Introduction
Moral hazard is often offered as an explanation for both low participation rates and high loss-
ratios (the ratio of indemnities to premiums) in agricultural crop insurance (Ahsan, Ali, and
Kurian; Nelson and Loehman; Chambers; Smith and Goodwin). However, there are many
definitions of moral hazard in the theoretical literature, and still more employed in empirical
studies of the demand for insurance. Arrow, for one, suggests a simple and compelling definition
of moral hazard as “hidden action” on the part of an insured agent. With the multiplicity of
definitions of moral hazard comes as many alternative methods of measuring or detecting its
presence in insurance markets.
Recent attempts include Quiggin, Karagiannis, and Stanton (QKS) who estimate a Cobb-
Douglas production function and input share system with a binary variable indicating insurance
participation. With this framework, they conclude that moral hazard exists if the coefficient on
the insurance dummy is negative in both output and chemical equations. Alternatively, Just and
Calvin (JC) use a mean-variance expected utility model to motivate two simple measures of moral
hazard and adverse selection. Moral hazard, they claim, must exist if realized yields fall below the
producer's subjective expectations. Furthermore, they interpret a deviation between Federal Crop
Insurance Corporation (FCIC) and expected yields as evidence of adverse selection. Others claim
that the use (or nonuse) of risk reducing inputs indicates the presence of moral hazard (Horowitz
and Lichtenberg). However, such a clear relationship between observable output or input levels
and the decision to insure must surely be evidence of violating the “best practice” condition of any
insurance contract. Therefore, evidence of moral hazard must be found in otherwise unobservable