The interest rate has been recognized as the main instrument of monetary policy of most
Central Banks to reach inflation stability, output stability and maybe exchange rate
stability. The main objective of this paper is to identify the targets of the monetary policy
conducted by the Bank of Japan (BOJ).
There is a proposition among Japanese scholars that the BOJ has given a major emphasis
on exchange rate targeting instead of inflation rate targeting. The empirical works of
Hutchison (1988), Okabe (1995) and McKinnon and Ohno (1997) are good examples of this line of research. Their results point to a minor role for output gap and a major role for
exchange rate deviations on the monetary policy of the BOJ.
McKinnon and Ohno’s argument is that exchange rate is a forcing variable and the
domestic prices level an adjusting variable. Accordingly, stabilization of the exchange rate
may lead to stabilization of the price level in the long run.
Taking an alternative view, Chinn and Dooley (1997) and Clarida et al. (1998) present
new evidences in favor of the inflation targeting approach.
Chinn and Dooley (1997) estimate an interest rate reaction function, using as targets the
forecasts of inflation and output gap obtained from a structural VAR of the main economic
aggregates that are related to monetary policy, as suggested by Clarida and Gertler (1996).
Their findings indicate the relevance of inflation and output stabilization to the BOJ policy.
They observe that the inclusion of real exchange rate deviations is not statistically
significant to explain the behavior of the interest rate. Much stronger evidence in favor of inflation targeting regime is found in Clarida et al.
(1998). They estimate a reaction function using a forward-looking framework, where the
arguments of the function are deviations of inflation and output gap in relation to their
target values. Their results would be consistent with the view that exchange rate contains
all the information on future inflation, as suggested by McKinnon and Ohno. Their
empirical analysis, however, did not consider adequately the properties of the time series.
They use non-stationary series, or integrated of first order, in a GMMmodel without testing
for the possibility of cointegration among them. This procedure limits seriously the results
of the asymptotic theory and may invalidate their estimation process (e.g., Davidson and
Mackinnon, 1993).
This paper aims to identify the behavior of the BOJ in the management of the monetary
policy using the methodology of cointegration analysis, impulse response functions, and
historicaldecompositionof theresidualsduringthecyclicalmovementsof thenominal interest
rate. The period of the analysis is the same as the one considered by Clarida et al. (1998).