Trade Balance -1

8/12/2019 Trade Balance -1

1/25

Sources of time-varying trade balance and real

exchange rate dynamics in East Asia q

Sohrab Rafiq

International Monetary Fund, United States

a r t i c l e i n f o

Article history:

Received 13 June 2012

Revised 22 May 2013

Available online 13 July 2013

JEL classification:

F14

F32C5

Keywords:

Exchange rate

Trade balance

Time-variation

Correlation

Asia

a b s t r a c t

Rafiq, SohrabSources of time-varying trade balance and real

exchange rate dynamics in East Asia

A sticky-price model with minimal assumptions for identification

is used to motivate a time-varying model that allows for state

dependent innovations to explore the trade balance dynamics of

a group of East Asian economies. This paper shows that the corre-

lation between the trade balance and the real exchange has histor-ically been highly conditional on the type of macroeconomic shock.

Permanent (transitory) shocks have historically produced a posi-

tive (negative) correlation between the trade balance and real

exchange rate over the last 20 years. Second, since the Asian finan-

cial crisis the real exchange rate dynamics of the East Asian coun-

tries have been dominated by persistent component(s), while the

dynamics of the trade balance have been more influenced by tran-

sitory factors. J. Japanese Int. Economies 29 (2013) 117141. Inter-

national Monetary Fund, United StatesInternational Monetary

FundUnited States.

2013 Elsevier Inc. All rights reserved.

1. Introduction

The tight management of the exchange rate in many Asian economies is often seen as part of an

export-led development strategy. However, the trade balance surpluses accumulated over recent

0889-1583/$ - see front matter 2013 Elsevier Inc. All rights reserved.

http://dx.doi.org/10.1016/j.jjie.2013.06.001

q The views expressed herein are those of the author and should not be attributed to the International Monetary Fund (IMF),

its Executive Board, or its management. The author would like to thank an anonymous referee for comments. The usual

disclaimer applies. Address: 700, 19th Street, N.W., Washington, DC 20431, United States.

E-mail address:[email protected]

J. Japanese Int. Economies 29 (2013) 117141

Contents lists available at ScienceDirect

Journal of The Japanese and

International Economiesj o u r n a l h o m e p a g e : w w w . e l s e v i e r . c o m / l o c a t e / j j i e
http://dx.doi.org/10.1016/j.jjie.2013.06.001mailto:[email protected]://dx.doi.org/10.1016/j.jjie.2013.06.001http://www.sciencedirect.com/science/journal/08891583http://www.elsevier.com/locate/jjiehttp://www.elsevier.com/locate/jjiehttp://www.sciencedirect.com/science/journal/08891583http://dx.doi.org/10.1016/j.jjie.2013.06.001mailto:[email protected]://dx.doi.org/10.1016/j.jjie.2013.06.001http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://crossmark.crossref.org/dialog/?doi=10.1016/j.jjie.2013.06.001&domain=pdfhttp://-/?-


2/25

years has led to large global macroeconomic imbalances vis--vis much of the advanced economies. In

the context of multilateral consultations and G20 meetings have given rise to questions regarding

whether large current account surplus countries should be called upon to appreciate their currencies.

Much of the focus has been on Chinas large trade balance surplus. This paper examines the trade bal-

ance dynamics of five large current account surplus East Asian economies: China, South Korea, Japan,Malaysia and Taiwan (seeFig. 1).

Understanding factors driving fluctuations in Asian trade is important for domestic policy consid-

eration and international policy coordination. The analysis also contributes to the question of whether

Asian exchange rate dynamics are better characterised by equilibrium or disequilibrium models.1

Recent analyses conducted in the framework of the new generation of macroeconomic models, while

providing a high degree of theoretical rigor to the debate, has often led to dissatisfying divergent results

(Chinn and Lee, 2007). Without claiming full generality, this paper adopts a parsimonious method to

decompose shocks in a manner relevant to the current global imbalance debate. This paper investigates

the evolution of key trade imbalances and real values of currencies by relying upon a minimalist set of

long-run identifying assumptions that are consistent with a wide range of models. With the role of the

exchange rate being a focal point for the Asian economies, shocks are decomposed according to theirlong-run effect on the real exchange rate.

The paper complements other papers that adopt a long-run identification strategy to study the ex-

change rate or current account.Clarida and Gali (1994),for example, applied a long-run identification

strategy to explore the role of monetary shocks in the exchange rate fluctuations, in a three-variable sys-

tem comprising bilateral real exchange rate, inflation differential, and relative output. In related work,

Chadha and Prasad (1997), Prasad (1999), Astley and Garratt (2000) and Lee and Chinn (2006) examine

the role of transitory and permanent shocks on the current account balances of the G7 economies.

These studies have been subject to criticism. Grier and Ye (2009)showed in a fixed coefficient mod-

el that without taking into account structural shifts (particularly in the mean) between the real ex-

change rate and the trade balance leads to spurious conclusions.2 Furthermore, they also show that

without accounting for changes in conditional heteroskedasticity in the dynamics of the trade balance

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010-14

-12

-10

-8

-6

-4

-2

0

2

4

6x 10

5

$

China, Japan, South Korea, Taiwan and Malaysia

US, UK and Euro zone (minus Germany)

Fig. 1. Global trade imbalance.

1 SeeDornbusch (1976)for an example of a disequilibrium exchange rate model, and Stockman (1987)of equilibrium models.2 Also seeRafiq (2010).

118 S. Rafiq / J. Japanese Int. Economies 29 (2013) 117141


3/25

would lead the ordinary least-squares estimate of the VAR coefficients to be inefficient, with the

subsequent variance decompositions and impulse response functions being sub-optimal. In the exchange

rate-trade balance literature, thus far, there has been little research accounting for such shifts in a unifiedmodel. Data for two Asian countries China and Malaysia inFig. 2show structural shifts in the mean of

the trade balance as well as the real exchange rate. Such shifts need to be accounted for in any structural

framework. The potential presence of structural instabilities may make the analysis of the trade balance

and real exchange rate in fixed coefficient models sample dependent.

In order to investigate the link between the real exchange rate and the trade balance, uniquely in

the literature, this paper marries a minimalist set of identifying long-run assumptions to an extended

by allowing for time-variation in the coefficients and stochastic volatility vector autoregression

(TVPVAR) model. The varying coefficients are meant to capture smooth transitions in the propagation

mechanism of exchange rate and trade balance shocks without imposing a specific breakpoint, while

the stochastic volatility component models change in the magnitude of structural shocks and their

immediate impact. The specification proposed in this paper has the advantage of being able to pickup subtle progressive long-run structural changes in the economy, shifts in the preferences of policy-

makers, as well the effect of exchange rate changes in the presence of state dependent macroeconomic

shocks, such as commodity prices disturbances, on the trade balance.

