This chapter describes Mexico’s experience with foreign exchange intervention since the free-floating exchange rate regime was adopted in December 1994. The first two sections describe the background of the different intervention mechanisms that have been used over time, along with the stated objectives of those interventions. The third section presents a set of analyses of the interventions’ effectiveness. The main takeaways from that section are that interventions with the objective to achieve a specific stock of international reserves have been very effective. However, it is difficult to assess the effectiveness of interventions if the objective is to temper foreign exchange volatility, because the central bank does not have the counterfactual case to compare what would have been the outcome without the intervention. Despite those difficulties, there is some statistical evidence to infer that the Bank of Mexico’s interventions in the foreign exchange market to temper volatility have been effective. The next sections describe how the size and timing of interventions were decided, as well as the communication policies that were followed. The last section provides final remarks on what has been learned in Mexico about foreign exchange interventions.
Introduction
For more than 20 years, the Mexican peso has been floating freely, based on market forces. For this reason, since 1995, foreign exchange interventions have not been used to set the exchange rate. Instead, interventions have aimed at other objectives, such as managing the stock of international reserves or reducing exchange rate volatility during unusual financial market situations. Over time, different mechanisms have been used, depending on these objectives and on market conditions. Before every intervention, authorities perform a rigorous analysis to decide how to intervene and how to communicate this decision to the market.
This chapter details this experience in Mexico. It describes the background and intervention mechanisms used and the stated objectives of those interventions. Also, the chapter reviews the interventions’ effectiveness and describes how the size and timing of interventions were decided, along with the communication policy. The final remarks of this chapter summarize the lessons learned.
From Crawling Peg to Free Floating
In December 1994, after the depletion of international reserves, the Foreign Exchange Commission (FX Commission)1—integrated by officials of both the Bank of Mexico and the Ministry of Finance—decided to move from a crawling peg regime to a free-floating regime. Since then, interventions in the foreign exchange market have responded to various scenarios by selling and buying international reserves. Most of these interventions were implemented through pre-announced auction-based mechanisms. The use of extraordinary auctions and discretionary interventions was restricted to extreme circumstances. Interventions have been designed to manage the stock of international reserves or to temper extreme exchange rate volatility, but not to set a level for the exchange rate. In this regard, foreign exchange interventions have been executed to be compatible with the free-floating exchange rate regime. (Figure 11.1 shows the trend of international reserves since 1994 and indicates some of the main foreign exchange intervention periods.)
Bank of Mexico’s International Reserves and the Mexican Peso Exchange Rate, 1993–2017
(Billions of US dollars, left scale; Mexican peso per US dollar, right scale)
Source: Bank of Mexico.Note: The Bank of Mexico’s law defines international reserves as foreign assets owned by the Bank of Mexico, excluding the government’s and other’s short-term deposits.Bank of Mexico’s International Reserves and the Mexican Peso Exchange Rate, 1993–2017
(Billions of US dollars, left scale; Mexican peso per US dollar, right scale)
Source: Bank of Mexico.Note: The Bank of Mexico’s law defines international reserves as foreign assets owned by the Bank of Mexico, excluding the government’s and other’s short-term deposits.Bank of Mexico’s International Reserves and the Mexican Peso Exchange Rate, 1993–2017
(Billions of US dollars, left scale; Mexican peso per US dollar, right scale)
Source: Bank of Mexico.Note: The Bank of Mexico’s law defines international reserves as foreign assets owned by the Bank of Mexico, excluding the government’s and other’s short-term deposits.After the liberalization of the exchange rate in 1994, the only local institution that was subject to restrictions on operating in the foreign exchange market was Pemex, the state-owned oil company. As mandated by the Bank of Mexico’s law, all nonfinancial entities in the Federal Public Administration must operate in the foreign exchange market according to the rules and policies set by the central bank, so that large trade surpluses or deficits do not distort the exchange rate through market dominance. Letting Pemex freely operate in the foreign exchange market was not viable in 1994, because it had a large trade surplus, which could distort the exchange rate and give Pemex a dominant position over foreign exchange policy. Because of the regulatory framework, Pemex was, for many years, the main source of international reserves accumulation that are administered by the central bank (Figure 11.2).
Bank of Mexico’s International Reserves: Sources of Accumulation, 1996–2017
(Billions of US dollars)
Source: Bank of Mexico.Note: The sixth edition of the Balance of Payments Manual defines official reserve assets as those external assets that are readily available to and controlled by monetary authorities for meeting balance of payments financing needs, for intervention in exchange markets to affect the currency exchange rate, and for other related purposes (such as maintaining confidence in the currency and in the economy, and serving as a basis for foreign borrowing).Bank of Mexico’s International Reserves: Sources of Accumulation, 1996–2017
(Billions of US dollars)
Source: Bank of Mexico.Note: The sixth edition of the Balance of Payments Manual defines official reserve assets as those external assets that are readily available to and controlled by monetary authorities for meeting balance of payments financing needs, for intervention in exchange markets to affect the currency exchange rate, and for other related purposes (such as maintaining confidence in the currency and in the economy, and serving as a basis for foreign borrowing).Bank of Mexico’s International Reserves: Sources of Accumulation, 1996–2017
(Billions of US dollars)
Source: Bank of Mexico.Note: The sixth edition of the Balance of Payments Manual defines official reserve assets as those external assets that are readily available to and controlled by monetary authorities for meeting balance of payments financing needs, for intervention in exchange markets to affect the currency exchange rate, and for other related purposes (such as maintaining confidence in the currency and in the economy, and serving as a basis for foreign borrowing).The Federal Government is the other institution that has executed foreign exchange operations—almost exclusively with the central bank. It buys US dollars from the central bank to service foreign debt. When the Federal Government issues debt denominated in foreign currency, depending on its currency needs, it can sell part or all of those resources to the central bank. Figure 11.2 presents the main sources of accumulation that explain the movements in international reserves.
Mechanisms and Objectives
Since 1996, all the mechanisms to intervene in the foreign exchange market were determined by the FX Commission and executed by the Bank of Mexico. In general terms, some mechanisms were used to aim for a certain level of international reserves—either to accumulate reserves or to reduce the pace of accumulation. In other cases, intervention mechanisms aimed to provide liquidity to the market, given that its main goal is to ensure proper operating conditions and to reduce volatility. In all cases, the level of the exchange rate has not been targeted.
The mechanisms used to accumulate or reduce the pace of accumulation of international reserves considered the level of reserves at the time of implementation as well as other criteria. There is no single uniquely accepted metric to assess the sufficiency of international reserves. The central bank uses several metrics to assess reserve adequacy, but because results can vary, it relies on the FX Commission’s judgment. The metrics used to assess sufficiency include (1) the ratio of reserves to GDP, compared with the ratio in other emerging markets; (2) the reserve adequacy metric proposed by Moghadam, Ostry, and Sheehy (2011) from the IMF; (3) cost-benefit approaches, such as the one proposed by Calvo, Izquierdo, and Loo-Kung (2012); and (4) utility-maximizing approaches, such as the one that Rancière and Jeanne (2006) developed.2 The following sections outline the foreign exchange intervention mechanisms used.
Accumulation of International Reserves through Put Options
The accumulation of international reserves through a put options mechanism was first used from August 1996 through June 2001. By 1996, the natural pace of reserve accumulation through Pemex’s sales of US dollars to the central bank was not enough to bring international reserves to what was deemed a sufficient level. If intervention was needed to contain extreme volatility, reserves would not suffice. In addition, after the 1994 financial crisis, international investor and credit rating agency confidence needed to be restored.
The FX Commission therefore decided to implement a reserve accumulation mechanism, under which the central bank sold US dollar put options to the market through monthly auctions. These options gave financial institutions the right to sell US dollars to the central bank during the following month, subject to certain conditions. The first condition was that the exchange rate of exercise (strike price) was the FIX exchange rate determined by the central bank the business day before the exercise3. As a second condition, the option could be executed only when the exercise exchange rate was below its 20-day moving average (Figure 11.3). This way, international reserve accumulation would occur during Mexican peso appreciation periods; that way, the mechanism did not put pressure on the exchange rate in a depreciation trend. The mechanism allowed the central bank to purchase US dollars with the least possible interference with the free-floating exchange rate regime. Total US dollars bought by the central bank during that period was $12.2 billion (Table 11.1, panel 1).
Mexican Peso to US Dollar Exchange Rate: Options Exercises, 1996–2001 and 2010–11
(Mexican peso per US dollar)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate: Options Exercises, 1996–2001 and 2010–11
(Mexican peso per US dollar)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate: Options Exercises, 1996–2001 and 2010–11
(Mexican peso per US dollar)
Source: Bank of Mexico.Options Auctions and Exercises, 1996–2001 and 2010–11 (Billions of US dollars)
Options Auctions and Exercises, 1996–2001 and 2010–11 (Billions of US dollars)
| 1. 1996–2001 | 2. 2010–11 | ||||
|---|---|---|---|---|---|
| Year | Purchased | Exercises | Year | Purchased | Exercises |
| 1996 | 0.9 | 0.9 | 2010 | 6.6 | 4.5 |
| 1997 | 5.2 | 4.5 | 2011 | 6.0 | 4.6 |
| 1998 | 2.8 | 1.6 | Total | 12.6 | 9.1 |
| 1999 | 3.0 | 2.0 | |||
| 2000 | 3.0 | 1.8 | |||
| 2001 | 1.5 | 1.4 | |||
| Total | 16.4 | 12.2 | |||
Options Auctions and Exercises, 1996–2001 and 2010–11 (Billions of US dollars)
| 1. 1996–2001 | 2. 2010–11 | ||||
|---|---|---|---|---|---|
| Year | Purchased | Exercises | Year | Purchased | Exercises |
| 1996 | 0.9 | 0.9 | 2010 | 6.6 | 4.5 |
| 1997 | 5.2 | 4.5 | 2011 | 6.0 | 4.6 |
| 1998 | 2.8 | 1.6 | Total | 12.6 | 9.1 |
| 1999 | 3.0 | 2.0 | |||
| 2000 | 3.0 | 1.8 | |||
| 2001 | 1.5 | 1.4 | |||
| Total | 16.4 | 12.2 | |||
This mechanism was used again, with the same conditions, from February 2010 through November 2011, to restore international reserves after the central bank sold more than $30 billion during the 2008–09 global financial crisis. During that period, the central bank accumulated $9.1 billion through the mechanism. Table 11.1 (panel 2) shows the results of these auctions during 2010 and 2011.
Foreign Exchange Sales to Slow the Accumulation of International Reserves
The foreign exchange sales to slow the accumulation of international reserves mechanism was in place from May 2003 through July 2008. After the suspension of the reserve accumulation program through put options, with the sole accumulation derived from Pemex’s US dollar sales, international reserves reached a level where the benefits of future expected accumulation were beginning to be outnumbered by their financial costs (cost of carry). Thus, the FX Commission introduced this mechanism by selling, in a given quarter, half the amount accumulated during the previous quarter. At the beginning of each selling period, the central bank announced the daily amount to be auctioned. To smooth the amount sold each quarter, the previous quarter’s accumulation reference was changed by a moving average of the previous four quarters’ accumulation, avoiding the volatility inherent in the seasonality of reserve accumulation. Because the amount of central bank US dollar sales was predetermined, and resulted from previous accumulation, the interference with the free-floating regime was minimized. During the years in which this mechanism was implemented, $30.1 billion were sold through auctions (Table 11.2).
International Reserves Sold, 2003–08 (Billions of US dollars)
International Reserves Sold, 2003–08 (Billions of US dollars)
| Year | Amount |
|---|---|
| 2003 | 3.2 |
| 2004 | 6.7 |
| 2005 | 4.4 |
| 2006 | 8.1 |
| 2007 | 4.2 |
| 2008 | 3.5 |
| Total | 30.1 |
International Reserves Sold, 2003–08 (Billions of US dollars)
| Year | Amount |
|---|---|
| 2003 | 3.2 |
| 2004 | 6.7 |
| 2005 | 4.4 |
| 2006 | 8.1 |
| 2007 | 4.2 |
| 2008 | 3.5 |
| Total | 30.1 |
US Dollar Daily Auctions with Minimum Bid Price
The US dollar daily auctions with minimum bid price mechanism was used during several periods to limit extreme exchange rate volatility and provide liquidity to ensure proper operating conditions in the foreign exchange market, but never to fix a prespecified level for the exchange rate. Under this mechanism, the central bank auctions a prespecified amount of US dollars every day and sets a minimum accepted price. This way, the trigger to sell US dollars to the market is predetermined by market conditions without meddling with the free-floating exchange rate regime. This tool was first used from February 1997 through June 2001; during that time, the minimum price accepted was set equal to the FIX exchange rate, which was determined the previous day plus a 2 percent depreciation factor. Therefore, only when the exchange rate depreciated by 2 percent or more, would the auctions be allotted (Figure 11.4, panel 1). A total of $2 billion were sold with this mechanism during this period (Table 11.3).
Mexican Peso to US Dollar Exchange Rate: Minimum Price Auctions, 1997–2001 and 2008–16
(Mexican peso per US dollar)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate: Minimum Price Auctions, 1997–2001 and 2008–16
(Mexican peso per US dollar)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate: Minimum Price Auctions, 1997–2001 and 2008–16
(Mexican peso per US dollar)
Source: Bank of Mexico.International Reserves Sold with Minimum Price Auctions, 1997–2016
(Billions of US dollars)
International Reserves Sold with Minimum Price Auctions, 1997–2016
(Billions of US dollars)
| Year | Sold |
|---|---|
| 1997 | 0.6 |
| 1998 | 0.9 |
| 1999 | 0.4 |
| 2000 | 0.1 |
| 2001 | 0 |
| 2008 | 4.2 |
| 2009 | 4.2 |
| 2011 | 0 |
| 2012 | 0.6 |
| 2013 | 0 |
| 2014 | 0.2 |
| 2015 | 3.8 |
| 2016 | 3.6 |
| Total | 18.6 |
International Reserves Sold with Minimum Price Auctions, 1997–2016
(Billions of US dollars)
| Year | Sold |
|---|---|
| 1997 | 0.6 |
| 1998 | 0.9 |
| 1999 | 0.4 |
| 2000 | 0.1 |
| 2001 | 0 |
| 2008 | 4.2 |
| 2009 | 4.2 |
| 2011 | 0 |
| 2012 | 0.6 |
| 2013 | 0 |
| 2014 | 0.2 |
| 2015 | 3.8 |
| 2016 | 3.6 |
| Total | 18.6 |
The mechanism was used again, with the same characteristics, during the global financial crisis (from October 2008 through April 2010; see Figure 11.4, panel 2) and during the European financial crisis (from November 2011 through April 2013; see Figure 11.4, panel 3). It was used also from December 2014 through February 2016 to provide the liquidity to ensure the proper operating conditions in the foreign exchange market and to temper the volatility in the exchange rate generated by the drop in oil prices. During this last period, the mechanism was modified several times to increase the amount of US dollars to be offered. Initially, in December 2014, the minimum accepted price was set at the FIX exchange rate determined the previous day plus a 1.5 percent depreciation factor (instead of the previous 2.0 percent depreciation factor). Then, in July 2015, the minimum price threshold was adjusted by using a 1.0 percent depreciation factor. And in that same month, supplementary daily auctions with a minimum price equivalent to a 1.5 percent depreciation of the previous day’s FIX were added to the toolkit (Figure 11.4, panel 4).
