The Estimation of the Equilibrium Real Exchange Rate for Romania

Transcription

1 119 European Research Sudies, Volume XII, Issue (1) 2009 The Esimaion of he Equilibrium Real Exchange Rae for Romania Absrac: By Bogdan Andrei DUMITRESCU 1, Vasile DEDU This paper aims o esimae he equilibrium real exchange rae for Romania, respecively he real exchange rae consisen wih he macroeconomic balance, which is achieved when he economy is operaing a full employmen and low inflaion (inernal balance) and has a curren accoun ha is susainable (exernal balance). This equilibrium real exchange rae is very imporan for an economy because deviaions of he real exchange rae from is equilibrium value can affec he compeiiveness of a counry. An overvalued real exchange rae will deermine a lack of exernal compeiiveness and deeriorae he counry s real aciviy. An undervalued exchange rae will increase on shor erm expors and i will lower he curren accoun defici bu, on he long erm i will increase he inflaionary pressures. The equilibrium real exchange rae is also a very imporan variable for a counry who wishes o join ERM II. In fac he cenral pariy should be chosen o reflec he equilibrium exchange rae. The conclusion is ha he real exchange rae had some imporan deviaions from is equilibrium value which were deermined by he liberalizaion of he prices and of he foreign exchange marke and by he flucuaions of he nominal exchange rae. These deviaions are no likely o pu a risk he enry in ERM II. Keywords: equilibrium real exchange rae, produciviy differenial, nominal convergence JEL Classificaion: M41, M42, H83 1. Inroducion This paper aims o provide esimaes for he equilibrium real exchange rae in Romania, respecively he real exchange rae consisen wih he macroeconomic balance, which is achieved when he economy is operaing a full employmen and low inflaion (inernal balance) and a curren accoun ha is susainable (exernal balance). This equilibrium real exchange rae is very 1 Dumirescu Bogdan Andrei, Dedu Vasile, The Academy of Economic Sudies, Faculy of Finance, Insurance, Banking and Sock Exchange, 6, Romana Square, Buchares, Romania; vdedu03@yahoo.com; bdumirescu@homail.com

2 120 European Research Sudies, Volume XII, Issue (1) 2009 imporan for an economy because deviaions of he real exchange rae from is equilibrium value can affec he compeiiveness of a counry. An overvalued real exchange rae will deermine a lack of exernal compeiiveness and deeriorae he counry s real aciviy. An undervalued exchange rae will increase on shor erm expors and i will lower he curren accoun defici bu, on he long erm i will increase he inflaionary pressures. The equilibrium real exchange rae is also a very imporan variable for a counry who wishes o join ERM II. In fac he cenral pariy should be chosen o reflec he equilibrium exchange rae. If no, i will be hard for he cenral bank o limi he flucuaion o +-15% and i could resul a delay in adoping he euro. This paper will use a BEER approach in order o esimae he equilibrium exchange rae. This approach was compued by Clark and MacDonald (1998) and consiss in explaining he acual behavior of he exchange rae in erms of relevan economic variables. The variables used in his paper are: he produciviy differenial beween Romania and he Euro area (produciviy differenial beween he wo secors of he economy), oal consumpion, ne foreign asses and he degree of openness. 2. Lieraure Review The firs aemp o deermine a counries equilibrium exchange rae was made by Gusav Cassel (1922) who inroduced he purchasing power pariy. The PPP heory saes ha exchange raes end o equalize relaive price levels in differen counries. This heory can be seen as a long erm endency for he exchange rae (he value prediced by he PPP heory is an equilibrium value). However, he convergence o PPP is a slow process. Consensus esimaes pu he half-life of deviaions from PPP a abou 4 years for exchange raes among major indusrialized counries. The heory is no valid on he shor erm for various reasons: firs, he exisence of non-radable secor (a secor where prices do no equalize because hey are no subjec of inernaional compeiion) creaes imporan deviaions from he level deermine by PPP. Second, exchange raes end o be higher in rich counries han in poor counries, and relaively fas growing counries experience real exchange rae appreciaion. The economeric esing of PPP evolved from linear regressions o uni roo and coinegraion ess. Balassa and Samuelsson were he firs who showed ha he PPP heory is no valid. They separae he economy in 2 secors: he radable secor (goods) ha is subjec o inernaional compeiion and non-radable secor (services). The produciviy ends o increase more in he radable secor han in he non-radable secor. As a resul, he wages in he radable secor increase and, wih labor being mobile, wages in he enire economy will rise. Producers of non-radables will be able o pay he higher wages only if he relaive price of non-radables rises. This will in general lead o an increase in he overall price level in he economy. For a cach-up counry he produciviy gains are higher so he effec is sronger. Wiliamson (1994) inroduced he concep of fundamenal equilibrium exchange rae (FEER) which is he exchange rae consisen wih he macroeconomic balance, boh inernally and exernally. Inernal balance is he

