Real Exchange Rates Behavior in Selected EU Member States: Assessment of the Financial Crisis Effect

Transcription

1 Real Exchange Raes Behavior in Seleced EU Member Saes: Assessmen of he Financial Crisis Effec Daniel Savárek Silesian Universiy in Opava School of Business Adminisraion in Karviná, Deparmen of Finance and Accouning Univerziní nám. 1934/3 Karviná, Czechia Absrac The real exchange rae is one of he crucial macroeconomic variables for all open economies. Therefore, analysis of is evoluion as well as volailiy and behavior of is componens (nominal exchange rae and relaive prices) is of criical imporance for boh he economic heory and economic policy. In his paper, we focus on he ineracion among he componen variables of he real exchange rae. The main objecive of his paper is evaluae how he relaive prices affec he exchange rae. We calculae volailiy measure and apply he Granger causaliy es, variance decomposiion and impulse-response funcion in he Vecor Auo Regression model for six seleced non-euro EU member saes (Czechia, Hungary, Poland, Denmark, Sweden and he Unied Kingdom). The calculaions are conduced for wo periods disinguished as he pre-crisis period and he pos-crisis period. The resuls differ subsanially beween he periods and provide evidence ha he relaive prices play more imporan role in explaining he exchange rae behavior in he pos-crisis period han before is origin. Keywords: real exchange rae, volailiy, Granger causaliy, variance decomposiion, impulse-response funcion JEL codes: E32, E44, F31 1. Inroducion The real exchange rae is one of he mos imporan indicaors in macroeconomics and economic policy as is changes and flucuaions have implicaions for boh exernal compeiiveness as well as resource allocaion wihin he economy. The real exchange rae also plays a crucial role in numerous models of he open economy. The quesions on real exchange rae developmen, deerminans, volailiy and effecs have been frequenly posed in economic research. The imporance of real exchange rae and is monioring even increased in recen years as a growing share of economic aciviies are direcly or indirecly affeced by economic developmen in oher economies. Moreover, all he issues associaed wih he real exchange rae behavior have aken on heighened imporance in he curren period of economic slowdown and recession. Since he real exchange rae is one of he mos comprehensive indicaors of he counry s compeiiveness on inernaional markes, he cross-counry comparisons poin o fuure prospecs of he counry s economy and provides a good guidance for he officials, businessmen and inernaional organizaions. This paper focuses on volailiy of he real exchange rae. The relevan lieraure in his area can be divided ino four caegories. According o Ouyang and Rajan (2013), he firs sream of lieraure links he volailiy o he exchange rae arrangemen and aribues he increase in volailiy o he shif from fixed o flexible exchange rae regime. The second se of sudies generally use Vecor Auo Regression (VAR) mehods and variance decomposiion procedures o idenify relaive conribuion of real and nominal shocks o he real exchange rae flucuaions. The hird caegory of lieraure deals wih he fundamenal deerminans of he long-run equilibrium real exchange rae such as produciviy, invesmen posiion, foreign invesmen or fiscal indicaors. The fourh line of lieraure employs various echniques o decompose real exchange rae volailiy ino is wo subcomponens exernal prices (deviaions from he Purchasing Power Pariy) and inernal prices (relaive price of radeable and nonradeable goods). 366

