CSU 101/2011: HEPATITIS B AND THE CASE OF THE MISSING WOMEN

Thursday, 24th of March 2011

Emily Oster proposes hep B infection as one explanation for the uneven sex ratios among newborns in several countries, where boys outnumber girls. If her hypothesis is true, then hep B vaccination will, over time, partially correct sex ratios among offspring of vaccinated mothers.

 Summary and introduction are below. The full text is at http://home.uchicago.edu/~eoster/hepb.pdf

 Good reading.

 BD

Hepatitis B and the Case of the Missing Women

Emily Oster

Harvard University

In many Asian countries the ratio of male to female population is higher than in the West: as high as 1.07 in China and India, and even higher in Pakistan. A number of authors (most notably Amartya Sen) have suggested that this imbalance reflects excess female mortality and have argued that as many as 100 million women are “missing.”

This paper proposes an explanation for some of the observed overrepresentation of men: the hepatitis B virus. I present new evidence, consistent with an existing scientific literature, that carriers of the hepatitis B virus have offspring sex ratios around 1.50 boys for each girl. This evidence includes both cross-country analyses and a natural experiment based on recent vaccination campaigns. Hepatitis B is common in many Asian countries, especially China, where some 10–15 percent of the population is infected. Using data on prevalence of  the virus by country and estimates of the effect of hepatitis on the sex ratio, I argue that hepatitis B can account for about 45 percent of the “missing women”: around 75 percent in China, between 20 and 50 percent in Egypt and western Asia, and under 20 percent in India , Bangladesh, Pakistan, and Nepal.

I. Introduction

The ratio of men to women in the Western world is close to unity. Sex ratios at birth are a bit higher (around 1.05 boys for each girl), but higher male mortality results in sex ratios around 1.00 by young adulthood and even lower among older adults. Although this pattern is universal in the West, population sex ratios in a number of Asian countries seem to be much higher: 1.07 in China, 1.08 in India, and as high as 1.11 in Pakistan. This imbalance has caused a number of authors to argue that there is substantial excess female mortality in Asia (Sen 1990; Coale 1991; Klasen 1994). In general, they argue that neglect of female children and poor conditions for women contribute to the higher sex ratio. These authors have estimated that there are between 60 and 100 million “missing women.”

This paper connects the puzzle of the missing women to a separate literature investigating a connection between the hepatitis B virus and offspring sex ratios. There is evidence that women who are carriers of hepatitis B give birth to a higher ratio of boys to girls than noncarriers.

Since many of the countries with missing women also have a relatively high prevalence of hepatitis B carriers, the naturally occurring higher sex ratio at birth could produce a higher population sex ratio even in the absence of excess female mortality. Indeed, I argue in this paper that after one adjusts for differences in the sex ratio at birth caused by hepatitis B, the number of missing women (based on population estimates from 1980–90) drops to 32 million from the 60 million calculated by Coale (1991) and the 107 million suggested by Sen (1992). There

is significant variation among countries in the share of the gender bias that is accounted for: I find that hepatitis B can explain 75 percent of the missing women in China but less than 20 percent in India, Pakistan, and Nepal. Other countries lie between these polar cases.

The issue of the missing women has become a flash point for development economists, demographers, and policy makers. For many, presumed excess female mortality is the most severe example of discrimination against women in general in these regions. The missing women

are emblematic of what seems to be a much larger problem, and understanding why these populations have such a widely skewed sex ratio seems vital to moving forward on issues of gender equality. The results here suggest that sex ratio bias may be less attributable to differences in tastes and preferences than many have suggested. However, there are

at least two important caveats to the results here.

