EXCHANGE RATE PASS-THROUGH TO DOMESTIC PRICES: THE CASE OF SOUTH AFRICA

Transcription

1 Aricles EXCHANGE RATE PASS-THROUGH TO DOMESTIC PRICES: THE CASE OF SOUTH AFRICA Mahew Kofi Ocran* Absrac: This paper examines he exchange rae pass-hrough o impor, producer and consumer prices in Souh Africa using monhly daa covering he period 2000M1 o 2009M5. The sudy uses innovaion accouning ools (impulse response and variance decomposiion) wihin he framework of an unresriced VAR o examine he degree of pass-hrough as well as he relaive imporance of a number of variables in explaining changes in domesic prices. The key fi ndings sugges ha afer 1 per cen shock o nominal effecive exchange rae, he level of CPI increases by per cen, giving a pass-hrough elasiciy of 13 per cen. However, he pass-hrough elasiciy of producer price is 20 per cen afer 24 monhs suggesing ha favourable shocks o producer price infl aion can have considerable moderaing effec on CPI inflaion. Keywords: exchange rae, pass-hrough, domesic prices, VAR, Souh Africa. JEL Classificaion: C32, E31, E37, F31, O52 1. Inroducion While he debae in developed counries regarding exchange rae pass-hrough (ERPT) has moved on from one of esimaion o he ask of explaining he observed low exchange rae pass-hrough, he discussion in mos developing counries revolves around he level of pass-hrough. The case of Souh Africa is even more perinen since he rae of inflaion has fallen considerably over he years. From an average rae of 19.2 per cen in 1986 he inflaion rae dropped o 9.8 per cen in 1993 and boomed * Deparmen of Economics, Nelson Mandela Meropolian Universiy, Souh Summer Srand Campus, Por Elizabeh 6031, Souh Africa (mahew.ocran@nmmu.ac.za). The auhor acknowledges very useful suggesions from hree anonymous referees of he journal on an earlier draf. Consrucive criicisms from paricipans of he biennial conference of he Economic Sociey of Souh Africa in Por Elizabeh, Sepember 2009, are also much appreciaed. PRAGUE ECONOMIC PAPERS, 4,

2 ou a 1.4 per cen in 2004 before peaking again a 11.5 per cen in Indeed if we consider he year 2001 when he counry adoped an explici inflaion argeing regime as reference poin we noe ha he rae of inflaion came down considerably unil he recen upsurge associaed wih he commodiy price super-cycle ha ended in For insance, he rae of inflaion on an annual basis, which averaged 6.0 per cen for he nine-year period afer adopion of explici inflaion argeing, compares favourably wih he high of 9.4 per cen observed for he period preceding he argeing regime. 2 Given ha he lieraure on developed counries appears o sugges ha exchange rae pass-hrough in an environmen of low inflaion is more subdued, i would be ineresing o ascerain he level of pass-hrough in Souh Africa over a period of low inflaion. This would help deepen he lieraure by bringing an emerging marke economy oucome o bear on he curren lieraure, and also o indirecly conribue o he debae on he usefulness of he inflaion argeing regime currenly raging in he counry. Exchange rae pass-hrough essenially deals wih he effec of a change in he exchange rae on domesic prices, be i consumer prices, producer prices or impor prices. Pass-hrough is deemed incomplee when he effec of a uni change in he exchange rae leads o less han a uni change in a given domesic price. Campa and Goldberg (2005) argue ha here are hree key facors ha drive he exen of exchange rae pass-hrough. As cied in Liu and Tsang (2008), hese are: he pricing behaviour of exporers in he producer counries, he sensiiviy of mark-ups o compeiive condiions and exisence of disribuion coss ha end o widen he difference beween impor and reail coss. Significanly, he las wo consiue a se of counervailing forces wih a ne effec ha can impac exchange rae pass-hrough in one way or he oher. Following he discussion above, he empirical quesion ha comes o he fore is: wha is he exen and naure of exchange rae pass-hrough o domesic prices in Souh Africa? The objecive of his paper herefore is o esimae he exchange rae pass-hrough o domesic prices in Souh Africa using an unresriced VAR model. In general, if domesic prices respond o nominal exchange rae depreciaion one-o-one hen any expor compeiiveness ha accrues from nominal depreciaion would be wiped ou, as he real exchange rae would no change a all (Io and Sao, 2008). Given ha he exchange rae is an insrumen for managing he rade balance, he pass-hrough o impor prices and producer prices may influence expendiure swiching while high pass-hrough o consumer prices can bring pressure o bear on inflaion levels. Increased undersanding of exchange rae pass-hrough in he economy can aid boh rade and moneary policy formulaions; his consiues he moivaion for he presen paper. The res of he paper is srucured as follows: Secion 2 reviews evoluion of Souh Africa s exchange rae policy as well as developmens in he consumer price inflaion over he recen pas. Secion 3 presens a brief overview of he proximae lieraure on exchange rae pass-hrough. In Secion 4, he mehodology and empirical model is discussed. The key findings are hen discussed in Secion 5, wih he conclusions and policy issues in Secion 6. 1 Consumer price index for all meropolian areas all imes; base year: 2008=100. Source: Saisics Souh Africa, Daa Files. 2 The difference in average inflaion beween he wo periods is found o be saisically significan a he 5 per cen level when a saisical es of he means was underaken. 292 PRAGUE ECONOMIC PAPERS, 4, 2010

