Is there an observed tendency for exchange rate regimes to drift to the polar extremes of hard pegs and free floats, with a hollowing of the middle between the two? Have regime changes become significantly more frequent in the post-Bretton Woods era? And have certain regimes historically proven more difficult to sustain, particularly in countries more open to capital flows? Policy debates centered around these questions have forced a growing recognition that the exchange rate regime a country actually operates (its de facto regime) often differs meaningfully from its announced (or de jure) regime. This divergence affects potentially the analysis of historical trends in exchange rate regimes, their macroeconomic performance, and the answers to salient policy questions.
In recognition of the divergence between actual and operational regimes, a number of efforts have been undertaken to develop a classification of de facto rather than de jure regimes. The IMF now publishes regime classifications that take into account the actual functioning of regimes; these are available from 1990, and findings based on this classification are reported in IMF (2003b). The Natural classification, developed by Reinhart and Rogoff (2004), extends back to the 1940s, and overlaps significantly with the IMF de facto classification in the 1990s. The Natural classification also draws analytically useful distinctions that facilitate the interpretation of countries’ economic behavior and performance.
This section describes the evolution of exchange rate regimes across the world using primarily the Natural classification, but provides also comparisons with other (including the de jure) classifications. Below are the main findings.
Historically, the actual operation of exchange rate regimes seems to have differed from the announced framework about 50 percent of the time. Many countries have exhibited a fear of floating; as a result, the actual flexibility of their exchange rate was substantially less than announced.
Intermediate regimes remain prevalent, especially among emerging markets and other developing countries. The so-called “middle” along the flexibility dimension continues to constitute half of all regimes, as it has throughout the past three decades. Freely floating regimes remain rare. The moderate increase in the number of pegs in the 1990s was mainly in the euro area and the transition economies.
The frequency of regime transitions today is similar to what it was 50 years ago. Since 1940, around 7 percent of all countries have changed their regime in a given year, with emerging markets tending to switch regimes more frequently than other countries. Apart from transitions related to major global or regional events in economies experiencing severe macroeconomic stress, changes in de facto regimes in the post-Bretton Woods period have been about as frequent as during the period of fixed parities.
This section also provides a brief discussion of the different approaches to exchange rate regime classification and documents the evolution of regimes across the world from 1940. It considers transitions across regimes, and concludes with some observations of how the choice of a classification system might affect the assessment of the performance of alternate regimes. Throughout the section, differences across economies that are at different stages of development and integration into global capital markets are highlighted by dividing countries into three groups—advanced, emerging market, and other developing economies.3
New Regime Classifications
Until the late 1990s, most empirical studies of exchange rate regimes relied on the de jure regime classification reported in the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions (AREAER), which was then based on countries’ official notifications to the IMF. The de jure classification distinguished between three broad categories—pegged regimes, regimes with limited flexibility (usually within a band or cooperative arrangement), and more flexible arrangements (those with managed or free floats)—which were then divided into 15 subcategories.4
Although comprehensive in terms of country and historical coverage, the de jure classification system had a serious drawback: in practice, exchange rate regimes often differed from what they were officially announced to be. For example, some pegged regimes devalued frequently, while many floats typically moved within a tight band. Consequently, the de jure classification characterized inaccurately the distribution of operative currency regimes across the world and over time. Moreover, empirical analyses employing this classification to test theories of regime choice or to assess the relationship between regime choice and economic performance risked reaching incorrect conclusions and drawing misleading policy implications.5
Recognizing the merits of classifying regimes more realistically, a number of new de facto classification systems have been proposed. Ghosh and others (1997) classify regimes on a de facto basis using information on actual exchange rate movements. Subsequently, the evidence on macroeconomic performance under alternative de jure regimes was reexamined by Ghosh, Guide, and Wolf (2003) by checking the robustness of these results against a hybrid de jure/de facto classification.6 Another classification system, devised by Levy-Yeyati and Sturzenegger (2002 and 2003), discards the de jure classification altogether and instead employs purely statistical techniques to exchange rate and reserves data to determine the de facto flexibility of exchange rate regimes.7 In addition, the IMF itself moved to a de facto classification system in 1999. The IMF de facto classification combines available information on the exchange rate and monetary policy framework and authorities’ formal or informal policy intentions with data on actual exchange rate and reserves movements to reach a judgment about the actual exchange rate regime.8
Despite these advances, analysis sometimes requires a more nuanced characterization of regimes. Countries experiencing episodes of macroeconomic instability often have very high inflation rates, which may be reflected in high and frequent exchange rate depreciation. Classification of such regimes as floating, intermediate, or pegged is problematic because the macroeconomic disturbances could be incorrectly attributed to the exchange rate regime. In addition, in countries with significant parallel foreign exchange markets, where rates differ substantially from official ones, movements in parallel rates rather than in official rates provide a more realistic barometer of underlying monetary policy. In particular, countries with a fixed official rate but with high inflation and a rapidly depreciating parallel rate cannot be considered as having a monetary stance that is consistent with a pegged regime. Moreover, to assess the relationship between regimes and longer-term economic performance, it is helpful to identify longer-term regimes rather than shorter-term spells within a regime, such as the widening of a horizontal band or a onetime devaluation followed by a re-peg. By employing a relatively short horizon over which the de facto regime is assessed, classification algorithms, such as the one employed by Levy-Yeyati and Sturzenegger, can record potentially a large number of regime changes that are related to short periods of disturbances—possibly transient economic or political shocks—and that do not involve a change in the regime itself.
Reinhart and Rogoff’s (2004) Natural classification addresses these shortcomings by separating episodes of severe macroeconomic stress and incorporating information on dual/parallel market exchange rates.9 Their classification distinguishes regimes that are freely falling as a separate category and, in cases where the dual/parallel exchange rate differs substantially from the official rate, uses movements in the former rate to classify the regime. Also, a five-year horizon is used to gauge the true flexibility of the longer-term exchange rate regime. The Natural classification divides de facto regimes into five coarse categories—fixed, limited flexibility, managed floating, freely floating, and freely falling—and into 14 fine subcategories. The Reinhart-Rogoff data set is comprehensive, covering virtually all IMF members, in most cases, back to 1946. Hence, it facilitates richer historical analysis of regime distributions, transitions, and performance than other de facto classifications.10
Some qualifications should be noted, however, with respect to de facto classifications, including the Natural classification. The absence of exchange rate variability that is used to classify regimes may reflect the absence of real shocks to the economy rather than a fixed exchange rate regime. Reinhart, Rogoff, and Spilimbergo (2003) find, however, that countries that have had relatively stable exchange rates have not been subjected to fewer or smaller terms-of-trade shocks.11 Also, de facto classifications are based on past movements of exchange rates as well as other variables. Hence, they are backward looking and do not incorporate information on policy intentions, which may in turn affect economic performance.12 This argument cuts both ways, however. Stated, and even informal, exchange rate policy intentions may be forward looking but may also be misleading.13 Finally, de facto classifications may result in a high frequency of recorded regime transitions because of changes in the pattern of actual exchange rate movements. The Natural classification addresses this issue by employing a five-year horizon to gauge actual exchange rate flexibility. While this helps to distinguish regimes from spells, it limits the Natural classification’s ability to detect short-term currency market pressures, such as those that culminated in the CFA franc devaluation in early 1994, that could have longer-term macroeconomic effects. Hence, the Natural classification is not necessarily appropriate for analyzing issues, such as the near-term impact of changes in a country’s exchange rate spell. From a global perspective, however, the Natural classification, with its special features and rich historical coverage, has the potential to yield important new insights into the history of regimes and their effect on macroeconomic performance.
