III Recent History and Geography of Exchange Rate Volatility

Peter Clark, Shang-Jin Wei, Natalia Tamirisa, Azim Sadikov, and Li Zeng
Published Date:
September 2004
  • ShareShare
Show Summary Details

Measuring Exchange Rate Volatility

In the voluminous literature on exchange rate volatility and trade, there is no consensus on the appropriate method for measuring such volatility. This lack of agreement reflects a number of factors. As noted in the section below, there is no generally accepted model of firm behavior that is subject to risk arising from fluctuations in exchange rates and other variables. Consequently, theory cannot provide definitive guidance as to which measure is most suitable. Moreover, the scope of the analysis will dictate to some extent the type of measure used. If the focus is on advanced countries, then one could take into account forward markets for the assessment of exchange rate volatility on trade, whereas this would not be possible if the analysis extended to a large number of developing countries. In addition, one needs to consider the time horizon over which variability is to be measured, as well as whether it is unconditional volatility or the unexpected movement in the exchange rate relative to its predicted value that is the relevant measure. Finally, the level of aggregation of trade flows being considered will also play a role in determining the appropriate measure of the exchange rate to be used.

This study provides a comprehensive picture of volatility in exchange rates across the entire IMF membership for which data are available. In the empirical analysis, the paper starts with an examination of the relationship between aggregate exchange rate volatility and aggregate trade. Recognizing the limitations of looking at the aggregate data, the study then turns to analyzing the effect of exchange rate volatility on trade across different country pairs and over time. Methodologically, the switch to bilateral trade and volatility allows one to better control a variety of factors other than volatility that could affect trade. As a consequence, the chance to detect an effect of exchange rate volatility on trade improves. Given this methodological approach, the basic building block in the analysis is the volatility in the exchange rate between the currencies of each pair of countries in the sample. For the descriptive part of the study below, which looks at the exchange rate volatility facing a country as a whole, it is necessary to aggregate the bilateral volatilities using trade shares as weights to obtain what is referred to as the “effective volatility” of a country’s exchange rates. This ensures that the measures of volatility in the descriptive and econometric parts of the study are fully consistent.

Such a measure of effective volatility presupposes that the exchange rate uncertainty facing an individual firm is an average of the variability of individual bilateral exchange rates (Lanyi and Suss, 1982). If a trading firm engages in international transactions with a wide range of countries, however, any tendency for exchange rates to move in offsetting directions would reduce the overall exposure of the firm to exchange rate risk. This would argue for using the volatility of a country’s effective exchange rate as the measure of exchange rate uncertainty facing a country. This would seem particularly appropriate for advanced economies, where much trade is undertaken by diversified multinational corporations. This was the approach taken in the original IMF study, which focused almost exclusively on the G-7 countries. The present study, however, covers nearly all developing countries, where the role of diversified firms is less pronounced. For this reason, as well as for consistency with the econometric analysis below, effective volatility is used in the descriptive part of the study.

It is important to realize that the degree of exchange rate variability to which a country is exposed is not necessarily closely related to the type of exchange rate regime it has adopted. A country may peg its currency to an anchor currency but will float against all other currencies if the anchor does as well. Thus, as with effective exchange rates, effective volatility is a multi-dimensional concept (Polak, 1988). Pegging can reduce nominal exchange rate volatility vis-à-vis one trading partner, but it can by no means eliminate overall exchange rate variability. This is shown below, where measured volatility is related to two different classifications of a country’s exchange rate arrangement.

The choice between using nominal and real exchange rates depends in part on the time dimension that is relevant for the economic decision being taken. In the short run, where costs of production are known and export and import prices have been determined, the exchange rate exposure of a firm is a function of the nominal exchange rate. The decision to engage in international transactions, however, stretches over a longer period of time, during which production costs and export and import prices in foreign currency will vary. From this perspective, exchange rates measured in real terms are appropriate. Nonetheless, as nominal and real exchange rates tend to move closely together, given the stickiness of domestic prices, the choice of which one to use is not likely to affect significantly measured volatility or the econometric results. Still, real rates are preferable on theoretical grounds and are used in the benchmark measures of volatility below. Consumer prices are used to construct the real rates because they are the most widely available measures of domestic prices. As a robustness check, results using nominal exchange rates are also reported.

