Exchange-Rate Policies in Emerging Countries: Eleven Empirical Regularities From Latin America and East Asia

Abstract

In this paper I discuss some of the most important lessons on exchange-rate policies in emerging markets during the last 35 years. The analysis is undertaken from the perspective of both the Latin American and East Asian nations. Some of the topics addressed include: the relationship between exchange-rate regimes and growth, the costs of currency crises, the merits of “dollarization,” the relationship between exchange rates and macroeconomic stability, monetary independence under alternative exchange-rate arrangements, and the effects of the recent global “currency wars” on exchange rates in commodity exporters.

For years scholars and policy makers have tried to understand the reasons long term economic performance has been so different in Asia and Latin America. A number of possibilities have been offered, including explanations based on culture, politics, colonial pasts, and institutions. There is little doubt that all of these are important factors that affect long-term growth and income distribution, but perhaps the most-important cause behind the different outcomes in these two regions has to do with economic policies. By and large, the Asian countries have maintained macroeconomic stability (the 1997–1998 currency crises being, of course, an exception) while the Latin American nations have had extremely volatile macroeconomies.¹

It is, possibly, in the area of exchange-rate policies where the contrast between the two regions has been most pronounced. While in the second half of the 20th century almost every Latin American country went from currency crisis to currency crisis, the Asian nations managed—with the major exception of 1997–1998—to maintain exchange-rate stability.

However, during the last decade or so, things have changed significantly. Most Latin American nations seem to have learned the lessons of the past and have avoided two perennial (and related) problems: pegging their currencies at artificially high levels; and defending these pegs even when it was apparent that major adjustments (e.g., depreciation) were needed. This change in policies became particularly evident in 2008–2011, during the so-called Great Recession. Contrary to what many observers feared, the vast majority of the Latin America nations were able to withstand major external shocks—including a sudden (and, as it turned out, short lived) reversal of capital inflows—without experiencing currency collapses or balance of payments crises.

As many authors have noted, a relatively stable (real) exchange rate that does not become overvalued is a key component of outward-oriented, export-based development strategies. In fact, some policy makers have even argued that, in order to encourage the export sector, the emerging economies should have an undervalued currency. This, indeed, has been the policy stance adopted by China. In addition, exchange-rate stability tends to be reflected in a lower “country risk” premium—that is, it is translated into a lower cost of capital.

The purpose of this paper is to analyze some of the most-important lessons pertaining to exchange-rate policies in emerging economies during the last 35 years. The discussion draws from the experiences of both the Latin American countries and the East Asian nations. Before proceeding, however, it is important to clarify that I do not attempt to provide an answer to the question of which is “the” optimal exchange-rate regime for emerging markets. Indeed, the point of departure of my analysis is the recognition that “one size does not fit all” and that different policies are likely to be appropriate for different nations.

I have organized the discussion around eleven empirical regularities, or lessons, on exchange rates in emerging countries. Although I don’t claim that these are the only regularities that apply to these countries, I do believe that these are the most-relevant factors to keep in mind when thinking about exchange-rate policies in developing countries. Some of these regularities are based on abundant historical evidence—mostly those related to currency misalignments and the costs of crises—while others are more recent, and, thus, are based on observations over a shorter time span. There is a broad literature on most of the long-standing regularities. On the other hand, there is very little work (or almost none) on the more-recent ones (Regularities 10 and 11, in particular).

The eleven regularities discussed in this paper may be classified in five areas: the first deals with currency crises (Regularities 1, 4 and 8). The second area (Regularities 2, 3 and 4) is related to the relationship between exchange-rate regimes and economic performance. The third area has to do with the effectiveness of macroeconomic policy under alternative nominal exchange-rate regimes (Regularities 5, 6, 7 and 9). The fourth is related to the costs and causes of exchange-rate misalignment (Regularities 8 and 9). The fifth area (Regularities 10 and 11) relates to the effects of the current “currency wars” on the emerging markets.

1 Regularity Number One: Exchange-Rate Crises are Very Costly

Existing empirical evidence—including the evidence in Reinhardt and Rogoff’s (2009) massive study—strongly suggests that currency crises are extremely costly in terms of growth, unemployment, and inflation.

I define an “exchange-rate crisis” broadly. A “crisis” occurs when there is a large depreciation in the nominal exchange rate (a depreciation that exceeds 20%, in a two-month period), and/or a “sudden stop” or a “current-account reversal,” where a country’s current-account deficit is reduced significantly in a short period of time.²

In Edwards (2004a, b) I analyze this issue using dynamic panel regressions and conclude that major current-account reversals have had “a negative effect on GDP per capita growth, even after controlling for investment, in excess of 4 percentage points.” Freund and Warnock (2005) use a multivariate statistical approach and find that reversals have been associated with a slowdown in economic growth. A similar conclusion is reached by Frankel and Cavallo (2007), using a somewhat different definition of crisis.

In Fig. 1 I present data on (median) GDP per capita growth in the periods surrounding “current-account-reversal” crises. In this Figure a “current-account reversal” is defined in two alternative ways: either as a situation where the current-account deficit declines by at least four percent of GDP in one year, or a situation where the deficit is reduced by at least two percent of GDP in one year. The data in Fig. 1 are broken down for three samples: “large countries”—countries in the top 25% of the world’s GDP Distribution, including a number of Latin American and Asian countries—, “industrial countries,” and “all countries.” As may be seen, in the three samples there is a rather pronounced decline in GDP growth in the year of the external crisis. It is interesting to notice, however, that the drop in the rate of GDP growth appears to be short lived. In the “large countries” and “all countries” samples there is a very sharp recovery in per capita GDP growth one year after the reversal. Non-parametric χ² tests indicate that, in the crisis countries, growth is significantly lower in the years surrounding the crisis than in a control group of counties that have not experienced a crisis (the p-values range from 0.07 to0.00).

Fig. 1

Evolution of per capita GDP growth (median)

In order to analyze this issue further, I estimate a number of regressions on the (potential) effects of “depreciation crises” on short-term growth. In this exercise, two alternative definitions of “crisis” are used: (a), a monthly nominal-exchange-rate depreciation that exceeds the average exchange-rate change for the country in question by three standard deviations; and, (b), a broader definition of “external crisis” that combines in one indicator changes in the nominal exchange rate (depreciation) with changes (declines) in the stock of international reserves. For details on this indicator see Eichengreen et al. (1996) and Edwards (2004a, b). The first variable is denoted Cri xr; the second is denoted Cri_index.

