Open economies review 7:219-236 (1996)
9 1996 Kluwer Academic Publishers. Printed in The Netherlands.
Simple Credibility Tests of the ERM Bands
for the Pound Sterling and the Italian Lira
ELIAS D. BELESSAKOS
Baruch College
MICHAEL G. PAPAIOANNOU
International Monetary Fund and Bank of Greece
Key words:
JEL Classification Number: t=31; F33
Abstract
This article presents evidence that the European Monetary System (EMS) bands for the Italian lira
and the pound sterling were not credible for most of the period 1990-1992, and especially during the
week prior to their withdrawal from the EMS system. Using a simple test, developed by Svensson,
domestic interest rates for both Italy and the United Kingdom have been found to be mostly outside
the rate-of-return bands implied by the official arrangements of the EMS target zone system. Furthermore, comparing implied forward rates for various maturities with the official EMS bands of both currencies, we again found that the followed monetary policies in both countries were not in general consistent with those needed to maintain an orderly functioning of the (EMS) system. The Svensson test
can further be used as an indicator of potential speculative attacks on an EMS currency, and, in turn,
as a signal of an emerging need to adjust the corresponding country's monetary policy.
This article presents the results of a simple test, developed by Svensson (1990),
on the credibility of the exchange rate mechanism (ERM) bands around the central parity rates of the pound sterling and the Italian lira against the Deutsche mark
during the period 1990-1992. Motivated by the ERM crisis of September 1992,
this study intends to examine whether early warnings concerning realignment pressures can be detected by the application of this test. In this sense, the Svensson
test could serve as a market signal to monetary authorities of imminent speculative
attacks on certain European Monetary System (EMS) currency. Our results indicate
that the ERM bands for the lira were not credible for the period January 1990 to
September 1992, when three-month maturities were used. Also, for three-month
maturities, it is shown ,that lack of credibility of the sterling exchange rate band
persisted for about five months after the pound joined the ERM. Thereafter, its credibility improved until the EMS turbulence of September 1992. However, for maturities longer than three months and up to five years, the band does not appear to
be credible for the period November 1990-September 1992.
220
BELESSAKOS AND PAPAIOANNOU
The test employed assumes no arbitrage and has two alternative forms, one
utilizing the domestic interest rate and another utilizing the implied forward rate
(Geadah, Saavalainen, and Svensson, 1992). The test in its first form, which examines whether the domestic interest rate lies outside certain rate-of-return bands for
a given maturity, can be interpreted as a sign that the monetary authorities of the
country in question may not be following the appropriate interest rate policies in
relation to foreign countries. In other words the test in this form may serve as a
simple indicator of whether a currency is a candidate for a speculative attack (Flood
and Garber, 1989). Using the second form of the test, which examines whether a
currency's implied forward rate is within a prespecified exchange rate band, we can
determine whether market participants expect with certainty that over a certain
investment period an exchange rate realignment will occur. If the implied forward
rate lies outside the band, then the band is considered not credible. Therefore the
followed monetary policy may not be compatible with the survival of the regime.
Note that the second form of the test utilizes the additional assumption of uncovered
interest rate parity, and has the benefit of quantifying the degree of credibility.
The notion of credibility underlying the article is similar to that employed in
Svensson's (1990) article. That is, an explicit target zone cannot be credible if
domestic interest rates or expected future exchange rates fall outside certain prespecified bands. There have been various other concepts of credibility analyzed
in the policy game and target zone literature. For example, credible exchange rate
pegging may be viewed as an instrument for gaining anti-inflation credibility in the
sense of Giavazzi and Pagano (1988). Especially for inflation-prone countries, credible pegging to a fixed exchange rate could substitute for a precommitment to all
announced monetary target. However, the advantages of pegging the exchange
rate rather than setting a monetary target lie in the easier quantification and testing
of the former policy undertaking. Furthermore, target zone credibility and its relation to realignment expectations (a change in the central parity) and within-band
movements are analyzed in Bertola and Cabalero (1991) and Bartolini and Bodnar
(1992a). Finally, the link between capital flow liberalization and exchange rate
credibility within the EMS is examined in Eichengreen and Wyplosz (1993).
