After years of calm, the international financial system recently has been hit by a succession of speculative attacks. Among the consequences were the abandonment in August 1993 of the narrow fluctuation bands of the European Monetary System, and the 1994 Mexican peso crisis, which prompted the International Monetary Fund, the United States and other G–7 countries to provide $50 billion of assistance for fear of serious repercussions on other countries. These events brought the issue of speculatively engendered crises to the forefront of academic and policy discussions. On 18/19 April 1997, a joint CEPR conference with the Banco de Portugal, held in Sesimbra, Portugal, on Speculative Attacks on Foreign Exchange Reserves, brought together leading researchers in the field. The conference was organised by Axel Weber (Universität Bonn and CEPR), with financial support from the European Commission. Seven papers were given, followed by a policy panel on how to deal with speculative attacks.

Nancy Marion (Dartmouth College) presented ‘Speculative Attacks: Fundamentals and Self-Fulfilling Prophecies’, a joint work with Robert Flood (International Monetary Fund). The authors developed a modified ‘first-generation model’ in which the monetary authorities fully sterilize the effect of reserve changes on the monetary base and price-setting behaviour is sticky. The introduction of a time-varying stochastic risk-premium dampens the interest-rate response to an impending attack and gives rise to self-fulfilling multiple equilibria. Flood and Marion showed that both inconsistent macroeconomic policies and large shifts in speculative opinion can trigger a crisis. Contrary to previous ‘first-generation models’, the post-attack growth rate of the monetary base is not required to change – a feature more in line with the Mexican crisis of 1994.

Hélène Rey (London School of Economics) suggested that the behaviour of the monetary authorities after the attack be endogenized and the robustness of the results with respect to an alternative modelling of the stochastic shocks be examined. Giancarlo Corsetti (Yale University) pointed out that, because agents’ anticipations are

unobservable, a signal-extraction problem arises
which prevents empirical verification. Peter Garber (Brown University) stressed the discretionary nature of the first Mexican devaluation and the importance of reserves for the second and final devaluation.

In ‘The Leading Indicators of Currency Crises’, Graciela Kaminsky (Federal Reserve Board, Washington), Saul Lizondo (The World Bank) and Carmen Reinhart (University of Maryland) examined the ‘symptoms’ of currency crises and their potential as ‘early warning signals’. They began by identifying a series of potential indicators suggested by the theoretical and empirical literature. By applying the leading-indicators technique from the business-cycle literature to speculative crises, the authors derived a warning system that involves monitoring a number of variables. Indicators judged particularly useful in predicting speculative crises included the behaviour of international reserves, the real exchange rate, domestic credit, credit to the public sector and domestic inflation.

Axel Weber (Universität Bonn and CEPR) emphasized the problems involved in using a panel, like the one considered by the authors, which mixes developing and industrialized countries. He noted that crises in the ERM succeeded one another at very short intervals, which made it difficult to attribute the behaviour of an indicator to a particular crisis. Paolo Pesenti (Princeton University) stressed the role of policy variables in the explanation of the 1992/3 ERM crises. Maurice Obstfeld (University of California, Berkeley, and CEPR) wondered about the influence of policy on the generation of false signals. Can differences in policy response to a speculative attack explain why, in some countries, no crisis occurred, despite strong signals from the indicators? Pierre-Richard Agénor (International Monetary Fund) remarked that controls on exchange rates and interest rates in some countries could distort the value of these variables as indicators.

In ‘Borrowing Risk and the Tequila Effect’, Pierre-Richard Agénor (International Monetary Fund) investigated contagion effects in an inter-temporal optimizing model of a small open economy. The Tequila effect was modelled as a temporary increase in the exogenous risk premium faced by domestic agents on imperfect world capital markets. The author then analysed the real and monetary effects of a shock on the risk premium by modelling portfolio decisions, real-wage rigidity and the link between bank credit and the supply side of the economy through firms’ demand for working capital. The model was able to reproduce the main characteristics of Argentina’s downturn immediately after the Mexican peso crisis. These effects included the sharp increase in interest rate spreads relative to the United States; the reduction in net capital inflows; the drop in official reserves; the reduction in bank deposits and

credit supply; the fall in private consumption; the contraction in output; and the increase in unemployment.

Bernard Bensaid (Université de Lille) asked why Agénor did not consider the level and structure of public debt, which had played a major role in triggering the crisis. He then noted that currency boards are not necessarily fully credible, as the model assumed. Graciela Kaminsky pointed to some empirical facts about Argentina, namely the real exchange rate appreciation and the loss in bank competitiveness prior to the crisis. Giancarlo Corsetti remarked that the default risk could be calculated using Brady Bonds.

Alan Sutherland (University of York and CEPR) presented ‘Currency Crisis and the Term Structure of Interest Rates’. He suggested using historical term-structure data to distinguish between crises based on fundamentals and those generated by self-fulfilling expectations, since the two types generated different term-structure behaviour in the periods preceding a crisis. Determining the nature of a crisis may be of great practical importance given that the appropriate policy response depends largely on the type of crisis. Paolo Pesenti (Princeton University) noted that the shape of the term structure depends crucially on the particular theoretical modelling and parameterization of the crisis type. Maurice Obstfeld emphasized the term-structure effects of a monetary policy aimed at fighting speculation, which could blur the outcomes described in the paper. Graciela Kaminsky enquired about the implications of an endogenous trigger for the results, while Axel Weber suggested the trigger could be inferred from the distribution of exchange rates.