2 This sudy examines behavior of he real exchange rae in seleced EU member saes ouside he euro area. In paricular, we invesigae he ineracion among he componen variables of he real exchange rae, i.e. exchange rae and relaive prices. The main aim of he paper is o find ou wheher he causaliy ha movemen of exchange rae is influenced by changes in relaive prices hold and o evaluae he degree of impac of he relaive prices on he exchange rae. Moreover, he paper brings a new perspecive ino he analysis as i compares he real exchange rae behavior and ineracion beween he componens in he pre-crisis and pos-crisis periods. The group of seleced counries includes six non-euro EU member saes ou of which hree are new member saes from he Cenral Europe (Czechia, Hungary, Poland) and hree counries are radiional developed EU members (Denmark, Sweden, Unied Kingdom). Hereafer, he counries are denoed as CZ, HU, PL, DK, SE, and UK, respecively. 2. Daa and Research Mehodology The real exchange rae can be expressed as he nominal exchange rae adjused for relaive price level differences beween domesic and foreign economy. In order o obain he real exchange rae in logarihmic form one can employ he sandard formula (1): q s + p p * (1) where q is he real exchange rae, s is he nominal exchange rae, p is he domesic price level and p * is he foreign price level. The daa are all colleced from he Economy and Finance daabase available on he Eurosa websie. All daa are on monhly basis and cover wo periods. In order o obain consisen resuls we exclude he crisis period (2008: :06) from our analysis and compare he real exchange raes behavior in he pre-crisis and pos-crisis period. The pre-crisis period covers 2002: :12 and he pos-crisis period spans from 2009:07 o 2015:06. Hence, boh periods include 72 monhly observaions. The nominal exchange rae represens monhly average of daily spo exchange raes of naional currencies agains he euro and i is quoed as he price of euro in naional currency unis. The price levels are HICP indices defined as 2005=100. The price level in he euro area is aken as he foreign price level for compuaion of he real exchange rae. The firs empirical ool o invesigae real exchange rae behavior is compuaion of volailiy. We apply he measure of volailiy used by Hausmann e al. (2006), which is he sandard deviaion of he growh rae of he real exchange rae. Formally, our volailiy measure is given by vol i = SD(ln( qi ) ln( qi, n, n )) (2) where n is he number of quarers. We experimen wih he one-monh and hree-monh volailiy indicaors and compare he resuls beween he periods and across he counries. In he nex sep of empirical analysis we examine he relaionship beween he wo componens of he real exchange rae, i.e. he nominal exchange rae and he relaive prices. This analysis is conduced by means of a VAR model. Before seing up he VAR model we verify he long-run sabiliy of he wo real exchange rae componens using wo alernaive uni roo ess. In paricular, we apply he augmened Dickey-Fuller (ADF) and he Phillips-Peron (PP) ess in order o examine saionariy of all series during he pre-crisis and pos-crisis periods. Since boh es have been exensively used in lieraure heir formal derivaion and formulaion are no presened in he paper. However, i is worh o noe ha he ADF es accouns for emporally dependen and heerogeneously disribued errors by including lagged innovaion sequences in he fied regression. By conras, he PP es accouns for nonindependen and idenically disribued processes using a non-parameric procedure. Since he ADF es relies on a parameric procedure o correc for auocorrelaion and heerogeneiy, he PP es is ofen favored over he ADF es in erm of power (Taguchi, 2010). By applicaion of he VAR model we can consequenly use number of relaed echniques o shed some ligh on he main channels of ineracion among he variables in he sysem, i.e. he nominal 367

3 exchange rae and he relaive prices. Namely we use he Granger causaliy es, variance decomposiion and impulse-response analysis. The Granger causaliy refers o a specific noion of causaliy in ime-series analysis. A ime series X is said o Granger-cause Y if i can be shown, usually hrough a series of -ess and F-ess on lagged values of X (and wih lagged values of Y also included), ha hose X values provide saisically significan informaion abou fuure values of Y. The variance decomposiion represens he proporion of he oal variance of each variable ha is aribuable o each of he orhogonalized innovaions. I measures he overall relaive imporance of an individual variable in generaing variaions due o is own shock as well as shocks due o oher variables in he sysem. Because he Granger causaliy may no show he full picure abou he ineracions beween he variables of he sysem we also apply he impulse response funcions. These funcions race he dynamic responses o he effec of a shock in one endogeneous variable on all endogeneous variables in he sysem. In oher words, he impulse response funcions map ou he dynamic response pah of a variable due o a one-period sandard deviaion shock o anoher variable. 3. Real Exchange Raes Developmen and Volailiy Before we sar examining volailiy of he real exchange raes i is crucial o demonsrae and discuss evoluion of real exchange raes in all analyzed counries. We presen he developmen in he new EU member saes in Figure 1 and he developmen in radiional member saes in Figure 2. For graphical convenience, we choose o sudy all currencies under he base 2002 = 100. An increase in he index represens a weakening of he local currency and srenghening of he euro. The wo dashed verical lines mark ou he pre-crisis period (2002: :12), he crisis period (2008: :06) and he pos-crisis period (2009: :06). Figure 1: Real Exchange Raes Evoluion in New EU Member Saes (2002: :06) Source: auhor s calculaions based on daa from he Eurosa Economy and Finance daabase One can disinguish very differen developmen of he real exchange raes in he new member saes during he pre-crisis period. While he Czech koruna experienced a gradual real appreciaion of 6.5%, he Hungarian forin depreciaed in real erms of abou 4%. The mos urbulen evoluion can be observed in he case of Polish zloy. Alhough he real exchange rae a he end of he pre-crisis period was almos idenical wih he value a he beginning he zloy iniially depreciaed of abou 22% over he firs wo years and hen appreciaed back during he remaining four years. The pos-crisis 368