Divergence Between Stated and Actual Policies
Comparison of the de jure and Natural classifications highlights the divergence between stated and actual policies, particularly at the polar extreme of flexibility. Focusing on the broad classification categories over the period 1973–99 (for which there are overlapping data), Figure 2.1 shows that only about half of the observations—where each observation corresponds to a given country’s regime in a particular year—were classified in the same broad category under both the de jure and the Natural classifications. The divergence was particularly striking among so-called floating regimes, where only 20 percent were de facto free floats while 60 percent were either intermediate or pegged regimes and another 20 percent had freely falling currencies.14 Although almost all de jure hard pegs were in fact operated as hard pegs, fewer than 40 percent of de jure soft pegs were de facto pegs, either hard or soft. About 60 percent of de jure intermediate regimes actually operated as intermediate regimes.15
Natural Classification Regimes by De Jure Category, 1973–99
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); Ghosh, Guide, and Wolf (2003); and IMF staff estimates.Natural Classification Regimes by De Jure Category, 1973–99
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); Ghosh, Guide, and Wolf (2003); and IMF staff estimates.Natural Classification Regimes by De Jure Category, 1973–99
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); Ghosh, Guide, and Wolf (2003); and IMF staff estimates.In the 1970s and 1980s, the differences between actual and stated policies reflected to a large extent the prevalence of dual/parallel foreign exchange markets. In the early 1970s, almost one-half of all countries and one-third of advanced economies had active dual/parallel markets with exchange rates that deviated substantially from official rates (Figure 2.2). Foreign exchange markets have since been unified in most countries. In emerging markets and other developing countries, the unification occurred mainly in the 1990s as capital flows to emerging market economies accelerated and efforts were intensified by the international community, including the IMF, to encourage countries to accept Article VIII of the IMF’s Articles of Agreement. Although the number of countries with dual/parallel exchange rates that deviated substantially from official rates declined to 9 in 2001 from 30 in 1995,16 the number of mismatches between countries’ classifications in the de jure and Natural classifications did not. This was due mainly to the increase in freely falling regimes in the 1990s, which included the transition economies of central and eastern Europe and the former Soviet Union, and the de jure classification of euro area currency regimes as intermediate until 1999.
Countries with Dual/Parallel Foreign Exchange Markets
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Countries with Dual/Parallel Foreign Exchange Markets
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Countries with Dual/Parallel Foreign Exchange Markets
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.The frequency of freely falling regimes is also on a declining trend, despite a brief resurgence following the breakup of the Soviet Union. Rogoff (2003) notes that this in turn reflects the decline in inflation across the world in recent years. Hence, accounting for dual/parallel markets and free falling regimes, while critical in drawing lessons from the history of regimes, is less likely to be as relevant in the future.
Differences Across Country Groups
As noted, compared to the Natural classification the de jure classification significantly overstates the number of true floats and pegs, suggesting that fewer countries are at the polar extremes than implied by their announcements. Figures 2.3 and 2.4 show that few countries, especially emerging markets and other developing countries, actually allow their exchange rates to float freely. Among emerging markets, the proportion of de facto free floaters has remained relatively small at 4-7 percent since the mid-1980s (Figure 2.5).17 Even among advanced economies, only about 20 percent allow their currencies to float freely, although close to 40 percent state that they have floating regimes. These figures also show that fewer countries actually peg their exchange rates than announcements would suggest. De facto pegs accounted for about one-third of all de facto regimes in recent years, while de jure pegs comprised about one-half of all de jure regimes. The number of hard pegs was significantly higher, however, under the Natural classification than under the de jure.18 While the proportion of de facto pegs has increased slightly since the early 1990s, this mainly reflects the monetary union in Europe and the adoption of pegs by some of the countries that were previously experiencing freely falling currency values. Interestingly, hard pegs accounted for most of the recent increase in pegs in other developing countries, while soft pegs accounted for much of the increase in emerging markets.
Natural Classification Regime Distribution
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Distribution
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Distribution
(In percent of annual observations)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.
De Jure Regime Distribution
(In percent of annual observations)
Sources: Ghosh, Guide, and Wolf (2003); and IMF staff estimates.De Jure Regime Distribution
(In percent of annual observations)
Sources: Ghosh, Guide, and Wolf (2003); and IMF staff estimates.De Jure Regime Distribution
(In percent of annual observations)
Sources: Ghosh, Guide, and Wolf (2003); and IMF staff estimates.
Natural Classification Regime Distribution by Country Group
(In percent of annual observations for each group)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Distribution by Country Group
(In percent of annual observations for each group)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Distribution by Country Group
(In percent of annual observations for each group)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Intermediate regimes have been, and continue to be, considerably more prevalent than suggested by the de jure classification. While de jure intermediate regimes rose to about a quarter of all exchange rate regimes in the late 1990s from around 10 percent in the mid-1970s, the proportion of de facto regimes with an intermediate degree of flexibility has remained at about one-half since the mid-1970s.19 Within intermediate regimes, however, managed floats have become more prevalent in emerging markets over the past decade, while other developing countries have tended to move in the opposite direction toward more limited flexibility.
A historical retrospective using the Natural classification also suggests that the breakup of the Bretton Woods system was much less of a watershed event for emerging markets and other developing countries than for advanced economies. De facto pegs in advanced economies declined sharply as the Bretton Woods system collapsed, while among emerging markets and other developing countries the decline in pegs was more gradual and continued through the 1980s.20
Even when compared with other de facto classifications, the Natural classification records fewer regimes near the polar extremes of full flexibility and rigid pegs. At a broad level, the IMF de facto classification yields similar results to the Natural classification—two-thirds or more of Natural classification free floats, pegs, and intermediate regimes are classified the same way by the IMF de facto classification. The IMF classification, however, picks up many more free floats than the Natural classification, especially among emerging markets, where as many as one-third were listed as freely floating regimes in 2001 (Figure 2.6).21 Similarly, the prevalence of pegs is higher than in the Natural classification, especially for other developing countries, of which about half were listed as pegged regimes in 2001.22 The Levy-Yeyati-Sturzenegger de facto classification also records many more free floats and pegs and, consequently, many fewer intermediate regimes than the Natural classification (Figure 2.7). Surprisingly, over half of emerging markets are classified as floats in the Levy-Yeyati-Sturzenegger classification in the late 1990s, both before and after the Asian crises, and free floats are more prevalent than in the de jure classification, drawing into question the degree to which the former presents a more accurate picture of actual regimes than the latter.
IMF De Facto Regime Distribution
(In percent of annual observations)
Sources: Bubula and Ötker-Robe (2002); and IMF staff estimates.IMF De Facto Regime Distribution
(In percent of annual observations)
Sources: Bubula and Ötker-Robe (2002); and IMF staff estimates.IMF De Facto Regime Distribution
(In percent of annual observations)
Sources: Bubula and Ötker-Robe (2002); and IMF staff estimates.
Levy-Yeyati-Sturzenegger Regime Distribution
(In percent of annual observations)
Sources: Levy-Yeyati and Sturzenegger (2003); and IMF staff estimates.Levy-Yeyati-Sturzenegger Regime Distribution
(In percent of annual observations)
Sources: Levy-Yeyati and Sturzenegger (2003); and IMF staff estimates.Levy-Yeyati-Sturzenegger Regime Distribution
(In percent of annual observations)
Sources: Levy-Yeyati and Sturzenegger (2003); and IMF staff estimates.Anchor Currency Choice
While there is a large empirical literature on the conditions under which countries adopt fixed or floating regimes (discussed in Appendix II), less has been written on the determinants of anchor currency choice. The question of interest is: Once countries choose to peg their exchange rates to an anchor currency—including by means of crawling pegs or bands—what determines the choice of this anchor?
The theory of optimal currency areas suggests that countries benefit from adopting the same anchor as a trade partner, because this reduces their bilateral exchange rate variability. Meissner and Oomes (2004) provide empirical evidence of these network externalities. The authors find that, after controlling for other factors—such as country size, openness, and colonial history—the probability of choosing a particular anchor currency increases with the amount of trade with other countries that use this same anchor. These externalities may explain why virtually all countries that have chosen to peg their exchange rates in some way to another currency have converged over the last 50 years to using either the U.S. dollar or the euro as their anchor currency (see figure below).
1940–72
Between 1940 and 1972, the U.S. dollar was the most popular anchor currency chosen by advanced countries, followed by the British pound and the German deutsch mark. For developing countries, the predominant anchor currencies were the U.S. dollar, the British pound, and the French franc, with the latter two choices being determined largely by colonial history.
1973–89
Following the collapse of the Bretton Woods system, the British pound disappeared entirely from the menu of anchor choices. Pegs to the U.S. dollar declined in popularity among advanced countries as an increased number of free and managed floaters emerged, and the majority of advanced countries that retained pegs ended up tying their currencies in some form to the deutsche mark, and later to the euro. Developing countries largely switched to using the U.S. dollar as anchor, except the group of former French colonies that continued to peg to the French franc.
