Intersectoral conflict and delays in macroeconomic stabilization

Razmi, A slan
Wo king Pape
In e sec o al con lic and delays in mac oeconomic
s abiliza ion
Wo king Pape , No. 2024-4
P o ided in Coope a ion wi h:
Depa men o Economics, Uni e si y o Massachuse s
Sugges ed Ci a ion: Razmi, A slan (2024) : In e sec o al con lic and delays in mac oeconomic
s abiliza ion, Wo king Pape , No. 2024-4, Uni e si y o Massachuse s, Depa men o Economics,
Amhe s , MA,
h ps://hdl.handle.ne /20.500.14394/22344
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/300486
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/4.0/
Uni e si y o Massachuse s Amhe s Uni e si y o Massachuse s Amhe s
Schola Wo ks@UMass Amhe s Schola Wo ks@UMass Amhe s
Economics Depa men Wo king Pape Se ies Economics
2024
In e sec o al Con lic and Delays in Mac oeconomic S abiliza ion In e sec o al Con lic and Delays in Mac oeconomic S abiliza ion
A slan Razmi
Economics Depa men , Uni e si y o Massachuse s Amhe s
Follow his and addi ional wo ks a : h ps://schola wo ks.umass.edu/econ_wo kingpape
Pa o he Mac oeconomics Commons
Recommended Ci a ion Recommended Ci a ion
Razmi, A slan, "In e sec o al Con lic and Delays in Mac oeconomic S abiliza ion" (2024).
Economics
Depa men Wo king Pape Se ies
. 356.
Re ie ed om h ps://schola wo ks.umass.edu/econ_wo kingpape /356
This A icle is b ough o you o ee and open access by he Economics a Schola Wo ks@UMass Amhe s . I has
been accep ed o inclusion in Economics Depa men Wo king Pape Se ies by an au ho ized adminis a o o
Schola Wo ks@UMass Amhe s . Fo mo e in o ma ion, please con ac schola wo ks@lib a y.umass.edu.
In e sec o al Con‡ic and Delays in
Mac oeconomic S abiliza ion
A slan Razmi
Ma ch 22, 2024
Abs ac
An impo an body o li e a u e explo es he poli ical economy easons
unde lying delays in mac oeconomic s abiliza ion. This pape de elops
a amewo k o analyze con‡ic be ween wo g oups o economic ac o s,
one ha has an endowmen o in e na ionally adable goods and ano he
ha is endowed wi h non- adable goods. The ocus is on he exchange
a e policy in a de eloping coun y se -up whe e he go e nmen employs
seignio age e enue o …nance spending p e-s abiliza ion, and aces …scal
and balance o paymen s p oblems ha necessi a e s abliza ion wi h a s ep
de alua ion. The p esence o exchange a e and endowmen unce ain y,
he ole o owa d-looking expec a ions, and he possibili y o IMF aid
in‡uence he likelihood, iming, and e ms o a na ional consensus on
s abiliza ion in in e es ing ways.
JEL classi…ca ion: E31, F34, F41
Keywo ds: Mac oeconomic s abiliza ion, seignio age, in‡a ion, de alu-
a ion, capi al ‡igh , IMF p og ams.
Depa men o Economics, Uni e si y o Massachuse s, Amhe s , MA 01003; email:
a [email protected]
1
When he [ illage ] was caugh s ealing onions, he Panchaya [ illage coun-
cil] ga e him he op ion o choosing be ween wo punishmen s: ea ing a 100
onions o su¤e ing a 100 lashes. He chose onions, ga e up a e ea ing 10 and
asked o lashes; hen changed his mind a e ecei ing 10 lashes, and so on.
Ul ima ely, he ended up bea ing he agony o bo h.
–Folk s o y om no he n India.1
1 In oduc ion
On Decembe 13, 2023 he A gen ine go e nmen announced a 50 pe cen de-
alua ion o he peso agains he US dolla (see Figu e 1). This was a e
a long pe iod o ela i ely gen le bu s eady dep ecia ion o he peso agains
majo in e na ional cu encies, high and inc easing in‡a ion, and la ge …scal
and cu en accoun de…ci s. A simila i sligh ly less d ama ic pa e n played
ou housands o miles away in Pakis an as he coun y nego ia ed e ms o a
new IMF s abiliza ion and adjus men p og am. Many i no mos economis s
would a gue ha hese de alua ions (and o he accompanying measu es) we e
long o e due.
Mo e b oadly, mac oeconomic s abiliza ion p og ams in de eloping coun-
ies ypically ollow sus ained pe iods o expanding …scal and cu en accoun
de…ci s. These de…ci s, in u n, a e o en mani es a ions o deepe unde lying
mac oeconomic p oblems such as an inabili y o collec ax e enues, inapp o-
p ia e agg ega e demand policies ha lead o exchange a e o e alua ion, and
a eso o seignio age o …nance go e nmen spending in he ace o declining
demand o domes ic money. The ques ion, howe e , is why don’ hese s a-
biliza ions occu soone han hey ypically do, especially once he necessi y o
s abiliza ion becomes ob ious o impo an segmen s o he economy? Why do
some ac o s, in o he wo ds, delay decisi e ac ion only o e en ually end up,
like he illage in he p o e bial s o y, ecei ing bo h onions and lashes (i.e.,
bea ing he cos s o delay and …nally ag eeing on less a o able e ms). This
pape con ibu es o he li e a u e add essing his ques ion by analyzing he
poli ical economy o delays in s abiliza ion om an unde -explo ed angle, i.e.,
con‡ic o e he desi able pos -s abiliza ion le el o he exchange a e and he
expec a ions and unce ain y in ol ed he ein.
This pape is mo i a ed by se e al obse ed empi ical egula i ies:
The abandonmen o a …xed exchange a e egime o c awling peg o en
accompanies mac oeconomic s abiliza ion in de eloping coun ies (see, o
example, Co nia (2020) o a de ailed discussion). Fu he , his aban-
donmen o en in ol es de aluing he cu ency o b ing i s alue close
o he open/pa allel ma ke a e. As no ed abo e, his mean a 50 pe -
cen de alua ion in he case o A gen ina whe e, acco ding o epo s, he
1This e sion is ci ed o i s b e i y om he Oc obe 7, 2013 edi ion o India To-
day (h ps://www.india oday.in/magazine/co e -s o y/s o y/19900131-mili an -mo emen -
holds-kashmi -in-a-s a e-o - iolen -siege-sepa a ism-ge s-new-legi imacy-812287-1990-01-30).