Macroeconomics typically focuses on overall trade balances rather than bilateral balances, which

has tended to be the primary focus in the political sphere; the ChinaUS trade balance being a prime

example. One way to illustrate this is that given a Chinese revaluation, the US trade deficit might be

re-apportioned to other countries. This paper arrives at a number of findings:

This paper finds that transitory (nominal) shocks have dominated trade balance, whilst permanent

(real) shocks have been the primary driver behind real exchange rate variation over the last

25 years. The time-varying results over the last 25 years for the real exchange rate are consistentwith equilibrium exchange rate models, and the observed time-series behaviour of the real

exchange rates for the Asian economies.

1990 2000 2010

0

2

4

x 104

$

China trade balance

1990 2000 2010

0

5000

10000

$

Malaysia trade balance

1990 2000 2010

100

150

200

Index

Malaysia REER

1990 2000 2010

100

150

200

Index

China REER

Fig. 2. Structural shift in the trade balance and real exchange rate.

S. Rafiq / J. Japanese Int. Economies 29 (2013) 117141 119
http://-/?-


4/25

Bivariate unconditional correlations and other unconditional measures are inadequate for

characterising the cyclical dynamics of the trade balance and the real exchange rate relation-

ship. This cautions against anticipating a specific correlation the trade balance and the real

exchange rate. The continuously changing roles of temporary and permanent shocks illustrated

by the time-varying conditional correlation estimates offer an explanation for the difficulty in

empirical attempts to uncover a stable relationship between the real exchange rate and the

trade balance.

The time-varying response functions show that real shocks, rather than nominal ones, will be

more effective in reducing trade imbalances. This implies the income channel in trade balance

adjustment is important. Permanent supply shocks have, in general, produced a positive rela-

tionship between the trade balance and real exchange rate over the last two decades. This

implies that supply shocks in the region mostly reflect preference rather than productivity

innovations.

This paper is organised as follows. Section 2lays out an open economy macro model, which ac-

counts for the dynamics of the trade balance. Section 3provides an exposition of the time-varying

model with long-run restrictions. Sections4 and 5illustrate the results from with model, with Sec-tion6 concluding.

2. Benchmark open economy model with trade dynamics

In order to derive some theoretically meaningful econometric estimates of the dynamics of the

trade balance one needs to begin with an analytical framework. The theoretical framework is based

onPrasad (1999), which is turn is a generalised version of the stochastic open economy models in

Obstfeld (1985), Clarida and Gali (1994) and Lee and Chinn (2006). The model is essentially a stochas-

tic version of the MundellFleming framework, which has been extended to incorporate sticky prices.

Thus, the advantage of the model is that it incorporates key assumptions and predictions from stan-

dard New Open Economy Macro (NOEM) models in the current literature.

yd dt gstpt rit Etpt1pt 1

pt 1 hEt1pethp

et 0< h 6 1 2

mtptyt cit 3

itEtst1st 4

The nominal exchange rate is denoted st, the price level is signified by pt,itis the nominal interest rate.

petis the flexible price level and m

tdenotes the money supply. Demand shocks are captured by d

twith

overall aggregate demand denoted relative to foreign demand denoted as yd. Eq.(1) represents a tra-

ditional IS curve. Eq.(2)measures price adjustment relative to its flexible price equilibrium. The speed

of adjustment is captured by h. Eq.(3)is a LM equation and Eq.(4)represents the uncovered interest

rate parity condition.

The trade balance is determined by real output and the real exchange rate

tbtwqtayt 5

where w and a denote the elasticities of the trade balance with respect to the exchange rate and rel-ative output. This assumption is consistent with partial equilibrium trade equations, which implies

that the composition of domestic output would be solely determined by the exchange rate in a world

characterised by perfect business cycle symmetry.3

Three stochastic shocks are introduced into the model. The three stochastic processes driving the

model are expressed as follows:

3 SeeCheung et al. (2010).



5/25

ystyst1

st Supply shock

dtdt1dt #dt1 Demand shock

mtmt1nt1 Nominal shock

The innovations st; dt and

nt are assumed to be serially and mutually uncorrelated. Demand shocks

may be interpreted as fiscal shocks and nominal shocks as monetary shocks. In Betts and Devereux(1996) and Chari et al. (1997) pricing-to-market effects cause monetary shocks to fluctuate the real

exchange rate in the short-run but not in the long-run. Therefore, identification is consistent with a

broad class of open economy macro models. A second advantage is that the model distinguishes be-

tween nominal and demand shocks.4

The model can be solved for the long-run flexible-price rational expectations equilibrium

yet yst 6

qet yst dt

g

1

gg rrcdt1 7

tbet y

st

w

ga

w

g dt

1

g rrcdt1

8

In the long run, relative output yet

, the real exchange rate qet

and the trade balance tbet

, are essen-

tially driven by two shocks: supply st

demand dt1

. These equations imply three long-run restric-

tions. Nominal or demand shocks do not have any long-run effect on relative real output. Additionally,

nominal shocks do not influence the long-run level of the real exchange rate. This does not preclude

nominal disturbances perhaps reflecting changes in monetary policy or expectational shocks from

influencing short-run movements in the exchange rate. The imposition of these restrictions on a time-

varying econometric framework are examined next.

3. Time-varying parameter model with long-run restrictions

To keep the analysis tractable in a TVPVAR model this paper develops a three-equation macro

model. The use of a VAR model over simple single equation regression estimation to examine factors

determining the trade balance should help reduce simultaneity bias, and allow for feedback effects.

The time-varying reduced form model is expressed as

yta0;t /1yt1 /pytp etX0th et 9

whereytis ak 1 vector of observations of the dependent variables, with h a k(kp+ 1) vector con-

taining the VAR reduced-form coefficients (/i) and the constant term a0; XtIk 1;y0t1;. . . ;y

0tp

h iand etis ak 1 vector of unobservable heteroskedastic disturbance terms with zero mean. The vector

ethas a time-varying covariance matrix

Varet Xt A1t HtA

1t

010

The matrixAtis a lower triangular matrix that models the contemporaneous interactions among the

endogenous variables and Htis a diagonal matrix which contains the stochastic volatilities,

At

1 0 0

a21;t 1 .

.

.

.

.

....

..

.0

ak1;t akk1;t 1

0BBBBB@

1CCCCCA

Ht

h1;t 0 0

0 h2;t .

.

.

.

.

....

..

.0

0 0 hk;t

0BBBBB@

1CCCCCA

11

4 Lastrapes (1992) and Robertson and Wickens (1997), for instance, lump nominal and demand shocks together and are referred

to as aggregate demand shocks.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


6/25

Cogley and Sargent (2005) showed that ignoring movements originating from the heteroskedastic

covariance structure would be picked up by the VAR coefficients leading to an upward bias.

htht1 vt

at

at1

nt

lnhi;tln hi;t1 gt

etvt

nt

gt

0BBB@

1CCCA N 0;

Xt 0 0 0

0 Q 0 0

0 0 S 00 0 0 Z

0BBB@

1CCCA

0BBB@

1CCCA 12

FollowingPrimiceri (2005)the time-varying parameters of the model,ht= {a0,/i,t}, evolve as driftless

random walks. Theatparameters, which also evolve as driftless random walks, are non-zero and non-

one elements of the matrixAtstacked by rows. The ln hi,tevolve as geometric random walks, indepen-

dent of one another, and contain the diagonal elements ofHt. The priors for the initial states of the

regression coefficients, the covariances and the log volatilities p(h0),p(a0) andp(lnh0) are assumed

to be normally distributed and independent of each other.