1990–2001
The overall distribution of anchor currencies did not change much in the 1990s, apart from the introduction of the euro in 1999. The behavior of transition economies during this period, however, is illustrative of the dynamics of anchor currency choice. Following the breakup of the Soviet Union in the early 1990s, most transition economies fell initially in the freely falling category for several years, and then increasingly started tying their currencies to the deutsche mark or the U.S. dollar. Interestingly, the choice of anchor was almost perfectly divided among regional lines: while Central and Eastern European countries chose to anchor to the deutsche mark, and later to the euro, most former Soviet Union republics chose the U.S. dollar as their anchor—with the exception of Estonia, which adopted a currency board arrangement with the deutsche mark, and Latvia, which chose the SDR. As Meissner and Oomes (2004) show, this divide between the euro and the dollar cannot be explained solely on the basis of trade flows with Europe or the United States but is partially the result of network externalities arising from trade partners’ anchor currency choices.
Bipolar Hypothesis and Fear of Floating
The Natural classification raises questions about the general validity of the bipolar hypothesis. Starting in the mid-1990s, some observers had predicted that emerging market countries would, over time, move to the polar extremes of exchange rate flexibility; that is, they would either adopt freely floating regimes or move to hard pegs.23 That speculative attacks against hard pegs were rare and could apparently be warded off seemed to lend support to the hypothesis.24 The increase in free floats and hard pegs since 1990 in the de jure—and to a smaller extent in the IMF de facto classifications, as illustrated in Figures 2.4 and 2.6, respectively—appeared to support the bipolar view. As noted above, however, the Natural classification indicates that there has been no “hollowing out of the middle.” While a few emerging markets indeed moved in the 1990s to de facto hard pegs (Argentina and Malaysia) or free floats (Indonesia, Korea, and South Africa), just as many transitioned from freely falling to intermediate regimes (Brazil, Peru, Poland, Russia, and República Bolivariana de Venezuela).25 As a result, the middle remained as large as it was a decade ago. Moreover, transitions since 1990 to de facto pegs among emerging markets have been more in the soft category (China, Egypt, Jordan, and Peru) rather than the hard category.26
The tendency of countries to allow less exchange rate flexibility in practice than in policy statements is consistent with the fear of floating. As Calvo and Reinhart (2002) argue, fear of floating—a reluctance to allow exchange rates to fluctuate freely—could arise for various reasons, including policy credibility concerns; fear of Dutch disease in case of large appreciations; and fear of inflation, currency mismatches, and/or balance sheet effects (on account of high liability dollarization) in case of large depreciations.27 As Figure 2.1 indicates, the vast majority of countries that say they float actually do not. Moreover, many countries that say they have intermediate regimes in fact have de facto pegs.
Regime Transitions
Major global and regional events have influenced exchange rate regime transitions. The collapse of the Bretton Woods system in 1973 was, of course, the outcome of pressures built up in a relatively rigid system of exchange rate regimes and was followed by a sharp increase in flexible arrangements (Figure 2.8). The debt crisis of the 1980s and the transformation of the economies of Central and Eastern Europe and the former Soviet Union in the early 1990s were also accompanied by a relatively high frequency of regime transitions, especially into and subsequently out of the freely falling category. In the latter half of the 1990s, as several large emerging markets faced external financing crises, the frequency of exchange rate regime transitions among this group rose once again. Then in 1999, a major transition occurred among advanced economies with the adoption of a monetary union in the euro area.
Natural Classification Regime Transitions
(Number of transitions)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Transitions
(Number of transitions)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Natural Classification Regime Transitions
(Number of transitions)
Sources: Reinhart and Rogoff (2004); and IMF staff estimates.Once the transitions into and out of the freely falling category—as well as those that occurred as a result of global events—are distinguished, it turns out that the frequency of changes in exchange rate regimes today is remarkably similar to that of 50 years ago. As Figure 2.8 illustrates, the average number of countries transitioning to a different regime (excluding transitions into and out of the freely falling category) in any given year since the collapse of the Bretton Woods system was about the same as during the Bretton Woods period.
Thus, the interesting finding is that countries have changed their de facto exchange rate regime relatively infrequently. On the basis of data going back to the 1940s, about 7 percent of all countries transitioned to a different regime in an average year, and the typical exchange rate regime had a duration of about 14 years (Table 2.1). If the 1970–75 period is excluded and Eastern and Central European and former Soviet Union countries along with the euro area countries are removed from the sample, transitions were even less frequent. In the adjusted sample, the average regime duration rises to just over 16 years, while the proportion of countries changing regime in any given year declines to about 6¼ percent.
Annual Transition Probabilities
Excludes euro area and former command economies in Europe and the former Soviet Union.
Natural classification transition rates for all regimes and pegs over the same period were 9.3 percent and 5.1 percent, respectively.
Annual Transition Probabilities
| Natural Classification | |||
| All countries, 1940–2001 | 7.0 | ||
| Pegs only | 3.5 | ||
| All countries, adjusted sample, 1940–2001 (excluding 1970–75)1 | 6.2 | ||
| Pegs only | 2.5 | ||
| Advanced economies, 1940–2001 | 7.0 | ||
| Pegs only | 5.1 | ||
| Emerging markets, 1940–2001 | 9.7 | ||
| Pegs only | 6.7 | ||
| Emerging markets, 1989–2001 | 14.4 | ||
| Pegs only | 9.8 | ||
| Developing countries, 1940–2001 | 6.1 | ||
| Pegs only | 2.4 | ||
| De jure Classification2 | |||
| All countries, 1973–2001 | 6.8 | ||
| Pegs only | 4.9 | ||
Excludes euro area and former command economies in Europe and the former Soviet Union.
Natural classification transition rates for all regimes and pegs over the same period were 9.3 percent and 5.1 percent, respectively.
Annual Transition Probabilities
| Natural Classification | |||
| All countries, 1940–2001 | 7.0 | ||
| Pegs only | 3.5 | ||
| All countries, adjusted sample, 1940–2001 (excluding 1970–75)1 | 6.2 | ||
| Pegs only | 2.5 | ||
| Advanced economies, 1940–2001 | 7.0 | ||
| Pegs only | 5.1 | ||
| Emerging markets, 1940–2001 | 9.7 | ||
| Pegs only | 6.7 | ||
| Emerging markets, 1989–2001 | 14.4 | ||
| Pegs only | 9.8 | ||
| Developing countries, 1940–2001 | 6.1 | ||
| Pegs only | 2.4 | ||
| De jure Classification2 | |||
| All countries, 1973–2001 | 6.8 | ||
| Pegs only | 4.9 | ||
Excludes euro area and former command economies in Europe and the former Soviet Union.
Natural classification transition rates for all regimes and pegs over the same period were 9.3 percent and 5.1 percent, respectively.
De facto pegged regimes have tended to change less frequently and last longer than other regimes. For all de facto pegs since 1940, the probability of exiting to a different regime in any given year was about 3½ percent.28 Since the Natural classification classifies as pegs only those that are successful, countries that attempt to peg but are able to sustain the peg only briefly tend not to be classified as pegs. This, together with the fact that the Natural classification does not treat onetime devaluations followed by a re-peg as a change in the longer-term regime,29 reduces the observed exit rate from de facto pegs. It is also worth noting that regime transitions are less frequent in the de jure classification than in the Natural classification, suggesting that countries tend to change their stated exchange rate policy objectives even less frequently than their de facto exchange rate policies. The average annual exit rate from de facto and de jure pegs during 1973–99 has been about the same; however, this is partly because the collapse of the Bretton Woods system accounted for a sizable portion of such exits during this period.30
Emerging markets, however, have tended to switch regimes more frequently, and have gone into the freely falling category more often, than other countries. Since 1940, the annual regime transition rate among emerging markets has averaged about 10 percent, compared with 7 percent for advanced countries and about 2 percent for other developing countries. On average, about 3 percent of emerging markets, excluding those already in the freely falling category, have transitioned to a freely falling regime every year. By contrast, only 0.5 percent of all advanced countries and less than 2 percent of other developing countries have switched to a freely falling regime in any given year. The transition rate out of pegged regimes among emerging markets has also been higher (about 7 percent) than in advanced and other developing countries (5 percent and 2½ percent, respectively).31
If historical transition rates continue and no further major global shocks occur, intermediate regimes will remain prevalent in the future, and the overall distribution of de facto regimes will be similar to that at present. Given that pegs have had a somewhat longer average duration than other regimes in the past, the historical transition rates imply that the proportion of pegs could increase slightly over time. Similarly, since relatively few countries, especially developing countries, have had true free floats in the past, the historical likelihood of transitioning into a free float has been low, implying that the share of free floats among all regimes is likely to remain modest in the future. As other developing countries become increasingly integrated into global financial markets, however, their regime transitions may well resemble those seen among emerging markets during the 1990s. In that case, the proportion of pegged regimes among developing countries will tend to decline gradually in the future, while managed floats and free floats will gradually increase. Over the longer term, of course, political economy considerations may guide regime choice decisions in some countries. For example, some may choose to join currency unions in the not-so-distant future. Prospects for transitions of that nature cannot be assessed on the basis of historical transition rates, however, and are clearly beyond the scope of this analysis.
