Building a New Framework for Analyzing Effects of Japanese Shocks on Asia

Transcription

1 Preliminary Building a New Framework for Analyzing Effecs of Japanese Shocks on Asia March 1, 2004 Esuro Shioi * (Yokohama Naional Universiy) * I would like o hank Kiyoaka Sao for valuable commens on he previous version of he paper.

2 Absrac: This paper is a firs sep oward building a new macroeconomic model ha is usable for analyzing he effecs of shocks ha originae in Japan on Asian economies. The new framework borrows is cenral ingrediens from he lieraure of he new open economy macroeconomics, ha is characerized by explici dynamic opimizaion, shor-run nominal rigidiy, and imperfec subsiuion beween producs. The las feaure of his approach enables us o analyze how he rade srucure beween counries influences inernaional ransmission of shocks. This paper builds a hree-counry model, where he hree counries are Asia, Japan, and US, which reflecs rade and producion paerns beween hem. Thus, he model is expeced o yield more realisic predicions abou how policy and produciviy shocks in Japan affec Asian economies, boh in he shor and he long runs. 2

3 1 Inroducion This paper is a firs sep oward building a new macroeconomic model ha is usable for analyzing he effecs of shocks ha originae in Japan on Asian economies. How shocks o he Japanese economy, such as produciviy slowdown and moneary expansion, affec Asian economies is of grea ineres o policy makers boh in Japan and in he res of Asia. Heurisically, i seems plausible ha he ransmission mechanism has undergone some changes due o shifs in rade srucure beween Japan and Asia in he pas weny years or so. However, sandard macroeconomic models, ypically wih he represenaive agen and homogeneous goods, are no suiable for invesigaing such possibiliy. Forunaely, recen progress in he lieraure of new open macroeconomics has made i possible o incorporae richer rade srucure ino he analysis of no only long run bu also shor run effecs of changes in policies and produciviy. This paper develops a hree counry model wih hree ypes of producs, called high-ech radables, low-ech radables, and non-radables (each of which consiss of many varieies), which reflecs realisic rade srucure beween Asia, Japan, and he US. The lieraure of he new open economy macroeconomics is characerized by explici dynamic opimizaion, shor-run nominal rigidiy, and imperfec subsiuion beween producs. The las feaure of his approach enables us o analyze how he rade srucure beween counries influences inernaional ransmission of shocks. This kind of model can be applied o many imporan inernaional policy issues. For example, some economiss have argued ha he depreciaion of he Japanese yen since was parly responsible for riggering he Asian currency crisis in According o hose views, he yen depreciaion made Eas Asian producs much less compeiive in he 1 In April 1995, he yen was a he hisorically highes level of 1$= Since hen, he yen depreciaed rapidly, o 1$= in December. The yen coninued o depreciae, and reached he level of 1$= in April 1997, us before he beginning of he Asian currency crisis. In Augus 1998, i hi he lowes value since 3

4 global marke, and pu pressures on Asian counries o devalue heir own currencies. The heoreical framework offered in his paper is suiable for analyzing quaniaive imporance of such effecs. The res of he repor is organized as follows. Secion 2 provides an overview of he relaed lieraure. Secion 3 describes he basic heoreical framework. Secion 4 presens he model. Secion 5 presens he resuls of numerical simulaions. Secion 6 concludes. 2 Overview of he model The model considered in his paper builds on he framework of Corsei e al. (2000). Their model in urn is based on a muli counry equilibrium model of Obsfeld and Rogoff (1995 and 1996). In he Obsfeld-Rogoff model, each counry produces one ype of goods (which consiss of many varieies). In each counry, here are consumers who live for infinie number of periods. They decide oday s consumpion and labor supply so as o maximize heir life-ime uiliy, aking ino accoun he ineremporal budge consrain. The model is characerized by nominal rigidiy: Nominal prices are assumed o be se in advance, and says unchanged during one period. This means ha a pure moneary expansion could have real effecs and could change he uiliy level of he locals and foreigners. Corsei e. al. (2000) develop a hree counry version of he Obsfeld-Rogoff model. In heir model, each counry is specialized in he producion of us one ype of producs (each of which consiss of many varieies) and hose goods are raded inernaionally. Consumers live for infinie periods and maximize heir life ime uiliy. They do no face any borrowing consrain. Their preferences are assumed o be symmeric across counries, in he sense ha consumers in any counry spend he same fracion of heir expendiure on goods produced in a paricular counry. Firms are monopolisically compeiive and se nominal prices one period in advance. mid 1990 a 1$= (All he numbers are monhly averages.) 4