4 developmen of he real exchange raes seems o be more homogeneous in he group of new member saes as he Polish zloy was oscillaing around a cerain and relaively sable level. The change of he real exchange rae during he pos-crisis period is no remarkably high in none of he counries. Whereas he Polish zloy appreciaed of 4%, he Czech koruna and Hungarian forin depreciaed of 1% and 3%, respecively. When comparing he real exchange rae evoluion in radiional non-euro EU member saes he exchange rae arrangemen in Denmark should be aken ino accoun. Denmark mainains a fixedexchange-rae policy vis-à-vis he euro area and paricipaes in he European Exchange Rae Mechanism, ERM 2, a a cenral rae of kroner per 100 euro wih a flucuaion band of +/- 2.25%. Therefore, he relaive sabiliy of he Danish kroner nominal exchange rae is ransferred ino sabiliy of he real exchange rae and one can see almos no exchange rae flucuaions over he whole period analyzed. During he pre-crisis period he Swedish korona and Briish pound also experienced a sable developmen. The only excepion was he 8% real depreciaion of he pound in The overall changes of he real exchange raes in he pre-crisis period are as follows: Danish kroner appreciaed of 1%, Swedish korona appreciaed of 0.5%, and Briish pound depreciaed in real erms of 7.5%. By conras, he pos-crisis period is more urbulen for boh he Swedish korona and he Briish pound. As ypically documened in currency and financial crisis, he real exchange rae overshoos a he shock and hen appreciaes afer some ime (Couder e al., 2011). However, his is he only one common feaure aribuable o boh currencies. The Swedish korona sared he pos-crisis period wih real appreciaion ha was replaced by depreciaion afer four years. As a resul, he korona appreciaed of 8% during he pos-crisis period. The Briish pound was oscillaing around he saring level during he firs four years and hen embarked on appreciaion pah, which resuled o overall appreciaion of 5.5%. Figure 2: Real Exchange Raes Evoluion in Tradiional EU Member Saes (2002: :06) Source: auhor s calculaions based on daa from he Eurosa Economy and Finance daabase As Mabin (2010) poins ou he shor-erm volailiy reflecs monh-o-monh changes in real exchange raes, up o a maximum of one year. We can observe his as he exchange rae moves around he cyclical exchange rae. These flucuaions in he real effecive exchange rae usually sem from changes in he nominal exchange rae. We compue and examine he volailiy of he real exchange rae by a measure formulaed in (1). Following he approach of Mollick (2009), he volailiy indicaor is calculaed for each exchange rae over hree differen periods, i.e. he whole period covered by he daase (2002: :06), he pre-crisis period and he pos-crisis period. Subsequenly, we compue 369

5 he growh rae in volailiy beween he wo sub-periods. Moreover, he volailiy was calculaed from one-monh and hree-monh changes of he real exchange raes. The resuls are summarized in Table 1. Table 1: Volailiy of Real Exchange Raes Change monh volailiy CZ % HU % PL % DK % SE % UK % 3-monh volailiy CZ % HU % PL % DK % SE % UK % Source: auhor s calculaions Two crucial findings can be revealed in he resuls. Firs, he volailiy of he real exchange rae is higher in he new member saes han radiional members. A subsanial difference in volailiy beween he wo groups of counries can be found paricularly in he pre-crisis period. The volailiy converged significanly during he pos-crisis period and volailiy of he Czech koruna real exchange raes is even lower han volailiy of he Swedish korona and Briish pound. Second, he growh rae of volailiy beween he pre-crisis and pos-crisis periods is considerably higher in he group of radiional EU member saes he in he newcomers. Similar finding for real effecive exchange raes are presened by Savárek and Migliei (2015). Focusing on he one-monh volailiy, one can observe ha he range of growh raes for he new member saes is from -5.68% in Poland o % in Czechia. By conras, he growh raes in he group of radiional EU members vary from % in he UK o % in Sweden. A very similar picure is revealed if one concenraes on he hree-monh volailiy. While he volailiy in he new members changed from % in Poland o % in Czechia he growh raes in he group of radiional members range from % in Denmark o % in Sweden. 4. Ineracion beween he Real Exchange Rae Componens In his secion we repor and discuss resuls of he Granger causaliy ess, variance decomposiions and impulse-response funcions in order o examine he ineracion beween he nominal exchange rae and relaive prices. The major concern in hese analyses is o prove he causaliy from he relaive prices o he exchange rae and o deermine he degree of impac of he relaive prices o he exchange rae. Before conducing all he menioned empirical procedures, we es individually for uni roos on all componens of he real exchange rae using he ADF and PP ess. These resuls are omied for space consrains bu are available upon reques. The ADF and PP ess equally do no rejec he uni roo null hypohesis in levels and does rejec i in firs-differences. This finding is revealed for he pre-crisis as well as he pos-crisis period. Therefore, one can conclude ha he nominal exchange raes and relaive prices follow I(1) processes a sandard significance level in all counries analyzed. Based on his conclusion we can proceed wih consrucion of VAR models and applicaion of associaed analyses. 370

