Sie sind auf Seite 1von 9

Exchange-Rate Volatility and Foreign Trade: Evidence from Thirteen LDC's

Author(s): Augustine C. Arize, Thomas Osang, Daniel J. Slottje


Source: Journal of Business & Economic Statistics, Vol. 18, No. 1 (Jan., 2000), pp. 10-17
Published by: American Statistical Association
Stable URL: http://www.jstor.org/stable/1392132
Accessed: 13/01/2009 01:30
Your use of the JSTOR archive indicates your acceptance of JSTOR's Terms and Conditions of Use, available at
http://www.jstor.org/page/info/about/policies/terms.jsp. JSTOR's Terms and Conditions of Use provides, in part, that unless
you have obtained prior permission, you may not download an entire issue of a journal or multiple copies of articles, and you
may use content in the JSTOR archive only for your personal, non-commercial use.
Please contact the publisher regarding any further use of this work. Publisher contact information may be obtained at
http://www.jstor.org/action/showPublisher?publisherCode=astata.
Each copy of any part of a JSTOR transmission must contain the same copyright notice that appears on the screen or printed
page of such transmission.
JSTOR is a not-for-profit organization founded in 1995 to build trusted digital archives for scholarship. We work with the
scholarly community to preserve their work and the materials they rely upon, and to build a common research platform that
promotes the discovery and use of these resources. For more information about JSTOR, please contact support@jstor.org.
American Statistical Association is collaborating with JSTOR to digitize, preserve and extend access to Journal
of Business & Economic Statistics.
http://www.jstor.org
Exchange-Rate Volatility
and
F or eign
T r ad e:
Evid ence F r om T hir teen L D C ' s
Augus tine
C . ARIZE
C ollege
of Bus ines s and
T echnology,
T exas A&M
Univer s ity, C om m er ce,
T X 75429
(chuck_ar ize@tam a-com m er ce.ed u)
T hom as OSANG
D epar tm ent
of
Econom ics ,
Souther nMethod is t
Univer s ity, D allas ,
T X 75275
(tos ang@m ail.s m u.ed u)
D aniel J. SL OT T JE
D epar tm ent
of Econom ics ,
Souther nMethod is t
Univer s ity, D allas ,
T X 75275
(d s lottje@m ail.s m u.ed u)
T his ar ticle
inves tigates em pir ically
the
im pact
of r eal
exchange-r ate volatility
onthe
expor t
flows
of 13 les s
d eveloped
countr ies (L D C ' s ) over the
quar ter ly per iod
1973-1996. Es tim ates of the
cointegr ating
r elations ar e obtained
us ing
Johans en' s m ultivar iate
pr oced ur e.
Es tim ates of the s hor t-
r un
d ynam ics
ar e obtained for each
countr y us ing
the er r or -cor r ection
technique.
T he
m ajor
r es ults
s how that incr eas es inthe
volatility
of the r eal effective
exchange r ate, appr oxim ating exchange-
r ate
uncer tainty,
exer t a
s ignificant negative
effect on
expor t
d em and inboththe s hor t-r unand the
long-r un
ineachof the 13 L D C ' s . T hes e effects
m ay
r es ult in
s ignificant
r eallocationof r es our ces
by
m ar ket
par ticipants .
KEY WORD S:
Exchange-r ate var iability; Expor ts ;
Er r or -cor r ection m od el; L es s
d eveloped
coun-
tr ies .
F or eign-exchange
r ates for
d eveloping
and
d eveloped
countr ies have been
highly
volatile s ince the aband onm ent
of fixed
exchange
r ates inMar ch
1973, following
the U.S.
d ollar d evaluationin
F ebr uar y
1973. A centr al
ques tion
has
beenthe effect of s uch
highexchange-r ate volatility
on
the
gr owth
of
for eign
tr ad e. Stud ies
by C howd hur y (1993),
C us hm an
(1988), T hur s by
and
T hur s by
(1987),
and Kenen
and Rod r ik
(1986), am ong other s ,
have
pr ovid ed em pir i-
cal evid ence for
d eveloped
countr ies
(m ainly
the United
States ,
the United
Kingd om , Italy, Ger m any, C anad a, etc.).
In
gener al,
thes e s tud ies have conclud ed that
exchange-r ate
volatility
inhibits the
gr owth
of
for eign
tr ad e. But little is
knownabout the extent to whichthis conclus ion
m ay
be
confir m ed for les s
d eveloped
countr ies
(or
L D C ' s for
s hor t).
Stud ies of the
exper ience
withL D C ' s have been
ver y few,
d ue
m ainly
to the
unavailability
of s ufficient tim e s er ies
d ata.
T his ar ticle exam ines the
im pact
of
exchange-r ate
volatil-
ity
onthe
expor t
flows of 13 L D C ' s . T he evid ence
pr es ented
will
ad d anextr a d im ens ionto this liter atur e and
pr ovid e
a
bas is withwhichfutur e s tud ies canbe
com par ed .
Knowl-
ed ge
of the
d egr ee
to which
exchange-r ate volatility
affects
expor ts
is
im por tant
for the
d es ign
of both
exchange-r ate
and tr ad e
policies .
F or
exam ple,
if
exchange-r ate volatility
has ad ver s e effects on
expor ts , tr ad e-ad jus tm ent pr ogr am s
inL D C ' s that have
s tr ongly em phas ized
the need for
expor t
expans ion
could be uns ucces s ful if
exchange
r ates ar e
ver y
volatile. In
ad d ition,
the intend ed effect of a tr ad e liber al-
ization
policy m ay
be d oom ed
by
a var iable
exchange
r ate
and could
pr ecipitate
a
balance-of-paym ents
cr is is .
T he r em aind er of the ar ticle is
or ganized
as follows . In
Section
1,
we exam ine the theor etical
r elations hip
between
volum e of inter national tr ad e and
exchange-r ate volatility,
followed
by
the d is cus s ionof the econom etr ic
m ethod ology
is s ues . InSection
2,
we d is cus s the
em pir ical
r es ults for 13
L D C ' s for the
pos t
Br ettonWood s
per iod .
C onclus ions ar e
d r awninSection3.
1. T HEORET IC AL C ONSID ERAT IONS
AND MOD EL SPEC IF IC AT ION
T heor etical
analys es
of the
r elations hip
between
higher
exchange-r ate volatility
and inter national tr ad e tr ans actions
have beencond ucted
by Hooper
and
Kohlhagen
(1978)
and
other s . T he
ar gum ent
is as follows :
Higher exchange-r ate
volatility
lead s to
higher
cos t for r is k-aver s e tr ad er s and
to les s
for eign
tr ad e. T his is becaus e the
exchange
r ate is
agr eed
onat the tim e of the tr ad e
contr act,
but
paym ent
is
not m ad e until the futur e
d eliver y actually
takes
place.
If
changes
in
exchange
r ates becom e
unpr ed ictable,
this cr e-
ates
uncer tainty
about the
pr ofits
to be m ad e
and , hence,
r ed uces the benefits of inter national tr ad e.
Exchange-r ate
r is k for L D C ' s is
gener ally
not
hed ged
becaus e for war d
m ar kets ar e not acces s ible to all tr ad er s . Evenif
hed ging
inthe for war d m ar kets wer e
pos s ible,
ther e ar e lim itations
and cos ts . F or
exam ple,
the s ize of the contr acts is
gener ally
lar ge,
the
m atur ity
is
r elatively s hor t,
and it is d ifficult to
plan
the
m agnitud e
and
tim ing
of all inter national tr ans ac-
tions to take
ad vantage
of the for war d m ar kets . F or fur ther
d etails onthe d ifficulties of
hed ging
in
L D C ' s ,
s ee Med hor a
(1990).
Onthe other
hand ,
r ecent theor etical
d evelopm ents s ug-
ges t
that ther e ar e s ituations inwhichthe
volatility
of ex-
? 2000 Am er icanStatis tical As s ociation
Jour nal of Bus ines s & Econom ic Statis tics
Januar y 2000,
Vol. 18, No. 1
10
Ar ize, Os ang,
and
Slottje: Exchange-Rate Volatility
and
F or eign
T r ad e 11
change
r ates could be
expected
to have either
negative
or
pos itive
effects ontr ad e volum e. D e Gr auwe (1988) s tr es s ed
that the d om inance of incom e effects over s ubs titutionef-
fects canlead to a
pos itive r elations hip
betweentr ad e and
exchange-r ate volatility.
T his is
becaus e,
if
expor ter s
ar e
s ufficiently
r is k
aver s e,
anincr eas e in
exchange-r ate
volatil-
ity
r ais es the
expected m ar ginal utility
of
expor t
r evenue
and ther efor e ind uces them to incr eas e
expor ts .
D e Gr auwe
s ugges ted
that the effects of
exchange-r ate uncer tainty
on
expor ts
s hould
d epend
onthe
d egr ee
of r is k aver s ion.
Recently,
theor etical m od els of
hys ter es is
ininter national
tr ad e
(s ee
Bald winand
Kr ugm an
1989, p. 635;
D ixit
1989,
p. 206)
have s hownthat incr eas ed
uncer tainty
fr om
high
volatility
in
exchange
r ates canals o influence
for eigntr ad e,
in
par ticular
if
s ignificant
s unk cos ts ar e involved ininter -
national tr ans actions . It is
d ifficult, however ,
to
id entify
how
tr ad e will be affected . F or
exam ple,
F r oot and
Klem per er
(1989, p. 643)
s howed that
exchange-r ate uncer tainty
can
affect the
pr ice
and
quantity
of
tr ad e,
either
pos itively
or
negatively,
whenm ar ket s har e m atter s und er an
oligopolis -
tic m ar ket
s tr uctur e, r egar d les s
of r is k choices .
In
s um m ar y,
the
im pact
of
exchange-r ate volatility
on
for eign
tr ad e is an
em pir ical
is s ue becaus e
theor y
alone
cannot d eter m ine the
s ign
of the r elationbetween
for eign
tr ad e and
exchange-r ate volatility.
T o cons er ve
s pace,
no
theor etical d is cus s ions onthe effects of
for eign
incom e or
r elative
pr ice
var iables ar e
pr es ented
her e. T hes e canbe
found inanar ticle
by
Ar ize
(1990).
A tr ad itional
s pecification
of the
long-r unequilibr ium
ex-
por t
d em and inthe flexible
exchange-r ate
envir onm ent is
that of Ar ize
(1995)
and
C howd hur y (1993),
Qd
=
T o +
T 1
-
Wt
+ T 2
'
Pt + 73 ' t + EC t, (1)
wher e
Qd
d enotes the
logar ithm
of d es ir ed volum e of a
countr y' s expor t good s , wt
is the
logar ithm
of a s cale var i-
able that
captur es
wor ld d em and
cond itions , pt
is the
loga-
r ithm of r elative
pr ices
and is m eas ur ed
by
the r atio of that
countr y' s expor t pr ice
inU.S. d ollar s to the wor ld
expor t
pr ice
inU.S.
d ollar s , at
is the
logar ithm
of a
m oving-s am ple
s tand ar d d eviation
(Jt+m ),
and
EC t
is a d is tur bance ter m .
It is
expected
that
7-
>
0, T 2
< 0,
and
T 3
>
0.
Befor e
pr es entation
of the
em pir ical r es ults ,
thr ee tech-
nical notes
r egar d ing Equation(1) and the m ethod of es -
tim ationar e inor d er . F ir s t, we m ake the
as s um ptionthat,
inthe
long r un, one would
expect
that
any
d eviationof ac-
tual (obs er vable) r eal
expor ts
fr om d es ir ed (unobs er vable)
s hould
d is appear (i.e.,
Qd
=
Qt).
Second , it is
neces s ar y
to d er ive an
oper ational
m eas ur e
of
exchange-r ate uncer tainty.
Inthis ar ticle we us e a tim e-
var ying m eas ur e of
exchange-r ate volatility
to account for
per iod s
of low and
highexchange-r ate uncer tainty.
T his
pr oxy
is cons tr ucted
by
the
m oving-s am ple s tand ar d d e-
viation
expr es s ed as
t+m =
(Rt+i-1
-
Rt+i-2)2
1
(2)
?7
i= 1
wher e
Ris the natur al
logar ithm
of r eal effective
exchange
r ate and m = 7 is the or d er of the
m oving aver age.
Wor k
by Baba, Hend r y,
and Star r (1992, pp.
34-36) gives
the ad -
vantages
of
em ploying
this m eas ur e. See als o
Kor ay
and
L as tr apes (1989). F ur ther m or e, this
m eas ur e
is s im ilar to
thos e
em ployed
inthe
exchange-r ate volatility
and inter na-
tional tr ad e liter atur e
(e.g.,
Kenenand Rod r ik
1986; Kor ay
and
L as tr apes 1989; C howd hur y 1993).
F inally,
to es tablis hwhether ther e is a
long-r unequilib-
r ium
r elations hip am ong
the var iables in
Equation(1),
we
m us t
em ploy
the
concept
of
cointegr ation. C ointegr ation
tes ts inthis ar ticle ar e cond ucted
by
m eans of the Johans en
m ethod intr od uced
by
Johans en
(1988)
and extend ed
by
Jo-
hans enand Jus elius
(1990).
T he m ethod us es two likelihood
r atio
(L R)
tes t
s tatis tics -nam ely,
the tr ace and the m axi-
m al
eigenvalue (A-m ax)
s tatis tics -to tes t for the num ber
of
cointegr ating
vector s in
nons tationar y
tim e s er ies . T he
num ber of
lags applied
ineach
cointegr ation
tes t is bas ed
oninfor m ation
pr ovid ed by
the Sim s L R
tes t,
the Akaike
infor m ation
cr iter ion,
and the
L jung-Box
tes t.
2. EMPIRIC AL RESUL T S
2.1
C ointegr ationAnalys is
T he fir s t
s tep
in
tes ting
for
cointegr ation
ina s et of var i-
ables is to tes t for s tochas tic tr end s inthe
autor egr es s ive
r epr es entation
of eachind ivid ual tim e s er ies
us ing aug-
m ented
D ickey
and F uller and Johans entes ts . F or
s pace
cons id er ation,
the
em pir ical
r es ults ar e not
pr es ented her e,
but
they s ugges t
that all the var iables in
Equation(1)
ar e
integr ated
of or d er 1. F or d etails onthes e as well as s om e
of the other r es ults d is cus s ed
later ,
s ee
Ar ize, Os ang,
and
Slottje (1997).
In
applying
the Johans en
pr oced ur e,
we allow for a d e-
ter m inis tic tr end becaus e the null
hypothes is
of aninter -
cept
inthe
cointegr ating
vector s
agains t
the alter native of
a linear tr end inthe var iables was
r ejected .
