Capital Inflows, Fiscal Discretion, and Exchange Rate Policy * David Cook Hong Kong University of Science and Technology

Transcription

1 Capital Inflows, Fiscal Discetion, and Exchange Rate Policy David Cook Hong Kong Univesity of Science and Technology Michael B. Deveeux Univesity of Bitish Columbia CEPR Revised, July 25 Keywods: Capital Inflows, Fiscal Policy, Exchange Rate Regime, Pe-commitment JEL Classification: F32, F4, F4 Abstact Many wites have agued fo the benefits of a cedible fixed exchange ate (a had peg) as a commitment device in an open economy. But histoically, fixed exchange ates have often been associated with lage cuent account deficits and episodes of ove-boowing. This pape develops a model of capital inflows that ae linked to the exchange ate egime because of endogenous fiscal policy. The key message of the pape is that a had peg is undesiable in the absence of commitment in fiscal policy. In face of a cedible fixed exchange ate, the fiscal authoity subsidizes capital inflows. The economy will engage in inefficiently high intenational boowing, and in welfae tems may end up wose off than unde capital maket autaky. To eliminate the incentive to subsidize boowing, the monetay authoity must follow a flexible exchange ate ule in which capital inflows lead to exchange ate appeciation. If fiscal policy must be financed by money ceation athe than diect taxation, then a fixed exchange ate ule may cause both ove-boowing and a subsequent exchange ate cisis. Deveeux thanks SSRHC, the Bank of Canada, and the Royal Bank of Canada fo financial suppot. We ae gateful to an anonymous efeee and the edito fo comments, as well as semina paticipants at the Univesity of Hong Kong, the Univesity of Hawaii, the Univesity of Bitish Columbia, the Univesity of Wisconsin, the Univesity of Mayland, Gezensee, and the Canadian Macoeconomics Study Goup. In addition, we would like to acknowledge helpful comments fom Patick Fancois, Pete Ieland, Hais Dellas, and Angelo Melino.

2 Many wites have agued fo the benefits of cedible fixed exchange ates (had pegs) in poviding a nominal ancho fo monetay policy in an open economy. `Had pegs ae seen as instilling a commitment in monetay policy (e.g. Calvo 2 fo emeging makets and Giavazzi and Pagano 989 fo Euope), paticulaly in counties with a ecod of high inflation. But the histoical ecod of had pegs is mixed. Many counties that have fixed thei exchange ates have expeienced vey lage cuent account deficits, and in some cases, subsequent exchange ate cises. In fact, most of the significant capital inflows into emeging maket counties have taken place unde fixed exchange ates. This is tue of both the Latin Ameican and East Asian economies. Figue shows that fo Agentina, Bazil, and Mexico, episodes of exchange ate stability have typically coincided with lage and sustained cuent account deficits. This pape exploes the linkage between the fixed exchange ate egimes and capital inflows in envionment of endogenous fiscal policy. The essential message of the pape is that the incentive fo govenment intefeence in capital makets (though implicit o explicit boowing guaantees o subsidies) will geneally depend on the exchange ate policy. If thee is cedible commitment in both monetay and fiscal policy, then the exchange ate policy is ielevant. But, absent commitment in fiscal policy, a cedible fixed exchange ate egime (a had peg ) encouages the fiscal authoity to subsidize intenational boowing, even in a small economy whee the cost of boowing is fixed by wold maket foces. In ode to avoid this excessive intenational boowing, the cental bank must follow a flexible exchange ate ule wheeby the cuency appeciates duing episodes of capital inflows. The link between fixed exchange ates and capital inflows has been noted in many pevious contexts. See Edwads (2) fo discussion of the exchange ate based stabilization liteatue. Fo discussion of oveboowing, see Mckinnon and Pill (998), Bunside, Eichenbaum, and Rebelo (2a) among othes. We discuss this efeences moe fully below. 2

3 An implication of ou agument is that it is impotant to combine the commitment to a had peg with limitations on fiscal policy. But this is fo quite diffeent easons than those emphasized in standad fist-geneation models of cuency cises. We need both a had peg and constaints on fiscal policy not because the pesence of govenment budget deficits theatens the sustainability of the peg, but athe because it is the had peg itself which gives ise to bad fiscal policy. Altenatively, if constaints on fiscal policy cannot be put in place, we ae best off without the had peg. The agument is developed a simple inte-tempoal open economy model. A fiscal authoity can tax o subsidize foeign boowing. We can think of these eithe as explicit subsidies to domestic boowes, o implicit guaantees that ultimately epesent a liability of govenment. By assumption, monetay policy is detemined by an independent cental bank that can pe-commit to an exchange ate ule. If the fiscal authoity can also pe-commit, it will neve wish to intefee with intenational capital flows, no matte what the exchange ate policy followed by the cental bank. But without pe-commitment, thee is an incentive to encouage boowing by subsidies o guaantees. The undelying distotion pushing the fiscal authoity to intevene in capital makets is the unemployment ate. Wages ae set in advance by monopoly supplies of labo so that employment is inefficiently low. By subsidizing capital inflow afte wages ae set, the fiscal authoity attempts to amelioate this distotion. With a fixed exchange ate ule, the fiscal authoity will always wish to subsidize foeign boowing. As a esult the county will have an excessive level of extenal debt. In fact, in ou model a fixed exchange ate leads the economy to end up in a wose welfae position than if it was unde complete capital maket autaky. 3