5 Shioi (2001) develops a modified version of his model and analyzes he welfare effec of a Japanese moneary expansion on Asia. He finds ha he overall welfare effec was posiive. Shioi (2002) generalizes his model significanly by incorporaing home bias in consumer preference and a fracion of agens ha are myopic (ha is, hey simply maximize heir periodic uiliy each period). He finds ha he welfare implicaion of he previous paper is weakened bu remains qualiaively similar. The assumpion ha each counry specializes in producion of us one ype of produc, however, may no be paricularly realisic. Some ype of goods produced by one counry may be beer subsiues for cerain ype of goods produced by anoher counry han anoher ype of goods produced by ha counry. For example, owels expored from China o Japan are probably beer subsiues for Japanese owels han, say, Japanese TV games. To beer reflec he realiy of he rade srucure, his paper abandons he one counry, one ype of goods specificaion. Insead, he model in his paper has hree ypes of goods ha are produced in all hree counries. They are called high-ech radables, low-ech radables, and non-radables. Counries differ in he relaive shares of each of hose hree ypes of producs in overall producion, consumpion, expors, and impors. The model inheris he imporan feaures of he model of Shioi (2002): (1) I allows for a possible asymmery in preferences across counries. For example, uiliy migh be characerized by home bias : spending shares may be higher for domesically produced goods han foreign goods. (2) I also incorporaes myopic (no forward looking) consumers who do no borrow or save. 2 The fracion of hose myopic agens is reaed as a parameer in he model. Models wih only forward looking consumers end o predic unrealisically srong responses of curren accouns in response o various shocks. Inroducion of myopic consumers makes curren accoun less responsive o shocks and is herefore appears o be more realisic. 2 I migh be more realisic o model hem as consumers who face borrowing consrains. However, i is more difficul o incorporae such agens, as heir behavior is asymmeric depending on which side of he borrowing consrain hey are in each period. 5

6 3 The Model The world consiss of hree counries, US (denoed by U), Japan (denoed by J), and Asia (denoed by A). Each counry is inhabied by a coninuum of households. The numbers of households in US, Japan, and Asia are all consan, and are denoed by γ U, γ J, and γ A, respecively. Time is discree and households live for infinie periods of ime. There is free flow of goods and bonds beween he counries. 3-1 Type of Goods Goods are classified ino hree ypes, called high-ech radables (denoed by subscrip H), low-ech radables (L), and non-radables (N). Those hree are imperfec subsiues. As he names sugges, H goods and L goods are raded inernaionally while N goods are consumed locally. Each of he hree counries produces all hree ypes of goods. Each ype of goods consiss of many brands, ha are imperfec subsiues beween each oher. Each household specializes in producion of us one brand of goods, over which i has a monopoly righ o produce. This means ha he number of brands produced is always equal o he number of households. There is no invesmen and all he goods are final consumer goods. We make an assumpion on he uiliy funcion so ha all he households decide o consume all brands of goods available o hem, ha is, all brands of radable goods as well as all non-radable goods produced in he counry hey live in. 3-2 Households In each period, each household obains uiliy from consuming a bundle of consumer goods. I derives disuiliy from working o produce is own brand of consumer goods. I also derives uiliy from holding real money balance. The one-period uiliy of he household x, ha produces ype k goods (k=h, L, or N) in counry in period is assumed o ake he following form: u 2 M ( ) ( x) Y + ( x) χ ln P κ ( x) = lnc ( x) 2 The firs par represens uiliy from consumpion. The variable C (x) (3-1) is a bundle of 6

7 consumer goods (or he composie consumpion index ) consumed by his household in period. The exac definiion of his index will be specified laer. The second par represens he disuiliy of work. The variable Y (x) is he amoun of oupu produced by his household in period, using labor as he sole inpu. The parameer (which is assumed o be posiive) describes how work effor is relaed o oupu: when is value is high, i means ha produciviy is low (more work effor is needed o produce he same amoun of oupu). The hird par corresponds o he uiliy from money holding, where M (x) in he uni of he local currency, while specified exacly laer. The parameer is he amoun of cash held by his household, denoed P budge consrain akes he following form: is he average price level of counry, o be χ is assumed o be posiive. The periodic E B ( x) M ( x) E B ( x) M (x) SR ( x) T ( x) + P P (3-2) C ( x) = ( 1+ i ) + + P P P P In he above, is he exchange rae of counry (=U, J, or A) in period. We shall ake he US dollar as he numeraire so ha T (x) =1. The oher exchange raes are defined as he value of a US dollar in he unis of local currency, so an increase in his variable means a depreciaion of he local currency agains he US dollars. he amoun of bond held by his household a he end of period, measured in US dollars. The nominal ineres rae ha accrues o holding his bond beween periods -1 and is denoed by, and his is also measured in he US dollars. The assumpion of free financial capial mobiliy implies ha his value will always be he same across he counries. SR E (x) is he revenue from sales of he goods produced by his household, defined in he unis of he local currency. In a flexible price equilibrium (long run), law of one price holds, and he sales revenue is equal o he price of his brand of goods charged by his monopolisically compeiive household (which will be denoed by In a fixed price equilibrium (shor run), he domesic price is fixed, while sales prices abroad vary depending on he pass-hrough rae beween he seller s counry and he buyer s counry. Finally, defined in he unis of he local currency. i U E is lump sum ax imposed by he governmen, also B 1( x) P (x) ), imes he quaniy of he goods sold world-wide ( SR ( x) = P ( x) Y ( x) ). + κ is 7