T he calculated
L Rvalues ar e 8.93 for
Ecuad or ,
9.74 for
Ind ones ia,
13.58
for
Kor ea,
27.33 for
Malays ia,
13.44 for
Malawi,
7.92 for
Maur itius ,
15.29 for
Mexico,
7.05 for
Mor occo,
8.59 for
the
Philippines ,
9.81 for Sr i
L anka, 8.99 for
T aiwan,
8.47
for
T hailand ,
and 12.98 for T unis ia. T he cr itical value is
6.25. T able 1
pr es ents
the
cointegr ation
tes ts
r es ults ,
wher e
r d enotes the num ber of
cointegr ating
vector s . F or A-m ax
and tr ace s tatis tics , the null
hypothes is is that ther e ar e, at
m os t, r
cointegr ating vector s , wher eas the alter native
hy-
pothes es ar e
r +
1
and at leas t r +
1 for the A-m ax and tr ace
s tatis tics , r es pectively.
Star ting
withthe A-m ax tes t r es ults , the null
hypothes is
r
= 0
(no cointegr ation) is
r ejected
infavor of r
=
1 ineach
countr y.
T he calculated tes t s tatis tics
r ange fr om a low of
28.26 inT unis ia to a
high
of 95.86 inMalawi. T he cr itical
value at the 5% level fr om Os ter wald -L enum (1992, p. 468)
is 27.07. F ur ther m or e, the null
hypothes es of r
1,
r
K
2,
and r
<
3 cannot be
r ejected
infavor of the alter native
hy-
pothes es of r =
2, r
=
3, and r
=
4, r es pectively. T hes e
r es ults ind icate the
pr es ence
of one
cointegr ating r elation-
s hip
for each
countr y.
12 Jour nal of Bus ines s & Econom ic Statis tics , Januar y
2000
T able 1. Res ults F r om
C ointegr ation
T es ts
Maxim um
eigenvalue
T r ace s tatis tics
Ho:
r = 0 r < 1 r <2 r <3 r = 0 r <1 r <2 r <3
C ountr y Ha:
r = 1 r = 2 r = 3 r = 4 r > 1 r > 2 r > 3 r = 4
Ecuad or 35.66 15.14 4.07 1.02 55.88 20.23 5.09 1.02
Ind ones ia 28.41 12.18 8.07 2.99 51.65 23.24 11.05 2.99
Kor ea 35.92 17.87 5.95 1.81 61.54 25.62 7.75 1.81
Malays ia
32.70 13.80 6.84 .18 53.52 20.82 7.02 .18
Malawi 95.86 14.90 13.54 .06 124.35 28.49 13.59 .06
Maur itius 39.18 20.90 7.41 .65 68.15 28.96 8.06 .65
Mexico 34.98 13.80 5.40 .02 54.20 19.22 5.42 .02
Mor occo 35.45 13.66 5.73 2.94 57.77 22.32 8.67 2.94
Philippines
31.62 10.84 4.33 2.80 49.58 17.96 7.13 2.80
Sr i L anka 57.04 17.59 4.73 .94 80.29 23.26 5.67 .94
T aiwan 56.01 12.83 6.47 .52 75.82 19.82 6.99 .52
T hailand 38.20 12.71 10.05 .01 60.97 22.77 10.06 .01
T unis ia 28.26 14.45 4.50 .73 47.95 19.69 5.23 .73
C r itical value
(5%)
27.07 20.97 14.07 3.76 47.21 29.68 15.41 3.76
NOT E: r d enotes thenum ber of
cointegr ating
vector s . T he
laglengths us ed ar e2 inMalawi, the
Philippines ,
and Sr i L anka; 3 inMexico; 4 inEcuad or , Malays ia, Maur itius , Mor occo, and T aiwan;
and 5 inKor ea, Ind ones ia, T hailand , and T unis ia.
F or the tr ace-tes t
r es ults ,
we obtains im ilar conclus ions
whenthe null
hypothes is
of r = 0 is tes ted
agains t
the alter -
native
hypothes is
of r > 1
ineach
countr y. Mor eover ,
the
null
hypothes es
r
<
1,
r
<
2,
and r < 3 cannot be
r ejected
inall countr ies . In
s um ,
we as s um e the
pr es ence
of one
cointegr ating
vector for each
countr y
inour
s am ple.
T his
find ing s ugges ts
that ther e is a
long-r unequilibr ium
r ela-
tions hip am ong
r eal
expor ts , for eign
econom ic
activity,
r el-
ative
pr ice,
and
exchange-r ate volatility
inall the countr ies
inour
s am ple.
It is wor th
m entioning
that when
exchange-
r ate
volatility
is om itted fr om the
cointegr ation' equation,
ther e is evid ence of
cointegr ationam ong
r eal
expor ts ,
for -
eign
econom ic
activity,
and r elative
pr ice
inall countr ies
(except
for
Ecuad or , Ind ones ia, Kor ea, Malays ia, Mexico,
T hailand ,
and
T unis ia).
T able 2
pr ovid es par am eter
es tim ates that
r epr es ent long-
r un
elas ticities , together
withtheir
r es pective as ym ptotic
s tand ar d er r or s . T hes e elas ticities ar e obtained
by
nor m al-
izing
the es tim ates of the uncons tr ained
cointegr ating
vec-
tor s onr eal
expor ts . Except
for Sr i
L anka,
the
L ,
tes t for
par am eter cons tancy
inthe
cointegr ationr elations hip pr o-
pos ed by
Hans en(1992a)
ind icates that eachnor m alized
equationcaptur es
a s table
r elations hip.
As canbe s eeninT able
2,
the es tim ated
for eign
econom ic
activity (wt) elas ticity
car r ies the
expected pos itive s ign
and
is
s ignificantly
d iffer ent fr om 0
(at
the 5%
level)
inall the
countr ies inour
s am ple.
T he
long-r un
incom e
elas ticity
is
gr eater
than
unity
inall
countr ies , gr eater
thantwo in10 of
the 13
countr ies ,
and
gr eater
thanthr ee in6 countr ies . T her e
ar e s ever al
explanations
for the
r elatively high
incom e elas -
ticities .
F ir s t,
and
for em os t,
note that the values for the
incom e elas ticities ar e cons is tent withes tim ates found in
other s tud ies . As noted
by
Ried el
(1988),
m os t es tim ates of
incom e elas ticities in
expor t-d em and equations ,
"whether
for
d eveloped
or
d eveloping
countr ies ,
or for
countr y ag-
gr egates
or inind ivid ual countr ies , gener ally
lie inthe
r ange
between2.0 and 4.0"
(p.
140).
Of the s ix s tud ies s ur veilled
by Mar quez
and
McNeilly (1988,
table
1, p. 307),
four r e-
por ted
incom e elas ticities
gr eater
than2 and thr ee
r epor ted
elas ticities
gr eater
thanthan3. Ried el
(1988)
es tim ated the
incom e
elas ticity
for
Hong Kong' s expor ts
of m anufactur es
to be
gr eater
than4.
Ried el
(1988, 1989) conjectur ed
that the
high
elas ticities
found inthe liter atur e r eflect the
inad equate
tr eatm ent of
boththe
s upply
s id e of
expor ts
and the nor m alizationis s ue.
His es tim ate of a
s im ultaneous -equation
m od el withex-
por t
d em and nor m alized as a
pr ice equationyield s
a lower
incom e
elas ticity.
