Trade Patterns and Exchange Rate Regimes: Testing the Asian Currency Basket Using an International Input-Output System

Transcription

1 Trade Paerns and Exchange Rae Regmes: Tesng he Asan Currency Base Usng an Inernaonal Inpu-Oupu Sysem Taash Yano a and Hroyu Kosaa b Absrac The Asan currency crses occurred n he year We focused on exchange raes of he Asan counres/regons among several causes and examned how hey were deermned wh a currency base peg framewor. We also consruced an nernaonal npu-oupu model lned wh macroeconomerc models. Usng he model, we have analyzed rade paern changes n relaon o he yen-dollar rae. Esmaon resuls of he Asan currency bases show ha he wegh of he yen was low, ndcang ha he Asan exchange rae polcy led o he de-faco dollar peg. Smulaon resuls ndcae ha he effecs of he yen-dollar rae on rade paerns depend on ndusral srucures. When a counry/regon s ndusral srucure s smlar o ha of Japan s and he yen s wea, he ncrease of he yen s wegh proves o hold s compeveness. As for he complemenary srucure, he wea yen wll mprove s curren accoun. If he Asan counres/regons had had nernaonal capal nflows conrol, her curren accoun defcs mgh no have spread as wdely. 1 Inroducon In July 1997, he adopon of a free floa exchange rae polcy n Thaland rggered off he Asan currency crses. The conemporary world economy s based on he nerdependence beween dfferen counres/regons. Hence, he currency crses n Asa spread over Russa and Brazl. There are so many sudes on he causes and polcy consderaons of he Asan currency crses. In hs paper, we wll focus on he Asan exchange rae polces and nvesmen booms whch se forh he man causes of he Asan currency crses. The Asan counres/regons moneary auhores had employed he de-faco US dollar peg polcy. The de-faco US dollar peg sysem had hree benefs: a conrol of mpored nflaon, smooh access o he US mare and accepance of foregn nvesmens; however, effeced he Asan real exchange raes o be overvalued n he process of he devaluaon of he Chnese yuan n 1994 and he srong US dollar endency agans he Japanese yen from he year The moral hazard problem of he Asan fnancal nsuons loans of large amouns of foregn capal also occurred because he currency peg sysem lowered exchange rae rss. Thus, we can fnd ha nflexbles of he Asan currences were one of he mporan facors o he Asan currency crses. In rerospec, wha nd of exchange rae regme s approprae for he Asan counres/regons? The Asan counres/regons are small The auhors are graeful o Professor Taao Fuuch for hs helpful commens on he earler draf. Needless o say, any remanng errors are solely our own. a Graduae School of Meda and Governance, Keo Unversy, Japan; e-mal: z02706@nfy.ne.p b Deparmen of Polcy Managemen, Keo Unversy, Japan; e-mal: hosaa@sfc.eo.ac.p 1

2 open economes whch are very dependen on rade. As a resul, exchange rae sably equals domesc prce sably. Under a free floa sysem, he exchange rae wll be volale o where leads o prce nsably. Hence, a base peg polcy ha s easy o conrol and wll no allow he real exchange rae o be overvalued s one of he alernave exchange rae polces for he Asan counres/regons. In hs paper, we wll nvesgae how he Asan currences are deermned n relaon o he movemens of maor currences usng he US dollar, he Japanese yen and he deuschemar as man examples. Then, we wll analyze how rade paerns are affeced when he Asan moneary auhores ncrease he wegh of he Japanese yen n her currency bases. 1 We wll also examne he effecs of nvesmens o rade paerns. To analyze hese ssues, we have consruced an nernaonal npu-ou pu model ha s lned wh macroeconomerc models. We use he Insue of Developng Economes he Asan Inernaonal Inpu- Oupu Table, 1990 ha has 78 ndusral secors 2 and covers 10 counres/regons (Indonesa, Malaysa, he Phlppnes, Sngapore, Thaland, Chna, Tawan, Korea, Japan and he Uned Saes). Ths paper s srucured as follows: The summarzaon of he leraure on he Asan currency crses n Secon 2. The explanaon of he model srucure n Secon 3. Smulaon resuls presened n Secon 4. Fnally, he conclusons n Secon 5. 2 Conroverses on he Asan Currency Crses There are wo nds of heorecal models have been developed for currency crses n general. One s he frs generaon model formulaed by Krugman [16], and Flood and Garber [06]. The frs generaon models suppose ha a small open economy adops he fxed exchange rae sysem and ha he moneary auhory fnances he governmen bonds ssued by he fscal auhory. Monezng he deb ncreases he money supply whch n urn causes he nflaon and decrease of neres raes. As a resul of hs process, currency value decreases. In hs suaon, he moneary auhory moves o susan he fxed rae, however, nvesors aac s currency before foregn reserve exhauson, creang a currency crss. The 1 As for smulaon analyses on he Asan currency devaluaons, here are few expermens. Far [04] uses a mul-counry macroeconomerc model o evaluae he effecs of devaluaons of Tha, Malaysan, Phlppne and Korean currences. McKbbn [20], and Noland, Lu, Robnson and Wang [21] examne he economc mpac of he producvy shocs or changes of rs premum usng compuable general equlbrum models. Our approach dffers from her analyses n he aspecs of nroducng and esmang currency bases n addon o usng an nernaonal npu-oupu sysem. 2 As for he secor classfcaon, see Table 1. 2

3 oher s he second generaon model developed by Obsfeld [22, 23]. As n he frs generaon model, he second generaon model also supposes ha a small open economy operaes s exchange rae polcy under he fxed sysem. An exchange rae polcy s conduced along he lnes of a ceran loss funcon. We assume ha a governmen mus endure he cos and he loss of rus n he fxed rae o adus s currency. Thus, a governmen wll evaluae he coss and benefs of a currency adusmen, and when he benefs are larger han he coss, a governmen wll abandon he fxed exchange rae sysem and vce versa. Wheher or no a governmen whdraws he fxed exchange rae depends on he expecaons of he prvae secor; hence, when people expec a currency adusmen, coss of mananng he fxed rae ncrease, resulng n a currency crss. The mare expecaon acually rggers a currency crss and he possbly of ha crss depends on he expecaon. From hs logc, we also call hs model a self-fulfllng model wh mul-equlbrums. The self-fulfllng model does no have any lmaons o he nernaonal reserves or any consderaon on macro-fundamenals concludng ha he lac of exchange reserves or poor macroeconomc condons wll no gve rse o a currency crss. Causes and polcy mplcaons on he Asan currency crses dffer beween hese models, however, boh of hem mplcae ha he drec cause of he Asan currency crses s he rapd nflow and ouflow of he nernaonal capal. As for he rapd capal nflow, fnancal lberalzaon wh nsuffcen supervson n addon o low neres raes n he U.S. and Japan generaed lendng booms n Asa. Ths s no he same regardng he causes of he nernaonal capal ouflows. Those who are based on he frs generaon model, e.g. Corse, Pesen and Roubn [02], Feldsen [05], Goldsen [09], and Noland, Lu, Robnson and Wang [21], focus on curren accoun defcs. They pon ou ha he followng wo problems are he man causes of he defcs, one of he causes beng he Asan exchange rae polces. 3 The Asan exchange rae polces were he de-faco dollar peg polcy. We consder ha here were hree reasons for conducng ha polcy. Frs, mos Asan counres/regons mpor her daly producs, e.g. foods and energy. The prces of he prmary producs are deermned a nernaonal mares n he US dollar. The Asan governmens red o conrol mpored nflaons by lnng he naonal currences o he US dollar. Second, he US mare s val o he Asan counres who have adoped he expor-orened developmen polcy. Thrd, remendous amoun of foregn capal has been nvesed n Asa snce he Plaza agreemen n he year 1985 whch has played an mporan role n Asa s developmen. To arac foregn nvesors o Asa, he sably of home currences agans he US dollar was necessary. Though he de-faco dollar peg sysem was worng well, he Asan real exchange rae had begun o be overvalued by he devaluaon of he Chnese yuan n 1994 along wh sronger dollar/weaer yen from he year Kohsaa [13] consders ha he exchange rae polcy s no a serous facor of he Asan currency crss based on he real effecve exchange rae sably n Indonesa, Korea, Malaysa, Thaland and Tawan from he laer half of he 80 s. 3

