Flexible exchange rate regime and forex intervention

Text-only Preview

Flexible exchange rate regime and forex intervention
José De Gregorio and Andrea Tokman R
This paper reviews the recent experience with a flexible exchange rate regime and forex interventions
in Chile. It discusses the state of the economy and the policy implications that arise in the new regime;
in particular, the reaction of the authorities to unexpected movements in the exchange rate, through
monetary policy and sterilised interventions. The low risks associated with financial and price instability
prevailing in Chile justify limiting policy reaction to exceptional circumstances in the exchange rate

A Managed exchange rate was a common feature of Chilean exchange rate policy during the 1990S
and before. The move towards a flexible exchange rate at the end of the decade implied a de facto
compromise away from exchange rate targeting. Macroeconomic stability, consolidated with low
inflation, sound fiscal policies and a strong financial system, made the compromise credible and
feasible in the eyes of the market and the authorities. However, an escape clause was kept open
when flexibility was introduced. The central bank reserved the right to intervene in the foreign
exchange market under exceptional circumstances, and it actually did so in two four-month episodes,
in 2001 and 2002.
The purpose of this paper is to analyse exchange rate management in the Chilean economy, within
the flexible exchange rate regime. First, in section 1, we briefly describe the conditions under which
the flexible exchange rate regime was implemented in September 1999. In section 2 we discuss policy
responses to exchange rate variations within the floating regime. We look at the rationale for
intervening through interest rates and/or directly in the foreign exchange market, the strategy and
instruments used and their effectiveness. In section 3 we deal explicitly with the Chilean experience
with intervention. Section 4 presents the conclusions and policy implications.
The implementation of the free float
Following a long history of managing the exchange rate, the Central Bank of Chile decided to let the
exchange rate float freely in September 1999. This was a reasonable thing to do; the coexistence of
two nominal anchors – inflation and exchange rate – eroded the credibility of the inflation-targeting
regime, and undermined its effectiveness. Moreover, although there were risks associated with the
float, the benefits far exceeded the potential costs.
Most of the crises and recessions in Chile have been associated with some rigidity in the exchange
rate. Most notably, in 1982, a fixed exchange rate, together with a bad international environment and a
fragile financial system led to a recession where output fell by 13%. The next recession, in 1999,
which caused a decline in output of 1%, was also linked to difficulties in adjusting the exchange rate to
the deteriorated international environment and heavy pressures on the peso during 1998. This had led
to a sharp tightening of monetary policy and a narrowing of the exchange rate band. The main
rationale provided for the monetary tightening on the eve of the Asian crisis was the need to reduce
expenditure and consequently the current account deficit, then larger than 5% of GDP, which were
viewed as difficult to finance in a scenario of a sharp decline in capital flows. In addition, reasons for
defending the currency were the fear that the inflationary repercussions from a sharp depreciation, and
the potential balance sheet effects on the corporate sector, could contaminate the financial system,
were the.
As financial turbulence passed and inflation declined to levels close to 2%, the authorities decided to
implement a fully fledged inflation target, which also included increased degrees of transparency of
monetary policy. Since then, the target has been to keep inflation within a range of 2 to 4% within a
horizon of twelve to twenty four months. As part of this strategy, the implementation of a flexible
exchange rate regime was central.
BIS Papers No 24

Arguments that favour flexible exchange rates are abundant in the economic literature. The most
traditional ones, associated with Mundell (1961), state that flexible exchange rates are key in easing
the adjustment to real shocks in the presence of price stickiness. In such case, real shocks will
generate movements in the exchange rate that will produce the necessary shift in resource allocation,
reducing the impact on output and employment. In contrast, a real shock that calls for a depreciation of
the currency is magnified in the presence of a fixed exchange rate. The only way to achieve a more
depreciated real exchange rate is through a recession that brings deflationary pressures with it.
Another conclusion arising from the Mundell-Fleming model is that with floating exchange rate
regimes, domestic authorities retain the flexibility to use independent monetary policy as a stabilising
tool. Thus, this preserves the possibility of conducting countercyclical monetary policies. Credibility is a
key factor for its effectiveness, which can be achieved with an independent central bank, as is the
case in Chile. In contrast, under a fixed exchange rate regime, monetary policy becomes subordinated
to the exchange rate commitment, and fiscal policy remains the only stabilisation tool. This has proven
to be a titanic task, as evidence shows that fiscal policy is highly procyclical in many countries due to
poor access to financial markets (Gavin and Perotti 1997). In Chile, there is still scope for stabilisation
through fiscal policy. Indeed, Chile is the only country in the region that can have countercyclical fiscal
policy. The government has implemented a rule based on a cyclically adjusted budget deficit. Although
the advantages of having fiscal rules and limiting discretionary fiscal policy are beyond the scope of
this discussion, the disadvantages of giving up monetary policy and relying only on fiscal policy are
Figure 1
Standard deviation of the nominal exchange rate
(30 day rolling window)