8 Also, noe ha, as a producer, each household faces a downward sloping demand curve, as differen brands of goods are assumed o be imperfec subsiues. Laer, we shall specify exacly how hose varieies of goods ener ino each household s uiliy. For he momen, i suffices o know ha, in a flexible price equilibrium (long run), each household faces he demand curve of he following kind: where θx Y ( x) P ( x) Z, (3-3) = θ x is a secor-specific consan larger han one, whose role in he uiliy funcion will be spelled ou laer. And of each household. Z is some variable ha is beyond he conrol I assume ha here are wo ypes of households, forward looking households and myopic ones. Forward looking ones maximize he following life ime uiliy: s= 0 s U ( x) = E β u ( x), (3-4) + s (where β is he subecive discoun facor) subec o he periodic budge consrain and a non-ponzi game condiion. Myopic ones simply maximize, period by period. This maximizaion is also subec o he same periodic budge consrain, hough i should be noed ha hey will opimally choose no o hold any bond a he end of each period, namely B (x) u (x) =0 for all,, and k. I will denoe he se of forward looking households as FL and ha of myopic households as NFL (for no forward looking). The populaion shares of each ype are fixed in each counry. I denoe he number of forward looking households ha produce ype k goods in counry π FL π NFL by and ha of non forward looking ones wih similar characerisics as. 3-3 Equilibrium condiions (forward looking households) Here, I will discuss equilibrium condiions ha have o be saisfied for forward looking households as a whole. For example, define he average consumpion of forward looking households producing ype k goods in counry in period as he inegral of C (x) over all x ha belongs o he forward looking group in he counry. Denoe such a variable as. Define,, and, in analogous ways for oupu, C FL Y FL M FL B FL 8

9 money holdings, and bond holdings, respecively. Then, by he assumpion of symmery wihin he forward looking group, we obain CFL = C (x), YFL = Y (x), M FL = M (x), BFL = B (x), (3-5) for all x FL,, k and. In equilibrium, he following hree condiions ha are derived from individual forward looking household s opimizaion condiions have o be saisfied a he aggregae level. Firs, he following Euler equaion has o be saisfied: C C P / E = β ( 1 i + 1) (for all,, and k). (3-6) P / E FL FL Second, he following money demand relaionship has o be saisfied: M P FL (1 + i ) E = χ CFL (for all,, and k). (3-7) (1 + i E ) E + 1 The previous wo condiions have o be saisfied a all imes. When prices are flexible, he following opimaliy condiion for he consumpion-leisure choice will have o be me as well: where P P FL FL, P θ κ = C θ 1 FL Y FL (for all,, and k), (3-8) is he average price index for he ype k goods produced and sold in counry by forward looking agens in counry (which will be equal o individual price P (x) for x FL, by symmery). 3-3 Equilibrium condiions (myopic households) Denoe average consumpion, oupu, money holdings and he price charged by myopic agens in heir own counry as,, and, respecively. Again, C NFL Y NFL M NFL P NFL, by he wihin-group symmery, consumpion ec. of individual household in his group is equal o hese group averages. In heir case, only he inra-emporal opimizaion condiions have o hold. Firs, M P NFL = χc (for all,, and k) (3-9) NFL has o always hold. Second, when prices are flexible, 9

10 P NFL, P θ κ = C θ 1 NFL Y NFL (for all and ) (3-10) also has o hold. 3-4 Equilibrium condiions (governmen) Nex, he governmen s budge consrain has o be saisfied in equilibrium. In his paper, i is assumed ha he governmen s only role is o prin money and o disribue i across households in a lump sum fashion. This implies: M M 1 + T = 0 (for all and ), (3-11) where and are money supply and ransfer, respecively, in counry in M period. I assume ha he governmen supplies he same amouns of money and ransfers o households wihin he same caegory, i.e., hose who produce he same ype of goods and have he same uiliy funcion (forward looking or no forward looking). Then, wriing such money supply and ransfers per capia o he forward looking group as M and T, and hose for he myopic group as and, we can wrie FL M FL T = π FLM FL + k k π NFL M NFL M NFL T NFL (3-12) and T π T π T. (3-13) = FL FL + k k NFL NFL 3-5 Equilibrium condiions (resource consrain) The aggregae resource consrain for counry can be wrien as: E ( B 1 = SR + i E B P C (for all and ), (3-14) + B ) where B, SR, and C are aggregae bond holding, sales revenue, and consumpion, respecively. Tha is, B = π FLBFL, (3-15) k SR = π FLSRFL + k k π NFL SR NFL (3-16) SR FL SR NFL (where and are sales revenue for forward-looking and myopic 10

