Faculty of Economics and Business University of Groningen, the Netherlands December 22, 2010 The role of currency unions in preventing currency crises: A comparison with fixed exchange rate regimes T M

M

T

The role of currency unions in preventing currency crises:

A comparison with fixed exchange rate regimes

Faculty of Economics and Business

University of Groningen, the Netherlands

December 22, 2010

Keywords: Currency union; currency crisis; exchange rate; speculative pressure JEL classification: F30; F33; F36; G01

Wilma Huitema∗

Supervised by Prof. Dr. K.H.W. Knot

Abstract

C

I. Introduction 4

II. Theoretical Background 8

2.1 History of International Financial Systems 8 2.2 Models of Currency Crises 11 2.3 The Optimal Choice of Exchange Rate Regime 14 2.4 Measurement of Exchange Rate Regimes 17

III. Comparative Analysis 20

3.1 First Generation Models 20 3.2 Second Generation Models 25

3.3 Banking Crisis 31

3.4 Cost of Leaving a Monetary Arrangement 32

3.5 Conclusion 35

IV. Empirical Analysis 38

4.1 Probit Model 38

4.2 Data 40

4.3 Results 41

4.4 Tests for Robustness 44

4.5 Conclusion 46

V. Conclusion and Limitations 47

References 49

I. I

The current financial and economic crisis, which erupted in the summer of 2007 with the United States (US) subprime mortgage crisis, affected the world economy drastically. The crisis has resulted in bail outs of banks, collapses of financial institutions, and declining economic growth. Europe is hit as well by this financial crisis. In 2008 and 2009 three European countries (Hungary, Latvia, and Romania) experienced a currency crisis. These crises were so severe that the European Union (EU) and the International Monetary Fund (IMF) had to provide financial assistance. This financial program had as purpose to ensure balance of payments stability and to ensure credibility in order to avoid other speculative attacks (IMF, 2009a). A currency crisis is generally defined as an increase in the speculative pressure index or as an sharp depreciation or devaluation of the exchange rate (Coulibaly, 2009). Hungary was one of the first developing countries which was hit by the financial crisis. Hungary has high levels of government and external debt and, therefore, it experienced immediate financing difficulties when the world started to deleverage in 2007. These difficulties resulted in, among others, high interest rates on government debt and downward pressure on the exchange rate (IMF, 2009b). In response to this, Hungary had to eliminate its exchange rate band against the euro in February 2008 (IMF, 2008). The Hungarian forint depreciated by seven percent against the euro (IMF, 2009b). The financial crisis tested the currency pegs of the Eastern Europe currencies to the euro. The Latvian lats showed to be volatile, however, financial assistance of the IMF and the EU prevented the currency from devaluation (ECB, 2010).

can, for example, not use a devaluation to stimulate its exports, which might improve its balance of payments. Though, one has to bear in mind that the effect of a nominal devaluation tends to dissipate over time. Nominal depreciations of the exchange rate only have a temporary effect on the real exchange rate since the nominal depreciation will lead to increases in domestic costs and prices which restores the initial relative prices (De Grauwe, 2009).

The definition of a currency crisis in a currency union is twofold. On the one hand, speculators could attack a country to force its departure from a currency union, on the other hand speculators could attack an entire currency union in order to force a devaluation against another currency (Coulibaly, 2009). However, is it likely that Greece will be forced to leave the Eurozone, or that the entire Eurozone will be in a currency crisis because of the problems in Greece? In other words, is the probability of a currency crisis smaller in a currency union than in a fixed exchange rate regime? A currency union could have for instance a higher level of foreign reserves with which it can defend its exchange rate. In addition, being a member of a currency union might lead to more fiscal discipline since currency union members are not able to monetize budget deficits anymore. More fiscal discipline will lead to lower fiscal deficits and a lower probability of a currency crisis. On the other hand, because of moral hazard a currency union might reduce fiscal discipline which in turn will lead to a higher probability of a currency crisis. Subsequently, the hypothesis that will be tested in this study is: the probability of a currency crisis is, ceteris paribus, significantly lower when a country is in a currency union than when that country has a fixed exchange rate regime.

characterized by fixed exchange rates. If membership of a currency union will lower the probability of a currency crisis this might move countries towards currency unions over time, like the Economic and Monetary Union (EMU). Besides, it can give an additional solution to prevent financial crises, which is important since financial crises are costly. When there is a sudden stop or an outflow of capital during a crisis this can slow down growth. In addition, a devaluation can lead to a rise in the external debt burden which can lower investment and, thus, growth. Gupta et al. (2003) estimated, for example, that during the East Asian crises the Indonesian and Thai economies contracted by 13 and 10 percent, respectively. In addition, in the Argentina crisis of 2002 output declined by nearly 11 percent. The IMF (1998) forecasted that currency crises result in a cumulative loss in output growth of 4.25 percentage points on average. For ‘severe’ crises this amounts to 8.25 percentage points. However, some studies argue that currency crises can have a positive effect on economic growth. Especially the early literature focuses on the beneficial effects of currency crises. An depreciation can make an exchange rate more competitive, which can expand the tradable goods sector and, hence, can stimulate growth (Gupta et al., 2003).

In order to test the hypothesis this study will perform a comparative analysis and an empirical analysis. The comparative analysis will compare the factors that trigger a currency crisis in a currency union with these factors in a fixed exchange rate regime. The empirical part will determine the change in probability of a currency crisis when a country moves from a fixed exchange rate to a currency union. This will be done by estimating a probit model using panel data of 193 countries over the years 1970-2000. An article by Kaminsky et al. (1998) gives an overview of the selected independent variables used in comparable currency crises models.

percent. This study will differ from Coulibaly (2009) since it focuses particularly on the difference between a currency union and a fixed exchange rate regime.

II. T

B

This chapter will discuss the theoretical background of exchange rate regimes and currency crises. The chapter starts with a brief history of international financial systems. Subsequently, various models of currency crises will be discussed. Related to the search for financial stability is the optimal choice of an exchange rate regime, which will be discussed in Section 2.3. This chapter will conclude with the measurement of exchange rate regimes in Section 2.4

2.1

H

I

F

S

One of the first attempts to enhance financial stability was the Bretton Woods system, which had as purpose to avoid a breakdown in international financial relations as occurred in the 1930s and during World War II. The initiators of the Bretton Woods system were the US and the United Kingdom. The Bretton Woods system commenced in March 1947. 44 countries joined the system which was characterized by fixed but adjustable exchange rates. Each currency had a central parity against the US dollar and was allowed to fluctuate with a margin of one percent. In addition, the US dollar was fixed to the price of gold at 35 US dollars per ounce. Besides these fixed exchange rates, the International Monetary Fund (IMF) and the International Bank for Reconstruction and Development (IBRD), also known as the World Bank, were set up.

