Professional Documents
Culture Documents
ETHAN ILZETZKI
CARMEN M. REINHART Downloaded
KENNETH S. ROGOFF
Thi s arti cle provi des a comprehens i ve his tory of anchor or ref erence curren-
cies, exchan ge rate arrangement s, and a new measure of fo reign exchan ge res tri c- fr
om
ti ons for 194 coun tri es and t errit ori es over 1946-2016. We fi nd t ha t t he often cited https
pos t-Bre tton Woods t ransition fr om fix ed t o fl exible arran gement s i s overs t a t ed;
regi mes with li mited fl exi bi lity rem ai n i n t he maj ority. Even if central bankers' :
//academ
communi cati ons j argon has evolved consi derably i n recent decades, it i s apparen t
t hat many s till place a large i mpli cit weigh t on t he exchange rat e. The U. S. dollar
scores as th e world's domi nan t anchor currency by a very large margi n. By some i
c
.
oup
met r i cs, its use i s fa r wi der t oday t han 70 years a go. In cont ras t, t he global role of
.
8m/
t he euro app ears t o have s t alled. We argue t hat in ad dition t o t he usual safe as-
set s s t ory, t he record accumulation ofr eserves si nce 2002 may also have t o do with qj
e/a
many coun tries'desi re t o s t abi li ze exchan ge rate s i n an envi ronment of markedly
rti
reduced exchan ge rat e rest r i cti ons or, more broadly, capit al con t rols: an i m portan t cle,abs
amendmen t t o t he conventi onal portrayal of t he macroeconomi c t r ilemma. J EL
Codes: E5, F3, F4, N2 t
racU134/2/599/527
exchange rat e res t r i cti ons and capital cont rols, t he shift t oward
by
i n fl a tion t argetin g, and great er exchange rat e fl exi bility i n a num- gues
her of key emergin g marke t s. In t eresti n gly, we fin d t hat despite
t he wi despread predi ction t hat t he world i s evolvi n g t oward a t
on
16
*We t hank t he e ditors, Pol An t ras and Robe rt Barro, and fo ur anonym ous Oc
refe rees fo r t hei r useful comment s. We t hank Xavi er Gabai x, Matte o Maggiori, t
ober
KevinO 'Rourke, Vi ncen t Rei nhart, and seminar parti cipan t s a t t he Bank ofl n te r-
nati onal Settlemen t s, t he In t ernati onal Monetary Fund, t he AEA mee ti n gs, t he 2019
NBER In t ernati onal F i nance and Monetary Economics March 2017 meetin g, Har-
vard Uni versity, t he Bank of England, and t he CEPR-SNB-Bol Confe rence on For-
eign Exchan ge Market In t erven ti on fo r helpfu l commen t s and sugges ti ons. Lukas
Alt hoff, Kenyi Cansi no, Duygu Guven, Gonzalo Huertas, Patrici o Merlani, Paul
Schmelzing, and J ordan War provided out st anding research assi s tan ce. Coun try
chronologies and dat a can be fo und a t h ttp ://www. i lzet zki .com/irr-dat a.
el The Au thor(s) 2019. Publi shed by Oxford Univers ity Press on beh alf of the Presi den t and
Fellows of Harvard College. All righ ts reserved. For Permi ssi ons, please email:
j ournals.permi ssi ons@o up.com
The Quarterly J ournal of Economi cs (2019), 599-646. doi :10.1093/qje/qjy033.
Advance Access publi cation on J anu ary 5, 2019.
599
600 THE QUARTERLY J OURNAL OF ECONOMICS
cle,abs
i n fluence of U. S. mone tary poli cy on asse t prices worldwi de (Rey
2013; Bruno and Shi n 2015; Gopi nat h 2015; Maggiori, Ni eman, t
racU134/2/599/527
and Schreger 2018).1 Indeed, t he fact t hat exchange rat e s t abi li za-
tion a pp aren t ly remai ns an i m port ant consi deration fo r so many
cent ral banks can be consi dered a port mant eau measure of t he
vari ous i mpact s of dollar volatility, i ncludi n g li abi lity dollari za-
tion and the sensitivity of a country's marke t s t o shifts i n global
r i sk. 4128
Our result s are based on a comprehensi ve his t ory of anchor
or reference currenci es, exchange rat e arrangement s, and a new by
measure of fo re ign exchange res t r i cti ons for 194 count r i es and t er- gues
cle,abs
tion. The analys i s goes beyond exchange ra te-based i ndi cat ors,
for example, i ncorporatin g reserve holdi n gs and goods prici n g fo r t
racU134/2/599/527
borderli ne cases. Section III summari zes result s for anchor classi-
fi cation and for measures of de fa ct o exchange rat e classification.
No t ably, alt hough t here i s some t endency t oward more i n t ermedi-
a t e re gimes, t he world remai ns heavi ly skewed t oward less fl exi-
ble exchange rat e re gimes i nst ead of managed fl oatin g and freely
fl oati n g. This secti on looks a t t he t heory of anchor choi ce and ex- 4128
plores why one does no t see more evi dence of an expansi on of t he
euro's i n fl uence or t he emergence of a nascen t renmi nbi bloc. We by
also lay out a new dat a set on exchange rat e cont rols, a key piece gues
.
oup
exchange rat e arrangement by me tri cs t hat primarily (but no t ex-
gr
elusi vely) measure t he de ee of fl exibi lity. The approach expands
.
8m/qje/a
cle,abs
4. The Rei nhart and Ro goff (2004) classifi ca ti on has been ext remely wi dely
used i n empirical macroeconomics across a wi de v ariety of t opics, for exam-
ple, Aghion et al. (2009) and Raj an and Subramani an (2005) on growth; Ai zen-
man and Lee (2007) and J eanne and Ranci ere (2006) on reserve accumulation;
Chin n and Wei (2013) and Ghosh, Qures hi, and Tsangari des (2013) on current
account adj ust ment; Hau and Rey (2006) on capital fl ows and equity prices;
EXCHANGE ARRANGEMENTS 603
cle,abs
has gone t hrough a relatively subdued peri od. F igure I shows t he
absolu t e value of t he mont hly change i n t he dollar-Deut schmark t
racU134/2/599/527
cross-rat e from t he end of Bretton Woods t o t oday (the German DM
i s replaced by t he euro a fter 1999). Despite volatility's count er-
cycli cal nat ure, a clear secular decline i n exchange rat e volatility
i s vi si ble. (The dollar-yen cross-rat e shows a s i mi lar t rend.) The
combi nation ofr elatively low exchange ra t e volatility among large
advanced economi es t hat do no t i n t ervene versus t he rela tively 4128
h igh volatility of some emergin g marke t s t ha t heavily i n t ervene
forces us t o broaden our algorithm t o draw on additional i n for- by
mation, parti cularly on t he nat ure and i n t ent of foreign exchange gues
market i n t ervention. t
on
Mendoza and Terrones (2008) and J orda, Schularick, and Taylor (2015) on credit
fl ows;
Ball, Lopez, and Reyes (2013) on t he effect of remittances on t he macroe-
conomy; and Habi b, Biitzer, and St racca (2016) on t he effect of oil shocks.
604 THE QUARTERLY J OURNAL OF ECONOMICS
3.2
2.8
Downloaded
2.6
2.4
2.2
fr
om
1. 8
https
1. 6
1.4 :
//academ
1. 2
1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 200120032005 2007 2009 2011 2013 2015
i
c
FIGURE I .
oup
cle,abs
The cat egori es are list ed i n Online Appendi x Table A. l. For ease
of exposition, t he descripti on here focuses primarily on the coarse
cat e gori es.
For t he 194 count r i es and t erritori es s t udi ed, t he raw da t a
i nclude t he mont h-on-mont h rat e of i n fl a ti on and t he absolut e
4128
value of t he mon thly change i n t he (average) spot exchange rat e.
We deno t e t he latter as Bn, t for count r y n i n mont h t. The ex- by
change rat e i s evaluat ed a ga i ns t 11 candi dat e anchor currenci es.5 gues
The candi da t e anchors were chosen based on h i s t ori cal prac ti ce
and currenci es t hat are wi dely i ncluded i n exchange rat e basket s. t
on
Chi nese renmi nbi i s no t consi dered a candi dat e anchor because it t
ober
cle,abs
change rat e undergoes large fluct uati ons due t o a lack of mone t ary
cont rol (often accompani ed by the deple ti on of foreign exchange t
racU134/2/599/527
reserves and ot her cri s i s s ympt oms) from currenci es t hat fluc t uat e
freely i n t ran qu il ti mes.9 Toge t her, t hese feat ures of t he al gorithm
provi de a classificati on t hat separat es severe currency cri si s pe-
r i ods (temporary or persi s t ent) fr om t ran qu i l ti mes. The d i s tinc-
tion bet ween a currency "crash" and "turbulence" i s a matter of
gr
degree. For example, E i chen een, Rose, and W losz (1996) con- yp 4128
si der bot h an ext reme defi n iti on of a currency cri si s (two or more
s t andard devi a tions i n t hei r exchange market t urbulence i ndex) by
gues
fo r nearly a decade.
9. Thi s dimensi on of t he algorithm, for i ns t ance, places t he Sout h Korean won
i n t he freely falli ng cat egory from December 1997 t hrou gh J une 1998, and more
recently, t he Russi an ruble from November 2014 t hrough t he beginni ng of 2016.