4 Asa s expor compeveness was damaged by he real exchange rae apprecaon causng a negave effec on expors hemselves, whch were one of he man engnes of he Asan economc growh. Anoher cause of curren accoun defcs was over-nvesmen. 4 Due o he closely fxed exchange rae, he Asan fnancal nsuons could borrow foregn capals whou any exchange rae rss. Furhermore, he Asan counres/regons lberalzed her fnancal sysems gnorng he use of supervson sysems generang he ncrease of easy borrowng of he foregn capals and easy nvesmens n real esaes or non-performng maers. Along hs lne, moral hazard problems n banng arose. These over-borrowngs and over-nvesmens ncreased boh nvesmens and mpors. In anoher opnon, Goldsen [09] quesoned he connuaon of he curren accoun defcs based on he over-producon of expor goods, he over-compeons n he expor mare and he proeconsm of he Uned Saes. Meanwhle, hose who nerpreed he Asan currency crses by he second generaon model, e.g. Furman and Sglz [08], Kohsaa [13] 5, and Radele and Sachs [25] 6, consdered ha hough he vulnerably of he Asan economy creaed panc, he macro-mbalances were no solely he reasons of he Asan currency crses. In her opnon, was he errors n he choce of polces whch made he crses worse. There s ye anoher opnon on he Asan currency crses. Krugman [18] argues ha he Asan currency crses are no conssen wh eher he frs generaon model or he second generaon model. There were no problems n fscal secors ha he frs generaon model supposes, and he Asan counres/regons dd no have any ncenves o devalue her currences as he second generaon model shows. He focuses on he rse and fall of he asse prces, and sresses ha he Asan currency crses are caused by he emergence and collapse of he asse bubbles. Though he holds a dfferen sance, he also pons ou ha he moral hazard problems n he banng sysem played a prncple role n he crses. As Krugman [18] menoned, neher he frs nor he second generaon models necessarly f he crcumsances of he Asan currency crses. When we ae a loo a economc fundamenals of he Asan counres/regons, e.g. he Tha curren accoun defcs, s dffcul o conclude ha here were no any problems n macro-fundamenals. Even f mare expecaons had rggered he crses, hose expecaons should have depended on he Asan macro economes. As many sudes show, s a fac ha macrombalances emerged n Asa. In he Asan case, s approprae for us o consder ha macro-mbalances, 4 Kohsaa [13] poned ou ha here were no only nvesmen booms bu also consumpon booms n Asa. 5 We caegorze Kohsaa [13] no he group ha s based on he second generaon model because hough he recognzes ha macro-mbalances were one of he causes of he crses, he sresses he mporance of he mare parcpans expecaons for devaluaons and polcy errors. 6 Radele and Sachs [25] ndcaes ha f he Tha governmen/moneary auhory had adoped he free floa sysem and ghened s macroeconomc polcy when he asse prces decreased n 1997, he Asan currency crses would no have occurred. 4

5 e.g. curren accoun defcs, are he man causes of he crses as he frs generaon model ndcaes. 7 Based on he frs generaon model, he man causes of he Asan currency crses are he exchange rae polces and over-nvesmens as we menoned above. The de-faco dollar peg sysem generaed he moral hazard problems of he fnancal nsuons, hence, he exchange rae polcy s he roo of he crses. The Asan naons/regons had adoped he de-faco dollar peg polcy n favor of prce sably, smooh access o he US mare and nernaonal capal nflows. Though hs aded Asa o develop a hgh speed, he fxed rae could no adus o he varaons of he exernal envronmens (e.g. he yen-dollar rae movemens). To solve hs problem, he Asan counres/regons mus possess exchange rae flexbly. However, s dffcul for a small and rade dependen counry, le ha of n Asa, o adop he free floa polcy because he exchange raes are much oo volale under ha sysem. Thus, we propose he currency base sysem ha enables ceran flexbly and s less volale han ha of he free floa polcy for Asan counres/regons o conduc her exchange rae polces. 8 In he case of Thaland and Korea, however, hey had n fac adoped he currency base sysems bu he weghs of he US dollar were very hgh. We conclude ha a currency base peg polcy wh much hgher wegh on he Japanese yen wll be more approprae han he de-faco dollar peg polcy or a free floa sysem for he Asan counres/regons. 3 The Srucure of he Model 3.1 Macroeconomerc Models There are several dfferences n he economc srucures beween developed and developng counres/regons. Of course, we can apply he same model o boh developed and developng counres/regons, ye s no an accurae approach n analyzng her economes. To acqure more precse oupu o each counry/regon s dfferences, we have consruced wo nds of macroeconomerc models Macroeconomerc Model for Developed Counres/Regons The macroeconomc bloc s consruced by he Keynesan framewor. Our benchmar model s he demand-deermned sysem developed by Klen [12] and ha s called he seleon model. The summaon 7 Krugman [17] crczes ha he second generaon model gnores he rend of fundamenals, and ndcaes ha when he fundamenals ge worse, he possbly of mulple equlbrum s lmed. 8 Hamada [10] analyzes he exchange rae polcy of a small counry whose nal exchange rae polcy s he fxed one by applyng he framewor of he hree-counres (wo large counres and one small counry) moneary polcy games. Hs analyss shows ha he free floa sysem or he currency base polcy s beer han he fxed rae, and ha he currency base polcy s more approprae han he free floa polcy under he srong wage demand pressure. 5

