How does monetary policy change? evidence on inflation targeting countries

How Does Monetary Policy Change? Evidence on Inflation Targeting Countries Jaromír Baxa Roman Horváth Borek Vasícek 10.07 De p artament d'Economia A p licada Facultat d'Economia i Empresa Aquest document pertany al Departament d'Economia Aplicada. Data de publicació : Departament d'Economia Aplicada Edifici B Campus de Bellaterra 08193 Bellaterra Telèfon: (93) 581 1680 Fax:(93) 581 2292 E-mail: [email protected] http://www.ecap.uab.es Setembre 2010 1 How Does Monetary Policy Change? Evidence on Inflation Targeting Countries Jaromír Baxa* Institute of Economic Studies, Charles University, Prague and Institute of Information Theory and Automation, Academy of Sciences of the Czech Republic, Prague Roman Horváth Czech National Bank and Institute of Economic Studies, Charles University, Prague Bořek Vašíček Universitat Autonoma de Barcelona Abstract We examine the evolution of monetary policy rules in a group of inflation targeting countries (Australia, Canada, New Zealand, Sweden and the United Kingdom), applying a moment-based estimator in a time-varying parameter model with endogenous regressors. Using this novel flexible framework, our main findings are threefold. First, monetary policy rules change gradually, pointing to the importance of applying a time-varying estimation framework. Second, the interest rate smoothing parameter is much lower than typically reported by previous time-invariant estimates of policy rules. External factors matter for all countries, although the importance of the exchange rate diminishes after the adoption of inflation targeting. Third, the response of interest rates to inflation is particularly strong during periods when central bankers want to break a record of high inflation, such as in the UK or Australia at the beginning of the 1980s. Contrary to common wisdom, the response becomes less aggressive after the adoption of inflation targeting, suggesting a positive anchoring effect of this regime on inflation expectations. This result is supported by our finding that inflation persistence as well as the policy neutral rate typically decreased after the adoption of inflation targeting. JEL Classification: E43, E52, E58. Keywords: Taylor rule, inflation targeting, monetary policy, time-varying parameter model, endogenous regressors. * We thank Aleš Bulíř, Peter Claeys, Øyvind Eitrheim, Michal Franta, Dana Hájková, Gabriel Perez-Quiros, Ekkehart Schlicht, Pierre Siklos, Miloslav Vošvrda and the seminar participants at the 7 th Norges Annual Monetary Policy Conference, the Czech National Bank, the Institute of Information Theory and Automation (Academy of Sciences of the Czech Republic), the Universitat Autonoma de Barcelona and the XXXIVth Symposium of the Spanish Economic Association for helpful discussions. The views expressed in this paper are not necessarily those of the Czech National Bank. Support from the Czech Science Foundation under grant 402/09/H045 and support from the Grant Agency of Charles University (GAUK) under project 46108 are gratefully acknowledged. Emails: [email protected], [email protected], [email protected]. 2 1 Introduction Taylor-type regressions have been applied extensively in order to describe monetary policy setting for many countries. The research on U.S. monetary policy usually assumes that monetary policy was subject to structural breaks when the Fed chairman changed. Clarida et al. (2000) claims that the U.S. inflation during the 1970s was unleashed because the Fed’s interest rate response to the inflation upsurge was too weak, while the increase of such response in the 1980s was behind the inflation moderation. Although there is ongoing discussion on the sources of this Great Moderation (Benati and Surico, 2009), it is generally accepted that monetary policy setting evolves over time. The evolution of monetary policy setting as well as exogenous changes in the economic system over time raises several issues for empirical analysis. In particular, the coefficients of monetary policy rules estimated over longer periods are structurally unstable. The solution used in the literature is typically sub-sample analysis (Clarida et al., 1998, 2000). Such an approach is based on the rather strong assumption that the timing of structural breaks is known, but also that policy setting does not evolve within each sub-period. Consequently, this gives impetus to applying an empirical framework that allows for regime changes or, in other words, time variance in the model parameters (Cogley and Sargent, 2001, 2005). Countries that have implemented the inflation targeting (IT) regime are especially suitable for such analysis because it is likely that the monetary policy stance with respect to inflation and other macroeconomic variables changed as a consequence of the implementation of IT. Moreover, there is ongoing debate of to what extent IT represents a rule-based policy. Bernanke et al. (1999) claim that IT is a framework or constrained discretion rather than a mechanical rule. Consequently, the monetary policy rule of an IT central bank is likely to be time varying. Our study aims to investigate the evolution of monetary policy for countries that have long experience with the IT regime. In particular, we analyze the time-varying monetary policy rules for Australia, Canada, New Zealand, Sweden and the United Kingdom. As we are interested in monetary policy evolution over a relatively long period, we do not consider countries where IT has been in place for a relatively short time (Finland, Spain), or was introduced relatively recently (such as Armenia, the Czech Republic, Hungary, Korea, Norway and South Africa). We apply the recently developed time-varying parameter model with endogenous regressors (Kim and Nelson, 2006), as this technique allows us to evaluate changes in policy rules over time, and, unlike Markov-switching methods, does not impose sudden policy switches between different regimes. 3 On top of that, it also deals with endogeneity of policy rules. Unlike Kim and Nelson (2006) we do not rely on the Kalman filter, which is conventionally employed to estimate time-varying models, but employ the moment-based estimator proposed by Schlicht and Ludsteck (2006) 1 for its mathematical and descriptive transparency and minimal requirements as regards initial conditions. In addition, Kim and Nelson (2006) apply their estimator to evaluate changes in U.S. monetary policy, while we focus on inflation targeting economies. Anticipating our results, we find that monetary policy changes gradually, pointing to the importance of applying a time-varying estimation framework (see also Koop et al., 2009, on evidence that monetary policy changes gradually rather than abruptly). When the issue of endogeneity in time-varying monetary policy rules is neglected, the parameters are estimated inconsistently, even though the resulting errors are economically not large. Second, the interest rate smoothing parameter is much lower than typically reported by previous time-invariant estimates of policy rules. This is in line with a recent critique by Rudebusch (2006), who emphasizes that the degree of smoothing is rather low. External factors matter for understanding the interest rate setting process for all countries, although the importance of the exchange rate diminishes after the adoption of inflation targeting. Third, the response of interest rates to inflation is particularly strong during periods when central bankers want to break a record of high inflation, such as in the UK at the beginning of the 1980s. Contrary to common wisdom, the response can become less aggressive after the adoption of inflation targeting, suggesting a positive anchoring effect of this regime on inflation expectations or a low inflation environment. This result is consistent with Kuttner and Posen (1999) and Sekine and Teranishi (2008), who show that inflation targeting can be associated with a smaller response of the interest rate to inflation developments if the previous inflation record was favorable. The paper is organized as follows. Section 2 discusses the related literature. Section 3 describes our data and empirical methodology. Section 4 presents the results. Section 5 concludes. An appendix with a detailed description of the methodology and additional results follows. 