F or a
cr itique
of Ried el' s
appr oach,
s ee
T able 2. Es tim ates of the
C ointegr atingRelations hips
Stability
tes t:
C ountr y
Nor m alized
cointegr ating
vector Hans en' s
L c
Ecuad or Qt
=
1.60wt
-
.115pt
-
.2lat
.398
(.60) (.04) (.05) [.20]
Ind ones ia Qt
=
3.37wt
+
.197pt
-
.24ot .378
(1.23) (.18) (.08) [.20]
Kor ea Qt
= 3.01 wt
-
1.673pt
-
.36ot .415
(.31) (.13) (.08) [.18]
Malays ia Qt
=
3.12wt
-
1.529pt
-
.55ct
.699
(.16) (.15) (.08) [.07]
Malawi Qt
=
1.0lwt
-
.839pt
-
.18ot
.550
(.42) (.09) (.08) [.20]
Maur itius Qt
= 2.31 wt
-
1.345pt
-
.59ot .575
(.73) (.28) (.21) [.20]
Mexico Qt
=
4.19wt
-
.452pt
-
.85ct
.233
(1.14) (.17) (.39) [.20]
Mor occo Qt
=
2.14wt
-
.669pt
-
.15ot
.740
(.34) (.09) (.03) [.06]
Philippines Qt
=
2.72wt
-
.686pt
-
.13ot
.439
(.25) (.05) (.04) [.20]
Sr i L anka Qt
=
2.01 wt
+
.881pt
-
.29ot
2.287*
(.21) (.30) (.08) [.01]
T aiwan Qt
=
3.75wt
-
.269pt
-
.10ct
.475
(.08) (.08) (.03) [.20]
T hailand Qt
=
3.69wt
-
1.290pt
-
.70at
.757
(.48) (.22) (.10) [.10]
T unis ia Qt
=
1.53wt
-
.420pt
-
.15ot .736
(.14) (.12) (.04) [.06]
NOT E: T henum ber s in
par enthes es
beneaththees tim ated coefficients ar ethes tand ar d er r or s .
Inthecas eof Mexico,
awas d eter m ined
by
a
m ovingaver age
of 5.
Pr elim inar y
exam ination
s ugges ts
that it is better thanthe
m ovingaver age
of 7
(on
thebas is of T heil' s r es id ual var iance
cr iter ion). L cis Hans en' s (1992a) par am eter s tability tes t, and thevalues inbr ackets ar ethep
values . Anas ter is k im plies s ignificance
at the5% level.
Ar ize, Os ang,
and
Slottje: Exchange-Rate Volatility
and
F or eign
T r ad e 13
Nguyen
(1989).
A d iffer ent
explanation
for
high
incom e
elas ticities was
givenby
Ar ize
(1990).
He
ar gued
that an
incr eas ed
penetr ation
of wor ld m ar kets over the
s am ple pe-
r iod
can,
in
par t,
be attr ibuted to the incom e elas ticities of
L D C ' s
being
s om e functionof the incom e elas ticities of
the
expor ts
of the
im por ting
countr ies . T his is
plaus ible
if
expor ts
ar e
lar gely com pos ed
of s em ifinis hed
pr od ucts
that
ar e us ed to
pr od uce
final
pr od ucts
inother countr ies . F i-
nally,
Ad ler
(1970)
s ugges ted
that d iffer ent incom e elas tici-
ties r eflect the extent to which
expor ts
have been
ad apted
to
the
im por ting countr y' s
local
tas tes ,
with
higher elas ticity
pr ovid ing
evid ence of
gr eater ad aptation.
T he es tim ated
pr ice (pt) elas ticity
has the
expected neg-
ative
s ign
in11 of the 13 countr ies s tud ied . F or Ind ones ia
and Sr i
L anka,
we obtain
pos itive pr ice
elas ticities that ar e
s tatis tically ins ignificant
inthe cas e of Ind ones ia and
s ignif-
icant at the 5% level inthe cas e of Sr i L anka.
Ins ignificant
pr ice
effects ar e
gener ally
attr ibuted to at leas t thr ee fac-
tor s . T he fir s t is the us e of unit-value
ind exes ,
whichar e
com puted
fr om obs er vationunits inwhichs om e
aggr ega-
tionhas taken
place
(s ee
L eam er and Ster n
1970).
T hey
ar e
accur ate
only
if the
com pos ition
of the unit r em ains the
s am e or if the net effect of s uch
changes
is
ins ignificant.
T he s econd is that s om e L D C ' s
m ay
have beenable to d if-
fer entiate their
expor ts (s ee
Gr os s m anand
Helpm an1992)
by focus ing
on
nonpr ice
factor s s uchas one-tim e
d eliver y,
d es ignim pr ovem ent, pr od uct var ieties ,
and
aggr es s ive
m ar -
keting.
Inother
wor d s ,
s om e L D C ' s
m ay
be
oper ating
lim it
pr icing
to d eter
entr y
while
com peting
inother ar eas . F i-
nally, pr ice
elas ticities that ar e
pos itive and /or ins ignificant
can
cer tainly
be the r es ult of
poor
d ata
quality typical
for
m any
L D C ' s .
An
appealing as pect
of the r es ults is that the
exchange-
r ate
volatility
(ct)
elas ticities have
negative s igns
and ar e
s tatis tically s ignificant
inthe r es ults for each
countr y.
T he
long-r un
elas ticities
r ange
fr om a low of
.10
inT aiwanto a
high
of .85 in
Mexico, im plying
that
exchange-r ate
volatil-
ity
exer ts a
s ignificant
ad ver s e
long-r un
effect on
expor t
volum e.
We als o exam ined whether
any
var iable inthe four -
var iable
s ys tem
canbe cons id er ed
weakly exogenous (s ee
Er ics s onand Ir ons
1994).
T he d ata inT able 3 s how the r e-
s ults for
tes ting
eachvar iable
ind ivid ually
and for
tes ting
incom e, r elative
pr ices ,
and
exchange-r ate volatility, jointly.
T he tes t r es ults s how that the null
hypothes is
of weak ex-
ogeneity
cannot be
r ejected
for
exchange-r ate
r is k
(except
Sr i L anka) and
for eign
econom ic
activity
in
any countr y,
wher eas the null for the r elative
pr ice
var iable cannot be
r ejected
in11 of the 13 countr ies . T he r es ults als o confir m
that the
expor ts
var iable s hould be cons id er ed
end ogenous .
Inad d ition, the null
hypothes is
of the
joint exogeneity
of
incom e, r elative
pr ices ,
and
exchange-r ate volatility
is con-
fir m ed in9 of the 13 countr ies at a
s ignificance
level of 5%.
T o exam ine the r obus tnes s of our
cointegr ationr es ults ,
Stock and Wats on' s (1993) d ynam ic or d inar y
leas t
s quar es
(OL S) pr oced ur e (D OL S), inwhichOL S is
applied
to
Equa-
tion(1) augm ented withcur r ent and two lead s and
lagged
d iffer ences of all the
r egr es s or s ,
was als o
em ployed .
T he
fact that the D OL S coefficient es tim ates ar e
ver y
s im ilar to
thos e
r epor ted
inT able 2 lets us conclud e that the coeffi-
cient es tim ates inT able 2 ar e not d r iven
by
our choice of
the m ethod of es tim ation.
2.2 Er r or -C or r ection Mod el
T he
Gr anger r epr es entation
theor em
pr oves that,
if a
cointegr ating r elations hip
exis ts
am ong
a s et of
I(1)
s e-
r ies ,
thena
d ynam ic er r or -cor r ection(EC )
r epr es entation
of the d ata als o exis ts . T he
m ethod ology
us ed to find this
r epr es entation
follows the
"gener al-to-s pecific" par ad igm
(s ee Hend r y 1987). Initially,
four
lags
of the fir s t d iffer -
ence of eachvar iable in
Equation(1),
a cons tant
ter m ,
and
one-lagged
EC ter m
(EC t-1) gener ated
fr om the Johans en
pr oced ur e
wer e us ed . T henthe d im ens ions of the
par am -
eter
s pace
wer e r ed uced to a final
par s im onious s pecifica-
tion
by s equentially im pos ing s tatis tically ins ignificant
r e-
s tr ictions or
elim inating ins ignificant
coefficients . Giventhe
pr es ence
of the
volatility
var iable inthe er r or -cor r ection
m od el
(EC M)
and the
end ogeneity
of s om e of the
r egr es -
s or s ,
we us e the ins tr um ental var iables
pr oced ur e s ugges ted
by Pagan
and
Ullah(1988).
