Professor Fletcher pointed us towards this and I had my own thoughts.

Stay in school, until you learn that correlation ≠ causation « Page 1 « Aid Thoughts.


I am bothered by this tweet by Justin Wolfers:

The answer to Wolfer’s question: you might convince a lot of children to get more (or less!) education than they need to.

It’s incredibly easy to estimate outcomes for people in given education categories, but this tells us very little about thecausal impact of education. More or less the first identification problem you learn in an econometrics class concerns estimating the returns to education when you can’t observe ability. The problem is that high ability people are likely to get more education and have better labour force outcomes than low ability people. Hence, the results above could be driven partially (or totally!) by selection.

Why is this a problem for convincing kids to stay in school? Well, let’s say I’m a low ability person (or, to be fair, lacking in the type of ability which facilitates traditional education) – I might recognise that I’m low ability and decide to find a decent job that I enjoy after high school, rather than getting more education. Now let’s rewind: after the 2012 Wolfers act is passed, I get this shock of misleading information that the returns to more education are really massive! I then blow more money (either in fees or in foregone wages) on education, only to find that A) it is still hard and B) because I’m low ability, my labour force outcomes are worse than all my similarly-educated peers.

There have been a few studies in developing countries where researchers approximations of the above chart are shown to children – one in Madagascar and one in the Dominican Republic. It turns out that showing children these relationshipsdoes cause them to get more education – but we don’t know yet whether what the resulting impacts are. If we think that the entire population is stuck in a low-education trap and needs to break out, this is a good thing. However, if we thought that people are making at least semi-rational choices about the returns to the education, we might be tricking some people into getting more education than they need to.

I’m being a little unfair here – there’s a lot of uncertainty out there: I’m a 28 year old PhD student who still doesn’t know where he is on the ability distribution (yay for 15% unemployment!). I think we should be showing graphs like these to kids, but be more honest about what could be driving the relationships.



I agree with your point in terms of the actual productivity of workers in the work force.  People definitely have different skill sets and generate different marginal benefits from additional education.  However, there are other reasons we want people to be educated.  Educated citizens can make decisions based on a wider knowledge base.  Whether its in politics, employment, community, or any number of other areas, education gives people the tools and information to become a more engaged citizen.  If we concede these additional private and social benefits to education we must also concede that additional education might be desirable across all ability levels?