3.1. Structural identification Long-run restrictions

The model is composed of three variables DXt= [Dyt,Dqt,Dtbt]. The data runs from 1981:1 till

2011:2 for all countries. The theoretical model expresses output relative to foreign output. Hence,

an index of external demand is constructed by taking the trade-weighted average of real GDP of the

main trading partners of each country. The index is then subtracted from the index of domestic output.

This variable implicitly controls for external demand. The trade balance, tbt, is expressed as a ratio of

total output in order to control for scale effects. Finally, the real exchange rate,qt, is the trade weighted

real effective exchange rate, which is consumer price index based. The exchange rate index is taken

from the Bank of International Settlements.

Since the stochastic process of the model is driven by three random walk processes suggests that

the three series are nonstationarity in levels but stationary in first differences. Additionally, that real

output, the real exchange rate and the trade balance are driven by different shocks implies that theseries are not cointegrated. To uncover the structural shocks identification is achieved by imposing

long-run constraints on the long-run multipliers in the system, leaving the short-run dynamics uncon-

strained. The approach is based onCanova and Gambetti (2010). Re-writing Eq.(1)in companion form

xt lt Atxt1 ut 13

wherext yt;y0t1;. . . ;y

0tp1

h i0; ut e0t; 0;. . . ; 0

; lt a0;t; 0;. . . ; 0andAtis the corresponding com-

panion matrix. FollowingCanova and Gambetti (2010)a local approximation of the implied response

of t+k is estimated. The companion form is transformed into its vector moving average (VMA)

representation

xt I AtL1

lt ut 14

This implies that the rows and columns ofAt identify the impulse response at t= 1, . . .,k to innova-

tions et. The impulse response is given by

@ytk@e0t

ei;jAkt At;k 8kP 0 15

where ei,j() is a selector function which selects the first rowi and column j and where Bt,0 I. Thek-period impulse response to structural shocks hitting the economy at time tare given by

@ytk@e0t

@ytk@e0t

@ytk@u0t

At;kX0:5t Ct;k 8kP 0 16

for k = 0,1, 2, . . ., 20.

DefineAt1 P1k0At;k and Ct1 P1k0Ct;k, which measure the long-run effect on variable xt. Thestochastic variances ofxt in the companion form is given by



7/25

Varxt At1 Xt 0

0 0

At1

0

Therestrictionsimplied by thetheoreticalmodel that nominal anddemandshocks have no long-run

effect on the level of output and that nominal shocks do not have a permanent effect on the level of the

real exchange rate implies that the matrix of long-run cumulative multipliers Ct(1) is lower triangular.This impliesCt(1) is uniquely identified by a Cholesky decomposition, which can be represented via

Varxt X1k0

At;k

!Xt

X1k0

At;k

!017

Varxt X1k0

Ct;k

!X1k0

Ct;k

! Ct1Ct1

018

The impulse responses to a structural shock at t+k= 1, . . ., 20 can be expressed through exploiting

Ct1 At1X0:5t

@ytk@e0t

At;kAt11Ct1 8kP 0 19

The model implies that the dynamics capture changes in the propagation of structural shocks. Simul-

taneously, by incorporating heteroskedastic shocks, the volatility in the underlying series are ac-

counted for. In sum, the model accounts for the possibility of structural breaks in the dynamics that

characterise the real economy whilst allowing policymakers to continuously update its knowledge

about the macroeconomic environment. The model is estimated following Primiceri (2005), Benati

and Mumtaz (2007) and Canova and Gambetti (2010). Details are provided in theAppendix.

4. Decomposing time-varying trade balance and exchange rate movements

This section presents the estimates from the time-varying model. Since the parameters are time-

varying the estimates are illustrated across the entire sample. The estimates run from 1987:1 till

2011:2. The first six years of the original sample are discarded as they are used to calibrate the priors

(seeAppendix).

The analysis focuses on the second moments conditional and unconditional of the trade bal-

ance, the real exchange rate and real output. Each variable is expressed as a time-varying distributed

lag of the three structural disturbances. Thus, lettingx i,trepresent one of the three variables

xi;tX1k0

Ci;edt;k

2X1k0

Ci;es

t;k

2X1k0

Ci;en

t;k

2From the estimates of the coefficient of the distributed lags it is possible to construct time-varying

measures of unconditional and conditional second moments of the three variables in the model.

The unconditional variance and covariance at timetof variable xi,tis calculated as:

Varxi;t X1k0

Ci;edt;k

2X1k0

Ci;est;k

2X1k0

Ci;ent;k

2Unconditional variance

where the three terms on the right represent the contribution of the three shocks demand, supply

and nominal to that variance, or the variances conditional on each of the shocks. The covariance at

timetof variable x i,tis calculated as:

Covxi;t;xj;t X1

k0

Ci;es

t;k 2

Cj;ed

t;k 2

Unconditional covariance

X1k0

Ci;est;k

2Cj;ent;k

2X1k0

Ci;edt;k

2Cj;ent;k

2

http://-/?-http://-/?-http://-/?-http://-/?-


8/25

where the three terms on the right representing the covariances at time tconditional on demand, sup-

ply and nominal shocks.

4.1. Conditional volatilities: Trade balance and exchange rate

A complementary perspective on the relative importance of shocks for trade balance dynamics is toexamine the importance of the derived structural innovations for the variance of the trade balance and

real exchange rate.

Fig. 3(a) shows that pre-2000, with Taiwan being the outlier, nominal shocks tended to be the most

important factor driving trade balance volatility for the Asian economies. To the extent that some of

the trade balance surplus of the Asian economies is driven by temporary (in this case nominal) shocks,

its correction will be accompanied by an appreciation in the real exchange rate. However, the uncon-

ditional volatilities for the trade balance inFig. 3(a) the estimates show that, with the exception of Tai-

wan, since early 2000 trade balance volatility has increased. In terms of magnitude, the largest rise in

volatility is reported for China. The time-varying estimates show that much of this rise for China is

down to an increase in the size of demand, supply and nominal shocks. Post-2005 there was a partic-

ularly large rise in the importance of supply (permanent) shocks in driving Chinese trade balance var-iation. Again, with the exception of Taiwan, all countries report a rise in the size of supply shocks for

trade balance variation.

Moving onto the dynamics of the real exchange rate, apart from Japan, changes in the uncondi-

tional volatility of the real exchange rate have not, in general, moved in phase with the trade balance.

Fig. 3(b) shows that the unconditional volatility in the Chinese real exchange rate has been lower than

that for the trade balance. This is consistent with the controlled exchange rate regime in China, sug-

gesting that the adjustment of the economy in the face of adverse shocks falls on the real side of the

economy. For Japan and Malaysia conditional real exchange rate volatility has been greater than the

trade balance. This is consistent with the relatively freer exchange rate regimes employed in these

countries since the early 1990s.

