years, 1974 and 1975, the results reject θ= n , θ= 1, and θ= 0, at a 5% significance level.
θt generally had a downward trend, with a spike in 1981, until 1984. It stayed around
zero for 1984-1989, and rose in the first half of the 1990s, rejecting perfect competition for
these years. In the middle of the 1990s, θt again took on a value around zero, while it
rose again in the final few years. In summary, competition improved until 1984, and the
competitive environment obtained until 1997. Cournot oligopoly was not rejected for most
of the period. Perfect competition was not rejected for about half of the period.
4.3 Estimation results of regional banks
Results of simultaneous estimation of equations (7), (8), and (9) for regional banks are
presented in Table 4. The results by MVR are shown in the left columns. All the
estimates of the cost function are significant except for that of the quadratic term of deposits.
A problem is that the coefficients of lnw (and (lnw)2 ) are negative. We suspected that
the wage rate might be an endogenous variable and conducted 3SLS estimation using a rank
variable of wi,t as an instrumental variable. However, the signs of the coefficients did not
change.
The estimates of the demand function significantly satisfy the expected signs except
for that of operation funds. The estimates of ɪ took on a value around 0.43 and were
ηt
significant for all the years. Fits are good for all the equations: the coefficient of
determination exceeds 0.93.
13