I Easy criticism 1
First, easy, and almost boring criticism: why is there such a big surprise that only a very small share of people who studied on average only two years of a foreign language in school happens to speak it very well? Is it necessary to argue that two years of foreign language study is thousands of miles away from being enough for acquiring a good command of a foreign language? Two years of a given foreign language usually correspond, at best, to the completion of the beginner's level.
Also: is it necessary to argue that mastering a foreign language (speaking it "very well") requires way more than school: it includes movies, music, meeting natives of that language, if possible spending time abroad?
II Easy criticism 2
Over 4000 respondents, «25.7% of respondents speak a language other than English», and the survey «only followed up with the other questions if the answer to the first question was "yes"». Then, people who actually stated their foreign language skills were about 1000. Can a 1000 individual sample be representative of a 312 million people country? Maybe it can... but I have huge doubts about that: it sounds like an epic assumption to me. But note that the starting point of Caplan's article are statistics based on such a sample, while Caplan's conclusions apply to the whole population (the «Americans»).
A related not-so-easy criticism: Caplan's conclusions apply to the «Americans», whether they actually speak a foreign language or not. However, the sample is (at best) representative of only those who do speak a foreign language. Do conclusions based on that sample apply, again, to the whole population? This is relevant: people who don't speak a foreign language may still have benefited with foreign language education, had it been, say, of higher quality - had they, actually, learnt some. The fact that someone does not speak a foreign language doesn't mean their demand for it is zero...
Relevant but not so obvious: it is possible that a respondent who does not speak a foreign language at the time of the survey did speak some in an earlier period of her life, benefited from it, and eventually lost it. This is specially likely in the case of old people (see more about age below), as foreign language fluency easily drops with time if there is no practice, and as personal interests change as one's life goes on. There's also the case of emigrants who stayed abroad for a relatively short period, and who have returned for good to their country of origin long time ago: they may not speak a foreign language now, but they did it in the past, and may have greatly benefited from it.
People who don't speak a foreign language are part of the debate - but they are not considered in Caplan's article.
III Economics criticism
Now, the economics criticism, or, what really matters is utility and demand.
Caplan's article is based on an assumption that is not stated. The assumption is: the utility derived from having a poor command of a foreign language is very low and, thus, the "demand for two years of foreign language education in a fully privatised world" would be very low. In this case, supplying two years of foreign language is inefficient - a waste of money.
There is an alternative assumption: even people with a poor command of a foreign language may well derive a high utility from that skill, to the point that their "demand for two years of foreign language education in a fully privatised world" is high or, even, very high. In that case, the present quantity of foreign language study at school is not inefficient, it is not a waste of money, in the sense that it might correspond, even if only roughly, to the equilibrium quantity that would emerge in a fully privatised world.
While I accept the first assumption has a fair probability of being true, I believe as well that both of them are reasonable. And it is because both of them are pretty much reasonable, that I find Caplan's article not that much convincing.
Let me be a bit more clear about the alternative assumption: the fully privatised society means a place where there are no State owned schools, which are funded by taxation, and which offer their services at price zero or at a "social price". Also, in that society, taxes are low, or there are no taxes at all, and no State owned anything. The utility of people who just speak "not well", or even "poorly", might already be high enough so that in such a society they would have demanded and paid for those two years of foreign language education. If this is true, then foreign education is efficient in our society, at least for them.
Since there is in reality no truly free market for foreign language education - because of compulsory schooling at price zero or social price - it is extremely difficult to estimate the "demand for two years of foreign language" that would exist in the fully privatised world. Thus, we are left with this uncertainty: two years of foreign language in school may be or may not be efficient.
Important too: what was the age of the 4000 respondents? The interpretation of survey's results should be different whether all the respondents were, say, 16, or all of them were, say, 81.
Speaking a foreign language poorly at 16 still leaves a high chance of enhancing that skill some time later. That means the claim "demand for foreign language education is high" has a higher likelihood than if we were talking about 81 year old people. Speaking a foreign language poorly at 81 indicates there might have not been much interest in foreign language learning through an entire life, which points (not necessarily) to a low demand for it and, thus, supports the claim of foreign language education being inefficient.