Duri ng its famous 2007-2008 cri si s, t he Icelandi c krona only j us t mi sses t he crash
cu t off.
606 THE QUARTERLY J OURNAL OF ECONOMICS
and a m i lder versi on (turbulence rat her t han cri s i s).1° For reasons
we d i scuss lat er, our algorithm does no t t reat pe ri ods of elevat ed
t urbulence (whi ch do no t mee t t he cri s i s de fin iti ons) as a separat e
cat e gory. Downloaded
bands and managed fl oati n g (cat e gory 3), and freely fl oatin g (ca t-
e gory 4). We describe i n Section II. C how t he d i s tincti on be t ween
mana ged fl oat s and fr eely fl oating i s made. i
c
.
oup
Currenci es whose absolut e change vi s-a-vi s t he anchor cur-
.
rency was less t han 1% are classified as pe gs. Thi s fo llows t he de 8m/qje/a
cle,abs
changes b y less t han 1% i n absolut e value fo r 80% of consecutive
mont hly observati ons (P(cn, t < 1%) > 80%) 卫 Bands are assessed t
racU134/2/599/527
i n a five-year rollin g w i ndow, but t wo-year w i ndows are used t o
dat e s t a rt and end poi n t s. 12
Narrow (wi de) bands are defin ed as allowi n g fl uct uati ons of
2% (5%) v i s-a-vi s t he anchor currency. These bands were chosen
based on common exchange rat e practices.13 Formally, we classify
4128
10. One and a half st andard devi ati ons in t he index i s consi dered i n t hei r
sensitivity anal ysi s. by
11. If t hese adj us t ment s are mai nly in one di recti on (70% of t he time), we gues
classify t he currency as havi ng a crawli ng peg (cat egory 7) rat her t han a hard peg
(ca t egory 4) i n t he fi ne classifi ca ti on. t
on
12. In addition, currenci es t ha t had exac t ly 0% change fo r fo ur consecu tive 16
mon t hs are classified as pegs i n t hese mont hs. Oc
13. For example, t he Nati onal Bank of Denmark has a de j ure band of t
ober
土2.25% agains t t he euro (al t hough we fi nd t hat t he band i s smaller i n prac-
tice). Several euro access i on coun tri es have had 2% de j ure bands (e. g., Cyprus, 2019
and t he Czech and Slovak Republi cs). Si milarl y, t he offi ci al poli cy of t he People's
Bank of China si nce 2015 has been t o li mit renmi nbi fl uct uati on with respect t o a
baske t of currenci es t o be within a 2% band. In t hi s peri od t he renmi nbi exchange
rat e vi s-a-vi s t hi s basket has always changed by less t han 2% i n absolut e t erms.
In cont ras t, t he renmi nbi saw mon t hly changes exceedi ng 1% i n a quarter of all
mon t hl y observati ons. Thus a de fa ct o classificati on t hat differenti at es be tw een a
1% and a 2% band would di scri mi nat e be tween t hese cases.
EXCHANGE ARRANGEMENTS 607
cle,abs
a w i de band w ith res pect t o the dollar i n t he early 2000s. Thi s
precedes a peri od when Sout h Korea had a narrower band and i s t
racU134/2/599/527
consi s t ent w ith volumi nous narra tive e vi dence t hat t he count r y
14. Thi s cat egory also i ncludes crawlin g pegs (cat egory 7 i n t he fine classi-
fi cati on), as described i n no t e 13. We not e t hat t he 80% crit eri on allows some
fl exibility for exchange rat e movements-an avera ge of tw o a year withi n an —
ot herwi se fi xed exchan ge rat e re gi me. Allowi n g thi s wiggle room i s i mportant for
4128
several reasons. Fi rst, some cent ral banks may have occasi onal parity chan ges
withi n an ot herwi se fi xed exchan ge rat e re gime. Second, cen t ral banks may eval-
by
uat e exchan ge rat e t arget s based on differen t criteri a t han we do (end of peri od
i nst ead of average exchan ge rat es or a t a differen t frequency t han our mont hly gues
dat a). This would lead t o some devi a ti ons from t he bands when evalua t ed by our
t
criteri a. Thi rd, parallel exchan ge rat es may devi a t e from t he band even i n t he on
con t ext of a largely fi xed exchan ge rat e re gime. If an exchan ge rat e i s within a 2% 16
band but adj ust ment s are m ai nly i n one di recti on (70% of t he ti me), we classify Oc
t he curren cy as havin g a crawli n g band, si mi lar t o t he definiti on of a crawli n g peg. t
ober
15. Rei nh art and Rogoff (2004) used a s i mi lar procedure to classify wi de bands
as refl ecti n g changes ofl ess t han 5% i n absolut e value a t leas t 80% of t he time. 2019
16. We do no t reclass ify tw enti et h-cent ury regimes re t ros pec tively based on
thi s criteri on. Thi s i s i n t en ti onal, as t he definiti on was adapted t o t he lower ex-
chan ge rat e volatility i n t he tw enty-first cen tury. Det ermini n g t he break poi n t
based on t he t urn of t he millenni um i s perhaps arbitr ary, but Figure I illus-
t rat es t hat t he decline i n volatility was a con tinuous process, givin g no obvious
breakpoi n t. Thi s cat egory also i ncludes crawli n g bands (cat egory 11 i n t he fi ne
classificati on) as defi ned i n no t e 14.
608 THE QUARTERLY J OURNAL OF ECONOMICS
y
or longer-las ti n g phenomenon. Al t hou gh it i s be ond t he sco e of p
this a rti cle t o i nves tiga t e its causes, among possi ble explanations
are t he t rend decli ne i n i n fl a ti on rat es t hat has charact eri zed t h i s
peri od possi bly rei n forced by t he unusually h igh de gree of syn- fr
om
chroni city i n mone t ary poli cy from t he erupti on of t he cri si s i n https
2008 up until the end of 2015, a t which poi n t t he Federal Re-
serve embarked on t he process of unwi ndi n g from t he post cri si s :
//academ
policy s timulus ahead of t he ot her maj or cent ral banks.17 Nei-
t her explanation res t s on a change i n t he exchange rat e arrange-
ment s of t hese count r i es. As t o how i n t ernati onal t urbulence (as i
c
.
oup
opposed t o a domes tic currency crash) a ffect s exchange rat e pol-
.
i cy i n emergin g and developin g economi es, t he range of observed 8m/qje/a
out comes i s broad and het erogeneous; some count r i es have used
t urbulent times t o allow for significant currency depre ci a tion as
a boost t o competitiveness, while ot hers have shown an i ncreased rti
cle,abs
"fe ar of fl oatin g" as concerns about capital fl ow reversals and
balance sheet effect s domi nat e t he poli cy response. This cross- t
racU134/2/599/527
country he t erogeneity i n t he behavi or of t he exchange rat e i n
peri ods of heigh t ened i n t ernational t urbulence suggest s t ha t a
separat e cat e gory i s unwarrant ed. Furthermore, our emphasi s i s
on t he exchange rat e re gime (as our use of rolli n g five-year wi n-
dows highligh t s), no t i n descri bi n g what are possi bly short -li ved
changes i n t he poli cy reacti on fu ncti on of cent ral banks duri n g 4128
bout s of t urbulence.
All rem ain i n g cases are classified as fl oatin g. The followi n g by
17. The positive relati onship be t ween i nfl ati on levels and t hei r volatility i n
F i scher and Modigli ani (1978) can plausi bly ex te nd t o exchange rat e volatility.
EXCHANGE ARRANGEMENTS 609
out exchange rat es t hat were managed amon g fl oat ers. However,
fl oati n g emergin g marke t currenci es appear t o be more volatile
on average t han t hose of advanced economi es despite subs t an ti al
i n t ervention i n fo reign exchan ge marke t s. Thi s makes it difficult fr
om
t o const ruc t a count erfac t ual as t o how volatile emergin g mar- https
ke t currenci es would have been if t hey had been allowed t o fl oat
freely. :
//academ
Our new a pp roach i s based on a narrative assessment of cen-
t ral bank practices. The narratives are fully summari zed i n On-
line Appendix l. Our poi n t of departure i s t hat a freely fl oatin g i
c
.
oup
exchange rat e regime i s i nconsi s t ent with fre quent exchange rat e
.
i n t ervention with t he expli cit a i m of managin g t he level, pa t h, 8m/qje/a
cle,abs
where t here i s a preponderance of evidence t ha t t he i n t erventi on's
i n t ent was exchange rat e management. Sources i nclude cent ral t
racU134/2/599/527
bank mi nut es, reports, and s t a t ement s; t he IMF's Annual Report
on Exchange Arrangement s and Exchange Res tri cti ons; OECD
and BIS reports; t he U. S. Treasury's bi annual Report t o Congress
on Int ernati onal Economi c Exchange Ra t e Poli ci es; cent ral bank
dat a on reserve holdi n gs and ne t foreign exchange purchases;
press report s; and country-specific research. These narrative as- 4128
sessment s are a ppli ed only t o peri ods when a currency fl uct uat ed
out s i de of t he 5% band described i n Section II.B. by
gori es. F i rs t, t he cent ral banks of Aus t rali a, Canada, Sout h Afr i ca,
and t he United Kin gdom virtually never i n t ervened i n foreign ex-
change marke t s durin g th i s peri od 18 and held reserves an order
of magnitude lower t han ot her cent ral banks i n t he sample. Sec- fr
om
ond, t wo cent ral banks (Brazi l and Turkey) a ppear t o have bri efly https
a ttempted t o fl oat t hei r currenci es freely i n t he peri od 2000-2010
but reanchored t hei r exchange rat es and i n t ervened regularly as :
//academ
t he global fin anci al cri si s unfolded. The t h i rd cat e gory i ncludes all
remai n i n g cases, where cent ral banks i n t ervened re gularly bo th
before and durin g t he global fi nanci al cri si s. i
c
.