The first is that technological developments in the last 15 years—in particular, ultrasound—have made selection of a child’s gender before birth possible in a number of the countries in question here. And, as has been pointed out by many authors, the sex ratios in India and China in particular have correspondingly increased (see, e.g., Hull 1990; Junhong 2001; Dre`ze and Sen 2003). As parents’ ability to select the sex of their child becomes less costly, any naturally occurring bias in the sex ratio becomes less important. In the extreme, if selection were costless,

these biases would have no effect. This suggests that, while there may well be a role for hepatitis B in understanding the existing gender imbalance in the countries with missing women, this role will decline in future generations as parental control improves. Indeed, this has already happened to some extent: the increases in the sex ratio at birth in China and India in the last 15 years cannot be attributed to hepatitis B. This is particularly true since vaccination for hepatitis B has become virtually universal, but it would be true regardless.

The second, and perhaps more important, caveat is that what is estimated in this paper is the total effect of hepatitis, not the partial effect. The estimates here assume that, in the absence of hepatitis B, the number of excess female deaths would be the same: there would be no

compensating behavioral response. This is unrealistic. It is clear that there is significant gender discrimination in these countries: the changes described in the above paragraph clearly point to a latent demand for male children, and further, even the estimates in this paper suggest that

about half of the women are missing as a result of discrimination. If the naturally occurring sex ratio at birth had been lower historically, we would expect some compensating behavioral response.

The correct counterfactual is therefore not the West, but a world in which there is no hepatitis but a similar level of gender bias. Creating this counterfactual is difficult. In Section V.B, I attempt to use crosscountry variation in hepatitis rates to explore this, but even this requires some significant assumptions. It seems likely that in the absence of hepatitis, sex ratios would be somewhat lower, but it is perhaps not clear exactly how much. For this reason, it is important to keep in mind exactly what the estimates represent. Although I shall refer to the share of the gender imbalance “explained by” hepatitis, this should be taken to refer to the total effect, not the partial one.

Section II of this paper discusses how the number of missing women is calculated. Section III then presents a discussion of the origin of the sex ratio bias. The hepatitis B hypothesis would suggest that the sex ratio bias arises at birth; to the extent that the bias arises later, there cannot be a role for hepatitis B. I present evidence that the sex ratio at birth is substantially higher in most of the missing women countries than it is in the West. In addition, sex ratio patterns by age vary across these countries. China, for example, exhibits very high sex ratios at birth and declining sex ratios over childhood, whereas India exhibits moderately higher sex ratios at birth but also sees increasing sex ratios during childhood. The overall high sex ratio at birth suggests a role for hepatitis B in the missing women puzzle, especially in China.1

1 Recognizing the recent increases in the sex ratio at birth in China, I present a number

of pieces of historical data and argue that the sex ratio was high in the past, even if not

as high as at present.

Section IV presents a variety of evidence pointing to an effect of hepatitis B on sex ratios at birth. Section IV.B discusses individual-level evidence drawn from the existing literature. In 1972, researchers studying hepatitis B in Greece noted that the sex ratio among 131 children

born to women with hepatitis B was 1.77 boys for each girl, versus only 1.13 among 542 children born to women without hepatitis. Five additional studies in different areas also suggested this connection. The evidence from the scientific literature is suggestive, but not necessarily conclusive. As will be discussed in more detail later, the sample sizes are quite small, and there are some issues with data quality. Further, the studies (for the most part) were not designed to test the hepatitis–sex ratio connection, and as a result, the methodologies are not perfectly suited to testing this theory. I therefore devote a significant portion of the paper to new analyses designed to test the hepatitis B–sex ratio connection.

First, Section IV.C takes advantage of the availability of a vaccine since the late 1970s as a natural experiment. I consider first offspring sex ratios for Alaskan Natives (historically high hepatitis) and non-Natives in Alaska (low hepatitis). Data on births to these groups are compared

before and after the vaccination campaign. I find that sex ratios among white Alaskans are unaffected, but sex ratios among Alaskan Natives drop dramatically over the period being considered. In addition, I consider the sex ratio among women in Taiwan who were affected by a

vaccination program that began in 1984; there is suggestive evidence of a drop in the sex ratio at birth.