US Dollar Daily Auctions without a Minimum Bid Price
Sometimes, a constant demand for US dollars that could distort operating conditions and generate volatility is expected to persist. Under those scenarios, the FX Commission has used daily auctions of US dollars for determined amounts and periods, without a minimum price. In 2009, the FX Commission decided to provide additional liquidity to improve the operating conditions in the foreign exchange market; hence, it introduced daily auctions of US dollars without a minimum price. This mechanism was in operation from March through September 2009, and a total of $10.3 billion was sold, with initial daily auctions of $100 million that were later reduced to $50 million.
This mechanism was used again from March through November 2015, with the total amount offered corresponding to the expected accumulation of international reserves during the following quarter. For the first three-month period, the daily amount offered was set at $52 million, but it later increased to $200 million, as market conditions demanded (Figure 11.5, panel 2). A total of $20.7 billion was sold through this intervention tool.
Mexican Peso to US Dollar Exchange Rate and Nonminimum Price US Dollar Daily Auctions, 2009 and 2015
(Mexican peso per US dollar, left scale; millions of US dollars, right scale)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate and Nonminimum Price US Dollar Daily Auctions, 2009 and 2015
(Mexican peso per US dollar, left scale; millions of US dollars, right scale)
Source: Bank of Mexico.Mexican Peso to US Dollar Exchange Rate and Nonminimum Price US Dollar Daily Auctions, 2009 and 2015
(Mexican peso per US dollar, left scale; millions of US dollars, right scale)
Source: Bank of Mexico.Extraordinary US Dollar Auctions
Even though the FX Commission has traditionally shown a preference to intervene in the market through rules-based and preannounced operations, it has also employed extraordinary auctions offering large amounts of US dollars. In October 2008, during the worst part of the global financial crisis, the Mexican peso depreciated more than 20 percent in one week alone, and the movement was not considered to be aligned with Mexico’s macroeconomic fundamentals. Moreover, operating conditions were dramatically distorted. To provide liquidity and restore the well-functioning of the foreign exchange market, therefore, the FX Commission, in five extraordinary auctions, sold $11 billion. One of those interventions amounted to more than $6 billion.
US Dollar Discretionary Outright Sales
Since the beginning of the free-floating exchange rate regime, the FX Commission has resorted to discretionary interventions, through outright sales of US dollars, only in four episodes of extreme volatility. As with the preannounced mechanisms, this type of intervention is meant to reestablish orderly market operating conditions and does not target a specific level for the exchange rate. When a discretionary intervention takes place, the FX Commission releases a press bulletin to inform the public immediately after the intervention. Even though the amount of the intervention is not usually disclosed in these press releases, the central bank publishes the stock and flows of the international reserves on a weekly basis, and the aggregated amount of intervention is disclosed. The first time that discretionary sales took place during the free-floating period was in September 1998, during the Russian and Brazilian crises. Almost a decade later, during February 2009, in the middle of the global financial crisis, conditions in the foreign exchange market justified direct intervention with local market participants. Then, in February 2016, when oil prices reached very low levels, worsening Mexico’s fiscal position, all predetermined mechanisms previously in place were suspended, and a discretionary US dollar sale was conducted. In January 2017, as market conditions deteriorated as a result of the new US administration’s inauguration, the FX Commission decided to intervene in a discretionary manner to restore operating conditions. This last intervention was the first time that the Bank of Mexico operated in the Mexican peso exchange market with nonresident counterparts.4
US Dollar Credit Auctions
To provide additional sources of US dollars to the Mexican private sector during the global financial crisis, the Bank of Mexico auctioned credit in US dollars to banks, which they, in turn, could lend to firms. International reserves were not used, as the resources came from a swap line established with the Federal Reserve Bank of New York. This mechanism was used only once, in April 2009. The central bank auctioned $4 billion, from which $3.2 billion were allotted.
Foreign Exchange Hedge Auction Program (Nondeliverable Forward Auctions)
Given the fall in oil production and prices of the past three years, Pemex has, since 2016, registered a negative trade balance. Also, since 2016, Pemex has not been selling US dollars to the Bank of Mexico, so the traditional main source to accumulate international reserves has disappeared, at least in the short and medium terms.
In this environment, the FX Commission needed an intervention mechanism that prevented the stock of international reserves from falling. At the same time, the FX Commission wanted a mechanism to provide a foreign exchange hedge to market participants; so it introduced the auctions of nondeliverable forwards settled in Mexican pesos in February 2017. Through this instrument, market participants bid for the forward exchange rate they need at the maturity of the contract. If the exchange rate at the time of maturity is higher than the forward exchange rate assigned in the auction (a depreciation takes place), the commercial bank that owns the contract makes a profit that is settled in pesos. In this way, the commercial bank ensures that it has the exact number of pesos it needs to buy the notional amount of US dollars embedded in the foreign exchange hedge contract. In contrast, if the exchange rate at the time of maturity is lower than the forward exchange rate assigned in the auction (an appreciation takes place), the commercial bank that owns the contract ends up with a loss that is settled in pesos. Even with this loss, the commercial bank ensures that it has the exact number of pesos it needs to buy the notional amount of US dollars embedded in the foreign exchange hedge contract.
The program was announced for a notional amount of up to $20 billion. An initial auction for $1 billion was allotted in March 2017. Afterward, during the fourth quarter of 2017, additional auctions of $4.5 billion took place. One important characteristic of this mechanism is that, at each expiration, all nonde-liverable forward positions are rolled over until the FX Commission deems it necessary to continue.
It is important to mention that, even though this mechanism does not undermine the amount of US dollars available to intervene in the spot market, it represents a short position of US dollars for the central bank. As a result, if the use of this mechanism begins to represent an important share of the stock of international reserves, even though their availability is not changed, there would be a signaling effect that many market analysts and rating agencies could begin to highlight in their credit assessments.
Flexible Credit Line with the IMF to Supplement Current International Reserves
In 2009, Mexico was one of three countries able to access the IMF Flexible Credit Line facility. The facility is only available for economies with sound economic fundamentals and policies, and its objective is to enhance their economic position in the event of external shocks. The credit line represents potential additional reserves for intervening in the foreign exchange market. Even though the credit line has not been used, it has been renewed several times. In the first installment (2009), the credit line was for special drawing rights (SDRs) 31.5 billion (about $47 billion), and it increased 50 percent in 2011. In 2016, the amount available in SDRs increased again, by more than 30 percent, reaching SDR 62.4 billion (close to $90 billion).
Intervention Effectiveness
To determine the effectiveness of a foreign exchange intervention, its objective has to be very well defined. In this regard, when the Bank of Mexico bought or sold US dollars to accumulate international reserves or to reduce the pace of accumulation, respectively, the interventions were quite effective. Every time the auction of put options described above was in place, the central bank accumulated more reserves, and every time a mechanism to reduce the pace of accumulation was used, the central bank accumulated fewer reserves than it would have without the mechanism. Hence, the Bank of Mexico’s foreign exchange interventions were quite effective.
When the aim of interventions is to provide liquidity to the market, to ensure operating conditions in the foreign exchange market, and to temper volatility, it is difficult to determine their effectiveness. For example, even if volatility does not go down after intervention, it is not necessarily true that the intervention was ineffective. In fact, central banks do not have the counterfactual case to compare what would have been the outcome without the intervention. Therefore, any result from an analysis of the efficiency of interventions should be taken with caution.
The theoretical literature on this topic proposes several channels through which sterilized central bank interventions could influence the economy. These channels are known as the “portfolio balance channel,” the “signaling channel,” and the “equilibrium selection channel.” The idea behind a portfolio balance channel is that if different currencies are not perfect substitutes, for investors to absorb an increase in the supply of an asset denominated in a currency, they will demand higher returns on that asset. Therefore, a sterilized intervention (an increase in the supply of US dollars) will result in an increase in the price of the currency relative to the US dollar (for more information, see Canales-Kriljenko, Karacadag, and Guimaraes 2003).
The second channel proposed in the theoretical literature is the signaling or “expectations” channel. This channel suggests that exchange market interventions do not have an effect on their own, but they do contain information about future monetary policy.
