Chapter

Chapter 10 Costa Rica: Learning to Float

Author(s):
Marcos Chamon, David Hofman, Nicolas Magud, and Alejandro Werner
Published Date:
February 2019
Share
  • ShareShare
Show Summary Details
Author(s)
Rodrigo Cubero, Valerie Lankester and Evelyn Muñoz 

This chapter was written while Rodrigo Cubero worked at the IMF. The views expressed are those of the authors and should not be attributed to the Central Bank of Costa Rica or to the IMF. The authors are grateful for comments from Marcos Chamon, Nicolás E. Magud, and participants at the conference “Foreign Exchange Intervention in Latin America: Learning from Experience,” International Monetary Fund, Washington, DC, February 27–28, 2018, and for research assistance from Diego Wachs.

Costa Rica provides a case study of how a small, highly dollarized and very open emerging market economy gradually learns to float. Since 2006, Costa Rica has transitioned from a crawling peg regime to a managed float as part of the move toward an inflation-targeting regime. Yet, the central bank has exhibited fear of floating: foreign exchange intervention has remained frequent and the nominal exchange rate relatively stable amid concerns about the impact of exchange rate fluctuations on price and financial stability. To learn from Costa Rica’s experience, this chapter first describes the structure of the exchange rate market and the modalities and motives of intervention. It then examines the effectiveness of intervention in changing the rate of variation or the volatility of the exchange rate during the float period by using two different methodologies over a high-frequency dataset: an event (or before-and-after) analysis and a standard econometric approach. It finds that, on balance, intervention has been effective.

Introduction

Over the past two decades, many emerging market and developing economies have modified their exchange rate regimes from some form of pegging toward greater flexibility. Monetary policy frameworks have moved from the exchange rate as a policy instrument to an ever-greater reliance on the interest rate, often in the context of adopting an inflation-targeting regime (IMF 2014; Laurens and others 2015). Yet, despite this move toward more flexible exchange rates, central banks in those countries have continued to engage frequently in foreign exchange intervention (Adler and Tovar 2011; Ostry, Gosh, and Chamon 2012; Benes and others 2013; IMF 2015).

That pattern is illustrated by the case of Costa Rica. Since 2006, the country has transitioned first from a crawling peg to a crawling band, and then to its current managed float, as part of its migration to an inflation-targeting regime. Since the last part of the crawling band regime in late 2013, the exchange rate has continuously floated—the exchange rate has moved away from the band limits and therefore the central bank was not forced to intervene to defend them.

Yet, under such a de facto float, the central bank has resorted to foreign exchange intervention for various purposes, including to contain excessive volatility. In fact, intervention has helped keep the exchange rate relatively stable during the float period, partly because of concerns over the effects of exchange rate fluctuations on price and financial stability, in the context of high trade and financial openness and high financial dollarization.

Costa Rica thus provides a case study of how a small, very open emerging market economy learns to float on its way to inflation targeting under significant structural constraints.

This chapter examines three main questions regarding foreign exchange intervention in Costa Rica:

  • First, what are the modalities of intervention, and how do they work in practice? The chapter notes that direct foreign exchange interventions take place exclusively in the spot market, are sterilized, are typically large relative to the daily market turnover, and are decided on a discretionary basis by a committee at the Central Bank of Costa Rica. They are guided by rules that remain undisclosed and motives only announced ex post (after the intervention takes place but on the same day). Communications by the central bank also provide an indirect and often effective intervention mechanism.
  • Second, what are the main motives and rationales of intervention? Three main motives exist: reserve accumulation, serving the needs of the public sector, and mitigating excessive fluctuations. In practice, however, all interventions seem intended to help stabilize the market, either by deliberately countering fluctuations or at least by avoiding exacerbating them. In effect, the central bank leans against the wind in its intervention decisions.
  • Third, has intervention been effective in changing the rate of variation or the volatility of the exchange rate? To address this final question, the analysis focuses on the period since late 2013, and it uses two different empirical methodologies over the same high-frequency dataset on foreign exchange transactions: a standard econometric approach and an event (or before-and-after) analysis. It finds that, on balance, central bank intervention is effective in mitigating exchange rate fluctuations in Costa Rica.

The chapter is structured as follows: It first describes the evolution of the exchange rate regime and then reviews the structure and main features of the foreign exchange market in the country. It discusses the modalities, motives, and decision-making process of official interventions, and empirically analyzes the effectiveness of them. The final section concludes.

Evolution of the Foreign Exchange Regime

Since the creation of the Central Bank of Costa Rica in 1950, the country has had four exchange rate regimes: fixed exchange rate (1950–80), crawling peg (1984–2006), crawling band (2006–15), and managed float (2015–present), with a disorganized, de facto floating regime during the debt crisis (1980–83). This section briefly examines each, as background for the analysis in the remainder of the chapter.

Fixed Exchange Rate Regime (1950–80)

Upon the establishment of the Central Bank of Costa Rica in 1950, Costa Rica maintained its prevailing (official) fixed exchange rate regime, along with a small free market and multiple exchange controls and restrictions.1 Until the end of 1980, the peg remained stable, with only two discrete, step devaluations—in 1961, from 5.57 colones per US dollar to 6.65 colones, and in 1974, to 8.6 colones. However, severe negative external shocks and large fiscal deficits in the late 1970s led to a significant real exchange rate appreciation and eventually to an external debt moratorium and a massive loss of international reserves. The official exchange rate was devalued to 20 colones per US dollar. Unable to further withstand the pressures, however, in December 1980, the central bank allowed most transactions to take place in the interbank and parallel markets at a market exchange rate, and the fixed exchange rate regime was effectively abandoned. The colón depreciated sharply. High inflation and a deep economic contraction ensued (Edwards 1990; Rodríguez Clare and Rodríguez Echeverría 1990).

Crawling Peg (1984–2006)

A top priority after the crisis was to bring order to and reunify the exchange rate market (Delgado 2000). This was achieved toward the end of 1983. As a mechanism to ensure a smooth external adjustment, the central bank in 1984 gradually introduced a crawling peg regime, whereby the pace of adjustment of the nominal exchange rate was determined by the difference between Costa Rica’s inflation rate and the weighted average of its trading partners’ inflation.2 The crawling peg helped the country maintain a stable real exchange rate—a crucial component of its transition to an export-oriented development strategy (Figure 10.1).

Figure 10.1.Nominal and Real Effective Exchange Rates under Different Regimes, 1957–2017

Source: Central Bank of Costa Rica.

Note: The bilateral real exchange rate between Costa Rica and the United States is shown for January 1957 through November 1979. Costa Rica’s consumer price index was interpolated between January 1973 and October 1974 because of data irregularities.

However, against an open capital account (whose liberalization was completed in March 1992), the peg limited the autonomy and effectiveness of monetary policy (the “impossible trilemma”). Moreover, because of its inherent inertia and predictability, this regime (as well as the central bank’s large sustained losses) led to high inflation (18.7 percent on average between 1982 and 2006), which consequently required a sustained nominal depreciation (from about 43 colones per US dollar in 1984 to about 520 colones by 2006). It also stimulated financial dollarization. This, along with an open capital account, large fiscal deficits (averaging about 2 percent of GDP in 1982–2006), and the central bank operating deficits, reduced monetary policy effectiveness to bring about price stability (Alfaro Ureña, Sánchez Wong, and Tenorio Chaves 2016).

Crawling Band (2006–15)

Explicitly acknowledging the drawbacks of the crawling peg, in October 2006 the central bank switched to a crawling band regime, with no central parity.3 The shift was presented as a transitory strategy to allow for a gradual move toward greater exchange rate flexibility and the eventual adoption of a floating regime. This, in turn, was justified as a precondition for an eventual shift to inflation targeting, with the policy interest rate as the instrument. The rates of crawl of the band limits were set such that the band would widen over time, although the crawl parameters were modified several times, and the floor was eventually fixed, in July 2008, at C500.

For much of the crawling band period, the exchange rate remained pegged to either end of the band (Figure 10.2). Initially, it was stuck to the floor. In 2008, with the financial shocks around the global financial crisis, it moved to the ceiling. As external conditions improved, and capital inflows returned toward the final quarter of 2009, the exchange rate moved gradually back to the floor of the band and remained there or close to it from about October 2010 to late 2013. Starting in December 2013, against expectations of US monetary policy tightening, the exchange rate moved away from the floor of the band and remained within the band (in a sort of float, although with active intraband foreign exchange intervention), until the central bank officially switched to a floating regime in January 2015.

Figure 10.2.MONEX Exchange Rate under the Crawling Band Regime, 2006–14

Source: Central Bank of Costa Rica.

Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica (more details are provided in the “Foreign Exchange Market: Structure and Characteristics” section).

The crawling band regime brought significant progress to the monetary policy framework (Central Bank of Costa Rica 2015; Alfaro Ureña, Sánchez Wong, and Tenorio Chaves 2016). It allowed a gradual move away from the exchange rate anchor and toward greater exchange rate flexibility, and it created the conditions for an active and meaningful use of a policy interest rate from June 2011. In addition, with on-the-band intervention centered, by definition, on buying US dollars at the floor and selling them at the ceiling, the regime helped reduce central bank financial losses (Muñoz 2012). This, in turn, along with the greater effectiveness of monetary policy, contributed to a significant reduction in inflation, despite a deteriorating fiscal situation (Figure 10.3). The scope for wider exchange rate fluctuations induced a higher perception of currency risk, helping to continue an already present trend toward lower financial dollarization: the share of foreign currency deposits went down over 2006–14, even when adjusting for the effect of exchange rate changes (Figure 10.4). The effect on credit dollarization is less clear.4 In addition, as the monetary policy framework strengthened, the crawling band regime helped reduce the exchange rate pass-through to inflation, from 0.6 at the beginning of the period to 0.2 (Rodríguez Vargas 2009; Brenes and Esquivel 2017).