1 The description of this estimator is also available in Schlicht (2005), but we refer to the more recent working paper version, where this estimator is described in a great detail. Several important parts of this framework were introduced already in Schlicht (1981). 4 2 Related Literature 2.1 Monetary policy rules and inflation targeting Although the theoretical literature on optimal monetary policy usually distinguishes between instrument rules (the Taylor rule) and targeting rules (the inflation-targeting based rule), the forward-looking specification of the Taylor rule, sometimes augmented with other variables, has commonly been used for the analysis of decision making of IT central banks. The existing studies feature great diversity of empirical frameworks, which makes the comparison of their results sometimes complicated. In the following we provide a selective survey of empirical studies aimed at the countries that we focus on. The United Kingdom adopted IT in 1992 (currently a 2% target and a ±1% tolerance band) and the policy of the Bank of England (BoE) is subject to the most extensive empirical research. Clarida et al. (1998) analyzed the monetary policy setting of the BoE in the pre-IT period, concluding that it was consistent with the Taylor rule, yet additionally constrained by foreign (German) interest rate setting. Adam et al. (2005) find by means of sub-sample analysis that the introduction of IT did not represent a major change in monetary policy conduct, unlike the granting of instrument independence in 1997. Davradakis and Taylor (2006) point to significant asymmetry of British monetary policy during the IT period; in particular the BoE was concerned with inflation only when it significantly exceeded its target. Assenmacher-Wesche (2006) concludes by means of a Markov-switching model that no attention was paid to inflation until IT was adopted. Conversely, Kishor (2008) finds that the response to inflation had already increased, especially after Margaret Thatcher became prime minister (in 1979). Finally, Trecroci and Vassalli (2009) use a model with time-varying coefficients and conclude that policy had been getting gradually more inflation averse since the early 1980s. New Zealand was the first country to adopt IT (in 1990). A particular feature besides the announcement of the inflation target (currently a band of 1–3%) is that the governor of the Reserve Bank (RBNZ) has an explicit agreement with the government. Huang et al. (2001) study the monetary policy rule over the first decade of IT. He finds that the policy of the RBNZ was clearly aimed at the inflation target and did not respond to output fluctuations explicitly. The response to inflation was symmetric and a backward-looking rule does as good a job as a forward-looking one at tracking the interest rate dynamics. Plantier and Scrimgeour (2002) allow for the possibility that the neutral real interest rate (implicitly assumed in the Taylor rule to be constant) changes in time. In this framework they find that the response to inflation increased 5 after IT was implemented and the policy neutral interest rate tailed away. Ftiti (2008) additionally confirms that the RBNZ did not explicitly respond to exchange rate fluctuations and Karedekikli and Lees (2007) disregard asymmetries in the RBNZ policy rule. The Reserve Bank of Australia (RBA) turned to IT in 1993 (with a target of 2–3%) after decades of exchange rate pegs (till 1984) and consecutive monetary targeting. 2 De Brouwer and Gilbert (2005) using sub-sample analysis confirm that the RBA’s consideration of inflation was very low in the pre-IT period and a concern for output stabilization was clearly predominant. The response to inflation (both actual and expected) increased substantially after IT adoption but the RBA seemed to consider exchange rate and foreign interest rate developments as well. Leu and Sheen (2006) find a lot of discretionality in the RBA’s policy (a low fit of the time-invariant rule) in the pre-IT period, a consistent response to inflation during IT, and signs of asymmetry in both periods. Karedekikli and Lees (2007) document that the policy asymmetry is related to the RBA’s distaste for negative output gaps. The Bank of Canada (BoC) introduced IT in 1991 in the form of a series of targets for reducing inflation to the midpoint of the range of 1–3% by the end of 1995 (since then the target has remained unchanged). Demers and Rodríguez (2002) find that the implementation of this framework was distinguished by a higher inflation response, but the increase in the response to real economic activity was even more significant. Shih and Giles (2009) model the duration analysis of BoC interest rate changes with respect to different macroeconomic variables. They find that annual core inflation and the monthly growth rate of real GDP drive the changes of the policy rate, while the unemployment rate and the exchange rate do not. On the contrary, Dong (2008) confirms that the BoC considers real exchange rate movements. Sweden adopted IT in 1993 (a 2% target with a tolerance band of 1 percentage point) just after the krona had been allowed to float. The independence of Sveriges Riksbank (SR) was legally increased in 1999. Jansson and Vredin (2003) studied its policy rule, concluding that the inflation forecast (published by the Riksbank) is the only relevant variable driving interest rate changes. Kuttner (2004) additionally finds a role for the output gap, but in terms of its growth forecast (rather than its observed value). Berg et al. (2004) provide a rigorous analysis of the sources of deviations between the SR policy rate and the targets implied by diverse empirical rules. They 2 In Australia, the adoption of inflation targeting was a gradual process. As from January 1990, the RBA increased the frequency of its communications via speeches and the style of the Bank’s Bulletin started to correspond to the inflation reports as introduced in New Zealand. The exact inflation target was defined explicitly later, in April 1993. Greenville (1997) describes the policy changes in Australia in great detail. 6 claim that higher inflation forecasts at the early stages of the IT regime (due to a lack of credibility) generate a higher implied target from the forward-looking rule and therefore induce spurious indications of policy shocks. Their qualitative analysis of SR documents clarifies the rationale behind actual policy shocks, such as more gradualism (stronger inertia) in periods of macroeconomic uncertainty. Finally, there are a few multi-country studies. Meirelles Aurelio (2005) analyzes the time-invariant rules of the same countries as us, finding significant dependence of the results on real-time versus historical measures of variables. Lubik and Schorfheide (2007) estimate by Bayesian methods an open economy structural model of four IT countries (AUS, CAN, NZ, UK) with the aim of seeing whether IT central banks respond to exchange rate movements. They confirm this claim for the BoE and BoC. Dong (2008) enriches their setting by incorporating some more realistic assumptions (exchange rate endogeneity, incomplete exchange-rate pass-through), finding additionally a response to the exchange rate for the RBA. 2.2 Time variance in monetary policy rules The original empirical research on monetary policy rules used a linear specification with timeinvariant coefficients. Instrument variable estimators such as the GMM gained popularity in this context, because they are able to deal with the issue of endogeneity that arises in the forwardlooking specification (Clarida et al., 1998). 