Academic journal article The Journal of Developing Areas

Is Monetary Policy Non-Linear in Latin America? a Quantile Regression Approach to Brazil, Chile, Mexico and Peru

Academic journal article The Journal of Developing Areas

Is Monetary Policy Non-Linear in Latin America? a Quantile Regression Approach to Brazil, Chile, Mexico and Peru

Article excerpt

(ProQuest: ... denotes formulae omitted.)


Many developing countries have struggled with high and persistent inflation for much of their post-WWII histories. This is particularly true in Latin America where, for example, annual consumer-price inflation averaged 27% from 1964 to 1998 (Loungani and Swagel, 2003). Over the last few decades, however, inflation for many developing countries has come down-indeed, the hyperinflation that ravaged Latin America now is a thing of the past. Given the importance of price stability, and the battles that these countries have fought in achieving low inflation, it is no surprise that a considerable body of research has developed which studies how monetary policy has been conducted in emerging markets (see, for example, Lin and Ye, 2009, Goncalves and Salles, 2008, Moura and de Carvalho, 2010).

One approach used in the literature to examine central bank behavior, and the approach used in this paper, is to estimate the Taylor Rule. This rule models the shortterm interest rate as a function of inflation and the output gap, and thus the rule can be used to measure the tightness of monetary policy (Taylor, 1993). For instance, Clarida, Gali and Gertler (2000) estimate the Taylor Rule for the United States and find that the impact of inflation on the interest rate was larger after 1979 than before, suggesting a "tougher" Federal Reserve policy after the appointment of Paul Volcker than in the 1970s. There is no a priori reason why the Taylor Rule should be estimated linearly. It is quite plausible that central bankers will be more aggressive with their interest rate response to greater inflation when inflation is already high than when inflation is low. Indeed, recent empirical research finds evidence of nonlinear policy by some central banks. For example, Cukierman and Muscatelli (2008) find for the United States that interest rates are raised more than proportionately when inflation rises from a high level than from a low starting point. In a paper similar in focus with ours, Moura and de Carvalho (2010) estimate the Taylor Rule for a sample of Latin American countries and find some tentative evidence policy is nonlinear, but the authors note that their findings are constrained by a relative lack of data.

Data availability is, of course, a perennial challenge for many studies involving developing countries. This is especially true when estimating the Taylor Rule because, unlike the US or other G-7 nations, inflation in some developing countries was quite high even into the 1990s, and thus there are fewer observations available than for industrialized countries. In addition, investigating an asymmetric policy response has typically meant adding dummies for different regimes, or the use of some sort of switching model, which requires more parameters than linear estimation, further exacerbating the degree of freedom problem.

Given the difficulties of detecting nonlinearity with a relatively short span of data, we employ an alternative technique, called quantile regression, to investigate asymmetric monetary policy in four Latin American countries. Unlike the conditional mean estimator, quantile regression allows for the use of the entire sample to obtain estimates of the impact of regressors-such as inflation-on different quantiles of the dependent variable-the interest rate. Chevapatrukul, Kim and Mizen (2009) employ quantile regression to Taylor Rules for Japan and the United States. The authors find that, in both countries, at higher quantiles of interest rates, when inflation is presumably higher than desired, inflation has a palpably larger effect than at lower quantiles of interest rates (where inflation is likely relatively low). These results are consistent with the findings of Cukierman and Muscatelli-US policy responds in a nonlinear fashion, with higher levels of inflation eliciting a more than proportional response. Put simply, this is a policy response that a linear conditional mean model could not detect. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed


An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.