COUNTRY RISK AND THE MUNDELL-FLEMING MODEL APPLIED TO THE ARGENTINE EXPERIENCE

Transcription

1 Journal of Appled Economcs, Vol. V, No. 2 (Nov 2002), COUNT ISK AND THE MUNDELL-FLEMING MODEL 327 COUNT ISK AND THE MUNDELL-FLEMING MODEL APPLIED TO THE AGENTINE EXPEIENCE JAVIE OTIZ AND CALOS ODÍGUEZ * Unversdad del CEMA In ths paper we propose a modfcaton of the tradtonal Mundell-Flemng model. The extended model ntroduces the mplcatons of ncludng the fscal defct and nternatonal reserves as determnants of the level of country rs. Ths slght modfcaton of the tradtonal paradgm leads to radcal changes n the effects that fscal and monetary polces have n economes wth hgh captal moblty under an extreme verson of a fxed exchange rate regme (Currency Board). We use the proposed model to evaluate some of the economc polces mplemented between December 1999 and March 2001 by the frst economc team under the Presdency of Fernando De la úa n Argentna. Addtonally, we suggest that some of the man results obtaned from the model are applcable to other emergng economes. JEL classfcaton code: F41 Key words: open economy macroeconomcs, country rs for Argentna, reserves, convertblty I. Introducton In a seres of artcles n the early sxtes, two brllant economsts,. Mundell and M. Flemng, 1 developed the bass for what has snce become nown among economsts as the Mundell-Flemng model, that enables analyss * E-Mals: jortzarg@aol.com.ar and nfo@cema.edu.ar, respectvely. We are grateful for the excellent techncal assstance provded by Gullermo Bermúdez, Marano Fernández and Pablo Palla. 1 Although ths lne of study was publshed n several boos and artcles, the most mportant are Mundell (1963) and Flemng (1962).

2 328 JOUNAL OF APPLIED ECONOMICS of dfferent macroeconomc aspects n economes open to nternatonal trade. The model began by recognzng that nternatonal marets n the sxtes were gradually becomng more ntegrated because of the new developments n technology that were tang place. In addton, t recognzed that a growng number of countres were n the process of dsmantlng the rgorous exchange controls that had become wdespread after the Great Depresson. On the bass of the foregong, these authors assumed that the world was made up of economes producng dfferentated goods (that s to say, goods that were not perfect substtutes) but whch were subject to a hgh level of moblty n ther captal flows. These authors proceeded to analyze the effects derved from alternatve economc polces, under the assumpton that the velocty of captal movements between countres was far hgher than that of ther trade flows for goods. In subsequent years, because of the characterstcs mentoned, ths model gradually became the professon s preferred tool for the analyss of a wde range of matters related to the macroeconomcs of open economes. Ths artcle s ntended as a contrbuton to the extensve lterature that followed the acceptance of the Mundell-Flemng model as the central paradgm for the analyss of open economes. It does so by ntroducng the possblty that one of the components determnng domestc nterest rates, the factor nown as country rs, s related to both the fscal stuaton of the country and ts level of nternatonal reserves. In ths regard, ths artcle represents the combnaton of two types of models, those n the Mundell-Flemng tradton and those taen from the more modern but equally extensve lterature on the topc of country rs. 2 However, even ths smple change to the Mundell- Flemng approach allows us to reach very relevant and novel conclusons on the possble effects of alternatve monetary and fscal polces on small, open economes wth fxed exchange rate arrangement whle facng a hgh nternatonal captal moblty. Whereas n the orgnal Mundell-Flemng model monetary polcy does not affect domestc nterest rate or the level of economc actvty, n our model 2 Eaton, Gersovtz, and Stgltz (1986) provde a comprehensve summary of such lterature.

