Literacy and the sex ratio

« previous post | next post »

[Guest post by Richard Sproat]

I was spending a pleasant portion of a Sunday morning reading a shocking article in The Economist on The Worldwide War on Baby Girls. One of the sad conclusions of that article is that the preference for male babies, which in some parts of the world is driving the ratio of male to female births to as high as 130 male births per 100 female, is actually getting worse as education gets better in some parts of the world. One of the points made is that "[i]n China, the higher a province’s literacy rate, the more skewed its sex ratio."

I was curious to see how this trend fared worldwide. I have data on literacy and other socioeconomic factors that I collected from the  United Nations Human Development Programme's set of economic indicators, which I had collected for my forthcoming Oxford University Press book Language, Technology, and Society.  Data on sex ratios is available from the CIA World Factbook.

The literacy data is for adults, defined as those 15 and over. A variety of sex-ratio data is available from the CIA page: I chose the ratios for the population 15 and under, as reflecting sex biases integrated over the past 15 years. Per The Economist, this underestimates the ratio for recent in several parts of the world, where things are only getting worse.

The data, for 176 countries, is plotted below. On the horizontal axis is the sex-ratio for the population under 15, with the higher numbers meaning more males per females. Bear in mind that a ratio of between 1.03 to 1.06 male/female is considered biologically normal. On the vertical axis is the literacy rate for those 15 and over. Note the positive correlation of 0.51: a higher literacy rate correlates with a higher male/female ratio.

For the raw data, sorted by decreasing male/female ratio, see the table.

To put this in perspective, the correlation for the percentage of the population employed in agriculture and literacy is -0.62: this is expected since in many parts of the world agriculture is highly labor-intensive, and children are often employed in agriculture at the expense of schooling. So this strong negative correlation is expected. The almost as strong positive correlation in the new data is both unexpected, and shocking. Note also that none of the countries with a male/female ratio of above 1.1 have a literacy rate that is less than 60%. Indeed, only India has a literacy rate that low; all the other countries with high male/female ratios have literacy rates above 90%. Indeed, as The Economist points out: "In India, some of the most prosperous states—Maharashtra, Punjab, Gujarat—have the worst sex ratios." So one suspects that if one were to plot these data not by country, but with finer-grained information at the state level (unfortunately I do not have those data) then the correlation would be even stronger.

Obviously literacy (and education) does not cause this trend. What it does suggest though is that if the social structure is in place to favor boys over girls, and if the economic means are there, then a better educated population is actually more at risk. This is surprising: one would have thought that a higher literacy rate (and thus better education) would correlate with a more enlightened attitude towards females. Sadly, this seems not to be the case.

[Guest post by Richard Sproat.]


  1. John said,

    March 7, 2010 @ 6:32 pm

    You should plot this the other way, no? With sex-ratio as the dependent variable.

    I suspect, as you perhaps suggest, that what you're seeing is literacy as a proxy for wealth.

    I would also suggest doing a multi-variable analysis on this, esp. as you note a stronger correlation with agricultural activity.

    (Your link to the table doesn't work.)

  2. John said,

    March 7, 2010 @ 6:33 pm

    Now the link works. :-)

  3. Eric said,

    March 7, 2010 @ 6:37 pm

    The geographical grouping of the outliers in sex ratio seems somewhat striking, particularly the Caucasian countries all being near the 100% literacy mark with very high sex ratios. It seems like the high literacy – high sex ratio group is largely composed of Eastern European and Southeast Asian countries while, for example, Caribbean countries seem to be skewed more toward high literacy and low sex ratio. Of course, it's difficult to tell what's going on in the mass of names around 80-100% literacy and sex ratio ~1.04. It might be interesting to see the same plot with some of the outliers cropped off to see what's happening in that confused region.

  4. John said,

    March 7, 2010 @ 6:55 pm

    Eric, if a sex ration of 1.04 is within the normal range, then nothing is likely going on there.

    If I plot the ratio vs adult literacy, I get an R-squared of 0.26 on your data. If I set a lower bound to the sex ratio in order to exclude countries where nothing seems to be going on, I get an R-squared of 0.

  5. David Eddyshaw said,

    March 7, 2010 @ 6:56 pm

    Unfortunately this trend is not too hard to explain on the basis of the availability of prenatal ultrasound and the social acceptability and availability of abortion.

    In most African countries of which I have experience, abortion is illegal, and prenatal ultrasound is available to very few.

    I suppose the positive way to look at this is that, at least, most people in most societies do not readily practice female infanticide.

