Agriculture Reference
In-Depth Information
Table 6.1
Summary of previous empirical contracting studies that consider risk sharing and transaction costs
Study
Topic
Risk sharing
Transaction cost
Cheung (1969)
Farmland
Limited
Limited
Rao (1971)
Farmland
No
Yes
Allen and Lueck (1992a)
Farmland
No
Yes
Umbeck (1977)
Gold mining
No
Limited
Hallagan (1978)
Gold mining
Ambiguous
Ambiguous
Martin (1988)
Franchising
No
Yes
Lafontaine (1992)
Franchising
No
Yes
Leffler and Rucker (1991)
Timber sales
Ambiguous
Yes
Mulherin (1986)
Natural gas
No
Yes
Lyon and Hackett (1993)
Natural gas
No
Yes
Most empirical land contracting studies have examined the effects of contract choice on
input use rather than estimating the factors that determine contract choice (Hayami and
Otsuka 1993).
5
In this chapter we test a standard risk-sharing model of contracts against our data on
individual farmland contracts. The chapter begins with a principal-agent model of share
contracting that generates predictions that are testable with our data. Unlike our earlier
transaction cost models that allowed for more than one unobservable margin of behavior,
the model we present only allows farmer effort to be unobservable. Next, we test the risk-
sharing implications of this model against our contract data, by linking the contract data to
reasonable measures of exogenous risk by using data on crop yield variability.
Modern North American agriculture is a particularly good test bed for the risk-sharing
theory compared to undeveloped agricultural economies. First, the property rights to land are
well defined and enforced, unencumbered by political land reforms. Second, the behavior
of farmers and landowners is not severely constrained by caste systems or other potentially
dominating cultural forces.
6
Third, asset values are relatively large, making the gains from
optimal contract design relatively large as well. Finally, there are reliable data on regional
crop yield variability that can be used to approximate exogenous risk. In general, we find
no support for predictions motivated by risk sharing. We finish the chapter by contrasting
the risk-sharing predictions with those derived from our earlier model where both parties
were risk neutral.
6.2
A Model with Risk-Averse Agents
With sharecropping, the typical risk-sharing model makes several routine assumptions: the
principal is a risk-neutral landowner and the agent is a risk-averse farmer;
7
the effort of the
