Analysis Of China’s PISA 2009 Results

As human capital is so important for prosperity, it behoves us to know China’s in detail to assess whether it will continue converging on developed countries. Until recently the best data we had were disparate IQ tests (on the basis of which Richard Lynn’s latest estimate is an IQ of 105.8 in his 2012 book Intelligence: A Unifying Construct for the Social Sciences) as well as PISA international standardized test scores from cities like Shanghai and Hong Kong. However, the problem was that they were hardly nationally representative due to the “cognitive clustering” effect. The Chinese did not allow the OECD to publish data for the rest of the country and this understandably raised further questions about the situation in its interior heartlands, although even in 2010 I was already able to report a PISA representative saying that “even in some of the very poor areas you get performance close to the OECD average.”

As regards Chinese intelligence

Happily (via commentator Jing) we learned that the PISA data for Zhejiang province and the China average had been released on the Chinese Internet. I collated this as well as data for Chinese-majority cities outside China in the table below, while also adding in their PISA-converted IQ scores, the scores of just natives (i.e. minus immigrants), percentage of the Han population, and nominal and PPP GDP per capita.

Reading Math Science Average (native) IQ (native IQ) %汉族 GDP/c (n) GDP/c (P)
China* 486 550 524 520  ~ 103.0  ~ 91.6% 5,430 8,442
China: Shanghai 556 600 575 577 589 111.6 113.4 99.0% 12,783 19,874
China: Zhejiang 525 598 567 563 ~ 109.5 ~ 99.2% 9,083 14,121
Hong Kong 533 555 549 546 557 106.9 108.6 93.6% 34,457 49,990
Macau 487 525 511 508 514 101.2 102.1 95.0% 65,550 77,607
Singapore 526 562 542 543 550 106.5 107.5 74.1% 46,241 61,103
Taiwan 495 543 520 519 534 102.9 105.1 98.0% 20,101 37,720

* Twelve provinces including Shanghai, Zhejiang, Beijing, Tianjin, Jiangsu totaling 621 schools, 21,003 students. Results have been released for Shanghai, and later on for Zhejiang (59 schools, 1,800 students – of which 80% were township-village schools) and for the 12-province average.

(1) Academic performance, and the IQ for which it is a good proxy, is very high for a developing nation. Presumably, this gap can largely be ascribed to the legacy of initial historical backwardness coupled with Maoist economics.

(2) The average PISA-converted IQ of the 12 provinces surveyed in PISA is 103.0. (I do not know if provincial results were appropriately weighed for population when calculating the 12-province average but probably not). We know the identities of five of the 12 tested provinces (Shanghai, Zhejiang, Beijing, Tianjin, Jiangsu). They are all very high-income and developed by Chinese standards. Furthermore, these five provinces – with the exception of Tianjin – all perform well above average according to stats from a Chinese online IQ testing website.

The provinces of Jiangsu and Zhejiang also have a reputation in China as gaokao powerhouses.

(3) The Chinese average as given by PISA therefore appears to have an upwards bias, as at least a third of the tested provinces – Shanghai, Jiangsu, Zhejiang, Beijing – are at the very top end of the Chinese IQ league charts. As such, the true IQ average for China is likely closer to 101-102.

(4) The very high score of Shanghai (111.6) is surely for the most part a reflection of its long status as a magnet of Chinese cognitive elites. This may well be true for Hong Kong (106.9) too although perhaps to a lesser extent. But the IQ of native Taiwanese is 105.1 even though the Han Chinese there are substantially interbred with lower-IQ aborigines. Singapore (107.5) too drew Chinese cognitive elites, and quite consciously too – their immigration policies were (are) de facto cognitively elitist – but on the other hand, this is counteracted by their large, lower-IQ Malay and Indian minorities. Regardless, one cannot escape the conclusion that with the (unexplained) exception of Macau, all developed Han majority regions have IQ’s in the 105-110 range. Likewise with other East Asians, such as native Koreans (106.6) and native Japanese (105.3). This means that there is a 5-10 point IQ gap between developed East Asian regions and the Chinese average.

(5) The biggest gaps between China and Chinese enclave regions are typically where we can reasonably hypothesize a “cognitive clustering” effect, so minus that the current gap is probably closer to 5 points. This means that China very likely still has the potential to raise its average IQ by c. 5 points via the Flynn Effect.

