Oil Price, Exchange Rates and Inflation in the Asian Countries: Endogenous versus Exogenous Regime Shift(s)

Transcription

1 Ol Prce, Exchange Raes and Inflaon n he Asan Counres: Endogenous versus Exogenous Regme Shf(s) Ahmad Zubad Baharumshah and Soon Sew-Voon Faculy of Economcs and Managemen Unvers Pura Malaysa, 434 UPM Selangor, Malaysa Absrac Ths sudy nvesgaes he relaonshp beween ol prce and exchange rae on nflaon for egh Asan counres, ncludng he wo ol-exporng counres (Indonesa and Malaysa). For hs purpose, we esmae degree of pass-hrough from he exchange rae and ol prce changes o he nflaon. The major fndngs are: frs, here s a moderaon n he prce levels n all bu one counry Indonesa. Prce levels have dropped sgnfcanly n he more recen years wh he mng of he regme shf(s) varyng across he counres. Second, here s a posve relaonshp beween oupu gap and he nflaon for Chna, Malaysa and Sngapore whle reverse relaon holds n Indonesa. Thrd, he degree pass-hrough from he ol prce and exchange rae changes o nflaon has declned somewha durng he low nflaon rae regme n mos of he Asan counres. All n all, he degree of ol prce pass-hrough s hgher n he non-nflaon argeers. KEY WORDS: Ol prce, fear o floa phenomenon, nflaon argeng, srucural breaks JEL CLASSIFICATIONS: C32, E31, E32, G1 1. Inroducon The bulk of leraure on nflaon has focusng on wheher nflaon s saonary or random walk varable. If he nflaon s saonary process hen shocks have only ransory effec and he nflaon s conrollable. Ths also means ha s volaly (rsk) effec on macroeconomc performance wll be less perssen. In he pas wo decades a number of researchers have sough o deermne he perssency of he nflaon for nflaon argeers and non-nflaon argeers. Gregorou and Kononkas (26), Narayan and Narayan (28), Basher and Weserlund (28), Kang e al. (29), and Cuesas and Harrson (21), among ohers have suggesed ha he nflaon s n fac conrollable. Ths evdence, however, s n sharp conras wh he works by Culver and Papell (1997), and more recenly Sregar and Goo (21) and Prasernukul e al. Correspondng auhor. Emal: baharumshah@yahoo.com. Tel: Fax:

2 (21) who found he nflaon seres s hghly perssence and dffcul o conrol. If he nflaon seres s less volale as clamed by some analyss hen he seres s easer o be forecas, and hs s mporan for counres adopng eher mplc or explc nflaon argeng (IT) as a moneary framework. The ueson wheher nflaon becomes less perssence wh he adopon of IT s becomng ncreasngly dffcul o gnore. I s well known ha any changes on he nomnal anchor of IT nomnal neres rae wll have rpple effecs on he nflaon hrough he exchange rae and aggregae demand channels. The leraure has also denfed currency, debs, and commody crses as mporan exernal facors nfluencng he nflaon. The degree a whch ol prce and he exchange rae are allowed o ransmed (pass-hrough) no he domesc economy has nfluence (bearng) on domesc nflaon. The leraure has found ha pass-hrough effecs of hese varables have changed over me, and s no surprsng o fnd ha he emprcal evdences have no been conclusve. Byand-large, he evdences of exchange rae pass-hrough (ERPT) and ol prce pass-hrough (OPPT) on he domesc nflaon reman nconclusve across counres wh varous daa span and echnues employed for he analyss (e.g., Prasernukul e al., 21; Sregar and Goo, 21; Campa and Goldberg, 25; Jongwanch and Park, 211; Chen, 29a, b). Much uncerany sll exss abou he relaon beween he exchange rae and he nflaon. There are several facors n explanng he delayng or declnaon on he pass-hrough of he exchange rae (see Taylor, 2; Pavasuhpas, 21; Reyes, 27; Choudhr and Hakura, 21; Schmd-Hebbel and Werner, 22; Gagnon and Ihrg, 24; Cornhas, 29). Mos of hese auhor clamed ha he degree of ERPT no he domesc nflaon end o be lower durng low nflaonary envronmen. Meanwhle, some sudes concenrae on ol (supply) shock raher 2

