رژیم ره گزینی سیاست های پولی در کانادا

This study captures regime-switching, monetary policy responses to financial market disturbances in Canada. Monetary policy is identified within a nonlinear, structural VAR framework with a regime-switching policy block that allows for contemporaneous policy reactions in a small open economy. The key finding is that monetary policy in Canada has undergone important changes in monetary policy regimes since the early 1970s that correspond to changes in operating procedures and medium-term inflation targets. Overall, the Bank of Canada is found to use a “hybrid” operating procedure in different monetary policy regimes, where it sets a monetary policy rate, while “leaning against the wind” to smooth exchange-rate fluctuations and to offset unexpected increases in long-term yields in order to maintain monetary conditions. Although the magnitude of the contemporaneous monetary policy responses are regime-dependent, the transmission of the responses through financial markets are regime independent. The study also constructs regime-switching measures of the overall level and relative volatility of monetary conditions that captures both the endogenous and exogenous components of the monetary policy response function.

Vector autoregression (VAR) models have been widely used to recover monetary policy innovations and policy responses to economic and financial disturbances. However, the VAR reaction functions generally rely on the strong assumptions of the linearity properties of moving average representations, and the time invariance of estimated coefficients and variance–covariance structures.1 These assumptions are particularly problematic in light of the growing literature documenting that the stochastic behaviour of interest rates varies over time.2 The important policy issue is the stability of the reaction function because central bank behaviour may change with different operating procedures or medium-term operating targets. Another common criticism of the VAR approach to the study of monetary policy is that it mainly captures policy innovations and not the important endogenous or systematic responses that rely on the central bank’s information set. The criticism implies that VAR response functions may generate monetary policy choices that are uncorrelated with current economic and financial surprises. An overall measure of monetary conditions should also include endogenous responses to important policy variables or indicators. This study overcomes some of the limitations of VAR models using the Bernanke and Mihov (1998) methodology that identifies monetary policy responses within a nonlinear, structural VAR framework with a regime-switching policy block. The structural VAR approach has the advantage of specifying an explicit monetary policy response function using non-recursive contemporaneous restrictions, rather than relying solely on reduced-form equations and a Choleski factorization that mechanically imposes a “semi-structural” interpretation. Instead, the methodology in this study allows for contemporaneous policy reactions in a small open economy. In addition, the regime-switching framework has the advantage of allowing parameters of the policy response function to vary over time in response to changes in operating procedures and targets.3 The regime-switching approach allows inferences about changes in central bank behaviour that are based entirely on the data. The empirical framework in this study extends the Bernanke and Mihov methodology by allowing the variances of the orthogonal shocks in the endogenous policy block of the model to vary between the states in order to provide for a more interesting shift in the structure of the policy block. Sims and Zha, 2009 and Valente, 2003 find that shifts in the variances of regime-switching VAR models are very important for improving fit. This study also constructs measures of the overall level and relative volatility of monetary conditions that captures the time-varying endogenous responses of monetary policy to important economic and financial disturbances, as well as exogenous innovations in macroeconomic variables. The key finding is that monetary policy in Canada has undergone significant changes in monetary policy regimes since the early 1970s that correspond to changes in operating procedures and medium-term inflation targets. In particular, the data choose two distinct high-variance, large-response regimes, both coinciding with changes in monetary policy operating procedures or targets and with sharp drops in the trend rate of inflation. The first regime occurs mainly during the first half of the 1980s, when the trend rate of inflation dropped from the double-digit range to slightly above 4%, and the Bank of Canada abandoned its policy of monetary gradualism tied to targets for M1 growth. The second regime occurs mainly during the first half of the 1990s, when the trend rate of inflation declined another step to around 2% and explicit inflation-control targets were adopted by the Bank and the Government of Canada. In short, the two high-variance, large-response regimes coincide with changes in monetary policy operating procedure and with shifts to lower inflation regimes. The data also choose two distinct low-variance, small-response regimes. The first regime spans the period between the two high-variance regimes during the latter half of the 1980s when the Bank of Canada maintained trend inflation around 4% and set the level of the overnight rate to achieve overall monetary conditions across the interest rate, exchange rate, and credit channels of the monetary transmission mechanism. The second regime occurs from the late 1990s to the present when the Bank explicitly targeted a 2% rate of inflation and continued to use the overnight rate to achieve an overall level of monetary conditions. Overall, the Bank of Canada is found to use what is often referred to as a “hybrid” operating procedure in both regimes, where it targets the overnight rate, while “leaning against the wind” to smooth exchange rate fluctuations and to offset unexpected increases in long-term yields in order to maintain monetary conditions. In particular, the estimated coefficients indicate that contemporaneous monetary policy responses to shocks to either the exchange rate or the long-term yield are about three times larger during volatile interest-rate periods than during the more “normal” periods. However, the contemporaneous responses of credit and exchange markets to monetary policy responses and shocks are found to be relatively similar across the different regimes. Thus, although the contemporaneous financial component of the monetary response function changes with regime switches, the transmission of monetary policy through financial markets does not change with regimes. The following section briefly reviews some previous research on regime-switching VAR approaches to monetary policy. Sections 3 and 4 outline the empirical methodology and the monetary policy model. Section 5 discusses the data. The estimated impulse responses for the structural VAR are presented in Section 6 and the regime-switching policy responses in Section 7. The monetary policy regimes and regime-switching measures of monetary conditions are discussed in Section 8. The final section outlines some limitations of the study and issues for further research.

This study constructs a model that identifies monetary policy responses within a regime-switching, structural framework that explicitly allows for contemporaneous policy reactions to financial disturbances in a small open economy. The key finding is that monetary policy in Canada, like the United States, has undergone important changes in monetary policy regimes since the early 1970s that correspond to changes in operating procedures and inflation targets. Overall, the Bank of Canada is found to use a hybrid operating procedure in both regimes, where it targets the overnight rate, while “leaning against the wind” to smooth exchange-rate fluctuations and to offset increases in long-term yields triggered by increases in risk premiums. The empirical results find that while the size of the policy responses to financial disturbances are regime-dependent, the transmission of policy shocks and responses through financial markets are regime independent. The study also constructs a regime-switching measure of the level of overall monetary conditions and the relative volatility of monetary conditions that captures both the endogenous and exogenous components of the monetary policy response function. The model in this study avoids some of the criticisms of the VAR approach by incorporating current information about financial disturbances into the endogenous component of the monetary policy response function. However, the information set should also allow for economic conditions, such as the output or inflation gap, that are also believed to affect monetary policy choices. Recent work on monetary policy rules by Sims and Zha, 2009 and Valente, 2003 using regime-switching VAR models are examples of future research directions that incorporates economic variables into their models in order to recover implicit monetary policy rules. Also, a more complex Markov-switching model would allow for the covariance matrix of the financial variables in the policy block, not just the variances, to vary between the states. This would allow for regime-dependent, impulse response functions, which could provide some insight into the extent to which the regime-specific, monetary policy behaviour is related to systematic responses to financial disturbances or the covariance–variance matrix of the financial innovations themselves. Similarly, a broader switching model could provide insight into whether the responses of the macroeconomy to monetary policy shocks differs between regimes, and whether the dynamic responses are due to changes in the propagation of the shocks or the degree of the variability of the policy shocks.