3 2. Evoluion of Exchange Rae Policy and Recen Developmens in Inflaion This secion provides a concise review of he developmens in exchange rae policy in Souh Africa over he pas hree or so decades as well as recen performance of inflaion in he counry. The discussion is mosly descripive in naure. 2.1 Exchange Rae Exchange rae regimes vary widely from counry o counry and even in counries from ime o ime. However, he wo exremes in erms of policy are he freely floaing and fixed regimes respecively. While he exchange rae level in he freely floaing regime is enirely lef o marke forces, wih no official direc inervenion excep indirecly by policy measures, he auhoriies may seek o keep he exchange rae fixed, as was he case prior o 1914 when he gold sandard broke down. Now, beween hese wo policy exremes here are also a number of variaions or possibiliies. The exchange rae policy in Souh Africa has evolved over he years. 3 For insance, up unil 1979, he exchange rae was largely fixed and pegged o eiher he Briish pound serling or he US dollar. Indeed, over he 1970s he peg even alernaed beween he dollar and he pound serling a few imes. As par of he exchange rae managemen, conrols on capial flows/foreign exchange ransacions were enforced. Afer ighening he conrols in he wake of he poliical urmoil of 1961 following he Sharpville Shooings, 1976 saw a moderaion of he conrols when a dual currency exchange rae sysem was pursued (De Kock Commission, 1984 as cied in Van der Merwe, 1996). The regime provided for a financial rand, which was applied o non-residen porfolio and direc invesmens, and a commercial rand for all oher foreign exchange ransacions. The commercial rand was quasi-marke deermined. The Reserve Bank on a daily basis in consonance wih marke forces announced he commercial rand exchange rae. While he financial rand was freely floaing he same canno be said of he commercial rand. The financial rand was dropped in 1983 in an aemp o liberalise he exchange rae and as a resul all conrols on non-residen capial movemens were removed. However, hose on residens remained in place bu were somewha relaxed. Due o he difficul economic siuaion precipiaed by he deb crisis of he mid 1980s, he financial rand was brough back in 1985 bu his was again removed afer 10 years (i.e. in 1995) when he exchange rae sysem was unified. Since hen a gradual approach has been adoped o liberalise he exchange rae o he exen ha by 2004 he exchange rae was characerised by he IMF as one of independen floa. Figure 1 presens he exchange rae developmens since I is apparen ha since 1980 here has been a general downward rend in he size of he NEER. However, while he fall was relaively sharp, beween he pre-1985 and he dual exchange raes regime he pos-1995 decline was somewha measured. On he maer of shor-erm variabiliy here appear o be marked differences in he degree of variabiliy beween he period of dual exchange rae regime and ha of he floaing exchange rae regime. 3 For a deailed descripion of he evoluion of exchange rae policies in Souh Africa up o 2004, see Farrell and Todani (2004). PRAGUE ECONOMIC PAPERS, 4,

4 Figure 1 The NEER under Differen Regimes (2005=100) 600 NEER 500 Dual Exchange Rae Regime wih conrols Unified and Independen Floaing Exchange Rae Apar from he wide variaions in he firs five years following he re-inroducion of he dual exchange rae, he las five years of he regime, , were characerised by relaively moderae variaions as compared o he floaing exchange rae regime. Figure 2 Shor-Term Variabiliy of he NEER Expressed as Percenage Change from Previous Quarer (1980:1 o 2008:3) 0.20 DLNEER Dual Exchange Rae Regime wih conrols Unified and Independen Floaing Exchange Rae PRAGUE ECONOMIC PAPERS, 4, 2010