3 han exchange rae movemen (exernal rade channels) no he domesc nflaon (e.g., Jongwanch and Park, 211; Chen, 29b). 1 Ther fndngs revealed ol prce has sgnfcan effec on domesc nflaon and ha hgh pass-hrough s recorded n he counres under suded To bes of our knowledge, no prevous work have no deal wh he above ssues for he Asan counres n much deal, especally he effecveness of IT and he exchange rae movemen durng exreme ol crss (see Ban and Tereanu, 29, 21). In hs sudy we are neresed n he followng uesons: Does (low) nflaonary envronmen maers n he degree of pass-hrough? Should emergng counres respond o he changes n he exchange rae and ol prce, especally for he nflaon argeers? Is he reducon n he degree of pass-hrough a feaure common only o he developed counres? I should be noe ha majory of he prevous pass-hrough sudes concenrae on ndusral counres. There are lmed passhrough effec sudes on he emergng markes, and all of hem have gnore possble endogenous regme change, wh few excepons (e.g., Cho e al., 211; Sregar and Goo, 21; Prasernukul e al., 21; Io and Sao, 27). In hs paper we nend o fll hs gap. Anoher conrbuon of hs paper s ha we look a he mpac of regme changes on he prce sably. Specfcally, we nend o examne how he regme changes nfluences he mpac of a changes n he exchange raes and ol prce movemens on he domesc nflaon n egh Asan counres over 1975 o 21 perods. Our sample ncludes four nflaon argeers, namely Souh Korea, he Phlppnes, 1 Governmen polcy (e.g., fuel or elecrcy subsdes or prce conrol) on he prce of basc necesses may delay or reduce he pass-hrough of ol and food prce ncrease o domesc prce n Asan counres (Jongwanch and Park, 211). Chen (29a, 19 ndusralzed counres) argues he apprecaon of he domesc currency, a more acve moneray polcy n response o nflaon and a hgher of rade openness lead ohe declne n OPPT. 3

4 4 Thaland and Indonesa 2. Ths paper akes a fuher sep o nvesgae wheher he hgh (low) nflaonary envronmen maers for he ERPT and OPPT. The oulne of he sudy s as follow. The nex secon provdes a bref dscusson on heorecal framework deployed n hs sudy. Secon 3 dscusses he daa and emprcal resuls and Secon 4 concludes he paper. 2. A Bref Theorecal Framework An augmened Phllps curve based on auoregressve (AR) dsrbuon lag model s used o esmae he degree of pass-hrough effec: 3 u x ol p p er p * 2 1, (1) Three neracve erms are added n E. (1) o nvesgae he exchange rae and ol prce changes n he degree of pass-hrough effec on he domesc consumer nflaon for he pre- and pos regme changes 4 :, ) ( ) ( ) ( * 2 1 u d ol d p d er x ol p p er p (2) 2 For he four nflaon argeers, Thaland (2M5) and Indonesa (25M7) are under managed floang, and he Phlppnes (22M1) and Souh Korea (21M1) are under ndependen floang. For non-nflaon argeers, Chna under crawng peg, and Malaysa, Inda and Sngapore are under managed floang. The nflaon arge bands for Indonesa s 5% (+/-1), he Phlppnes s 5-6%, Souh Korea s %, and Thaland s -3.5%. 3 For more deals, see Campa and Goldberg (2), Gagnon and Ihrg (24), Edwards (26), Chen (29b), Prasernukul e al. (21), and Sregar and Goo (21) who reled on AR dsrbuon lag esng wh he general-ospecfc approach for pass-hrough analyss. 4 Some sudes found he nflaon seres n fac negraed of order one [I(1)] and ncorporae error correcon componen n he ARDL (o avod poenal endogeney) esng for pass-hrough effec (e.g., Sregar and Goo, 21; Chen, 29b).

5 where s CPI denoed as Consumer Prce Index (CPI) nflaon rae calculaed as p log CPI 1 he change n he log of CPI, backward-lookng nflaon expecaon and nflaon perssence are capured wh lags of he dependen varables (see Hooker, 22), er s nomnal effecve exchange rae (or nomnal exchange rae) 5, * p s foregn nflaon/ world nflaon (rae of change of US producer prce ndex), ol s ol prce, and he conrol varables ( x) are he varables ha expeced o affec a change n he rae of nflaon. Followng prevous sudes (e.g. Chen, 29b; Prasernukul e al., 21), hs sudy uses oupu gap ( y y) as proxy for naural rae of unemploymen and o capure he effecs of busness cycle. The oupu gap s defned as he devaon of log ndusral producon ndex from s Hodrck-Presco flered rend seres. The role of money and foregn nflaon n nfluencng nflaon are capured by money growh and he changes n US producer prce ndex. Gven he fac ha he moneary polcy accommodaes he nflaon wh lag, and any changes of he sablzaon polcy wll no have mmedae effec on he rae of nflaon. As argued by Edwards (26, p. 1): akes me for he publc o undersand a new polcy regme, he srucural change n he pass-hrough coeffcen wll no be nsananeous, bu raher. Depar from prevous sudes, hs sudy wll depend on Ba and Perron (1998, 28, hereafer Ba- Perron) procedure o deec srucural shf(s) nsead of solely reled on he offcal dae announcemen. To compare he effec of endogenous break(s) versus exogenous break(s), wo se dummy wll be generaed. Exogenous dummy wll be creaed based on he offcal announcemen dae of IT, akng value of one snce he offcal announcemen dae of IT, and 5 Edwards (26) conended he lmaon of nonradables prces for he emergng counres. Auhor found producer prce ndex (proxy for domesc prce of radables) o consumer prce ndex (proxy for domesc prce of nonradables) rao provded farly good proxy for (effecve) real exchange rae. 5

6 zero oherwse. Endogenous dummy wll be generaed by usng Ba-Perron seuencal procedure for he nflaon seres. The shor-run ERPT pre-regme shf s, whle long-run pass-hrough s 1 /. For he pos-regme shf, he shor-run ERPT s long-run s /. Turnng o he OPPT, he shor-run OPPT preregme shf s , and he long-run OPPT s / 4 1 6, and Furhermore, when he effec of regme change s aken no accoun, he shor-run and long-run OPPT durng posregme change become 4 and / , respecvely. 3. Resul and Dscusson 3.1 Perssency of nflaon seres Ths sudy covered he monhly daa from 1975M1 o 21M12 for egh Asan counres, ncludng ASEAN+5 (Indonesa, he Phlppnes, Malaysa, Sngapore, and Thaland), Souh Korea, Inda and Chna. 6 The sample perods vares for each counry accordng o he avalably of daa. The seres for consumer prce ndex (CPI, 25=1), nomnal effecve exchange rae (NEER, 25=1), blaeral US dollar (Inda, Indonesa, Thaland, and Souh Koera), foregn prce (US ndusral producon ndex, 25=1), and UK Bren are drawn from Inernaonal 6 The samples perods are ddeced by he manufacurng/ndusral producon volume ndex, and vares across counres. The sample perod for Indonesa s 1994M1-21M12 (24 observaons); for he Phlppnes, 1985M1-21M12 (312 observaons); for Malaysa, 199M1-21M12 (252 observaons); for Sngapore, 1983M1-21M12 (336 observaons); for Thaland, 1976M1-21M12 (42 observaons); for Souh Korea, 1975M1-21M12 (432 observaons), for Inda, 1981M4-21M12 (357 observaons), and for Chna, 199M1-21M12 (252 observaons). All daa for Chna, excep for NEER are sourced from Daasream. All seres are seasonal unadjused excep he NEER for he Phlppnes, Malaysa and Chna. 6

7 Fnancal Sascs (Inernaonal Moneary Fund, IMF). Money supply seres for Malaysa, Thaland, Souh Korea and Inda s drawn from IMF, whereas ohers are drawn from Daasream. A sngle world real crude ol prce s consruced by deflaed he nomnal prce of UK Bren wh US producer prce ndex for all commodes. The man purpose of hs sudy s o compare he pass-hrough effec for nflaon argeers and non-nflaon argeers. To acheve hs ask, we appled he Ba-Perron procedure o denfy endogenous break pons n he nflaon seres for nflaon argeers (IDN, PHL, THA, and KOR) and non-nflaon argeers (CHN, IND, MYS, and SGP). Here, we have consdered he pure srucural changes model, whch allows he enre coeffcen o change. We allowed up o fve break-pons n our model wh he value of he rmmng (eps1) = Accordng o he Ba-Perron, break-pons denfed by he nformaon crera are based and hey have herefore recommended he seuenal procedure nsead. Based on he procedure, we conclude ha one (wo) break(s) pon s deeced for nflaon seres n hree nflaon argeers (hree non-nflaon argeers), whls one break for Malaysa s nflaon seres. However, we faled o deec any srucural shfs for he case of Indonesa. The dfferences n he esmaed mean over each segmen has pon o a large declne for all he counres (PHL, %; THA, %; KOR, %; MYS; %). For CHN, IND and SGP, he esmaed mean has a large declne n he second regme (CHN, %; IND, %; SGP, %), bu he esmaed mean somehow has ncreased of 14.62%, % and 55% for CHN, IND, and SGP, respecvely n he hrd regme. Fgure 1 dsplayed he 7 By usng Ba-Perron echnue, Russell (211) fnd egh nflaon regmes (seven shfs n he mean rae of nflaon) for US nflaon seres (1952M3 o 24M9) and conend ha he resuls repored n he modern Phllps curve leraure are domnaed by, and are as a drec resul of, he shf n he mean rae of nflaon ha are no accouned for n he analyss. 7

8 nflaon seres ogeher he regme shfs n he nflaon volaly. A number of shfs n he volaly of nflaon seres can be seen n he Fgure 1. The plos n Fgure 1 clearly showed ha he nflaon argeers are adoped IT polcy regme durng low nflaon envronmen. The volaly of nflaon seres has been reduced followng he declnaon n he nflaon seres. The volaly seres however has ncreased n he lae 199s for CHN and around 26 for IND and SGP. 8 [Inser Fgure 1] Havng evdence of srucural shf(s) n he nflaon seres, we proceed wh Narayan and Popp (21, NP hereafer) wo-break Augmened Dckey Fuller (ADF) un roo ess for he nflaon seres 9. The convenonal un roo ess used o nvesgae he negraon properes mgh suffer from loss of power when srucural break(s) s gnored n he daa generang process (DGP) (Perron, 1989). NP proposed a new ADF-ype es whch was approxmaely nvaran o level and slope breaks n fne samples. As n Lee and Srazcch (23) mnmum Lagrange Mulpler, he NP ADF es recognzed naccurae esmaon of break daa as he source of spurous rejecons. The key feaures of hs es are an ADF-ype nnovaonal ouler (IO) un roo es n whch he DGP s specfed as an unobserved componen model and he breaks are allowed under boh he null and alernave hypohess. 8 Table 1 repors he esmaed daes for srucural breaks n he nflaon raes. We noe he mng of he break pons for IT s no maches wh he offcal adopon daes of IT. Ths s n lne wh he fndngs by Cho e al. (211) ha he esmaed break pons are close o offcal adopon daes of IT n some counres bu some IT counres have a declne n mean nflaon a b earler han he formal announcemen daes of IT adopon. In order o conserve place, he resuls for Ba-Perron are reuesed upon auhors. 9 In hs sudy, we consder he wo-break models. As ndcaed by NP, model selecon (Model 1 or Model 2) should be dcaed by he economc heory. Addonally, Narayan and Smyh (25) argued ha he greaer he numbers of breaks ha are added o he model, he closer are he macroeconomc me seres o random walks and he less relevan are un roos wh srucural breaks. 8

9 Followng Hall s (1994) general-o-specfc approach we seleced a maxmum lagged erm o when he seral correlaon. The resuls dsplayed n Table 1. We found ha he breaks daes denfed by boh models are vares across he counres and occurred around he 1997 Asan crss excep for Sngapore and Chna. For Inda and Chna, s apparen ha he wo break daes denfy by he ess are closely conneced o her own polcal and economc evens. Chna, for nsance, he frs break dae was deeced n 1994: M11 and a second break occurred n 1995: M1 (Model 1, allows for wo breaks n level). The mng of he break daes are conneced wh sharp devaluaon of he Chnese yuan agans he US dollar. The NP es whch allowed for wo breaks n level (labeled as Model 1 n Table 1) found ha he seres for IDN [ ˆ =-4.85], PHL [ ˆ =-4.267], CHN [ ˆ =-6.356], IND [ ˆ =-4.243], MYS [ ˆ =-6.7], and SGP [ ˆ =-4.59] rejec he un roo null hypohess a he 5% sgnfcance level or beer. Meanwhle, allowng for wo breaks n he level and slope (labeled as Model 2, allows for wo breaks n level as well as slope), here s no addonal saonary seres. Taken ogeher, he resuls from hese wo models rejec he un roo null for sx ou of 8 seres. Table 3 presens he pon esmaes of half-lves along wh her 95% confdence nervals (CIs) for he nflaon seres. The perssency of nflaon seres for non-nflaon argeers s lower compare wh he nflaon argeers. Ths ndcaes ha he nflaon seres for nflaon argeers s more perssen han non-nflaon argeers. Addonally, he CIs for nflaon argeers are wder han non-argerers. Indeed, he CIs for nflaon (non-nflaon) argeers are ranges from 1.3 monhs o 4.72 monhs (.48 monhs o 1.16 monhs) for Model 1 and.99 monhs o 2.56 monhs (.27 monhs o 1.14 monhs) for M2. Togeher wh he prelmnary daa analyss (Fgure 1), he fall n he nflaon volaly s more sgnfcanly n hgh perssen nflaon counres. [Inser Tables 1 and 2] 9

