The table 2 shows the results of the political pressure model. Signs of all explanatory
variables in each year are consistent with theoretically expected ones. Moreover, the
values of adjusted R square are more than 0.6 from 1988 to 1993. These levels are
judged to be a considerably good performance, considering average scores of
cross-section analyses. All figures of t-statistics of the two explanatory variables are
also highly significant. Overall, hypothesis 1 was empirically supported.
Consequently, these suggest that bureaucrats consider the balance of farmers’
grievances against the set-aside program among regions and tend to revise the allocation
of set-aside acreage in inverse proportion to the degree of farmers’ grievances. Thus,
political pressure can be succinctly explained by two variables: expected income, which
represents a loss of rent, and thereby the degree of grievance; and the share of part-time
farmers, which represents a majority of farmers who tend to express their complaints.
If we take a look at time-series figures in detail, a further discovery can be found.
From 1978 to 1986, the scores of adjusted R square had not been high, while, in the
middle part of the observation period, they had been adequately high. Then after 1994,
they have gradually been decreasing. This fact can be well explained by the policy
change which happened during this period. This will be explained in detail in the next
section.
Table 3 shows the results of the multiple regressions consisting of several selected
factors which are made public to be used in bureaucrats’ seemingly discretionary
allocation formula. According to these results, some explanatory variables in some years
were insignificant or showed the wrong sign. In particular, FTF showed the worst
statistical performance. Signs consistently contradicted expected ones and these wrong
signs and the figures of coefficients were highly significant. Since this explanatory
variable is one of the factors rarely advocated by political actors, except for bureaucrats,
and its features are contrary to the market mechanism, hypothesis 2 was empirically
supported.
On the other hand, COST showed good statistical performance. Signs are the same
17