(6) A side-consequence is that this presents a serious challenge to Ron Unz’s theory of The East Asian Exception to Socio-Economic IQ Influences.

As regards Chinese intelligence in global perspective

Below is another table with a list of countries representing a typical sample of the developed countries that China is striving to become; and the emerging nations (BRIC’s and SE Asian) with which China is typically compared.

Reading Math Science Average (native) IQ (native IQ)
Korea 539 546 538 541 544 106.2 106.6
Japan 520 529 539 529 535 104.4 105.3
China 486 550 524 520 ~ 103.0 ~
Germany 497 513 520 510 533 101.5 105.0
United States 500 487 502 496 502 99.5 100.3
Russia 459 468 478 468 477 95.3 96.6
Thailand 421 419 425 422 422 88.3 88.3
Malaysia 414 404 422 413  ~ 87.0 ~
Brazil 412 386 405 401 399 85.2 84.9
Indonesia 402 371 383 385 378 82.8 81.7
India* 327 345 337 336 ~ 75.4 ~

* Average of Tamil Nadu and Himachal Pradesh.

(1) Assuming that average Chinese IQ is now 101-102:

  • Means that it is approximately equivalent to the German IQ of 101.5 (with the typical East Asian bias towards better numerical and worse verbal scores).
  • As of today, this IQ level is still somewhat below those of other developed East Asian nations be they Korean, Japanese, or Han majority. It is also slightly below the results of Australians, Canadians, native Germans and white Americans; and approximately equal to the results of native Britons and French.
  • It is head and shoulders above other SE Asian “tigers” whose average IQ’s are in the high 80’s (Thailand, Malaysia) or low 80’s (Indonesia).
  • Relative to the BRIC’s, the Chinese average IQ is substantially ahead of Russia (95.3) and greatly ahead of Brazil (85.2). As for India, whose average IQ is 75.4 according to PISA results from two fairly rich provinces, there is simply no comparison whatsoever. As I have indeed pointed out on numerous occasions.

(2) Needless to say this is an extremely good result that practically ensures convergence to developed country levels within a reasonable time frame. This is especially true because as is almost always the case, there exists a positive feedback loop with greater development pushing average Chinese IQ to its genetic “ceiling” of approximately 105-108. That in turn will further raise the capacity of Chinese labor for skills absorption and even greater productivity.

Addendum 8/15: The commentator Jing graciously provided the list of all the 12 Chinese provinces that participated in the PISA 2009 study. They were: Tianjin, Shanghai, Beijing, Jiangsu, Zhejiang, Jilin, Hubei, Hebei, Hainan, Sichuan, Yunnan, Ningxia.

This allowed me to make an interesting conclusion. No matter whether you weigh the provincial IQ scores above by population or not, the difference between the 12 provinces and China on average is only about 0.5 points in favor of the 12 provinces. This means that the PISA sample is actually pretty good – and that China’s PISA-derived IQ is in fact about 102.5 or so.


  1. Fantastic post! If the New York Times or the Wall Street Journal were even remotely competent, they’d be racing each other right now to turn your data into an immediate front-page story, and maybe even give you full credit for it.

    One of the astonishing things that really jumps out from this data is that although Europe still has 5-10 times China’s per capita GDP, China’s average PISA scores are higher than that of almost any European country. In fact, even if we use your analysis from May, and look at only the native-born European PISA results, China’s still ahead of everyone in Europe except Finland, Germany, Switzerland, and the Low Countries. And I think a substantial fraction of Chinese probably still have an income of less than $1000 per year.

    I don’t know if you ever saw my “China’s Rise, America’s Fall” article from a few months ago, but I don’t think this data will persuade me to retract my analysis:

    • Thanks. It would be nice if NYT/WSJ noticed but not waiting with baited breath for that… 🙂

      It is of course remarkable. That said I would caution against using 520/103 as the definitive estimate of Chinese IQ. As noted above, what are probably the four cleverest Chinese provinces (1/3 of the 12 surveyed) go into that number, whereas in China itself they together only account for 10% of the population. So the real average China IQ is likelier comparable to that of native French, Britons, Poles, Swedes. Which is still very impressive of course.