5 Recen Developmens in Souh Africa s Inflaion Generally, he inflaion level in Souh Africa has been on a seady decline since 1990 apar from he spike in 2002 due o economic challenges ha included he large flucuaions in he exernal value of he rand 4 (see Bhundia, 2002). Boh he GDP deflaor and CPI measures of inflaion respecively have followed a downward rend. However, inflaion in he counry sared picking up seadily from 2005 afer reaching more han a 20-year low of 1.4 per cen in The average CPI inflaion for 2008 was 11.5 per cen, he highes since Since 1985 boh he GDP deflaor and headline inflaion as measured by he CPI have largely moved ogeher unil 2004 when hey diverged again bu he wo have since come close o convergence. A look a he GDP deflaor 5 a raher broad measure of inflaion also sugges a subdued inflaionary rend similar o ha observed in he headline inflaion, CPI. Thus from Figure 3 i can be argued ha while shor-run variabiliy in he NEER increased in he pos-1995 period he impac on inflaion appear o have been quie subdued a leas from descripive analysis poin of view. The empirical response o he quesion hough is he subjec of his paper. Figure 3 Trends in Inflaion, Inflaion, consumer price index (% change) Inflaion, GDP deflaor (annual %) For insance, he rand los abou 40 per cen of is value in he las four monhs of The developmens in domesic prices are bes analysed wih he CPI insead of he GDP deflaor. Even hough i does no ake accoun of invesmen spending and exernal rade as in GDP deflaor, i is deemed more useful for analysis of domesic price movemens because i dwells on domesic consumpion behaviour only. In any even, Figure 3 suggess ha boh measures have moved ogeher over he pas decades or so. PRAGUE ECONOMIC PAPERS, 4,

6 3. Lieraure Review This secion of he paper presens a brief overview of he lieraure regarding exchange rae pass-hrough (ERPT). While noing ha he lieraure is quie exensive i is possible o make some generalisaions ha characerise he body of lieraure around ERPT. The lieraure may be considered a wo main levels of aggregaion. Again, while mos sudies consider ERPT a he macro level here is a subsanial lieraure ha looks a he issues a he micro level. The micro level analyses may be indusry or produc specific (Corsei and Dedola, 2005). These sudies examine he pass-hrough o impor prices of differen indusries or producs raher han dwelling on he effecs a aggregae price levels. On he heoreical fron here appears o be no consensus wih respec o he appropriae heoreical underpinnings regarding ERPT modelling effors. Prominen among he heoreical frameworks is he Law of One (LOP). Criics of he LOP have assered he presence of price sickiness in some cases invalidaes he LOP framework and ha ERPT is imperfec. In he conex of sicky prices, Bes and Devereux (2000) proposed he heory of local currency pricing o explain he breakdown of he PPP heory inheren in he law of one price. Ohers have also argued in favour of producer currency pricing. The argumen here is ha exchange rae changes are refleced in one-for-one foreign currency prices, i.e. perfec pass-hrough. Then here is he Phillip curve framework, which has aained populariy in recen ime (e.g. Liu and Tsang, 2008). The economeric ools used in esimaion of ERPT have evolved over ime wih a number of accompanied moivaions, from single equaions o simulaneous equaions and o a limied exen seemingly unrelaed regressions (SUR). Various ypes of VARs are also in use. There are sudies such as hose ha use recursive VARs, srucural and unsrucured VARs. The lieraure can also be looked a in erms of wo main srands regarding characerisics based on he naure of modelling as o wheher one is using a srucural or reduce-form modelling effor. The firs sage examines he relaionship beween exchange rae and impor prices while he second sage looks a he pass-hrough from impor prices o domesic prices. Noneheless, here are a few papers ha aemp o examine hese sages separaely wihou regard o he oher. These papers have mosly analysed he firs sage wih only a handful of sudies dwelling direcly on he laer (Gagon and Ihrig, 2004). The VAR approach has a marked advanage over he single equaion framework; his explains he populariy of he framework in he lieraure (McCarhy, 2000; Hahn, 2003; Faruqee, 2006; Io and Sao, 2006). Among he key advanages of he VAR framework is he opporuniy o idenify srucural shocks via Cholesky decomposiion of innovaions. Again, while a single equaion framework allows for jus one domesic price index, he VAR framework makes room for a se of domesic prices, hus making i possible for an evaluaion of ERPT considering a se of domesic prices wihin he pricing chain from he imporer/producer o consumer levels. The single equaion framework on he oher hand is based on he assumpion of causaliy from exchange rae o inflaion and hus ignoring he possibiliy of reverse causaliy from inflaion o exchange rae. Consequenly in he presen sudy we adop he VAR approach as he analyical framework for esimaion. 296 PRAGUE ECONOMIC PAPERS, 4, 2010

