Capital flows, exchange rate

regime and monetary policy

Sweta C Saxena 1

Introduction
Financial globalisation can provide significant benefits to developing countries but at the
same time poses significant risks. There is strong evidence to suggest that developing
economies could benefit from financial globalisation, given that certain framework conditions
are fulfilled. 2 Hence, there is a trend towards open capital accounts, as illustrated by
Malaysia, which recently shed controls that had been brought in 1998 in the aftermath of the
Asian crisis. The move towards higher capital mobility confronts central banks with some
difficult choices in implementing monetary policy:
1.

Control of exchange rate or interest rate? If central banks want to stabilise exchange
rates, they have to accept the consequences for domestic interest rates. If they wish
to gain control over their domestic interest rates, then they have to accept higher
volatility in their exchange rates. Hence, their independence to choose interest rates
can be constrained under an open capital account.

2.

Exchange rate or inflation as the nominal anchor? The move towards inflation
targeting implies giving up the exchange rate as the nominal anchor for monetary
policy, which means floating exchange rates with higher volatility. Does this mean
that the central bank should not care about exchange rate stability as such?
Conventional wisdom would have central banks pay attention to the exchange rate if
it interferes with the price stability goal. But what happens when the economy is
dollarised (Peru) or some contracts are denominated in foreign currency (Israel)?
What should countries that are building net foreign liabilities (denominated in foreign
currency) do when faced with the choice of exchange rate stability vis--vis price
stability? This paper will address some of these issues.

To foreshadow the main results, the paper finds that the emerging markets have become
more financially globalised, as can be seen in a build-up of gross foreign asset and liability
positions, increased presence of foreign investors in local currency bond markets and
increasing correlations of stock markets in the emerging markets with those of the industrial
countries. In fact, some countries have been able to issue longer-term local currency bonds
in the international markets, in spite of so-called original sin. Such an integration is
desirable as it increases international risk-sharing, but it can also increase the impact of
foreign shocks on domestic economies. The recent MayJune sell-off is a testimony to this.
Although many emerging markets (mainly Asia) improved their net external positions
between 1996 and 2004, the situation has worsened for others (mainly CEE countries due to
deteriorating current account balances). In the light of significant external liabilities

Extremely useful comments from Valerie Cerra, Jose de Gregorio, Mr Gudmundsson, Madhu Mohanty,
Ramn Moreno, Philip Turner and Bill White, expert research assistance from Clara Garcia, Marjorie Santos
and Gert Schnabel and proficient secretarial assistance from Clare Batts and Marcela Valdez-Komatsudani
are gratefully acknowledged.

See Kose et al (2006), for details.

BIS Papers No 35

81

(denominated in foreign currencies), CEE economies in particular are exposed to substantial

exchange rate risk.
On the impact of capital flows and the exchange rate regime on monetary policy, the paper
finds that domestic short-term interest rates are significantly affected by foreign interest
rates, especially for countries with high capital mobility and less than fully floating exchange
rates. The link between domestic and foreign interest rates is also in line with Moreno (2008)
that finds that the foreign long-term interest rate affects the domestic long-term interest rate
more than the domestic policy rate. The results also indicate that the credibility gained by
central banks in keeping inflation low and maintaining a stable macroeconomic environment
is helping to stabilise long rates more generally.
The rest of the paper is organised as follows. Section 1 discusses the constraints imposed by
capital flows on macroeconomic policy (the so-called impossible trinity or trilemma).
Section 2 investigates the impact of foreign interest rates on domestic interest rates under
various exchange rate and capital mobility regimes. Section 3 analyses the indicators of
financial globalisation and the issues related to exchange rate stability vis--vis price stability
(especially in the light of balance sheet effects and dollarisation issues). Section 4 concludes.

1.

The impossible trinity

The transmission of monetary policy depends on the openness of the capital account and the
exchange rate regime. The famous trilemma from the Mundell-Fleming model states that
countries cannot simultaneously fix their exchange rate, have an open capital account and
pursue an independent monetary policy. Only two out of these three objectives are mutually
consistent. 3 If the capital account is closed, then domestic interest rates would transmit to
domestic demand, irrespective of the exchange rate regime. However, if the capital account
is open, then domestic monetary policy will be determined by the exchange rate regime and
the degree of substitutability between domestic and foreign financial assets. Under a floating
regime, monetary policy can work either through the interest rate and liquidity channel or
through the exchange rate channel. Under the latter channel, the impact of monetary policy
on aggregate demand is larger if domestic and foreign assets are substitutable, as policyinduced changes in interest rates affect the exchange rate, which in turn affects output and
inflation. However, the higher substitutability between domestic and foreign assets offsets the
impact of monetary policy through capital flows in a fixed exchange rate regime. Hence,
monetary authorities can move domestic interest rates independently of foreign rates only if
there is a lesser degree of substitutability under a fixed exchange regime.
The foregoing analysis suggests that the exchange rate channel of monetary policy
transmission is hampered if the exchange rate is not allowed to move freely. Indeed, nine out
of 13 Asian and Latin American emerging economies actually use foreign exchange
intervention to complement their conduct of monetary policy. Hence, the impact of capital
flows on exchange rates may be offset through foreign exchange intervention. For instance,
Malaysia intervenes in the foreign exchange market to prevent large changes in exchange
rates that are not supported by fundamentals (Ooi (2008)).

82

Obstfeld et al (2005) find that this trilemma has been largely borne out by history. They find considerable
monetary autonomy for non-pegged regimes in the presence of capital mobility, but loss of this independence
for countries with pegged regimes.

BIS Papers No 35

2.

Do foreign interest rates influence domestic short-term interest

rates?

