مراحل عملکرد بانک ژاپن و شناسایی شوک سیاست پولی: یک بازنگری با استفاده از روش برنانکه می هو

This paper reexamines the operating procedures of the Bank of Japan (BOJ) using the structural VAR approach of Bernanke and Mihov [Measuring Monetary Policy, Quart. J. Econ. 113 (1998) 869–902]. This approach identifies the exogenous components of monetary policy by establishing equilibrium models of the reserve market. In this paper, two equilibrium models are presented: the Implicit Cost (IC) model and the Credit Rationing (CR) model. These particular models are distinguished by opposing views about the BOJ's discount-window borrowing policy. The IC model is characterized by the assumption that the BOJ endogenously accommodates the demand for discount-window borrowing by private banks, whereas the CR model assumes that the BOJ exogenously controls the level of discount-window lending. The results indicate that the CR model is superior to the IC model in describing the operating procedures used by the BOJ up to June 1995. J. Japanese Int. Economies20 (3) (2006) 406–433.

The analysis of monetary policy indicators that are controlled by central banks, and assumed to be closely related to the real economy, is one of the most important issues in macroeconomics. Some economists have examined short-term interest rates in this regard, including the call rate and the federal funds rate, while others have considered money supply variables such as M2 + CD and high-powered money. However, before the analysis of any particular monetary variable as a prospective policy indicator can be undertaken, a number of questions arise. How can policy indicators be specified which precisely reflect the central banks' past policy decisions? Can policy indicators be specified using single monetary variables? If not, how should they be specified? The current paper is motivated by these empirical questions. In particular, the paper explores the best policy indicator of the Bank of Japan (BOJ) and identifies policy shocks by employing the structural VAR approach of Bernanke and Mihov (1998). This approach allows the specification of central bank policy indicators and the clarification of operating procedures by formulating equilibrium econometric models of the reserve market. In general, central banks aim to stabilize the macroeconomy by intervening in the reserve market and setting short-term interest rates or reserves within a target range. The approach selected is convincing because it assumes that monetary variables that are affected by the operating procedures of central banks in the reserve market embody the decisions of central banks. In a traditional structural VAR framework, the most convenient identification scheme for the exogenous components of monetary policy since Sims (1980) is based on Cholesky decompositions. If this identification scheme is employed, we must select a priori a single measure of monetary policy and also specify a recursive structure for the macroeconomy. In the VAR literature concerning Japanese monetary policy, Miyao, 2002 and Miyao, 2000, Ogawa (1999), and Hatakeda (1997) assume that the BOJ's policy stance can be measured by the call rate. They then examine business fluctuations and the role of monetary policy in Japan.1 In more recent work, Sims and Zha (1998) suggest an identifying methodology that does not depend on the recursive assumption, and which imposes a contemporaneous restriction on all economic variables in a VAR system. As an alternative, Sims (1986), Gordon and Leeper (1994), and Leeper et al. (1996) impose a contemporaneous restriction on all economic variables in a VAR system, assuming that at least a subset of goods market variables are predetermined. In this paper, an identification scheme that imposes a contemporaneous restriction on all economic variables in a VAR system is titled the “Sims scheme.” By way of contrast, Bernanke and Mihov (1998) suggest an identifying methodology that divides the macroeconomy into a policy sector and a non-policy sector, and after assuming a block recursive structure between the two sectors, imposes a contemporaneous restriction on monetary variables in the policy sector. In this paper, such identification is called the “Bernanke–Mihov scheme.” These identification schemes are more sophisticated than the Cholesky approach, in that they quantitatively clarify a central bank's operating procedures, and specify the policy indicator in the process of identifying the policy shock. Shioji (2000), for example, used the Sims scheme to examine Japanese monetary policy, while Kasa and Popper (1997) employed the Bernanke–Mihov scheme. In terms of results, Shioji found that the BOJ had targeted both money supply, such as M2 + CD, and short-term interest rates, including the call rate, until June 1995, while Kasa and Popper argued that the BOJ had targeted both non-borrowed reserves and the call rate; that is, both concluded that the BOJ had not targeted a single variable, but a mixed monetary variable. According to this view, if a single monetary variable is specified as a BOJ policy indicator, conjectures about the monetary transmission mechanism would be erroneous. This paper presents an econometric model of the Japanese reserve market that is significantly different from Kasa and Popper, and reexamines the policy indicators of the BOJ. When examining the BOJ's operating procedures and modeling the Japa-nese reserve market, it should be noted that short-term interest rates such as the call rate remained above the discount rate until June 1995.2 This led to two differing views on the BOJ's discount-window policy and caused difficulties in understanding its operating procedures, and thereby modeling, the reserve market. One is the “Implicit Cost Hypothesis,” which assumes that the BOJ endogenously accommodates the demand for borrowing by private banks. The other is the “Credit Rationing Hypothesis,” which assumes that the BOJ exogenously controls the level of discount-window lending. From the 1980s to the early 1990s, economists and policy makers discussed which of these views was true, but failed to reach agreement. Moreover, since lowering the call rate below the discount rate in July 1995, the BOJ has adopted the new framework for its operating procedures. Therefore, there is no consensus about how the BOJ implemented its operations in the reserve market up to June 1995, and we have very little knowledge about how the BOJ influenced the macroeconomy by operating in this market.3 In this paper, we present two equilibrium models of the Japanese reserve market, which are differentiated by opposing views about the BOJ's discount-window policy, and deal with the following three problems: (1) How did the BOJ implement its operations in the reserve market until June 1995? (2) How can we characterize the policy shock and indicator of the BOJ up to June 1995? (3) To what extent did the policy shock of the BOJ influence the macroeconomy? The paper itself is organized as follows. Section 2 discusses the BOJ's discount-window policy up to June 1995, in order to model the reserve market. This section also briefly reviews the studies of Shioji and Kasa and Popper. Section 3 demonstrates the VAR-based methodology suggested by Bernanke and Mihov. Section 4 presents two alternative equilibrium models of the Japanese reserve market, which are differentiated by opposing views about the BOJ's discount-window policy. Section 5 discusses the estimation method and the results, and identifies the best BOJ policy indicator. Section 6 discusses the extent to which our identified policy shock influences the macroeconomy by deriving impulse response functions. Some concluding remarks are made in the final section.

This paper draws three substantive conclusions. Firstly, with regard to its operating procedures up to June 1995, the Bank of Japan (BOJ) fully offset demand shocks in the market for bank reserves and aimed to stabilize the call rate by using both open-market operations and discount-window lending. Secondly, the BOJ's policy shock up to June 1995 emerges as unanticipated changes in the call rate. Therefore, we suggest that the call rate should be identified as the best policy indicator of the BOJ up to June 1995. This suggestion is especially noteworthy because Shioji and Kasa and Popper point out that no single operating procedure can fully explain the behavior of the BOJ. Thirdly, with regard to the effects of the BOJ's monetary policy shock to the macroeconomy, the response of output reaches its peak after about one year before declining back to zero after about three years. The response of price exhibits the nominal rigidity of about one year and reaches its peak after some 30 months. The response of reserves exhibits the liquidity effect in the market for bank reserves for about one year following the shock to the call rate. These conclusions are obtained without splitting the sample from January 1975 to June 1995. However, the most attractive feature of the Bernanke–Mihov approach is that the approach can determine how central banks make decisions in response to the institutional and macroeconomic changes surrounding monetary policy in each period, and what combinations of monetary variables can explain these facts. Future refinement requires that Japanese monetary policy in each period be statistically conceptualized by splitting the sample according to the institutional and macroeconomic changes surrounding the BOJ.