Güneş, Didem; Ozkan, Ib ahim; E den, Lü i
A icle
Economic sen imen and o eign po olio lows: E idence
om Tü kiye
Cen al Bank Re iew (CBR)
P o ided in Coope a ion wi h:
Cen al Bank o The Republic o Tu key, Anka a
Sugges ed Ci a ion: Güneş, Didem; Ozkan, Ib ahim; E den, Lü i (2024) : Economic sen imen and
o eign po olio lows: E idence om Tü kiye, Cen al Bank Re iew (CBR), ISSN 1303-0701, Else ie ,
Ams e dam, Vol. 24, Iss. 1, pp. 1-9,
h ps://doi.o g/10.1016/j.cb e .2024.100147
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/297969
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by-nc-nd/4.0/
CBREV 24 (2024) 100147
A ailable online 16 Janua y 2024
1303-0701/© 2024 The Au ho s. Published by Else ie B.V. on behal o Cen al Bank o The Republic o Tu key. This is an open access a icle unde he CC
BY-NC-ND license (h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/).
Economic sen imen and o eign po olio lows: E idence om Tü kiye
Didem Günes
¸
a
,
*
, ˙
Ib ahim ¨
Ozkan
b
, Lü i E den
c
a
Hace epe Uni e si y, Depa men o Economics, Me kez Bankası Üm aniye Ye les¸kesi, Sa ay Mah. Küçüksu Cad. No:64, Üm aniye/˙
Is anbul, Tu key
b
Çankaya Uni e si y, Depa men o Managemen In o ma ion Sys ems, Yuka ıyu çu Mahallesi, Mima Sinan Caddesi No:4, 06790, E imesgu , Anka a, Tu key
c
Hace epe Uni e si y, Depa men o Economics, Hace epe Üni e si esi Bey epe, 06800, Anka a, Tu key
ARTICLE INFO
JEL classi ica ion:
E7
F3
G4
G15
Keywo ds:
Po olio lows
Economic sen imen
ARDL
Tex analy ics
Sen imen analysis
ABSTRACT
The no able su ge in capi al lows in ecen yea s has eme ged as a key ac o shaping he dynamics o in e -
na ional inancial ma ke s and in luencing economic pe o mance o eme ging economies. E en hough mac-
oeconomic undamen als o an economy can explain some o he pa e ns in in e na ional capi al lows,
beha io al ac o s also seem o be essen ial o posi ioning capi al lows ac oss coun ies. In his s udy, we aim o
examine whe he o e all economic sen imen owa ds Tu kish economy plays a signi ican ole on ne po olio
lows o Tü kiye. To his end, we i s cons uc a no el ex -based sen imen index called "Tu kish Economic
Sen imen Index (TESI)", o cap u e he beha io al endencies o in e na ional in es o s and media owa ds
Tü kiye. Ou subsequen s ep in eg a es TESI in o au o eg essi e dis ibu ed lag models (ARDL) alongside majo
pull-push de e minan s o assess whe he ma ke sen imen holds disce nible in luence on capi al in lux in o
Tu key. The esul s e eal ha he TESI and VIX s and ou as pi o al de e minan s in luencing in e na ional
po olio lows. The TESI has a posi i e impac on po olio low dynamics, whe eas he deg ee o global isk
a e sion in e sely a ec s hese lows. These indings align wi h he con en ion ha a a o able sen imen can
boos po olio in lows o eme ging ma ke s. Con e sely, heigh ened ola ili y expec a ions in global ma ke s can
p omp ou lows om hese economies.
1. In oduc ion
The ma ked inc ease in capi al lows o e ecen decades appea s o
be a signi ican ac o in de ining in e na ional inancial ma ke dy-
namics and in luencing economic ajec o ies. While capi al lows can be
ha m ul o bene icial depending on he size, ola ili y, and iming o in/
ou lows, in an inc easingly in e connec ed global economy, hey
appea o a ec economic de elopmen , echnology ans e , employ-
men c ea ion, isk di e si ica ion, s abiliza ion o local inancial ma -
ke s, and in eg a ion in o he global economy, pa icula ly o eme ging
coun ies. The e o e, he e has been a g owing academic a en ion o he
ques ion o wha ac o s d i e in e na ional capi al lows.
In a ounda ional s udy by Cal o e al. (1993), a lis o pull (domes ic)
and push (global) ac o s is pu o wa d as he majo de e minan s. Thei
esea ch sugges s ha global de e minan s, such as subdued US in e es
a es and global cyclical ends, a e pi o al in shaping capi al lows.
Howe e , subsequen esea ch employing a simila lens o e s a ied
conclusions. While some s udies (Fe nandez-A ias 1996; Albuque que
e al., 2005) suppo Cal o e al. (1993)’s ini ial asse ions, o he s
spo ligh he impo ance o domes ic elemen s such as domes ic money
supply and p oduc i i y g ow h (Mon iel and Reinha 1999; De Vi a
and Kyaw 2008). O he esea ch also emphasizes he conjoined signi -
icance o global and domes ic in luence s (Chuhan e al., 1998; Baek,
2006; F a zsche , 2012).
While ea ly empi ical s udies wi hin he con en ional pull-push
amewo k p ima ily examine he in luences o quan i a i e economic
de e minan s, subsequen esea che s u n hei a en ion o quali a i e
ac o s such as ins i u ional in eg i y (Binici e al., 2010; De San is and
Lüh mann, 2009) and beha io al dynamics such as ma ke expec a ions
and in es o sen imen (Te lock, 2007; Hwang e al., 2019). No ably,
he e has been ising in e es in he li e a u e sugges ing ha beha io al
ac o s a e also essen ial o posi ioning capi al lows in oday’s global
inancial landscape. These s udies, concen a ing on ma ke psychology,
posi ha sen imen ega ding he o igin o he des ina ion coun y
plays a signi ican ole in di ec ing in e na ional po olio lows. Spe-
ci ically, i is p oposed ha a posi i e sen imen owa ds a des ina ion
economy se es as a pull mechanism. Con e sely, an ad e se sen imen
ega ding an o igin coun y can ac as a push mechanism, p omp ing
Pee e iew unde esponsibili y o he Cen al Bank o he Republic o Tu key.
* Co esponding au ho .
E-mail add esses: [email p o ec ed] (D. Günes
¸), [email p o ec ed] (˙
I. ¨
Ozkan), [email p o ec ed] (L. E den).
Con en s lis s a ailable a ScienceDi ec
Cen al Bank Re iew
jou nal homepage: www.jou nals.else ie .com/cen al-bank- e iew/
h ps://doi.o g/10.1016/j.cb e .2024.100147
Recei ed 6 Oc obe 2023; Recei ed in e ised o m 28 No embe 2023; Accep ed 8 Janua y 2024
Cen al Bank Re iew 24 (2024) 100147
2
inc eased capi al ou lows om ha na ion o o he s wi h a mo e
a o able ma ke sen imen .
Ma ke sen imen , also known as in es o sen imen , embodies he
collec i e eelings and pe cep ions o ma ke playe s abou o eign in-
es men oppo uni ies. I in eg a es psychological and emo ional
d i e s ha can in luence decisions o buy o sell inancial asse s,
causing luc ua ions in global po olio lows. In ac , he ela ionship
be ween po olio lows and sen imen is no unidi ec ional; i is a
eedback loop whe e each ac o a ec s and ein o ces he o he . To
illus a e, la ge o eign capi al in lows can imp o e sen imen , leading
o u he in es men s and po en ially d i ing up asse p ices.