Arguments against a shift towards floating exchange rate regimes are the costs associated with an
unsurprising increase in exchange rate volatility (Figure 1). Still, the threat that such increase inflicts
on the Chilean economy is limited. First, the relative volatility of the Chilean peso remains within
normal levels by international standards (Table 1). Additionally, price and financial stability have not
been affected since the exchange rate was floating. The pass-through from depreciation to domestic
inflation was relatively small and has declined since the peso began floating. Regarding financial
fragility, in late 1999 there was only slight liability dollarisation in the banking system, and the
corporate sector was adequately hedged against exchange rate risk. Finally, the derivatives market
has deepened substantially since the exchange rate was allowed to float. Thus, Chile was well
BIS Papers No 24

prepared to float, with little reason to fear floating.1 In such a scenario, the flexible exchange rate
system should have operated smoothly, as it has done for most of the floatation period. Nevertheless,
there have been a few episodes where the central bank has reacted to movements in the exchange
rate. These episodes have been extremely rare and constrained enough as to keep Chile within the
limited group of countries that are qualified as both de facto and de jure floaters in all international
classifications (see, for example, IMF or Levy-Yeyati and Sturzenegger, 2004).
Policy response to exchange rate fluctuations: an overview
Policy responses
The Central Bank of Chile, like most central banks around the world, has responded to exchange rate
fluctuations in several ways during the past years: first, during the1990s, with capital controls and
reserve accumulation; then, during the Asian crisis, with a combination of monetary and intervention
policies. Finally, after the exchange rate was allowed to float, there have been two periods of sterilised
intervention, when the currency was under extreme stress and the monetary authority believed that it
was becoming misaligned and the market was overreacting. As such, these reactions have been rare
and not automatic.

Table 1
Currency volatility

NZ $
1990-1995 1.71
25.25 8.90 7.44 10.67 8.14 6.67 5.93 4.44 11.94
0.45 6.68 8.08 4.99 11.79 9.94 10.12 9.58 4.80 9.97 10.62
2001-2004 (Oct)
15.67 16.63 8.02 9.25 9.43 11.41
18.55 7.18 11.17 10.57
Source: Central Bank of Chile, based on data from Reuters.
Note: The volatilities are computed with respect to the US dollar. The methodology used corresponds to “Riskmetrics”,
proposed by JP Morgan, where a Garch (1,1) approximation is achieved by modifying its specification, so that the standard
deviation depends on the first lag in levels and variance of the first differences.
3. Monetary
With respect to the reaction through interest rates, the empirical evidence for developed (Clarida et al,
1998) and emerging countries (Mohanty and Klau, 2004) confirms the pattern observed in Chile:
inflation-targeting countries do react to exchange rate misalignment. And, as in the Chilean case, it is
rarely done in a mechanical way. This is wise, given the extreme difficulties that authorities have in
predicting future exchange rates and identifying movements away from their equilibrium levels.
Policy rule estimates by Schmidt-Hebbel and Tapia (2002) and Caputo (2003) confirm that there has
been a reaction to fluctuations in the exchange rate, over and above its effects on expected inflation,
but smaller than the reaction to expected inflation and output. It appears as if the exchange rate
equilibrium were a target per se, which is inconsistent with having only one nominal anchor. In such a
case, even if expected inflation were lined up with the target, monetary policy would react to deviations
of the exchange rate from its equilibrium. This result is not surprising since it was obtained for