3 While a time-invariant policy rule may be a reasonable approximation when the analyzed period is short, structural stability usually fails over longer periods. The simplest empirical strategy for taking time variance into account is to use sub-sample analysis (Taylor, 1999; Clarida et al., 2000). The drawback of this approach is its rather subjective assumptions about points of structural change and structural stability within each sub-period. An alternative is to apply an econometric model that allows time variance for the coefficients. There are various methods dealing with time variance in the context of estimated monetary policy rules. The most common option is the Markov-switching VAR method, originally used for business cycle analysis. Valente (2003) employs such a model with switches in the constant term 3 One exception is when a researcher uses real-time central bank forecasts for Taylor-type rule estimation, i.e. the data available to the central bank before the monetary policy meeting. In such case, the endogeneity problem will not arise and least squares estimation may perform well (Orphanides, 2001). However, as we will discuss in more detail below, the use of real-time data may not solve the issue of endogeneity completely. 7 representing the evolution of the inflation target (the inflation target together with the real equilibrium interest rate makes the constant term in a simple Taylor rule). Assenmacher-Wesche (2006) uses the Markov-switching model with shifts both in the coefficients and in the residual variances. Such separation between the evolution of policy preferences (coefficients) and exogenous changes in the economic system (residuals) is important for the continuing discussion on the sources of the Great Moderation (Benati and Surico, 2008; Canova and Gambetti, 2008). Sims and Zha (2006) present a multivariate model with discrete breaks in both coefficients and disturbances. Unlike Assenmacher-Wesche they find that the variance of the shock rather than the time variance of the monetary policy rule coefficient has shaped macroeconomic developments in the U.S. in the last four decades. The application of Markov-switching VAR techniques turns out to be complicated for IT countries, where the policy rules are usually characterized as forward-looking and some regressors become endogenous. The endogeneity bias can be avoided by means of a backward-looking specification (lagged explanatory variables), but this is very probably inappropriate for IT central banks, which are arguably forward-looking. 4 However, there is another distinct feature of the Markov-switching model that makes its use for the analysis of time variance in the monetary policy rule rather questionable. The model assumes sudden switches from one policy regime to another rather than a gradual evolution of monetary policy. Although at first sight one may consider the introduction of IT to be an abrupt change, there are some reasons to believe that a smooth monetary policy transition is a more appropriate description for IT countries (Koop et al., 2009). Firstly, the IT regime is typically based on predictability and transparency, which does not seem to be consistent with sudden switches. Secondly, it is likely that inflation played a role in interest rate setting even before the IT regime was introduced, because in many countries a major decrease of inflation rates occurred before IT was implemented. Thirdly, the coefficients of different variables (such as inflation, the output gap or the exchange rate) in the monetary policy rule may evolve independently rather than moving from one regime to another at the same time (see also Darvas, 2009). For instance, a central bank may assign more weight to the observed or expected inflation rate when it implements IT, but that does not mean that it immediately disregards information on real economic activity or foreign interest rates. Finally, there is relevant evidence, though mostly for the U.S., that monetary policy evolves rather smoothly over time (Boivin, 2006; Canova and Gambetti, 2008; Koop et al., 2009). Therefore, based on this research, a smooth transition seems to be a more appropriate description of reality. In a similar manner, it 4 Psaradakis et al. (2006) proposed a solution to the endogeneity problem in the context of the Markov-switching model in the case of the term structure of interest rates. 14 that the Kalman filter is initialized with the correct initial conditions. Yet in this case, the VC estimator has a slightly lower mean squared error and this difference is more pronounced for small samples. 17 We assume that the variance of the disturbance term in Eq. (12) is not time-varying. Nevertheless, there is an ongoing discussion about to what extent changes in the macroeconomic environment are driven by changes in the variance of the disturbance term (i.e. exogenous changes in the economic system) vis-à-vis the variance in the coefficients of the monetary policy rule (see, for example, Benati and Surico, 2008, Canova and Gambetti, 2008, or Sims and Zha, 2006). One can also think about Eqs. (4), (10) and (11) in terms of the New Keynesian model, with Eqs. (10) and (11) representing the Phillips and IS curves. It should be noted that our framework is in general less restrictive and imposes less structure than the full-blown New Keynesian model. We expect t β to be positive, as the central bank is likely to react to an increase in expected inflation by increasing its policy rate. In particular, t β should be greater than one in the long-run solution of Eq. (4) if monetary policy is stabilizing. The development of t β over time may be driven by a number of factors, such as changes in monetary policy regime or institutional constraints (Adam et al., 2005). The effect of the adoption of inflation targeting on t β is ambiguous. As put forward by Kuttner and Posen (1999), t β can both increase and decrease. They show that under a conservative central bank the response of short-term interest rates is greater than under discretion or the optimal state-contingent rule (inflation targeting), 18 while the strength of the response under inflation targeting as compared to discretion depends on the credibility of the regime. Credible monetary policy does not have to react so strongly to inflation surprises, as inflation expectations are likely to remain anchored. Sekine and Teranishi (2008) provide a new Keynesian model that reaches to the same conclusions. Siklos and Weymark 17 For comparison, we estimated equation (12) using the conventional Kalman filter in the GROCER software using the function tvp (Dubois-Michaux, 2009). We parameterized the model with initial conditions taken from the OLS estimates of the parameters on the full sample and the initial forecast error covariance matrix set to 0. The matrix of the residuals of the time-varying coefficients is assumed to be diagonal as in the VC method. The results were very similar to those obtained from the VC method, with the estimated variances being the same in both methods. The only country where the estimated variance was different, was Sweden, with a lower variance in smoothing parameter ρ and higher a variance in β. Still, the results were consistent with ours. These results are available upon request. 18 See King (1997) on how inflation targeting allows one to come close to the optimal state-contingent rule. 15 (2009) estimate that inflation targeting in Australia, Canada and New Zealand reduced the magnitude of the interest rate changes to needed to maintain a low inflation environment. Similarly, ρ , a measure of interest rate smoothing, is expected to be positive with values between zero and one. Many time-invariant estimates of monetary policy rules find the value of this parameter to be about 0.7–0.9, implying a substantial degree of interest rate smoothing. Rudebusch (2006) claims that such figures are clearly overestimated in the face of very low interest rate forecastability in the term structure of interest rates. On the contrary, the timevarying model in principle enables some variables to affect interest rate setting in one period but not in another, and is less prone to autocorrelated shocks. Next, the effect of the output gap, t γ , on interest rates is expected to be positive or insignificant. In the first case, the central bank may have an explicit concern for real activity or understand the output gap as a useful predictor of future inflation. In the latter case, the insignificant coefficient may suggest that the central bank is primarily focused on inflation and does not consider the output gap to be important in delivering low inflation. There is a debate in literature about whether other variables should be included in the monetary policy rule. This is especially appealing for small open economies, which may be concerned with exchange rate fluctuations as well as the evolution of foreign interest rates. Taylor (2001) puts forward that even if the exchange rate or foreign interest rates are not explicitly included in the policy rule, they still remain present implicitly, as the exchange rate influences the inflation forecast, to which the inflation-targeting central bank is likely to react. It is also worth emphasizing that significance of the exchange rate or foreign interest rates does not necessarily mean that the central bank targets some particular values of these variables, but rather that the bank considers foreign developments to be important for its inflation forecast. On the other hand, empirical studies often favor the inclusion of these variables in the estimated policy rule. Having these considerations in mind, we decided to include the exchange rate and foreign interest rates, too, in order to assess whether these two variables carry any additional information for understanding the interest rate setting process in our sample countries. 