Howe e , ab up capi al ou lows o a sen imen shi can igge
ma ke down u ns, magni ying pessimis ic in es o eelings and
spawning a downwa d spi al o leeing capi al and heigh ened ma ke
ins abili y. Gi en i s po en ial impac on asse alua ions and o e all
ma ke s abili y, comp ehending he in e ela ion be ween in e na ional
lows and sen imen becomes pa amoun o in es o s, policymake s,
and ma ke pa icipan s. Pa icula ly, acking sen imen is pi o al o
policymake s aiming o coun e balance he possible pe ils posed by
capi al low ola ili y, ensu ing esilience agains sen imen -induced
po olio shi s. Mo eo e , policymake s can in eg a e sen imen me -
ics in o mac oeconomic p ojec ions and isk e alua ions, aiding in he
o mula ion o mo e in o med policies ha bols e ma ke anspa ency,
s eng hen in es o p o ec ion, and imp o e egula o y amewo ks.
Ye , gauging sen imen p esen s challenges, and adi ionally hinges on
su ey-o ien ed app oaches, which o en ace c i icisms due o hei
high cos s and po en ial sampling issues, as Lud igson (2004) no ed.
Thanks o imp o emen s in compu a ional da a analy ics and ising
awa eness o al e na i e da a o ms, ex -based measu es ha e ga ne ed
in e es among academics and policymake s alike. While some no able
s udies ha e c a ed coun y-speci ic ex -based sen imen indica o s
(Shapi o e al., 2020; Ga cia, 2013; F aibe ge , 2016), he e has been a
no iceable gap in de ising such a sen imen me ic o he Tu kish
economy h ough ex analy ics, especially conce ning i s in luence on
capi al in lows o Tu key. As an eme ging ma ke , Tu key s ands as he
19 h la ges economy in he global a ena, boas ing a GDP o app oxi-
ma ely 906 billion US dolla s and being an ac i e membe o in e na-
ional o ganiza ions.
1
Tü kiye also embodies qui e a ac i e inancial
ma ke s o no only domes ic bu also o eign in es o s. Acco ding o
he Wo ld Fede a ion o Exchanges,
2
he Tu kish capi al ma ke anked
20 h in e ms o equi y ading olume and 1s in e ms o u no e a io
in 2021, while he Tu kish bond ma ke anked 15 h based on he
numbe o lis ed co po a e bonds in 2022.
To his end, we ini ially es ablish a no el ex -based sen imen index
o he Tu kish economy, as a beha io al measu e o in e na ional in-
es o s, o he pe iods o Janua y 2004–June 2022 and e alua e i s ime
a ying pa e n along wi h majo global and domes ic e en s. The
sen imen index, namely "Tu kish Economic Sen imen Index (TESI)", is
de eloped using compu a ional ex analysis echniques and elying on
ex ensi e ex ual da a. In con as o adi ional su ey-based indica o s,
he TESI concen a es on sen imen e lec ed wi hin he news a icles
and epo s ocused on Tü kiye’s economic landscape. Ou subsequen
s ep in ol es in eg a ing he TESI in o models alongside majo pull-push
de e minan s o assess whe he ma ke sen imen holds disce nible in-
luence beyond con en ional ca alys s on capi al in lux
3
in o Tü kiye.
Gi en he mixed in eg a ion deg ees o he a iables, we u ilize he
bounds es ing me hodology de ised by Pesa an e al. (2001) using an
ap Au o eg essi e Dis ibu ed Lag (ARDL) amewo k, aiming o
disce n he exis ence o a coin eg a ing ela ion. Upon es ablishing a
signi ican long- un ela ionship be ween capi al in lows and ma ke
sen imen , we e alua e he na u e and scale o hei sho - e m in-
e ac ions ia an e o co ec ion model.
One o he p ima y con ibu ions o his s udy lies in he in oduc ion
o a no el ex -based sen imen indica o o he Tu kish economy,
which add esses he limi a ions o exis ing app oaches. The s udy in-
eg a es machine lea ning algo i hms wi h lexical echniques, o e ing a
mo e accu a e and ad anced analysis o he sen imen owa ds he
Tu kish economy. The TESI could be used as an al e na i e measu e o
su ey-based indica o s, o could con ibu e o he s udies in es iga ing
he beha io al/expec a ional aspec s o he Tu kish economy and
inancial ma ke s. Mo eo e , since he TESI p o ides p ac ical in o -
ma ion abou he cu en and p ospec i e s a e o he Tu kish economy,
i could be a use ul ool o esea che s and policymake s o analyze and
an icipa e po en ial shi s in c oss-bo de inancial lows.
Addi ionally, his esea ch p o ides empi ical insigh s in o analysis
o po olio lows o Tü kiye, inco po a ing adi ional de e minan s o
capi al lows ac oss bo de s and beha io al ac o s such as in es o
sen imen and isk appe i e. The indings e eal ha , in he long e m,
he TESI and VIX s and ou as pi o al de e minan s in luencing in e -
na ional po olio lows. The TESI exhibi s a posi i e co ela ion wi h
po olio low dynamics, whe eas, as an icipa ed, he deg ee o global
isk a e sion in e sely a ec s hese lows. This aligns wi h he con en-
ion ha a a o able sen imen can boos po olio in lows in o eme ging
ma ke s. Con e sely, heigh ened ola ili y expec a ions in global ma -
ke s can p omp ou lows om hese economies.
The s uc u e o his pape un olds as ollows: The subsequen sec ion
o e s an o e iew o he ele an li e a u e. This is ollowed by he
explana ion o adop ed me hodology. A e wa d, we p esen and del e
in o he da a selec ion p ocess and empi ical indings. The pape cul-
mina es wi h a summa ion o he key indings and an explo a ion o hei
implica ions o policy-making.
2. Li e a u e Re iew
Unde s anding he a ionale behind he de e minan s o in e na-
ional capi al lows has gene a ed ex ensi e academic li e a u e. In hei
seminal pape , Cal o e al. (1993) in oduce an analy ic amewo k o
explain he beha io o capi al lows and conclude ha global ac o s, in
pa icula low in e es a es in he US and global cyclical condi ions, a e
he main d i e s o he su ge in capi al lows o de eloping economies.
Th ough an analy ic amewo k based on coun y isk, Fe nandez-A ias
(1996) sc u inizes he ques ion o whe he a o able domes ic ac o s in
an eme ging economy pull capi al lows o una ac i e o eign condi-
ions in an ad anced economy push hese lows. The au ho inds ha a
decline in in e na ional in e es a es is he key ac o o capi al lows o
eme ging coun ies. I is also no ed ha he policy eac ions o ad anced
economies du ing inancial ins abili y pe iods shi ed he academic
a en ion o a new push ac o , namely he impac o expansiona y
mone a y policies by de eloped coun ies on capi al lows o eme ging
economies. F a zsche e al. (2012) unde line he ole o loose mone a y
condi ions in di ec ing capi al lows o eme ging coun ies and ind an
economically meaning ul e ec o asse pu chases o mone a y au ho -
i ies in ad anced coun ies on po olio lows o eme ging economies.
Despi e hei pa icula ocus on he policy ac ions by he Fede al
Rese e, Koepke (2019), and Dahlhaus and Vasish ha (2014) add a new
dimension o he adi ional push-pull amewo k by including ma ke
expec a ions o p ospec i e Fede al Rese e policy decisions and sug-
ges ha ma ke expec a ion is one o he impo an ac o s in de e -
mining a cou se o in e na ional capi al lows.