0
200
400
600
800
Jan-20 Jan-21 Jan-22 Jan-23 Jan-24
Mon h-yea
Figu e 1: A gen inian peso pe US dolla (mon hly a e).
Sou ce: The Cen al Bank o he A gen ine Republic
((h ps://www.bc a.gob.a /PublicacionesEs adis icas/P incipales_ a iables_da os_i.asp)).
gap be ween he wo a es was as high as 150 pe cen a he ime o he
de alua ion.
IMF-suppo ed p og ams ha e ypically in ol ed a signi…can nominal de-
p ecia ion agains he US dolla . O en his is mo i a ed by IMF condi-
ionali ies ha equi e uni…ca ion o he o¢ cial and uno¢ cial exchange
a e. Gündüz and Da ius (2021) epo , o example, ha he bila e al
exchange a e is-a- is he U.S. dolla dep ecia ed, on a e age, by abou
13 pe cen o e he pe iod be ween 6 mon hs p io o and 36 mon hs a e
IMF p og am app o al.2This issue was also salien du ing ecen IMF
nego ia ions wi h coun ies such as A gen ina and Pakis an.
De alua ions a e unpopula among impo an segmen s o socie y since
hese esul in income edis ibu ion. Mo e han hal a cen u y ago,
Coope (1971) e med de alua ions “one o he mos d ama ic – e en
auma ic –measu es o economic policy ha a go e nmen may unde -
ake.” A i id illus a ion o his auma is he calcula ion by S einbe g
and Malho a (2014) ha mili a y dic a o s los powe du ing 17 pe cen
o hei 48 de alua ion episodes and democ a ic leade s in 38 pe cen o
hei de alua ions be ween 1973 and 2006.
In pa icula , p oduce s in he non- adable goods p oducing sec o s op-
pose de alua ions. B oz e al. (2008) use … m le el da a om he Wo ld
Bank’s Wo ld Business En i onmen Su ey and …nd ha owne s and
manage s o … ms p oducing adable (manu ac u ing) goods a e mo e
2See also K uege e al. (2003) and Ageno (2008, chap. 11).
3

likely o epo an exchange a e app ecia ion as a business conce n com-
pa ed o p oduce s o non- adables (cons uc ion and se ices). Based on
OLS and GMM es ima es o a baseline panel o 68 de eloping and 39 de-
eloped coun ies be ween 1989-2013, Ugu lu and Razmi (2023) …nd ha
a highe sha e o non- adable sec o ou pu is co ela ed wi h less unde -
alued RERs. This …nding is consis en wi h he iew ha unde alued
RER policies a e unpopula among non- adable indus ies
The e is some e idence ha his dislike o de alua ions inc eases wi h
he sha e o impo ed in e media e cos s in p oduc ion. Again, (Ugu lu
and Razmi (2023)) p o ide some econome ic e idence in his ega d. The
… m-le el empi ical e idence p esen ed by Egan (2017) suppo s he iew
ha … ms wi h a high dependence on impo ed inpu s a e unhappy wi h
eal dep ecia ion. In e es ingly, as no ed by S einbe g (2015), e en he
p oduce s o highly adable goods some imes oppose de alua ions (o , he
case o ‡exible exchange a es, unde alua ion), i hei impo ed in e me-
dia e con en is high. Consis en wi h hese …ndings, Weldzius (2021) a -
gues ha an inc ease in adable inpu s s emming om he globaliza ion
o p oduc ion ne wo ks has lowe ed suppo o unde alua ion e en in
coun ies ha ha e in e ened in o eign exchange ma ke s o unde alue
hei cu encies.
The e is signi…can unce ain y abou no jus he magni ude o e¤ec s
ha a gi en deg ee o exchange a e change will induce bu also he deg ee
o which an o¢ cial announcemen o change ac ually pins down he ma ke
exchange a e. This unce ain y, in u n, ends o in‡uence how e en
p oponen s and opponen s o s abiliza ion a e in hei a i udes. See
McNama a (2001) o a mo e de ailed discussion.
Finally, heo y and e idence sugges ha expec a ions o de alua ion ha
esul om high in‡a ion inc ease ade misin oicing and o he o ms o
e asion o axes on e enues om adable p oduc ion (Pa naik e al.
(2012)). T ade misin oicing, in o he wo ds, is endogenous o expec ed
exchange a e policy.
This pape analyzes he poli ical economy o s abiliza ion wi h he le el o
he exchange a e as he cen al ocus. This is impo an since, apa om being
one o he mos impo an mac oeconomic p ices in he economy, he exchange
a e is also di¤e en as a a iable om ax a es o go e nmen spending since
i is di ec ly a¤ec ed ahead o he ime o ac ual policy change by expec a ional
changes. This ocus on he exchange a e and i s in e play wi h he ex en o
dependence on impo ed inpu s helps shed in e es ing new ligh on he possible
easons o en unde lying delays. The analysis also demons a es he ambiguous
ole o o eign aid and aid- ela ed condi ionali ies –in pa icula hose a¤ec ing
u u e exchange a e le els –in acili a ing o inhibi ing delay.
The wo k he e is ela ed o se e al s ands o li e a u e including hose an-
alyzing: (1) he ole o expec a ions and ins umen unce ain y in in‡uencing
4
economic agen s wi h di e se p e e ences o e a gi en choice se , (2) he poli ical
economy o delayed s abiliza ions, (3) he p e e ence among la ge sec o s o pop-
ula ion o o e alua ion, (4) he p e e ence o unde alua ion among adable
sec o agen s, (5) he opposi ion o unde alua ion some imes e en among man-
u ac u e s, (6) he poli ical economy o cu ency c ises and IMF s abiliza ion
p og ams.
