Agriculture Reference
In-Depth Information
share of output as our dependent variable. Because this variable ranges from zero to one,
the model for each share contract
i
for crop
j
is
2
s
ij
=
σ
ij
γ
j
+
Z
j
ξ
j
+
ij
;
if
s
ij
<
0;
and
(6.11)
s
ij
=
1
otherwise
i
=
1,
...
,
n
j
;
j
=
1,
...
, 14,
th
contract governing the
th
crop;
2
ij
where
is the
crop-specific variability of the random input for a given plot of land (again measured by CV
or STD);
s
ij
is the farmer's share of the crop for the
i
j
σ
γ
j
is the corresponding crop-specific coefficient;
Z
j
is a row vector of explanatory
variables including the constant;
is
a crop-specific error term. The control variables are nearly identical to those used in the
estimation of equations (6.9) and (6.10) and are explained in appendix A.
We use a right-censored tobit model to generate maximum likelihood estimates of the
model given by equation (6.11) for the same fourteen crop-specific samples of cropshare
contracts from the Great Plains (9 crops) and Louisiana (5 crops) data.
30
Table 6.6 presents
the tobit estimates of 6.11 and shows the estimates of
ξ
j
is a column vector of unknown coefficients; and
ij
γ
j
from forty-six estimated equations,
thirty-six using Great Plains crop samples and ten using Louisiana crop samples. Like table
6.5, each entry in table 6.6 is an estimated CV or STD coefficient—that is, an estimate
of
, derived from a separate estimated equation. Accordingly, the entry in the upper
left cell
γ
j
is the estimated tobit coefficient for REGIONAL CV from equation 6.11
using a sample of cropshare contracts for dryland corn (
(
21.96
)
= 521) in Nebraska and South
Dakota. The rest of the table is organized like table 6.5. Prediction 6.3 implies a
negative
coefficient for the CV and STD variables; that is
n
j
crops. Of the forty-six
estimates, twenty-eight are not significantly different from zero, thus failing to support
the risk-sharing prediction. More than half (30) of the estimated coefficients actually have
a positive coefficient, and fourteen are significant. Only four estimates are negative and
significant.
31
γ
j
<
0 for all
j
Wealth, Risk, and Contract Choice
It is often assumed that as wealth increases, individuals become less risk averse in absolute
terms. The assumption of declining absolute risk aversion (DARA) for farmers is so routine
among agricultural economists that Pope and Just (1991) note, “Decreasing absolute risk
aversion has emerged as a 'stylized' fact or belief” (743). In our model, DARA implies
that wealthier farmers should cash rent more often than poor farmers (prediction 6.6). This
follows, because as wealth increases, the amount of exogenous risk the farmer is willing to
bear should rise. A corollary to this prediction is that the share the farmer receives should
also rise with his wealth (prediction 6.4). Larger output shares mean that the farmer is
bearing more of the exogenous variability.