oup
We classify J apan as a freely fl oatin g currency, but t he narra-
.
tives illus t rat e t hat t h i s i s a borderli ne case. The Bank of J a pan 8m/qje/a
cle,abs
However, it i s very possi ble that foreign exchange i n t erventi ons
around 2012 were expli citly t arge t ed a t cappi n g yen a ppreci a ti on. t
racU134/2/599/527
In concludi n g t h i s secti on, it i s i mportant t o no t e t hat t he
di stincti on bet ween a managed fl oat and freely fl oati n g i s no t
t erribly consequential for many macro i ssues, and i n cases where
it i s i m portant, our fine 15-bucket classificati on can be used where
t he weigh t s are very si mi lar, with a freely fl oatin g bei n g a 13 and
managed fl oatin g a 12. 4128
18. There were rare and brief excepti ons in t he days fo llowi ng September 11,
2001, and t he collapse of Lehman Brot hers i n Septe mber 2008.
EXCHANGE ARRANGEMENTS 611
Alt hou gh t hese res tri cti ons may appear arbitrary, we fin d
t hem i n formative fo r a number of reasons. A more vari ed set of
candi dat e basket s would lead t o spu ri ous classificati on of curren-
ci es t o narrow bands. For any currency, one can fi nd a basket with fr
om
some anchor currency weigh t s, t o whi ch a given currency i s virtu- https
ally pegged with i n a short enough wi ndow. 19 Alt hough i n principle
basket s may cont a i n any number of currenci es, it i s i mportant t o :
//academ
reco gni ze t hat our classificati on shows t hat only seven are act u-
ally freely floatin g. Thi s means t hat any currency anchored t o a
basket contain i n g many currenci es i s i m plicitly anchori n g t o a f ar i
c
.
oup
smaller number. For example, t he People's Bank of Chi na (PBC)
.
publi ci zed t he weigh t s i n its re ference basket i n lat e 2015. The 8m/qje/a
cle,abs
and Si nga pore dollars, Myanmar rin git, and Thai baht). Hence
a classificati on t hat t ook t he basket weigh t s a t fa ce value would t
racU134/2/599/527
grossly unders t a t e t he ext ent t o whi ch t he renmi nbi continues
t o be anchored t o t he dollar. 2° F i nally, t hese si m ple basket s are
i n formative and i n t u itive. It i s i n t eres tin g, for example, t hat t he
dollar-euro basket was t he only one t o which even a small num-
ber of currenci es were anchored. No country has had a de fact o
dollar-yen basket or a dollar-euro-yen basket as its anchor. The 4128
rare cases of de fact o basket anchors were also i n t u itive: for exam-
ple, t he Icelandi c krona over t he pas t decade and t he Polish zlo ty by
16
19. Our a i m i s si mi lar t o Frankel and We i (1994), Frankel (2008), and Frankel Oc
and Xi e (2010), who a ttem pt to esti mat e baske t weigh t s. We re qu i re a different t
ober
approach due t o t he j oi n t t ask of classifyin g anchors and exchange rat e fl exibi lity.
20. In t he parti cular case of China, we also allowed fo r t he possi bility t hat t he 2019
renm i nbi was anchored de fact o t o t h i s de j ure baske t in t he years befo re it was
offi ci all y announced. It was not .
2 1. In fut ure research, it will be i n t eres ti n g t o assess t he de fact o basket
correspondi n g t o t he renmi nbi 's curren t de j ure one. At t he time of writin g, it i s
t oo soon t o t ell: i n t he t hree years since t he basket was announced, t he de fact o
dollar-euro baske t performs sligh t ly better t han t he offi ci al baske t itself, but one
should not e t hat thi s was a period of dollar a ppreci a ti on and t he al gorit hm would
be best t es te d over a longer peri od, i ncludin g dollar weakness.
612 THE QUARTERLY J OURNAL OF ECONOMICS
cle,abs
with 50% relative t o t he basket. Georgi a was t here fore assigned
a dollar anchor. In cont ras t, the Icelandi c krona was w ith i n a 5% t
racU134/2/599/527
band of bo t h t he euro and t he dollar-euro basket i n t he peri od
2009-2015. However, it was with i n a 2% band only of t he basket
and was t here fore assigned a dollar-euro basket anchor. In t hese
borderli ne cases, we check whe t her t he algorithm provi des an-
swers consi s t ent with ot her sources of evi dence. For example, t he
Georgian economy i s h ighly dollari zed, whereas Icelandi c t rade 4128
i s denomi nat ed i n dollars and euros i n roughly equal shares. 23
Beyond provi di n g a consi s t ent summary s t a tis ti c of t he anchor by
16
22. As i n t he discussi on of exchange rat e fl exibi lity, a 1% or 2% band re qui res
Oc
t he currency t o fl uct uat e withi n t he band for 80% of observati ons, while t he 5%
band requires t he currency to be withi n t he band 100% of t he ti me. t
ober
t ral bank fo reign reserves? F i nally, which was t he mos t recent an-
chor currency? Conveni ent ly, all fo ur i ndi cat ors poi n t t o t he same
reference currency i n almos t all countri es i n t he t able. In Online rti
cle,abs
Appendi x 2, we propose an i ndi cat or fo r refe rence classifi cati on
t ha t a ggrega t es t hese four measures. However, t he four measures t
racU134/2/599/527
are st rat egic complement s, and we t h i nk it i s no coi nci dence t hat
t he y give consi st ent predi ctions. 26
For complet eness, we assess t he robus t ness of our anchor
choi ce by s t udyin g t wo recent nat ural experi men t s. There have
been tw o large recent swi n gs i n t he bi lat eral dollar-euro exchange
rat e (see Onli ne Appendi x F igure A. 1). These movemen t s can be 4128
t raced back t o monet ary policy shocks i n Europe and t he United
St a t es. Firs t, on J uly 22, 2012, ECB Pres i den t Mario Dragh i made by
TABLE I
CLASSIFYING REFERENCE CURRENCIES FOR 胚AGED FLOATING CURRENCIES W兀'H
SUPPLEMENTARY DATA
Downloaded
Country (refe rence
currency) Years Indi cat ors
Brazil (dollar) 2000-2002 94% of exports and 84% of i mports
2008- pr i ced i n dollars. 90% of fo reign
fr
currency publi c and publ i cly om
guarant eed (PPG) deb t in dollars.
https
Pri or anchor: dollar.
Chile (dollar) 2008- Invoi ci n g dat a n.a., bu t given large :
//academ
share of copper i n export s and t he
denomi nation of copper pri ces i n
dollars, t he li on's share of export s are
likely denom i nat ed i n dollars. Pri or i
c
.
oup
anchor: dollar.
Colombi a (dollar) 2008- g
Close t o 100% of i nvoici n and 100% of .
8m/qje/a
PPG debt i n dollars. Previous anchor:
dollar.
Iceland (dollar-euro) 200 6--2009 Very divers ified i nvoici n g be t ween
dollar, Briti sh pounds, and euro. rti
cle,abs
Cen t ral bank FX reserves very
diversified. Anchored t o dollar-euro
baske t subse quently. t
racU134/2/599/527
Sou t h Korea (dollar) 2004-2009 Previ ous and subse quent anchor:
dollar. Ot her dat a unav ai lable.
Mexico (dollar) 2009- Invoicin g dat a n.a, bu t with more t han
80% of exports and nearly half of
i m port s from t he United St a t es, t he
dollar i s certa i nly t he mai n i nvoi cing
4128
currency. Maj ority of debt i nvo i ced i n
dollars. Previ ous anchor: dollar.
Paragua y (dollar) 2011-2013 Invoici n g dat a n.a. Nearly 90% of PPG by
Sources. Gopi nath (2015), World Bank interna tional deb t statistics, national cent ral banks, and aut hors'
calculations.
EXCHANGE ARRANGEMENTS 615
de t a i ls for t he full exerci se, we h ighligh t here t hat all i n all, t hese
t wo event s t udi es s t rongly corroborat e our anchor and refe rence
classifi cati ons.
fr
om
.
