Syndicate content

Do Things Have to Get Worse for Women Before They Get Better?

Berk Ozler's picture

While the U.S Presidential Debate on Tuesday night brought to the fore issues of gender equity in the U.S. (Binders Full of Women has more than 5,000 members on FB and @RomneyBinders has more than 33,000 followers on Twitter), in particular with respect to their place in the labor market and their reproductive rights (see Washington Post here and NYT here), women in developing countries continue to face restrictions with more basic rights, such as inheritance, land ownership, and the very basic right of safety from domestic violence. A recent paper by Anderson and Genicot examines the relationship between the introduction of property rights and suicides in India and provides some food for thought – theoretically and empirically…

Exploiting the fact that women had no land rights prior to 1956 under customary Hindu law and that things changed at different times in different states, they are able to examine the effect of giving women more property rights on the suicide rates of males and females. They find that increased property rights for women lead to a rise in suicide rates for both males and females. The faster increase in the suicide rate among males makes further reduces the female-male suicide gap (or ratio). Obviously, not great news...

The authors argue that the increase in suicides is due to increased conflict in the household over resource allocation. They provide some evidence by using suicide rates due to household conflict rather than all suicides. They also have some data on the prevalence of wife beating from DHS data. They then conclude the paper with a theoretical model that predicts separations (divorce) as well as suicides.

The paper has some empirical limitations. First, the identification is not experimental. However, given that suicide is a rare event (about 10 in 100,000) and it takes time for these types of law changes to show effects, any experiment would have to conduct a very long-term follow-up – much longer than the time horizon of field experiments that we’re used to conducting. The authors are hence utilizing data that are well-suited for trying to tackle the question they’re interested in and obtain findings that are robust to changes in definitions and specifications. (There is actually a randomized law change in India’s recent history, but it provided exogenous variation in women’s political voice and representation rather than their inheritance and tenancy rights. See a recent paper by Beaman et al. on the effect of this change on aspirations and educational attainment of girls. Furthermore, another recentpaper by Iyer et al.,again using a similar identification strategy to that employed by Anderson and Genicot, examines the effect of increased political voice (rather than property rights) on suicides and finds no effects on suicide rates for either men or women. Obviously, the channels of impact from intervention to suicide are likely to be different in each case.)

More worrisome, as the authors admit themselves, is the reporting issue. If the reforms are correlated with an increased likelihood of reporting suicides (or with the male-female differential), then the identification strategy fails. The Iyer et al. paper cited above finds that increases in political voice for women led to increases in reporting of crimes against women. I have no idea how the suicide of someone is reported to the authorities (and by whom) in India (or, for that matter, anywhere else) so I have no guesses about the direction of any bias. But, measurement issues regarding crimes, suicides, rape, domestic violence, etc. plague the field especially when the data come from police records.

The cross-sectional association between pro-female property rights laws and women’s acceptance of being beaten by husbands is confusing. Perhaps, as women are beaten more, they rationalize the violence by adopting these views – I don’t know. But, I did not find these findings to be helpful in supporting the main hypothesis in the paper. Finally, it’s probably better to cluster the standard errors at the state level: this will lead to higher standard errors (as shown in one column in one table) but it seems that the results stay qualitatively the same.

To me, the paper’s contribution is as much (if not more) in its model of intra-household bargaining as it is empirical. By adding asymmetric information and costly conflict within a marriage, the authors are able to improve on (or tweak) the earlier models (e.g. by Lundberg and Pollak 1993; Ligon et al. 2004). When a change in bargaining power (through a change in relative resources controlled by each spouse) causes conflict over sharing rules within a marriage, the threat of separation causes discord and is costly to either side. But, as separation is almost never swiftly achieved, one party may choose the “ultimate exit and commit suicide.”

The real question is how long it takes for the change to reach a new equilibrium where men accept the increased power of women and are content to live as equals to their wives. The authors acknowledge that the empirical accounts from developed countries suggest that such undesirable effects in the short and medium run disappear in the longer run. The authors’ empirical strategy does not address this question and looks at the “average” effect of changes in property rights over time. It might be a good idea to examine the heterogeneity of the effect as the duration under more equitable laws increases (treatment interacted with year dummies). Is it the case that there is an initial spike in suicides after which this effects starts to dissipate, or is the effect more or less constant over a long period of time?

The authors are clear in the paper that gender equity is a worthy pill to swallow for societies, even if there are some serious side effects for some people, such as increased suicide rates in this case. Perhaps, direct policy implications take a back seat here. Rather, the paper helps us think about other policies where women receive preferential treatment: cash transfers being given to women in the household, focus on adolescent girls rather than boys, etc. It’s wise to keep in mind that while those policies might be performingwell on average, they might have some unintended consequences…