T he lis t of ins tr um ental var i-
ables cons is ts of the cons tant
ter m ,
the
lagged
EC
ter m ,
and four
lags
inthe d iffer ences of all var iables includ ed
inthe
long-r un
s olution. Intheir
ar ticle, Pagan
and Ullah
r ecom m end ed the us e of a
heter os ced as ticity-
and s er ial-
cor r elation-cons is tent es tim ator of the covar iance m atr ix.
T o ens ur e that the covar iance is
pos itive s em id efinite,
we
ad jus t Pagan
and
Ullah' s
covar iance es tim ator as
s ugges ted
by Newey
and Wes t
(1987).
T he r es ults ar e s um m ar ized in
T able 4.
C ons id er ing
that each
r egr es s and
inT able 4 is cas t in
fir s t
d iffer ence,
the
em pir ical
r es ults
s ugges t
that the s ta-
tis tical fit of eachm od el to the d ata is
s atis factor y,
as in-
d icated
by
the values of
ad jus ted R2,
which
r ange
fr om
a low of .43 inSr i L anka to a
high
of .83 in
Malays ia.
Mor eover ,
the s tatis tical
appr opr iatenes s
of the
equations
is
s uppor ted by
the
d iagnos tic
tes ts . In
par ticular ,
the
s tability
of eaches tim ated EC M is confir m ed
by
Hans en' s
(1992b)
joint par am eter noncons tancy
and var iance
noncons tancy
T able 3. Res ults of
Exogeneity
T es t
Real
F or eign
Relative
Exchange-
Joint
Sam ple
C ountr ies expor ts incom e pr ices r ate r is k tes t per iod
Ecuad or 3.97* 1.53 1.13 .32 7.87* 73q2-96q1
Ind ones ia 3.02* .92 4.26* 1.89 14.6* 73q2-93q4
Kor ea 4.26* 1.63 1.83 1.76 4.02 73q2-96q1
Malays ia 3.25* .47 .38 1.20 5.04 73q2-94q1
Malawi 12.49* .57 1.24 .99 1.21 73q2-93q1
Maur itius 3.14* 1.75 1.09 .77 3.13 73q2-95q1
Mexico 3.48* 1.54 1.21 1.03 7.83* 73q2-95q1
Mor occo 4.85* 1.78 1.47 1.52 .99 73q2-93q1
Philippines 3.23* 1.56 .45 1.09 3.88 73q2-93q1
Sr i L anka 6.07* 1.34 8.19* 2.77* 69.3* 73q2-96q1
T aiwan 3.80* 1.86 1.49 .89 3.87 73q2-96q1
T hailand 3.67* .92 1.38 1.64 3.56 73q2-96q1
T unis ia 3.96* 1.81 1.13 .87 7.48 73q2-93q2
NOT E: T hed atafor r eal
expor ts ,
wor ld incom e, r elative
pr ices ,
and
for eign-exchange
r is k var i-
ables ar eabs olute t values .
Significance
of the
joint exogeneity
of wor ld incom e, r elative
pr ices ,
and
exchange
r is k is
pr es ented by m eans of Wald s tatis tics ; thus the
d egr ees
of fr eed om ar e3.
Anas ter is k im plies that t-r atio or X2 valueis s ignificant at the5% level.
14 Jour nal of Bus ines s & Econom ic Statis tics , Januar y
2000
coC D oI. ---C P
(D N
V)
"-It I-
C D C ' -
'
-NO-C D C ' OO0 "
C ucli
-
C i a
C
,)0)
C 0 . . . c.
a
D C
o
C D (D
C ~
N-
z-D
' -D
It
-D
C
- - 0ioi -I
r
co
(
OC
D D C O 0' -00
P
- '
- L 6
9 9 cr ' -
I I I7I -I
I t
NC D r - coC D C D -0N
C u C ' v-- L j~I) C l=I-I I
~ ~ D C D - 04
N-C ' O
. ? " 04
o 9- r -,0 9C ,--.C .
D
r o. oc.o
N
E ' -
II "- I I c9.
"m . I I
t
I 0I )I-
C o c
SC ' - OD C "
r
I1 o
o-
C D
C D i C
C D C OC ' -n 0) m
i
' )cD r -
C
o 00 w00
o o0 .
co
w' z - -
-.
-
,- "0. . .
I(0.N N r -C nJN C D
m
-
- -
S
i
' c
,i-
N
.-
0) )
.C ' ) 04i 04-
>0
C D C
C C C D -' -' C D C D C C
7
C 0)
It
0m C n
m ,--jjAII*II _., Oa
It
C D C D
oi
C ?
' >O- co 00
)D C co' NN- 0. D ' -
0
o
oC D
-
C ) 0 C D
?
-0
w
C ' ) -
-04
~ 2U-0) 0C D - C
C 0-
C O C C D C
S,9 ' -:
C D
9. q C q
J
ooC -2
Sn (0
.n- o O- .
0)-
00
L ... o' . "
o. -C . JC ' .) ,-( ,
-
c,-o
-C
C ' .
"-"iiI1I
04 C 1W)1~ ' IttcIt 4 i 0 - C C "C
0)IN-0, C C
C C 0
N-
C O O. C
C -N-C
C o,- C C "
L --N
C
C o
U)z m C C
lb- , C D O
N-
.C ..
liii
?
C C
W
'
(0co 0 (0C D C D
0 )
N-C D U) 0
I
0
0I I
IVC !
(-) C z
co
c D o-C D
(0c999 0 9a
N-r -C D C D C D It0
C D C D (0 C N-C D C D
cli
7
I,
k C C JO D ' -' - D
99 ' C OC t 99' -' -O
- -0
C D ~ C J ' -C J~ ~C D
C C C C ' ) C ' C ' )C C D U- D C C D
0
C ' )
n
00C ' )C D D )0C M10 r 0
co 0) T - L n N C 0C 9(D C D D ' C D D
(64 C C ) C ' ) C ' `4C
--D C C -C D D
N-co 0) 99 (0 (0
co ItIItIcoIco 0)
OC O
n' 1 C D C O
~lco
-
-
C D
L I) C ' L I
U~' ) M M 0 X
0 14, L n
C C P-L 0( )
C O
C C N C D C li 9 9 *M X
0 c' cD t ocD 00 (0 co
k
0IC ' C
C D D C ~~C
(0
~
l ) C D C D N-0)N- C ' - -r
co r -co
L
I6 v
-li i ci T - -
Ar ize, Os ang,
and
Slottje: Exchange-Rate Volatility
and
F or eign
T r ad e 15
(jt
and
var )
tes ts for
s tationar y
d ata.
Mor eover ,
eaches ti-
m ated m od el fulfills the cond itions of s er ial
noncor r elation,
hom os ced as ticity,
zer o d is tur bance m ean
(i.e.,
no
s pecifica-
tion
er r or s ),
and
nor m ality
of r es id uals .
F inally,
we als o tes t
the
as s um ption
of s hor t-r un
pr ice hom ogeneity.