The article is organized as follows: In section 2, we present a brief survey of
the literature; in section 3, we formulate the tests of target zone credibility; in section 4, we discuss the data and analyze the empirical results; section 5 concludes
by discussing the relevant monetary policy implications.
1. A brief survey of the literature
A significant part of the target zone literature examines the question of how credible a target zone is. In many studies the analysis of the credibility has employed
a simple test developed by Svensson (1990). This test incorporates the face that
an assessment of the degree of exchange rate policy credibility cannot be based
only on the interest rate differential, since the relationship between credibility and
interest rate differentials varies not only with the respective maturities, but also
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
221
with the actual level of the exchange rate. By applying this test to Swedish data,
he shows that the unilaterally defended Swedish target zone was not credible for
a 24-month horizon or longer.
Along this line of research, Svensson (1993) presents estimates of devaluation
expectations for six EMS currencies relative to the Deutsche mark between March
1979 and April 1992. He finds that exchange rates within the ERM bands display
mean reverison rather than random walk (unit root) behavior. He shows that expected rates of devaluation have generally fallen during the period when the simple
test and a "drift adjustment" method (Bertola and Svensson, 1993) are used. Still,
he finds that some devaluation expectations relative to the mark remain in 1992,
and that the expected rates of devaluation have fluctuated more than interest rate
differentials. By applying an empirical model of time-varying realignment risk in
an exchange rate target zone, Rose and Svensson (1991) estimate expected rates
of devaluation for the French franc/Deutsche mark exchange rate as the difference
between interest rate differentials and estimated expected rates of depreciation
within the exchange rate band. During the EMS, it is shown that actual realignments
are able to be predicted with some success. Using daily data since the inception
of the EMS, Rose and Svensson (1993) analyze realignment expectations for various European exchange rates. They assert that there are few indications of poor
ERM credibility before late August 1992, thus making the currency crisis of September 1992 a rather unexpected event.
Bartolini and Bodnar (1992b) also use the Svensson test to examine the credibility
of the Italian lira's ERM band. They derive indicators which show that the lira's
target zone credibility increased substantially around mid-1989, along with Italy's
increased commitment to the ERM. Also, Molho (1991) argues that, under free
capital mobility, a high inflation country pursuing a nonaccommodating exchange
rate policy will have higher interest rates than its lower inflation trading partners
as long as that policy is not credible. He investigates developments in interest rate
differentials and capital and reserve flows for Italy, and suggest that this country's
nonaccommodating exchange rate policy has become significantly more credible
in 1989-1990 due to the authorities' reliance on fiscal and income policies to promote disinflation rather than on exchange rate policy.
Koen (1991) also applies the Svensson test for the Belgian franc band and concludes that the announcement and active implementation of a tighter exchange
rate policy generates credibility gains. Geadah, Saavalainen, and Svensson (1992)
use the simple test developed by Svensson (1990) to examine the credibility of
the exchange rate bands for the Nordic currencies during 1987-1991. They conclude that the credibility hypothesis of Finland's exchange rate band could not be
rejected except for the fall of 1991; Denmark's and Norway's bands lacked credibility
in the beginning of the period, but the test does not find lack of credibility for various
subsequent periods. The credibility of Sweden's band within a one-year horizon
could not be rejected up to fall 1989, but thereafter its credibility deteriorated sharply.
In a similar application of the Svensson test to the French franc/Deutsche mark
exchange rate band, Caramazza (1993) finds that the corresponding EMS fluctuation band may or may not have been credible for the period 1987-1991 and for
three months maturity.