In short: "speaking a foreign language poorly at 16" does not specifically or strongly point at any of the two alternatives, low demand or high demand; the same at 81 does. But Caplan is silent about the age structure of the respondents.
Furthermore, individuals are different (an à la Palice truth even libertarians sometimes seem to forget). Some of them do enjoy a great utility from
foreign language skills even when the skill is low - Caplan himself provides his own example in his article. This is specially true for those
individuals who continue learning that language. Then, it would be interesting to have at hand a survey that asked the same respondents at two different points in their life, to check whether they continued or not language education beyond those first two years in high school.
People who only poorly spoke a foreign language at 16, but continued their foreign language education, would provide a signal that their demand at 16 might have actually been high, which would give support to the claim that "for some individuals, even a low level foreign language skill might correspond to a very high utility".
I guess the "mistake" in Caplan's article, if I can talk about mistakes in an opinion matter, is assuming (again, not explicitly) that there is a very simple
correspondence between foreign language fluency and utility. If fluency is low, then the utility must necessarily be so low that demand
for it would not justify the supplied levels we actually see in the
real life. While I agree there should be a positive
relationship between fluency and the utility derived (if higher skill
then higher utility), I disagree that a low skill necessarily
means such a low utility that such a person would never buy foreign
language education in the fully privatised world (in that world, people
pay low taxes or no taxes at all, in which case there would be higher
And this is even more likely if one keeps in mind, again!, that individuals
are different: while it is very likely that for some people a low skill corresponds to
low demand (and vice-versa), for some others a low skill might correspond to a pretty high demand.
After all, we have all to start from a zero skill level, even those who mastered a foreign language started from that point. And isn't it accepted (at least among economists) that marginal utility is decreasing? Then, why should anyone assume that the demand for foreign language education after only two years of it is low? Finally, and taking Caplan's assertion on diminishing returns (though he is talking about diminishing returns of money, not of a service's quantity): if returns to education diminish with quantity, what is it more likely at low education levels: high demand or low demand?
Of course, obviously, evidently: if the aim in Bryan Caplan's article is to fight against compulsory education and government spending on education, then one should prefer the assumption "if low skill, then extremely low utility, hence, extremely low demand, and this is true for most if not all individuals, which happen to be not that different from one another after all".
It then should be evident for anyone defending a given political objective: as one qualifies and qualifies even more their favourite assumption, so that it perfectly fits one's political aim, the less the assumption sounds reasonable, or the smaller becomes the validity scope of the assumption, or both (this was essentially the reason I disagreed with one article written by Caplan also on education, see somewhere
below in this blog).
Also, one should prefer the 4000 respondents were all old - but were they? One should prefer that the "demand for two years of foreign language in a fully privatised world" is low - but is it? And how can anyone tell? Where is such estimation?
Under those preferred assumptions, inefficiency of "two years of foreign language education" should be the conclusion. But as I have shown here, that conclusion rests on too many assumptions (and the way these are chosen, and whether to make them explicit or not, depends on each one's own political aims, which is just fine and legitimate).
Of course and obvious as well: if people could only express an opinion after having at hand a complete and sound cost-benefit analysis based on the best possible demand function estimations, only a few would feel their opinions were justified. Opinions are legitimate "even" if they are not supported by the biggest and best economic study, which is usually not available. But, then, opinions have to rest on assumptions. And assumptions can be explicitly stated or not, can be reasonable or not, can be strongly substantiated or not.
The basic assumption in Caplan's article is not stated, it was not substantiated, and while it has a fair chance to be true, I find the alternative at least as reasonable as the reference one. Thus, the article is not much convincing: one has only to keep in mind that what matters is utility and demand.
Foreign language education is a waste of time and money? That depends on the assumptions! If the share of people such that "low skill implies low demand" is big enough, then yes. But how do we know? The survey on which Caplan bases his article is not the answer because it asks about skills, not about the utility derived from those skills. So we are left with two alternative assumptions, both reasonable and, thus, we are left not with one conclusion but still with a question to be answered: is foreign language education efficient?