3 The Esimaion of he Equilibrium Real Exchange Rae for Romania 121 level of oupu consisen wih full employmen (NAIRU) and low inflaion. The exernal balance is he desirable ne flow of resources beween counries when hey are in inernal balance. The FEER exchange rae measure is a normaive one; i is an equilibrium exchange rae consisen wih ideal economic condiions. Clark and MacDonald (1998) inroduced he BEER approach (Behavioral Equilibrium Exchange Rae) which is he mos used echnique for esimaing he equilibrium real exchange rae. Basically, i consiss in esimaing a reduced-form model, which explains he behavior of he real exchange rae on medium and long erm. The model allows calculaing he curren misalignmen and he oal misalignmen from he equilibrium exchange rae. The variables used for explaining he behavior of he real exchange rae are no he same for each counry. Generally speaking, hese variables are: he produciviy differenial (an increase will resul in real appreciaion), he degree of openness (posiive or negaive relaion), erms of rade (an increase will resul in real appreciaion), consumpion (posiive or negaive relaion), real ineres rae differenial (an increase will resul in appreciaion). The BEER approach was used in he esimaion of equilibrium real exchange raes for cenral and easern european counries by Halpern and Wyplosz (2001), De Broeck and Slok (2001), Eger (2002). Sein (1994) inroduced he noion of Naural Real Exchange Rae (NATREX). This heory saes ha he desired capial flows beween wo counries depends on he difference beween invesmens and savings. The variables used are produciviy and savings which influence he capial accoun. The laer influences he real exchange rae by changes in he curren accoun. The equilibrium is obained when he domesic capial sock and ne foreign asses are se o heir equilibrium values. The economeric echniques used are uni roo and coinegraion ess which show he long-erm behavior of he series. Because of he limied number of observaions some auhors use panel analysis. (Eger 2004, Couder and Couharde, 2006). 3. The Model This paper uses a BEER model in order o esimae equilibrium real exchange rae. The saring poin in his model consiss in expressing he real exchange rae as a funcion of he expeced value of he real exchange rae a mauriy +k, he real ineres rae differenial and a ime-varying premium-risk: (nominal exchange rae is expressed in unis of foreign currency for one uni of local currency so an increase sands for depreciaion) * q = E ( q+ k ) + ( r r ) Π (1) The ime-varying risk premium is a funcion of relaive governmen deb. Because of he low level for his variable for Romania I considered ha he risk premium is 0. Also I ried o implemen he real ineres rae differenial bu he resuls were no saisically significan.

4 122 European Research Sudies, Volume XII, Issue (1) 2009 I assume ha he unobservable expecaion of he exchange rae is deermined solely by he long run economic fundamenals Z 1. I denoe he long run equilibrium as qˆ and assume ha qˆ = E[ β 1 Z1 ] = β1z1 The oal misalignmen from he equilibrium real exchange rae can be expressed: m = q β 1 Z1 (2) The long-run economic fundamenals used in his paper ) are q = f ( prod, cons, nfa, open). The coefficiens are found o be saisically significan and correcly signed. The seps in esimaing he equilibrium exchange rae will be described in Secion Daa Descripion The source of daa is Eurosa and The Naional Bank of Romania daabase. The economies and periods covered are Romania (1998:1-2006:3) and Euro area (1998:1-2006:3). The frequency of observaions is quarerly and, in he economeric work, all series are seasonally adjused using TramoSeas. A firs problem is how o consruc radable and non-radable secors. I have considered boh he suggesions presen in he lieraure and he characerisics of he Romanian economy. The radable-secor is composed from indusry and consrucion while he non-radable is residual (oal - radable agriculure). I didn include agriculure because rade is disored by conrolled prices. A very imporan fac is ha when I calculaed he produciviy differenial I have also considered he share of radables (calculae as he share of he radable secor in oal value added). Also, he produciviy is proxied by labor produciviy because daa on capial socks are unavailable. All variables are in consan prices (1998=100). Descripion of variables: Quarerly observaions of value added from he producion side GDP esimaes (decomposed ino radables and non-radables) CPI raes of inflaion wih subcomponens enabling a breakdown ino raded and non-raded goods and services Nominal exchange raes of domesic currency agains he euro (quarerly averages) Employmen (quarerly averages) in raded and non-raded indusries. Consumpion, ne foreign asses, openness as a share of GDP.