Second, South Korea and Malaysia experienced a large rise in the unconditional volatility of the realexchange rate during the Asian financial crisis in 1997/1998. In both cases the rise in volatility was the

result of large demand shocks. The time-series data shows that both countries endured a very persis-

tent downward (devaluation) jump in their real effective exchange rates going forward.

Third, with the exception of Taiwan, demand shocks have tended to dominate in driving real ex-

change rate volatility since the late 1980s. In Taiwan, nominal (transitory) shocks have tended to play

a more prominent role in real exchange rate fluctuations. That the Chinese, Korean and Japanese real

exchange rate dynamics are dominated by permanent (demand and supply) shocks is consistent with

what would be expected from the time series literature on exchange rate behaviour. The real exchange

rates for these countries demonstrate high persistence and long-run mean shifts, which make them

non-stationary.5 From a long-run purchasing power parity (PPP) perspective the real exchange rate

should be stationary.6

Prasad (1999)reports an important role for nominal shocks in driving the variance of the trade bal-

ance. Results for the G7 countries in Lee and Chinn (2006) also showed permanent shocks to have been

more important than temporary ones for the real exchange rate.Clarida and Gali (1994)based on the

MundellFlemmingDornbusch model that demand shocks dominate in the forecast error variance of

the real exchange rate for the G7 countries. Clarida and Gali (1994)discussed the role of supply, de-

mand, and monetary shocks, and found that the demand shock is the most important source of real

exchange rate fluctuations, which is consistent with our results.

4.2. Correlation between the trade balance and exchange rate

The relationship that tends to get most attention in the global imbalances literature is the tradebalance-real exchange rate relationship. Another perspective on the likely evolution of trade imbal-

5 In early workMesse and Rogoff (1983)found that a substantial fraction of real exchange rate fluctuations are quite persistent.6 Clarida and Gali (1994) also found that nominal shocks were unimportant in determining real exchange rate fluctuations.



9/25

1990 1995 2000 2005 2010

1

2

3

4

5

6

7

8

1990 1995 2000 2005 2010

0.2

0.4

0.6

0.8

1

1.2

1990 1995 2000 2005 2010

0.60.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

2.6

1990 1995 2000 2005 2010

0.5

1

1.5

2

2.5

1990 1995 2000 2005 2010

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

2.2

2.4

1990 1995 2000 2005 2010

1.5

2

2.5

3

3.5

4

4.5

5

5.5

6

6.5

Fig. 3. Variance decomposition for the trade balance and exchange rate: supply (blue), demand (blue), nominal (green),

unconditional standard deviation.

http://-/?-


10/25

ances with respect to the exchange rate has been provided through unconditional correlations, often

viewed through the prism of event studies.7 This section also reports on the time-varying unconditional

correlation. Contributing to the literature this paper goes onto derive time-varying correlations between

the trade balance and real exchange rate conditional on shocks relevant to the current debate. The his-

torical correlations between the trade balance and the real exchange rate generated by the three shocks

identified from the model demand, supply and nominal are illustrated in Fig. 4.

4.2.1. Trade balance and real exchange rate

From a simple income-absorption perspective it is expected that the correlation between the trade

balance and real exchange rate is negative.8 Lee and Chinn (2006)found that the shocks with only tem-

1990 1995 2000 2005 2010

0.5

1

1.5

2

2.5

3

1990 1995 2000 2005 2010

1

1.5

2

2.5

3

3.5

4

4.5

1990 1995 2000 2005 2010

0.5

1

1.5

2

2.5

3

1990 1995 2000 2005 2010

0.04

0.06

0.08

0.1

0.12

0.14

Fig. 3 (continued)

7 SeeGalati and Debelle (2005) and Croke et al. (2006).8

Theoretically, if foreign currency prices are assumed to remain fixed (full path-through of exchange rate changes onto importprices), import prices will decrease one-for-one in response to an exchange rate revaluation. The relative price of imports to the

price level of all domestically consumed goods will fall in lower proportion since the consumer price index also falls after a

currency revaluation. Still, the lower relative price of imports induces substitution away from domestic products towards imports

so that the volume of imports increases. The fixed foreign currency prices and the volume increase unambiguously leads to an

increase in the foreign currency value of imports, reducing the current account surplus.



11/25

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Fig. 4. Conditional correlations for the trade balance and exchange rate: supply (blue), demand (blue), nominal (green),

unconditional correlation.

http://-/?-


12/25

porary exchange rate effects brought about a negative correlation between the current account and the

real exchange rate for the G7, while shocks with permanent exchange rate effects did not necessarily

bring about such a negative correlation.For China, Japan and Taiwan the time-varying unconditional correlation between the trade balance

and the real exchange rate has hovered around zero for the last two decades. For South Korea the

unconditional correlation reached 0.4 in 1999, just after the country experienced a large exchange

rate devaluation due to the Asian financial crisis in 1997/1998. Since then the size of the unconditional

correlation coefficient has progressively approached zero. Finally, from the late 1990s, the uncondi-

tional correlation has become progressively more negative for Malaysia. For instance, in 1999 the

unconditional correlation between the trade balance and the real exchange rate was around 0.05.

By the end of the sample in 2011 the analogous figure was around0.5.

The conditional correlation estimates between the trade balance and the real exchange rate helps

provide more nuance to the reduced form correlation estimates. The estimates show that since the

early 1990s, with the exception of Taiwan, nominal (transitory) shocks have tended to induce a neg-

ative correlation between the trade balance and the real exchange for all countries. The correlation

generated by a nominal shock invites us to interpret it as a monetary shock. A money shock induced,

for example, by quantitative easing depreciates the currency so much that the trade balance im-

proves, inducing a negative correlation. For Japan, only nominal shocks have historically precipitated

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

1990 1995 2000 2005 2010-1

-0.8

-0.6

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Fig. 4 (continued)



13/25

the expected negative correlation between the trade balance and the real exchange rate over the last

20 years. In terms of magnitude, a positive nominal demand shock pre-financial crisis induced a neg-

ative correlation between the trade balance and the real exchange rate of between0.8 and0.6 dur-

ing 1990 till 2007. Since the financial crisis, however, nominal shocks have stopped generating a

negative coefficient value for Japan. Finally, much of the increased negativity in the unconditional cor-

relation between the real exchange rate and trade balance for Malaysia is due to the increased impor-

tance of nominal shocks. By 2010 the correlation coefficient induced by a nominal shock was around

0.8, compared with a value of zero in 1994.The estimates for China illustrate the importance of allowing for time-variation. For much of the

1990s nominal and supply shocks induced a negative correlation between the trade balance and

the real exchange rate, whilst demand shocks did the opposite. In terms of numbers, the correlation

peaked at 0.5 for nominal and 0.6 for supply innovations during the mid-to-late 1990s. Since

2000, however, these trends have slowly reversed. By 2011 a demand shock induced a negative cor-

relation of around 0.4, while nominal shocks went to positive 0.4. In traditional Keynesian analysis, a

demand shock induced, for example, by a fiscal expansion leads to a real exchange rate appreci-

ation that deteriorates the trade balance. Since the early 1990s demand shocks have been the only

type of shock that has generated a negative coefficient for Taiwan over the last two decades.