3 COUNT ISK AND THE MUNDELL-FLEMING MODEL 329 an expansve monetary polcy leads to a contracton n the output level by nducng an ncrease n country rs and therefore n the domestc nterest rate. In addton, n our model the tradtonally postve mpact on economc actvty from an expansve fscal polcy would be weaened by the contractonary effect from the mpact of a deterorated fscal stuaton on country rs and consequently on domestc nterest rates. In recent years t has become a habt among a wdenng crcle of economsts, partcpants n the fnancal sector and even poltcans and journalsts nvolved n mang or analyzng proposals for the emergng country, to resort mechancally to the transmsson mechansm mentoned prevously as the only sgnfcant mechansm operatng n such economes. The transmsson mechansm mples that a country faced by a recesson (boom) must proceed to mplement contractve (expansve) fscal polces n order to reduce (ncrease) country rs and thus nduce an ncrease (declne) n the output level. One notable example of the above proposton was the sharp ncrease n taxaton (the so-called Tax eform) ntroduced by the new Argentne economc team n December 1999 wth the support of most of Argentna s nternatonal credtors (ncludng the IMF and almost all the prvate credt agences). The assumpton behnd these restrctve fscal polces was that they would lead to a recovery n the level of economc actvty, severely affected snce md by negatve external shocs. The only grounds for such a strategy were that a stronger fscal poston would lead to a sharp drop n country rs, whch would result n an equvalent reducton n nterest rates and therefore a sgnfcant ncrease n domestc spendng. However, and n spte of recent pror experence n Argentna, the rse n taxes turned out to be hghly recessve and brought the mld economc recovery that had begun n md-1999 to a complete standstll. The model presented here, by ncludng n the analyss the tradtonally contractve effect that hgher taxes have on domestc demand, provdes a possble explanaton for such epsode. In the rest of the paper we wll proceed as follows. Secton II presents our model. The man mplcatons for economc polcy are dscussed n the three

4 330 JOUNAL OF APPLIED ECONOMICS followng sectons. In Secton VI we expand on the basc model wth the am of ntroducng a compettve domestc fnancal sector, n an attempt to capture the prncpal mechansms that come nto play when the Central Ban decdes to modfy the mnmum reserve requrements on deposts, the only monetary polcy nstrument that remans under the monetary system we analyze (Currency Board). We suggest that n cases such as that of Argentna durng the last decade, expansve monetary polces very probably ended up by ncreasng nterest rates, further accentuatng the declne n levels of actvty. Secton VII contans our conclusons. II. The Model Followng the orgnal Mundell-Flemng model, our model assumes the exstence of an aggregate demand that depends negatvely on nterest rate (nomnal and real, as an expected zero nflaton s assumed) and postvely on government defct D. It s assumed that the so-called IS curve behaves as follows: = (, D), where < 0 and D > 0 (1) In an economy wth a fxed exchange rate, nternatonal captal moblty mples that the domestc nterest rate should be equal to the nternatonal nterest rate * plus a country-rs premum (where t s assumed that the expected exchange rate devaluaton s zero): = * + (2) n turn, n lne wth the extensve lterature on country rs and the currently accepted fact that t grows as the publc sector defct rses and falls when nternatonal reserves rse: = (, D), where < 0 and D > 0 (3)

5 COUNT ISK AND THE MUNDELL-FLEMING MODEL 331 under a Currency Board scheme (such as that adopted by Argentna at the begnnng of 1991 wth the launch of a monetary reform nown as the Convertblty Plan), the monetary base s equal to total nternatonal reserves. The base s a fracton e of total money supply M, because of the exstence of a fractonary cash reserve system: 3 = e M (4) n turn, money supply equals demand through the change n the stoc of reserves, so that equlbrum n the money maret (the LM curve) determnes the level of reserves as a functon of the arguments n the money demand and the fractonal reserve requrements: = el(, ), where L < 0 and L > 0 (5) therefore, gven the values of the three exogenous varables D, * and e, the equlbrum values of our three endogenous varables, and are determned exactly by the followng three equatons: (IS) = (, D) (6) (FF) = * + (, D) (7) (LM) = el(, ) (8). The Effects of Monetary Polces Under a Convertblty System such as the one we are analyzng, the 3 For the sae of smplcty we have assumed that throughout the analyss there s a full concdence between the legal cash reserve requrements and actual amounts, so that at no tme do bans mantan techncal reserves n excess of those requred by the monetary authorty. Lewse, there are never any shortfalls n cash reserve requrement complance.

