Agriculture Reference
In-Depth Information
First, consider tests using FUTURES MARKET. This dummy variable equals one when
a crop is traded in a futures market. Because futures markets are an alternative method of
sharing risk, when they are present the farmer is, ceteris paribus, less likely to choose a share
contract.
32
In all six of the equations in table 6.7 we included FUTURES MARKET, which
should be negatively correlated with share contracts. The predicted coefficient estimate for
FUTURES MARKET is, once again, negative for the logit equations but is positive for the
tobit equations. The top panel of the table shows that none of the four logit estimates support
this version of risk sharing—all four estimates are positive and significantly different from
zero. The bottom panel in table 6.7 is also unfavorable to this prediction, with only one of
the four estimates to be positive, but statistically insignificant.
33
Next, consider the variable INSTITUTION, which is used to isolate the effects of
landowner wealth on contract choice. By definition, INSTITUTION identifies large,
wealthy landowners, thus isolating the cases when, by all traditional measures, the land-
owner should be less risk averse than the farmer. These landowners should be more likely
than smaller landowners to share contract with farmers. Risk sharing thus predicts positive
INSTITUTION coefficients for the logit estimation of contract choice and negative IN-
STITUTION coefficients for the tobit estimation of the farmer's cropshare. The coefficient
estimates for INSTITUTION are reported in table 6.7, along with the wealth and futures
market variables, and do not support the risk-sharing prediction. In the logit estimates in
the upper panel, only one estimated coefficient is positive but still statistically insignifi-
cant, while the other three coefficients are statistically insignificantly different from zero.
In the tobit estimates in the lower panel, all of the four coefficients are positive but none
are significantly different from zero.
Other estimates of INSTITUTION coefficients using the equations in tables 6.5 and 6.6
further undermine support for this risk-sharing hypothesis.
34
INSTITUTION was included
in the crop-specific logit equations used to estimate the coefficients for CV and STD in table
6.5. For each crop sample we estimated four equations, corresponding to the four different
measures of exogenous variability. For Louisiana, we estimated these equations for only
three crops (soybeans, rice, sugarcane) resulting in just six estimated coefficients. We find
that four of the six are insignificantly different from zero, but that both rice coefficients are
positive.
35
In order to estimate these equations for the Nebraska-South Dakota data, we
used a smaller sample for which the variable INSTITUTION was available (see appendix
A). Because there were no institutional landowners for sorghum (dryland and irrigated), we
estimated the crop-specific contract choice equations for only seven crops. This resulted in
a total of twenty-eight estimated coefficients (7 crops times 4 risk measures). None of the
estimated coefficients are significantly different from zero, indicating that INSTITUTION
does not affect contract choice as predicted by the risk-sharing model.
