Exchange Rate Regimes and Economic Growth: Evidence from Developing Asian and Advanced European Economies Haizhou Huang and Priyanka Malhotra 1 Preliminary. Comments Welcome. September 8, 2004

Abstract The choice of exchange rate regime remains an important issue not only in international finance but also in development economics. While both the theoretical and empirical literatures have attempted to identify the most appropriate regime by examining the effect of regime choice on various macroeconomic and financial variables, neither has been able to provide a conclusive answer. In this paper, we pay special attention to effects of the level of development and use the new classification of de facto exchange rate regime by Reinhart and Rogoff (2004) to investigate the relationship between the choice of regime and the subsequent economic growth rate for developing Asian and advanced European countries. We discover two interesting regularities. First, for advanced European countries, the choice of regime does not seem to affect the rate of economic growth or its variability, although more flexible exchange regimes are associated with slightly higher growth rates. Second, for developing and emerging Asian economies, the choice of regime does affect the economic growth rate non- linearly–managed float outperforms other regimes – but do not affect the variability of growth. Our findings suggest that not only whether but also how the choice of exchange rate regime affects economic growth critically depends on the level of development of that economy.

JEL Classification Numbers: F31; F43; O11; O53 Key words: exchange rate regime, economic growth, economic development, financial Authors’ Email Addresses: [email protected]; [email protected] crisis 1

Research Department, International Monetary Fund, Washington, DC 20431, U.S.A. For helpful comments and suggestions the authors wish to thank Michael Bordo, Eduardo Borensztein, Ping Chen, Peter Clark, Allessandro Zanello and especially Justin Lin. The views expressed are those of the authors and do not necessarily reflect the views of the IMF or IMF policy.

-2-

I. INTRODUCTION With the emergence of the European Union as a single currency zone, the reversion of the East Asian crisis countries towards managing their exchange rates closely with the U.S. dollar soon after their economies began stabilizing, and finally, calls for moving towards a revived Bretton Woods system (Dooley et al, 2003), the choice of exchange rate regime remains an important issue in international finance and development economics. While both the theoretical and empirical literatures have attempted to identify the most appropriate exchange rate regime for an economy by examining the effect of regime choice on various macroeconomic and financial variables, neither has been able to provide an unambiguous conclusion. At the theoretical level, the difficulty lies in resolving the partly offsetting and partly reinforcing relationship between the nominal exchange rate regime and macroeconomic variables. At the empirical level, an additional difficulty lies in accurately classifying the de factoOur exchange paper makes rate regimes. an empirical contribution to this literature by examining the relationship between the choice of the de facto exchange rate regime and the subsequent economic growth rate of a country, with special focus on the level of development. We empirically investigate this relationship using the newly available de facto classification by Reinhart and Rogoff (2004) for 12 developing and emerging Asian economies and 18 advanced European countries from 1976-2001. Comparing the industrial countries in Europe with the developing and emerging market economies in Asia allows us to capture the effect of different levels of development on the link between regime choice and growth. We discover two interesting regularities: the choice of regime does not matter for advanced European economies but is important for developing Asian economies in terms of their economic growth rates. More specifically, for advanced European countries, the choice of regime does not affect the growth rate or its variability, although more flexible exchange regimes are weakly associated with higher growth rates. For developing and emerging Asian economies, however, the choice of regime does matter for economic growth but non- linearly, i.e. fixed and managed float outperform other regimes; however, the regime does not affect the variability of growth. Our findings highlight that not only whether, but also how significantly, the

-3-

choice of exchange rate regime affects economic growth rate critically depends on an economy’s development level. The earliest and leading theoretical foundation for the choice of exchange rate regimes lies in the optimal currency area (OCA) theory, developed by Mundell (1961) and McKinnon (1963). This literature focuses on trade, and stabilization of the business cycle. It is based on concepts of the symmetry of shocks, the degree of openness, and labor market mobility. However, since the links between the nominal exchange rate regime and macroeconomic performance both counterbalance and reinforce each other, the OCA theory is unable to present an unambiguous proposal for the optimal exchange rate regime. For example, according to the theory, a fixed exchange rate regime can increase trade and output growth by reducing exchange rate uncertainty and thus the cost of hedging, and also encourage investment by lowering currency premium from interest rates. However, on the other hand it can also reduce trade and output growth by stopping , delaying or slowing the necessary relative price adjustment process. It is well known that these offsetting effects led Mundell (1961) to suggest a dual currency system for Canada, an eastern and a western Canadian dollars. However, later empirical evidence showed that the transaction costs for trade within Canada are much lower than between Canada and the U.S. (Engel and Rogers, 1996), invalidating In theMundell’s 1990s, a new suggestion. theoretical literature emerged that focused on financial markets and stabilization of speculative financial behavior, particularly for emerging market economies. According to the theory, a fixed regime can increase trade and output growth by providing a nominal anchor and the often needed credibility for monetary policy by avoiding competitive depreciation, and enhancing the development of financial markets (see Barro and Gordon (1983), Calvo and Vegh (1994), Edwards and Savastano (2000), Eichengreen et al (1999), and Frankel (2003) among others). On the other hand, however, the theory also suggests that a fixed regime can also delay the necessary relative price adjustments and often lead to speculative attacks. Therefore, many emerging economies suffer from a “fear of floating,” in the words of Calvo and Reinhart (2002), but their fixed regimes also often end in crashes when there is a “sudden stop” of foreign investment (Calvo, 2003) and capital flight follows, as we saw in the East Asian and Latin American crises.