Bretton Woods system were reluctant to devalue or revalue their currencies or to take other measures to ensure sustainable balance of payments positions. Countries with an undervalued currency were reluctant to revalue their currency since this would have a negative effect on their exports. Countries with an overvalued currency would not devalue since this was a sign of governmental weakness. In the long run it was impossible to maintain the fixed exchange rate parities as well as balance of payments balances. The exchange rate parities had to be realigned. In 1971 it was clear to speculators and governments that the US dollar was overvalued and other currencies were undervalued. Countries allowed their exchange rate to float against the US dollar (Pilbeam, 2006).

After the collapse of the Bretton Woods system, the exchange rates in Europe were determined by market forces. However, the countries in the European Economic Community (EEC) were concerned that large movements in the exchange rate could undermine their international competitiveness and could hamper the development of free trade. Therefore, the six member countries of the EEC (Belgium, France, Italy, Luxembourg, the Netherlands, and Germany) decided at The Hague summit in 1969 that they would establish a complete economic and monetary unification by 1980. So, the idea of a single European currency goes back to 1969. In order to establish a monetary union, the EEC members set up a so-called ‘Snake in the Tunnel’ system to provide exchange rate stability. The ‘Snake in the Tunnel’ comprised two agreements. Firstly, the member countries agreed that they could float a maximum of ± 1.125 percent against each other (the Snake) and secondly, the countries could float against the US dollar by a maximum of ± 2.25 percent (the Tunnel). In 1972 the United Kingdom, Denmark, and Norway joined the system, however, after six weeks the United Kingdom abandoned its membership. Italy abandoned its membership in February 1972 and in March 1973 the remaining countries decided to let their currencies float against the US dollar, which resulted in ‘The Snake in the Lake’. In the late 1970s France and Norway both left the system. In addition, the ‘Snake in the Tunnel’ was characterized by devaluations and revaluations of, among others, the deutschmark and the Norwegian krone (Pilbeam, 2006).

all member countries with margins of fluctuation of ±2.25 percent. To overcome the problems of the Bretton Woods system, the EMS system differed from it in several ways. Firstly, the EMS had a wider band of fluctuation than the Bretton Woods system. Secondly, EMS member countries where obliged to intervene when there bilateral exchange rate reached the margin whereas the Bretton Woods system did not have an intervention mechanism. In addition, each currency had an individual band of fluctuation against the European Currency Unit (ECU), which was an artificial currency composed as a weighted basket of the 12 member currencies. Since the currencies were fixed against the ECU instead of a single currency, like the US dollar in the Bretton Woods system, the EMS overcame the liquidity problem of the Bretton Woods. The first years of EMS (1979-1984) showed frequent and substantial realignments. After 1984 the realignments became less frequent, however, in 1992 the pressure on EMS increased and a currency crisis emerged. The United Kingdom and Italy left the EMS system in 1992 and Spain, Portugal, and Ireland had to devalue their currencies. In 1993 the margins of fluctuations widened drastically to 15 percent (Pilbeam, 2006). The widening of the margins of fluctuations of 15 percent can be seen as the collapse of the EMS, since these margins might be too large to be effective. On the other hand, Knot and De Haan (1994) show that after 1993 most currencies of the EMS system fluctuated again in the more narrow margins of before the crisis. Labhard and Wyplosz (1996) demonstrate as well that the average bandwidth of the EMS currencies closely resembles the bandwidths of before the crisis. In addition, Germany and the Netherlands agreed to keep their bilateral exchange rate between a narrow band of ± one percent.

2002 the euro was introduced in 12 European countries. At this moment the euro is the common currency of 16 European countries. On 1 January 2011 the euro will be introduced in Estonia as well.

2.2

M

C

C

There is a vast literature about currency crises, which can be classified into first, second, and third generation models of currency crises. The first generation models were mostly developed in reaction to the Latin American crises in the 1960s and 1970s (Kaminsky, 2003). These models stress that currency crises are caused by governments who pursue fiscal policies that are inconsistent with maintaining a fixed exchange rate. For example, governments that monetize large fiscal deficits. An influential article by Krugman (1979) describes the basics of these models. Krugman (1979) describes an economy with a fixed exchange rate which will be defended by the government with the use of foreign exchange reserves. The government expands domestic credit primarily to finance a budget deficit. There will be an imbalance between the increasing domestic credit and the stable money demand. These imbalances lead to a downward pressure on the interest rate, capital outflows, and a pressure for devaluation of the currency. Since the peg is defended with foreign exchange reserves, the reserves will be depleted. When the level of foreign reserves reaches a critical value, rational investors will start a speculative attack on the currency. They will sell the domestic currency and buy the foreign currency in anticipation of an devaluation. Foreign reserves will be depleted and the government cannot sustain the peg (Kaminsky, 2003). For other first generation models see, for example, Flood and Garber (1984). In conclusion, first generation models explain currency crises by fundamental discrepancies between fiscal and monetary policies on the one hand and the commitment to a fixed exchange rate on the other hand. This discrepancy is observed by speculators and they start a speculative attack which might result in a currency crisis (Pilbeam, 2006).

can anticipate that this might lead to a speculative attack and a currency crisis. Finally, the currency crises in the EMS in the 1990s and the Mexican crisis of 1994 could not be explained on basis of these first generation models. These crises have, for example, not showed a dramatic fall in foreign reserves (Pilbeam, 2006).

The limitations of the first generation models in explaining the EMS crisis and the Mexican crisis led to a new generation of models. The second generation models of currency crises are characterized by a government who continually compares the benefits and losses from defending the peg against changing the exchange rate peg. The loss function of maintaining a fixed exchange rate is defined, for example, by deviations of other domestic macroeconomic fundamentals like output or inflation. Governments face a trade-off between maintaining a fixed exchange rate and keep domestic macroeconomic fundamentals around a target. If maintaining a fixed exchange rate is at the expense of macroeconomic fundamentals the cost of defending the peg rises (Obstfeld, 1996). Another feature of second generation models is the possibility of multiple equilibria. This means that when the expectations of devaluation are low the costs of maintaining the peg are low. When the expectations of devaluation are high the costs of defending the peg are high and the probability of a devaluation will become higher. Therefore, the expectations of speculators can become self-fulfilling. A shock that increases the expectations of devaluation, increases the costs of defending this peg and this leads to a higher probability of devaluation. Important to note is that the shocks that alter the expectations of devaluation need not to be linked with macroeconomic fundamentals (Pilbeam, 2006). Second generation models argue that even in credible fixed exchange rate regimes speculative attacks can occur by altered expectations of speculators (De Grauwe, 2009). However, second generation models are limited by their lack of explaining the change in speculators’ expectations (Pilbeam, 2006).

capital outflows and, hence, a downward pressure on the exchange rate or even a currency crisis (Pilbeam, 2006).