6 of each fnal demand componens equals gross naonal producs (GNP) and deermnes naonal producon oupus. Endogenous Varables C = Prvae fnal consumpon (Real) π = Corporae profs (Nomnal) I = Gross fxed capal formaon (Real) Z = GNP (Real) T 1 = Indrec ax (Nomnal) EX = Expors (Real) IM = Impors (Real) Y = Dsposable ncome (Nomnal) P = Prces D = Deprecaon (Real) r = Ineres rae (Nomnal) K = Capal soc (Real) T 2 = Personal ax (Nomnal) L = Employmen T 3 = Corporae ax (Nomnal) LF = Labor force w = Wage rae T r = Transfer paymens (Nomnal) Exogenous Varables G = Governmen spendng (Real) N = Populaon PM = Impor prce WT = World rade ransacons (Real) PW = World rade prce MS = Money supply (Nomnal) 6

7 Defnonal Equaons Real GNP (3.1) C + I + G + EX IM = Z The real GNP s explaned by hs defnonal equaon. From he prncple of effecve demand, he lef sde of he gross naonal expendure componens deermnes he producon oupu of a counry. Nomnal GNP (3.2) P Z T P D = Y + T + T Tr Ths equaon defnes he nomnal GNP. By ransformng hs equaon we explan he nomnal dsposable personal ncome. Naonal Income (3.3) w L π = Y + T + T Tr Though hs s he defnonal equaon of he naonal ncome, he nomnal corporae profs are accouned for by hs equaon. Capal Soc (3.4) K K + I D = 1 The ncrease of he capal soc for hs perod equals he gross fxed capal formaon mnus he deprecaon. The end of las perod s capal soc plus he ncrease of hs perod become he capal soc of hs perod. Behavor and Technologcal Equaons Consumpon 7

8 (3.5) = a + a ( Y P ) C a2c- 1 We adop he Brown-ype consumpon funcon whose dependen varables are he real dsposable ncome and he lagged consumpon expendure. Invesmen (3.6) = b + b Z + b r b K I The end perod s capal soc and he real GNP explan nvesmens. In hs funcon he neres rae s added as one of he explanaory varables because mos companes nves by borrowng. Expor (3.7) = c + c WT + c ( PW P ) EX c3ex - 1 Flucuaons of rade ransacons depend on changes of ncome and relave prces. The more rade parners ncome ncrease, he more expors ncrease. If expor prces are lower han mporer s prces, expors ncrease. World rade ransacons represen mporers ncome and he world rade prces sand for he oher counres prces. Impor (3.8) = d + d Z + d ( P PM ) IM d 3IM - 1 The real GNP, he rao of prces and mpor prces explan mpors. Employmen (3.9) ln L = e0 + e1 ln Z + e2 ln K -1 + e3 ln L -1 In hs model, employmen s deermned by usng he producon funcon. 8

9 Prces (3.10) P = f + f ( w L Z ) f 2 PM Prces depend on he un labor cos (marup prcng) and mpor prces (cos-push nflaon). Wages (3.11) w = g + g [ LF ( LF L )] + g ln P ln The Phllps curve, whch explans he relaon beween he unemploymen rae and nflaon, deermnes wages. Labor Force (3.12) N = h + h [( LF L ) LF ] h ( w P ) LF The unemploymen rae and he real wages explan he labor force. Velocy of Crculaon of Money (3.13) ( P Z MS ) ln = 0 + 1r + 2 ln P The neres rae and nflaon explan he velocy of money crculaon. From hs funcon, he neres rae s deermned. Deprecaon (3.14) D = 0 K -1 The deprecaon s esmaed a a ceran rae of he las perod s capal soc. 9

10 Insuonal Equaons Indrec Tax (3.15) = ( P Z ) T Indrec axes are explaned by he nomnal GNP. Personal Tax (3.16) T2 = l0 + l1y Personal axes depend on he nomnal dsposable personal ncome. Corporae Tax (3.17) T = m m π Corporae profs explan corporae axes. Transfer Paymens (3.18) T = n + n ( LF -L ) r n 2 w In reference o he unemploymen nsurance, ransfer paymens depend on unemploymen and wages. The exchange rae funcon s no n he Klen s seleon model. In hs sudy we use he Flaov-Klen exchange rae funcon. 9 Ths funcon s wren as: (3.19) ln = o + o ( P P ) + o ( r r ) + o ( Bal P Z ) e 0 1 ln US, 2 US, 3 where P US, s he prces of he US, r US, s he nomnal US neres rae and Bal s he curren accoun. The Flaov-Klen model explans he exchange rae by relave prces beween he home counry and he 10

11 US, he neres rae dfferences and he nomnal curren accoun per he nomnal GNP. The exchange rae s bascally deermned by he purchasng power pary. Is shor-erm flucuaon depends on he neres rae dfferences and he curren accoun Macroeconomerc Model for Developng Counres/Regons Developng counres bascally have labor surplus-capal shorage economes. They are excess demand economes. Hence, we analyze macroeconomc srucures of developng counres by he supplydeermned sysem. We use he UNCTAD s macroeconomerc model as he proo-ype model ha s n Ball [01]. Ths model does no have an exchange rae funcon, as n he Klen s seleon model. Thus, we use he Flaov-Klen exchange rae funcon o deermne endogenously. Endogenous Varables C = Prvae consumpon (Real) I = Invesmen (Real) J = Invenory Z a, = Agrculural producon Z na, = Non-agrculural producon Z P na, = Non-agrculural producon poenal X = Expor (Real) P = Prces PE = Expor prces (US dollar) PE d, = Expor prces P a, = Agrculural prces P na, = Non-agrculural prces R = Inernaonal reserve D f, = Exernal ndebedness (Nomnal) M = Impor (Real) Z n, = GNP (Real) K = Capal soc (Real) Z d, = GDP (Real) Z f, = Ne facor paymens abroad (Real, US dollar) Z * f, = Ne facor paymens abroad (Nomnal, US dollar) 9 See De Grauwe and Peeers [03] 11

12 Exogenous Varables N = Populaon PM = Impor prces TW = World expor volume ndex F = Ne capal nflow MS = Money supply = Tme PW = World expor prce ndex (US dollar) E = Exchange rae (US dollar per Home currency) Defnonal Equaons Real GDP (3.20) Z d, = Z a, + Z na, Agrculural producon plus non-agrculural producon deermne he real GDP because hs s he supply deermned model. Real GNP (3.21) Z n, = Z d, Z f, By defnon, he real GNP equals he real GDP mnus ne facor paymens abroad. Invenory (3.22) J = Z d, + M C I X By he defnon of he gross domesc expendure, we deermne nvenory as he resdual. Capal Soc (3.23) K = I 12