-4-

Not surprisingly, there is little theoretical consensus on this question of regime choice and subsequent economic growth in the development economics literature as well. While the role of a nominal anchor is often emphasized, factors ranging from market depth (or the lack of it), political economy, institutions and so on often lead to inclusive suggestions as to which exchange rate regime is appropriate for a developing country (Frankel et al (2001), Montiel (2003), Montiel and Ostry (1991)). 2 While the literature in development economics acknowledges the importance of the effects of the level of development to the relationship between regime and growth (see Berg et al (2002), Borensztein and Lee (2002), Frankel (1999), Lin (2001), McKinnon and Schnabl (2003), and Mussa et al (2000) among others), these effects have nevertheless been empirically The empirical under explored. literature on the choice of the optimal exchange rate has concentrated on evaluating the economic performance of different de jure and de facto regimes. While there has been considerable research on the topic using de jure regime classifications, such as the official IMF classification, and the more detailed taxonomy of Ghosh, Gulde, Ostry and Wolf (1997), the strong divergence of de jure regimes from the de facto ones poses a challenge to the findings in these papers and has necessitated a new look at the relationship between regime choice and economic growth using de facto classifications. As Obstfeld and Rogoff (1995) show, most de jure fixed regimes are not de facto fixed. They find that most countries abandon a declared peg within five years or less. Klein and Marion (1997) also show that the mean duration of pegs in the western hemisphere is only 10 months. Complimentary to this evidence is the literature on the “fear of floating” which contends tha t most de jure floating regimes are not de facto floating (Calvo and Reinhart 2002). This is particularly prevalent in emerging economies. For instance, contrary to popular assertion, the Asian crisis countries had not declared dollar pegs pre-1997, although they managed their exchange rates closely in line The with explicit the dollar. advantages of using de facto classifications for studies on the optimal choice of exchange rate regimes raises the issue of the most effective method to 2

For see Bordo (2003) for a historical perspective on exchange rate regime and literature cited therein.

-5-

estimate the actual regime. Two approaches have been used so far 3 , and each one puts forth a different optimal exchange rate regime. First, Levy-Yeyati and Sturzenegger (2002, thereafter LY-S) use statistical methods to evaluate key variables and group observations into four regimes. They demonstrate that floats are optimal. Second, and more recently, Reinhart and Rogoff (2004, thereafter R- R) have used market determined parallel market rates and detailed country specific chronologies of exchange arrangements to create another de facto classification. Based on their new classification, they argue that limited flexibility performs best. To our knowledge, Rogoff et al (2003) is the only paper that extensively analyzes the effect of regime choice on macroeconomic performance using the new R- R classifications for over 150 countries. This paper finds that countries at a relatively early stage of financial development and integration are better off choosing fixed or relatively rigid regimes, while developed countries are better off choosing relatively flexible exchange rate regimes. However, results in this paper do not support the bipolar view of exchange rate regimes (Fischer, 2001). In this paper we investigate the relationship between the choice of de facto exchange rate regime and the subsequent economic growth rate, paying special attention to the effects of the level of development. We use data from both the R- R and LY-S de facto classifications for 12 developing Asian economies and 18 advanced European countries from 1976 to 2001. 4 We focus our study on these particular groups of economies because of the differences in the level of economic development between the two, and our ability to subject the relationship between the two variables to a series of global, regional and count ry-specific shocks. First, prior to the adoption of the euro by 12 of the advanced 3

The approach taken by Ghosh et al (1997) and Ghosh, Gulde and Wolf (2003) is a hybrid of de jure and de facto classifications. 4

Due to its advanced level of development, we exclude Japan from our analysis. However, since Japan has had a flexible regime and low growth rates in the 90s including it does not alter our results for the Asian countries.

-6-

European economies, they had diverse choices of exchange rate regimes, and large volumes of trade among them. If the exchange rate regime has any effect on a country’s economic growth rate, we should be able to capture the effect of the adoption of the euro in our analysis of the European countries. Second, well before the East Asian crisis, in the 1970s some of the Asian countries began growing rapidly, while others did not. As the choice of exchange rate regimes varied immensely amongst these countries, we should be able to capture the effect of regime choice on economic growth just by focusing on the Asian countries. Third, both the Asian economies and the advanced European countries faced a major region-specific shock to their exchange rate regimes in the 1990s. The former were hit by the East Asian crisis, an unexpected shock that had changed the regime of many Asian economies during and after the crisis. In the latter case, twelve countries adopted the same currency in January 1999. This was an ambitious undertaking and largely an anticipated shock. Therefore, if exchange rate regimes have any significant effect on economic growth rates, this effect should be evident around the two specific shocks to the economies in the two regions. Fourth, the effects of financial crises should not bias our results because the East Asian crisis adversely affected several Asian economies but not all, while the Exchange Rate Mechanism (ERM) crisis and the Nordic banking crisis affected some but not all European countries. And finally, our analysis is not biased by the level of regional coordination in either region. Only 15 of our 18 European countries are members of the European Union, of which 12 chose to adopt the euro, while in Asia, only the East Asian economies have strong trade agreements, and closely manage their exchange rates with the dollar. While there exists an institutionalized coordination among euro-area countries, there exis ts formal and informal regional coordination mechanisms in Asia (Frankel and Wei (1993), McKinnon (1999)). In an attempt to evaluate the link between regime choice and economic growth rate in a thorough yet simple empirical framework, we employ both the method used by Alesina and Summers (1993) in examining the relationship between central bank independence and economic performance to form hypotheses, and regressions to verify our hypotheses.