222
BELESSAKOS AND PAPAIOANNOU
In an innovative study, Frankel and Phillips (1991) use a survey of exchange rate
forecasts, as well as interest rate differentials, to measure exchange rate expectations. Using these measures, they demonstrate the empirical failure of the standard target zone models to generate credible band, but find no evidence that such
inadequacy can be attributed to mismeasurement of expectations. They then consider a time-varying credibility measure, which is able to capture the importance
of possible realignments in forming overall expectations of exchange rate changes,
and show that the target zone model now becomes capable of producing sensible
bands. Lindberg and Soderlind (1991) test the first generation of exchange rate
target zone models using Swedish exchange rate data, and conclude that these
models do not adequately describe the Swedish exchange rate band. The results
of the application of the test to Swedish data show that the target zone lacked credibility for a 24-month horizon and beyond.
2. Tests of target zone credibility
This section presents the test of an explicit target zone credibility suggested by
Svensson (1990). Assuming no arbitrage, the test of an exchange-rate target-zone
credibility examines whether domestic interest rates fall within certain derived rateof-return bands. Under the assumptions of free capital mobility and no major capital
flows, an increase in the domestic interest rate above the corresponding rate-ofreturn band for any given term would signal the possibility of a devaluation. This
is an asymmetric test in the sense that a move of domestic interest rates inside
the rate-of-return band does not constitute a sufficient condition for credibility of
the target zone. If we further assume uncovered interest rate parity, the credibility
of a target zone can be tested by examining whether future spot exchange rates
fall within the officially announced exchangerate band.
A closer look of the incorporated assumptions will show some of the limitations
of Svensson's test and point out the sensitivity of this credibility measure with
respect to violations of such assumptions. First, if full capital mobility does not
prevail, the proposed first form of the test is not valid because international arbitrage can take place. Therefore, examination of whether domestic interest rates
fall inside or outside the exchange rate band becomes meaningless because their
relative position may be the sheer outcome of capital market constraints and not
the result of market conditions. Second, if no major capital flows accompany interest rate movements outside the band, then such an event may indicate that market
participants perceive a devaluation risk within the maturity period of the bond; thus
the explicit exchange-rate target zone is not considered credible. Therefore, if the
domestic interest rate is outside the rate-of-return band for a given maturity under
conditions of sufficient international capital mobility and thin capital markets (i.e.,
small capital inflows), then we conclude that the target zone is not credible.
If uncovered interest parity does not hold, the second form of the test that examines whether the expected future exchange rates fall within the exchange rate band
cannot be performed. Instead, other theories of short-term exchange rate determi-
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
223
nation (like the portfolio balance approach) could be employed. These, however,
lack the simplicity and precision of uncovered interest rate parity. Some methods
of estimating expected exchange rates under target zones are provided in Chen
and Giovannini (1992). In addition, if the foreign exchange risk premium is zero,
then uncovered interest rate parity holds. Thus the credibility of exchange rate target
zones could be assessed through the proposed second form of the test. However,
if the foreign exchange risk premium is not zero, then we need to estimate such
a premium (Caramazza, 1993) and adjust appropriately the interest rate differential for any given maturity in order to compute expected future exchange rates.
Subsequently, we may utilize the methodology of the proposed second form of
the test. Finally, it has been shown that the foreign exchange risk premium in target
zones is small and "can hardly exceed a fifth of observed interest rate differentials" even for high risk-averse investors (Svensson, 1991).
Formally, the test proceeds by defining the annualized, ex post domestic currency return in period t on a duration-T, foreign-currency-denominated investment,
R T, as:
R~ = (1
+
i:T)(~t+T/St) 12/T --
].
