So I come back from vacation to find out that I was part of a randomized experiment in my absence. No, this had nothing to do with the wonders of airline travel in Europe (which don’t add that frisson of excitement through random cancellations like their American brethren), but rather two of our co-bloggers trying to figure out if the blog actually makes people recognize me and Jed more (here are links to parts one  and two , and three  is where they use us for their experiment). At any rate, well rested and better analyzed, I wanted to follow up this week on an earlier discussion on evaluating land projects.
In a recent working paper , Daniel Ali, Klaus Deininger and I evaluate the impacts of a pilot land tenure regularization initiative in Rwanda. The background to this work, in my experience, speaks to the difficulty of doing rigorous impact evaluations of the formalization of property rights. As many developing countries don’t have a systematic registration of rights and rely on citizens to come and register when they want, endogenity is a big problem. The immediate options: either try to convince a government to randomize this (very hard) or use some careful institutional knowledge to identify things – as Sebastian Galiani and Ernesto Schargrodsky did in Argentina (you can find an ungated version of one of their papers here ).
When we first talked about this with the Rwandan government, they were contemplating a national land tenure regularization program, but had done a pilot in a couple of diverse but representative areas a couple of years back. Now that some time had passed, they were interested to learn about whether or not there had been any significant investments and what impacts there had been on the land market as a result of increasing the formality of property rights. But the methodology wasn’t immediately obvious.
However, a very interesting approach taken by Jeremy Magruder, in his work  looking at the effects of centralized bargaining on employment in South Africa provided an idea. Jeremy uses the boundaries of administrative units to implement spatial regression discontinuities to look at impacts. This alone isn’t new (a significant number of papers use border discontinuities) but Jeremy adds spatial fixed effects estimation which, while raising the data requirement (you need GIS information), allows basically for a kind of spatially continuous control group and for the control of local level unobservables as everyone has their own geographic neighborhood.
Given that the Rwandan government had chosen areas that were not contiguous (indeed, they were all over the map), we were able to apply this approach in Rwanda and collect a set of post-intervention data. What we find is rather significant effects on investment (primarily in the form of the construction and maintenance of soil conservation structures). Moreover, there is a strong gender dimension to these results – female headed households invest significantly more than their male-headed counterparts. Women also seem to be getting stronger property rights out of the registration and formalization of rights – married women report higher ownership rights after registration. This isn’t a surprise – the laws that provide the context for this registration specify that legally married wives should be co-owners and the results indicate that the registration process is making these rights effective. However, some folks (a distinct minority) are not formally married (which is a fairly easy process in Rwanda). And these women seem to see a decline in their property rights. The final result of note is that there is actually a decline in land market activity within the areas undergoing land title registration. Given that this was explicitly (and publicly) a pilot, there a lot of good and interesting explanations for this – but not one that is definitive. What we can say for sure is that there is no evidence of a (short-term) significant sell-off of land. So this provides some early evidence of better property rights having significant payoffs. It will be interesting to see what happens as this program goes nationwide.