While exchange rates are often highly volatile, the extent to which they are a source of uncertainty and risk depends on the degree to which exchange rate movements are foreseen. When hedging instruments are available, the predicted part of exchange rate volatility can be hedged away and hence may not have much effect on trade. This suggests that the appropriate measure of risk should be related to deviations between actual and predicted exchange rates. One possibility along these lines would be to use the forward rate as a prediction of the future spot rate and to use the difference between the current spot rate and the previous period forward rate as an indicator of exchange rate risk. One problem with this approach is that the forward rate is not a good predictor of future exchange rates. In addition, quotations are available only for the major currencies. More generally, there are a wide variety of methods—ranging from structural models to time series equations using autoregressive conditional heteroskedasticity (ARCH)/generalized ARCH (GARCH) approaches, for example—that could be used to generate predicted values of exchange rates (McKenzie, 1999). As pointed out by Meese and Rogoff (1983), however, there are inherent difficulties in predicting exchange rates. Therefore, this study adopts the approach followed in much of the work on the topic and uses a measure of the observed variability of exchange rates as the benchmark. GARCH estimates are included as an alternative measure of volatility.

The most widely used measure of exchange rate volatility is the standard deviation of the first difference of logarithms of the exchange rate.10 This measure has the property of being equal to zero if the exchange rate follows a constant trend, which presumably could be anticipated and therefore would not be a source of uncertainty. Following the practice in most other studies, the change in the exchange rate is computed over one month using end-of-month data. The standard deviation is calculated over a one-year period as an indicator of short-run volatility, as well as over a five-year period to capture long-run variability.

Finally, it is useful to take note of the role of currency invoicing here. Very often trade between a pair of countries, especially between two developing countries, is not invoiced in the currency of either country. Instead, a major currency—especially the U.S. dollar—is often used as the invoicing currency. It might appear that the volatility of the exchange rate between the currencies of the two trading partners is not the relevant volatility to consider. For example, if Chinese exports to India are invoiced in U.S. dollars, it might seem that the Chinese exporters would only care about the fluctuations between the U.S. dollar and the Chinese yuan, but not between the Indian rupee and the Chinese yuan. This view, however, is not correct. Any fluctuation between the Chinese yuan and the Indian rupee, holding constant the Chinese yuan/U.S. dollar rate, must reflect fluctuations in the Indian rupee/U.S. dollar rate. As the latter could affect the Indian demand for Chinese exports, fluctuations in the Chinese yuan/Indian rupee exchange rate would also affect Chinese exports to India even if the trade is invoiced in U.S. dollars. Generally speaking, the choice of invoicing currency does not alter the effect of exchange rate volatility on trade.

Comparisons Using the Benchmark Measure of Volatility

It is useful to begin the analysis of exchange rate volatility over time and across countries by examining the evolution of fluctuations in exchange rates for broad groups of countries as shown in Figure 3.1.11 This shows the short-run effective volatility since 1970 of exchange rates reported in the IMF’s International Financial Statistics (IFS), converted to real terms using consumer prices for advanced, transition, emerging market, and developing economies.12 As noted in Section I, there were several developments in the international monetary system over this period, including crises in emerging market economies, capital account liberalization, and the breakup of the former Soviet Union, all of which tended to be associated with an increase in exchange rate volatility.

Figure 3.1.Short-Run Effective Volatility of the Real Exchange Rate by Country Groups

(In percent)

First, looking at the how variability has changed over the sample period, it is noteworthy that there is no obvious trend increase over time. In the first three years of the sample period 1970–72, lower-than-average effective volatility is evident for the advanced economies, which reflects the fixed-rate system of most of these countries. Since then, the exchange rates of these countries have exhibited greater volatility, but not markedly so. In fact, the average effective volatility from 1991–2002 is about the same as in 1970–80. There is also no clear upward trend in exchange rate volatility in emerging market economies and developing countries over the entire period. While transition economy exchange rates exhibited much greater variability on average in the 1990–2002 period, this reflects the very large change in exchange rates associated with the breakup of the former Soviet Union and the shift to market economies from 1989 to 1993. The unprecedented high level of volatility during these years was a reflection by and large of adjustments in real exchange rates that were needed to accommodate the structural transformation of these economies. These adjustments now appear to be essentially complete, and in recent years (1999–2002) the effective volatility in their real exchange rates has been less than that of emerging and developing countries.