4 While a had peg should be avoided in an envionment whee fiscal authoities lack cedibility, thee is an exchange ate ule which exactly nullifies the lack of fiscal pe-commitment. If this ule is followed by a cental bank, it offsets the incentive fo the fiscal authoity to intevene in capital makets, allowing the economy to fully exploit the benefit fom intenational capital flows. The key ingedient behind this ule is that the exchange ate should appeciate in esponse to capital inflows. We extend the model to show that the combination of fixed exchange ates, absence of pe-commitment in fiscal policy, and a equiement that the govenment needs to finance its activity with money ceation can geneate a joint capital-maketcuency cisis. The key featue of this extended model is that this joint cisis is caused solely by the initial imposition of a fixed exchange ate egime. Without an initial peg, thee is no ove-boowing, and no subsequent cuency collapse. In fact, without the peg, the exchange ate can be kept stable thoughout. The pape is elated to a numbe of liteatues on exchange ates and fiscal policy. Many wites have noted that exchange ate pegs used to eliminate high inflations may give ise to ovevaluation and excessive cuent account deficits (Edwads 2). Othes have emphasized the tendency fo fixed exchange ates to give ise to excessive, unhedged foeign cuency boowing (McKinnon and Pill, 2, Bunside et al. 2a). This in tun may give ise to `Fea of Floating, as in Calvo and Reinhat (22). Fom a diffeent pespective, thee is a debate about whethe a cedible fixed exchange ate can itself instil fiscal discipline (see Tonell and Velasco 996). We discuss the elationship of the pape to these questions in moe detail below. The pape is oganized as follows. The next section develops the basic model. Section 2 shows the esults with pe-commitment. Section 3 deals with the no 4

5 commitment case. Section 4 extends the analysis to show that excessive capital inflows may geneate subsequent cuency cises. Section 5 discusses the elationship of the pape to the liteatue on exchange ate egimes and fiscal policy. Some conclusions ae then offeed.. The Model Take a simple two-peiod model of an open economy 2. In each peiod thee is a single good poduced by competitive fims. Consume-wokes supply labo to fims, and set wages monopolistically in advance, as in Obstfeld and Rogoff (2). Fims Each fim has a poduction function given by Y = F( H ), whee FH ( ) is inceasing, concave, and satisfies Inada conditions. We define the labo composite as t t t λ () λ H = H i di, so that the fim uses diffeentiated labo in poduction, and the elasticity of substitution between types of labo is λ. Fo the fim, the pofit maximizing ule is: (.) λ Ht () i SF t '( Ht) = Wt( i) Ht so that the wage elasticity of demand fo each type of labo is equal to λ. Consume-Wokes Consumes have utility ove peiods and given by: (.2) t i = β t t t= E U E ( U( C ()) i V( H ())) i 2 The esults do not depend on the two peiod stuctue. The model could be e-famed in a dynamic setting to show the same aguments, but at the cost of analytical tactability. 5

6 whee Ct () i ( Ht () i ) is consumption (employment) of household i in peiod t. The functions U (.) and V (.) ae inceasing, concave, and continuously diffeentiable. The household can lend o boow (if negative) amount B( i) at ate, eceives wage income whee the nominal wage is W, and eceives pofits Π ( i) fom owneship of the fim. Household budget constaints ae then (.3) PC() i + PB () i = W() i H() i +Π PT t t (.4) PC () i = ( + ) PB () i + W() i H() i +Π PT whee T t is a tax fom the govenment, and P t is the pice of the consume good. Assume that puchasing powe paity (PPP) holds, and the foeign pice level is nomalized at unity so that Pt = S. t The fist ode conditions fo each consume-woke i, taking the demand fo labo schedule (.) into account, ae: (.5) U'( C( i)) = β EU '( C( i))( + ) W t (.6) Et U'( Ct) Ht( i) λv '( Ht( i)) Ht( i) =, t =,. St whee λ λ =. Equation (.5) is the Eule equation fo optimal consumption λ gowth, while (.6) implicitly detemines the pe-set wage fo each peiod 3. Now assume all wokes ae alike, so that Wi () = W, Hi () = H. 3 Equation (.6) emphasizes a cental featue of the model; aveage eal wages will be set so that the maginal poduct of labo exceeds the maginal ate of substitution between consumption and leisue. As in Bao and Godon (983), thee is a eal distotion that leads a benevolent policy-make to wish to aise the level of employment. Although the fomal model taces this eal distotion to monopoly powe in the labo maket, we could altenatively think of it as aising fom monopoly contol of a nontaded input into poduction. 6

7 Fiscal Authoity The fiscal authoity may intefee in capital makets by subsidizing intenational boowing. If the wold eal inteest ate is subsidy to foeign boowing as, we define the effective, so that epesents the contol vaiable of the fiscal authoity. The fiscal authoity then taxes households to finance any boowing subsidies 4. Then the fiscal authoity s budget constaint will be (.8) T T ( ) B = whee B < if the pivate secto ae boowes. The left hand side is the pesent value of tax evenue, while the ight hand side is the pesent value of expenditue on capital subsidies. Combining (8) and the consume budget constaint implies that the economy s oveall budget constaint must be: (.9) C = F( H ) + ( + )( F( H ) C ). Cental Bank We have not allowed fo an explicit ole fo money in the model. But this doesn t matte, as we can instead simply assume that the monetay authoity diectly chooses the exchange ate ule, as a policy tool. We assume that the monetay authoity faces no cedibility poblem, and can commit to a peiod exchange ate ule 5. The time exchange ate is assumed to follow a policy ule given by (.) S = S() 4 While we descibe these as tax financed subsidies, thee is nothing in the model that ules out an altenative intepetation as lending guaantees to domestic boowes that ae implicitly financed by govenment. This less tanspaent view of the ole of govenment bias in capital makets may be moe in accod with the actual expeience of emeging maket economies. Since taxes ae non-distotionay, thee is an equivalence between the two intepetations. 5 The consequences of the lack of cedibility in monetay policy have been extensively investigated befoe, both fo open and closed economies (Bao and Godon, 983, Lane 998). 7

