We propose a method that enables one to test efficient risk sharing even when households have different risk preferences. The method is composed of three tests. The first one determines whether in the data households have homogeneous risk preferences. The second and third tests evaluate efficient risk sharing when the hypothesis of homogeneous risk preferences is rejected. We use this method to test efficient risk sharing in rural India. Using the first test, we strongly reject the hypothesis of identical risk preferences. Using the second and third tests, we reject efficiency at the village but not at the caste level. (JEL D12, D86, G22, O12, O18, R23, Z13)
展开▼