- Email: [email protected]
Currency boards, credibility and crises
Economic Systems 26 (2002) 381–386
Currency boards, credibility and crises Marcella Mulino∗ Faculty of Economics, University of L’Aquila, L’Aquila, Italy Received 6 September 2002; accepted 20 September 2002
Abstract The paper calls into question the proposition that currency boards are a solution in preventing currency crisis. On the basis of a “second generation” model, it shows that, in the presence of unemployment persistence, a currency board system can become vulnerable to a currency crisis, as well. The model underlies the role played in triggering the crisis both by expectations of exchange rate realignments and by fundamentals. As the persistence of unemployment has a feedback effect on subsequent periods’ expectations, the credibility of a currency board may decrease over time, eventually inducing a self-fulfilling crisis. © 2002 Elsevier Science B.V. All rights reserved. JEL classification: F33; F41 Keywords: Currency boards; Crises; Credibility
1. Introduction The currency crises of the 1990s have once again raised a deep interest in the issue of the choice of the exchange rate regimes. The growing integration and liberalisation of world capital markets has made it fashionable to argue, especially for emerging countries, that only extreme exchange rate regimes are sustainable. The “two-corner” approach states that a currency crisis can be avoided either by letting the exchange rate float freely or by a final commitment to a fixed exchange rate. Short of adopting a common currency, currency board arrangements represent the most extreme form of exchange rate pegs. The paper calls into question the proposition that currency boards are the solution in preventing currency crises. I elaborate a model of the “second generation” type, in which both realignment expectations and economic fundamentals matter, showing that, in the ∗ Tel.: +39-862-434890; fax: +39-862-434803. E-mail address: [email protected] (M. Mulino).
0939-3625/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0939-3625(02)00062-6
382
M. Mulino / Economic Systems 26 (2002) 381–386
presence of unemployment persistence, a currency board system can become vulnerable to a currency crisis, as ordinary pegs do. A currency board strengthens policymaker’s credibility, as it signals a very firm commitment to a fixed exchange rate, backed by institutional arrangements, and involves a high political cost to the policymaker abandoning it. However, the inherent rigidities of a currency board may have some dangerous side effects that magnify the shortcomings of fixed rates. The most important one is the constraint it imposes on the policymaker with respect to her ability to react to real shocks, because of the absence of policy instruments for stabilisation purposes. In a currency board regime, autonomous monetary policy is ruled out, and budget deficits are run under very strict conditions, so that the required adjustment process may entail large output and unemployment losses, which could in turn undermine the confidence in the sustainability of the peg. Even economists favouring currency boards admit that policy inflexibility has a cost and countries adopting currency boards often do worse than alternative regimes when facing shocks. The reason lies in the basic features of a currency board that make the effects of an adverse shock much more painful than in ordinary pegs, as there are no means by which the policymaker can act to support economic activity, driving down unemployment, other than deflation, which is often a slow and highly painful process. As the Mexican Peso collapsed in 1994, the currencies of nearly all Latin American countries were also subject to speculative attacks. The attack implied that agents perceived that the probability of devaluation was positive and high. In Argentina, the only Latin American country with a currency board system, capital outflows led to an automatic and dramatic cut in the monetary base, putting under considerable stress domestic banks and firms, while domestic interest rates rose as high as 20% in real terms. Albeit, the pressure to devalue was resisted and the expectation of a devaluation subsided, Argentina was pushed into a severe recession, with GDP falling by 6% and unemployment rising to 18%. Owing to the lack of policy instruments to support the economy, the rise in unemployment persisted over time. After the Asian and Russian crises, a major slowdown in 1999 affected the whole Latin American region, but “the fact that Argentine did worse than other countries after 1999 must be attributed to her higher vulnerability to shocks, weaker policy responses or a combination of both” (Perry and Servén, 2002, p. 3). Devaluation expectations and capital outflows got strength, eventually leading to the currency crisis in late 2001.