-7-

Our findings highlight the importance of the level of economic development for the choice of the appropriate regime, but do not lend support for the bipolar view of exchange regimes. For the advanced European countries, the choice of regime does not seem to affect economic growth, although more flexible regimes are weakly associated with higher growth rates. For developing and emerging Asian economies, however, the choice of regime does affect economic growth rate but non- linearly – fixed and managed float outperform other regimes. The choice of regime does not affect variability of economic growth rates for either advanced European or developing Asian economies. Our findings suggest that the importance of exchange rate regimes for economic growth depends on the level of development of the economy(ies) in question. As industrial economies in general have more complete markets and in particular, deeper financial markets, they are better able to deal with real and financial shocks, and hence their economic growth rates and their variability do not depend on their choice of exchange regime. Although, a more flexible exchange rate regime can permit an economy to make the necessary adjustments more rapidly and thus on the margin more flexible regimes are weakly associated with slightly higher growth rates. As developing and emerging economies, such as those in Asia, have less depth in their financial markets and more incomplete markets, however, they are less able to deal with real and financial shocks, and hence the choice of exchange rate regime is more important. Despite the partly reinforcing and partly offsetting effects of exchange rate regime choice, credibility effects stand out – fixed and managed float regimes outperform Our results other regimes are largely – and consistent the “fear with of float” Reinhart’s is for real. and Rogoff’s (2004) assertion in several aspects. First, developed countries are better off choosing relatively flexible exchange rate regimes, but we find this effect to be fairly weak. Second, countries at a relatively early stage of financial development and integration are better off choosing fixed or relatively rigid regimes. Our result of exchange rate regime choice having a weak effect on economic growth for advanced economies also lends some support for Baxter and Stockman (1989), who show that business cycles do not seem to be affected by the choice of exchange rate regime.

-8-

The paper is organized in four sections as follows. The following section describes the data set. The key results are shown in section III. Section IV makes some conclusion remarks. II. DESCRIPTION OF DATA In this paper, we create a dataset using two different exchange rate regime classifications and per capita GDP growth rates for 12 developing and emerging Asian economies and 18 advanced European countries for the period 1976-2001.5 For data on exchange rate regimes, we use the two currently available de facto exchange rate classifications: the R- R course classification, and the LY-S classification. 6 While both datasets reclassify the exchange rate regimes for more than 150 countries based on market- determined rates and are significantly different from the official de jure IMF classification, there are significant differences between the two, which are discussed below. The R- R data is our primary classification because it presents a more detailed regime classification, is more closely reflective of the market perceived regime, and finally, the LY-S taxonomy suffers from missing data for some key countries. Nevertheless, we check our results using LY-S data. 7 Our primary source for per capita GDP growth rates is the PPP adjusted per capita GDP series from the World Development The R-Indicators R dataset is 2004, the most whichextensive begins inexchange 1976. rate regime classification available. It differs from all other datasets both in terms of scope and methodology. The dataset spans 153 countries during the 1946-2001 period. In addition, the authors 5

Asian economies: China, Hong Kong SAR, India, Indonesia, Korea, Malaysia, Nepal, Pakistan, Philippines, Singapore, Sri Lanka and Thailand. Advanced European countries: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom. 6

There are no country- year observations in our dataset that have been classified as inconclusive by Reinhart and Rogoff (code 15). However, we have to drop 32 observations from the LY-S series as they are classified as inconclusive. 7

Interestingly, the correlation between the 13-point Reinhart- Rogoff and the 5-point Levy-Yeyati -Sturzenegger classifications is only 0.53.

-9-

differentiate between high inflation regimes and floating regimes. They introduce a new regime category called “freely falling” which describes those instances when the twelve month inflation rate is equal to or greater than 40 percent per annum i.e. a signal of a complete lack of monetary control. This distinction between high inflation floating regimes and low inflation floating regimes is essential and very useful because of the economic distortions introduced by the former. As Reinhart and Rogoff show, plotting a floating regime against the scale of a country undergoing a free fall makes the former look like a fixed regime! Since “freely falling” is not a conscious regime choice by a country, and cannot be classified as any other regime, we choose to drop these observations from our dataset. 8 In addition to extending the detail of exchange rate classifications, the authors also use an innovative methodology, which enables them to capture the “real” regime of a country as closely as possible. In a radical departure from prior classifications, the authors incorporate data on parallel and dual exchange rate markets to estimate the de facto regime. This is essential because as the authors show, market determined parallel and dual market rates are important indicators of monetary policy, and future official exchange rate regime changes. Also, in many instances, the authors find that only few transactions have taken place at the official rate. They use estimates of export misinvoicing going back to 1948 to evaluate the importance of secondary market rates. Another highlight of Reinhart’s and Rogoff’s approach, is the construction and use of detailed historical chronologies of exchange rate arrangements, and related information such as currency reforms, to distinguish between de jure and de facto regimes. While Levy-Yeyati and Sturzenegger also use market data to classify countries’ de facto exchange rate regimes, their dataset is smaller and uses theoretical predictions and statistical methods to classify regimes. The authors reclassify the exchange rate

8

These observations are: Indonesia 1998, Korea 1998, Thailand 1997, and Iceland 1976-1983. To ensure that our findings are not driven by the elimination of these eleven observations, we repeat our analysis with them included and find no change in our results.