Consider the following Eq. (1) for growth dynamics:

$$ \Delta {g_{{jt}}} = \lambda [{\tilde{g}_j} - {g_{{jt - 1}}}] + \phi {v_{{jt}}} + \gamma {u_{{jt}}} + {\varepsilon_{{jt}}}. $$

(1)

Where ${\tilde{g}_j}$ is the long run rate of real per capita GDP growth in country j; the terms ν _jt and u _jt are shocks, assumed to have zero mean, finite variance and to be uncorrelated among them. More specifically, ν _jt is assumed to be an external terms-of-trade shock, while u _jt captures other shocks, including currency crises. Equation (1) has the form of an equilibrium correction model and states that the rate of growth in period t will deviate from its long run trend because of three types of shocks: ν _jt , u _jt and ε _jt . Over time, however, the rate of growth will tend to converge towards its long run value, with the rate of convergence given by λ. Parameter φ, in Eq. (1), is expected to be positive, indicating that an improvement in the terms of trade will result in a (temporary) acceleration in the rate of growth and that negative terms of trade shocks are expected to have a negative effect on g _jt .³

I estimate Eq. (1) using a GLS two-step procedure. In the first step I estimate a long-run growth equation using a cross-country data set. I use these first-stage estimates to generate long-run predicted growth rates to replace ${\tilde{g}_j}$ in the equilibrium error-correction model (1). In the second step, I estimate Eq. (1) using GLS for unbalanced panels; I used both random-effects and fixed-effects estimation procedures.⁴ The data set used covers 157 countries for the 1970–2006 period.

In Table 1 I present the results from the second-step estimation of the growth dynamics Eq. (1), using random effects.⁵ The estimated coefficient of the growth gap is, as expected, positive, significant, and smaller than one. The point estimates are on the high side, suggesting that, on average, deviations between long-run and actual growth get eliminated at a steady pace. In addition, as expected, the estimated coefficients of the terms-of-trade shock are always positive and statistically significant, indicating that an rise (fall) in the terms of trade results in an acceleration (de-acceleration) in the rate of growth of real per- capita GDP. The point estimate is, in both regressions, 0.08, indicating that a fall of ten percent in the terms of trade of results in a temporary slowdown in the rate of growth of slightly less than one percent.

Table 1 Exchange rate crisis and GDP per capita growth regressions (random effects GLS estimates)

As may be seen from Table 1, in both regressions the coefficient of the currency- crisis variable is significantly negative, indicating that crises result in a significant decline in GDP growth. The point estimates suggest that this decline in growth per capita ranges from 0.91 to 1.27 percentage points in one year. This decline in growth continues until short-term growth converges to its long term value. It is possible that the regression results reported in Table 1 are subject to endogeneity. After all, it may be the case that “devaluation crises” are more likely to occur in countries that have experienced a slowdown in growth than in countries that have not. In order to address this issue I re-estimated the regressions in Table 1 using a GLS random-effects instrumental-variables technique. The results obtained, not reported here because of space considerations, are consistent with those in Table 1: currency crises are costly—the IV point estimates (t-statistics) are −2.11 (2.35) and −1.42 (3.19), respectively.

The results presented here, then, support the idea that external-sector crises—and in particular currency crises—have a significant negative effect on GDP growth. From a policy perspective the message is rather simple: an important objective of macroeconomic policy should be to avoid situations that evolve into currency crises and/or current-account reversals.

2 Regularity Number Two: Countries with More-Flexible Exchange Rates Have Tended to Grow Faster in the Long Run than Countries with Rigid Currency Pegs

Existing empirical evidence suggests that, over the long term, countries with more-flexible exchange-rate regimes—either floating rates or “intermediate regimes” that allow the exchange rate to act as a shock-absorber—have tended to outperform, in terms of GDP growth, countries that have more-rigid nominal exchange rates. Reaching this conclusion, however, has not been easy, nor has it been free of controversies. Until recently, research on the issue of exchange-rate regimes and economic performance was subject to two related limitations. First, the official data—that is, the data provided by the countries, or by international institutions such as the International Monetary Fund (IMF)—are subject to a serious “survival bias.” The problem is that only countries that have successfully defended their peg are included in the “fixed-exchange-rate” category. On the other hand, countries that adopted—but failed to sustain—a fixed exchange rate have usually been classified (at least in the period following the devaluation crisis) as having a “flexible regime.” This situation means that high inflation rates that follow exchange-rate “crashes” are frequently incorrectly attributed to a flexible-rate system, rather than to the failed pegged system. Similarly, a growth de-acceleration that follows a currency crisis has often (and incorrectly) been associated with the new post-fixed rate exchange-rate regime.

A second limitation of traditional studies on the relationship between exchange- rate systems and economic performance is that, for many years, some countries misclassified their exchange-rate regimes. Indeed, some countries that informed the IMF that they had adopted a flexible-exchange-rate regime had a de facto pegged rate. In addition, some countries that, in reality, had flexible regimes some times were labeled as peggers. This misclassification of regimes means that it is not uncommon for analytic results to be incorrectly attributed to a particular regime.

Recent research has dealt with both of these issues. Perhaps the best-known study along these lines is by Levy-Yeyati and Sturzenegger (2003). These authors use data on the volatility of international reserves, the volatility of exchange rates, and the volatility of exchange-rate changes for 99 countries, during the period 1990–1998 to determine the “true” exchange-rate regimes of the countries. The authors undertake a series of cluster-analysis exercises to classify the countries in their sample into five categories: (1) fixed; (2) dirty float/crawling peg; (3) dirty float; (4) float; and (5) inconclusive exchange rate regimes.⁶

Using these de facto exchange-rate classifications, Levy-Yeyati and Sturzenegger (2003) estimated a traditional cross-country growth model to analyze whether the exchange-rate regime affects long term growth. Their results indicate that emerging countries with more-rigid exchange rates experienced slower growth and higher output volatility than countries with more-flexible exchange-rate regimes. They also found that the exchange-rate regime had no effect on output growth or output volatility in industrial countries. Other authors that have reached similar conclusions include Edwards and Levy-Yeyati (2005) and Reinhart and Rogoff (2004).⁷

3 Regularity Number Three: Dollarized Countries do not Outperform Countries with a Currency of Their Own

The recurrence of currency crises in emerging countries during the 1990s and early 2000s generated an intense debate on exchange-rate policies. A number of economists argued that (many) emerging nations should completely give up their national currencies and adopt an advanced nation’s currency as legal tender. This policy proposal has come to be known by the general name of “dollarization.” ⁸ The debate over “dollarization” is, of course, closely related to that on currency unions. Should two or more countries have a common currency? This question has moved into the fore of the policy discussion with the recent—first half of 2010—difficulties faced by Greece, Ireland, Portugal and Spain, and (potentially) other members of the Euro-zone.

