ττr* , O/ , *
( W = α β Xi ε ) exceeds a certain limit value - zero in this case. In order to pool
insurance valuations across commodities, the dependent variable (W) is expressed as a percentage
of the total cost of production per acre. Among the explanatory variables, yield variability is
measured as the five-year coefficient of variation, and the expected price is simply taken to be a
naive forecast - growers are assumed to expect last year’s price to prevail next period. This
model is estimated using the maximum likelihood procedure in LIMDEP 7.0.
Results and Discussion
Table 1 provides parameter estimates that are defined as the marginal effect of a change in each
element of X on the expected, conditional willingness to pay. Estimates of the Cobb-Douglas
production frontier used to derive the farm-specific measure of efficiency are available from the
authors, so this section focuses on the insurance demand results. [table 1 in here]
Whether reflecting inherent managerial ability (adverse selection) or input choice once
the insurance decision is made (moral hazard), these results support the central hypothesis of the
paper that more efficient growers are willing to pay less for insurance. Table 1 also shows that
growers who contract their production are willing to pay more for each level of coverage,
suggesting that these growers are particularly risk averse. Similarly, older growers’ greater
willingness to pay may indicate that they tend to be more risk averse than younger growers.
Although growers who are less concerned with yield risk appear to be willing to pay more for
insurance, suggesting that adverse selection may indeed present a problem for fruit and vegetable
insurance, the effect of historical yield variability on the willingness to pay for insurance is
statistically insignificant. Likewise, growers who expect a higher price for their produce are not
willing to pay more for insurance than others. Although two methods of providing self-insurance,
10