8 Hence, the monetay authoity follows a ule elating the exchange ate to the domestic eal inteest ate. Since in ou model, capital flows ae diven by the ate of etun, if this is an inceasing function, it implies that when the economy expeiences capital outflows (inflows), the exchange ate depeciates (appeciates). In the case of a had peg, by definition the ule would imply that S'( ) =, and in addition, that S = S. In this section, the peiod exchange ate plays no ole in the analysis, since thee is no estiction on diect taxation, and afte the fiscal authoity has acted, thee is complete monetay neutality. In section 4 below, we investigate the implications of having an uppe bound on tax collection. Equilibium Fo a given value of, an equilibium is vey easy to define. In an equilibium; a) consume wokes maximize utility subject to thei budget constaints and thei individual labo demand constaint, b) fims maximize pofits, and c) the labo maket cleas. As a benchmak, assume that β ( + ) =. This ensues that in an equilibium without capital contols, the cuent account is zeo, and consumption and employment ae equal in both peiods. Detemination of boowing subsidies Assume that the fiscal authoity is benevolent, choosing a subsidy to maximize utility of the epesentative household. But the authoity may o may not be able to commit. With commitment, the subsidy is chosen befoe the wage fo peiod is set. Then the authoity takes account of how its cuent choice of affects both consumption and employment in both peiods and. But without commitment, it takes the nominal wage as given in peiod, and chooses a subsidy to maximize the household s utility. 8

9 2. Fiscal policy with commitment Unde full commitment, the fiscal authoity chooses to maximize utility taking into account the consumes Eule equation, labo maket cleaing in both peiods, and the economy s extenal budget constaint. Table illustates these constaints. The thee conditions in Table implicitly detemine the values of C, H, H. Table : Constaints facing the fiscal authoity: commitment Eule equation U'( C ) = β EU'( F( H ) + ( + )( F( H ) C ))( + ) Labo maket cleaing F'( H) U'( C) = λv '( H), F'( H) U '( C) = λv '( H) Implicitly, we may wite these as functions of the authoity s decision, so that C = C (), H = H (), H = H (), whee these functions satisfy the popety that C '( ) <, H '( ) >, and H '( ) < 6. A ise in the eal inteest ate educes peiod consumption, inceases peiod consumption, which, though adjustment in the eal wage, inceases peiod employment and educes peiod employment. Moeove, unde the assumptions made above, we must have C ( ) = F( H ( )) = F( H ( )). Note that unde commitment, the fiscal authoity s poblem is independent of the exchange ate ule. When the authoity can pe-commit, it takes account of how the nominal wage set by wokes will adjust to exchange ate. The fiscal authoity s poblem is then defined by the poblem max (P) + β β UC ( ( )) VH ( ( )) UC ( ( )) VH ( ( )) subject to C ( ) = F( H ( )) + ( + )( F( H ( )) C ( )). 6 Unde the assumptions on the cuvatue of pefeences and technology, an equilibium exists. 9

10 Poposition Unde commitment, the optimal subsidy on foeign boowing is zeo. Poof. Using Table, the fist ode condition fo poblem P is (2.) U '( C) βu '( C)( + ) C'( ) + U '( C ) F'( H ) V '( H ) H '( ) + U'( C ) F'( H ) V '( H ) H '( ) = [ ] β [ ] The expession on the top line captues the effect on utility of the evision of intetempoal consumption geneated by a subsidy. A ise in will educe consumption in peiod, so that C '( ) <. If > ( < ) it must be that U C > U C +, ( U'( C ) < βu'( C )( + ) ), since the fist ode condition '( ) β '( )( ) fo the consume implies that U'( C) = βu'( C)( + ). Then, if this was the full effect on utility, it is clea that the fiscal authoities would set =, because they wish to educe (incease) when > (<). As egading the two expessions on the second line, at the initial point = and with β ( + ) =, it is the case that β H '( ) = H '( ) 7. Then, since when =, we have C C, H H = =, it follows that these two expessions also sum to zeo when capital contols ae zeo. The explanation fo poposition is quite staightfowad. With monopoly wage setting, thee is a labo maket distotion in each peiod, due to the fact that the maginal poduct of labo exceeds the maginal disutility of woking. That is F'( H) > V '( H) / U'( C). Hence, employment is inefficiently low in each peiod. The fiscal authoity would like to incease employment. But unde pe-commitment, a boowing subsidy will have opposite effects on employment in each peiod. Fo instance, a subsidy will educe consumption in peiod, and incease consumption in

11 peiod. This will lead to a ise in employment in peiod, as wage settes will educe thei desied wage, given that the disutility of woking in peiod has fallen. But by the same token, employment in peiod will fall. The ise in employment in peiod inceases household utility (the second expession in the above fist ode condition), since employment is inefficiently low to begin with. But the fall in employment in peiod will educe household utility, since the distotion of inefficiently low employment in peiod is exacebated. Unde the symmety assumptions we have made, these two effects exactly offset each othe. Hence thee is no benefit to the fiscal authoity, acting unde commitment, fom subsidizing boowing. As a esult, unde commitment, the fiscal authoity has no incentive to intefee with intenational capital flows Fiscal policy without commitment Without pe-commitment, the fiscal authoity chooses the boowing subsidy afte the peiod wage has been set. Table 2 illustates the set of conditions facing the authoity. The diffeence fom the case with commitment is that employment in time peiod is not diectly dependent on the domestic inteest ate, given that the nominal wage is pedetemined. Nevetheless, employment depends on indiectly though the exchange ate ule S( ). If a ise in the domestic eal inteest ate geneates a nominal exchange ate depeciation, this inceases employment in peiod, fo a given nominal wage. Hence the exchange ate ule followed by the monetay authoity becomes a citical deteminant of the decision to subsidize boowing. 7 This can be shown by diffeentiating the economy s budget constaint (.9) fo changes in aound =, taking into account the two labo maket cleaing conditions in the second panel of Table. 8 When the initial cuent account is unbalanced, so that β ( + ), the fiscal authoity may have an incentive to levy capital taxes (o gant subsidies) even in the pecommitment case. Nevetheless,

