Was Japan's Real Interest Rate Really Too High During the 1990s? The Role of the Zero Interest Rate Bound and Other Factors

Japan's more than a decade long quot;Great Recessionquot; has presented a disconcerting case of what could happen if interest rates are bounded by zero and deflation sets in. Since Krugman (1998), the commonplace observation is that the deflationary situation combined with the zero nominal interest rate has caused elevated real interest rates, thereby nullifying monetary policy. This paper investigates this oft-cited claim and examines whether it is associated with anomalies in the way real interest rates are determined by employing an error correction model (ECM) based on the time-varying parameter model with Markov-switching variances. Using this model it is revealed that during the 1980s both ex ante and ex post rates were often lower than the equilibrium rates, indicating strong and persistent optimism among agents. However in the 1990s the ex ante real interest rate was persistently higher than the equilibrium, indicating the pessimistic expectations among agents. The time-varying speed of convergence to the equilibrium appears to slow down considerably in 1996-99, making the misalignment in the real interest rate process last twice as long as in the 1980s. In addition the analysis using the Smooth Transition Regression (STR) model shows a regime shift in the real rate process in mid-1995, three years before the implementation of the zero interest rate policy. This result suggests that a situation with an extremely low nominal interest rate, even before it reaches the zero bound, may create anomalies or nonlinearity in the effectiveness of monetary policy