13 I s dffcul for economercans o collec capal soc daa of developng counres. Hence, we use he summaon of pas nvesmens as s proxy. Nomnal Exernal Indebedness * [ Z ] + (3.24) D = ( PM M ) ( PE X ) f, f, Ne mpor plus ne facor paymens abroad equal he flow of he exernal ndebedness. Hence, he soc of he exernal ndebedness s he summaon wh respec o me. Inernaonal Reserve * (3.25) + + ( ) ( ) R = R 1 F PE X PM M Z f, Changes of nernaonal reserves wll be explaned as ne capal nflow plus ne expor mnus ne facor paymens abroad. Inernaonal reserves are equal o s flow plus he lagged nernaonal reserves. Expor Prces n US Dollar (3.26) PE = PEd E Expor prces n he US dollar can be defned as expor prces n he home currency by he exchange rae (he US dollar per he home currency). Ne facor paymens abroad n US dollar (Real) * (3.27) Z = ( Z /PM ) f, f, Nomnal ne facor paymens abroad s deflaed by mpor prces. Behavor Equaons 13

14 Agrculural Producon (3.28) = a a Z a, We assume ha agrculural producon depends on me. Consumpon (3.29) C N b + b ( Z N ) + b ( C N ) = 0 1 n, We explan per capa real consumpon by per capa GNP and he lagged value of per capa real consumpon. Invesmen (3.30) I = c0 + c1z d, + c2z d, 1 Invesmens are explaned by he real GDP and he lagged real GDP. Expor (3.31) = d + d TW d ( PE /PW ) X Expors are deermned by he world rade volume ndex and he expor prces relave o he world expor prces. Impor (3.32) e + e Z + e ( R /PM ) e ( PM P E ) M = 0 1 d, Impors are explaned by explanaory varables; he real GDP, he mpor prces relave o he domesc prces n he US dollar and he nernaonal reserves deflaed by he mpor prces. 14

15 Non-agrculural Producon Poenal P (3.33) Z na, = f 0 + f1k 1 Mos developng counres have he capal shorage economes. Hence, non-agrculural producon poenal depends on he capal socs. Non-agrculural Producon - Type I P Z (3.34) na, Z na, = g + g M g Z a, Ths non-agrculural producon funcon focuses on he supply sde. The non-agrculural producon poenal and he real mpors explan he non-agrculural producon. Mos developng counres mpor capal goods and nermedae npus, herefore we assume ha he real mpors also deermne he nonagrculural producon. Non-agrculural Producon - Type II (3.35) na, = g + g C + g I g X Z We can also formulae he non-agrculural producon by he effecve demands. In hs case, explanaory varables are he real consumpons, he real nvesmens and he real expors. Agrculural Prces (3.36) a, h + h Z a, + h Z a, h Z na, P = Agrculural prces are deermned by he agrculural producon and non-agrculural producon. Nonagrculural producon s a proxy varable of he demand for he agrculural producs. Non-agrculural Prces - Type I 15

16 P P = Z Z + 2 MS Z + 3P + 4 PM (3.37) na, ( na, na, ) ( d, ) a, In one of he formulaons of non-agrculural prces, can be explaned by he capacy ulzaon rao, he rao of he money supply o he real GDP, and he agrculural prces plus he mpor prces as producon coss. Non-agrculural Prces - Type II (3.38) + Z + Z ( MS Z ) Pna, = 0 1 na, 2 na, d, In anoher formulaon of he non-agrculural prces, we assume ha here s a non-agrculural producon lmaon. In hs case, wll depend on he non-agrculural producon, he lagged nonagrculural producon and he raon of he money supply o he real GDP. Prces (3.39) = + Pa, Pna, P Ths model s he wo-secor supply-deermned sysem. Hence, Agrculural prces and non-agrculural prces deermne he general prces. Expor Prces n he Home Currency (3.40) PEd = 0 + 1P + 2 X + 3PEd, 1 Expor prces are deermned by he general prces, expors and he lagged dependen varable. Nomnal Ne Facor Paymens Abroad n he US Dollar * (3.41) l + l D l ( PE X ) Z f, = 0 1 f, Nomnal ne facor paymens abroad s explaned by he neres paymens of he exernal ndebedness as he paymen facor and expors as he recevable facor. 16

17 3.1.3 Esmaon Resuls Srucural equaons of he en counres/regons are esmaed by me-seres daa. 10 Table 2 shows several feaures of our macroeconomerc models. Though we provde only exchange rae funcon resuls ha are mos mporan n hs analyss, we noe ha all macroeconomerc models passed he fnal ess wh good resuls. General noaons on esmae resuls of exchange raes are as follows: he upper parenhess = -sasc, he lower brace = p-value, Ad. R 2 = he adused R-squared, S. E. = he sandard error, D. W. = he Durbn-Wason sasc, OLS = he ordnary leas squares, NLS = he nonlnear leas squares. A subscrp s a degree of lags. Abbrevaons of a counry/regon are as follows: IDN = Indonesa, MLS = Malaysa, PHL = he Phlppnes, SGP = Sngapore, THA = Thaland, CHN = Chna, TWN = Tawan, KOR = Korea, JPN = Japan, USA = he Uned Saes. Indonesa The Indonesan exchange rae funcon s as follows: (3.42) EIDN = *(D86LATER*E) *LOG(PEX738393IDN/PEX92USA) (13.108) (3.477) (5.216) [0.0000] [0.0046] [0.0002] *((EXIDN -1 -IMIDN -1 )/GDPIDN -1 ) *D *D79_82 (-2.239) ( ) (-9.728) [0.0448] [0.0000] [0.0000] *D83_ *D88_96 (-8.671) (-2.432) [0.0000] [0.0316] Ad. R 2 = S. E. = D. W. = 1.22 Sample: Mehod: OLS where EIDN s he Indonesan exchange rae, E s he Japanese exchange rae, PEX738393IDN s he expor deflaor of Indonesa (1973, 1983, 1993 = 100), PEX92USA s he expor deflaor of he Uned Saes (1992 = 100), (EXIDN -1 -IMIDN -1 )/GDPIDN -1 s he lagged rao of he nomnal curren accoun agans he nomnal GDP n Rupah, D86LATER s a dummy varable (1986-he end of he smulaon perod = 1; oherwse 0), D78 s a dummy varable (1978 = 1; oherwse 0), D79_82 s a dummy varable ( = 1; oherwse 0), D83_87 s a dummy varable ( = 1; oherwse 0) and D88_96 s a dummy varable ( = 1; oherwse 0). The Indonesan Rupah s deermned by he Japanese yen 17

