could also have derived much the same results by claiming more traditional
assumptions from the mechanism design literature, cf. e.g. Laffont and
Tirole(1993). Specifically, we could have derived similar sufficiency and ne-
cessity results by assuming that the cost of farmer i*s production is Cγ‰,
where θi is the privately known productivity parameter, θi ∈ [⅛,0⅛7 , that
costs are increasing and convex, ∂Ci∣∂qi > O and ∂2Cx∕∂2ql > O, that costs
and marginal costs are increasing in types ∂Ci∕∂θ{ > O and ∂2Cx∕∂qi∂θi > O
to ensure the so-called single crossing property, and more technically that
∂3C*/ (∂2Qi∂qi} ≥ O and (PCi/(∂2qi∂θi) ≥ O to avoid mixed strategy and
pooling equilibria, cf. Jensen(1998). We suggest that the assumptions we
have inwoked are as realistic and certainly simpler to understand.
8 Conclusions
We have shown that with asymmetric information about farm level produc-
tion costs, the only way to ensure socially optimal production levels may be
to organize processing as a cooperative. This gives an information economic
rationale for cooperatives. Specifically, we have shown that a cooperative is
necessary when fanners’ marginal production costs are independent, the net
average revenue product from sales is constant, and the income distribution
does not matter.
We have shown also that the relative advantage of cooperatives (com-
pared to investor owned processors) is largest, when the cost uncertainties
are large and when the profitability is limited, i.e. when the net marginal
product is small compared to the primary production cost. In these cases
the investor owned processor tends to ration away more social value to gain
private value. Since the agricultural sector may have these properties, we
suggest that our results may in part explain the apparent success of cooper-
atives among fanners.
In reality, the choice of organizational structure is not only determined by
incentive costs. The resulting market behavior should be taken into account
as well. Most likely, we would get different results if the market conditions
are not - as assumed he - characterized by perfect competition. We leave the
analysis of some such cases to future research.
Acknowledgments: This paper was written while Peter Bogetoft vis-
ited Department of Agricultural and Resource Economics, University of Cal-
17