1 For further details, see De Gregorio and Tokman (2004).
BIS Papers No 24

monetary policy reaction functions estimated during the1990s, where there was a declared objective of
exchange rate stability, implemented through an exchange rate band.
However, the above-mentioned results must be analysed with caution. In particular, a significant
reaction to exchange rate deviations might reflect an overestimation of its effects on future inflation. In
fact, if the inflation model used by the monetary authority is unable to fully anticipate a permanent
reduction in the pass-through, the policy reaction to a given depreciation will be larger, as the
expected inflation effect will also be larger. As a result, the ex-post reaction will appear to be bigger
than it really is, suggesting a very reactive monetary authority. Alternatively, responding to exchange
rate deviations on top of its effects on expected inflation might be reflecting reactions to expected
inflationary effects that are farther away in time than the policy horizon. In such a case, the exchange
rate reaction term is the response to an omitted variable (inflation in an excluded time horizon).
There is also an issue of endogeneity in the estimation of Taylor rules, as the exchange rate is
determined by the interest rate, thus reducing the robustness of results obtained from reduced-form
equations. In the same sense, there is an endogeneity problem in estimating whether monetary policy
can have an effect on the exchange rate. Indeed, using SVARs for the Chilean economy,
Parrado (2001) estimates that a contractionary monetary policy of 100 basis points produces a
significant instant real appreciation close to 1%, which is undone by the twentieth month.2
The extent to which monetary policy responds to exchange rate shocks should depend on whether the
authority has alternative instruments, such as capital controls and forex intervention, their relative
efficacy under different circumstances, and the nature and persistence of the exchange rate shock.
For example, there should be no monetary policy responses to temporary shocks, as they will not
produce lasting effects and therefore will not modify inflation expectations in the policy horizon.
However, this depends on the credibility of monetary policy. If credibility is low, temporary shocks will
have more persistent effects on inflation, as the public will expect authorities to accommodate to
higher inflation. Conversely, the effects will be minimised if the public perceives that monetary policy
will be tightened if inflationary repercussions are significant.
One of the main benefits of having a flexible exchange regime is that it allows fast adjustments in
relative prices in the face of real shocks, thus reducing their costs. Therefore, it is unlikely that interest
rates will need to be moved in the face of exchange rate movements that are a response to real
shocks (ie terms of trade or productivity shocks), since it is unlikely that they will have an impact on
inflation. Indeed, movements in the exchange rate that respond to adjustments in the equilibrium real
exchange rate will have smaller inflationary effects than movements that are not a response to
changes in fundamentals. This is the main reason, as we argue below, for exceptional sterilised
Ultimately, in a flexible exchange rate regime, the authorities should react to exchange rate
movements only if they impact the rate of inflation in the policy horizon, which, given the estimates of
pass-through, is limited. Attempting to target a misaligned exchange rate, for example to artificially
reduce inflation, may only bring costs, as the Chilean experience of limiting exchange rate adjustments
has shown. However, preventing overreactions in the exchange rate from leading to inflationary
pressures through monetary tightening, may reduce the costs of achieving the inflation target.
Exchange rate intervention
Alternatively, or complementarily, many countries react to exchange rate movements through some
type of intervention policy. In fact, most countries classified as free-floaters intervene in the forex
market.3 And even those that do not intervene have retained the option to do so in particularly stressful

2 This reaction is found to be small, nearly half of the one found for Australia, Canada and New Zealand (Zettelmeyer, 2000).
3 Only New Zealand and Poland have abstained from intervening and can be considered pure floaters. The US, Japan and
the EU have intervened in the market at different points in time, but their interventinos have shown a diminishing trend in
frequency, while increasing in size. Other examples are the diminishing interventions required by European central banks

