شرح مکانیسم نرخ ارز مجدد : نکاتی درباره مدل های بهینه سازی بحران ارز

This paper attempts to provide empirical evidence on the determinants of the realignments throughout the European exchange rate mechanism (ERM). Motivated by the implications of optimising currency crisis models, we relate the probability of “crises” to a set of macroeconomic fundamentals. By using a conditional binominal logit model we show that regime switches are strongly influenced by movements in industrial production, foreign interest rates, competitiveness and imports as well as in foreign exchange reserves. These findings are consistent with the general propositions of recent currency crises models.

This paper attempts to provide more insights into the understanding of the determinants of currency crises following the recent turmoil in international financial markets. Currency and financial crises experienced by Mexico in 1994–1995, by Indonesia, Korea, Malaysia, the Phillippines and Thailand in 1997 and more recently by Turkey in 2001 and Argentina in 2002 all revived the interest in understanding the roots of currency collapses. The issue of currency crises has already been a lively research area and has been a subject of investigation in international finance literature since the seminal paper by Krugman (1979). The main proposition of the early literature (generally referred to as first generation models) is that balance of payments crises are precipitated by policy imbalances such as lax monetary and fiscal policies. In addition, the level of foreign exchange reserves is argued to be the main determinant of the duration of a fixed exchange rate regime. The events in the European exchange rate mechanism (henceforth the ERM) in 1992 and 1993 posed serious problems for the propositions of the early models. Although the crises which preceded the exit of the Pound and the Italian Lira from the ERM involved some loss of reserves reserve adequacy was clearly not an issue for these countries which could borrow almost unlimited amounts of foreign exchange due to the provisions of the European exchange rate system. In addition, it might have been possible for these governments to raise interest rates to whatever level necessary to avoid the crises. The fact that these countries chose to leave the system instead was widely interpreted as an indication that the authorities’ objective functions included some other key variables.1 These observations might indicate that the decision to leave the ERM was to some extent an optimising decision on the part of those countries rather than an action only forced upon them by either policy imbalances or speculators. This has been at the centre of the new approach to the currency crises. These so-called optimising models of currency crises view regime changes as conscious decisions taken by policy makers who continuously weigh the marginal costs versus benefits of being in a fixed exchange rate regime. Main examples of these are Obstfeld, 1994 and Obstfeld, 1996a, Ozkan and Sutherland, 1995 and Ozkan and Sutherland, 1998, Davies and Vines (1995), Bensaid and Jeanne (1997) and Andersen (1998) among others. These second generation models all share the common feature of modelling policy makers as having well-defined utility functions and thus clear preferences in terms of being in a particular exchange rate regime. The main implication of these models is that a fixed exchange rate regime lasts as long as the policy maker derives more utility from being in it than in either changing the rate or leaving the regime altogether. In other words, a country’s commitment to the fixed exchange rate regime is not state-invariant (see, for example, Flood and Marion, 1998). Motivated by the implications of these new currency crises models, this paper empirically investigates the roots of ‘currency crises’ during the ERM period by extending the set of potential determinants, as suggested by this recent literature. Our data set covers the ERM experience between 1979 and 1992. The reasons for constraining the panel to the ERM members and leaving out the Latin American and East Asian experiences are two-fold. First of all, in the case of recent Latin American and East Asian crises, currency collapses can not be analysed in isolation from financial and banking crises (see, for example, Radelet and Sachs, 1998; Kaminsky and Reinhart, 1998). Secondly, although empirical studies on the ERM abound most of the existing research on the European experience focuses on the determinants of expected changes in the exchange rate (see, for example, Chen and Giovannini, 1993; Caramazza, 1993; Thomas, 1994; Rose and Svensson, 1994). 2 Here we explore the determinants of actual rather than expected realignments. 3 Therefore, by incorporating an extended set of potential variables we provide fresh evidence on the causes of crises during one of the important examples of pegged exchange rate regimes. The first practical task in carrying out this exercise is how to define ‘currency crises’ and distinguish between different crises events. Although it is straightforward to classify the events leading to the decisions of the UK and Italy to leave the ERM in September 1992 as “currency crises”, the same is not true of a large number of other changes in central parities throughout the period 1979–1992. Despite this, in this paper we aim to study the sources of all regime switches. Therefore, we attempt to empirically analyse the macroeconomic determinants of pressure leading to all realignments of central parities during the life of the ERM. This is because all of these changes signal regime switches though to a varying degree. As stated above, the contribution of the optimising models of currency crises is to broaden the set of fundamentals that are derived from the objectives of and the constraints faced by the policy makers that affect their decision to leave the existing exchange rate regime. It is crucial to point out, however, that it is also possible for a shift in market expectations to push an otherwise sound economy into a crisis. As Flood and Marion (1998) point out, in such a situation anything that helps to coordinate speculators’ actions may bring about a sudden attack on the currency. This highlights the difficulty in empirically isolating between the implications of the first and second generation models.4 This is, however, not the purpose of the current study. In this paper, we are mostly concerned with identifying common characteristics of devaluations that could represent sources of macroeconomic trade-offs that were faced by policy makers during the ERM. Affirmative evidence to this would be in line with general propositions of these new models that ‘currency crises’ are precipitated by a broader set of ‘fundamentals’ that affect the interactions between policy makers and the markets.5 Our findings suggest that regime switches in a particular period are influenced by the movements in industrial production, foreign interest rates, competitiveness and imports as well as the foreign exchange reserves. The rest of this paper is organised as follows. Section 2 explains the empirical specification. The main findings are discussed in Section 3 and finally Section 4 concludes the paper. A brief definition of data is provided in the Appendix A.

There are two main findings of this study. First of all, devaluations during the ERM were strongly influenced by movements in industrial production, competitiveness, foreign interest rates and imports as well as in foreign exchange reserves. This piece of evidence is consistent with optimising models of currency crisis where the well-being of policy makers are related to these variables. In fact, this link may be stronger in the empirical analyses of non-ERM pegs. This is because the link between macroeconomic variables and the optimal response by the authorities is weakened by the institutional framework of the ERM which required difficult negotiations with other participants at times of parity changes. Having such an institutional framework is, however, precisely the factor that renders this kind of systems more credible and less vulnerable. In addition, we show that pressure on currencies develops over time. More specifically, the probability of devaluation in the present period is influenced by macroeconomic developments in up to past three months. These findings lead one to believe that a country’s ability to operate a successful exchange rate policy is closely linked to macroeconomic variables and to the economic and political structure required to implement such a policy. Having established the link between the probability of regime switches and the evolution of macroeconomic variables that defines the well-being of the policy maker, the obvious policy implication stands out. Fixed exchange rate regimes can be prolonged by raising the benefits and/or reducing the costs associated with them. One can argue that the reason why it took so long for the ERM crises of 1992–1993 to emerge after the start of the Europe-wide recession was high benefits of maintaining the status quo especially in order to qualify for joining the core countries of the system in shaping the future of an integrated Europe.