There is wide agreement among economists that countries that give up their currencies, and delegate monetary policy to an advanced country’s (conservative) central bank, will tend to have lower inflation than countries that pursue an active domestic monetary policy. Indeed, Engel and Rose (2002), Eichengreen and Haussmann (1999), and Edwards (2001) found that dollarized countries had a significantly lower rate of inflation than countries with a domestic currency.⁹ Moreover, there is agreement that, in countries with perennial macroeconomic instability—including bouts of hyperinflation—, dollarization is likely to end inflationary pressures and provide price stability. This has been the case, for example, in Ecuador and Zimbabwe. Moreover, in these countries, the adoption of dollarization—and the price stability that follows—is very likely to restore incentives and economic growth. The case of Zimbabwe, where the monetary system was de facto “dollarized” in early 2009, is a good illustration of this phenomenon. ¹⁰

There is much less agreement, however, on the effects of dollarization on real economic variables, such as growth, employment and volatility, in more “normal” countries that have not been subject to major and chronic imbalances. According to its supporters, dollarization will positively affect growth through two channels: First, dollarization will tend to result in lower interest rates, higher investment and faster growth (Dornbusch 2001). Second, by eliminating currency risk, a common currency will encourage international trade; this situation, in turn, will result in faster growth. Rose (2000), and Rose and Van Wincoop (2001), among others, have emphasized this trade channel. Other authors, however, have been skeptical regarding the alleged benefits of dollarization. Indeed, according to a view that goes back at least to Meade (1951), countries with a hard peg—including dollarized countries—will have difficulties accommodating external shocks. This fact, in turn, will be translated into greater volatility, and in many cases into slower economic growth.

Until a decade or so ago, there had been few comparative analyses on economic performance under dollarization. Most empirical work on the subject had been restricted to the experience of a single country—Panama; see Guidotti and Olivares (2001), Moreno-Villalaz (1999), Bogetic (2000) and Edwards (2001). Cross-country studies on currency unions have included very few observations on strictly dollarized countries. For instance, the Engel and Rose (2002) data set includes only seven countries that use another nation’s currency, and only two—Panama and Puerto Rico—that use a convertible currency as legal tender, and are thus “strictly dollarized” countries. The study on exchange-rate regimes by Ghosh et al. (1995) does not include nations that do not have a domestic currency. The IMF (1997) study on exchange-rate systems excluded dollarized countries, and the paper by Levy-Yeyati and Sturzenegger (2003) does not include any nation without a central bank.

In a series of papers Edwards and Magendzo (2003, 2006) analyze empirically the historical record of strictly dollarized economies. They investigate whether, as argued by its supporters, dollarization is associated with superior macroeconomic performance, as measured by faster GDP growth and lower GDP-growth volatility. The reason for focusing on strictly dollarized countries is simple: the policy debate in the emerging and transition world focuses on whether these countries ought to adopt an “advanced” country's currency as a way of achieving credibility. For Argentina, for example, delegating monetary policy to the Federal Reserve is very different from delegating it to a Mercosur central bank run by Brazilians and Argentineans.¹¹

In their analyses Edwards and Magendzo (2003, 2006) use treatment-regressions techniques that estimate jointly the probability of being a dollarized country and outcome equations on GDP per capita growth and on GDP growth volatility.¹² See Table 2 for a list of dollarized nations that have enough data available for undertaking this analysis of economic performance. This Table also contains key information on the most- important economic variables for these strictly dollarized countries as well as for a control group of nations with currencies of their own.

Table 2 Dollarized and non-dollarized countries: basic data

The results obtained from these studies may be summarized as follows: (a) with other things given, dollarized countries have had a slightly lower rate of growth than countries with a domestic currency; this difference, although small, is statistically significant. (b) GDP volatility has been significantly higher in dollarized economies, than in with-currency countries.

These results are robust to the technique being used; they hold when instrumental variables are used and when a “matching-coefficients” technique (that pairs every dollarized country with one or more non-dollarized “neighbors” that share their most- important structural characteristics) is implemented. These results, then, indicate that the alleged superiority of “dollarized” regimes is not supported by the data; on the contrary, the data suggest that, when both long-run trend rates of growth and variability around those trends are considered, dollarized nations have fared, on average, more-poorly than countries that have a currency of their own.

4 Regularity Number Four: Countries with Flexible Exchange Rates are Able to Accommodate External Shocks Better than Countries with Rigid Rates

Supporters of flexible exchange rates have argued that, under this type of regime, it is possible to buffer real shocks stemming from abroad. This fact, in turn, allows countries with floating rates to avoid costly and protracted adjustment processes.¹³ Determining whether flexible-exchange-rate regimes are able to insulate the economy from external shocks and, thereby, contribute to improved economic performance is, ultimately, an empirical issue that can be elucidated only by analyzing the historical evidence.¹⁴ This issue has been investigated by, among others, Broda (2004), and Edwards and Levy-Yeyati (2005). See, also, Aghion et al. (2009).

In Table 3 I report the results obtained from the estimation of equations similar to Eq. (1) for countries that fall under four different exchange-rate regimes: (a) flexible; (b) intermediate; (c) pegged; and (d) hard-pegged (including dollarized and currency-union countries). The main purpose of this analysis is to investigate whether the estimated regression coefficients for the terms-of-trade shocks differ across these exchange-rate regimes.¹⁵ In particular, I am interested in finding out whether, as claimed by supporters of flexible rates, this coefficient is smaller for countries with flexible exchange rates than for countries with higher degrees of exchange-rate rigidity. (A smaller coefficient would support the hypothesis that flexible exchange rates act as shock absorbers).

Table 3 Growth dynamics under alternative exchange rate regimes (GLS)

As may be seen from Table 3, the results obtained do support this view: the sum of the contemporaneous and lagged terms of trade coefficients is lowest for flexible- exchange-rate nations, and highest for hard peggers. Intermediate and pegged regimes fall neatly in the middle of these two extreme results.

To summarize, these results, as well as those in Broda (2004) and Edwards and Levy-Yeyati (2005) among others, provide support to the notion that flexible-exchange regimes allow countries to accommodate external shocks, including shocks to their terms of trade.¹⁶

5 Regularity Number Five: Under Capital Mobility and Fixed Exchange Rates, There Is no Room for (Fully) Independent Monetary Policy (The Impossibility of the Holy Trinity)

One of the fundamental propositions in open-economy macroeconomics is that under free capital mobility, the exchange-rate regime determines the ability to undertake independent monetary policy.¹⁷ According to this view, a fixed regime implies giving up monetary independence while a freely floating regime allows for a national monetary policy (Summers 2000). This principle has received the name of the “Impossibility of the Holy Trinity,” and in its simplest incarnation may be stated as follows: it is not possible simultaneously to have free capital mobility, a pegged exchange rate, and an independent monetary policy.

Some authors, however, have argued that, from a strict policy perspective, this is a false dilemma since there is no reason for emerging economies to have free capital mobility. Indeed, the fact that currency crises are often the result of capital-flow reversals—or “capital flight”—has led some observers to argue that capital controls—and, in particular, controls on capital inflows—can reduce the risk of a currency crisis. Most supporters of this view have based their recommendation on Chile’s experience with capital controls during the 1990s. In the aftermath of the East Asian crisis, Joseph Stiglitz was quoted by the New York Times (Sunday February 1, 1998) as saying:

“You want to look for policies that discourage hot money but facilitate the flow of long-term loans, and there is evidence that the Chilean approach or some version of it, does this.”