While hese empi ical s udies sugges ha global ac o s play a
dominan ole, some s udies documen ha domes ic ac o s a e mo e
p ominen in d i ing capi al lows. Fo ins ance, in he s udy o De Vi a
and Kyaw (2008), p oduc i i y g ow h in eme ging economies is sug-
ges ed o be he main engine behind he low o o eign di ec in es -
men o hose economies, while domes ic mone a y agg ega es a e he
1
h ps://www.wo ldbank.o g/en/coun y/ u key/o e iew.
2
h ps://www.wo ld-exchanges.o g/.
3
The s udy akes in o accoun ne po olio equi y and deb lows, because
hey a e subjec o sho - e m inancial luc ua ions and economic news (see, o
example, Mon iel and Reinha 1999; Baek 2006).
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
3
undamen al de e minan s o po olio in lows. Simila ly, He nandez
e al. (2001) analyze he de e minan s o capi al lows owa d some
de eloping coun ies in he 70s and 90s, ocusing on po en ial con agion
de i ed om he ade linkages and simila i ies in e ms o hei mac-
oeconomic ou look. The au ho s ind ha coun y-speci ic ac o s o he
ecipien coun y de e mine he beha io o capi al in lows, pa icula ly
in he 90s. Ca allo and F ankel (2008) unde line he impo ance o
ins i u ional quali y in hos economies and a gue ha be e ins i u-
ional quali y is associa ed wi h a highe deg ee o s abili y in capi al
in lows and mo e in low. The e a e also s udies ha aim a connec ing
bo h global and domes ic ac o s o he mo emen o capi al lows ac oss
bounda ies. Fo example, Chuhan e al. (1998) examine he beha io o
bond and equi y lows om he US o nine La in Ame ica and nine Asia
coun ies. Thei esul s indica e ha no only global ac o s a e essen ial
in explaining capi al lows, bu also domes ic ac o s mo i a e capi al
lows o hose economies. The s udy o Baek (2006), which examines he
ole o push and pull ac o s in de e mining po olio in es men s o-
wa d some Asian and La in Ame ican economies, sugges s ha he
push-pull amewo k a ies be ween he economies o he wo egions.
While ex e nal ac o s dominan ly push po olio in es men s in Asian
economies, hose in La in Ame ica a e de e mined by bo h ex e nal and
domes ic ac o s.
Conside ing he possible unpleasan e ec s o po olio lows on
domes ic cycles, he numbe o s udies digging in o he main de-
e minan s o capi al lows o Tü kiye has also gained momen um. Çulha
(2006) examines he de e minan s o capi al lows in o Tü kiye o he
p e- and pos -banking c isis pe iods. Acco ding o he au ho ’s esul s,
he pull ac o s become essen ial o e ex e nal ones in de e mining
capi al lows in o he Tu kish economy. Kucukkaya (2011) in es iga es
he push-pull amewo k o he Tu kish economy by employing he
VAR Me hodology and inds ha he domes ic in e es a e, US in e es
a e, s ock exchange p ice index, USDTRY exchange a e, and capi al
accoun balance ha e an in luence on he po olio in es men s o
Tu key. By in eg a ing he po en ial impac o expec a ions on FED’s
u u e policy ac ions in o hei eg ession model, Ak as e al. (2018)
in es iga e he eac ion o eme ging economies’ sha e in o al po olio
lows o FED’s signals on u he decisions. In hei s udy, Ak as and Eksi
(2020) analyze he d i e s o po olio lows o Tü kiye and o he
eme ging economies, ocusing on cyclical e sus s uc u al d i e s.
Thei esul s sugges ha he isk appe i e o in es o s is he main global
d i e , while g ow h pe o mance, so e eign isk, and eal in e es a e
a e iden i ied as he main domes ic ac o s a ec ing in e na ional lows.
Akdo˘
gan e al. (2021) in es iga e he e ec o capi al lows ac oss na-
ions wi h a special ocus o c oss-bo de banking sec o liabili ies o
global banks. The au ho s sugges ha he disagg ega ion o he
non-co e liabili ies o he banking sec o in o hei demand and supply
pilla s holds alue in dis inguishing he global liquidi y impac om
domes ic condi ions. This di e en ia ion yields aluable insigh s in o
he ap s uc u ing o coun e cyclical mac op uden ial policies. Çepni
e al. (2021) explo e he impac o unce ain ies ela ed o global
economy and ade-policies o he US on p edic ing he po olio lows o
Tü kiye. Thei esul s e eal a causal ela ionship be ween capi al lows
and unce ain y, especially du ing he global inancial c isis and he
elec ion o he T ump adminis a ion. Gemici e al. (2023) a gue ha he
isk appe i e o in es o s has a causal ela ionship wi h bo h local and
global ac o s and ob ain ha he main d i e s o isk appe i e a e
changes in CDS sp eads, inancial s ess and VIX. Thei esul s imply ha
all a iables exhibi a nega i e in luence on changes in isk appe i e
indices and ha in es o s end o adjus hei o e all in es men posi-
ions in esponse o ad e se shocks s emming om bo h local and global
ac o s.
As no ed abo e, esea ch on he de e minan s o capi al lows has
mainly paid a en ion o he quan i a i e ac o s. Ne e heless, ecen
s udies aim o add ess he issue wi h a pe spec i e ha ocuses on
quali a i e ac o s such as ma ke sen imen . The e is no doub ha
sen imen is no a new concep in he li e a u e, bu i occupies a
pa icula place in he ield o inance o in es iga e he ole o in es o
sen imen in explaining and p edic ing asse p ices (Bake and Wu gle ,
2006; Te lock, 2007, 2011; Te lock e al., 2008; Lough an and McDo-
nald, 2011; Soo, 2013). The e a e a ew s udies in he li e a u e
ega ding he examples ha combine his new pe spec i e wi h he
adi ional li e a u e on in e na ional po olio lows. Fo example,
Hwang e al. (2019) analyze he ela ionship be ween global in es o
sen imen and capi al lows ac oss he Ko ean economy and o he
Asia-Paci ic coun ies. The au ho s build a global in es o sen imen
indica o using an a e age sen imen sco e based on o eign edi o ials
abou Ko ea. Thei esul s sugges ha he sen imen indica o p o ides
impo an insigh s in o he de e minan s o in e na ional capi al lows o
he Ko ean ma ke s. The s udy o Hwang (2011), explo ing whe he
e ail in es o s’ non- undamen al easoning can a ec in e na ional
capi al mo emen s, e eals ha indi idual in es o s’ po olio holdings
in a o eign coun y a e highly ela ed o he popula i y o ha o eign
coun y in he Uni ed S a es. Bi u and Wyn e (2023) demons a e ha
economies wi h highe sen imen no only expe ience highe po olio
in lows o hei local ma ke s bu also ace lowe po olio ou lows om
ha coun y. The au ho s also elabo a e ha sen imen in an economy
has spillo e impac s on demand o asse s in o he coun ies. Despi e
hese e i ing academic e o s on in e na ional capi al lows, no s udies
examining he lows o po olio in es men s o Tü kiye u ilizing ma ke
psychology. This s udy con ibu es o he ela ed li e a u e by in e-
g a ing a new a iable ep esen ing in es o mood and ma ke expec-
a ions owa d he Tu kish economy in o he adi ional push-pull
amewo k.
3. Me hodology
This esea ch examines he in luence o bo h ex e nal and domes ic
undamen als on po olio lows, paying pa icula a en ion o ma ke
sen imen owa ds he Tu kish economy. To unde s and he ela ionship
be ween economic sen imen conce ning Tü kiye and in e na ional
po olio lows, we employ he Au o eg essi e Dis ibu ed Lag (ARDL)
me hod, a coin eg a ion app oach pionee ed by Pesa an e al. (2001).