BIS Papers No 24

Two interesting issues emerge from the observed reactions to exchange rate movements through
interventions. The first has to do with the high support they have received at the policymaking level,
given the much more sceptical view in the academic world of interventions effectively producing a
change in the exchange rate, its trend, or its volatility. The second is related to the decision to react
through the forex market and not through monetary policy. We will leave this second issue for the next
sections and will concentrate now on a broad review of the efficacy of interventions. In the next
sections we will discuss the intervention experience in Chile and conclude with a discussion on the
optimal response to exchange rate fluctuations.
In the context of a flexible exchange rate regime, there are three channels through which interventions
in the forex market affect the exchange rate.4 The first one is the portfolio channel, whereby changes
in the desired allocation of currencies in the portfolios of investors could cause large swings in the
exchange rate. In this case the intervention of the central bank could reduce fluctuations by providing
the necessary supply of currency to reduce fluctuations. The research that has analysed the
effectiveness of intervention through the portfolio channel concludes that, more often than not, that
sterilised interventions have very small short-run effects, mainly because intervention volumes are
A second channel through which intervention affects exchange rates is signaling. The idea is that an
intervention provides signals about the future course of monetary policy, which in turn affects asset
prices. For example, when intervention is undertaken to avoid depreciation, the next step would be to
tighten monetary policy, which should strengthen the domestic currency. This view has received some
empirical support, but we do not believe it is very relevant, because many times, and as we argue
below in the case of Chile, intervention is done precisely to prevent a monetary tightening to avoid
inflationary pressures stemming from excessive depreciation.
There is a third channel, called the information channel. In this case the authorities transmit certain
information to the market via an intervention and its announcements. In the Chilean case, for example,
this was that the exchange rate was moving out of line with the evolution of fundamentals. The
empirical analysis of the information channel has centred on the microstructure of the foreign
exchange market, concluding that the impact of the interventions is bigger, the larger the uncertainty in
the market, as measured by exchange rate volatility.
While most central bank officers believe that interventions may have an effect on the exchange rate
(Neely, 2001), the empirical evidence has been unable to provide robust support for that notion. In
fact, although there have been swings in the prescribed efficacy of interventions through time, today
the issue is still open to debate. The relative consensus reached by earlier studies regarding the small
effect of sterilised interventions on the exchange rate (Jurgensen, 1983), was at conflict with the
apparent success of coordinated interventions that followed the Plaza and Louvre agreements.5
Moreover, numerous recent studies have arrived to contradictory results.6
The disparity of results can be partly attributed to the presence of two empirical problems. The first
one is lack of data, and stems from the reluctance of central banks to publish official intervention
information, which makes the task of gathering statistics tedious and deficient.7 The second one is the

with the introduction of the euro in 1999, while the European Central Bank has intervened on only two occasions, both in the
year 2000; the UK and Switzerland have not intervened since 1992, except in 2000, in a coordinated action to support the
euro; and Canada abandoned its mechanical intervention rule, reducing its intervention activity substantially.
4 In many Latin American countries, including Chile in the 1990s, interventions were made to target a specific level or path for
the exchange rate, but we do not examine them here because they are not consistent with the free floating we are
discussing here. For further discussion see De Gregorio and Tokman (2004).
5 See, for example, Domínguez and Frankel, 1990 and 1993.
6 For a description of the later advances, see Sarno and Taylor (2001) and Ramaswamy and Samiei (2000).
7 Central banks normally do not make public announcements of their interventions, let alone disclose the amounts involved.
Even when present, disclosures are few and infrequent in comparison to the time span in which one expects the market to
adjust to intervention, often days or even hours. This deficiency has forced researchers to build indirect intervention series,
resorting to sources such as media news, surveys and movements in international reserves. Since these proxies are far
from perfect, it is possible that the intervention series built upon them are inadequate to estimate the true effects of
exchange rate interventions.
BIS Papers No 24