- 10 -

regime of 154 countries during the 1974-2000 period into five regime types based on exchange rate behavior in a particular year. The LY-S classification is based on the idea that for each exchange rate regime, there is an expected behavior of exchange rate volatility, volatility of exchange rate changes and the volatility of reserves. For instance, the model flexible exchange rate regime is typified by high volatility in exchange rates but little intervention in the exchange rate market, while an intermediate regime is characterized by relatively high volatility across the three variables. Using the measures of their three characteristic variables, and the K-Means Cluster Analysis procedure, the authors classify each country- year observation into one of the five groups: flexible, dirty floats, dirty/crawling pegs, fixed and inconclusive. Inconclusive observations are then re-classified using the same method. A drawback of this method is that countries which do not exhibit “significant” variability in either variable are classified as inconclusive. While the authors present a sound theoretical argument for the inconclusive category, it decreases the size and variance in our dataset, as we are unable to use these 32 observations in our analysis. 9 In addition, there are two more drawbacks of the LY-S dataset for our empirical exercise: First, the LY-S dataset has missing data (including inconclusive) for important countries such as China (2 observations), Finland (9 observations), Singapore (14 observations), UK (14 observations), Iceland (11 observations), and Norway (13 observations). Out of 31 countries, eleven have 35 percent or more observations missing. Second, as Reinhart and Rogoff explain, by focusing on the behavior of exchange rates in each year individually, the authors are more likely to characterize one time episodes such as one-time repeg or devaluation, or an economic or political shock as a regime change, when there is no regime change at all. In our dataset, we do

9

These observations include: Austria 1981, 1983, 1997, 1998; Belgium 1991, 1994-1998; France 1997; India 1992, 1994, 1999; Indonesia 1994; Italy 1986; Nepal 1999; Netherlands 1985, 1986, 1990, 1997, 1998; Pakistan 1994; Philippines 1977-1979; Sri Lanka 1986, 1990; Thailand 1996.

- 11 -

find that regime changes occur twice as often in the LY-S taxonomy compared with the R- R classification (199 in LY-S vs. 80 in R- R). III. K EY R ESULTS Our findings show that the importance of exchange rate regime choice for economic growth depends on the level of development of an economy. For developing and emerging market countries, there is a non- linear relationship between economic growth and regime choice, with fixed and managed float regimes associated with the highest rates of growth. However, regime choice does not affect the rate of economic growth for the advanced European countries, though more flexible regimes are associated with slightly higher growth rates on average. We also find that the variability of economic growth rates are unaffected by the choice of exchange rate regime. To highlight these significant differences between the optimal regime choice for emerging market and industrialized countries, we present our results in two sub- sections: the first focusing on the Asian countries, and the second, on the European countries. Our empirical strategy is simple and implemented in three stages: We first present scatter plots of the rate of growth of per capita GDP against the R- R course index of exchange rate regimes for European and Asian countries during the period 1976-2001. Next, we investigate this relationship more thoroughly by presenting frequency-weighted plots of the mean and variance of the economic growth rate against the mean exchange rate regime for each country- regime-episode during the entire 19762001 period. Finally, we present regressions of economic growth rate on regime choice controlling for other determinants of economic growth. After examining individual country plots of regime choice, we found that most countries adopt more than one regime for periods of at least three years (see Appendix II for some examples). Since regime choice could be endogenous to the economic and political conditions of a country, we feel that each country- regime-episode should represent a different economic and political entity. Additionally, aggregating solely by country eliminates the variance created by the different regime-episodes for each country, and therefore does not allow for an efficient estimation of the mean growth

- 12 -

effect of regime choice. To create country- regime-episodes, we aggregate the observations by regime-episodes for each country. Regime-episodes are defined as those regimes that a country has chosen for at least three years.10 For instance, if out of 25 country- year observations, Indonesia has 3 years under a freely floating regime, 3 years under a de facto crawling band narrower than +/- 5 percent and 19 years under a de facto crawling peg, then we would have three regime-episodes for Indonesia. Combining our observations by country- regime-episodes give us 82 observations (34 for Asia, and 48 for Europe) instead of the 31 observations we would have if we grouped by country. We then plot the mean and variance of the economic growth rate against the mean exchange regime weighing each country- regime observation by the length of the regime-episode as a fraction of the total number of country- year observa tions for that country. We verify the robustness of our results using the LY-S 4 point classification for the 1976-2000 time period. In this section we only present the plots between the course R- R exchange rate regime classification and rate of growth of per capita GDP (PPP adjusted). The remaining plots are presented in the Appendix I. III.1. Developing and Emerging Asian Economies In figure 1, we plot the rate of economic growth against the exchange rate regime for each country- year observation of the 12 developing and emerging Asian economies in our dataset. As the figure shows, the higher the flexibility of the regime, the lower the growth rate of the economy. Since the slope is not very steep, it remains to be seen if the negative linear correlation between the choice of exchange regime and the economic growth rate would be significant statistically or not.