5 The Esimaion of he Equilibrium Real Exchange Rae for Romania Empirical Resuls In order o esimae he equilibrium real exchange rae using a BEER approach I have followed more seps. Firs I checked if he series used are saionary, using Augmened Dickey- Fuller and Phillips-Perron ess. The resuls are in he appendix (able 1.a) As i can be seen from he able all variables are inegraed of order 1. Nex I ried o deermine a long-erm relaion beween variables by using coinegraion ess. Firs I esimaed a VAR. The lag lengh was choosing in order o minimize he informaion crierions. The resuls are in he appendix (able 2.a) Afer examining he resuls from able 2.a, I have esimaed a VAR wih 3 lags. The ess performed on he residuals revealed a normal disribuion, no auocorrelaion and he absence of heeroskedasiciy. The resuls are presened in he appendix (able 6.a) Nex I performed a Johansen coinegraion es. The es showed he presence of wo coinegraions vecors a boh 1% and 5% level. The resuls are presened in he appendix (able 3.a) If we normalize he coinegraing vecor wih respec o RER, we can obain he following expression (sandard errors in ( ), -saisics in [ ]): RER= *PROD_DIF *CONS *NFA *OPENESS (0.1545) (0.8637) (0.029) (0.1273) [9.2795] [5.6158] [ ] [ ] All coefficiens are saisically significan and correcly signed. An increase in he produciviy differenial beween Romania and he euro area will lead o an appreciaion of he real exchange rae, as prediced by he Balassa- Samuleson effec. An increase in he produciviy differenial will increase he relaive price of non-radables. Given he fac ha non-radable goods and services represen abou 65% of he consumpion baske, domesic prices will have a superior dynamic o inflaion of he euro area. As a consequence he real exchange rae appreciaes. An increase in consumpion will lead o real exchange rae appreciaion because i is mainly direced owards non-radable goods. An increase in ne foreign asses will lead o real exchange rae appreciaion which is characerisic for a ransiion counry. A higher value for his variable will lead o higher yield for domesic savings, higher levels of foreign currency enering he counry and as a consequence, real exchange rae appreciaion. An increase in he degree of openness will deermine an increased demand for foreign radable goods, he curren accoun defici will widen so a depreciaion of he real exchange rae will be required. The nex sep in he BEER approach requires he esimaion of he long run susainable values for he variables used. In order o do ha, I have used a Hodrick-Presco filer on he exend ARIMA series. (I have exended he series because of he problems of he Hodrick-Presco filer a he beginning and a he end of he series). The resuls are in he appendix. (Figures 1.a o 4.a)

6 124 European Research Sudies, Volume XII, Issue (1) 2009 In order o deermine he equilibrium real exchange rae, I have replaced he values obained by filering he series in he coinegraing relaionship esimaed in sep 1. The resuls were used o consruc he nex figure: Figure 1: The real exchange rae and he real equilibrium exchange rae RER_ECH RER_ACTUAL The oal misalignmen of he real exchange rae from is equilibrium value was obained by using he nex formula: Toal misalignmen= (Real exchange rae-equilibrium Real Exchange Rae)/Equilibrium Real Exchange Rae The resuls are shown below: Figure 2: Toal misalignmen real exchange rae from is equilibrium value TOTAL MISALIGNEMENT_ECH_RER As i can be seen from figure 2, he real exchange rae had some imporan deviaions from is equilibrium value. During 1998:Q1-2001:Q1 he real exchange rae was undervalued from is equilibrium value, wih a maximum misalignmen of 28,96% in he firs quarer of This misalignmen was caused by he rapid depreciaion of he nominal exchange rae induced by he liberalizaion of prices