The third channel revealed in the literature is the “equilibrium selection channel.” This channel relies on the deviation of the exchange rate from the level that fundamental conditions warrant and can be underpinned (Shleifer and Vishny 1986). In that case, expected future actions of a large market participant can convince fundamentally driven investors to enter the market and bet on the right currency level (Shleifer and Vishny 1986).
The empirical literature has not arrived at a consensus about the effectiveness of currency market interventions. The literature that focused on developed economies has found mixed results (Neely 2005). While interventions have generally been found to have small effects, some interventions seem to have been effective, particularly those associated with large interventions coordinated among several central banks. The literature that focuses on developing economies has not reached a consensus either. Although evidence suggests that some interventions have been effective, no regular circumstances or intervention forms have been deemed more effective than others (Broto 2012).
García-Verdú and Ramos-Francia (2014) evaluate the intervention effects for Mexico on the expected exchange rate, using derivatives market data through a linear regression model. The authors find little evidence of an effect on the expected exchange rates’ distribution. In some cases, they find statistical evidence of a certain effect. However, it tends to be short-lived or not economically significant.
To summarize, this section assesses the efffectiveness of foreign exchange interventions to provide liquidity to the foreign exchange market for ensuring adequate operating conditions and tempering foreign exchange volatility. To make an adequate assessment, the first step is to determine whether the foreign exchange interventions altered the behavior of the exchange rate following a principal component approach. The second step is to use traditional econometric methodologies to assess the effectiveness of those interventions. The section considers the effect of these interventions over the different moments of the volatility and bid-ask spread distributions.
Measuring the Effect of Foreign Exchange Interventions over the Behavior of the Exchange Rate with a Principal Component Approach
This section analyzes the effects on the exchange rate before and after specific interventions. The observed exchange rate is compared with a portfolio of twelve currencies that displayed a high correlation with the Mexican peso before each intervention.5 The underlying assumption is that, in the absence of idiosyncratic shocks to the Mexican peso, it should follow the movements of this currency portfolio (referred to as synthetic MXN hereafter). The synthetic MXN is based on a principal component approach in which currency weights are determined by the linear combination of the three first principal components that explain 85 percent of the variance of these currencies against the US dollar.6
The methodology is applied to five different episodes in which the Bank of Mexico intervened in the foreign exchange market, using a time frame of 96 hours (4 days) before and after each of these foreign exchange policy actions.7 The five episodes represent three different kinds of foreign exchange interventions:
1. The announcement of US dollar auctions in December 2014 and March 2015, with and without a minimum price, respectively;
2. The implementation of two direct interventions in the foreign exchange market, one in February 2016 and the other one in January 2017; and
3. The announcement of the foreign exchange hedge auctions program of February 2017.
The results of the analysis are in the following section.
Announcement of US Dollar Auctions
As noted, on December 8, 2014, the FX Commission began the “USD daily auctions with minimum bid price” mechanism. On March 11, 2015, the FX Commission announced the “USD daily auctions without a minimum bid price.”
Under the methodology, a synthetic MXN was calculated for the four days after each of those two announcements of interventions. Figure 11.6 shows that before the intervention, in both cases, the observed Mexican peso was depreciated relative to the synthetic MXN; after the intervention, the observed Mexican peso turned out to have appreciated relative to the synthetic MXN—meaning that it is likely that those announcements affected the exchange rate. In both cases, in the four days after the announcement, the observed Mexican peso was approximately 1.5 percent more appreciated than the synthetic MXN. The outperformance of the Mexican peso relative to the synthetic MXN could be explained by US dollar liquidity injected by the central bank. However, it is not possible, with a high degree of confidence, to derive this conclusion from this analysis, because idiosyncratic factors of the Mexican economy may have affected the Mexican peso behavior.
Mexican Peso and Synthetic MXN with Auction Announcements
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Auction Announcements
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Auction Announcements
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Direct Interventions in the Foreign Exchange Market
In February 2016 and January 2017, the FX Commission instructed the Bank of Mexico to sell US dollars outright in the market to supply liquidity and ameliorate observed volatility; it did so without targeting a determined level for the exchange rate. The outright intervention of February 2016 was followed by a very benign context in the global financial markets, driven by a recovery in oil prices after a sharp decline. Figure 11.7 shows that the Mexican peso outperformed the synthetic MXN in the four days after the intervention.
Mexican Peso and Synthetic MXN with Direct Interventions
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Direct Interventions
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Direct Interventions
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.The outright intervention of January 2017 was followed by a negative environment in the global financial markets, ahead of the inauguration of a new US government. In this case, the Mexican peso underperformed the synthetic MXN in the first days after the intervention. By no means can this behavior be interpreted as an ineffective intervention. As noted, it is not straightforward to measure what would have been the behavior of the Mexican peso without the US liquidity provided by the central bank.
On February 2017, the FX Commission announced the implementation of the foreign exchange hedge auctions program described earlier. In this case, the Mexican peso clearly outperformed the synthetic MXN (Figure 11.8). As in the previous cases, this behavior could be explained by the implementation of the foreign exchange mechanism. However, that cannot be strongly concluded, because of possible idiosyncratic factors in the Mexican economy during those days.
Mexican Peso and Synthetic MXN with Hedge Auctions Announcement, February 2017
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Hedge Auctions Announcement, February 2017
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Mexican Peso and Synthetic MXN with Hedge Auctions Announcement, February 2017
(Mexican peso per US dollar)
Source: Bank of Mexico’s estimates, using Bloomberg Finance L.P. data.Plenty of evidence in the economic literature proves that the stock of international reserves enhances resilience to external shocks. In general, the economic literature related to adequate international reserves establishes that a larger amount of reserves diminishes the probability of a sudden stop; for example, cost-benefit approaches, such as the one proposed by Calvo, Izquierdo, and Loo-Kung (2012). Therefore, using a mechanism that does not reduce the amount of US dollars available to intervene in a shock to the balance of payments, may send a signal that the external accounts remain strong, and therefore, the impact of the intervention may be more effective.
Even though other idiosyncratic factors could be affecting the behavior of the Mexican peso, it might also be suggested that foreign exchange interventions tend to temper strong depreciations of the currency.
Regression Analysis of the Effect of Interventions on Option-implied Volatility and Skewness
Another measure of the effectiveness of foreign exchange interventions is the analysis based on an exercise similar to García-Verdú and Ramos-Francia (2014), using regression analysis to evaluate intervention effects on different features of market expectations about the Mexican peso. Different models were estimated for the options-implied volatility and skewness of the currency (in levels and differences). The specification for each model is given by the following:
where Intt represents the intervention dummy,
The main result of the estimation using this approach (Table 11.4) is that, controlling for the behavior of the currencies of similar countries (Brazil, Chile, Colombia, and Peru) as well as Canada, the average effect of interventions during 2010–17 has been a significant decrease in the options-implied, short-run volatility of the exchange rate. Quantitatively, the estimates suggest that an intervention is, on average, followed by a decrease of half a percentage point in the options-implied one-week volatility. However, the result is not statistically significant for the three-month implied volatility, which supports the idea that there is no strong statistical evidence to suggest that interventions always affect the whole structure of the volatility curve.