8m/qje/a
or fai l as anchors. Alt hough t he reruni nbi i s not an anchor cur-
rency i n our h i st ori cal analysi s, we nevertheless devo t e a ttenti on
t o t he possi bi lity t hat it m igh t be emergin g as a new pole of t he
i n t ernational mone t a ry order. We t hen shift our fo cus t o t he evo-
rti
cle,abs
luti on of exchange rat e arrangement s i n t he seven decades si nce
World War II. Of pa rticular i n t erest i s t he emergence of new typ es t
racU134/2/599/527
of de j ure mone t ary and exchange rat e arrangement s and t hei r
de gree of exchange rat e fl exibi lity. We ask whe t her t hese arran ge-
ment s are charact eri zed by t rends t oward great er exchange rat e
y
fl exi bility or b long cycles with no clear-cut t endency. We sup-
plement our di scussi on with an ext ensi ve measure of exchange
rat e res t r i cti ons and cont rols, which helps capt ure t he ext ent t o 4128
27. Thi s differs from t he proverbi al "ta per t an t rum" of t he previ ous year, when
t he Federal Reserve i ndi cat ed plans t o slow down and event ually reverse asse t
purchases as part of its quan tita tive easi n g poli ci es. Alt hough thi s announcemen t
di d creat e some volatility i n emergin g marke t currenci es, it had a relatively mut ed
e ffect on t he bi lat eral euro-dollar exchan ge ra t e.
616 THE QUARTERLY J OURNAL OF ECONOMICS
60 UK.pound
fr
om
50
https
40
:
//academ
30
20
US dollar
10 i
c
.
oup
。
1946 1956 1966 1976 1986 1996 2006 .
8m/qje/a
Number of coun tr ies we ighted by thei r share i n world GDP, 1950--2015, excludes freely
f alling cases
rti
I 00 , Percent cle,abs
90
t
80 racU134/2/599/527
70
60
50
40
4128
US dollar
30
by
20
gues
10
t
on
。
1950 1960 1970 1980 1990 2000 2010 16
Oc
FIGURE II t
ober
Pos t-World War II Maj or Anchor Currenci es
2019
To p: Share of countries, 1946-2015, excludes freely falli ng cases. Bottom: Num-
her of coun tries we igh t ed by t hei r share i n world GDP, 1950- 2015, excludes freely
fal lin g cases.
Sources: The Conference Board, To tal Economy Da tabase; Int ernati onal Monet a ry
Fund, In ternational 几nancial S t a tis tics; Rei nhart and Rogoff (2004) and sources
cited t here in; and aut hors'calculati ons. Currency baske t s were allocat ed propor-
tionally t o t he anchors represen t ed i n t he baske t. The Coun try Chronologies t hat
supp lemen t t h i s a rti cle show t he evoluti on of t he anchor currency on a country-
by-coun try basi s.
EXCHANGE ARRANGEMENTS 617
t he years a fter World War II, accent uat ed by t he United Kin gdom's
peri odic economi c cri ses. At t he end of t he Bretton Woods era, t he
DM emerged as t he domi nant European currency i n t he 1970s, rti
cle,abs
as many European countri es began t o shadow t he ac tions of t he
Bundesbank, expli citly or ot herwi se. t
racU134/2/599/527
Whi le t he U. S. dollar was t he currency of choi ce among t he
former Briti sh coloni es exitin g t he s t erlin g zone, t he loss of com-
paratively high-i ncome Europe t o t he DM led t o a shrin kin g of t he
dollar zone by 1980. Thi s dent t o t he U. S. dollar zone i s a pp aren t
i n F igure II. At t h i s time, ano t her t rend was t he falli n g share of
count r i es with a tigh t exchange rat e li nk t o t he U. S. dollar. 4128
The dollar's anchor currency s t a t us recei ved a boost i n
t he early 1990s a fter t he collapse of t he fo rmer Sovi et Uni on by
gues
28. Thi s and subse quent figures i nclude only i ndependen t (sovereign) s t a t es, t
on
which means t hat t he total number of en titi es more t han doubles over t he course
16
of t he sample, from 79 in 1946 to 188 in 2016. There are 6 te rritories fo r which
we have a full classifi cati on, bri n gi n g t he t ot al t o 194. The dat a se t also i ncludes Oc
pre-i ndependence de facto exchange rat e arran gement s so t hat it i ncludes monthly t
ober
dat a fo r 194 coun tri es from 1946 t o 2016.
29. The figure combi nes anchor and reference currenci es. A figure res tri ct ed 2019
t o anchors (i .e., excludi n g mana ged fl oati n g curren ci es) looks almos t i denti cal
and i s shown i n Onli ne App en dix F igu re A. 3. The share of world GDP anchored
t o t he dollar i s 65%, and an additional 5% of world GDP has t he dollar as a
refe rence currency. Onli ne App endix F igure A. 4 reweigh t s anchors based on t he
de gree of exchange ra t e fl exi bi lity, givi n g a h igher weigh t t o more fi xed exchan ge
arrangemen t s. Gi ven t hat t he eurozone members recei ve t he h ighes t we igh t, t he
euro bloc i ncreases i n si ze, but t he domi nance of t he dollar bloc rem ai ns.
618 THE QUARTERLY J OURNAL OF ECONOMICS
all ot her currenci es. As i n fl a tionary cri ses became much less com-
mon i n t he t wenty-fi rst cent u ry (to dat e), nearly all t he countri es
with freely falli n g currenci es i n t he lat e 1970s-1990s have rean- rti
cle,abs
chored t o t he dollar. Arguably, one can plausi bly rei n t erpre t the
h i st ory of t he freely falli n g cases and conclude t ha t t hese coun- t
racU134/2/599/527
tries mai n t a i ned a U. S. dollar anchor even duri n g t he years of
very h igh i n fl a ti on, as t hei r t rade, deb t s, and hard-currency re-
serves conti nued t o be denomi nat ed i n U. S. currency. Duri n g t hese
long and chroni c high i n fl a ti on bout s, many of t hese countri es be-
came significantly doll arized domesti cally. 31 The one place t hat
t he dollar li nk was no t a pp arent was i n t he s inkin g value of t hei r 4128
currenci es.
The French franc zone, whi ch had its larges t roo t s i n French by
Afri ca, mos t explicitly i n t he CFA currency arran gement, held gues
30. As not ed i n Secti on 11.B, freely falli ng captures all t he cases where t he
12-mont h i nfl ati on rat e exceeds 40%. The i nci dence of freel y f alling over 1946-
2016 i s di scussed lat er i n t hi s section.
31. On de fact o dollari zation see, for inst ance, Calvo and Ve gh (1999), Ize and
Levy-Yeyati (2003), and Rei nh art, Rogoff, and Savast ano (2014).
EXCHANGE ARRANGEMENTS 619
fr
Ill. B. Dri vers of Anchor Currenc i es om
https
The role of a domi nant i n t ernati onal currency has evolved
over time along with the comi n g and goi n g of great powers and t he :
//academ
cle,abs
pa rtly owed t o t he quality of t he coi nage but also t o t he fact t hat
her coloni es, i ncluding Mexi co, Bolivia, and Peru, account ed fo r t
racU134/2/599/527
fo ur-fifths of global silver production bet ween 1493 and 1850 (see
Chalmers 1893; P i a tt 1904). But t he i n t ernational role of Spa i n's
currency was also cat alyzed b y t he country's role as t he domi-
nant global t radi n g and mi lita ry power i n t he sixteent h cent u ry.
Once ent renched, t he cent rality of t he Spa nish si lver dollar (and
lat er Mexican silver dollar) persi s t ed long a fter Spai n had ot her- 4128
wi se faded t o t he periphery of Europe a fter t he Napoleoni c Wars.
Spani sh coi nage di d have competition a t ti mes. Durin g part s of by
t he sevent eent h and eigh t eent h cent uri es, t he Dut ch guilder ci r- gues
y,
Duri n g t he nine t eent h cent ur t he United Kin gdom emerged
as t he world's domi nan t t rade, m ilitary, and bankin g superpower.
The gold-backed Briti sh pound i n t urn became t he i n t ernati onal
currency, a position it ret a i ned long a fter t he United St a t es had
far surp assed t he United Ki n gdom i n global t rade and manufa c-
t uri n g production. Even a fter t he devas t a tion of World War I, t he
620 THE QUARTERLY J OURNAL OF ECONOMICS
currenci es, mai nly t he pound s t erlin g. The modern const ruct of
t he anchor or reference currency, however, i s much broader and
also re fl ect s a currency's role in t rade i nvoicin g, global bond i s- i
c
.
oup
suance, and exchange rat e management. Our a rti cle parti cularly
.
emphasi zes measures of the latter, because t he currency countri es 8m/qje/a
cle,abs
What i s t he gravit a ti onal pull t oward a (si n gle) anchor cur-
rency? Why do so man y cent ral banks choose t o place a significant t
racU134/2/599/527
weigh t on dollar exchange rat e s t ability, fa r out of proportion t o
t he si ze of t he United St a t es i n global t rade or financi al market s?
The out si ze role of t he dollar re fl ect s a number of mu t ually rei n-
forci n g fa ct ors, mos tly (but no t entirely) deri vi n g from t he fa ct t hat
t he global currency i s a nat ural monopoly i n bot h goods and fl.n an-
ci al market s. Economi s t s have long no t ed t he convenience fact or 4128
firms and consumers derive when t he bulk of t rade i s i nvoi ced i n
a common currency (e. g., Swoboda 1969). Recent t rends i n global by
gues
t rade t oward i ncreasi n gl y large multicountry supply chai ns, mos t
not ably within 知ia, have rei n fo rced t he convenience advan- t
on
t a ge of dollar prici n g. Bacchetta and van Wi ncoop (2005) and 16
Goldberg and Tille (2008) emphasi ze t he i m portance of s t ra t egic Oc
complement ariti es and pri ci n g t o marke t i n t he choi ce of currency t
ober
32. Ahamed (2009) emphas i zes how t he United Kingdom's domi nance i n bank- 2019
cle,abs
ity mi smat ch. By shadowi n g a large count r y's exchange rat e, a
small count r y can effectively make domes tic currency bonds less t
racU134/2/599/527
r i sky i n real t erms fo r global i nves t ors, allowi n g t he small country
t o enj oy some of t he "saf e asse t" advant a ges t hat nat urally accrue
t o a large currency (see Hassan 2013).