The question of monetary policy independence is closely linked to the choice of exchange
rate regime. If it is credible, a fixed exchange rate provides a nominal anchor for monetary
policy; if not, monetary policy is dictated by the need to attract capital flows to finance the
current account imbalances. If policymakers float their currency, then they gain control over
their monetary policy. The central bank can use domestic interest rates to respond to shocks
if the exchange rate is floating. Hence, domestic short-term interest rates in countries with
floating exchange rates should be less sensitive to changes in international interest rates.
But certain factors (eg foreign currency liabilities) prevent countries from following
independent monetary policies despite adopting a flexible exchange rate regime.
The relationship between exchange rate regime and monetary policy independence has
been tested in a few papers. For a large sample of industrial and developing countries,
Frankel et al (2004) show that domestic short-term interest rates, even in countries with
floating exchange rates, are linked with international interest rates in the long run. Only a
couple of large industrial countries can choose their own interest rates over time. However,
Frankel et al (2004) also find that the adjustment of floaters interest rates to international
interest rates is rather slow, implying some monetary independence in the short run. Unlike
Frankel et al (2004), Shambaugh (2004) finds that domestic interest rate behaviour is
different between pegged and non-pegged regimes: countries with pegged exchange rates
follow the base country interest rate more than others.
There is little empirical research linking capital mobility to monetary independence.
Shambaugh (2004) and Obstfeld et al (2005) do include a dummy variable for capital
controls to study the link between domestic and foreign interest rates. But this measure
cannot capture the intensity of capital controls or liberalisation. To address this, this paper
introduces a measure of international capital mobility which gauges the intensity of capital
liberalisation. A variable for interest rate liberalisation is also introduced.
These academic papers are an interesting line of research, but all of them are dated. The
data go up to 2000, but much has changed since then. During the last five to six years, the
emerging market economies have become more open on capital account and are following
freer exchange rate policies. For example, on a scale of 3, the average index of capital
mobility increased from 1.61 during 197599 to 2.59 between 2000 and 2006 for a group of
17 emerging economies in Asia and Latin America and including South Africa. The
proportion of observations on exchange rate regimes classified as floating increased from
68% to 73% between 197599 and 200006. Against this background, it would be interesting
to see if:
1.

higher capital mobility has increased the impact of foreign interest rates on domestic
rates; and

2.

floating the exchange rate helps reduce the impact of foreign interest rates on
domestic rates.

So what should we expect? Consider the following four scenarios and the expected domestic
interest rate link with the foreign interest rate:

Capital immobility
Capital mobility

BIS Papers No 35

Fixed exchange rate

Flexible exchange rate

No link

No link

Positive link

83

Under no capital mobility, monetary policy would be independent irrespective of the

exchange rate regime, implying that domestic interest rates can move independently of
foreign interest rates (and hence no link between the two). However, under full capital
mobility, the link between the domestic interest rate and the foreign interest rate would be
positive under a fixed exchange rate regime, as higher foreign interest rates would induce
capital outflow and a depreciation of the domestic currency. In order to prevent depreciation,
domestic interest rates should rise.
However, the link between domestic and foreign interest rates is not so clear when capital is
mobile and exchange rates are floating. The difficulty in determining the effect arises
because central banks often intervene in foreign exchange markets, even when their
exchange rates float, to smooth exchange rate fluctuations or accumulate foreign reserves
(see BIS Papers no 24). If the central bank does not allow full adjustment of the exchange
rate by intervening in the foreign exchange market even when the exchange rate is floating,
the reaction of the domestic interest rate to a foreign interest rate shock can be large. Hence,
we would expect a significantly positive link between domestic and foreign interest rates.
More precisely, in order to answer these questions, I use two techniques. I estimate the
following regression as well as the impulse response functions: 4
r it = + 1r t* + 2Capmobit + 3 r t*Capmobit + 4 Float it + 5 rt* Float it + u it

If coefficient 1 > 0 (significantly greater than zero), domestic short-term interest rates are
correlated with foreign interest rates. The correlation may arise because of common shocks
that require a common interest rate response, because of high capital mobility that imposes
an interest parity condition, or because of attempts to fix the exchange rate. For countries
with floating exchange rate regimes, any linkage may also provide evidence that the country
does not allow the exchange rate flexibility that it claims to (fear of floating) or intentionally
follows the foreign country.
A1: If higher capital mobility increases the impact of foreign interest rates on domestic
interest rates, the interaction of foreign interest rates and capital flows should be high and
significant ( 3 > 0 ). So, I test:
H0: 3 = 0 against H1: 3 > 0
A2: If exchange rate flexibility has reduced the impact of foreign interest rates on domestic
interest rates, then the relationship between local interest rates and foreign interest rates
should be weaker than for countries that continue to fix their exchange rates ( 5 < 0 ).
Therefore, I test the following hypothesis:

H0: 5 = 0 against H1: 5 < 0

In the light of the recent tightening of monetary policy in the United States, I also examine the
asymmetry of the interest rate linkage when US monetary policy is tight.
In addition, I estimate impulse response functions from the following regression: 6

Following Frankel et al (2004) and Shambaugh (2004), r = ln(1 + i), where i of 10% is represented as 0.10.
Also, the regression in changes is estimated as a pooled OLS (Shambaugh (2004) Obstfeld et al (2005)),
unlike the regression in levels with fixed effects as done in Frankel et al (2004).

While other factors can influence domestic interest rates, Shambaugh (2004) controls for time, trade shares,
debt exposure, capital controls and level of industrialisation, and finds that, during the post-Bretton Woods era,
with the exception of capital controls, the exchange rate regime tends to be the major determinant of how
closely domestic interest rates follow foreign interest rates.

This work is in the same spirit as Romer and Romer (1989), who identify the impact of monetary shocks on
US output in the postwar period.