inability to control for the endogeneity of interventions in the estimations of their effect on the
exchange rate.
Fortunately, availability of information and reduced endogeneity bias due to the specific characteristics
of the intervention policy followed in Chile during the floating exchange rate regime has allowed for an
evaluation of its impact (Tapia and Tokman, 2004). The two intervention episodes after 1999 were
found to have effects on the exchange rate, and are discussed below.
Chile’s intervention experience
Chile is one of the countries that reserved the right to intervene when it adopted the floating regime in
1999. The monetary authority declared that, during exceptional episodes of uncertainty and volatility,
under which there might be adverse economic effects of an overreacting exchange rate, it was
desirable that the central bank intervene in the exchange rate market.8 Two such episodes occurred in
2001 and 2002, where the central bank, motivated by excessive volatility of the international financial
markets and the potentially adverse effects, announced a package of intervention measures to provide
more liquidity and foreign currency coverage. The first episode coincided with financial turmoil
stemming from the convertibility crisis in Argentina, aggravated by the events of 11 September 2001,
and the second with turbulence in Brazil during the presidential elections of 2002.
In both cases, there were clear indications that exchange rate depreciation was excessive, given the
evolution of fundamentals. Chile’s trade and financial links with Argentina and with Brazil are small.
For example, trade with both countries combined is less than 20% of overall Chilean trade. The sharp
depreciations clearly indicated that the market had lost its anchor, and hence they could have had
adverse effects on inflation. These would have required tightening monetary policy in a period in which
the economy was growing slowly and thus there were no inflationary pressures. Then, the intervention
was seen as a first line of defense against inflation coming from excessive depreciation. The chance of
a bubble dominating the market would have required actions to verify whether this was truly an
overreaction. Had the central bank not intervened, the excess depreciation more than that required for
adjusting the real exchange rate would have resulted in inflation undoing the real effects of the
nominal depreciation. Indeed, it is likely that a depreciation that pushes the real exchange rate above
its equilibrium level will bring inflation. This inflation, in turn, will validate an initially excessive
depreciation. Before tightening monetary policy or giving up on the inflation target it could be advisable
to intervene. This intervention does not pursue a particular level of the exchange rate, but aims rather
to avoid an excessive weakening of the currency. If intervention is not effective, it is an indication that
exchange rate movements could be the result of a need for a real depreciation. Given this reason,
intervention should last for a limited period, and must be oriented at providing liquidity and
reestablishing an orderly working of the forex market, rather than looking for a particular level for the
exchange rate or aiming to reduce fluctuations. The purpose of the intervention is to prevent a rapid
The success of sterilised interventions in Chile showed that market reactions indeed were unfounded.
Otherwise, intervention would have been ineffective, calling for monetary tightening if inflation
expectations had been inconsistent with the target.
The first intervention started on 16 August 2001, when the central bank communicated that spot
market interventions could occur up to a maximum of US$2 billion, over the following four months.
Additional sales of US$2 billion of dollar-denominated central bank bills (BCD) were also announced.10
During that period, spot market interventions totaled US$803 million, less than half the maximum
announced, which represented nearly 5% of the total stock of international reserves. The spot trades

8 See box II.4 in the Monetary Policy Report of January 2003.
9 Given that the inflation target in Chile is symmetric, all the arguments given for excessive depreciation discussed in the text
are also valid for excessive appreciation. We focus on depreciations since they have been the relevant issue in Chile in
recent years.
10 This amount was in addition to that of the regular program of renewal of BCD’s issued in 1998.
BIS Papers No 24