12 Table 2: Constaints facing the fiscal authoity: no commitment Eule equation U'( C ) = β EU'( F( H ) + ( + )( F( H ) C ))( + ) Labo maket Wage detemination SF H U'( C ) F'( H ) = λv '( H ) '( ) = W W E U '( C) λv '( H) H = S Exchange ate ule S = S() Fom Table 2, we may define the implicit functions undelying the authoity s decision without commitment as C = C(,()), s H = H(()) s and H = H (,()) s, S() whee s () =. It is staightfowad to show that these functions satisfy the W conditions C <, Cs >, H'( s) >, H <, and H s <. A peiod eal depeciation (defined as a fall in the peiod eal wage) inceases employment. The incease in income diectly inceases peiod consumption. But, though the income effects on labo supply, this aises the peiod eal wage, and educes employment in time peiod. Without commitment the fiscal authoity s poblem is defined as follows max UC ( (,)) s VH ( ()) s + βuc ( (,)) s βvh ( (,)) s (P2) the incentives to deviate fom the = case ae always geate without commitment (as shown below). 2

13 subject to C (, s) = F( H (, s)) + ( + )( F( H ( s)) C (, s)), and s () S () W =. The choice of the fiscal authoity depends citically upon the fom of the exchange ate ule followed by the cental bank. We fist establish the following esult Poposition 2. Unde a fixed exchange ate the fiscal authoity subsidizes intenational boowing, so that <. Poof. A fixed exchange ate ule implies that s'( ) =. Then the fist ode condition fo the fiscal authoity unde poblem P2 is given by (3.) β β[ ] U'( C) U'( C)( + ) C + U'( C) F'( H) V '( H) H = The fist expession is the same as that in poposition, captuing the impact on utility of distoting the inte-tempoal consumption choice. The second expession captues the effect on utility fom the evision of peiod employment following the boowing subsidy. This expession is always negative, since fom the wokes optimal wage setting decision, we have U'( C) F'( H) = λv '( H), with λ >, and H <. If the sum of the two expessions is to equal zeo, it must be that the fist expession is positive, which can only be the case if <. Hence, unde a fixed exchange ate, the fiscal authoity subsidizes intenational boowing. Consumes boow at below the wold eal ate of inteest, and the economy uns a cuent account deficit. The intuition behind this poposition is quite easy to see. Imagine fo a moment that =, so that the economy faces the wold ate of inteest ate, and thee is zeo intenational boowing. Then, since the 3

14 fist tem in the above fist-ode condition is zeo at =, the impact on utility of a small fall in the eal inteest ate is [ ] U '( C ) F'( H ) V '( H ) H. This is unambiguously positive. In contast to the case with commitment, a small boowing subsidy inceases utility. Without commitment, the fiscal authoity takes the peiod eal wage as given, unde fixed exchange ates, so it cannot affect the fist peiod level of employment. But if it educes the domestic eal inteest ate slightly, this means that it will aise cuent consumption, and educe futue consumption. At the initial point, whee =, the welfae cost of this inte-tempoal consumption distotion is negligible. But the eduction in peiod consumption leads to an incease in peiod employment. Because peiod employment is inefficiently low to begin with, this has fist ode positive welfae effects. Thus the fiscal authoity has an incentive to subsidize boowing, essentially because it can incease peiod employment without peceiving a concomitant eduction in peiod employment. In an equilibium, it will subsidize boowing so much that the welfae benefits fom an incease in peiod employment ae offset by the welfae costs of the inte-tempoal consumption distotion. Theefoe, unde fixed exchange ates, the equilibium configuation of capital contols implies that <. Welfae is lowe in an equilibium without commitment,. Given that wage settes will adjust peiod wages to account fo the highe peiod consumption, the net impact of the excess boowing on the labo maket distotion is zeo, because in equilibium, the benefit of highe employment in peiod is offset by lowe employment in peiod. But the inte-tempoal consumption distotion geneated by the capital subsidy leaves the home consume wose off. In fact because we have 4

15 stated fom a baseline case without net boowing, the home consume is wose off in this equilibium than she would be unde complete capital maket autaky. A fixed exchange ate ule theefoe gives ise to `ove-boowing in this model, in the absence of fiscal pe-commitment. Excessive boowing takes place simply because unde fixed exchange ates, optimal govenment policy educes the pivate cost of funds below the wold oppotunity cost 9. It is woth noting that the impotant popety of the exchange ate ule is the S elasticity of the facto maket eal exchange ate W to changes in the domestic inteest ate. Given that we have not defined a sepaate demand fo money in the model so fa, the level of the nominal exchange ate is ielevant. We now focus on the case whee the monetay authoity follows a moe geneal exchange ate ule. Poposition 3. When the exchange ate ule satisfies s() = s (), such that (3.2) s [ '( ) '( ) '( )] β U C F H V H H '( ) = > whee = [ ] + β [ ] U'( C) F'( H) V '( H) H' U '( C) F'( H) V '( H) Hs, then the optimal boowing subsidy is zeo, and =. Poof. The fist ode condition fo the fiscal authoity unde a geneal exchange ate ule is given by 9 Although ou model abstacts fom investment, exactly the same channel would be pesent in an economy with endogenous capital accumulation and investment. A subsidy on cuent capital inflows would incease domestic investment and the futue capital stock, geneating a highe level out futue output, in this way acting so as to minimize the undelying distotion faced by the fiscal authoity. 5

