Real Effective Exchange Rate Imbalances and Macroeconomic Adjustments: evidence from the CEMAC zone

Transcription

1 MPRA Munich Personal RePEc Archive Real Effecive Exchange Rae Imbalances and Macroeconomic Adjusmens: evidence from he CEMAC zone Simplice Asongu 14 January 2014 Online a hps://mpra.ub.uni-muenchen.de/63154/ MPRA Paper No , posed 22 March :30 UTC

2 AFRICAN GOVERNANCE AND DEVELOPMENT INSTITUTE A G D I Working Paper WP/14/019 Real Effecive Exchange Rae Imbalances and Macroeconomic Adjusmens: evidence from he CEMAC zone Simplice A. Asongu African Governance and Developmen Insiue, Yaoundé, Cameroon. asongusimplice@yahoo.com 1

3 2014 African Governance and Developmen Insiue WP/19/14 AGDI Working Paper Research Deparmen Real Effecive Exchange Rae Imbalances and Macroeconomic Adjusmens: evidence from he CEMAC zone Simplice A. Asongu 1 January 2014 Absrac We assess he behavior of real effecive exchange raes (REERs) of members of he CEMAC zone wih respec o heir long-erm equilibrium pahs. A reduced form of he fundamenal equilibrium exchange rae (FEER) model is esimaed and associaed misalignmens are derived for he period 1980 o Our findings sugges ha for majoriy of counries, macroeconomic fundamenals have he expeced associaions wih he exchange rae flucuaions. The analysis also reveals ha, only he REER adjusmens of Cameroon and Gabon are significan in resoring he long-erm equilibrium in even of a shock. The Cameroonian economic fundamenals of erms of rade, governmen expendiure and openness have differen long-erm relaions wih he REER in comparison o hose of oher member saes. Ulimaely, here is no need for an adjusmen in he level of he peg based on he presen quaniaive analysis of REER pahs. JEL Classificaion: F31; F33; F42; F61; O55 Keywords: Exchange rae; Macroeconomic impac; CEMAC zone 1 Simplice A. Asongu is Lead economis in he Research Deparmen of he AGDI (asongus@afridev.org). 2

4 1. Inroducion The euro area has come under immense pressure afer is firs decade. Triggered by adjusmens in he fiscal accouns of Greece, he crisis which iniially spread o Ireland and Porugal has become a hrea o he euro zone s exisence afer Spain and Ialy s sovereigns began experiencing funding pressures. This crisis reflecs banking secor fragiliies and inerwined public deb made worse by apparenly weak growh oulooks, as well as subsanial gross and ne exernal liabiliies 2. German firms have coninued heir ouward inegraion by seing-up producion plaforms in emerging Europe in order o ake advanage of higher reurn on capial and lower wage coss which boosed compeiiveness and expors. This easy financing coninued unil he crisis eruped and allowed defici counries o susain appreciaing real effecive exchange raes (which have also been driven by he nominal appreciaion of he euro) and has delayed he adjusmen needed o end he growing divergence of rade performance wihin he moneary union (Chen e al., 2012). The debae in he lieraure on srucural adjusmen and macroeconomic sabilizaion has emphasized he crucial role played by real exchange rae (owing o he imporance of expor promoion) for he generaion of opimal pahs of employmen and oupu (Mussa, 1974; Edwards & Van Wijnbergen, 1986; Obsfeld & Rogoff, 1996; Acemoglu e al., 2003; Abdih & Tsangarides, 2010). The mainenance of exchange raes a an appropriae level has been credied for he success of cerain developing counries. In he same vein, i is believed ha a disinguishing feaure of Eas and Souheas Asia s success wih susainable growh has been he consisen avoidance of overvaluaion (Abdih & Tsangarides, 2010). The specre of he Euro crisis is huning exising and embryonic moneary unions. Recen sudies have shown ha proposed African moneary unions of Eas and Wes Africa are no opimal currency areas (Asongu, 2012a). In he same ligh; real, moneary and fiscal policy 2 For insance, ne exernal liabiliies of close o 100% of GDP in Greece, Ireland, Porugal and Spain. 3