2. The model The model is based on standard models of monetary policy choice in a small open economy, whose basic framework is drawn from Barro and Gordon (1983). The model is an elaboration of that in Obstfeld (1996), but I assume that shocks and government policies have persistent effects on unemployment. This feature, common to Masson (1995) and Irwin (2001), captures a well-known characteristic of labour markets, which is the existence of a high degree of unemployment persistence. As is standard in the Barro–Gordon framework, the policymaker enjoys an information advantage over the private sector, so that, while the private sector must forecast next period’s inflation in order to specify the nominal wage contract, the government observes the shock hitting the economy and then decides on the
M. Mulino / Economic Systems 26 (2002) 381–386
383
exchange rate policy. Its choice is between a fixed rate in the form of a currency board and a perfectly flexible one. Assuming purchasing power parity, exchange rate policy is directly linked to inflation, as devaluations are equivalent to a positive inflation rate. On any date t, the policymaker minimises a loss function1 which depends on the square both of the deviation of unemployment from the natural rate ut and of the inflation rate:2 Lt = u2t + θε2t .
(1)
I assume that θ is relatively high, reflecting the policymaker’s concern for exchange rate and prices stability. The assumption is that a policymaker adopting or considering to adopt a currency board is a “tough” policymaker, aiming at making more quickly credible her commitment against inflation, and at decreasing the cost of acquiring the reputation connected to this attitude against inflation. The deviation of unemployment from the natural rate is given by: ut = α1/2 [(εet − εt ) + k + δut−1 + zt ],
(2)
with α > 0 and where εet is the private sector’s expectation of exchange rate changes conditional on information available prior to t, so that unexpected devaluations have an expansionary effect, while expected but unrealised devaluations increase ut ; k > 0 is a fixed distortion in the economy that causes unemployment systematically to be above the natural rate; δ is a measure of the persistence of unemployment deviations and zt is an unemployment shock, serially uncorrelated, which is uniformly distributed in the interval [−µ, µ]. In the model, centred only on the supply side of the economy, the shock plays the role of representing the (stochastic) demand side, in a way as simple as possible. Comparing the policy losses in the case of a flexible exchange rate and in the case of a currency board, I obtain that the policymaker will choose a flexible exchange rate regime when LCB − LFL > c, that is when: α2 (α + θ)−1 (εet + k + ut−1 + zt )2 > c,
(3)
where c is the political cost of exiting the currency board.3 Accordingly, she will set εt to its optimal value: εt = α(α + θ)−1 (εet + k + δut−1 + zt ).
(4)
For any εet , and given the values of the fundamentals of the economy, the higher the absolute value of the random shock, the higher will be the relative costliness of the commitment to a currency board, as fixed rates prevent the policymaker from absorbing the shock’s effect on the unemployment rate. 1 This hypothesis implies that the policymaker does not take into consideration the effects her current exchange rate decision has on subsequent periods’ unemployment rate and on market expectations of her future behaviour. Albeit, the model contains inter-temporal links, the above assumption—as stressed by Masson (1995)—greatly simplifies the analysis, allowing to shed light on the interrelations among economic variables. Later on, we are going to consider a multi-period objective function, in order to show how, even in a currency board regime, the policymaker’s credibility may decrease over time. 2 All variables are in logs. 3 I assume that c is exogenously given at a high level, known by the public.
384
M. Mulino / Economic Systems 26 (2002) 381–386
For the policymaker to choose a flexible rate regime, the shock must exceed the following threshold values: √ z¯ t = α−1 c(α + θ) − εet − k − δut−1 , (5) √ zt = −α−1 c(α + θ) − εet − k − δut−1 . ¯ Given the private sector’s expectation of exchange rate realignments, an unemployment shock such that zt > z¯ t induces the government to devalue, while for zt < zt it will revalue. ¯ of the econThe threshold values of the shock depend on the state of the fundamentals omy (as reflected in the previous unemployment deviations ut −1 , given the degree of the persistence δ, and in the economy’s structure k) and on the private sector’s expectation of exchange rate changes εet , as well as on the given political cost of exiting a currency board and the policymaker’s anti-inflationary stance.