Based on the theoretical model outlined a supply shock should induce a permanent depreciation in

the real exchange rate, which in turn should improve the trade balance. With the exception of Japan,consistent with theory outlined, supply shocks have produced a negative correlation between the

1 5 9 13 17

-0.4

-0.2

0

0.2

0.4

0.6

0.8

1

Supplyshock

1 5 9 13 17

-1.4

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

Demandshock

1 5 9 13 17

-0.4

-0.2

0

0.2

0.4

0.6

1 5 9 13 17

-1

-0.5

0

0.5

1

1 5 9 13 17

-1.5

-1

-0.5

0

0.5

1 5 9 13 17

-1.2

-1

-0.8

-0.6

-0.4

-0.2

0

0.2

1 5 9 13 17

-0.05

-0.04

-0.03

-0.02

-0.01

0

0.01

1 5 9 13 17

-0.03

-0.02

-0.01

0

0.01

0.02

0.03

JapanChina S. Korea Taiwan

Fig. 5. Trade balance dynamics (fixed coefficient model).

http://-/?-


14/25

trade balance and the real exchange rate. Temporary shocks (in this case nominal shocks) find an easycandidate in monetary shocks.9 However, permanent shocks in this case supply shocks are more dif-

2 4 6 8 10 12 14 16 18 20

-1

-0.8

-0.6

-0.4

-0.2

0

0.2 pre-1999post-1999

2 4 6 8 10 12 14 16 18 20

0.2

0.4

0.6

0.8

1

1.2

2 4 6 8 10 12 14 16 18 20

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

2

Fig. 6. Trade balance dynamics.

9 These are often viewed as having only temporary effects on the real exchange rate.



15/25

ficult to pin down. Typically, permanent shocks are associated with productivity innovations. This is of-

ten thought of as inducing a negative correlation between the trade balance and the exchange rate. In

contrast, Blanchard et al. (2005) provide an analysis of preference shocks that have permanent effects

on the real exchange rate. A preference shock in favour of home exports would have a long-run effect

Fig. 6(continued)



16/25

on the real exchange rate, while inducing positive comovement between the trade balance and the real

exchange rate.10

Consistent with this interpretation, a preference shock in favour of home goods generates a realappreciation and an improvement in the trade balance along the adjustment path. This highlights

Fig. 6 (continued)

10 It might be more useful to think of a negative value of this shock a shock against home exports.



17/25

the importance of allowing for state-dependent shocks on the trade balance and real exchange rate,

and implies that fixed coefficient models are more likely to conflate shocks (particularly supply inno-

vations). For instance, the time-varying estimates for China show that supply shocks went from pro-

ducing a positive correlation between the trade balance and the real exchange rate to a negative

coefficient, before turning positive again during the mid-2000s.

4.2.2. Trade balance and business cycles

The unconditional correlations between the trade balance and output are close to zero for all coun-

tries. This suggests that the trade balance is not strongly affected by changes in relative domestic busi-

ness cycle conditions. In contrast, unconditional correlation evidence documented for the G7 countries

inPrasad (1999) and Lee and Chinn (2006)show trade balance dynamics to be countercyclical.

With the exception of Taiwan, the time-varying estimates show that nominal shocks have gener-

ated a consistent negative correlation between the trade balance and output over the last 20 years.

This contrasts with results from a fixed coefficient VAR model inKim and Lee (2008), which showed

that nominal shocks produced a positive correlation between real output and the current account for

Japan.It is difficult to make any generalisations regarding the correlation generated by demand and sup-

ply shocks for the five countries. Since 2000 demand and supply shocks have, in general, led to pro-

cyclical trade balance dynamics in Japan, Malaysia and Taiwan. If supply shocks represent

preference shocks then this will induce a positive correlation between macroeconomic aggregates

and the trade balance.

5. Time-varying response functions

This paper computes the impulse responses using a local approximation of the impulse responses

at timet. FollowingCanova and Gambetti (2010), this is calculated by taking the posterior distribution

from the reduced-form covariance matrix Xt A1t HtA1t 0 in Eq.(2). The elements contained in thecovariance matrix provide the contemporaneous impact. In order to calculate the response functions

at horizonst+h, whereh is the horizon, the posterior distribution of the VAR coefficients in Eq.(1)is

applied for all t, whilst setting et+h= 0. This provides a median posterior distribution of impulse re-sponses where the responses of all variables at time tare allowed to vary. Since the autoregressive

coefficients are time-varying the impulse responses are based on parameter values drawn from the

median of the posterior distribution along the entire sample period.11

Given the pivotal role of China in current discussions regarding the trade balance and real exchange

rate adjustment the estimates for China are presented in conjunction with standard response func-

tions.12 Since Chinas trade balance surplus grew rapidly post-1999 the mean impulse responses based

on the time-varying parameters for the pre- and post-1999 period are explicitly illustrated.

5.1. Fixed coefficient structural model

Before preceding to the results derived from the time-varying parameter model, response functions

based on a fixed coefficient VAR model estimated using ordinary least squares (OLS) are illustrated in

Fig. 5. The identification scheme remains as explained in the previous section. For reasons of brevity,

only the response of the trade balance to demand and supply responses of China, Japan and South Kor-

ea are shown.13 The average response are estimated over two sample periods, using the Asian financial

crisis in 1997 as the breakpoint. The estimates represent the average response across both subsamples

and, therefore, ignores within subsample variation.

11 The 3D-graphs of the time-varying relationship graphs are to be read in the following way: along the x-axis the starting

quarters are aligned from 1987Q1 to 2011Q2, on the y-axis the quarters after the shock are displayed, and the z-axis represents the

value of the response of the variable to the shock.12 Standard response functions for the other countries are available upon request from the author.13 Full set of responses are available upon request from the author.

http://-/?-http://-/?-http://-/?-


18/25

The subsample estimates well illustrate the importance of allowing for state-dependency when

analysing trade balance dynamics over time. The response of the trade balance to a demand shock

pre- and post-1997 differ markedly. The responses are not only consistent with a change in the

2 4 6 8 10 12 14 16 18 20

0.15

0.2

0.25

0.3

0.35

0.4

0.450.5

0.55

0.6pre-1999

post-1999

2 4 6 8 10 12 14 16 18 20

-0.1

-0.08

-0.06

-0.04

-0.02

0

2 4 6 8 10 12 14 16 18 20

-0.18

-0.16

-0.14

-0.12

-0.1

-0.08

-0.06

-0.04

-0.02

Fig. 7. Real exchange rate dynamics.



19/25

stochastic volatility, as exemplified by the shift in the magnitude of the responses, but also by achange in the autoregressive coefficients, illustrated by a shift in the signs of the response functions.