10 3.2 Shor-run and long-run exchange rae and ol prce pass-hrough Is he degree of ERPT and OPPT sensve wh he degree of nflaon perssency? These have movaed us proceed o nvesgae he degree of ERPT and OPPT effecs for he hgh and low perssen envronmen. We nvesgae how he domesc nflaon responds o he exchange rae and ol prce shocks usng he mpulse response funcons. Snce he ol prce and exchange rae may no be exogenous n E. 2, we conduc he generalzed mpulse response funcon based on he work by Pesaran and Shn (1998). I s amed o crcumven he problem of he dependence of orhogonalzed mpulse responses o he orderng of varables n he VAR. Gaphcal up o 15 monhs are presened n Fgure 2. The pos-it regme s used o analyss how OPPT and ERPT changes afer he adopon of IT. The lag lengh s chosen based on lag lengh crera. For he pos-it perod, he domesc nflaon o exchange rae shocks (ERPT) s low and negave. The nflaon s responded negavely o he shock n exchange rae and ol prce mplyng he exsence of a negave relaonshp. Ths ndcaed ha a deprecaon of he exchange rae wll reduce he domesc nflaon. However, some response of nflaon o he exchange rae s posve (e.g., SGP), mplyng ha a deprecaon cause a rse n he nflaon. Turnng o he response of domesc nflaon o ol prce shock (OPPT), he domesc nflaon o ol prce shock s lower compare wh he domesc nflaon o exchange rae shocks excep for Inda. However, majory of he shocks are dyng ou afer 2 years for boh nflaon argeers and non-nflaon argeers. For Thaland, Souh Korea, and Malaysa, he degree of ERPT and OPPT does no exs whn 2 years. The response of nflaon caused by he exchange rae s larger han he ol prce. 1 [Inser Fgure 2] 1 Prasernukul e al. (21) fnd he resuls from VAR models almos conssen wh sngle-euaon approach. However, he mehods employed by he auhors may sensve o he orderng of varables n VAR models. 1

11 Nex, we nvesgae degree of ERPT and OPPT for he pre- and pos- regme shf(s) and for hs purpose he Ba-Perron procedure s appled o daa. The analyss s based on auomercs ree search procedure (auomac model selecon). Snce some of he hgher lagged varables are sgnfcan, we sar wh 12 monh lag. To accoun for seral correlaon, we re-ran he model by usng Newey and Wes (1987) correcon prescrbed by Sock-Wason o oban heeroscedascy and auocorrelaon conssen sandard error. The ERPT and OPPT s coeffcens for he pre- and pos shf 1 (or 2) are summarzed n Tables 3 and 4. For nflaon argeers, we noe ha he degree of ERPT has been reduced n he shor-run and long-run excep for KOR (shor-run), PHL (long-run) and THA (long-run). The fear of floang phenomenon does no exs n he hree (IDN, PHL, THA) of he IT counres n he shor-run and wo IT (IDN, KOR) counres n he long-run. For CHN and IND, he degree of ERPT has been ncreased for he pos-1994m11 and pos-1998m11, respecvely. In he long-run, he ERPT for MYS s ncreased for he pos 1999M1. Turnng o he OPPT, we observe ha he OPPT for CHN and MYS has ncreased n he longrun, bu he OPPT for CHN s ncreased for boh shor-run and long-run for pos-1997m12. For he case of SGP, he OPPT s ncreased n he pos-shf 2 (26M6). Overall, he degree of ERPT does no sensve wh he degree of perssency for domesc nflaon n one counry. I s hgher n wo Bg Emergng Markes CHN and IND. However, he degree of OPPT somehow s hgher n he low perssen envronmen (non-nflaon argeers). I s ncludng he CHN (pos-1994m11 and pos-26m3), MYS (pos-1999m1) and SGP (pos-26m6). The money supply and foregn prce also wll cause a hgh nflaon. Prasernukul e al. (21) and Sregar and Goo (21) conend ha he long-run ERPT also depends on nflaon nera whch can be capured by coeffcen and d. We noe ha he lagged nflaon has evdence p p 11

12 pass-hrough no domesc nflaon, and he nflaon nera has rsen n he local economy. The nflaon nera has caused he long-run ERPT and OPPT are greaer han shor-run passhrough. [Inser Tables 3 and 4] Ths paper fnds eher he level or he sandard devaon of he nflaon have been declned subsanally from he frs o second regme. The declne n OPPT excep for Chna was caused by he move o a low nflaon envronmen ha n urn resuled from he regme change. For he case of Chna, s ol self-suffcency ndex s negave snce Addonally, as compare o he year 28 wh 2, he oal prmary energy consumpon has ncreased around 134% n Chna, followed by Sngapore (56%), Thaland (53%), Indonesa (48.5%), Inda (48.2%), Souh Korea (26%), Malaysa (24%) and he Phlppnes (3%). Hence s no surprsng he negave self-suffcency ndex and he huge ncreases n he oal prmary energy consumpon have helps o explan he ncreases n he long-run and shor-run OPPT no he Chna domesc nflaon. The hgh OPPT mply ha global prce shocks are a sgnfcan deermnan of nflaon for he case of Chna. However, oher regons prces seem solaed from he global prce shocks and exchange rae shocks. Jongwanch and Park (211, p.92) ndeed argue ha he degree of OPPT s hgher for counres wh lmed fuel subsdes and he governmen nervenon for polcal reasons may bounder consumers o adjus o hgher global prces. By-and-large, majory of he pass-hrough ended o declne, suggesng ha he moneary polcy ended o shf owards sablzng nflaon. 4. Concluson Ths sudy presens he effecs of ol prce and exchange rae changes on he domesc nflaon for egh Asan counres, ncludng Chna and Inda. For hs purpose, we esmae degree of 12