Policy makers in Asia have, in general, been receptive to the view that emerging countries should be extremely careful in removing capital controls. This view became particularly strong in the aftermath of the Asian crisis of the late 1990s. Consider the following quote from the Asian Policy Forum (2000):

“If an Asian economy experiences continued massive capital inflows that threaten effective domestic monetary management, it may install the capability to implement unremunerated reserve requirements (URR) and a minimum holding period on capital inflows.” (Page 5).

More recently—in January 2011—in the context of large capital inflows into the emerging countries, Olivier Blanchard, the Chief Economist of the International Monetary Fund, said that capital controls on capital inflows, similar to those in place in Chile from 1990 through 1998, could “sometimes” play a positive role in slowing down speculative international flows.¹⁸ Indeed, during 2010 and early 2011 a number of emerging nations, including Brazil, Colombia, and Thailand, took steps towards restricting capital inflows.

From a conceptual perspective the argument for some form of (market-based) controls on capital inflows is simple. It is likely that free capital mobility—where domestic residents can borrow freely from abroad—will generate a “congestion” externality. Borrowers do not realize that by increasing their foreign exposure, they are generating an increase in country risk, and in the cost of borrowing, for everyone. Likewise, one can think that this “congestion” effect results in higher vulnerability for the economy as a whole, and an increase in systemic risk. According to a long tradition in applied welfare economics, it is possible to deal with these types of distortions by imposing a Pigovian tax that moves the economy closer to the undistorted equilibrium. In this context, a tax on borrowing, similar to the controls on capital inflows in Chile, would be warranted.¹⁹

At a more practical level, the question is how successful are these types of controls. Most empirical work—see, for example, Valdés-Prieto and Soto (1998), De Gregorio et al. (2000), Forbes (2005)—conclude that this policy is not overly effective. It affects macroeconomic variables for only short periods of time and only partially. In a recent study, however, Edwards and Rigobon (2009) use a model of exchange-rate behavior under (implicit) bands—such as the ones that Chile had during the controls period—and find that these capital controls did help reduce nominal- exchange-rate volatility over the long run. Quantitatively, speaking the effect was rather small.

Many of the early empirical works on the “Impossibility of the Holy Trinity” relied on the estimation of “offset coefficients.” Although these have tended to be lower than one, suggesting less-than-complete loss in monetary independence, they are significantly positive. This fact suggests that attempts to maintain an exchange rate that is undervalued in real terms—as has been the case of the yuan since the late 1990s—has important implications for monetary policy. An undervalued currency will tend to result in a current-account surplus and, in most cases, in the accumulation of international reserves. This outcome will generate monetary and inflationary pressures. The traditional way of dealing with this problem is by sterilizing the international-reserves changes. A limitation of this approach, however, is that sterilizing reserve changes may result in (rather large) financial costs. The magnitude of these costs will depend on a number of factors, including the extent of real undervaluation, the monetary-policy stance as reflected by interest rate differentials on domestic and foreign securities, the extent of capital mobility, and the degree of substitutability of domestic and foreign financial assets.

It is important to notice that situations of accumulation and decumulation of international reserves are asymmetric. While the former will come to an end when the country “runs out of reserves,” and, thus, faces a (major) devaluation crisis, reserve accumulation may continue for a long period of time. As pointed out above, this fact does not mean that maintaining an undervalued currency is a costless policy; indeed, since international reserves earn a very low rate of return and the interest costs of sterilization tend to be significant, the net effect is a loss for the sterilizing central bank. The recent (2010 and 2011) increases in bank reserve requirements by China’s central bank is an attempt to deal with the (expansive) monetary consequences of reserves accumulation.

6 Regularity Number Six: Exchange-Rate Policies Based on Inflation-Rate Differentials Result in a Loss of Anchor and in Macroeconomic Instability

An exchange-rate policy that consists of adjusting the nominal exchange rate in proportion to (lagged) inflation-rate differentials is inherently unstable and results in a loss of the macroeconomic anchor. Although this type of exchange rate system, known as “crawling peg,” is currently out of vogue, a number of voices,—including voices in Latin America, Asia and international think tanks—clamor for its return.

At the simplest possible level a “crawling-peg” policy may be summarized as follows.

$$ d\log {E_t} = \phi {(}d \log {P_{{t - 1}}} - d\log P_{{t - 1}}^{ * }{)} $$

(2)

Where d logE _t is the rate of (policy-determined) nominal-exchange-rate adjustment, d logP _t−1 is lagged domestic inflation, $d\log P_{{t - 1}}^{ * }$ is lagged foreign inflation, and ϕ is a factor of proportionality.

From the late 1960s through the 1980s, it was thought that, by adjusting the nominal exchange rate by the difference between domestic and international inflation, it was possible to avoid disequilibria—and, in particular overvaluation—, and achieve stability. The policy that implies ϕ = 1 in Eq. (2) is known as a “strict backward-looking crawling peg,” and is sometimes referred to as “maintaining a realistic real exchange rate.”

There is abundant empirical evidence, however, suggesting that this policy tends to generate significant inflationary inertia. More specifically, if a regime that fully adjusts the nominal exchange rate to past inflation—one where ϕ = 1—is implemented alongside a backward-looking wage-indexation system, the economy will lose its anchor. From a technical point of view, this situation means that the inflationary process will be characterized by a unit root, and will have an infinite variance. In this case inflation may wander around aimlessly, and achieve any level; this point was made by Felipe Pazos (1972), one of the fathers of the Latin American economics profession. A good example of a country that lost its anchor, and had trouble re-establishing it, is Brazil before the stabilization program and monetary reform of the mid 1990 s that introduced the real as a new currency. This Regularity is intimately related to the next one, on how to combat inflationary inertia.

7 Regularity Number Seven: Exchange-Rate-Based Stabilization Programs Are Dangerous, and Have Often Generated Severe Currency Overvaluation

Historically, many cases of inflationary inertia have been tackled by implementing “exchange-rate-based stabilization programs.” These programs consist of pegging the nominal exchange rate as a way of introducing price discipline. This approach to reducing inflation has been particularly popular—and costly—in the Latin American nations (see Edwards 2010a, b, for an extensive discussion and for detailed analysis of several case studies).