7 On he subsanive maer of esimaion of ERPT, mos sudies focusing on developed counries conclude ha ERPT has fallen over he years paricularly in an environmen of low inflaion. Consequenly he bulk of he recen lieraure on ERPT has been aimed a explaining away he reasons underlying he seemingly low ERPT figures (Marazzi e al., 2006; Campa and Goldberg, 2005; Taylor, 2000). However, in he case of emerging economies, such as Souh Africa, no much has been accomplished by way of research. Tha nowihsanding, here is a growing number of sudies on emerging economies in Asia (e.g. Io and Sao, 2008). While many of hese have focused on individual counries a good number of hem have dwel on cross-counry samples. The presen sudy herefore aemps o conribue o he lieraure on developing counries by focusing on Souh Africa. In erms of previous work on Souh Africa, Bhundia (2002) is he only sudy ha came o our noice. Unlike Bhundia (2001) we use an exensive daa series ha gives a beer coverage of he inflaion-argeing regime. 4. Mehodology This par of he paper discusses he analyical framework, inuiion and assumpions underlying he empirical model. Also discussed are he relevan daa issues. 4.1 Model Following MacCarhy (2000) and ohers, i is firs assumed ha he various prices involved in he ERPT consiue a disribuion chain, hence he modelling effor is geared a modelling he various prices along he disribuion chain. The asserion here is ha inflaion a each poin or sage of he chain has several componens. For insance, he expeced inflaion a a given poin or sage is deemed o be based on he available informaion a he close of ime -1. The second and hird sages are he effecs of period domesic supply and demand shocks on inflaion a ha phase. The effec of exchange rae shocks on inflaion a a paricular sage consiues he fourh componen. This is followed by he shocks of he previous sages of he disribuion chain. We also have an own shock componen, i.e. ha is a paricular sage s own shock. Now, he poins along he disribuion chain namely, impor, producer and consumer prices may be formally presened as follows: m m s d e m E 1 ) ( (1) w w s d e m w E 1 ) ( (2) c c s d e m w c E 1 ) ( (3) m w c Where,, and denoe impor price, producer price and he consumer price inflaion respecively. One imporan assumpion a his poin is ha he shocks a each sage are par of he sage s inflaion ha canno be accouned for by drawing on informaion from period 1, as well as conemporaneous informaion on local demand and supply variables, exchange raes and inflaion a he preceding sages of he disribuion chain. These shocks may be seen as having resuled from pricing PRAGUE ECONOMIC PAPERS, 4,

