|Abstract or Summary
- Agricultural revenues, the product of stochastic prices and yields, lead to markets which are
incomplete, thereby entreating and complicating economic inquiry. The following three essays
explore the incomplete nature of agricultural markets and consider the implications of
incompleteness for a range of policy questions and economic tools.
The first essay, "Cooperative Pricing Policy Under Stress: The Case of Tn Valley
Growers," explores the incomplete contract markets that arise from unobservable yield processes.
I formalize a common class of forward contracts, decompose them into a convex combination of
yield derivatives, then derive the arbitrage-free forward price bounds. These bounds are used to
show how the Board of Directors of a large agricultural cooperative, Tri Valley Growers,
overstated earnings in order to liquidate financial equity.
The second essay, "Adapting Cooperative Structure for the New Global Environment,"
follows up on the first by showing that the liquidating strategy Tn Valley's Board pursued was
rational in terms of maximizing expected net present value of future cash flows, I derive a
condition under which optimal equity retention is strictly greater for investor-owned than for
cooperatively owned firms. Finally, I use ruin probabilities associated with the standard firstcrossing-
time problem, together with numerical integration methods, to verify that this condition
held under the market conditions in which Tn Valley and its investor-owned rivals operated.
The third essay, "DEA and The Law of One Price," explores the effect of variable prices
on technical efficiency estimation. Data commonly are furnished in value, rather than factor
terms. This raises the question of how value-based DEA models coincide with factor-based models. A sufficient condition for the two models to coincide is that all firms face the same set of
prices. In practice, however, prices commonly vary across firms. I show that, unless an
unreasonable restriction holds, the two models do not coincide. I decompose the resulting
estimation error into its technology and firm-related components. Using Farrell's original 1957
data set to illustrate, the resulting estimation error is found to be both systematic and one-sided.