5 convergence is absen wihin he Cenral African Economic and Moneary Communiy (CEMAC) CFA franc zone (Asongu, 2012b). The CFA franc arrangemen daes back o he mid-1940s and is one of he longes-sanding fixed exchange rae regimes in he world. This currency was devalued once in 1994, when a 50% adjusmen in he nominal rae reversed domesic and exernal disequilibria ha had buil-up since he mid-1980s. Over he period 2002 o 2005, he CFA franc appreciaed (by more han 30% in nominal erms vis-à-vis he US dollar) along wih he euro o which i is pegged. This appreciaion has led o a renewed ineres in he prospecs of and oulook for he CFA franc and reignied he debae on he susainabiliy of he peg, including wheher he same peg o he euro for boh moneary unions coninues o be appropriae (Abdih & Tsangarides, 2010). In ligh of he above, assessing he compeiiveness of he CFA zone requires quaniaive analysis of he acual and equilibrium exchange raes as well as adjusmens of macroeconomic fundamenals o corresponding exchange imbalances. Previous empirical work on he CFA franc currency valuaion has focused on he firs few years afer he 1994 devaluaion, wih paricular ineres of some sudies in how real exchange rae misalignmens were correced by he 1994 change in pariy (Clémen e al., 1996; Devarajan, 1997; Baffes e al., 1999; Ahlers & Hinkle, 1999). To he bes of our knowledge, sudies on he effecs of he recen appreciaion of he euro on he CFA franc are scany. Abdih & Tsangarides (2010) have recenly evaluaed wheher he CFA franc is currenly significanly over-valued from is equilibrium rae and concluded ha he currency is no over-valued. While panel-based resuls as presened by Abdih & Tsangarides (2010) are imporan, hey fail o accoun for counryspecific dynamics ha could have relevan policy implicaions. This is because, like in he curren euro crisis, macroeconomic fundamenals maybe differen across CFA franc counries. Hence, he need for counry-specific analyses o complemen exising lieraure. 4

6 In his aricle, we analyze he movemens of he acual real effecive exchange rae (REER) for counries wihin he CEMAC zone vis-à-vis long-erm values. The res of he paper is organized as follows. Secion 2 presens he heoreical framework, daa and mehodology. Empirical analysis and corresponding discussion are covered in Secion 3. Secion 4 concludes. 2. Theoreical framework, Daa and Mehodology 2.1 FEER model specificaion We esimae he reduced form of he Fundamenal Equilibrium Exchange Rae (FEER) model (Edwards, 1989) using he Johansen s (1995) coinegraion mehodology in order o derive he equilibrium pahs and corresponding misalignmens for he period The FEER approach is paricularly appropriae in assessing if he movemen of he REER represens a misalignmen or if he Equilibrium Real Effecive Exchange Rae (EREER) iself has shifed because of changes in he macroeconomic fundamenals (Abdih & Tsangarides, 2010). Consisenly, we define equilibrium as he rae ha resuls in he simulaneous aainmen of inernal and exernal equilibrium in he economy. Hence, inernal equilibrium is achieved when he marke for non-radable goods clears in he presen (and expeced o clear in he fuure) as price and wage flexibiliy ensure ha he condiion of inernal balance is saisfied (ha is, demand equals o supply). Conversely, exernal equilibrium is achieved when he curren accoun balance is a a susainable level as given by a susainable hreshold of capial flows. Accordingly, he long-run deerminans of he EREER are defined by he following fundamenals. (1) Governmen spending of which he expeced sign is ambiguous in he absence of a breakdown of governmen spending ino radable and non-radable goods 3. (2) 3 If governmen spending is primarily direced owards non-radable (radable) goods, an increase in governmen consumpion will resul in an appreciaion (depreciaion) of he REER. 5

7 Produciviy wih an expeced posiive sign ha capures he Balassa-Samuelson effec 4. (3) Terms-of-rade (TOT) in goods wih also an expeced posiive sign ha capures he wealheffec 5. (4) Invesmen (of an ambiguous sign) which is included in he heoreical model because of supply-side effecs ha are dependen on he relaive facor inensiies across secors 6. (5) Degree of rade conrols/resricions for which he sign is ambiguous Daa and Mehodology We examine a sample of 4 CEMAC counries (Cameroon, Cenral African Republic, Gabon and Equaorial Guinea) due o consrains in daa availabiliy. The daa include he following variables: he REER, produciviy, governmen expendiure, TOT, invesmen and openness. The firs wo variables are presened in erms of naural logarihm o ease comparabiliy and compaibiliy. Definiion of variables (wih corresponding) sources is presened in he Appendix. Consisen wih Abdih & Tsangarides (2010), we employ a Vecor Error Correcion Model (VECM). Applicaion of he VECM presupposes he exhibiion of uni roos in levels and he exisence of a long-run equilibrium (coinegraion). Firs, we employ he Augmened Dickey Fuller (ADF) es for uni roos or he order of inegraion of he series. Nex, he Johansen (1988, 1991, 1995) maximum likelihood procedure is used o es for he corresponding long-run coinegraion relaionships beween he exchange rae and is fundamenals. Then, he equilibrium levels of he fundamenals are compued specifically by exracing he permanen componen from he fundamenals series. Finally, he vecor of long- 4 An increase in he produciviy of radables vis-à-vis non-radables of one counry relaive o a foreign counry increases is relaive wage, which leads o an increase in he relaive price of non-radables and hence, causes a REER appreciaion. 5 An appealing TOT shock induces an increase in he domesic demand, a corresponding increase in he price of non-radable goods which leads o a REER appreciaion. Alernaively from an inernal-exernal balance angle, an increase in he TOT leads o an increase in real wages of he expor secor and a rade surplus. For he exernal balance o be resored, he REER mus appreciae. 6 Since invesmen in a developing counry may have a high impor conen, a rise in he invesmen share of GDP could shif spending owards raded goods, hus depreciae he REER. Hence, we expec a negaive sign. 7 As rade conrols and barriers are lifed, increase in rade may eiher be impor or expor skewed and hence he need for depreciaion or appreciaion of he REER respecively. 6

