The Nature of Aggregate Demand and Supply Shocks in ASEAN Countries

Transcription

1 MPRA Munich Personal RePEc Archive The Naure of Aggregae Demand and Supply Shocks in ASEAN Counries Omar H M N Bashar Deakin Universiy. December 9 Online a hp://mpra.ub.uni-muenchen.de/19881/ MPRA Paper No , posed 11. January 1 17:16 UTC

2 The Naure of Aggregae Demand and Supply Shocks in ASEAN Counries Omar HMN Bashar * Absrac: This paper revisis he issue of idenificaion of macroeconomic shocks in ASEAN counries using an alernaive idenificaion scheme where he aggregae demand and supply shocks are allowed o be correlaed. Applying he echnique of Cover e al (6) wihin a bivariae Srucural VAR model, his paper showed ha aggregae demand and supply shocks are inerrelaed (posiively) in ASEAN counries. Unlike previous sudies, i is found ha changes in he oupu level are mainly driven by aggregae demand shocks, whereas supply shocks are playing he dominan role in affecing inflaion in ASEAN counries. The correlaions of he shocks across he counries are found o be quie small, suggesing ha ASEAN is sill no ready o form a common currency union like Europe. Keywords: ASEAN, Aggregae Demand and Supply, Srucural VAR, Blanchard-Quah Decomposiion JEL Codes: C3, E31, E3 * School of Accouning, Economics and Finance, Deakin Universiy, Ausralia 1

3 The Naure of Aggregae Demand and Supply Shocks in ASEAN Counries 1. Inroducion For he purpose of designing and implemening appropriae macroeconomic policies, i is essenial o idenify various shocks and analyse heir impacs on he economy. In paricular, in he wake of recen global recession, i is ime o reassess he convenional beliefs abou he naure of macroeconomic shocks ha are playing crucial roles in an economy. This paper revisis he issue of idenificaion of macroeconomic shocks in ASEAN counries, using an alernaive idenificaion scheme where he aggregae demand (AD) and aggregae supply (AS) shocks are allowed o be correlaed. The origin of he recession lies in he adverse aggregae demand shock and he persisence of curren recession essenially depends on how he demand shocks affec an economy. If he demand shock has only emporary effecs on he employmen and oupu level, hen we may expec an economy o recover quickly, even wihou any demand managemen policy. The main quesion hen is: Do aggregae demand shocks affec he oupu level permanenly? Convenional macroeconomic analysis suggess No, and as a maer of fac, many previous prominen empirical sudies were based on he assumpion ha he aggregae demand shock has only a emporary effec on oupu. The classic paper by Blanchard and Quah (1989) firs applied his assumpion wihin he SVAR framework in order o isolae a demand shock from he supply shocks. To dae we can find a huge amoun of lieraure ha applied he Blanchard-Quah (BQ, henceforh) approach for he purpose of idenifying macroeconomic shocks. The main idea is ha he long run aggregae supply curve is verical, and a shif in he aggregae demand curve will increase he inflaion (or he price level) proporionaely, bu will no aler he oupu level. The BQ echnique furher assumes ha he shocks are uncorrelaed. Recenly Cover e al (6) quesioned his assumpion, and using he US daa, hey showed ha his assumpion leads o a complee isolaion in he dynamics of inflaion and oupu. Wih demand and supply shocks being uncorrelaed, changes in oupu were found o be driven mainly by he