3.2 The dataset We use quarterly data. The sample period varies from country to country owing to data availability (the UK 1975:1Q–2007:4Q, Australia 1975:1Q–2007:4Q, Canada 1975:1Q–2007:4Q, 16 New Zealand 1985:1Q–2007:3Q, Sweden 1982:2Q–2007:3Q), but on average the time coverage is about three decades. Following Clarida, Galí and Gertler (1998), the dependent variable is the short-term interest rate, which is typically closely linked to the monetary policy rate. The reason for choosing the shortterm interest rate rather than the monetary policy rate is the fact that the monetary policy rate and the change therein are censored (Podpiera, 2008). Therefore, the dependent variables capturing the policy rate are the discount rate (3-month Treasury bills) for the UK, the interbank 3-month interest rate for Australia, the 3-month Treasury bills rate for Canada, the overnight interbank 90-day interest rate for NZ and the interbank 3-month interest rate for Sweden. We choose the interest rate so as to be closely linked to monetary policy, but also to be available for a sufficiently long period. The foreign interest rate is the German 3-month Euribor for the UK and Sweden and the U.S. 3-month interbank interest rate for Australia, NZ, and Canada. The inflation is measured as the year-on-year change in the CPI, except for the UK, where we use the RPIX (the retail price index excluding mortgage interest payments), and the NZ, where we use the CPIX (the CPI without interest payments). 19 The output gap is taken as reported in the OECD Economic Outlook (the production function method based on the NAWRU – the non-accelerating wages rate of unemployment), except for NZ, where this series is short and where we use the output gap derived from the HodrickPrescott filter applied to the GDP series (constant prices, seasonally adjusted). 20 The exchange rate is measured by the chain-linked nominal effective exchange rate (NEER), except for Canada, where we use the bilateral USD/CAD exchange rate. For the regressions we use the deviation of the index from the HP trend (first differences were used for a robustness check) – see also Lubik and Schorfheide (2007). 19 We use year-on-year data, as the inflation target is also defined on a year-on-year basis. 20 There is no agreement on what is the best method for extraction of the unobserved output gap (Billmeier, 2009). We prefer to use the output gaps obtained by the OECD by means of the production function approach because they are based on a substantially richer information set than simple statistical detrending. Somewhat surprisingly, the OECD output gap and the simple HP gap evolve very closely. 17 4 Results This section presents the country-specific estimates of the time-varying monetary policy rules in sub-sections 4.1–4.5. Sub-section 4.6 contains the estimates of time-varying inflation persistence and sub-section 4.7 provides a summary of the main policy-relevant findings. 4.1 United Kingdom Our results show that the BoE significantly increased its response to inflation from the late 1970s till the mid-1980s. This overlaps with the Thatcher government and its major priority of inflation control. The overall decline of the response since 1985 can be related to the dismissal of the medium-term financial strategy (adopted in 1979). We find that the response of interest rates to inflation was gradually decreasing during the 1990s in spite of the introduction of IT. Although this finding may seem at first sight counterintuitive, it is important to keep in mind that, unlike in some emerging countries, IT was not implemented in the UK as a strong anti-inflation strategy. Inflation had already been contained in the 1980s and the very benign inflation environment was also supported by declining prices of raw materials on world markets. This corroborates with the findings of Kuttner and Posen (1999) and Sekine and Teranishi (2008), who show that inflation targeting can be associated with less aggressive monetary policy. The effect of the output gap is estimated as positive (although the confidence intervals are rather large, probably reflecting the fact that the gap is an unobserved variable and calculated ex post) and does not vary substantially over time. The interest rate smoothing parameter is found to have values between 0.1 and 0.3, which is much lower than typically reported by time-invariant estimates of monetary policy rules (Clarida et al., 1998, 2000). Our estimates seem to be reasonable in the face of the recent critique by Rudebusch (2006). Finally, in this basic model the intercept can be interpreted as the policy neutral (nominal) interest rate. We can see that it has steadily declined over time, which is consistent with the low inflation environment that prevailed in the UK in the 1990s. 18 Figure 1 – Time-varying response coefficients in baseline (closed economy) policy rule, UK -.4 -.2 .0 .2 .4 .6 .8 1980 1985 1990 1995 2000 2005 RHO_UB RHO_LB RHO Note: 95% confidence bands; model with bias correction terms, i.e. dealing with endogeneity in monetary policy rules. The upper-left graph depicts the evolution of the neutral rate. The upper-right graph depicts the evolution of the response of interest rates to inflation. The lower-left graph depicts the evolution of the response of interest rates to the output gap. The lower-right graph depicts the evolution of the interest rate smoothing parameter. The results of our augmented model show that the monetary policy of the BoE was influenced by external factors, although their importance was greater in the 1980s than recently. 21 In particular, we find evidence that the BoE decreased its policy rate as the nominal effective exchange rate (NEER) strengthened during the 1980s even before the pound officially joined the ERM (1990). Yet, once the UK abandoned the ERM and introduced IT, the BoE no longer seemed to react explicitly to the exchange rate. Obviously, it considered the exchange rate indirectly, as exchange rate fluctuations influence the inflation forecast (for more on this see Taylor, 2001). The same reasoning applies to the response to the foreign interest rate (Euribor). 