84

BIS Papers No 35

j =1

s =0

r it = i + j r i ,t j + r t*s + it

The impulse response functions are shown with one standard error bands drawn from 1,000
Monte Carlo simulations. 7

Results 8, 9, 10
The regressions show a mostly significantly positive 1 coefficient (Tables 1 and A1A3),
implying that changes in domestic interest rates in these emerging markets do move in line
with the interest rate changes in the United States. This could be because of fear of floating
(for flexible exchange rate regimes) or because of the interest rate parity condition (for fixed
exchange rate regimes) or due to common shocks. The inclusion of world oil or food prices in
these regressions does not change the sign or the significance on the change in US interest
rates, implying that these common global shocks cannot be the reason for the positive sign.
But this linkage with the United States is stronger for the entire sample and early part (197589)
when the Fed tightens its monetary policy than when it eases it (Table 1). However, the
relationship between domestic and US interest rates is stronger in the recent period when
there is global tightening rather than easing 11 (Tables A1 and 2), ie interest rates in these
emerging markets move with the US interest rates when there is a general global tightening
which could occur due to common shocks requiring a common response. Perhaps the recent
oil shock is one example.
For countries with high capital mobility, the coefficient 3 is normally positive (when
significant), which implies that higher capital mobility reduces these countries ability to
change their interest rates independently (Tables 2, A2 and A4). 12 Countries with more
flexible exchange rates see a downward trend in their interest rates relative to countries with
fixed exchange rates (Tables 2, A3 and A4). But the coefficient on 5 is positive (when
significant), implying that flexibility in the exchange rate has apparently not bought these
countries any independence in setting their own domestic interest rates. Tables 2 and A4
show that 1 is negative during the 200006 period, but 3 and 5 are positive, implying
some delinking between the domestic and the US interest rate in general in recent times,
except for countries with high capital mobility and a flexible exchange rate regime. This result
is, however, counterintuitive as a flexible exchange rate regime in principle gives a central
bank greater room to manoeuvre and so makes monetary policy more independent. But as
discussed above, countries with flexible exchange rates can still have their domestic interest
rates move with the foreign interest rate under a higher level of capital mobility. This point is
brought out in Table 3.

I use four lags for domestic and US interest rates as the lags beyond that were mostly insignificant and the
DW stat shows no sign of serial correlation.

Data construction and some tables and graphs are provided in the Annex.

When I exclude periods of high inflation in Argentina, Brazil and Chile, capital mobility for the entire period
(19752006) becomes insignificant. All other results hold.

10

The results remain qualitatively unchanged even when a variable is introduced to capture the business cycle.

11

Global tightening refers to the periods when interest rates increase in the United States, the United Kingdom
and Japan simultaneously.

12

The results from interest rate liberalisation equations (not reported here) are similar. Countries with fully
market-determined interest rates have their interest rates move with US interest rates during 19752006 and
200006.

BIS Papers No 35

85

Table 1
Impact of US monetary policy tightening1
19752006

19751989

199099

200006

0.04

0.02

0.40

0.17

0.04

0.24

0.01

0.00

0.10

0.07

0.06

0.01

0.01

0.06

0.82

0.81

Rsq

0.00

0.00

0.00

0.03

DW statistic

1.80

1.87

1.73

1.51

6,902

2,360

2,671

1,871

24

18

24

24

r*
US tight MP * r*

Total observations
Cross sections
1

P-values are below the coefficients.

Table 2
Impact of capital mobility, exchange
rate regime and global tightening
19752006
r*

19751989

199099

200006

0.01

0.01

0.10

0.36

0.82

0.61

0.81

0.08

0.00

0.00

0.00

0.00

0.56

0.74

0.90

0.19

0.04

0.03

0.12

0.16

0.05

0.15

0.39

0.02

0.0004

0.00

0.001

0.00

0.03

0.36

0.01

0.30

0.00

0.01

0.12

0.17

0.95

0.78

0.60

0.00

0.04

0.05

0.38

0.24

0.52

0.42

0.31

0.02

Rsq

0.00

0.00

0.01

0.07

DW statistic

1.72

1.84

1.67

1.37

5,398

2,069

2,003

1,326

17

16

17

17

Capmob
Capmob * r*

Float
Float * r*
Global Tight * r*

Total observations
Cross sections
1

86

P-values are below the coefficients.

BIS Papers No 35

Table 3 illustrates the impact of foreign interest rates on domestic interest rates classified by
exchange rate regime and level of capital mobility. 13 When capital mobility is low, there is no
link between domestic and US interest rates, irrespective of the exchange rate regime. But,
as expected, the link between domestic and US interest rates is significantly positive for
countries with a fixed exchange rate and mobile capital. In addition, countries with flexible
exchange rate regimes have their domestic interest rate linked to the US interest rate only
when capital is mobile. Of the 6,273 observations on exchange rates and capital mobility
between 1975 and 2006, 40% represent a flexible exchange rate and high capital mobility
against 21% with a fixed exchange rate and high capital mobility. During 200006, the
proportion of observations with floating exchange rates and mobile capital is 72% against
23% with a fixed regime and mobile capital. The implication is that the proportion of
economies influenced by high capital mobility has risen sharply in recent years. Moreover,
this has coincided with a greater reliance on floating and intervention in the foreign exchange
markets.
Table 3
Impact of capital mobility and exchange rate regime

Fix*No Capmob
Fix*No Capmob* r*

Fix*Capmob
Fix*Capmob * r*

Float*No Capmob
Float*No Capmob* r*

Float*Capmob
Float*Capmob* r*

Rsq
DW statistic
Total observations
Cross sections
1

13

19752006

19751989

199099

200006

0.0002

0.0002

0.0005

0.0021

0.2954

0.3815

0.6640

0.8782

0.0002

0.0002

0.0076

0.0190

0.1794

0.2236

0.1529

0.3090

0.0001

0.0003

0.0004

0.0001

0.5580

0.4248

0.1314

0.0137

0.0017

0.0012

0.0028

0.0013

0.0220

0.1705

0.0756

0.0041

0.0002

0.0001

0.0011

0.0049

0.5320

0.8474

0.2391

0.5727

0.0003

0.0003

0.0043

0.0897

0.3442

0.4696

0.4327

0.1037

0.0004

0.0001

0.0008

0.0002

0.0191

0.7162

0.0204

0.0002

0.0014

0.0012

0.0041

0.0029

0.0057

0.0173

0.0212

0.0000

0.004

0.001

0.003

0.077

1.71

1.84

1.68

1.36

5,414

2,069

2,003

1,342

17

16

17

17

P-values are below the coefficients.

Here I distinguish between countries with low capital mobility (values 0 and 1) and those with high capital
mobility (values 2 and 3).