16 (3.3) U'( C) βu'( C)( + ) ( C + Css'( )) + ' + U '( C) F'( H) V '( H) Hs'( ) + + β U'( C ) F'( H ) V '( H ) ( H + H s'( )) = [ ] s This extends the condition of Poposition 2 in two ways. Fist, an inteest ate change now affects peiod employment, if it affects the nominal exchange ate. A fall in the inteest ate educes fist peiod employment if the exchange ate ule is inceasing in the inteest ate. This follows, because with a fixed money wage, a nominal exchange ate appeciation aises the eal wage and educes employment. Second howeve, an inteest ate change inceases fist peiod consumption and second peiod employment though sepaate channels, since the ise in fist peiod income (following the incease in output) leads to a ise in consumption in peiod (and peiod ), and a fall in employment in peiod. Note howeve, that if the exchange ate ule satisfies the condition given above, then the second two expessions in the above fist ode condition sum to zeo. But then the fist expession must be always zeo (unde the condition that ( C + Css'( )) < ), since we have shown above that this expession is always positive (negative) if < ( > ). In othe wods, the fiscal authoity would neve want to intevene in capital makets only to distot the inte-tempoal consumption decision alone. The ole of the exchange ate ule (3.2) is to eliminate the incentive fo the fiscal authoity to intevene in intenational capital makets so as to incease peiod output. If the authoity educes the domestic eal inteest ate below the wold ate, this geneates a eal exchange ate appeciation (i.e. a eal wage incease), which educes peiod employment and output. This offsets the incentive to subsidize 6

17 boowing. In fact, the ule (3.2) estoes the full commitment case, eliminating the existence of govenment subsidized ove-boowing. This exchange ate ule is esonant of the ecent debate on the ole of the exchange ate in the pocess of capital inflows in emeging maket economies. Many wites have agued that it is unwise fo policy-makes to keep the exchange ate fixed in the pesence of lage capital inflows, because this uns the isk of high domestic inflation and oveheating of the domestic economy, which may pecipitate subsequent cises. It is best to allow the exchange ate to appeciate as a natual bake on the domestic economy. Ou esults ae in accod with this ecommendation. In ou model, because the economy expeiences capital inflows (outflows) when < ( > ), the policy pesciption given by the s() ule implies that the cuency should appeciate in esponse to capital inflows. But the logic behind the need fo appeciation is quite diffeent fom the standad agument. Hee the agument is stategic. An appeciating exchange ate emoves the incentive fo the fiscal authoity to stimulate a consumption boom though boowing subsidies. It limits capital inflows, not due to the usual effects of eal exchange ate appeciation, but by focing fiscal policy to intenalise the tue welfae costs of intevention in capital makets. It is clea that the ule given by (3.2) is also an optimal exchange ate ule that would be chosen by a cental bank, unde commitment, wee it to design a ule to maximize the utility of the epesentative individual. If the cental bank could design a ule ex ante, it would face the same objective as would the fiscal authoity, if the fiscal authoity wee acting with commitment. In that case the cental bank would choose the ule to sustain the commitment outcome. If the exchange ate ule is not designed optimally howeve, thee is no guaantee that it will do bette than a fixed exchange ate. While the paticula ule 7

18 (3.2) ensues fully open capital makets in this economy, if the esponse of the exchange ate to the eal inteest ate is too high, then it may actually outweigh the incentives to subsidize boowing. Then the fiscal authoity will begin to subsidize lending (i.e. > ). This will happen if the impact of a ise in on the exchange ate is so geat that the ise in peiod employment esulting fom an exchange ate depeciation is moe impotant in utility tems than the fall in peiod employment. In this case, an equilibium without commitment involves ovelending. 4. A Cuency Cisis Extension In the model so fa a fixed exchange ate policy gives ise to ove-boowing, but does not geneate an exchange ate cisis. The fixed exchange ate egime is not itself theatened by the pesence of ove-boowing. This is because of a simplifying assumption that we ve made; namely that the fiscal authoity has unesticted access to lump-sum taxes. Clealy this is an unealistic assumption the public finances of many emeging maket economies ae esticted in ways that substantially limit the collection of diect taxes by any means. Moeove, as agued by Kugman (998) and Cosetti et al. (999), the scale of implicit lending guaantees that lie behind oveboowing episodes may fa outstip any possibility of tax finance. Bunside, Eichenbaum and Rebelo (2b) ague that the lage public secto liabilities geneated by capital inflows into East Asia inevitably equied monetay financing. We can extend the model to put a limit on evenues that can be aised by diect taxation, and assume that expenditues in excess of this will foce the govenment to sell bonds to the cental bank, as in Kugman (979). Assume now that the cental bank can make only a tempoay commitment to a fixed exchange ate. By this we mean that it can commit to a ule in which S is fixed, so that S'( ) =. But it cannot necessaily ensue that S = S. Since the fixed exchange ate again ceates an 8

19 incentive fo the govenment to subsidize domestic boowes, pat o possibly all of these subsidies must be financed by money ceation, theatening the sustainability of the fixed exchange ate in the second peiod. The cental bank s commitment ability is only tempoay, since ultimately the govenment budget constaint takes pecedence ove monetay policy. In ode to pusue the money-financing implications of the exchange ate egime we need to intoduce a demand fo money. This is done by extending the utility function to encompass a utility benefit of eal balances. The details of the extended model ae vey standad, and elegated to the appendix. Thee, it is shown that even in the extended model, Popositions and 2 apply exactly as befoe. This means that the model is entiely ecusive. The eal side of the economy; i.e. the vaiables C,, C and B, ae detemined as in section 3. The level of the nominal exchange ate depends on the extent of money issue, which is dependent on the limits on diect taxation. The intuitive eason that Popositions -2 emain unchanged is that a) the economy s inte-tempoal budget constaint is still given by (.9), and b) fom the pespective of the fiscal authoity, both money ceation and diect taxation ae lump-sum taxes, and hence do not affect thei incentives to subsidize boowing, given the exchange ate policy followed by the cental bank. Nevetheless, given an uppe limit on diect taxation, the outcome of the fiscal policy game will have implications fo the size of money ceation, and theefoe fo the level of the exchange ate in peiod. Given the equilibium configuation of C,, C and B, fom section 3, we may descibe the nominal side of the economy by the following thee conditions While this is specific set of assumptions ove the elative pedominance of the cental bank and fiscal authoity, it accods quite well with the oiginal Kugman (979) model of speculative attacks and the lage liteatue which followed it. 9