We find that
the
as s um ption
cannot be
r ejected except
for two countr ies
(Ecuad or
and
Malays ia).
F or thes e
countr ies ,
we us e
expor t
pr ice
and wor ld
expor t pr ice
as
s epar ate
var iables in
place
of the r elative
pr ice
var iable.
Having pr ovid ed
evid ence
s uppor ting
the
ad equacy
of the
es tim ated
equations ,
we canm ake the
following
obs er va-
tions
r egar d ing
the obtained es tim ates :
F ir s t,
the EC ter m ' s coefficient is
s tatis tically s ignifi-
cant ineachof the 13 cas es and is
always negative.
T hes e
find ings s uppor t
the
valid ity
of an
equilibr ium r elations hip
am ong
the var iables ineach
cointegr ating equation.
T his
im plies
that
over looking
the
cointegr ated nes s
of the var i-
ables would have intr od uced
m is s pecification
inthe und er -
lying d ynam ic s tr uctur e,
and note als o that
higher power s
of
EC t_1
wer e includ ed ineach
countr y' s equation
but
pr oved
s tatis tically ins ignificant.
Second ,
the
change
inr eal
expor ts per quar ter
that is at-
tr ibuted to the
d is equilibr ium
betweenthe actual and
equi-
libr ium levels is m eas ur ed
by
the abs olute values of the
EC ter m of each
equation.
T her e is cons id er able inter coun-
tr y
var iation,
inthe
ad jus tm ent s peed
to the las t
per iod ' s
d is equilibr ium ,
with
Malawi having
the
lar ges t
value and
Mexico the s m alles t. T his
im plies
that the
ad jus tm ent
of
expor t
volum e to
changes
inthe
r egr es s or s m ay
take about
one
quar ter
inMalawi to
s lightly
m or e than16
quar ter s
in
Mexico. T he r es ults
point
to the exis tence of m ar ket for ces
inthe
expor t
m ar ket that
oper ate
to r es tor e
long-r unequi-
libr ium after a s hor t-r und is tur bance.
T hir d ,
the
d ynam ics
of the
equation
s how that
changes
in
for eign
econom ic
activity,
r elative
pr ice,
and
exchange-
r ate
volatility
have
s ignificant
s hor t-r uneffects on
expor ts .
T hes e r es ults ar e s um m ar ized inT able 5.
T he evid ence s hows
that,
for the
m ajor ity
of countr ies
inthe
s am ple, expor t
volum e
r es pond s
fas ter to
for eign
econom ic
activity changes
thanto r elative
pr ice changes .
Mor e
im por tantly,
the r es ult als o ind icates that
exchange-
r ate
volatility, ignor ed
in
s ever al
pr evious
L D C ' s
expor t-
T able 5. MeanT im e
L ags for Ad jus tm ent of Expor ts *
Exchange-r ate
C ountr y F or eignincom e Relative pr ice volatility
Ecuad or .77 6.54 5.86
Ind ones ia 23.89 .30 3.01
Kor ea 17.91 10.76 5.87
Malays ia 4.69 6.29 4.87
Malawi .47 1.50 .68
Maur itius .61 4.74 2.47
Mexico 24.69 33.88 15.37
Mor occo 1.20 3.08 1.64
Philippines 1.13 2.81 1.46
Sr i L anka 4.35 18.14 13.85
T aiwan 3.49 2.65 1.85
T hailand 1.98 18.16 10.23
T unis ia 1.39 2.78 1.80
*
Abs olutevalues .
d em and
s tud ies ,
has a s ubs tantial s hor t-r uneffect on
expor t
d em and ,
inad d itionto its
long-r un
effect.
T hus , neglect
of
s ucha var iable can
pr od uce
bias ed r es ults . It canbe d ed uced
that tr ad itional
expor t-d em and
s tud ies for other econom ies
that d o not includ e a var iable
r epr es enting
the influence of
exchange
r is ks ar e
potentially m is s pecified .
T his evid ence
fur ther
s ugges ts
that
exchange-r ate volatility m ay
have
s ig-
nificant effects onthe allocationof r es our ces as m ar ket
par -
ticipants attem pt
to m inim ize
expos ur e
to the effects of ex-
change
r is k.
3. SUMMARY AND C ONC L USIONS
Our r es ults
concer ning
the effects of
exchange-r ate
volatility
on
expor t
flows
s ugges t
that ther e is a
negative
and
s tatis tically s ignificant long-r unr elations hip
between
expor t
flows and
exchange-r ate volatility
ineachof the 13
L D C ' s . Inm os t of the
countr ies , exchange-r ate volatility
has a s hor t-r uneffect on
expor t
flows and ther e is s ub-
s tantive caus al
r elations hip
inwhich
changes
in
exchange-
r ate
volatility Gr anger -caus e changes
inr eal
expor ts .
F ur -
ther m or e,
and
per haps
m or e
s ubs tantive,
evid ence for the
contr ibutionof the
exchange-r ate volatility
is that its in-
clus ion
appear s neces s ar y
inm os t of the countr ies for the
es tim ated
expor t
d em and
equation
to exhibit the d es ir ed
pr oper ty
of
cointegr ation.
T hes e r es ults
ar e,
onthe
whole,
cons is tent,
at leas t in
s pir it,
withthos e obtained
by pr evious
s tud ies that have exam ined the
expor t
behavior of L D C ' s .
T he wor k
by
C oes
(1981)
us ed a
log-level s pecification
to
exam ine Br azilian
expor ts (annual
d ata for
1965-1974)
and
conclud ed that a
s ignificant
r ed uctionin
exchange-r ate
un-
cer tainty
inthe
countr y' s econom y d ur ing
the
cr awling-peg
er a had a
pos itive
effect onthe
countr y' s expor ts
after the
cr awling peg
was
ad opted
in1968. T he
s tud y by
Br ad a and
Mend ez
(1988) includ ed
14
d eveloping
countr ies and con-
clud ed that
exchange-r ate uncer tainty
inhibits bilater al ex-
por ts . T hey
d id not us e a m eas ur e of
exchange-r ate
volatil-
ity
but ins tead r elied ona
d um m y
var iable that was
as s igned
to countr ies that fixed or floated their
exchange
r ates . C a-
baller o and C or bo
(1989)
us ed a
Koyck-type
m od el and
r eal bilater al
exchange-r ate volatility
m eas ur e to es tim ate an
expor t-d em and equation
for
C hile, C olom bia, Per u, Philip-
pines , T hailand ,
and
T ur key. T hey
conclud ed that ther e is a
s tr ong negative
effect of r eal
exchange-r ate uncer tainty
on
the
expor ts
of all thes e countr ies .
T he
appr oach
we have us ed her e to
inves tigate
the r ela-
tions hip
between
expor t
flows and
exchange-r ate volatility
for 13 L D C ' s is char acter ized
by
two
im por tant
elem ents .
F ir s t, the d atas et for each
countr y
cover s the cur r ent float-
ing exchange-r ate
er a and allows the
analys is
to ad d r es s the
s tability
over tim e of the es tim ated
d ynam ic
m od els
d ur ing
this
per iod .
T his is es s ential for
appr opr iate policy
conclu-
s ions to be infer r ed fr om the es tim ated r es ults . Second , by
cons id er ing
anEC M, this
s tud y pr ovid es
es tim ates of the
s peed
of
ad jus tm ent
or the
aver age
tim e
lag
for
ad jus tm ent
of
expor ts
to
changes
inthe
explanator y
var iables as well as
the s hor t-r uneffects of
exchange-r ate volatility
on
expor ts .