8 power and mark-up of firms a various sages. The hird key assumpion underlying he modelling effor is ha impored inflaion shocks do affec local consumer inflaion direcly and indirecly via heir impacs on producer price inflaion. I is also assumed ha here is no room for conemporaneous feedback in he model. The oher assumpions are as follows: he supply shocks are idenified from he dynamics of oil price inflaion; demand shocks are generaed from oupu gap afer considering conemporaneous effec of supply shocks; exchange rae shocks are obained from he dynamics of exchange rae changes afer considering he conemporaneous effecs of he supply and demand shocks. These addiional se of assumpions give rise o equaions (4) o (6). e y 1 ( ) (4) oil oil s E E ( y a (5) 1 ) 1 E s s 1 ) 1 2 d ( e b b (6) Following McCarhy (2000), we inroduce wo more equaions in he form of a cenral bank reacion funcion and a money demand equaion along he lines of Chrisiano e al (1996). The reacion funcion esablishes he relaionship beween shor-erm ineres raes o he earlier idenified variables since he moneary auhoriy uses shor-erm ineres rae as a pre-eminen policy insrumen. The money demand 6 funcion ies he changes in money supply o he oher variables in he earlier models specified. These las wo assumpions give rise o equaions (7) and (8). d c i E ( i c c c c c c (7) s d e m w 1 ) e MP ( (8) s d e m w e m E 1 m) d1 d 2 d 3 d 4 d 5 d 6 d 7 Finally, we assume ha he condiional expecaions inheren in equaions (1) o (8) can be subsiued by linear projecion of he lags of he eigh variables represened in he sysem of equaions. Drawing on he above se of assumpions, we fashion a model in he form of a VAR using Cholesky decomposiion. Consequenly, he impulse response of PPI and CPI inflaion o he orhogonalized shocks of exchange rae changes and impor price inflaion provide esimaes of he effec of hese variables on domesic inflaion. The ordering of he VAR is based on he assumpion menioned earlier regarding he disribuion chain argumen in erms of he pricing srucure. Hence he equaions ener he VAR in he following order: oil price, oupu gap, exchange rae, ineres rae, money supply, impor prices, producer price and consumer price. This may be schemaically represened as: oil gap e i m m w c The oil price and oupu gap are conrols for supply and domesic shocks respecively. Also assumed in he modelling effor is ha causaliy runs from exchange rae o prices and ha he exen of endogeneiy increases from op o boom of he order. MP MD 6 Addiional papers ha discuss he role of moneary policy in he esimaion of exchange passhrough include bu limied o Pigo e al. (1985) and Parsely and Popper (1998). 298 PRAGUE ECONOMIC PAPERS, 4, 2010

9 4.2 Daa Issues The paper uses monhly daa from 1998M1 o 2009M5. The nominal effecive exchange rae (NEER) index 7 is used o accoun for he exchange rae variable. The consumer, producer and impor price indices are obained from daa files of Saisics Souh Africa. The volume of oupu generaed by he manufacuring secor of he economy is used as a proxy for indusrial oupu. Wih he aid of he sandard Hodrick-Presco Filer for daa smoohing we obained he oupu gap. This is defined as he difference beween poenial oupu and acual oupu. The fob price of UK Bren crude was used as a proxy for he world price of crude oil. The US dollar/rand exchange rae was used o conver he crude oil price in US dollar ino he rand price. While he oil price was obained from he IMF s commodiy price daabase, he NEER was aken from he daa file of he Souh Africa Reserve Bank. The prime rae is used as a proxy for he shor-erm ineres rae. Even hough he policy rae, he cenral bank s repurchase rae would have been preferred, he series only sars in 1999M11. The choice of he prime rae is based on he assumpion ha, i is closely linked wih he policy rae. The seasonally adjused money supply series was also obained from he daa files of he Reserve Bank. The series represens broadly defined money supply, M3. Before underaking he empirical analysis we examined he saisical properies of he daa by esing for uni roo o ascerain wheher or no he series was saionary. The ADF es saisic was used in he uni roo analysis. The resuls are presened in he able below. The resuls show ha all he variables are non-saionary in levels bu saionary in heir firs differences excep he oupu gap variable. Table 1 Uni Roo Tes Resuls ADF Tes Variable Levels Firs Difference Oil ** Oupu gap -3.22* -3.22* Exchange Rae ** Impor ** Producer Index ** Consumer price Index ** Ineres rae ** Money supply ** Noe: *** denoes 1 per cen level of significance and for ** 5 per cen level of signifi cance. 7 The rand NEER is compued using he rade weighs of Souh Africa s main rading parners. The weighs are: for he euro area; US, 14.88; China, 12.49; UK, 10.71; Japan, 10.12; Swizerland, 2.83; Ausralia, 2.04; Sweden, 1.99; India, 2.01; Hong Kong, 1.48; Singapore, 1.40; Brazil, 1.37; Israel, 1.11 and Zambia, 0.80 (SARB (2008). PRAGUE ECONOMIC PAPERS, 4,