11 households, respecively), and C π C π C. (3-17) = FL FL + k k NFL NFL The world wide ne supply of bonds has o be equal o zero: U J A B + B + B = 0 (for all ). (3-18) The amoun of oupu produced by each ype of household has o equal he demand for he good. Tha is, Y ( x) = D ( x) + D ( x) + D ( x) (for k=h or L, for all x, and ), (3-19a) for radable goods, U, J, A, and Y ( x) D, ( x) (for all x, and ), (3-19b) = for non-radable goods, where ( ), (, and ( are demand for oupu D U, x D J, x) D A, x) produced by household x in counry ha come from he US, Japan, and Asia, respecively. Those demands will be specified in deail laer. 3-6 Composie consumpion indices Now I move on o specify conens of each consumpion index. In his secion, ime subscrip is omied for he sake of exposiion. The overall consumpion index, C 1/ ρ ( ρ / 1/ ( ( ) = ( ( )) 1) ρ ρ C x ω C x + ω ( C ( x) ) where (x) C N (x), is assumed o ake he following form: C HL(x) ρ / /( 1) [ ] 1) ρ ρ ρ HL N N HL, (3-20) is iself a composie consumpion index of H goods and L goods, and is an index of non-radable goods consumpion. The parameer ρ is he elasiciy of subsiuion beween radable goods as a whole and non-radable goods, and ω s are he expendiure share parameers. The index 1/ ψ ( ψ / 1/ ( and ( ) = ( ( )) 1) ψ ψ C x ω C x + ω ( C ( x) ) C HL(x) ψ / /( 1) [ ] 1) ψ ψ ψ H H L L is defined as HL. (3-21) The parameer ψ is he elasiciy of subsiuion beween high-ech radable goods as and low-ech radable goods. Each of he above indices are hemselves composie consumpion indices. For example, in he case of high-ech radable goods, 11

12 where C H i, ( x) H /( H 1) 1/ H 1/ H 1/ H C ( ),, ( ),, ( ) H x = ω θ H U CH U x + ω θ H J CH J x + ω θ H, A CH, A( x) θ θ (3-22) θ H is he elasiciy of subsiuion beween brands wihin ype H goods, and (i=u, J, or A) is an index of consumpion of high-ech radable goods produced in counry i : C ( ( C ( zh i x ) H,, 1/ θ H H, i( x) = ωh, i ) z H, i θ H 1) / θ H (3-23) where summaion inside he brackes is aken over all he high-ech radable brands produced in counry i. Likewise, for low-ech radable goods, we define: L /( L 1) 1/ L 1/ L 1/ L C ( ),, ( ),, ( ) L x = ω θ L U CL U x + ω θ L J CL J x + ω θ L, A CL, A( x) θ θ ( ( L i ) L, 1/ θ L, i L, i ) z L, i θ 1) / θ, (3-24) L L L and C ( x) = ω C ( z, x. (3-25) For non-radable goods, ( θ ) θn / N 1 1/ θ ( 1)/ N θn θn N( ) = ωn ( N( N, )). (3-26) zn C x C z x 3-7 Price indices and demand funcions The above definiions of consumpion indices allow us o appropriaely define composie price indices. Also, we can derive demand funcions ha each household faces as a producer of goods. Those are summarized in he mahemaical appendix 1 (o be added laer). 4 Descripion of he Numerical Exercise 4-1 Dynamics of he Model In he following analysis, i is assumed ha he world economy sars from a flexible 12