6 332 JOUNAL OF APPLIED ECONOMICS monetary authorty s prohbted by law from ssung domestc credt. Therefore, the only monetary polcy possble conssts of modfyng the cash requrements, whch leads to: / L e = 0 1 e ( L + L ) > (9) / e = 1 L e ( L + L ) < 0 / e = 1 L e ( L + L ) > 0 where the notaton x z means x / z. Under the assumpton that the system s stable (see Appendx), then 1 e(l + L ) would be postve and consequently expansve monetary polces would lower reserves, ncrease domestc rates of nterest and lower the level of economc actvty. Therefore, whle n Mundell (1963) an expansve monetary polcy under a fxed exchange rate and perfect captal moblty lowers nternatonal reserves n the same magntude, wthout affectng ether the nterest rate or the level of ncome, here the drop n reserves has a contractve effect on the output level snce country rs levels rse. The drop n reserves would tae place due to: 1. A drect effect, whch mpacts by causng reserves to declne because of equaton 8 (the LM curve). 2. Because of equaton 7, the reducton n would ncrease the country-rs premum rasng domestc nterest rates (the FF curve). In turn, hgher nterest rates mply a lower level for because of equaton 6 (the IS curve). 3. A reducton n the money demand because of the ncrease nduced n and the reducton n. Ths drop n desred monetary balances would also reduce because of equaton 8. Therefore, the tradtonal effect ponted out by Mundell-Flemng (pont 1)

7 COUNT ISK AND THE MUNDELL-FLEMING MODEL 333 has now been strengthened by two other ndrect effects, both arsng from the hgher rs levels nduced by the expansonary monetary polcy. Graphcally, a reducton n cash reserves shfts the FF and LM curves upward and to the left (see Fgure 1), 4 movng the equlbrum from E 0 to E 1. The new equlbrum s characterzed by a hgher nterest rate and consequently a lower level of output. Fgure 1. The Effects of an Expansve Monetary Polcy LM 1 LM 0 E 1 *+ 1 *+ 0 E o FF 1 FF 0 IS d 1 The slope of the IS curve has a negatve sgn: = < 0. The LM curve, for ts d IS d L part, has a postve slope: = > 0. In addton, t can be verfed that the FF d L curve has a slope equal to zero. LM

8 334 JOUNAL OF APPLIED ECONOMICS IV. The Effects of Fscal Polces Other man concluson of the tradtonal Mundell-Flemng model s that wth a fxed exchange rate arrangement and perfect captal moblty, expansve fscal polces have fully postve mpact on economc actvty. However, n our model the effect of fscal polcy on the level of economc actvty s undetermned. Whle n the Mundell-Flemng model a worsenng of the budget defct does not alter the nterest rate, n our model t would deterorate country rs, thus ncreasng domestc nterest rates. The fnal effect on the level of output wll depend on the relatve magntude of both effects (.e. the expansve effect from the ncreased level of absorpton and the contracton nduced by a hgher nterest rate). We would therefore have: / D = D 1 L e e ( L + L D + ) D (10) / D = e L 1 D + L e e ( L + L D + L ) D e / D = D 1 + e L e ( L + L D ) Even under the assumpton that the expresson 1 e(l + L ) were to be postve, the three sgns reman ndetermnate. These results arse because an ncrease n D would generate: 1. The shft of the IS to the rght, therefore nducng an ncrease n both and. 2. However, at the same tme, an upward shft of the FF curve nduces an ncrease n and a contracton n. 5 5 It could be argued that the expansve effects of a fscal stmulus are mmedate, whle the effects of a government defct on the stoc of debt and the rs premum tae tme to