In addition, Eichengreen and Hausmann (1999) examine the relationship between exchange rate regimes and financial fragility. They discuss a moral hazard hypothesis as well. Like the third generation models of currency crises, this hypothesis focuses on the distorting features of implicit or explicit guarantees. The moral hazard hypothesis of Eichengreen and Hausmann (1999) states that fixed exchange rates are a form of an implicit guarantee. Investors do not hedge their foreign currency position since the government promises a fixed exchange rate. In addition, capital flows become more short term since the peg is more credible in the short run. This results in unhedged short term foreign currency denominated liabilities which creates financial fragility. More flexible exchange rates are a solution to this problem since they provide an incentive to hedge foreign currency positions. Dollarization or a currency union are mentioned as well as a solution since it removes the probability of exchange rate movements, which were not foreseen and thus not hedged by investors.

In addition to these generation models, there are other models as well that explain currency crises. Eichengreen et al. (1996) were among the first who performed a broad econometric analysis of currency crises caused by contagion. They argue that various forms of financial interdependence exist between countries that can trigger contagion from one country to another. The literature distinguishes two different ways in which contagion can occur: ‘real’ contagion and information-based contagion. Real contagion takes place through trade linkages. Gerlach and Smets (1995) show that when a country is hit by a speculative attack this will boost the competitiveness of that country at the expense of the country with which it directly trades. When the level of reserves is low in this neighbouring country the exchange rate peg can come under pressure through this contagion effect. Information-based contagion arises because of imperfect information. When investors are not properly informed about the economic structure of the neighbour country and the physical exposure to each other, this might incorrectly affect the credibility of this country and can lead to a speculative attack (De Bandt and Hartmann, 2000). In addition, Baig and Goldfajn (1999), Fratzscher (2003), and Caramazza et al. (2004) show that contagion effects are import in explaining the Latin American crisis of 1994 and the Asian crisis of 1997.

foreign currency liabilities, foreign investors might lose confidence in the domestic bank and they want to withdraw their foreign currency. When the commercial bank has not enough foreign reserves to pay out the holders of the foreign currency deposits, the national bank can sell foreign reserves to the commercial bank. The central banks performs its function of lender of last resort. However, the foreign reserves, with which the central bank defends its peg are being depleted. This will increase the likelihood of a speculative attack and a currency crisis. In order to defend the peg the central bank has to raise the interest rate, which will in turn worsen the problems of the commercial banks (De Grauwe, 2009).

Other models of currency crises focus on herd behaviour as an explanation for currency crises (Pilbeam, 2006). See for example Masson (1998) and Calvo and Mendoza (2000).

2.3

T

O

C

E

R

R

nowadays China still has capital controls in order to keep the Chinese renminbi pegged to the US dollar while following an independent monetary policy (Pilbeam, 2006).

However, since the beginning of the 1990s many countries have liberalized their capital movements. Free movement of capital becomes more important since countries financially integrate more and more with each other. For developing countries free movement of capital is important in order to enhance development (De Grauwe, 2009). So, when a country has an open capital market and it wants to have a fixed exchange rate, it cannot pursue an independent monetary policy. In this situation authorities are tempted to devalue the domestic currency by surprise in order to create some monetary autonomy. This leads to problems of credibility. Governments prefer some monetary autonomy, for example, to stabilize unemployment, output, or to finance the government budget deficit. First and second generation models of currency crises underline the weights governments attach to domestic objectives. The more weight the government attaches to these objectives, the higher the incentives of the government to devalue the exchange rate by surprise (De Grauwe, 2009). Kydland and Prescott (1977) and Barro and Gordon (1983) have illustrated these problems with reputation and credibility. As demonstrated in first generation models, fixed exchange rates can lose their credibility because the economic fundamentals are not consistent with the current peg. In second generation models fixed exchange rates can lose credibility when high expectations of devaluation leads to high costs of defending the peg at the expense of domestic objectives. When a peg is not credible anymore, speculators will foresee that the authorities will have to devalue the currency in the future. Therefore, they attack the currency and speculators will sell their domestic currency for the foreign currency in anticipation of an devaluation. This leads to a downward pressure on the exchange rate. If the attack is fierce enough the government has to devalue the domestic currency and a currency crisis occurs (Dooley, 1998). In addition, a world with increased liberalization of capital movements makes fixed exchange rates more fragile. The reason for this is that the expectation of an devaluation leads to a large capital outflow and an increase in the domestic interest rate. This will raise the cost of defending the peg and, therefore, the probability that the authorities will defend the peg will decrease. The peg will be become less credible and the fixed exchange rate regime will be more sensitive to speculative attacks.

speculative attacks. This ‘target zone’ model shows that when the exchange rate comes to close to, for example, the upper limit of the band, speculators know that the authorities will prevent the exchange rate to move above the band. So the closer the exchange rate comes to its upper band, the greater the probability that the exchange rate will decline in the future. Speculators will anticipate to this and will sell the currency resulting in a decline of the exchange rate. In this case, speculation would be stabilizing (De Grauwe, 2009).

One often mentioned solution to the credibility problem of fixed exchange rates is called the bipolar view. The bipolar view states that countries should either allow for more flexible exchange rates or should adopt a hard peg such that speculative attacks are not possible anymore (De Grauwe, 2009). The bipolar view includes monetary unions, currency unions, dollarized countries, and currency boards in the definition of a hard peg (Fisher, 2001). Proponents of this bipolar view are, among others, Eichengreen (1994) and Fischer (2001). They argue that over time countries will adopt either a flexible exchange rate or a hard peg.

currency unions completely from the credibility problem. Whether a currency union is more credible, or less sensitive to speculative attacks and currency crises than countries with a fixed exchange rate will be examined in Chapter 3 and Chapter 4.

Another form of a hard peg is a currency board. A currency board is characterized by a fixed exchange rate and the local currency is 100 percent backed by the central bank’s holdings of the foreign currency. So, when the central bank wants to increase the domestic money supply, it has to increase their holdings of foreign currency by an equivalent amount. In addition, the central bank is not allowed to hold domestic currency denominated assets so that it cannot monetize fiscal deficits by purchasing governments bonds. Proponents of the bipolar view believe that a currency board is resistant to speculative attacks since the local currency is 100 percent backed by holdings of foreign currency. The domestic central bank cannot run out of reserves with which it defends the currency. However, since a currency board is based on a peg there is still the possibility that this peg will be altered (Pilbeam, 2006). The financial crisis of Argentina in 2002 illustrates this. Argentina had pegged its peso to the US dollar by means of a currency board, however, because of the strength of the US dollar Argentina lost its competitiveness and it ran large current account deficits. Increasing concerns about the sustainability of the peg raised speculative pressure on the peso and in 2002 the peso was forced to float (Pilbeam, 2006).