      I would also add that IMO the actual economic difference between Europe and China today isn’t of the order of 5-10x. Chinese GDP is probably underestimated, on consumption indicators like meat or Internet penetration they are fast becoming an upper middle country, and their CO2 per capita emissions recently converged with those of the EU (in large part because they have a lot more polluting factories/coal plants but still). Two years ago average Chinese manufacturing wages were $400 / month. So I would guess there are now very few people with less than $1,000 / year income in the cities. Poverty remains substantial in the countryside however, although nothing on the scale of what you’d see in a place like India.

      I did see and read your article. It was very cogent and like you I have little patience for the China doomers who are saying it will fall any day now while studiously not noticing far bigger problems in the West. I am a “Sino Triumphalist” too 🙂 and I fully believe that the US will be knocked down to second place by China economically very soon (2015?), in terms of the global military balance by 2025-30 perhaps. That said, I part with you on some of the causes. IMO in most cases corruption is more symptom than cause. I don’t want to write a mini-essay on this but IMO the crux of the matter is that what we have now is a doubling of the so-called Golden Billion of privileged consumers (because China is joining it) while the global stock of resources remains more or less constant. Something has to give and that something is the living standards citizens and governments of developed nations living absurdly beyond their means.

      • Yes, that’s a perfectly valid point about the numbers being from just a few of China’s brighter provinces. I’ll admit I hadn’t closely read the details of your entire posting before I jumped the gun and added my enthusiastic comment.

      • “Chinese GDP is probably underestimated,”

        Very good point. I have been arguing the same based on meat consumption ect.

        Just add another factor into equation. Supply/demand dilemma for large nations. Any thing China sell become cheap. Any thing China buy become expensive. Small nations have easier time to become rich for opposite reason.

        When we consider value added manufature, it is very hard for China to achieve that. As result, smar phone and computer should be very expensive but not. It is because China sell them in large quantities. Natural resource like oil, should be cheap but not. Again China buy them in large demand.

        • All of these developing countries have burgeoning underground economies, which in many cases are actually larger than the formal economy. Even developed countries like Italy have very large underground economies (20 to 30%, I last read). We can only imagine how large China’s is.

  2. Good posting. Correlating with this is an article this week in the economist about manufacturing in India. It seems that with a billion in population, the types of manufacturing that actually succeeded in India are the automated kind with high tech machines that employs few higher quality people, which jives with the high tech outsourcing theme.

    I agree that while corruption is huge in China, it has not and will not impede their progress. Lacking in natural resources should also not be as big a problem as you think, We humans are very clever in coming up with substitutions. Take Chrome, it was a dwindling resource in the seventies as the only major mines were found in South Africa and Russia. With the prices shooting up, the auto industry came up with uni-body construction and no longer use much Chrome in their bumpers. I think the cars look better as a result. There are numerous examples of this as prices change.

  3. ironrailsironweights says:

    I’m surprised that Brazil does so poorly.

    • I am too: Brazil is a fairly wealthy country with a burgeoning middle class. That being said, being half Black/Mulatto doesn’t help. Still, Brazil SHOULD have way more high scorers than it currently has. I mean, it is building a burgeoning industry, and it can compete with other nations in certain areas of high technology (notably aircraft production). Why does it have so few high scorers?? Are the samples being tested just not representative??

      I’m particularly shocked at Chile and Argentina’s performance (among others): these countries are both majority White, and IQ tests in general show an IQ in the mid 90’s. These countries are pretty wealthy, with a high standard of living. I mean, how is it that these countries perform poorer than Turkey?? Something just ain’t right…

      • I read this in the recent FP report about Brazil – Chinese investor was observing Brazilian kids playing football and he remarked – “in China this wouldn’t happen, in China these kids would be studying”. This is mainly why I don’t believe PISA scores are a good estimate of IQ. Brazilian kids would rather play football than study and from what I’ve read Brazil has a high dropout rate. Brazil need educated people, of course, and she is atracting a lot of young graduates from Portugal.

