Is Everything We Know about Asia's Missing Women Wrong? (And Do More Women Go Missing in Africa?)

Most readers of this blog are probably aware that a substantial number of women are said to be "missing" in India and China. That is, if you look carefully at the demographics, you find there are roughly 100 million to 200 million fewer females than you'd expect in those countries. Most readers are probably also aware that there's not a similar gap in Africa.

Amartya Sen first reported on Asia's "missing women" in a 1990 article in the New York Review of Books. Because there is no ready biological explanation for the gap, and because--as the example of Africa appears to demonstrate--there is no simple correlation between the male-female ratio and the extent of absolute poverty or sexism in a society, the issue became a sort of high-stakes version of Clue. What--or more ominously, who--was killing the girls? Why, and when, and how? An entire academic cottage industry grew up around those questions.

Sen himself speculated that girls were comparatively less likely to find "gainful employment" in Asian societies than in African ones, and that this tended to discourage parents from providing as much care for their girls as they did their boys. A Chicago economist suggested that half the difference might be due to the rate of Hepatitis B infection, which appeared to reduce the incidence of female conception. (And later, in a rare bit of intellectual honesty, repudiated that finding.) My own pet theory, which I never hesitated to bring up despite having no relevant expertise, had to do with the relative prevalence of bride-prices and dowries.

Whatever the explanation, I never much doubted that the findings were accurate. I never doubted, that is, that some appreciable number of Asian women were missing. It wasn't hard to imagine that a poor peasant family in India would be more willing to spend their last rupee on medicine for the (male) child they expected would care for them in their old age than for the (female) child they would soon marry off to another family. Add in China's one-child policy and the ready availability of sex testing and selective abortion, and it all seemed to make terrible sense. The proliferation of stories about missing women in the media--The Economist, the NY Times, and the Today Show (if memory serves, though I can't find the link)--seemed to leave little doubt about the facts.

So I was surprised today to read in Tyler Cowen's Marginal Revolution that much of what I thought I knew about missing women may NOT be true. He links to a 2010 paper in the Review of Economic Studies by Siwan Anderson and Debraj Ray that finds that many of the "missing women" are adults, not children; that Africa appears to have an even higher incidence of missing women than Asia; and that America had a similar gender ratio to Asia's in the year 1900.

What does all this mean? Well, as the authors themselves emphasize, if the findings hold then the number of missing women--defined as women who have died as a result of discrimination--needs to be seriously revised downwards.

Like Cowen, I don't know whether these findings are true or not, and unlike him, I'm not even remotely equipped to referee the debate. I'm linking mostly because these findings don't seem to have permeated into the public discussion, even as a caveat. (For example, neither of the articles in the Times or the Economist mentioned that the number of missing women is in dispute, although both date from after the study's publication. My brief survey of the web suggests that this is typical. As far as I can tell, almost no one is disputing the notion that large numbers of Asian women are missing.)

It's possible, of course, that Anderson and Ray are wrong: In fact, it's more than likely, given the preponderance of studies to the contrary. Still, it highlights the importance of careful and parsimonious data collection to public policy, which, as readers know, is something of a fetish idee fixe obsession of mine.