The ARDL bounds es me hod o e s se e al ad an ages o e o he
coin eg a ion es s. No ably, i acili a es he examina ion o he
long- e m ela ionship be ween a iables, e en i hey a e in eg a ed in
a ying o de s. Speci ically, he echnique is e ec i e ega dless o
whe he he model’s a iables a e pu ely I(0), solely I(1), o a combi-
na ion o bo h. This e sa ili y ci cum en s po en ial challenges a ising
om uni oo es s. Addi ionally, he ARDL app oach add esses he
endogenei y conce ns o independen a iables and p o es sui able o
s udies wi h limi ed sample sizes. One o i s dis inc bene i s is i s ca-
paci y o concu en ly es ima e he sho - and long- e m in e play be-
ween a iables.
In o de o in es iga e i sen imen owa ds he Tu kish economy has
a bea ing on he beha io o in e na ional po olio lows in o Tü kiye,
we conside he Fede al Rese e Sys em’s asse alue as a global
liquidi y indica o . Addi ionally, we ac o in he p oduc ion and in-
e es a e dispa i ies be ween Tü kiye and he US o e lec eal and
mone a y di e gences be ween eme ging and ad anced economies
(Cal o e al., 1993; Ahmed and Zla e, 2014).
The ollowing eg ession is es ima ed using he ARDL app oach:
ΔNe lows =β0+∑p−1
i=1βiΔNe lows −i+∑q−1
i=0δiΔTESI −i
+∑q−1
i=0
τ
iΔVIX −i+∑q−1
i=0θiΔIn Ra e Di −i+∑q−1
i=0γiΔIP Di −i
+∑q−1
i=0
α
iΔFED BS −i+φ1Ne Flows −1+φ2TESI −1+φ5VIX −1
+φ3In Ra e Di −1+φ4IP Di −1+φ5FED BS −1+
μ
(1)
whe e "Ne lows" e e s o he ne po olio lows di ec ed owa ds
Tü kiye, "TESI" signi ies he sen imen indica o pe inen o he Tu kish
economy, "VIX" ep esen s he global isk a e sion le els, "In Ra e_Di "
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
4
ep esen s he disc epancy in in e es a es be ween Tü kiye and he US,
"IP_Di " indica es he g ow h a e di e en ial be ween Tü kiye and he
US and "FED_BS" s ands o he agg ega e asse alue held by all Fede al
Rese e Banks. β, δ,
τ
, θ, γ and
α
co espond o he sho - e m co-
e icien s, while he φ alues ep esen he long- e m coe icien s.
μ
e-
e s o he e o e m. Las ly, "p" and "q" deno e he lag leng hs associa ed
wi h dependen and independen a iables, espec i ely.
Coin eg a ion is assessed using he Bounds es ing app oach, whe e
he join signi icance o he lagged alues o he a iables is examined
ia an F- es . The null hypo hesis, sugges ing no coin eg a ion, is
deno ed as H
0
: φ
1
=φ
2
=φ
3
=φ
4
=φ
5
=0. We conside wo se s o
c i ical alues de i ed by Pesa an e al. (2001) and Na ayan (2005).
Pe aining o he in e p e a ion o Bounds es ing, i he F-s a is ic ex-
ceeds he uppe c i ical bound, i p o ides s a is ical e idence e u ing
he null hypo hesis o no coin eg a ion, ega dless o he in eg a ion
o de o he a iables in he model (be I(0) o I(1)). Con e sely, i he
F-s a is ic alls below he lowe c i ical bound, he null hypo hesis e-
mains uncon es ed. In ins ances whe e he compu ed F-s a is ic is posi-
ioned be ween he c i ical bounds, i necessi a es a uni oo es o
asce ain he in eg a ion o de o he a iables, whe he hey a e I(0) o I
(1).
I he e is e idence o a coin eg a ing ela ionship, an e o co ec-
ion model can be employed o cap u e he sho - e m dynamics o his
long- e m associa ion, which is de ined as:
ΔNe lows =
α
0+∑p
i=1
α
iΔNe lows −i+∑q
i=0γiΔTESI −i
+∑q
i=0βiΔVIX −i+∑q
i=0δiΔIn Ra e Di −i+∑q
i=0θiΔIP Di −i
+∑q
i=0
π
iΔFED BS −i+
∂
ECT −1+
μ
(2)
whe e ECT ep esen s he e o co ec ion e m, cap u ing he speed a
which adjus men s a e made o e u n o equilib ium in he endu ing
ela ionship among ne po olio lows, economic sen imen , and o he
con ol a iables.
4. Da a and empi ical analysis
This sec ion ou lines he da a u ilized o he s udy, including con-
s uc ion o a composi e sen imen indica o o he Tu kish economy,
"Tu kish Economic Sen imen Index (TESI). I de ails he me hodology
employed o examine he ela ionship be ween sen imen ega ding he
Tu kish economy and he ecen ends in p i a e po olio lows.
4.1. Cons uc ion o sen imen index
Recen ad ancemen s in ex analy ics p o ide a ga eway o mine
in o ma ion om uns uc u ed ex ual sou ces and o nume ically
ep esen hese ea u es. Sen imen analysis, a pilla o ex analy ics, is
cha ac e ized as a me hod o gauge he emo ional ones p esen in
quali a i e da a, using an a ay o echniques, including na u al lan-
guage p ocessing, compu a ional linguis ics, machine lea ning, and
s a is ical me hods.
While sen imen analysis has been discussed in he li e a u e o
some ime, only a hand ul o s udies ocus on gauging he o e all eco-
nomic mood o an economy. Shapi o e al. (2020) c a an economic
sen imen me ic, oo ed in economic and inancial jou nalism, o
explo e i s co ela ion wi h he economic ac i i y in he US. They
sc u inize he in e play be ween hei news-o ien ed sen imen index
and US economic ac i i y by employing compu a ional ex analysis,
p edominan ly lexical me hods. Ga cia (2013) cons uc s a ma ke
sen imen indica o based on a icles om he New Yo k Times. Using a
lexical me hod, he au ho allies posi i e and nega i e e ms o e e y
a icle column, e e encing he lexicon cu a ed by Lough an and
McDonald (2011). Simila ly, F aibe ge (2016) dissec s he in o ma-
ional con en o news-based sen imen me ics and hei link wi h
economic ac i i y luc ua ions ac oss 12 na ions. The au ho in eg a es
he posi i e and nega i e wo d dic iona ies om bo h Lough an and
McDonald (2011) and Young and So oka (2012) o he analysis.
To cons uc ha no el sen imen me ic o he Tu kish economy,
we sou ce 31,084 news a icles conce ning he Tu kish economy om
Thomson Reu e s and he Financial Times. Fu he mo e, we ob ain
12,900 epo s om majo in e na ional ins i u ional in es o s
enowned o hei deep acumen and amilia i y wi h he Tu kish eco-
nomic landscape. These epo s hail om he op 15 global in es men
banks ha ou inely pen insigh s on Tu key’s key economic and inan-
cial e olu ions. These epo s we e accessed om he espec i e in-
s i u ions’ online pla o ms wi h explici pe missions. To p ese e
con iden iali y, he iden i ies o bo h he ins i u ions and he au ho s
emain undisclosed in his con ex .
A e he da a ga he ing phase, he news a icles and in es o epo s
p oceed o a p e-p ocessing s age in which noisy elemen s and i ele an
in o ma ion, such as special cha ac e s, punc ua ion, e c., a e elimi-
na ed. Following his, he e ined ex ual da a is di ided in o okens,
o ming he undamen al basis o he subsequen sen imen analysis.