13 price equilibrium wih consan money supply. I is also assumed ha all households had zero foreign bonds or deb a he ouse. All he counries are in he seady sae in which all he variables remain consan over ime. Then a permanen shock his he Japanese economy. In he shor run, here is price rigidiy: nominal prices quoed by he producers are suck a he previous levels in heir own counry. Prices in foreign markes migh sill change in response o flucuaions in he foreign exchange raes. Here, i is assumed ha he nominal exchange rae pass-hrough is no necessarily complee: foreign prices may no fully reflec changes in he exchange rae beween he seller s and he buyer s counry. On he oher hand, hose prices may also be influenced by flucuaions in he exchange rae beween he buyer s counry and he hird counry: for example, when prices of expors from Japan o Asia are quoed in US dollars, i is conceivable ha heir prices measured in he unis of he Asian currency migh change when he US-Asia exchange rae changes. In any case, as a consequence of he shor run price rigidiy, he world economy deviaes from he long run equilibrium. I is assumed ha, in he shor run, oupu is demand deermined. Afer one period, prices become fully flexible. The world economy arrives a a new flexible price equilibrium, which is likely o be differen from he old one. In a case wihou myopic households ( π = 1 for all ), he world economy will auomaically ump o he new long run equilibrium immediaely. This is he beauy of he approach of Corsei, e.al. (2000): i convers an infinie period model ino a virual wo period model, and researchers have o worry abou only he shor run (period 1) and he long run (period 2 onwards). This is no necessarily he case when myopic households are presen. Due o he asymmery in he demand for money (refer o equaions (3-7) and (3-9)), money holdings a he end of period 1 by forward looking and myopic households do no necessarily coincide heir new long run equilibrium levels. In such a case, here will be a ransiion o he new seady sae and he analysis would be far more complicaed. To avoid such complicaion, I inroduce he following governmenal re-disribuion policy. I assume ha, a he beginning of period 2, he governmen in each counry re-disribues money hrough lump sum ransfers so ha he amouns of moneary wealh held by each ype of households a 13

14 he beginning of period 2 would be equal o heir respecive long run values. In his case, he world economy will ump o he new long run seady sae immediaely, us as in he model wihou myopic households. This assumpion is admiedly arificial bu i simplifies he analysis enormously wihou alering he essenial aspecs of he conclusions. The effecs of he policy change are analyzed by log-linearizing he equilibrium condiions around he seady sae wih zero bond holding. As i is difficul o obain analyical resuls, I repor resuls from numerical exercises in he nex secion. 4-2 Calibraion The model is calibraed o fi characerisics of daa for he US, Japan, and Asia on producion and spending paerns, such as relaive secoral produciviy and secoral shares in expendiure. Daa for Asia is compued by aggregaing values for Hong Kong, Indonesia, Korea, Malaysia, he Philippines, Singapore, and Thailand (Taiwan is omied due o missing daa). In compuing secoral saisics from daa, I inerpre high-ech radables secor as he machinery (including ranspor equipmen) indusry, low-ech radables secor as agriculure, mining and manufacuring (oher han machinery), and non-radables secor as he res. The acual numbers employed are summarized in Table 1-3. Populaion World populaion is normalized o equal 1, and each counry s populaion is chosen o mach is acual share (among he hree economies) in he number of persons employed, as is shown in Table 1 3. Secoral allocaion of workers In he base-line case, oal populaion of a counry is allocaed o each secor so as o mimic acual secoral allocaion of labor in each counry in recen years. In he pas decades, Eas Asia has become one of he mos prominen areas for IT producion. In Table 1, his is refleced in he size of high ech radables employmen in Asia, which 3 Toal numbers and secoral allocaion of workers are esimaed by combining informaion from he Key Indicaors web sie of he Asian Developmen Bank and he INDSTAT CD-ROM (UNIDO). I use daa from year 2000 whenever available. 14

15 is comparable o ha of Japan and he US (hough smaller in erms of populaion shares). In he oher case, called he hisorical case, I consider Asia before he hisorical srucural change. Back hen, Eas Asia was predominanly agrarian and very lile high-ech indusries were presen. To sudy how his ransformaion of Asia has changed inernaional policy ransmission channel, I se he share of households allocaed o he high-ech secor o us 0.01% in his case. The difference in he shares beween he wo cases is allocaed o he low-ech secor (hink of agriculure). Produciviy The produciviy parameers in he las row of Table 1 are chosen o mach observed GDP per worker as well as daa on relaive secoral produciviy 4. Produciviy in he non-radables secor in Asia is normalized o be 1. Noe ha Asia s high-ech radables secor is much more producive han he oher wo secors, especially in comparison wih he low-ech radables secor. On he oher hand, GDP per worker is relaively similar across secors in he US and Japan. This means ha, in he model, Asia has a very srong comparaive advanage in high-ech radables secor. Uiliy Weighs The values of he expendiure share parameers, ω s, are chosen o equal acual spending shares of Asia, Japan, and US, summarized in he upper panel of Table Noe ha counries end o spend disproporionaely large shares of heir expendiure allocaed o radable goods on domesically produced radables (home bias). This paper s flexible specificaion of preference makes i possible o incorporae such feaures ino he model. An imporan excepion o his general endency of home bias 4 Labor produciviy is esimaed from combining informaion in World Developmen Indicaors 2002 CD-ROM wih ha in Key Indicaors and INDSTAT. 5 Oupu shares and expendiure shares in Table 4 and Table 5 are compued from he hree sources menioned in he previous foonoes and he COMTRADE web sie of he Unied Naions. 6 In compuing hose shares, I ignore rade wih he res of he world, such as EU and China. This has an inconvenien consequence ha he imporance of domesic consumpion in he relaive shares of spending is exaggeraed. Anoher minor problem wih his omission is ha expendiure shares do no exacly add up o 100%, as can be seen in he upper panel of Table 5. In he calibraion exercise, he share parameers are adused slighly so ha hey would always sum up o 100%. 15