A drawback of currency unions and currency boards is that countries cannot follow an independent monetary policy, therefore, they cannot determine the quantity of money in circulation, the short term interest rate, or rate of inflation. They lose a policy instrument to influence economic activity. Additional disadvantages for currency boards are that central banks cannot lend to domestic banks in times of crises since they are not allowed to hold domestic securities. In addition, the central bank is not able to stimulate the economy in times of recession since it is not able to perform a surprise devaluation or finance fiscal deficits (Pilbeam, 2006).

2.4

M

E

R

R

classification, has been a major challenge for empirical analyses of exchange rate regimes (Bubula and Ötker-Robe, 2002).

There are different ways to measure or classify an exchange rate regime. Originally, most empirical studies relied on the official exchange rate classification of the IMF. The IMF classified their member countries according to a country’s official, or de jure, notification to the IMF. This classification recognized three categories: pegged exchange regimes, regimes with limited flexibility, and regimes with more flexibility. This de jure classification summarized the exchange rate regimes for all IMF member countries from 1975 till 1998 and are reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions. Since 1999, the IMF changed its de jure classification to a so called de facto classification. The main reason for this was the discrepancy between the exchange rate regime the countries claimed to have and their actual exchange rate regime policies. For example, countries who claimed to have a pegged exchange rate regime devalued their currencies frequently, or floating exchange rate regimes appeared to float in a band. To determine the de facto arrangements the IMF takes into account information on a country’s real exchange rate, monetary policies, and foreign reserves (Rogoff et al., 2003). A second drawback of the de jure classification was the categorization into three broad categories. For example, rigid forms of pegs and soft pegs were placed in the same group. However, by doing this one disregards the fact that both types of pegs have different degrees of monetary autonomy (Bubula and Ötker-Robe, 2002). The de facto classification now distinguishes between eight different categories. In 2009 the IMF again modified its de facto classification. These modifications are not a fundamental change of the system like in 1999, however, it comprised changes in the categorization of the different regimes (Habermeier et al., 2009).

others by taking into account multiple or parallel markets and, in addition, they focus on market-determined rates instead of official rates. When there are multiple or parallel markets, the actual regime of a country might be different from its official label. In addition, the official rate can be far removed from the rate at which transactions take place (Reinhart and Rogoff, 2002).

This study will make use of the natural classification of Reinhart and Rogoff (2002) and the update of this by Ilzetki et al. (2008). This classification is the most accurate since it is a de facto classification instead of a de jure and it takes into account multiple or parallel markets and market-determined rates.

III. C

A

This chapter will examine the hypothesis on basis of a comparative analysis. The theoretical background in chapter two discussed various models of currency crises. These models illustrate several factors that can trigger a currency crisis in a monetary arrangement like the level of foreign reserves, fiscal and monetary policies that do not comply with the peg, and a banking crisis. This chapter will compare the differences between these effects in a currency union and a fixed exchange rate. In addition, the costs of leaving a monetary arrangement are important as well since the higher the costs of leaving a monetary arrangement, the more credible this arrangement is and, thus, the lower the probability of a speculative attack. The degree of political integration also is an important determinant for the credibility of currency unions (Disyatat, 2001). The degree of political integration depends on the extent to which countries have transferred sovereignty to the supranational level of the union and in which areas they have transferred sovereignty (De Grauwe, 2009). Since the definition of a currency crisis is twofold - an attack to force a country’s departure from a currency union or an attack on the entire union – the features that influence the probability of a currency crisis can be divided into characteristics that influences the probability of a speculative attack on one country to force its departure from a currency union and into characteristics that influences the probability of an attack on the union as a whole.

3.1

F

G

M

What is the difference between the level of foreign reserves in a currency union and in a country with a fixed exchange rate? On the one hand, since a currency union is composed of several individual countries it has a bigger balance sheet than most individual countries. Therefore, it has more foreign reserves than an individual country with a fixed exchange rate. A currency union can combine all the individual countries reserves. However, the degree of integration in a currency union is important for the combination of foreign reserves. For example in the EMU, national central banks are obliged to contribute only a share of their foreign exchange reserves to the ECB. National central banks are still owner of these reserves, however, the ECB is authorized to manage the reserves (Scott, 1998). So, the more the members of currency union are politically integrated, the more they are willing to combine their foreign exchange reserves and the better the union can defend the exchange rate. On the other hand, OCA theory states that it is optimal to form a currency union of countries who trade intensively with each other. Therefore, currency unions are in general more closed than the individual countries it is composed of and this can depress the level of foreign reserves. For example, the individual countries who are in the EMU can be considered as relatively small and open, however, the EMU as a whole is a large and more closed economy (ECB, 2004). In general, the level of foreign reserves will increase when a country joins a currency union since the effect of the combination of all individual balance sheets is bigger than the effect that a currency union is often more closed than the individual countries. On the contrary, one should bear in mind that critics of first generation models criticize the crucial role of foreign reserves to defend the peg. It is argued that governments could also defend the currency by raising the domestic interest rate. Selling the domestic currency will, in that case, lead to losses for speculators if they borrowed large amounts of domestic currency.

union instead of an attack on the entire union. The reason for this is that fiscal discipline affects the relative competitiveness of the countries in a currency union and might lead to discrepancies between them. If one country is not in line with the other countries a speculative attack can arise to force that country out of the currency union.

Which effect is larger, the moral hazard effect or the no-monetization effect, is being debated in the literature. Beetsma and Bovenberg (1998) and De Grauwe (2009) argue that the no-monetization effect is larger and, thus, a currency union increases the fiscal discipline of countries compared to a ‘stand-alone’ country with a fixed exchange rate. In addition, Restoy (1996) argues that fiscal discipline is dependent on the level of debt. When the level of debt is moderate the market mechanisms are relatively powerful in preventing countries to take on excessive debt. However, when countries have a large amount of debt these mechanism are relatively weak. Finally, critics of first generation models argue that rational governments would not pursue fiscal policies that are inconsistent with maintaining a fixed exchange rate, since they can anticipate that this might lead to a speculative attack and currency crisis (Pilbeam, 2006). In addition, Aizenman (1992) examines the relationship between fiscal discipline and the degree of political integration in a currency union. He reaches the conclusion that currency unions with decentralized fiscal budgets generate excessive public spending. When countries are more politically integrated the more fiscal budgets will be centralized. The more fiscal budgets are centralized the more the decision making will take place on the supranational level which will lower the discrepancies between fiscal and monetary policies and the commitment to a fixed exchange rate.

pressure was not as pronounced in the non-peripheral countries the peripheral countries faced loss of competitiveness (Lane, 2006). This decline in interest rates and the associated lending boom in, for example, Greece and Portugal indicates as well that the moral hazard effect might be underestimated in the literature. Further, in the spring of 2010 it appeared that Greece’s budget deficit was more than twice as high as was announced earlier and did not comply with the Stability and Growth pact (Reuters, 2010). At last, the ‘no-bailout’ clause has lost credibility since the EU stepped in to help Greece. In the spring of 2010 the 16 countries of the EMU decided to raise a 500 billion euro fund to help countries who cannot pay their debts anymore (NRC Handelsblad, 2010).