of foreign exchange were made in 15 interventions (15% of working days), and were substantially
smaller than those made in the interventions during the crawling peg period, and less than half the
amount exchanged during the unsterilised intervention to defend the peso in 1998.
The sale of BCDs added up to US$3.04 billion, including the BCDs that were part of the regular
rollover program. These were more frequent than interventions in the spot market and even than sales
of BCDs in previous intervention periods. The amount above the regular program of BCDs sales was
US$ 2.3 billion, which led to a total intervention of US$3.1 billion. During that time, the exchange rate
appreciated 3.9% (partly reversing the depreciation observed until August), although it had
accumulated a depreciation of nearly 5% in September. The maximum daily devaluation was 2.8%
(September), and the maximum appreciation in one day was 1.8%, in October.
On 10 October 2002, the central bank announced a period of interventions very much like that of
2001, two billion dollars in spot and US$ 2 billion in BCDs, to end on 10 February 2003. This
intervention occurred with the Brazilian country risk rate climbing and a complex global scenario. The
peso/dollar exchange rate depreciated 7% in one month, showing an accelerating trend, and without
similar deterioration in fundamentals, except for turmoil coming from Brazil. Thus, these developments
suggested that the exchange rate depreciated more as a result of contagion than of fundamentals.
Contrary to the previous experience, however, the central bank actually did not intervene in the spot
market. However, there were interventions, with the issue of dollar denominated debt. Furthermore, in
December 2002, the Governor of the central bank announced the possibility of redefining the
intervention strategy for the second half of the intervention period. A few days later, the BCD sale
calendar was cut by half. Five hundred million US dollars in BCDs were sold in each of the first two
months, October and November. Subsequently, the central bank considered that a milder intervention
would suffice, and sold US$250 million in each of the following months, December and January. Total
intervention in this episode was US$1.5 billion, without spot interventions. This episode involved much
less intervention than the first one. Reserves did not change and the total stock of BCDs increased to
US$5.8 billion.
During this second episode, the exchange rate appreciated by 2.1% (partly reversing the previous
depreciation), although by mid-December it had appreciated by 8.8%, to relapse in the following
months. The biggest depreciation in one day was 1.3%, and a 2.3% appreciation occurred the day
after the intervention announcement (see Figure 2).
During the second episode, the exchange rate gradually approached its initial level (Figure 2). The
intervention was prompted by unusual increases in spreads in Brazil and emerging markets, but after
the announcement that intervention would lessen because financial turmoil in emerging markets was
diminishing, the exchange rate began depreciating again, and at a slower pace than the one that
triggered the interventions (August-October). This episode shows, first, that the purpose of the
interventions was not to target a specific level for the exchange rate, but rather to reduce the speed of
depreciation. And, second, that the reaction of the Central Bank was based on turmoil in financial
markets rather than, again, on the exchange rate reaching a certain level.
It is interesting to compare the evolution of the exchange rate in both episodes. As Figure 2 shows, the
turmoil was much more intense in the first period, as even with significant spot market intervention
there were strong pressures against the peso. These pressures came not only from the deterioration
of the Argentine economy, but were also reinforced by the further weakening of the peso (by nearly
2%) that followed the terrorist attacks of 11 September 2001. In contrast, during the second episode
the reaction to the announcement may, despite lower effective intervention, be explained by the
credibility gained during the first episode. Nevertheless, the first impact, associated with the
intervention announcement was, after controlling for other fundamental movements, not as strong as
that of the first episode. This could also indicate that the market already assigned a high probability to
the assumption that all resources committed to intervene would not be used, as actually happened.
Therefore, the lesson is that announcements not followed by actions could reduce their impact on the
exchange rate.
BIS Papers No 24

Figure 2
Exchange rate and spot foreign exchange market interventions
US$ millions
Pesos per dollar
Pesos per dollar
Free float period
Intervention threat
Free float period
Intervention threat period
Massad Speech
Reduction of BCD
Jul Aug Aug Sep Sep Oct Oct Nov Nov Dec Jan Jan Feb

In terms of frequency and magnitude of interventions, Chile’s intervention policy has been modest. It
has only intervened in the spot market on fifteen occasions during the last five years,11 and amounts
have been low, both in absolute terms and relative to the market’s average turnover. Spot
interventions in 2001 averaged less than 5% of the daily volumes traded in the market, with no spot
market interventions in 2002 or 2003. Most interventions were made by issuing dollar denominated
As for transparency, the two announced intervention periods were very transparent by international
standards. Contrary to the worldwide trend towards transparent public policies – following the good
behaviour codes of the IMF – intervention policies around the world have not been very transparent.
Secrecy is still the rule in most countries, although there is a trend towards more public interventions
(Chiu, 2003). Canada, Hong Kong and the European Central Bank have compromised to provide
information whenever interventions occur (typically through press releases). The US has maintained
the possibility to intervene secretly, but has intervened publicly in its most recent episodes. In Japan,
the degree of transparency has fluctuated substantially over time. Moreover, even in those countries
that have moved towards more transparent interventions, the amounts involved have been kept secret
(only Hong Kong makes a real time amount announcement). Most report amounts with delays or in
monthly (as has Japan since June 2003) or quarterly (as have Canada, ECB and US) aggregate
amounts. But in every case the information is provided ex-post or, at best, simultaneously. There is, in
general, no evidence of ex-ante information on amounts and dates. In this sense, the last two
intervention episodes in Chile were exceptions.
The specific form of Chile’s intervention packages is of special interest, as it is not commonly observed
in other countries. In Chile, interventions have been announced in advance, and the beginning and
end of the intervention period have been made public. In addition, the maximum amount of the
intervention in the spot and BCD markets, and the calendar for the monthly BCD placements, have
also been made public. In practice, the only unknown intervention is the daily spot intervention, which
is published as official data from the central bank with a two week lag. The rationale for the choice of
high transparency is twofold. First, because the authorities have made a commitment to intervene in a
transparent manner, rather than surprising the market, in order to work through the information
channel. Indeed, intervention is done to let the market know that the authorities consider the evolution
of the exchange rate to be unjustified by fundamentals (short- and long-term). Second, and more
important, because intervention is intended to provide liquidity and stabilise the market, rather than
being a fight against speculators. This strategy avoids creating an “addiction to intervention”,
especially in a country with a high level of reserves. There must be a full assessment before
intervening, and this requires introducing some costs to the decision-making process. Thus,