Figure 10.3.Consumer Price Index Inflation under Different Exchange Rate Regimes, 1957–2017

(Percent)

Source: Central Bank of Costa Rica data.

Note: Costa Rica’s consumer price index was interpolated between January 1973 and October 1974 because of data irregularities.

Figure 10.4.Financial Dollarization: Foreign Exchange Deposits and Credit to the Private Sector, 1997–2017

(Percent)

Source: Central Bank of Costa Rica data.

Note: To control for the effect of exchange rate changes on the local currency values of foreign exchange variables, these values are computed at the January 2014 mean exchange rate. Data shown are for December of each year.

Managed Float (2015–Present)

Considering these achievements and that the exchange rate had remained away from the band limits since December 2013, the Central Bank of Costa Rica formally adopted a managed floating regime in January 2015, as a necessary step to complete the transition to inflation targeting.

Under the new regime, the central bank commits to “allow the exchange rate to be determined freely by the market, but will participate in the market to meet its own foreign currency needs and those of the non-financial public sector and, at its discretion, to prevent sharp (literally, “violent”) fluctuations in the exchange rate” (Central Bank of Costa Rica 2015).

However, since the adoption of a de jure floating regime, the nominal exchange rate has remained relatively stable. The IMF classified Costa Rica’s de facto regime as “crawl-like” in 2017 and “stabilized” during 2015 and 2016 (Table 10.1). Both categories are under the “soft peg” type. This category applies when the spot market exchange rate has remained within a margin of 2 percent for six months or longer and is not floating; that is, the exchange rate remains stable “as a result of official action.”5 We review this topic again in the section on the anatomy of official foreign exchange intervention.

Table 10.1.IMF’s Classification of De Facto Exchange Rate Arrangements, 2006–17
YearCrawling PegCrawling BandCrawl-LikeStabilizedOther Managed
2006×
2007×
2008×
2009×
2010×
2011×
2012×
2013×
2014×
2015×
2016×
2017×
Source: IMF’s AREAER, 2006–17.
Source: IMF’s AREAER, 2006–17.

The managed float regime has furthered the transition toward inflation targeting that began in 2006 and has strengthened monetary policy.6 In particular, the scope for greater exchange rate flexibility under the managed float (even if dampened by intervention) has allowed the use of the policy interest rate as a policy instrument.7 Inflation has further declined and remained at or below target during this period, while inflation expectations—although sticky—have moved into the inflation target range, as shown in Figure 10.5.

Figure 10.5.Inflation Target, Expectations, and Observed Inflation, 2006–18

(Percent)

Source: Central Bank of Costa Rica data.

Foreign Exchange Market: Structure and Characteristics

Foreign exchange transactions in Costa Rica take place almost entirely in the spot markets—despite the authorities’ efforts to foster development of foreign exchange derivatives, the use of these instruments remains limited and largely incipient.

There are two main foreign exchange spot markets: over-the counter windows (“ventanillas”) and an organized electronic foreign exchange market (Mercado de Monedas Extranjeras, known as MONEX).

Over-the-Counter Windows (Ventanillas)

The ventanillas comprise over-the-counter foreign exchange transactions between financial institutions and the public, whether they are individuals or corporations, performed through physical or electronic platforms. Smaller, nonfinancial participants in this market tend to be price takers, and they face relatively high transaction costs, with spreads of around 2 percent between sale and purchase rates.8 While the traded amounts are generally small (90 percent are smaller than $1,000), the market accounts for about 89 percent of the total value of foreign exchange transactions. The average turnover in the over-the-counter market was about $141.5 million a day in 2017. This is very small by international standards (Figure 10.6).

Figure 10.6.Over-the-Counter Daily Average Foreign Exchange Turnover, 2016

(Percent of GDP)

Sources: Central Bank of Costa Rica; and Bank for International Settlements.

Electronic Foreign Exchange Market (MONEX)

This is an electronic platform provided by the central bank in which the central bank itself, financial intermediaries, and any individual or corporation with registered accounts can participate.9 To improve the price-making process, negotiating sessions have been gradually shortened, from an initial eight-and-a-half hours to three hours, at present. The minimum transaction amount is $1,000, and the cost for participants is 0.20 percent of the traded amount. Agents place their bids and offers on the electronic platform, and the transaction is completed and settled automatically in real time when a match emerges. An important feature of MONEX transactions is that they are anonymous: participants do not and cannot know the identity of the counterpart; only the central bank does.

MONEX represents only around 11 percent of the country’s total foreign exchange transactions in value terms, and this share has not changed significantly since the opening of this platform in 2006. The mean turnover value was $17 million per day in 2017 (Table 10.2). However, MONEX plays a critical role as the market in which price discovery takes place. Because financial institutions are subject to prudential limits on their net open foreign exchange positions, any excess demand or supply of foreign exchange in their over-the-counter transactions will typically be reflected in their positions in MONEX. As such, MONEX acts as the clearing and price-making channel for the country’s overall foreign exchange flows, despite its small share in total transactions. It is also the market in which the central bank’s interventions take place.

Table 10.2.Structure and Size of the Foreign Exchange Market, 2012–17 (Daily average turnover)
MONEXOver the CounterTotal
YearThousands of US DollarsPercent of TotalThousands of US DollarsPercent of TotalThousands of US dollars
201219,45414.2117,48285.8136,935
201317,58512.7120,88987.3138,474
201416,45512.1119,39587.9135,850
201515,27010.9124,65689.1139,926
201613,5519.3132,87190.7146,422
201717,03510.7141,54689.3158,581
Source: Central Bank of Costa Rica.Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.
Source: Central Bank of Costa Rica.Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

In 2017, there were 633 participants in the MONEX market, comprising the central bank, 4 state-owned banks, 11 private banks, 20 nonbank financial institutions (stock brokerage houses, credit unions, and nonbank intermediaries), and 597 private firms and individuals.

MONEX is characterized by a relatively high degree of concentration. The central bank accounts for about one-quarter of total turnover, but it faces powerful counterparts in MONEX (Figure 10.7). Excluding the central bank, the share of the top four participants (all private banks) in turnover was 57 percent in 2017, up from 47 percent in 2014 (Figure 10.8).10 It is widely recognized that a few large banks wield a disproportionately large power in the foreign exchange market (see, for example, the views from different economists cited in Lizano 2018). The share of individuals and nonfinancial corporations in the value of MONEX transactions has been declining, to less than 4 percent in 2017. Deterred by moderate barriers to entry (for example, traded amounts must be multiples of $1,000, with a minimum of $1,000 per transaction; and the know-how to use the electronic platform), smaller private sector agents may prefer the over-the-counter ventanillas, despite their higher spreads. This creates market segmentation.

Figure 10.7.MONEX Market: Structure, by Category, 2014–17

(Percent)

Source: Central Bank of Costa Rica data.

Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

Figure 10.8.MONEX Market: Share of Top Four Participants in Turnover, 2014–17 (Percent)

Source: Central Bank of Costa Rica data.

Note: The figure excludes the Central Bank of Costa Rica. MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

These features, along with the residual nature of MONEX (to which surplus or deficit positions in the over-the-counter market are shifted by financial institutions for clearing), create a risk that sudden shifts in overall foreign exchange flows or in sentiment by large players may inundate the MONEX market, fuel destabilizing speculation, and lead to excessive price volatility.

Anatomy of Official Foreign Exchange Intervention

This section studies the procedural and institutional aspects of foreign exchange intervention by the Central Bank of Costa Rica: the process for making intervention decisions, the modalities of intervention, the motives, and communications about intervention. The analysis focuses on interventions since December 2013; that is, in the latter part of the crawling band regime, when the exchange rate was not on the limits of the band and under the current managed floating regime. Interventions under any variant of a fixed exchange rate regime (including when the exchange rate was pegged to the band limits) lack analytical interest: they are not discretionary, but rather forced by the very nature of the regime, because the central bank is committed to buying or selling all foreign exchange needed to maintain the exchange rate at the given rate.

The analysis in this section is based on daily intervention data. Of course, the central bank can conduct multiple discrete interventions within one day, but these are all aggregated into daily observations for the purposes of this section.

Modalities of Intervention

As noted, official central bank intervention in Costa Rica takes place entirely in the MONEX (spot) market. There are no well-developed derivatives markets or derivative instruments that the central bank can use for intervention.

Between December 13, 2013, and March 9, 2018, there were 1,077 trading sessions in MONEX. During this period, central bank interventions in MONEX (sales and purchases) amounted to $6.5 million per day, on average, over all days of session, and to a mean of $8.8 million per day for the days when central bank interventions took place. The latter were 789 out of the 1,077 days (and because some sessions had interventions for different motives, the period comprised 832 daily interventions). The average daily value of transactions in MONEX for the period was $15.6 million; thus central bank interventions account for 57 percent of total MONEX turnover on days when interventions take place. This is clearly significant in value terms.