16 is Asia s expendiure on high-ech radables. I purchases only a small fracion of high ech goods produced domesically, and buys far more high ech goods from abroad. This aspec of he daa is replicaed in he model by seing he uiliy weigh of Asian consumers on domesically produced high ech goods very low. Subecive Discoun Facor and he Uiliy Weigh on Money As is shown in Table 2, I se he subecive discoun facor a β = The parameer for money in he uiliy, χ, is somewha arbirarily se a 1. Elasiciies Assumpions on he elasiciies of subsiuion are summarized in Table 2. High-ech goods end o be highly differeniaed, and hus he wihin-ype elasiciy ends o be low. This idea is refleced in he small value of θ H. On he oher hand, low-ech goods and non-radable goods are assumed o be highly subsiuable wih he oher goods of he same ype. Share of myopic households Choice of his imporan parameer will be discussed in deail in he nex secion. Exchange rae regimes I is assumed ha all hree counries are under flexible exchange rae regimes. In fuure revision, I plan o sudy he case in which Asia fixes is exchange rae agains he US. Rae of nominal exchange rae pass-hrough I is difficul o deermine he exen of exchange rae pass hrough empirically. In his exercise, I assume ha hose raes are deermined by he shares of currencies used in rade beween each pair of wo economies. Those shares, esimaed from daa provided in he web sie of he Minisry of Finance of Japan, are presened in Table 3. For example, he able shows ha, in he oal value of expors from Asia o Japan, 2% is mediaed by Asian currencies, while he shares of he Japanese yen and he US dollars are 27% and 71%, respecively. In such a case, in he model, shor run prices of goods expored from Asia o Japan are assumed o increase by 0.02 imes he rae of depreciaion of he Asian currency agains he Japanese yen, plus 0.71 imes he rae of depreciaion of he US dollars agains he Japanese yen. Those pass hrough raes are assumed o be equal beween high-ech radables and low-ech radables. 16

17 Table 1: Parameer values for he calibraion exercise (A) Populaion and Produciviy (Secoral variables are lised in he order of high-ech, low-ech, non-radable.) Asia Japan US Populaion Populaion shares of secors (%) <Base-line case> 2.6, 49.8, , 18.6, , 12.4, 83.4 <Hisorical case> 0.01, 52.39, 47.2 Same as above Same as above Secoral Produciviy (square roo of 1/κ ) 2.90, 0.38, , 7.59, , 9.37, 8.82 Table 2: Parameer values for he calibraion exercise (B) Preference parameers Preference parameers: Discoun facor ( β ) 0.9 Uiliy weigh on money ( χ ) 1 Elasiciies: Beween radables and non-radables ( ρ ) 2 Beween high-ech and low-ech (ψ ) 2 Wihin high-ech ( θ H ) 3 Wihin low-ech ( θ L ) 10 Wihin non-radables ( θ N ) 10 Share parameers (ω s): Se o equal acual expendiure shares ha appear in he upper panel of Table 5. Share of myopic households: See secion 5. Table 3: Parameer values for he calibraion exercise (C) Value shares of currencies used for ransacion In he order of Asian, Japanese, and US currencies. To Asia To Japan To US From Asia - 2%, 27%, 71% 2%, 0%, 98% From Japan 3%, 48%, 49% - 0%, 16, 84% From US 0%, 0%, 100% 0%, 17%, 83% - 17

