The other day, Tyler Cowen challenged me to name any country that I consider under-educated. None came to mind. While there may be a country on earth where government doesn't on net subsidize education, I don't know of any.
…This analysis holds in the Third World as well as the First. The fact that Nigerians and Bolivians don't spend more of their hard-earned money on education is a solid free-market reason to conclude that additional education would be a waste of their money.
I would first note that many parts of many poor countries, today, receive de facto zero government subsidies for education. Or put aside the issue of government provision and ask if you were a missionary and could inculcate a few norms what would they be? Many regions – in particular Latin America — are undereducated for their levels of per capita income. I view this as a serious cultural failing, most of all in terms of its collective social impact. In contrast, Kerala, India is very intensely educated for its income level and that brings some well-known benefits in terms of social indicators and quality of life.
If I think of the Mexican village where I have done field work, the education sector "works" as follows. No one in the village is capable of teaching writing, reading, and arithmetic. A paid outsider is supposed to man the school, but very often that person never appears, even though he continues to be paid. Children do have enough leisure time to take in schooling, when it is available. I am told that most of the teachers are bad, when they do appear. You can get your children (somewhat) educated by leaving the village altogether, and of course some people do this. In the last ten years, satellite television suddenly has become the major educator in the village, helping the villagers learn Spanish (Nahuatl is the indigenous language), history, world affairs, some science from nature shows, and telenovela customs. The villagers seem eager to learn, now that it is possible.
That scenario is only one data point but it is very different than the "demonstrated preference" model which Bryan is suggesting. Bolivia and Nigeria are much poorer countries yet and they have dysfunctional educational sectors as well, especially in rural areas. Bad roads are a major problem for "school choice" in these regions, just as they are a major problem for the importation of teachers.
A simple model is that underinvestment in infrastructure results in a high shadow value for marginal increments of education. Model = high fixed costs, liquidity constraints if you wish, high shadow values for lots of goods and services, toss in social externalities to raise the size of the distortion. I read Bryan as focusing on "the fixity of the fixed costs" and claiming it is too costly to get the service through, relative to return.
Of course Bryan favors rising wealth and falling fixed costs, as do I. But in the meantime he also should admit that a) education "parachuted" in from outside can have a high marginal return, b) collectively stronger pro-education norms raise demand and can alleviate the high fixed costs problem, c) there are big external benefits, some operating through the education channel, to lowering the fixed costs, d) stronger pro-education norms put a region closer to a "big breakthrough" and weaker education norms do the opposite, and e) a-d still impliy "too little education" is the correct judgment. On b), some evangelical groups in Latin America do seem to have stronger pro-education norms in their converts and it appears to be much better for the children of these families and no I'm not going to buy any response which ascribes the whole effect to selection.
I believe that Bryan's own work on voting suggests significant positive social external benefits from education, although he is not happy with how I characterize his view here. I also believe his views on children suggest strong peer effects across children (parental effort doesn't matter so much in his model and the rest of the influence has to come from somewhere), though in conversation I am again not sure he accepts this characterization.
I consider most countries in today's world to be undereducated.
Signaling models are important but they are not the only effect and of course a lot of signaling is welfare-improving for reasons of screening and sorting and character reenforcement. The traditional story of high social returns to education is supported by evidence from a wide variety of different fields and methods, including cross-sectional growth models, labor economics, political science, public opinion research, anthropology, education research, and much more. You can knock some of this down by stressing the endogeneity of education, but at the end of the day the pile of evidence, and the diversity of its directions, is simply too overwhelming.