20 M S (4.) = L( C, C, ) S S M (4.2) L ( C ) S = (4.3) M M M M ( ) B. + Tp + = S ( + ) S + Equation (4.), we defines the implicit demand fo money in peiod, whee M is the nominal money supply in peiod. The function L is inceasing in C, deceasing S in C, and deceasing in S. Likewise, (4.2) defines the implicit peiod money demand schedule. Finally, (4.3) descibes the consolidated public secto budget constaint. This says that total subsidies to households (the ight hand side) must be financed with money ceation and diect taxes, whee T p denotes the pesent value of diect taxes. Fo given values of S and S, equations (4.)-(4.3) detemine the thee vaiables M, M, and T p. If thee is no constaint on the value of diect taxes, (4.)-(4.3) ae always consistent with a fixed exchange ate; i.e. with S = S. But if we assume an uppe bound on feasible tax evenue, so that Tp Tp, then it may not be possible to maintain ensue that S = S, because the money supply has to adjust to ensue that (4.3) is satisfied. To show this, substitute (4.) and (4.2) into (4.3), giving the condition: (4.4) S S M ( ) B L C C T L C. (,, )( ) + p + ( ) = S ( + ) S S + 2

21 This equation epesents the govenment budget constaint, combined with money maket equilibium in both peiods. Is a fixed exchange ate sustainable ove both peiods and? Define T ˆp as the pesent value of diect tax evenue that satisfies (4.4) given an exchange ate S, such that S '( ) =, and S = S (i.e. an exchange ate ule that is independent of, and the nominal level equal ove both peiods). Unde full fiscal pe-commitmentt ˆp is Tˆ C p M L ( F( H ), F( H ),) L ( F( H )) = +. S C C C ( + ) + whee a C supescipt denotes the equilibium descibed by poposition. In the absence of pe-commitment, we have (whee NC denotes the equilibium without pecommitment) Tˆ NC p L ( F( H ), F( H ),) = + + NC NC NC ( ) B M L( F( H )) + S ( + ) +. Unde a vey weak additional assumption, we have T ˆ Poposition 4. C p T ˆ NC p. p P p If ˆ C ˆ NC T T T, then an exchange ate peg is sustainable unde full pecommitment, but is not sustainable in the absence of fiscal pe-commitment. Poof: IfT ˆ C p T, then lump sum taxes (o tansfes) may be used without limitation P to balance the inte-tempoal govenment budget constaint without equiing money NC C NC C ( ) B L L L L The condition is that + >, whee the notation is selfexplanatoy. This assumption says that the pesent value of the seignoiage evenue eaned by the + ( + ) + incease in demand fo money in the fist peiod aising fom the incease in boowing, net of the fall in seignioage evenue coming fom the decease in demand fo money in the second peiod (given the lowe second peiod consumption in the no-commitment case), cannot be so geat as to offset the incease in diect financing equiements coming fom the boowing subsidies. 2

22 finance. Then the money supply may adjust so as to keep the exchange ate fixed in the second peiod. But if T ˆ NC p > T, then the diect tax evenue equied to finance p capital subsidies exceeds the uppe limit on the pesent value of taxation. In that case, the cental bank is foced to use money ceation to finance the govenment spending equiement in excess of tax evenue, and this is inconsistent with a fixed second peiod exchange ate. Hence, if the govenment can pe-commit, thee is no excess extenal boowing, and the fixed exchange ate is sustainable. But if it cannot pecommit, the economy does engage in excessive boowing, and the fixed exchange ate egime must collapse in the second peiod. Thus we have an outcome whee capital subsidies ae tied to fixed exchange ates, the economy ove-boows, and the fixed exchange ate egime collapses at the same time as the economy s cuent account switches fom deficit to suplus. Hence, we have the chaacteistics of a joint exchange ate and capital maket cisis. Moeove, the key featue of this outcome is that the `cisis is geneated solely by the fixed exchange ate ule. Without an exchange ate peg, thee is no ove-boowing. In fact, without an exchange ate peg, the exchange ate can emain stable - thee is no exchange ate cisis at all! But an explicit policy of pegging the exchange ate geneates a fiscal policy that encouages ove-boowing. The financing of this oveboowing then equies an abandonment of the fixed exchange ate. Paadoxically, it is the imposition of a fixed exchange ate ule itself that geneates the ove-boowing, which then leads to the collapse of the fixed exchange ate. Thus, in a novel sense, the fixed exchange ate policy sows the seeds of its own demise 2. 2 It should be noted that ou cuency cisis hee is based on a `fist-geneation undepinning, focusing only on the effects of fiscal deficits. Thee ae many aspects of the expeience of cuency cises that ae not captued by ou model. Fo instance, the poblems of financial fagility (see Chang and 22