The nex sec ion p o ides some mo e backg ound. Sec ion 3 de elops he o -
mal analy ical amewo k s a ing wi h he baseline case whe e no o eign/IMF
aid is in ol ed, hen inco po a es he possibili y o IMF aid in “bad” imes,
and …nally analyzes he case whe e p e-announced IMF ecommenda ions ake
away unce ain y abou he u u e, pos -s abiliza ion exchange a e. Sec ion 4
concludes.
2 Backg ound and Li e a u e
One could a gue, wi h a lo o jus i…ca ion, ha he delay in mac oeconomic
s abiliza ion commonly expe ienced in coun ies on he e ge o …scal and bal-
ance o paymen s c ises lea es almos e e yone wo se o¤. This poses in e es ing
ques ions o ponde o social scien is s. No su p isingly, hen, in e es in he
poli ical economy o mac oeconomic s abiliza ion su ged du ing he 1990s ol-
lowing he limi ed success o nume ous p og ams ac oss he de eloping wo ld.
D azen (2000) p o ides a comp ehensi e, i da ed, su ey o he ini ial li e a u e.
Among models ha explici ly inco po a e empo al dynamics o add ess his
ques ion, Alesina and D azen (1991) p o ide an ea ly ea men . In hei model,
ising …scal de…ci s equi e a change in he policy egime which is delayed by
an ongoing “wa o a i ion”be ween wo sides. Each side p e e s o shi he
bulk o he adjus men cos s o he o he side bu has impe ec in o ma ion
abou he la e ’s cos s uc u e. Delay esul s as each side hopes o wai i
ou .
Casella and Eicheng een (1996) ex end he Alesina and D azen model o
inco po a e expec a ions o o eign aid, and show ha an announcemen o aid
ha lowe s he expec ed bu den o adjus men o p i a e g oups has ambigu-
ous e¤ec s on he likelihood o delay. The lowe cos o adjus men acili a es
expedi ion. On he o he hand, i he e is a lag be ween he announcemen o
he aid and i s disbu semen , his will end o delay s abiliza ion. Wi h an
ongoing wa o a i ion, an ea ly announcemen se es he pu pose o accel-
e a ing in o ma ion dispe sal, encou aging he lose o concede soone . Lags
be ween announcemen s and dispe sals, by con as , p esen incen i es o delay
admission o de ea un il he a i al o aid. Thus he ole o aid depends on he
iming o he ans e .
The p e ious wo pape s assumed an exogenous …scal de…ci . Velasco (1999)
changes his in a dynamic con ex . The bene… s o he …scal spending a e g oup-
speci…c while he cos s a e join ly bo ne. This dis o s incen i es, leading o
o e spending and deb accumula ion.
5
D azen (1996) p o ides a use ul big pic u e axonomy by poin ing ou ha
he e a e wo b oad app oaches o analyzing delays in mac oeconomic policy
change and s abiliza ion, each based in a di¤e en manne on he e ogenei y o
in e es s. One app oach ocuses on impe ec in o ma ion abou he ne bene… s
ha he o he g oup will ecei e pos -s abiliza ion, al hough he ne bene… s
o e o m hey hemsel es would ecei e a e known o each g oup be o ehand.
This lack o in o ma ion c ea es unce ain y abou he willingness o he o me
o pay o e o ms which, in u n, means ha e o ms may be shunned e en
i hey would bene… a majo i y i unde aken. The o he app oach cen e s
a ound he s a us quo bias c ea ed by unce ain y abou he bene… s o e o ms
ha indi idual g oups will hemsel es ecei e.
The p esen con ibu ion alls in he second ca ego y, al hough I ocus on he
con‡ic caused by he he e ogenous e¤ec s o exchange a e changes o di¤e en
g oups. In addi ion o i s p ac ical ele ance as a mac oeconomic a iable –
exchange a e and balance o paymen s issues ha e been a he co e o many a
ound o IMF nego ia ions and mac oeconomic s abiliza ion p og ams – ocusing
on i also enables me o inco po a e he ole o expec a ions in a quali a i ely
di¤e en manne . This comes ou s a kly in sec ion 3.3 whe e knowledge abou
he pos -s abiliza ion le el o he exchange a e in‡uences he alue o he open
ma ke exchange a e oday.
The wo pape s closes o my amewo k a e Al a o (2002) and Laban and
S u zenegge (1994). Laban and S u zenegge (1994) model delay as he a-
ional ou come o dis ibu ional con‡ic be ween wo isk-a e se g oups, he
“ ich”and he “poo ”in a wo-pe iod amewo k. S abiliza ion in ol es ans-
e o esou ces om he o me o he go e nmen o onwa d ans e o he
la e . Also, he o me ha e access o echnology ha allows hem o hide
income h ough …nancial adap a ion; a skill ha de elops o e ime. I pos -
s abiliza ion payo¤ unce ain y is su¢ cien ly high, he wo g oups may end up
delaying adjus men e en hough he ela i e posi ion o he poo de e io a es
o e ime in he absence o an ag eemen . This is ue e en hough, unlike in
he wa o a i ion amewo k, he iden i y o he side ha loses mo e om
unde alua ion is known ex-an e o all, and seems o be mo e in line wi h he
expe iences o de eloping coun ies such as A gen ina and Pakis an discussed
b ie‡y ea lie .
I de elop a wo-pe iod amewo k o a gue, in he spi i o Laban and
S u zenegge (1994), ha delays in unde aking much-needed and much-an icipa ed
exchange a e e o ms a e no bes unde s ood as a planne a emp ing o max-
imize he wel a e o a ep esen a i e indi idual. Ins ead in e -g oup con‡ic o -
en plays he lead ole. The abo e-men ioned pape di ec ly inspi es my model
and he logical s uc u e ollowed by my a gumen , e en hough he con‡ic in
my case is be ween wo g oups ha ha e endowmen s ha di¤e in a di¤e en di-
mension. One g oup has an endowmen ha can be aded eely in e na ional
ma ke s a while he o he has a non- adable endowmen . Mo eo e , he e is
no di ec ax ans e om one g oup o he o he . The endogenous a iable o
in e es in my case is he nominal exchange a e a he han he ax a e. These
changes a e mo i a ed by my in e es in he scena io whe e he coun y aces
6
se ious p oblems on bo h he …scal and balance o paymen s on s, and wishes
o e o m by shi ing o ax collec ion a he han seignio age-based …nancing
o de…ci s. The e is a con‡ic o in e es be ween he adable sec o (which
expe iences a boos in income ollowing a de alua ion) and he non- adable
sec o (which expe iences a loss o income), and his leads o delay in s abiliza-
ion i he wo g oups a e su¢ cien ly isk-a e se. The ocus on he nominal
exchange a e and he balance o paymen s adds o he exis ing li e a u e in
impo an ways. Fi s , i is o signi…can p ac ical impo ance. The con‡ic
o in e es s be ween he adable and non adable sec o s, lies a he hea o
F ieden (1991)’s seminal “in e es g oup heo y o exchange a e p e e ences”.
