V. Explanations for the Recent Behavior of the Mexican Peso1
This chapter looks at the behavior of the peso/U.S. dollar exchange rate since 1999, examining to what extent Mexico’s trade links with the U.S. economy, regional shocks in Latin America, and changes in global risk perception toward emerging market economies, explain the evolution of the Mexican peso. In this context, it pays particular attention to the close correlation between the peso/dollar and dollar/euro rates since 1999, compared with the behavior of selected other currencies. It also looks at the recent breakdown in the close co-movement of the peso/dollar and dollar/euro exchange rates, focusing on the role of changes in reserve management policy and international competitiveness.
118. Following the abandonment of the exchange rate target band in late 1994 and subsequent turbulence in local financial markets, Mexico adopted a flexible exchange rate system. Since then, the exchange rate has been allowed to move freely, apart from temporary periods when the Bank of Mexico has participated in the foreign exchange market through well-defined rules. This chapter looks at the behavior of the peso/U.S. dollar exchange rate (MEX-USD) in recent years to shed light on the factors that may have driven exchange rate movements in this floating-rate environment.
119. A phenomenon that we pay particular attention to is the close correlation analysts have noted between movements in the peso/dollar exchange rate and the U.S. dollar/euro (USD-EUR) exchange rate, especially in the past 2–3 years. Specifically, the peso has tended to depreciate against the dollar when the dollar has depreciated against the euro, reflecting in some sense an “over-shooting” by the peso of dollar movements against the euro. This chapter examines various factors that could account for this apparent regularity, including: Mexico’s strong trade links with the U.S. economy; shocks in Latin American financial markets that may have affected both the peso/dollar and dollar/euro exchange rates; and changes in global market appetite for risk that may have affected emerging market exchange rates and the dollar/euro rate. The paper then turns to the more recent apparent “disconnect” between movements in the peso/dollar and dollar/euro rates, assessing the importance of the change in reserves management strategy announced in the spring of 2003, and also concerns about Mexico’s international competitiveness.
B. Correlation With the Dollar/Euro Rate
120. Figure 1 shows the movements in the peso/dollar rate from January 1999 to end-August 2003.2 The peso gradually appreciated for much of the period from 1999 until early 2002, even though economic activity in Mexico slowed sharply in 2001 and inflation remained higher than in trading partners. There was a marked shift in direction starting in April 2002, however, with the peso more than fully reversing its earlier appreciation over the following 12 months. Market sentiment shifted again in late March 2003, partly coinciding with the reduction of risk aversion in international markets due to apparent resolution of the U.S.-Iraq conflict and partly coinciding with the announcement by the Bank of Mexico (BOM) that it would automatically sell half of its reserve accumulation over the previous quarter to the market, starting in May 2003 (for details see Box 1, EBS/03/326). Finally, the summer of 2003 witnessed renewed weakness in the peso that took it close to previous lows.
Figure 1.Mexico: Mexican Peso Versus the U.S. Dollar: 1999-2003
Source: Mexican Authorities.
121. The top panel of Figure 2 shows these movements in the peso/dollar rate in relation to those in the dollar/euro rate over the same period. It is apparent that the two have behaved similarly over much of the period, both in terms of low-frequency trends and higher-frequency movements around these trends. The obvious exception is the period since March 2003, when first the strength and then the weakness of the peso/dollar rate contrasted with the opposite movements in the dollar/euro rate. The tendency for the two exchange rates to move in tandem in recent years is supported by the correlation coefficients shown in Table 1 and the rolling correlations shown in the second panel of Figure 2. The correlation between movements in the peso/dollar and dollar/euro rates over the period as a whole is 0.68, with the relationship being stronger in the post-2000 period (with a correlation coefficient of 0.91) than during 1999–2000 (with a correlation of 0.40).3
|January 1999 - End-August 2003|
|January 1999 - End-December 2000|
|January 2001 - End-August 2003|
|CZK-USD||1.00|Figure 2.Mexico: Movements in the Mexican Peso and the Canadian Dollar vs. the Euro
Source: Bloomberg; and Fund staff estimates.
122. The co-movements in the peso/dollar and dollar/euro rates over this period could reflect various factors. Distinguishing among these factors and determining their relative importance is of interest, in part because the underlying sources of the correlation are likely to determine how durable it will be in the future. The following are three explanations that could account for this phenomenon:
- Shocks to U.S. activity. When U.S. activity slows, the dollar tends to depreciate against the euro, and at the same time Mexico experiences a negative shock to exports. With the U.S. being Mexico’s main trading partner, the peso tends to depreciate against the dollar in response to this deterioration in external demand.
- Volatility in Latin American financial markets. Financial turbulence in the region, particularly in 2001–02, affected both the bilateral exchange rates of regional currencies (including the peso) against the U.S. dollar, and possibly also the value of the dollar versus the euro. In this explanation, the causation is the reverse of that described above, being from Latin American shocks to the dollar/euro exchange rate.
