Farm size is often used in profitability studies (Whittaker, Lin and Vasavada,
1995). If sales are used as a proxy for farm size, one could surmise that while revenues
from large farms would be more sensitive to external shocks brought on by
macroeconomic conditions, smaller farms would also be affected by such shocks. For
example, to the extent that trade restrictions result in retaliation, more agricultural
products would remain on the domestic market and depress the price of the affected
commodity. In that case, all farm operators, irrespective of farm size would be affected
by the fall in price of the affected commodity. Thus the coefficient of the farm size
variable is hypothesized to have a negative sign.
Although average annual gross sales provide a measure of farm size, it does not
necessarily follow that such sales provide most of the income of farm households.
Tavernier, Temel and Li (1997) show that off-farm income plays a major part in the
income of farm households. Thus a farm operator with sales of $10,000 - $$49,999 per
year and off-farm income of over $150,000 per year generates at most 25% of income
from farming. Moreover, the farm operation where 51-75% of family income is earned
from farming may fall in any of the farm sales categories. Thus the sign of the
coefficient of the income (inc) variable is unclear.
Empirical Results
The estimated coefficients with t-ratios from the logit models that provide the best model
fit are presented in Table 3. One of the variables from each category is dropped to avoid
multicollinearity. In addition to those results, Table 3 also presents the marginal effects
and goodness of fit measures such as the Chi-square test statistic, the Mc Fadden R2 test
statistic, and the percent of successful predictions. These measures are discussed in
14