4 supply shocks and inflaion by he demand shocks. Cover e al (6) showed ha his finding can be reversed if we allow he supply shock o be affeced by he demand shock. The main purpose of he curren paper is o idenify he aggregae demand and supply shocks in ASEAN region using he echnique used in Cover e al (6). Looking beyond he US daa o a se of Souheas Asian counries will provide us more robus conclusions regarding he dynamics and inerrelaion of he shocks. Moreover, he examinaion of he shocks in ASEAN counries will help us analyse he prospecs of forming a Euro-syle currency union in he Souheas Asian region. Essenially, he idenificaion scheme used in his paper allows he case where a shif in he aggregae demand curve induces he long run aggregae supply curve o shif. The main quesion, however, remains: can an aggregae demand shock change he poenial level of oupu, which in urn would shif he long run aggregae supply curve? The answer is yes, if he aggregae demand shock affecs echnology and produciviy. There indeed exis a few significan lieraures ha draw aenion o he role of demand in affecing innovaion in echnology in he producion process and showed ha an aggregae demand shock ha raises he level of oupu emporarily can, hrough a number of channels, exer a permanen influence on he supply side. For example, Sadler (199) argues ha.changes in he uilizaion of facor inpus when demand changes can resul in reorganizaion and he acquisiion of new skills; or a higher level of oupu may make innovaion more profiable and resul in he allocaion of more resources o R&D. Wih he use of heoreical models wih endogenous echnology, his sudy essenially showed ha Changes in he supply side of he economy are no independen of changes on he demand side. The empirical work of Uerback (1974) also poins o he same idea, which showed ha majoriy of he innovaions in indusry akes place in response o marke demand condiions. Lucas (197, 1973) showed how an unanicipaed change in money (an aggregae demand shock) could affec oupu. This effec occurs because firms misperceive he aggregae demand shock for he relaive demand shock. However, his effec should no be persisen, as firms would have perfec informaion abou he demand shock over ime. Blanchard (1989) surveyed he lieraure on he effecs of money on oupu and observed ha if he iniial mispercepion abou he shock led firms or workers o change a sae variable, which affec heir decision in subsequen periods, hen he iniial changes in money or demand would have persisen effec. For insance, if he iniial mispercepion leads he firms o inves in 3

5 produciviy enhancing echnology, hen we may expec a shif in he long run aggregae supply curve, which in urn will change he oupu level permanenly. The induced shif in he long run aggregae supply curve can also be explained by he concep of hyseresis in unemploymen 1. The main idea is ha he cyclical movemen in unemploymen rae, which is mainly caused by demand shocks, may in fac cause he naural rae of unemploymen o change. As he naural level of unemploymen rae is relaed o an economy s long run employmen, his in urn will shif he long run aggregae supply curve. Ball (1999), using he daa from OECD counries showed ha moneary policy and oher deerminans of aggregae demand have long run effecs on unemploymen. This implies ha an aggregae demand shock may cause a supply shock by affecing he naural level of unemploymen. There are a few previous sudies ha aemped o analyse he naure of macroeconomic shocks in Souheas and Eas Asian counries. Examples include Bayoumi and Eichengreen (1994) and Ng (). These papers basically uilized basic BQ decomposiion in order o idenify he shocks and found ha each ype of shocks in differen counries in Souheas and Eas Asia are highly correlaed, in general. The main drawback of hese sudies is ha hey did no ake ino accoun he possibiliy ha he aggregae demand and supply shocks can be inerrelaed. Thus he findings of hese papers can be misleading if he shocks are acually correlaed. For example, if he demand shocks invie some shocks o aggregae supply, hen he idenified supply shocks in BQ model would conain boh he demand and supply componens. And if his is he case we may see some ambiguiy in he effecs of an idenified supply shock on inflaion (as demand and supply shocks affec he inflaion rae in opposie direcions). In he curren paper I have used a bivariae SVAR model wih long run resricion. However, unlike BQ sraegy, I assumed ha aggregae demand shock does no have any long run effec on oupu only when he shifing of aggregae demand curve does no induce he long run aggregae supply curve o shif. In oher words, following Cover e al (6), I have allowed he causaliy running from demand o supply shocks while imposing he long run neuraliy resricion. Having been idenified he shocks; I have shown ha an aggregae demand shock may have a permanen effec on he oupu level in Souheas Asian Counries. I have also 1 This erm was firs used formally by Blanchard and Summers (1987) 4