23 5. Discussion These esults ae elated to a lage liteatue that daws a link between pegged exchange ates, fiscal imbalances, and excessive cuent account deficits. One taditional agument is that pegged exchange ates foste excessive eal appeciation and tade deficits in face of domestic inflation. Many Latin Ameican counties used an exchange ate peg as pat of a disinflation stategy. The agument fo exchange ate pegs was based on the belief that they offeed moe monetay cedibility than managed floating exchange ate egimes. A simila agument fo the cedibility of exchange ate pegs in the case of Euopean economies was made in Giavazzi and Pagano (989). In the Latin Ameican expeiences, howeve, inetia in domestic inflation gadually led to an ovevalued eal exchange ate. Combined with lax fiscal policy, this led to a lage and unsustainable cuent account deficits, eventually pecipitating a cisis (see Edwads 2). To explain the Asian cisis of , whee neithe ovevaluation o fiscal imbalances played a lage ole, wites began to ague that the mix of fixed exchange ates and implicit govenment guaantees on pivate boowing geneated a `moal hazad, leading to the gowth of un-hedged dolla denominated debt (see McKinnon and Pill, (2), Bunside et al (2b), Kugman (998), Cosetti et al (999)). In this view, when pivate boowes in emeging maket economies believe that the public secto will absob any losses, they effectively take the exchange ate peg as pemanent, and fail to hedge against possible depeciation. This leads to inefficiently lage boowing and a subsequent cisis. Velasco 2), o cuency mismatch (Aghion et al 2) ae clealy impotant featues of a complete 23

24 A elated liteatue views the causation unning in the othe diection fom balance sheet positions to exchange ate pegs. Calvo and Reinhat (22) document the eluctance of many emeging maket counties to allow exchange ate flexibility, thus exhibiting `Fea of Floating. They and othes (Aghion et al. 2, Chamon et al. 2) ague that, in the pesence of weak domestic financial secto and lage foeign cuency denominated liabilities, exchange ate depeciation may play a pevese ole, eliminating the standad agument fo flexible exchange ates as a stabilization mechanism. Finally, Chang and Velasco (25) develop a model whee the linkage between pegged exchange ates and foeign cuency boowing both ways. In thei model, the expectation of a fixed exchange ate can give ise to lage foeign cuency liabilities, which then lead monetay authoities to maintain a pegged exchange ate and thus validate expectations 3. Ou pape contasts with this liteatue in a numbe of ways. Fist, since we deal with an envionment whee PPP holds continually, so thee cannot be any eal exchange ate ovevaluation. While ou extended model of section 4 does offe a diect causal link fom an exchange ate peg to a futue exchange ate cisis, this occus not because of the diect impacts of fiscal deficits on the cuent account, but due to the effect of capital subsidies on pivate boowing. As in the moal hazad liteatue, ou model also implies that a combination of fixed exchange ates and govenment subsidies to pivate boowing leads to an excessive cuent account deficit and may lead to cuency cises. Howeve, the cuency denomination of boowing is ielevant in ou model 4, and in contast to the pevious liteatue, in ou theoy of cuency and financial cises in emeging makets. 3 Thei model also allows fo anothe equilibium wheeby if agents anticipate flexible exchange ates, they insue themselves against exchange ate volatility though capital makets, and the monetay authoities then choose to float the exchange ate. 4 This is because a) we have continual PPP, so that neithe monetay o fiscal policy can affect the eal exchange ate, b) monetay policy is based on a ule, athe than discetion, and c), we do not 24

25 pape the subsidies epesent an ex-post optimal fiscal esponse to an existing distotion. Finally, unlike Chang and Velasco (25), we do not endogenize the exchange ate egime choice itself. Rathe ou aim is to highlight a key link between the exchange ate egime and the incentive stuctue facing fiscal policy. We have focused on a paticula stategic envionment wheeby the monetay authoity can fully commit to an exchange ate peg, but the fiscal authoity cannot commit to an optimal ule. How ealistic is this assumption? In fact, a taditional agument in favou of exchange ate pegs was that they enhanced not just monetay cedibility but also impoved fiscal discipline (see Tonell and Velasco 996). The agument is that substantial fiscal imbalances must lead to losses of foeign exchange eseves and/o eventual monetization of debt, which ae inconsistent with the long un maintenance of the peg 5. But the evidence on fiscal discipline and exchange ate pegs is quite mixed. In emeging maket economies Hamann (2) finds no link between fixed exchange ates and eductions in fiscal imbalances. Likewise thee is little evidence that the ERM led to impoved fiscal discipline in Euopean economies (Wyploz, 997). Indeed, Tonell and Velasco (996) ague that theoetically, a flexible exchange ate may povide moe fiscal discipline. Hence, the assumption of full commitment in monetay policy, but lack of commitment in fiscal policy, may epesent quite a ealistic pespective. One could also ague that the widely obseved gowth of cental bank independence, combined with gowing fiscal imbalances undewites this assumption. Moe geneally, ou pape elates to the substantial liteatue on monetayfiscal inteactions and the debate on ules vesus discetion. Dixit and Lambetini (2) develop a model in which monetay commitment is negated by fiscal incopoate financial fagility o balance sheet effects at the copoate level, which would intoduce a 25

26 discetion. While in a vey diffeent famewok, ou esults ague that monetay commitment (o at least commitment to an exchange ate peg) is undesiable in face of fiscal discetion. 6. Conclusions Thee have been many papes witten on cuency and capital maket cises in emeging maket counties. The pevious section showed that elements of ou analysis have been stessed in many othe papes; e.g. the subsidization of capital inflows, ove-boowing, the ole of fixed exchange ates, and the fiscal foundations of cuency cises. The key diffeentiating featue of ou pape is to tie all these pieces togethe in a way that eveals a simple message a fixed exchange ate can cause excessive boowing and subsequent cuency cises. This occus because fixed exchange ates encouage bad fiscal policy choices. So the essential conclusion of the pape is simila that of othe commentatos a had peg should be accompanied by coodination between monetay and fiscal authoities. On the othe hand, ou analysis also suggests that an appopiate exchange ate ule, allowing capital inflows to be matched by appeciation, can in itself impove the pefomance of fiscal policy. Of couse, an immediate implication of the pape is that a had peg is not necessaily optimal policy in itself. But as discussed in Section 5, thee ae many easons (not explicitly incopoated in the model) fo emeging maket counties eluctance to allow exchange ate flexibility. Nevetheless, even allowing fo these, it is impotant to undestand the boade stategic context in which both monetay and fiscal policy ae chosen. sepaate `cedit channel fo the exchange ate. 5 Canzonei et. al. ague that fiscal discipline is also a equiement of a viable cuency union. 26

