Two weeks ago, David flagged an interesting paper  by Bendavid, Avila and Miller in the Bulletin of the WHO which reminded me of a paper  I had been following by Kelly Jones, a revised version of which has just been posted. Both of these papers look at the effect of the U.S. Mexico City Policy (a.k.a. the Global Gag Rule) and are interesting not only for their results, but also because they provide a nice example of how we can do evaluations of wide-sweeping policies.
So what do they find? Before I turn to this, a description of the Mexico City Policy is in order. This policy is basically a removal of U.S. international funding for family planning for any organization which performs, lobbies for, or educates clients about abortion. (So why, if it is a US policy is it called the Mexico City Policy? Apparently because then President Reagan announced it at a conference in Mexico City in 1984). This policy tracks U.S. political shifts perfectly – it was first put in place by Reagan, then rescinded by Clinton, put back in place by Bush and, most recently, rescinded by Obama. And the last three changed the policy within 2 days of being inaugurated.
So this clearly creates discrete breaks in the funding for international NGOs that provide family planning as some of the larger ones (e.g. the International Planned Parenthood Federation) don’t change their practices when the U.S. changes presidents. And both papers exploit this. Bendavid and co. use cross country data, while Jones creates a panel using DHS data from Ghana. Both papers show that the policy not only does not reduce abortion (as it is intended to do) but actually increases it, significantly. They both posit that this is due to the fact that these cuts end up reducing the availability of a wide range of contraception (it seems that U.S. funding is quite significant in the scheme of things).
Of course, even though abortion is now going up, not every additional pregnancy due to this cut in contraception is aborted. Indeed, Jones estimates that 5 out of 6 of these pregnancies actually result in unwanted or unplanned births. And she shows that these kids are more likely to be born to poorer, less educated mothers (after all, they are less likely to be able to afford abortions in the absence of aid). The outcomes for these kids look worse – both in terms of height for age (0.7 of a standard deviation worse) and weight for age (0.6 of a standard deviation). So not only are we seeing the effects of a policy intended to cut abortion here, we’re also seeing what happens to unwanted kids.
The methodology of both papers is interesting. Bendavid and co. are using DHS data for 20 African countries, and abortions aren’t explicitly identified so they have to infer induced abortions. They then separate countries into high or low dependence on the U.S. for family planning funding and then do a difference in difference around the 2001 policy change. Jones does most of her analysis with the 2007 Ghana DHS, which compiled a complete pregnancy history of each woman, including a direct measure of those which ended in abortion. Using this, Jones creates a woman-by-month panel, which allows her to look at a number of periods where the policy was in place or not. She can also use the other variables in the DHS to look for heterogeneous effects. Indeed, this is where her results get interesting – on average she doesn’t find a significant impact of the policy on abortion. It’s only when she breaks the sample into rural and urban areas that she finds the impact – for the rural areas there is a 50% increase in the use of abortion and, within the rural areas, this is being driven by the top 3 quintiles (most of whom are still poor).
So overall these papers provide nice examples of how evaluations can be done of larger policy changes. And the results give us plenty of food for thought.
P.S. Markus wrote this while sick and ailing at home today, so I helped him post what he wrote (without any edits) so please excuse any typos, omissions, or errors. Berk.