21 In what follows, we present the evolution of the coefficients for the response to the exchange rate and foreign interest rates; the other coefficients remain largely unchanged and are not reported for the sake of brevity. -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1980 1985 1990 1995 2000 2005 GAMMA GAMMA_UB GAMA_LB -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 1980 1985 1990 1995 2000 2005 BETA BETA_LB BETA_UB 0 2 4 6 8 10 12 1980 1985 1990 1995 2000 2005 ALPHA ALPHA_LB ALPHA_UB 19 It was particularly strong during the 1980s and subsequently its importance somewhat declined. Our results show little support for the hypothesis that the monetary policy of the BoE follows that of the ECB, as the estimated response of the coefficient declined and the confidence intervals widened after the launch of the euro. Figure 2 – Time-varying response coefficients in augmented (open economy) policy rule, UK Note: 95% confidence bands; model with bias correction terms, i.e. dealing with endogeneity in monetary policy rules. The left-hand graph depicts the evolution of the response of interest rate to the nominal effective exchange rate (the deviation from the HP trend). The right-hand graph depicts the evolution of the response of the interest rate to the foreign interest rate. There are two studies directly comparable with this paper. Kishor (2008) obtains results similar to ours in spite of using monthly data known to have slightly different dynamics. He finds that the anti-inflation stance peaked in the mid-1980s and tended to decline from then onwards in spite of the adoption of IT. Similarly, his finding that the response to the foreign interest rate significantly declined after the ERM crisis is complementary to our result that the BoE gave much less consideration to the evolution of the exchange rate (the NEER gap). Trecroci and Vassalli (2010), who, unlike Kishor (2008) and this paper, do not correct for endogeneity in the time-varying model, come to the opposite conclusion that the BoE’s response to inflation increased over time. Yet, some counterintuitive results of their study point to the possibility of endogeneity bias. First, the interest rate smoothing parameter takes on significantly negative values from 1980 till 1995. This would imply not only that policy was not inertial, but also that there was actually a negative correlation between the present and past interest rate, which is inconsistent even in the face of a simple visual inspection of the interest rate series. -0.4 0.0 0.4 0.8 1.2 1.6 2.0 1980 1985 1990 1995 2000 2005 UBDELTA LBDELTA DELTA -.5 -.4 -.3 -.2 -.1 .0 .1 .2 .3 1980 1985 1990 1995 2000 2005 LB_NEER UB_NEER A_NEER 20 When we estimate our model without the endogeneity-correcting coefficients we obtain a similar result (see Appendix, Figure A.1). 22 Second, their coefficient for the foreign (German) interest rate peaks in 1990 and is de facto invariant since then, which the authors interpret as implicit exchange-rate targeting. This finding is doubtful given the pound’s exit from the ERM and the implementation of IT from 1992 onwards. In fact, British and German short-term rates, which were almost at par in 1992, diverged and the interbank interest rate in the UK exceeded the German one by almost 4% on the eve of euro adoption. 4.2 New Zealand New Zealand was the first country in the world to introduce inflation targeting, doing so by means of the Federal Bank Act signed in March 1990. 23 Our results indicate that the response of the RBNZ to expected inflation was very close to unity throughout the sample period (1985– 2007). In fact, it is clearly visible that the interest rate and inflation series move together very closely. However, in Figure 3 we can also see that the official introduction of IT does not seem to have engendered a significant change in interest rate setting (if anything there is very slight decrease of the response coefficient on inflation). Unlike in the UK, the response coefficient does not decrease substantially. This may be related to the fact that at the time IT was introduced in New Zealand the inflation rate was not far from double-digit values. Therefore, this policy was implemented in a different context than, say, in the UK, where single-digit inflation had already been achieved during the 1980s. 24 This result, together with the estimated insignificant response to the output gap, is consistent with the findings of time-invariant studies (Huang et al., 2001; Plantier and Scrimgeour, 2002) that the RBNZ applied a rather strict version of inflation targeting. Finally, we find that the interest rate smoothing parameter is again rather modest. 22 The bias correction terms are significant when we estimate the model addressing endogeneity, even though the economic significance is in general not large – see the Appendix. 23 Huang et al. (2001) argue that this policy was in effect since the end of 1988, when the RBNZ abandoned both monetary and exchange rate targeting. They also point to a specific feature of RBNZ monetary policy that could be referred to as “Open Mouth Operations”. Between 1989 and 1999 the RBNZ specified a 90-day bank bill rate consistent with price stability and threatened to use quantitative controls to achieve the desired market rate if it deviated from the target. Therefore, the RBNZ did not control this interest rate permanently and directly. 24 At the end of 1986 New Zealand introduced VAT, which had a direct impact on the inflation rate in the following two quarters. Consequently, we include a time dummy in Q1 and Q2 1987, whose coefficient is estimated as positive and significant. 21 Figure 3 – Time-varying response coefficients in baseline (closed economy) policy rule, New Zealand -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 86 88 90 92 94 96 98 00 02 04 06 RHO RHO_UB RHO_LB Note: 95% confidence bands; model with bias correction terms, i.e. dealing with endogeneity in monetary policy rules. The upper-left graph depicts the evolution of the neutral rate. The upper-right graph depicts the evolution of the response of interest rates to inflation. The lower-left graph depicts the evolution of the response of interest rates to the output gap. The lower-right graph depicts the evolution of the interest rate smoothing parameter. When we estimate the augmented model for New Zealand, the evidence for the exchange rate is not conclusive. We find a positive response to the NEER, which is rather counterintuitive in terms of the Taylor rule. However, the coefficient is significant only before the introduction of IT and its positive sign is probably related to currency appreciation following the interest rate increase. In that period the RBNZ aimed to keep the exchange rate within a predefined range and the interest rate was probably set so as to influence the exchange rate ex ante rather than ex post. Consistent with this finding, Ftiti (2008) in a time-invariant model rejects the hypothesis that the RBNZ responded to the exchange rate. On the other hand, we find some evidence in favor of -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 86 88 90 92 94 96 98 00 02 04 06 GAMMA UBGAMMA LBGAMMA 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 86 88 90 92 94 96 98 00 02 04 06 BETA UBBETA LBBETA -4 0 4 8 12 16 20 24 86 88 90 92 94 96 98 00 02 04 06 ALPHA LBALPHA UBALPHA 22 consideration of the foreign interest rate, although its response coefficient generally decreased after the launch of IT. Figure 4 – Time-varying response coefficients in augmented (open economy) policy rule, New Zealand Note: 95% confidence bands; model with bias correction terms, i.e. dealing with endogeneity in monetary policy rules. The left-hand graph depicts the evolution of the response of interest rate to the nominal effective exchange rate (the deviation from the HP trend). The right-hand graph depicts the evolution of the response of the interest rate to the foreign interest rate. 4.3 Australia Our results for Australia are available in Figures 5 and 6. The response of the interest rate to inflation is strongest in the 1980s, which is very similar to the UK experience. This period was characterized by inflation rates of around 10% and central bankers had to be quite aggressive in interest rate setting in order to break the record of high inflation deeply ingrained in public expectations. Neither monetary targeting (employed until 1984) nor the checklist approach (1985–1990) seemed to be successful in this regard. The fluctuation of the inflation response coefficient points to the discretionary nature of policy decisions (making this finding consistent with Leu and Sheen, 2006). The response coefficient peaks in 1990 on the eve of IT but declines after the adoption of this regime. It is again arguable whether it was the credibility of this regime that anchored inflation expectations and allowed the RBA to behave less aggressively. The original inflation decline may also have been related to the world recession in the early 1990s. Our results dispute the finding of De Brouwer and Gordon (2005) that the inflation response of the RBA increased as a result of the launch of inflation targeting. -0.5 0.0 0.5 1.0 1.5 2.0 2.5 86 88 90 92 94 96 98 00 02 04 06 UBDELTA LBDELTA DELTA -.2 -.1 .0 .1 .2 .3 .4 