These okens can encompass wo ds, ph ases, o sen ences, wi h he
choice dependen on he desi ed le el o g anula i y o he sen imen
analysis. Wi hin his s udy, he e alua ion o he seman ic o ien a ion o
he news a icles and in es o epo s is conduc ed a bo h wo d and
sen ence le els. In he wo d-le el sen imen sco ing, each indi idual
wo d wi hin he news a icles and in es o epo s is assigned a sen i-
men sco e (posi i e, nega i e, o neu al). In he sen ence-le el sen i-
men sco ing, he o e all sen imen o emo ional one exp essed wi hin
a sen ence is assessed, aking in o accoun he con ex and in e ac ion
be ween wo ds.
Fu he mo e, a a ie y o sen imen sco ing models a e applied,
conside ing he ad an ages and limi a ions associa ed wi h di e se
sen imen analysis me hodologies. Speci ically, bo h lexical and ma-
chine lea ning echniques a e employed o quan i y he sen imen
exp essed wi hin each news a icle and epo . The lexical app oach
elies on a p ede ined dic iona y in which sen imen sco es a e assigned
o okens. This app oach u ilizes a lis o opinion wo ds (lexicons) and
hei associa ed alence shi e s, as well as a se o ules o de e mining
he sen imen o ien a ion. In he ini ial phase, wo ds a e compa ed wi h
opinion wo ds ound in he ele an lexicon. Each wo d ha ma ches an
opinion wo d is gi en a sen imen sco e, ca ego izing i as posi i e,
nega i e o neu al. The subsequen s ep add esses alence shi e s,
which a e wo ds ha al e o in ensi y he meaning o opinion wo ds.
"Nega o s" a e ypically ad e bs ha nega e he meaning o wo ds, while
"ampli ie s" a e gene ally ad e bs o adjec i es ha enhance he
meaning o wo ds. The inal s age in ol es sen imen agg ega ion o
de e mine he o e all sen imen o ien a ion wi hin he ex ual da a. I
he cumula i e sen imen sco e is posi i e, he ex is classi ied as
"posi i e," and ice e sa.
Wi hin he scope o he lexical app oach, he wo d-le el sen imen
sco ing is applied using he lexicons o Bing, Lough an & McDonald
(LM), A inn and N c lexicons. Respec ing calcula ion me hods o sen i-
men sco es, o he Bing and LM lexicons, he sen imen sco es a e
ob ained by he di e ence be ween posi i e and nega i e sen imen
coun s. Fo he A inn and N c lexicons, o e all sen imen sco e is
measu ed by he sum o sen imen sco es. Thus, eigh wo d-le el
sen imen se ies a e yielded. Fou o hem is de i ed om he news
a icles abou he Tu kish economy and he es is ob ained om he
in es o epo s.
Mo eo e , he sen imen analysis is en iched by adding a sen ence-
le el analysis based on he s udy o Silge and Robinson (2017) in
which he au ho s p opose ha sen ence-sized o pa ag aph-sized
sen imen analysis gene ally displays be e pe o mance. To ex ac
o e all sen imen om he ex da a, he sen ences a e in eg a ed wi h
“Jocke s”, “Sen icne ”, “Sen iwo d”, “Syuzhe ”, “Bing”, “A inn”, “LM”
and “N c” dic iona ies. Hence, eigh een sen ence-le el sen imen indices
a e ob ained. Nine o hem a e de i ed om he news a icles abou he
Tu kish economy and he es is ob ained om he in es o epo s.
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
5
As a esul o lexical p ocess, a o al o wen y six sen imen indices
o he Tu kish economy is de i ed. Ha ing such a g ea a ie y o
indices enables o imp o e he obus ness o he analysis and o e s a
subs an ial compa ison o he accu acy o he di e en me hods and
lexicons.
Recognizing ha he sen imen analysis can be en iched by
combining al e na i e app oaches o le e age he s eng hs o each, he
news a icles and in es o epo s a e also analyzed by he machine
lea ning (ML) algo i hm. Conside ing i s lexibili y and ined- uned
ouch in he sen imen analysis, he da ase unde goes analysis using
Google’s Cloud Na u al Language API algo i hm. While i s use - iendly
in e ace is o en unde sco ed, he p ima y a ionale o selec ing Goo-
gle’s API is i s p o ision o p e- ained powe ul NLP models. This comes
a a ela i ely low cos o such an in ica e p ocess, owing o he
declining a e age cos s o cloud se ices. The e icacy o he ML models
wi hin he API is ypically deemed commendable, especially o English-
language da ase s, gi en ha he models bene i om aining on as
documen co po a.
4
In o al, wen y-eigh aw sen imen sco es a e calcula ed— ou een
om news a icles and he emaining om in es o epo s—all o
which a e on di e en scales. To ha monize hese aw sen imen sco es,
we adop he me hodology pu o h by Bake e al. (2016). This in ol es
scaling he aw sco es by he o al coun o ex documen s in a gi en
mon h, ollowed by s anda diza ion. Gi en ha e e y sen imen sco ing
model possesses dis inc s eng hs and limi a ions, we employ p incipal
componen analysis (PCA) o de i e a composi e index om he
wen y-eigh sen imen sco es. The i s componen , accoun ing o 52%
o he a iance, is designa ed as he mon hly composi e sen imen index,
e med he "Tu kish Economic Sen imen Index (TESI)" ( e e o Fig. 1).
As seen in Fig. 1, he TESI e lec s nume ous spikes and d ops in
o e all sen imen owa ds he Tu kish economy ac oss he imeline. To
assess he TESI’s abili y o de ec he main de elopmen s a ec ing he
o e all economic clima e, an e en analysis is conduc ed below o
explo e he pe iods in which he TESI exhibi s luc ua ions and he
signi ican e en s ha align wi h hese pe iods.
⁃ I would no be misleading o s a e ha he pe iod spanning om
2004 o 2007 was p edominan ly shaped by i s economic e o m
agenda unde he IMF S and-by p og am and Tü kiye’s EU accession
p ocess. O he pi o al momen s om his e a, which could po en-
ially sway he b oade sen imen owa ds he Tu kish economy,
include he discussions conce ning he euni ica ion o G eek and
Tu kish Cyp io s in 2004, he dep ecia ion o he Tu kish Li a (TL)
in luenced by global ma ke ola ili y a ising om s ic e mone a y
policies in ad anced economies in 2006, and he inc eased s ess in
he global inancial ma ke in la e 2007 due o issues in he US
subp ime ma ke s.
⁃ Be ween 2008 and 2012, he mos signi ican e en was he subp ime
mo gage c isis ha began in he US in 2007. This localized c isis
escala ed in o a global inancial and economic down u n ollowing
he collapse o Lehman B o he s in Sep embe 2008. Howe e , by he
la e hal o 2009, he e was a ma ked and ab up up ick in global
sen imen and in es o isk appe i e, leading o subs an ial capi al
in lows in o eme ging ma ke s, Tü kiye included, in sea ch o highe
yields. Al hough 2011 saw he de imen al e ec s o he Eu opean
Deb C isis, he onse o 2012 ma ked a esu gence in global isk
appe i e coupled wi h expansiona y s ances by cen al banks.