- Shocks in global capital markets. Recent years have witnessed significant swings in global market appetite for risk, affected yield spreads and demand for assets in both developed and emerging capital markets. To the extent that these shocks have induced co-movements in the dollar/euro rate and the exchange rates of emerging market currencies against the dollar, they could explain the correlation between the peso/dollar and dollar/euro rates.
123. To shed light on the possible role of these factors in explaining movements in the peso/dollar rate, we examine the correlations between other selected currencies and the dollar versus the dollar/euro rate. The following currencies are included: the Canadian dollar, Brazilian real, Chilean peso, Thai baht, South African rand, Polish zloty and the Czech koruna. Canada is chosen because, like Mexico, it is a NAFTA member with a high share of exports destined to the U.S. market. If peso/dollar movements are largely driven by U.S. shocks through trade channels, one might expect to see a similar pattern for the Canadian dollar as for the peso. In contrast, if financial shocks in Latin America have driven movements in the peso/dollar rate, one might expect to see the Brazilian real and Chilean peso behaving similar to the peso. Finally, if the peso/dollar rate has been driven by changes in global market sentiment, this would likely be reflected in similar movements in the currencies of other emerging market economies with floating exchange rates, including Thailand, South Africa, Poland, and the Czech Republic.
124. Figures 3a to 3c provide a visual impression of the extent to which these other currencies have moved in line with the dollar/euro rate over the 1999–2003 period. While there are brief periods when the currencies exhibit similar movements, they tend to be short-lived. This general pattern is confirmed by the predominantly negative correlation coefficients in the first row of the correlation matrices in Table 1, although there are some notable exceptions, as discussed below. The rest of this section assesses the various explanations for the behavior of the peso/dollar rate in light of these stylized facts.
Figure 3a.Bilateral Exchange Rates and 90-day Rolling Correlation with the USDEUR, Brazilian and Chilean Currencies
Source: Bloomberg; and Fund staff estimates.
Figure 3b.Bilateral Exchange Rates and 90-day Rolling Correlation with the USD-EUR, Thai and South African Currencies
Source: Bloomberg; and Fund staff estimates.
Figure 3c.Bilateral Exchange Rates and 90-day Rolling Correlation with the USDEUR, Polish and Czech Currencies
Source: Bloomberg; and Fund staff estimates.
Mexico’s trade links with the United States
125. Figure 4 shows various indicators of Mexico’s trade dependence on the U.S. economy. The share of Mexican exports in GDP has more than doubled from about 12 percent in 1992 to 25 percent in 2002, with almost 90 percent of non-oil exports going to the U.S. market, and most oil exports marketed to U.S. oil companies. At the same time, Mexico’s share in U.S. imports has doubled to nearly 12 percent during this period. The lower panel of Figure 3 indicates the close relationship between movements in Mexican exports and U.S. industrial production since the late 1990s.4 In terms of broader measures of activity, a regression of four-quarter growth in Mexican real GDP on growth in U.S. real GDP (lagged one quarter) from 1999Q1 to 2003Q2 yields a slope coefficient, or “beta,” of 1.52, with an R2 of 0.76. This is consistent with the view that Mexican activity is “leveraged” by a factor of more than one to the U.S. economy, possibly explaining an “over-shooting” of the exchange rate in response to shocks to actual or expected U.S. activity.
Figure 4.Mexico: Export Dependence on the United States
Sources: Authorities; and Fund staff estimates.
126. In the case of Canada, about 85 percent of exports are destined to the U.S. market, similar to the Mexican share. Canadian and U.S. GDP growth rates are also closely related, with a beta from the regression described above of 0.97, with an R2 of 0.84. But, as shown by the correlations in Table 1, the behavior of the Canadian dollar has been opposite to that observed for the peso — that is, when the U.S. dollar has weakened against the euro, the Canadian dollar has strengthened against the U.S. dollar, with the negative correlation being of roughly the same magnitude as the positive correlation for the peso/dollar rate over the period as a whole. The correlation coefficient for the Canadian dollar is also negative over the two sub-periods. The contrast in behavior could reflect the somewhat lower dependence of Canadian growth on U.S. growth, with a beta of slightly less than one as opposed to well over one for Mexico. This, in turn, could result from Canada’s greater scope for using countercyclical monetary and fiscal policies to offset negative shocks from the United States, given the well-established credibility of the fiscal and monetary policy framework.
