Discover more from Just Emil Kirkegaard Things
Scientific productivity by country
The relatively subpar performance by USA and China
I've previously written about scientific misconduct by country, but what about good conduct, that is, (good) science? There are various ways to count this. The simplest is to count the number of publications published by researchers working in a given country. Wikipedia has a list of these based on Scimagojr which counts "citable documents". This probably means those with DOIs assigned or something like that. The top list won't surprise you too much:
So in terms of who is leading the world in science, one might say it is China and the USA. China has been gaining rapidly in scientific productivity over the last few decades.
Here we might say that China produces a lot of science, but not so much relative to how many residents it has. China has about 4.2 times more people than USA, so one could simply adjust for the data by dividing the publications by the population in the same years. Doing this gives this different top 25 (which I also posted a variant of on Twitter yesterday):
Some things to note. First, there's a few tiny city or islands states near the top. One might suspect the Vatican City produced mostly theology, but if one checks, this turns out to be not true:
I think rather the explanation is that the Vatican city has a very elite population of men, and no women or children to drag down the average.
Second, skipping the not-so-real countries, we get to Switzerland in position 4, the real number 1. Their residents produced 5.9k citable documents per million people. The next few countries are the Nordics and other west Europeans. But no USA or China. In fact, USA is in position 45 with a rate of 1.8k and China is in position 77 with 690. In other words, though these countries produce the most science, their rate of output is actually not very impressive, being 0.31 and 0.12 of Switzerland's. This was quite surprising to me. I had expected that given the massive amount of science funding in the USA and their numerous elite institutions, they would be quite productive on average, but that's not at all the case.
Next up we might ask, what are the historical trends? Scimagojr provides data from 1996 to 2022, so let's look at the per capita rates over time. I've aggregated countries to reduce clutter:
I used the existing regions in the dataset, but split out countries of special interest (Nordics, USA, China, India, Japan/South Korea). It can be seen that USA was close to the other West Europeans in 1996, but has since declined in relative terms, and been overtaken by the rest of West Europe. The Nordics have skyrocketed, and the other Anglos (UK, Canada, New Zealand, Australia) are also doing well. China's productivity is unimpressive, though they have almost caught up to the East Europeans and the Middle East. Probably China will catch up to Japan and South Korea in the coming few decades and stabilize there. East Europe experiences extreme brain drain to the richer parts of Europe, which is probably why it's lacking behind so much. Still, USA's relatively subpar performance is extraordinary (to me). America ranks together with Taiwan (rank 43), Lithuania (44), and France (48). Germany isn't far ahead either (39), but is somehow still beat by Greece (36).
What about the quality?
It is possible to game the above statistics by producing tons of publications in paper mill journals, or other "citable documents" to one's list. China's poor position suggests this strategy isn't too effective and probably doesn't affect the rankings much. But maybe we need a better index of quality. One idea is to look at the "Nature index". Anatoly Karlin analyzed this data in 2020, but didn't use per capita adjustments. Here's how the absolute values look like over time:
The Nature count index is defined as:
The Nature Index is one indicator of institutional research performance. The metrics of Count and Share used to order Nature Index listings are based on an institution’s or country’s publication output in 82 natural-science journals through 2022, in 2023 64 health-science journals were added to the Index. The 146 journals in the Nature Index were selected on reputation by an independent panel of leading scientists in their fields.
So, it doesn't have that much to do with Nature, the journal, except for the name. Still, these are not just any journals, but those "selected on reputation", i.e., not paper-mill journals. It's also only natural science, so no theology or Islamic studies, or gender studies, or autoethnography. Note that the version of the dataset used previously by Anatoly Karlin started in 2012, but these data are now missing from the website. The data starts in 2015, but the 2015 data are not on the same scale either, so I omitted them.
Anyway, the more interesting version is looking at the per capita data. These were computed as before:
We see Switzerland, Iceland, Denmark in top again, but this is still very cluttered, here's a version with aggregated data:
There's a spike in the data for 2022. I speculate this may be related to COVID publications, which were maybe more easy to do in regions with good government health data. Nordic registers provide exceptionally good COVID data, hence their big spike. City-state Singapore manages to beat the Nordics, mainly because Norway and Finland perform relatively poorly compared to Iceland, Denmark, and Sweden.
But the important question here was whether USA still performs poorly when we use more quality controlled scientific publications. The answer is yes. It performs slightly worse than "rest of Western Europe" which is France, Belgium, Netherlands, Luxembourg, and Ireland.
Why isn't USA number 1?
Why is America doing so relatively poorly? It enjoys a huge brain drain from the rest of the world, which should give it a big advantage. What about funding? Wikipedia has the numbers:
Apparently, the US doesn't actually spend that much money on research in comparison to their GDP as other countries do. India spends a lot per capita, but its results are very poor. The top performers in productivity don't seem to spend a particularly large amount on research: Switzerland 3.3%, Iceland 2.4%, Denmark 3.3%, which is about average for the top spenders. Israel spends the most, probably because of military related research (all it's neighbors hate it).
My guess is that the US suffers from having a quite ignorant culture in general, with a focus on sportsball and making money, as well as racial conflicts and low skill Latin American immigration. This offsets it's considerable advantages in brain draining other countries. China performs poorly for the usual East Asian reason of low individualism and sky-high general dishonesty. I speculate a large chunk of this has to do with communism. Japan, Korea, Taiwan, Singapore etc. never had communism and they do far better, but still not up to North European levels, despite having a higher average intelligence. One could probably try to test various of these explanations, but that is left as an exercise to the reader.