The estimates for South Korea and Taiwan show the sign of the trade balance to a supply shock has

changed direction. With the exception of Taiwan, the responses of the trade balance to a demand

shock have been, relatively speaking, more homogenous. However, the responses show a significant

Fig. 7(continued)



20/25

difference in the magnitude of the responses. This finding signifies the importance of allowing for

shifts in the stochastic volatility. The following section goes onto the explore the nature of the

time-variation more thoroughly.

Fig. 7 (continued)



21/25

5.2. Trade balance and real exchange rate dynamics

The time-varying responses of the trade balance and real exchange rate for the constituent coun-

tries are shown inFigs. 6 and 7. A policy prescription often forwarded in dealing with Chinas large

current account surplus is for an increase in consumption from its currently low base.14 The estimates

show that the effect of a demand shock on Chinas trade balance has increased in magnitude over the last

two decades. Since the late 1990s the time-varying response functions show that demand shocks have

led to a decline in Chinas trade balance as a percentage of GDP. The response of the trade balance to

a demand shock demonstrates a high level of persistence. This is despite the time-varying responses

showing that, since the late 1990s, the effect of a demand shock on the real exchange rate has consider-

ably diminished. That the effect of demand shocks are larger on the Chinese trade balance than the real

exchange rate implies that the income channel is central to trade balance adjustment. This finding is con-

sistent with partial equilibrium estimates for China inCheung et al. (2010).

Consistent with indications from the conditional correlation estimates the time-varying estimates

for South Korea and Taiwan illustrate a rise in the trade balance in response to a demand shock. The

estimates for Taiwan show that since early 2000 the effectiveness of demand disturbances in curbing

trade balance increased. In terms of magnitude, of all three countries (China, South Korea and Taiwan),a demand shock would have the largest impact in reducing trade surpluses for China. For Japan and

Malaysia it is not clear that a demand shock leads to a deterioration in the trade balance. The

short-run positive response of the Japanese trade balance to a demand shock is consistent with esti-

mates derived from a fixed coefficient VAR in Prasad (1999). One explanation is the elasticity of im-

ports with respect to transitory output fluctuations is much smaller than with respect to persistent

changes in output. Finally, the real exchange rate also appreciates for all countries in response to a de-

mand shock. This is consistent with a plethora of NOEM models: seeObstfeld (1985), Clarida and Gali

(1994), Chadha and Prasad (1997), Prasad (1999), Astley and Garratt (2000) and Lee and Chinn

(2006).15

Over the last 20 years the real exchange for the East Asian economies to nominal shocks have

tended to experience a transitory short-run depreciation before then appreciating towards its trendlevel. This result is consistent with the overshooting hypothesis in Dornbuschs (1976) sticky-price

model. Nominal shocks have tended to have persistent effects on the trade balance.16 This indicates

that the effect on the trade balance of the real exchange rate depreciation caused by relative monetary

easing outweighs the effects of the expansion of relative output. This implies the presence of hysteresis

and beach-head effects in Asian trade dynamics, which could translate temporary exchange rate shifts

into persistent changes in the trade balance.17

The NOEM model outlined indicated the effects of supply and nominal shocks on the trade balance

to be ambiguous, depending on the elasticities of the trade balance with respect to the real exchange

rate and output. With the exception of supply shocks in Japan, in all countries, permanent supply

shocks have historically appreciated the real exchange rate.18 The long-run appreciation of the ex-

change rate to permanent supply shocks is consistent with an extendedClarida and Gali (1994)model,which includes money and productivity shocks on the current account and real exchange rate inLee and

Chinn (2006). However, the positive response of the trade balance for the other countries pose a puzzle.

As the real exchange rate appreciates to permanent shocks the trade balance also improves. This positive

14 Based on the most recent data household consumption as a share of GDP was around 35% in 2011.15 Intuitively, this is due to a rise in interest rates as a result of an outward shift in the IS curve precipitating capital flows.16 Corsetti et al. (2006) show explicitly that the effects of supply shocks on the real exchange rate may differ depending on

whether the technology shocks fall more on the tradables or nontradables sector. The cases where the current account deteriorates

despite a real exchange rate depreciation may also reflect the role of investment. An investment boom following a technology

shock can lead the current account to deteriorate, while causing a real exchange rate depreciation concurrently.17 Similar evidence has been reported byPrasad (1999)for the G7 countries, and Kim and Lee (2009) for the Euro zone, Japan and

the US. Also seeBaldwin (1988).18 Clarida and Gali (1994)also find the dollaryen exchange rate depreciates in response to a supply shock.

http://-/?-


22/25

comovement between the exchange rate and the current account does not accord well with predictions

of single-sector models.19

A productivity shock is probably the first to be counted among permanent shocks, but is not the

only one. For example, a permanent preference shock in favour of home exports would also have a

long-run effect on the real exchange rate. As noted in Lee and Chinn (2006) and Blanchard et al.

(2005), such a preference shock is more likely to lead to the positive comovement between current

account and the real exchange rate.

6. Conclusion

This paper provides some unique empirical estimates on the trade balance and real exchange rate

dynamics for a group of East Asian economies that maintain large current account surpluses. This pa-

per presents the first attempt to document these links in a Bayesian time-varying model which em-

ploys minimal long-run identifying restrictions that are consistent with both the theoretical and

empirical NOEM trade balance-exchange rate literature.

The time-varying estimates show that real (permanent) shocks have tended to dominate in driving

variation in the real exchange rate. Equilibrium models rely on permanent real shocks to explain

movements in real and nominal rates. In contrast, disequilibrium models of exchange rate determina-

tion generally assume that variation in both real and nominal exchange rates is due primarily to nom-

inal disturbances, which can be expected to have a transitory effect on real exchange rates. The

importance of persistent shocks is consistent with the observed characteristics of the real exchange

rate time series for the Asian countries.

The response functions imply that the income channel has become important in trade balance

adjustment for China. The time-varying response functions also show that nominal shocks have a per-

sistent effect on the trade balance, despite a transitory impact on the real exchange rate. Supportive of

the MarshallLearner condition this implies that a revaluation would have a real impact on Asian

trade balances. Additionally, that much of the variance of the trade balance has been driven by nom-inal (transitory) shocks suggest it is not just long-run structural factors that explain the trade balance

surpluses of the Asian economies, as is sometimes suggested. Short-term (transitory) factors are also

significant.

Taken at face value the results guard against projecting a specific correlation between the trade bal-

ance and the real exchange rate independent of the largest source of shocks. Attempts to establish

tight evidence on the effect of the real exchange rate on the trade balance will generate mixed results

if models do not successfully control for persistent shocks that drive the bulk real exchange rate move-

ments. The results in this paper demonstrate that insufficient disentanglement of shocks in models

may lead to conflations in the structural innovations for example, productivity and preference

shocks in fixed coefficient models.