Mentions hould als o be m ad e that eaches tim ated m od el
s atis fies
s ever al
r ecently d eveloped econom etr ic tes ts inthe
16 Jour nal of Bus ines s & Econom ic Statis tics , Januar y
2000
analys is
of tim e s er ies d ata for is s ues s uchas
cointegr a-
tion, s tationar ity, s pecificationer r or s ,
r es id ual autocor r ela-
tion, heter os ced as ticity,
r es id ual
nor m ality,
and s tr uctur al
s tability.
T hes e
em pir ical r es ults ,
d er ived as
they
ar e fr om the
L D C ' s
d ata,
ar e cons is tent withthe theor etical cons id er -
ations d is cus s ed inSection1
and als o confir m
pr evious
r e-
cent r es ear chd one for
d eveloped
countr ies
[e.g.,
C howd -
hur y 1993;
Ar ize
1995],
which
s ugges ts
that
exchange-r ate
volatility
has a
s ignificant negative im pact
on
expor t
flows
inboththe s hor t r unand the
long
r un.
F inally,
to the extent that one can
gener alize
fr om a s am -
ple
of 13
L D C ' s ,
the r es ults
pr es ented
inthis ar ticle ar e
im por tant. T hey s ugges t
that
exchange-r ate volatility
con-
s id er ations ar e
im por tant
for
m od eling expor t
behavior in
L D C ' s . As a
r es ult,
the
d es ign
and
im plem entation
of tr ad e
and
exchange-r ate policies
inL D C ' s s hould benefit fr om
knowled ge
of boththe exis tence and the
d egr ee
of
for eign
exchange-r ate volatility.
It is clear that
policy
actions aim ed
at
s tabilizing
the
expor t
m ar ket ar e
likely
to
gener ate
r es ults
that
ar e,
at
bes t, uncer tain,
if
policy
m aker s
ignor e
the s ta-
bility,
as well as the
level,
of the r eal effective
exchange
r ate. Onthe other
hand , given
the
high
r ate of inflationin
m os t
L D C ' s ,
an
exchange-r ate policy by
its elf would not
elim inate all
exchange-r ate volatility; ther efor e,
L D C
pol-
icy
m aker s s hould als o
pay
attentionto fis cal
policy.
AC KNOWL ED GMENT S
T his is a cond ens ed ver s ionof
Ar ize, Os ang,
and Slot-
tje (1997),
whichintur nis an
extens ively
r evis ed ver s ion
of a
pr evious paper .
We would like to thank Ed Man-
ton,
Keith
McF ar land ,
L ee
Schm id t,
and Wend ell Ed war d s
for
helpful s ugges tions .
We
gr eatly appr eciate
com m ents
fr om a
coed itor ,
anas s ociate
ed itor ,
and a r efer ee that
s ig-
nificantly im pr oved
the
quality
of the ar ticle. Pee
Yong-
vanich,
Nukor nT her eer attanwibool, Vor apoj Pr as anpanich,
Pr akai
C hooekawong,
Jittad a
Kitiyam ongkol,
and Kathleen
Sm ith
pr ovid ed
excellent r es ear chas s is tance. T his r es ear ch
is fund ed
by
a GSRF -T AMU-C
gr ant.
APPEND IX:
D AT A, D EF INIT ION,
AND SOURC ES
T his
append ix
d es cr ibes the r aw
d ata, s our ces ,
and con-
s tr uctionof var iables us ed inthe
em pir ical
tes ts . All d ata
wer e obtained fr om the Inter national
Monetar y
F und ' s
(IMF ' s ) Inter national F inancial Statis tics (IF S), IMF ' s C en-
tr al Statis tics Office, Or ganization
for Econom ic
C ooper a-
tionand
D evelopm ent
MainEconom ic Ind icator s , and D i-
r ections of T r ad e.
F or eign
econom ic activity
is
pr oxied by
"wor ld " r eal in-
com e
expr es s ed
as anind ex, 1980 = 100. Wor ld incom e
for each
countr y
was cons tr ucted as a
geom etr ic aver age
of the ind ex of the r eal incom e in17 countr ies . T he 17
countr ies us ed ar e as follows : Aus tr alia, Belgium , C anad a,
D enm ar k, F inland , F r ance, Ger m any, Ir eland , Italy, Japan,
New Zealand , the Nether land s , Nor way, Swed en, Switzer -
land , the United
Kingd om ,
and the United States , r es pec-
tively. F ollowing Gold s teinand Khan(1978, p. 285), this
s er ies was calculated fir s t as anannual s er ies
(d ue
to lack
of
quar ter ly
d ata onr eal incom e in
m any countr ies )
and
thenconver ted to a
quar ter ly
bas is
by us ing
a
quad r atic
inter polation
m ethod
they
r ecom m end ed .
D ata for ind ivid ual
countr y' s expor t
volum e and unit
values wer e takenfr om
IF S,
and the wor ld
expor t pr ice
ind ex, PWt,
is a
geom etr ic weighted aver age
of
expor t
pr ices .
T he
weights
ar e
wji,
and the bas e
per iod
is 1980
=
100. T he r elative
pr ice
r atio was calculated as
Pt
=
InPXt
-
InEt
-
In
PWt,
wher e PX is inlocal
cur r ency,
E
is
exchange-r ate ind ex,
and PW is wor ld
pr ice
ind ex.
T o
com pute
m eas ur es for
exchange-r ate volatility,
tr ad e
weighted
effective
exchange
r ate
(eer )
and r eal effective ex-
change
r ate
(r eer )
wer e
com puted .
F or
exam ple,
for Ko-
r ea
they
wer e cons tr ucted as follows : T he
per iod aver age
exchange
r ates ar e inunits of d om es tic
cur r ency per
d ol-
lar . T hes e
per iod aver ages
wer e then
expr es s ed
inind ex
for m
(1980
=
1.0).
T he eer var iable was calculated as
exp[[Ewji
In
E(i, $, t)
- In
E(J, $, t)],
wher e
exp
=
expo-
nent, In
= natur al
logar ithm , E(i, $, t)
=
exchange-r ate
in-
d ex of
countr y
i at tim e
t,
and
E(J, $, t)
=
exchange-r ate
in-
d ex of Kor ea at tim e t. T he r eal effective
exchange
r ate was
calculated as r eer
(J, t)
=
exp[-
In
P(J, t)
+ In
E(J, $, t)
+
Ewji InP(i, t)-E wji InE(i, $, t)], wher e J s tand s for Ko-
r ea and the
exchange-r ate
ter m s ar e inunits of J or i cur -
r ency per
U.S. d ollar s inind ex for m
(1980
=
1.0).
P is the
cons um er
pr ice
ind ex of
countr y
J or i inind ex for m
(1980
=
1.0).
[Received Septem ber
1996. Revis ed
Apr il 1999.]
REF ERENC ES
Ad ler , M. F . (1970), "T he
Relations hip
Betweenthe Incom e and Pr ice
Elas ticities of D em and for United States
Expor ts ,"
Review
of
Econom ics
and Statis tics , 52, 313-319.
Ar ize, A. C . (1990),
"AnEconom etr ic
Inves tigation
of
Expor t
Behavior in
SevenAs ian
D evelopingEconom ies ," Applied Econom ics , 22, 891-904.
-
(1995), "T he Effects of
Exchange-r ate Volatility
onU.S.
Expor ts :
An
Em pir ical Inves tigation,"
Souther nEconom ic Jour nal, 62, 34-43.