⁃ Du ing he ime ame om 2013 o 2018, i is pe inen o emphasize
a signi ican junc u e ma ked by he announcemen made by FED
Go e no Be nanke ega ding an exi s a egy om quan i a i e
easing policies. This announcemen led o a swi con ac ion in
global inancial condi ions, esul ing in capi al wi hd awals om
eme ging ma ke s. In his pe iod, o en e e ed o as he " ape
an um," Tü kiye is pe cei ed as one o he mos ulne able
eme ging ma ke s due o i s signi ican ex e nal inancing e-
qui emen s and a high p opo ion o p i a e ex e nal deb . The
sen imen sco es exhibi ed a no able plunge, which was o se when
he Cen al Bank o he Republic o Tu key (CBRT) ins i u ed a sig-
ni ican in e es a e su ge and s eamlined i s mone a y policy
amewo k du ing an eme gency mee ing on Janua y 28, 2014,
whe e he epo a e leaped om 4.5 pe cen o 10 pe cen . Ano he
c i ical junc u e ha igge ed one o he mos p o ound d ops in
sen imen sco es was in 2018, when he ensions escala ed be ween
Tü kiye and he US go e nmen . This ension esul ed in he TL’s
signi ican dep ecia ion agains he dolla due o a massi e sell-o in
Augus 2018. A se ies o TL/FX liquidi y measu es we e enac ed o
coun e he TL’s de alua ion in Augus , ollowed by agg essi e a e
hike decisions on he mone a y on in Sep embe .
⁃ Subsequen ly, sen imen owa d he Tu kish economy eached a
nadi , p ima ily in luenced by he onse o he Co id-19 pandemic.
The close o 2019 saw he global communi y g appling wi h he
eme gence o he i s Co id-19 case in Wuhan, China. By Ma ch 11,
2020, he Wo ld Heal h O ganiza ion o icially ecognized i as a
pandemic. Tü kiye, on ha e y day, announced he i s co ona i us
case. By Ma ch 17, wi h he pandemic p o oundly a ec ing bo h he
global landscape and he Tu kish economy. Howe e , he CBRT,
alongside o he ele an en i ies, quickly eac ed and ini ia ed
measu es o elimina e he pandemic’s nega i e ami ica ions and
bols e inancial s abili y. Hence, in his pe iod, he sen imen o-
wa ds Tü kiye displayed a s ong eco e y hanks o hese policy
s eps aken. Following a signi ican decline in he second qua e o
2020, due o diminished domes ic and in e na ional demand, he
Tu kish economy wi nessed a swi esu gence pos -May, in line wi h
he g adual ollback o pandemic- ela ed es ic ions. This s ong
eco e y led o a 1.8 pe cen g ow h in Tu key o 2020, posi ioning
i as he second as es -g owing economy a e China. The sen imen
owa ds Tu key ebounded om i s his o ic lows, hi ing a zeni h in
No embe 2020.
⁃ Howe e , he Tu kish economy wi nessed a subsequen dep ecia ion
su ge in TL in he ollowing pe iod, coinciding wi h he downwa d
mo emen in he sen imen . Agains his backd op, he CBRT
emba ked on a mone a y policy igh ening jou ney, coupled wi h
simpli ying i s mone a y policy amewo k. The ma ke la gely
pe cei ed hese con en ional policy measu es, as o e whelmingly
posi i e, leading sen imen indices o c es in he inal qua e o
2020 and he ini ial qua e o 2021. Ye , as o he second qua e o
2021, he sen imen index began a descen , e lec ing ma ke pa -
icipan s ega ding a shi back o uno hodox economic policy
choices. Addi ionally, he Russia-Uk aine wa became ano he sig-
ni ican e en ha led o a downwa d mo emen in he TESI in he
i s qua e o 2022.
4.2. Da a se ies
As adi ional pull-push ac o s, he VIX index as a p oxy indica o o
global isk appe i e, he in e es a es di e ences be ween Tü kiye and
he US, he di e ence in indus ial g ow h a es o hese wo economies
and he agg ega e asse alue held by all Fede al Rese e Banks as a
global liquidi y indica o a e included. In addi ion o hese ac o s, he
TESI is in eg a ed in o he analysis as a me ic o ma ke psychology
ha can signi ican ly impac in es men decisions.
As dependen a iable, he ne po olio low in o Tü kiye is es ab-
lished, because we posi ha he o e all economic sen imen owa ds
Tu key migh in luence po olio lows in wo ways, as a icula ed by
Bi u and Wyn e (2023). Fi s ly, an ele a ed economic sen imen abou
Tü kiye could lead o inc eased po olio in lows. Secondly, du ing imes
o highe sen imen , he e migh be educed po olio ou lows om
Tu key. Finally, we inco po a e wo dummy a iables ep esen ing he
Global Financial C isis and he Co id pandemic as non-dynamic ac o s.
4
NLP wi h Google Cloud Na u al Language API by Maximilian Hop .
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
6
The aw da a o indus ial p oduc ion, in e es a e, VIX, and he
asse alues o he Fede al Rese e Sys em a e acqui ed om he
Bloombe g Te minal. Con e sely, he da a pe aining o ne po olio
lows o Tü kiye a e sou ced om he Ins i u e o In e na ional Finance’s
da a po al. The sen imen me ic conce ning he Tu kish economy is
de i ed om he au ho s’ compu a ions, wi h he me hodology de ailed
in he subsequen sec ion. All da a se s a e on a mon hly basis, spanning
om Janua y 2004 o June 2022 (see Table 1).
4.3. Model es ima ion and empi ical esul s
In o de o examine i he TESI and po olio lows o m a meaning ul
and s able ela ionship, one needs o check he o de o in eg a ion o
he a iables in ques ion. To his end, h ee commonly used uni - oo
es s a e employed: Augmen ed Dickey-Fulle (ADF), Phillips-Pe on
(PP), and Kwia kowski–Phillips–Schmid –Shin (KPSS). The esul s a e
p esen ed in Table 2. As seen, he in e es a e di e en ials and he
agg ega e alue o he FED’s balance shee achie e s a iona i y a e he
i s di e ence. Meanwhile, he emaining a iables exhibi s a iona i y
a he le el. In o he wo ds, he a iables in he model a e ei he I(0) o I
(1).
Thus, we apply he bounds es ing echnique o coin eg a ion using
he ARDL model. To do so, one needs o de e mine he op imal lag o -
de s, p and q, o he dependen and independen a iables in he ARDL
eg ession, as shown in equa ion (1), he Akaike In o ma ion C i e ia is
u ilized. The esul s o he op imal lag o de s a e p esen ed in Table 3.
Fu he mo e, diagnos ic es s such as no au oco ela ion, homosce-
das ici y, no mali y, and s abili y es s a e conduc ed o ensu e ha he
eg essions wi h speci ied lag o de s adhe e o he classical eg ession
assump ions. The esul s a e epo ed in Table 4. As i is seen, he e is no
e idence o se ial co ela ion o he e oscedas ici y in he model e-
siduals. Fu he mo e, he esiduals appea o ollow a no mal dis ibu-
ion, and he models seem o be co ec ly speci ied.
As a esul , we can un bounds es s o coin eg a ing ela ionships
using he op imal lag leng hs epo ed in Table 3. The F-s a is ics o
bounds es is displayed in Table 5. The esul s indica e he coin eg a ing
ela ionship be ween he a iables.
4.3.1. Analysis o he long- un ela ionship
Conside ing he coin eg a ion be ween he ne po olio lows o
Tu key and de i ed sen imen indica o alongside o he a iables, he
long- un coe icien s a e es ima ed. F om he ou comes o he model
es ima ions, i becomes appa en ha , o e he long un, he sen imen
owa ds he Tu kish economy and he ex en o global isk a e sion a e
pi o al de e minan s in luencing ne po olio in lows o Tu key.