Section IV.D presents aggregate evidence on the cross-country relationship between hepatitis B and sex ratio. This subsection takes advantage of both a categorical distribution of hepatitis B and continuous estimates of the prevalence of hepatitis B across countries. Both types

of evidence suggest that hepatitis B and the sex ratio at birth are strongly related, and this finding is robust to restricting to countries within a particular region, countries within the Organization for Economic Cooperation and Development (OECD), and to controlling for income levels and regional fixed effects. I find that about 40 percent of the cross-country variation in the sex ratio at birth can be explained by hepatitis B. In addition, the magnitude of the effect implied by the

cross-country data is very close to the size of the effect in the individual level data. Taken together with the original scientific data, this new evidence is more strongly supportive of the hepatitis–sex ratio connection.

In Section V, I combine data on the prevalence of hepatitis B and estimates of the effect of hepatitis on the sex ratio at birth to estimate and adjust the number of missing women. The baseline results suggest that around 45 percent of the missing women estimated by Coale (1991)

hepatitis b are explained by hepatitis. The hepatitis B–adjusted number of missing women is 32 million: 8 million in China, 19 million in India, and smaller numbers elsewhere. In Section V.B, I attempt to use within-country variation in China to create a more realistic counterfactual for that country.

Although the number of missing women under this counterfactual is smaller, the results still suggest that at least 62 percent of the missing women are explained (down from 75 percent in the baseline specification).

First and foremost, the findings in this paper have implications for the puzzle of the missing women. However, the work here is also closely related to work on the economics of the family. In particular, understanding the contribution of biology to gender imbalance necessarily

requires understanding the contribution of behavior, and vice versa.

Drawing on Becker (1981), a number of papers have developed theories that provide a rational motivation for sex-selective mistreatment, infanticide, and abortion, based on future earnings capacity of children, and have done empirical work to test those theories (see Rosenzweig and

Schultz 1982; Klasen 1996; Agnihotri 1999; Edlund 1999; Kojima 2005).

For the most part, the empirical work in these papers relies on regressing sex ratios by region on economic factors of interest (maternal education, earning power, etc.). Others (Sen 1998; CIA World Factbook) have suggested that the sex ratio may be used as a direct measure of gender

discrimination. Clearly, the validity of the previous regressions and the value of sex ratios as a measure of discrimination may be affected by the contribution of biology to the gender imbalance.

Similarly, good estimates of the effect of behavior on sex ratios may speak to the potential importance of biology. Qian (2005) provides substantially better-identified estimates of the effect of economic incentives on sex ratios of surviving children in China. Using variation across

provinces and time in the value of female labor, she shows that increases in the value of female labor increase the share of girls who survive. Qian’s results strongly support the existence of behavioral responses to economic incentives over and above any biological differences. These

results are important to keep in mind in the context of this paper, in particular when thinking about appropriate counterfactual sex ratios.

However, the results in the two papers are not necessarily at odds; among other things, the hepatitis B hypothesis is about the average sex ratio, and the estimates in Qian’s article represent marginal changes.

A bit further afield, the results in this paper may also be related to economists’ study of marriage and labor market outcomes. In his work on the economics of the family, Becker (1981) highlights the importance of population sex ratios in the formation of polygamous societies, as

well as in the determination of dowry and bride-price levels. Building on this, others have argued that high sex ratios (many men in the population) lead to higher rates of marriage among women and lower outside–labor market participation (Angrist 2002; Chiappori, Fortin, and Lacroix 2002). Movements in the population sex ratio during wars have also been connected to female labor market participation and wages (Goldin 1991; Acemoglu, Autor, and Lyle 2004). The connection between hepatitis and the sex ratio therefore has the potential to provide a cleaner test for the relationship between population sex composition and market outcomes, although obviously this potential may be limited by the existence of other factors influencing the sex ratio.