10 Because he variables are saionary in heir firs differences, i means he variables are inegraed wih order one [I (1)]. Enders (2004) suggess 8 ha we differeniae he variables ha ener he VAR only if hey are no coinegraed. However, our es for coinegraion suggesed ha he I (1) variables are indeed coinegraed. 9 Hence we do no difference he variables ha ener he unresriced VAR. Before he esimaion we also deermined he appropriae lag lengh o use in he esimaion. Here we use a baery of saisics. An inspecion of he various ess indicaed lag hree as he opimal lag lengh (see able below). Table 2 Lag Order Selecion Crieria Lag LogL LR FPE AIC SC HQ NA 5.20e e * * e * 1.85e-21* * e e e e e * indicaes lag order seleced by he crierion. LR: sequenial modifi ed LR es saisic (each es a 5 per cen level). FPE: Final predicion error. AIC: Akaike informaion crierion. SC: Schwarz informaion crierion. HQ: Hannan-Quinn informaion crierion. 5. Main Findings As menioned in Secion 3, he model is esimaed wih he aid of VAR involving eigh variables: oil price, oupu gap, exchange rae, consumer price, producer price, impor price, shor-erm ineres rae and money supply. The implemenaion of he model is such ha he reduced form residuals from he VAR are orhogonalised using a Cholesky decomposiion o idenify he srucural shocks wih he variables in he order given above. The Model is esimaed over he period 2000M1-2009M5 (113 monhs). Following McCarhy (2000) we use innovaion accouning echniques (i.e. impulse response funcions and variance decomposiions) o ascerain he degree of pass-hrough from exchange rae flucuaions o domesic inflaion and also o figure ou he imporance of he various variables in explaining flucuaions in he domesic prices. These exercises are underaken over a 24 monh-ime horizon. 8 See Enders (2004) pp. 288 for he consequences of differencing he variables used in he VAR analysis when hey are coinegraed. 9 The coinegraion es saisics are no repored for he sake of breviy. 300 PRAGUE ECONOMIC PAPERS, 4, 2010

11 5.1 Response o Exchange Rae Shocks Panel 1 shows he responses of he impor price index, he PPI, and he CPI o an exchange rae shock. In his model he exchange rae shock is examined given pas values of all variables as well as curren values of oil prices and he oupu gap. The solid line in each graph indicaes he esimaed response while he broken lines describe a wo-sandard error confidence band around he esimae. The general observaion here is ha impor prices respond more rapidly o exchange rae shocks as compared o he oher wo domesic prices, PPI and CPI. However, beween he wo, PPI is also ahead of CPI. The iniial impac of an exchange rae appreciaion on impor prices is negaive as suggesed by he lieraure and remains so for he enire 24-monh horizon. The response peaks a 0.02 per cen by he 7h monh and hen drops back o 0.01 per cen by he 24h monh. The response o he exchange rae shocks was mosly saisically significan. In he case of he PPI, he response does no appear o be sable. Firs, i reaches a peak a less han per cen, receding o per cen in he 15h monh, before reaching nearly per cen by he 20h monh. I should be noed, however, ha he changes in he exen of response is marginal. The accumulaed response of CPI o shocks from he exchange rae appears o be relaively smaller as compared o he oher wo domesic prices. However, while he response in he second monh was saisically, insignifican responses from he 6h monh were all saisically significan even hough he responses were in hemselves sill marginal. Figure 4 Response of Domesic Indices o 1 per cen Increase in Exchange Rae Accumulaed Response o Cholesky One S.D. Innovaions ± 2 S.E..01 Accumulaed Response of DLMP o DLE.005 Accumulaed Response of DLPPI o DLE.004 Accumulaed Response of DLCPI o DLE This is indeed consisen wih he disribuion chain assumpion underlying domesic pricing oulined earlier oulined in he paper (see Figure 4 above). Esimaion of he Pass-Through Elasiciy Using he impulse response funcion, we can derive he cumulaive pass-hrough coefficien by dividing he cumulaive response of each price index afer m monhs by he exchange rae shock afer m monhs. The exchange rae pass-hrough elasiciy a ime, ERPT is herefore defined as: PRAGUE ECONOMIC PAPERS, 4,

12 P ERPT, m where P is he cumulaive change in price level afer m periods and E cumulaive change in he exchange rae a ime, = 0. The numeraor represens he percenage change in he level of he CPI beween period 0, when he iniial exchange rae shock occurs, and ime. The denominaor is also he percenage change in he nominal effecive exchange rae a ime 0. The exchange rae shocks ge hrough o consumer prices gradually while he effec on impor prices is more pronounced and promp. The impac on producer price is beween he oher wo domesic prices in erms of magniude. Ineresingly, he oucomes here appear o be consisen wih he findings from Bhundia (2002). While in Bhundia s paper he pass-hrough o consumer price inflaion peaks a 12.3 per cen afer eigh quarers (24monhs), here we aained a pass-hrough of 12.5 per cen in 24 monhs. However, wha is mos insighful is he fac ha he pass-hrough o producer price has fallen significanly from 72 per cen afer 24 monhs in he Bhundia paper o 20 per cen in he presen work afer a peak of 40 per cen afer 7 monhs. Table 3 Time Profile for ERPT Elasiciy for Domesic s, m Period Impor Producer Consumer Afer 3 monhs Afer 6 monhs Afer 7 monhs Afer 8 monhs Afer 12 monhs Afer 24 monhs E The figure below shows he comparaive ime profile of he exchange rae pass-hrough seen in all hree domesic prices over 24 monhs (or 2 years). 302 PRAGUE ECONOMIC PAPERS, 4, 2010