9 COUNT ISK AND THE MUNDELL-FLEMING MODEL 335 If and can n prncple change n any drecton, the same wll happen wth the money demand and consequently wth reserves. V. The Argentne Experence As mentoned above, because of external factors that ncluded a steady deteroraton n terms of trade snce 1996, and a sharp rse n country rs for emergng natons snce 1998 because of the ussan crss (see Fgures 2 and 3), the Argentne economy began to declne. Ths drop, smlar to that observed n many other emergng countres, contnued untl md-1999, when a slght recovery began (see Fgures 4 and 5). Fgure 2. Terms of Trade (1991 = 100) 125,00 120,00 123,1 121,1 115,00 110,00 105,00 100,00 103,2 95,00 90,00 I I I I I I I I I I I I-02 Source: Argentne Mnstry of Economy (2001) 1800 Fgure 3. Emergng Maret s (EMBI +) (Bass Ponts)

10 336 JOUNAL OF APPLIED ECONOMICS 1800 Fgure 3. Emergng Maret s (EMBI +) (bass ponts) (EMBI +) (Bass Ponts) Jan-94 Apr-94 Jul-94 Oct-94 Jan-95 Apr-95 Jul-95 Oct-95 Jan-96 Apr-96 Jul-96 Oct-96 Jan-97 Apr-97 Jul-97 Oct-97 Jan-98 Apr-98 Jul-98 Oct-98 Jan-99 Apr-99 Jul-99 Oct-99 Jan-00 Apr-00 Jul-00 Oct-00 Jan-01 Apr-01 Jul-01 Oct-01 Source: J. P. Morgan (2001) Source: J.P.Morgan (2001) At the end of 1999 the recently elected government of Argentna, concerned about strengthenng the recovery n the real economy, mplemented a Tax eform amed to mprove fscal solvency, whch ncluded greater progresson n ncome tax scales, a rse n excse tax rates and a seres of new taxes. Ths tax ncrease was decded on the assumpton that wde-rangng tax ncreases, develop, as they occur only wth the accumulaton of successve defcts. However, ths would be true only n a world of statc expectatons and not n a world of ratonal expectatons, where nvestors would perceve an ncrease n the defct as a sgn that the debt would ncrease, actng therefore on the bass of such expectatons. Ths case was analyzed n detal by odríguez (1979).

11 COUNT ISK AND THE MUNDELL-FLEMING MODEL 337 Fgure 4. Seasonally Adjusted GDP (n mllons of pesos at 1993 prces) I I I I I I I I I Source: Argentne Source: Argentne Mnstry Mnstry of Economy of Economy (2001) (2001) by assurng the solvency of the state, would help to strengthen the ncpent economc recovery that had begun n August of that year. 6 The eform, t was argued, would help to speed up economc recovery, snce by reducng the fscal defct t would cause a drop n the country-rs level, therefore 6 Two experences of the recent past weghed partcularly heavly on Argentne economc analysts, as n both cases contractve fscal polces were seen to have strengthened the process of economc recovery. The frst of these experences was the tax ncrease that was decded on followng the so-called Tequla Effect at the begnnng of Ths ncluded a 3-percentage pont ncrease n the value-added tax rate, hgher labor taxes and reductons n publc sector wages. The second experence too place n md-1996 when, followng the replacement of Economy Mnster Cavallo by oque Fernández, a he too place n taxes on ncome, fuels and certan mport categores, whch helped to overcome wthout major dffculty the frst changes n economc leadershp snce the launch of the Convertblty Plan.