10

Not necessarily three consecutive years. Also, if a country does not adopt a regime for at least three years over the sample, we do not include that observation in our aggregated data.

- 13 -

-10

Per Capita GDP Growth Rate 0 10 20

30

Figure 1. Asia 1976-2001

1

2 3 Exchange Rate Regime (R-R course classification)

4

In figure 2 we present a frequency-weighted plot of the mean economic growth rate against the exchange rate regime for each country- regime-episode during the entire 1976-2001 period. The size of the points represents the difference in the weights given to each observation. Figure 2 shows that three regimes, namely fixed (regime 1), crawling peg (regime 2) and managed float (regime 3), may be significantly associated with higher growth rates, but perhaps in a non- linear manner, and free floating may be associated with lower growth.

- 14 -

0

Mean Per Capita GDP Growth Rate 5 10

15

Figure 2. Asia, Mean 1976-2001

1

2 3 Mean Exchange Rate Regime (R-R course classification)

4

Table 1 below presents regression results that confirm our hypothesis from figures 1 and 2. The first column on the right hand side shows that the linear relationship between the rate of economic growth rate and the exchange regime is not significant at any reasonable confidence level, pointing out to a possible nonlinear relationship between the two variables. Results in the second column show that the nonlinear relationship does indeed exist. We find that all but the freely floating regime are positively statistically significant, however the magnitudes of their effect on growth differ. Managed float is significant at the 5 percent level and has a coefficient of 9.84, while the more fixed regimes are significant at the 10 percent level with slightly smaller coefficients. Financial crisis (dcrisis) affects growth negatively also at the 1 percent confidence level, and on average it lowers growth rate by about 5.5 percent. In figure 3, we estimate the effect of regime choice on the variability of economic growth using frequency-weighted plot for our country- regime-episodes. Interestingly, we find that more flexible exchange rate regimes are associated with slightly higher variability of economic growth, however, this result is not statistically significant, and therefore we cannot reject the hypothesis that regime choice has no effect on the variability of economic growth rates.

- 15 -

Table 1. Asian Countries regime

0.23 (0.51)

fixed crawling peg managed float freely floating dcrisis openness govt. consumption initial GDP per capita fertility rate secondary school enrollment ratio Constant Observations No. of group(country) R-squared

-5.29*** (1.35) 0.02* (0.01) 0.06 (0.13) -0.00 (0.00) 0.64 (0.48) -9.55** (4.03) 7.59* (3.94) 296 12 0.18

9.05* (4.09) 8.04* (4.06) 9.84** (3.93) 8.27 (4.63) -5.46*** (1.36) 0.02 (0.01) 0.05 (0.13) -0.00* (0.00) 0.54 (0.48) -8.18** (4.09)

296 12 0.29

Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%

- 16 -

0

Variance of per capita growth rate 20 40 60 80

100

Figure 3. Asia, Variance 1976-2001

0

5 10 Exchange rate regime (R-R natural taxonomy)

15

The higher growth rates associated with fixed and managed float regimes help explain why emerging market economies such as those in Asia are reluctant to move towards flexible exchange rate regimes, and exhibit the “fear of floating” phenomenon.

III.2 Advanced European Countries While the choice of exchange rate regime matters for Asian developing economies and managed float and fixed exchange rate regimes appear to be obvious policy choices for them, the choice of regime does not seem to matter for advanced European countries. Figure 4 shows that there may be a positive relationship between a more flexible exchange rate regime and the subsequent rate of economic growth, and once again as the slope is not very steep we need to check the significance and robustness of the relationship using regressions.

- 17 -

-5

Per Capita GDP Growth Rate 0 5 10 15

20

Figure 4. Advanced European Countries 1976-2001

1

2 3 Exchange Rate Regime (R-R course classification)

4

In figure 5, we examine the robustness of this result using frequency-weighted plots of the rate of economic growth rate versus the exchange regime for each countryregime-episode.

2

Mean Per Capita GDP Growth Rate 4 6 8 10

Figure 5. Advanced European Countries, Mean 1976-2001

1

2 3 Mean Exchange Rate Regime (R-R course classification)

4

Table 2 below presents the results of our regressions of regime choice on economic growth rates. The first column on the right hand side shows that there is positive relationship between the rate of economic growth rate and the exchange regime

- 18 -

at the 5 percent confidence level. However, the second column shows that no particular regime has a statistically significant effect on growth. Therefore, we can conclude that on average, more flexible regimes are weakly associated with higher growth rates. Financial crises also have a negative and statistically significant (at the 1 percent level) effect on economic growth.