13 Figure 5 Esimaed Cumulaive Exchange Rae Pass-Through 0.40 Impor Producer Consumer Variance Decomposiion The error forecas variance decomposiion essenially shows he proporion of he movemens in a sequence due o is own shocks as compared o shocks o oher variables in he sysem. If shocks from oher variables in he sysem do no accoun for any par of he variance of a given variable a all forecas ime horizons, i means he variable is exogenous. On he conrary, if he shocks from he oher variables in he sysem accoun for all he error forecas variance decomposiion, hen he variable we are dealing wih is enirely endogenous (Enders, 2004). In pracice, i is no uncommon for a variable o explain almos compleely is enire forecas error variance, paricularly a shor horizons and relaively smaller proporions a longer horizons. Thus he variance decomposiion provides insigh ino he relaive imporance of each of he random innovaion in affecing he variables in he VAR. As an addiional effor o help ascerain relaive imporance of exchange rae shocks on he various domesic prices we sough o use variance decomposiion o help separae he variaions in he domesic prices, ino he componen shocks o he VAR. Following he discussions above, i can be observed ha hough changes in he exchange rae are imporan in explaining changes in all hree domesic prices: he effecs appear o be subdued as compared wih oil price shocks (see column 3 in Tables 4, 5 and 6). While he role of he exchange rae seems bigger in he producer price index, he effec on consumer prices is relaively mued. Wha is significan hough is PRAGUE ECONOMIC PAPERS, 4,

14 he preeminen role played by supply shocks (due o changes from world oil prices) in he various domesic prices (see column 1 of Tables 4, 5 and 6). Table 4 Variance Decomposiion of Impor Inflaion Period Oil price Oupu gap Exchange rae Ineres rae Money supply Impor Producer Consumer Table 5 Variance Decomposiion of PPI Inflaion Period Oil price Oupu gap Exchange rae Ineres rae Money supply Impor Producer Consumer Table 6 Variance Decomposiion of CPI Inflaion Period Oil price Oupu gap Exchange rae Ineres rae Money supply Impor Producer Consumer I is worh noing he imporance of shor-erm ineres rae in explaining variaions in he CPI. Ineresingly, he impac of ineres rae changes more han doubles afer 3 monhs as he share of he variance decomposiion increases from 3 per cen o almos 9 per cen. The demand shocks from he oupu gap play a modes role in explaining variaions in each of he domesic prices (see columns 5 o 7 in Tables 4, 5 and 6). In sum, i may be argued ha while exchange rae pass-hrough o domesic prices is significan bu small, oil price shocks have a bigger role for explaining variaion in all domesic prices. 304 PRAGUE ECONOMIC PAPERS, 4, 2010