Second, looking across the major country groupings, it is not surprising that measured volatility is lowest for the advanced economies. This reflects the fact that these countries trade relatively more with each other and that their bilateral exchange rates with each other tend to exhibit smaller fluctuations than with other countries, as discussed below. The lower volatility within the group arises presumably from the greater stability in the economic policies of the advanced economies, as well as from their ability to adjust relatively smoothly to shocks. In addition, the foreign exchange markets in which these currencies are traded are very large and liquid, with instruments available to hedge volatility that enable these markets to clear quickly, thereby dampening potentially large fluctuations in exchange rates.

Figure 3.2 shows the same measure of volatility for the G-7 countries individually, as well as for the group as a whole. While variability is, on average, very similar to that for advanced countries as a whole, there are notable differences. The high average volatility for Japan, at 3.50 percent, is double that of Canada, at 1.75 percent. This latter low figure would appear to reflect the close integration of the Canadian and U.S. economies, as well as the strategy of the Canadian authorities to avoid large swings in the Canadian-U.S. dollar exchange rate over part of the sample period. Also noteworthy is the increased volatility for France, Germany, and Italy surrounding the turmoil in the exchange rate mechanism of the European monetary system in 1991–93 (which also affected the United Kingdom in 1992), as well as a noticeable reduction in effective volatility in the exchange rates of these three countries with the introduction of the euro in 1999.

Figure 3.2.Short-Run Effective Volatility of the Real Exchange Rate for the G-7 Countries

(In percent)

To illustrate the reasons for the relatively low effective volatility of the advanced economies, it is useful to decompose the variability in their exchange rates into the contributions of each of the major country groups. This is done in Table 3.1 for the G-7 countries. First, the table decomposes the effective volatility of each of the G-7 into the share of volatility from each group for the years 1970, 1980, 1990, and 2000, so that the row sum equals the overall effective volatility for that country, shown in the last row. It is clear that with two exceptions, Japan and the United States in 1970, the largest component of volatility is that arising from the exchange rates of the other advanced economies. This reflects, in part, the fact that the trade weights of the industrial countries are very high, as well as the lower volatility of the individual bilateral exchange rates among the advanced countries. This is shown in the second portion of the table, which gives the volatility of the exchange rates of the G-7 within each of the major country groupings, computed with the trade weights summing to unity for each group. It shows that, with only a few exceptions, the volatility of the exchange rates of the G-7 with other advanced economies was less than that of the G-7 exchange rates with the other country groups.

Table 3.1.Short-Run Effective Volatility of Real IFS Exchange Rates in G-7 Countries by Major Country Groups
CountryYearAdvanced EconomiesTransition EconomiesEmerging EconomiesOther CountriesTotal Effective Volatility
Decomposition of Volatility
United States19700.6160.0051.8080.1012.529
United States19801.7510.0210.3640.3212.457
United States19901.3970.0450.6480.2832.372
United States20001.3850.0280.8110.1112.335
United Kingdom19700.6020.0060.3210.1761.105
United Kingdom19801.9570.0290.2510.4152.651
United Kingdom19901.8270.0610.3010.2082.396
United Kingdom20002.5690.0850.3340.1323.119
Volatility Within Groups
United States19700.8061.35811.5891.3182.529
United States19802.9136.2311.5651.9722.457
United States19902.13411.4442.4093.8822.372
United States20002.4782.3462.2361.6702.335
United Kingdom19700.7671.2003.2861.5701.105
United Kingdom19802.4426.6563.7393.2562.651
United Kingdom19902.1378.9543.6603.7092.396
United Kingdom20003.1623.0792.9092.9053.119

As noted above, of the major country groups, the transition economies have had the highest level of exchange rate variability, which was associated with the breakup of the former Soviet Union. Data for this group are shown only starting in 1988 because most of these countries did not exist in the 1970s and 1980s. Only starting in 1995 are data available for all 22 transition countries, and over the period 1995–2002 the exchange rate volatility of this group was comparable to that of emerging market economies and developing countries. Volatility in these latter two groups, while on average not quite double that of the advanced countries for the period as a whole, nonetheless declined between the 1980s and 1990s, especially for the emerging market economies.