Second, ocusing on he exchange a e ins ead o axes sheds ligh on addi ional
new aspec s o he issue since, unlike axes: (1) he le el o he exchange a e
is less de e mina e; he e o en exis gaps be ween he o¢ cial and open/black
ma ke s exchange a es – he la e being mo e widely accessible – e en un-
de …xed exchange a e egimes in de eloping coun ies acing ex e nal accoun
p oblems, and (2) changing expec a ions can modi y he exchange a e ele an
o ma ke pa icipan s e en be o e a change in policy is ac ually implemen ed.
The o he pape close o he p esen one, Al a o (2002), has a simila se -up
o he ex en ha i analyzes an economy wi h a adable sec o and a non-
adable one. Howe e , he pape signi…can ly di e ges in ha i employs a cash-
in-ad ance se -up and analyzes he poli ical economy o exchange a e based
s abiliza ions. In con as o he case he e, ha pape analyzes a empo a y
s abiliza ion ha in ol es a empo a y educ ion in he a e o de alua ion and
a esul ing eal app ecia ion o con ol in‡a ion. This policy hu s he adable
goods owne s and has ambiguous e¤ec s on he non- adable good owne s. In
he p esen pape , a pe manen de alua ion hu s he ela i e posi ion o non-
adable p oduce s, who may s ill ag ee o s abiliza ion unde ce ain condi ions,
albei wi h a delay ha u he weakens hei ela i e posi ion.
Finally, as men ioned ea lie , Casella and Eicheng een (1996) analyze he
ole o he iming o he announcemen o o eign aid in in‡uencing he likelihood
o delay in s abiliza ion. Howe e , unlike he p esen con ibu ion, hey do so
in a con inuous ime “wa o a i ion”se -up. A majo implica ion is ha aid
educes he bu den o adjus men on he g oup ha is he … s o concede, which
expedi es concession. In he p esen se -up, each g oup has pe ec in o ma ion
abou he cos s uc u e o he o he g oup so ha ea ly announcemen does
no esol e in o ma ional ambigui ies and does no induce an ea lie concession.
Also, hei ocus is on axes a he han he exchange a e so ha he policy
ins umen ( he ax a e) is comple ely non-s ochas ic.
3 In e sec o al Con‡ic : The Model and Ana-
ly ics
As in Laban and S u zenegge (1994), conside a wo-pe iod se -up bu wi h an
economy consis ing o wo ypes o agen s, one called T( o owning a adable
7
In ui i ely, ecall ha a de alua ion educes N-agen consump ion by eN
while inc easing T-agen consump ion by (1  )eT. Thus, he e¤ec o he
elimina ion o in ela i e p e-s abiliza ion dis o iona y ax bu dens depends on
he sha e o impo ed inpu s. The highe his sha e, he g ea e he ela i e
damage o N-agen s om a de alua ion, and hence he less likely i is ha hey
will ag ee o s abilize.
Co olla y 2: P o ided ha : (a) ("2) + (1 2)g > eN, i.e., he dis o iona y
ax bu den p e-s abiliza ion exceeds he pos -s abiliza ion ax cos s o he N-
sec o , and (b) S2>1, i.e., a de alua ion ollows an ag eemen , a highe sha e
o impo ed inpu s sh inks he possible se o ag eemen s in Pe iod 2 o a gi en
isk p emium.
P oo : Pe haps he bes way o show his is by ecalling ha policy make s make
de alua ion a p econdi ion o s abiliza ion in ligh o he balance o paymen s
si ua ion, and by hen conside ing (12) w i en in a sligh ly di¤e en o m:
R < 1 + ("2) + (1 2)geN
eN
R+ 1 ("2) + 2g+c(h
2; K)eT
(1  )eT
(13)
Gi en a s ep de alua ion as a necessa y backg ound condi ion, he e m in
he squa e b acke s on he igh hand side –which is also he igh hand side o
inequali y (11) –mus by necessi y be g ea e han one. Since R > 0, his means
in u n ha ("2) + (1 2)g > eNis a necessa y condi ion o s abiliza ion.
Gi en he sa is ac ion o his condi ion (and ha S2>1), he second e m
on he igh hand side o (13) implies ha a highe sha e o impo ed inpu s
educes he likelihood o an ag eemen .
N-agen s a e hu by a de alua ion in p opo ion o he in e media e sha e
o ou pu . I he sha e o impo ed inpu s is high enough, he cos s o a
s ep de alua ion o hem ou weigh he cos s om in‡a ion in he absence o
s abiliza ion.
Fo la e e e ence, le ’s deno e he exchange a e ag eed on o s abiliza ion
in Pe iod 2 by S
2(>1). Thus, o s abiliza ion o occu in Pe iod 2, S2N>
S
2>S2T>1.
P oposi ion 3: Gi en ha "2> "1, and 2< 1 om P oposi ion 1, he ela i e
equilib ium posi ion o N-agen s in e ms o he exchange a e de e io a es o e
ime (i.e., S
2> S
1).
P oo : The p oo ollows om he esul ha posi i e …nancial adap a ion
(g ea e misin oicing) in Pe iod 1 esul s in g ea e in‡a ion in o de o …nance
he highe ax bu den in Pe iod 2. To unde s and his, conside ha N-agen s
will be willing o s abilize in Pe iod 1 a he han Pe iod 2 i hei discoun ed
expec ed u ili y om doing so is g ea e . Tha is,
14

(1 + )U [1  (S1N+R)]eNg  U[(1  )eN("1)(1 1)g]
+U [(1  (S
2+R)]eN
Le 
S1Nbe he maximum jump de alua ion ha N-agen s a e willing o
accep in Pe iod 1 gi en ha he e will be s abiliza ion in Pe iod 2. Then he
exp ession abo e will hold wi h equali y.
