Economists' often start off analysing a hypothetical 'consumer choice' problems without stating that the choices are assumed to be known to the hypothetical choice maker. Setting aside such trivial choices as 'wine' vs 'beer', which I presume are commodities known even to the least skilled in the USA, in reality people don't know until late in life what jobs there are. (Experimental economics might get to this stage some time in the future; I am aware of at least one result where the assumption of 'common knowledge' was explicitly tested and found to be important).

e) I am aware that the current Australian government is in favour of supporting pre-school assistance but not revising its policy on private schools. But, surely, this is not your reason for your dichotomous answer.

So, why do you think my last paragraph is not sensible?

hc said...

There are no immaculate conceptions but my understanding is that the fathers don't stick around to help with children's upbringing. If they do then, yes, they can be included in the effort.

I favour spending more on pre-schools but a general increase not an increase at the expense of the so-called wealthy. I support the private schools - currently me enrolling my kids there and getting relatively low subsidies releases funds net to the public system. If I sent my kids to public schools average taxes would need to be higher.