In this paper, we examine the relative efficiency of various targeting regimes by using linear quadratic dynamic programming given estimated parameters. This analysis yields two notable implications. First, in Japan, money supply growth targeting generates higher inflation volatility and higher GDP-gap volatility than other targeting regimes including inflation targeting and GDP-gap targeting. This conclusion differs from that of Rudebusch and Svensson [Rudebusch, G.D., Svensson, L.E.O., 2002. Eurosystem monetary targeting: lessons from U.S. data. European Economic Review 46, 417–442] for the US. Second, strict inflation targeting and flexible inflation targeting are more efficient than other targeting regimes including strict GDP-gap targeting.
The adoption of various targeting regimes, especially inflation targeting, has been studied and discussed. Rudebusch and Svensson (2002) compare the relative performance of various targeting regimes (inflation targeting, output-gap targeting, and money growth targeting) under an empirical model in the US. Adopting their procedure, in this paper, we compare the relative performance of the various targeting regimes to determine which targeting regime is desirable in Japan.
The result of this analysis is interesting and yields two notable implications. First, in Japan, money supply growth targeting generates higher inflation volatility and higher GDP-gap volatility than other targeting regimes including inflation targeting and GDP-gap targeting. This conclusion differs from that of Rudebusch and Svensson (2002) for the US. Second, strict inflation targeting and flexible inflation targeting are more efficient than other targeting regimes including strict GDP-gap targeting.
This paper comprises four sections. In Section 2, we present the estimated model for Japan. In Section 3, we describe the theoretical method. In this section, we rewrite the model in matrix form, define the targeting regime of the central bank, and present an overview of the use of linear quadratic dynamic programming as a tool for evaluating the relative efficiency of targeting regimes. In Section 4, we evaluate the relative efficiency of the various targeting regimes by using the efficiency frontier based on the criterion of social loss. We then derive conclusions.
Recently, the adoption of various targeting regimes, especially that of inflation targeting, has been studied and discussed as an alternative regime in Japan. In this paper, we examined the relative performance of various targeting regimes in Japan by using the method of Rudebusch and Svensson (2002).
We obtained two notable conclusions. The first is that, in Japan, money supply growth targeting produces higher inflation volatility and higher GDP-gap volatility than other targeting regimes, including inflation targeting and GDP-gap targeting. This result differs from the finding of Rudebusch and Svensson (2002) for the US.
Second, Rudebusch and Svensson (2002) found the trade-off relation of the efficiency frontier, which connects SIT and SGT, for the US economy, even if the interest rate smoothing term is included in their loss function. This paper finds the similar efficiency frontier for the Japanese economy. However, our frontier curve also includes some part where the trade-off between inflation variability and GDP-gap variability disappears.
Finally, we suggest three avenues for further research on this topic. The first involves expanding our model. In this paper, we examined three types of targeting regime (inflation targeting, real GDP-gap targeting, and monetary growth targeting). Expanding our model would enable discussion of other targeting regimes, such as targeting of the nominal exchange rate. As McCallum (1999) has argued that depreciation of the yen may overcome deflation, it is worth analyzing the exchange rate.
Second, comparisons could be made with a forward-looking model. The analysis of this paper is based on a backward-looking approach. Although the backward-looking model is valid, it does not allow derivation of the social welfare function. Cristiano et al. (2005) adopt a forward-looking approach. By adopting such an approach, it would be possible to derive the social welfare function and avoid internal inconsistency.
Third, in this paper, the loss function of the central bank depends on variability in the inflation rate, the GDP gap, monetary aggregates, and interest-rate smoothing. However, there is some evidence that Bank of Japan reacted to movements in stock prices in the 1990s (see Bernanke and Gertler, 1999). Such a reaction might have distorted our estimates of the social loss function. To address this issue formally, we must explicitly include variability in stock prices in the central bank's loss function. This is beyond the scope of this paper but is worthy of future research.