6 shown ha his sraegy can help explain he anomaly in he effec of supply shocks on inflaion observed in he BQ model. Res of his paper is organised as follows. In Secion, basic resuls in he BQ model are repored. The daa descripion and a few economeric issues involving VAR analysis are also explained in his secion. In Secion 3, using he AD-AS model inroduced in Cover e al (6), an aemp is made o explain how his model can be convered o a bivariae SVAR model where he demand and supply shocks are allowed o be correlaed. Then he resuls of his modified model are discussed. Secion 4 concludes he paper.. BQ Model and he Oupu and Inflaion Dynamics in ASEAN Counries.1 The BQ Approach Consider he following bivariae Srucural VAR Model: AX = A 1(L)X + Bε (1) where X (, ), and y and are measures of real oupu and inflaion, respecively. y s d s d Here ε (, ), wih and being one sandard deviaion supply and demand shocks, respecively. A is a x marix, q i 1 A1iL shows marices of lag coefficiens of he i= 1 A (L) SVAR sysem. The BQ approach assumes ha wo shocks are no correlaed, and hence, B is a diagonal marix. Le us denoe he diagonal elemens in B by b 11 and b, which essenially are he sandard deviaions of he wo shocks. The srucural shocks in (1) are no direcly observable. I is he usual pracice o esimae he reduced form VAR and use he esimaed parameers and residuals in he reduced form VAR o rerieve he srucural shocks. The reduced form VAR has he following form X = C(L)X + e () Here y and are assumed o be I(1) and hence he VAR model uses firs differences of he respecive variables. 5

7 wih q C(L) i=1 i CiL being he marices of esimaed lag coefficiens and e being he vecor of wo residual series. Equivalenly, his reduced form VAR can be expressed in more simple way as q q i i y 11 i 1 i i1 i1 y c y c e q q i i c1 y i c i e i1 i1 (3) The relaion beween he srucural shocks and he reduced form VAR-residuals is crucial in idenifying he srucural shocks. This relaion can be expressed as e = G ε (4) where -1 G = AB is a x marix represening he conemporaneous effecs of he one sandard deviaion shocks on he wo variables. I is sraighforward o see ha for above specificaions, one canno make a disincion beween he supply and demand equaions, and hence he wo shocks, unless we impose some qualificaion(s) for defining he shocks. In order o separae ou he demand shocks from he supply shocks, he BQ mehod defines a demand shock as he one ha does no have any long run effec on he oupu level. If we denoe G as g11 g 1 G, g1 g hen he long run resricion essenially implies ha g c i 1 i 1 g i 1 c i (5) Imposiion of his resricion makes he srucural VAR sysem exacly idenified, and one will now be able o idenify he srucural shocks, he esimaed reduced form VAR. s and d, by using he informaion from 6

8 . Daa Descripion and VAR Specificaion Taking ino accoun he availabiliy of daa, I have considered 5 ASEAN counries for he SVAR analysis. They are Indonesia, Malaysia, The Philippines, Singapore and Thailand. Bivariae SVAR models have been esimaed separaely for each of hese 5 ASEAN counries. The variables included in he VAR models are: dly = Growh rae of real GDP. This variable is defined as dly = 1[ln(y ) ln(y -4 )], where y denoes real GDP volume ( =1). dinfl = Firs difference of he log of inflaion rae. The inflaion rae is defined as infl = 1[ln(P ) ln(p -4 )], where P denoes he Consumer Price Index ( = 1). The variable dinfl is defined as dinfl = infl - infl -1. The samples range from 1981:Q1 o 8:Q4. DaaSream, IMF and Abeysinghe and Rajaguru (4) are he main sources of daa. 3 As he GDP daa exhibis high seasonaliy, I applied Census X1 mehod on y o de-seasonalize his series. ADF uni roo es has been applied on ln(y) and infl (no repored here). For each of he counries, he es resuls suggesed ha we canno rejec he null of uni roo unambiguously for ln(y) and infl. However, firs differences of hese variables have been found o be saionary 4. To induce saionariy, I herefore use he firs difference forms of ln(y) and infl in VAR models. The VAR lag orders used for each counry are provided in Table 1. To decide on he lag order, I looked a LR, AIC, SC, FPE and HQ crieria for up o 16 lags. As usual, hese crieria suggesed differen lag orders. In general, SC suggess he shores lag order, whereas AIC suggess he longes lag orders. From he suggesed specificaions by differen crieria, I have 3 For Malaysia, Indonesia and hailand, quarerly raw daa for much earlier periods are no available. I hus depended on disaggregaed quarerly series esimaed in Abeysinghe and Rajaguru (4) for hese daa. 4 The finding ha inflaion is a nonsaionary variable is somewha conrary o he general percepion ha inflaion is mean revering. The inuiion behind he nonsaionariy of he inflaion rae can be jusified by he heory of opimal collecion of seigniorage proposed by Mankiw (1987). According o his heory inflaion is one kind of ax on holding money balances. One would expec ha an opimal fiscal policy involves smoohing of he ax raes over ime. Jus as he smoohing of consumpions by he consumers makes consumpion a random walk, he smoohing of ax rae by he Governmen makes ax rae a random walk (Mankiw 1987). Applying he same principle, he heory of opimal seigniorage suggess ha inflaion should be smoohened, and such smoohing essenially makes his series approximaely a random walk. 7