Effect of Interventions on the One-Week and Three-Month Options: Implied Volatility and Skewness of the Exchange Rate
Effect of Interventions on the One-Week and Three-Month Options: Implied Volatility and Skewness of the Exchange Rate
| Statistic | Maturity | Intt | β0 | st-1 | Intt-2 | ||
|---|---|---|---|---|---|---|---|
| Volatility | 1-week significance level | -0.5039*** | -0.3536*** | 0.8867*** | -0.0664** | 0.0758*** | 0.1717*** |
| 3-month significance level | -0.0469 | 0.0276 | 1.1827*** | -0.2253*** | 0.0148* | 0.0322*** | |
| Skewness 1-week significance level | 0.0135 | -0.0961*** | 0.8211*** | -0.0315 | 0.1348*** | 0.2167*** | |
| 3-month significance level | -0.0047 | -0.0258* | 0.8211*** | -0.0630*** | 0.0356*** | 0.1141*** |
Effect of Interventions on the One-Week and Three-Month Options: Implied Volatility and Skewness of the Exchange Rate
| Statistic | Maturity | Intt | β0 | st-1 | Intt-2 | ||
|---|---|---|---|---|---|---|---|
| Volatility | 1-week significance level | -0.5039*** | -0.3536*** | 0.8867*** | -0.0664** | 0.0758*** | 0.1717*** |
| 3-month significance level | -0.0469 | 0.0276 | 1.1827*** | -0.2253*** | 0.0148* | 0.0322*** | |
| Skewness 1-week significance level | 0.0135 | -0.0961*** | 0.8211*** | -0.0315 | 0.1348*** | 0.2167*** | |
| 3-month significance level | -0.0047 | -0.0258* | 0.8211*** | -0.0630*** | 0.0356*** | 0.1141*** |
Table 11.4 also shows the effects of interventions over the options-implied skewness. However, in this case no statistically significant evidence exists to infer that this variable is affected by foreign exchange interventions.
Although the results found for the implied volatility are interesting and some of them statistically significant, Table 11.5 shows that they are not robust to estimating the same relationship using the first difference of the variables instead of the level of the variables. The coefficient on the effect of interventions on one-week volatility is not significant, and it is of the opposite sign of that in Table 11.4, again suggesting that the effect on volatility is only temporary.
Effect of Interventions on the First Difference of the Options: Implied One-Week and Three-Month Volatility and Skewness of the Exchange Rate
Effect of Interventions on the First Difference of the Options: Implied One-Week and Three-Month Volatility and Skewness of the Exchange Rate
| ΔStatistic | Maturity | Intt | β0 | Δt-1 | ||
|---|---|---|---|---|---|---|
| Volatility | 1-week significance level | 0.0700 | -0.0040 | -0.0289* | 0.3421*** | 1.4846*** |
| 3-month significance level | 0.0016 | 0.0030 | 0.0142 | 0.3421*** | 1.2658*** | |
| Skewness | 1-week significance level | 0.0011 | 0.0012 | -0.0499 | 0.2893 | 1.6137 |
| 3-month significance level | 0.0070 | 0.0001 | -0.0752*** | 0.1896*** | 1.2713*** |
Effect of Interventions on the First Difference of the Options: Implied One-Week and Three-Month Volatility and Skewness of the Exchange Rate
| ΔStatistic | Maturity | Intt | β0 | Δt-1 | ||
|---|---|---|---|---|---|---|
| Volatility | 1-week significance level | 0.0700 | -0.0040 | -0.0289* | 0.3421*** | 1.4846*** |
| 3-month significance level | 0.0016 | 0.0030 | 0.0142 | 0.3421*** | 1.2658*** | |
| Skewness | 1-week significance level | 0.0011 | 0.0012 | -0.0499 | 0.2893 | 1.6137 |
| 3-month significance level | 0.0070 | 0.0001 | -0.0752*** | 0.1896*** | 1.2713*** |
Overall, the evidence suggests a negative effect, on average, of interventions on the options-implied, short-run volatility, which is consistent with the previous discussion. Interventions of different types, or in different periods, might have different effects, but the results suggest that central bank interventions affect markets in a way consistent with the stated objectives of improving the short-run functioning of the foreign exchange market.
Measuring the Effect of Interventions Over the Different Moments of Foreign Exchange Differences, Volatility, and Bid-Ask Spread Distributions
Traditional approaches to assessing the effectiveness of foreign exchange interventions, such as the econometric model described earlier, measure the impact of interventions during the first moment of the foreign exchange distribution for several indicators, such as exchange rate levels, average volatility, or bid-ask spread levels. However, these approaches do not look at the possible effects over the different moments of their respective distributions.
This section uses an approach based on the Mexican peso density functions related to the changes in the exchange rate, volatility, and liquidity to assess possible intervention effects that go beyond the first moment of the distribution.
Volatility Analysis Based on Intraday data for the Exchange Rate
Mexican peso volatility density functions four days before and four days after the day of the announcement of interventions were estimated using high-frequency intraday data and a Gaussian kernel approach. Although the comparison of these density functions shows that interventions have no significant effect on average volatility, they tend to reduce the presence of extremely high volatility before the policy actions. Figure 11.9 displays the density function of Mexican peso volatility during four days before and after the three kinds of interventions analyzed in the previous section: the announcement of US dollar auctions in December 2014 and March 2015, the implementation of two direct interventions in February 2016 and January 2017, and the announcement of the foreign exchange hedge auctions program.8
Mexican Peso Intraday Volatility Density Function, by Intervention
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of the distributions (mean and variance) were tested to compare the values before and after the policy actions for both episodes. For panels 1 and 2, a Welch’s t-test was used to compare the means and prove that they were statistically different. For panel 3, a Welch’s t-test was used to compare the means and prove that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were not statistically different; however, in March 2015, the variances were proven to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different. Intervention day’s data were eliminated to avoid momentary distortions because of this action. The vertical line shows the average of the observed data.Mexican Peso Intraday Volatility Density Function, by Intervention
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of the distributions (mean and variance) were tested to compare the values before and after the policy actions for both episodes. For panels 1 and 2, a Welch’s t-test was used to compare the means and prove that they were statistically different. For panel 3, a Welch’s t-test was used to compare the means and prove that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were not statistically different; however, in March 2015, the variances were proven to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different. Intervention day’s data were eliminated to avoid momentary distortions because of this action. The vertical line shows the average of the observed data.Mexican Peso Intraday Volatility Density Function, by Intervention
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of the distributions (mean and variance) were tested to compare the values before and after the policy actions for both episodes. For panels 1 and 2, a Welch’s t-test was used to compare the means and prove that they were statistically different. For panel 3, a Welch’s t-test was used to compare the means and prove that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were not statistically different; however, in March 2015, the variances were proven to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different. Intervention day’s data were eliminated to avoid momentary distortions because of this action. The vertical line shows the average of the observed data.In each panel of Figure 11.9, the density function for the days before the interventions is depicted with a green line, and the density function for the days after the intervention with a blue line.
In some cases, the volatility density function seems to have a minor drift with no clear direction for assessing the net effect of interventions (similar to some of the results presented in the literature previously mentioned). Because, the counterfactual case for comparing the average volatility at a given point in time with and without foreign exchange interventions is not available, it is not possible to make a strong conclusion from the analysis. Nonetheless, most of the interventions seem to reduce the presence of extremely high volatility, which is shown in Figure 11.9 as the long right-hand tail of the Mexican peso volatility density function that disappears after the majority of the interventions.9 Overall, the three intervention mechanisms analyzed seem to have a similar effect on volatility.
Liquidity Analysis Based on Intraday Data for Bid-Ask Spreads
Another implication of foreign exchange interventions is their effect on market liquidity, as measured by bid-ask spreads. After all, central bank interventions aim to provide liquidity and to foster stable market operating conditions. For the Mexican peso, evidence suggests that interventions tend to prevent these spreads from increasing significantly, given that the relative scarcity of foreign currency is compensated by the central bank.
Analogous to the density function analysis for Mexican peso volatility, density functions for bid-ask spreads were estimated before and after each of the interventions mentioned above. The results, depicted in Figure 11.10, suggest the following:
In the announcement of US dollar auctions, the March 2015 event reduced the long right-hand tail of the bid-ask spread density function. However, the intervention of December 2014 coincided with an increase in the right-hand tail.
The direct intervention in January 2017 slightly reshaped the density function, which reduced extreme values. However, evidence for the intervention of February 2016 does not show major changes in the distribution.
The announcement of the foreign exchange hedge program in February 2017, despite reducing some extreme values, clustered more observations on the right side, which suggests an overall increase in bid-ask spreads.
