Agriculture Reference

In-Depth Information

the positive finding in the second specification is the different type of irrigation present in this

data. In Louisiana, flood irrigation is exclusively for rice, where it is used to control pests.

This type of system is quite homogeneous across the rice growing region of south Louisiana

and may not be a good measure of specific assets. In the British Columbia-Louisiana

equation, the estimated coefficient for TREES is negative and statistically insignificant,

as predicted.
22

Because the type of contract (cropshare or cash rent) is likely to be a determinant

of contract length we include a variable CONTRACT that indicates the presence of a

cropshare arrangement. Cropshare contracts have a built-in adjustment mechanism, because

the payment (from farmer to landowner) is in terms of the crop and not dollars. This means

that changes in market prices do not require changes in the term of the contract, making

adjustment clauses unnecessary and multiyear contracts less costly. This implies a negative

coefficient for the variable CONTRACT, which we find in both equations. Both coefficients

are statistically significant in a one-tailed test.

The better the reputation of the farmer or of the landowner, the more likely the contract

will be annual, as implied by prediction 3.1b, because the simple annual contract can be

relatively easily enforced. Again, the age of the farmer (AGE) can be used to measure repu-

tation, and we expect a quadratic relationship between a farmer's age and probability of an

annual contract. In both equations we find, as expected, the estimated coefficients for AGE

to be positive and for AGE
2
to be negative. All coefficients are statistically significant in a

one-tailed test. As with the oral-written contract estimates, we used several variables to mea-

sure information about reputations. The estimated coefficients for CHANGED PARTIES

and YEARS DURATION are expected to be negative. YEARS DURATION is expected to

be negative because an increase in contracting duration makes it more likely the parties will

trust each other enough to enter into a longer term agreement. The estimated coefficients

for YEARS DURATION are negative as predicted, but not statistically significant. The es-

timated coefficient for CHANGED PARTIES (only available for Nebraska-South Dakota)

however, is positive and significant, refuting our prediction. Except for this anomaly, though,

these estimates of the effects of reputation offer broad support for prediction 3.1b.

Finally, as in the logit estimates of ORAL, we use ACRES, FAMILY, INFORMATION,

and VALUE as variables that control for the value of the contracted plot of land and the

information among the parties. In this setting we have no predictions for the estimated co-

efficients. We do find that the estimated coefficient for ACRES is negative and statistically

significant for British Columbia-Louisiana but is not significantly different from zero for

Nebraska-South Dakota. VALUE, available only for the Nebraska-South Dakota data, is

positive and statistically significant. The estimated coefficients for FAMILY and INFOR-

MATION also vary from sample to sample.