86 88 90 92 94 96 98 00 02 04 06 LB_NEER UB_NEER A_NEER 23 As for other countries, the neutral rate declines in the 1990s, reflecting the global low inflation environment. The output gap is not found to be significant and the estimated interest rate smoothing is again rather low. Figure 5 – Time-varying response coefficients in baseline (closed economy) policy rule, Australia -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1975 1980 1985 1990 1995 2000 2005 RHO RHO_LB RHO_UB Note: 95% confidence bands; model with bias correction terms, i.e. dealing with endogeneity in monetary policy rules. The upper-left graph depicts the evolution of the neutral rate. The upper-right graph depicts the evolution of the response of interest rates to inflation. The lower-left graph depicts the evolution of the response of interest rates to the output gap. The lower-right graph depicts the evolution of the interest rate smoothing parameter. We find that the exchange rate does not have a significant effect on the short-term interest rate (besides the NEER we use also the trade-weighted index – TWI, which is an exchange rate measure reported and often referred to by the RBA) except in the period of 1985–87, when the currency depreciation (the Australian dollar was allowed to float in 1983 after a period of a moving peg vis-à-vis the TWI) was offset by an interest rate increase so as to curb the inflation -.6 -.4 -.2 .0 .2 .4 .6 .8 1975 1980 1985 1990 1995 2000 2005 GAMMA UBGAMMA LBGAMMA -0.5 0.0 0.5 1.0 1.5 2.0 1975 1980 1985 1990 1995 2000 2005 BETA LBBETA UBBETA 0 4 8 12 16 20 1975 1980 1985 1990 1995 2000 2005 ALPHA LBALPHA UBALPHA 30 Australia Canada 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1975 1980 1985 1990 1995 2000 2005 INFLAG INFLAG_LB INFLAG_UB 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1975 1980 1985 1990 1995 2000 2005 INFLAG INFLAG_LB INFLAG_UB Sweden 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 82 84 86 88 90 92 94 96 98 00 02 04 06 INFLAG INFLAG_LB INFLAG_UB Note: 95% confidence bands; the coefficient of the AR(1) term (i.e. model 1 t t t t t π α ρ π ε − = + + ) for the inflation series employed in the previous analysis for each country. 4.7 Monetary Policy Rules – Wrap-Up of Main Policy Findings This sub-section summarizes the main policy-relevant findings of this paper. We focus on the following four issues: 1) monetary policy aggressiveness and the inflation rate, 2) monetary policy aggressiveness and inflation targeting, 3) interest rate smoothing and 4) inflation persistence and inflation targeting. 31 Figure 12 – Monetary Policy Aggressiveness and Inflation Targeting -0.5 0.0 0.5 1.0 1.5 2.0 2.5 0.7 0.8 0.9 1.0 1.1 1.2 1.3 AUS CAN NZ UK SWE IT Adoption Note: The y axis depicts the evolution of the estimated parameter ( β ) (the response of interest rates to expected inflation) and the x axis represents time, with the year of inflation targeting adoption denoted by a black vertical line. The value of β for New Zealand is plotted on the right-hand axis. Figure 12 presents the link between monetary policy aggressiveness (defined as the estimate of the response of interest rates to expected inflation) and inflation targeting. It can be seen that in no country did the aggressiveness parameter increase after the adoption of inflation targeting. In fact, this aggressiveness substantially decreased in the UK, Australia and Sweden. If we look at the link between aggressiveness and expected inflation within the IT period, the results suggest that in most countries the aggressiveness is higher the more expected inflation deviates from its target. This broadly corresponds to the findings of Davradakis and Taylor (2006), who document a non-linear policy rule for the UK of a similar pattern. Figure 13 documents a relatively strong link between the level of inflation and monetary policy aggressiveness for the UK, Australia and Sweden. This is because 1) the response of interest rates to inflation has been particularly strong during periods when central bankers want to break a record of high inflation and 2) the response has been less aggressive after the adoption of inflation targeting with well-anchored inflation expectations. In contrast, New Zealand and Canada do not seem to exhibit a link between aggressiveness and inflation. 32 Figure 13 – Monetary Policy Aggressiveness and Inflation Rate United Kingdom -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0 4 8 12 16 20 24 I n f l a t i o n Aggressiveness New Zealand 0.7 0.8 0.9 1.0 1.1 1.2 1.3 -1 0 1 2 3 4 5 6 7 8 I n f l a t i o n Aggressiveness Australia 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 -2 0 2 4 6 8 10 12 14 I n f l a t i o n Aggressiveness Canada 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 -2 0 2 4 6 8 10 12 14 I n f l a t i o n Aggressiveness Sweden 0.0 0.4 0.8 1.2 1.6 2.0 2.4 0 2 4 6 8 10 12 I n f l a t i o n Aggressiveness Note: The figure presents the scatter plots between the inflation rate and the response of interest rates to expected inflation ( β ), labeled as aggressiveness, for each country individually. The evolution of the estimated interest rate smoothing parameter in comparison to the timeinvariant estimates for the UK in 1979–1990 by Clarida et al. (2000) is available in Figure 14. Our time-varying estimates of interest rate smoothing are well below the time-invariant one, which seems reasonable in the light of the recent critique by Rudebusch (2006), who puts forward that 33 the degree of interest rate smoothing is actually low. While omitted variables or persistent shocks were deemed to be behind the implausible degree of policy inertia, our empirical results suggest that omission of the time-varying nature of the response coefficient may be another reason for the overestimation of smoothing coefficient ρ in time-invariant policy rules. Moreover, given that the policy rule is modeled as a partial adjustment (see Eq. (3)), the overestimation of smoothing coefficient ρ may drive the upper bias in all the other coefficients. This seems to explain why our long-term inflation multiplier β is substantially lower than found in timeinvariant studies. Though a high β is consistent with common wisdom, we suspect that it is a byproduct of the upper bias in ρ . Figure 14 – Interest Rate Smoothing -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 76 78 80 82 84 86 88 90 92 94 96 98 00 02 04 06 AUS CAN Clarida et al. NZ SWE UK Note: The figure presents the evolution of the estimated interest rate smoothing parameter ρ over time in comparison to the interest rate smoothing parameter estimated in the time-invariant model of Clarida et al. (2000) for the UK. 34 Figure 15 – Inflation Targeting and Inflation Persistence -0.8 -0.4 0.0 0.4 0.8 1.2 UK SWE AUS CAN NZ IT Adoption Note: The y axis depicts the evolution of the estimated inflation persistence parameter and the y axis represents time, with the year of inflation targeting adoption denoted by a black vertical line. Finally, the results in Figure 15 plot the estimates of inflation persistence over time for all countries with respect to the inflation targeting adoption date. The results suggest that inflation persistence decreased after the adoption of inflation targeting, with a very distinct fall in the UK and New Zealand. 