18 (1986-), he home expor deflaor relave o he US expor deflaor and he rao of curren accoun agans he nomnal GDP. All sascs perform well excep for he Durbn-Wason sasc. Based on he Durbn-Wason sasc, we can recognze he posve auoregressve process. Malaysa The Malaysan exchange rae funcon s esmaed as follows: (3.43) EMLS = *(D85LATER*E) *(D86LATER*EGER) (64.271) (3.310) (9.515) [0.0000] [0.0048] [0.0000] *LOG(PEX7078MLS/PEX92USA) *D79_ *D86_88 (4.859) (-4.446) (-4.823) [0.0002] [0.0005] [0.0002] *D92_ *D95_96 (-3.648) (-4.549) [0.0024] [0.0004] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where EMLS s he Malaysan rngg per he US dollar, EGER s he deuschemar per he US dollar, D85LATER s a dummy varable (1985-he end of he smulaon perod = 1; oherwse 0), D79_80 s a dummy varable ( = 1; oherwse 0), D79_80 s a dummy varable ( = 1; oherwse 0), D92_93 s a dummy varable ( =1; oherwse 0) and D95_96 s a dummy varable ( = 1; oherwse 0). The Malaysan rngg depends on he Japanese yen (1985-), he deuschemar (1986-) and he relave prces of radable goods. We can fnd ha sascs provde suffcen resuls. The Phlppnes The esmaon resul of he Phlppne exchange rae s as follows: (3.44) EPHL = *(D86LATER*E) *LOG(PEX85PHL/PEX92USA) (47.432) (1.969) (23.056) [0.0000] [0.0706] [0.0000] *D *D90_ *D93 (-3.098) (3.999) (2.032) 10 As for daa sources, see Appendx A. 18

19 [0.0085] [0.0015] [0.0631] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where EPHHL s he exchange rae of he Phlppnes, D79 s a dummy varable (1979 = 1; oherwse 0), D90_91 s a dummy varable ( = 1; oherwse 0) and D93 s a dummy varable (1993 = 1; oherwse 0). The Phlppne peso s deermned by he Japanese yen (1986-) and he relave expor deflaors. The Japanese yen s no sascally sgnfcan a he 95 percen level bu s a he 90 percen level. In excepon o ha, we can conclude ha hs funcon s well-esmaed. Sngapore The Sngaporean exchange rae funcon s esmaed as follows: (3.45) ESGP = *(D85LATER*E) (19.738) (4.779) [0.0000] [0.0004] *LOG(PGDP8590SGP -2 /PGDP92USA -2 ) ( 4.251) [0.0011] *(NEXSGP -1 /GDPSGP -1 ) *D79_ *D *D85 (-2.893) (3.862) (1.982) (-2.108) [0.0135] [0.0023] [0.0709] [0.0567] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where ESGP s he Sngaporean dollar per he US dollar, PGDP8590SGP s he GDP deflaor of Sngapore (1985, 1990 = 100), PGDP92USA s he US GDP deflaor (1992 = 100), NEXSGP s he nomnal ne expor of Sngapore, D79_89 s a dummy varable ( = 1; oherwse 0), D84 s a dummy varable (1984 = 1; oherwse 0) and D85 s a dummy varable (1985 = 1; oherwse 0). The Japanese yen (1985-), he relave GDP deflaors and he curren accoun deermne he value of he Sngaporean dollar. Though mos sascs show good resuls, he Durbn-Wason sasc shows he posve auoregressve process. Thaland The esmaed exchange rae funcon of Thaland s as follows: (3.46) ETHA = *E *(RGBTHA -2 -RGBUSA -2 ) 19

20 (45.047) (2.544) (-2.112) [0.0000] [0.0273] [0.0584] *((EXTHA -3 -IMTHA -3 )/GDPTHA -3 ) *D78_ *D81_83 (-2.677) ( ) ( ) [0.0215] [0.0000] [0.0000] *D *D *D86_96 ( ) ( ) ( ) [0.0000] [0.0000] [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where ETHA s he Tha exchange rae, RGBTHA s he governmen bond yeld of Thaland, RGBUSA s he US governmen bond yeld, EXTHA s he nomnal expor of Thaland, IMTHA s he nomnal mpor of Thaland, GDPTHA s he Tha nomnal GDP, D78_80 s a dummy varable ( = 1; oherwse 0), D81_83 s a dummy varable ( = 1; oherwse 0) and D86_96 s a dummy varable ( = 1; oherwse 0). The Tha bah s deermned by he neres rae dfference and he curren accoun. In he Tha case, he relave prce s no sascally sgnfcan. Chna We esmae he Chnese exchange rae funcon as follows: (3.47) ECHN = *(D85LATER*E) *LOG(PGDP95CHN/PGDP92USA) (22.279) (2.112) (1.970) [0.0000] [0.0584] [0.0745] *D80_ *D88_ *D90_93 + [AR(1)=0.297] (-3.169) (-3.703) (-2.911) (2.196) [0.0089] [0.0035] [0.0142] [0.0504] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where ECHN s he Chnese exchange rae, PGDP95CHN s he Chnese GDP deflaor (1995 = 100), D80_87 s a dummy varable ( = 1; oherwse 0), D88_89 s a dummy varable ( = 1; oherwse 0), D90_93 s a dummy varable ( = 1; oherwse 0) and AR(1) s he frs order auoregressve process. Though he Japanese yen (1985-) and he relave GDP deflaors deermne he Chnese yuan, he coeffcen of he relave prce s sascally sgnfcan a he 90 percen level. 20

21 Tawan The Tawanese exchange rae funcon s esmaed as follows: (3.48) ETWN = *E *LOG(PEX8691TWN/PEX92USA) (20.390) (4.074) (9.948) [0.0000] [0.0015] [0.0000] *(RSTWN -3 -RSUSA -3 ) (-2.558) [0.0251] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where ETWN s he exchange rae of Tawan, PEX8691TWN s he expor deflaor of Tawan (1986, 1991 = 100), RSTWN s he Tawanese shor-erm neres rae and RSUSA s he US shor-erm neres rae. In he Tawanese case, he Japanese yen, he relave expor deflaors and he neres rae dfference explan he exchange rae. The curren accoun does no affec he exchange rae. All sascs are accepable. Korea The exchange rae funcon of Korea s esmaed as follows: (3.49) EKOR = *E *LOG(PEX95KOR/PEX92USA) ( ) (4.619) (7.307) [0.0000] [0.0006] [0.0000] *(RSKOR-RSUSA) *((EXKOR -1 -IMKOR -1 )/GDPKOR -1 ) (-1.851) (-2.714) [0.0889] [0.0188] + [AR(1)=0.339] (4.425) [0.0008] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where EKOR s he Korean exchange rae, PEX95KOR s he expor deflaor of Korea (1995 = 100), RSKOR s he shor-erm neres rae of Korea, EXKOR s he Korean nomnal expor, IMKOR s he nomnal mpor of Korea and GDPKOR s he Korean nomnal GDP. In hs esmaon, we assume he frs order auoregressve process. The Korean won s deermned by he Japanese yen, he relave expor 21

