Itaú BBA - Assessing the undervaluation of the Mexican peso

Macro Vision

< Back

Assessing the undervaluation of the Mexican peso

September 27, 2016

We expect the Mexican peso to end 2016 and 2017 at 17.5 to the dollar

Please open the attached pdf to read the full report .

• Although 2016 has been a good year for emerging-market assets, the Mexican peso weakened substantially, standing as one of the worst-performing currencies.

• Investors have set forth a number of arguments to explain the peso’s underperformance: a combination of high liquidity and low carry (which makes the Mexican currency a tool to hedge against emerging-market and other risks); the deterioration of fiscal accounts and balance of payments associated with the fall of oil prices; and, more recently, the rise of Donald Trump in the U.S. presidential election polls.   

• Mexico’s fundamentals are clearly weaker than they were a few years ago, when oil prices were high and the government was advancing a number of structural reforms. Even so, our models show that the depreciation of the exchange rate has been excessive, meaning that the currency is undervalued – provided there is some turnaround in energy exports. More precisely, we estimated that  the peso should be trading at 17.1 to the dollar in September, in contrast with the 19 average level observed in the first 21 days of the month.

• In our view, a (still likely) defeat of Donald Trump in the U.S. presidential elections would be the key trigger for appreciation. But other factors can contribute to improving sentiment on the currency too: additional interest-rate hikes by Banxico (Mexico’s central bank) in a world of loose monetary policy; higher oil prices; progress in the implementation of the fiscal-consolidation targets set by the government; and a correction of the current-account deficit. We expect the Mexican peso to end both this year and the next at 17.5 pesos to the dollar, which means that the bulk of the undervaluation would vanish.  

The Mexican peso (MXN) has performed poorly against the U.S. dollar in 2016, standing out as one of the most battered currencies in the world. The nominal USDMXN exchange rate recently hit its historically weakest level, while the real bilateral exchange rate (vs. the U.S. dollar) and the real effective exchange rate (multilateral, traded-weighted) have reached levels only comparable to those recorded in the aftermath of the Tequila crisis in 1994.

The poor performance of the currency was in spite of the measures taken by the central bank to support the currency. Since December 2015, the central bank has raised interest rates by 125 bps (including a 50-bp increase in an extraordinary meeting) and the communication of the central bank continues to emphasize the behavior of the currency and the Fed’s decisions as the key drivers for future policy actions. While in February the Foreign Exchange commission (formed by members of the central bank and the Ministry of Finance) suspended dollar sales through rules-based auctions, authorities made clear that discretionary intervention may be used if necessary.

What are the most-often cited causes for the peso weakness?

The Mexican peso is very liquid (it is the second most-widely traded emerging-market currency in the world, after the Chinese Renminbi), which coupled with its low (albeit rising) carry makes it an attractive instrument to hedge against losses in other emerging-market asset investments – and also risks derived from developments in some mature markets, such as those attributed to the US election. Few other floating currencies – if any - strike such a balance of high liquidity and low carry. 

With the fall of oil prices and the strong increase in public expenditures over the past couple of years, fiscal accounts have deteriorated. Considering that oil accounts for a significant share of government revenues (31% in 2014, when oil prices were still high), the fall of the crude price has widened the fiscal deficit and pushed net public debt up (to 45.7% of GDP by mid-2016, from 36.9% in 2013) – gross debt stands at 47.3%. Expansion of public expenditures is also to blame – programmable expenditures, a proxy for primary expenditures, grew 4.1% in real terms, on average, in the past five years. In this context, during 1H16 Moody’s and S&P have revised their outlook on Mexico’s sovereign debt (A3 and BBB+, respectively) from stable to negative, citing concerns over the path and level of public debt. Since the beginning of the year, the government has announced expenditure cuts, but the deficit will remain wide (the public-sector borrowing requirements – the broadest measure of fiscal deficit – would reach 3.0% of GDP this year), in spite of a non-recurrent large dividend received from the central bank (1.2% of GDP) due to valuation gains on international reserves (a consequence of last year’s peso depreciation). Importantly, as many of the budget cuts are carried by Pemex, oil production is falling, which, in turn, adds more uncertainty to the outlook for oil-related revenues. 

Besides fiscal accounts, the drop in oil prices and oil production has widened the current-account deficit. Mexico has turned into a net energy importer: apart from lower oil output, the price of Mexico’s energy exports (largely crude oil) has decreased by more than the price of its energy imports (mostly refined oil products). In 2Q16, the four-quarter rolling energy balance reached a deficit of 1.1% of GDP (contrasting with a 0.7% surplus recorded in 2013). So, in 2Q16 the current-account deficit reached 2.9% of GDP, the highest since the first quarter of 1999. PEMEX’s investment cuts – as dictated by the government’s fiscal consolidation plan – is a risk for oil output recovery and, consequently the energy balance, going forward.

The wider current-account deficit comes at a time when financing is more challenging. In 2Q16, foreign portfolio investment was negative by USD 4.4 billion, as foreigners disinvested USD 7.0 billion from the public-sector domestic debt market. At the same time, direct investment in Mexico is not visibly benefiting from the structural reforms: foreign direct investment is currently not very far from the average level recorded over the past 10 years.   