9 chosen he lowes lags ha ensures he propery of serially uncorrelaed residuals and his propery has been verified mainly by examining he auocorrelograms of he residuals. Table 1: VAR lag Specificaions Counry VAR Lag Orders Indonesia 9 Malaysia 4 The Philippines 1 Singapore 5 Thailand 11.3 Impulse Response Funcions in he BQ Model The dynamic impacs of oupu and inflaion o a posiive one sandard deviaion demand and supply shocks in each of he 5 ASEAN counries are shown in figure 1. As he VAR models use firs-difference forms of oupu levels and inflaion raes, here I have compued he accumulaed responses of he changes in wo variables o he shocks in order o see he dynamics of he variables in heir levels. By consrucion, he oupu response in each counry approaches o zero wihin a few quarers of he demand shocks in each counry. In all he counries, he supply shocks appear o cause permanen increases in he oupu levels. In line wih he convenional view, he demand shocks appear o have a permanen effec on he inflaion rae in each of he 5 counries considered. The main anomaly, however, is observed in he responses of he inflaion raes o he posiive supply shocks. Excep for Indonesia, we observe ha a posiive supply shock causes he inflaion rae o increase (insead of o decrease) in he long run. Indonesia is he only counry which can jusify he BQ decomposiion in explaining he effecs of an aggregae supply shock on he inflaion rae. For res of he counries, his anomaly migh have been caused by he fac ha he idenified supply shock in BQ model conains he demand componen. In oher words, par of he supply shocks (which is supposed o be an independen shock) is induced by a demand shock, and as posiive demand shocks unambiguously increase he inflaion rae, he ne effec will depend on he relaive magniudes of he effecs of demand and supply shocks. In he nex Secion i has been shown how we can resolve his anomaly by allowing he demand and supply shocks o be correlaed. 8

10 Figure 1: IRF s in he BQ Model: Oupu Response Inflaion Response Supply Shock Demand Shock Indonesia: 1 Supply Shock Demand Shock Supply Shock Demand Shock Malaysia: Supply Shock Demand Shock Supply Shock1 Demand Shock Philippines: 1..5 Supply Shock Demand Shock Singapore: Supply Shock Demand Shock Supply Shock Demand Shock Thailand: Supply Shock Demand Shock Supply Shock Demand Shock

11 3. Alernaive Idenificaion 3.1 The Model The idenificaion scheme used here is based on a simple AD-AS model 5 : y E [ y ] ( E [ ]) v s s 1 1 y E [ y ] v y d d d 1 s y d (6) Here, as before, y and are measures of real oupu and inflaion, while supply and demand shocks, respecively. s v and d v are he E 1 represens he expeced value of a variable in period, given he se of informaion available a he end of period (-1). The superscrips s and d represen supply and demand respecively. The firs equaion is a version of Lucas Supply Curve, where real oupu depends on is expeced value, unanicipaed inflaion and a random supply shock. The second equaion represens a simple aggregae demand funcion, where aggregae demand (expressed in money erm) consiss of is expeced value and a random aggregae demand shock. The hird equaion represens he equilibrium condiion. s s d d We can express he AS and AD shocks as v b11 and v b, wih uni variance and b 11 and b being he sandard deviaion of acual shocks, s and s v and d having d v. These expressions of shocks assume ha he wo shocks are uncorrelaed o each oher. If we allow supply shocks o be affeced by he demand disurbances, hen he wo shocks can be represened as: v ( b ) b v s d s 11 d d b (7) 5 This model is similar o he one used in Cover e al (6). 1