22 deflaors, he neres rae dfference and he curren accoun. The coeffcen of he neres rae dfference s no sascally sgnfcan a he 95 percen level, ye he Korean exchange rae funcon s esmaed well. Japan The Japanese exchange rae funcon s esmaed as follows: (3.50) LOG(E) = *LOG(PGDP90/PGDP92USA) *(RGB -2 -RGBUSA -2 ) (54.278) (4.926) (-1.806) [0.0000] [0.0001] [0.0887] *((EX-IM)/GDP) *D *D *D952 (-2.788) (-3.568) (-1.920) (-3.506) [0.0126] [0.0024] [0.0717] [0.0027] + [AR(1)=0.608] (4.670) [0.0002] Ad. R 2 = S. E. = D. W. = Sample: 1990:1-1996:1 Mehod: NLS where LOG(x) s he log of he varable x, PGDP90 s he Japanese GDP deflaor (1990 = 100), RGB s he governmen bond yeld of Japan, EX s he nomnal expor of Japan, IM s he Japanese nomnal mpor, GDP s he nomnal GDP of Japan, D904 s a dummy varable ( he fourh quarer of he year 1990 = 1; oherwse 0), D932 s a dummy varable ( he second quarer of he year 1993 = 1; oherwse 0), D952 s a dummy varable ( he fourh quarer of he year 1995 = 1; oherwse 0). The relave GDP deflaor, he long-erm neres rae dfference and he curren accoun explan he Japanese yen - he US dollar rae. We esmae hs funcon wh he assumpon of he frs order auoregressve process. Alhough he neres rae dfference s sascally sgnfcan a he 90 percen level, oher sascs show good resuls Polcy Reacon Funcons Polcy varables, e.g. he shor-erm neres rae and he governmen expendure, are exogenous ones n he general macroeconomerc models. In hs model, we esmae polcy reacon funcons o deermne macroeconomc polcy nsrumens endogenously. The moneary polcy reacon funcon s esmaed for boh developed and developng counres/regons, and he fscal one s esmaed for only Japan and he Uned Saes. As for he moneary polcy, we assume ha he moneary polcy nsrumen of he developed counres s he shor-erm neres rae and ha of he developng counres/regons s he money 22

23 supply. The general noaons, subscrps and abbrevaons on a counry/regon are he same as hose n Secon Indonesa The moneary polcy reacon funcon of Indonesa s esmaed as follows: = *(@PCH(PCP738393IDN -1 )) *EIDN -1 (3.736) (-2.106) (-2.086) [0.0028] [0.0570] [0.0590] *(@PCH((EXIDN-IMIDN)/GDPIDN)) *RSIDN -2 (-5.600) (-1.666) [0.0001] [0.1215] *D75_ *D75_ *D *D83_87 (2.927) (-2.304) (3.974) (-2.511) [0.0127] [0.0399] [0.0018] [0.0273] + [MA(1)=-0.990,BACKCAST=1977] ( ) [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS s he one-perod percenage change of he varable x, M2IDN s M2 of Indonesa, PCP738393IDN s he prvae consumpon deflaor of Indonesa (1973, 1983, 1993 =100), RSIDN s he shor-erm neres rae of Indonesa, D75_79 s a dummy varable ( = 1; oherwse 0), D75_82 s a dummy varable ( = 1; oherwse 0), D80 s a dummy varable (1980 = 1; oherwse 0), and MA(1) s he frs order movng average erm. The Indonesan moneary polcy depends on he prvae consumpon deflaor change, he exchange rae, he curren accoun and he shor-erm neres rae. They are sascally sgnfcan a he 90 percen, he 90 percen, he 99 percen and he 85 percen levels, respecvely. Therefore, he Indonesan moneary auhory focuses on he curren accoun, n parcular. Ths s because as Corese, Pesen and Roubn [02, Table 23] shows, he rao of foregn debs agans he nomnal GDP s relavely hgh n Indonesa, hence, he curren accoun s he mos mporan for repayng hem. As for oher sascs, resuls are accepable n excepon o he low adused R-squared. Malaysa The Malaysan moneary polcy reacon funcon s esmaed as follows: 23

24 (3.52) LOG(M2MLS) = *((EXMLS -2 -IMMLS -2 )/GDPMLS -2 ) ( ) (-2.577) [0.0000] [0.0180] *(@TREND(1975)) + [MA(1)=0.953,BACKCAST=1975] (33.884) (12.749) [0.0000] [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where M2MLS s M2 of Malaysa, EXMLS s he Malaysan nomnal expor, IMMLS s he nomnal mpor of Malaysa, GDPMLS s he nomnal GDP of Malaysa s a me rend (1975 = 0). The Malaysan moneary polcy s deermned by s curren accoun. The prce level/change s no sascally sgnfcan. The Malaysan moneary auhory also pays aenon o balances beween Malaysa and foregn counres/regons. The Phlppnes We esmae he Phlppne moneary polcy reacon funcon as follows: = *(@PCH(PCP85PHL -2 )) *EPHL (3.754) (-1.390) (-2.199) [0.0024] [0.1876] [0.0466] *((EXPHL-IMPHL)/GDPPHL) *D75_ *D79 (-3.772) (-0.643) (-2.887) [0.0023] [0.5317] [0.0127] *D *(D88_89) (8.058) (4.479) [0.0000] [0.0006] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where M2PHL s M2 of he Phlppnes, PCP85PHL s he prvae consumpon deflaor of he Phlppnes (1985 = 100), EXPHL s he nomnal expor of he Phlppnes, IMPHL s he nomnal mpor of he Phlppnes, GDPPHL s he Phlppne nomnal GDP, D75_92 s a dummy varable ( = 1; oherwse 0) and D83 s a dummy varable (1983 = 1; oherwse 0). The Phlppne moneary polcy s 24

25 explaned by he prvae consumpon deflaor, he exchange rae and he curren accoun. Though he prvae consumpon deflaor s sascally sgnfcan a he 80 percen level, he exchange rae and he curren accoun are sascally sgnfcan a he 95 percen and 99 percen levels, respecvely. The Phlppne moneary auhory pays more aenon o he curren accoun han he domesc nflaon. Though D75_92 s no sascally sgnfcan, we use as an explanaory varable because he Phlppne M2 daa s no connuous before/afer he year Sngapore The Sngaporean moneary polcy reacon funcon s esmaed as follows: = *(@PCH(PCP8590SGP)) *(RTBUSA-RMMSGP) (10.595) (-4.707) (-1.404) [0.0000] [0.0011] [0.1938] *D75_ *D *D *D *D89 (8.449) (-3.103) (-7.382) (-5.911) (3.485) [0.0000] [0.0127] [0.0000] [0.0002] [0.0069] *D *D98 + [AR(3)=-0.666] (2.742) (6.397) (-6.392) [0.0228] [0.0001] [0.0001] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where M2SGP s M2 of Sngapore, PCP8590SGP s he prvae consumpon deflaor of Sngapore (1985, 1990 = 100), RTBUSA s he US reasury bll rae, RMMSGP s he money mare rae of Sngapore, D75_89 s a dummy varable ( = 1; oherwse 0), D86 s a dummy varable (1986 = 1; oherwse 0), D89 s a dummy varable (1989 = 1; oherwse 0), D94 s a dummy varable (1994 = 1; oherwse 0), D98 s a dummy varable (1998 = 1; oherwse 0) and AR(3) s he hrd order auoregressve erm. The domesc prvae consumpon deflaor and he neres raes dfference beween Sngapore and he Uned Saes explan he money supply of Sngapore. The p-value of he prvae consumpon deflaor s only , hence, hs s sascally sgnfcan a he 99 percen level. The neres rae dfference s sascally sgnfcan a he 80 percen level. The Sngaporean moneary auhory conducs s moneary polcy o conrol he domesc nflaon. Thaland The esmaon resul of he Tha moneary polcy reacon funcon s as follows: 25