(1 + )U1 
S1N+ReN=U[(1  )eN("1)(1 1)g]
+U [1  (S
2+R)]eNg(14)
Inequali y (10) ells us ha :
1 
S2N+ReN= (1  )eN("2)(1 2)g
Using his equa ion and he esul s om P oposi ion 1, namely, "2> "1, and
2< 1, allows us o ew i e and simpli y exp ession (14):
(1 + )U1 
S1N+ReN>(1 + )U[1  (S
2+R)]eN
o ,

S1N< S
2
and, since 
S1Nis he maximum de alua ion accep able o N-agen s, S
1<
S1N,
so ha
S
1< S
2(15)
Wi h ime, he T-agen s a e able o conceal a g ea e p opo ion o hei
income ab oad so ha , ba ing s abiliza ion, hei sha e o he dis o iona y
ax bu den is lowe in Pe iod 2, imp o ing hei ba gaining posi ion.
P oposi ion 4: I 2R > 
(1 )+("1)+(11)g
eN+("1)+1g+c(h
1;
H1)
(1 )eT,i is an
equilib ium s a egy o delay ag eemen e en as he N-agen s lose ba gaining
powe o e ime and he e is a non-emp y se o possible ag eemen s in Pe iod
2.
P oo : Again, le 
S1Nbe he implied de alua ion ha lea es N-agen s indi¤e -
en be ween ag eeing o no o s abiliza ion in Pe iod 1, gi en ha s abiliza ion
is desi able in Pe iod 2. Then 
S1Nis he solu ion o:
(1 + )U1 
S1N+ReN=
U (1  )eN("1)(1 1)gg+U [1  (S
2+R)]eNg(16)
15
Simila ly, he implied de alua ion ha lea es T-agen s indi¤e en , S1Tis
gi e by:
(1 + )U [(1  ) (S1TR)]eTg=
U(1  )eT("1)1gc(h1;
H1)+U [(1  ) (S
2R)]eTg
(17)
Linea izing a e assuming ha , o each g oup, he di¤e ence be ween ne
endowmen s is small unde he di¤e en ax egimes, and hen sol ing o he
wo exchange a es, yields,

S1N1
1 + R+ 1 
+("1) + (1 1)g
eN
+S
2(18)
S1T1
1 + R+1+ 
1 ("1) + 1g+c(h
1;
H1)
(1  )eT
+S
2(19)
and ecognizing ha a delay in s abiliza ion equi es ha 
S1N< S1T, yields
he ollowing condi ion o delay o be an equilib ium s a egy:
2R > 
(1  )+("1) + (1 1)g
eN
+("1) + 1g+c(h
1;
H1)
(1  )eT
(20)
Again, he deg ee o exchange a e unce ain y plays a key ole in delay.
In he case o low unce ain y, delay is unlikely. In ui i ely, o wa d-looking
N-agen s ealize ha hei ba gaining posi ion de e io a es o e ime in he
absence o s abiliza ion and, he e o e, ha e an incen i e o each an ag eemen .
Wi h high unce ain y abou he pos -s abiliza ion exchange a e, howe e , he
cos s o in‡a ion a e domina ed by he po en ial cos o ag eeing o a de alua ion
which aises impo ed inpu cos s.
How does he sha e o impo ed inpu s in‡uence he likelihood o delay?
P oposi ion 5: As long as S >1, ha is, s abiliza ion in ol es a de alua ion,
a highe impo ed inpu sha e inc eases he likelihood o delay in Pe iod 1 i
("1) + (1 1)g > eN( he cos s o in‡a ion exceed he pos -s abiliza ion ax
bu den o N-agen s).
P oo : Exp ession (20) can be ew i en as:
R > 1 + ("1) + (1 1)geN
eN
+S
2(R+1+eT("2) + 1g+c(h
1;
H1)
(1  )eT
+S
2)
(21)
F om (19), we know ha he exp ession in he cu ly pa en heses mus be
g ea e han 1 + as long as he e is a de alua ion (S >1). A look a he
16
exp ession in he … s se o squa e pa en heses on he igh hand side e eals
ha , as long as ("1) + (1 1)g > eN, his e m is g ea e han 1. This,
in u n, means ha a highe alue o ende s he sa is ac ion o (21) mo e
likely.
Exp essions (18) and (19) ell us ha an inc ease in lowe s he le el o S
a which bo h pa ies a e indi¤e en i p e-s abiliza ion in‡a ion cos s exceed
hei pos -s abiliza ion ax bu den, and educe ha le el in he opposi e case.
In ui i ely, highe impo in ensi y aises he cos o a de alua ion o N-agen s
ela i e o he dis o iona y cos s o in‡a ion, making hem mo e likely o delay
ag eemen .
To see he ole o unce ain y s a kly, conside he case whe e he bu den o
pos -s abiliza ion axes on N-agen s exac ly equals ha o he p e-s abiliza ion
dis o iona y cos s (i.e., eT=("2) + 1g+c(h
1;
H1)). Then, i R= 0,
exp ession (21) ells us ha he e is no delay in he p esence o a de alua ion
as long as ("1) + (1 1)g > eN. I s abiliza ion in ol es a de alua ion, delay
will no happen unless he e is exchange a e unce ain y.