13 pass-hrough from he exchange rae and ol prce changes no he nflaon by usng an augmened Phllps curve. We fnd srucural break(s) for all counres wh sole excep of Indonesa, an ol exporng counry. We observed a noceable declne n nflaon volaly afer he frs break pon for all he Asan counres. Ineresngly, our resuls reveal ha he volaly for all of he non-nflaon argeers ncreased afer he second shfs (excep for Malaysa). The degree pass-hrough from he ol prce and exchange rae changes o he nflaon has somewha declned durng he low nflaon rae (less volale) regme n mos of he counres (7 n he shor-run and 6 n he long-run). The degree of ERPT appears o be nsensve wh he degree of perssency n he domesc prce level. However, he degree of OPPT somehow s hgher n he low perssen envronmen (non-nflaon argeers). There s ncreasng OPPT n Chna, suggesng ha here s ncrease n oal prmary energy consumpon whch helps o explan he ncreases n he OPPT, however hs s no he case for he IT counres. 11 For some of he cases, he nflaon was drven by demand pull facors snce lmed mpacs of ol prce shock (supply shock) on domesc nflaon, suggesng ha moneary auhores can launch some polcy o accommodae hese home growh facors. Addonally, here s a posve relaonshp beween oupu gap and he nflaon for Chna, Malaysa and Sngapore whle reverse (negave) relaon holds for Indonesa. Overall, he resuls show ha wha maers for ERPT and OPPT s no he moneary regme,.e., IT per se bu raher he nflaon regme (nflaon envronmen) pror o he mplc and explc adopon of IT or polcy changes. 11 The oal prmary energy consumpon (ncludes he consumpon of peroleum, dry naural gas, coal, and ne nuclear, hydroelecrc, and non-hydroelecrc renewable elecrcy) n Chna s 85.6 Quadrllon Bu (Brsh Thermal Un) Bu n 28 (17.25% of he world oal prmary energy consumpon). 13

14 References Ba, J. and Perron, P. (1998). Esmang and esng lnear models wh mulple srucural changes. Economerca, 55(1), Ba, J. and Perron, P. (23). Compuaon and analyss of mulple srucural change models. Journal of Appled Economercs, 18, Basher, S.A. and Weserlund, J. (28). Is here really a un roo n he nflaon rae? More evdence from panel daa models. Appled Economcs Leers, 15, Ban, N. and Tereanu, E. (29). Wha should nflaon argeng counres do when ol prces rse and drop fas? IMF Workng Paper, WP/9/11. Ban, N. and Tereanu, E. (21). Inflaon argeng durng asse and commody prce booms. Oxford Revew of Economc Polcy, 26(1), Campa, J.M. and Goldberg, L.S. (25). Exchange rae pass-hrough no mpor prces. Revew of Economcs and Sascs, 87, Chen, S.-S. (29a). Revsng he nflaonary effecs of ol prces. The Energy Journal, 34(4), Chen, S.-S. (29b). Ol prce pass-hrough no nflaon. Energy Economcs, 31, Cho, C.-Y., Lm, Y.S. and O Sullvan, R. (211). Inflaon argeng and relave prce varably: wha dfference does nflaon argeng make? Souhern Economc Journal, 77(4), Choudhr, E.U. and Hakura, D.S. (21). Exchange rae pass-hrough o domesc prces: does he nflaonary envronmen maer? IMF Workng Paper, No. WP/1/194, Corhnhas, C. (29). Exchange rae pass-hrough n ASEAN: mplcaons for he prospecs of moneary negraon n he regon. The Sngapore Economc Revew, 54(4), Cuesas, J.C. and Harrson, B. (21). Inflaon perssence and nonlneares n Cenral and Easern European counres. Economcs Leers, 16, Culver, S.E. and Papell, D.H. (1997). Is here a un roo n he nflaon rae? Evdence from seuenal break and panel daa models. Journal of Appled Economercs, 12(4), Edwards, S. (26). The relaonshp beween exchange raes and nflaon argeng revsed. NBER Workng Paper Seres Gagnon, J.E. and Ihrg, J. (24). Moneary polcy and exchange rae pass-hrough. Inernaonal Journal of Fnance and Economcs, 9(4), Gregorou, A. and Kononkas, A. (26). Inflaon argeng and he saonary of nflaon: new resuls from an ESTAR un roo es. Bullen of Economc Research, 58(4), Hall, A. (1994). Tesng for a un roo n me seres wh prees daa-based model selecon. Journal of Busness and Economc Sascs, 12(4), Hooker, M.A. (22). Are ol shocks nflaonary? Asymmerc and nonlnear specfcaons versus changes n regme. Journal of Money, Cred and Bankng, 34(2), Io, T. and Sao, K. (27). Exchange rae pass-hrough and domesc nflaon: a comparson beween Eas Asa and Lan Amercan counres. Research Insue of Economy, Trade and Indusry (RIETI) Dscusson Paper Seres, No. 7-E-4. 14