Implications for Assessing Regime Performance
Empirical analysis seeking to uncover the link between countries’ exchange rate regimes and their macroeconomic performance depends critically on how regimes are classified. The wide variation between countries’ stated exchange rate regimes and their actual practice suggests that results obtained by employing the de jure classification could be off the mark and that use of a classification that more accurately captures true regime flexibility can lead to different conclusions. The Natural classification, with its special features and historical coverage, is a promising candidate for such analysis.32
The persistent popularity of intermediate regimes—especially among emerging markets and other developing countries—as identified by the Natural classification, suggests that such regimes may provide important advantages. Indeed, the absence of a general bipolar tendency may indicate that intermediate regimes are able to capture some of the benefits of both extremes while avoiding many of the costs.
Finally, the relatively long average duration of Natural classification regimes may suggest that regime transitions involve significant costs. The higher transition rates for emerging markets indicate, however, that either these costs decline as countries experience higher capital flows or, more likely, that higher capital flows in the absence of adequate financial infrastructure and safeguards make it harder to sustain regimes, particularly pegged regimes. Again, evidence in support of this channel may be obtained potentially by assessing the (historical) likelihood of crises under alternative exchange rate regimes across different types of economies.
Appendix I. The Natural Classification
This appendix summarizes the data and algorithm used to construct the Natural classification and provides a brief summary of the main features of various de facto classifications (see Table A2.1).
Main Features of Various De Facto Classifications
Main Features of Various De Facto Classifications
| Ghosh, Guide, and Wolf (2003) | IMF (1999, 2003b); Bubula and Ötker-Robe (2002) | Levy-Yeyati and Sturzenegger (2003) | Reinhart and Rogoff (2004) | |
|---|---|---|---|---|
| Period | 1973–99 | 1990–present | 1974–2000 | 1940–2001 |
| Frequency | Annual | Annual and monthly | Annual | Annual and monthly |
| Number of countries | 165 | 190 | 156 | 153 |
| Number of regime types | 25 fine, 9 coarse | 15 fine, 8 coarse | 4 | 14 fine, 5 coarse |
| Advantages | Uses quantitative and qualitative information (survey of IMF desk economists) Fine taxonomy | Uses quantitative and qualitative information (survey of IMF desk economists; discussions with authorities; news articles; press reports) All IMF member countries classified; classification continuously updated | Uses information on volatility of foreign exchange reserves Systematic approach; no judgment needed | Uses dual/parallel exchange rate information Separates freely falling episodes Long time series; monthly exchange rate movements to identify regime Systematic approach; no judgment needed |
| Disadvantages | Relies to large extent on stated policy intentions, which may deviate substantially from actual practice Requires subjective judgment, which may differ across countries and over time Not all countries are classified for all time periods | Requires subjective judgment, which may differ across countries and over time | Exchange rate stability or reserve changes may occur for reasons other than policy intervention Reserves data may not cover derivatives Many observations not classified—only 15 years per country classified on average Other countries affect classification (due to cluster analysis) | Exchange rate stability may occur for reasons other than policy intervention A few countries are not classified for all years |
Main Features of Various De Facto Classifications
| Ghosh, Guide, and Wolf (2003) | IMF (1999, 2003b); Bubula and Ötker-Robe (2002) | Levy-Yeyati and Sturzenegger (2003) | Reinhart and Rogoff (2004) | |
|---|---|---|---|---|
| Period | 1973–99 | 1990–present | 1974–2000 | 1940–2001 |
| Frequency | Annual | Annual and monthly | Annual | Annual and monthly |
| Number of countries | 165 | 190 | 156 | 153 |
| Number of regime types | 25 fine, 9 coarse | 15 fine, 8 coarse | 4 | 14 fine, 5 coarse |
| Advantages | Uses quantitative and qualitative information (survey of IMF desk economists) Fine taxonomy | Uses quantitative and qualitative information (survey of IMF desk economists; discussions with authorities; news articles; press reports) All IMF member countries classified; classification continuously updated | Uses information on volatility of foreign exchange reserves Systematic approach; no judgment needed | Uses dual/parallel exchange rate information Separates freely falling episodes Long time series; monthly exchange rate movements to identify regime Systematic approach; no judgment needed |
| Disadvantages | Relies to large extent on stated policy intentions, which may deviate substantially from actual practice Requires subjective judgment, which may differ across countries and over time Not all countries are classified for all time periods | Requires subjective judgment, which may differ across countries and over time | Exchange rate stability or reserve changes may occur for reasons other than policy intervention Reserves data may not cover derivatives Many observations not classified—only 15 years per country classified on average Other countries affect classification (due to cluster analysis) | Exchange rate stability may occur for reasons other than policy intervention A few countries are not classified for all years |
The Natural classification, which classifies exchange rate regimes into fine and coarse categories (as summarized in Table A2.2), employs monthly data on official and market-determined exchange rates for the period 1940–2001.33 The data on market-determined exchange rates are drawn from various issues of Pick’s Currency Yearbook, Pick’s Black Market Year-book, and Pick’s World Currency Report, while the official rate data are from the same sources as well as the IMF’s International Financial Statistics. The quotes are end-of-month exchange rates. Annual classifications are simply the modal monthly classifications for each country in each year.
Natural Classification Categories
Natural Classification Categories
| Fine | Coarse | Description |
|---|---|---|
| 1 | 1 | No separate legal tender |
| 2 | 1 | Preannounced peg or currency board arrangement |
| 3 | 1 | Preannounced horizontal band that is narrower than or equal to ±2 percent |
| 4 | 1 | De facto peg |
| 5 | 2 | Preannounced crawling peg |
| 6 | 2 | Preannounced crawling band that is narrower than or equal to ±2 percent |
| 7 | 2 | De facto crawling peg |
| 8 | 2 | De facto crawling band that is narrower than or equal to ±2 percent |
| 9 | 3 | Preannounced crawling band that is wider than or equal to ±2 percent |
| 10 | 3 | De facto crawling band that is narrower than or equal to ±5 percent |
| 11 | 3 | Moving band that is narrower than or equal to ±2 percent (i.e., allows for both appreciation and depreciation over time) |
| 12 | 3 | Managed floating |
| 13 | 4 | Freely floating |
| 14 | 5 | Freely falling |
| 15 | 6 | Dual market in which parallel market data are missing |
Natural Classification Categories
| Fine | Coarse | Description |
|---|---|---|
| 1 | 1 | No separate legal tender |
| 2 | 1 | Preannounced peg or currency board arrangement |
| 3 | 1 | Preannounced horizontal band that is narrower than or equal to ±2 percent |
| 4 | 1 | De facto peg |
| 5 | 2 | Preannounced crawling peg |
| 6 | 2 | Preannounced crawling band that is narrower than or equal to ±2 percent |
| 7 | 2 | De facto crawling peg |
| 8 | 2 | De facto crawling band that is narrower than or equal to ±2 percent |
| 9 | 3 | Preannounced crawling band that is wider than or equal to ±2 percent |
| 10 | 3 | De facto crawling band that is narrower than or equal to ±5 percent |
| 11 | 3 | Moving band that is narrower than or equal to ±2 percent (i.e., allows for both appreciation and depreciation over time) |
| 12 | 3 | Managed floating |
| 13 | 4 | Freely floating |
| 14 | 5 | Freely falling |
| 15 | 6 | Dual market in which parallel market data are missing |
The procedure employed by the Natural classification to classify regimes is as follows:
First, a separation is made between countries with either official dual or multiple rates or active parallel (black) markets.