Some additional detail is shown in Figure 3.3, which gives a geographic breakdown of developing countries (WEO classification), and in Figure 3.4 for two analytical groups, fuel exporters and exporters of non-fuel primary products.13 Among the geographic regions, sub-Saharan Africa (excluding South Africa and Nigeria) shows the highest average level of volatility of real exchange rates over the sample period, although this may reflect the unusually high 14.5 percent figure in 1994, which is related in large part to the dramatic devaluation of the CFA franc that year. By contrast, the developing countries in Asia have exhibited fairly consistently below-average volatility, especially if one excludes the exceptionally high degree of variability associated with the Asian crisis in 1997–98. For the developing countries in the Western Hemisphere, exchange rate fluctuations have been below average, except for the turbulence associated with the lost decade of the 1980s. Regarding the analytic groupings shown in Figure 3.4, fuel exporters have experienced increasing exchange rate volatility over the sample period, and exporters of non-fuel primary products have had the highest average level of real exchange volatility over the entire period, which likely reflects the effects of movements in the terms of trade of these countries.

Figure 3.3.Short-Run Effective Volatility of the Real Exchange Rate in Developing Countries Grouped by Geographic Region

(In percent)

Figure 3.4.Short-Run Effective Volatility of the Real Exchange Rate in Two Developing Country Groups by Source of Export Earnings

(In percent)

The average figures for the country groups embody wide variations in the level of exchange rate volatility of the individual countries in each group. It is therefore useful to examine the variation across the members in each group. This is done in Table 3.2, which presents figures for the average effective volatility of real exchange rates over the entire sample period 1970–2002 for the five countries with the highest and lowest volatilities.14 As expected, the dispersion of exchange rate volatility across the advanced economies is quite low, compared with the other groups. It is noteworthy, however, that Japan has the highest measured volatility in this group, with another G-7 country—the United Kingdom—ranking fifth. The dispersion is much higher within the other country groups.

Table 3.2.Average Effective Volatility Ranking, 1970–2002
Advanced (Avg = 2.42)Emerging (Avg = 4.43)Other (Avg = 4.59)Sub-Sahara (Avg = 5.89)Developing Asia (Avg = 3.66)
Top Five (Avg = 3.21)Top Five (Avg = 6.70)Top Five (Avg = 16.05)Top Five (Avg = 15.46)Top Five (Avg = 5.44)
Japan3.50Argentina9.36Angola27.32Angola27.32Afghanistan, I.S. of6.82
Australia3.23Mexico5.92Nicaragua13.51Congo, Dem. Rep. of13.07Lao People’s Dem. Rep5.43
New Zealand3.03Peru5.89Congo, Dem. Rep. of13.07Uganda10.77Indonesia4.87
United Kingdom2.81Uruguay5.80Uganda10.77Sudan10.53Sri Lanka4.37
Bottom Five (Avg = 1.78)Bottom Five (Avg = 2.33)Bottom Five (Avg = 1.30)Bottom Five (Avg = 2.60)Bottom Five (Avg = 2.43)
Bel_Lux1.77China, Hong Kong SAR2.38Aruba1.33Cape Verde2.72Malaysia2.40
Netherlands1.81Malaysia2.40Guiana, French1.34Mauritius2.72Fiji2.55
Denmark1.91Thailand2.80China, Macao1.55Cameroon2.75Thailand2.80
Middle East and Turkey (Avg = 4.28)Western Hemisphere (Avg = 4.54)Fuel-Exporting (Avg = 6.18)Non Fuel-Exporting (Avg = 6.15)
Top Five (Avg = 6.65)Top Five (Avg = 9.55)Top Five (Avg = 11.25)Top Five (Avg = 11.65)
Iran, I.R. of8.39Nicaragua13.51Angola27.32Zambia15.59
Lebanon8.27Bolivia10.26Iran, I.R. of8.39Congo, Dem. Rep. of13.07
Yemen, Republic of6.07Argentina9.36Equatorial Guinea7.86Uganda10.77
Syrian Arab Republic5.48Suriname8.11Nigeria6.61Bolivia10.26
Turkey5.04Chile6.52Yemen, Republic of6.07Ghana8.56
Bottom Five (Avg = 2.46)Bottom Five (Avg = 2.11)Bottom Five (Avg = 2.67)Bottom Five (Avg = 2.90)
Malta2.15Panama1.89Bahrain, Kingdom of2.22Mali2.16
Bahrain, Kingdom of2.22Netherlands Antilles2.13Kuwait2.51Liberia2.63
Kuwait2.51Bahamas, The2.14Saudi Arabia2.58Solomon Islands3.10
Saudi Arabia2.58Barbados2.14Oman2.93Guyana3.30
Yemen, Republic of2.85Dominica2.24Gabon3.13Côte d’Ivoire3.31