With the interest rate as policy instrument, all foreign exchange interventions must be sterilized to isolate the monetary effect of intervention and safeguard the independence and effectiveness of the policy rate (see Ostry, Gosh, and Chamon 2012; Benes and others 2013; and chapter 2 of this book). In the case of Costa Rica, all foreign exchange interventions are sterilized through daily liquidity auctions. These operations are approved daily by a committee (different from the market committee making intervention decisions), and their magnitude is determined depending on, among other factors, the effects of central bank intervention on the monetary base. As a result, during periods of sustained foreign exchange purchases (sales) by the central bank, its domestic liabilities increase (decrease) because of sterilization operations, in line with its net international reserves.

In addition to direct interventions through participation in MONEX, the central bank conducts more subtle or indirect intervention through two other mechanisms: (1) the decision not to intervene in MONEX to replenish reserves after fulfilling the foreign exchange needs of the public sector; and (2) the communications with the public to influence exchange rate expectations.

Decision-Making Process

Intervention decisions are guided by a general framework approved by the central bank’s Board of Directors (Figure 10.9).

Figure 10.9.Decision Makers for the Central Bank of Costa Rica’s Foreign Exchange Intervention

Source: Authors, based on information from the Central Bank of Costa Rica.

Note: ALMD = Assets and Liabilities Management Division; BCCR = Central Bank of Costa Rica.

In this framework, the strategy and specific criteria for intervention are defined by the Policy Committee, comprised of four members: the central bank president, the general manager, the chief economist, and the director of the Asset and Liability Management Division. A quorum is reached with the president and two other members of the committee, and decisions are made by simple majority. In a tie, the president’s vote counts twice.

The committee meets every Monday afternoon and several times per week, depending on the circumstances. In times of stress or high volatility, the committee meets every day, and sometimes even several times per day. The frequency of meetings is determined by daily information from staff, who continuously monitor foreign exchange market behavior, including developments in prices (for instance, trend and volatility of the exchange rate), trading volumes, and the overall deficit or surplus of foreign exchange in the market, and whether it is consistent with identified seasonal patterns. To back up this analysis, a model is used to predict foreign exchange volumes and trends. If staff identify significant movements that are unusual (once seasonality factors are considered) or not justified by fundamentals, it proposes a meeting of the Markets Committee to decide whether stabilization interventions are required.

The Markets Committee makes actual intervention decisions (timing and magnitude), and comprises the general manager, the chief economist, the director of the Asset and Liability Management Division, and the chief of the open market operations department. The decisions are made based on the general framework provided by the central bank board, the criteria and strategy set by the Policy Committee, and the information provided continuously by staff.

Any intervention decided by the Markets Committee will be executed by the trading desks in the Asset and Liability Management Division. They have little room for discretion, given that the relevant parameters of intervention (including amounts, timing, and total available budget) will have been set by the committee.

Intervention Motives

The central bank participates in the foreign exchange market for three main, publicly articulated reasons (Central Bank of Costa Rica 2015):

  • To “satisfy its own foreign exchange needs”
  • To “fulfill the foreign exchange needs of the nonfinancial public sector”
  • To “prevent violent exchange rate fluctuations”11

This subsection looks at each one of these in turn.

Intervention to Satisfy the Central Bank’s Foreign Exchange Needs

The central bank needs to accumulate reserves for precautionary purposes. To fulfill this need, starting in 2010, the central bank’s board approved and publicly announced five net international reserve accumulation programs (Table 10.3 and Figure 10.10). The announcements included the expected accumulation amounts and the time window for implementation, but they did not dictate the specific timing or magnitude for each intervention, which was left to the central bank’s discretion. The central bank board, however, prescribed that foreign exchange purchases for reserve accumulation be undertaken when there is a foreign exchange surplus, to prevent creating excessive volatility in the market. Against increased instability in the foreign exchange market, the fourth program was not completed, and the fifth was not implemented at all. There is no active reserve accumulation program at present.

Table 10.3.Central Bank of Costa Rica’s Recent International Reserve Accumulation Programs
ProgramExpected Implementation PeriodActual Implementation PeriodExpected Amount (in $ million)Actual Amount (in $ million)
ISep. 2010–Dec. 2011Sep. 6, 2010–Apr. 18, 2011600.0600.0
IIFeb. 2012–Dec. 2013Mar. 12, 2012–Jan. 11, 20131,500.01,500.0
IIIAug. 2014–Dec. 2015Oct. 21, 2014–Jul. 27, 2015250.0250.0
IVFeb. 2015–Dec. 2016Feb. 27, 2015–Dec. 31, 2016800.0698.1
VJan. 2016–Dec. 20171,000.00.0
Total4,150.03,048.0
Source: Central Bank of Costa Rica data.
Source: Central Bank of Costa Rica data.

Figure 10.10.Central Bank of Costa Rica’s Reserve Accumulation Programs: Cumulative Effect on Net International Reserves, 2010–15

(Thousands of US dollars)

Source: Central Bank of Costa Rica data.

The stated objective of these programs has been to strengthen Costa Rica’s net international reserve position on several adequacy metrics, such as GDP, short-term external debt, total external debt, monetary aggregates, or months of imports, sometimes using relevant comparator countries as a benchmark. The central bank has often defined its desired reserves as a range, typically as a ratio to GDP. For example, in 2006, it established a target minimum net international reserves at 11 percent of GDP (Laverde and Muñoz Salas 2006). More recently, it announced a target range for a net international reserve of 11–15 percent of GDP, including in its Strategic Plan for 2015–2019 (Central Bank of Costa Rica 2016). Some of the program announcements also indicated that the purchases should be made at the lowest price possible (that is, when the domestic currency has appreciated), and should balance between costs (inflation, exchange rate, and interest rates) and benefits (stronger external position).

Reserve accumulation is, of course, not discretionary during periods when the exchange rate is pegged to the bottom of the crawling band. This made Programs I and II somewhat redundant: Accumulation had to occur under the rules of the regime, regardless of whether a deliberate program existed. Programs to guide reserve-boosting foreign exchange purchases are more relevant when the exchange rate is fluctuating within a band or floating, such as for Programs III through V in the period after the end of 2013.

Looking at standard reserve adequacy metrics, including those cited by the central bank, Costa Rica’s net international reserve position has been strong since 2014; it comfortably exceeds commonly used thresholds, and it is well within the central bank’s target adequacy range of 11–15 percent of GDP (Figure 10.11). This may, in part, explain why reserve accumulation programs have been modest in size and why their implementation has not always been completed during the past few years.

Figure 10.11.Standard Reserve Adequacy Metrics, 2014–17

(Percent)

Sources: Central Bank of Costa Rica; and the IMF.

Note: For the ratio to GDP, there is no conventional threshold; for all other metrics, the standard threshold is 100.

Figure 10.12 shows central bank intervention for reserve accumulation purchases (by definition, they are all foreign exchange purchases, shown as positive bars), as well as the evolution of the weighted-average MONEX exchange rate. As the figure suggests, these reserve accumulation programs have been executed during times when the exchange rate was stable or amid appreciation pressures, to help stabilize the market. The mean value of these interventions was $8.9 million per day from late December 2013 through February 2018, which is sizable compared with a mean total turnover in MONEX of $15.6 million per day during the period. Also, the intervention amounts for this motive have tended to be more uniform (a lower standard deviation) than for the other two motives.

Figure 10.12.Central Bank of Costa Rica’s Foreign Exchange Intervention to Accumulate Reserves, 2013–18

(Thousands of US dollars, left scale; colones per US dollar, right scale)

Source: Central Bank of Costa Rica data.

Note: The figure presents data collected daily. MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

Intervention to Fulfill the Foreign Exchange Needs of the Nonfinancial Public Sector

In Costa Rica, the nonfinancial public sector is a net demander of foreign exchange. While some public sector entities supply foreign currency, others are among the country’s largest importers, including the state-owned hydrocarbons and electricity and telecommunications enterprises (RECOPE and ICE, respectively). Until 2014, large nonfinancial public sector agencies participated in MONEX to meet their foreign exchange needs directly. Despite central bank attempts to schedule and regulate their purchases, occasionally these entities placed disproportionately large demands relative to MONEX’s thin trading volumes, which have potentially destabilizing effects for price formation.

To stop this source of volatility, in June 2014, the central bank decided to supply the nonfinancial public sector foreign exchange needs directly, using its net international reserves. It then resorts to MONEX, at its own pace and discretion, to replenish the reserve position. Or it purchases public sector agencies’ foreign exchange surpluses, when they occur, and might, on occasion, sell them in MONEX. The timing of these interventions is guided by market conditions, including seasonal factors, which the interventions help offset.

Figure 10.13 shows the central bank’s participation in MONEX for this motive. It is, by far, the most frequent type of intervention: 534 daily events of a total of 832 during the period of analysis. The figure shows that most cases were purchases; only 11 were sales. The mean value for central bank purchase interventions for this motive was $7.8 million per day from late December 2013 through February 2018, about half the mean daily turnover in MONEX.

As the figure illustrates, foreign exchange purchases to meet the needs of the nonfinancial public sector have taken place at times when the domestic currency’s exchange rate is either stable or appreciating. However, amid depreciation pressures in the market, this type of intervention is postponed. The absence of participation by the central bank may thus be interpreted as an intervention by omission to stem the depreciation pressures. Lizano (2018) makes this case. On the other hand, the central bank’s sporadic sales in MONEX because of transactions with the nonfinancial public sector have taken place when the exchange rate was depreciating.

Figure 10.13.Central Bank of Costa Rica’s Foreign Exchange Interventions to Fulfill Nonfinancial Public Sector Needs, 2013–18

(Thousands of US dollars, left scale; colones per US dollar, right scale)

Source: Central Bank of Costa Rica data.