12 338 JOUNAL OF APPLIED ECONOMICS Fgure 5. Seasonally Adjusted Gross Domestc Fxed Investment (n mllons of pesos at 1993 prces) I I I I I I I I I Source: Argentne Mnstry of Economy (2001) lowerng nterest rates and encouragng a strong ncrease n the level of prvate spendng. However, the result was qute the opposte. The country rs rate declned from 610 bass ponts n the second wee of December 1999 to 522 n the same wee of March Baced by the mentoned reducton n rs, the nterest rate declned. However, the economc recovery that was under way came to a halt, n spte of mproved external ndcators, begnnng a lengthy recesson. Ths too place because the tax ncrease, combned wth a government dscourse that dramatzed the prevously exstng fscal stuaton, led to a sharp reducton n perceved dsposable ncome by consumers. The consequent declne n consumpton was aggravated by a sgnfcant drop n the margnal effcency of captal, provong a strong downward shft n the IS. Clearly, nvestor pessmsm prevaled over the reducton n nterest rates, provong a strong

13 COUNT ISK AND THE MUNDELL-FLEMING MODEL 339 contracton n nvestment, whch also contrbuted to wpng out the ncpent economc recovery (see Fgure 6). In short, our approach dffers sgnfcantly n ts prncpal predctons not only from the tradtonal Mundell-Flemng approach but also from the consensus that currently exsts n fnancal crcles. As shown n Table 1, whereas the tradtonal Mundell-Flemng model predcts that a restrctve fscal polcy would fully mpact by lowerng the output level, the consensus vew predcts that t has a reactvatng effect on output, because t would generate lower publc sector nsolvency rss and therefore lower nterest rates. The model shown here s of a more general nature, as t allows both results to tae place (output can ether rse or fall), dependng on the relatve strength of the shft of the FF and IS curves. In addton, ths new approach renforces the negatve mpact of expansve monetary polcy, as n addton to generatng the famlar effect of reducng Fgure 6. The Effects of a Contractve Fscal Polcy E 0 *+ 0 FF o *+ 1 E 1 FF 1 IS 1 IS 0 1 0

14 340 JOUNAL OF APPLIED ECONOMICS Table 1. Alternatve Effects of Monetary and Fscal Polces under Fxed Exchange ate Arrangements Alternatve Vews Polces esults Mundell-Flemng Model D e "Consensus Vew" D e??? odríguez & Ortz D +/- +/- +/- e (whch n ths new approach s renforced by two new effects) t ncreases domestc nterest rates and reduces the level of ncome. VI. Includng a Fnancal System nto the Model In the mddle of 2000 the Argentne economc authortes found themselves trapped n a lengthy recesson of ther own mang. In these crcumstances, they argued that f the Central Ban were to mplement a cut n reserve requrements on deposts, the consequent reducton n nterest rates would stmulate a recovery n economc actvty levels. The mechansm for transmsson would be the reducton n costs that lower reserves would represent for the banng system, whch would n turn lead to a fall n nterest rates on ban loans to companes (the lendng rate). Wth the am of ncludng n the analyss the prncpal effects of a change n mnmum reserves requrements on deposts, we wll proceed to wor wth a smple extenson of the model shown n the prevous sectons. We wll do so followng Calvo and Fernández (1983), assumng the exstence of a compettve domestc fnancal system. We wll suppose that there are a large number of

15 COUNT ISK AND THE MUNDELL-FLEMING MODEL 341 dentcal bans operatng on the bass of constant returns to scale, n a compettve envronment. Under these condtons, we can expect bans to obtan profts equal to zero. We assume for the sae of smplcty that the operatng costs of the bans are also equal to zero. If borrowers (companes) are unable to gan access to the nternatonal credt maret, then we can wrte: d S = l (1 - e)s (11) where d s the nterest rate pad to depostors and l s the lendng rate. S n turn represents total deposts. It mmedately follows that: ( 1 e ) d = l (12) so that, wth ths modfcaton, our system of equatons s: (IS) = ( l, D) (13) (FF) d = * + (, D) (14) (LM) = el ( d, ) (15) d = l (1 e) (16) Ths system determnes the equlbrum values of the four endogenous varables:,, d, l ; that s to say, ncome, reserves and the two nterest rates. Now, we can try to analyze the effects of monetary polces. In ths context, a reducton n e, such as that proposed by the Argentne government, would result n:

16 342 JOUNAL OF APPLIED ECONOMICS 1. A drect effect (by means of equaton 15) on, nducng a lower level of reserves and consequently hgher rates of nterest pad to depostors (to compensate for the hgher rs). 2. Ths rse n depost nterest rates would lead to a drop n the money demand and therefore n. 3. Hgher depost rates would have an ambguous effect on, snce even though the lendng rate would have to ncrease to reflect the ncrease n the depost rate (n response to the ncreased rs provoed by lower rates for ) ths effect s offset by the lower opportunty cost mplct n lower cash reserves. Therefore, lendng rates (those charged by bans to ther customers) may be hgher, equal to or lower than before. 7 And as ths s the rate that comes nto play n the demand for goods, then can fall, rse or reman the same. It s because of the latter that t s not possble to predct the drecton n whch the demand for real balances and thus wll move. Under these scenaro therefore, expansve monetary (fnancal) polces would affect not only but also and domestc nterest rates. In order to dentfy the relatve magntudes of these effects, t s necessary to estmate the above-ndcated group of equatons for each of the countres under analyss. However, t s plausble to speculate that n the cases of countres such as Argentna, where country rs responds abruptly to changes of nternatonal reserves and the fscal defct, the contractonary mpacts may predomnate over the mproved lqudty (or credt avalablty). If ths were the case, expansve monetary polces would have a restrctve effect on the level of economc actvty, as a reducton n nternatonal reserves would ncrease both passve and lendng nterest rates (.e. that pad to depostors and that collected from ban customers). Ths result s of great sgnfcance as t means that countres wth fxed 7 It was ths latter effect that was nvoed by the Argentne economc authortes when they recommended a reducton n cash reserves.

17 COUNT ISK AND THE MUNDELL-FLEMING MODEL 343 exchange rate arrangements face country-rs levels that are very senstve to varatons n nternatonal reserves. And hence they wll not be able to ncrease the level of economc actvty by expandng ther monetary aggregates. In such cases an expansve monetary polcy would be contractonary, as the ncrease n soveregn rs would predomnate over the effect from greater lqudty generated by a reducton n cash reserves. Ths s a stuaton we presume could exst n many other developng countres. Gven the relevance of ts mplcatons, we consder that t merts a hghly detaled statstcal exploraton n the future. VII. Conclusons In the case of tradtonal models based on fxed exchange rates and free captal moblty, the usual concluson s that monetary polcy has no nfluence on output level, whle the opposte s the case of fscal polcy. Monetary polcy does not affect nterest rates, and therefore t does not generate any crowdng out of prvate spendng, and therefore fscal polcy s fully effectve n modfyng output. In ths paper we present a varant of the Mundell-Flemng model, whch by ntroducng the commonly accepted effects of reserves and the fscal stuaton on the level of soveregn rs, radcally alters these conventonal results. Now, expansve monetary polces are contractve, because by encouragng a lower level of reserves they ncrease country rs, thus stmulatng a rse n domestc nterest rates. In addton, the conventonal results of ths type of model, whch suggest that expansve monetary polces result n a loss of reserves, are strengthened under our approach by two new transmsson channels. On the other hand, fscal polces lose part of ther effectveness because hgher fscal defcts now mply hgher rates of nterest, generatng a crowdng out of prvate spendng. Thus t becomes perfectly plausble that a relaxaton of fscal polces could have a contractve effect on the level of output.

