Agriculture Reference
In-Depth Information
Two types of data are used to test these enforcement cost predictions. The predictions
about specific assets and reputation are tested against contract data using regression analysis.
The predictions about the common law are tested against the historical record of farmland
contracts around the turn of the twentieth century.
The regression analysis data come from the
1986 Nebraska and South Dakota Land
Leasing Survey
, and the
1992 British Columbia and Louisiana Leasing and Ownership
Survey
. These surveys collected data from farmers and landowners engaged in farmland
leasing during the 1986 and 1992 crop seasons. Each observation is a single contract between
a farmer and a landowner, where each contract is an exchange of rights to a tract of land.
Before beginning the analysis, we find it useful to examine some features of the data used
in this chapter. The definitions of the variables are provided in table 3.1. The definitions and
summary statistics for all variables used in the topic are found in appendix A. Our empirical
focus is to explain whether a contract is oral or written and whether a contract is annual or
multiyear. ORAL is a dummy variable that identifies oral contracts, and ANNUAL is a
dummy variable that identifies annual contracts. They are used as dependent variables in
the analysis that follows. Independent variables can be grouped into those that measure
specific assets and those that measure reputation. IRRIGATION and TREES are dummy
variables that indicate the presence of irrigation systems and tree crops, respectively, and
both are used as specific asset variables. Several variables are used to measure reputation and
the market enforcement of contracts: AGE (age of farmer), AGE
2
, CHANGED PARTIES
(dummy indicating new contracting party), and YEARS DURATION (duration of contract
in years).
16
The size of the leased plot (ACRES), a dummy indicating parties knew each
other (INFORMATION), a dummy for leases between family members (FAMILY), and the
per acre value of the plot (VALUE) are used as control variables.
We use our contract data to estimate the determinants of contract structure (oral versus
written, annual versus multiyear contracts) and test the previous predictions. We use the
following empirical specification, where for any contract
i
the complete model is
C
i
=
X
i
β
i
+
i
i
=
1,
...
,
n
;
and
(3.1)
1,
C
i
>
if
0
C
i
=
(3.2)
C
i
≤
0,
if
0,
C
i
where
is an unobserved farmland contract response variable;
C
i
is the observed dichoto-
mous choice of contract structure for plot
(in table 3.2 this is equal to 1 for oral contracts
and equal to 0 for written contracts and in table 3.3 this is equal to 1 for annual contracts
and equal to 0 for multiyear contracts);
i
X
i
is a row vector of exogenous variables including
the constant;
β
i
is a column vector of unknown coefficients; and
i
is a plot-specific error