18 Table 4: Oupu shares, by ype of goods produced and by counry of desinaion Table 5: Expendiure shares, by ype of goods purchased and by counry of origin Daa ASIA o ASIA o JPN o USA oal ASIA from Asia from JPN from USA oal H 0.5% 2.7% 7.0% 10.2% H 0.5% 6.7% 4.9% 12.1% L 16.7% 3.8% 5.4% 25.8% L 16.7% 3.9% 3.1% 23.7% N 63.9% 63.9% N 63.9% 63.9% sum 81.1% 6.5% 12.4% 100.0% sum 81.1% 10.6% 8.0% 99.7% JPN o ASIA o JPN o USA oal JPN from Asia from JPN from USA oal H 1.5% 4.8% 2.2% 8.4% H 0.7% 4.8% 0.6% 6.0% L 0.9% 13.4% 0.7% 14.9% L 1.0% 13.4% 0.8% 15.1% N 76.6% 76.6% N 76.6% 76.6% sum 2.4% 94.8% 2.9% 100.0% sum 1.7% 94.8% 1.3% 97.8% USA o ASIA o JPN o USA oal USA from Asia from JPN from USA oal H 0.8% 0.4% 4.1% 5.3% H 0.9% 1.1% 4.1% 6.1% L 0.5% 0.5% 11.8% 12.9% L 0.7% 0.4% 11.8% 12.9% N 81.8% 81.8% N 81.8% 81.8% sum 1.3% 0.9% 97.7% 100.0% sum 1.6% 1.5% 97.7% 100.8% Model Seady Sae Daa Model Seady Sae ASIA o ASIA o JPN o USA oal ASIA from Asia from JPN from USA oal H 0.4% 3.1% 6.2% 9.6% H 0.4% 6.0% 5.2% 11.5% L 16.4% 5.4% 4.0% 25.8% L 16.4% 3.1% 4.4% 24.0% N 64.6% 64.6% N 64.6% 64.6% sum 81.4% 8.5% 10.2% 100.0% sum 81.4% 9.1% 9.6% 100.0% JPN o ASIA o JPN o USA oal JPN from Asia from JPN from USA oal H 1.4% 4.6% 1.6% 7.6% H 0.7% 4.6% 0.7% 6.0% L 0.7% 13.5% 0.3% 14.5% L 1.2% 13.5% 1.4% 16.2% N 77.8% 77.8% N 77.8% 77.8% sum 2.1% 96.0% 2.0% 100.0% sum 1.9% 96.0% 2.1% 100.0% USA o ASIA o JPN o USA oal USA from Asia from JPN from USA oal H 0.6% 0.4% 3.9% 4.9% H 0.8% 0.9% 3.9% 5.6% L 0.5% 0.8% 11.9% 13.2% L 0.5% 0.2% 11.9% 12.6% N 81.8% 81.8% N 81.8% 81.8% sum 1.2% 1.1% 97.7% 100.0% sum 1.2% 1.1% 97.7% 100.0%

19 4-3 Seady Sae of he Model I firs derive values of various shares and raios in he iniial seady sae wih zero bond holding for he base-line case. By comparing hose wih acual saisics, we can sudy how closely he model replicaes he acual paerns of producion and spending. Firs, in Table 6, I compare acual produciviy (relaive o Asia and relaive o non-radables secor) wih he produciviy prediced by he model. I can be seen ha he model follows he acual paerns fairly closely. Table 6: Relaive produciviy, acual and seady sae ACTUAL Asia Japan US GDP per capia (Asia=1) H relaive o N L relaive o N MODEL Asia Japan US GDP per capia (Asia=1) H relaive o N L relaive o N Nex, he lower panel of Table 4 repors he model s predicion for he secoral composiion of goods produced in each counry as well as where hose goods are sold o. Those values can be compared wih he acual numbers presened in he upper panel of he same able. Also, he lower panel of Table 5 displays he prediced secoral composiion of expendiure on various ypes of goods as well as where he goods come from. Those numbers can be conrased wih he acual ones shown in he upper panel of he same able. In general, he model replicaes he acual paerns very well. 5 Main findings 5-1 Effecs of Japanese Moneary Expansion in he base-line case Before moving ono deailed analysis of he numerical resuls, I will invesigae how he resuls are sensiive o differen assumpions abou he share of myopic households. Suppose ha, in he base-line case, here was a once-and-for-all moneary expansion in Japan, which increases is money supply by one percen. Figure 1 plos shor run responses of curren accouns of he hree counries, measured as percenages of he 18

20 original levels of GDP of respecive counries, under differen assumpions on he populaion share of myopic households (he share is assumed o be equal across he secors and across he counries). When he share is only 1%, Japanese curren accoun increases by almos 0.2% (of GDP), while ha of Asia declines by abou 0.4%, and ha of he US decreases by 0.06%. These reacions seem oo large, considering ha hey are responses o us 1% increase in money. As he share of myopic households increases, hese reacions become weaker. When he share reaches 99%, he response of Japanese curren accoun is only 0.01%. As we lack obecive crieria o choose an appropriae value for his share, in wha follows, I will simply se his value equal o an inermediae value of 0.5 for all he secors and he counries. 0.3 Figure 1: Effecs of Japan's Moneary Expansion on Curren Accouns curren accoun as % of GDP Asia Japan US -0.5 share of myopic households 5-2 Effecs of Japanese Moneary Expansion in he base-line case (Coninued) Table 7 summarizes effecs of a one percen increase in money supply in Japan on imporan variables in hree counries, under he base-line case in which Asia is assumed o be reasonably high-ech. All he numbers are percenage changes (wih he excepion of he curren accoun, which is denoed as a percenage of he original level of GDP). 19