‘Contagious Currency Crises’ was the title of the paper presented by Andrew Rose (University of Berkeley, California) – a joint work with Barry Eichengreen (Center for Advanced Study in the Behavioral Sciences and CEPR) and Charles Wyplosz (Institut Universitaire des Hautes Études Internationales, Genève, and CEPR). The authors explored the observation that speculative attacks tend to be temporally correlated, in that an attack elsewhere increases the likelihood of an attack at home. Having confirmed empirically the existence of contagion effects, the authors examined two potential transmission channels for speculative attacks: first, trade links – the hypothesis being that attacks spill over more easily the more intensive the countries’ trade links; and second, similarities in macroeconomic policies and conditions. Empirically, the first channel seemed more important.

Mark Salmon (European University Institute) stressed the importance of the market microstructure. Contagion effects could be desirable if they represent agents’ optimal portfolio rebalancing responses to exogenous shocks. Salmon also wondered about the sample sensitivity of the results and the difficulties in defining a crisis in empirical applications. Graciela Kaminsky objected to treating all crises and countries identically and suggested replacing the binary crisis variable by a more refined index. Peter Garber argued that the breakdown of an exchange rate system – such as Bretton-Woods or the ERM – affects all participating countries without constituting a contagion effect. Consequently, the sensitivity of the results to the exclusion of those periods should be examined. William Allen (Bank of England) was concerned about the choice of the centre currency. He suggested replacing the Deutsche mark with the dollar. Helmut Fritsch (Deutsche Bundesbank) remarked that currencies of countries that exhibit strong trade links tend also to appreciate together owing to the intervention behaviour of monetary authorities. In this case contagion effects are beneficial. Nick Vidalis (European Monetary Institute) added that contagion effects should depend on the degree of financial liberalization.

The extent to which speculation against the French franc in 1992/3 was driven by fundamentals was investigated by Olivier Jeanne (University of California, Berkeley) and Paul Masson (International Monetary Fund). In ‘Was the French Franc Crisis a Sunspot Equilibrium?’, Jeanne and Masson developed a model of currency crises that encompassed both hypotheses about the origin of speculation. They found that the model provided a better description of the crisis when it gave a role to sunspots. In this case, the model not only tracked speculative episodes better, by interpreting them as self-fulfilling jumps in the beliefs of foreign-exchange market participants, but also provided a better account of the empirical relationship between fundamentals and devaluation expectations.

Berthold Herrendorf (University of Warwick) observed that the ERM crisis should not be viewed as the collapse of a bundle of unilateral pegs, but as the collapse of an exchange rate system. This systemic view may foster better understanding of some important features of the French franc crisis. A further – and so far neglected – cost of a devaluation consists in the implied reduction in the likelihood that the country will achieve membership of the first-stage EMU. This factor could have played an important role in the crisis. Peter Garber enquired whether the augmented degrees of freedom in the sunspot model could account for the better performance of this model. He noted that, generally, the two models are observationally equivalent. Axel Weber pointed to the potential empirical relevance of non-stationarities in the fundamentals.

‘Interpreting the ERM Crisis: Country-Specific and Systemic Issues’, was written by Willem Buiter (University of Cambridge and CEPR), Giancarlo Corsetti (Yale University) and Paolo Pesenti (Princeton University). In a multi country monetary and exchange rate game, in which they combined country-specific and systemic factors, they argued that the role of strategic interactions among national policy-makers hitherto had been overlooked. In contrast to a single-country analysis, they were able to explain the ‘puzzles’ of the ERM crisis. Peter Garber stressed the importance of the very short-term financing facilities and the ‘microcosmos’ of exchange rate defence for the course of the 1992/3 EMS crisis. Helmut Fritsch remarked that an early realignment could have smoothed the ERM crisis.

The conference closed with a policy panel on ‘Dealing with Speculative Attacks’. Maurice Obstfeld discussed the transition to EMU and highlighted some potential dangers associated with the choice of the conversion rate. Market-based schemes perpetuate the exchange rates on the last day, which may not be desirable. A superior solution consists in fixing the conversion rates early on and then letting exchange rates float until the last day. José Viñals (Banco de España and CEPR) discussed the policy responses to speculative attacks and their relation to the type of crisis. In this context he stressed the role of the widened bands in restoring two-sided risk in financial markets. Hervé Carré (European Commission) noted that the move towards EMU represented the only definitive means for preventing speculative attacks. There was a need for credible and consistent policies and for collective surveillance and cooperation. Helmut Fritsch argued that the selection of first-stage participants would be of the utmost importance for the success of EMU and suggested that central rates should be chosen as conversion rates. With regard to the 1992/3 EMS crises, the Bundesbank had no alternative to the restrictive policy it conducted. Fritsch emphasized the huge interventions of the Bundesbank in the exchange rate markets, and added that the introduction of the euro could be expected to stabilize the international monetary system.