Table 3.3 provides information on the frequency (number of years) that each country appeared in the top five or bottom five in terms of effective real exchange rate volatility. This indicates which countries exhibited persistently high or low variability over the sample period. The results are often similar to what is shown in Table 3.2; for example, Japan is in the top five advanced economies in 30 out of the 33 years in the sample. Similarly, in the emerging markets group, Argentina is in the top five in 20 of the 33 years.15

Table 3.3.Number of Years in Average Effective Volatility Ranking 1970–2002
AdvancedEmergingDevelopingSub-SaharaDeveloping Asia
Frequency in Top FiveFrequency in Top FiveFrequency in Top FiveFrequency in Top FiveFrequency in Top Five
Japan28Argentina21Congo, Dem. Rep. of17Congo, Dem. Rep. of23Sri Lanka22
New Zealand16Chile14Bolivia9Angola10Samoa17
United Kingdom15Indonesia13Ghana8Uganda9Indonesia16
Frequency in Bottom FiveFrequency in Bottom FiveFrequency in Bottom FiveFrequency in Bottom FiveFrequency in Bottom Five
Belgium Luxembourg28Singapore21Guiana, French22Gabon16Thailand23
Canada19Malaysia18Réunion20Côte d’Ivoire13Fiji22
Netherlands17Venezuela, Rep. Bol.18Netherlands Antilles14Madagascar12Philippines13
Denmark13Mexico17Bahamas, The9Mauritius12Samoa12
Middle East and TurkeyWestern HemisphereFuel-ExportingNon Fuel-Exporting
Frequency in Top FiveFrequency in Top FiveFrequency in Top FiveFrequency in Top Five
Turkey30Argentina19Nigeria25Congo, Dem. Rep. of24
Syrian Arab Republic29Paraguay15Iran, I.R. of20Bolivia12
Iran, I.R. of19Haiti11Venezuela, Rep. Bol.16Sierra Leone11
Jordan17Bolivia11Angola12Burkina Faso11
Frequency in Bottom FiveFrequency in Bottom FiveFrequency in Bottom FiveFrequency in Bottom Five
Malta32Netherlands Antilles20Kuwait25Côte d’Ivoire19
Bahrain, Kingdom of26Panama16Bahrain, Kingdom of24Rwanda17
Kuwait23Bahamas, The13Gabon23Togo15
Saudi Arabia21Mexico13Venezuela, Rep. Bol.22Liberia15
Egypt12Trinidad and Tobago12Saudi Arabia21Bolivia15

Alternative Measures of Volatility

It is useful to compare the benchmark measure of volatility with a number of alternatives. Figure 3.5 provides figures for the short-run effective volatility of the nominal official exchange rate. A comparison with Figure 3.1 shows that, while there are no major differences between these two measures, generally real exchange rate volatility is somewhat higher than nominal volatility. This is particularly the case in 1970, when fixed nominal rates were more widespread and inflation differentials generated larger movements in real rates.16 Lower volatility in nominal exchange rates is also more pronounced among developing countries over the entire sample period, which would appear to reflect the fear of floating described by Reinhart and Rogoff (2002).

Figure 3.5.Short-Run Effective Volatility of the Nominal Exchange Rate by Major Country Groups

(In percent)

Figure 3.6 shows the longer-run measure of exchange rate volatility—namely, the standard deviation of monthly log differences in exchange rates calculated over the five years preceding the year in question. As one would expect, the measured volatility is larger than the average of the short-run volatilities over the same years. Figure 3.7 shows a measure of conditional volatility—namely, that estimated for each currency, assuming it follows a GARCH process. The underlying idea is that part of the volatility can be forecasted, based on past values of the exchange rate, and firms engaged in trade would naturally make an effort to develop such a forecast. Figure 3.7 plots the conditional—or forecasted—exchange rate volatility (for a description of this methodology, see the appendix). A comparison with Figure 3.6 shows that this measure is in general somewhat lower than the standard measure, which is particularly the case for the transition economies in 1995. Figure 3.8 gives the long-run volatility for the G-7 countries.