BIS Papers No 35

87

The results from the impulse response functions support the regression results. 14 During the
period 19752006, a 1% change in US interest rates leads to a 22.5 basis point change in
domestic interest rates in the next 10 months (Graph 1). 15 Here again, we would expect the
link between domestic and US interest rates to be higher during periods of fixed exchange
rates and/or high capital mobility. Graph 1 shows that the interest rate pass-through from the
US to emerging markets was about 70 basis points during 199099, a period characterised
by a de facto pegged regime. But as flexibility in exchange rates has increased, the response
rate has decreased to 30 basis points. The higher pass-through during the 1990s reflects the
fixed exchange rate regime in most of these economies and/or higher capital mobility.
Domestic interest rates also respond positively to global tightening (Graph A1). However, the
response during 200006 is half of that during 199099. The link can decline either because
of a greater willingness to let the exchange rate move or recourse to some other means than
monetary policy (ie foreign exchange intervention) to stabilise it or because some other
factors (exogenous to capital flows) are helping the exchange rate from falling. This may
reflect the recent phenomenon where, despite interest rate hikes in the United States, capital
still flowed to the emerging markets. Hence, these economies did not need to raise their
interest rates to the same extent to prevent capital outflows and depreciations. Rather, they
have been engaged in foreign exchange intervention to stabilise their exchange rates and
prevent them from appreciating. Bank of Thailand (2008) notes that, despite a stable interest
rate differential with the United States, the Thai baht has appreciated since 2004 because of
deterioration in market sentiment over the US twin deficit and hence of the dollar.
Consequently, large inflows into the region led to trend appreciation.
Graph 1
Impulse response of domestic interest rate to US interest rate
19752006

0 1 2 3 4 5 6 7 8 9 10

120

199099

120

200006

120

90

90

90

60

60

60

30

30

30

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

Capital mobility diminishes the ability of these economies to conduct an independent monetary
policy. 16 Countries with intermediate or no capital mobility have very little or an insignificant
link between the US interest rate and the domestic interest rate (Graphs A2 and A3). For

14

Here again, to check for robustness, I exclude the high-inflation periods for Argentina, Brazil and Chile, and
the results remain largely unchanged, except that the impulse responses for countries with low capital mobility
(Graph A2, 19752006) and with floating exchange rates and mobile capital (Graph 7, 199099) become
insignificant.

15

The response increases to 50 basis points during 19832006 (since the Fed officially started targeting interest
rates).

16

I create dummies for no, middle and high capital mobility. No capital mobility means that the value of the
capital mobility variable is 0; medium capital mobility is represented when capital mobility takes on the values
1 and 2. Full capital mobility means that the variable value is 3.

88

BIS Papers No 35

countries with full capital mobility, the interest rate pass-through was 70 basis points in
199099, but it declined to about 30 basis points in 200006 (Graph 2). Similarly, the results
from countries with fully liberalised interest rates show that the response of domestic interest
rates to US interest rates halved between 199099 (87 basis points) and 200006 (42 basis
points) (Graph A4).
Graph 2
Graph 2
Impulse response of domestic interest rate to
Impulse response of domestic interest rate to US interest rate for countries with full capital
US interest rate for countries
mobilitywith full capital mobility
19752006

0 1 2 3 4 5 6 7 8 9 10

199099

200006

100

100

100

50

50

50

50

0 1 2 3 4 5 6 7 8 9 10

50

0 1 2 3 4 5 6 7 8 9 10

50

The impulse response functions in Graphs 3 and 4 indicate that pegged regimes show a
greater interest rate pass-through (27 basis points) than flexible regimes (20 basis points) for
the entire sample (19752006). However, during the recent periods (199099 and 200006),
flexible regimes tend to exhibit greater co-movement with US interest rates (77 basis points
and 34 basis points respectively) than the pegged ones (60 basis points and 25 basis
points). This may reflect a fear of floating. While the classification of a country as a floater in
this paper is based on the actual behaviour of the exchange rate, central banks that float still
intervene in response to exchange rate movements that are perceived as excessive or to
accumulate foreign reserves (see BIS Papers no 24). If the exchange rate is not allowed to
adjust fully, domestic interest rates can still be affected. For example, if the foreign interest
rate falls and the exchange rate appreciates less than required to achieve equilibrium,
domestic interest rates can still fall even under a (de facto) floating regime. Although
domestic interest rates have responded less to US interest rates since 2000, they still exhibit
a high co-movement, implying that central bankers in emerging markets have still not gained
full autonomy over their monetary policies, despite adopting inflation targeting and moving to
exchange rate regimes that can be classified as flexible but not necessarily a free float.
The impulse responses in Graphs 58 shed light on what constrains monetary policy. 17, 18
For pegged regimes, the response of domestic interest rates to US interest rates is
significant during all periods for countries with high capital mobility (Graph 5) and only
significant during 199099 for those with low capital mobility (Graph 6). The high response
during 199099 may reflect the absolutely higher volume of capital flows, even for
economies with relatively low capital mobility, that may have required significant changes in
domestic interest rates to match the changes in US interest rates to maintain the exchange
rate pegs. In addition, most of these emerging markets experienced currency crises during
this period, which may have exaggerated the response.

17

Here I distinguish between countries with low capital mobility (values 0 and 1) and those with high capital
mobility (values 2 and 3).

18

There were insufficient data to estimate the impulse responses during 200006 for exchange rate regimes and
immobile capital.