The time-varying estimates show that the recent financial crisis has not dramatically altered thedynamics or the real exchange rate or the trade balance, or the correlation between the two. Allowing

for a few caveats, the relatively small heterogeneity in the trade balance dynamics of the East Asian

countries to macroeconomic shocks reported is evidence in favour of regional and international policy

coordination.

This paper highlights the importance of incorporating state-dependent shocks on the real exchange

rate and the trade balance. The findings in this paper are relevant for future studies. Allowing for a few

caveats, the progressive as opposed to abrupt change in the size of the response functions imply that

using crude split sample or rolling window models may not adequately capture the evolving effect of

macroeconomic shocks on the trade balance. Current structural break tests are unlikely to capture

progressive structural changes. Using split sample models may also ignore changes within the sub-

sample. The findings in this paper suggest that allowing for time-variation in the parameters provides

19 Regardless of whether the permanent shock captures the productivity shock or the portion of monetary shock that affects the

long-run real exchange rated in single-sector models, current account improvement is associated with real exchange rate

depreciation. SeeLee and Chinn (2006).



23/25

a more nuanced picture regarding the changing relationship between macroeconomic shocks and the

trade balance over the course of a few decades.

Appendix A. Markov Chain Monte Carlo estimation

A.1. Priors

Let zTdenote a sequence ofzs up to time T. The conditional prior density ofhTis given by

phTjaT;h

T;Q;H;N /IhThTja

T;hT;Q;H;N A:1

where IhT QT

t0IhT,

fhTjaT;h

T;Q;H;N fh0

YTt1

fhTjhT1;aT;h

T;Q;H;N A:2

andI(hT) takes a unit value if all the roots of the VAR polynomial associated with htare larger than one

in modulus and 0 otherwise, ruling out a non-stationary processes.FollowingBenati and Mumtaz (2007) and Primiceri (2005), the following assumptions on the prior

distributions and its hyperparameters are made

ph0 /Ih0NhOLS;bVhOLS A:3plogh0 Nlog ^hOLS; 10 I A:4

pa0 NaOLS; jaOLSj A:5

pQ IWQ1; T0 A:6

pH IWH1; 2 A:7

pNi;i IG 0:0001

2 ;

1

2

A:8

hOLSis a vector of OLS estimates of the reduced form VAR coefficients.

bVhOLSrepresents the OLS estimate of the covariance matrix using the training sample. hOLSis a vector containing the stochastic volatility in the diagonal elements of the matrix Ht.

aOLSis the element (2,1) of the lower triangular matrixbAt.

Q0:005 bVhOLS. T0 is the number of observations in the initial training sample, in this case 6 years (24

observations).

H 0:0012 jaOLSj.

A.2. Estimation

The model is estimated using a Markov Chain Monte Calro (MCMC) method based on Gibbs sam-

pling. The Gibbs sampler partitions the vector of unknowns into blocks and the transition density is

defined by the product of conditional densities.

Step 1: p(hTjxT,aT,hT, Q,H,N) Conditional on xT,aT, hT, Q, H, N the unrestricted posterior of the states is normal.

http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


24/25

The conditional mean and variance of the terminal statehT is computed using standard

Kalman filter recursions while for all the other states the following backward recursions

are employed:htjt1 htjtPtjtP

1tjt1ht1htjt A:9

Ptjt1 PtjtPtjtP

1

t1jtPtjt A:10

where p(hTjxT,aT, hT, Q,H,N)N(htjt+1, Ptjt+1)Step 2: p(aTjxT,hT, hT, Q,H,N)

Conditional on hT; ^ytxt /0;t /1;txt1 /p;txtp is observable.

Rewriting the system of equations as AtytHtvt where vt N(0, I).

Conditional onhT, theCarter and Kohn algorithm (1994)is used to estimate a draw for attaking the above system as observational equations.

atand vtare independent across equations and, thus, the algorithm can be used to estimateequation by equation.

Step 3: p(hTjxT,hT,aT, Q,H,N) This step is estimated using the date-by-date blocking system inJacquier et al. (1994).

Step 4: p(HjxT,hT,aT, hTQ,N), p(Ni,ijxT,hT,aT, hTQ,H), p(Q, jxT,hT,aT, hT,H,N) Conditional onxT, hT, aT,hT all the remaining hyperparameters, under conjugate priors, can

be sampled from an inverted Wishart and Inverted Gamma densities.

The estimates are retrieved from 100,000 burn-in draws and then saving every fifth draw from an-

other 50,000 draws. This is enough to ensure that the Gibbs sampler converges to the ergodic distri-

bution, whilst preventing autocorrelation in the Gibbs chain.

References

Astley, M., Garratt, A., 2000. Exchange rates and prices: sources of sterling real exchange rate fluctuations 19731994. Oxford

Bulletin of Economics and Statistics 62 (4), 491509.Baldwin, R., 1988. Hysteresis in import prices: the beachhead effect. American Economic Review 78, 773785.Benati, L., Mumtaz, H., 2007. U.S. Evolving Macroeconomic Dynamics: A Structural Investigation, ECB Working Paper Series

Number 746.Betts, C., Devereux, M., 1996. Exchange Rate Dynamics in a Model of Pricing-to-Market. University of British Columbia, Mimeo.Blanchard, O., Giavazzi, F., Sa, F., 2005. The U.S. Current Account and the Dollar. NBER Working Paper No. 11137.Canova, F., Gambetti, L., 2010. Do expectations matter? The great moderation revisited. American Economic Journal:

Macroeconomics 2 (3), 183205.Carter, C.K., Kohn, R., 1994. On Gibbs sampling for state space models. Biometrika 81, 541553 .Chadha, B., Prasad, E., 1997. Real exchange rate fluctuation and the business cycle: evidence from Japan. IMF Staff Papers 44 (3),

328354.Chari, V.V., Kehoe, P., McGrattan, E., 1997. Monetary Shocks and Real Exchange Rates in Sticky Price Models of the International

Business Cycle. NBER Working Paper No. 5876.Cheung, Y.-W., Chinn, M.D., Fujii, E., 2010. Chinas current account and exchange rate. NBER chapters. In: Chinas Growing Role

in World Trade. National Bureau of Economic Research, Inc., pp. 231271Clarida, R., Gali, J., 1994. Sources of real exchange-rate fluctuations: how important are nominal shocks? Carnegie-RochesterConference Series on Public Policy 41, 156.

Cogley, T., Sargent, T.J., 2005. Drifts and Volatilities: Monetary Policies and Outcomes in the Post WWII US. Review of EconomicDynamics 8, 262302.

Corsetti, G., Dedola, L., Leduc, S., 2006. Productivity, External Balance and Exchange Rates: Evidence on the TranmissionMechanism among G7 Countries. NBER Working Paper No. 12483.

Croke, H., Kamin, S.B., Leduc, S., 2006. Financial market developments and economic activity during current accountadjustments in industrial economies. International Finance 9 (1), 3762.