  6. Tamara said,

    March 7, 2010 @ 6:58 pm

    I think John (above) is right that literacy is just a proxy for wealth. I read somewhere that there are at least two mechanisms at play. The obvious one, especially for a poor country like India, is that wealthier people are more likely to get prenatal care and find out the sex before birth (in India, that involves bribing the doctor since it is illegal to tell parents the sex of the child before birth).

    The less obvious mechanism involves how many children people expect to have. Wealthy people have fewer kids, so the opportunity cost of having a daughter is more. If you expect to have say, six kids, then you can tolerate a few daughters in there. If you're only planning on having two, however, you can only tolerate one daughter. The article said this idea was borne up by the fact that the sex ratio gets more skewed for later children.

    Sorry I totally can't remember where or when I read this.

  7. JS Bangs said,

    March 7, 2010 @ 7:06 pm

    You said: Obviously literacy (and education) does not cause this trend.

    I don't really doubt this statement, but I do feel that it should be argued for rather than merely asserted. Are there mechanisms by which education might contribute directly to a preference for boys over girls? And what evidence do we have that contradicts that notion?

  8. Eric said,

    March 7, 2010 @ 7:11 pm

    Eric, if a sex ration of 1.04 is within the normal range, then nothing is likely going on there.

    Sure, I wouldn't expect to see much of great interest to the question at hand in there. I'm just curious as to how well the geographical patterns that seem to be apparent on the edges (e.g. South American and Caribbean countries toward the low end of normal for sex ratio, Eastern European countries toward the high end) hold up in the middle. It's not something that I expect to deepen my understanding of the origin of the correlation at issue in the original post, just something that struck me as interesting on looking at the plot.

  9. eloriane said,

    March 7, 2010 @ 7:22 pm

    I almost wonder if it's related to the fact that literacy for women is often less than literacy for men. If there are fewer women, they might bring down the average numbers less.

  10. Bill Bennett said,

    March 7, 2010 @ 7:38 pm

    OK, I realise this is a global survey, but I interested to know if these results connect to the idea that in rich countries, women have higher education and literacy levels than men.

  11. Jarek Weckwerth said,

    March 7, 2010 @ 7:44 pm

    Aren't we reading too much into this?

    Firstly, the countries in the top right-hand corner aren't particularly affluent. So I'm not so sure that it's a clear question of the "economic means being there". Of the three Caucasus states, Azerbaijan has by far the highest GDP, and the lowest M/F ratio.

    Secondly, comparing Hong Kong to China also seems to suggest that higher literacy and better economic conditions in fact mitigate the sex ratio. (Assuming, of course, that HK and PRC share at least some "social structure".) Pity Taiwan isn't in the table.

    Thirdly, the graph looks the way it does partially because there are two very long country names in the top right-hand corner (Hong Kong China (SAR) and Korea (Republic of)), making it look busier than it really is.

    Finally, India clearly doesn't fit the pattern, either. For example, Sri Lanka, with a literacy rate of ~90%, and a higher GDP, is squarely in the 1.04 group. And within India, Kerala, with a literacy rate of over 90%, has a sex ratio of about 1.05.

    There are cultural, even regional, factors, but I don't think literacy is one of them.

  12. Melissa K Fox said,

    March 7, 2010 @ 8:04 pm

    Out of tangential curiosity, how accurate do we think the literacy rates are – and does that depend on who's doing the reporting? And on how they're defining "literacy", I suppose – 15+ can spell, or 15+ can read the Times? There are different metrics in different countries, I learn from idly perusing the CIA World Fact Book, which I sometimes do for fun (don't you?), which is why I bring it up – I am wildly skeptical that 100% of adults *anywhere* are literate by any meaningful standard, no matter what the Republic of Georgia says.

  13. D said,

    March 7, 2010 @ 8:21 pm

    Hm, women having their pick of an abundance of men vs. men having their pick of an abundance of women. You'd think that patriarchal cultures would want it the other way around.

  14. Jason said,

    March 7, 2010 @ 8:29 pm

    eloriane: It would be interesting to separately correlate male literacy to sex ratio and female literacy to sex ratio.

    I suspect this may be confounded by socioeconomic status. One could imagine that higher income families often fewer children, raising the average literacy rate and gender-specific abortion rate.

  15. David G said,

    March 8, 2010 @ 3:19 am

    I think that at least some of the correlation can be explained biologically:
    I assume that in more literate countries, men (and women) marry/have children in older age. It is a known fact that males accumulate more inheritable mutations with age (due to the ongoing meiosis, this is known as "male driven evolution"). Now, mutations to the Y chromosome are more deadly to a just-conceived baby than mutations in the X chromosome, because a lethal mutation in the X chrom can be covered up by the X chrom from the mother while for the Y chrom this is not the case. In such a scenario the ratio of conceived babies is normal, but more females are born.