7 The Esimaion of he Equilibrium Real Exchange Rae for Romania 125 and of he foreign currency marke. Also, he inflaions expecaions were a a high level which deermined he populaion o keep savings in foreign currency, puing even more pressure on he nominal exchange rae. During 2001:Q2-2003Q4 he real exchange rae was fairly valued while in 2005Q3-2006Q4 i was slighly undervalued. In he las period, 2005Q3-2006Q4, he real exchange rae was overvalued. This siuaion was caused by he increase in he produciviy differenial and he appreciaion of he nominal exchange rae caused by he large speculaive funds araced by he ineres rae differenial. The overvaluaion can have he effec of losing exernal compeiiveness inducing a larger curren accoun defici. The shor erm dynamics of he real exchange rae can be obained by esimaing a vecor error correcion model: RER = *( *PROD_DIF(-1) CONS(-1) *NFA(-1) + (0.0463) (0.1545) (0.8637) (0.029) [2.0333] [9.2795] [5.6158] [ ] *OPENESS ) * RER(-1) * CONS(- 1) (0.1273) (0.1818) (0.4028) [ ] [2.6988] [3.8292] The adjusmen parameer is saisically significan and has a value of Concluding Remarks This paper has esimaed he equilibrium real exchange rae for Romania using a BEER approach. This mehod is commonly used for ransiion counries (Halpern and Wyplosz (2001), De Broeck and Slok (2001), Eger (2002)). In he covered period he real exchange rae had some imporan deviaions from is equilibrium value which were deermined by he liberalizaion of he prices and of he foreign exchange marke and by he flucuaions of he nominal exchange rae. The nominal exchange rae had a high volailiy because of is use by he cenral bank in order o mainain a low inflaion and because of he speculaive funds araced by he ineres rae differenial. In he las period covered, he real exchange rae seems o be fairly valued which will lead o exernal compeiiveness. The nominal appreciaion of he exchange rae in 2007 can cause an overvaluaion of he exchange rae puing a risk he exernal compeiiveness and he abiliy of Romania o join ERM II. If he cenral pariy is no chosen in order o reflec he equilibrium exchange rae, he cenral bank is likely o face speculaive aacks ha could resul in delaying he inroducion of he single currency. This paper found a negaive relaion beween produciviy differenial, oal consumpion, ne foreign asses and he real exchange rae which is consisen wih he lieraure. Also, an increase of he degree of openness is likely

8 126 European Research Sudies, Volume XII, Issue (1) 2009 o cause a depreciaion of he real exchange rae because of he increased demand for radable goods from abroad. This paper has found evidence of he Balassa-Samuelson effec in Romania. The coefficien of he produciviy differenial is saisically significan bu he effec does no explain he large inflaion differenials beween Romania and he euro area. The produciviy differenial explained on average only 0.5% of he inflaion differenial in he period covered wih a higher impac in 2005 and 2006 (1.17% and 1.31%). The conclusion is ha facors differen from he produciviy differenial are responsible for he high inflaion differenial and ha he Balassa-Samuleson effec is no likely o pu a risk he Maasrich inflaion crierion. References 1) Bergin, P., R. Glick and A.M. Taylor, 2006, Produciviy, radabiliy and he long-run price puzzle, Journal of Moneary Economics, 53, ) De Broek, M. and T. Slok, 2001, Inerpreing real exchange rae movemens in ransiion counries, IMF Working Paper 01/56. 3) Brooks, C., 2002, Inroducory economerics for finance, Cambridge Universiy Press. 4) Candelon, B., K. Raabe, T. van Veen and C. Kool, 2006, Longrun real exchange rae deerminans: Evidence from eigh new EU member saes, , Journal of Comparaive Economics, 35, ) Canzoneri, M. B., R.E. Cumby and B.Diba, 1999, Relaive labor produciviy and he real exchange rae in he long run: evidence for a panel of OECD counries, Journal of Inernaional Economics, 47, ) Clark, P. B. and R. MacDonald, 1998, Exchange rae and economic fundamenals: A Mehodological Comparison of BEERs and FEERs, IMF Working Paper 98/67. 7) Coricelli, F. and B. Jazbec, 2004, Real exchange rae dynamics in ransiion economies, Srucural Change and Economic Dynamics, 15, ) Couder, V. and C. Couharde, 2006, Real equilibrium exchange rae in European Union New Members and Candidae Counries, Conference on economic policy issues in he EU, Berlin. 9) Froo, K. A. and K. Rogoff, 1994, Perspecives on PPP and long run real exchange raes, NBER Working Paper ) Lommazsch, K. and S. Tober, 2005, Wha is behind he real appreciaion of he Accession counries currencies? An invesigaion of he PPI-based real exchange rae, Economic Sysems, 28, ) MacDonald, R., 1997, Wha deermines real exchange raes? The long and shor of i, IMF Working Paper 97/21.