15 Jongwanch, J. and Park, D. (211). Inflaon n developng Asa: pass-hrough from global food and ol prce shocks. Asan-Pacfc Economc Leraure, 25(1), Kang, K.H. Km, C.-J. and Morley, J. (29). Changes n U.S. nflaon perssence. Sudes n Nonlnear Dynamcs & Economercs, 13(4). Km, S. and Lma, L.R. (21). Local perssence and he PPP hypohess. Journal of Inernaonal Money and Fnance, 29, Lee, J. and Srazcch, M.C. (23). Mnmum Lagrange mulpler un roo es wh wo srucural breaks. The Revew of Economcs and Sascs, 85(4), Narayan, P.K. and Narayan, S. (28). Is here a un roo n he nflaon rae? New evdence from panel daa models wh mulple srucural breaks. Appled Economcs, Frs, 1-1. Narayan, P.K., Popp, S., 21. A new un roo es wh wo srucural breaks n level and slope a unknown me. Journal of Appled Sascs, 37(9), Narayan, P. and Smyh, R. (25). Srucural breaks and un roos n Ausralan macroeconomc me seres. Pacfc Economc Revew, 1(4), Newey, W.K. and Wes, K.D. (1987). A smple, posve sem-defne, heeroskedascy and auocorrelaon conssen covarance marx. Economerca, 55(3), Pavasuhpas, R. (21). Should nflaon-argeng cenral banks respond o exchange rae movemens. Journal of Inernaonal Money and Fnance, 22, Perron, P. (1989). The grea crash, he ol prce shock and he un roo hypohess. Economerca, 57, Pesaran, M and Shn, Y. (1998). Generalzed mpulse response analyss n lnear mulvarae models, Economc Leers, 58(1), Pesaveno, E. and Herrera, A.M. (29). Ol prce shocks, sysemac moneary polcy, and he Grea Moderaon. Macroeconomc Dynamcs, 13, Prasernukul, W., Km, D. and Kaknaka, M. (21). Exchange raes, prce levels, and nflaon argeng: evdence from Asan counres. Japan and he World Economy, 22(3), Reyes, J. (27). Exchange rae pass-hrough effecs and nflaon argeng n emergng economes: wha s he relaonshp? Revew of Inernaonal Economcs, 15(3), Russell, B. (211). Non-saonary nflaon and panel esmae of Uned Saes shor and longrun Phllps curves. Journal of Macroeconomcs, 33, Schmd-Hebbel, K. and Werner, A. (22). Inflaon argeng n Brazl, Chle, and Mexco: performance, credbly, and he exchange rae. Economa, 2(2), Sregar, R.Y. and Goo, S. (21). Effecveness and commmen o nflaon argeng polcy: evdence from Indonesa and Thaland. Journal of Asan Economes, 21(2), Taylor, J.B. (2). Low nflaon, pass-hrough, and he prcng power of frms. European Economc Revew, 44,

16 Fgure 1: Inflaon and regme shfs n volaly for nflaon and non-nflaon argeers Inflaon Targeers Non-Inflaon Targeers IDN CHN Pos-IT: 25M PHL Pos-IT: 22M IND THA MYS Pos-IT: 2M KOR Pos-IT: 21M SGP Noes: Inflaon seres defned as CPI * 1. The sragh dash lnes shown are referrng o bands a 3 log CPI 1 sandard devaon durng each regme whch esmaed usng Ba and Perron (1998, 23). Gray areas refer o pos- IT regme. 16

17 Fgure 2: Generalzed mpulse responses of nflaon o exchange rae and ol prce shocks Inflaon Targeers IDN (Pos-IT) Non-Inflaon Targeers CHN er op er op PHL (Pos-IT) IND er op er op THA (Pos-IT) MYS er op er op KOR (Pos-IT) SGP er op er op Noes: The lag lengh s chosen based on lag lengh crera. Exogenous break(s) ncorporaed n he non-nflaon argerers analyss. 17