  16. J. Goard said,

    March 8, 2010 @ 4:40 am

    This is surprising: one would have thought that a higher literacy rate (and thus better education) would correlate with a more enlightened attitude towards females.

    I know very little about the specifics of this issue, but it strikes me that many kinds of trend might exacerbate sex ratio and nevertheless reflect "more enlightened attitude toward females". Stuff like less frequent use of dowry, lower societal tolerance for grooming brides as commodities, and especially the right of a woman rather than her parents to choose the most desirable husband. Theoretically, any change which makes a girl less of a vehicle for her parents' interests might increase the sex ratio. That includes a lot of important feminist goals, don't you think?

  17. Anita said,

    March 8, 2010 @ 5:20 am

    I encountered the same fact myself but on a much more micro level when I visted Punjab in 2006. Having grown up believing Punjab was a successful, prosperous and progressive agrarian economy (a rarity in India) I expected to encounter higher than average levels of enlightenment and female emancipation even in villages. I was dismayed to find that one of the biggest issues the government was trying to tackle was that of female foeticide and infanticide. This persisted despite foetus gender checlks being made illegal some years ago.

    Questioning of local officials revealed that a more prosperous rural population meant they could get around the law by bribing private clinics to establish gender of a child and abort it if it was a girl.

    Sadly also given that most women will have an occasion to medically terminate a pregnancy during their life time for social, medical or health reasons – this planned termination of female feotuses was by and large treated with emotional detachment by the people I spoke to.

  18. Jen said,

    March 8, 2010 @ 5:55 am

    I need to point out that in many of these countries with supposed high male-to-female birth ratios, the ratio is actually much lower. For example, in China (PRC), people often wonder about the missing girls…what happened to them? The answer is nothing…the ratio has changed very little since the implementation of the One -child Policy in 1980; it's just that the birth of females is heavily underreported since families want to leave reporting of their one official birth to a male child; otherwise they would be stuck with just one female child. In China, giving birth to females is viewed as sowing crops for other people's families, since females marry and move in with their husbands' families, leaving the woman's parents without any means of support once they reach old age. Moreover, females cannot do farmwork and defend the family against intruders like men can.

  19. Josh said,

    March 8, 2010 @ 6:07 am

    It should not be assumed that birth ratios are the same everywhere unless there is a preference for boys (or girls). Sex ratios vary by latitude: Considering that in the US black people have a consistently lower sex ratio than whites and Asians, one cannot rule out genetic factors.

  20. John Atkinson said,

    March 8, 2010 @ 6:34 am

    @Jen: "Females cannot do farmwork and defend the family against intruders like men can."

    In many peasant societies, women do most of the farmwork. This is especially the case in Africa; I don't know for sure that it is also the case in modern China, but I suspect it is.

    No doubt you're right about women being less able to fight against intruders. This may be an important consideration in some societies where bandits are rife and the police are ineffectual, but this is not the case in either India or China.

    The rest of your post makes better sense, ISTM.

  21. Jen said,

    March 8, 2010 @ 7:17 am

    Thanks John! I was just restating the view that I have heard others express…but you make a good point, especially now with the availability of modern machinery (though many rural Chinese farmers would not be able to afford this luxury).

    I found this in wikipedia, which states Chinese view sons as more helpful for farm work:

    China, like many other Asian countries, has a long tradition of son preference.[27] The commonly accepted explanation for son preference is that sons in rural families may be thought to be more helpful in farm work. Both rural and urban populations have economic and traditional incentives, including widespread remnants of Confucianism, to prefer sons over daughters. Sons are preferred as they provide the primary financial support for the parents in their retirement, and a son's parents typically are better cared for than his wife's. In addition, Chinese traditionally hold that daughters, on their marriage, become primarily part of the groom's family. High male-to-female sex ratios in the current population of China do not occur only in rural areas; the ratio is nearly identical in rural and urban areas.

  22. Vinaigrette Girl said,

    March 8, 2010 @ 8:09 am

    You need to examine the sex ratio in respect of * female * literacy rates, not overall adult literacy. The UN council for Women and Children, as well as a number of other research bodies, link female literacy to lower rates of female infanticide and abortion of female foetuses, along with lower birth rates overall and greater economic independence for women. Girls who are educated are better assets to their families; the link between female literacy and drops in female infanticide is well-established, as is the underlying gender bias in the reported "adult literacy" statistic.


    @Jen, what's your evidence for under-reporting of female births in China? This doesn't square with any demographic evidence I have ever seen published, either in paper or online, so I'm interested.