27 Appendix To obtain equations (4.)-(4.3), assume that the home esident has the utility function in each peiod given by M t UC ( t( i)) + G( ) VH ( t( i)), t=,, P t whee G(.) is inceasing, concave, and continuously diffeentiable, and M t epesents nominal money balances. Household budget constaints ae now amended to encompass money holdings: P C () i + M () i M () i + PB () i = W () i H () i +Π PT (A.) (A.2) PC () i + M() i = ( + ) PB () i + W() i H() i +Π + M() i PT Now households ae taxed both diectly though lump sum taxes T and T and indiectly though cental bank money issue. Initial money holdings ae M () i. The optimality conditions fo consume i ae as in (.5) and (.6), with the addition of the conditions detemining money demand in peiods and. These ae descibed as () (A.3) '( M i ) '( ( )) P G = U C i P P( + ) M() i (A.4) G'( ) = U'( C ( i)) P Conditions (A.3) and (A.4) implicitly give the demand fo money schedules of household i. Again, we assume that all households ae alike, and dop the i notation heeafte. The fiscal authoity eceives income in the fist peiod fom the cental bank s money ceation; M M, and fom savings of consumes PB (which may be negative). With this, it puchases foeign bonds. In the second peiod it eceives 27

28 income fom its holdings of foeign bonds, money ceation, and with this finances payments to households. Thus, the two budget constaints ae: P B = M M + PB + PT g PB ( + ) = M M + ( + ) PB g+ PT. Again, given PPP, we have P = S, P = S. Putting these togethe, we have (A.5) M M M M T ( ) B. + T + + = S ( + ) S ( + ) + Thus, total subsidies to households (the ight hand side) must be financed with money ceation and diect taxes. Assume that the fiscal authoity is concened with household utility net of the utility of eal balances. Now combining (A.), (A.2), and (A.5), it is staightfowad to see that the economy s inte-tempoal budget constaint is still given by (.9). This has an immediate implication; Popositions and 2 still hold as befoe. The eason that Poposition holds is immediate. If the fiscal authoity acts with commitment, it ignoes the pice level implications of its subsidy choice, because fom a welfae pespective, it is ielevant whethe subsidies ae financed with diect taxes o money issue. Poposition 2 holds because the fiscal authoity s poblem is still defined S exactly as in Table 2, because it cannot affect the eal exchange ate W, and again it is ielevant to its optimal decision whethe its subsidies ae financed by diect taxation o money issue. Fom (A.3), (A.4), and (A.5), we can obtain equations (4.)-(4.3). 28

29 Refeences Aghion, Phillipe, Philippe Bacchetta, and Abijit Banejee (2) Cuency Cises and Monetay Policy in an Economy with Cedit Constaints, Euopean Economic Review, 44, Bao, Robet J. and David B. Godon, (983) A Positive Theoy of Monetay Policy in a Natual Rate Model, 9, Bunside, Caig, Matin Eichenbaum, and Segio Rebelo (2a) Hedging and Financial Fagility in Fixed Exchange Rate Regimes, Euopean Economic Review, 4, 7-2. Bunside, Caig, Matin Eichenbaum, and Segio Rebelo (2b) Pospective Deficits and the Asian Cuency Cisis, Jounal of Political Economy, 9, Calvo, Guillemo A. (2) The Case fo Had Pegs in the Bave New Wold of Global Finance, mimeo, Univesity of Mayland. Calvo, Guillemo A. and Camen Reinhat (22) Fea of Floating, Quately Jounal of Economics. 67, Chang, Robeto, and Andes Velasco (2) A Model of Financial Cises in Emeging Makets, Quately Jounal of Economics, 6, (25) Cuency Mismatches and Monetay Policy: A Tale of Two Equilibia, Jounal of Intenational Economics, fothcoming. Cosetti, Giancalo, Paolo Pesenti, and Nouiel Roubini, (999) What Caused the Asian Cuency Cisis? view, Japan and the Wold Economy, Dixit, Avinash, and Luisa Lambetini (23) Inteaction of Commitment and Discetion in Monetay and Fiscal Policies, Ameican Economic Review, 93,

30 Edwads, Sebastian, (2) Exchange Rate Systems in Emeging Economies, mimeo, UCLA. Lane, Philip R., (997) Inflation in Open Economies, Jounal of Intenational Economics, 42, McKinnon, Ronald and Huw Pill (999) Exchange Rate Regimes fo Emeging Makets: Moal Hazad and Intenational Oveboowing, Oxfod Review of Economic Policy, 5, Giavazzi, Fancesco, and Mato Pagano (989) The Advantage of Tying One s Hands: EMS Discipline and Cental Bank Cedibility, Euopean Economic Review, 32, Hamann, A. Javie (2), Exchange Rate Based Stabilization: A Citical Look at the Stylized Facts, IMF Staff Papes, 48, Hausmann, Ricado, Ugo Panizza, and Enesto Stein, (2) Why do Counties Float the Way the Float?, Jounal of Development Economics, 66, Kugman, Paul (979) A Model of Balance of Payments Cises Jounal of Money Cedit and Banking,, (998) What Happened to Asia, Obstfeld, Mauice, and Kenneth Rogoff (2) New Diections fo Stochastic Open Economy Models Jounal of Intenational Economics, 5, Tonell, Aaon, and Andes Velasco (995) Fixed vesus Flexible Exchange Rates: Which Povides moe Fiscal Discipline?, NBER WP 58. 3

31 3 Agentina CA/GDP Changes of FX Bazil CA/GDP Changes of FX Figue

32 Mexico CA/GDP Changes of FX Figue 32