Are Flexible Exchange Rate Regimes more Volatile? Panel GARCH Evidence for the G7 and Latin America

• DOI: http://doi.org/10.1111/rode.12143
• Published date: 01 May 2015
• Date: 01 May 2015
• Author: María Eugenia Sanin,Rodolfo Cermeño

Are Flexible Exchange Rate Regimes more

Volatile? Panel GARCH Evidence for the G7 and

Latin America

Rodolfo Cermeño and María Eugenia Sanin*

Abstract

This paper investigates empirically the relationship between exchange rate (ER) regimes and volatility of

real exchange rate depreciation (RERD), comparing the G7 and 17 Latin American (LA17) countries,

during 1970–2010. We estimate a panel autoregressive model with generalized autoregressive conditional

heteroskedasticity (GARCH) errors and regime-specific effects on both the conditional mean and condi-

tional variance. For the G7, we find that, relative to the fixed ER regime, only the freely floating regime

shows higher RERD volatility; under the managed floating regime the RERD is equally volatile and under

the crawling peg it is actually less volatile. Instead, in the case of the LA17, more flexible ER regimes are

associated with more volatile RERD rates, with higher volatility under the managed floating regime than

under the crawling peg and with extremely high volatility under the “freely falling” ER regime.

1. Introduction

In today’s globalized economy understanding the dynamics of the real exchange rate

(RER) and its changes as well as its associated volatility is of great relevance since

they may have important real economic effects. In particular, currency volatility may

itself act as a barrier to international trade as shown by Obstfeld and Rogoff (2000).

Moreover, there is considerable empirical literature that shows that RER volatility

affects not only trade but also consumption and output growth.1This paper studies

the relationship between exchange rate (ER) regimes and volatility of real exchange

rate depreciation (RERD) in the G7 and 17 Latin American (LA17) countries.

Although the number of empirical papers on ER regimes and RER volatility is size-

able, most efforts are focused on flexible ER regimes. Only a few papers study the dif-

ferences in RER volatility across regimes.

This study attempts to contribute to the related empirical literature in two impor-

tant ways. First, it uses the typology of ER arrangements proposed by Reinhart and

Rogoff (2004), updated by Ilzetzki et al. (2011), which provides an accurate charac-

terization of de facto ER regimes. Second, volatility is explicitly modeled as a time

dependent variance process within a panel data framework. Specifically, we consider

a panel autoregressive (AR) model with generalized autoregressive conditional

* Cermeño: Centro de Investigación y Docencia Económicas (CIDE), División de Economía, Carretera

México-Toluca 3655, Lomas de Santa Fe, Delegación Álvaro Obregón, México D.F. 01210. Tel: +52-

(55)57279806; Fax: +57279878; E-mail:rodolfo.cermeno@cide.edu. Sanin: EPEE, Université d’Evry Val

d’Essonne and Ecole Polytechnique, Paris, France. The authors would like to thank participants at the

DEGIT XVIII Conference, Lima, Perú, for their useful comments. Special thanks are due to Paula

Hernández, for her numerous and valuable comments. Cermeño acknowledges the Department of Eco-

nomics of the Pontificia Universidad Católica del Perú for the excellent hosting of his sabbatical stance and

the Consejo Nacional de Ciencia y Tecnología (CONACYT, México) for providing financial support. The

authors are responsible for all remaining errors.

Review of Development Economics, 19(2), 297–308, 2015

DOI:10.1111/rode.12143

© 2015 John Wiley & Sons Ltd