        • We seem to see plenty of countries (Chile, Argentina, Romania, Moldova, Russia, Kazakhstan, etc.) that perform well below what their IQ’s would predict. We don’t see any countries performing well ABOVE what their IQ’s would predict (save possibly Finland). But I agree with you: using PISA as a direct measure of IQ, right off the bat, is probably not such a great idea…

        • This assumes that Brazilians are less bright and drop out more than Chinese simply because they prefer playing football to studying. On the other hand it is just as possible that they prefer play football because they are less bright in the first place.

          The latter is indeed my view.

          Contrary to popular perceptions, it is actually duller people (the people who score less on tests) who use far more brainpower for intellectual work. This has been established with MRI brain scans. As such they find it far less enjoyable to do brain work, so they play more football. Quite simply people do what they enjoy doing.

          @Ryan G.,
          The PISA/IQ correlation is 0.7. This is quite solid. PISA does not in my opinion underestimate Russia to any significant extent. As for Romania/Moldova, one has to bear in mind that lots of their brightest citizens have been leaving those places for two decades, and many decades, respectively.

      • Argentina took the effort to track down and test the truants and drop-outs who constitute 39% of its school-age population. Without them its PISA average would be at around 440 (=91) right where Chile and Bulgaria are. That is quite logical.

        I’ll have a post on Brazil. But as far as I can see, one has to adjust for the pardos and Blacks who make up 50% of the population (though with the adjustment that Brazilian pardos and even Blacks are far more Europeanized than US Blacks). Then there’s the as yet uncompleted Flynn Effect.

        Turkey does indeed do surprisingly well for its economic position. It’s now close to catching up with Greece. This is especially surprising as the rate of consanguineous marriage there is significant, which has a crippling effect on Arab IQ’s. Maybe as elsewhere it’s a matter of sampling. Obviously performance will be much higher in the cosmopolitan west of the country as opposed to its Third World Kurdish east.

  4. The 12 provinces/municipalities tested are as follows followed by / population, million / nominal per capita gdp, current USD

    Tianjin / 12.9 / 13058
    Shanghai / 23 / 12783
    Beijing / 19.6 / 12456
    Jiangsu / 78.7 / 9545
    Zhejiang / 54.4 / 9083
    Jilin / 27.5 / 5933
    Hubei / 57.2 / 5434
    CHINA AVERAGE / 5432
    Hebei / 71.9 / 5198
    Hainan / 8.7 / 4459
    Sichuan / 80.4 / 4046
    Yunnan / 46 / 2935
    Ningxia / 6.2 / 5015

    • Thanks a lot for this Jing. Do you know if the calculation of China’s average (520) was weighed by each province’s population or not?

      • Nope sorry, can’t help you there.

      • Richard Sharpe says:

        Note, that as Jing points out, but you seem to have elided in your table caption above, Beijing, Tianjin and Shanghai are not provinces. They are municipalities/cities.

        I tend to think that the downward mobility effect and selection caused by 2000 years (approximately) of the civil service exams has flushed a lot of alleles causing IQ lowering out of the population.

        Do you have any numbers for the variance?

        • Marie Arouet says:

          These four “municipalities” are bigger than any but a handful of states in the U.S. with populations of 28.8, 19.6, 12.9, and 23 millions.

  5. A great entry , AK!

    Han Chinese, like the term “white Europeans”, are not all the same. Thus population-weighted average could also be misleading along with simple average in calculating the scores. Actually these 12 regions were excellent choices. Some background knowledge (feel free to correct me if I am wrong here, Jing):

    1. Tianjin / 12.9 / 13058 : average northern Han Chinese. Mid level in Gaokao. China’s “Norway”?

    2. Shanghai / 23 / 12783 : majority Yue (or Wu) Han Chinese + minority immigrants. Mid level in Gaokao. China’s “Belgium“?

    3. Beijing / 19.6 / 12456: average northern Han Chinese+ minority elites from allover China. Upper-mid level in Gaokao.

    4. Jiangsu / 78.7 / 9545: typical Yue Yue (or Wu) Han Chinese + minority immigrant (at Northern Jiangsu). One of the toppers in Gaokao.