3. Expectations and self-fulfilling equilibria Market expectations εet cannot be taken as exogenous, as expectations are determined rationally by agents that understand the choice the government faces. Accordingly, εet is obtained from the optimal rate of exchange rate changes (Eq. (4)), where εet is weighted with the probabilities that the realisation of the shock zt exceeds the above threshold values: z2t − z¯ 2t α α e . (6) εt = (k + δut−1 ) + ¯ θ θ 2[µ − (¯zt + zt )] ¯ Eq. (6) shows that the expectation of exchange rate realignments depends on the state of fundamentals and on the threshold values for the shock. In turn, regime switch points depend on private sector expectations; there is thus an interdependence that creates the potential for a range of self-validating rational expectations equilibria to exist, as discussed at length in several papers by Obstfeld (1996, 1997). In this context, even a currency board is not a fully binding exchange rate regime, notwithstanding the tough attitude towards inflation and the high political cost faced by the policymaker if the currency board collapses. As the commitment may be not fully credible, it is possible that changes in realignment expectations lead to closer threshold values and eventually to self-fulfilling realignments.4 For the commitment to be fully credible, it is necessary that agents’ rational expectations are εet = εt = 0, that fundamentals are in a good state (low k and ut −1 ) and that both c and θ are high, so that the interval between regime switch points is the maximum one. I assume that in a currency board both c and θ are higher than in ordinary pegs. However, the latter is a necessary but not a sufficient condition for the existence of a full credibility equilibrium, owing to the influence of fundamentals. On the other hand, when expectations are positive, the policymaker will maintain the fixed rate only when zt < zt < z¯ t . If this is the case and the policymaker does not realign, the ¯ maintained at the expenses of higher unemployment. As unemployment currency board is 4
The existence of multiple equilibria depends on the distribution function of the shock zt .
M. Mulino / Economic Systems 26 (2002) 381–386
385
deviations are persistent, this will tend to reduce the credibility of the exchange rate peg, augmenting the relative burden of the commitment to a fixed rate.
4. Inter-temporal linkages and currency crisis Unemployment persistence makes the outcome in one period have implications for the outcome in subsequent periods. A simple way to see it is to consider that an increase in ut −1 shortens the interval of the shock’s threshold values zt and z¯ t , for which the policy¯ two channels: a direct one, maker chooses to fix the exchange rate. This occurs through as the threshold values are inversely proportional to ut −1 and an indirect one, through an increase in realignment expectations, as higher unemployment deviations inherited from the previous period raise the expected εet . Analytically, from Eq. (5), we have ∂¯zt /∂ut−1 = ∂zt /∂ut−1 < 0. The level of unemployment the government chooses to leave behind, not ¯ moving to flexible exchange rates when unemployment grows, is detrimental to credibility, inducing a growth in inflationary expectations in the future, thereby increasing unemployment deviation and further reducing the threshold values of the shock. Suppose a two-period setting in which revaluations are ruled out. At the start, unemployment is equal to its steady-state value. The policymaker decides the exchange rate regime to be a currency board for the next two periods; on the basis of the delay implied by parliamentary and constitutional restrictions, I assume that this choice cannot be altered in the first period, so that for the first period the commitment is fully credible. In the second period, instead, the government may devalue (or not). Suppose now that the economy is hit FL by a shock in period 1. The government will devalue in period 2 when LCB 2 − L2 > c; the threshold value for the shock in period 2: z¯ 2 = α−1 c(α + θ) − εe2 − k − δu¯ − α1/2 δz1 , (7) shows a negative correlation to period 1 shock. In addition, z¯ 2 is negatively linked to εe2 , as unfulfilled private sector’s expectations imply an unexpected low inflation and a worsening of fundamentals. The probability of devaluation perceived by the market in the second period is correlated to z1 as well, meaning that the influence of period 1 shock on z¯ 2 and εe2 is mutually reinforcing. Even if there is no shock in the second period, the effect of past unemployment on current unemployment persists, so that the economy moves—after the first period shock—from a full credibility equilibrium to a partial one. In turn, unrealised devaluation expectations influence unemployment and feed back expectations, further lowering the threshold value for the shock. Thus, the realisation of a positive shock moves the economy from a situation of full credibility to one of partial credibility, in which expectations may trigger a speculative attack and eventually induce a self-fulfilling currency crisis. References Barro, R.J., Gordon, D.B., 1983. A positive theory of monetary policy in a natural rate model. J. Political Econ. 91, 589–610.
386
M. Mulino / Economic Systems 26 (2002) 381–386
Irwin, G., 2001. Currency boards and currency crises. Oxford Applied Economics Discussion Paper Series No. 65, January. Masson, P., 1995. Gaining and losing ERM credibility: the case of the United Kingdom. Econ. J. 105, 571–582. Obstfeld, M., 1996. Models of currency crises with self-fulfilling features. Eur. Econ. Rev. 40, 1037–1047. Obstfeld, M., 1997. Destabilizing effects of exchange-rate escape clauses. J. Int. Econ. 43, 61–77. Perry, G., Servén, L., 2002. The anatomy of a multiple crisis: why was Argentina special and what can we learn from it. Mimeo, The World Bank, May.