The rationale underlying this policy is simple: if exchange-rate indexation contributed to inflationary inertia (this is Regularity 6), pegging the nominal exchange rate will help eliminate it. Chile and Mexico are two examples of the many Latin American countries that during the 1970s, 1980s and early 1990s tried to deal with a very high degree of inflationary inertia by adopting pre-announced exchange-rate paths. In both of these countries the pre-announced rate of devaluation was set at a significantly slower pace than the ongoing rate of inflation. While the Chilean system converged to a strict peg exchange rate (at 39 pesos per dollar), the Mexican system was characterized by a narrowing nominal exchange-rate band.²⁰ Neither of these programs, however, succeeded. In both countries inflation continued at a significant pace, pushing domestic costs—including wages—upward. This process generated overvalued real exchange rates and a decline in international competitiveness. In both cases the final outcome was a major and costly currency crisis—in Chile in 1982 and in Mexico in 1994–95. In many ways, Argentina’s experience with its quasi-currency board during 1991–2001 is an extreme case of an exchange rate-based stabilization program.

The rationale behind these exchange programs—and their likely failure—may be explained as follows: Initially the economy is characterized by a high and persistent inflationary process, fed by a crawling peg exchange rate rule such as the one summarized in Eq. (2). This rule, in turn, has been put in place as a way of avoiding real exchange rate overvaluation. At some point, however, there is a switch in the government preferences, and the authorities announce that from that point on their main priority is defeating inflation. In order to achieve the goal of eliminating inflation, a fixed exchange rate regime is put in place.

Under these circumstances, the time path followed by inflation will depend on the degree of credibility of the nominal-exchange-rate anchor policy. It is likely that, at least initially, the public will have some doubts about what the new regime will be—either a genuine fixed-rate system, or a pegged-exchange-rate regime with an escape clause. If the public doubts the sustainability of the fixed exchange rate, it will consider that there is a positive probability that the authorities will abandon the pegged rate and will revert to the old crawling-peg regime. It can be formally shown that, after the exchange rate has been pegged, the (remaining) degree of inertia will be (approximately) equal to the perceived probability that the program will be abandoned. Indeed, this situation prevailed in both Chile and Mexico; in both countries credibility was low and inertia was barely affected by the adoption of the exchange-rate-based program. As a result, inflation persisted, even if the nominal exchange rate was fixed, and an acute degree of overvaluation emerged.²¹

Pegging the currency value at the wrong level was a recurrent event throughout Latin America during the 1990s and early 2000s.²² Some of the countries that made that mistake during the last 25 years include Argentina, Brazil, Colombia, the Dominican Republic, Mexico, Uruguay and Venezuela. In many ways, clinging to the notion of fixed exchange rates in the 1990s was similar to the obsession with the gold standard during the interwar period. As Liaquat Ahamed pointed-out in his 2009 book Lords of Finance, the attachment to fixed currency values during the interwar period—in those years currencies were fixed to gold—was at the heart of the economic maladies of the time, including the triggering and magnification of the Crash of 1929 into the Great Depression.²³

In the case of Latin America, what makes this recurrent currency-policy mistake surprising is that a number of prominent economists had argued, during the late 1980s and early1990s, that pegging the exchange rate to reduce inflation was a policy fraught with dangers. For example, referring to Mexico’s policies in the early 1990s, Rudi Dornbusch wrote:²⁴

“Exchange rate-based stabilization goes through three phases: The first one is very useful (…) [It] helps bring under way a stabilization…In the second phase increasing real [currency] appreciation becomes apparent, it is increasingly recognized, but it is inconvenient to do something (…) Finally, in the third phase, it is too late to do something. Real [currency] appreciation has come to a point where a major devaluation is necessary. But the politics will not allow that. Some more time is spent in denial, and then—sometime—enough bad news pile up to cause the crash.”

One could argue that Mexican and other Latin American policy-makers should have been aware of the dangers of fixed exchange rates during a disinflation effort. After all, during the early 1980s, Chile had a traumatic experience with that policy. As discussed in great detail in Edwards and Edwards (1991), during 1979 the so-called “Chicago Boys” tackled Chile’s stubborn inflation—which at the time lingered around 35 percent per year—by pegging the value of the currency at 39 pesos per U.S. dollar. During the next 24 months the country went through the three phases laid down by Rudi Dornbush in the above quote. Inflation declined slowly, capital inflows skyrocketed, exports struggled, the inflation-adjusted value of the currency strengthened significantly, and a huge trade deficit developed. This assault on competitiveness was compounded by labor legislation passed in 1981 that, literally, outlawed reductions in inflation-adjusted wages. In early1982, partially in response to a slowdown of the global economy, international investors abruptly reduced their exposure to Chile. This sudden stop of capital inflows was followed by a major currency devaluation, negative growth, a significant increase in unemployment—in 1983 the rate of unemployment exceeded 20 percent—, and massive bankruptcies. The three key lessons from this episode were that an artificial strengthening of the currency had to be avoided, that rigidly fixed exchange rates were dangerous during a disinflation process, and that this danger was extreme if wages were mandated to increase at an unsustainable pace.

8 Regularity Number Eight: Real Exchange Rate Misalignment Can Be Very Costly; Central Bank Intervention May Be Justified from Time to Time

The emerging countries’ currency crises of the 1990s and early 2000s underscored the need of avoiding overvalued real exchange rates—that is, real exchange rates that are incompatible with maintaining sustainable external accounts. This important point has once again become very relevant in light of the recent (2010) Greek crisis and of the difficulties faced by other Eurozone nations, including Portugal, Ireland and Spain.

As pointed-out above, one of the most significant historical cases of costly exchange-rate overvaluation is that of Mexico during the first half of the 1990s. The overvaluation of the Mexican peso before the December 1994 crisis has been documented in a number of post-crisis studies. According to Sachs et al. (1996), for example, during the 1990–94 period the Mexican peso was overvalued, on average, by almost 29 percent (see their Table 9). An ex-post analysis by Ades and Kaune (1997), using a detailed empirical model that decomposed changes in fundamentals into permanent and temporary changes, indicated that, by the fourth quarter of 1994, the Mexican peso was overvalued by 16 percent. According to Goldman-Sachs, in late 1998—a few months before its crisis—the Brazilian real was overvalued by approximately 14%.²⁵

After the Mexican and East Asian crises, analysts in academia, the multilaterals, and the private sector redoubled their efforts to understand real-exchange- rate behavior in emerging economies. Generally speaking, the RER is said to be “misaligned” if its value exhibits a (sustained) departure from its long-run equilibrium. The latter, in turn, is defined as the real exchange rate that, for given values of “fundamentals”—terms of trade, interest-rate differentials, productivity differentials, fiscal stance, degree of openness, and so on—, is compatible with the simultaneous achievement of internal and external equilibrium.²⁶

There have been two important, and related, developments in recent analyses of real-exchange-rate misalignment: First, there is a generalized recognition that, even under flexible regimes, the exchange rate may become misaligned. There are several reasons for this possible outcome: (a) market participants may have incomplete information; (b) there may be a “herd instinct” among investors; and/or (c) in the short term there may be significant departures of “fundamentals” from their long-run equilibrium values. It is important to notice, however, that the extent of misalignment is much smaller under flexible regimes than under pegged systems. Second, most recent efforts to assess misalignment have tried to go beyond simple versions of purchasing-power parity (PPP), and to incorporate explicitly the behavior of variables such as terms of trade, openness, real interest rates and productivity growth.²⁷

One of the most common methods for assessing real exchange rates is based on single-equation, time-series econometric estimates. In the late 1990s Goldman-Sachs (1997) implemented a real-exchange-rate model (largely) based on this methodology.

