Figure 3 shows the MVR estimates of θt together with their 95% confidence interval.
During the entire sample period, θ= n , θ= 1, and θ= 0 were all rejected. This implies
that the degree of competition was less severe than Cournot oligopoly, although it was more
severe than joint profit maximization. θt declined from 1974 to 1987, and the decline was
especially rapid in the 1970s, implying that competition improved during this period. θt
then rose to around 0.35 and did not change from 1988 to 1998. It rose again markedly
after 1998.
In summary, the degree of competition improved from the 1970s to 1987, and was stable
during the late 1980s and the first half of the 1990s. Competition has recently become
laxer.
The 3SLS estimates shown in the right columns of Table 4 are not remarkably
different from those generated by MVR. However, the 3SLS estimates of θt are smaller
than the MVR estimates in the early period.
Comparing Figure 3 with Figure 2, we find that competition among city banks has
been stronger than among regional banks. The result is plausible because the business
areas of regional banks are limited, even though they overlap, so that they may possess
monopolistic power in their areas.
The results of the basic estimations are summarized as follows: First, competition in
the Japanese banking sector had improved in the 1970s and in the first half of the 1980s.
Second, the degree of competition is higher for city banks than for regional banks. Third,
14