to different types of contracts. Tenants with highest skill levels choose fixed-rent contracts, those with
medium-skills self-select into sharecropping and the least skilled ones remain as agricultural laborers.
However, the screening model is of limited use because the adverse selection argument falls apart in
village economies where social networks are especially strong and information on tenant attributes is easily
available.2 This model is also theoretically flawed because it fails to address the landlord’s optimization
problem.3
More recent studies of land leasing have emphasized that moral hazard is a greater problem than
adverse selection in village economies. The reasons for the severe moral hazard (or agency) problems are
twofold: 1) many farm tasks are characterized by long time lags between inputs and the outcome and 2) a
variety of random shocks (weather etc.) can make output uncertain and the contribution of workers to
output difficult to measure.4 An obvious solution to the moral hazard problem, especially in a modern
industrial context, is the direct supervision of workers. However, direct supervision is a directly
unproductive activity and can incur heavy costs especially for tasks with the large time lags and greater
exposure to production shocks.5 Some examples of such tasks are farm management and decision-making,
application of chemicals and fertilizer, irrigation control, seed preparation and care and the supervision of
2 Bardhan [1984], Bell [1988] and Lanjouw [1999] find that information on the attributes of farmers is widely available
in village communities.
3 See Allen [1982] for a discussion of this problem. Allen [1985] attempts to salvage the screening model by looking
at the optimization behavior of both parties under the assumption that contracts are designed to prevent default in
rent payments. However, he succeeds only in showing that both parties are indifferent between share and fixed rents
for certain skill levels. In addition, the advanced payment of a fixed-rent has been shown to be a widely practiced
first-best solution for the default problem [Otsuka and Hayami 1993].
4 See Lanjouw [1999] for empirical evidence on the existence of moral hazard problems.
5 For an empirical analysis of farm supervision activity, see Evenson, Kimhi and DeSilva [2000]