Tuesday, September 30, 2008

Human capital

The Role of Education Quality for Economic Growth
Eric A. Hanushek and Ludger Woessmann (http://ssrn.com/abstract=960379)

The figure shows, by country, the share of students that scored very low (< 400 rough PISA equivalent, "scientifically and mathematically illiterate") or very high (> 600) on cognitive tests administered over the last 40 years. The results give a good indication of the quality of human capital in the country's workforce. Click for larger version.

From the paper:

...To create our measure of quality of education employed in this study, we use a simple average of the transformed mathematics and science scores over all the available international tests in which a country participated, combining data from up to nine international testing occasions and thirty individual test point observations. This procedure of averaging performance over a forty year period is meant to proxy the educational performance of the whole labor force, because the basic objective is not to measure the quality of students but to obtain an index of the quality of the workers in a country.

If the quality of schools and skills of graduates are constant over time, this averaging is appropriate and uses the available information to obtain the most reliable estimate of quality. If on the other hand there is changing performance, this averaging will introduce measurement error of varying degrees. [i.e., younger workers in developing countries probably have better skills than indicated in the data.]

More PISA fun.


Dog of Justice said...

The Iran result boggles me a bit. Have practically all the smart people left the country already?

Simfish InquilineKea said...

Finland's results are especially surprising. One would expect that its egalitarian school system would also simultaneously depress exceptionally high scores on such tests, but this doesn't seem to be the case here - Finland is still higher than most European countries.

Steve Hsu said...

DoJ: I don't know what the literacy rates, etc. are for Iran. They might be in pretty bad shape. Note I cut the figure in half -- there are many countries below Zimbabwe :-(

simfish: yes, there is something mysterious about those Finns :-)

Luke said...

I see our ally Saudi Arabia is the absolute bottom of the barrel.

That figure is poorly done. I wish they'd given a table so I could plot it myself.

Anonymous said...

I think your fascination with test scores is a bit excessive. While they do mean something, there are many examples where countries that don't test well succeed. Notice Israel is at the bottom of the list. They also don't do well on the same math and science olympiads that correlate with your list. But they do produce top-notch scientists.. it's just exams aren't too big over there.

Also these just measure math and science skills. While this might appeal to you personally, other forms of "intelligence" might contribute to the success of a society. Lastly, there are big regional differences within a country. I've heard Iowa would beat every country on the list.

Steve Hsu said...

I've never claimed tests tell the whole story -- they are an imperfect tool. But they are all we have for quantitative comparisons.

I don't know why jews in the US test so well but Israelis don't. I suspect it has to do with the population breakdown (Mizrahim and Sephardim = about 39% of the national population; Arabs 20% of the population; Ashkenazim are only 37% -- according to Wikipedia). If some of these subgroups don't perform well it can change the national averages a lot. Keep in mind the tests I have looked at here look at nationally representative samples.

I grew up in Iowa! I don't think they would beat every country on the list, but they would do pretty well.

The main question is whether the tests are statistically predictive (in this case on a country by country basis), and I suspect they are even if there are some exceptional cases. Of course for economic growth there are many other factors like culture, institutions, historical factors, etc. But I think "human capital" is a fairly well defined input.

Anonymous said...

I'm not sure if I'm buying this bimodal theory for Israel. If that were the case there would be large bars on both sides... but it looks pretty similar to some of the other ones near them on the list. Even if you take 1/3 of them to be Askenazi that's over 2 million people. Post-world war II that's about how many Jews there were in the Soviet Union. The supposedly race-free society made them carry identity cards, so there was a pretty accurate count. Now they were pretty good at these exams in their day, dominated the olympiads and so on.. but now most of them have emigrated to Israel where their kids don't do so great on the same tests. But the young generation is much more technology-savvy and generally more valuable employees than their parents were. High-tech is booming over there. I'd consider them more valuable as "human capital".

Steve Hsu said...

I don't know the full answer to your question, but you might find this of interest:


Note admission to Talpiot is based on --gulp -- tests!

Suppose I have a country with a high performing group that is 1/3 of the total population and the rest a low performing group. Let's say that 15% of the high performing group scores above 600 on PISA, but only a few percent of the others. Then the blended result will be a bit over 5% success, which would bring this hypothetical country in line with many other countries.

Regarding Olympiads, you need a national program for training the teams. Talent is not sufficient to do well unless you also have institutional support for training, which the US, Russia, China, etc. all have, but not all countries (e.g., India).

Seth said...


This is OT, but I thought you might enjoy reading -- and perhaps blogging about -- this op-ed piece in the NYT.

Anonymous said...

One thing that this test certainly did not take into account is the portion of people who did not even take the test. For example, India's score is somewhat respectable in the low end (even though the high end is bad). However, it has 40% illiteracy rate. Surely no one who is illiterate will be taking the test. By this count, some of the countries with high illiteracy rate will fair worse than the scores indicate.

Blog Archive