(1)
where i [ r and St denote the annualized, foreign currency interest rate in period t
for term-Tloans in foreign currency and the amount of the domestic currency per
unit of the foreign currency in period t, respectively. The test then investigates
whether the annualized, domestic currency interest rate in period t on term-T
domestic-currency loans, i dr, is within the rate-of-return band for that term. It is
assumed that the exchange rate fluctuates between a lower bound SL and an
upper bound Su such that: SL < St < S U. The latter inequality implies that there
are bounds on the amount of depreciation and appreciation of the domestic currency. The lower and upper bounds for the rate of return are then determined by:
Rrt = (1 + ifr)(sL/st) 12/r- 1
(2)
i:T)(Su/St) 12/T- 1
(3)
RTt = (1
+
where RTt < R T <Rrt. In order to test if the rate-of-return band is credible, we simply compare the domestic currency interest rate with the band itself. If RLTt < i dT
< RTt, then we cannot reject the hypothesis that the exchange rate band is credible. However, it does not imply that the exchange rate band is credible. Therefore,
the test is inconclusive since the "true" expected spot rate may not be the same
as the forward rate (for further discussion see Geadah, Saavalainen, and Svensson,
1992). If RLTt > i dT or i dr > RTt, we conclude that the exchange rate band is not
credible.
By making the additional assumption of uncovered interest parity, the test implies that a positive difference between the interest rate on an asset denominated
in the domestic country's currency and that of a similar asset denominated in a
foreign country's currency is equal to the expected rate of depreciation in the spot
224
BELESSAKOS AND PAPAIOANNOU
exchange rate of the domestic country's currency (Isard, 1991). Then the essential
implication of this assumption is the equality between the expected rates of return
on a domestic and a foreign currency investments (Svensson, 1990). We define
uncovered interest parity as:
S: +T= St[(1 + i~T)/(1 + i[T)] T/2
(4)
where S[ +~ denotes the expected value in period t of the prevailing exchange rate
in period t + Z For a given maturity, the distance of the expected value of the exchange rate from the band can be used as a measure of the lack of credibility:
the further the expected future exchange rate is from the band, the less credible
the band is for any given intervention rule (Flood and Garber, 1989). If the expected
future exchange rate lies inside the exchange rate band, then the test is not inconclusive in the sense that the exchange rate band is not 100% credible (Geadah,
Saavalainen, and Svensson, 1992, p. 5).
3. Data and empirical results
In this section we report the results of the test applied to Italy and the United Kingdom, which are summarized in table 1. The calculated interest rate bands and the
forward exchange rates are presented in figures 1 through 12. The data used are
average weekly observations of spot Deutsche mark exchange rates and three-,
six-, twelve-, thirty-, and sixty-month Euro-deposit rates for the January 1990-September 1992 period. Forward exchange rates are calculated implicitly from (4),
assuming covered interest parity, for the various maturities.
We should mention here that the assumption of free capital mobility for Italy and
U.K. held true during our sample period, as both countries' capital markets were
completely liberalized in the middle and late 1980s (Mussa and Goldstein, 1993).
As far as the assumption of the volume and direction of capital flows when domestic interest rates for some term lie outside the corresponding rate-of-return bands,
we examined quarterly portfolio investment data for the relevant periods using IFS
numbers. Finally, with respect to the assumption of zero foreign exchange risk premium and thus of uncovered interest parity prevailing, we note that the risk premia
for relatively narrow target zones are likely to be small and ineffect make no difference, even when there is devaluation risk (Svensson, 1991a).
Examination of the credibility of the Italian lira target zone in the period of the
narrow band, January 1990-September 1992, i.e., until the time that the authorities
suspended the currency's participation in the ERM, indicates that the lira's narrow
target zone was never credible during that period. The test shows a dramatic decrease in credibility during the last week before the crisis (figures 1 and 7). It also
becomes evident (appendix) that during the period in question domestic interest
rates were outside the three-month rate-of-return band and there was a large and
increasing portfolio investment outflow, thus lending support to the interpretation
that these months might have been a period with a lack of credibility of the target
zone on a three-month horizon.
225
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
.,..-k
O,1
e-
.m
~|
E
II)
0
~D
tO
O
0
O~
ee--
~o
f0
(0
8
E
tO
eo
:,~
e-
0
m
~
I.,1.1
(0
ID
e-
u)
ie-
r
0
&
~
~
(~
"-- ~
~
O
N
.,
r
ffl
E
._~
__
~
r
e-
E
o
<O
r
"I
E~3
~
N
ul
0
r
~E
o
kU.I
ID
r
.~.om
.~_ ~
o
b.,
O)
o
~m
O
226
BELESSAKOS AND PAPAIOANNOU
20
15
10
5
0
Im R
LOW ~"R HIGH mi(t)
JAN 1990- SEPT 1992
Figure 1. 3-month lira rate of return band.