12 s Now he AS shock has wo componens; b11 is he independen AS shock, whereas ( b d ) is he induced (by an independen AD shock) AS shock. The coefficien measures he exen o which an AS shock is influenced by an AD shock. If we assume ha he expeced value of oupu and inflaion are jus he acual respecive values in he immediae las period; i.e.; if E 1[ y ] y 1 and E 1[ ] 1, hen he above model (6) can be modified o y ( b ) b y b d s 11 d In general, he acual and expeced values of oupu and inflaion also depend on he sequences of pas realizaions of hese variables. We can hus ransform he above model o a sandard Srucural VAR form as q y y ( b ) b yi i i i i1 i1 q q d y yi y i i i b i1 i1 q d s 11 (8) Having been specified he Srucural VAR in (8) and he reduced form VAR in (3), i is now sraighforward o show he following relaionship beween he reduced form VAR residuals y ( e, e s d ) and he uni variance srucural shocks (, ): y 1 s e 1 b11 b e 1 1 b d (9) which can be simplified o obain: e 1 b ( ) b y s 11 d e 1 b11 (1 ) b (1) The variance-covariance marix of he reduced form VAR residuals can be defined as 11

13 b11 ( ) b b11 b11 y y var( e ) cov( e, e ) y (11) cov( e, e ) var( e ) b11 (1 ) b ( ) b (1 ) b The hree elemens in he lef hand side of (11) can be obained from he esimaed reduced form residuals. However, here are four elemens in he righ hand side o be rerieved from hese hree lef hand side elemens. Technically ha is how he idenificaion problem arises and we need o impose a resricion o solve his idenificaion problem. This addiional resricion comes from he BQ long run resricion, which in his case saes ha he aggregae demand shock does no have any long run effec on he oupu level as long as his does no induce he long run aggregae supply curve o shif. Following Cover e al (6), wih reference o Srucural VAR sysem (8) and reduced for VAR sysem (3), his resricion essenially implies ha i c1 i 1 c i i (1) As a maer of fac, we are assuming a verical long-run aggregae supply curve, however, we are no ruling ou he case in which he long-run aggregae supply curve may shif due o a shif in aggregae demand curve. Now he value of from (1) can be subsiued in (11) o solve for he remaining hree parameers. 3. Esimaed parameers Based on (1), I have compued from he esimaion of reduced form VARs for each of he 5 counries. Here is a measure of slope of he shor run aggregae supply curve. The higher he, he larger will be he effec of changes in inflaion on aggregae supply in he shor run (A flaer AS curve). Table 3 below repored he compued values of in 5 ASEAN counries 1

14 Table 3: Compued Slopes of he Shor Run Aggregae Supply Curve Indonesia Malaysia Philippines Singapore Thailand Excep for Singapore, he compued is very small, suggesing a relaively seeper shor run aggregae supply curve for hese counries. In Singapore, is greaer han 1, which suggess ha an aggregae demand shock would have larger effec on oupu and smaller effecs on inflaion in he impac period, provided ha AD shock is no correlaed wih he supply shock. However, his need no be he case in our modified model, as we have allowed for correlaion beween he demand and supply shocks. Table 4 shows he esimaes of he res of he hree coefficiens for each of he 5 ASEAN counries. Excep for Indonesia, i is observed ha is significanly posiive, implying ha a demand shock affecs he supply shocks in he same direcion. The esimaes of in hese four counries range beween.4 and.81. A shif in he aggregae demand curve causes he larges shif in he aggregae supply curve in Malaysia and smalles shif in he Philippines. Accordingly, remaining oher hings same, we may observe a larger effec of AD shock on he oupu level in Malaysia, and relaively smaller effec in he Philippines. Table 4: Esimaed Coefficiens γ b 11 b Malaysia.81 (4).7 (5) 1.6 (.11) Singapore.6 (5) 1.1 (7) 1.99 (.14) Thailand.69 (4) 1.7 (7) 1.9 (.13) Philippines.4 (6) 1.1 (7) 1.37 (.1) Indonesia -1 (7) 1.53 (1.11).1 (.16) Noes: Figures in parenheses are relevan sandard errors. Indonesia is he only counry for which we observe a negaive coefficien 6 (hough insignifican). However, his negaive sign, o some exen, helps us explain he response of inflaion o a supply shock in Indonesia observed in he BQ model. I can be recalled ha Indonesia exhibied he righ (negaive) long run response of inflaion o a posiive supply shock. In BQ model, we didn see any anomaly for Indonesia eiher because he correlaion 6 I is hard o explain he negaive correlaion beween demand and supply shocks. I may be quie ineresing o analyse he case of Indonesia individually o explore his issue furher. One possible explanaion could be ha during he recession, economic agens re-evaluae heir pas decisions and misakes, and based on his, adop beer organizaional and managerial sysem, which may in urn increase he produciviy in he long run. 13