21 Table 7 Effecs of a one percen increase in money supply in Japan, A. Shor Run Base-line case (Asia is high ech). Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Toal High-ech Goods Low-ech Goods Curren Accoun B. Long Run Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Toal High-ech Goods Low-ech Goods Curren Accoun There is a srong yen depreciaion righ afer he moneary expansion. This creaes a big boom in expors, and oupu expands in Japan. Noe ha low-ech expors respond more srongly: his is because he wihin-ype elasiciy of subsiuion is assumed o be much higher for low-ech goods han for high-ech goods. Likewise, loss of expors in Asia comes more from he low-ech secor han he high-ech secor. In he long run, as Japan can enoy ineres paymens on is foreign bonds ha i accumulaes during he shor run, households work for less hours and hus oupu and expors conrac. 5-3 Effecs of Japanese Moneary Expansion in he Hisorical Case Table 8 summarizes effecs of he same shock under he hisorical case in which Asia is assumed o be predominanly agrarian. 20

22 Table 8 Effecs of a one percen increase in money supply in Japan, A. Shor Run Hisorical case (Asia is low ech). Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Toal High-ech Goods Low-ech Goods Curren Accoun B. Long Run Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Toal High-ech Goods Low-ech Goods Curren Accoun Noe ha, compared o he base-line case, Asian expors decline more srongly in response o a yen depreciaion in he shor run. This is because, in his case, Asia is more specialized in exporing low-ech goods, whose markes are more compeiive (he wihin-ype elasiciy of subsiuion is higher). As a consequence, curren accoun of Asia deerioraes more srongly and oupu declines more. Thus, comparing he wo cases, i can be seen ha he advance of high-ech, more differeniaed secors in Asia has conribued o parial insulaion of shocks from Japan. 5-4 Effecs of a Produciviy Increase in Japan Nex, consider wha happens when he overall produciviy in Japan increases by 1% (ha is, is κ declines by 0.5%). Table 9 repors he resuls for he base-line case, and Table 10 repors hose for he hisorical case. As is ypically he case wih his class of models, a permanen produciviy increase causes a shor run reducion in oupu in he counry ha experiences he produciviy 21

23 surge. As oupu is demand deermined in he shor run, beer produciviy does no simulae producion immediaely. A he same ime, Japanese households (forward looking ones) perceive ha heir permanen income has become higher, so hey increase consumpion and reduce work effor. This resuls in a emporary curren accoun defici in Japan, which is accompanied by an appreciaion of he yen. This causes expors and oupu of he oher wo counries o rise in he shor run. Comparing he wo ables, i can be seen ha he shor run expansion of Asian oupu and expors, as well as is improvemen in curren accoun, are all sronger under he hisorical case. Thus, again, we find ha he recen rise of he high-ech secor in Asia has played he role of a parial sheler from shocks ha originae in Japan. Table 9 Effecs of a one percen produciviy increase in Japan A. Shor Run Base-line case (Asia is high ech). Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Curren Accoun B. Long Run Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Curren Accoun Table 10 Effecs of a one percen produciviy increase in Japan A. Shor Run Hisorical case (Asia is low ech). Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Curren Accoun

24 B. Long Run Asia Japan US Exchange rae GDP in cons. US $ Expors in cons. US $ Curren Accoun Conclusions This paper has developed a new macroeconomic model for analyzing policy effecs of Japan on Asian economies. The model is rich enough o incorporae various feaures of indusrial (as well as rade) srucure in Asia, Japan, and he US. In paricular, i has been shown ha he emerging high-ech secor in Eas Asia has alered he ransmission mechanism of Japanese policy on Asia subsanially. In a fuure version of he paper, I will explore he possibiliy of incorporaing inermediae goods and FDI ino his analyical framework. References Corsei, Giancarlo, Paolo Peseni, Nouriel Roubini and Cédric Tille (2000). Compeiive devaluaions: oward a welfare-based approach. Journal of Inernaional Economics 51(1), pp Obsfeld, Maurice and Kenneh Rogoff (1995). Exchange rae dynamics redux. Journal of Poliical Economy 103, pp Obsfeld, Maurice and Kenneh Rogoff (1996). Foundaions of inernaional macroeconomics. Cambridge, MA: MIT Press. Shioi, Esuro (2001). Welfare implicaions of he yen depreciaion on Asia. EMEAP Sudy on Exchange Rae Regimes (hp:// Shioi, Esuro (2002). Consumpion Smoohing, Home Bias in Preferences, and Welfare Effecs of a Yen Depreciaion on Asia. Mimeo. 23