BIS Papers No 35

89

Graph 3
Graph 3
Impulse response of domestic interest rate to
Impulse response ofUS
domestic
interest
rate toexchange
US interestrates
rate when
exchange rates are fixed
interest
rate when
are fixed
19752006

0 1 2 3 4 5 6 7 8 9 10

150

199099

150

200006

150

100

100

100

50

50

50

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

Graph 4
Graph 4
Impulse response of domestic interest rate to
Impulse response
domestic
interest
to US interest
exchange rates are
USofinterest
rate
whenrate
exchange
rates rate
are when
floating
floating

19752006

0 1 2 3 4 5 6 7 8 9 10

150

199099

150

200006

150

100

100

100

50

50

50

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

Graph 5
Graph 5
Impulse response of domestic interest rate to
Impulse response of domestic interest rate to US interest rate when exchange rates are fixed
US interest rate when exchange
rates
are fixed and capital is mobile
and capital
is mobile
19752006

0 1 2 3 4 5 6 7 8 9 10

90

199099

200006

100

100

100

50

50

50

50

0 1 2 3 4 5 6 7 8 9 10

50

0 1 2 3 4 5 6 7 8 9 10

50

BIS Papers No 35

We expect interest rates in countries with flexible exchange rates to be more independent of
US interest rates and this is the case, especially since 1990 (Graph 8) for countries with low
capital mobility. However, the link between the US interest rate and the domestic interest rate
is significantly higher in countries with high capital mobility (Graph 7), implying that capital
mobility may increase exchange rate fluctuations and induce central banks in these emerging
markets to move their interest rates with foreign interest rates to cushion these movements.
Graph 6
Graph 6
Impulse
response
of
domestic
interest
rateexchange
to
Impulse response of domestic interest rate to
US interest
rate when
rates are fixed
US interest rate when exchange
rates
are
fixed
and
capital
is immobile
and capital is immobile
19752006

199099

10

100

400

50

200

50

10

200

Graph 7
Graph 7
Impulse
response
of
domestic interest rate to
Impulse response of domestic interest rate to US interest rate when exchange rates are
US interest rate when exchange
rates
areisfloating
floating and capital
mobile and capital is mobile
19752006

0 1 2 3 4 5 6 7 8 9 10

199099

200006

100

100

100

50

50

50

50

0 1 2 3 4 5 6 7 8 9 10

50

0 1 2 3 4 5 6 7 8 9 10

50

To summarise, unlike Frankel et al (2004), Shambaugh (2005) and Obstfeld et al (2005), the
results in this paper suggest that domestic interest rates in countries with flexible exchange
rate regimes exhibit a higher co-movement with US interest rates than in countries with
pegged exchange rate regimes. This conundrum is resolved from Table 3 and Graphs 58,
which show the link between domestic and foreign interest rates under different exchange
rate regimes and levels of capital mobility. The response of domestic interest rates to
changes in US interest rates is higher for countries with flexible exchange rates and higher
capital mobility. Although exchange rates should undertake part of the burden of adjustment
to a foreign interest rate shock and hence we should observe a lower link between domestic
and foreign interest rates in case of floating regimes, this mechanism may be hampered

BIS Papers No 35

91

when central banks intervene. The intervention may not necessarily reduce the volatility in
the exchange rate to less than 2%, so that the exchange rate regime would still be
considered a de facto float according to the classification used in the paper. Indeed, the
probability that central banks intervene in a floating regime is about 40% in the period
2000-06 against 50% for 19752006. Another explanation is that the regimes classified as
floats also have better developed financial markets, where the pass-through from foreign
interest rates to domestic interest rates would be higher. These two factors could explain why
the floating exchange rate regimes have been associated with a bigger link between
domestic and foreign interest rates. Lastly, the estimation does not account for the possibility
that central banks might choose to move their domestic interest rates in line with foreign
interest rates due to fear of excessive volatility in foreign exchange markets or for business
cycle reasons.
Graph 8
Graph 8
Impulse response of domestic interest rate to
Impulse response of domestic interest rate to US interest rate when exchange rates are
US interest rate when exchange
areisfloating
floating andrates
capital
immobileand capital is immobile
19752006

199099
250
100

50

250

3.

10

50

10

500

Financial globalisation and its (dis)content

Since 2000, there has been an increasing trend towards flexibility in exchange rates and the
opening of capital accounts. For example, the proportion of observations with floating
exchange rates and mobile capital is 72% during 200006 against 23% with fixed regime and
mobile capital. These economies are also becoming financially globalised. Graph 9 shows
foreign assets and liabilities as a percentage of GDP for various regions. This variable has
grown at an unprecedented rate for all the regions. Ball (2006) argues that this form of
globalisation has implications for monetary policy because it affects the behaviour of both
interest rates and asset prices. Graph 10 shows that the five-year rolling correlations
between regional and G7 stock markets fell after the Asian crisis of 199798, but began to
rise again after the technology bubble burst in 2001. In fact, the correlations have ranged
between 0.8 and 0.95 in recent times.
This financial integration can be a boon or a bane, depending on the circumstances and the
kind of external positions these countries hold. Such internationalisation of portfolios is
desirable as it increases international risk-sharing, but at the same time shocks in one
country can be immediately transferred to foreign holders of financial instruments issued by
that country. Hence international events can have stronger domestic repercussions (Lane

92

BIS Papers No 35

and Milesi-Ferretti (2006b)). The recent MayJune 2006 sell-off is a testimony to the
vagaries of such shocks. 19 In the face of a slowdown of capital inflows, emerging markets
can also be more vulnerable to a crisis if they have debt liabilities denominated in foreign
currency (servicing costs rise with a depreciation) than if they rely on FDI (where returns are
linked to the performance of the domestic economy).

Graph99
Graph
1
Financial
Financial globalisation
globalisation1
175
Asia2, 3
Latin America2, 4
Central and eastern Europe2, 5
South Africa

1970

1975

1980

1985

1990

1,500
Hong Kong SAR
Singapore

150

1995

2000

1,250

125

1,000

100

750

75

500

50

250

25
2005
1970

1975

1980

1985

1990

1995

2000

0
2005

11

2
3
Foreign
assets
foreign
liabilities,
as a percentage
of GDP.
Simple
average
of or
thecited.
economies
Foreign
assets
plusplus
foreign
liabilities,
as a percentage
of GDP. 2 Simple
average of
the economies
listed
China, listed
India, or
3
4
4
Argentina,
Chile, Colombia,Argentina,
Mexico,
Indonesia, Korea,
Malaysia,
the Philippines,
Singapore,
Taiwanthe
(China)
and Thailand.
China,
India, Indonesia,
Korea,
Malaysia,
Philippines,
Taiwan
(China)Brazil,
and Thailand.
cited.
5
5
The Mexico,
Czech Republic,
Poland, Russia
and
Venezuela.
Peru andChile,
Brazil,
Colombia,
Peru Hungary,
and Venezuela.
TheTurkey.
Czech Republic, Hungary, Poland, Russia and
Turkey.
Source: Lane and Milesi-Ferretti (2006).