Currency boards, credibility and crises Marcella Mulino∗ Faculty of Economics, University of L’Aquila, L’Aquila, Italy Received 6 September 2002; accepted 20 September 2002
Abstract The paper calls into question the proposition that currency boards are a solution in preventing currency crisis. On the basis of a “second generation” model, it shows that, in the presence of unemployment persistence, a currency board system can become vulnerable to a currency crisis, as well. The model underlies the role played in triggering the crisis both by expectations of exchange rate realignments and by fundamentals. As the persistence of unemployment has a feedback effect on subsequent periods’ expectations, the credibility of a currency board may decrease over time, eventually inducing a self-fulfilling crisis. © 2002 Elsevier Science B.V. All rights reserved. JEL classification: F33; F41 Keywords: Currency boards; Crises; Credibility
1. Introduction The currency crises of the 1990s have once again raised a deep interest in the issue of the choice of the exchange rate regimes. The growing integration and liberalisation of world capital markets has made it fashionable to argue, especially for emerging countries, that only extreme exchange rate regimes are sustainable. The “two-corner” approach states that a currency crisis can be avoided either by letting the exchange rate float freely or by a final commitment to a fixed exchange rate. Short of adopting a common currency, currency board arrangements represent the most extreme form of exchange rate pegs. The paper calls into question the proposition that currency boards are the solution in preventing currency crises. I elaborate a model of the “second generation” type, in which both realignment expectations and economic fundamentals matter, showing that, in the ∗ Tel.: +39-862-434890; fax: +39-862-434803. E-mail address: [email protected] (M. Mulino).
0939-3625/02/$ – see front matter © 2002 Elsevier Science B.V. All rights reserved. doi:10.1016/S0939-3625(02)00062-6
382
M. Mulino / Economic Systems 26 (2002) 381–386
presence of unemployment persistence, a currency board system can become vulnerable to a currency crisis, as ordinary pegs do. A currency board strengthens policymaker’s credibility, as it signals a very firm commitment to a fixed exchange rate, backed by institutional arrangements, and involves a high political cost to the policymaker abandoning it. However, the inherent rigidities of a currency board may have some dangerous side effects that magnify the shortcomings of fixed rates. The most important one is the constraint it imposes on the policymaker with respect to her ability to react to real shocks, because of the absence of policy instruments for stabilisation purposes. In a currency board regime, autonomous monetary policy is ruled out, and budget deficits are run under very strict conditions, so that the required adjustment process may entail large output and unemployment losses, which could in turn undermine the confidence in the sustainability of the peg. Even economists favouring currency boards admit that policy inflexibility has a cost and countries adopting currency boards often do worse than alternative regimes when facing shocks. The reason lies in the basic features of a currency board that make the effects of an adverse shock much more painful than in ordinary pegs, as there are no means by which the policymaker can act to support economic activity, driving down unemployment, other than deflation, which is often a slow and highly painful process. As the Mexican Peso collapsed in 1994, the currencies of nearly all Latin American countries were also subject to speculative attacks. The attack implied that agents perceived that the probability of devaluation was positive and high. In Argentina, the only Latin American country with a currency board system, capital outflows led to an automatic and dramatic cut in the monetary base, putting under considerable stress domestic banks and firms, while domestic interest rates rose as high as 20% in real terms. Albeit, the pressure to devalue was resisted and the expectation of a devaluation subsided, Argentina was pushed into a severe recession, with GDP falling by 6% and unemployment rising to 18%. Owing to the lack of policy instruments to support the economy, the rise in unemployment persisted over time. After the Asian and Russian crises, a major slowdown in 1999 affected the whole Latin American region, but “the fact that Argentine did worse than other countries after 1999 must be attributed to her higher vulnerability to shocks, weaker policy responses or a combination of both” (Perry and Servén, 2002, p. 3). Devaluation expectations and capital outflows got strength, eventually leading to the currency crisis in late 2001.