9 The Esimaion of he Equilibrium Real Exchange Rae for Romania ) Myjajima, K., 2005, Real exchange raes in growing economies: How srong is he role on he nonradables secor? IMF Working Paper 05/ ) Thalassinos E., Kiriazidis Th., 2003, Degrees of Inegraion in Inernaional Porfolio Diversificaion: Effecive Sysemic Risk, European Research Sudies Journal, Vol. VI, issue 1-2. Appendix Table 1.a Uni roo ess for variables incuded in BEER approach Series ADF Tes Phillips-Perron Tes Conclusion Level Firs difference Level Firs differnce LRER I(1) (0.3354) (0.0023) (0.2758) (0.0025) LPROD_DIF I(1) (0.0578) (0.0002) (0.3922) (0.0001) LCONS I(1) (0.2454) (0.0014) (0.3196) (0.000) LOPENESS I(1) (0.1629) (0.0041) (0.092) (0.000) LNFA I(1) (0.1087) (0.0002) (0.147) (0.000) Table 2.a Choosing he lag lengh in VAR Lag LR AIC SC HQ 0 NA * * * * Table3.a Coinegraion es for BEER Model Sample(adjused): 1998:4 2006:3 Included observaions: 32 afer adjusing endpoins Trend assumpion: Linear deerminisic rend Series: LCURS_REAL L_DIF_PROD LCONSUM LNFA LOPENESS Lags inerval (in firs differences): 1 o 2 Unresriced Coinegraion Rank Tes Hypohesized Trace 5 Percen 1 Percen No. of CE(s) Eigenvalue Saisic Criical Value Criical Value

10 128 European Research Sudies, Volume XII, Issue (1) 2009 None ** A mos 1 ** A mos A mos A mos 4 * *(**) denoes rejecion of he hypohesis a he 5%(1%) level Trace es indicaes 2 coinegraing equaion(s) a boh 5% and 1% levels Hypohesized Max-Eigen 5 Percen 1 Percen No. of CE(s) Eigenvalue Saisic Criical Value Criical Value None ** A mos 1 ** A mos A mos A mos 4 * *(**) denoes rejecion of he hypohesis a he 5%(1%) level Max-eigenvalue es indicaes 2 coinegraing equaion(s) a boh 5% and 1% levels Tabel 4.a VEC-VEER Model Vecor Error Correcion Esimaes Sample(adjused): 1998:4 2006:3 Included observaions: 32 afer adjusing endpoins Coinegraing Eq: CoinEq1 LCURS_REAL(-1) L_DIF_PROD(-1) ( ) [ ] LCONSUM(-1) ( ) [ ] LNFA(-1) ( ) [ ] LOPENESS(-1) ( ) [ ] C

11 The Esimaion of he Equilibrium Real Exchange Rae for Romania 129 Table 5.a : Residual ess on VAR -saisic Probabiliy LM auocorrelaion es LM(1) (0.6455) LM(2) (0.1635) LM(3) (0.8804) LM(4) 9.32 (0.4078) Jarque-Berra Normaliy es (0.8078) Whie`s heeroskedasiciy es (0.3237) Table 6.a : Checking VAR sabiliy Roos of Characerisic Polynomial Endogenous variables: LCURS_REAL L_DIF_PROD LCONSUM LNFA LOPENESS Exogenous variables: C Lag specificaion: 1 3 Roo Modulus i i i i i i i i i i ` i i i i No roo lies ouside he uni circle. VAR saisfies he sabiliy condiion.

12 130 European Research Sudies, Volume XII, Issue (1) 2009 Figure 1.a : Equilibrium values for oal consumpion -Hodrick-Presco filer Presco filer Figure 2.a : Equilibrium value for produciviy differenial - Hodrick DIF_PROD_ECH_ARIMA L_DIF_PROD CONSUM CONS_ECH Figure 3.a : Equlilibrium value for ne foreign asses- Hodrick-Presco filer Figure 4.a : Equilibrium value for openness- Hodrick-Presco filer NFA_ECH_ARIMA NFA OPENESS OPENESS_ECH_ARIMA