Source: Lane and Milesi-Ferretti (2006).

Graph 10
10
Graph
1
Emerging
stock
withthe
the
1
Emerging
stockmarkets
markets co-movement
co-movement with
G7G7
1.2

1.2
Emerging Asia
Latin America
Emerging Europe

0.8

0.8

0.4

0.4

0.4
0.8

0.4

1995

1996

1997

1998

1999

2000

2001

2002

2003

2004

2005

2006

0.8

Five-year
rolling
correlations
equity
prices
G7 and
the regions.
Thetoshaded
refer
Five-year
rolling
correlations
of equityofprices
between
thebetween
G7 and thethe
regions.
The shaded
areas refer
the Asianareas
crisis and
the to the
Asian crisis
the bursting
of the technology bubble, respectively.
bursting
of the and
technology
bubble, respectively.

Sources:
MSCI;
BIS calculations.
Sources:Datastream;
Datastream;
MSCI;
BIS calculations.

19

Basci et al (2008) show that the bond holdings of foreigners exhibited an upward trend after the 2001 Turkish
crisis. In the recent MayJune 2006 turbulence, the bond holdings of foreign investors declined sharply, while
stock portfolios remained unaffected. As a result, both interest rates and exchange rates increased sharply in
a very short time period.

BIS Papers No 35

93

Although not a problem for developed countries whose liabilities are denominated in
domestic currencies, but for emerging markets that are net debtors and whose external
liabilities are primarily denominated in foreign currency, an exchange rate depreciation raises
the domestic currency burden of foreign liabilities. The adverse effect of depreciation on the
valuation of the external position can outweigh the gains in trade accruing from such
depreciation. This is one reason why emerging markets are concerned about exchange rate
volatility (Lane and Milesi-Ferretti (2006b)). 20 And this may perhaps explain why they
intervene so frequently (40% of the time during 200006) even when the exchange rate is in
fact floating.
Can they borrow in their own currency?

In order to shield themselves from the vagaries of exchange rates, Lane and Milesi-Ferretti
(2006b) argue that emerging markets should promote local currency debt markets and
increase the role of FDI and portfolio equity inflows. In fact, some of these countries (Brazil,
Colombia, Thailand, the Czech Republic, Mexico and Malaysia) have been successful in
issuing domestic currency denominated bonds in the international market (Graph 11). 21

Graph 11
Local currency bonds and notes outstanding issued
1
in notes
international
markets
Local currency bonds and
outstanding
issued in international markets1
5,000
Brazil
Colombia
Thailand

2003
1

1 By

4,000

2004

2005

2006 Q3

1,000
Czech Republic
Mexico
Malaysia

800

3,000

600

2,000

400

1,000

200

2003

2004

2005

2006 Q3

nationality of issuers, in millions of US dollars; end of period.

By nationality of issuers, in millions of US dollars; end of period.

Source: BIS.

Source: BIS.

20

Lane and Milesi-Ferretti (2006a) maintain that, with the increase in gross assets and liabilities, the valuation
effects induced by changes in exchange rates and asset prices have become an important source of
fluctuations in countries external portfolios, often swamping the effects of the underlying capital flows.

21

The reasons for issuing global bonds vary. Tovar (2005) considers the issuance of global bonds in three Latin
American economies (Brazil, Colombia and Uruguay) and finds that the financial crises of the 1990s and early
2000s forced the governments to search for alternative financing to reduce their vulnerabilities. The crises also
gave impetus to structural reforms that attracted investors. Global factors have also aided the issuance of
bonds in domestic currency. For example, the success of emerging markets in reducing inflation in line with
the global trend and their growing integration with developed financial markets has broadened the range of
investors investing in emerging market securities.

94

BIS Papers No 35

Should exchange rate stability receive any attention? Issues in a dollarised economy?

As most emerging markets have moved towards floating exchange rates, monetary policy
requires an anchor. Hence, there has been a shift towards inflation targeting. 22 How far such
an approach can work in a dollarised economy is unclear; some have argued that, in such
circumstances, the exchange rate should enter the central banks objective function.
Calvo (2006) argues in favour of aiming at exchange rate stabilisation (to the extent of
outright pegging) during sudden stops and when there is liability dollarisation. 23 There is fairly
good evidence suggesting that dollarised countries have more fragile corporate sectors; are
more exposed to contractionary devaluations, devastating sudden stops and banking crises;
and exhibit more output volatility (see Ize and Levy-Yeyati (2006) for references). Hence, the
contractionary impact of real exchange rate depreciations limits the effectiveness of
countercyclical monetary policy under large shocks.
In fact, dollarisation of the Peruvian economy has made monetary policy under inflation
targeting less effective whenever there are stronger balance sheet effects (Rossini and
Vega (2008)). Exchange rate volatility can also create a problem for countries where
contracts are indexed in foreign currency. For example, rental contracts in Israel are indexed
to the dollar (a tradition from the days of hyperinflation that still exists), hence the exchange
rate pass-through into inflation of non-tradable goods is high and can lead to high costs in
the event of a sharp depreciation of the shekel (Eckstein and Soffer (2008)). Morn and
Winkelried (2003) argue that inflation targeting may be useful in guiding inflation expectations
but, a priori, not to solve liability dollarisation issues, and hence suggest that it might be
optimal to follow a non-linear policy rule that defends the real exchange rate in extreme
circumstances in a financially vulnerable economy.

4.