6 CZ HU PL DK SE UK Table 2: Granger Causaliy Tes F saisics Probabiliy F saisics Probabiliy ER -/- Pdif * Pdif -/- ER * ER -/- Pdif Pdif -/- ER ** ER -/- Pdif *** Pdif -/- ER ER -/- Pdif ** Pdif -/- ER ER -/- Pdif Pdif -/- ER ER -/- Pdif Pdif -/- ER ** Noe: ER is he nominal exchange rae, Pdif is he price differenial (relaive prices), *,**,*** denoe significance on 1%, 5% and 10% level, respecively Source: auhor s calculaions The resuls of he Granger causaliy es are given in Table 2. One can find only one example of he Granger causaliy during he pre-crisis period. The causaliy from he relaive prices o he exchange rae is revealed in Hungary. More evidence on Granger causaliy beween he componens of he real exchange rae was discovered in he pos-crisis period. There are wo cases of causaliy in which he relaive prices Granger-cause he exchange rae (Czechia and Unied Kingdom). There are hree more examples of he reverse causaliy, i.e. he exchange rae Granger-causes he relaive prices (Czechia, Poland, Denmark). One can conclude ha he wo examined periods yield compleely differen resuls as regards he srengh and direcion of he causaliy beween he variables. However, no general conclusion can be drawn as he resuls differs across he counries. The oucomes of he variance decomposiion analysis are graphically depiced in Figure 3. In accordance wih he main objecive of he paper we only repor he variance decomposiion of he exchange rae in order o realize how much of he exchange rae variaions can be explained by he price differenial. The share of variance explained by he relaive prices usually rises wih he increasing ime lag. In he pre-crisis period he relaive prices explain afer 12 monhs 4.9% of he exchange rae variance in Czechia, 14% in Hungary, 8.1% in Poland, 3.2% in Denmark, 1.5% in Sweden, and 3.7% in he UK. I is eviden ha he highes conribuion of he relaive prices was revealed in Hungary, which is he only case of he Granger causaliy leading from relaive prices o exchange rae idenified in he precrisis period. As can be seen in he graphs he share of he price differenial in he exchange rae variance decomposiion increased in all counries during he pos-crisis period. Specifically, i was 13% of he variance explained afer 12 monhs in Czechia, 16.3% in Hungary, 15.8% in Poland, 10.8% in Denmark, 5.4% in Sweden and 14.9% in he UK. A noeworhy join resul of he Granger causaliy ess and variance decomposiion analysis appears o be ha in all sample counries he relaive prices play a more significan role in explaining he exchange raes afer he financial crisis han before is oubreak. 371

7 Figure 3: Variance Decomposiion in he VAR Model Noe: ER is he nominal exchange rae, Pdif is he price differenial (relaive prices) Source: auhor s calculaions Figure 4 shows he dynamic response paern of he exchange rae o innovaion in he price differenial by using he impulse-response funcions wihin he consruced VAR model. I is apparen from he funcions ha he response of he exchange rae o a shock in he relaive prices changed considerably in he pos-crisis period. In all analyzed counries one can observe ha he pos-crisis response is more inense, more dynamic and less permanen han he pre-crisis reacion. Addiionally, he iniial response (1-3 monhs) was found o be compleely opposie when comparing he wo periods. For insance, in Czechia and Sweden he shock o relaive prices led o depreciaion of he naional currency in wo consecuive monhs in he pre-crisis period bu conribued o is appreciaion in he pos-crisis period. By conras, he pre-crisis appreciaion in Poland, Denmark and UK was convered ino depreciaion in he pos-crisis period. There is one more remarkable finding ha almos all 372