15 6. Conclusions This paper sough o invesigae he exchange rae pass-hrough o domesic prices including CPI, wih he aid of innovaion accouning echniques wihin he framework of an unresriced VAR ha incorporaes a disribuion chain. The findings provide evidence of a link beween exernal facors and consumer price inflaion. Noneheless, as far as he effec of exchange rae shocks is concerned he impac on consumer price inflaion, a major focus of moneary policy, is very modes. Tha nowihsanding some oher insighs is obained. For insance, he effec of oil price shocks seems o be more pronounced a each domesic price level, o he exen ha i is even a par wih shorerm ineres raes in explaining flucuaions in he consumer price inflaion. While he pass-hrough coefficien for consumer price inflaion (12.5 per cen afer 24 monhs) is no very differen from he figure of 12.3 per cen obained by Bhundia (2002), here are sark differences beween he values for he PPI. We had 19 per cen agains Bhundia s 72 per cen, suggesing ha he pass-hrough o PPI has reduced considerably over he years, paricularly over a period of relaively low inflaion. Now, given he closeness of he values beween he PPI and CPI pass-hrough, i can be argued ha favourable shocks o producer price inflaion can have subsanial moderaing effec on CPI inflaion and vice versa. Again, given he relaively low inflaion regime of he period and he associaed low ERPT, i may be argued ha he volailiy of he rand does no pose any serious hrea o inflaion. I is herefore suggesed ha, he Reserve Bank focuses on price sabiliy and no be unduly worried abou he volailiy of he rand. References Bes, C., Devereux, M. B. (2000), Exchange Rae Dynamics in a Model of Pricing o Marke. Journal of Inernaional Economics, Vol. 50, pp Bhundia, A. (2002), An Empirical Invesigaion of Exchange Rae Pass-Through in Souh Africa. IMF Working Paper No. 02/165. Campa, J., Goldberg, L. (2005), Exchange Rae Pass-Through ino Impors s. The Review of Economics and Saisics. Vol. 87, pp Chrisiano, L. J., Eichenbaum, M., Evans, C. L. (1996), The Effecs of Moneary Policy Shocks: Evidence from he Flow of Funds. Review of Economics and Saisics, Vol. 78, pp Corsei, G. and Dedola, L. (2005), The Macroeconomics of Inernaional Discriminaion. Journal of Inernaional Economics, Vol. 67, pp Enders, W. (2004), Applied Economeric Time Series. New York: John Wiley & Sons. Farrell, G., Todani, K. R. (2004), Capial Flows, Exchange Conrol Regulaions and Exchange Rae Policy: The Souh African Experience. Background Paper prepared for he OECD seminar on how o reduce deb coss in Souhern Africa Bond Exchange of Souh Africa, 24 h and 26 h March. Faruqee, H. (2006), Exchange Rae Pass-Through in he Euro Area. IMF Saff Papers, Vol. 53, pp Gagnon, J., Ihrig, J. (2004), Moneary Policy and Exchange Rae Pass-Through. Inernaional Journal of Finance and Economics, Vol. 9, pp Hahn, E. (2003), Pass-Through of Exernal Shocks o Euro Area Infl aion. ECB Working Paper 243. PRAGUE ECONOMIC PAPERS, 4,

16 Ihrig, J., Marazzi, M., Rohenberg, A. (2006), Exchange Rae Pass-Through in he G-7 Counries. Board of Governors of he Federal Reserve Sysem Inernaional Finance Discussion Papers, No Io, T., Sasaki, Y.N., Sao, K. (2005), Pass-Through of Exchange Rae Changes and Macroeconomic Shocks o Domesic Inflaion in Eas Asian Counries, RIETI Discussion Paper Series 05-E-020. Io, T., Sao, K. (2008), Exchange Rae Changes and Infl aion in Pos-Crisis Asian Economies: VAR Analysis of he Exchange Rae Pass-Through. Journal of Money, Credi and Banking, Vol. 44, No 7, pp Liu, L., Tsang, A. (2008), Exchange Rae Pass-Through o Domesic Infl aion in Hong Kong. Hong Kong Moneary Auhoriy Working Paper, 02/2008. Marazzi, M. e al. (2006), Exchange Rae Pass-Through o U.S. Impor s: Some New Evidence. Inernaional Finance Discussion Papers 833 (Washingon, D.C.: Board of Governors of he Federal Reserve Sysem). McCarhy, J. (2000), Pass-Through of Exchange Raes and Impor s o Domesic Infl aion in Some Indusrialized Economies. Working Paper No. 79, Bank for Inernaional Selemens, Basel. Parsley, D. C., Popper, H. A. (1998), Exchange Raes, Domesic s, and Cenral Bank Acions: Recen U.S. Experience. Souhern Economic Journal, Vol. 64, pp Pigo, C., Ruledge, J., Wille, T. D. (1985), Esimaing he Infl aionary Effecs of Exchange Rae Changes. in Sven W. Arnd, Richard J. Sweeney, Thomas D. Wille, eds., Exchange Raes, Trade, and he U.S. Economy, Cambridge, MA: Ballinger, pp Taylor, J. B. (2000), Low Infl aion, Pass-Through, and he Pricing Power of Firms. European Economic Review, Vol. 44, No. 7, pp Van der Merwe, E. J. (1996), Exchange Rae Managemen Policies in Souh Africa: Recen Experiences and Prospecs, Souh Africa Reserve Bank, Occasional Paper No PRAGUE ECONOMIC PAPERS, 4, 2010