2. The model The model is based on standard models of monetary policy choice in a small open economy, whose basic framework is drawn from Barro and Gordon (1983). The model is an elaboration of that in Obstfeld (1996), but I assume that shocks and government policies have persistent effects on unemployment. This feature, common to Masson (1995) and Irwin (2001), captures a well-known characteristic of labour markets, which is the existence of a high degree of unemployment persistence. As is standard in the Barro–Gordon framework, the policymaker enjoys an information advantage over the private sector, so that, while the private sector must forecast next period’s inflation in order to specify the nominal wage contract, the government observes the shock hitting the economy and then decides on the
M. Mulino / Economic Systems 26 (2002) 381–386
383
exchange rate policy. Its choice is between a fixed rate in the form of a currency board and a perfectly flexible one. Assuming purchasing power parity, exchange rate policy is directly linked to inflation, as devaluations are equivalent to a positive inflation rate. On any date t, the policymaker minimises a loss function1 which depends on the square both of the deviation of unemployment from the natural rate ut and of the inflation rate:2 Lt = u2t + θε2t .
(1)
I assume that θ is relatively high, reflecting the policymaker’s concern for exchange rate and prices stability. The assumption is that a policymaker adopting or considering to adopt a currency board is a “tough” policymaker, aiming at making more quickly credible her commitment against inflation, and at decreasing the cost of acquiring the reputation connected to this attitude against inflation. The deviation of unemployment from the natural rate is given by: ut = α1/2 [(εet − εt ) + k + δut−1 + zt ],
(2)
with α > 0 and where εet is the private sector’s expectation of exchange rate changes conditional on information available prior to t, so that unexpected devaluations have an expansionary effect, while expected but unrealised devaluations increase ut ; k > 0 is a fixed distortion in the economy that causes unemployment systematically to be above the natural rate; δ is a measure of the persistence of unemployment deviations and zt is an unemployment shock, serially uncorrelated, which is uniformly distributed in the interval [−µ, µ]. In the model, centred only on the supply side of the economy, the shock plays the role of representing the (stochastic) demand side, in a way as simple as possible. Comparing the policy losses in the case of a flexible exchange rate and in the case of a currency board, I obtain that the policymaker will choose a flexible exchange rate regime when LCB − LFL > c, that is when: α2 (α + θ)−1 (εet + k + ut−1 + zt )2 > c,
(3)
where c is the political cost of exiting the currency board.3 Accordingly, she will set εt to its optimal value: εt = α(α + θ)−1 (εet + k + δut−1 + zt ).
(4)
For any εet , and given the values of the fundamentals of the economy, the higher the absolute value of the random shock, the higher will be the relative costliness of the commitment to a currency board, as fixed rates prevent the policymaker from absorbing the shock’s effect on the unemployment rate. 1 This hypothesis implies that the policymaker does not take into consideration the effects her current exchange rate decision has on subsequent periods’ unemployment rate and on market expectations of her future behaviour. Albeit, the model contains inter-temporal links, the above assumption—as stressed by Masson (1995)—greatly simplifies the analysis, allowing to shed light on the interrelations among economic variables. Later on, we are going to consider a multi-period objective function, in order to show how, even in a currency board regime, the policymaker’s credibility may decrease over time. 2 All variables are in logs. 3 I assume that c is exogenously given at a high level, known by the public.
384
M. Mulino / Economic Systems 26 (2002) 381–386
For the policymaker to choose a flexible rate regime, the shock must exceed the following threshold values: √ z¯ t = α−1 c(α + θ) − εet − k − δut−1 , (5) √ zt = −α−1 c(α + θ) − εet − k − δut−1 . ¯ Given the private sector’s expectation of exchange rate realignments, an unemployment shock such that zt > z¯ t induces the government to devalue, while for zt < zt it will revalue. ¯ of the econThe threshold values of the shock depend on the state of the fundamentals omy (as reflected in the previous unemployment deviations ut −1 , given the degree of the persistence δ, and in the economy’s structure k) and on the private sector’s expectation of exchange rate changes εet , as well as on the given political cost of exiting a currency board and the policymaker’s anti-inflationary stance.