18 Table 1: Un roo ess for nflaon seres Model 1 Model 2 ˆ ˆ k TB1 TB2 ˆ ˆ k TB1 TB2 Inflaon Targeers IDN a M1 a 25M9 a a M1 a 25M9 a PHL b M12 a 1999M12 a M12 a 1999M12 a THA M9 a 1997M7 a M9 a 1997M7 a KOR M7 a 1997M11 a M7 a 1997M11 a Non-Inflaon Targeers CHN a M11 a 1995M1 a a M11 a 25M1 a IND b M11 a 21M1 a a M12 a 1998M11 a MYS a M1 a 26M2 a a M1 a 26M2 a SGP b M8 b 24M1 a b 13 24M1 a 24M5 b Noes: ( a ) and ( b ) ndcae sgnfcance a 1% and 5% sgnfcance levels, respecvely. TB1 and TB2 are he daes of srucural breaks deeced. ˆ s he es sasc of ˆ for Model 1 and Model 2, respecvely. Lengh lag (k) s seleced based on he general-o-specfc procedure proposed by Hall (1994) for every cases. Crcal values for T=3 are abulaed n Narayan and Popp (21). Table 2: Perssency, half-lves (n monhs) and confdence nervals for nflaon seres Model 1 Model 2 dˆ HL(M) se (ˆ ) 95%CI dˆ HL(M) se (ˆ ) 95%CI Inflaon Targeers IDN [1.886, 2.176] [.9898, ] PHL [1.279, 1.633] [1.16, 1.62] THA [1.344, 2.168] [1.129, 1.688] KOR [2.5113, 4.724] [1.5873, ] Non-Inflaon Targeers CHN [.6535, 1.75] [.2656, ] IND [.7767, 1.159] [.4934,.9566] MYS [.4838,.9863] [.4613,.9949] SGP [.7397, ] [.71, 1.786] Noes: The HL(M) s he half-lfe measured n monhs by ln(.5 b(1ˆ)) ( 1 n ). The wo-sded 95% confdence nervals (CIs) measured n uarers are consruced accordng o h 1.96se( ˆ )([ ln.5 ˆ ][ln( ˆ )] ), where ˆd ˆ dˆ 2 2 d se ( ˆ ) 2 ( n ), and perssence parameer d ˆ 1 ˆ n, <d<1. (see also Km and Lma, 21) 18

19 Ol Prce Table 3: Pass-hrough effecs based on auomercs ree search procedure Inflaon Targeers Non-Inflaon Targeers Perod IDN PHL THA KOR CHN IND MYS SGP Passhrough Effecs Exchange Rae Pre- Shf 1 Pos- Shf 1 Pos- Shf 2 Pre- Shf 1 Pos- Shf 1 Pos- Shf 2 Shorrun Longrun Shorrun Longrun Shorrun Longrun Shorrun Longrun Shorrun Longrun Shorrun Longrun (.419) (-.892) (.5898) (-.1423) (.1347) (-.747) (-.3249) ( ) (-.579).57 (.334) (-.831).91 (.533) (.64) (.353) (-.272) (.8445) (.326) (.736) -.1 (.759).28 (-.5414).92 (.126).954 (.284).6 (-.251) (.534) (.334) (.1548).14 (.6) (-.196).597 (.257).717 (.1191).14 (-.113) (.369) Inflaon nera-shf 1 (.9322) (.2311) (.9132) (.5218) Inflaon nera-shf Oupu gap (-.189) -.35 (-.7) -.3 (-.95) -.1 (-.29) Noes: The pass-hrough coeffcen s based on auhor s own calculaon. The las hree rows represen he coeffcen for nflaon nera and oupu gap from ARDL model whch based on auomercs ree search procedure sa dcussed n ex. The values n ( ) ndcaes he pass-hrough effec by usng offcal IT adopon dae. 19

20 Table 4: Summary of pass-hrough effecs Inflaon Targeers Non-Inflaon Targeers Perod IDN PHL THA KOR CHN IND MYS SGP Pos- Shf 1 Passhrough Effecs Exchange Rae Shorrun (+ ) Longrun (+ -) - (- ) - (- ) - (+ ) - + (+ -) - + (+ ) - (+ -) Pos- Shf 2 Shorrun Longrun Ol Prce Pos- Shf 1 Shorrun (- +) Longrun (- ) + - (+ ) + - (+ ) + (+ -) + - (+ ) + (+ -) + - (+ ) Pos- Shorrun Shf 2 Longrun Noes: The pass-hrough coeffcen s based on auhor s own calculaon. Ths able summarzed he pass-hrough mpacs from Table 3. The arrow n ( ) ndcaes he pass-hrough effec by usng offcal IT adopon dae.,, + - and - + denoes declne, ncrease, change from posve mpac o negave mpac, and change from negave mpac o posve, respecvely. 2