Mexican Peso Intraday Bid-Ask Spread Density Function, by Intervention Day, 2014–17
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of these distributions (mean and variance) were tested to compare the values before and after the policy actions. The intervention day’s data were eliminated to avoid momentary distortions due to this action. The vertical line shows the average of the observed data. For panel 1, a Welch’s t-test was used to compare the means and proved that they were statistically different. For panel 2, a Welch’s t-test was used to compare the means, and in February 2016 proved that they were statistically different; however, in January 2017 they did not prove to be different. In panel 3, a Welch’s t-test was used to compare the means and proved that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were statistically different; however, in March 2015 the variances were proved not to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different.Mexican Peso Intraday Bid-Ask Spread Density Function, by Intervention Day, 2014–17
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of these distributions (mean and variance) were tested to compare the values before and after the policy actions. The intervention day’s data were eliminated to avoid momentary distortions due to this action. The vertical line shows the average of the observed data. For panel 1, a Welch’s t-test was used to compare the means and proved that they were statistically different. For panel 2, a Welch’s t-test was used to compare the means, and in February 2016 proved that they were statistically different; however, in January 2017 they did not prove to be different. In panel 3, a Welch’s t-test was used to compare the means and proved that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were statistically different; however, in March 2015 the variances were proved not to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different.Mexican Peso Intraday Bid-Ask Spread Density Function, by Intervention Day, 2014–17
Source: Bank of Mexico’s estimates, using Thomson Reuters Worldscope data.Note: For all panels, the first and second moments of these distributions (mean and variance) were tested to compare the values before and after the policy actions. The intervention day’s data were eliminated to avoid momentary distortions due to this action. The vertical line shows the average of the observed data. For panel 1, a Welch’s t-test was used to compare the means and proved that they were statistically different. For panel 2, a Welch’s t-test was used to compare the means, and in February 2016 proved that they were statistically different; however, in January 2017 they did not prove to be different. In panel 3, a Welch’s t-test was used to compare the means and proved that they were not statistically different. For panel 1, an F-test was used to demonstrate that the variances in December 2014 were statistically different; however, in March 2015 the variances were proved not to be different. For panels 2 and 3, an F-test was used to demonstrate that the variances were statistically different.The evidence is not clear whether interventions help reduce extreme values of the bid-ask spread.10 However, given the lack of a counterfactual, it is difficult to claim categorically that interventions had no positive impact on the bid-ask spreads. Further analysis is required to confirm this preliminary evidence.
Kolmogorov-Smirnov Test for the Exchange Rate Volatility Distributions Based on Intraday Data
In line with previous sections, and using the same high-frequency intraday data, a two-sample Kolmogorov-Smirnov Test was performed to examine whether exchange rate interventions changed the distribution of the volatility of the Mexican peso. Table 11.6 summarizes the results when contrasting the null hypothesis H 0:F prior(Vol MXN/USD) = F post(Vol MXN/USD)vs. H a:F prior(Vol MXN/USD) ≠ F post(Vol MXN/USD), where F prior(Vol MXN/USD) and F post(Vol MXN/USD) correspond to the distribution of the volatility of the exchange rate for four days before and four days after each intervention, respectively. For the interventions considered, the Kolmogorov-Smirnov statistic is small when compared with the three critical values presented in four of the five interventions; therefore, the null hypothesis is not rejected in the majority of cases. With these results we can only confirm that the volatility of the exchange rate was modified with the direct intervention of January 2017.
Results for the Kolmogorov-Smirnov Test for the Difference in the Mexican Peso Distribution before and after Interventions
Results for the Kolmogorov-Smirnov Test for the Difference in the Mexican Peso Distribution before and after Interventions
| Type of Intervention | Intervention Date | Kolmogorov-Smirnov Statistic | Critical Values | Reject H0? | ||
|---|---|---|---|---|---|---|
| a = 0.01 | a = 0.05 | a = 0.10 | ||||
| Announcement of US dollar auction | December 2014 March 2015 |
0.3333 0.2188 |
0.4569 0.4447 |
0.3812 0.3711 |
0.3435 0.3344 |
No No |
| Direct intervention on the foreign exchange market | February 2016 January 2017 |
0.1277 0.3187 |
0.4037 0.4143 |
0.3368 0.3457 |
0.3035 0.3115 |
No Reject at 5% |
| Foreign exchange hedge auctions announcement | February 2017 | 0.2876 | 0.3949 | 0.3295 | 0.2970 | No |
Results for the Kolmogorov-Smirnov Test for the Difference in the Mexican Peso Distribution before and after Interventions
| Type of Intervention | Intervention Date | Kolmogorov-Smirnov Statistic | Critical Values | Reject H0? | ||
|---|---|---|---|---|---|---|
| a = 0.01 | a = 0.05 | a = 0.10 | ||||
| Announcement of US dollar auction | December 2014 March 2015 |
0.3333 0.2188 |
0.4569 0.4447 |
0.3812 0.3711 |
0.3435 0.3344 |
No No |
| Direct intervention on the foreign exchange market | February 2016 January 2017 |
0.1277 0.3187 |
0.4037 0.4143 |
0.3368 0.3457 |
0.3035 0.3115 |
No Reject at 5% |
| Foreign exchange hedge auctions announcement | February 2017 | 0.2876 | 0.3949 | 0.3295 | 0.2970 | No |
Kolmogorov-Smirnov Test for the Bid-Ask Spread Distribution Based on Intraday Data
Similarly, a two-sample Kolmogorov-Smirnov Test was performed to examine whether exchange rate interventions changed the bid-ask spread distribution of the Mexican peso. Table 11.7 summarizes the results when contrasting the null hypothesis.
Results for the Kolmogorov-Smirnov Test for the Bid-Ask Spread Distribution of the Exchange Rate before and after Interventions
Results for the Kolmogorov-Smirnov Test for the Bid-Ask Spread Distribution of the Exchange Rate before and after Interventions
| Type of Intervention | Intervention Date | Kolmogorov-Smirnov Statistic | Critical Values | Reject H0? | ||
|---|---|---|---|---|---|---|
| a = 0.01 | a = 0.05 | a = 0.10 | ||||
| Announcement of US dollar auction | December 2014 March 2015 |
0.3458 0.1718 |
0.1605 0.1546 |
0.1339 0.1290 |
0.1207 0.1163 |
Yes Yes |
| Direct intervention on the foreign exchange market | February 2016 January 2017 |
0.1367 0.1774 |
0.1410 0.1405 |
0.1176 0.1172 |
0.1060 0.1056 |
Reject at 5% Yes |
| Foreign exchange hedge auctions announcement | February 2017 | 0.1336 | 0.1375 | 0.1147 | 0.1034 | Reject at 5% |
Results for the Kolmogorov-Smirnov Test for the Bid-Ask Spread Distribution of the Exchange Rate before and after Interventions
| Type of Intervention | Intervention Date | Kolmogorov-Smirnov Statistic | Critical Values | Reject H0? | ||
|---|---|---|---|---|---|---|
| a = 0.01 | a = 0.05 | a = 0.10 | ||||
| Announcement of US dollar auction | December 2014 March 2015 |
0.3458 0.1718 |
0.1605 0.1546 |
0.1339 0.1290 |
0.1207 0.1163 |
Yes Yes |
| Direct intervention on the foreign exchange market | February 2016 January 2017 |
0.1367 0.1774 |
0.1410 0.1405 |
0.1176 0.1172 |
0.1060 0.1056 |
Reject at 5% Yes |
| Foreign exchange hedge auctions announcement | February 2017 | 0.1336 | 0.1375 | 0.1147 | 0.1034 | Reject at 5% |
H 0:F prior(Vol MXN/USDB-A) = F post(Vol MXN/USDB-A)vs. H a:F prior(Vol MXN/USDB-A) ≠ F post(Vol MXN/USDB-A), where F prior(Vol MXN/USDB-A) and F post(Vol MXN/USDB-A) correspond to the bid-ask distribution of the exchange rate for four days before and four days after each intervention, respectively. For the interventions considered, the Kolmogorov-Smirnov statistic is high when compared with the critical values presented; therefore, the null hypothesis is rejected in all cases at some level. These results confirm that exchange rate interventions modified the bid-ask spread distribution of the exchange rate.