Note: The figure presents data collected daily. MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

Therefore, in practice, discretion on the timing and magnitude of MONEX transactions to compensate for the net international reserve effects of fulfilling the nonfinancial public sector’s foreign exchange needs allows the central bank to wait until market conditions are favorable to conduct these operations: purchases when there are foreign exchange surpluses and appreciation pressures and sales when the opposite occurs. The intervention amounts for these purposes are also typically smaller and more homogenous in size, to avoid disrupting the market.12 Thus, participation for this motive seems to have stabilization purposes, as in the case of interventions for reserve accumulation. The implication is that the central bank is allowing its net international reserve position to become more volatile in the short run to accommodate its implicit foreign exchange market stabilization decisions.

Intervention for Stabilization and Intervention Rules

The central bank’s third motive for intervention is to prevent extreme or “violent” fluctuations in the exchange rate. On the basis of its own communications (see, for instance, Central Bank of Costa Rica (2017a and 2017b), the central bank has two main concerns regarding the effect of sharp exchange rate fluctuations:

  • Financial stability: While the degree of dollarization of the financial system has moderated over the past decade, it remains high (Figure 10.14). Furthermore, it is estimated that about 71 percent of foreign exchange denominated credit to the private sector has been granted to households or corporations that do not earn foreign exchange (Central Bank of Costa Rica 2017c). These balance sheet mismatches mean that a sharp depreciation could lead to the materialization of large foreign exchange–induced credit risks in the financial system.
  • Price stability: Although the exchange rate pass-through coefficient has declined, as argued earlier, it is still high, and the Central Bank of Costa Rica has expressed concern that excessive foreign exchange volatility might affect exchange rate expectations, and thereby, expected and actual inflation.

Figure 10.14.Central Bank of Costa Rica’s Foreign Exchange Stabilization Interventions, 2013–18

(Colones per US dollar)

Source: Central Bank of Costa Rica data.

Note: The figure presents data collected daily. MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

These concerns create, and help explain, a clear antidepreciation bias in the central bank’s stabilization interventions. Moreover, because interventions for the other two motives are mostly purchases, they cannot directly help to stem a depreciation (except, implicitly, by postponing the purchases until the pressures have ended or eased). Stabilization interventions as sales are thus the central bank’s only direct, active intervention tool to contain strong depreciation pressures and the stability risks associated with them.

Figure 10.14 confirms this. It shows the interventions in MONEX for stabilization purposes. As the figure illustrates, the majority of such interventions have been foreign exchange sales, and some very large (with a mean of $9.9 million per day on days when they occur). This includes sales of $71.7 million on June 21, 2017, the largest daily intervention of any type in the period of analysis, and more than four times larger in size than the average total daily foreign exchange turnover in MONEX. By contrast, only six isolated foreign exchange purchases have been made for stabilization, most amid strong appreciation trends in 2014.

To guide the implementation of foreign exchange interventions for stabilization, the central bank board has defined two intervention rules (Central Bank of Costa Rica 2015), whose specific parameters are kept confidential, as explained in the next section:

  • An intraday rule: Adopted early in the crawling band regime period, this rule is triggered by excessive or “violent” exchange rate volatility within the same negotiation day in MONEX.13 The rule defines excessive fluctuation, with the volatility on the previous several days as a benchmark (with the number of days decided by the Markets Committee). The purpose of the resulting intervention is precisely to mitigate such sharp fluctuations, but not to affect the trend of the exchange rate. If the rule is triggered, the central bank assesses first whether the large intraday exchange rate movements can be explained by fundamentals (for example, changes in interest rate differentials or in the terms of trade) or are consistent with seasonal events. If not, the Markets Committee might decide to intervene, using its discretion to determine the details of the intervention within the broad contours defined by the central bank board. There is an assigned budget for this type of intervention, with daily and annual limits. Once this budget is exhausted, price adjustments may be allowed.
  • An interday rule: This rule was introduced in March 2014, following several weeks of sustained, sharp depreciation pressures (Box 10.1). This episode made it clear that a rule based on intraday volatility might not be sufficient to stem persistent pressures. This rule aims to set the criteria for intervention to prevent sustained “deviations of the exchange rate from the level consistent with medium- and long-term economic fundamentals” (Central Bank of Costa Rica 2014c, 2015). Instead of focusing on the intraday volatility relative to the last few days’ volatility, the interday rule is based on the cumulative deviations (over several days or weeks) of the exchange rate from the long-term path predicted by its fundamentals. This is determined based on an equilibrium real effective exchange rate framework, with a confidence band. If the exchange rate is deemed misaligned (that is, outside the confidence bounds), the rule may be triggered. This rule would entail interventions that involve much larger amounts, and that last for longer (several days or weeks) than those triggered by the intraday rule.

The interday rule was triggered for the first time in May 2017, after another episode of sustained sharp depreciation (Box 10.1). Between May 16 and July 7, 2017, the central bank deployed $477 million in foreign exchange sales (around 7 percent of the stock of international reserves at the beginning of the episode) to withstand the pressures. These started abating in July 2017.

In its official communications the central bank indicates that, under its managed floating regime, it does not have a commitment to a particular level or range for the exchange rate, the interday rule does suggest a concern with keeping the nominal exchange rate at a level consistent with real effective exchange rate fundamentals (allowing for a confidence band around it).14 That level, of course, will vary continuously, but the concern remains about a (moving) target level or, considering the confidence interval, a target range. Therefore, the articulation of the intraday and interday rules in official documents makes it possible to conclude that the central bank’s intervention for stabilization purposes attempts to affect both the level (insofar as it is deemed inconsistent with fundamentals) and the volatility (insofar as it is deemed “violent” or excessive) of the exchange rate.

The Stabilizing Role of Official Foreign Exchange Intervention

It could be argued that, in analyzing intervention as an instrument of monetary and exchange rate policy, the interventions that matter are those conducted for exchange rate stabilization, not those aimed at accumulating reserves or satisfying the needs of the nonfinancial public sector, which are motivations that lie far from exchange rate policy. However, as argued earlier, interventions for all three motives, in practice, share a stabilizing role in Costa Rica. First, the timing and magnitude of interventions to satisfy the central bank’s or the nonfinancial public sector’s needs are discretionary for the central bank and chosen according to market circumstances. They are therefore unlikely to be used in a way that might contribute to a destabilizing trend. On the contrary, the specifics of these interventions, or even the lack of intervention, seem motivated by a desire to reduce volatility or unwanted fluctuations in the exchange rate. The central bank has publicly recognized this (see, for example, Central Bank of Costa Rica 2014b). Second, all three types of intervention take place in an (ex ante) anonymous, unannounced fashion on MONEX (the intervention information is published only at the end of the trading session). Therefore, from the perspective of market participants, they are observationally equivalent.

Foreign Exchange Intervention: Transparency and Communication

Official foreign exchange intervention in Costa Rica is subject to a mixed regime in terms of transparency. On the one hand, the central bank board decided that the intervention rules must be kept strictly confidential. Only a handful of high-level officials have access to these rules, and any disclosure is considered a serious fault subject to administrative and criminal liabilities. The policy of confidentiality has been justified on the argument that the small size of the domestic foreign exchange market and its high degree of concentration entail a risk that some large actors may use informational advantages about the rules to obtain excessive profits, to the detriment of smaller agents or the central bank’s international reserves. The confidentiality of the intervention rules was upheld by the Constitutional Chamber of the Supreme Court of Justice.15 Moreover, contrary to other central banks (see chapter 3 of this book), the central bank does not disclose the motives, timing, or magnitude of its interventions ex ante, not even for interventions with the purpose of accumulating reserves or meeting the needs of the nonfinancial public sector.

On the side of transparency, however, the central bank publishes very detailed data on its interventions ex post, at the end of each MONEX session. Therefore, the public has full access to data on official intervention patterns up to the day before. Moreover, the central bank has complemented its intervention policies with press releases to explain the purposes of its participation in MONEX and to try to influence expectations, particularly during periods of market stress. Box 10.1 summarizes the experience with three episodes, which suggest that communications may have helped stabilize the exchange rate.

In this sense, communications themselves may be an additional intervention tool (just like, as argued earlier, the omission to intervene to meet the needs of the central bank of Costa Rica, or the nonfinancial public sector may itself be a form of intervention).

Intervention Effectiveness: Empirical Analysis

This section analyzes the effectiveness of foreign exchange intervention in the period from December 2013 through February 2018, when the exchange rate was in a de facto float, and all official intervention was therefore discretionary (rather than mandated by the need to defend the limits of the band). The section begins with a discussion of the framework for assessing effectiveness, continues with a brief description of the data used, and then presents the methodology and results on the effectiveness of intervention on the variation rates and volatility of the exchange rate, for two different empirical approaches: an event analysis and an econometric analysis.

Empirically Assessing the Effectiveness of Intervention in Costa Rica: A Framework

In theory, sterilized foreign exchange intervention may affect the exchange rate through portfolio balance (relative supplies of domestic and foreign currency assets) and signaling (market expectations) channels.16 However, empirical analyses of the effectiveness of intervention face inherent endogeneity problems: interventions themselves are motivated by exchange rate developments. Failing to account for this endogeneity when central banks lean against the wind biases the coefficients toward zero (Kearns and Rigobón 2005; and Chapter 4 of this book). One approach to address this issue is to use high-frequency data to help isolate the intervention variable from contemporaneous exchange rate developments (Payne and Vitale 2003; Benes and others 2013). It also allows the analysis to abstract from other factors, potentially driving the exchange rate over longer-term horizons. This chapter adopts this approach by exploiting a second-by-second dataset for MONEX trades, and it applies the two empirical methodologies referred to earlier.