If there is no dual or black market, a check is done to see if there is an official preannounced arrangement, such as peg, crawling peg, or band. If there is, the announced regime is verified by examining the mean absolute monthly change over the period following the announcement.34 If the regime is verified according to rules analogous to those described in step 3 below, it is then classified according to the announcement.35
If there is no preannounced exchange rate path, if the announced regime cannot be verified by the data (which is often the case), and if the 12-month rate of inflation is below 40 percent, the regime is classified on the basis of actual exchange rate behavior as follows:
If the absolute monthly percent change in the exchange rate is equal to zero for four consecutive months or more, that episode is classified (for however long it lasts) as a de facto peg, if there are no dual or multiple exchange rates in place.36
If the probability is 80 percent or higher that the monthly exchange rate change remains within a plus/minus 1 percent band over a rolling five-year period, then the regime is classified as a de facto peg or crawling peg over the entire five-year period. If the exchange rate has no drift, it is classified as a fixed parity; if a positive drift is present, it is labeled a crawling peg; and, if the exchange rate also goes through periods of both appreciation and depreciation, it is a moving peg.
The approach regarding de facto bands, as well as preannounced bands, follows a parallel two-step process. Thus, if there is more than an 80 percent probability that the monthly exchange rate change remains within a plus/minus 2 percent band over a rolling five-year period, then the regime is classified as either a de facto narrow band, a narrow crawling band, or a moving band throughout the entire period during which it remains continuously above the 80 percent threshold.
If the 12-month rate of inflation exceeds 40 percent, the episode is classified as freely falling.37
The remaining regimes—those that have not already been classified by steps one through four—become candidates for managed or freely floating. To distinguish between the two, the degree of exchange rate flexibility is measured by a composite statistic.
Appendix II. Determinants of Exchange Rate Regime Choice
The Natural classification data show some links between de facto regime flexibility and certain macroeconomic and financial variables, such as trade openness and dollarization. A review of the literature suggests, however, that it is difficult to find empirical regularities between potential exchange rate regime determinants and actual regimes that hold consistently across all countries, time periods, and regime classifications. Systematic robustness checks of the determinants of regime choice employing the Natural classification support this result.
Macroeconomic and Financial Characteristics of Regimes
Optimum currency area (OCA) theory holds that variables, such as large size and low openness to trade, are likely to be associated with floating exchange rates. One reason for this may be that trade openness raises the transactions benefits from common currencies, and should be expected to lead, therefore, to a decline in the number of independent currencies. The data appear to support the OCA theory prediction that countries that trade a lot will tend to have less flexible exchange rate regimes. Advanced economies that have a high trade openness ratio have tended to have pegged regimes, while the prevalence of free floats has been notably higher in advanced countries with low external trade ratios, such as Australia, Japan, and the United States. A similar pattern holds among other developing countries, where the prevalence of managed floats has been markedly higher and pegs significantly lower in the countries that rely less on external trade. The pattern among emerging markets has been less clear, although relatively closed economies in this group have had a much higher likelihood of being in the freely falling category.
Higher dollarization appears to be associated with less flexible exchange rate regimes among emerging markets, consistent with fear of floating. Fear of floating appears to be stronger in highly dollarized emerging markets, where pegged regimes are more prevalent, than in less-dollarized countries in the group. Conversely, emerging markets with low and medium degrees of dollarization are more likely to have managed or freely floating regimes. Fear of floating does not explain, however, why other developing countries with high dollarization ratios appear to prefer regimes with limited flexibility to pegs. A possible explanation for this could be that many of these countries became highly dollarized following a freely falling episode and lacked the credibility necessary to defend a peg. A regime with limited flexibility allowed them to obtain the benefits of a relatively stable currency, while at the same time maintaining some ability to adjust to shocks.
There is little systematic relation, however, in the degree of capital account openness across de facto regimes. Emerging markets and other developing countries tend to have more capital controls and lower capital flows in relation to GDP than advanced economies. Nevertheless, the variation in capital account openness does not appear to be related to the flexibility of countries’ currency regimes. Among advanced economies, the volume of capital flows in countries with de facto pegged regimes tends to be higher than in those with intermediate regimes and significantly higher than in those with freely floating regimes. The relationship is more mixed, however, for emerging markets and other developing countries, possibly because capital controls are often in-effective, so the expected inverse relation between controls and observed capital flows may not hold.
Empirical Findings on Factors Affecting Regime Choice
Systematic prediction of exchange rate regime choice is elusive. A review of a reasonably broad collection of previous studies shows that different empirical studies using the de jure and other de facto regime classifications have often obtained different results, suggesting that it is very difficult to draw general conclusions about how countries choose their exchange rate regimes. Although certain characteristics have been shown to be important in determining exchange rate regime choice in some groups of countries, and certain characteristics may distinguish countries in some regimes from those in different regimes, no result appears fully robust to changes in country coverage, sample period, estimation method, and exchange rate regime classification.
Several empirical studies have analyzed the determinants of exchange rate regime choice in a cross section of countries. Among the first studies of this kind are Heller (1978), which analyzes the determinants of exchange rate regimes with data from the mid-1970s, soon after the generalized floating that followed the breakup of the Bretton Woods system, as well as Dreyer (1978); Holden, Holden, and Suss (1979); Melvin (1985); Bosco (1987); Savvides (1990); Cuddington and Otoo (1990 and 1991); Rizzo (1998); and Poirson (2001). Some studies, such as those by Collins (1996), Edwards (1996 and 1999), and, more recently, Frieden, Ghezzi, and Stein (2001), have used random effects panel data to analyze also the determinants of changes in exchange rate regime. As such, they can be seen as somewhat related to the recent literature on predicting exchange rate crises. Nevertheless, these studies are included in this review because they report findings on the role of country characteristics that are relatively stable over time (such as openness) in determining exchange rate regime choice. Another recent study, by Berger, Sturm, and de Haan (2000), uses panel data in an attempt to identify the long-run determinants of exchange rate regime choice. Additional studies addressing changes in exchange rate regimes include Masson (2001), Klein and Marion (1997), and Duttagupta and Ötker-Robe (2003).
The vast majority of previous studies have attempted to explain countries’ de jure exchange rate regime choice. A few studies have constructed and used measures of the degree of de facto flexibility on the basis of the actual observed volatility of exchange rates and reserves, including Holden, Holden, and Suss (1979) and, more recently, Poirson (2001). Table A2.3 summarizes the approaches and findings of these studies with regard to the impact of several variables on observed exchange rate regime choice. Most studies considered some of the optimum currency area variables, such as trade openness (typically measured as imports plus exports, divided by GDP), the size of the economy (gross domestic product in common currency), the degree of economic development (GDP per capita), and geographical concentration of trade (the share of trade with the country’s main partner). Among macroeconomic variables, several studies included inflation (whether the country’s own inflation or inflation in excess of partner countries) and foreign exchange reserves. Many studies included an indicator of either capital controls, which were typically also drawn or constructed from the IMF’s Annual Report on Exchange Arrangements and Exchange Restrictions, or de facto capital openness (e.g., the ratio of foreign assets of the banking system to the money supply). Some studies included measures of volatility of domestic output, exports, domestic credit, or the real exchange rate, although no two studies seem to have looked at the same measure of volatility. A few studies considered variables related to political economy or institutional strength. Most studies analyzed some variables that were not included in any preceding (or subsequent) studies. Collectively, the studies considered more than 30 potential determinants of exchange rate regime choice. (Only the variables considered by more than one study are included in Table A2.3.)