15 beween he AD and AS shocks is insignifican or simply because boh posiive supply and negaive demand shocks work in he same direcion in affecing inflaion. The sizes of he shocks as measured by heir respecive sandard deviaions are found o be quie large, compared o wha were claimed in he previous sudies on ASEAN. The sandard deviaion of he aggregae demand shocks is found o be higher han ha of aggregae supply shocks, in general. The mean of he aggregae demand shock is 1.8, whereas he corresponding figure for he aggregae supply shocks is only 1.6. This finding suggess ha a demand shock is much more volaile han a supply shock in ASEAN counries. Indonesia is appeared o be facing he larges shocks, as his counry has he larges sandard deviaions for boh AS and AD shocks. In erms of boh AD and AS shocks, Singapore and Thailand are exposed o he shocks ha are of similar size. The size of he AS shock in he Philippines is also similar o ha of Singapore and Thailand. Malaysia and he Philippines are experiencing he smalles AS and AD shocks, respecively. 3.3 Impulse Response Funcions and Variance Decomposiions Figure and 3 below show he impulse responses of oupu and inflaion o a one sandard deviaion aggregae demand and supply shocks, respecively. The 95 percen confidence inervals of he responses are also shown in he figures. The responses of oupu o he supply shocks and inflaion o he demand shocks are qualiaively similar o hose in BQ model. A posiive supply shock has permanen effec on oupu and a posiive demand shock has a permanen posiive effec on inflaion. However, compared o he BQ model, he responses of oupu o demand shocks and inflaion o supply shocks are quie differen in his modified model. For majoriy of he counries, we observe ha an aggregae demand shock leads o a permanen shif in he oupu level. This is in line wih our earlier posulaion ha suggess ha a posiive aggregae demand shock induces he long run aggregae supply curve o shif righward and causes he oupu level o increase permanenly. Again Indonesia shows he almos zero response of oupu o a demand shock, which, given our insignifican esimae of, is no surprising. The mos sriking conribuion of his modified model is visible in he response of inflaion o he aggregae supply shocks. The poin esimaion of long run response of inflaion o he 14

16 aggregae supply shock is negaive for majoriy of he counries. The poin esimaes of he long run responses of inflaion o a supply shock in he Philippines are sill posiive, however, he responses are insignifican and much smaller compared o hose in he BQ model. The posiive response of inflaion o a posiive supply shock, which we observed for Singapore and Thailand in he BQ model, has disappeared in his modified model. The negaive long run response is even significan in Malaysia and Singapore. This finding suggess ha his alernaive idenificaion scheme is able o isolae he demand and supply shocks in some wha beer way han wha he BQ model does. 15

17 Figure : IRFs o a one Sandard Deviaion Aggregae Demand Shock Oupu Response Inflaion Response Indonesia: Malaysia: Philippines: Singapore: Thailand: Noe: The solid lines give he esimaes of he accumulaed responses of he changes in respecive variables, while he doed lines represen he boosrap 95 percen confidence bound. 16

18 Figure 3: IRFs o a one Sandard Deviaion Aggregae Supply Shock Oupu Response Inflaion Response Indonesia: Malaysia: Philippines: Singapore: Thailand: Noe: The solid lines give he esimaes of he accumulaed responses of he changes in respecive variables, while he doed lines represen he boosrap 95 percen confidence bound. 17