Table 1: Real interest rates: Pre- and post-EMU.

Countries 1993-1998 1999-2004 Differential Germany 2,5 1,7 -0,8 Austria 2,5 1,2 -1,2 Netherlands 2,1 0,7 -1,4 Belgium 3,1 1,2 -1,9 Finland 3,7 1,6 -2,1 Luxembourg 3,1 0,9 -2,2 France 3,9 1,4 -2,4 Italy 4,9 0,8 -4,1 Spain 4,3 0,0 -4,3 Portugal 4,7 0,2 -4,5 Ireland 4,5 -0,6 -5,1 Greece 7,4 1,0 -6,4 Source: Lane (2006).

no-monetization effect in the EMU. In addition, the higher the level of political integration the lower the probability of a currency crisis. The higher the level of political integration the more currency union members are willing to combine their foreign exchange reserves and the better the union can defend the exchange rate or mitigate abrupt exchange rate misalignments. Besides, when countries are more politically integrated the more fiscal budgets will be centralized. The more decision making will take place on the supranational level which will lower the discrepancies between fiscal and monetary policies and the commitment to the ‘irrevocably’ fixed exchange rate associated with a currency union.

As stated before, the features that influence the probability of a currency crisis – the level of foreign reserves and fiscal discipline – can be divided into characteristics that influences the probability of a currency crisis one country to force its departure from a currency union and into characteristics that influence the probability of an attack on the union as a whole. The level of a country’s foreign reserves influences the probability of an attack on the entire union since reserves will be used to defend the exchange rate of the common currency with regard to other currencies. Foreign reserves cannot be used to restore competitiveness misalignments between the currency union members. In contrast, fiscal discipline affects the probability of a speculative attack to force one country out of the currency union instead of an attack on the entire union since fiscal discipline affects the relative competitiveness of the countries in a currency union and might lead to discrepancies between them. If one country is not in line with the other countries a speculative attack can arise to force that country out of the currency union.

3.2

S

G

M

target which raises the cost of defending the peg and increases the probability on a speculative attack. If countries in a currency union or fixed exchange rate regime are hit by a symmetric shock, this shock will affect the economies in the same way. In both monetary arrangements, countries can take the same fiscal and monetary measures in response to the shock, the relative competitiveness of the countries will not be affected and the exchange rate does not come under pressure. However, this is different when countries in a monetary arrangement are hit by an asymmetric shock: their domestic macroeconomic fundamentals react differently. When one country experiences, for example, an increase in demand the other country faces a decline in demand. Therefore, the relative competitiveness between countries will change. The fixed exchange rate or the common currency may be incorrect compared with the underlying economic fundamentals. The fixed exchange rate or the common currency will come under pressure which might lead to a speculative attack or even a currency crisis.

Essential here is the question whether a currency union will increase or decrease the likelihood of asymmetric shocks between the countries. If joining a currency union reduces asymmetric shocks between the member countries, this will reduce the probability of a currency crisis in a union compared to a fixed exchange rate since asymmetric shocks might lead to discrepancies between the relative competitiveness of the currency union members. If one country is not in line with the other countries a speculative attack can arise to force that country out of the currency union.

produce the same sort of products and, therefore, this trade structure leads to a situation in which countries will be affected in the same way by a shock. De Grauwe (2009) argues as well that asymmetric shocks occur less frequently in currency unions than in fixed exchange rate regimes, however, he focuses on the fact that more integration leads to national borders becoming less important and, therefore, agglomeration will more and more cross national borders and countries will be hit by symmetric shocks. Nevertheless, De Grauwe (2009) does not take into account the fact that countries in currency unions might have different legal systems or tax systems that might lead to concentration of an industry in one country. Differences in legal and tax systems can only decrease by political integration between the member countries.

a global process according to the European Commission (2008a). In addition, the level of synchronization in the Eurozone is higher compared to outside countries.

Another determinant of business cycle synchronization has to do with the extent to which countries are politically integrated. The more they are politically integrated the more decision making will be at supranational level instead of national level. This reduces the probability of asymmetric shocks originating from different political and fiscal policies and different institutions, like labour unions. An asymmetric shock can occur if, for example, one country unilaterally decides to moderate wages which will increase this country’s competitiveness position compared to other countries. The same is true, for instance, if one country decides to unilateral lower its taxes. The more countries are integrated, the more decision making will take place on supranational level which will reduce these asymmetric shocks.

Figure 1: Relative unit labour costs in the Eurozone. 80 90 100 110 120 130 140 150 2000 2001 2002 2003 2004 2005 2006 2007 2008 R e a l e ff e c ti v e e x c h a n g e r a te Ireland Italy Greece Spain Netherlands Portugal France Belgium Finland Austria Germany

Source: European Commission (2008b).

Figure 2: Highest marginal corporate tax rate for countries in the Eurozone. 0 10 20 30 40 50 60 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Belgium France Italy Spain Germany Luxembourg Finland Netherlands Austria Greece Portugal Ireland

Source: World Bank (2010a).

competitiveness of the currency union members. If one country is not in line with the other countries a speculative attack can arise to force that country out of the currency union.

3.3

B

C

Another channel through which currency crises can occur is a banking crisis. Suppose a bank has foreign currency activities. When this bank makes losses on the foreign currency liabilities, foreign investors might lose confidence in the domestic bank and they want to withdraw their foreign currency. When the commercial bank has not enough foreign reserves to pay out the holders of the foreign currency deposits, the national bank can sell foreign reserves to the commercial bank. The central banks performs its function of lender of last resort. However, the foreign reserves, with which the central bank defends its peg are being depleted. This will increase the likelihood of a speculative attack and a currency crisis. In order to defend the peg the central bank has to raise the interest rate, which will in turn worsen the problems of the commercial banks. Hungary and Latvia experienced a similar crisis in 2008. The banks in these countries had expanded their foreign currency activities. However, when the world started to deleverage in 2007 banks faced a liquidity crisis. The central banks of these countries could not issue foreign currency and were not able to perform their function of lender of last resort (De Grauwe, 2009). The IMF and EU had to provide financial assistance to avoid speculative attacks on the Hungarian forint and Latvian lat (IMF, 2009a).

union. The reason for this is that a currency union has more reserves than a ‘stand-alone’ country and is, therefore, more able to perform its function of lender of last resort. Since this higher level of foreign reserves, the reserves will be depleted less quickly than in ‘stand-alone’ countries and therefore reduces the probability of an attack on the entire union.