Studies on Determinants of Exchange Rate Regimes(Likelihood to Float)
+ indicates that the coefficient of explanatory variable is positive and - that it is negative; ± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
Studies on Determinants of Exchange Rate Regimes(Likelihood to Float)
| Author | Heller (1978) | Dreyer (1978) | Holden, Holden, and Suss (1979) | Melvin (1985) | Sawides (1990) | Cuddington and Otoo (1990,1991) | Honkapohja and Pikkarainen (1994) | Collins (1996) | Edwards (1996) | Edwards (1999) | Rizzo (1998) | Frieden, Ghezzi and Stein (2000) | Berger, Sturm and de Haan (2000) | Poirson (2001) | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sample | 86 countries | 88 developing countries | 76 countries | 64 countries | 39 developing countries | 66 countries | 125 countries | 24 Latin American and Caribbean countries | 63 countries | 49 developing and middle income | 123 countries | 26 Latin American countries | 65 developing countries | 93 countries | |
| Time frame | 1976 | 1976 | 1974–75 | 1976–78 | 1976–84 | 1980, 83, 86 | 1991 | 1978–92 | 1980–92 | 1980–92 | 1977–95 | 1960–94 | 1980–94 | 1990–98 | |
| Methodology | Discriminant analysis | Probit | OLS on a continuous measure | Multinomial logit | Two-stage probit | Ordered/non-ordered Mult./bin. logit | Logit and probit | Probit (panel) | Probit (panel) | Probit (panel) | Probit | Ordered logit (panel) | Probit (panel) | Ordered probit | |
| Explanatory variables | |||||||||||||||
| (OCA factors)1 | |||||||||||||||
| Openness | – | – ^^ | – ^^ | • | – | ± ^ | – ^ | + ^ | + ^ | – ^ | + ^ | – | |||
| Economic development | + ^^ | + ^^ | – ^ | • | + ^ | – ^^ | ± ^ | ± | |||||||
| Size of economy | + | + | + ^^ | + | + | + ^^ | + ^^ | + ^ | |||||||
| Inflation differential | + | + ^^ | + | + ^ | |||||||||||
| Capital mobility | – | – ^^ | |||||||||||||
| Geographical trade concentration | – | – ^^ | – | – | + | – ^^ | |||||||||
| International financial integration | + | ± | |||||||||||||
| (Other macro/external/structural factors)1 | |||||||||||||||
| Growth | + ^^ | + ^^ | + | ||||||||||||
| Negative growth | – ^ | – ^^ | |||||||||||||
| Inflation | + ^^ | + ^^ | + ^^ | + ^ | |||||||||||
| Moderate-to-high inflation | + ^^ | – ^^ | |||||||||||||
| Reserves | – ^^ | – ^ | ± ^^ | + ^^ | + ^ | – ^ | |||||||||
| Capital control | ± ^ | + ^ | – ^ | ||||||||||||
| Terms-of-trade volatility | + | + ^^ | – ^^ | + ^ | |||||||||||
| Variability in export growth | + ^^ | + | |||||||||||||
| External variability/openness | – ^^ | – ^^ | |||||||||||||
| Real exchange rate volatility | + ^^ | + ^ | + ^^ | ||||||||||||
| Product diversification | – ^^ | – ^^ | + ^ | ||||||||||||
| Current account | – ^^ | ± ^^ | |||||||||||||
| External debt | + ^ | + ^^ | |||||||||||||
| Growth of domestic credit | + ^^ | + ^ | |||||||||||||
| Money shocks | – ^^ | – | |||||||||||||
| Foreign price shocks | + ^^ | + | |||||||||||||
| (Political/historical factors)1 | |||||||||||||||
| Political instability | + ^^ | + ^^ | – ^^ | + ^ | + ^^ | ||||||||||
| Central bank independence | + | + ^ | |||||||||||||
| Party in office has majority | – ^ | – ^ | |||||||||||||
| Number of parties in coalition | + | + | |||||||||||||
| Coalition government | – | – | |||||||||||||
| (OCA factors)1 |
+ indicates that the coefficient of explanatory variable is positive and - that it is negative; ± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
Studies on Determinants of Exchange Rate Regimes(Likelihood to Float)
| Author | Heller (1978) | Dreyer (1978) | Holden, Holden, and Suss (1979) | Melvin (1985) | Sawides (1990) | Cuddington and Otoo (1990,1991) | Honkapohja and Pikkarainen (1994) | Collins (1996) | Edwards (1996) | Edwards (1999) | Rizzo (1998) | Frieden, Ghezzi and Stein (2000) | Berger, Sturm and de Haan (2000) | Poirson (2001) | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Sample | 86 countries | 88 developing countries | 76 countries | 64 countries | 39 developing countries | 66 countries | 125 countries | 24 Latin American and Caribbean countries | 63 countries | 49 developing and middle income | 123 countries | 26 Latin American countries | 65 developing countries | 93 countries | |
| Time frame | 1976 | 1976 | 1974–75 | 1976–78 | 1976–84 | 1980, 83, 86 | 1991 | 1978–92 | 1980–92 | 1980–92 | 1977–95 | 1960–94 | 1980–94 | 1990–98 | |
| Methodology | Discriminant analysis | Probit | OLS on a continuous measure | Multinomial logit | Two-stage probit | Ordered/non-ordered Mult./bin. logit | Logit and probit | Probit (panel) | Probit (panel) | Probit (panel) | Probit | Ordered logit (panel) | Probit (panel) | Ordered probit | |
| Explanatory variables | |||||||||||||||
| (OCA factors)1 | |||||||||||||||
| Openness | – | – ^^ | – ^^ | • | – | ± ^ | – ^ | + ^ | + ^ | – ^ | + ^ | – | |||
| Economic development | + ^^ | + ^^ | – ^ | • | + ^ | – ^^ | ± ^ | ± | |||||||
| Size of economy | + | + | + ^^ | + | + | + ^^ | + ^^ | + ^ | |||||||
| Inflation differential | + | + ^^ | + | + ^ | |||||||||||
| Capital mobility | – | – ^^ | |||||||||||||
| Geographical trade concentration | – | – ^^ | – | – | + | – ^^ | |||||||||
| International financial integration | + | ± | |||||||||||||
| (Other macro/external/structural factors)1 | |||||||||||||||
| Growth | + ^^ | + ^^ | + | ||||||||||||
| Negative growth | – ^ | – ^^ | |||||||||||||
| Inflation | + ^^ | + ^^ | + ^^ | + ^ | |||||||||||
| Moderate-to-high inflation | + ^^ | – ^^ | |||||||||||||
| Reserves | – ^^ | – ^ | ± ^^ | + ^^ | + ^ | – ^ | |||||||||
| Capital control | ± ^ | + ^ | – ^ | ||||||||||||
| Terms-of-trade volatility | + | + ^^ | – ^^ | + ^ | |||||||||||
| Variability in export growth | + ^^ | + | |||||||||||||
| External variability/openness | – ^^ | – ^^ | |||||||||||||
| Real exchange rate volatility | + ^^ | + ^ | + ^^ | ||||||||||||
| Product diversification | – ^^ | – ^^ | + ^ | ||||||||||||
| Current account | – ^^ | ± ^^ | |||||||||||||
| External debt | + ^ | + ^^ | |||||||||||||
| Growth of domestic credit | + ^^ | + ^ | |||||||||||||
| Money shocks | – ^^ | – | |||||||||||||
| Foreign price shocks | + ^^ | + | |||||||||||||
| (Political/historical factors)1 | |||||||||||||||
| Political instability | + ^^ | + ^^ | – ^^ | + ^ | + ^^ | ||||||||||
| Central bank independence | + | + ^ | |||||||||||||
| Party in office has majority | – ^ | – ^ | |||||||||||||
| Number of parties in coalition | + | + | |||||||||||||
| Coalition government | – | – | |||||||||||||
| (OCA factors)1 |
+ indicates that the coefficient of explanatory variable is positive and - that it is negative; ± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
Table A2.3 (concluded)
+ indicates that the coefficient of explanatory variable is positive and – that it is negative;± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
Table A2.3 (concluded)
| Author | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Coillins (1996) | Edwards (1996) | Edwards (1999) | Rizzo (1998) | Frieden, Ghezzi and Stein (2000) | Berger, Sturm and de Haan (2000) | Poirson (2001) | ||||
| Sample | 24 Latin American and Caribbean countries | 63 countries | 49 developing and middle income | 123 countries | 26 Latin American countries | 65 developing countries | 93 countries | |||
| Time frame | 1978-92 | 1980-92 | 1980-92 | 1977-95 | 1960-94 | 1980-94 | 1990-98 | |||
| Methodology | Probit (panel) | Probit (panel) | Probit (panel) | Probit | Ordered log it (panel) | Probit (panel) | Ordered probit | |||
| Explanatory variables | ||||||||||
| (OCA factors)1 | ||||||||||
| Openness | + ^ | + ^ | _ ^^ | + ^^ | ||||||
| Economic development | • | + ^ | _ ^^ | ± ^ | ± | |||||
| Size of economy | + ^^ | + ^^ | + ^ | |||||||
| Inflation differential | ||||||||||
| Capital mobility | ||||||||||
| Geographical trade concentration | _ ^^ | + | ||||||||
| International financial integration | ||||||||||
| (Other macro/external/ structural factors)1 | ||||||||||
| Growth | + ^^ | + ^^ | + | |||||||
| Negative growth | _ ^ | _ ^^ | ||||||||
| Inflation | + ^^ | + ^^ | + ^^ | + ^ | ||||||
| Moderate-to-high inflation | + ^^ | _ ^^ | ||||||||
| Reserves | _ ^^ | _ ^ | ± ^^ | + ^^ | + ^ | _ ^ | ||||
| Capital controls | ± ^ | + ^ | _ ^ | |||||||
| Terms-of-trade volatility | + ^^ | _ ^^ | + ^ | |||||||
| Variability in export growth | + ^^ | + | ||||||||
| External variability/ openness | _ ^^ | _ ^^ | ||||||||
| Real exchange rate volatility | + ^ | + ^^ | ||||||||
| Product diversification | + ^ | |||||||||
| Current account | _ ^^ | ± ^^ | ||||||||
| External debt | + ^ | + ^^ | ||||||||
| Growth of domestic credit | + ^^ | + ^ | ||||||||
| Money shocks | ||||||||||
| Foreign price shocks | ||||||||||
| (Political/historical factors)1 | ||||||||||
| Political instability | + ^^ | + ^^ | _ ^^ | + ^ | + ^^ | |||||
