Empirics of currency crises: A duration analysis approach

• Date: 01 July 2019
• DOI: http://doi.org/10.1002/rfe.1056
• Published date: 01 July 2019

428

|

wileyonlinelibrary.com/journal/rfe Rev Financ Econ. 2019;37:428–449.

© 2019 University of New Orleans

1

|

INTRODUCTION

Currency crises have been a recurrent feature of the international economy from the invention of paper money. They are not

confined to particular economies or a specific region. They take place in developed, emerging, and developing countries and are

spread all over the globe. Some have scattered over time and some are clustered in points of time. They play an important role in

the world economy's turmoil. Countries that experience currency crises face economic losses that can be huge and disruptive.1

Yet, the exacted toll is not only financial and economic, but also human, social, and political.

In the recent decades, while the frequency of currency crises has increased2, the globalization process and the emergence

of integrated international financial markets have propagated domestic crises beyond the borders of individual countries.

Now, it is clear that a currency crisis is a real threat to financial stability and economic prosperity. As a result, studying

currency crises to find out what drives them and through which channels they spread is of great interest to policy makers,

academics, and market participants. Such studies should illustrate the mechanism of the crisis and forecast whether or not,

and when, an individual country might experience a currency crisis. Credible studies would help policy- makers to come up

with solutions for crisis prevention, crisis management, and crisis resolution.

The main objective of this paper is to analyze the determinants of currency crises in 20 OECD countries and South Africa

from 1970 through 1998. It systematically examines the role of economic fundamentals and contagion in the origins of currency

Received: 15 March 2018

|

Revised: 8 October 2018

|

Accepted: 20 November 2018

DOI: 10.1002/rfe.1056

ORIGINAL ARTICLE

Empirics of currency crises: A duration analysis approach

MohammadKarimi

|

Marcel-CristianVoia

Carleton University, Ottawa, Ontario,

Canada

Correspondence

Marcel-Cristian Voia, Laboratoire

d’Économie d'Orléans, Faculté de Droit

d’Économie et de Gestion, Rue de Blois -

BP 26739, 45067, ORLÉANS and Carleton

University 1125 Colonel By Drive, Ottawa,

ON K1S 5B6, Canada.

Emails: marcel.voia@univ-orleans.fr and

marcel.voia@carleton.ca

Abstract

This paper empirically analyzes the origins of currency crises for a group of OECD

economies from 1970 through 1998. We apply duration analysis to examine how the

probability of a currency crisis depends on the length of non- crisis periods, contagion

channels, and macroeconomic fundamentals. Our findings confirm the negative

duration dependence of a currency crisis—the likelihood of speculative attack

sharply increases at the beginning of non- crisis periods and then declines over time

until it abruptly rises again. The results also indicate the hazard of a crisis increase

with high values of the volatility of unemployment rates, inflation rates, contagion

factors—which mostly work through trade channels, unemployment rates, real effec-

tive exchange rate, trade openness, and size of economy. To address concerns re-

garding validity of the identified crisis episodes, we exploit crisis episodes that are

identified by a more objective approach based on the extreme value theory. Our re-

sults are robust under various specifications including two different crisis event sets

that are identified on monthly and quarterly basis.

JEL CLASSIFICATION

F31, F37, F 47, G01

KEYWORDS

crises, exchange rate, International finance

|

429

KARIMI And VOIA

crises and empirically attempts to identify the channels through which the crises are being transmitted. Our goal is to shed light

on the mechanisms of the crises by studying the realization of time- varying explanatory variables, constructed with quarterly

data, as well as the duration pattern of non- crisis periods.

There is an extensive literature on currency crises that empirically evaluates the roots and causes of the crises. Despite

the interesting results of these studies, only very few of them account for the influence of time on the probability of crises.

In a pioneering work, Klein and Marion (1997) provided a key early contribution on the duration of fixed exchange rates and

showed that, time matters as a determinant of the exchange rate survival. They introduced the duration of exchange rates of 17

emerging and developing countries as an explanatory variable in a logit specification.3 Tudela (2004) adopted a more sophis-

ticated approach—duration models—to study the determinants of currency crises in 20 OECD countries. With the help of this

methodology, she incorporated the length of time that a currency had already spent in non- crisis periods as a determinant of

the likelihood of moving into a crisis state. There are also a few other papers that apply duration analysis to study some related

topics such as exchange rate regimes and financial stability.4

We employ duration models to study the probability of a currency exiting a tranquil state into a crisis state. It appears that

maintaining currency credibility gets harder over time.5 Duration models can help us to examine how the passing of time can

affect the stability of a currency. The starting point is that each crisis episode can be treated as a random process. By incorporat-

ing the randomness, we recognize that some important determinants of currency stability remain unobservable at the aggregate

time series level. The unobservable factors can be embodied systematically in the baseline hazard of the attacks, which is easily

captured by duration models. Furthermore, we check whether there is a common pattern for the duration of non- crisis periods

among countries, and whether the timing of crises significantly differs across countries.

The contribution of this paper to the literature is threefold. First, following Eichengreen, Rose, and Wyplosz (1996), we test

for contagious currency crises and attempt to recognize potential contagion channels while controlling for a set of macroeco-

nomic fundamentals. We apply duration analysis, with focus on semi- parametric models, to estimate a model with unrestricted

baseline hazard. These models enjoy the important advantage of not requiring any assumptions on the distribution of the time

of failures. This advantage, on one hand, allows us to capture both the monotonic and the non- monotonic nature of duration

dependence and improve the efficiency of our model. On the other hand, they allow us to remove the risk of a biased coefficient

and provide estimations that are more precise.6 Second, we exploit the identified crisis episodes by a relatively more objective

method—Karimi and Voia (2015)—which is based on the extreme value theory, to minimize concerns regarding the accuracy

of the identified crisis episodes that directly affect the final results of the model. Third, we make use of several robustness

checks, including running our models on two different crisis episode sets that are identified based on monthly and quarterly

type spells.

We found that the probability of speculative attack sharply increases at the start of the tranquil period for three quarters, then

it declines over the time and abruptly rises again after the 20th quarter. Further, the probability of currency crisis rises with

undesired changes in job markets as well as increase in values of inflation rates, real effective exchange rate, size of economy,

trade openness, and trade linkages. Among the three contagion channels that are considered in this paper, the estimation results

for trade linkages were constantly significant in all of our models.

The remainder of the paper is organized as follows. Section 2 concisely reviews the literature on currency crises. Section 3

presents a brief review of theoretical and empirical contagion models. Section 4 concentrates on the methodology and related

issues. Section 5 introduces the variables and describes the data. Section 6 presents the main empirical findings and robustness

tests. Section 7 discusses the results and concludes. Methodology details and some technical results are presented in appendices.

2

|

LITERATURE REVIEW

2.1

|

Theory

The early work on currency crises, now known as first-generation, begins with the seminal work of Krugman (1979).7 The

first- generation models outline how inconsistent domestic monetary and fiscal policies as well as international commitments

(e.g. a fixed exchange rate) push the economy into the crisis. Flood and Garber (1984) develop a comprehensive analytical

framework to examine the speculative attacks by modeling these stylized facts. Flood and Marion (1999) provide a detailed

review of first- generation models.

The first- generation models do not fit very well with what actually happened during most of currency crises. Consequently,

the second-generation models were designed to analyze speculative attacks. These models show even in the presence of policies

that are consistent with the fixed exchange rate, attack- conditional policy changes can pull the economy into an attack. These

models allow speculative attacks to be self- fulfilling8 and set forth the possibilities of multiple equilibria.9 Obstfeld (1996)