Fu he mo e, i is also no ed ha he signs o he es ima ed coe icien s
align seamlessly wi h heo e ical expec a ions.
Del ing deepe in o he esul s, he model e eals ha , o e he long
un, TESI, VIX, IP_Di , and FED_BS eme ge as s a is ically signi ican
d i e s o he ne po olio low o Tü kiye, as displayed in Table 6.
Speci ically conce ning he TESI, a di ec and posi i e co ela ion exis s
be ween o e all sen imen owa ds he Tu kish economy and ne in lows
om non- esiden s o Tü kiye, as expec ed. Con e sely, he deg ee o
global isk a e sion, ep esen ed by he VIX, exe s an expec ed in e se
in luence on ne po olio lows. Such indings align wi h he idea ha
a o able sen imen can s imula e in e na ional po olio in lux in o
eme ging ma ke s, while heigh ened global ma ke ola ili y expec a-
ions can induce ou lows om hese economies. Ano he a iable
posi i ely co ela ed wi h he po olio lows is he o al asse alue o all
Fede al Rese e Banks. (Fo obus ness check o he long- un model
es ima ions, see Appendix B).
4.3.2. Analysis o he sho - un ela ionship
Following he analysis o he long- un ela ionship be ween he
economic sen imen owa ds Tü kiye and ne po olio lows, we p o-
ceeded wi h he es ima ion o e o co ec ion model o examine he
sho - e m dynamics and he speed o adjus men o he long- un equi-
lib ium. The es ima ions om he e o co ec ion model unde sco e
ha he e o co ec ion e m (ECT) is nega i e and s a is ically
signi ican .
Fig. 1. E olu ion o he TESI
Table 1
De ini ion o a iables.
Va iable Sho De ini ion
Ne Po olio Flows Mon hly ne o eign lows o Tü kiye (in billion US
dolla s)
TESI Sen imen index ep esen ing he o e all pe cep ion o
he Tu kish economy
VIX Global isk a e sion indica o
Indus ial P oduc ion
Di e en ial
Na u al log o Tu kish Indus ial P oduc ion Index -
Na u al log o US Indus ial P oduc ion Index
In e es Ra e Di e en ial Tu kish ON In e es Ra e - Wu-Xia shadow ede al unds
a e
FED_Balance Shee Na u al log o o al alue o he asse s o all Fede al
Rese e Banks
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
7
The coe icien sugges s ha any depa u e om he equilib ium
le el o ne po olio lows in he cu en pe iod would be ec i ied in
oughly 1.2 mon hs. Addi ionally, he esul s indica e ha he TESI has a
s a is ically signi ican and posi i e ela ionship wi h he ne po olio
lows o Tü kiye. Ano he e idence is such ha al hough i s impac ades
ou o e ime, he sen imen exe s a posi i e sho - e m in luence on ne
po olio lows o he Tu kish economy, bo h in he cu en and p e-
ceding pe iods. I is also wo h no ing ha he deg ee o global isk
a e sion plays a s a is ically signi ican ole in explaining he sho - e m
a ia ions in po olio lows o he Tu kish economy ( e e o Table 7).
5. Conclusion
This esea ch examines he in luence o bo h ex e nal and domes ic
undamen als on po olio lows, paying pa icula a en ion o he
ma ke sen imen owa ds he Tu kish economy. To analyze he ela-
ionship be ween economic sen imen conce ning Tü kiye and in e na-
ional po olio lows, we in oduce a no el ex -based sen imen
indica o o he Tu kish economy, in eg a ing lexical echniques and
machine lea ning algo i hms. To assess whe he ma ke sen imen holds
disce nible in luence beyond con en ional ca alys s on capi al in lux
in o Tu key, he ob ained sen imen indica o , TESI is analyzed along-
side majo pull-push de e minan s.
Ou esul s sugges he coin eg a ion be ween he ne po olio lows
o Tü kiye and he de i ed sen imen indica o along wi h o he con-
olling a iables. The ou comes o he analysis e eal ha , in he long
e m, he sen imen owa ds he Tu kish economy and deg ee o isk
appe i e o in es o s s and ou as pi o al de e minan s o in e na ional
po olio lows. In ac , he sen imen indica o , TESI exhibi s a posi i e
co ela ion wi h po olio low dynamics; whe eas, as in ui ion sugges s,
he deg ee o global isk a e sion in e sely a ec s hese lows. This
aligns wi h he con en ional iew ha a o able global ma ke sen i-
men can boos po olio in lows in o eme ging ma ke s, whe eas de e-
io a ed sen imen and nega i e expec a ions can p omp ou lows.
The signi icance o sen imen in in luencing po olio lows un-
de sco es i s c i ical ele ance o bo h policymake s and ma ke pa -
icipan s, gi en i s p o ound implica ions o economic s abili y, ma ke
dynamics, and o e all inancial well-being. The po en ial impac o
ma ke psychology on guiding po olio lows ex ends beyond sho -
e m ma ke dynamics, exe ing a las ing in luence on ends in inan-
cial and capi al ma ke s, and consequen ly shaping long- e m in es -
men s a egies. Hence, by moni o ing and unde s anding sen imen
Table 2
Uni oo es esul s.
ADF Tes PP Tes KPSS Tes O de o In eg.
Le el 1s Di Le el 1s Di Le el 1s Di
Po olio Flows Ne Flows 0.01 0.01 0.05 0.1 I(0)
*** *** * ***
Sen imen TESI 0.01 0.01 0.05 I(0)
*** *** *
Global Risk VIX 0.18 0.01 0.01 0.1 I(0)
*** *** ***
In e es Ra e Di e en ial In Ra e_Di 0.01 0.13 0.01 0.04 0.1 I(1)
*** *** * ***
Indus ial P oduc ion Di e en ial IP_Di 0.01 0.01 0.01 0.1 I(0)
*** *** ***
Global Liquidi y FED_BS 0.55 0.01 0.96 0.01 0.01 0.1 I(1)
*** *** ***
Table 3
Op imal lag o de s.
Va iable Ne Flows TESI VIX In Ra e Di IP_Di FED_BS
Lag O de 2 5 1 2 0 2
Table 4
Resul s o diagnos ics es s.
TEST P-VALUE NULL HYPOTHESIS
AUTOCORRELATION - BREUSCH-
GODFREY
0.87 (up
o 1)
0.27 (up
o 2)
The e is no se ial
co ela ion o any o de up
o p
HETEROSKEDASTICITY -
STUDENTIZED BREUSCH-PAGAN
0.50 The e o a iances a e all
equal
NORMALITY - SHAPIRO-WILK 0.07 The popula ion is no mally
dis ibu ed
STABILITY - RAMSEY’S RESET 0.22 (d 1
=1)
0.36 (d 2
=2)
The model is co ec ly
speci ied.
No e: The igu es a e he p- alues o he co esponding es s a is ics.
Table 5
Bounds es esul s o coin eg a ion.
F-
S a is ic
p-
alue
C i ical Values Lowe
Bound
C i ical Values Uppe
Bound
12.89 0.000 4.26 6.04
Table 6
Long- un coe icien s.
Va iables Coe icien p- alue Signi icance
In e cep −9.00 0.072 *
TESI 4.47 0.008 ***
VIX −0.08 0.002 ***
IP_Di −3.51 0.050 **
In Ra e_Di −0.03 0.428
FED_BS 1.051 0.081 *
Table 7
Sho - un coe icien s.