Box 10.1.Central Bank Communications on Foreign Exchange Intervention Amid Market Stress: Three Episodes

January–March 2014:

Following days of sustained depreciation pressures, the central bank issued a press release in January 2014 (Central Bank of Costa Rica 2014a) explaining that it had decided to intervene the day before to stabilize the market in light of what it considered “excessive” volatility in the exchange rate market, following announcements by the US Federal Reserve on a future tightening of policy. The central bank issued another release in February 2014, explaining in detail the logic and mechanics of its intraday intervention rule (Central Bank of Costa Rica 2014b). Pressures continued: The average exchange rate on MONEX moved from C501.8 per US dollar on January 9 to C569.1 on March 11 (a 13.4 percent depreciation). On March 12, it announced and justified the introduction of a new, interday rule (Central Bank of Costa Rica 2014c). Markets reacted immediately: The exchange rate strengthened by over C25 (4.5 percent) from March 12 to March 13. The central bank continued to issue releases that month and in April, and although pressures resumed somewhat, they remained contained.

April–May 2017:

Following sharp depreciation pressures, the central bank issued a press release on May 22, 2017 (Central Bank of Costa Rica 2017a) explaining its interpretation of foreign exchange market developments and announced that it had resorted to intraday interventions to contain volatility. It also explained that, in its view, the real exchange rate was not misaligned. Foreign exchange interventions were complemented by a significant hike in the policy rate. However, markets remained unsettled, with the average MONEX exchange rate increasing from C573.8 per US dollar to C595.3 (3.8 percent) in just eight sessions between May 14 and May 24, and by over C9.0 (1.5 percent) in the two days up to May 24. As a result, on May 25 the central bank issued a new press release (Central Bank of Costa Rica 2017b), stating that the recent evolution of the exchange rate was not consistent with fundamentals, and it announced that it had decided to trigger its interday intervention rule for the first time and to dedicate a budget of up to $1 billion to stem the pressures. This reassured the markets, and the exchange rate started strengthening that same day.

Figure 10.1.1.MONEX Weighted-Average Exchange Rate, 2013–18

(Colones per US dollar)

Source: Central Bank of Costa Rica data.

Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

January–February 2018:

Between January 29 and February 7, 2018, the exchange rate depreciated by about 1.1 percent. The central bank issued a press release on February 7 (Central Bank of Costa Rica 2018) indicating that the recent movements were not consistent with fundamentals, and it warned that similar abrupt depreciations in the past (including those in 2014 and 2017) were followed by equally sharp appreciations. Market pressures eased starting the next day.

Effectiveness must be assessed relative to the central bank’s purported objectives for intervention. As described earlier, the central bank has three distinct motives for its foreign exchange intervention. In practice, however, all three are oriented toward reducing (or at least to avoid exacerbating) excessive fluctuations in the exchange rate. In line with this, the analysis in this section focuses on the effect of intervention, on the rate of variation, and on the volatility of the exchange rate. These two—growth rates and volatility—are alternative, but not equivalent, measures of fluctuations. While the rate of variation is signed, volatility measures are not. Using variation rates allows us to assess whether interventions affect the exchange rate in a specific direction—purchases would be expected to have a positive (depreciation) effect on variation rates, and sales a negative (appreciation) effect. By contrast, the effect of any intervention, whether a purchase or a sale, on volatility would be expected to be negative. Variation rates also allow us to assess how effective interventions are in containing deviations from long-run fundamentals, which is, as explained earlier, a further motive for central bank interventions for stabilization. Moreover, variation rates can capture a deceleration of an appreciation or depreciation process as a result of an intervention, whereas using exchange rate levels as a dependent variable may lead to finding a “wrong” sign in these cases (see Chapter 4 of this book).

There has only been one published empirical study of the effectiveness of foreign exchange intervention in Costa Rica: Espinoza and Valerio (2016). Using a generalized autoregressive conditional heteroskedasticity (GARCH) analysis on daily data for the period January 2014 through April 2015, they find that interventions have a positive, but very small, effect on the level of the exchange rate and a negative effect on (conditional) volatility. In addition, a chart in a study by the Organisation for Economic Co-operation and Development (OECD) shows that, in the first year of the managed float period, interventions reduced the volatility of the exchange rate in the three-hour period after the intervention, compared with the three hours before the intervention (OECD 2016).

The empirical analysis in this chapter complements and updates Espinoza and Valerio (2016) and OECD (2016) in a number of dimensions: (1) it covers a much longer sample period (through early 2018); (2) it includes a broader analysis of the effectiveness of intervention, which separates intervention events and amounts, and aggregates intervention from purchases and sales; (3) it uses different empirical approaches; and (4) it adds a number of robustness tests. Moreover, contrary to Espinoza and Valerio (2016), who use daily data, the empirical analysis in this chapter is based on high-frequency data, which can deal with simultaneity problems.

Data and Key Variables

The empirical analysis focuses on a high-frequency dataset of MONEX trades, for the period December 13, 2013, through February 28, 2017. For each trade, it includes information on amount, price (exchange rate), and parties involved.17

The original dataset contains second-by-second, transaction-level data, with a total of 200,000 observations. To make the time frame more meaningful for empirically assessing the effectiveness of intervention, the data are aggregated into 15-minute intervals, following Payne and Vitale (2003). This yields a total of 28,940 observations.

The main variables used for the analysis are two measures of the exchange rate (its rate of variation and its volatility) and several indicators of intervention, including intervention dummies and amounts, whether aggregate or broken down into purchases and sales (see Table 10.4).

Table 10.4.Definition of Variables
VariableDefinition
rRate of variation (or rate of return, as it is frequently referred to in the finance literature) of the colones/US dollar exchange rate over 15 minutes.
r-anAnnualized r, expressed in basis points.1 This is used for regression analysis to make the coefficients easier to interpret.
r-sqSquared returns, estimated as rsq=ln(r2+1), as a first measure of volatility.
r-devStandard deviation of the exchange rate level, defined over alternative windows (15 minutes, 60 minutes), as another measure of volatility.
ISigned intervention indicator dummy, which, for any 15-minute interval, takes the value of 1 if the central bank purchased US dollars in that interval, –1 if it sold US dollars, and 0 otherwise.2
POfficial foreign exchange purchase indicator, which, for any 15-minute interval, takes the value of 1 if the central bank purchased US dollars in that interval, and 0 otherwise.
SOfficial foreign exchange sale indicator, which, for any 15-minute interval, takes the value of 1 if the central bank sold US dollars in that interval, and 0 otherwise.
XCentral bank foreign exchange intervention amounts, in millions of US dollars, where purchases take a positive sign and sales a negative sign.
XPCentral bank foreign exchange purchases, in millions of US dollars.
XSCentral bank foreign exchange sales, in millions of US dollars.
Source: Authors.Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

Annualization takes into account the several changes in MONEX’s opening times during the sample period.

In the sample period, there are only four 15-minute intervals during which the central bank both sold and purchased dollars. In these cases, the indicator takes the value of zero.

Source: Authors.Note: MONEX (Mercado de Monedas Extranjeras) is an organized electronic foreign exchange market provided by the Central Bank of Costa Rica.

Annualization takes into account the several changes in MONEX’s opening times during the sample period.

In the sample period, there are only four 15-minute intervals during which the central bank both sold and purchased dollars. In these cases, the indicator takes the value of zero.

Methodology

To assess empirically the effectiveness of foreign exchange interventions, this chapter uses two different methods: an event analysis and an econometric approach. This subsection discusses the two approaches in turn; the next one presents the key results.

Event Analysis

The event analysis compares the average variation rate (or volatility) of the exchange rate before and after a central bank foreign exchange intervention event, to see whether intervention affected the exchange rate in the expected (desired) direction. The approach in this chapter follows broadly the methodology laid out in the seminal study by Fatum and Hutchison (2003).18 The advantage of this approach is that it does not rely on a theoretical model of exchange rate determination. There are three critical steps in an event analysis:

  • 1. Identification of the event: To identify intervention events, this chapter considers only interventions that are isolated; that is, where no other interventions take place in the reference windows before or after the intervention. This allows more precise assessment of the effects of the specific intervention event on the exchange rate path following the intervention, and it can treat the intervention as unanticipated. The intervention event itself is a 1-minute window.
  • 2. Definition of the length of the pre- and post-intervention windows: These windows should be long enough to identify relatively persistent, more sustained effects, but also short enough to isolate the analysis from the influence of third factors on the exchange rate. For robustness, we use, alternatively, 60-, 120-, and 180-minute windows.
  • 3. Classification of events as effective (or not): To gauge the success of an intervention, we use different criteria depending on whether the impact variable is the rate of variation or the volatility of the exchange rate.For the variation rate, an intervention at time t (for 1-minute intervals) is considered effective when at least one of the following three conditions is met:
    • Moderation: The rate of appreciation (depreciation) is slower after an official foreign exchange purchase (sale). Thus, for example, if the exchange rate is appreciating (note that an appreciation is a negative rate of change, and therefore a moderation implies that the rate of change increases):
    • Reversal (change of trend): intervention reverses the direction of exchange rate variation. Thus, if the exchange rate was appreciating (depreciating), it depreciates (appreciates) after a purchase (sale). For example, in the case of an appreciation before a purchase intervention:
    • Intensification: Depreciation (appreciation) increases after an official foreign exchange purchase (sale). For example, if the exchange rate is depreciating, and there is a foreign exchange purchase:

If none of these three conditions is met (for example, the exchange rate is appreciating and a purchase intervention leads to a faster appreciation, or it is depreciating and a sale intervention leads to a faster depreciation), then the intervention is deemed to be ineffective.