Dornbusch, R., 1976. Expectations and exchange rate dynamics. Journal of Political Economics 84, 11611176 .Galati, G., Debelle, G., 2005. Current Account Adjustment and Capital Flows. BIS Working Paper No. 169.Grier, K., Ye, H., 2009. Twin sons of different mothers: the long and short of the twin deficits debate. Economic Inquiry 47 (4),

625638.Jacquier, E., Polson, N.G., Rossi, P.E., 1994. Bayesian analysis of stochastic volatility models. Journal of Business and Economic

Statistics 12, 371417.

Kim, S., Lee, J., 2008. International Macroeconomic Fluctuations: A New Open Economy Macroeconomics Interpretation. HongKong Institute of Monetary Research, Working Paper No. 23/2008.

Lastrapes, W., 1992. Sources of fluctuations in real and nominal exchange rates. Review of Economics and Statistics 74, 530539.

Lee, J., Chinn, M.D., 2006. Current account and real exchange rate dynamics in the G7 countries. Journal of International Moneyand Finance 25, 257274.

http://refhub.elsevier.com/S0889-1583(13)00029-4/h0005http://refhub.elsevier.com/S0889-1583(13)00029-4/h0005http://refhub.elsevier.com/S0889-1583(13)00029-4/h0010http://refhub.elsevier.com/S0889-1583(13)00029-4/h0015http://refhub.elsevier.com/S0889-1583(13)00029-4/h0015http://refhub.elsevier.com/S0889-1583(13)00029-4/h0020http://refhub.elsevier.com/S0889-1583(13)00029-4/h0025http://refhub.elsevier.com/S0889-1583(13)00029-4/h0025http://refhub.elsevier.com/S0889-1583(13)00029-4/h0030http://refhub.elsevier.com/S0889-1583(13)00029-4/h0030http://refhub.elsevier.com/S0889-1583(13)00029-4/h0050http://refhub.elsevier.com/S0889-1583(13)00029-4/h0050http://refhub.elsevier.com/S0889-1583(13)00029-4/h0035http://refhub.elsevier.com/S0889-1583(13)00029-4/h0035http://refhub.elsevier.com/S0889-1583(13)00029-4/h0040http://refhub.elsevier.com/S0889-1583(13)00029-4/h0040http://refhub.elsevier.com/S0889-1583(13)00029-4/h0045http://refhub.elsevier.com/S0889-1583(13)00029-4/h0045http://refhub.elsevier.com/S0889-1583(13)00029-4/h0055http://refhub.elsevier.com/S0889-1583(13)00029-4/h0055http://refhub.elsevier.com/S0889-1583(13)00029-4/h9950http://refhub.elsevier.com/S0889-1583(13)00029-4/h9950http://refhub.elsevier.com/S0889-1583(13)00029-4/h9950http://refhub.elsevier.com/S0889-1583(13)00029-4/h9855http://refhub.elsevier.com/S0889-1583(13)00029-4/h9855http://refhub.elsevier.com/S0889-1583(13)00029-4/h9855http://refhub.elsevier.com/S0889-1583(13)00029-4/h9855http://refhub.elsevier.com/S0889-1583(13)00029-4/h9950http://refhub.elsevier.com/S0889-1583(13)00029-4/h9950http://refhub.elsevier.com/S0889-1583(13)00029-4/h0055http://refhub.elsevier.com/S0889-1583(13)00029-4/h0055http://refhub.elsevier.com/S0889-1583(13)00029-4/h0045http://refhub.elsevier.com/S0889-1583(13)00029-4/h0045http://refhub.elsevier.com/S0889-1583(13)00029-4/h0040http://refhub.elsevier.com/S0889-1583(13)00029-4/h0035http://refhub.elsevier.com/S0889-1583(13)00029-4/h0035http://refhub.elsevier.com/S0889-1583(13)00029-4/h0050http://refhub.elsevier.com/S0889-1583(13)00029-4/h0050http://refhub.elsevier.com/S0889-1583(13)00029-4/h0030http://refhub.elsevier.com/S0889-1583(13)00029-4/h0030http://refhub.elsevier.com/S0889-1583(13)00029-4/h0025http://refhub.elsevier.com/S0889-1583(13)00029-4/h0025http://refhub.elsevier.com/S0889-1583(13)00029-4/h0020http://refhub.elsevier.com/S0889-1583(13)00029-4/h0015http://refhub.elsevier.com/S0889-1583(13)00029-4/h0015http://refhub.elsevier.com/S0889-1583(13)00029-4/h0010http://refhub.elsevier.com/S0889-1583(13)00029-4/h0005http://refhub.elsevier.com/S0889-1583(13)00029-4/h0005http://-/?-http://-/?-


25/25

Messe, R.A., Rogoff, K., 1983. Empirical exchange rate models of the seventies: do they fit out of sample. Journal of InternationalEconomics 14, 324.

Obstfeld, M., 1985. Floating exchange rates: experience and prospects. Brookings Papers on Economic Activity 2, 369450 .Prasad, E.S., 1999. International trade and the business cycle. Economic Journal 109, 588606.Primiceri, G.E., 2005. Time varying structural vector autoregressions and monetary policy. Review of Economic Studies 72, 821

852.

Rafiq, M.S., 2010. Fiscal stance, the current account and the real exchange rate: some empirical estimates from a time-varyingframework. Structural Change and Economic Dynamics 21, 276290.

Robertson, D., Wickens, M.R., 1997. Measuring real and nominal macroeconomic shocks and their transmission under differentmonetary system. Oxford Bulletin of Economics and Statistics 59, 527.

Stockman, A., 1987. The Equilibrium Approach to Exchange Rates. Federal Reserve Bank of Richmond Economic Review, March/April.

http://refhub.elsevier.com/S0889-1583(13)00029-4/h0060http://refhub.elsevier.com/S0889-1583(13)00029-4/h0060http://refhub.elsevier.com/S0889-1583(13)00029-4/h0060http://refhub.elsevier.com/S0889-1583(13)00029-4/h0065http://refhub.elsevier.com/S0889-1583(13)00029-4/h0065http://refhub.elsevier.com/S0889-1583(13)00029-4/h0070http://refhub.elsevier.com/S0889-1583(13)00029-4/h0075http://refhub.elsevier.com/S0889-1583(13)00029-4/h0075http://refhub.elsevier.com/S0889-1583(13)00029-4/h0080http://refhub.elsevier.com/S0889-1583(13)00029-4/h0080http://refhub.elsevier.com/S0889-1583(13)00029-4/h0085http://refhub.elsevier.com/S0889-1583(13)00029-4/h0085http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0085http://refhub.elsevier.com/S0889-1583(13)00029-4/h0085http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0080http://refhub.elsevier.com/S0889-1583(13)00029-4/h0080http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0075http://refhub.elsevier.com/S0889-1583(13)00029-4/h0075http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0070http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0065http://-/?-http://refhub.elsevier.com/S0889-1583(13)00029-4/h0060http://refhub.elsevier.com/S0889-1583(13)00029-4/h0060http://-/?-

Date post:	03-Jun-2018
Category:	Documents
Upload:	phuonglinh93
View:	220 times
Download:	0 times

Trade Balance -1

Documents