Figure 3.6.Long-Run Effective Volatility of the Real Exchange Rate by Major Country Groups

(In percent)

Figure 3.7.Long-Run Effective Conditional Volatility of the Real Exchange Rate by Major Country Groups

(In percent)

Figure 3.8.Long-Run Effective Volatility of the Real Exchange Rate in the G-7 Countries

(In percent)

Up to this point, exchange rates in the IFS, i.e., those compiled and reported by the authorities to the IMF, have been used in the analysis. Recently, however, attention has focused on the classification of exchange rate regimes and the appropriateness of using these exchange rates as the basis for such a classification. In particular, Reinhart and Rogoff (2002) have put together an extensive data set for 153 countries of monthly parallel exchange rates that are market determined going back to 1946. They find striking and widespread differences between the official de jure regime, as reported in the IMF’s Annual Report on Exchange Rate Arrangements and Exchange Restrictions (AREAER), and that implied by the information they gathered on actual de facto exchange rate practices.17 As the exchange rates reported by Reinhart and Rogoff may be more representative of the price of foreign exchange at which international trade transactions were conducted, it would also appear worthwhile to calculate exchange rate volatility using these market-determined rates.

In order to compare the volatility implied by both IFS and market-determined rates, it is necessary to use the same set of countries. Because the usable data for real market-determined rates are significantly smaller (from 107 countries) than what is available for real IFS rates (from 172 countries), the benchmark measure of volatility for the latter had to be recomputed.18 This is shown in Appendix Table A7, where the sample period extends only through 1998 because of data limitations. Comparing the benchmark measure of exchange rate volatility with the same measure, but using the larger sample of countries, the evolution of exchange rate volatility over time and between major country groupings is quite similar. The difference in measured volatility for the same country group reflects only the difference in the sample of countries and the fact that the variability of the currencies of the countries included in the larger sample is not the same as that of the currencies in the smaller sample.

Appendix Table A8 shows the benchmark measure of volatility using parallel market exchange rates, which can be compared directly with Appendix Table A7, as both use the same list of countries. It is immediately clear that, in almost all cases, the volatility of parallel market rates is larger than that of IFS rates.19 This is true for advanced countries as well. Even though there are unlikely to be significant differences between IFS and market quotations for the bilateral rates between advanced countries, there would tend to be much larger differences for the bilateral rates between the advanced economies and countries in other groups. The only exceptions occur in 1991, 1992, 1997, and 1998 for transition economies, when movements in IFS rates exceeded those in parallel market rates. It should be noted, however, that the difference between the two measures of volatility declined from the 1970s to the 1990s for all the country groups except emerging markets, where there was a slight increase. This largely reflects the fact that, except for transition economies, the effective volatility of the market exchange rate declined between the 1970s and the 1990s, whereas the volatility of the IFS rate increased for transition and developing economies, remained almost unchanged for advanced countries, but decreased for emerging market economies.

In comparing the volatility of currencies across countries, it is relevant to consider the type of exchange rate regime because this would likely have a bearing on the degree of variability of a country’s currency against other currencies. This is done in Table 3.4, which shows the real effective exchange rate volatility across country groupings in terms of both the official IMF exchange rate classification as well as the Reinhart-Rogoff Natural classification. It is noteworthy that a currency classified as pegged is by no means insulated from exchange rate fluctuations. Indeed, the average effective volatility of freely floating advanced countries (2.94 percent with the IMF classification and 3.09 percent with the Natural classification) is less than the average volatility of pegged currencies of other country groups, except for the emerging market countries in the Natural classification. Also, looking across types of currency regimes within country groupings, limited flexibility confers less exchange rate volatility than pegged, except for the advanced countries under the Natural classification; and managed floating is not associated with a great deal more volatility than pegged regimes. Only freely floating and freely falling regimes have distinctly greater average volatility; the latter category in the Natural classification includes those countries that had annual inflation rates exceeding 30 percent, which not surprisingly caused considerable exchange rate volatility.