It is interesting to consider what this type of models said with respect to the extent of overvaluation in East Asia, just before the 1997–1998 crisis. In this regard, the Goldman-Sachs model suggested that overvaluation had been persistent in most East Asian countries for a number of years: in Indonesia the real exchange rate had been overvalued since 1993, in Korea in 1988, in Malaysia in 1993, in the Philippines in 1992, and in Thailand since 1990. In 2000 J.P. Morgan (2000) unveiled its own real-exchange-rate model. In an effort to capture better the dynamic behavior of real exchange rates this model went beyond the “fundamentals,” and explicitly incorporated the role of monetary variables in the short run. During the last few years JP Morgan has continued to improve its analyses, and introduced more-accurate point estimates of the coefficients for different fundamentals.

An alternative approach to evaluate the appropriateness of the real exchange rate at a particular time consists of calculating the sustainable current-account balance as a prior step to calculating the equilibrium real exchange rate. Deutsche Bank (2000) used a model along these lines to assess real-exchange-rate developments in Latin America. According to this model, the sustainable level of the current account is determined, in the steady state, by the country’s rate of (potential) GDP growth, world inflation, and the international (net) demand for the country’s liabilities. If a country’s current-account deficit exceeds its sustainable level, the real exchange rate will have to depreciate in order to help restore long-run sustainable equilibrium. Using specific parameter values, Deutsche Bank (2000) computed both the sustainable level of the current account and the degree of real-exchange-rate overvaluation for a group of Latin American countries during early 2000. This approach has also been used in recent efforts to compute the degree of misalignment of the U.S dollar—see, for example, Obstfeld and Rogoff (2005), Mann (1999), Edwards (2005), Williamson (2007), Cline and Williamson (2010), and Cline (2010).

There are three fundamental costs of (real) exchange rate misalignment: First, it results in resource misallocation. Second, misalignment will also affect monetary policy. Situations of overvaluation result in losses of international reserves and, thus, in a decline in the monetary base; undervaluation, on the other hand, usually results in international reserves accumulation. In order to maintain monetary policy within reason the central bank needs to sterilize these reserves changes, usually at a non-trivial cost. Third, and most importantly, situations of severe and persistent overvaluation often lead to deep currency crises and to a precipitous drop in growth and employment.

The fact that misalignment may be costly even under a flexible-exchange-rate regime provides justification for central-bank intervention in the foreign-exchange market. A key question—and one that has not been resolved fully—refers to the appropriate extent, frequency, and modality of intervention (sterilized or non-sterilized; spot or forward; direct or indirect). Generally speaking, the authorities have to balance the costs and benefits of active intervention and keep in mind that massive interference with market forces will transform a flexible regime into a disguised pegged regime—this phenomenon is what has been called “fear of floating.”²⁸ For intervention to be effective under a flexible regime, it needs to be infrequent, well justified, fully explained to the public, and based on a firm belief that the market exchange rate is (significantly) out of line with respect to its long-term equilibrium value.²⁹

In a series of important contributions, John Williamson and his colleagues at the Petersen Institute for International Economics have used a variety of methods to calculate what they call the FEER, or Fundamental Equilibrium Exchange Rate—see, for example, Cline and Williamson (2010). Naturally, once an estimate of the long-term value of the equilibrium exchange rate is available, it is possible to determine, using simple comparisons, whether a particular country’s currency is out-of-line or overvalued (see Williamson 2007). This type of analysis has been used both to assess economic policies and to guide political discussions on the international-economic relationship between countries, including between the United States and China. (See the discussion on Regularity Number 9, below, for more on the Chinese yuan).

9 Regularity Number Nine: There is an Asymmetry Between Overvaluation and Undervaluation

On July 21st, 2005, China announced that it was abandoning its decade-long policy of pegging its currency to the U.S. dollar at 8.28 yuan per dollar. In the following weeks the yuan rapidly appreciated relative to the greenback. A new regime based on a basket peg, a maximum daily exchange rate adjustment of 0.3% and bands of unknown width was put in place—or so it seemed. For years the Chinese authorities had been under pressure from the United States and European nations to reform their currency regime. There was consensus among analysts that the yuan was undervalued and that it provided China an unfair advantage in world international markets.

Initially there was some confusion about what China’s “currency reform” really meant. Some analysts argued that it was the beginning of a new era, where the yuan would (mostly) reflect market forces; others argued that the extent of the reform was limited and that a significant change in the value of the yuan was unlikely. A few weeks after the reform was unveiled, the Chinese authorities stated that there would be no further adjustments of the yuan-dollar exchange rate in the immediate future. In spite of this announcement, the yuan continued to strengthen relative to the US dollar until the eruption of the sub-prime global financial crisis in 2007.

In mid 2008 the yuan/USD rate was, de facto, pegged at 6.8 yuan per dollar. At the time almost every observer agreed that the Chinese currency was undervalued by at least 20%—see the studies in Goldstein and Lardy (2009) for a series of estimates.

During early 2010 China’s exchange-rate policy once again became politicized. A number of members of the U.S. Congress asked the Treasury Department to label China an “exchange-rate manipulator.” At the same time, senior Chinese officials accused the U.S. of protectionism and of being an international bully. Currency-value issues have indeed been at the center of political discussions between the leadership of the U.S. and China. Using an eclectic approach, Subramanian (2010) calculated that in mid 2010 the extent of undervaluation of the yuan was close to 30%.