  23. Jen said,

    March 8, 2010 @ 8:57 am

    Hi Vinaigrette Girl, I found this link for you:'s+boy+bias+means+millions+of+%60missing'+girls-a080509707

    Normally, women give birth to about 107 boys for every 100 girls. But in China, 118 boys were born last year for every 100 girls. The ratio is creeping up: In 1990, it was 112 boys per 100 girls.

    The World Health Organization, in a 1997 report, estimated as many as 50 million Chinese girls are missing since the problem appeared in the early 1980s.

    The births of many girls are simply not reported, Chinese experts say. Instead, parents give them away or abandon them. Girls fill China's 900 state-run orphanages.

    When the communists took power in 1949, they virtually eradicated the previously common practice of killing unwanted newborn girls. Officials plastered the countryside with slogans like "A girl is worth as much as a boy" and "Firmly uphold equal rights for girls and boys."

  24. Vinaigrette Girl said,

    March 8, 2010 @ 11:23 am

    @Jen, you didn't quote this: "Sex-selective abortions account for at least a third and maybe more than half of the missing girls," said Zeng Yi, a population expert at Peking University." Further, the article states that infanticide is viewed by some as a late-term abortion, which is still carried out by state-appointed medics in China.

    However, an article in the Seattle Times still isn't my top choice for a wholly reliable statement about female survival in modern China. I studied Chinese history at university and although the early Communists did a great deal – ironically – for some cadres of Chinese women ("Women hold up half the sky") these advantages tended to be predominantly urban, after the initial period when China was at last able to feed itself and basic nutritional standards were raised for both sexes. Amongst the rural poor, abortions of female embryos either by choice or imposed by the state remain a consistent feature of life.

  25. Jesse said,

    March 8, 2010 @ 12:44 pm

    Love the alt text on the picture. =]

  26. Ellie said,

    March 8, 2010 @ 1:25 pm

    Interesting, but maybe too simplistic? How is literacy defined, and does that really correlate to increased access to education (meaning, perhaps some cultures traditionally value literacy over others)? What if sex ratio was looked at in light of change in literacy over the past 50 years? Do Portugal and Swizerland have a culture pf male preference? And Jen's note on the "missing girls" is an additional confounding factor.

    So good food for thought, but it clearly raises more questions than it answers.

  27. Chandra said,

    March 8, 2010 @ 2:55 pm

    Indeed, I would be very interested to know how literacy is being defined here. I work as a literacy program coordinator in Canada, and anyone familiar with the International Adult Literacy Survey ( would question how any country, developed or otherwise, could be scoring at a near-100% literacy rate.

  28. Stephen Jones said,

    March 8, 2010 @ 3:57 pm

    This is surprising: one would have thought that a higher literacy rate (and thus better education) would correlate with a more enlightened attitude towards females. Sadly, this seems not to be the case.

    No, it's exactly what you would expect, as the Economist article does explain. The main cause of a decline of high birth rates is increased income, and female literacy.

    If there is a pressure for smaller families and this correlates with a social demand for boys then the following applies
    increased literacy and or prosperity > lower birth rate > skewed gender ratio.

    If there is no pressure for smaller families, as is common where there is widespread illiteracy, then there's not the pressure to get rid of girls.

  29. Nathan Myers said,

    March 8, 2010 @ 4:42 pm

    To me the big mystery is Sierra Leone. Never mind the trend line, how are they, er, achieving a sub-unit ratio? Is this really reporting births, or children who survive to some recordable age? I can imagine various ways to get that number if the latter, but prefer not to.

  30. Mr Punch said,

    March 8, 2010 @ 5:00 pm

    It's alarming that the countries that "stick out" on the right side of the chart (unduly high male percentages) include some very large ones — China, India, Pakistan, and to a lesser extent Vietnam, Egypt, Nigeria.

  31. Michael Turton said,

    March 8, 2010 @ 8:27 pm

    Taiwan data are consistent with this. Funny because i was just going to blog on it today and saw this post at LL. Thanks…

    Over the past five years, the imbalance between the sexes at birth in Taiwan ranged between 108 and 109 boys to 100 girls. The figure is far higher than the ratio of 105 to 106 suggested by the United Nations, according to the Ministry of the Interior.

    MOI data show the sex ratio at birth in Taiwan from 1955 to 1986 was within the U.N. scale. It began rising in 1987 and in 2004 was 110.7 on average, the highest in the ROC’s history and topping countries around the world.