16
14
12
~
~
iiiiii_iii~i-iii...........iii ....
10
8
.... :-iiiiii_iiiiiiiiiiiiiiiiiiii
6
4
2
0
m R LOW " ' R HIGH w i ( t )
NOV 1990- SEPT 1992
Figure 2. 3-month pound rate of return band.
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
16
14
12
10
8
6
4
2
0
" - R LOW mR HIGH mi(t) I
NOV 1990- SEPT 1992
Figure 3. 6-month pound rate of return band.
14
12
10
8
6
4
2
0
'--R LOW mR HIGH mi(t)
NOV 1990- SEPT 1992
Figure 4. 12-month pound rate of return band.
227
228
BELESSAKOS AND PAPAIOANNOU
12
10
8
6
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
IIR
LOW i R
HIGH ,~i(t) I
NOV 1990- SEPT 1992
Figure 5. 30-month pound rate of return band.
14
12
10
8
6
4
2
0
1 R LOW I R
NOV 1990- SEPT 1992
Figure 6. 60-month pound rate of return band.
HIGH Bi(t)
.
.
.
.
.
.
.
.
.
.
.
.
.
229
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
1200
1000
800 . . . . . .
~
--
600
400
200
I- ' F O R W A R D
B S P O T LOW ~ S P O T HIGHJ
JAN 1990- SEPT 1992
Figure 7. 3-month forward lira/mark rate.
0.4
0.3
0.2
0.1
I--'FORWARD
"SPOT
NOV 1990- SEPT 1992
Figure 8. 3-month forward pound/mark rate.
LOW B S P O T HIGH]
230
BELESSAKOS AND PAPAIOANNOU
0.5
0.4
0.3
0.2
0.1
0
I='FORWARD
='SPOT LOW t S P O T HIGH 1
NOV 1990- SEPT 1992
Figure 9. 6-month forward pound/mark rate.
0.5
0.4
0.3
0.2
0.1
I==FORWARD ='SPOT LOW u S P O T HIGH I
NOV 1990- SEPT 1992
Figure 10. 12-month forward pound/mark rate.
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
" F O R W A R D "-SPOT LOW --SPOT HIGH t
NOV 1990- SEPT 1992
Figure 11. 30-month forward pound/mark rate.
1.2
1
0.8
0.6
0.4
0.2
0
" F O R W A R D "-SPOT LOW --SPOT HIGH
NOV 1990- SEPT 1992
Figure 12. 60-month forward pound/mark rate.
231
232
BELESSAKOS AND PAPAIOANNOU
The sample for the pound sterling spans the period November 1990, when it
joined the ERM, until September 1992, when it exited the system. For the threemonth maturity, the test implies that the pound joined at a time when the commitment to the band was not perceived to be credible, but that it became more credible later on. In September 1992, the domestic interest crossed the upper bound,
signaling the lack of credibility of the band (figures 2 and 8). Using six-month interest rates, we obtain results similar to the three-month maturity case (figures 3 and
9). In figures 4 and 10, utilization of twelve-month rates implies that the commitment was only rarely perceived to be credible, and the devaluation could have easily
been predicted in early September 1992. Figures 5 and 6, as well as 11 and 12,
show that the sterling's target zone completely lacked credibility for 30- and 60month horizons, and that the divergence of the domestic interest rate from the upper
bound increases sharply right before the crisis. As becomes apparent from figures
2-6 and 8-12, the band loses credibility as maturity increases.
For the pound sterling, when the three-month interest rate was outside its rateof-return band from November 1990 to March 1991, there were observed large
capital outflows. This indicates a definite lack of credibility of the target zone. The
six-month interest rate was outside the corresponding band from November 1990
to August 1991, October 1991 to M arch 1992, and August 1992 to September 1992.