How to Intervene
As noted, the FX Commission decides when to intervene in the foreign exchange market. Its considerations depend on the objective of the intervention. For example, if the Commission wants to accumulate additional international reserves or to reduce its pace of accumulation, then the obvious variable to look at is the stock of international reserves and to evaluate its marginal costs versus its marginal benefits.
If the decision is to provide liquidity to the market to ensure proper operating conditions and to temper the volatility of the exchange rate, then the variables to assess are those related to the operating conditions of the Mexican peso. Other variables, such as the real effective exchange rate or the nominal effective exchange rate, should also be considered. However, it is important to note that the FX Commission uses no single rule to decide the adequate time to intervene.
This section describes analytical tools the Bank of Mexico’s staff has provided to the FX Commission to facilitate the decision process. First, it briefly describes some of the models used to assess the level of international reserves. Second, it looks at some of the metrics used to assess the operating conditions of the Mexican peso. Those metrics are used to calculate an index of operating conditions that clearly indicates those days of poor operating conditions.
Models to Assess the Sufficiency of International Reserves
Bank of Mexico staff works with several indicators to assess the sufficiency of international reserves. As already mentioned, some of those metrics are (1) the ratio of reserves to GDP, compared to the ratio observed in other emerging markets; (2) the reserve adequacy metric proposed by Moghadam, Ostry, and Sheehy (2011) (the IMF metric); (3) cost-benefit approaches, such as the one proposed by Calvo, Izquierdo, and Loo-Kung (2012); and (4) utility-maximizing approaches, such as the one developed by Jeanne and Rancière (2006).
1. Ratio of international reserves to GDP: A straightforward metric used by rating agencies, market analysts, and the Bank of Mexico to judge how international reserves compare to other emerging market economies in terms of GDP. After the global financial crisis, this metric determined that the Bank of Mexico needed to accumulate more international reserves. Figure 11.11 shows how Mexico’s reserves compared to other countries’ reserves by the end of 2009.
2. IMF metric: This metric proposes a level of international reserves based on a weighted sum of variables related to potential US dollar outflows during periods of exchange market pressure. The metric considers that international reserves should cover between 100 percent and 150 percent of those potential outflows. After the 2008–09 financial crisis, it was appropriate to accumulate more reserves. After the put options mechanism was implemented in 2010, the stock of international reserves increased, reaching the middle point of the range suggested by the metric (Figure 11.12).
3. Cost-benefit approaches: Calvo, Izquierdo, and Loo-Kung (2012) propose an optimization approach that maximizes the benefits of holding international reserves subject to the costs that would arise under a sudden stop of capital inflows and the marginal cost of accumulating reserves (cost of carry). This metric shows that, in May 2003, when the mechanism to reduce the pace of accumulation of international reserves was put in place, the actual stock of reserves was above the level suggested by this model. Moreover, at the time the mechanism was put in place, the amount of US dollars Pemex was typically selling to the central bank was quite significant (see Figure 11.13).
4. When the put option mechanism was implemented after the “tequila crisis” of 1994–95, and after the global financial crisis, the actual stock of reserves (Figure 11.13) was below the level suggested by this model.
5. Utility-maximizing approach: The model proposed by Jeanne and Rancière (2006) determines the adequate level of reserves by maximizing the utility of a representative agent of the domestic economy during a sudden stop of capital inflow. This model is known as an extension of the Greenspan-Guidotti rule. It considers the total amount of short-term debt to calculate the sufficient level of reserves, as well as the potential drop of GDP during the sudden stop period and the cost of carry of holding international reserves. Figure 11.14 shows that the model has recently suggested a stock of reserves considerably above the one observed. However, once the resources from the flexible credit line with the IMF are considered, the stock of reserves seems appropriate. In view of the various approaches, the decision to buy or sell US dollars cannot be based on one single model or metric; even though this model may suggest higher reserves, it has to be put into perspective with the results from the other metrics.
International Reserves at the End of 2009: A Country Comparison of Stocks
(Percentage of GDP)
Sources: Bank of Mexico; Haver Analytics; and National Institute of Statistics and Geography.Note: The light sections of the bars indicate flexible credit lines with the IMF.International Reserves at the End of 2009: A Country Comparison of Stocks
(Percentage of GDP)
Sources: Bank of Mexico; Haver Analytics; and National Institute of Statistics and Geography.Note: The light sections of the bars indicate flexible credit lines with the IMF.International Reserves at the End of 2009: A Country Comparison of Stocks
(Percentage of GDP)
Sources: Bank of Mexico; Haver Analytics; and National Institute of Statistics and Geography.Note: The light sections of the bars indicate flexible credit lines with the IMF.
International Reserves and the IMF Metric, 2007–17
(Billions of US dollars)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.International Reserves and the IMF Metric, 2007–17
(Billions of US dollars)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.International Reserves and the IMF Metric, 2007–17
(Billions of US dollars)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.
International Reserves and the Calvo and Others’ Metric, 1993–2017
(Percentage of GDP)
Source: Bank of Mexico.International Reserves and the Calvo and Others’ Metric, 1993–2017
(Percentage of GDP)
Source: Bank of Mexico.International Reserves and the Calvo and Others’ Metric, 1993–2017
(Percentage of GDP)
Source: Bank of Mexico.
International Reserves and Jeanne-Rancière Metric, 2007–17
(Percentage of GDP)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.International Reserves and Jeanne-Rancière Metric, 2007–17
(Percentage of GDP)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.International Reserves and Jeanne-Rancière Metric, 2007–17
(Percentage of GDP)
Sources: Bank of Mexico; and the National Institute of Statistics and Geography.Metrics to Assess the Operating Conditions of the Mexican Peso
Given that the FX Commission has stated several times that the level of the exchange rate is not targeted, and that what matters is a liquid and well-functioning foreign exchange market, the Bank of Mexico’s staff follows several indicators that reflect the operating conditions of the Mexican peso. The variables that reflect these conditions are (1) observed volatility of the Mexican peso; (2) skewness and kurtosis of the Mexican peso; and (3) bid-ask spreads, among other factors.
With those indicators, the bank’s staff calculates an index that provides timely information about all trading and operating conditions of the foreign exchange market (hereafter referred to as the index of FX operating conditions). To calculate the index, each variable is normalized to its mean and a historical percentile is obtained. Then an index is constructed using the average of the percentiles computed in the previous step. This information shows current conditions of the foreign exchange market and helps identify periods when the market is operating normally.
Figure 11.15 shows that in February 2016 and January 2017, the FX Commission instructed direct interventions, given that operating conditions of the foreign exchange market were deteriorating. The index is only one of many measures the FX Commission frequently uses to decide when to intervene.
Index of Foreign Exchange Market’s Operating Conditions, 2016–17
(Five-day moving average)
Source: Bank of Mexico, using Thomson Reuters Worldscope data.Note: The index presents higher values when the operating conditions are deteriorated.Index of Foreign Exchange Market’s Operating Conditions, 2016–17
(Five-day moving average)
Source: Bank of Mexico, using Thomson Reuters Worldscope data.Note: The index presents higher values when the operating conditions are deteriorated.Index of Foreign Exchange Market’s Operating Conditions, 2016–17
(Five-day moving average)
Source: Bank of Mexico, using Thomson Reuters Worldscope data.Note: The index presents higher values when the operating conditions are deteriorated.Process for Deciding How Much to Intervene
There is no specific rule that determines how much to intervene. For interventions to accumulate reserves or reduce the pace of accumulation of reserves, the amount is determined based on the objective for the stock of reserves. Given that those amounts tend to be relatively high, a rules-based mechanism is used to smooth the intervention through time and to avoid altering the liquidity conditions of the foreign exchange market (that is, a put options mechanism to accumulate international reserves or daily auctions without a minimum price to sell US dollars).