5 Concluding Remarks In this paper, we shed light on the evolution of monetary policy in the main inflation targeting central banks during the last three decades. The evolution of monetary policy is evaluated within a novel framework of a time-varying parameter model with endogenous regressors (Kim and Nelson, 2006), further addressing small sample issues (Schlicht, 1981; Schlicht, 2005; Schlicht and Ludsteck, 2006). In our view, the results point to the usefulness of this econometric framework for analysis of the evolution of monetary policy setting. The estimation of standard monetary policy rules reveals that policy changes gradually and the changes coincide with several important institutional reforms as well as with the periods when the central banks successfully decreased double-digit inflation rates to rates consistent with their definitions of price stability. In this respect, our results suggest that the response of interest rates to inflation is particularly high during periods when central bankers want to break a record of high inflation, such as in the 35 UK in the early 1980s. Contrary to common wisdom, the response is often found to be less aggressive after the adoption of inflation targeting, suggesting a positive anchoring effect of this regime on inflation expectations. In other words, monetary policy need not be as aggressive as under a discretionary regime in order to achieve price stability (Kuttner and Posen, 1999). This result is supported by our finding that inflation becomes less inertial and the policy neutral rate decreases after the adoption of inflation targeting. We find that external factors matter for interest rate setting in all our sample countries. To be more precise, the foreign interest rate is found to enter the monetary policy rule significantly. The importance of the exchange rate varies, being apparently more important before the countries adopted inflation targeting than afterwards. Our results also indicate that interest rate smoothing is much lower than typically reported by time-invariant estimates of monetary policy rules (see, for example, Clarida et al., 1998, 2000). Our estimates support the recent critique by Rudebusch (2006), who argues that the degree of interest rate smoothing is rather low. We suggest that neglect of changes in monetary policy setting over time is the reason for the implausible degree of policy inertia previously found. Moreover, the fact that upper bias in the smoothing parameter affects all the estimates may explain some fundamental differences between our findings and those established in the literature, such as the size of the inflation response coefficient. In terms of future research we believe it would be worthwhile to apply this framework to better understand whether and how monetary policy reacts to periods of financial instability and which types of financial instability are the most worrying for central banks. In consequence, this would improve the understanding of both the interest rate setting process and the reaction of monetary policy makers to the current global financial crisis in a more systematic manner. 36 References Adam, Christopher; Cobham, David; Girardin, Eric, 2005. Monetary Frameworks and Institutional Constraints: UK Monetary Policy Reaction Functions, 1985–2003. Oxford Bulletin of Economics and Statistics 67, 497–516. Assenmacher-Wesche, Katrin, 2006. Estimating Central Banks’ Preferences from a Time-varying Empirical Reaction Function. European Economic Review 50, 1951–1974. Ball, Laurence, 1999b. Policy Rules for Open Economies, in J. Taylor (ed.): Monetary Policy Rules. University of Chicago Press for NBER, Cambridge, 127–144. Benati, Luca; Surico, Paolo, 2009. VAR Analysis and the Great Moderation. American Economic Review 99(4), 1636–1652. Benati, Luca, 2008. Investigating Inflation Persistence across Monetary Regimes. The Quarterly Journal of Economics 123, 1005–1060. Berg, Claes; Jansson, Per; Vredin, Andres, 2004. How Useful Are Simple Rules for Monetary Policy? The Swedish Experience. Sveriges Riksbank Working Paper Series, No. 169. Bernanke, Ben S.; Laubach, Thomas; Mishkin, Frederic; Posen, Adam S., 1999. Inflation Targeting: Lessons from the International Experience. Princeton University Press, Princeton. Billmeier, Andreas, 2009. Ghostbusting: Which Output Gap Really Matters? International Economics and Economic Policy 6(4), 319–419. Blinder, Alan, 2007. Monetary Policy by Committee: Why and How? European Journal of Political Economy 23, 106–123. Boivin, Jean, 2006. Has U.S. Monetary Policy Changed? Evidence from Drifting Coefficients and Real-Time Data. Journal of Money, Credit and Banking 38(5), 1149–1173. Canova, Fabio; Gambetti, Luca, 2008. Structural Changes in the US Economy: Is There a Role for Monetary Policy? Journal of Economic Dynamics & Control 33, 477–490. Clarida, Richard; Galí, Jordi; Gertler, Mark, 1998. Monetary Policy Rules in Practice: Some International Evidence. European Economic Review 42, 1033–1067. Clarida, Richard; Galí, Jordi; Gertler, Mark, 2000. Monetary Policy Rules and Macroeconomic Stability: Evidence and Some Theory. The Quarterly Journal of Economics 115, 147–180. Cogley, T.; Sargent, T., 2001. Evolving Post World War II U.S. Inflation. NBER Macroeconomics Annual, Vol. 16. Cogley, T.; Sargent, T., 2003. Drifts and Volatilities: Monetary Policies and Outcomes in the Post WWII U.S. Review of Economic Dynamics 8(2), 262–302. 37 Darvas, Zsolt, 2009. Monetary Transmission in Three Central European Countries: Evidence from Time-Varying Coefficient Vector Autoregressions. DNB Working Paper No. 209. Davradakis, Emmanuel; Taylor, Mark P., 2006. Interest Rate Setting and Inflation Targeting: Evidence of a Nonlinear Taylor Rule for the United Kingdom. Studies in Nonlinear Dynamics & Econometrics 10(4), Article 1. De Brouwer, Gordon; Gilbert, James, 2005. Monetary Policy Reaction Functions in Australia. The Economic Record 81(253), 124–134. Demers, Fréderick; Rodríguez, Gabriel, 2002. Estimation of the Taylor Rule for Canada Under Multiple Structural Changes. Working Paper No. 107E, University of Ottawa. Dolado, Juan José; María-Dolores, Ramon; Ruge-Murcia, Francisco, 2004. Nonlinear Monetary Policy Rules: Some New Evidence for the U.S. Studies in Nonlinear Dynamics & Econometrics 8(3), Article 2. Domenech, Rafael; Ledo, Mayte; Taguas, David, 2002. Some New Results on the Interest Rate Rules in EMU and in the U.S. Journal of Economics and Business 54, 431–446. Dong, Wei, 2008. Do Central Banks Respond to Exchange Rate Movements? Some New Evidence from Structural Estimation. Working Paper No. 24, Bank of Canada. Dubois, É.; Michaux, E., 2009. Grocer 1.4: an econometric toolbox for Scilab, available at http://dubois.ensae.net/grocer.html. Elkhoury, Marwan, 2006. A Time-Varying Parameter Model of a Monetary Policy Rule for Switzerland. Working Paper No. 01, HEI Geneva. Ftiti, Zied, 2008. Taylor Rule and Inflation Targeting: Evidence from New Zealand. International Business & Economics Research Journal 7, 131–150. Granger, Clive W. J., 2008. Non-Linear Models: Where Do We Go Next – Time-Varying Parameter Models? Studies in Nonlinear Dynamics & Econometrics 12(3), Article 1. Grenville, Stephen, 1997. The Evolution of Monetary Policy: From Money Targets to Inflation Targets, in P. W. Lowe (ed.), Monetary Policy and Inflation Targeting. Reserve Bank of Australia, Sydney, 125–158. Horváth, Roman, 2009. The Time-Varying Policy Neutral Rate in Real Time: A Predictor for Future Inflation? Economic Modelling 26(1), 71–81. Hondroyiannis, George; Swamy, P. A. V. B.; Tavlas, George, 2009. The New Keynesian Phillips Curve in a Time-Varying Coefficient Environment: Some European Evidence. Macroeconomic Dynamics 13, 149–166. Huang, Angela; Margaritis, Dimitri; Mayes, David, 2001. Monetary Policy Rules in Practice: Evidence from New Zealand. Multinational Finance Journal 5(3), 175–200. 38 Jansson, Per; Vredin, Anders, 2003. Forecast-Based Monetary Policy: The Case of Sweden. International Finance 6(3), 349–380. Karedekikli, Ozer; Lees, Kirdal, 2007. Do the Central Banks of Australia and New Zealand Behave Asymmetrically? Evidence from Monetary Policy Reaction Functions. The Economic Record 83(261), 131–142. Kim, Chang-Jin, 2006. Time-Varying Parameter Models with Endogenous Regressors. Economics Letters 91, 21–26. Kim, Chang-Jin; Nelson, Charles R., 2006. Estimation of a Forward-Looking Monetary Policy Rule: A Time-Varying Parameter Model Using Ex-post Data. Journal of Monetary Economics 53, 1949–1966. Kim, Chang-Jin; Kishor, Kundan; Nelson, Charles, 2006. A Time-Varying Parameter Model for a Forward-Looking Monetary Policy Rule Based on Real-Time Data, mimeo. King, Mervyn, 1997. Changes in UK Monetary Policy: Rules and Discretion in Practice. Journal of Monetary Economics 39, 81–87. Kishor, Kundan, 2008. International Evidence on Time Variation in a Forward-Looking Monetary Policy Rule, mimeo. Koop, Gary; Leon-Gonzalez, Roberto; Strachan, Rodney W. 2009. On the Evolution of the Monetary Policy Transmission Mechanism. Journal of Economic Dynamics and Control 33, 997–1017. Koopman, Siem J.; Shephard, Neil; Doornik, Jurgen A., 1999. Statistical Algorithms for Models in State Space Using SsfPack 2.2. Econometrics Journal 2(1), 107–160. Kuttner, Kenneth N., 2004. The Role of Policy Rules in Inflation Targeting. Federal Reserve Bank of St. Louis Review 86(4), 1–23. Kuttner, Kenneth N.; Posen, Adam S., 1999. Does Talk Matter After All? Inflation Targeting and Central Bank Behavior. Federal Reserve Bank of New York, Staff Reports 88 Leu, Shawn Chen-Yu; Sheen, Jeffrey, 2006. Asymmetric Monetary Policy in Australia. The Economic Record 82 (Special Issue), 85–96. Lubik, Thomas A.; Schorfheide, Frank, 2007. Do Central Banks Respond to Exchange Rate Movements? A Structural Investigation. Journal of Monetary Economics 54, 1069–1087. Meirelles Aureliano, Marcela, 2005. Do We Really Know How Inflation Targeters Set Interest Rates? Federal Reserve Bank of Kansas City, Research Working Paper No. 02. Orphanides, Athanasios, 2001. Monetary Policy Rules Based on Real-Time Data. American Economic Review 91, 964–985. 