3.4

C

L

M

A

This section will compare the costs of leaving a currency union with the costs of abandoning a peg. The higher the costs to renege on a monetary arrangement the more credible this monetary arrangement is (Stasavage and Guillaume, 2002). In addition, the higher the credibility of a currency union, the lower the probability of a speculative attack to force one country out of the union. Menoncin and Tronzano (2005) argue that there are two different costs aspects when leaving a monetary arrangement: the foregone benefits of being in the monetary arrangement; and the operational exit costs. This section will focus on the differences between these costs for currency unions and fixed exchange rate regimes.

regime. In return, when the probability of a currency crisis is lower, this will increase the cost of leaving a currency union with respect to the costs of leaving a fixed exchange rate regime. This will lower the probability of a currency crisis in a currency union, and so on.

Secondly, operational and political exit costs determine the costs of leaving a monetary arrangement. The operational cost of leaving a currency union differs from the cost of abandoning a fixed exchange rate in several ways. To examine these differences one has to consider political costs of leaving a monetary arrangement, the technical costs of introducing a new currency against abandoning a fixed exchange rate, and the threat of financial crises (Eichengreen, 2007).

To begin with, leaving a monetary arrangements entails political costs. These political costs differ between currency unions and fixed exchange rates for several reasons. When a currency union member leaves the union it reneges on the bilateral or multilateral agreement it made with the other member countries. However, this is different for countries with a fixed exchange rate: most fixed exchange rates are not based on bilateral agreements but on an agreement of one country to peg its exchange rate, for example, to the US dollar to create monetary credibility. Reneging on this type of agreement leads to lower political costs. In addition, Stasavage and Guillaume (2002) argue that the political costs of leaving a monetary arrangement are higher the more the countries are integrated on other policy areas like trade, finance, or security. According to them, reneging on the monetary arrangement leads to losses in these areas as well. In general, currency unions tend to cooperate more on different areas than sovereign countries and therefore, their political costs will be higher than countries with a fixed exchange rate.

Subsequently, abandoning a currency and introducing a new one entails substantial technical and legal costs. Computers have to be programmed and notes and coins have to be distributed around the country. In addition, Scott (1998) argues that the cost of introducing a new currency also depends on the preservation of the national central bank, the preservation of national payment systems and national debt instruments, and to the extent the pooling of foreign exchange reserve is limited. The more a currency union is economically and politically integrated, the more national central banks and debt instruments are abolished and the more complicated a break-up will be.

be redenominated in the new currency. Since it takes time to arrange and execute the transition, economic agents will anticipate that their contracts will be redenominated and the new currency will lose value against the old currency. Households and firms would deposit their money in other currency union member countries and a bank run would follow. Government bonds would lose value as well and investors will sell their bonds. This could lead to bank runs and bond-market crises (Eichengreen, 2007).

common currency and the benefits of a fixed exchange rate is an endogenous feature as well. Finally, the operational costs introducing a new currency are higher than the costs of abandoning a fixed exchange rate. The reason for this is that, firstly, currency unions tend to cooperate more on different areas than sovereign countries and therefore, their political costs will be higher than countries with a fixed exchange rate. Secondly, abandoning a currency and introducing a new one entails substantial technical and legal costs that are higher than abandoning a fixed exchange rate. In addition, introducing a new currency with the threat of devaluation might lead to severe financial distress. Whether these costs of leaving a currency union are prohibitively costly is ambiguous. Eichengreen (2010) argues that leaving the Eurozone is not an option since it will trigger severe financial crises. However, examples of currency unions exits indicate that entering a currency union is reversible. On the other hand, this might depend on the financial situation in the currency union. If there is a severe threat of devaluation before the exit, leaving a currency union might be very costly.

3.5

C

Models of currency crises illustrate several factors that can trigger a currency crisis in a monetary arrangement. This chapter has compared the working of these channels in currency unions and in fixed exchange rate regimes. From this analysis one can conclude that being a member of a currency union will reduce the probability of a speculative attack and a currency crisis.

a debate ongoing which effect is larger, however, the Stability and Growth pact, the ‘no-bailout’ clause, and the lending boom in Greece and Portugal indicate that in the EMU the moral hazard effect might be larger than the no-monetization effect. The synchronization of business cycles in currency unions affects the probability of a speculative attack to force one country out of the currency union instead of a speculative attack on the whole currency. Thirdly, countries in a currency union tend to have more synchronized business cycles than sovereign countries who have pegged their exchange rate. The more business cycles are synchronized the less countries are hit by asymmetric shocks and the lower the probability of a speculative attack to force one country out of the currency union. Finally, the costs of leaving a monetary arrangement are higher for a currency union than for a fixed exchange rate since the foregone benefits of leaving a currency union are higher than for a fixed exchange rate and the operational costs of introducing a new currency are higher than abandoning a fixed exchange rate peg. These characteristics influence the probability of a speculative attack on one country to force its departure from a union.

Taking these characteristics of political integration into account, it appears that the Eurozone is not significantly integrated. The members of the Eurozone have transferred a substantial part of their sovereignty in agriculture, competition policy, and external trade policy to the supranational level (De Grauwe, 2009). However, there are substantial differences in, for example, wage and tax policies and the Eurozone has not centralized its national budgets and cannot use its automatic stabilizers to alleviate the effects of asymmetric shocks. Besides, the members of the Eurozone are obliged to contribute only a share of their foreign exchange reserves to the ECB (Scott, 1998).

IV. E

A

There is a growing literature on currency crises and the role of currency unions to prevent these crises. Eichengreen (1994), Krugman (1997), and Coulibaly (2009) argue that currency unions reduce the likelihood of a currency crisis. However, the empirical literature is limited. Coulibaly (2009) is the first who performed an empirical study to test whether a currency union can prevent a currency crisis. He uses data from 192 countries over the period 1970-1999 to estimate a probit model. The results of this probit model indicate that becoming a member of a currency union reduces the probability of a currency crisis with three percent. However, Coulibaly (2009) does not differentiate between other exchange rate regimes like fixed or floating regimes. This study will differ from Coulibaly (2009) since it explicitly focuses on countries that move from a fixed exchange rate to a currency union. In order to test this, a probit model is estimated using panel data of 193 countries from 1970 until 2000. The results suggests that for a country that moves from a fixed exchange rate to a currency union the probability of a currency crisis reduces with 4.44 percent. Section 4.1 will explain the probit model that will be estimated. The data will be explained in Section 4.2. The results of the probit estimation will be discussed in Section 4.3. Section 4.4 will test for robustness of the results. Finally, Section 4.5 will give a conclusion.