18 344 JOUNAL OF APPLIED ECONOMICS In ths paper we use ths new approach to analyze the economc polces adopted between December 1999 and March 2001 by the frst economc team apponted under the admnstraton of Fernando de la úa, when Argentna stll appled a monetary system nown as Convertblty (a Currency Board). Based on our model, we offer a new and plausble vew of the falure of the monetary and fscal polces appled durng that perod. More sgnfcantly, we establsh how small changes to the conventonal Mundell-Flemng model can lead to mportant dfferences n the results to be expected from fscal and monetary polces. The model we present also dffers very consderably n ts prncpal predctons from those derved from the consensus vew of fnancal crcles and economc analysts. In addton, we present an extenson of the model that ncludes the exstence of a compettve fnancal system. The ncluson of such a system ntroduces a new transmsson mechansm, whch by means of a reducton n the system s reserve requrements, all thngs beng equal, nduces a correspondng reducton n lendng rates. We suggest however that n Argentna durng the perod expansve monetary polces, even tang nto account ths last transmsson channel, would have lead n any case to lower levels of reserves, hgher nterest rates and a lower level of output, gven the heghtened senstvty of country rs to nternatonal reserves and fscal solvency. Fnally, we suggest that ths s the result to be expected n many other developng countres that have adopted a fxed exchange rate system n a context of hgh nternatonal captal moblty such as currently prevals. In all of them, a reducton n reserve requrements, by nducng a loss of reserves, would generate a rse n depost rates through the percepton of ncreased levels of rs, whch would preval over the reducton of spreads from the lower level of bloced deposts. Ths mples that expansve monetary polces would contnue to rase nterest rates (ncludng lendng rates) and would cause a fall n actvty levels.

19 COUNT ISK AND THE MUNDELL-FLEMING MODEL 345 Appendx We wll assume that the three endogenous varables, and move over tme as follows: d dt [ (, D ] = α ) wth α > 0 α( 0 ) = 0 d dt d dt = β = γ [ + (,D ) ] [ e L(, ) ] wth β > 0 β( 0 ) = 0 wth γ > 0 γ ( 0 ) = 0 where α, β and γ are functons wth frst postve dervatves and a steady-state value equal to zero. If we tae the frst-order term n a Taylor expanson, we can lnearze these equatons close to ther equlbrum values, resultng n: d dt = α ( 0) + α ( o) d = β ( o) + β ( 0) dt d = γ e L( o) + γ e L( o) ( o) dt where the zero sub-ndexes ndcate ntal equlbrum values. At the steady state, we can compute:

20 346 JOUNAL OF APPLIED ECONOMICS < 0 γ β α ( ) [ ] < + = = L L e L e L e A Det γ β α γ γ γ β β α α Therefore, the characterstc equaton would be as follows: where λ are the roots of the system and a, b and c are coeffcents. So that ths system s stable (wth roots wth real parts wth a negatve sgn) t s necessary that the trace and the determnant of the above matrx system should have a negatve sgn. Gven that α, β and γ are all postve, the necessary and suffcent condton for the local stablty of the system s: 0 ) ( 1 > + L L e Trace A = A = & & & & & & & & & = = c b a L e L e λ λ λ λ γ γ γ β λ β α λ α

21 COUNT ISK AND THE MUNDELL-FLEMING MODEL 347 Graphcally: Fgure 7. Stablty of the Ortz & odríguez Model & = 0 & = 0 eferences Calvo, G., and Fernández,. (1983), Compettve Bans and the Inflaton Tax, Economc Letters 12: Eaton, J., Gersovtz, M., and J. Stgltz (1986), The Pure Theory of Country s, European Economc evew 30: Flemng, M. (1962), Domestc Fnancal Polces under Fxed and under Floatng Exchange ates, Internatonal Monetary Fund Staff Papers 9 (3): Mundell,. (1963), Captal Moblty and Stablzaton under Fxed and Flexble Exchange ates, Canadan Journal of Economcs and Poltcal Scence 29: odríguez, C. (1979), Short and Long un Effects of Monetary and Fscal

22 348 JOUNAL OF APPLIED ECONOMICS Polces under Flexble Exchange ates and Perfect Captal Moblty, Amercan Economc evew 69: