exogenous variables, wi,t , ASLi,t, SMSFi,t , OPLi,t , and Qt , and year dummies.16
4.2 Estimation results of city banks
Results of the simultaneous estimation of equations (7), (8), and (9) for city banks are
presented in Table 3. In the left columns, the estimates by MVR are shown. As for the
cost function, four coefficients out of ten are significant at a 5% level of significance. As
for the demand function, the coefficient of IIPt is significantly positive and that of lnQt is
significantly negative, confirming that the demand function (9) is identified from the supply
relation (8). The estimates of ɪ took on a value of around 0.7 and were significant for
ηt
all the years. However, the estimate of b4 unexpectedly took on a negative value. This
may be interpreted as meaning that the small- and medium-sized firms that could obtain
loans from city banks were promising firms and not riskier than large firms.
In the right columns of Table 3, the estimates by 3SLS are shown. They are similar
to those by MVR, but the results of the cost function are not convincing: the coefficient of
lnd is non-positive, casting some doubt on the results.
The estimates of θt by MVR are depicted in Figure 2 together with their 95%
confidence interval. The value of 1/n is also depicted as a proxy of Cournot oligopoly.
The results from 3SLS are quite similar so they are not presented here. For the first two
important role in identifying the marginal cost part in equation (8).
16 The rank variable of Xi is the variable whose value is the order of Xi ranked in ascending order.
(See Maddala, 1977, pp. 297-298.)
12