4.1

P

M

Prob(CCit =1)=Φ(γCUit +β'Xit) Prob(CCit =0)=1−Φ(γCUit +β'Xit) where 2 ) ' ( 2 2 1 ) ' ( it it X CU it it X e CU β γ π β γ + − = + Φ

CCi is 1 if country i experiences a currency crisis and 0 otherwise. CUi is the exchange rate regime variable which takes on the value 1 if country i is a member of a currency union and 0 if country i has a fixed exchange rate regime. In this manner, the effect on the probability of a currency crisis can be estimated when a country with a fixed exchange rate becomes a member of a currency union. γ represents the coefficient on the currency union binary variable and is therefore the parameter of interest. A negative sign of γ is expected since a significant negative value implies that membership of a currency union instead of having a fixed exchange rate is associated with a lower probability of currency crises. Xi is a set of independent variables explaining currency crises. The constant term is also included in Xi. The choice of the independent variables in Xi explaining currency crises is based on the literature and data availability. In accordance with Kaminsky et al. (1998) and Coulibaly (2009) the independent variables are banking crisis, import growth, inflation, M2(broad money) to total reserves, deposit interest rate, and government expenditure to GDP.

individuals rush to the bank to convert their domestic currency in foreign currency the bank will be less able to do this. Therefore a higher ratio of M2 to total reserves is associated with a higher probability of a currency crisis. Finally, according to first and second generation models, higher government expenditure results in a higher probability of a currency crisis. Import growth and reserves to import are expected to have a negative sign. Kaminsky and Reinhart (1999) argue that the poor performance of the terms of trade preceding a currency crisis erodes the purchasing power which lowers imports. However, Coulibaly (2009) argues that low imports are likely to be the result of a currency crisis than the cause. Finally, according to Gosh et al. (2000), Disyatat (2001), and Coulibaly (2009), the lower the reserves, the less governments are able to defend the peg or to mitigate abrupt exchange rate misalignments which increases the probability of a currency crisis.

4.2

D

fixed exchange rate regimes. Examples are Australia, Bangladesh, Barbados, Cyprus, Madagascar, and Portugal. See Table A2 in the Appendix for an overview. Section 4.3 will discuss the effects of this on the results. Finally, Table 2 illustrates a list of independent variables included in the probit estimation. The data for the other independent variables are obtained from the World Development Indicators 2010 of the World Bank (2010a) and from Glick and Hutchison (2000). Table A3 in the Appendix provides an overview of the used variables and data resources.

4.3

R

This section will estimate the probit model as described in Section 4.1. The model is estimated using maximum likelihood and the results of the probit model are illustrated in Table 2. Since a probit function is not a linear function, its coefficients are not easily interpretable. To examine the effect of a one unit increase in a variable one has to consider the marginal effects by taking the derivative of the probit function. For calculations of marginal effects for probit models see Hill et al. (2001). The coefficients as well as the marginal effects and the z-statistics are demonstrated in the table. As can be seen from Table 2, the transition from a fixed exchange rate regime to a currency union reduces the probability of a currency crisis with 4.44 percent. This result is significant at the one percent level. Banking crisis, import growth, reserves to imports, and deposit interest rate all have the expected sign as discussed in Section 4.1 and show a significant relation with currency crises. Government expenditure shows a significant negative relationship, however, a positive relationship was expected. In addition, there appears to be no statistically significant relationship between inflation, M2 to total reserves, and money growth.

addition, Coulibaly (2009) includes currency unions in his sample that Reinhart and Rogoff (2002), Ilzetki et al. (2008), and the de facto classification by Gosh et al. (2002) classify as a currency board or as a fixed exchange rate regime. This might give a downward bias to the result since a part of the fixed exchange rate regimes are classified as currency union. In this case, the currency union variable of Coulibaly (2009) also includes fixed exchange rate regimes, who are associated with a higher probability of a currency crisis. Therefore, the difference between the probability of a currency crisis will be smaller between the different exchange rate regimes.

Table 2: Probit model estimates.

Variable Coefficient dФ(x)/dx z-statistic Exchange rate regime -0.8464 -0.0444*** -3.0632 Banking crisis 0.6529 0.0329*** 2.6437 Import growth -0.0204 -0.0011*** -2.7500 Inflation -0.0010 -0.0001 -0.4963 M2 to total reserves -0.0021 -0.0001 -0.3598 Reserves to imports -0.0120 -0.0006** -1.9810 Deposit interest rate 0.0627 0.0033*** 3.5538 Money growth -0.0029 -0.0001 -0.7238 Government expenditure to GDP -0.0371 -0.0019* -1.6519

Observations 536

Note: (*), (**), and (***) denotes coefficients significant at the 0.01, 0.05, and 0.10 level respectively.

heteroskedasticity. When the model exhibits heteroskedasticity the constant variance assumption is violated. The constant variance assumption implies that for each value of the independent variable the variance of the dependent variable is the same. So, for each value of the independent variable it is equally uncertain how far the values of the dependent variables deviate from their mean value (Hill et al., 2001). The model is being tested for heteroskedasticity by means of a LM test using the artificial regression method described by the Eviews 6 User’s guide II (2006). On basis of the LM statistic of 14.8368 with a p-value of 0.0955 the null-hypothesis of homoskedasticity cannot be rejected at a one and five percent significance level. Finally, a test for multicollinearity is performed using the method of ‘auxiliary regressions’ as described by Hill et al. (2001). Multicollinearity exists when the variables move together in a systematic way. This means that a large part of the variation in one variable is explained by the variation in the other independent variables. When this is the case it is not possible to interpret the parameters of interest separately (Hill et al., 2001). As can be seen in the table, no R2 _{value is above 0.80 and therefore there is no reason to suspect} multicollinearity.

Table 3: model specification tests.

p-value H0: slopes are zero

LR statistic (χ2 ) 46.9517 0.0000 Goodness-of-fit tests Hosmer-Lemeshow statistic (χ2) 6.8044 0.5579 Andrews statistic (χ2) 114.3939 0.0000 Heteroskedasticity LM test statistic 14.8368 0.0955 Multicollinearity

Exchange rate regime R2 0.3083 Banking crisis R2 0.0691 Import growth R2 0.0506 Inflation R2 0.3536 M2 to total reserves R2 _0.2166 Reserves to imports R2 0.3385 Deposit interest rate R2 0.3857 Money growth R2 0.3471 Government expenditure to GDP R2

4.4

T

R

This section will perform several robustness tests. The results are displayed in Table 4. The table demonstrates the coefficients, the marginal effects, and the z-statistic only for the exchange rate regime variable: the variable that measures the effect on the probability of a currency crisis when a country with a fixed exchange rate becomes a member of a currency union. The same independent variables are used in these robustness tests as in the model in Section 4.3. The results for the other variables are not reported. The last column shows the number of observations in the regression. It is being tested whether the results in Section 4.3 are delivered by the level of income of the countries, a specific time period, or region. The model specification tests for the models excluding certain income levels, time periods, or regions are displayed in the Appendix, Table A4 until Table A14.