During these periods, significant portfolio investment outflows were recorded, consistent with credibility deterioration. Note that, although monthly statistics for August
and September are not available, our data for the third quarter of 1992 indicate
sizable portfolio capital inflows. Indeed, this is not in direct conformity with the previously stated and generally accepted lack of credibility of the target zone during
the period of the pound sterling withdrawal from the EMS in September 1992. Such
inflows may only be interpreted as a possible temporary improvement in the target
zones credibility owing to the strong official defense of the EMS in June and July
1992. Moreover, monthly data for August and September 1992 will be able to shed
more light on this issue. The twelve-month interest rate was outside of its rate-ofreturn band from November 1990 to June 1992, and from August 1992 to September 1992. Capital outflows were large during most of these periods, indicating again
lack of target zone credibility. Finally, the thirty- and sixty-month interest rates were
outside the respective bands for the entire sample period, with portfolio capital
outflows being large and increasing in 1990, 1991, and the first quarter of 1992,
thus providing additional evidence of a lack of credibility of the target zone on the
corresponding horizons (appendix).
From figures 1-6, it becomes quite clear that the actual interest rate is either
outside or close to the upper bound of the rate-of-return bands for all maturities.
This phenomenon may indicate an inherent instability of the EMS system, which
may stem from the fact that participating countries do not face an enforcement
mechanism for coordination of their policies. The EMS, like any other fixed exchange
rate regime, encounters the n - 1 problem in the sense that only one member
country can maintain monetary policy autonomy. Traditionally, this role is assumed
by the reserve-center country. Furthermore, the reserve center is expected to live
up to the expectations of the other members in fulfilling the responsibility of effectively maintaining reasonably stable exchange rate relationships.
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
233
It is widely believed that the September 1992 turmoil in the ERM was principally
due to the high German interest rates, in response to the excessive unification
financing needs and the insistence of the Bundesbank on its strict antiinflationary
policy, as well as the inability of many member countries to maintain comparable
interest rate levels owing to high domestic unemployment rates and concerns over
national debt servicing. In this case, the credibility of the underlying policies in
sustaining such an entrusted exchange rate stability was challenged by markets,
by large shocks, since they disapproved of the followed policies. When a currency
comes under speculative attack, ERM participants, including the reserve center,
traditionally act swiftly and intervene in a coordinated fashion as a first measure,
while preparing to take appropriate interest rate actions if they consider the parity
viable. Cooperation of member central banks and governments may calm the markets for a short while, but it is only consistent macroeconomic policies that will
avert the markets from aggressively testing the viability of the system.
When announced objectives are not accompanied by supportive macroeconomic
policies, foreign exchange markets may react abruptly and force a certain exchange
parity to alter in line with market perceptions. Speculative attacks may occur and
cause the parity to be abandoned, even though economic fundamentals may be
consistent with the infinite maintenance of the fixed exchange rate system (Obstfeld,
1986). On the other hand, markets may successfully attack the parity because fundamentals are out of line (Krugman, 1979). 1 Often, successful defense against a
speculative attack depends critically on the size of available reserves and the readiness and determination of central banks to intervene. The effectiveness of central
bank interventions is determined by the extent of their sterilization and the consequent effect on interest rates (O'Connell, 1988). If intervention is sterilized, there
will be no monetary or interest rate effects, and, ordinarily, there will be no sustained
change in an exchange parity. If nonsterilized, intervention will move interest rates
in the right direction, and may even move them sufficiently to fight off the attack.
Note that if nonsterilized intervention occurs, the preintervention interest rate differentials and hence the credibility measure have no implications whatsoever for the
potential success of the defense. This is not true for sterilized intervention.
4. Concluding remarks
The test used in this study can serve as an additional indicator for assessing the
consistency of monetary and exchange rate policy targets. In effect, this test can
be utilized as an early warning signal of a potential devaluation for a currency participating in an exchange rate target zone arrangement, like the ERM of the EMS.