To summa ize he esul s om his sub-sec ion, in ou baseline se -up, a high
le el o unce ain y will lead o delay in Pe iod 1 and ag eemen on a le el o
he exchange a e in Pe iod 2 in he p esence o high in‡a ion i he unce ain y
associa ed wi h he le el o he exchange a e is high enough o ou weigh he
cos s o in‡a ion in Pe iod 1 bu no su¢ cien ly high o ou weigh hose cos s in
Pe iod 2. The abili y o he T-agen s o adap o he in‡a iona y ax unde lies
he dynamics o he model. Inc eased impo ed inpu use lowe s he ceiling
o he exchange a e ha is accep able o he N-agen s, making delay mo e
likely a a gi en le el o unce ain y. In he p esence o a s ep de alua ion, and
ela i ely high in‡a ion cos s, a non-ze o isk p emium is a necessa y condi ion
o a delay.8
As men ioned in ea lie sec ions, IMF p og ams o en accompany mac oeco-
nomic adjus men in de eloping coun ies. Such p og ams, mo eo e , ypically
come wi h an ag eemen in a o o a la ge de alua ion o add ess balance o
paymen conce ns. The nex sec ion explo es hese impo an aspec s.
3.2 Income unce ain y and he possibili y o o eign aid
The las subsec ion ocused on he ole o unce ain y and impo ed inpu s. I
now add wo new elemen s: (1) he possibili y o o eign (say IMF) aid ollowing
a s abiliza ion ag eemen , and (2) an expec ed possible ecession (decline in
endowmen ) in Pe iod 2.
Many coun ies e en ually op o an IMF adjus men p og am once hey a e
a he p ecipice o a majo down u n. Suppose ha he e is a p obabili y o a
symme ical all in endowmen s in Pe iod 2. Tha is, he expec ed endowmen s
8No ice ha , as in Laban and S u zenegge (1994), a social planne who weighs he u ili y
o bo h g oups will ha e no incen i e o pu sue …nancial adap a ion since i simply inc eases
he a e o dis o iona y in‡a ion.
17
in ha pe iod a e gi en by:
e2i=ei+ (1 )eiAei;i=N; T (22)
whe e  < 1and Ai+(1) ep esen s he expec ed endowmen in Pe iod
2. Mo eo e , suppose ha i he e is a ecession (i.e., hings a e bad), and i
s abiliza ion is ag eed upon be ween he wo g oups, hen IMF aid a i es. Pu
di¤e en ly, agen s nego ia e he exchange a e knowing ha IMF aid will a i e
ollowing an ag eemen and condi ional on he occu ence o a ecession. This
aid uni o mly educes he pos -s abiliza ion ax bu den by a ac o 1(whe e
0<  < 1). Thus, i he e is no ecession, (4) will con inue o apply. O he wise,
in Pe iod 2, wi h IMF aid,
g =(eN+eT)
Thus, he expec ed e¤ec i e ax a e mul iple condi ional on ag eemen being
eached in Pe iod 2 is gi en by:
B= [1 (1 )]
No e ha , 0<B<A<1and keep in mind ha 6= 1 ii  < 1. No hing
else changes ela i e o he baseline se -up and he wo se s o agen s nego ia e
o e he le el o he exchange a e ollowing s abiliza ion, and in an en i onmen
o unce ain y ega ding he pos -s abiliza ion alue o he exchange a e. In
he nex sub-sec ion, we will s ipula e ha he IMF dic a es he le el o he
nominal exchange a e which is p e-announced and known o e e yone.
I can be shown o s ill be he case ha in‡a ion inc eases while he p e-
s abiliza ion dis o iona y ax sha e o he N-agen s inc eases o e ime; no hing
changes he e. Mo eo e , P oposi ion 3 om he p e ious subsec ion con inues
o hold wi h he di¤e ence ha he equilib ium posi ion o N-agen s in e ms
o he exchange a e now de e io a es e en mo e om Pe iod 1 o Pe iod 2 (see
he Appendix o he p oo ). As shown below, o he e o be a non-emp y
se o ag eemen s in Pe iod 2 (see P oposi ion 2 in he p e ious subsec ion),
he espec i e exchange a es and he agg ega e condi ion a e now somewha
di¤e en .
All o which leads o ou nex wo p oposi ions.
P oposi ion 6: The possibili y o o eign aid in Pe iod 2 (i.e.  < 1): (i) expands
he possible se o alues o Swhich allow an ag eemen in Pe iod 2, and (ii)
p o ides he necessa y medium h ough which expec a ions o a ecession ( i.e.,
 < 1)allow a simila expansion o he possible se o alues o S,and (iii)
inc eases he likelihood o an ag eemen a a gi en le el o R.
P oo : To p o e ha he possibili y o o eign aid expands oom o ag ee-
men in Pe iod 2, conside ha he modi…ed condi ions o such an ag eemen
co esponding o (10), (11), and (12) a e as ollows. Fo N-agen s:
U [1  (S2N+R)AB]eNg>
U (1  )AeN("2)(1 2)gg
18
)S2N R+ 1 
B
A+("2) + (1 2)g
AeN
(23)
Fo T-agen s,
U [(1  )(S2TR)AB]eTg>U(1  )AeTc(h2;
H2)("2)2g
)S2TR+1+ 
1
B
A("2) + 2g+c(h
2; K)
(1  )AeT
(24)
and he agg ega e condi ion co esponding o (12) becomes:
2R  
(1  )
B
A+("2) + (1 2)g
eN
+("2) + 2g+c(h
2; K)
(1  )eT
(25)
The di¤e ence om he co esponding exp essions (10), (11), and (12) in
he p e ious subsec ion is he p esence o he e m B=A (<1) on he igh
hand sides. The possibili y o IMF aid in he case o an ag eemen aises
he maximum le el o S ha N-agen s a e willing o allow while educing he
minimum le el accep able o T-agen s. This expands he se o possible alues o
he exchange a e o e which an ag eemen can be eached, making s abiliza ion
in Pe iod 2 mo e likely, and p o ing (i).
To es ablish (ii), no ice ha , once he possibili y o o eign aid is emo ed,
i.e., = 1, hen B=A = 1, which yields he same exp ession as he co esponding
exp ession (12) in he p e ious subsec ion whe e o eign aid was absen . Thus,
expec a ion o a ecession does no in i sel in‡uence he likelihood o ag eemen
in Pe iod 2 unless  < 1
Finally, he p esence o he e m B=A (<1) in (25), bu no in (12), p o es
(iii).
In ui i ely, he p obabili y o a ecession, does no by i sel a¤ec he alues
o he exchange a e ha lea e he wo se s o agen s indi¤e en unless o eign
aid is condi ional on he occu ence o ha ecession. This means ha he
wo sides elax hei ba gaining posi ions in Pe iod 2 i and only i o eign aid
(which lowe s hei ax bu dens) is a possibili y.
