more variability after 1995 than does Figure 3. Together with the fact that the current
profits of many banks were negative in these periods, these results may imply that it is
inappropriate to use the cost of capital as calculated above. However, the conclusions in
the previous section are confirmed with this analysis.
5.2 Bank reserves
The model in this paper disregards the existence of reserves for deposits. If we introduce
bank reserve R and assume that it is a constant ratio of deposit, i.e., R = βD , equation (8)
becomes:
θ ridtqit
Ri,t = t Ri,t + , , + Ci,t(a1 +2a2lnqi,t +a7lnwi,t +a8lndi,t)
ηt 1 - β
(8)’
qi t S
+ Ci , (a3 + 2a4lndi + a8lnq + a9lnwit) + εiS .
i,t 3 4 i,t 8 i,t 9 , i,t
(1 - β)di,t
Estimation of equations (7), (8)’, and (9), θt for regional banks is shown in Figure 6.
Figure 6 resembles Figure 3, but close inspection reveals that the curve shifts downwards,
so that θt took on a negative value in 1987. Estimates of θt for city banks are not
reasonable and not presented: θt is negative for most of the period, and although it
decreases until 1984, it rises after 1990 and takes almost the same value in the 1990’s as in
the 1970s.
We construct the data of β as money plus reserves divided by deposits. The mean
of β is about 0.1, which is much larger than the required reserve ratio. This implies that
16