11 The median intervention frequency among the central banks surveyed by Neely (2001) was 25%of trading days throughout
the 1990s.
BIS Papers No 24

intervention is more of a test for a potential bubble than an attempt to manage the exchange rate at
levels that could end up in a misalignment. For this reason, it has never been ruled out that an
intervention could be ineffective. If this were the case, the conclusion would be that the movement in
the exchange rate is much closer to an equilibrium phenomenon than an overreaction.
For the Chilean case, Tapia and Tokman (2004) show that the authorities are capable of affecting the
market not only through actual intervention operations, but also by public announcements and
commitments regarding them. This is, after all, the mechanism operating in two of the classical
channels through which interventions are effective: information and signaling. The information channel
refers to views of the authorities that differ from those implicit in asset prices, while the signaling effect
is related to the future course of monetary policy, ie after an intervention comes an interest rate hike.
Public announcements, whether formal or informal, reveal relevant information to the asset market,
which should adjust when news arrives.12 Indeed, in the case of Chile, the announcement provides
information that the central bank estimates that the depreciation is excessive.
However, the potential use of announcements as an effective intervention tool depends critically on
the credibility associated with them. Empty promises that are not backed by actions (in the case of the
portfolio or signaling channels) or that are made by authorities that are not considered reliable (in the
case of the information channel) should have no effect or, if the market was misled this time, weaken
the effect of future announcements. As the empirical methodology cannot distinguish the specific
channel through which interventions operate, it is not easy to say if announcements must be followed
by actual interventions. Under the portfolio channel, the announcement must necessarily be followed
by interventions and, in fact, the central bank must have enough reserves for the announcement to be
credible to begin with.
More debatable is whether the central bank must intervene when this is used as a signaling device or
to provide information. Under the signaling channel, there is no clear need for intervening after making
an announcement, because when intervention is used to signal a likely future tightening of monetary
policy, tightening must happen. Credibility would diminish if monetary policy (not interventions) did not
behave as implicitly suggested by the announcement. When the central bank uses intervention to
inform the market that it believes there is an overreaction, at some point it will have to “put its money
where its mouth” is. Therefore, via all channels through which intervention affects the exchange rate,
authorities should take action, either through sterilised intervention or monetary tightening.
The findings in Tapia and Tokman (2004) suggest that Chile’s relatively infrequent and unique
intervention strategy has succeeded in altering the exchange rate through its effect on expectations
caused by the (credible) policy announcements made for both periods. Obviously, this result is
conditional on the specific characteristics of the Central Bank of Chile, an independent institution with
high credibility and a large stock of international reserves. This suggests that these results, or the
policy prescriptions that might be derived from them, cannot be directly extended to other countries.
Nor can they be directly extrapolated to other periods in time for the Chilean economy.
In the case of Chile the most important impact of intervention came from the announcement, but in
both cases this was followed by action. We do not have evidence with which vto evaluate what would
have happened if the central bank had not validated the announcement with spot or BCD sales.
However, it is most likely that its credibility would have weakened. The absence of spot and/or BCD
interventions would have signaled lack of conviction accompanying the announcement. Moreover, it
was a reasonable thing to do; considering the availability of reserves and an excessive depreciation,
there was no reason for not shortening the central bank’s already very long position in dollars.
Furthermore, as the comparison of the two intervention episodes shows, an “open mouth” intervention
is not enough and may lose effectiveness if spot and/or BCD interventions do not follow. In any case,
their timing and amounts will depend on the initial effects of the announcement and on the evolution of
conditions that triggered the intervention.