Do he wo sides eac simila ly in Pe iod 1 by elaxing hei ba gaining
posi ions in esponse o he possibili y o o eign aid condi ional on a ecession
in Pe iod 2?
P oposi ion 7: Gi en ha o eign aid is condi ional on a ecession, i.e., 6= 1
ii  < 1, he possibili y o a ecession in Pe iod 2 (i.e.,  < 1) inc eases he
likelihood, a a gi en le el o R, ha an ag eemen o s abilize will be delayed
in Pe iod 1.
The p oo builds on exp ession (25) and ollows he p oo o P oposi ion 4
om he p e ious subsec ion. Gi en he possibili y o aid, N- and T-agen s,
19

espec i ely, will be indi¤e en i :
U1 
S1N+ReN+U 1 
S1N+RAeN=
U (1  )eN("1)(1 1)gg+U [1  (S
2+R)AB]eNg
U [(1  ) (S1TR)]eTg+U [(1  ) (S1TR)A]eTg=
U(1  )eT("1)1gc(h1;
H1)+U [(1  ) (S
2R)AB]eTg
Linea izing and simpli ying as in he p e ious sec ion, yields he indi¤e -
ence le els co esponding o (18) and (19) and he condi ion unde which delay
occu s.

S1N=1
1 + A R+ 1 [1 + (1 B)]
+("1) + (1 1)g
eN
+S
2(26)
S1T=1
1 + A R+1+[1 + (1 B)]
1 ("1) + 1g+c(h
1; K)
(1  )eT
+S
2
(27)
2R > [1 + (1 B)]
(1  )+("1) + (1 1)g
eN
+("1) + 1g+c(h
1;
H)
(1  )eT
(28)
To es ablish he p oposi ion, compa e he exp essions (20) whe e B= 1 and
(28), ecall ha  < 1) < 1so ha B < 1, which implies ha , condi ional
on a ecession, he igh hand side o (28) is lesse in magni ude. Al e na i ely,
i = 1 so ha = 1, hen he alue o B e u ns o 1 and he likelihood o
delay is unchanged.
In sum, his subsec ion has es ablished ha , while o eign aid condi ional on
a ecession expands he se o exchange a e alues ha allow o an ag eemen
in Pe iod 2, i also inc eases he chances o a delay in s abiliza ion in Pe iod 1.
In ui i ely, he expec a ion ha hings will be bad enough in Pe iod 2 so as o
necessi a e o eign aid inc eases he emp a ion o delay an ag eemen . Once
delayed, howe e , i also makes an ag eemen mo e likely in Pe iod 2 since aid
alle ia es some o he pain o adjus men .
3.3 Exchange Ra e Expec a ions and P e-Announced Aid
Some coun ies ha e in e nal nego ia ions in condi ions whe e i is al eady clea
ha he IMF will ha e o be app oached in he nea u u e o s abiliza ion
unds. Mo eo e , he b oad con ou s o IMF condi ionali y, such as s eps o
change he …scal ou look and o add ess balance o paymen s p oblems wi h he
help o exchange a e changes a e mo e o less known due o pas his o y wi h
he ins i u ion and ongoing nego ia ions.
20
Conside a si ua ion o o eign exchange sca ci y ha o ces he go e nmen
o s ic ly limi o eign cu ency access a he o¢ cial a e. A (possibly il-
legal) pa allel open ma ke exis s whe e expec a ions o jump changes in he
exchange a e can in‡uence i s cu en le el. To keep hings simple, suppose
ha N-agen s a e o ced by egula ions o pu chase o eign cu ency o pay o
impo s om his ma ke while T-agen s can pu chase cu ency a he o¢ cial
a e (which, a S= 1, is lowe han he ma ke a e in pe iod 1). This could be
ei he due o laws ha allow p i ileged access o o eign cu ency o adable
p oduce s –a no uncommon phenomenon –o because he T-agen s can em-
ploy hei o eign cu ency sa ings deposi ed in o eign banks (pa ly h ough
misin oicing). Suppose oo ha i is essen ially ce ain ha , unless he e is s a-
biliza ion in Pe iod 1, he IMF will ha e o be app oached in Pe iod 2, and ha
he IMF will demand a de alua ion o a le el SIMF
2(which could co espond
o he le el ha elimina es he pa allel ma ke p emium) and consequen uni…-
ca ion o he dual cu ency ma ke . Thus, he e is now no ba gaining o e he
exchange a e be ween T- and N-agen s in Pe iod 2, and he e is no unce ain y
abou he u u e le el o he exchange a e.
How will his a¤ec he pa allel ma ke exchange a e and he chances o an
ag eemen in Pe iod 1? We … s need o de e mine wha he ma ke exchange
a e in Pe iod 1 will be gi en expec a ions o de alua ion in Pe iod 2. Some e -
sion o in e es pa i y would be a good candida e bu ha d o jus i y igo ously
in he absence o domes ic and o eign asse s …nancial asse s in ou amewo k.
As an app oxima ion, suppose ha he le el o he exchange a e in he pa allel
ma ke is 1 + x, whe e xis he open ma ke p emium, and 1<1 + x<SIMF
2.
Assume also, again plausibly, ha his expec ed dep ecia ion is a posi i e unc-
ion o he exchange a e ha he IMF is known o p e e o Pe iod 2 i he e
is no s abiliza ion in Pe iod 1, i.e., x=x(SIMF
2),x0>0.
To summa ize, he se -up now is di¤e en om he p e ious wo subsec ions
in wo espec s: (1) he unce ain y abou S2is now gone, and (2) he open
ma ke exchange a e in Pe iod 1, S1, is now a unc ion o he exchange a e
le el ha he IMF is known o ecommend as pa o s abiliza ion in Pe iod 2.
In o de o isola e he e¤ec s o he emo al o exchange a e unce ain y, I se
and  om sec ion 3.2 back o 1 (as in sec ion 3.1).