Conclusions

This paper presents the challenges faced by central banks in the face of highly mobile capital
flows. The trilemma states that in such circumstances countries cannot simultaneously
control their exchange rates and their interest rates. In order to gain monetary independence,
countries either have to adopt a free float or impose full capital controls. Since emerging
markets are moving towards higher capital mobility, they need to adopt a free floating
exchange rate regime in order to gain any monetary independence. The econometric results
from the paper indicate that although exchange rates have become more flexible in these
economies, they are nonetheless not free floats and accordingly the interest rates of these
economies do still respond to foreign rates to some degree. Nevertheless, the impulse
response functions show a decreased response of domestic interest rates to changes in US
interest rates since 2000, which might suggest that as these emerging economies gain
credibility with their newer forms of monetary policies (a move away from fixed to flexible
exchange rate regimes with inflation targeting), there may be further delinking between these

22

Under inflation targeting, central banks may be tempted to stabilise exchange rates even. For example, the
central bank of the Philippines intervenes to dampen sharp fluctuations in the exchange rate, which tend to
feed into domestic prices and hence affect both actual inflation and inflation expectations (Guinigundo (2008)).

23

According to Bernanke (2005), the combination of an inflation target, central bank independence and a
market-determined exchange rate tends to reduce variability in both inflation and output even in small open
economies like Finland and New Zealand. However, these economies are financially robust and not dollarised
like some emerging markets. Ball (1998) argues that, even for developed small open economies, if the
policymakers minimise a weighted sum of output and inflation variance, then the optimal policy instrument is a
Monetary Conditions Index based on both the interest rate and the exchange rate, while long-run inflation
(an inflation variable purged of the transitory effects of exchange rate fluctuations) should be the target
variable.

BIS Papers No 35

95

interest rates. Indeed, several central banks have stated that inflation targeting has helped
bring expectations of inflation down and the expectation channel is becoming stronger,
whereby stronger anticipated effects of monetary policy require less aggressive interest rate
changes. 24 Such credibility will help monetary policy become more independent of external
influences.
In addition, these economies have significantly increased their financial integration with the
global economy. Such changes can impact the transmission mechanism of monetary policy.
Most central banks have given up their exchange rates and moved towards formal/informal
inflation targeting. In such a scenario, central banks need to reassess the importance to be
assigned to exchange rate stability vis--vis price stability. Exchange rates may still play an
important role in these economies if they are dollarised or have a substantial part of their
debt denominated in foreign currencies. In order to increase the effectiveness of monetary
policy in such economies, de-dollarisation should be encouraged by developing local
currency debt markets and encouraging prices to be set in local currency (as in Peru).

24

96

See Sidaoui and Ramos Francia (2008) for Mexico, where the authors show that the expectations channel
helps monetary policy reduce inflation pressures with a reduced output cost.

BIS Papers No 35

Annex
Data construction: Monthly data on short-term interest rates are from IMF International
Financial Statistics. They cover 24 emerging markets from 1975 to 2006. 25 The variable on
capital mobility and interest rate liberalisation is from Omori (2005). These variables are
coded on a scale of 03. For capital mobility, 0 signifies a completely closed capital account
and 3 the most open. For interest rate liberalisation, 0 means completely controlled interest
rates, while 3 means market-determined. Since this dataset goes up to 2002, I extrapolate
the 2002 observation for the following years, assuming that capital account liberalisation and
interest rate liberalisation have neither progressed nor regressed from the 2002 level. The
dummy variable for the float was constructed in the same spirit as Obstfeld and Rogoff
(1995) and Shambaugh (2004). The idea is to see if the exchange rate remained within a
2% band in a given year. Hence, the dummy variable for the float takes the value of 1 if the
exchange rate is outside the 2% band over the last 12 months, otherwise it is 0. The US
interest rate (r*) is taken to represent the foreign or world interest rate. 26 The dummy for tight
US monetary policy takes the value of 1 when the US interest rate rises, otherwise it is 0.
The dummy for global tightness assumes the value of 1 if interest rates in the United States,
the United Kingdom and Japan rise at the same time, otherwise it is 0.

Table A1
Impact of global tightening1
19752006

19751989

199099

200006

0.08

0.05

0.43

0.18

0.00

0.00

0.00

0.00

0.03

0.03

0.01

0.20

0.53

0.54

0.96

0.01

Rsq

0.00

0.00

0.00

0.04

DW statistic

1.80

1.87

1.73

1.52

6,882

2,360

2,671

1,851

24

18

24

24

r*
Global Tight * r*

Total observations
Cross sections
1

P-values are below the coefficients.

25

Asia: CN = China; HK = Hong Kong SAR; ID = Indonesia; IN = India; KR = Korea; MY = Malaysia; PH = Philippines;
SG = Singapore; TH = Thailand; TW = Taiwan, China. Latin America: AR = Argentina; BR = Brazil; CL = Chile;
CO = Colombia; MX = Mexico; PE = Peru; VE = Venezuela. Others: CZ = Czech Republic; HU = Hungary;
PL = Poland; RU = Russia; SA = Saudi Arabia; TR = Turkey; ZA = South Africa.

26

Since eastern European countries may not be so linked with the United States, I re-estimate the regressions
without these countries. The results are robust.

BIS Papers No 35

97

Table A2
Impact of capital mobility
19752006

19751989

199099

0.00

0.00

0.22

0.18

0.95

1.00

0.56

0.33

0.00

0.00

0.00

0.00

0.51

0.93

0.40

0.28

0.04

0.03

0.07

0.13

0.05

0.14

0.62

0.04

Rsq

0.00

0.00

0.00

0.06

DW statistic

1.71

1.86

1.67

1.35

5,454

2,109

2,003

1,342

17

16

17

17

r*

Capmob
Capmob * r*

Total observations
Cross sections
1

200006

P-values are below the coefficients.

Table A3
Impact of exchange rate regime
19752006

19751989

199099

0.11

0.03

0.36

0.21

0.00

0.08

0.01

0.00

0.0003

0.00

0.001

0.0002

0.07

0.51

0.02

0.05

0.05

0.02

0.11

0.04

0.17

0.39

0.54

0.49

Rsq

0.00

0.00

0.01

0.04

DW statistic

1.80

1.87

1.74

1.52

6,862

2,320

2,671

1,871

24

18

24

24

r*

Float
Float * r*

Total observations
Cross sections
1

98

200006

P-values are below the coefficients.