Therefore, if monetary authorities monitor the credibility of an explicit exchange
rate target zone through Svensson's simple test, speculative attacks could be prevented through appropriate and timely changes in their policies; in turn, foreign
exchange parities could be maintained.
Svensson's simple test also has implications in the policy area of defending a
target zone. Specifically, if the exchange rate target zone is credible in the Svensson
234
BELESSAKOS AND PAPAIOANNOU
sense, then it might be sensible for monetary authorities to attempt to defend the
target zone through nonsterilized interventions. In case the target zone is not credible, efforts to defend the zone might not have a reasonable chance of success,
and thus should be limited and rather reserved. Otherwise, foreign exchange market interventions might prove costly in terms of reserves and consequent interest
rate adjustments very disruptive to the orderly functioning of the financial system,
without in the end avoiding the change or the abandonment of the existing target
zone. Therefore, Svensson's simple test could be used to distinguish between potentially successful (i.e., hard to defend) and unsuccessful (i.e., reasonable to defend
a target zone from) speculative attacks.
By applying this test to the Italian lira and the pound sterling, we found that the
followed monetary policies were not in general consistent with the EMS exchange
rate bands for the period January 1990-September 1992. For three-month maturities, the official EMS band for the lira was proven to be not credible for the entire
sample period. In the case of sterling, the official EMS band was found to be not
credible for the first five months after it joined the (EMS) system, when we used
three-month interest rate maturities. In addition, the credibility of sterling's EMS
band was proven to be decreasing with longer maturities. From our results, it becomes apparent that the EMS bands for both currencies had turned completely
noncredible during the week prior to their withdrawal from the EMS system. In this
case, the Svensson test could be used as a sign of potential speculative pressures
on these EMS currencies, and therefore as a hint for necessary changes in the
respective country's monetary policy.
Appendix: Portfolio investment of Italy and the United Kingdom, January 1990
to September 1992
Italy
1st Quarter 1990
2nd Quarter 1990
3rd Quarter 1990
4th Quarter 1990
Year 1990
United Kingdom
- 3,097
6,218
- 3,122
-391
- 393
- 3,713
- 1,145
3,393
- 17,916
- 19,380
-3,560
1st Quarter 1991
2nd Quarter 1991
3rd Quarter 1991
4th Quarter 1991
Year 1991
-1,851
1,317
-6,170
-5,377
-2,943
-9,591
- 6,396
-24,307
1st Quarter 1992
2nd Quarter 1992
3rd Quarter 1992
- 16,362
- 8,656
- 13,033
-3,262
757
1,681
-2,077
Source: IMF, International Financial Statistics, January 1994.
SIMPLE CREDIBILITY TESTS OF THE ERM BANDS
235
Acknowledgments
The views expressed in this article are those of the authors and do not necessarily
represent those of the IMF or the Bank of Greece. We thank George Tavlas and
a referee for helpful comments and suggestions.
Note
1. Relative economic fundamentals should not be considered as absolute criteria for a currency's
strength. This is the case for the French franc at the time of its attack on September 22, 1992, when
French short-term interest rates were 100-125 basis points above those in Germany, despite the
fact that France's inflation rate had been significantly below that of West Germany's for the previous
two years.
References
Bartolini, L. and M.G. Bodnar (1992a) "Target Zones and Forward Rates in a Model with Repeated
Realignments;' Journal of Monetary Economics 30(3), 373-408.
Bartolini, L. and M.G. Bodnar (1992b) "An Analysis of the Process of Capital Liberalization in Italy."
WP/92/27, International Monetary Fund, Washington.
Bertola, G. and R. Caballero (1991) "Target Zones and Realignments," American Economic Review
82(3), 520-536.
Bertola, G. and LE.O. Svensson (1993) "Target Zones and Realignments" American Economic Review
82(3), 520-536.
