برآورد اولویت های بانک مرکزی از یک تابع واکنش های تجربی متغیر با زمان

We estimate monetary policy reaction functions for the United States, the United Kingdom, and Germany, using a Markov-switching model that allows for shifts in the coefficients of the central bank's reaction function as well as for independent shifts in the residual variance. The results indicate that central bank policy can be characterized as falling into a low- and a high-inflation regime. Over time all central banks have assigned changing weights to inflation and the output gap. Switching in the residual variance turns out to be important for the fit of the model. As a reaction function is a reduced form, coefficients embody preference parameters as well as parameters characterizing the structure of the economy. To recover the central banks’ preferences, we estimate the preference parameters jointly with a small model of the economy. The results show that the Bundesbank has placed a relatively higher weight on inflation than the Fed. Moreover, for the Bundesbank and the Fed the differences between both regimes seem to originate mainly from a changing preference for interest rate smoothing.

Since Taylor (1993) showed that interest rate policy in the United States can be described by a simple mathematical function comprising an interest rate, inflation, and the output gap, it has become common practice to use estimated central bank (CB) reaction functions for summarizing monetary policy behavior. Empirical studies of reaction functions, however, are typically confronted with the problem that CB policy changes over time, making parameter estimates unstable, especially when considering a longer sample period.1 Moreover, estimation of a CB reaction function does not allow us to draw conclusions on the preferences of the CB, as the reaction function is only a reduced form. Coefficients in the reaction function depend on the CB's preference parameters as well as parameters characterizing the structure of the economy. In this paper we try to separate both to investigate shifts in the CB's preferences under the assumption that the economic structure has remained constant. According to Orphanides (2004), CBs in the 1970s put a high weight on output while inflation was allowed to rise, whereas in the more recent past most CBs concentrated on achieving low inflation, and output goals received less attention. Such changes in CB policy may either take the form of a gradual shift or a sudden switch to another regime. While one could alternatively analyze a gradual policy shift in the framework of a smooth transition model, this paper is concerned with more abrupt changes of regime that are suitable for investigation with a Markov switching model. Experience shows that such shifts in CB policy have occurred, see Romer and Romer (1989). The CB's preference towards inflation could also shift with increasing CB independence. Fischer (1994, p. 293) presumes that dependent central bankers suffer from an inflationary bias, while independent central bankers develop a deflationary bias. Developments like, e.g. in the United Kingdom, where the introduction of the Monetary Policy Committee in 1997 marked a major change in monetary policy making, are also likely to have influenced the preferences of the CB. The Markov-switching model used in this paper characterizes CB policy as falling into two different regimes.2 One regime is expected to correspond to a high-inflation regime with a high weight on output stabilization and a low weight on inflation. In the second, low-inflation regime the CB should take a more aggressive position towards inflation with a high weight on inflation and a lower weight on output stabilization. Switching models are also able to deal with the changing variability of economic time series. The oil-price shocks in the 1970s increased the volatility of interest rates, inflation, and output. Other turbulent episodes in the stock or foreign exchange markets or the terrorist attacks on September 11, 2001 could induce volatility in interest rates. As these episodes are not necessarily related to the monetary policy regime in place, they are assumed to switch independently of the switching process driving the coefficients in the reaction function. The paper is structured as follows. Section 1 reviews the existing literature on time-varying CB reaction functions. Though empirical research on this topic is still tentative, evidence that CB reaction functions are non-linear is increasing rapidly. Since we are concerned with shifts in the CB's preferences here, we concentrate on studies investigating changing CB behavior as opposed to changes in the structure of the economy. Section 2 discusses the specification of the Markov-switching model and the estimation procedure. It is assumed that the monetary policy regime switches according to a first-order Markov process, while a second independent Markov process determines switching in the residual variance of the reaction function. Section 3 discusses the data and Section 4 presents the estimates. The countries investigated are the United States (US), the United Kingdom (UK), and Germany, which are large economies that pursued an independent monetary policy over the post-Bretton Woods era. Assuming that CB policy can be characterized by two different regimes, the results indicate that these regimes are significantly different from each other and correspond to a low and a high inflation target rate. Switching in the residual variance is relevant for all countries and contributes significantly to an improved fit as compared to a simple linear model. Svensson (1999) showed that the coefficients in the CB's reaction function are convolutions of the CB's preferences and other parameters describing the structure of the economy. Section 5 tries to isolate the CB's preference parameters from the estimated reaction function. Rudebusch and Svensson, 1999 and Rudebusch and Svensson, 2002 describe the economy by an IS and a Phillips curve and a CB's loss function. Under the assumption that the economic structure is constant over time but that the parameters of the loss function may shift, this model is used to derive the weights the CB assigns to the inflation and the output target. For the UK, output and interest rate smoothing receive a higher weight in the CB's loss function in the high-inflation regime. For the US and Germany, the main difference between both regimes relates to the weight on interest rate smoothing. Section 6 concludes.

This paper adds to the growing literature on time-varying monetary policy rules. Switching models are an alternative to a conventional linear specification of a monetary policy rule, as the effects of a changing economic environment on monetary policy can be investigated without having to determine the dates of the changes exogenously. In this paper CB reaction functions for US, the UK, and Germany have been estimated, using a Markov-switching model with independent switching processes for the coefficients in the reaction function and the residual variance. The results show that the reaction function can be better modelled with a two-state switching model than with a linear model using the same variables. While switching in the residual variance improves the distribution of the residuals, coefficient switching has an independent, significant effect for the US and Germany and is able to produce implied interest rates that follow actual interest rates more closely. In interpreting the results one has to keep in mind that the CB reaction function was estimated under the assumption that the CBs follow a truly independent monetary policy. The deviation of the implied interest rates from the actual interest rate for the UK during the membership in the EMS indicates that for this period the assumption of an independent monetary policy apparently was violated. While for the US the estimated weight on output lies at the higher end of the estimates reported in existing studies, the weight on interest rate smoothing is much lower and thus more in line with theoretical models. Our results show that the Bundesbank has attached a higher relative weight to inflation than the Fed. Moreover, for the Bundesbank and the Fed the differences between both regimes seem to originate mainly from different preferences for interest rate smoothing. In contrast, the Bank of England shows a markedly reduced weight on output in the later part of the sample period. All countries experience transitions from relatively volatile regimes at the beginning of the sample period to more stable regimes thereafter. We did not attempt to explore if the reduced volatility in interest rates, inflation and the output gap was caused by a lower frequency of external shocks, a more stable economic structure, or a better design of monetary policy (see, e.g. Ahmed et al., 2002). Here, we consider the economic structure as constant and investigate how CB's preferences have changed to explain the observed reaction function. This is in line with papers finding support for a change in CB policy (Clarida et al., 1998 and Orphanides, 2004). While the hypothesis of a change in policy preferences is confirmed in the case of the Bank of England, differences in policy attitudes are less pronounced in the case of the Fed and the Bundesbank, which indicates that the improved macroeconomic performance might to some extent be due to smaller external shocks or a better coordination of monetary and fiscal policy.