39 Plantier, L. Christopher; Scrimgeour, Dean, 2002. Estimating a Taylor Rule for New Zealand with a Time-Varying Neutral Real Rate. Reserve Bank of New Zealand, Discussion Paper No. 06. Podpiera, Jiri, 2008. The Role of Ad Hoc Factors in Policy Rate Settings. Economic Modelling 25(5), 1003–1010. Psaradakis, Zacharias; Sola, Martin; Spagnolo, Fabio, 2006. Instrumental-Variables Estimation in Markov Switching Models with Endogenous Explanatory Variables: An Application to the Term Structure of Interest Rates. Studies in Nonlinear Dynamics & Econometrics 10(2), Article 1. Rudebusch, Glenn, 2006. Monetary Policy Inertia: Fact or Fiction? International Journal of Central Banking 2(4), 85–136. Schlicht, Ekkehart, 1981. A Seasonal Adjustment Principle and a Seasonal Adjustment Method Derived From this Principle. Journal of the American Statistical Association 76(374), 374– 378. Schlicht, Ekkehart, 2005. Estimating the Smoothing Parameter in the So-called Hodrick-Prescott Filter. Journal of the Japan Statistical Society 35(1), 99–119. Schlicht, Ekkehart; Ludsteck, Johannes, 2006. Variance Estimation in a Random Coefficients Model. IZA Discussion Paper No. 2031. Siklos, Pierre; Weymark, Diana, 2009. Has Inflation Targeting Improved Monetary Policy? Evaluating Policy Effectiveness in Australia, Canada, and New Zealand. Vanderbilt University Working Paper 0906. Sekine, Toshitaka; Teranishi, Yuki, 2008. Inflation Targeting and Monetary Policy Activism. IMES Discussion Paper Series 2008-E-13, Bank of Japan. Shih, Rudy; Giles, David E., 2008. Modelling the Duration of Interest Rate Spells under Inflation Targeting in Canada. Applied Economics 41, 1229–1239. Sims, Christopher; Zha, Tao, 2006. Were There Regime Switches in US Monetary Policy. American Economic Review 96(1), 54–81. Taylor, John B., 1993. Discretion Versus Policy Rules in Practice. Carnegie-Rochester Conference Series on Public Policy 39, 195–214. Taylor, John B. (Ed.), 1999. Monetary Policy Rules. The University of Chicago Press: Chicago. Taylor, John, B., 2001. The Role of the Exchange Rate in Monetary-Policy Rules. American Economic Review 91(2), 263–267. Trecroci, Carmine; Vassalli, Matilde, 2010. Monetary Policy Regime Shifts: New Evidence from Time-Varying Interest-Rate Rules. Economic Inquiry, forthcoming. 46 Figure A.2 – Time-varying response coefficients in baseline (closed economy) policy rule, New Zealand -4 0 4 8 12 16 20 24 86 88 90 92 94 96 98 00 02 04 06 ALPHA ALPHA_LB ALPHA_NOEND ALPHA_UB 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 86 88 90 92 94 96 98 00 02 04 06 BETA BETA_LB BETA_NOEND BETA_UB -1.2 -0.8 -0.4 0.0 0.4 0.8 1.2 86 88 90 92 94 96 98 00 02 04 06 GAMMA GAMMA_LB GAMMA_NOEND GAMMA_UB -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 86 88 90 92 94 96 98 00 02 04 06 RHO RHO_LB RHO_NOEND RHO_UB Note: Model with bias correction terms (solid line); model not dealing with endogeneity in monetary policy rules (dashed line). The 95% confidence bands correspond to the model with bias correction terms. 47 Figure A.3 – Time-varying response coefficients in baseline (closed economy) policy rule, Australia 0 4 8 12 16 20 1975 1980 1985 1990 1995 2000 2005 ALPHA ALPHA_LB ALPHA_NOEND ALPHA_UB -0.5 0.0 0.5 1.0 1.5 2.0 1975 1980 1985 1990 1995 2000 2005 BETA BETA_LB BETA_NOEND BETA_UB -.6 -.4 -.2 .0 .2 .4 .6 .8 1975 1980 1985 1990 1995 2000 2005 GAMMA GAMMA_LB GAMMA_NOEND GAMMA_UB -0.4 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1975 1980 1985 1990 1995 2000 2005 RHO RHO_LB RHO_NOEND RHO_UB Note: Model with bias correction terms (solid line); model not dealing with endogeneity in monetary policy rules (dashed line). The 95% confidence bands correspond to the model with bias correction terms. 48 Figure A.4 – Time-varying response coefficients in baseline (closed economy) policy rule, Canada -4 0 4 8 12 16 1975 1980 1985 1990 1995 2000 2005 ALPHA ALPHA_LB ALPHA_NOEND ALPHA_UB -0.4 0.0 0.4 0.8 1.2 1.6 2.0 1975 1980 1985 1990 1995 2000 2005 BETA BETA_LB BETA_NOEND BETA_UB -.8 -.6 -.4 -.2 .0 .2 .4 .6 .8 1975 1980 1985 1990 1995 2000 2005 RHO RHO_LB RHO_NOEND RHO_UB 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1975 1980 1985 1990 1995 2000 2005 GAMMA GAMMA_LB GAMMA_NOEND GAMMA_UB Note: Model with bias correction terms (solid line); model not dealing with endogeneity in monetary policy rules (dashed line). The 95% confidence bands correspond to the model with bias correction terms. 49 Figure A.5 – Time-varying response coefficients in baseline (closed economy) policy rule, Sweden 0 2 4 6 8 10 12 82 84 86 88 90 92 94 96 98 00 02 04 06 ALPHA ALPHA_LB ALPHA_NOEND ALPHA_UB -1 0 1 2 3 4 82 84 86 88 90 92 94 96 98 00 02 04 06 BETA BETA_LB BETA_NOEND BETA_UB -0.8 -0.4 0.0 0.4 0.8 1.2 82 84 86 88 90 92 94 96 98 00 02 04 06 GAMMA GAMMA_LB GAMMA_NOEND GAMMA_UB -0.2 0.0 0.2 0.4 0.6 0.8 1.0 82 84 86 88 90 92 94 96 98 00 02 04 06 RHO RHO_LB RHO_NOEND RHO_UB Note: Model with bias correction terms (solid line); model not dealing with endogeneity in monetary policy rules (dashed line). The 95% confidence bands correspond to the model with bias correction terms. 50 Table A.1 Estimated Coefficients of Endogeneity Correction Terms UK NZ Aus Can Swe Inflation mean -0.0183 -1.06 -0.17 -0.873 -0.329 s.e. 0.269 0.337 0.0912 0.146 0.119 GDP gap mean 0.0745 -0.031 0.014 -0.127 0.0781 s.e. 0.125 0.27 0.092 0.0559 0.0808 Note: Bold: sign. at 5%, italic: sign. at 10%. TÍTOL NUM AUTOR DATA Setembre 2010 How Does Monetary Policy Change? Evidence on Inflation Targeting Countries 10.07 Jaromír Baxa Roman Horváth Borek Vasícek Juliol 2010The Wage-Productivity Gap Revisited: Is the Labour Share Neutral to Employment? 10.06 Marika Karanassou, Hector Sala Juliol 2010Oil price shocks and labor market fluctuations10.05 Javier Ordoñez, Hector Sala, Jose I. Silva Juliol 2010Vulnerability to Poverty: A Microeconometric Approach and Application to the Republic of Haiti 10.04 Evans Jadotte Maig 2010Nuevos y viejos criterios de rentabilidad que concuerdan con el criterio del Valor Actual Neto. 10.03 Joan Pasqual, Emilio Padilla Març 2010Memory in Contracts: The Experience of the EBRD (1991-2003) 10.02 Lionel Artige, Rosella Nicolini Febrer 2010Language knowledge and earnings in Catalonia10.01 Antonio Di Paolo, Josep Lluís Raymond-Bara Desembre 2009 Inflation dynamics and the New Keynesian Phillips curve in EU-4 09.12 Borek Vasicek Novembre 2009 Venezuelan Economic Laboratory The Case of the Altruistic Economy of Felipe Pérez Martí 09.11 Alejandro Agafonow Novembre 2009 Determinantes del crecimiento de las emisiones de gases de efecto invernadero en España (1990-2007) 09.10 Vicent Alcántara Escolano, Emilio Padilla Rosa Novembre 2009 Heterogeneity across Immigrants in the Spanish Labour Market: Advantage and Disadvantage 09.09 Catia Nicodemo Novembre 2009 A sensitivity analysis of poverty definitions09.08 Nicholas T. Longford, Catia Nicodemo Setembre 2009 Emissions distribution in postKyoto international negotiations: a policy perspective 09.07 Nicola Cantore, Emilio Padilla Juliol 2009Selection Bias and Unobservable Heterogeneity applied at the Wage Equation of European Married Women 09.06 Catia Nicodemo Maig 2009La desigualdad en las intensidades energéticas y la composición de la producción. Un análisis para los países de la OCDE 09.05 Juan Antonio Duro Moreno, Vicent Alcantara Escolano, Emilio Padilla Rosa