For the analysis of volatility, there is one single criterion: whether it declines or increases after the intervention, regardless of whether it is a sale or a purchase. An intervention is deemed effective if the volatility in the period after the intervention is lower than in the period before. Two measures of volatility are used: the standard deviation of a 15-minute rolling window of the level of the exchange rate (by minute), which gives a sense of the volatility in colones; and the squared returns of 15-minute time frames, estimated as (ln(R_15^2 + 1))^1/2.19

Regression Analysis

To complement the event analysis, a regression approach is also undertaken. As discussed earlier, econometric assessments of the effectiveness of the exchange rate are plagued by endogeneity issues, which bias standard ordinary least squares (OLS) coefficients. This chapter uses a high-frequency dataset to help address this problem.

Following Payne and Vitale (2003), the effectiveness of interventions is studied through a linear regression of the 15-minute percentage variation of the C/US$ rate (rt), or its volatility (r-devt) on leads and lags of alternative intervention indicators (as well as lags for the dependent variable):

  • Intervention dummy indicators, whether aggregate (It) or broken down into sales (St) and purchases (Pt): For example, the specification for rt and It is as follows:
  • Intervention amounts: The size of intervention may convey a clearer signal about the authorities’ intention (although it is likely also endogenously associated with the strength of the exchange rate trend it attempts to counter, which may bias the results). As for the intervention dummies, the amount indicators are used first in aggregate (Xt), where purchases take a positive sign and sales a negative one, or separating between these two (XPt and XSt, respectively). The specification takes the following form:

Eight leads and lags of the intervention variable are considered to capture the effects of intervention in the 2-hour period before and after any intervention event. The idea of introducing leads is to test whether interventions are anticipated by the market. Different lead and lag lengths were used alternatively, but this did not affect the results. The number of lags of the dependent variable was chosen using partial autocorrelation tests and Akaike and Schwartz information criteria. The regression coefficients are estimated using OLS, with Newey-West standard errors.

Results

This section summarizes the main results for the two methods: event analysis and regressions. As will be shown, the event analysis suggests that foreign exchange interventions tend to be effective. The results for regression analysis are consistent with this, if somewhat less conclusive.

Results: Event Analysis

The event analysis suggests that central bank interventions have been effective in moving the rate of variation of the exchange rate in the desired direction. As Table 10.5 shows, interventions are highly successful, with the success rate increasing at 120-minute windows and then decreasing. Purchases are more effective than sales at shorter intervals, but sales show a more lasting effect, with a success rate of 72 percent at 180-minute time frames.

Table 10.5.Effects of Intervention on the Rate of Variation of the Exchange Rate, 2013–18
Time Frame
60 minutes120 minutes180 minutes
TotalSuccessful%TotalSuccessful%TotalSuccessful%
Purchases57468125228816956
Moderation of appreciation10181400
Reversion of appreciation26461248744
Intensification of depreciation1018936213
Sales433172242083181372
Moderation of depreciation1023417317
Reversion of appreciation19441250844
Intensification of depreciation25417211
Source: Authors’ calculations.
Source: Authors’ calculations.

Table 10.6 shows the results for the success rate of intervention in reducing volatility. It indicates that central bank interventions reduce the volatility of the exchange rate in less than half of the cases (except in purchases in longer time frames).

Table 10.6.Success Rate of Intervention in Moderating Exchange Rate Volatility, 2013–18
Time Frame
60 minutes120 minutes180 minutes
TotalSuccessful%TotalSuccessful%TotalSuccessful%
Total1004934
Purchases572516
Sales432418
Standard deviation
Purchases22391040850
Sales1740938633
Returns
Purchases25441144956
Sales18421042739
Source: Authors’ calculations.
Source: Authors’ calculations.

However, a different conclusion is reached if the effect on the average volatility of the exchange rate is assessed. Figure 10.15 shows that mean volatility falls after an intervention, for both purchases and sales and regardless of the time frame considered. This implies that, although only a moderate proportion of interventions reduce the volatility of the exchange rate, they reduce it by a large margin.20

Figure 10.15.The Effect of Intervention on Mean Volatility of the Exchange Rate

(Percent)

Source: Authors’ calculations.

Note: Volatility is measured by squared returns.

Overall, event analysis suggests that central bank foreign exchange interventions are fairly effective in moving the variation of the exchange rate in the desired direction and in reducing its volatility.

Results: Regression Analysis

This section provides the results for the empirical estimation of equations (10.1) and (10.2), identified in the next six tables as model I and model II, respectively. Results are shown in aggregate (panel A) and breaking down interventions into sales and purchases (panel B). Only the coefficients for the relevant variables that turned out statistically significant are reported.21

The results for the rate of variation of the exchange rate (rt) as dependent variable are shown in Tables 10.7 and 10.8. Table 10.7 presents the results for the intervention indicator (model I). It suggests that interventions generally have the desired (positive) effect on the variation rate, with a cumulative effect of 12.7 basis points for the aggregate intervention dummy (panel A). However, while purchases do have the expected positive sign, sales also have a positive sign (panel B). This suggests that sales tend to exacerbate, rather than mitigate, depreciation pressures on the exchange rate (although it may also be that the econometric methodology is unable to fully control for strong depreciation trends). It is also interesting to note that, in this and other specifications shown later, some leads tend to have statistically significant coefficients, which indicate that the effects of intervention are anticipated by the market.

Table 10.7.Effects on the Rate of Variation of the Exchange Rate: Model I
Panel APanel B
VariableCoefficientVariableCoefficient
I (+3)16.98**P (+1)27.70***
I (0)24.82***P (0)38.33***
I (–7)-14.26**P (–7)-18.59***
I (–8)-14.69**P (–8)-14.70**
S (+2)54.58**
S (+1)47.59**
Adjusted R20.015Adjusted R20.018
Akaike information criterion14.606Akaike information criterion14.603
Cumulative effect12.67Cumulative effect (P)30.79
Cumulative effect (S)96.05
Source: Authors’ estimation.Note: The dependent variable is the rate of variation of the exchange rate (rt). The total number of observations after adjustment is 25,646.**p < .05; ***p < .01.
Source: Authors’ estimation.Note: The dependent variable is the rate of variation of the exchange rate (rt). The total number of observations after adjustment is 25,646.**p < .05; ***p < .01.
Table 10.8.Effects on Rate of Variation of the Exchange Rate: Model II
Panel APanel B
VariableCoefficientVariableCoefficient
X (+8)-4.82*XP (+3)12.53***
X (+2)-5.6*XP (+2)9.23**
XP (0)26.48***
XP (-2-14.98**
XP (-3)-19.89***
XS (+7)9.09**
XS (0)12.39***
XS (-3)-10.35**
XS (-7)-13.47***
Adjusted R20.012Adjusted R20.015
Akaike information criterion14.61Akaike information criterion14.61
Cumulative effect (P)12.56
Cumulative effect (S)-2.2
Source: Authors’ estimation.Note: The dependent variable is the rate of variation of the exchange rate (rt). The total number of observations after adjustment is 25,646.*p < .1; **p < .05; ***p < .01.
Source: Authors’ estimation.Note: The dependent variable is the rate of variation of the exchange rate (rt). The total number of observations after adjustment is 25,646.*p < .1; **p < .05; ***p < .01.

Table 10.8 presents the results of the intervention amounts (model II). The coefficients show the effect of a $1 million intervention over the 15-minute variation rate of the exchange rate, annualized and measured in basis points.22 For the aggregate intervention amount, the effects are negative but barely statistically significant at the 10 percent confidence level. For the disaggregated intervention amounts, on the other hand, the effects are statistically significant and in the expected direction in cumulative terms, though the magnitude is asymmetric for purchases and sales. Thus, a $1 million purchase by the central bank leads to a depreciation of about 13 basis points, while a sale of an equivalent amount leads to an appreciation of about 2 basis points.

Table 10.10.Effects on Exchange Rate Volatility: Model II
Panel APanel B
VariableCoefficientVariableCoefficient
X (+7)0.004**XP (+7)0.005**
XP (+4)-0.005**
XP (0)-0.010***
XP (-1)-0.014***
XP (-2)0.006**
XP (-3)0.008***
XP (-4)0.007***
Adjusted R20.270Adjusted R20.272
Akaike information criterion-0.438Akaike information criterion-0.441
Cumulative effect0.011Cumulative effect (P)-0.0062
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–std). The total number of observations after adjustment is 28,937.**p < .05; ***p < .01.
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–std). The total number of observations after adjustment is 28,937.**p < .05; ***p < .01.

The analysis now reviews the effect of intervention on exchange rate volatility, using the standard deviation (Tables 10.9 and 10.10) and then the squared returns (Tables 10.11 and 10.12) as alternative measures of volatility.