26 = *(@PCH(PCP7288THA)) *(@PCH(ETHA)) (22.069) (-1.683) (-3.187) [0.0000] [0.1117] [0.0057] *D75_ *(D82_83) *(D89_90) (6.088) (8.202) (8.809) [0.0000] [0.0000] [0.0000] + [MA(1)=-0.989,BACKCAST=1976] (-3.376) [0.0038] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where M2THA s M2 of Thaland, PCP7288THA s he prvae consumpon deflaor of Thaland (1972, 1988 = 100), D82_83 s a dummy varable ( = 1; oherwse 0) and D89_90 s a dummy varable ( = 1; oherwse 0). In Thaland, he moneary polcy s deermned by he prvae consumpon deflaor and he exchange rae. The prvae consumpon deflaor s sascally sgnfcan a he 85 percen level and he exchange rae s a he 99 percen level. The man polcy arge of he Tha moneary auhory s conrollng he value of he Tha bah per he US dollar. The adused R-squared s less han 0.800, however, oher sascs show good resuls. Chna We esmae he Chnese moneary polcy reacon funcon as follows: = *(@PCH(AGR95CHN -3 )) *ECHN -2 (0.809) (-1.668) (-3.546) [0.4450] [0.1392] [0.0094] *(NEXCHN/GDPCHN) *D *D88 (-3.618) (1.554) (-2.348) [0.0085] [0.1640] [0.0512] *D *(@TREND(1975)) (-2.936) (2.953) [0.0218] [0.0213] + [MA(1)=-0.926,BACKCAST=1983] (-9.561) 26

27 [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where M2CHN s M2 of Chna, AGR95CHN s he real agrculural producon of Chna (he base year = 1995), NEXCHN s he nomnal ne expor of Chna, GDPCHN s he nomnal GDP of Chna and D88 s a dummy varable (1988 = 1; oherwse 0). The Chnese moneary polcy s explaned by he agrculural oupu, he exchange rae and he curren accoun. The exchange rae and he curren accoun are more sascally sgnfcan han ha of he agrculural producon. The GDP deflaor s no sascally sgnfcan. The Chnese moneary auhory focuses on he exchange rae and he curren accoun raher han he nflaon. Tawan The moneary polcy reacon funcon of Tawan s esmaed as follows: = *(@PCH(GDP8691TWN -1 )) *(@PCH(ETWN)) (13.122) (-8.462) (-2.361) [0.0000] [0.0000] [0.0322] *RTBUSA *D75_ *D89_90 (-4.679) (-0.617) (-2.226) [0.0003] [0.5465] [0.0418] *(@TREND(1975)) (-9.862) [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: OLS where M2TWN s M2 of Tawan, GDP8691TWN s he real GDP of Tawan (base year = 1986 and 1991) and D75_77 s a dummy varable ( = 1; oherwse 0). The economc growh, he exchange rae and he US reasury bll rae explan he money supply of Tawan. The Economc growh and he US reasury bll rae are sascally sgnfcan a he 99 percen level, hence, he Tawanese moneary auhory focuses on boh he domesc sably and he US moneary polcy. Korea The esmaon resul of he Korean moneary polcy reacon funcon s as follows: 27

28 = *(@PCH(PCP95KOR -3 )) *(@PCH(GDP95KOR)) (3.534) (-3.165) (-3.281) [0.0033] [0.0069] [0.0055] *(@PCH(M2KOR -1 )) *(@TREND(1975)) (4.545) (-2.078) [0.0005] [0.0566] + [MA(1)=-0.989,BACKCAST=1979] ( ) [0.0000] Ad. R 2 = S. E. = D. W. = Sample: Mehod: NLS where M2KOR s M2 of Korea, PCP95KOR s he prvae consumpon deflaor of Korea (1995 = 100), GDP95KOR s he real GDP of Korea (base year = 1995). The moneary polcy of Korea s explaned by he prvae consumpon deflaor and he economc growh. The Korean moneary auhory arges he economc growh whou he nflaon. As for sascs, we mus noe ha he adused R-squared s relavely low. Japan Moneary Polcy The Japanese moneary polcy reacon funcon s esmaed as follows: (3.59) RMM = *PCP *RMM *GR_M2CD (-3.506) (3.318) (28.606) (2.160) *GR_M2CD *GR_M2CD *GR_M2CD -3 (1.148) (1.608) (2.217) *GR_M2CD *GR_M2CD -5 Sum of lags = (1.658) (0.020) (5.811) Ad. R 2 = S. E. = D. W. = Sample: 1986:3-1996:1 Mehod: OLS where RMM s he overngh call rae of Japan, PCP90 s he prvae consumpon deflaor of Japan (1990 = 100) and GR_M2CD s he percenage change of M2 plus cerfcaes of deposs of Japan. The prvae consumpon deflaor and he money growh explan he Japanese moneary polcy. For he esmaon, we apply he Almon polynomal lag o he money growh (fve lags and he hrd degree of polynomals). The 28

29 money growh can be recognzed as one of he nermedae arges of he moneary polcy. Therefore, he Japanese moneary auhory focuses on he domesc nflaon. Fscal Polcy The Japanese fscal polcy reacon funcon s esmaed as follows: (3.60) GR4_IG90 = *GR4_GDP *(@MOVSUM(DFBAL,5)) (1.307) (-3.328) (0.598) [0.2045] [0.0031] [0.5557] *GR4_IG90(-1) *D *D *D941 (9.001) (2.382) (-1.980) (-2.279) [0.0000] [0.0263] [0.0603] [0.0327] *D *D *D *D961 (-3.291) (-2.322) (-3.435) (4.278) [0.0033] [0.0299] [0.0024] [0.0003] Ad. R 2 = S. E. = D. W. = Sample: 1988:1-1996:1 Mehod: OLS where GR4_IG90 s he four-perod percenage change of he real publc nvesmens (he base year = 1990) of Japan, GR4_GDP90 s he four-perod percenage change of he real GDP (he base year = 1990) of s he fve-perod bacward movng summaon of he Japanese fscal defcs, D902 s a dummy varable (he second quarer of he year 1990 = 1; oherwse 0), D941 s a dummy varable (he frs quarer of he year 1994 = 1; oherwse 0), D943 s a dummy varable (he hrd quarer of he year 1994 = 1; oherwse 0), D944 s a dummy varable (he fourh quarer of he year 1994 = 1; oherwse 0), D951 s a dummy varable (he frs quarer of he year 1995 = 1; oherwse 0) and D961 s a dummy varable (he frs quarer of he year 1996 = 1; oherwse 0). The Japanese fscal polcy s deermned by he GDP growh and he fscal defc per he nomnal GDP, however, he fscal defc erm s no sascally sgnfcan. The Japanese governmen has conduced s fscal polcy whou any consderaon o he defcs. Alhough hs s he case, we explan he fscal polcy by he fscal defcs erm o negrae he fscal adusmen mechansm no our model. USA Moneary Polcy The esmaon resul of he US moneary polcy reacon funcon s as follows: 29

