محاسبه توابع واکنش سیاست پولی برای اقتصادهای بازار نوظهور: مورد برزیل
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|16452||2011||9 صفحه PDF||سفارش دهید||7894 کلمه|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Economic Modelling, Volume 28, Issue 4, July 2011, Pages 1730–1738
The paper investigates monetary policy in Brazil following a shift to a floating exchange rate alongside inflation targeting adoption. The benchmark reaction function reveals that the Central Bank behaves according to the Taylor principle by raising the overnight Selic policy interest rate more than the amount by which expected inflation exceeds the target. The investigation also considers a data-rich environment via an excess policy response containing information from a panel of 45 economic time series. The excess policy response carries a positive and significant coefficient in the reaction function including only an inflation gap variable.
Understanding monetary authorities' behaviour is fundamental and often appears to be a forlorn task. In reality, central bankers consider a significant amount of statistical information when deliberating on monetary policy's future course of actions, but empirical studies about central bank performance generally limit their scope to analysing a small number of variables (e.g., Taylor, 1993). The practice is gradually changing as research focusing on advanced economies incorporate a data-rich environment in modelling monetary policy (e.g., Bernanke and Boivin, 2003, Favero et al., 2005, Giannone et al., 2005 and Shibamoto, 2008). In contrast, there is a paucity of investigations on emerging market economies considering that critical aspect (see Frankel, 2010). The paper aims at advancing our understanding of monetary policy in emerging market economies by investigating Brazil. The investigation addresses the following questions: Are Taylor-type reaction functions practical for understanding how monetary authorities in Brazil behave following inflation targeting adoption and switching to a floating exchange rate regime? Can considering a data-rich environment lead to a better knowledge about the Central Bank of Brazil's conduct of policy? Brazil is an interesting case to study, inter alia, because it has made considerable efforts to build credibility on the sustainability of its economic policies (e.g., Giavazzi et al., 2005). Brazil is an early example of a country negotiating the perils of implementing a fully-fledged inflation targeting regime and concomitantly administering an IMF-backed stabilisation programme (Fraga et al., 2004).1 With reference to Brazil's monetary policy decisions in the crucial years following inflation targeting adoption in 1999, Mishkin (2004, page 20) notes that “The procedure followed by the Central Bank do Brazil was a textbook case for central bank response to shocks in emerging market countries.” Therefore the findings in the paper are potentially relevant for Brazil and other economies fostering policymaking institutions and consolidating macroeconomic stability. The literature on empirical monetary policy reaction functions mostly springs from Taylor's (1993) contribution to policy evaluation (e.g., Clarida et al., 1998). Research findings on emerging market economies generally highlight the potential usefulness of interest rate reaction functions in reaching a systematic understanding about monetary policy behaviour.2 Along this line of enquiry Minella et al. (2003) estimate Taylor-type reaction functions for Brazil and show that, in addition to the inflation and output gaps, the exchange rate enters significantly (e.g., Taylor, 2001). Minella et al.'s evidence shows that the Central Bank of Brazil reacts in line with the Taylor Principle and that the monetary authorities practise instrument smoothing (e.g., English et al., 2003).3 Focusing on Brazil and three other Latin American countries, de Mello et al. (2009) explore the relevance of unit roots, co-integration, and non-linearities when estimating monetary policy reaction functions. For Brazil their exercises reveal a positive coefficient on deviations of inflation from target and on the exchange rate depreciation term, but their reaction functions do not contain an output gap measure. The paper contributes by specifying and estimating alternative dynamic monetary policy reaction functions for Brazil.4 Estimating a battery of specifications applying automatic econometric model selection techniques helps in addressing the sensitivity of the relationship connecting developments in the economy and subsequent policy actions. The research strategy involves running regressions based on a benchmark Taylor-type reaction function linking monetary policy's interest rate instrument to developments in inflation, output, and the exchange rate. The exercises also investigate a reaction function only considering inflation, and equations incorporating a data-rich environment using a factor synthesising information from a panel of 45 economic time series, in the spirit of Bernanke and Boivin (2003). The investigation discusses the outcome from the battery of models, addressing the role of the exchange rate in estimated monetary policy reaction functions, the use of alternative inflation and output gap measures, and the magnitude of policy responses to the gaps. The paper finds that the Central Bank adjusts the Selic interest rate in line with the Taylor principle but that it does not systematically react to exchange rate developments. The investigation also explores the relevance of considering a data-rich environment via an ‘excess policy response’ (EPR). The exercises show that the EPR carries a positive and significant coefficient in the reaction function only including an inflation gap. When including an output gap term, the EPR is not statistically significant. The paper proceeds as follows. Section 2 overviews monetary policy and inflation targeting in Brazil. Section 3 explains monetary policy reaction functions and why considering a data-rich environment is relevant. Section 4 describes the time series data feeding the empirical modelling. Section 5 lays out the econometric specification, and discusses the results from estimating benchmark and alternative monetary policy reaction functions. Section 6 contains concluding remarks.
نتیجه گیری انگلیسی
The paper estimates alternative dynamic monetary policy reaction function specifications for Brazil during the floating exchange rate regime period following inflation targeting adoption in 1999. The benchmark, automatically reduced, model shows that the Central Bank adjusts the Selic interest rate in line with the Taylor principle but that it does not systematically react to exchange rate developments. The investigation also explores the relevance of considering a data-rich environment via an ‘excess policy response’ (EPR). The exercises show that the EPR is positive and significant in a reaction function including only an inflation gap, but it is not statistically significant in the benchmark reaction function incorporating inflation and output gaps. So the output gap appears to be sufficient for capturing inflationary pressure in Brazil, at least during the time period under investigation. In concluding, it is worth noting that central bankers in Brazil and elsewhere could benefit from using the approach explored in the paper and alternative means for incorporating the rich array of statistics available to them when deliberating about policy actions. Bernanke and Boivin (2003) also investigate what they label an ‘expert system for monetary policymaking’. Such a mechanical system (not intended to replace human input) would help in using real-time data on the economy to feed forward-looking monetary policy reaction functions that subsequently inform decision-making. There is also an emerging literature estimating dynamic stochastic general equilibrium models considering a data-rich context (e.g., Boivin and Giannoni, 2006). Other authors explore additional ways of incorporating rich data sets in understanding monetary policy. For instance, Giannone et al. (2008) develop a method for extracting information from large time series data sets released at different times and with diverse lags that can be used in nowcasting, e.g. in assessing current quarter GDP growth to inform monetary policy decisions.