    5. Zhejiang / 54.4 / 9083. typical Yue Yue (or Wu) Han Chinese. One of the toppers in Gaokao. China´s “Germany“? (together with Jiangsu)

    6. Jilin / 27.5 / 5933 by and large ethnic Korean Majority + minority Manchus and Han. Mid level in Gaokao. The ideal case for expolarating China´s “average” Flynn Effect points and current IQ “average” , given its relations to both South Korean scores and those of other regions in Chinese Gaokao.

    7. Hubei / 57.2 / 5434 typical central Han Chinese. Upper middle level in Gaokao. China’s “France”?

    CHINA AVERAGE / 5432

    8. Hebei / 71.9 / 5198 typical northern Han Chinese. Mid level in Gaokao. China’s “Poland”?

    9. Hainan / 8.7 / 4459 typical southern Chinese close to Northern Vietnam, not typical Han on average. Low level in Gaokao. China’s “Portugal”?

    10. Sichuan / 80.4 / 4046 typical Western Han Chinese + minority Tibetans. Mid level in Gaokao? China’s “ Britain+France”?

    11. Yunnan / 46 / 2935 majority 3 o 4 dozens non-Han + minority Southern Han Chinese. Very low level in Gaokao. China’s “Greece”?

    12. Ningxia / 6.2 / 5015: Majority Muslin and tiny minority Han. Very low level in Gaokao. China’s “Albania”?

    13. would be interesting to see Fujian: which forms the majority of Han Chinese of Taiwan, and is one of the core forces of “historical commercial legacy of Han Chinese in SE Asia”. High level in Gaokao. China’s “Italy”?

    14 also would be nice to see Shangdong: typical North-eastern Han Chinese, many of them have been migrating to Japan historically. High level in Gaokao. China’s “Scandinavia”?

    It could be not accurate, AK, to estimate China’s current “average”IQ as 102.5 or 103 IMO, since there is an obvious “problem” in PISA data set, namely:

    * Shanghai has always been at mid level in China’s Gaokao.

    * Shanghai has never surpassed either Zhejiang or Jiangsu in Gaokao (in both Science and Art). In fact, Zhejiang & Jiangsu have always be way way above Shanghai.

    Therefore, since Shanghai’s score is honest one ( proven by its relations to HK, Taiwan, SK and Japanese scores), Zhejiang’s PISA scores are largely underestimated! – it should have been higher, and likely much higher (surely in Math) than those of Shanghai, agree?

    The reason for this “inconsistency” is simple, as PISA footnotes disclosed, that 80% of the samples were taken in rural areas of Zhejiang, subjectively (and I suspect the similar happened in the next door Jiangsu).

    Therefore, using Zhejiang’s PISA 2009 scores to represent China’s topline IQ scores to get China’s current “average” score, as you did, is not entirely accurate. This is even more so considering that PISA 2009 missed Gaokao heavyweights such as Shangdong and Fujian, both being heavily populated Han Chinese heartlands.

    Logically, if both PISA 2009 scores and decades of Gaokao stats are to be trusted, then the true PISA scores (not 80% subjective selections but random samples) of both Zhejiang and Jiangsu (particularly the southern part) should reflect IQ of about 115 or even more. Yes, I believe that it would logically* surpass that of the Ashkenazi Jews. I am not surprised then. Zhejiang and Jiangsu have enjoyed being “the genius capitals of China” for thousands of years.

    * As Caucasoid, Ashkenazi Jews should have smaller average cranial capacity, higher hormone level, faster growth rates (e.g.bones, dental ,etc. etc.) than the East Asians, if Phillips Rushton’s 3-way theory is correct, which should be. Thus there is no reason to believe, theoretically speaking, that the average IQ (either current or +Flynn pints) of arguablely the smartest sub tribe of 1.3 Han people like Jiangsu and Zhejiang (Wu/Yue tribe Han Chinese) can not surpass that of Ashkenazis.

    • Thanks, SP.

      Is there any place on the Chinese web that has regional averages for Gaokao scores? I’ve looked for half an hour today but didn’t find anything relevant.

      Maybe it’s like Russia and they don’t reveal regional averages. Perhaps not as according to this post and the one in the other thread gaokao scores are only given in terms of some regions being reputed as doing better than others.

      • You’re welcome AP.

        No idea on Gaokao detail stats though. What I know is from my parents and some of my Chinese friends.

