Ebook Regionality Revisited: An Examination of the Direction of Spread of Currency
Several waves of excess volatility in currency markets in the last decade of the twentieth century have drawn attention to the fact that currency crises appear to “spread” regionally. In repeated instances, from the early 1970s to the late 1990s, we have observed that when speculative pressure leads to a crisis in one country, market volatility spreads to other countries in the region and elsewhere. If we agree that currency crises tend to spread from one country to another, it is of interest to learn something about the mechanism that drives this spread. Do currency crises spread along geographical lines? Do they spread along trade channels? Is the spread of currency crises simply an example of irrational panics on the part of exchange rate speculators? The answers to these questions are of significance from a theoretical standpoint. In addition, from a policy perspective, understanding the mechanism of spread of currency crises is important in anticipating and preventing them.
In this paper we have two goals. First, we attempt to provide partial answers to some of these questions. While doing so, we also seek to apply a set of relatively new Bayesian econometric techniques hitherto unused in the empirical literature on currency crises. However, before we begin to summarize what we do, it is important to point out what we don’t. The alleged phenomenon of the spread of currency crises between countries is often referred to as contagion. There is some debate in the literature as to whether contagion exists (see, for example, Forbes and Rigobon 1999). It is impossible to “prove” the existence of contagion using econometric techniques because we can never rule out some common, global shock in some unobserved (possibly unknown) variable or linear (or non-linear) combination of variables. While attempts have been made to do so, notably by Eichengreen, Rose, and Wyplosz (1996), and also in a preliminary study by Dasgupta (1998), all of this work is subject to the “common shock in unobserved variables” critique. The best one can do is to claim that it is very unlikely that scattered currency crises can be attributed to global shocks. Most market participants, however, agree on the existence of contagion. Some recent micro-theoretic work (Allen and Gale 2000, Morris 1997, and Dasgupta 2000 for example) provide (albeit highly stylized) structural backing to the existence of contagion in market equilibrium. In this paper, we do not seek to enter the existential debate on contagion. We assume that it exists. Let us agree at the outset that there is some mechanism by which currency crises tend to spread from one country to another. The analysis that follows seeks to learn about this mechanism.
We are not the first to attempt this. Glick and Rose (1999) address this question. Our analysis follows closely in the spirit of theirs, and overlaps non-trivially in content. A large part of our data is taken from the data used for that paper and we are exclusively dependent on them for the identification of currency crises. Part of our exercise, therefore, is a Bayesian reworking of part of their results. However, we do not stop there. The rest of our analysis extends and sharpens the analysis of Glick and Rose (1999), in a manner that we explain below. From a technical econometric standpoint, a major distinction between their paper and ours is in the choice of Bayesian versus Classical technique. We shall motivate and explicate this point in detail later.
Macroeconomic theories differ on the channels by which currency crises spread between countries. One set of theories (e.g. Corsetti, Pesenti, Roubini, and Tille 1998, Gerlach and Smets 1994) point to trade linkages as being the determinant of the direction of spread of currency crises. Their basic argument is that once a particular country experiences a currency crisis and devalues its domestic currency, its exports become more competitive and thus hurts the economies of its trade competitors. Financial speculators may then find it more attractive to attack the currencies of these potentially weakened countries. Another set of theories (e.g. Sachs, Tornell, and Velasco 1996) posit that it is macroeconomic structural similarity between the countries that render them equally vulnerable to coordinated attack by currency traders. Thus, according to these theories, crises spread to countries with similar fundamentals. Other theories point to other forms of links or perceived links between countries. For example, Goldfajn and Valdes (1997) posit that institutional connections vis `a vis cross-country financial links explain contagion. Rigobon (1998) suggests that the regionality of currency crises can be caused by investors learning about a given model of development, since development models tend to be regional.
Given this multiplicity of explanations for the mechanism of contagion, we shall attempt to empirically distinguish between the theories. Glick and Rose (1999) provide evidence in favor of the trade-link theories, but do not explicitly demonstrate that their conclusions favor one class of theories over the others. We shall provide evidence to corroborate their results on trade links using Bayesian techniques. Then we shall extend their data and analysis to provide (albeit indirect) evidence for or against the other two classes of theories. In particular, we shall concentrate on how the various explanations fare when they are considered simultaneously on the same data set.
Download
PDF Ebook Regionality Revisited: An Examination of the Direction of Spread of Currency
Posted in :