19 The variance decomposiion resuls repored in Table 5 sugges ha aggregae demand shock has larger conribuion in he variaion of changes in oupu han wha were repored in mos of he sudies ha applied BQ decomposiion. Cover e al (6) has also poined ou o his issue and repored he similar resuls for he USA. As can be seen in able 5, in Malaysia, Singapore and Thailand demand shocks can explain larger share of he error variances in oupu flucuaions. In he Philippines boh supply and demand shocks conribued quie a subsanial porion of he variaions in oupu. Again Indonesia is he only excepion, where a supply shock is he dominan facor in explaining he oupu error variance. The variance decomposiions of he changes in inflaion sugges ha he supply shocks accoun for larger variaions in he changes in inflaion in all he 5 ASEAN counries. Table 5: Variance Decomposiions in he Modified Model Indonesia Malaysia Philippines Singapore Thailand Variance Decomposiion Variance Decomposiion Period of dly due o of dinfl due o Supply Demand Supply Demand Shock (%) Shock (%) Shock (%) Shock (%) Table 6 repors he simple correlaions of he AD shocks in 5 ASEAN counries. In general, he correlaion coefficiens are considerably small compared o wha were repored in similar previous sudies on ASEAN counries. In erms of saisical significance, one may esablish some kind of relaion beween he demand shocks in Malaysia, Philippines, Thailand and Singapore. However, he larges coefficien obained is only.31, and based on hese small numbers, i canno be concluded ha he aggregae demand shocks are synchronized in he ASEAN counries. Table 7 repors he corresponding correlaions for supply shocks. The siuaion is even worse in he case of aggregae supply shocks. Even hough here are a few significan correlaion coefficiens, none of hese are greaer han.5. 18

20 Table 6: Correlaion of he Demand Shocks Indonesia Malaysia Philippines Singapore Thailand Indonesia Malaysia 3 1. (.73) Philippines -8.31*** 1. (.79) () Singapore 6.3***.8*** 1. (.54) (3) (5) Thailand 1.9***.4**.8*** 1. (.89) (5) () (5) Noe: Figures in paranheses are he respecive p-values. The 5% and 1% significance levels are indicaed by *** and **, respecively. Table 7: Correlaion of he Supply Shocks Indonesia Malaysia Philippines Singapore Thailand Indonesia Malaysia.1 1. (.3) Philippines (.63) (.1) Singapore (.54) (.78) (.76) Thailand.19*.1**.5*** (7) (4) (1) (.15) Noe: Figures in paranheses are he respecive p-values. The 1%, 5% and 1% significance levels are indicaed by a ***, ** and *, respecively. These numbers are much smaller han similar previous sudies. For example, Bayoumi and Eichengreen (1994) repored ha he correlaion of aggregae supply shocks beween Malaysia and Indonesia is.5 and ha beween Malaysia and Singapore is.71. The same sudy repored corresponding figures for demand shocks as.58 and.67, respecively. The correlaion coefficiens repored in Bayoumi e al () are also larger han he ones 19

21 repored in he curren paper. Tha paper found ha Disurbances are highly correlaed across cerain ASEAN counries, and based on his, claimed ha Evidence on macroeconomic disurbances does no obviously indicae ha ASEAN is furher han Europe from saisfying he symmerical-disurbances crierion. The small correlaion coefficiens obained in he curren paper are in sharp conras wih he findings in hese sudies and sugges ha ASEAN region is sill far away from meeing he requiremens for a common currency union. 4. Conclusion This paper uilizes he idenificaion scheme proposed by Cover e al (6) o examine he naure of aggregae demand and supply shocks in 5 ASEAN counries. I is found ha he aggregae demand and supply shocks are posiively correlaed in Malaysia, he Philippines, Singapore and Thailand. The aggregae demand shock is found o be playing he dominan role in affecing he oupu levels in boh shor and long run in hese counries. This finding is conrary o he convenional belief ha he aggregae demand shock has only emporary effec on he oupu level. The idenificaion scheme applied here allows he shocks o be ransmied from he demand o he supply side and he findings of he paper sugges ha a shif in he aggregae demand curve may have induced he long run aggregae supply curve o shif in he same direcion in hese counries. This paper furher exend he analysis of Cover e al (6) o show ha he BQ approach may lead o anomalies in he idenificaion of macroeconomic shocks, if we don allow he correlaion beween he wo shocks. I showed ha in he BQ model, he idenified aggregae supply shock leads o he posiive responses in inflaion, which is hard o jusify by any macro heory. This anomaly is basically caused by he fac ha aggregae demand and supply shocks are correlaed. This is likely ha he idenified supply shocks in he BQ model conains some demand driven componen, which causes he inflaion rae o go up (insead of going down), when here is a posiive supply shock. This paper showed ha he apparen anomaly in he inflaion dynamics disappears when we allow he correlaion beween he wo shocks and he causaliy running from demand o supply shocks.