On the other hand, when the objective of the intervention is to provide liquidity to the market, decisions have been ad hoc. For example, in October 2008, the Bank of Mexico sold more than $6 billion in a single day. That number was based on market intelligence about the shortage of foreign exchange liquidity in the market during that day. In outright interventions to sell US dollars (February 2016 and January 2017), the amount was determined by taking into account the total daily turnover of the Mexican peso, and that it had to be large to send a strong signal to the market.
For daily auctions of US dollars with and without a minimum price, the amount is based on the potential amount of US dollars that would be sold during a particular period. Even though those mechanisms are usually announced through these mechanisms without a specific expiration date, the FX Commission keeps in mind the potential amount of US dollars that could be sold if the mechanism prevailed for several months.
No specific rules determine the amount of foreign exchange interventions; the decision is taken on a case-by-case basis. The decision must consider liquidity conditions in the foreign exchange market and daily turnover of the Mexican peso, and the signal the FX Commission wants to send to the market, among other things.
Intervention Policy Communication
When considering intervention policy communication and transparency, several views should be considered:
1. Chiu (2003) concludes that the benefits of transparency (enhancement of policy effectiveness, strengthening of governance by promoting accountability, signaling, and reduced speculation, among others) in foreign exchange interventions appear to outweigh the risks (erosion of credibility if interventions fail to achieve their objective, reduced ability to surprise speculators, impression of distress by market participants and the public).
2. On the other hand, Chutasripanich and Yetman (2015) remark that adding an element of “opaqueness” to foreign exchange interventions tends to reduce the size of speculative flows and the cost of carrying reserves, despite increasing volatility in exchange rates, current account balances, and reserves.
Since December 1994, the Bank of Mexico has been very transparent in its foreign exchange interventions. In fact, all interventions are communicated to the public through its website.11 For the majority of interventions, information is available on amounts, dates, and reasons for intervention. Every intervention is accompanied by a press release from the FX Commission in which it clearly explains the intervention (objective, amounts, operating details, and so on).12 It is only when a discretionary intervention takes place, that the press release will omit specific details, such as amounts. However, the amount of the intervention can be inferred ex post from the information published in the weekly financial statement of the central bank. In that report, the stock of international reserves is presented with the weekly flows, and it breaks down the factors that explain the changes (market interventions).
In addition, from all the information the FX Commission makes public, and the operating conditions index the central bank publishes every now and then in its minutes for monetary policy decisions or in other reports, it is relatively predictable when the central bank may act. Sometimes, before an intervention, market analysts begin to report that a Bank of Mexico intervention is likely. However, they do not precisely predict the size or timing.
The Main Lessons from Mexico
The main lesson from more than 20 years of intervening in the foreign exchange market under a flexible exchange rate regime, and from interventions before the free-floating regime, is that there is no use targeting a specific level for the exchange rate. All interventions by the Bank of Mexico are respectful of the floating regime, and the prices at which they are conducted are either auction-based, or they are closed for discretionary direct interventions and consider prevailing market prices.
Preannounced mechanisms are preferred over interventions without previous notice, but the discretion to use the latter should always be considered when circumstances warrant it. In this regard, market participants, who understand the settings of the prevailing mechanism, have, on a few occasions, played with the rules to manipulate the exchange rate in an undesirable way. In those circumstances, an extraordinary measure or change of mechanism would take place.
No unique tool fits all situations. Even if the main objective of the intervention is the same in two different episodes, the different conditions can call for different tools. Regarding the objectives of foreign exchange intervention, it is difficult to statistically prove the effectiveness of an intervention, as measured by its effect on market operating conditions, because intervention counterfactuals are unknown. The FX Commission has therefore been careful to limit interventions to episodes in which they are deemed extremely necessary. Transparency has always been a good policy; it is important that markets and the public are informed of any type of intervention as long as it does not weaken its effectiveness.
References
Broto, C. 2012. “The Effectiveness of Forex Interventions in Four Latin American Countries.” Working Paper 1226, Bank of Spain, Madrid.
Calvo, Guillermo A., Alejandro Izquierdo, and Rudy Loo-Kung. 2012. “Optimal Holdings of International Reserves: Self-Insurance against Sudden Stop.” NBER Working Paper 18219, National Bureau of Economic Research, Cambridge, MA.
Canales-Kriljenko, Jorge I., Cem Karacadag, and Roberto Pereira Guimarães. 2003. “Official Intervention in the Foreign Exchange Market: Elements of Best Practice.” IMF Working Paper WP/03/152, International Monetary Fund, Washington, DC.
Chiu, Priscilla. 2003. “Transparency versus constructive ambiguity in foreign exchange intervention.” Working Paper 144, Bank for International Settlements, Basel.
Chutasripanich, Nuttathum, and James Yetman. 2015. “Foreign exchange intervention: strategies and effectiveness.” Working Paper 499, Bank for International Settlements, Basel.
García-Verdú, Santiago, and Manuel Ramos-Francia. 2014. “Interventions and Expected Exchange Rates in Emerging Market Economies.” Quarterly Journal of Finance 4 (1): 34.
Jeanne, Olivier D., and Romain Rancière 2006. “The Optimal Level of International Reserves for Emerging Market Countries: A New Formula and Some Applications.” IMF Working Paper 06/98, International Monetary Fund, Washington, DC.
Moghadam, Reza, Jonathan D. Ostry, and Robert Sheehy. 2011. “Assessing Reserve Adequacy.” International Monetary Fund, Washington, DC.
Neely, Christopher J. 2005. “An Analysis of Recent Studies of the Effect of Foreign Exchange Intervention.” Working Paper Series 2005–030B, Federal Reserve Bank of St. Louis, MO.
Shleifer, Andrei, and Robert W. Vishny. 1986. “Large Shareholders and Corporate Control.” Journal of Political Economy 94 (3): 461–88.
The FX Commission, which oversees Mexican foreign exchange policy, comprises three members of the Ministry of Finance (the minister and two deputy ministers) and three members from the Bank of Mexico’s board (including the governor). The minister of finance chairs the Commission and has the casting vote. In the absence of the minister of finance, the central bank’s governor presides over the session.
These metrics are described later in this chapter.
The FIX exchange rate is determined by the Bank of Mexico as an average of quotes in the wholesale foreign exchange market for operations payable in 48 hours. The central bank informs the FIX from 12 o’clock onward each banking day. It is published in the Official Gazette (Diario Official de la Federación) one banking business day after its determination date, and it is used to settle liabilities denominated in US dollars payable in Mexico on the day after its publication in the Official Gazette.
Previous interventions had been conducted exclusively with banks licensed and domiciled in Mexico. In this last intervention, for the first time, the Bank of Mexico operated in the Mexican peso exchange market with institutions located outside Mexico. The objective was to show the market that the central bank was capable of intervening at any time, even during hours of poor liquidity.
The currencies included are the South African rand, Canadian dollar, Colombian peso, Turkish lira, Brazilian real, Chilean peso, Singapore dollar, Peruvian sol, Polish zloty, Australian dollar, New Zealand dollar, and the Czech koruna.
With this approach, the main factors that explain the joint movement of all the currencies in the basket are captured in one single variable. Tis approach also rules out idiosyncratic effects of a particular currency included in the synthetic portfolio—for instance, the effect of any exchange rate intervention in any of the currencies included in the portfolio or any other relevant episode.
The base time is the observation in the window period (the fifth day before the intervention). A longer window period was not considered, as it could cause issues by including other episodes that may have affected the exchange rate.
Intraday data is grouped every 15 minutes.
The first and second moments of these distributions (mean and variance) were tested. In most cases, the test suggested a change in the distribution.
For the bid-ask spread distributions, the first and second moments were tested in means and variance. Similar to volatilities, the test suggested a significant change in the distribution.
FX Commission bulletins are available in Spanish at http://www.banxico.org.mx/informacion-para-la-prensa/comunicados/politica-cambiaria/comision-de-cambios/index.html.
Most of the press releases are published at the same time of the intervention. When the intervention is a rules-based mechanism, the press release comes before the intervention.