P oposi ion 8: Gi en ,x > 0, he p esence o a known, condi ionali y- equi ed
le el o he exchange a e SIMF
2 ha would emo e exchange a e unce ain y
in Pe iod 2 has ambiguous e¤ec s on he likelihood o delay.
P oo : Since he e is now no nego ia ing o e , o unce ain y abou , S2, he only
in e es ing change is ha P oposi ion 4 om Sec ion 3.1 needs o be al e ed.9
The exp essions ha de…ne indi¤e ence o each se o agen s now become:
9Tha is, P oposi ions (1)-(3) a e una¤ec ed and con inue o hold.
21
(1 + )U1 
S1N+ReN=
U1 (1 + x(SIMF
2))eN("1)(1 1)g+U 1 SIMF
2eN
(29)
(1 + )U [(1  )(S1TR)]eTg=
U(1  )eT("1)1gc(h
1;
H1)+U (1  )SIMF
2eT(30)
Again, linea izing and sol ing o he wo exchange a es ( 
S1Nand S1T)
yields,

S1N=R+1
1 + (1 + x)
+("1) + (1 1)g
eN
+SIMF
2(31)
S1T=R+1
1 + 1 + 
1 ("1) + 1g+c(h
1; K)
eN
+SIMF
2(32)
and ecognizing ha a delay in s abiliza ion equi es ha 
S1N< S1T, yields
he ollowing condi ion:
2R > 1
1 + xSIMF
2
(1  )+("1) + (1 1)g
eN
+("1) + 1g+c(h
1; K)
(1  )eT
(33)
As long as he discoun a e and he p emium xa e posi i e, he igh
hand side o he exp ession abo e may be lesse , equal o, o g ea e han ha
om (20).
Compa ing wi h he co esponding exp essions in Sec ion 3.1, i.e., (18) and
(19), e eals ha he emo al o unce ain y abou he exchange a e in Pe iod
2 has ambiguous e¤ec s on 
S1Nwhile inc easing S1T.
The o e all impac on he likelihood o delay is ambiguous. Fi s ecall ha
he unce ain y abou he le el o he exchange a e should s abiliza ion occu in
Pe iod 1 s ill exis s. In ui i ely, now he e a e wo opposing e¤ec s o emo ing
unce ain y abou S2and, he e o e, causing mo emen s oday in he pa allel
ma ke . The p esence o an open ma ke p emium in Pe iod 1 on he exchange
a e makes i less cos ly o N-agen s o ag ee o a dep ecia ion in Pe iod 1.
I makes less sense o desi e a delay i pa o he cos s o an ag eemen ha e
al eady been bo ne in Pe iod 1. On he o he hand, he emo al o exchange
a e unce ain y in Pe iod 2 and he esul ing highe open ma ke exchange a e
p e alen in Pe iod 1 educes he po en ial cos o delaying.10 The ne impac on
10 Tha is, ins umen unce ain y now only has an e¤ec on he cu en pe iod u ili y bu
no on he discoun ed second pe iod u ili y.
22
he likelihood o delay will be he sum o hese wo o ces. Pe haps su p isingly,
he highe he exchange a e known o be equi ed unde he condi ionali ies
o he p og am, he less likely a delay. In ui i ely, i he pa allel ma ke a e
a ailable o N-agen s al eady p ices in he possibili y o an ag eemen in Pe iod
2, he cos o ag eeing o s abiliza ion oday declines.
4 Concluding Rema ks
La ge sus ained …scal and balance o paymen s de…ci s culmina ing in home-
g own o IMF-in‡uenced s abiliza ion p og ams a e a common occu ence in
de eloping coun ies. O en hese p og ams seem o come a he las possible
momen when he economy is on he e ge o o in he middle o a c isis. This
applies o …scal c ises as well as exchange a e c ises. In e es ingly enough, his
happens e en hough almos e e yone can see he c ises app oaching, and some
o he pa icipan s opposed o changes in exchange a e egimes oppose e o ms
in spi e o he ecogni ion on hei pa ha delay could make hei posi ion
e en wo se when he much needed e o ms a e e en ually ca ied ou . Why
hen he push o delay?
This pape builds on exis ing li e a u e by ocusing on he poli ical econ-
omy o changes in exchange a e le els and he esul ing asymme ic e¤ec s on
di¤e en g oups o p oduce s. P oduce s o adables and non- adables ha e
a ying p e e ences o e his key p ice in he mac oeconomy. I show ha ,
exchange a e unce ain y, e en when di¤e en isk-a e se g oups ha e pe ec
o esigh o he wise, can domina e he cos s o in‡a ion o an ex en whe e i
causes delays in s abiliza ion. I in‡a iona y cos s a e high, ea o he de alu-
a ion ha ollows s abiliza ion inc eases he likelihood o delay and sh inks he
space o ag eemen in he p esence o a high sha e o impo ed inpu s. IMF
aid ha educes he bu den o adjus men makes i mo e appealing o delay an
ag eemen o he da e when he aid is dispe sed e en i ha happens ollowing
an economic down u n. An IMF condi ionali y ha uni…es he o¢ cial and
pa allel exchange a e ma ke s a a widely expec ed le el and akes unce ain y
ou o he pic u e o he u u e exchange a e, on he o he hand, could ha e
ambiguous e¤ec s on he likelihood o delay. Tha he de alua ion is al eady
pa ly p iced in o he cu en exchange a e inc eases he cos o delay o he
g oup ha does no ha e p i ileged access a he o¢ cial exchange a e. Tha
he discoun ed cos o exchange a e unce ain y in he u u e is ze o encou ages
delay. Thus, he highe he p emium al eady inco po a ed in o oday’s pa allel
ma ke exchange a e, he smalle he likelihood o delay. In sum, he o m
ha o eign aid akes ma e s.
I would be use ul o end by no ing some limi a ions o he analysis he e.
An impo an impac o ex e nal suppo is ha i can in and o i sel inc ease
he c edibili y o a p og am. I only ake his in o accoun in a limi ed sense
in sec ion 3.3 by pos ula ing ha he IMF p og am emo es exchange a e
unce ain y in he second pe iod . Also, IMF p og ams could cause mo al haza d
p oblems (D ehe and Vaubel (2004)) ha a e beyond he scope o his pape .
23