From an economic perspective, two important, and interrelated, points may be made: first, there is a clear asymmetry between situations of currency undervaluation and currency overvaluation. The latter, simply cannot be sustained, and a country that pursues such policy will have to impose increasingly tighter protectionist measures and will, eventually, experience a currency collapse. Countries with currency undervaluation, on the other hand, will suffer inflationary pressures (as discussed above) and will face the costs of sterilization. They could, however, sustain such a situation for a relatively long period of time. Second, and as it has been recently pointed-out by a number of analysts, strongly undervalued currencies in very large countries—such as China and Germany—impede the achievement of global balances. The reason is simple: large current-account surpluses—which are usually associated with undervalued currencies—need to be offset in the global economy by large deficits. The problem, of course, is that large deficits result in a dynamic of net international investment positions (NIIPs) that is not sustainable in the long run. This fact suggests that, eventually, China’s currency will have to move closer to its long-term equilibrium. Although it is not completely clear by how much it would have to strengthened, some calculations suggest that the most reasonable figure would be in the 15%– to −20% range. (Goldstein and Lardy 2009; Cline and Williamson 2010)

10 Regularity Number Ten: Even under Flexible Exchange Rates the Scope of Independent Monetary Policy Is Limited

According to traditional views—going back, at least, to the Mundell-Fleming model—, countries with fixed regimes cannot undertake independent monetary policy (this is Regularity 5 in this paper); on the other hand, and according to this view, countries with flexible exchange rates are able to conduct independent monetary policy. More recently, however, some authors, including Frankel et al. (2004) and Edwards (2010a, b), have argued that countries with flexible exchange rates do not have true monetary autonomy.³⁰

One way of assessing the degree of monetary independence under flexible- exchange-rate regimes is to analyze the extent (and speed) with which changes in policy interest rates in the United States—or another advanced nation, for that matter—are transmitted to emerging countries. Consider the following equation for the dynamic behavior of interest-rate differentials:

$$ {x_t} = \alpha \theta + (1 - \theta ){x_{{t - 1}}} + \sum\limits_{{i = 1}}^k {{\gamma_i}{y_{{it}}} + \lambda \Delta {x_{{t - 1}}}} + \sum\limits_{{j = 1}}^m {{\varphi_i}{z_{{it}}}} + {\psi_t}. $$

(3)

Where x _t is the differential between the (short-term) domestic interest rate in an emerging country and the short-term interest rate in an advanced nation, properly adjusted by expectations of depreciation and risk:³¹

$$ {x_t} = {r_t} - r_t^{*} - {\delta_t} - {\rho_t}, $$

(4)

r _t is the domestic-currency nominal interest rate for securities of a certain maturity, $r_t^{*}$ is the (international) nominal interest rate on foreign-currency-denominated securities of the same maturity, δ _t is the expected rate of depreciation of the domestic currency, and ρ _t is a measure of country risk. y _it is a vector of zero-mean, finite-variance shocks. In this analysis the most important (but not the only) y _it refers to changes in the Federal Funds interest rate.

In long-run equilibrium, and with perfect capital mobility, this interest rate differential should be close to zero.³² The speed at which convergence to long-term equilibrium takes place is captured by parameter θ, and will depend on specific countries’ conditions, but under free capital mobility it should be rather fast. γ _i , and λ are parameters, α, and ϕ _i are coefficients, the z _it are other possible level (as opposed to zero-mean shocks) determinants of the equilibrium long-run interest-rate differential, and ω _t is an error term with the usual characteristics.

The coefficient of the change in the Federal Funds interest rate in Eq. (3) captures the extent of monetary independence. If the Fed’s actions have no effect on the emerging country, the coefficient of the change in the Fed Funds rate will be −1.0. If the Fed’s policy affects domestic interest rates the coefficient will be negative but smaller than one. In the limit, if the Fed’s policy is fully transmitted into the emerging nation, the coefficient will be zero.

In Table 4 I report separate results from pooled regressions, using weekly data, for two groups of countries: four from Latin America—Brazil, Chile, Colombia, and Mexico—, and three from East Asia—Indonesia, Korea, and the Philippines. These regressions use data for the period spanning from the first week of January, 2000, to the last week of September of 2008 (just before the global financial crisis). All seven of these countries had some form of flexible exchange rates during the period under study.

Table 4 Interest rate differentials panel estimates: Latin America and Asia

As may be seen, for most coefficients the results are very similar across the two regions. There is an important difference, however, in the coefficient of the change in the Federal Funds policy rate. While it is negative and significant at conventional levels for the Latin American countries—with a point estimate of −0.75—, it is insignificantly different from zero for the East Asian countries (with a point estimate 0.225). The rest of the coefficients of interest, however, are significant in both equations, and the point estimates are not significantly different across the two regions.

These differences across the Latin American and Asia results have implications for the dynamics of interest rate differentials in response to an increase in the Fed’s policy rate. For example, the results in Equation 4.1 of Table 4 indicate that the dynamic adjustment in Latin America will be characterized by an immediate decline in the interest-rate differential, followed by a cyclical convergence to the new longrun equilibrium. That is, in Latin America there is downward overshooting. The estimates in Equation 4.2 of Table 4, on the other hand, indicate that in East Asia the adjustment toward a new equilibrium will be gradual and smooth; in contrast with the case of Latin America, there will be no downward overshooting or cyclical adjustment.

Even if the dynamic adjustment path is quite different, the final impact on interest-rate differentials turns out to be very similar in both regions. Indeed, the following conclusions emerge from these estimates:

• In the case of Latin America, a fifty- basis-point increase in the Fed’s policy rate results in a decline in the long-term equilibrium interest rate differential of 25 basis points.

• In East Asia, on the other hand, the same policy shock by the Fed results in a decline in the equilibrium interest-rate differential of 26 basis points.

These results indicate quite clearly that in recent times, and in spite of having adopted exchange-rate flexibility, these seven Latin American and Asian countries have not enjoyed complete monetary independence. The estimates reported here indicate that there is a rapid—although not complete—“pass through” from the Fed’s policy interest rate to these nations’ short term domestic interest rates.

11 Regularity Number Eleven: There Has Been an Important Structural Break in the Relationship Between the USD Real Effective Exchange Rate and the Real Effective Exchange Rates of the Emerging Commodity Exporters

In the aftermath of the 2007–2009 global financial crisis, policy-makers throughout Latin America (and other commodity exporting countries, for that matter) have become concerned about the (real) strengthening of their currencies. Real-exchange-rate (RER) appreciations throughout the region have reduced competitiveness, hurt export, and generated political pressure from tradable-goods’ producers. Some countries ( e.g., Argentina, Brazil, Colombia, Peru) have reacted to these developments by intervening heavily in the foreign exchange market, and some (Argentina, Brazil, Colombia) have introduced or stepped-up capital controls.. Of course, real currency appreciation is not a new phenomenon in Latin America. Indeed, the region’s economic history has been punctuated by wide currency cycles. This issue was investigated in detail in, among other places, the celebrated paper by Calvo et al. (1993).

A number of policy analysts have argued that these recent real appreciations are very different from historical experiences of Latin American overvaluation. Indeed, an increasingly accepted story about recent episodes goes as follows: for a variety of reasons—including an unsustainable current-account deficit and an overly expansive monetary policy by the Fed (i.e., QE2)—the USD needs to depreciate with respect to a basket of the currencies of U.S. trading partners. However, because of China’s policy of controlling the value of the renminbi, the USD cannot drop in value relative to the currency of its second-most-important partner. Consequently, and in order to achieve the required trade-weighted correction, the USD needs to over-depreciate significantly with respect to other currencies—including those in Latin America. In this story, then, the strengthening of the Latin currencies is largely exogenous and does not respond to particular policies or attributes of the Latin American countries themselves. Under these circumstances, the losses in competitiveness are the result of “collateral damage” from the global currency wars (mostly—but not exclusively—between the US and China).³³

Two important questions emerge in this context: First, is there, indeed, a negative relationship between the USD real effective (basket) exchange rate, and the Latin American countries’ REER’s? And second, if this negative relationship does exist, is it a new phenomenon?