      • Hi AK,
        I came across this website showing the result of 2012 Gaokao scores in 31 provinces, hope this will help!
        Note: it’s not ranked by highest scores, but the highest average scores to get to a A-class university. For example, 78.53%(文科一本得分率) for the first line should mean the average score to get to A-class university art faculty is 78.5.

        From this news though, it’s actually 广东GuangDong (Canton) Province eats the cake in 2012 Gaokao lol. ZheJiang ranks 9 in art, 5 in science (yeah it’s still good), Shanghai is 10 in art and 16 in science (yes it IS around average).

        • Thanks, but I don’t think it tells us very much. Shanghai (and Beijing, which is marked absurdly low) have a lot of good universities themselves so I would imagine that locals would have an much easier time getting in than the rest of the country?

        • “谈到腐败我想发表点小意见:中国的教育改革一直嚷着很凶,但是落到实处是很难的,比如外地学生要想进入清华北大这样的学校要比北京多出200分来,大家说说这是多大的差距,这个且不说。”

        • Marie Arouet says:

          You mistake the numbers. They represent the cut off for admissions to a university inside the province. The more university positions in that province, the lower the score needed to get into one.

  6. I’m not sure that your estimate of china’s average is correct, as provinces with large populations seem all to have higher than 104 IQs. (70-80% of China’s population). So I guess 105 is more possible than 102.

    • Marie Arouet says:

      I actually plugged into a spreadsheet and crunched the numbers using the IQ map given above and the population of each provinces of China. The IQ for China today is 105.3. Once China’s GDP rises to to the level of Hong Kong, you can probably add another 3 to 6 points to that.

  7. Marie Arouet says:

    If you take out the reading score for China, you basically have China and Korea with essentially the same top score with IQ of 106. The reason this is reasonable is that in rural China they don’t know what is an ATM or an airbag (two questions in the 2009 PISA) where as the Koreans all know what these are. The rural Chinese (80% of the tested subjects) have a comparative disadvantage in this area.

  8. You might want to compare/contrast the China IQ map with China’s population density map. Then you might discover your 101-012 IQ estimate is low, considering China’s most densely populated costal area have average IQ of 104-108.

  9. The Zhejiang number from the 2012 PISA has once again been leaked. This time it’s 623 in math, 570 in reading, and 582 in science. Better than Shanghai. But here is the shocker from article.

    “The provincial Education Examination handled organization implementation of PISA test measured the province. The spokesman Feng Chenghou said, the 2012 PISA test, the same school in 2009 are tested, sampling tests are ordinary high school student, none of the key schools participated, rural and township schools still account for about 80%, and the students are mostly from the 9th grade.”

    Note: China’s compulsory education ends at 9th grade, afterwards student can choose to enter college bound high schools, career bound vocational schools, or directly into apprenticeships. The last two options has gotten a lot flak in the international media as child labor abuses. Since the sampling is mostly at 8th grade, the Zhejiang numbers are not liked to be skewed by filtration effects at 9th-10th grade divide.

  10. Your analyzed data is probably far from the actual figure of China. In 2005, China randomly selected around 1200 children from age 8-10 from each province and tested the children’s average IQ as part of an iodine deficiency study. According to the results:

    Inner Mongolia——105.1±13.8
    Hainan——90.7±15.9 (iodine deficient)
    Tibet——77.3±16.8 (iodine deficient)

    In my opinion, the result has more to do with the level of economic development rather than demographics.

    • Based on your provincial IQ scores, I’ve made a prediction of China’s full country PISA score.
      (With a fair number of assumptions : The IQ numbers are relatively accurate, IQ scores correlate with PISA scores, educational quality is similar throughout China, and IQ scores are gaussian)

      It’s not too different from the US except for math.

      Test Avg Level1 Level2 Level3 Level4 Level5 Level6 Level7
      Science 514 3.39 10.92 22.53 29.52 23.91 8.87 0.97

      Test Avg Level1a Level1 Level2 Level3 Level4 Level5 Level6 Level7
      Reading 503 1.59 2.50 10.93 23.71 29.43 23.14 7.94 0.86

      Test Avg Level1 Level2 Level3 Level4 Level5 Level6 Level7
      Math 531 7.05 10.71 16.54 18.89 19.48 15.75 11.68