3. Expectations and self-fulfilling equilibria Market expectations εet cannot be taken as exogenous, as expectations are determined rationally by agents that understand the choice the government faces. Accordingly, εet is obtained from the optimal rate of exchange rate changes (Eq. (4)), where εet is weighted with the probabilities that the realisation of the shock zt exceeds the above threshold values: z2t − z¯ 2t α α e . (6) εt = (k + δut−1 ) + ¯ θ θ 2[µ − (¯zt + zt )] ¯ Eq. (6) shows that the expectation of exchange rate realignments depends on the state of fundamentals and on the threshold values for the shock. In turn, regime switch points depend on private sector expectations; there is thus an interdependence that creates the potential for a range of self-validating rational expectations equilibria to exist, as discussed at length in several papers by Obstfeld (1996, 1997). In this context, even a currency board is not a fully binding exchange rate regime, notwithstanding the tough attitude towards inflation and the high political cost faced by the policymaker if the currency board collapses. As the commitment may be not fully credible, it is possible that changes in realignment expectations lead to closer threshold values and eventually to self-fulfilling realignments.4 For the commitment to be fully credible, it is necessary that agents’ rational expectations are εet = εt = 0, that fundamentals are in a good state (low k and ut −1 ) and that both c and θ are high, so that the interval between regime switch points is the maximum one. I assume that in a currency board both c and θ are higher than in ordinary pegs. However, the latter is a necessary but not a sufficient condition for the existence of a full credibility equilibrium, owing to the influence of fundamentals. On the other hand, when expectations are positive, the policymaker will maintain the fixed rate only when zt < zt < z¯ t . If this is the case and the policymaker does not realign, the ¯ maintained at the expenses of higher unemployment. As unemployment currency board is 4
The existence of multiple equilibria depends on the distribution function of the shock zt .
M. Mulino / Economic Systems 26 (2002) 381–386
385
deviations are persistent, this will tend to reduce the credibility of the exchange rate peg, augmenting the relative burden of the commitment to a fixed rate.
4. Inter-temporal linkages and currency crisis Unemployment persistence makes the outcome in one period have implications for the outcome in subsequent periods. A simple way to see it is to consider that an increase in ut −1 shortens the interval of the shock’s threshold values zt and z¯ t , for which the policy¯ two channels: a direct one, maker chooses to fix the exchange rate. This occurs through as the threshold values are inversely proportional to ut −1 and an indirect one, through an increase in realignment expectations, as higher unemployment deviations inherited from the previous period raise the expected εet . Analytically, from Eq. (5), we have ∂¯zt /∂ut−1 = ∂zt /∂ut−1 < 0. The level of unemployment the government chooses to leave behind, not ¯ moving to flexible exchange rates when unemployment grows, is detrimental to credibility, inducing a growth in inflationary expectations in the future, thereby increasing unemployment deviation and further reducing the threshold values of the shock. Suppose a two-period setting in which revaluations are ruled out. At the start, unemployment is equal to its steady-state value. The policymaker decides the exchange rate regime to be a currency board for the next two periods; on the basis of the delay implied by parliamentary and constitutional restrictions, I assume that this choice cannot be altered in the first period, so that for the first period the commitment is fully credible. In the second period, instead, the government may devalue (or not). Suppose now that the economy is hit FL by a shock in period 1. The government will devalue in period 2 when LCB 2 − L2 > c; the threshold value for the shock in period 2: z¯ 2 = α−1 c(α + θ) − εe2 − k − δu¯ − α1/2 δz1 , (7) shows a negative correlation to period 1 shock. In addition, z¯ 2 is negatively linked to εe2 , as unfulfilled private sector’s expectations imply an unexpected low inflation and a worsening of fundamentals. The probability of devaluation perceived by the market in the second period is correlated to z1 as well, meaning that the influence of period 1 shock on z¯ 2 and εe2 is mutually reinforcing. Even if there is no shock in the second period, the effect of past unemployment on current unemployment persists, so that the economy moves—after the first period shock—from a full credibility equilibrium to a partial one. In turn, unrealised devaluation expectations influence unemployment and feed back expectations, further lowering the threshold value for the shock. Thus, the realisation of a positive shock moves the economy from a situation of full credibility to one of partial credibility, in which expectations may trigger a speculative attack and eventually induce a self-fulfilling currency crisis. References Barro, R.J., Gordon, D.B., 1983. A positive theory of monetary policy in a natural rate model. J. Political Econ. 91, 589–610.
386
M. Mulino / Economic Systems 26 (2002) 381–386
Irwin, G., 2001. Currency boards and currency crises. Oxford Applied Economics Discussion Paper Series No. 65, January. Masson, P., 1995. Gaining and losing ERM credibility: the case of the United Kingdom. Econ. J. 105, 571–582. Obstfeld, M., 1996. Models of currency crises with self-fulfilling features. Eur. Econ. Rev. 40, 1037–1047. Obstfeld, M., 1997. Destabilizing effects of exchange-rate escape clauses. J. Int. Econ. 43, 61–77. Perry, G., Servén, L., 2002. The anatomy of a multiple crisis: why was Argentina special and what can we learn from it. Mimeo, The World Bank, May.