| Central bank independence | + | - ^ | ||||||||
| Party in office has majority | _ ^ | _ ^ | ||||||||
| Number of parties in coalition | + | + | ||||||||
| Coalition government | - | - |
+ indicates that the coefficient of explanatory variable is positive and – that it is negative;± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
Table A2.3 (concluded)
| Author | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Coillins (1996) | Edwards (1996) | Edwards (1999) | Rizzo (1998) | Frieden, Ghezzi and Stein (2000) | Berger, Sturm and de Haan (2000) | Poirson (2001) | ||||
| Sample | 24 Latin American and Caribbean countries | 63 countries | 49 developing and middle income | 123 countries | 26 Latin American countries | 65 developing countries | 93 countries | |||
| Time frame | 1978-92 | 1980-92 | 1980-92 | 1977-95 | 1960-94 | 1980-94 | 1990-98 | |||
| Methodology | Probit (panel) | Probit (panel) | Probit (panel) | Probit | Ordered log it (panel) | Probit (panel) | Ordered probit | |||
| Explanatory variables | ||||||||||
| (OCA factors)1 | ||||||||||
| Openness | + ^ | + ^ | _ ^^ | + ^^ | ||||||
| Economic development | • | + ^ | _ ^^ | ± ^ | ± | |||||
| Size of economy | + ^^ | + ^^ | + ^ | |||||||
| Inflation differential | ||||||||||
| Capital mobility | ||||||||||
| Geographical trade concentration | _ ^^ | + | ||||||||
| International financial integration | ||||||||||
| (Other macro/external/ structural factors)1 | ||||||||||
| Growth | + ^^ | + ^^ | + | |||||||
| Negative growth | _ ^ | _ ^^ | ||||||||
| Inflation | + ^^ | + ^^ | + ^^ | + ^ | ||||||
| Moderate-to-high inflation | + ^^ | _ ^^ | ||||||||
| Reserves | _ ^^ | _ ^ | ± ^^ | + ^^ | + ^ | _ ^ | ||||
| Capital controls | ± ^ | + ^ | _ ^ | |||||||
| Terms-of-trade volatility | + ^^ | _ ^^ | + ^ | |||||||
| Variability in export growth | + ^^ | + | ||||||||
| External variability/ openness | _ ^^ | _ ^^ | ||||||||
| Real exchange rate volatility | + ^ | + ^^ | ||||||||
| Product diversification | + ^ | |||||||||
| Current account | _ ^^ | ± ^^ | ||||||||
| External debt | + ^ | + ^^ | ||||||||
| Growth of domestic credit | + ^^ | + ^ | ||||||||
| Money shocks | ||||||||||
| Foreign price shocks | ||||||||||
| (Political/historical factors)1 | ||||||||||
| Political instability | + ^^ | + ^^ | _ ^^ | + ^ | + ^^ | |||||
| Central bank independence | + | - ^ | ||||||||
| Party in office has majority | _ ^ | _ ^ | ||||||||
| Number of parties in coalition | + | + | ||||||||
| Coalition government | - | - |
+ indicates that the coefficient of explanatory variable is positive and – that it is negative;± indicates the coefficient is either positive or negative depending on the specification or method used;^^ indicates the coefficient is statistically significant in most cases;^ indicates the coefficient is statistically significant in some specifications; and • indicates not significant but sign not reported by the author.
No result appears to be reasonably robust to changes in country coverage, sample period, estimation method, and exchange rate regime classification. For example, openness—the most frequently analyzed variable—is found to be significantly associated with floating regimes by three studies, significantly associated with fixed exchange rates by three studies, and not significantly associated with any particular exchange rate regimes by another five studies. Per capita GDP is found to be significantly associated with floating regimes by three studies, significantly associated with fixed exchange rates by two studies, and not significantly associated with any particular exchange rate regime by another three studies.
There are a few possible exceptions, notably size of the economy and inflation. Size of the economy turns out to be positively associated with floating in almost all studies, though not always significantly. Inflation is almost always positively and significantly associated with floating. In the case of inflation, however, there are serious questions regarding the functional form of the relationship. In a number of studies, the authors use the inflation rate or the inflation differential rather than their logarithms or similar transformations, leaving open the possibility that the results might be driven by a few influential observations. Moreover, Collins (1996) finds that high inflation affects exchange rate regime choice in the opposite direction than that of low/moderate inflation, and significantly so.
New empirical tests using the Natural classification confirm that it is difficult to explain how countries choose their exchange rate regimes on the basis of simple empirical regularities. These results are consistent with previous work based on other exchange rate regime classifications (Juhn and Mauro, 2002). For a number of potential determinants of regime choice—including economic size, trade openness, and capital controls—the variation across regimes is statistically significant. With the possible exception of economic size and trade openness, however, none of the variables is consistently significant across varying specifications in probit and multinomial logit regression analysis. This suggests that the macroeconomic, structural, and institutional variables postulated in various theories are not robust predictors of exchange rate regime choice. Of course, this does not preclude the potential importance of certain variables for specific groups of countries, in certain time periods, or across some of the regime categories.
Emerging market economies are those that are included in the Morgan Stanley Capital International (MSCI) index, and comprises Argentina, Brazil, Chile, China, Colombia, Czech Republic, Egypt, Hungary, India, Indonesia, Israel, Jordan, Korea, Malaysia, Mexico, Morocco, Pakistan, Peru, the Philippines, Poland, Russia, South Africa, Thailand, Turkey, and Republica Bolivarian de Venezuela. With the exception of Israel, advanced economies are those that are classified as upper-income economies by the World Bank. All other economies constitute the other developing countries group. Small variations in the composition of the emerging markets group do not alter the thrust of the findings reported below on the evolution of regimes and regime transitions. Recognizing the significant variation in financial integration across countries and over time within the emerging markets group, this study also reports results for the 1990–2001 period where relevant.
See Ghosh, Guide, and Wolf (2003) for a description of the de jure classification system, as well as historical data on countries’ classification under this system.
For an early recognition of this concern, see Edwards and Savastano (1999).
The hybrid classification—referred to as the “consensus” classification by Ghosh, Guide, and Wolf (2003)—discards observations in which the de jure classification does not match a de facto one, based on actual exchange rate movements. Effectively, this procedure narrows the sample by 35 percent over the 1970–99 period.
The Levy-Yeyati-Sturzenegger data set, which goes back to 1974, attempts to classify, on an annual basis, about 180 countries in terms of actual flexibility. About one-third of the observations in their sample cannot be classified by their algorithm, however, because of missing data or because the regime was a peg to an undisclosed basket.
See IMF (1999), Section IV, for details. The IMF de facto classification is, in effect, a hybrid classification system that combines data on actual flexibility with information on the policy framework. Using historical data and information on countries’ exchange arrangements, Bubula and Ötker-Robe (2002) put together a database containing IMF de facto classifications going back to 1990.
The Natural classification relies on a broad set of descriptive statistics and detailed country chronologies of exchange rate arrangements to group regimes. As noted by Reinhart and Rogoff, this is analogous to natural taxonomic schemes in biology, where species are grouped according to their characteristics.
Technical aspects of the fine and coarse versions of the Natural classification system are described in Appendix I, which also contains a summary comparison of the various regime classification systems.
In principle, of course, countries with relatively stable exchange rates may have been subject to fewer or smaller other real shocks, including policy shocks, or to shocks that happened to offset the terms-of-trade shocks they experienced.
For example, de facto classifications (other than that of the IMF) do not distinguish unsuccessful pegs—those regimes where the authorities try to peg the exchange rate but are unable to do so. The IMF de facto classification, by contrast, incorporates information on policy intentions and, in principle, retains a forward-looking element.