Table 2. Advanced European Countries regime fixed crawling peg managed float freely floating dcrisis openness govt. consumption initial GDP per capita fertility rate secondary school enrollment ratio Constant

Observations No. of group(country) R-squared

0.88** (0.41) 0.62 (4.45) 1.68 (4.47) 2.87 (4.53) 0.30 (3.53) -2.24 -2.23*** (0.82) (0.83) 0.09*** 0.09*** (0.02) (0.02) 0.05 0.05 (0.05) (0.05) -0.00*** -0.00*** (0.00) (0.00) 0.88 0.74 (0.80) (0.81) -2.13 -2.14 (1.60) (1.62) -0.40 (2.50) 443 18 0.14

443 18 0.14

Standard errors in parentheses * significant at 10%; ** significant at 5%; *** significant at 1%

Finally, in figure 6, we investigate the effect of exchange rate regimes on the variance in the rate of economic growth by examining frequency-weighted plots of data aggregated by country- regime-episodes. Interestingly, we find that the choice of exchange rate regime has no effect on the variability in economic growth rates.

- 19 -

0

Variance of Per Capita GDP Growth Rate 10 20 30

40

Figure 6. Advanced European Countries, Variance 1976-2001

1

2 3 Mean Exchange Rate Regime (R-R course classification)

4

Our findings suggest that for the European countries, more flexible exchange rate regimes are associated with higher growth rates.

III.3 Robustness Checks Despite the shortcomings of the LY-S classification discussed earlier, we find that it does lend some support to our findings, especially for the Asian economies. As figures I.1 and I.2 show, we do find that countries with fixed exchange rate regimes have a higher rate of economic growth than countries with flexible regimes. The inclusion of China, which LY-S would probably classify as a fixed regime during its years of highest growth, the 1990s, and a flexible regime during its lower growth years before that, would further strengthen this result. However, figure I.3 contradicts our finding that the variability in growth rates is unaffected by regime choice. We believe this result could in part be driven by the absence of China. Before the 1990s, when its variability of growth was higher than 30-40 percent China would probably have been classified as a flexible regime, while after that China would have been classified as fixed regime with variability of growth at less than 10 percent. Thus the inclusion of China would probably reverse the LY-S result and confirm our earlier finding. However, without the data one cannot know this for sure.

- 20 -

In the case of Europe, we find support for our result that there is no effect of exchange rate regimes on the variability of economic growth rates, however, we also find that the former has no significant effect on the rate of economic growth. Again, we believe that this might be driven by the large amount of missing data for the UK, Finland, Iceland, Belgium, Norway, and Austria prevents us from conducting a sound analysis for the European countries.

IV. CONCLUDING R EMARKS The choice of exchange rate regime is an important issue in international finance as well as development economics, despite that the existing theoretical and empirical literatures have been unable to provide unambiguous answers on the effect of regime choice on various macroeconomic and financial variables. On the one hand, Baxter and Stockman (1989) show that the choice of regime seems not to affect business cycles. On the other hand existing researchers have shown that regime matters but the results range from fixed works the best (Ghosh et al (1997)), to limited flexibility works the best (Rogoff et al (2004)), and to flexible works the best (Levy-Yeyati and Sturzenegger (2002)). As the links between exchange rate regime and macroeconomic performance are partly reinforcing and partly offsetting, the whole range of the existing results are not surprising but they call for more empirical investigations and better regime classifications. Levy-Yeyati and Sturzenegger (2002) and especially the most recent classification by Reinhart and Rogoff (2004) are important contributions for classifying the de facto regime more precisely, our paper tries to contribute to this literature by empirically investigate the link between the de facto choice of exchange rate regime and economic growth, paying particular attention to the effects of the level of development on the link. While the importance of the level of development on this link has been long acknowledged in the literature, its effects have nevertheless been under explored. Our empirical investigation of the relationship between the choice of de facto exchange rate regime and the subsequent economic growth rate for developing Asian and advanced European economies, highlights two interesting regularities. For the

- 21 -

advanced European countries, the choice of exchange rate regime does not matter for economic growth, although a more flexible exchange regime is weakly associated with a slightly higher growth rate. In contrast, for developing and emerging Asian economies, the choice of exchange rate regime does matter for economic growth rate but nonlinearly. Fixed and managed float regimes are associated with higher growth rates than other regimes that may fall, between or outside of, fixed and managed float regimes in terms of flexibility. Interestingly, the choice of exchange rate regime does not affect the variability of economic growth rates for either advanced or developing economies. Our findings demonstrate that not only whether but also how the optimal choice of regime affects an economy’s growth rate critically depends on the level of development of the economy. Our results suggest regime choice largely does not matter for advanced European economies but developing and emerging Asian economies should pay more attention to their choice of exchange rate regimes, taking into consideration of their the level of development, capital market development, capital account and other important factors.

- 22 -

APPENDIX I: ROBUSTNESS CHECKS Using the Levy-Yeyati -Sturzenegger Exchange Rate Regime Classification 2 = Float; 3 = Dirty Float; 4 = Dirty/Crawling Peg; 5 = Fixed

-10

Per capita growth rate 0 10 20

30

Figure I.1. Asia, 1976-2000

2

3 4 Exchange rate regime (LY-S taxonomy)

5

0

Mean per capita growth rate 5 10

15

Figure I.2. Asia, Mean 1976-2000

2

3 4 Exchange rate regime (LY-S taxonomy)

5

- 23 -

0

Variance of per capita growth rate 20 40 60

80

Figure I.3. Asia, Variance 1976-2000

2

3 4 Exchange rate regime (LY-S taxonomy)