Ar ize, A. C ., Os ang, T ., and
Slottje,
D . J. (1997), "Exchange-r ate Volatility
and
F or eign
T r ad e: Evid ence F r om T hir teenL D C s ," d is cus s ion
paper ,
T exas A & M
Univer s ity-C om m er ce, C ollege
of Bus ines s and T echnol-
ogy.
Baba, Y., Hend r y,
D . F ., and Star r ,
R. M. (1992), "T he D em and for Ml in
the U.S.A., 1960-1988," Review
of
Econom ic Stud ies , 59, 25-61.
Bald win, R., and
Kr ugm an,
P.
(1989), "Per s is tent T r ad e Effects of
L ar ge
Exchange
Rate Shocks ," Quar ter ly
Jour nal
of Econom ics , 104, 635-665.
Br ad a, J. C ., and Mend ez, J. A.
(1988), "Exchange
Rate Ris k, Exchange
Rate
Regim e
and the Volum e of Inter national T r ad e," Kyklos , 41, 263-
280.
C aballer o, R. J., and C or bo, V. (1989), "T he Effect of Real
Exchange
Rate
Uncer tainty
on
Expor ts : Em pir ical Evid ence," T he Wor ld Bank Eco-
nom ic Review, 3, 263-278.
C howd hur y,
A. R. (1993), "D oes
Exchange
Rate
Volatility D epr es s
T r ad e
F lows ? Evid ence F r om Er r or -C or r ection Mod el," Review
of
Econom ics
and Statis tics , 75, 700-706.
C oes , D . V. (1981), "T he
C r awlingPeg
and
Exchange
Rate
Uncer tainty,"
in
Exchange
Rates Rules , ed . J. William s on, New Yor k: St. Mar tins , pp.
113-136.
C us hm an, D . 0. (1988), "U.S. Bilater al T r ad e F lows and
Exchange
Ris k
D ur ingthe F loatingPer iod ," Jour nal
of
Inter national Econom ics , 25,
317-330.
D e Gr auwe, P. (1988), "Exchange
Rate
Var iability
and the Slowd ownin
Gr owthof Inter national T r ad e," IMF
Staff Paper s , 35, 63-84.
Ar ize, Os ang,
and
Slottje: Exchange-Rate Volatility
and
F or eign
T r ad e 17
D ixit, A. (1989), "Hys ter es is , Im por t Penetr ation, and
Exchange-Rate
Pas s -
T hr ough," Quar ter ly
Jour nal
of Econom ics , 104, 205-227.
Er ics s on, N. R., and Ir ons , S. I. (ed s .) (1994), T es tingExogeneity,
L ond on:
Oxfor d
Univer s ity
Pr es s .
F r oot,
K.
A.,
and
Klem per er ,
P. (1989), "Exchange
Rate
Pas s -thr ough
WhenMar ket Shar e Matter s ," Am er icanEconom ic Review, 79, 637-
654.
Gold s tein, M., and Khan,
M.
(1978),
"T he
Supply
and D em and for
Expor ts :
A Sim ultaneous
Appr oach,"
Review
of
Econom ics and Statis tics , 60,
275-286.
Gr os s m an, G.,
and
Helpm an,
E. (1992), Innovationand Gr owthinthe
Global
Econom y, C am br id ge,
MA: MIT Pr es s .
Hans en,
B. E. (1992a), "T es ting
for Par am eter
Ins tability
inL inear Mod -
els ," Jour nal
of Policy Mod eling, 14,
517-533.
( (1992b), "T es ting
for Par am eter
Ins tability
in
Regr es s ions
With
I(1)
Pr oces s es ," Jour nal
of
Bus ines s & Econom ic Statis tics , 10, 321-335.
Hend r y,
D . F . (1987), "Econom etr ic
Method ology:
A Per s onal
Per s pec-
tive,"
inAd vances inEconom etr ics , 2, ed . T .
Bewley, C am br id ge,
U.K.:
C am br id ge Univer s ity Pr es s , pp.
29-48.
Hooper , P.,
and
Kohlhagen,
S. W. (1978),
"T he Effect of
Exchange
Rate
Uncer tainty
onthe Pr ices and Volum e of Inter national T r ad e," Jour nal
of
Inter national Econom ics , 8, 483-511.
Johans en, S. (1988), "Statis tical
Analys is
of
C ointegr ating Vector s ," Jour -
nal
of
Econom ic
D ynam ics
and C ontr ol, 12, 231-254.
Johans en, S., and Jus elius , K. (1990),
"Maxim um L ikelihood Es tim ation
and Infer ence on
C ointegr ation-WithApplications
to D em and for D e-
m and ," Oxfor d
Bulletin
of
Econom ics and Statis tics , 52, 169-210.
Kenen,
P. T ., and Rod r ik,
D . (1986), "Meas ur ing
and
Analyzing
the Effects
of Shor t-ter m
Volatility
inReal
Exchange Rates ," Review
of
Econom ics
and Statis tics , 68, 311-315.
Kor ay, F ., and
L as tr apes ,
W. D . (1989), "Real
Exchange
Rate
Volatility
and U.S. Bilater al T r ad e: A VAR
Appr oach,"
Review
of
Econom ics and
Statis tics , 71, 708-712.
L eam er , E. E., and
Ster n, R. M. (1970), Quantitative
Inter national Eco-
nom ics , Bos ton:
Allyn
and Bacon.
Mar quez, J., and
McNeilly,
C . (1988), "Incom e and Pr ice Elas ticities for
Expor ts
of
D evelopingC ountr ies ," Review
of
Econom ics and Statis tics ,
70, 306-314.
Med hor a, R. (1990), "T he Effect of
Exchange
Rate
Var iability
onT r ad e:
T he C as e of the Wes t Afr ican
Monetar y
Union' s
Im por ts ,"
Wor ld D e-
velopm ent, 18, 313-324.
Newey, W.,
and
Wes t, K. (1987), "A
Sim ple,
Pos itive Sem i-D efinite,
Heter os ked as ticity
and Autocor r elation C ons is tent C ovar iance Matr ix,"
Econom etr ica, 55, 703-708.
Nguyen,
D . T .
(1989), "T he D em and for L D C
Expor ts
of Manufactur es :
Es tim ates F r om
HongKong:
A C om m ent," Econom ic Jour nal, 99, 461-
466.
Os ter wald -L enum , M.
(1992),
"A Note With
Quantiles
of the
As ym ptotic
D is tr ibutions of the Maxim um L ikelihood
C ointegr ation
Ranks T es t
Statis tics : F our C as es ," Oxfor d
Bulletin
of
Econom ics and
Statis tics , 54,
461-472.
Pagan, A., and Ullah, A. (1988), "T he Econom etr ic
Analys is
of Mod els
WithRis k
T er m s ," Jour nal
of Applied Econom etr ics , 3, 87-105.
Ried el, J. (1988), "T he D em and for L D C
Expor ts
of Manufactur es : Es ti-
m ates F r om
HongKong,"
T he Econom ic Jour nal, 98, 138-148.
-
(1989),
"T he D em and for L D C
Expor ts
of Manufactur es : Es tim ates
F r om
HongKong:
A
Rejoind er ,"
T he Econom ic Jour nal, 99, 467-470.
Stock, J.,
and
Wats on,
J.
(1993),
"A
Sim ple
Es tim ator of
C ointegr ating
Vector s in
Higher
Or d er
Integr ated Sys tem s ," Econom etr ica, 61, 783-
820.
T hur s by,
M. C ., and
T hur s by,
J. G. (1987), "Bilater al T r ad e F lows , L end er
Hypothes is ,
and
Exchange Ris k," Review
of
Econom ics and Statis tics ,
69, 488-495.

Das könnte Ihnen auch gefallen