Table 3.4.Real Effective Volatility Across Country Groups by Type of Exchange Rate Regime1
Official IMF Classification2
Country GroupsLimited FlexibilityManaged FloatingFreely Floating
Natural Classification3
Country GroupsLimited FlexibilityManaged FloatingFreely FloatingFreely Falling

Based on a sample of 150 countries for the period 1970–2001.

Based on the IMF’s annual publication Exchange Arrangements and Exchange Restrictions, various issues.

Based on a sample of 150 countries for the period 1970–2001.

Based on the IMF’s annual publication Exchange Arrangements and Exchange Restrictions, various issues.

Table 3.5 shows how effective volatility has varied over time by exchange rate regime. Again, limited flexibility is associated with less variability than a pegged regime. If one ignores the 1970s, when the major industrial countries were pegged early in the decade, volatility declined from the 1980s to the 1990s, except in the category freely floating in the Natural classification.

Table 3.5.Real Effective Volatility Across Regimes and Time Periods1
Official IMF Classification2
Limited flexibility2.
Managed floating4.934.754.184.43
Freely floating3.056.955.015.22
Natural Classification3
Limited flexibility2.582.972.882.83
Managed floating3.484.274.164.02
Freely floating3.324.114.644.26
Freely falling7.9913.049.3110.56

Based on a sample of 150 countries for the period 1970–2001.

Based on the IMF’s annual publication Exchange Arrangements and Exchange Restrictions.

Based on Reinhart and Rogoff (2003).

Based on a sample of 150 countries for the period 1970–2001.

Based on the IMF’s annual publication Exchange Arrangements and Exchange Restrictions.

Based on Reinhart and Rogoff (2003).


Figure 3.1 shows equal-weighted averages of the effective volatilities of the exchange rates of the countries in each group, as each individual country is viewed as the unit of interest. Alternatively, one could weight the effective volatility of each country by its trade share. This weighted-average volatility was computed for each group, and the results are not markedly different from what is shown in Figure 3.1.


The list of countries in each group is given in Appendix Table A1. The list of advanced countries follows that in the World Economic Outlook, Table A in the Statistical Appendix, except that the four newly industrialized Asian economies are included in the group of emerging markets. The transition economies comprise the countries in transition in Table A. The group of emerging market economies is a fairly narrow list of 20 countries. All other countries are included in the list of developing countries.


The list of countries in each group is given in Appendix Table A2.


As noted above, the group of transition countries only attained its full complement of 22 in 1995, and so the ranking is only relevant for the 1990s.


The results for Myanmar in Table 3.3 need to be interpreted with caution, given that the bulk of trade appears to occur at the unofficial parallel market rate. Only public sector enterprises, accounting for about 30 percent of reported trade, conduct transactions at the official rate. The parallel market rate, as reported by Reinhart and Rogoff (2002), however, exhibits somewhat greater volatility than the official rate.


It is also interesting to note that the introduction of the European Monetary Union (EMU) in 1999 reduced, but by no means eliminated, effective nominal exchange rate volatility of its three G-7 members. Average nominal effective volatility from 1995–98 before the EMU was 1.91, 2.07, and 2.34 percent, in France, Germany, and Italy, respectively, whereas in 1999–2002, their average effective nominal volatility was 1.41, 1.68, and 1.63 percent, respectively.


The correspondence between the official IMF and the Reinhart/ Rogoff “Natural” regime classifications is shown in Appendix Table A3. Also shown are the distributions of the major country groups by type of exchange rate regime for the IMF classification (Appendix Table A4) and for the Natural classification (Appendix Table A5). It should be noted that since 1998, the IMF’s AREAER reports exchange rate classifications that are based on de facto, rather than de jure exchange rate arrangements. For an analysis that applies retroactively to the de facto classification back to 1990, see Bubula and Ötker-Robe (2002).


The list of countries in each group is given in Appendix Table A6.


The behavior of the two measures of volatility is quite different; the average of the simple correlation coefficient between the official and the parallel real exchange rate volatility measure for each bilateral exchange rate over the entire sample was 0.58. The correlation coefficient between the two measures of one-year volatility in the nominal exchange rate was even lower at 0.45.

    Other Resources Citing This Publication