8 currencies share in common. While he exchange rae response in he pre-crisis period gradually dies ou and is very close o zero afer 12 monhs such a fading in he pos-crisis period is apparen only in Sweden. The response of exchange rae of remaining currencies show no negligible values even 12 monhs afer he shock o relaive prices. Figure 4: Impulse-Response Funcion of Exchange Rae o Shock in Relaive Prices 5. Conclusion Source: auhor s calculaions The aim of he paper was o find ou wheher he causaliy ha movemen of exchange rae is influenced by changes in relaive prices hold and o evaluae he degree of impac of he relaive prices on he exchange rae. The analysis was conduced on a diverse sample of six non-euro EU member saes. Since our expecaion was ha he global financial crisis affeced behavior of he real exchange 373

9 raes as well as ineracion among he componen variables we run all he ess and esimaions for he pre-crisis and pos-crisis period. This crisis period (2008: :06) was excluded from our analysis. The resuls obained confirm our expecaions as hey show subsanial differences in findings from he pre-crisis and pos-crisis periods. During he pre-crisis period he real exchange raes in he new EU member saes exhibi considerably higher volailiy han he exchange raes in he radiional members. Alhough he financial crisis brough a growh of volailiy o five of he six counries examined one can idenify an unequal effec of he crisis. The pos-crisis volailiy of he real exchange raes in he radiional member saes increased remarkably and reached he level usual in he new members. For insance, he one-monh and hree-monh volailiy measures in Sweden increased of 70% and 106% beween he periods. By conras, he same indicaors in Poland decreased of 5% and 10%. As a resul, he real exchange rae volailiy in Czechia was lower han he volailiy in Sweden and he Unied Kingdom during he pos-crisis period. I is eviden ha he crisis changed economic environmen more considerably in he radiional member saes and he real exchange rae responded by a growing volailiy. We applied he Granger causaliy es, variance decomposiion and impulse-response funcions o examine he ineracion beween he nominal exchange rae and relaive prices. Similarly wih findings on he volailiy one can conclude ha he role of relaive prices in explaining he exchange rae evoluion and behavior is remarkably differen in each of he periods analyzed. Afer he crisis, we revealed significanly more cases of Granger causaliy beween he componens of he real exchange rae including he examples where he pas values of relaive prices help in predicion of fuure values of he exchange rae. Likewise, he conribuion of relaive prices in explaining variance of he exchange rae increased remarkably in all sample counries in he pos-crisis period. The more pronounced role of relaive prices is confirmed also by a shape of he impulse-response funcions. Afer he crisis, he response of he exchange rae o a shock in he relaive prices was greaer in inensiy, dynamics and persisence. The paper, hence, provides empirical evidence ha paricularly in he pos-crisis period we canno rejec he assumpion ha movemen of exchange rae is influenced by changes in relaive prices. Acknowledgemen Publicaion of his paper was suppored by he Suden Gran Sysem of Silesian Universiy (projec SGS/7/2013). The suppor is graefully acknowledged. References COUDERT, V., COUHARDE, C., MIGNON, V. (2011). Exchange rae volailiy across financial crises. Journal of Banking & Finance, vol. 35, no. 11, pp HAUSMANN, R., PANIZZA, U., RIGOBON, R. (2006). The long-run volailiy puzzle of he real exchange rae. Journal of Inernaional Money and Finance, vol. 25, pp MABIN, G. (2010). New Zealand s Exchange Rae Cycles: Evidence and Drivers. New Zealand Treasury Working Paper 10/10. Wellingon: New Zealand Treasury. MOLLICK, A. V. (2009). Crisis and volailiy in Asian versus Lain American real exchange raes. Économie inernaioanale, vol. 117, pp OUYANG, A. Y., RAJAN, R. S. (2013). Real exchange rae flucuaions and he relaive imporance of nonradables. Journal of Inernaional Money and Finance, vol. 32, pp STAVÁREK, D., MIGLIETTI, C. (2015). Effecive Exchange Raes in Cenral and Easern European Counries: Cyclicaliy and Relaionship wih Macroeconomic Fundamenals. Review of Economic Perspecives - Národohospodářský obzor, vol. 15, no. 2, pp TAGUCHI, H. (2010). The pre- and pos-crisis real exchange rae behavior in seleced Eas Asian counries. Sudies in Regional Science, vol. 40, no. 1, pp