127. To further assess the importance of trade links in explaining currency movements, we consider the correlation of the two European currencies in the sample —the Polish zloty and the Czech kurona—with the dollar/euro rate. Both countries have a relatively high trade dependence on the euro area, with about 70 percent of their exports being destined to the European Union. Nevertheless, the betas with respect to euro-area GDP, at 0.41 for Poland and 0.50 for the Czech Republic, are considerably lower than those discussed above for Mexican and Canadian activity with respect to U.S. GDP. In any case, if these currencies exhibited the same “overshooting” as the Mexican peso, we would expect them to appreciate against the euro when the euro appreciates against the dollar, and vice versa. In fact, the correlation coefficient between the zloty/euro rate and the dollar/euro rate is consistently positive at 0.84 for the full period, and 0.72 and 0.95 for the first and second sub-periods respectively. These positive correlations indicate a response opposite to the “overshooting” behavior observed for the peso—i.e., the zloty tends to depreciate against the euro at the same time as the dollar depreciates. A similar pattern holds for the Czech koruna: the correlation between the koruna/euro rate and the dollar/euro rate is 0.07 for the period as a whole, although it switches sign from 0.69 for the first sub-period to -0.41 in the second period.
128. The absence of supporting evidence for these other economies does not necessarily undermine the hypothesis that the behavior of the peso/dollar rate is driven, at least in part, by trade linkages. None of the other three economies considered above has a “beta” with respect to U.S. activity that exceeds unity, as does Mexico, which could explain why the peso is the only currency to overshoot. Furthermore, there are differences in policy regimes and various other factors that could lead to alternative exchange rate behavior. On balance, we conclude that trade linkages with the U.S. could explain part of the behavior of the peso/dollar rate, although there is not strong corroborating evidence from the behavior of other currencies.
Shocks in Latin American financial markets
129. We now examine whether financial market shocks in Latin America, particularly during the turbulent period from 2001 to early 2003, can explain the correlation between the peso/dollar and euro/dollar exchange rates. These shocks put pressure on yield spreads and currencies throughout the region, because of both trade and financial market interdependencies.5 They may also have driven movements in the dollar/euro rate, either because of a view that the U.S. economy would be more affected than Europe by problems in Latin America, or because of portfolio effects on exchange rates given that most of the crisis countries had large external liabilities denominated in dollars.
130. To explore this story, we examine the correlations between two other regional currencies with floating exchange rates—the Brazilian real (BRL-USD) and Chilean peso (CLP-USD)—with the dollar/euro rate. As shown in Table 1, the correlations between these currencies against the dollar and the dollar/euro rate were negative (highly so for the CLP) during the 1999–2000 sub-period, opposite to the positive correlation observed for the MEX-USD rate. But the correlations for the real and the Chilean peso become large and positive in the period beginning in 2001, around the time of the Argentinean crisis. It is notable that this also corresponds to a rise in the correlation between the peso/dollar and euro/dollar rates to 0.91 from 0.40 during the first sub-period. The other interesting point to note is that the correlations between these currencies and the dollar all became negative after March 2003, the time at which financial market pressures in the region had largely subsided.6
131. It appears, then, that spillovers within the region may have a significant effect on exchange rates during periods of market turbulence, even though Mexico and Chile both enjoy investment-grade credit ratings. To the extent that the recent (relative) stability in regional markets continues, it may be the case that the correlation between peso/dollar and dollar/euro exchange rates will be less strong than during the period from 2001 to early 2003 of high market volatility.
Global capital market shocks
132. If changes in risk perception and/or appetite by global investors underlie the comovements in the peso/dollar and euro/dollar rates, one might expect to see similar correlations in the currencies of other emerging market countries with floating exchange rates, such as Thailand and South Africa. These countries are also of interest because their trade flows are relatively diversified, meaning that any correlations with the dollar/euro rate are more likely to arise from financial market shocks than the strength of direct trade linkages. In the event, however, Table 1 indicates that the correlations between the Thai baht and the dollar and the South African rand and the dollar with the dollar/euro rate are consistently negative, both for the sample as a whole and for the two sub-periods.
133. The implication that shocks in global capital markets are not responsible for the correlation between the peso/dollar and euro/dollar rates is further supported by the negative correlation (-0.21) found between the EMBI spread for emerging markets as a whole and the peso/dollar rate. Thus, as the EMBI spread for emerging markets rises, the peso tends to appreciate against the dollar. It appears, then, that portfolio diversification works in Mexico’s favor compared with emerging markets as a whole, probably reflecting its investment-grade credit rating.
C. Behavior of the Peso in Recent Months
134. While the U.S. dollar continued to depreciate against the euro in the spring of 2003, it reversed course during June to August, perhaps reflecting the end of the Iraq war and improving signs of U.S. recovery, while the euro-area economy has remained sluggish. Unlike the previous experience, however, when peso appreciation accompanied dollar appreciation, the peso moved in opposite directions over this period, appreciating against the dollar between March and mid-May, and subsequently reversing course. As a result, the previous positive relationship between the peso/dollar and dollar/euro rates has broken down, with the recent correlation standing at -0.55. To understand these more recent movements in the peso, we look at the new reserve management policy of the BOM and changing views about Mexico’s external competitiveness.