12 There is very scarce literature on the role of communication or official central bank statements. Some exceptions are
Tivegna (2001), Fatum and Hutchinson (2002) and Jansen and De Haan (2003).
BIS Papers No 24

Conclusions and lessons on the optimal policy response to exchange
rate shocks

Chile has moved gradually to a floating exchange rate regime. This has been a reasonable thing to do,
especially because most of the recessions that the economy has endured have been associated with
exchange rate rigidities: the fixed exchange rate of the early 1980s and the narrowing of the exchange
rate band in the aftermath of the Asian crisis. Additionally, given the significant exposure to external
shocks – particularly terms of trade shocks and fluctuating capital inflows – a regime that allowed a
fast adjustment of relative prices was desirable and achievable through the flexibilisation of the
exchange rate. Instead of being subjected to extreme fluctuations in interest rates and financial
conditions as a means to control – often unsuccessfully – the exchange rate, it is sensible to let it
adjust, especially because it allowed expenditure switching and resource reallocation in the presence
of external shocks.
The float was implemented in accordance with the development of the financial markets and at a
moment when such movement appeared to convey more benefits than costs. Financial stability was
not threatened: the extent of currency mismatches in the banking and corporate sectors was small,
and liability dollarisation of the banking system was also unimportant. Firms have been increasingly
able to hedge their currency exposure in the derivatives market. Therefore, balance sheet effects
stemming from exchange rate fluctuations pose a threat to neither the financial nor the corporate
sectors. In addition, price stability was not at risk, given the low pass-through from depreciation to
From a macroeconomic point of view, moving to a flexible exchange rate regime was a necessary step
to implement a credible inflation target. In Chile during the 1990s, there was exchange rate targeting
with enough flexibility to accommodate the inflation target. This was the reason why, in the context of
an exchange rate band, the widths and central parity were frequently adjusted, this being also an
incentive for capital inflows as the authorities pursued a strategy of gradual appreciation (Cowan and
De Gregorio, 2004). Thus, allowing the exchange rate to float would make the inflation target more
credible and the economy more resilient to external shocks, as has proven to be the case in recent
Nevertheless, there have been some instances in which the central bank has intervened in the
exchange rate market, but they have been exceptions. The most significant drawback of intervention is
that authorities start intervening too often, denaturalising the float, which becomes a de facto managed
system. In some sense, authorities could become “addicted” to intervention. This could work for a
while, but in a country like Chile, with bad experiences in times of severe external turmoil, this is
dangerous and has proven to be very costly. For this reason, interventions must be rare events and
occur only in extreme circumstances. Their credibility, and thus their effectiveness, depends on them
being only occasional. Moreover, they should become even more infrequent, as the reasons to fear
floating appear to be fading with time.
For the above reasons, the method for conducting intervention is very relevant. Transparency and
clear rules of the game are necessary to make this an exceptional policy. Accordingly, in Chile it is the
Board of the Central Bank that decides exactly when intervention will occur and the maximum
amounts of intervention, and explains this clearly to the public. In contrast, if interventions are secret,
there is much more temptation to intervene whenever the market becomes volatile.
The question is now, what determines whether the central bank should react and how? There is little
evidence on such issues, but some things appear to be a natural part of the decision process. In the
first place, the potentially adverse effect of the shock must be substantial. In this sense, if the threats
of financial instability and inflation are low, as is the case with Chile (De Gregorio and Tokman, 2004),
then the need to react is diminished.
Also, the nature and the persistency of the exchange rate shock are important determinants of
whether there should be a reaction at all. Real and temporary shocks are, most probably, better left
alone. And, since a larger proportion of shocks is believed to be temporary in a floating regime, there
is a lower probability of having to react to them.
The size of the shock may be important too. As Lahiri and Vegh (2001) suggest, forex interventions
may be cost-effective in the presence of large shocks, as their fixed costs are lower than the costs
associated with interest rate policies for large shocks. If authorities react by tightening monetary policy
when facing a transitory and significant exchange rate shock, they may need to undo the tightening
BIS Papers No 24

Document Outline

  • ??
  • ??
  • ??
    • ??
  • ??
  • ??
  • ??
  • 6. Conclusions and lessons on the optimal policy response to exchange rate shocks
  • ??