Caramazza, F. (1993) "French-German Interest Rate Differential and Time-Varying Realignment Risk;'
Staff Papers (International Monetary Fund) 40(3), 567-583.
Chen, Z. and A. Giovannini (1992) "Estimating Expected Exchange Rates under Target Zones." Working Paper No. 3955, National Bureau of Economic Research, Inc., Cambridge, MA.
Eichengreen, B. and C. Wyplosz (1993) "The Unstable EMS" Brookings Papers on Economic Activity
24(1), 51-125.
Flood, P.R. and MP. Garber (1989) "The Linkage Between Speculative Attacks and Target Zone Models
of Exchange Rates." Working Paper No. 2918, National Bureau of Economic Research, Inc., Cambridge, MA.
Frankel, J. and S. Phillips (1991) "The European Monetary System: Credible at Last?" Working Paper
No. 3819, National Bureau of Economic Research, Inc., Cambridge, MA.
Geadah, G., T. Saavalainen, and L.E.O. Svensson (1992) "The Credibility of Nordic Exchange Rate
Bands 1987-91"' WP/92/3, International Monetary Fund, Washington.
Giavazzi, E and M. Pagano (1988) "The Advantage of Tying One's Hands: EMS Discipline and Central
Bank Credibility" European Economic Review 32, 1055-1082.
Isard, P. (1991) "Uncovered Interest Parity." WP/91/51, International Monetary Fund, Washington.
Koen, R.V. (1991) "Testing the Credibility of the Belgian Hard Currency Policy." WP/91/79, International
Monetary Fund, Washington.
Krugman, R.R (1988) "Target Zones and Exchange Rate Dynamics;' Working Paper No. 2481, National
Bureau of Economic Research, Inc., Cambridge, MA.
Krugman, R.R (1979) "A Model of Balance-of-Payments Crises," Journal of Money, Credit and Banking 11 (3), 311-325.
Lindberg, H. and R S~derlind (1991) "Testing the Basic Target Zone Model on Swedish Data." Seminar
Paper No. 488, Institute for International Economic Studies, Stockholm.
Molho, E.L. (1991) "The Italian Lira in the Narrow ERM Band: The Challenge of Credibility." WP/91/19,
International Monetary Fund, Washington.
236
BELESSAKOS AND PAPAIOANNOU
Mussa, M. and M. Goldstein (1993) "The Integration of World Capital Markets." In Changing Capita/
Markets: Implications for Monetary Policy. A Symposium Sponsored by the Federal Reserve Bank
of Kansas City, Wyoming, August 19-21, pp. 245-313.
Obstfeld, M. (1986) "Rational and Self-Fulfilling Balance-of-PaymentsCrises," American Economic
Review 76(1), 72-81.
O'Connell, J. (1988) "Sterilization and InterestRAtes," Journal of Intemational Money and Finance 7(4),
425-428.
Rose, A.K. and L.E.O.Svensson (1991) "Expected and Predicted Realignments:The FF/DM Exchange
Rate During the EMS." Working Paper No. 3685, National Bureau of Economic Research, Inc., Cambridge, MA.
Rose, A.K. and L.E.O. Svensson (1993) "European Exchange Rate Credibility Before the Fall." Working Paper No. 4495, National Bureau of Economic Research, Inc., Cambridge, MA.
Svensson, L.E.O. (1990) "The Simplest Test of the TargetZone Credibility," Staff Papers (International
Monetary Fund) 38(3), 655-665.
Svensson, L.E.O. (1991) "The Foreign Exchange Risk Premium in a TargetZone with DevaluationRisk."
CEPR Discussion Paper No. 494.
Svensson, L.E.O. (1992) "Why Exchange Rate Bands? Monetary Independence in Spite of Fixed Exchange Rates" Seminar Paper No. 521, Institute for International Economic Studies, Stockholm
University.
Svensson, L.E.O. (1993) "Assessing TargetZone Credibility: Mean Reversionand DevaluationExpectations in the ERM, 1979-1992," European Economic Review 37(4), 763-793.