Table 10.9Effects on Exchange Rate Volatility: Model I
Panel APanel B
VariableCoefficientVariableCoefficient
I (+7)0.008**P (+6)0.010**
I (+6)0.009**P (+4)-0.010**
I (+4)-0.010**P (+3)-0.012***
I (+3)-0.010***P (+2)-0.006**
I (+2)-0.006**P (+1)-0.009***
I (0)-0.012***P (0)-0.012***
I (–1)-0.023***P (–1)-0.036***
I (–3)0.010***P (–3)0.015***
I (–4)0.012**P (–4)0.013***
I (–5)0.006**P (–5)0.007**
I (–8)0.006**P (–7)0.006**
P (–8)0.009***
S (+8)0.043**
S(+1)-0.023***
S(-1)-0.042***
S(-7)0.014**
Adjusted R20.275Adjusted R20.281
Akaike information criterion-0.45Akaike information criterion-0.45
Cumulative effect-0.025Cumulative effect (P)-0.0822
Cumulative effect (S)-0.0255
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–std). The total number of observations after adjustment is 28,937.**p < .05; ***p < .01.
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–std). The total number of observations after adjustment is 28,937.**p < .05; ***p < .01.
Table 10.11.Effects on Exchange Rate Volatility: Model I
Panel APanel B
VariableCoefficientVariableCoefficient
I (+2)-13.54*P (+2)-13.97 ***
I (+5)12.03**P (–5)12.45**
P (–6)-10.35**
P (–7)12.12**
P (–8)14.93***
Adjusted R20.333Adjusted R20.335
Akaike information criterion13.962Akaike information criterion13.962
Cumulative effect-5.43Cumulative effect (P)55.45
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–sq). The total number of observations after adjustment is 20,160.**p < .05; ***p < .01.
Source: Authors’ estimation.Note: The dependent variable is the standard deviation of the exchange rate (r–sq). The total number of observations after adjustment is 20,160.**p < .05; ***p < .01.
Table 10.12.Effects on Exchange Rate Volatility: Model II
Panel APanel B
VariableCoefficientVariableCoefficient
X (+8)-7.18**XP (+8)-9.01**
X (+3)-9.28**XP (+2)-10.36**
X (0)-13.17**XP (0)-10.23**
XS (-7)-10.43**
XS (0)-18.64**
Adjusted R20.330Adjusted R20.330
Akaike information criterion13.969Akaike information criterion13.970
Cumulative effect-13.68Cumulative effect (P)-37.85
Cumulative effect (S)-120.38
Source: Authors’ estimation.Note: The dependent variable is the squared returns of the exchange rate (r–sq). The total number of observations after adjustment is 16,886.**p < .05.
Source: Authors’ estimation.Note: The dependent variable is the squared returns of the exchange rate (r–sq). The total number of observations after adjustment is 16,886.**p < .05.

We first review the results of the standard deviation; Table 10.9 presents the results for the intervention indicator. The results are clear: Intervention reduces the volatility of the exchange rate, and both purchases and sales contribute in this direction, but purchases have a much stronger effect.

If intervention amounts are used, however, the evidence is less conclusive (Table 10.10). The standard deviation of the exchange rate seems to increase with the amount of intervention, although the effect seems small (about 1 céntimo per $1.0 million intervention). The amount of sales does not have a statistically significant effect on volatility, but amounts purchased do have a small negative effect.23

The results are somewhat different if the squared returns of the exchange rate are used as a measure of volatility (Tables 10.11 and 10.12). In this case, interventions and intervention amounts do have a negative and statistically significant effect on the measured volatility, but sales and purchases only seem to reduce volatility when amounts are considered.

Overall, the evidence from regression analysis suggests that central bank foreign exchange intervention is effective in both pushing the exchange rate in the desired direction and reducing its volatility. This is particularly true of US dollar purchase interventions. The evidence on the effectiveness of sales, on the other hand, is more mixed, with some specifications suggesting that selling does not help appreciate the exchange rate nor mitigate its volatility.

Conclusions

Data from Costa Rica clearly illustrates that learning to float is not easy, particularly amid significant structural constraints. Although it transitioned to a managed floating regime (effectively since late 2013, and de jure since early 2015), the central bank has continued to engage in active sterilized exchange rate interventions—so actively that the de facto exchange rate regime has been classified by the IMF as “crawl-like” or “stabilized” over the past few years.

This chapter has argued that two factors contribute to such fear of floating by the central bank:

  • The likelihood of sharp fluctuations, in the context of a thin and relatively concentrated spot market (MONEX), where a few powerful participants can wield a disproportionately high influence on price formation, and therefore have an incentive to engage in self-fulfilling speculation. With still very underdeveloped derivatives markets, MONEX is the market where all official foreign exchange intervention takes place.
  • The likelihood of potentially deleterious effects of such fluctuations on price expectations and financial stability, against the background of a (still) considerable pass-through coefficient and a high degree of currency risk in household and corporate balance sheets, with limited availability of hedging instruments.

It is therefore understandable that the main rationale for official intervention in Costa Rica be to contain “violent” fluctuations in the exchange rate. As this chapter has shown, even interventions in MONEX, which are ostensibly intended for reserve accumulation or to meet the needs of the nonfinancial public sector, are, in practice, conducted in such a way as to help mitigate excessive volatility or at least to avoid exacerbating it. Moreover, while all intervention information is published ex post, interventions are not announced ex ante, and central bank rules for stabilization interventions are kept confidential. Official communications, particularly through press releases, are used to support actual interventions in MONEX, and effectively represent another channel of intervention. The central bank’s justification of this mixed approach to transparency is that advance information to markets may be used by large players for speculative gain, with potentially destabilizing effects.

Has the Costa Rican central bank been successful in its stabilization attempts? The answer is not straightforward, given that the formal empirical analyses are fraught with endogeneity and identification problems. This chapter has sought to overcome the standard endogeneity and identification problems in empirical analyses by deploying two different approaches on a high-frequency dataset. It finds that official foreign exchange intervention has been broadly effective in containing exchange rate fluctuations, whether measured by the rate of variation of the exchange rate, or its volatility. The ultimate evidence in favor of the effectiveness of intervention is in the observed path of the exchange rate itself: a simple chart shows that it has been relatively stable in the floating period—hence the soft peg classification by the IMF.

Going forward, the central bank could consider preannouncing interventions to meet the needs of the nonfinancial public sector. This may help to increase the potency and signaling effects of foreign exchange interventions for stabilization purposes. The central bank could also consider allowing greater exchange rate flexibility. As extensively pointed out (IMF 2017; Lizano 2018), high financial dollarization is endogenous to the exchange rate regime, and greater de facto flexibility would stimulate agents to internalize exchange rate risks more fully, thereby reducing financial vulnerabilities. In addition, greater exchange rate flexibility would help buttress the credibility of the monetary policy framework, as confidence would be strengthened in the primacy of the price stability objective over exchange rate stability. This, in turn, would contribute to a continued decline in the pass-through coefficient from exchange rate shocks to inflation. As in swimming, the best way to learn to float may be to let go gradually.

References

    AdlerGustavo andCamiloTovar. 2011.“Foreign Exchange Market Intervention: How Good a Defense Against Appreciation Winds?IMF Working Paper 11/165International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    AlfaroUreñaAlonsoBettySánchez Wong andEdwinTenorio Chaves. 2016. “Migración de la banda cambiaria hacia un régimen de flotación administrada.” BCCR Working Paper 03/2016Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    BarqueroJoséPablo andDavidMora. 2014. “El efecto traspaso de la tasa de interés de los instrumentos del Banco Central en Costa Rica hacia las tasas del interés del sistema financie-ro.” BCCR Research Paper DI-01-2014Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    BarqueroJoséPablo andAngeloOrane. 2015. “Orden de propagación de cambios en la tasa de política del Banco Central sobre las tasas de interés del sistema financiero en Costa Rica.” BCCR Research Paper DI-03-2015Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    BenesJaromirAndrewBergRafael A.Portillo andDavidVavra. 2013. “Modeling Sterilized Interventions and Balance Sheet Effects of Monetary Policy in a New-Keynesian Framework.” IMF Working Paper 13/11International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    BrenesCarlos andManfredEsquival. 2017. “Asimetrías en el traspaso del tipo de cambio a los precios en Costa Rica durante el período de flexibilidad cambiaria.” BCCR Research Paper DI-06-2017Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2010. “Mercado cambiario en Costa Rica y pasivos con costo en colones del Banco Central de Costa Rica octubre 2006–marzo 2010.” Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2014a. “Comportamiento reciente del mercado cambiario.” Press release Central Bank of Costa Rica San José January 30.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2014b. “La intervención cambiaria del Banco Central por fluctuaciones violentas del tipo de cambio.” Press release Central Bank of Costa Rica San José February 18.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2014c. “El Banco Central amplía sus políticas de intervención cambiaria.” Press release Central Bank of Costa Rica San José March 12.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2015. “Artículo 5 del Acta de la sesión 5677–2015 de la Junta Directiva del Banco Central de Costa Rica, celebrada el 30 de enero del 2015.” Central Bank of Costa Rica San José.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2016. “Programa macroeconómico 2016–2017.” Central Bank of Costa Rica San José.

    Central Bank of Costa Rica. 2017a. “Comportamiento reciente del mercado cambiario.” Central Bank of Costa Rica San José May 22.

    Central Bank of Costa Rica. 2017b. “Banco Central activa intervención cambiaria entre días.” Press release Central Bank of Costa Rica San José May 25.

    • Search Google Scholar
    • Export Citation

    Central Bank of Costa Rica. 2017c. “Revisión del Programa Macroeconómico 2017–2018.” Central Bank of Costa RicaSan José.

