Determinants of Sex Selective Abortions

I presented my paper "The Determinants of Sex Selective Abortions" at the NEUDC conference at Tufts University in Boston 7 November and at University of Copenhagen and University of Aarhus in Denmark in the beginning of December. The paper is still work in progress but a working paper version should be available soon. Until then the presentation is here and the abstract below.

Over the last decades many countries have seen a significant shift in the sex ratio at birth as techniques for pre-natal sex determination have become more widely available. Despite this there has been relatively little research on what determines the use of sex selective abortions at the individual level. A major impediment to analysing determinants of sex selective abortions is the absence of reliable direct information on individuals' use of pre-natal sex determination and abortions, which means that information have to be inferred from other observed outcomes. Previous studies have used the sex of children born and estimated which factors affect the likelihood of having a boy at a given parity. I argue that this method fails to address the close relationship between fertility, birth spacing and the use of sex selective abortions, and leads to biased estimates and low power in the estimations. To examine the determinants of the use of sex selective abortions I therefore estimate a competing risk hazard model, which directly incorporates both fertility and abortion decisions and the potential censoring of birth spacing. This is done using data from the three rounds of the Indian National Family and Health Survey. The results show that women who are likely to want fewer children are significantly more likely to be using sex selective abortions and that these numbers are substantially larger than previously found. Furthermore, contrary to resent research this paper finds no evidence of declining use of sex selective abortions; in fact, sex selective abortions appear to have increased for parity two, once one controls for censoring.