Gopinat h and St ei n (2018) develop a comprehensi ve model
t ha t emphasi zes t he self-rei n forci n g advant ages of t he domi nant
currency i n goods and asse t prici n g. Thei r model demons t rat es 4128
how h igh demand fo r "safe" dollar-denomi nat ed U. S. Treasury
bonds leads t o a reduced i n t erest rat e t hat i nduces non-U. S. by
corp orat es t o i ssue dollar debt as well. As :firms i ssue dollar- gues
p
has much t o do with olitical i ns t abi lity with i n t he euro syst em
and li n gerin g uncerta i n ty over t he fut ure of t he si n gle currency.
The nat ural advant a ges of t he dollar m igh t seem much less so
i n a world where national political shifts di d no t continuously Downloaded
cle,abs
of world reserves held i n t h i s currency, t he share of ext ernal deb t
t ha t i s denomi na t ed i n t he anchor currency, and an i ndex t hat t
racU134/2/599/527
summari zes t he ext ent t o whi ch world t rade i s denomi nat ed i n
t he anchor currency.33
The pict ure emergin g from Table II i s fai rly consi s t ent across
i ndi cat ors and confirms our assessmen t t hat t he U.S. dollar st ands
out as t he domi nant anchor. Based on t he classificati on approach
outlined i n Secti on II, t he dollar serves as t he anchor or re ference 4128
currency for 60% of t he countries i n our s t udy i n 2015. About t wo-
thi rds of t he world's foreign exchange reserves are held i n U.S. dol- by
33. The t rade-i nvoi ci n g i ndex i s based on t rade-i nvoicin g da t a from Gopina t h 2019
(2015) and i s det a i led i n Onli ne App endix 4. It averages t he percent of countries
with any t rade i nvo i ced in a given anchor currency with t he share of all t rade
i nvoi ced in t hat currency.
34. Currency denomi nati on of reserves i s available i n t he a ggrega t e bu t not
on a coun try-by-country basi s.
EXCHANGE ARRANGEMENTS 623
TABLE II
MARKERS OF AN ANCHOR CURRENCY
Percen t
Downloaded
Panel A: Anchor measure or criteri a : U. S. dollar
Share of countries with a U. S. dollar anchor i n t he i r exchan ge 59
rat e arrangemen t s
Share of world's reserves (excludi ng gold) i n U. S. dollars 65
fr
Share of developi n g country external deb t denom ina t ed i n U. S. 64 om
dollars. (Thi s does not i nclude debt owed t o Chi na
https
denomina te d i n U. S. dollars)
Trade i nvoici n g "i nde x'' 69 :
//academ
Memorandum item:
Share of t he U. S. i n world GDP 18
Panel B: Anchor measure or criteri a : euro i
c
Share of countries with a euro anchor in t hei r exchan ge rat e 29 .
oup
arrangemen t s
.
8m/qje/a
Share of world's reserves (excludi ng gold) i n euros 20
Share of develo pin g country external deb t denomina t ed i n 13
euros
Trade i nvoici n g "index" 56 rti
cle,abs
st a t us as world anchor currenci es duri n g t h i s peri od. For each of
t he fact ors, t he score i s less t han one-sevent h of t he dollar's com- t
racU134/2/599/527
bi ned score. The lat e Ronald McKinnon, i n several of h i s papers on
what he called East Asi a's dollar s t andard, emphasi zed t h i s poi n t
(see. e.g., McKi nnon and Schnabl 2004). As Gopinat h's (2015) dat a
h ighligh t, about 50% of J apan's exports and over 70% of its i m-
port s are denomi nat ed i n U. S. dollars. Furthermore, J a pan's li nk
t o t he U. S. dollar last ed longer t han Europe's and well pas t t he 4128
breakdown of t he Bretton Woods arrangement s, as J a pan m ain-
tained a narrow de fact o band un til 1977. (See t he compani on by
medi a t ely a pp arent. The renmi nbi has been s t rongly anchored https
t o t he dollar, so t hat shadowi n g t he dollar and t he renmi nbi
:
are observati onally e qu i valent. In 2005, t he PBC sligh t ly loos- //academ
.
oup
i n g it nearly i mpossi ble t o d i s ti n guish anchorin g t o t he renmi nb i .
8m/qje/a
from anchorin g t o t he dollar. 36 On ot her d i mensi ons, t here are
s till no s igns t hat t he renmi nbi has emerged as an a ltema-
tive anchor. Only 30% of Chinese t rade i s denomi nat ed i n ren- rti
m i nbi, and it has no t served as a vehi cle currency for any ot her cle,abs
35. Prasad (2016) discusses China's ambitions and poli ci es t o make t he ren-
2019
mi nbi an i nt ernati onal currency over t he lon g run.
36. Applying our anchor classificati on al gorithm over t he bri ef peri od si nce
2016 det ect s t wo candi dat e renminbi anchors: t he Malaysi an ringgit and Thai
baht. Thi s may be spurious and result from t hese currenci es havi ng some weight s
i n t he PBC's offici al baske t (4.6% and 3.3%, respectivel y), or because t hese Sout h-
eas t Asi an cent ral banks are putti ng larger euro weight s i n t hei r i mpli cit basket s,
rat her t han a renmi nbi anchor.
626 THE QUARTERLY J OURNAL OF ECONOMICS
cle,abs
goes beyond t he scope of t h i s work. Applying our anchor classifica-
tion algorithm t o the bri e f peri od si nce 2016 rai ses t he possi bi lity t
racU134/2/599/527
t hat t he renmi nbi i s now anchored t o a dollar-euro basket (th i s
50-50 basket performs sligh tly better t han t he de j ure baske t).
We no t e t hat th i s has been a peri od of st rong dollar a pp reci a ti on
q
and a longer wi ndow i s re u i red t o fully classify t he PBC's new
exchange rat e practi ces. Gi ven t he si ze of t he Chi nese economy,
reclassifyin g Chi na t o a basket will have some effect on t he share 4128
of world GDP anchored t o t he dollar and euro (with t he dollar bloc
declin i n g from 70% t o 61 % and t he euro bloc i ncreasi n g from 15% by
t o 23%) and even more so if ot her maj or emergin g marke t s follow gues
suit. We ret urn t o t he possi bi lity of a lat ent renmi nbi anchor i n t
on
t he conclusi ons.
16
Oc
III. D. Exchange Ra te Arrangemen ts t
ober
Havi n g described t he t rends i n anchor currenci es, we t urn t o
t he global evolution of exchange arrangement s. F igure III shows 2019
37. The Belgian franc was t he si ngle European currency t hat shadowed t he
German revaluati on of 1969.
EXCHANGE ARRANGEMENTS 627
(A) Groups I and 2: Less flexibility, primarily nominal exchange rate anchors
IOO l pereent Group 2. Graduahst adJUS tment
"午^”、^~``'l ;mn 九^”“、“
90
80
Downloaded
70
60
50
40
fr
30 om
Group I: Least fl e xi ble
20 (from no separate legal tender to de facto pegs) https
10
:
//academ
。
1940 1950 1960 1970 1980 1990 2000 2010
70
60
rti
50 cle,abs
IO
。
1940 1950 1960 1970 1980 1990 2000 2010
80 by
70 Group 5: Freely fa lling, gues
i nfl ati on > 40%
60 Group 6: Multiple, dual, or or currency crash
parallel markets with lim ited or t
50 no data on the structure o f on
exchange rates 16
40
30 Oc
t
20 ober
10
2019
。
1940 1950 1960 1970 1980 1990 2000 2010
FIGURE III
De Fact o Exchan ge Ra t e Arran gemen t s, Coarse Classifi cati on, 1946-2016: Share
of (Independen t) Coun tries i n Each Group
628 THE QUARTERLY J OURNAL OF ECONOMICS
cle,abs
The i nci dence of managed and free fl oat s (F igure III, Panel
B) rei n forces t he fi ndi n g t hat fl exible arran gement s are no t as t
racU134/2/599/527
common s i nce t he breakdown of Bretton Woods as one m igh t have
t hou gh t . Freely fl oati n g exchange rat es are still largely confined
t o a handful of economies. True, if i ns t ead of focusi n g on t he share
of count r i es i n each cat egory, we weigh t t he a ggre ga ti on by t he
country shares i n world GDP, as i n F igure IV, t hen t he share of
fl oat ers nearly doubles t o somewhere bet ween 30% and 40%. Si nce 4128
t he United St a t es and J a pan fl oat freely, t h i s already accoun t s fo r
23% of world GDP. Thus, t o develop a sense of count r y practi ces by
100
Freely Falling
90 、
80 Freely Floating
Downloaded
70
60
50 fr
om
40
https
30
:
//academ
20
10
。
i
c
.
oup
1950 1960 1970 1980 1990 2000 2010
.