Firstly, the countries are divided in low, lower middle, upper middle, and high income countries according to the definition used by the World Bank (2010b). The World Bank (2010b) defines low income countries as countries with GNI per capita below 995 US dollars. Lower middle income countries have a GNI per capita between 996 and 9,945 US dollars, whereas upper middle incomes are between 3,946 and 12,195 US dollars. High income countries are defined as countries with a GNI per capita of 12,196 US dollars or more. Table 4 illustrates that excluding countries with different levels of income does not affect the results. In all cases, the marginal effect is still negative and significant.

reject the null-hypothesis that the deviations between expectations and actual values are zero. This means that the model does not sufficiently fit the data.

Table 4: Probit estimates for various samples.

Sensitivities Coefficient dФ(x)/dx z-statistic Observations Excludes low income countries -0.6729 -0.2589* -1.6442 394

Excludes lower middle income

countries -0.9332 -0.0187** -2.3033 330 Excludes upper middle income

countries -0.7384 -0.0522*** -2.5737 419 Excludes high income countries -1.0095 -0.0507*** -3.3432 483 Exclude 1970s -0.8523 -0.0484*** -3.0552 443 Exclude 1980s -0.5476 -0.0332 -1.2091 304 Exclude 1990s -1.4958 -0.0766*** -3.3003 236 Exclude Americas -1.3595 -0.0794*** -3.8387 388 Exclude Asia and Pacific -0.7396 -0.0397*** -2.6145 496 Exclude Europe -1.1397 -0.0517*** -3.6677 511 Exclude Africa1 _{-0.5476 -0.0332 -1.2091 304} Note: (*), (**), and (***) denotes coefficients significant at the 0.01, 0.05, and 0.10 level respectively. 1: the estimation of the model excluding Africa excludes the independent variable deposit interest rate.

4.5

C

IV. C

L

This thesis examined the role of currency unions in preventing currency crises. The hypothesis being tested was: the probability of a currency crisis is, ceteris paribus, significantly lower when a country is in a currency union than when that country has a fixed exchange rate regime. In order to test this hypothesis a comparative analysis and an empirical analysis is performed. Both analyses indicate that moving from a fixed exchange rate regime to a currency union reduces the probability of a currency crisis.

Firstly, the comparative analysis compared factors that trigger currency crises in a fixed exchange rate with these factors in a currency union. Models of currency crises illustrate several factors that can trigger a currency crisis in a monetary arrangement. The comparative analysis has compared the working of these channels in a currency union and in fixed exchange rate regimes. The results indicates that becoming a member of a currency union increases the level of foreign exchange reserves, leads to less asymmetric shocks between countries, and leads to higher costs of leaving the monetary arrangement. These characteristics reduces the probability of a currency crisis. The more the countries of the currency union are politically integrated, the stronger these effects are. On the other hand, becoming a member of a currency union leads to less fiscal discipline which increases the probability of a currency crisis. All things considered, one can conclude that becoming a currency union member reduces the probability of a currency crisis. Moreover, since the definition of a currency crisis in a currency union is twofold, one can divide these characteristics into characteristics that influences the probability of a speculative attack on one country to force its departure from a currency union and into characteristics that influences the probability of an attack on the union as a whole. The level of foreign reserves influences the probability of a speculative attack on the entire union, whereas the level of fiscal discipline, the symmetry of shocks, and the costs of leaving a monetary arrangement influence the probability of a speculative attack to force one country out of the currency union.

the results of both the comparative analyses and the empirical analysis indicate that moving from a fixed exchange rate to a currency union reduces the probability of a currency crisis.

The results of this thesis are consistent with the existing literature on the role of currency unions to prevent currency crises. Eichengreen (1994), Krugman (1997), Sandemann Rasmussen (2002), and Coulibaly (2009) argue as well that currency unions reduce the likelihood of a currency crisis. In addition, the results have implications for academic research as well as for policies concerning financial stability. For academic research, the results lead to an additional argument in the theory of OCA. When the probability of a currency crisis is lower in a currency union, this will increase the benefits of a currency union and this may alter the optimal currency area. In addition,if membership of a currency union lowers the probability of a currency crisis this affects policies for financial stability. The global financial crisis and the currency crisis has led to, among others, a debate on how to improve financial stability in the world. The results of this thesis suggest that moving from a fixed exchange rate to a currency union reduces the probability of a currency crisis which enhances financial stability.

R

Aizenman, J. 1992. “Competitive Externalities and the Optimal Seigniorage.” Journal of Money, Credit, and Banking, 24(1): 61-71.

Baig, T. and I. Goldfajn. 1999. “Financial Market Contagion in the Asian Crisis.” IMF Staff Papers, 46(2): 167-195.

Barro, R.J. and D.B. Gordon. 1983. “Rules, Discretion, and Reputation in a Model of Monetary Policy.” Journal of Monetary Economics, 12: 101-121.

Baxter, M. and M.A. Kouparitsas. 2005. “Determinants of Business Cycle Co-movements: a Robust Analysis.” Journal of Monetary Economics, 52(1): 113-157.

Bayoumi, T. and B. Eichengreen. 1994. One Money or Many? Analyzing the Prospects for Monetary Unification in Various Parts of the World. Princeton N.J.: Princeton University. Beetsma, R.M.W.J. and A. L. Bovenberg. 1998. “Monetary Unions Without Fiscal

Coordination May Discipline Policymakers.” Journal of International Economics, 45: 239-258.

Brooks, C. 2008. Introductory Econometrics for Finance. 2th ed. Cambridge: Cambridge University Press.

Bubula, A. and I. Ötker-Robe. 2002. “The Evolution of Exchange Rate Regimes Since 1990: Evidence From De Facto Policies.” IMF Working Paper WP/02/155. Calvo, G.A. and E.G. Mendoza. 2000. “Rational Contagion and the Globalization of

Securities Markets.” Journal of International Economics, 51: 79-113.

Caprio, G. and D. Klingebiel. 1996. “Bank Insolvencies. Cross-Country Experience.” World Bank Policy Research Working Paper 1620.

Caramazza, F., Ricci, L., and R. Salgado. 2004. “International Financial Contagion in Currency Crises.” Journal of International Money and Finance, 23: 51-70.

CNN. 2010. “Q&A: Greece’s Financial Crisis Explained.”

http://edition.cnn.com/2010/BUSINESS/02/10/greek.debt.qanda/index.html (accessed September 16, 2010).

Coulibaly, B. 2009. ‘Currency Unions and Currency Crises: An Empirical Assessment.’ International Journal of Finance and Economics, 14: 199-221.