This paper proposes a new empirical methodology to measure the effect of monetary policy, which is applicable even under a low interest rate
environment, and applies it using Japanese data from 1998 to the present. First, I use the principal component analysis to extract the common features in financial market during the monetary policy board meeting days, and find that there are at least 3 common factors to characterize the daily changes in various kinds of financial market variables in response to the policy announcement. Second, I impose some restrictions to identify three kinds of structural monetary policy shocks, namely `current policy', `long-run expectation on future economic condition', and `future policy path' surprises. I find that (1)`long-run expectation' and `future policy path' surprises have a significant effect on real output and prices as well as stock and exchange rate market, while `current policy' surprises play a limited role in influencing the real economy, and (2) `current policy' surprises contribute to the financial market stability. This implies that the central bank can not only mitigate the malfunction in financial market, but also stimulate the real economy even when it face on the zero lower bound of short-term interest rate.