At the present juncture, another appealing argument that has come to the fore to explain the MXN’s weakness is the rise of Trump in the U.S. elections. To be sure, Trump’s campaign proposals – such as the re-negotiation of the North American Free Trade Agreement (NAFTA), taxes/restrictions on remittances, and tougher anti-immigration policies – would be very harsh for the Mexican economy if implemented.

Is the exchange-rate weakening backed by fundamentals?

Our statistical models for the Mexican peso indicate the currency is undervalued. Even when controlling for the interest-rate differential between Mexico and the U.S., the terms of trade and the sovereign credit default swap (which is supposed to capture the deterioration of fiscal accounts, balance of payments and, to some extent, the rise of Trump), the Mexican peso is too weak. All the model estimations are in the appendix.  

First, we estimate a “long-run” relationship between the real effective exchange rate and its determinants. Specifically, we model Mexico’s multilateral and trade-weighted real exchange rate (REER), as calculated by the Bank of International Settlements (BIS). As a large share of Mexico’s trade is with the U.S., the REER is not substantially different from the bilateral real exchange rate. So, as explanatory variables, we use the relative productivity between Mexico and the U.S. and the short-term interest-rate differential between the two countries, as well as Mexico’s terms of trade and the five-year credit default swap.

The equation shows that the REER was significantly undervalued (by 3.6 standard deviations of the residual) in September 2016 (assuming the USDMXN rate remains constant from September 21 until the end of the month). Put differently, the REER would need to appreciate by 14.2% to reach a level that is consistent with the fundamentals[1].

The misalignment of the Mexican peso vis-à-vis its fundamentals – is also considerable, once we use a Vector Error Correction (VEC) equation to estimate monthly changes in the REER. In the model, the explanatory variables are the first differences (i.e.: one-period changes) of the same variables we used in the first equation, besides a one-month lagged residual of the “long-run” equation[2]. The advantage of this model vs. the “long-run” equation is greater stability of the estimated coefficients (that is, coefficients do not change as much as in the first equation throughout the sample). Using the VEC model, the estimated undervaluation of the REER is smaller, as the REER would need to appreciate by 8.9% to catch up with the fair value (equilibrium REER) in September 2016.

Finally, we model the nominal exchange rate (Mexican pesos per dollar) as a function of two explanatory variables: the predicted REER obtained from the VEC equation (a proxy for the equilibrium real exchange rate) and the consumer price index ratio between Mexico and the U.S. (the inflation differential).[3]

Based on this equation, we estimate that the Mexican peso should have been trading at an average of 17.1 pesos to the dollar in September 2016, which is 11.1% stronger than the average level of 19.2 expected for this month (assuming that the USD/MXN stays constant from September 21 to the end of the month).

We acknowledge that the “Trump effect” could be harsher on the Mexican peso than on the CDS, as policies advocated by Trump during the presidential race would be especially harmful for manufacturing exports and remittances. So, we estimate a daily model for the USDMXN with the sample dating back to July 2015 only, using as explanatory variables the support for Trump in opinion polls; the 10-year U.S. Treasury yield and oil prices. Although the rise of Trump in opinion polls is statistically significant, this model, too, fails to fully explain the recent behavior of the Mexican peso. That is, the model continues to show that the Mexican peso is undervalued by 3.3 standard deviations of the residual, with fair value at 18.4 (7% stronger than the actual level of 19.7 recorded on September 21, 2016). However, this is not conclusive that the underperformance is not due to Trump. In fact, the relationship between Trump’s rise in opinion polls and the MXN could be non-linear, as it is extremely hard to measure at this point what the impact of his anti-Mexico/protectionist policies would be.  

The outlook for the Mexican peso

We expect the Mexican peso to end this year and the next at 17.5 to the dollar, which means that the bulk of the undervaluation would vanish. As it is not clear what drove the Mexican peso’s weakness, it is also unclear what exactly would trigger the appreciation. But there are many potential factors that would be conducive to a stronger peso. A still-likely defeat of Donald Trump in the elections would be the most important one. One additional factor is oil prices: we expect the price of Brent to stabilize slightly above USD 50 per barrel as the global supply of oil adjusts to lower prices. Progress on fiscal consolidation can also help, by removing the risk of a sovereign-rating downgrade: the government has an ambitious fiscal target for next year (a primary surplus of 0.4% of GDP, which would be the first positive primary result since 2008). An improvement in external accounts would improve sentiment on the currency: in fact, the more recent trade-balance figures already suggest a turnaround, as manufacturing exports recover and imports remain sluggish. Finally, additional interest-rate hikes in Mexico will also help. We expect the central bank to raise rates by 50 bps this week, in response to the peso’s overshooting, and further hikes are likely, in lockstep with the Fed.


Alexander Müller

João Pedro Resende

[1] In our model, we assume that productivity in Mexico and the U.S. stay at the same level they were in June (latest data point available), while we approximate the terms of trade for August and September based on the price of oil. We assume that the CDS and interest differential remain constant from September 21 to the end of the month.

[2] In order to avoid the bias of ex-post residuals, we tweak the long-run equation by calculating a rolling regression (window size of 60 months) which only uses past information that was available at each point in time. The above mentioned lagged residual is obtained from this modified equation.

[3] This modeling technique – specifically, using the predicted equilibrium exchange rate as an explanatory variable in the nominal exchange rate equation – is similar to the framework presented in Arend M. 2010. Forecasting Performance of Fundamental Models of Exchange Rates: A New Set-up. Department of Economics, Boston College.


Please open the attached pdf to read the full report .


< Back