5

-5

Per capita GDP growth rate 0 5 10 15

20

Figure I.4. Europe 1975-2001

2

3 4 Exchange rate regime (LYS taxonomy)

5

- 24 -

2

4

Mean per capita GDP 6 8

10

12

Figure I.5. Europe, Mean 1975-2000

2

3 4 Exchange rate regime (LYS taxonomy)

5

0

10

Variance of per capita GDP 20 30 40

50

Figure I.6. Europe, Variance 1975-2000

2

3 4 Exchange rate regime (LYS taxonomy)

5

- 25 -

APPENDIX II. COUNTRY-REGIME-EPISODES

15

Italy, 1975-2001

Exchange rate regime 5 10

1982

1993 1983

1990 1989 1996

1981 1976

1977

1992 1994 1991

1978

1980 1979

1986 1995

1985 1984

1987 1988

1997 1998

1999

2000

0

2001

0

5

10 Per capita GDP growth rate

15

15

Germany, 1975-2001

1990

1980 1995 1992 1981 1977 1984 1986 1988 1994 1976 1978 1987 1991 1985

1979

Exchange rate regime 5 10

1997 1983 1982 1996 1993 1989 1998

1999

2000

0

2001

0

5 10 Per capita GDP growth rate

15

- 26 -

8

10

France, 1975-2001

Exchange rate regime 4 6

1976 1983

1982

1996 19981990

1978

1985 1984 1981

1986

1995 1992

1980

1979

1987 1988

19911994

2

1993 1997 1989

1977

1999

2000

0

2001

0

5 10 Per capita GDP growth rate

15

12

Sweden, 1975-2001 1996 1998 2001

1994

19991995

2000

Exchange rate regime 6 8

10

1993 1997

1982

1981 1985

1988 1984 1978 1986

1980

1979

1987

2

4

19831991 1977 1976 1992 1989 1990

0

5 10 Per capita GDP growth rate

15

- 27 -

12

China, 1975-2001

Exchange rate regime 8 10

1989

1990

1983 1982 1981

1991

1992 1985 1984 1988 1987

1977 1993

1979

1978 1980

4

6

1976

1986

1998

-5

0

19992001 1997 1996 2000 1994 1995

5 10 Per capita GDP growth rate

15

20

8

India, 1975-2001 1976

1982 1989 1979

1983

1984

1985 1978 1986 1977 1981

Exchange rate regime 4 6

1998 2000 19911997 1990 1999 2001 1996

1987

1988 1980

1995

1994 1992

2

1993

0

5

10 Per capita GDP growth rate

15

- 28 -

12

Philippines, 1975-2001 2001

1999

2000

1994 1993 1982

1983

1981 1976

1978

1980

1995 1977

1979

1992 1988 1985

1991

1990

1989

4

Exchange rate regime 6 8

10

1998

1986

1987

1996

2

1997

-5

0

5 Per capita GDP growth rate

10

15

15

Indonesia, 1975-2001

Exchange rate regime 5 10

2001

1999

2000

1976

1982

1989 19931988 1983 1994 1985 19961990

1991 1995 19921981

1978

1987

1979 1980

1984 1986

0

1997

1977

0

5

10 Per capita GDP growth rate

15

20

- 29 -

12

Singapore, 1975-2001 1999 2000

Exchange rate regime 6 8

10

2001

1997 1985 1983 1989 199219951991 1976 1990 19841981 1982 1996 1994 1986 1977

1993 1978 1988

1987 1979 1980

2

4

1998

-10

0

10 Per capita GDP growth rate

20

15

Korea, 1975-2001

2001

1999

Exchange rate regime 5 10

2000

1989

1997

19821996

1983 1980 1992 1993 1994 1990

1995 1985

1988 1984

1986

1987

1981 1991

1976

1977

1978

0

1979

0

5

10 Per capita GDP growth rate

15

20

- 30 -

10

Hong Kong, 1975-2001

Exchange rate regime 6 8

1982 1981

1978 1976 1977 1979

1980

4

1983

2

1989 1998

2001

-10

1993 1994 19901997 1996 1985 1999 1995 1991

1992 1984

2000 1988

1986

1987

0 10 Per capita GDP growth rate

20

15

Japan, 1975-2001 1998

1983

Exchange rate regime 5 10

20011997

19951991 1989 1982 1994 1990 19931999 19961992 2000

1988 1986

1978

1980 1979 1987

1977

0

1976

1984 1981 1985

-5

0

5 Per capita GDP growth rate

10

15

- 31 -

References Alesina, Alberto and Lawrence H. Summers, 1993. Central Bank Independence and Macroeconomic Performance: Some International Evidence. Journal of Money, Credit, and Banking, Vol. 25, No. 2, pp. 151-162. Barro, Robert and David Gordon, 1983. Rules, Discretion, and Reputation in a Model of Monetary Policy. Journal of Monetary Economics, 12: 101-22. Baxter, Marianne and Alan C. Stockman, 1989. Business Cycles and the Exchange- Rae Regime: Some International Evidence. Journal of Monetary Economics, 23, 377-400. Berg, Andrew, Eduardo Borensztein and Paolo Mauro, 2002. An Evaluation of Monetary Regime Options for Latin America. The North American Journal of Economics and Finance, 13: 213-235. Bordo, Michael, 2003. Exchange Rate Regime Choice in Historical Perspective. NBER Working Paper 9654. Borensztein, Eduardo and Jone- Wha Lee, 2002. Financial Crisis and Credit Crunch in Korea: Evidence