BIS Papers No 35

Table A4
Impact of capital control and exchange rate regime
19752006

19751989

199099

0.00

0.01

0.07

0.34

0.91

0.71

0.87

0.09

0.00

0.00

0.00

0.00

0.57

0.72

0.86

0.16

0.05

0.04

0.12

0.15

0.05

0.12

0.40

0.02

0.0004

0.00

0.001

0.00

0.03

0.34

0.01

0.22

0.00

0.01

0.13

0.18

0.97

0.89

0.58

0.00

Rsq

0.00

0.00

0.01

0.07

DW statistic

1.72

1.84

1.67

1.38

5,414

2,069

2,003

1,342

17

16

17

17

r*

Capmob
Capmob * r*

Float
Float * r*

Total observations
Cross sections
1

200006

P-values are below the coefficients.

Graph A1
Impulse response of domestic interest rate to global tightening
19752006

0 1 2 3 4 5 6 7 8 9 10

BIS Papers No 35

120

199099

120

200006

120

90

90

90

60

60

60

30

30

30

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

99

Graph A2
Graph A2
of domestic
interest
rate
to
Impulse response ofImpulse
domesticresponse
interest rate
to US interest
rate for
countries
with no capital
US interest rate for countries
mobilitywith no capital mobility
19752006

30

199099

200
100

20

0
10
100
0

10

10

200

10

300

Graph A3
A3
Graph
Impulse interest
response
interest
to
Impulse response of domestic
rateoftodomestic
US interest
rate forrate
countries
with intermediate
US interest rate for countries
with
intermediate
level of
capital
mobility level of capital mobility
19752006

0 1 2 3 4 5 6 7 8 9 10

199099

200006

100

100

100

50

50

50

50

0 1 2 3 4 5 6 7 8 9 10

50

0 1 2 3 4 5 6 7 8 9 10

50

Graph A4
Graph A4
response
domestic
interest
rate
to interest rates are
Impulse responseImpulse
of domestic
interestofrate
to US interest
rate
when
US interest rate when
interest rates are market-determined
marketdetermined
19752006

0 1 2 3 4 5 6 7 8 9 10

100

150

199099

150

200006

150

100

100

100

50

50

50

0 1 2 3 4 5 6 7 8 9 10

0 1 2 3 4 5 6 7 8 9 10

BIS Papers No 35

Reference
Ball, L (2006): Has globalisation changed inflation?, NBER Working Paper 12687,
November, Cambridge, Massachusetts.
(1998): Policy rules for open economies, Working Paper 6760, October, Cambridge,
Massachusetts.
Bank of Thailand (2008): Changes in monetary transmission mechanism in Thailand,
BIS Papers, no 35, Bank for International Settlements.
Ba, E, zel and Sarkaya (2008): The monetary transmission mechanism in Turkey:
new developments, BIS Papers, no 35, Bank for International Settlements.
Bernanke, B (2005): Monetary policy in a world of mobile capital, Cato Journal, 25(1):
Winter: 112.
Calvo, G (2006): Monetary policy challenges in emerging markets: sudden stops, liability
dollarization and lender of last resort, paper presented at the 10th Annual Conference at the
Central Bank of Chile, Santiago, 910 November 2006.
Eckstein, Z and Y Soffer (2008): Exchange rate pass-through implications for monetary
policy: the Israeli case, BIS Papers, no 35, Bank for International Settlements.
Frankel, J, S Schmukler and L Serven (2004): Global transmission of interest rates:
monetary independence and currency regime, Journal of International Money and Finance,
23(5): 701734 (September).
Guinigundo, D (2008): Transmission mechanism of monetary policy in the Philippines,
BIS Papers, no 35, Bank for International Settlements.
Ize, A and E Levy-Yeyati (2006): Financial de-dollarization: is it for real?, in Financial
dollarization: the policy agenda, (eds) A Armas, A Ize and E Levy-Yeyati, Palgrave
MacMillan, NY.
Jeanneau, S and C Tovar (2006): Domestic bond markets in Latin America: achievements
and challenges, BIS Quarterly Review, June, 5164.
Kose, M, E Prasad, K Rogoff, and S Wei (2006): Financial globalization: a reappraisal, IMF
Working Paper 06/189, Washington DC.
Lane, P and G Milesi-Ferretti (2006a): The external wealth of nations mark II: revised and
extended estimates of foreign assets and liabilities, 19702004, IMF Working Paper 06/69,
Washington DC.
(2006b): Examining global imbalances, Finance and Development, 43(1): March,
Washington DC.
Moreno, R (2008): Monetary policy transmission and the long-term interest rate in emerging
markets, BIS Papers, no 35, Bank for International Settlements.
Morn, E and D Winkelried (2003): Monetary policy rules for financially vulnerable
economies, IMF Working Paper 03/39, Washington DC.
Obstfeld, M and K Rogoff (1995): The mirage of fixed exchange rates, Journal of Economic
Perspectives, IX, 7396.
Obstfeld, M, J Shambaugh and A Taylor (2005): The trilemma in history: tradeoffs among
exchange rates, monetary policies and capital mobility, Review of Economics and Statistics,
(August), 87(3): 42338.
Omori, S (2005): Financial reform database: measuring seven dimensions of financial
liberalization (Washington: International Monetary Fund), unpublished manuscript.

BIS Papers No 35

101

Ooi, Sang Kuang (2008): The monetary transmission mechanism in Malaysia: current
developments and issues, BIS Papers, no 35, Bank for International Settlements.
Romer, C and D Romer (1989): Does monetary policy matter? A new test in the spirit of
Friedman and Schwartz, NBER Macroeconomics Annual 4, 121170.
Rossini, R and M Vega (2008): The monetary policy transmission mechanism under
financial dollarization: the case of Peru 19962006, BIS Papers, no 35, Bank for
International Settlements.
Shambaugh, J (2004): The effect of fixed exchange rates on monetary policy, Quarterly
Journal of Economics, 301352 (February).
Sidaoui, J and M Ramos Francia (2008): The monetary transmission mechanism in Mexico:
recent developments, BIS Papers, no 35, Bank for International Settlements.
Tovar, C (2005): International government debt denominated in local currency: recent
developments in Latin America, BIS Quarterly Review, December, 10918.

102

BIS Papers No 35