    Central Bank of Costa Rica. 2018. “Comportamiento reciente del mercado cambiario.” Press release Central Bank of Costa Rica San José February 7.

    • Search Google Scholar
    • Export Citation

    ContrerasGabrielaAlfredoPistelli andCamilaSáez. 2013. “Efecto de intervenciones cambi-arias recientes en economías emergentes.” Economía Chilena16 (1): 122137.

    • Search Google Scholar
    • Export Citation

    DelgadoFélix. 2000. “La política monetaria en Costa Rica: 50 años del Banco Central de Costa Rica.” Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    EchavarríaJuanJoséLuisFernando Melo andMauricioVillamizar. 2014. “The Impact of Foreign Exchange Intervention in Colombia: An Event Study Approach.” Revista Desarrollo y Sociedad73: 731.

    • Search Google Scholar
    • Export Citation

    EdwardsSebastián. 1990. “Ajuste cambiario y equilibrio macroeconómico en Costa Rica: Lecciones y perspectivas.” In Políticas Económicas en Costa Rica edited by ClaudioGonzález Vega andEdnaCamacho Mejía. San José: Academia de Centroamerica.

    • Search Google Scholar
    • Export Citation

    EspinozaJulioCésar andMarco V.Valerio. 2016. “Efectividad de las intervenciones del Banco Central de Costa Rica en el Mercado de Monedas Extranjeras (MONEX).” Economía y Sociedad21 (49): 127.

    • Search Google Scholar
    • Export Citation

    FatumRasmus andMichael M.Hutchison. 2003. “Is Sterilized Foreign Exchange Intervention Effective After All? An Event Study Approach.” Economic Journal113: 390411.

    • Search Google Scholar
    • Export Citation

    FratzscherMarcel. 2005. “How Successful Are Exchange Rate Communications and Interventions: Evidence from Time-Series and Event-Study Approaches.” European Central Bank Working Paper Series 528European Central BankFrankfurt.

    • Search Google Scholar
    • Export Citation

    IMF(International Monetary Fund) . 2014. “Conditionality in Evolving Monetary Policy Regimes.” IMF Policy PaperInternational Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    IMF (International Monetary Fund). 2015. “Evolving Monetary Policy Frameworks in Low-Income and Other Developing Countries.” International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    IMF (International Monetary Fund). 2017. “Costa Rica: Staff Report for the 2017 Article IV Consultation.” International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    KearnsJonathan andRobertoRigobón. 2005. “Identifying the Efficacy of Central Bank Interventions: Evidence from Australia and Japan.” Journal of International Economics66 (1): 3148.

    • Search Google Scholar
    • Export Citation

    LaurensBernardJ.KellyEckholdDarrylKingNilsMaehleAbdulNaseer andAlainDurré. 2015. “The Journey to Inflation Targeting: Easier Said than Done: The Case for Transitional Arrangements along the Road.” IMF Working Paper 15/136International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    Laverde MolinaBernal andEvelynMuñoz Salas. 2006. “Modelo monetario de inflación apli-cado a la capitalización del Banco Central de Costa Rica.” Technical Note DIE-06-2006-NTCentral Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    LizanoEduardo. 2018. “Comentarios sobre política monetaria y cambiaria: Un poco más de leña a la hoguera.” Análisis 12. Academia de CentroamericaSan José.

    • Search Google Scholar
    • Export Citation

    MuñozEvelyn. 2012. “Costa Rica en la ruta hacia metas de inflación.” BCCR Working Paper 14–2012Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    MuñozEvelyn. 2018. “Adopción de un régimen monetario de metas de inflación en Costa Rica.” Central Bank of Costa Rica Working Paper 01–2018Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    OECD (Organisation for Economic Co-operation and Development). 2016. “OECD Economic Surveys: Costa Rica Economic Assessment.” OECDParis.

    • Search Google Scholar
    • Export Citation

    OstryJonathanAtish R.Gosh andMarcosChamon. 2012. “Two Targets, Two Instruments: Monetary and Exchange Rate Policies in Emerging Market Economies.” IMF Staff Discussion Note SDN/12/01International Monetary FundWashington, DC.

    • Search Google Scholar
    • Export Citation

    PayneRichard andPaoloVitale. 2003. “A Transaction-Level Study of the Effects of Central Bank Intervention on Exchange Rates.” Journal of International Economics61: 331352.

    • Search Google Scholar
    • Export Citation

    RodríguezClareAndrés andMiguel ÁngelRodríguez Echeverría. 1990. “Política cambiaria en Costa Rica: Dimensiones monetaria y real.” In Políticas Económicas en Costa Rica edited by ClaudioGonzález Vega andEdnaCamacho Mejía. San José.

    • Search Google Scholar
    • Export Citation

    RodríguezVargasAdolfo. 2009. “Evaluación del modelo lineal de pass-through para la proyección de inflación dentro del régimen de banda cambiaria.” BCCR Research Paper DI-10-2009Central Bank of Costa RicaSan José.

    • Search Google Scholar
    • Export Citation

    SarnoLucio andMark P.Taylor. 2002. The Economics of Exchange Rates. Cambridge, England: Cambridge University Press.

1Costa Rica’s central banking history might be said to have begun in 1921, when the monopoly of money supply was granted to Banco Internacional de Costa Rica, and more formally with the banking reforms of 1936, which brought about the creation of a Mint Department at the Banco Nacional de Costa Rica. From 1921 to 1950, the country had different forms of fixed exchange rate regimes in place, including the gold standard, the dollar-gold standard, and even a currency board from 1922 to 1930. In line with the Bretton Woods agreements, the country formally adopted a fixed exchange rate regime in 1947 (Delgado 2000).
2The adoption of the crawling peg was an evolutionary, de facto process: no explicit central bank board decree introduced the regime. The crawling peg regime was suspended between March and July 1992, when the exchange rate was allowed to float.
3Article 5 of the Board of the Costa Rican central bank’s decision to adopt a crawling band (Session 5300–2006, October 13, 2006) states: “The crawling peg regime has contributed to maintaining relative external stability over the last two decades and allowed a successful integration of the Costa Rican economy into global markets. However, it has also generated costs and vulnerabilities to the national economy, and reduced the effectiveness of the Costa Rican central bank’s monetary policy, whose primary objective is the achievement of a low and stable inflation.”
4In 2013–14, authorities also took prudential measures to induce a decline in the degree of financial dollarization, including higher risk weights for capital adequacy purposes for loans to foreign exchange borrowers who are not naturally hedged.
5For specific definitions, see IMF’s AREAER Database, multiple years. The IMF also classified the de facto regime as “stabilized” during 2014, when the central bank considered the exchange rate to be under a de facto managed float, because it remained within the limits of the crawling band.
6Muñoz (2018) provides details on this transition until the formal adoption of an inflation- targeting regime.
7Transmission from the policy rate to other rates in the system has been incomplete, in part because of financial dollarization and the oligopolistic structure of the financial system (Barquero and Mora 2014; Barquero and Orane 2015).
8Agents with very large transaction volumes often negotiate directly with financial intermediaries and obtain better rates. These transactions represent about 3 percent of the total, in value terms.
9MONEX started in 2006. It was initially restricted to authorized entities (mainly financial institutions). From 2009, corporations and individuals were also allowed to participate, which helped reduce concentration in MONEX as well as intermediation spreads in ventanillas (Central Bank of Costa Rica 2010).
10The Herfndahl-Hirschmann Index for traded values in MONEX (excluding the central bank’s interventions) gives a concentration figure of 1,000 for 2017, which corresponds to the threshold between low and moderate concentration. However, the index was designed for product markets, and yields low values when the largest firms have relatively equal market shares, as is the case in MONEX (19 percent, 14 percent, 13 percent, and 11 percent, for the four largest firms, respectively), instead of having one or two firms with very large shares, which is the main concern that the index addresses.
11Under the crawling band regime, a fourth motive for the central bank to intervene was to defend the limits of the band. As discussed, this type of intervention is not considered in the analysis.
12For example, the mean and standard deviation of daily purchases for this motive were $7.8 million and $6.3 million, compared with $9.9 million and $11.8 million, respectively, for foreign exchange sales for stabilization.
13The objectives and broad criteria for this rule were explained in detail in Central Bank of Costa Rica 2014b.
14See, for instance, Central Bank of Costa Rica 2015 (http://www.bccr.f.cr/politica_cambiaria).
15Vote 2014–00–7938 of June 6, 2014.
16See the brief review in Chapter 2 of this book. For a comprehensive survey of the earlier literature, see Sarno and Taylor (2002).
17To maintain the confidentiality of the information at the transaction level, all the estimations in this chapter were performed at the central bank.
18Other studies that have used event analysis for the assessment of intervention effectiveness are Fratzscher (2005) for advanced economies; Contreras, Pistelli, and Sáez (2013) for emerging markets; and Echavarría, Melo, and Villamizar (2014) for Colombia.
19Between these measures, there is a correlation of 0.86.
20The fgure is based on squared returns, but the same result obtains if the standard deviation of the exchange rate is used.
21The following notation is used: *p < .1, **p < .05, and ***p < .01. Lagged dependent variables and the constant are omitted.
22This measure is based on the yearly schedule of MONEX and the number of negotiation days throughout the year.
23If the standard deviation of the exchange rate over a 60-minute rolling window is used (instead of over a 15-minute window), the results are qualitatively similar, with interventions and intervention amounts contributing to a decline in volatility, and the effect of purchases being negative and statistically significant (while sales seem to have a small positive effect on the standard deviation).

    Other Resources Citing This Publication