from Firm- Level Data. Journal of Monetary Economics, 49(4): 853-75 Calvo, Guillermo, 2003, Explaining Sudden Stop, Growth Collapse and BOP Crisis: The Case of Discretionary Output Tax, The Mundell Fleming Lecture for the Third Annual IMF Research Conference, Washington, DC. Calvo, Guillermo and Carmen Reinhart, 2002. Fear of Floating. Quarterly Journal of Economics, 117, no. 2, May, 379-408. Calvo, Guillermo and Carlos Vegh, 1994. Inflation Stabilization and Nominal Anchors. Contemporary Economic Policy, 12 (April), 35-45. Dooley, M.P., D. Folkerts- Landau and P. Garber, 2003, An Essay on the Revived Bretton Woods System, NBER working paper 9971. Edwards, Sebastian and Miguel A. Savastano, 2000. Exchange Rates in Emerging Economies: What Do We Know? What Do We Need to Know? in Economic Policy Reform: The Second Stage, ed. by Anne O. Krueger, pp. 453-510. Chicago: University of Chicago Press. Eichengreen, Barry, Paul Masson, Miguel Savastano, and Sunil Sharma, 1999. Transition Strategies and Nominal Anchors on the Road to Greater Exchange Rate Flexibility,” Essays in International Finance, No. 213 (Princeton: Princeton University Press).

- 32 -

Engel, Charles and John H. Rogers, 1996. How wide is the Borders? American Economic Review, 86(5): 1112-1125. Fischer, Stanley. 2001 . Exchange Rate Regimes: Is the Bipolar View Correct? Journal of Economic Perspectives 15(2): 3-24. Frankel, Jeffrey, 1999. The International Financial Architecture, Brookings Institution, Washington, DC. Frankel, Jeffrey, 2003. Experience of and Lessons from Exchange Rate Regimes in Emerging Economies, in Monetary and Financial Cooperation in East Asia, Asian Development Bank, Macmillan, 2003. Frankel, Jeffrey A, Eduardo Fajnzylber, Sergio L. Schmukler and Luis Servén, 2001. Verifying Exchange Rate Regimes. Journal of Development Economics 66(2): 351-86. Frankel, Jeffrey and Shang- Jin Wei, 1993. Emerging Currency Blocs. NBER Working Paper 4335. Ghosh, Atish, Anne- Marie Gulde, Jonathan D. Ostry and Holger C. Wolf, 1997. Does the Nominal Exchange Rate Regime Matter? NBER Working Paper 5874. Ghosh, Atish, Anne- Marie Gulde, and Holger C. Wolf, 2003. Exchange Rate Regime: Choices and Consequences. Cambridge: MIT Press. Klein, Michael W. and Nancy P. Marion, 1997. Explaining the Duration of ExchangeRate Pegs. Journal of Development Economics 54(2): 387-404. Levy-Yeyati, Eduardo and Federico Sturzenegger, 2002. Classifying Exchange Rate Regimes: Deeds versus Words. Mimeo, Universidad Torcuato Di Tella. Lin, Justin Yifu, 2001. WTO Accession and Financial Reform in China. Cato Journal, 21(1), (Spring-Summer): 13-18. McKinnon, Ronald, 1963. Optimal Currency Areas. American Economic Review, 53, pp. 717-724. McKinnon, Ronald, 1999. Exchange Rate Coordination for Surmounting the East Asian Currency Crisis. Journal of International Development 11(1): 95-106. McKinnon, Ronald and G. Schnabl, 2003. The East Asian Dollar Standard, Fear of Floating, and Original Sin, in: G. Ortiz, ed. Macroeconomic Stability, Financial Markets, and Economic Development, Bank of Mexico.

- 33 -

Montiel, Peter J, 2003. Macroeconomics in Emerging Markets. Cambridge: Cambridge University Press. Montiel, Peter J. and Jonathan Ostry, 1991. Macroeconomic Implication of Real Exchange Rate Targeting in Developing Countries. IMF Working Paper 91/29. International Monetary Fund. Mundell, Robert, 1961. The Theory of Optimal Currency Areas. American Economic Review, 51, pp. 657-661. Mussa, M., P. Masson, A. Swoboda, E. Jadresic, P. Mauro & A. Berg, 2000, Exchange Rate Regimes in an Increasingly Integrated World Economy. IMF Occasional Paper No. 193, Washington, DC. Obstfeld, Maurice and Kenneth Rogoff, 1995. The Mirage of Fixed Exchange Rates. Journal of Economic Perspectives, Vol. 9, No. 4 (Fall), pp. 73-96. Reinhart, Carmen M. and Kenneth Rogoff, 2004. The Modern History of Exchange Rate Arrangements: A Reinterpretation. Quarterly Journal of Economics, Vol. 119(1), pp. 1-47. Rogoff, Kenneth, Aasim M. Husain, Ashoka Mody, Robin J. Brooks, and Nienke Onmes, 2003. Evolution and Performance of Exchange Rate Regimes. IMF Working Paper 03/243, International Monetary Fund. World Bank, 2004. World Development Indicators 2004. Washington D.C.