30 (3.61) RFFUSA = *GR4_PCP92USA *URUSA *GR4_M2USA (7.303) (15.739) (-4.872) (6.662) [0.0000] [0.0000] [0.0000] [0.0000] *D *D *D *D *D95 (-2.448) (-3.059) (-3.857) (-3.093) (2.810) [0.0187] [0.0039] [0.0004] [0.0036] [0.0076] Ad. R 2 = S. E. = D. W. = Sample: 1986:1-1998:2 Mehod: OLS where RFFUSA s he federal funds rae, GR4_PCP92USA s he four-perod percenage change of he prvae consumpon deflaor of he Uned Saes (base year = 1992), URUSA s he US unemploymen rae, GR4_M2USA s he four-perod percenage change of he US money supply (M2), D90 s a dummy varable (4 quarers of he year 1990 = 1; oherwse 0), D91 s a dummy varable (four quarers of he year 1991 = 1; oherwse 0), D92 s a dummy varable (four quarers of he year 1992 = 1; oherwse 0), D93 s a dummy varable (four quarers of he year 1993 = 1; oherwse 0) and D95 s a dummy varable (four quarers of he year 1995 = 1; oherwse 0). The US moneary polcy s deermned by he domesc nflaon, he unemploymen rae and he money growh. In he case of he Uned Saes, raher han he GDP growh, he unemploymen rae s sascally more suable. The unemploymen rae represens boh he economc growh and he wage movemens. Thus, he US Federal Reserve arges boh he prce sably and he GDP growh. Fscal Polcy The US fscal polcy reacon funcon s esmaed as follows: (3.62) LOG(IG92USA) = *(URUSA-URUSA -4 ) *LOG(GDP92USA) (5.479) (10.568) (-2.722) [0.0000] [0.0000] [0.0098] *LOG(GRUSA/PCG92USA*100) *LOG(IG92USA -1 ) (3.899) (16.192) [0.0004] [0.0000] *D *D *D *D943 (-4.622) (-3.431) (-3.330) (2.741) [0.0000] [0.0015] [0.0020] [0.0094] + [MA(1)=-0.989,BACKCAST=1986:1] 30

31 (-7.815) [0.0000] Ad. R 2 = S. E. = D. W. = Sample: 1986:1-1997:3 Mehod: NLS where IG92USA s he real publc nvesmens of he Uned Saes (1992 prces), GDP92USA s he real US GDP (1992 prces), GRUSA s he governmen revenue of he Uned Saes, PCG92USA s he US governmen consumpon deflaor (1992 = 100), D911 s a dummy varable (he frs quarer of he year 1991 = 1; oherwse 0) and D913 s a dummy varable (he hrd quarer of he year 1991 = 1; oherwse 0). The unemploymen, he GDP growh and he real governmen revenue explan he US fscal polcy. Due o he huge defcs, he US fscal polcy has no played an mporan role as one of he macroeconomc polces, however, we fnd ha he US fscal auhory conducs s polcy wh consderaon o he GDP growh and he level of he governmen revenue. 3.2 Inernaonal Inpu-Oupu Model 11 An nernaonal npu-oupu model s formulaed as below. 12 Varables a = Inpu coeffcen of he h secor n he h secor of he h counry/regon X = Toal oupu n he h secor of he h counry/regon X ( h) = Impors of he h commody from he hh counry/regon n he h secor of he h counry/regon s ( ) = Each counry/regon s elascy of subsuon of he h commody n he h secor of he h * counry/regon m ( h) = Inal share coeffcen of he h commody of he hh counry/regon n he h secor of he h counry/regon PX ( h) = Expor prces of he h commody of he hh counry/regon n he h secor of he h counry/regon 11 Ths nernaonal npu-oupu model was orgnally developed by Kosaa [14, 15]. 12 For anoher formulaon of an nernaonal npu-oupu model, see Tor and Ayama [28]. 31

32 = Tarff rae of he h commody of he h counry/regon (=Tarff on he h commody of he h counry/regon / mpor of he h commody of he h counry/regon) F ( h) = Fnal demand n he h secor of he hh counry/regon X ( h) = Toal oupu n he h secor of he hh counry/regon C 1 = Tarff on he mpor of he h commody n he h secor of he h counry/regon C 0 = Inernaonal fregh and nsurance, and ohers. n he h secor of he h counry (= Toal cos of c o he nernaonal fregh and nsurance, arff and ohers. - arff) = Inal rao of nernaonal fregh and nsurance, and ohers. o he oal oupu n he h secor of he h counry/regon F = Fnal demand of he h counry/regon o F = Fnal demand of he res of he world H = Converer marx of he fnal demand of he h counry/regon (= f g ) g = Fnal demand of he h counry/regon ransferred from s macroeconomerc model q P = Prces n he h secor of he qh counry/regon (he base year = 1) sf ( g ) = Elasces of subsuon of he fnal demands of he h counry/regon * f g = Inal shares of fnal demands of he h counry/regon P = Prces n he h secor of he h counry/regon (he base year = 1) v = Value added coeffcen n he h secor of he h counry/regon W = Wage n he h secor of he h counry/regon V 0 = Value added mnus wage n he h secor of he h counry/regon w = Wage rae n he h secor of he h counry/regon L = Employmen n he h secor of he h counry/regon b = Elascy of labor npu n he h secor of he h counry/regon w = Wage rae of he h counry/regon v 0 = Rao of value added mnus wage o he oal oupu n he h secor of he h counry/regon 32

33 h τ = Expor subsdy of he h expored commody of he hh counry/regon Inermedae Inpu and Techncal Coeffcen (3.63) X = a X (3.64) a = X ( h) h X, : secor, h: counry/regon Inermedae npus for he un producon of he h secor n he h counry/regon, n oher words, he echncal coeffcens, are defned n he same manner as hose of a domesc npu-oupu model. Inernaonal Transacon of Inermedae Inpu (3.65) * * X ( h) X = m ( h) ( 1 + ) PX ( h) m ( q )( 1 + ) PX ( q ) q: compeors q h s ( ) Hcman-Lau [11] developed a rade lnage sysem for he Proec LINK, hence, we can use her model o explan nernaonal ransacons of nermedae npus. The rao of he h commody of he hh counry/regon n he h secor of he h counry/regon o he h commody n he h secor of he h counry/regon s deermned by he nal rade share, he expor prces of he hh counry/regon relave o ha of he qh counry/regon and he elascy of subsuon. Toal Oupu = (3.66) X ( h) + F ( h) X ( h) = 1,!, M ; h = 1,!, N The hh counry/regon s oupu of he h secor equals s nermedae npus plus s fnal demands. Fnal Demand 33