8m/qje/a
FIGURE IV
De Fact o Exchan ge Ra t e Arran gemen t s, Coarse Classifi cati on, 1946-2016: Share
of World GDP i n Each Group
rti
cle,abs
l 098765432
Average Fine Classification
t
racU134/2/599/527
4128
Excluding Freely by
Falling gues
t
on
16
。 Oc
1940 1950 1960 1970 1980 1990 2000 2010 t
ober
FIGUREV 2019
cle,abs
out; now i n t he third year of a sharp decli ne i n oi l and commod-
ity pr i ces, losses i n revenues, deplet ed foreign exchange reserves, t
racU134/2/599/527
and markedly slowi n g economi c ac tivity may d rive some count r i es
back t o i nfla ti onary fi nance. The most ext reme case i s Venezuela
(where estimat ed i nfla ti on exceeded 1,000% i n 2017 and ent ered
severe h yp eri nfla ti on t erritory i n 2018) but a number of Afri can
countri es are also experi enci n g i nfla ti on deep i n t o double digits,
i ncludi n g the Democratic Republi c of t he Congo, Sout h Sudan, and 4128
An gola.
g
As no t ed earlier, our classificati on al orithm t akes i n t o ac- by
count parallel exchange rat e marke t s, whi ch have a t times been gues
qu ite i m port ant. In t he early post -World War II years, t he maj or- t
on
ity of count r i es pa rticip a t ed i n t he Bretton Woods sys t em, peggin g
16
t hei r currenci es t o t he U. S. dollar, whi ch itself was convertible t o Oc
t
ober
40. Of course it i s possi ble t ha t some count ri es may have relatively st able ex-
change rat es for ext ended periods si mply because t hey face rela tively small global 2019
and domes tic macroeconomi c shocks. Al t hough we do no t have space t o t ackle t he
i ssue here, it has been addressed i n a number of papers i n t he classificati on liter-
at ure, i ncludi ng Calvo and Rei nhart (2002) and Rei nh art and Ro goff (2004). The
general fi ndi ng i s t ha t mos t countri es t hat st abi li ze t hei r exchange rat es do so
despite faci ng consi derable volati lity i n external conditions (especi ally commodity
export ers) t ha t would normally i nduce large adj us t ment s i n a fl exible exchange
rat e envi ronment.
EXCHANGE ARRANGEMENTS 631
cle,abs
t he Sovie t Uni on and Yugoslavia could, a t any poi n t i n ti me, li st
a dozen admi n i s t ered exchange rat es. Continued scarcity meant t
racU134/2/599/527
t ha t black currency marke t s were active despite repressi on. Un-
like Wes t ern European countri es, for whi ch we have t he parallel
market exchange rat e dat a , we do no t have as complet e a dat aset
i n t he Sovi et bloc. Hence, we leave t hese cases under t he label
"p arallel marke t s—no dat a".
Las t, si nce t he eurozone compr i ses more t han 15% of world 4128
GDP, any conclusi on about the evolution of global exchange rat e
arran gement s and t hei r de gree of fl exi bility i n recent decades by
depends i m port ant ly on how t he exchange rat e prac ti ces of eu- gues
fi scal-monet ary i n t eractions, deb t cri ses). Obvi ously t he euro itself
i s a freely fl oati n g currency and i s i n fa ct a maj or anchor currency,
as our previ ous analysi s sugges t s. For comple t eness, Online Ap-
pendix F igure A. 8 repeat s F igure III, while replaci n g eurozone
members with t he eurozone as a whole, classified as freely fl oat-
i n g. The general pa tterns remai n i n t act.
632 THE QUARTERLY J OURNAL OF ECONOMICS
with differences across re gions, i ncome levels, and exchange rat e https
polici es. The proliferation of IT as a de j ure mone t ary re gime has
:
been a development of t he pas t t wo t o t hree decades, with a more //academ
.
oup
an IT policy fr amework, t he dat es of its i nception, and t he de fact o .
8m/qje/a
exchange rat e re gime classificati on on t he basi s of exchange rat e
behavi or.
As t he t able h ighligh t s, t here i s consi derable vari a tion i n de rti
4 1. A moving band refe rs to t he cases where peri ods of sus t ai ned appreci ations
are also evi dent; with crawli ng bands, changes are always i n t he di recti on of
depreci ati on.
EXCHANGE ARRANGEMENTS 633
Ill.F. Capital Mob i l ity, Multiple Exchange Ra tes, and Parallel Downloaded
Marke ts
In mos t of t he literat ure on classifyin g exchange rat e arrange-
ment s, t he relat ed i ssue of capital mobility has been ignored al- fr
t oge t her. Thi s omi ssi on i s a t odds with recurrent d i scussi ons of om
cle,abs
row compared with earli er i ndexes t hat a ggre ga t e a wi de vari-
e ty of (mostly) de j ure capit al fl ow and exchange res t r i cti ons.42
However, because our measure i s market based (drawi n g on our t
racU134/2/599/527
42. Kose et al. (2003) develop a de fact o model of capital marke t openness bu t
do no t fo cus on exchange rat e rest ri cti ons.
43. The IMF report i s complemen t ed with t he publi cati ons by Franz Pi ck and
lat er hi s firm (In te rnati onal Currency Analysi s) and P i ck and Sedillot (1971). The
AREAER report also offers det ai led information on a wi de class of capital and
634 THE QUARTERLY J OURNAL OF ECONOMICS
can also pot enti ally overst a t e t he e ffectiveness of cont rols and
t here fore underes timat e t he act ual de gree of capital mobility.
In t h i s cont ext, i n formation on parallel market exchange rti
cle,abs
rat es and premi a can provi de a "de fa ct o" sense of t he e ffective-
ness of cont rols and complement t he chronology on t he de j ure t
racU134/2/599/527
unita ry/dual/multiple exchange rat es. The mont hly i ndex we con-
st ruc t fo r 192 countri es or t erritori es from 1946 t hrough 2016 i s
t herefo re based on bot h a de j ure and de fact o component and on
t he answers t o t hree questi ons: (i) i s t here a de j ure (o ffi ci al) dual
marke t?; (ii) i s t here a de j ure sys t em of multiple exchange ra t es?;
(iii) i s t here an i n formal parallel market (tolerat ed or outrigh t i l- 4128
lega l), and, if t here i s, i s t he parallel marke t premi um above 10%
over t he maj ority of a movi n g 12-mont h peri od? If t he answer i s by
i s O ot herwi se.45 t
on
How do t he v arious a pp roaches t o measuri n g capital mobi lity
16
compare? Conceptually a t leas t, a count r y can have a plet hora of Oc
capital account res tri cti ons and still have a de fa ct o and a de j ure t
ober
unified exchange rat e; t he converse i s no t t rue. If t he answers
2019
current account res tri cti ons, which serve as a common st arting poi nt for many of
t he avai lable i ndexes of capital mobi lity, as di scussed in Onli ne App en dix 7.
44. The premi um i s defi ned as 午,the percent age difference bet ween t he
parallel market and the offici al exchange rat e.
45. Re fi nement s t o t he i ndex t hat allow for values bet ween O and 1 are cer-
t ai nly a possi ble ex t ensi on.
EXCHANGE ARRANGEMENTS 635
t o ques tions (i) and (ii) are yes, t hese are de j ure cont rols. If t he
answer t o question (iii) i s yes, it i s difficult t o see how a signifi-
cant and sust ai ned ga p bet ween t he offi ci al and t he parallel mar-
ke t exchange rat e can persi s t i n a country where capital moves Downloaded
cle,abs
wi se globally i n t egra t ed.47
With t hese caveat s i n mi nd, t he t op panel of F igure VI plo t s t
racU134/2/599/527
t he i ndex (as a share of all i ndependent count r i es) with and with-
out weigh t s t hat reflect country shares i n world GDP. Alt hou gh
t he i ndex i s available si nce 1946, GDP weigh t s are only avail-
able for many count r i es si nce 1950. About 70% of all count r i es
di d no t mee t t he criteri a of a unified exchange marke t i n 1950.
In t he 1960s, t hat share drops t o around 50%, as many advanced 4128
economi es moved t o elim i na t e multip le exchange rat e prac ti ces
(an i mportant goal of t he IMF a t t he time, as di scussed by De by
Vri es 1969, and shown i n F igure III, Panel C). The ne xt round of gues
46. De fact o capital mobi lity re fers here t o cases where t he e xi s ti n g de j ure
con t rols are no t bindi n g, either because t hese are bei n g ci rcumvent ed or because
t hey have become out moded or obsolet e.
47. See Onli ne App endi x 7 and Onli ne App en dix F igu re A. 13.
636 THE QUARTERLY J OURNAL OF ECONOMICS
`........ .. / https
\
`.--`“'寸 ..
.. J,...`_,俨
:
//academ
..._,、己一一·-I..
II
Share we ighted b y GDP (dashed line) -,
.
I i
c
.
` .....俨 ...
oup
。 .......-"、一--...- .
8m/qje/a
1950Ml 1960Ml 1970Ml 1980Ml 1990Ml 2000Ml 2010Ml
Advanced econom i es
rti
000000000
98765432l cle,abs
t
racU134/2/599/527
4128
。
`',__
、一 t
16
on
cle,abs
A. 12). We suggest t he CI i ndex should be i n t erpre t ed as an u pper
bound on t he presence of cont rols, whi le the IRR i ndex i n t roduced t
racU134/2/599/527
here provi des t he lower bound (see Onli ne Appendix 7 for de t a i ls).
global financi al syst em. One appli cation i s t o a t opic t hat has t
on
a ttrac t ed the a ttenti on of scholars and poli cy makers ali ke for
16
more t han a decade now: t he surge i n reserve accumulati on si nce Oc
t he early 2000s by emergin g market s i n general and by China i n t
ober
pa rti cular. The self-insurance rationale fo r accumulatin g reserves
(often refe rred t o as t he demand for safe asset s, e. g., Caballero, 2019
Farhi, and Gouri nchas 2017) i s well known. We conj ect ure t ha t
t he persi st ence of less flexi ble exchange arran gement s may be
an i mport ant contributi n g fa ct or t o t he saf e asset buildup, whi ch
has li kely been amplified i n recen t decades b y t he t rend loosen-
i n g of capital and exchange cont rols (until recently). As exchange
and capital marke t res tri cti ons are reduced, it probably re qu i res
638 THE QUARTERLY J OURNAL OF ECONOMICS
capital cont rols become easi er t o evade, and t here fore much less
effective. That count r i es with relativel y i n fl exible exchange rat es
would hold subst an ti al fo reign exchange reserves i s hardly a con- rti
cle,abs
t roversi al proposition. Wha t has become a pp arent i n recen t years
i s t ha t even countries with relatively more fl exi ble managed fl oat- t
racU134/2/599/527
i n g exchange rat e syst ems find it helpful t o have significant re-
serves on hand, particularly t o help s t abili ze exchange rat es dur-
i n g peri ods of duress. (As Gabai x and Maggiori 2015 poi n t out, a
small count r y peggin g t o t he dollar does not necessari ly have t o
hold all its hard currency reserves i n dollars.) Of course, i n many
ways, J a pan's pre-2003 successf ul foreign exchange i n t erventi on 4128
has been a model for man y managed fl oatin g re gimes.
by
gues
IV.A. The Imp ossi ble Tri n ity and t he 2003-2013 Reserve Surge
t
Si nce t he IMF was es t ablished a t t he end of World War II, on
48. Obs tfeld, Shambaugh, and Taylor (2010) and Ai zenman, Chi nn, and Ito
(2013, 2016) also i nves tiga t e how t he tri lemma i n t era cts with t he Triffin di lemma
i n ligh t of t he h igh demand fo r exchan ge rat e managemen t.
49. Virtually all coun tri es with relativel y infl exi ble exchange ra t e re gi mes i n
our dat a set m ain tain signifi can t levels ofr eserves, and t hese levels have i ncreased
as capital account res tri cti ons have fallen in recent years.
EXCHANGE ARRANGEMENTS 639
ean
arch
90 un .tnes wm
cor
e a re g es
, ___ _ - - , - _ _,
te
、
80
『
、
, ,,
、
、
、
-- ,,
\`
、
7060 ,
,,' - ,
, ,_
、
\`
__
、
、
、
,_ __` _ ( ,'
Downloaded
、
、、
、
-, -
g ~, g >
50
40
fr
30 om
literat ure has exami ned t he causes of t hat growth. Some pa- rti
cle,abs
pers have s t ressed t he precautionary, self-insurance mo tive (see
Gouri nchas and Obstfe ld 2012), while ot hers have h ighligh t ed t he
mercanti li s t mo tive and t he desi re t o avoi d or li mit exchange rat e t
racU134/2/599/527
exchange rat e fl exibi lity (top panel) and t he decli ne i n exchange 2019
cont rols, or r i si n g capital mobi lity (bo ttom panel). 50
50. The bui ldup i n reserves may have been especi ally acute durin g thi s period
when many emergi n g market s were also dealin g with a severe capit al infl ow
problem (as documen t ed by Re i nhart, Re i nhart, and Trebesch 2016 and tryi n g t o
avoi d excessi ve currency a ppreci a ti on as s t ressed by Levy-Yeya ti, St urzenegger,
and Gluzmann 2013).
640 THE QUARTERLY J OURNAL OF ECONOMICS
90
Share of countries with less flexi ble
80 exchange rat e re gimes
Downloaded
,、,-_..`一一、 ,-..---_,一、
: 『«',了','````--:--,,、、~~'__,,产',', /1~一、、
70
fr
om
40 、_(\.._.,/ https
30
\~一^v/ \ _ /广 :
//academ
20
Share o f emerging and developing colllltry
10 reserves in total world reserves
i
c
。 .
oup
1950 1960 1970 1980 1990 2000 2010
.
8m/qje/a
80 rti
cle,abs
70
Share o f countries with dua l/multiple
60 parallel exchange rates t
racU134/2/599/527
50
40
30
20 4128
10
by
。 gues
1946 1956 1966 1976 1986 1996 2006 2016
t
on
FIGURE VIII
16
Reserves and Two Si des of t he Impossi ble Tri n ity, 1950-2015 Oc
Sources: Int ernati onal Monet a ry Fund, In ternational F i nanc i al S t a tis tics; Bu- t
ober
reau of Economi c Analy s i s; and au t hors'calculati ons.
2019
cle,abs
q uantita tive h i s t ori cal perspective, thi s a rti cle offers new i nsigh t s
i n t o cont emporary global finance i ssues ran gin g from t he i m pos- t
racU134/2/599/527
si ble trin ity t o t he modem-day Triffin dilemma t o t he renewed
primacy of t he dollar as t he world's mos t i m port an t anchor cur-
rency. Perhaps because of wi despread hard-currency t rade i nvoi c-
i n g and deb t denomi nation, t he revealed pre ference of many of
t he world's cent ral banks i s t hat exchange rat e s t abi lity remai ns
a fu ndament al priority t hat has broad i mpli cati ons fo r how shocks 4128
t ransmit domes tically and i n t ernationally. Countries'continui n g
desi re t o s t abilize exchange rat es despite generally reduced ex- by
change rat e res tri ctions and i ncreasi n g capital mobi lity i s pot en- gues
cle,abs
many's DM separat ed from t he dollar that it became evi dent t hat
t he European economi es had already t ransitioned from a dollar t o t
racU134/2/599/5274128
a DM anchor. (An i n t eres tin g parallel i s t hat even as t he res t of Eu-
rope followed t he dollar until t he end of Bretton Woods, Germany
was already maki n g subs t anti al government-t o-government loans
i n DM, layi n g the groundwork fo r t he DM bloc t hat emerged as
Bretton Woods f ell a part.)
If i ndeed count r i es'desi re t o st abilize exchange rat es i s a
cont r i buti n g fa ct or t o emergin g marke t s'reserve accumulation,
t hen an emergence of t he renmi nbi as a maj or reserve currency by
could have i m port an t i m pli cati ons for t he value of t he dollar and gues
advanced count r y i n t erest rat es, and pot enti ally even deb t sus- t
on
tainability fo r some advanced economi es.
16
Las t, we reiterat e t hat our al gorithm for j oi n tly de t ermi n- Oc
i n g a count r y's anchor currency and its de gree of exchange rat e t
ober
fl exibi lity shows a world where relativel y inflexi ble exchange rat e
regimes remai n ext remely i m port ant, and where t he dollar's dom- 2019
SUPPLEMENTARY MATERIAL
REFERENCES i
c
.
oup
Aghion, Phi lippe, Philippe Bacchett a, Romain Ranc i ere, and Kenne t h Rogoff, "Ex-
change Rat e Volatility and Productivity Growt h : The Role of F i nanci al Devel- .
8m/qje/a
opment," J ournal of Monet a ry Econom i cs, 56 (2009), 49 4-513.
Ahamed, Li a quat, Lords of F i nance: The Bankers Who Broke the World (New York:
Pen gui n Press, 2009).
Ai zenman, J oshua, Menzi e D. Chinn, and Hiro Ito, "The'Impossi ble Trinity' H y- rti
pot hes i s i n an Era of Global Imbalances: Measurement and Tes ti n g," Rev i ew cle,abs
of In terna tional Economi cs, 21 (2013), 44 7-4 58.
, "Monet a ry Policy S pillovers and t he Tri lemma i n t he New Normal: Periph-
ery Country Sensitivity t o Core Count r y Conditions," J ournal of In terna tional t
racU134/2/599/527
Money and F i nance, 68 (2016), 29 8-330.
Ai zenman, J oshua, and J aewoo Lee, "In t ernational Reserves: Precautionary ver-
sus Mercan tili s t Vi ews, Theory and E vi dence," Op en Economy Rev i ew, 18
(2007), 191-214.
Bacchett a, Phi lippe, and E ri c van Wi ncoop, "A Theory of t he Currency Denomi na-
tion of Int ernati onal Trade," J ournal of In terna tional Economi cs, 67 (2005),
295-3 19
Ball, Chris t opher P., Claude Lopez, and J avi er Rey es, "Remittances, Infl a ti on 4128
and Exchan ge Rate Re gi mes in Small Open Economi es," World Economy, 36
(2013), 487-5 07. by
Bruno, Valenti na, and H yun Song Shin, "Global Dollar Credit and Carry Trades:
A F i rm-Level Anal ysi s," BIS Workin gPaper no. 510, 2015. gues
Bussi ere, Matthi eu, Gong Cheng, Menzi e D. Chi nn, and Noemi e Lisack,''For a
Few Dollars More: Reserves and Grow t h i n Times of Cri ses," J ournal of In ter- t
on
na ti onal Money and F i nance, 52 (2015), 127-145.
16
Caballero, Ri cardo, Emmanuel Farhi, and P i erre-Oli vi er Gouri nchas, "The
S afe Asset s Conundrum," J ournal of Economi c Persp ec ti ves, 31 (2017), Oc
29-4 6. t
ober
Calvo, Guillermo A., and Carmen M. Re i nhart "Fear of Floati n g," Quarterly J our-
nal of Economi cs, 117 (2002), 379-408. 2019
Calvo, Guillermo A., and Carlos A. Ve gh, "Infl a ti on S t abili za ti on and BOP
Crises i n Developi n g Count r i es," i n Handbook of Macroeconom i cs, J ohn
B. Taylor and Mi chael Woodfo rd eds. (Amst erdam: Elsevi er, 1999), vol. 1,
1531-1614.
Chalmers, Robert, A H i s tory of Currency i n t he British Coloni es (London: E yre and
Spotti swoode, 1893).
Chin n, Menzi e D., and H iro Ito,''What Matters fo r Fi nanci al Development? Capital
Con t rols, Ins titu ti ons, and In t eractions," J ournal of Develop men t Economi cs,
81 (2006), 163-192.
644 THE QUARTERLY J OURNAL OF ECONOMICS
Faudot, Adrien, and Jean-Fran f oi s Ponsot, "The Dollar Domi nance: Recen t cle,abs
E pisode of Trade Invoi ci n g and Debt Issuance," Journal of Economi c lnte-
gra tion, 31 (2016), 41-64.
F i scher, St anley, and Franco Modiglian i, "Towards an Unders t an di n g of t he Real t
racU134/2/599/5274128
E ffect s and Cos t s of Infl a ti on," Wel t wi r tschaftl i ches Arch i v, 114 (1978), 810-
833.
Frankel, Je ffrey, "Estimation of De Fac to Exchange Rat e Regi mes: Synthesi s of t he
Techniques fo r Infe r ring Flexi bility and Baske t Weigh ts," IMF St a ff Papers
55, 2008.
Frankel, Je ffrey, and Shang-Ji n Wei,"Yen Bloc or Dollar Bloc? Exchange Rat e
Polici es of the Eas t As i an Economi es," in Macroeconomi c L i nkages: Sav i ngs,
Exchange Ra tes, and Capital Flows, NBER East As i a Semi nar on Economi cs,
Takat oshi I t o and Anne Krueger, eds. (Chi cago: Un i vers ity of Chi cago Press,
1994), volume 3. by
Frankel, Jeffrey, and Dani el Xi e, "Es timati on of De Fact o Fle xi bility Paramete r
gues
and Baske t Weigh t s in Evolvin g Exchange Rat e Re gimes," Ameri can Economi c
Rev i ew Pap ers and Proceedi ngs, 100 (2010), 568-572.
t
Gab aix, Xav i er, and Ma tteo Maggiori, "In t erna ti onal Liqu idity and Exchange Rat e on
Dyn ami cs," Qua rterly J ournal of Economi cs, 130 (2015), 1369-1420.
16
Ghosh, Ati sh R., Mahvash S. Qureshi, and Charalambos G. Tsan garides, "Is t he
Exchan ge Ra t e Re gi me Really Irrelevant fo r E xternal Adj us t men t?," Eco- Oc
nom i cs Letters, 118 (2013), 104- 109. t
ober
Goldberg, Li nda S., and Ce dric Tille, "Vehi cle Currency Use i n In t ernational
Trade," Journal of In terna tional Economi cs, 76 (2008), 177-192. 2019
Go pinat h, Gita, "The Int erna tional Pri ce Sys t em," Jackson Hole Symposi um (Fed-
eral Reserve Bank of K皿sas City), 2015.
Go pinat h, Gita, and Jeremy S t ei n, "Bankin g, Trade, and t he Maki n g of a Domi nan t
Currency," Mi meo, Harvard Un i vers ity, 2018.
Go uri nchas, P i erre Oli vi er, and Mauri ce Obstfeld, "S t ories of t he Twentie t h Cen-
t ury fo r t he Twenty-Fi rs t," Ameri can Economi c J ournal: Macroeconomi cs,
4 (2012), 226-265.
EXCHANGE ARRANGEMENTS 645
Habi b, Mauri zi o Mi chael, Sascha Biitzer, and Livio S t racca, "Global Exchan ge
Ra t e Configura ti ons: Do Oil Shocks Ma tter?," IMF Economi c Rev ie w, 64
(2016), 443-4 70.
Hassan, Tarek A., "Country S i ze, Currency Uni ons and Int ernational Asse t Re-
t urns," J ournal of凡nance, 68 (2013), 2269-2308. Downloaded
Hassan, Tarek A., Thomas M. Mertens, and Tony Zhan g, "Currency Manip ulati on,"
Mi meo, Un i vers ity of Chi cago, 2016.
Hau, Harald, and H啦ne Rey, "Exchange Rat es, E qu ity P rices, and Capital Flows,"
Rev i ew of凡nancial S t ud i es, 19 (2006), 273-317.
He, Zhiguo, 如ndK百shnamurthy, and Kons t an ti n Mi lbradt, "A Model of Save
Asse t Det ermi nation," NBER Worki n g Paper no. 22271, 2016. fr
om
Heat hcote, Jonat han, and Fabri zi o Perri, "On t he Desi rability of Capital Cont rols,"
IMF Economi c Rev i ew, 64 (2016), 75-102. https
Ilze t zki, E t han, Carmen M. Re i nhart, and Kenne t h S. Rogoff, "Coun t r y Chronolo-
gies and Background Mat erial t o Exchan ge Rat e Arran gemen t s i n t o t he 21s t :
//academ
Cen t ury: Will t he Anchor Currency Hold?," NBER Workin g Paper no. 23135,
2017.
, "Repli cati on Dat a fo r:'Exchange A订angements En t eri n g t he Twen ty-
F i rs t Cen t u ry: Whi ch Anchor Will Hold?'," Harvard Dat averse (2018), do i: i
c
10.7910/DVN/I DEXPY. .
oup
In te rnati onal Currency Analy si s, World Currency Yearbook (Brooklyn, NY, vari ous
.
years). com/
Ito, Taket oshi, and Tomoyoshi Yabu, "Wha t Prompts Ja pan t o Inte rvene i n t he
Forex Marke t? A New Approach t o a Reacti on Functi on," Journal of ln terna- qj
e/a
tional Money and F i nance, 26 (2007), 193-212.
Ize, Al ain, and Eduardo Levy-Yeya ti, "Financi al Doll ari zati on," Journal of In ter- rti
Obs tfeld, Mauri ce, and Kenne t h Rogoff, Founda tions of Int erna tional Macroeco-
nom i cs (Cambri dge, MA: MIT Press, 1996).
Obs tfeld, Maurice, J a y C. Shambaugh, and Alan M. Taylor, "The Trilemma i n
H i s t ory: Tradeo ffs Among Exchange Rat es, Mone t a ry Polici es, and Capital
Mobility," Rev i ew of Economi cs and S ta ti s tics, 87 (2005), 423-4 38. Downloaded
, "Fi nanci al Stability, t he Trilemma, and In te rnati onal Reserves," Ameri can
Economi c J ournal: Macroeconom i cs, 2 (2010), 57-94.
Obs tfeld, Mauri ce, and Alan M. Taylor, "Globali zati on and Capit al Market s," i n
Globaliza tion i n H i s tori cal Persp ec ti ve, Mi chael B. Bordo, Alan M. Taylor and
J e ffrey G. Williamson, eds. (C hicago: Un i vers ity of Chi cago Press, 2003).
P i a tt, Andrew A., "The End of t he Mexi can Dollar," Quarterly J ournal of Eco- fr
om
nom i cs, 18 (1 904), 321-356.
P i ck, Franz, P i ck's World Currency Rep orts (New York: P i ck Publishi n g, va ri ous https
years).
P i ck, Franz, and Rene Sedillo t ,All t he Mon i es of t he World: A Chroni cle of Currency :
//academ
Values (New York: Pick Publishing, 1971).
Prasad, Eswar, Ga i n ing Currency: The R i se of the Renmi nb i (O xford: Oxford Uni-
vers ity Press, 2016).
Raj an, Raghuram G., and Arvind Subram ani an, "Wha t Underm ines Ai d's Impac t i
c
on Growth?," NBER Workin g Paper no. 11657, 2005. .
oup
Re i nhart, Carmen M., "The Re t urn of Dollar Shorta ges," Proj ect Syndi cat e, 2016.
.
Re i nhart, Carmen M., Vi ncent Reinh art, and Chri st oph Trebesch, "Global Cycles: com/qje/a
Capital Flows, Commodities, and Sovereign Defa ult s, 1815-2015," Ameri can
Economi c Rev i ew, 106 (2016), 57 4-580.
Re i nhart, Carmen M., and Kenne t h S. Rogoff, "The Modern Hi s tory of Exchange
Rat e A汀angements : A Reinterp re t a ti on," Quarterly J ournal of Economi cs, rti
2019