Change in reserve management strategy
135. The BOM started selling foreign exchange in the interbank market starting May 2, in line with the new policy announced on March 20. From May to July, the BOM sold an average of US$32 million per day, or a total of US$2.1 billion, equal to half of the reserves accumulated in the first quarter. During the current quarter, the BOM will sell an additional US$910 million. The peso appreciated by about 10 percent between early March and mid-May, influenced by reduction of risk aversion in international markets and by the new policy announcement on March 20. From mid-May to early September, the peso depreciated sharply.
136. Figure 5 shows the evolution of the nominal and real effective exchange rates (NEER and REER) since 1995 for the Mexican peso, the Chinese yuan and selected other currencies. While the NEER of the Mexican peso has depreciated quite significantly, in real terms it shows a significant appreciation for much of the period, with some reversal since early 2002. In contrast, for the other currencies, real effective appreciation has been rather modest. Since early 2002, China’s real effective exchange rate has also depreciated, moderating the gain in Mexico’s competitiveness from the depreciation in its real effective exchange rate over this period.
Figure 5.Effective Exchange Rates, Mexico and Selected Countries
Source: Mexican authorities; and IFS.
137. Against the background of these longer-term movements in real exchange rates, the more recent failure of exports to respond to positive U.S. economic indicators has raised concerns that Mexico may be losing competitiveness, especially to China. The continued sluggishness of Mexican exports to the U.S. and rapid growth in China’s exports to the U.S. has increased China’s market share in the U.S. For the first seven months of this year, China’s market share was marginally higher at 11.2 percent compared with Mexico’s at 11.0 percent. While U.S. manufacturing growth is critical to Mexican exports and analysts believe that the projected recovery in U.S. manufacturing will boost Mexico’s exports, increasing emphasis has been placed on structural reforms to help ensure Mexico’s competitiveness. In the last few years, China has undertaken major structural reforms, and further ongoing reforms of the state-owned enterprise sector combined with effective currency depreciation is likely to boost China’s competitiveness further. On the other hand, rigidities in the labor market in Mexico, the lack of investment in infrastructure, low fiscal revenues to invest on education and health, and stalled electricity and telecommunications sector reforms have inhibited productivity growth. The increasing focus by market analysts on these longer-term competitiveness issues in recent months may well underlie the weakening in the peso.
D. Concluding Remarks
138. Market analysts have noted a high correlation between the peso/dollar and dollar/euro exchange rates in recent years. Based on the observed movements in other currencies against the dollar over this period, the evidence supports the view that financial shocks in Latin America can explain much of the co-movement in the two currencies. To the extent that these shocks are not as pronounced in the period ahead, the correlation between the two exchange rates may not be as high as in the recent past. Mexico’s continuing strong trade and economic dependence on the U.S. economy, though, is still likely to have an ongoing effect on peso/dollar developments, at least until policy credibility is fully established.
139. Independent of these external shocks, developments in recent months underscore the role that domestic factors can play in driving peso/dollar movements. First, the change in the mechanism used to determine international reserves accumulation announced in March seemed to have some transitory effect on the exchange rate. Subsequently, concerns about Mexico’s longer-term competitiveness in the face of increasing international competition and the slow pace of structural reforms have likely contributed to a reversal in the strength of the peso. These developments serve to illustrate that empirical regularities in the historical data reflect the nature of the underlying shocks; as these shocks change, so will the pattern of movements in exchange rates.
BrooksR.H.EdisonM.Kumar and T.Sl⊘k2001“Exchange Rates and Capital Flows,”IMF Working Paper No. 01/190.
DabósM. and V.H.Juan-Ramón2000“Real Exchange Rate Response to Capital Flows in Mexico: An Empirical Analysis,”IMF Working Paper No. 00/108.
JinZ.2003“The Dynamics of Real Interest Rates, Real Exchange Rates and the Balance of Payments in China: 1980–2002IMF Working Paper No. 03/67.
MeredithG.2001“Why has the Euro been so Weak?,”IMF Working Paper No. 01/155.
Prepared by N. Thacker.
Daily exchange rate data from Bloomberg are used for the analysis.
Although, as noted above, the correlation has broken down in recent months, with a correlation coefficient from May to end-August 2003 of -0.55.
See Economic Integration in the Americas: Lessons from NAFTA in United States—Selected Issues (EBS/03/253).
The latter would be more important in the case of Mexico, as its trade shares with the rest of the Latin American region are relatively low. Financial market interdependencies could arise from investor aversion to taking risks in the region, either because of herding effects or the negative impact on portfolios of institutions with investments across the region of losses in specific countries.
The correlation between the BRL-USD and USD-EUR rate from April–August 2003 is -0.64, and for the CLP-USD and USD-EUR rate is -0.24.