duce the degree of hidden information. Over the years, program administrators have chosen
the latter course. The USDA uses soil maps and local land rents to establish county-level
bid caps rather than the previous regional ones.
I calculate potential savings to taxpayers from these two policy alternatives. First,
I evaluate the cost reduction from changing the bidding process so that it functions as a
second-best mechanism rather than a Pigouvian subsidy. Second, I obtain an indication of
the usefulness of efforts undertaken by the government to overcome information asymmetries
by comparing the cost of the first-best (full information) mechanism to the second-best
mechanism.
In Section II, I describe and implement the econometric strategy for developing
consistent beliefs regarding the production technology and distribution of agent types when
agent heterogeneity is unobservable. In Section III, I characterize the theoretical least-cost
land set aside program. In Section IV, I use this information to simulate policy alternatives
for reducing the cost of the CRP.3 Section V contains concluding comments.
II. Empirical Model
A. Specification
Producers are characterized by two exogenously fixed factors, one observable and
one not. The observed fixed factor is acres of land, denoted a ∈ <++ . The unobserved
factor is a type productivity index θ ∈ Θ ≡ (0, 1]. Let x ∈ <+N denote the variable-
input vector and q ∈ <+ denote aggregate output. The variable-input requirement set
is V (q, a, θ) ≡ {x : x can produce q given a, θ}. In addition to V (q, a, θ) being a closed,
convex set, it is assumed to satisfy the following properties: