Consumption Insurance with Heterogeneous Preferences. Can Sharecropping Help Complete Markets?

Empirical tests of the complete markets hypothesis generally reject full insurance and stimulated the study of optimal risk sharing. However, most of the empirical tests of the complete markets hypothesis make strong assumptions concerning the specification of preferences. In particular, the usual rejection of full insurance may be affected by the incapacity to account for heterogeneity of risk aversion of households. Underlining the methodological problems on the identification of preferences, the direction of the tests and their power, we study the risk sharing and consumption insurance achieved by rural households from three provinces of Pakistan. We show how to deal with these methodological problems in studying consumption smoothing and implement some non directional and directional tests of market completeness. We actually find that markets are incomplete in Pakistan even at the village level and look for a possible explanation of how households try to complete markets. The risk sharing property of sharecropping is actually often called upon, but its effect on consumption smoothing has never been tested directly. We therefore present a reduced form evidence that, in Pakistan, the sharecropping institution allows to complete markets. Households able to use this contractual choice, which permits to share production risk, are better insured against idiosyncratic shocks. It seems that sharecropping provides a contingent claim that other accessible markets do not allow to replicate. This empirical fact shows that agricultural contracts play an important role sharing production risk but also through informal insurance mechanisms when markets are incomplete.