This does not mean, of course, that formal announcement of a de facto regime does not affect macroeconomic performance. Indeed, as the results described in Section III indicate, the effect of announcing the true de facto regime has been significant for certain regimes.
Unless otherwise noted, all subsequent references to de facto regimes and regimes’ actual operations are to the Natural classification.
Surprisingly, during the run-up to the European Monetary Union, all the euro area countries were listed as intermediate regimes in the de jure classification until 1999.
These data are based on Reinhart and Rogoff (2004), and are not identical to the IMF’s classification of unified versus dual/multiple rates. By multiple exchange rates, Reinhart and Rogoff refer to cases where one or more rates is market determined, as opposed to cases where multiple official rates are all fixed and simply act as a differential tax on a variety of transactions. Another important difference is that dual/multiple markets are typically legal, whereas parallel markets may or may not be legal.
For other developing countries, the increase in de jure floats in the late 1980s and early 1990s was in reality a rise in freely falling regimes, and part of the decline in free floats since the mid-1990s reflected a reduction in freely falling currency values as macroeconomic stabilization progressed in many of these countries (e.g., Azerbaijan, Bulgaria, Iran, the Kyrgyz Republic, and Ukraine).
The definition of hard pegs differs slightly across classifications. In the de jure classification, such pegs constitute monetary unions and currency boards. The Natural classification also includes preannounced pegs. Of the 43 countries listed as hard pegs by the Natural classification in 2001, only five had preannounced pegs, of which only one (Malaysia) was in the emerging markets group. Excluding preannounced pegs from the hard peg category does not affect the finding that hard pegs are more prevalent under the Natural classification than under the de jure. The finding of a general absence of a bipolar tendency among emerging markets in the 1990s (discussed below) is actually accentuated by such an adjustment, however.
Among advanced economies, the proportion of intermediate de facto regimes expanded sharply around the time of the collapse of the Bretton Woods system but shrunk steadily in the 1980s and 1990s as the euro area countries moved toward monetary union. Among emerging market economies and other developing countries, the proportion of intermediate regimes rose markedly in the 1970s, but has remained relatively flat since then.
As the prevalence of de facto pegged regimes has evolved, the choice of anchor currency among peggers has undergone significant change, with virtually all peggers now anchoring to either the dollar or the euro (Box 2.1).
Of all the observations classified as free floats by the IMF de facto regime that were also classified by the Natural classification, only about 27 percent were classified by the latter as free floats, while 18 percent were freely falling regimes, 33 percent were managed floats, 18 percent were limited flexibility regimes, and 3 percent were pegs. About 30 percent of the IMF de facto free floats were not classified by the Natural classification, usually because qualitative evidence suggested the presence of a significant parallel market, but parallel exchange rate data were not available. That said, Bubula and Ötker-Robe (2002) also find, using the IMF de facto classification, that intermediate regimes have been more prevalent than suggested by the de jure classification.
Among advanced countries, however, euro area economies are listed as limited flexibility regimes rather than pegs in the IMF de facto classification, as they were listed in the de jure classification, until 1999 (until 2001 for Greece).
For example, Eichengreen (1994, pp. 4—5) argues that countries “will be forced to choose between floating exchange rates on the one hand and monetary unification on the other.” Obstfeld and Rogoff (1995, p. 74) claim that for countries with an open capital account, “there is little, if any, comfortable middle ground between floating rates and the adoption of a common currency.” More recently, Summers (2000, p. 8) argued that, for economies with access to international capital markets, “the choice of appropriate exchange rate regime . . . increasingly means a move away from the middle ground of pegged but adjustable fixed exchange rates towards the two corner regimes.” Fischer (2001, p. 22) concluded on the basis of the IMF de facto classification that “In the last decade, there has been a hollowing out of the middle of the distribution of exchange rate regimes in a bipolar direction, with the share of both hard pegs and floating gaining at the expense of soft pegs.”
According to the Natural classification, Brazil, Korea, and Malaysia had limited flexibility regimes prior to their recent capital account crises, while Mexico, the Philippines, and Thailand had de facto pegs (but not hard pegs) before their respective crises. Russia was not classified in 1997–98, while Argentina was classified as a hard peg through 2001. Of all the major recent crisis countries, only Turkey had a managed floating regime prior to its crisis.
Hernandez and Montiel (2001) argue that, while several Asian countries have increased the flexibility of their exchange rates in the postcrisis period, generally they have not adopted truly freely floating regimes.
Peru was classified as a de facto soft peg during 1999–2001 by the Natural classification on the basis of a two-year rather than a five-year window to allow for a possible structural break in the variability of the exchange rate toward the end of the sample period. Peru would fall just short of the criteria for a de facto peg in 1999–2001 if a five-year window, which would also span the period prior to 1999, were used.
See also Reinhart (2000). Hausmann, Panizza, and Stein (2001) find that exchange rate volatility declines with the decrease of amounts countries can borrow internationally in their own currency, which the authors consider an indicator of a country’s ability to avoid currency mismatches. The extent of exchange rate pass-through turns out to be less significant. Alesina and Wagner (2003) identify conditions under which countries declare a de jure float but, because of fear of floating, restrict exchange rate flexibility.
These conclusions contrast with the results obtained by Klein and Marion (1997), Eichengreen and others (1998), and Duttagupta and Ötker-Robe (2003) among others, who find the longevity of pegs to be much shorter. This is mainly because the Natural classification attempts to identify longer-term regimes rather than short-term “spells,” which are analyzed in the other studies.
For example, the Natural classification does not treat the 1994 CFA franc devaluation as a change in regime. By contrast, the Levy-Yeyati—Sturzenegger classification, which uses a one-year horizon to measure the variability of the official exchange rate, picks up significantly more transitions: for example, a switch from peg to “dirty float” for each of the CFA franc zone countries in 1994 with a switch back to peg in 1995.
Masson (2001) obtains very similar results for regime transition rates and regime duration using the Ghosh and others (1997) classification, but finds that transitions using the Levy-Yeyati-Sturzenegger classification are considerably more frequent. Masson suggests that the difference in historical transition rates may arise from sampling problems—a fair number of observations are inconclusive in the Levy-Yeyati-Sturzenegger data and thereby omitted—and methodological differences that tend to accentuate de facto flexibility (and hence transition rates) in the Levy-Yeyati-Sturzenegger algorithm in periods of heightened exchange market pressures. Using the IMF de facto classification, Bubula and Ötker-Robe (2002) conclude that intermediate regimes are unlikely to disappear in the future.
These calculations do not treat switches within the pegged category (e.g., from hard to other pegs) as a transition. The average duration of pegs in other developing countries is strongly affected by the CFA franc zone countries, many of which have retained de facto pegs throughout the sample period.
The issue of causation affects potentially the analysis of regime performance: better macroeconomic performance may be associated with certain regimes because countries with strong performance may choose systematically to adopt those regimes. As discussed in Appendix II, however, it is difficult to find empirical regularities between a large set of potential determinants of regime choice—including standard measures of the broader policy context—and between countries’ actual regimes.
While data on market-determined exchange rates are available only for the period 1946–98, Reinhart and Rogoff (2004) were able to classify most countries for the years 1940–45 and 1999–2001 on the basis of official exchange rate data only because few countries had active parallel markets in those years. Observations where the parallel market was known to be substantial but where parallel rate data were not available are marked “unclassified” by the Natural classification.
The advantage of using mean absolute deviations, rather than variances or standard deviations, is that this minimizes the impact of outliers. For example, when the exchange rate is fixed but subject to periodic large devaluations, the variance or standard deviation would overstate the extent of exchange rate flexibility in the period around the devaluation.
When the announced regime is a peg to an undisclosed basket of currencies, tests are done to see if the basket peg is really a de facto peg to a single dominant currency (or to the SDR). If no dominant currency can be identified, the episode is not labeled as a peg. While this suggests that the Natural classification could miss some de facto basket pegs, Reinhart and Rogoff (2004) argue that this is almost certainly not a major issue.
This allows for the identification of relatively short-lived de facto pegs as well as those with a longer duration. For instance, this exercise allowed for identification of the Philippines’ de facto peg to the U.S. dollar during 1995–97 in the run-up to the Asian crisis, as well as the numerous European de facto pegs to the deutsche mark prior to 1999.
In the rare cases where inflation is over 40 percent but the market rate nevertheless follows a confirmed, preannounced band or crawl, the preannounced regime takes precedence.