22 I is also found ha he supply shocks are playing he dominan role in affecing he inflaion rae in ASEAN counries. This finding is also somewha conrary o he convenional belief ha inflaion is always a moneary phenomenon. We even observed ha he supply shock has a permanen effec on inflaion in Malaysia and Singapore. The finding of his paper has imporan policy implicaion, especially, in he wake of recen global recession. The finding ha a shif in he aggregae demand curve induces a shif in he long run aggregae supply curve simply assers ha because of he recession an economy will reach he higher level of unemploymen and lower level of oupu permanenly if no policy acion is aken. On he oher hand an expansionary demand managemen policy (o comba recession) may no be as harmful for inflaion as i was previously hough, as his paper suggess ha supply shocks are he main conribuors for he variaion in inflaion in mos of he ASEAN counries considered. This finding also suggess ha some supply side policies may be needed along wih he inflaion argeing policy o comba inflaion. This paper also found ha he correlaions of shocks across he 5 ASEAN counries are quie small, alhough some of hese are saisically significan. This finding suggess ha shocks are no synchronized in hese counries and ASEAN counries are ye o be ready o form a common moneary union like Europe. References: Abeysinghe, Tilak and Gulasekaran Rajaguru (4). Quarerly GDP Esimaes for China and ASEAN4 wih Forecas Evaluaion. Journal of Forecasing 3, Ball, Laurence (1999). Aggregae Demand and Long Run Unemploymen. Brookings Papers on Economic Aciviy, Bayoumi, Tamim and Barry J. Eichengreen (1994). One Money or Many? Analysing he Prospecs for Moneary Unificaion in Various Pars of he World. Princeon Sudies in Inernaional Economics 16, Princeon Universiy. 1

23 Bayoumi, Tamim; Barry J. Eichengreen, and Paolo Mauro (). On Regional Moneary Arrangemens for ASEAN. Journal of he Japanese and Inernaional Economies 14, Blanchard, Olivier J (1987). Why Does Money Affec Oupu: A Survey. NBER Working Paper Series, Paper No. 85. Blanchard, Olivier J. and Danny Quah (1989). The Dynamic Effecs of Aggregae Demand and Aggregae Supply Disurbances. American Economic Review 79, Blanchard, Olivier J. and Summers, Lawrence H. (1987). Hyseresis in Unemploymen. Europian Economic Review 31, Cover, James P., Waler Enders, and C. James Hueng (6). Using he Aggregae Demand Aggregae Supply Model o Idenify Srucural Demand-side and Supply-Side Shocks: Resuls Using a Bivariae VAR. Journal of Money, Credi and Banking 38, Lucas, Rober E, Jr (197). Expecaions and he Neuraliy of Money. Journal of Economic Theory 4, Lucas, Rober E, Jr (1973). Some Inernaional Evidence on Oupu-Inflaion Tradeoffs. American Economic Review 63, Mankiw, N. Gregory (1987). The Opimal Collecion of Seigniorage: Theory and Evidence. Journal of Moneary Economics, Ng, Thiam Hee (). Should he Souheas Asian Counries form a Currency Union? Developing Economies 4, Sadler, George W. (199). Business Cycle Models wih Endogenous Technology. American Economic Review 8, Uerback, James, M. (1974). "Innovaion in Indusry and he Diffusion of Technology," Science 183, 6-6.