Va iables Coe icien p- alue Signi icance
ΔNe Flows
-1
0.05 0.472
ΔTESI
4.60 0.000 ***
ΔTESI
-1
2.17 0.058 *
ΔTESI
-2
−0.29 0.800
ΔTESI
-3
1.68 0.128
ΔTESI
-4
1.12 0.257
ΔVIX
−0.16 0.000 ***
ΔIP_Di
−3.69 0.031 **
ΔIP_Di
-1
2.90 0.083 *
ΔFED_BS
7.25 0.060 *
ΔFED_BS
-1
−10.49 0.028 **
ECT −0.85 0.000 ***
D. Günes¸ e al.
Cen al Bank Re iew 24 (2024) 100147
8
indica o s, policymake s can c a imely and e ec i e esponses aimed
a p omo ing economic s abili y, cul i a ing in es men , and
sa egua ding inancial ma ke s agains unwa an ed dis up ions.
Appendix
Appendix A: Lexicons Used in he Analysis.
Lexicon # o Wo ds Sco e/Value Calcula ion Me hod
A inn 2477 be ween −5 and +5 Sum o indi idual sco es
Bing 6789 Posi i e and Nega i e (# o posi i e wo ds - # o nega i e wo ds)
N c 5555 Posi i e and Nega i e (# o posi i e wo ds - # o nega i e wo ds)
LM 3917 Posi i e and Nega i e (# o posi i e wo ds - # o nega i e wo ds)
Sen icne 23,626 be ween −1 and +1 Sum o indi idual sco es
Jocke s 11,710 be ween −1 and +1 Sum o indi idual sco es
Sen iwo d 20,093 be ween −1 and +1 Sum o indi idual sco es
Syuzhe 10,748 be ween −1 and +1 Sum o indi idual sco es
Appendix B. Robus ness check o he long- un model es ima ed using he DOLS
Conside ing po en ial biases in he es ima ion o a coin eg a ed ela ionship due o endogenei y and se ial co ela ion, he long- e m ela ionship
be ween sen imen owa d he Tu kish economy and in e na ional po olio lows is also es ima ed using he Dynamic O dina y Leas Squa es (DOLS)
echnique. The esul s om he DOLS-based models sugges ha he coe icien s de i ed om he ARDL es ima ion align closely wi h hose om he
DOLS es ima ion. Ano he consis en inding ac oss bo h me hodologies is he s a is ical signi icance o sen imen indica o s and he VIX ac oss all
models.
Table B1
Resul s o DOLS Es ima ion o Model 1
Va iables Coe icien p- alue Signi icance
In e cep −8.92 0.001 ***
TESI 6.20 0.000 ***
VIX −0.07 0.000 ***
IP_Di −2.91 0.001 ***
In Ra e_Di −0.05 0.004 **
FED_BS 0.92 0.002 **
Table B1 exhibi s he esul s om he DOLS es ima ion, inco po a ing he composi e TESI in o he model. The empi ical e idence implies ha TESI
and ne po olio lows ha e a s ong and posi i e ela ionship. In con as , a su ge in global isk a e sion migh esul in a dec emen in ne po olio
lows. Fu he mo e, he ou pu and in e es a e dispa i ies be ween Tu key and he US in e sely a ec ne po olio in lows, while an inc ease in he
FED’s balance shee alue could po en ially boos ne po olio in lows.
Re e ences
Ahmed, S., Zla e, A., 2014. Capi al lows o eme ging ma ke economies: a b a e new
wo ld? J. In . Money Finance 48, 221–248.
Akdo˘
gan, K., Eks¸i, N.K., Eks¸i, O., 2021. De e minan s o non-co e liabili ies o banks in
eme ging ma ke s in he pos -c isis e a. Bogazici Jou nal: Re iew o Social,
Economic & Adminis a i e S udies 35 (1).
Ak as, Z., Eksi, N.K., 2020. Wha d i es po olio lows o Tu key? The dynamics and a
his o ical accoun ing o he lows (No. 2010. Resea ch and Mone a y Policy
Depa men , Cen al Bank o he Republic o Tu key.
Ak as, Z., E duman, Y., Eksi, N.K., 2018. The E ec o Fed’s Fu u e Policy Expec a ions
on Coun y Sha es in Eme ging Ma ke Po olio Flows.
Albuque que, R., Loayza, N., Se ´
en, L., 2005. Wo ld ma ke in eg a ion h ough he lens
o o eign di ec in es o s. J. In . Econ. 66 (2), 267–295.
Baek, I.M., 2006. Po olio in es men lows o Asia and La in Ame ica: pull, push o
ma ke sen imen ? J. Asian Econ. 17 (2), 363–373.
Bake , S.R., Bloom, N., Da is, S.J., 2016. Measu ing economic policy unce ain y. Q. J.
Econ. 131 (4), 1593–1636.
Bake , M., Wu gle , J., 2006. In es o sen imen and he c oss-sec ion o s ock e u ns.
J. Finan. 61 (4), 1645–1680.
Binici, M., Hu chison, M., Schindle , M., 2010. Con olling capi al? Legal es ic ions and
he asse composi ion o in e na ional inancial lows. J. In . Money Finance 29 (4),
666–684.
Bi u, J., Wyn e , M., 2023. The Role o Domes ic and Fo eign Sen imen o C oss-
Bo de Po olio Flows. Fishe College o Business Wo king Pape (2023-03), 016.
Ca allo, E.A., F ankel, J.A., 2008. Does openness o ade make coun ies mo e
ulne able o sudden s ops, o less? Using g a i y o es ablish causali y. J. In . Money
Finance 27 (8), 1430–1452.
Cal o, G.A., Leide man, L., Reinha , C.M., 1993. Capi al in lows and eal exchange a e
app ecia ion in La in Ame ica: he ole o ex e nal ac o s. S a Pape s 40 (1),
108–151.
Chuhan, P., Claessens, S., Mamingi, N., 1998. Equi y and bond lows o La in Ame ica
and Asia: he ole o global and coun y ac o s. J. De . Econ. 55 (2), 439–463.
Çepni, O., Çolak, M.S., Hacıhasano˘
glu, Y.S., Yılmaz, M.H., 2021. Capi al lows unde
global unce ain ies: e idence om Tu key. Bo sa Is anbul Re iew 21 (2), 175–185.
Çulha, A., 2006. A s uc u al VAR analysis o he de e minan s o capi al lows in o
Tu key. Cen al Bank Re iew 2 (2), 11–35.
Dahlhaus, T., Vasish ha, G., 2014. The Impac o US Mone a y Policy No maliza ion on
Capi al Flows o Eme ging-Ma ke Economies (No. 2014-53). Bank o Canada
wo king pape .
De San is, R.A., Lüh mann, M., 2009. On he de e minan s o ne in e na ional po olio
lows: a global pe spec i e. J. In . Money Finance 28 (5), 880–901.
De Vi a, G., Kyaw, K.S., 2008. De e minan s o capi al lows o de eloping coun ies: a
s uc u al VAR analysis. Jou nal o Economic S udies 35 (4), 304–322.
Fe nandez-A ias, E., 1996. The new wa e o p i a e capi al in lows: push o pull? J. De .
Econ. 48 (2), 389–418.
F aibe ge , S., 2016, No embe . News sen imen and c oss-coun y luc ua ions. In:
P oceedings o he Fi s Wo kshop on NLP and Compu a ional Social Science,
pp. 125–131.
F a zsche , M., 2012. Capi al lows, push e sus pull ac o s and he global inancial
c isis. J. In . Econ. 88 (2), 341–356.
Ga cia, D., 2013. Sen imen du ing ecessions. J. Finance 68 (3), 1267–1300.
D. Günes¸ e al.