In order to address these issues I analyze the correlation between the USD real effective exchange rate and the REER of eight Latin American countries—Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, Peru and Venezuela—in the period spanning January 1976 and November 2010. In particular, I am interested in investigating whether periods of real appreciation (depreciation) of the (trade-weighted) USD have been associated with periods of real depreciation (appreciation) in the Latin American currencies. Of course, for this question to be meaningful—and not a mere tautology—, each real exchange rate has to be defined relative to country-specific baskets. Moreover, the weights in the different RER basket indexes have to be different across countries.

In performing this analysis I break up the 1976–2010 period into three sub periods:

• (a) January 1976–December 1989: This period is characterized throughout Latin America by slow growth, external shocks, rapid inflation, and major external crises (including large devaluations). This period culminated with the so-called “Lost Decade,” where the region experienced a decade of negative per capita growth. The implementation of the Brady Plan marked the end of this period.³⁴

• (b) January 1990–December 2003: This period corresponds to the so-called “Washington Consensus” reforms. In most Latin American nations trade was opened to international competition, basic market-oriented reforms were enacted, sweeping privatization programs were put in place, fiscal deficits were clipped, and inflation was greatly reduced. However, during most of this period many Latin American countries maintained some form of predetermined (or pegged) nominal-exchange-rate regimes. These included narrow bands, fixed exchange rates, currency boards, and crawling pegs. One of the most salient characteristics of this period is that many countries (Argentina, Brazil, the Dominican Republic, Ecuador, Mexico, and Uruguay) experienced major currency crises; and, . (c), January 2004–November 2010: This sub-period may be called the “New Epoch.” During this period the Latin American countries experienced a revival and posted solid growth. One of the region’s most important achievements during these years is that it sailed through the global financial crisis without experiencing major setbacks. This ability to survive global financial upheaval was new to Latin America and was the result of a combination of factors, including the abandonment (in most countries) of rigid nominal-exchange-rate regimes, the sizable accumulation of international reserves in the previous decade, and prudent fiscal policies. In addition, during this period commodity prices soared, and most Latin American countries experienced significant improvements in their terms of trade.³⁵

In Table 5 I present correlation matrixes for (the logs) of nine trade-weighted RER indexes—those of eight Latin America nations and the United States. The number below each correlation coefficient is a t-test statistic for the null hypothesis that the coefficient is zero. The data are provided for the three sub samples under consideration. In interpreting the results I concentrate on the first column for each panel of Table 5, on the (partial) correlation between the log of each Latin currency and the USD. The correlations across the RERs for the Latin countries are of interest in themselves and reveal significant information. Analyzing them is, however, beyond the scope of this paper and, thus, will not be discussed here.

Table 5 Coefficients of correlation between nine RER indexes in the Americas: three sub-samples, 1976–2010

The results in this Table show that through time there has been an important change in the extent and direction of the correlation between the trade-weighted RER in the U.S. and the Latin American nations. The results may be summarized as follows:

• In the initial period (1976–1989) four of the correlation coefficients are significantly positive (those of Chile, Colombia, Ecuador and Venezuela), three of the coefficients are significantly negative (those of Argentina, Brazil and Mexico), and one (Peru) is insignificantly different from zero. That is, during this period neither positive nor negative correlation dominates. Further, the magnitude of each of the coefficients is quite small.

• During the middle sub period (1990–2003) there are still four coefficients that are significantly positive (those of Chile, Ecuador, Mexico and Venezuela); two are significantly negative (those of Brazil and Colombia), one is marginally negative (Argentina), and Peru remains insignificant. Once again, as in the earlier sub sample, there is a slight domination of the positive co-movements.

• For the most recent period—January 2004 through November 2010—the pattern and magnitude of the correlation coefficients are very different, however. As may be seen from Table 5, seven out of the eight coefficients of correlation are significantly negative. Only one (that of Argentina) is significantly positive.³⁶ Moreover, the absolute values of these negative correlations coefficients are quite large. For instance, the coefficient for Brazil is −0.704, that for Colombia is −0.6, and the coefficient for Mexico is −0.515. In the two earlier sub samples the negative coefficients never exceeded 0.3 in absolute terms.

These results provide support to the notion that in the last few years – since the mid-2000s, there has been an important break in the relationship between the U.S.’s RER and those of the Latin American countries considered here. From that date, the RER’s in the vast majority of the larger Latin American countries have been negatively correlated with the RER in the United States. This negative relationship is statistically significant, and for some countries—Brazil, Chile, Colombia, and Mexico—it is quite large. This result means that, since that time, real depreciations (appreciations) of the trade-weighted USD have been associated with real appreciations (depreciations) of the trade-weighted of the Latin American currencies. What is interesting—even surprising—is that this situation did not prevail until 2004. An important matter of future research—and one that is beyond the scope of this paper—is the identity of the channels through which the USD REER may exert an influence on Latin American REERs.

12 Concluding Remarks

In this paper I have presented a series of propositions that are pertinent to exchange-rate policies in the emerging markets. These lessons have been extracted from Asia and Latin America.

• First, exchange-rate policy should be pragmatic. “One size does not fit all”, so that different policies are likely to be appropriate for different countries.

• Second, rigid policies aimed at defending a specific currency value are dangerous.

• Third, there is abundant evidence that (more) flexibility is conducive to faster growth and a greater ability to accommodate exogenous shocks.

• Fourth, if fiscal policy is sustainable and central banks are independent (and focus on achieving their inflation targets), the fear that flexible rates will led to high inflation is misplaced.

• Fifth, even under floating rates it is possible for the real exchange rate to become overvalued.

• Sixth, there is ample evidence suggesting that overvaluation is very costly.

• Seventh, there is an important asymmetry between situations of over and undervaluation.

• Eighth, for most countries “dollarization” is not the most appropriate monetary system. This type of arrangement may work well, however, in countries with a long history of imbalances and stability.

• Ninth, given the above, occasional central bank intervention to avoid over valuation—or an overly appreciated real exchange rate relative to its long run equilibrium—is justified. Intervention, however, should be infrequent, well justified, fully explained to the public, and based on a firm belief that the market exchange rate is (significantly) out of line with respect to its long term equilibrium value.

• Finally, there is evidence suggesting that there has been an important change in the relationship between the real exchange rate of commodity exporting Latin American countries, and the RER in the US. While historically, there was no strong correlation—one way or another—between these variables, since the mid 2000 s there has been a significant and strong negative relationship. This situation suggests that the recent appreciation experienced by the commodity currencies is largely the result of the UDS weakness in global markets.