    “This figure was even higher than that in China, which has a one-child policy,” said Ho Bih-jen, secretary-general of the National Alliance of Taiwan Women’s Associations, March 5. “The proportion dropped in 2008, but was still the third-highest in the world.”

    The discrepancy in the ratio between boys and girls at birth has women’s groups in Taiwan worrying about the local practice of sex selection.

    According to the MOI, in 2004, the sex ratio at birth for first child was 108.7; for second, 109.4; for third, 122.6; and for the fourth, 122.8.

    The View from Taiwan

  32. Kenny Easwaran said,

    March 8, 2010 @ 10:11 pm

    It looks to me like the chart shows that a high literacy rate is necessary for a skewed sex ratio, but not sufficient, or even necessarily a positive causal factor in some senses. As others have mentioned, although some high-educated states of India like Gujarat have extreme sex ratios, others like Kerala don't at all.

  33. Doctor Science said,

    March 9, 2010 @ 6:01 am

    The crucial test case IMHO is South Korea, where there has been a strong, abrupt change from a "son preference" pattern to a natural sex ratio.

    The change has nothing to do with literacy or even the position of women. What it reflects is *social security*. If each family is responsible for its own retirement fund, for its own social security, then parents will favor whichever sex traditionally provides the social secuity. Wealth and literacy make it easier for parents to bias the gender of their children in societies where families don't feel the government is a reliable source of social security. But that sense of security — or its lack — is the crucial element.

    Putting on my Evolutionary Biologist Hat™, social security/old age planning is a uniquely human need: no other creature plans to become old and dependent; they just die. From a biologist's POV, human parents get their old age care by giving up some grandchildren. A male-biased mammal population will always have a much lower reproductive rate than a female-biased one, so for many human cultures to prefer sons means that they will have fewer grandchildren. From a biological POV this makes no sense at all; it only makes sense if humans are having children for social security, instead reproduction.

    I think the reason male-biased human societies and female infanticide/neglect are more common than the other way around — a pattern that makes no biological sense — is because it's easier for parents to make sure that their sons devote themselves to taking care of the old folks, to the neglect of the grandchildren (if any). Daughters are too likely to become attached to their children, to the neglect of the older generation.

    One thing that surprises me about these results is China. The huge, ongoing migration of rural Chinese young people to the industrial zone has let many of them send money back to support their parents, instead of having to live with them as social security. For the first time in Chinese history, huge numbers of young women are able to get reliable, paid work to help support their parents. I had expected that this unprecedented influx of daughterly support would start to normalize the Chinese sex ratio, as parents realize that daughters can be social security , too. It doesn't seem to be happening, and I don't know why.

  34. [links] Link salad is green and crunchy | said,

    March 9, 2010 @ 9:39 am

    […] Literacy and the sex ratio — Sigh. […]

  35. Sue Sims said,

    March 9, 2010 @ 12:07 pm

    Could the Kerala exception be partly/wholly because that state is over 40% Christian and Muslim? In both faiths, attitudes to abortion range from disapproval to an outright ban, and infanticide is completely out.

  36. et said,

    March 9, 2010 @ 12:22 pm

    Interesting times ahead when all these sons want to marry…

  37. Melanie S said,

    March 9, 2010 @ 12:34 pm

    There's evidence that good nutrition increases the likelihood of a woman having male babies, so that may also be having an effect on the numbers above…

  38. Emily said,

    March 9, 2010 @ 5:06 pm

    I'm not at all surprised by the results, because if you take a look around you (in academia), the higher the level of education, the lower the overall birthrate. This holds statistically throughout the US, and probably elsewhere, as highly-educated children either choose (or put it off too late, and have choice but) to have fewer children–below the replacement rate, in fact. The overall US birthrate is _above_ the replacement rate.

    Combine the fact of highly-educated below-replacement-rate birthrates, with a culture that values a son over a daughter (to the point of having a history of infanticide), and you end up with sex ratios correlating to literacy.

    A horrid effect, but not actually surprising. :-(

  39. michael farris said,

    March 9, 2010 @ 5:39 pm

    Doctor Science: "The change has nothing to do with literacy or even the position of women. What it reflects is *social security*. If each family is responsible for its own retirement fund, for its own social security, then parents will favor whichever sex traditionally provides the social secuity."

    One point not made about China and India is that while the son is responsible for supporting elderly parents, it's the daughter-in-law that actually provides the bulk of care. One reason people want sons is because they need a son to get the daughter-in-law who will take care of them.

    It's not that female children aren't wanted in general, they just want other people to have them. It's kind of a tragedy of the commons situations where sex-selective abortion is a way to try to be a free rider and beat the system. But if everybody beats the system it collapses.

RSS feed for comments on this post