8 erm parameers and he exraced permanen componen are combined o calculae he EREER, wih misalignmen esimaed as he shif of he REER from is value in equilibrium. 3. Empirical analysis 3.1 Inegraion analysis We perform he sandard ADF ess boh in levels and firs differences of he variables under consideraion. Opimal lengh selecion for goodness of fi in he ADF specificaion is by he Akaike Informaion Crierion (AIC). From he repored -ADF saisics in Table 1, bu for invesmen in he Cenral African Republic (CAR), we canno rejec he null hypohesis of a uni roo for all he variables in levels. However, he null of a uni roo in firs difference is srongly rejeced. Hence, we conclude ha he variables are overwhelmingly inegraed in he firs order; his is, hey can be differenced once o obain saionariy. Table 1: ADF uni roo es for variables Counries Variables Level Firs difference c c c c Cameroon ( ) Cenral African Republic ( ) Gabon ( ) Equaorial Guinea ( ) REER(ln) *** ** TOT * *** *** Gov Spending *** *** Openness ** ** Produciviy(ln) ** * Invesmen ** ** REER(ln) *** *** TOT *** *** Gov Spending *** *** Openness *** *** Produciviy(ln) ** * Invesmen ** ** n.a n.a REER(ln) *** *** TOT * *** *** Gov Spending * *** ** Openness *** ** Produciviy(ln) ** * Invesmen *** *** REER(ln) ** *** TOT * Gov Spending *** *** Openness ** Produciviy(ln) ** * Invesmen Noes. ***, **, *: denoe significance a 1%, 5% and 10% respecively. c and c : consan and consan and rend respecively. REER: Real Effecive Exchange Rae. TOT: Terms of Trade. Gov : Governmen. ln: logarihm. 7

9 3.2 Coinegraion analysis form: Y Le us specify a vecor of variables Y as a vecor auoregressive (VAR) equaion in he p 0 iy i D (1) i 1 Where Y is a 6 1 vecor: Y = REER Terms-of-rade of goods Governmen spending Openness Produciviy Invesmen where 0 is a 6 1 vecor of consans; i are 6 6 marices of coefficiens of lags of Y ; D is a vecor of he dummy-ype variables; p is he lag lengh; and is he 6 1 vecor of independen and idenically disribued error erms ha are assumed o be normal wih zero mean and covariance marix. As such, he VAR is made up of a sysem of six equaions where he righ-hand side of each equaion comprises a common se of lagged and deerminisic regressors. The VAR specificaion in Eq. (1) provides he basis for coinegraion analysis. Hence, adding and subracing various lags of Y yields an expression for he VAR in firs difference: Y p 1 0 Y 1 i Y i D e (2) i 1 is a 6 where denoes he difference operaor,,..., ) P i i and 1 I i ( i 1 p 6 coefficien marix, (i) If (rank) 6 or (rank) 0, hen coinegraion exis among he variables. In his case, i is appropriae o esimae he model in levels for [for rank n ] and firs difference [for rank 0]. 8

10 (ii) If (rank) 0 or (rank) r 6, hen here are r coinegraing vecors/relaionships. In his case, marix can be expressed as he ouer produc of wo full column rank 6 r marices and where. From he resuls presened in Table 2 below, he second condiion (ii) for coinegraion above is saisfied. Hence, here is evidence of a long-run relaionship beween he REERs and heir idenified fundamenals for all he four counries under consideraion.. Table 2: Johansen coinegraion es Cameroon Cenral African Republic Gabon Equaorial Guinea Eigenvalue Trace es Max-Eigen es None *** *** A mos *** *** A mos ** ** A mos A mos A mos None ** A mos A mos A mos A mos A mos 5 n.a n.a n.a None *** *** A mos ** A mos * A mos A mos A mos None *** *** A mos *** *** A mos *** *** A mos *** *** A mos *** ** A mos Noes. ***, **, *: denoe significance a 1%, 5% and 10% respecively. c and c : consan and consan and rend respecively. Model specificaion is by AIC wih 2 maximum lags. Wih condiion (ii) saisfied, he VAR can be expressed as a VECM: Y p 1 Y Y D e 0 ' 1 i i (3) i 1 The marix ' conains he coinegraing vecor(s) and he marix has he weighing elemens for he rh coinegraing relaion in each equaion of he VAR. The marix rows of 9

11 ' Y 1 are normalized on he variable(s) under consideraion in he coinegraing relaion(s) and inerpreed as deviaions from he long-erm equilibrium condiion(s). In his conex, he column of represens he speed of adjusmen o he long-erm equilibrium. The esimaed vecor can be used o provide a measure of he EREER and also quanify he misalignmen margin beween he prevailing exchange rae and is equilibrium level. The esimaed associaed wih he REER capures he speed a which he real exchange rae convergences o he equilibrium sae. Opimal lag selecion for goodness of fi in model specificaions is in line wih he recommendaions of Liew (2004) Coinegraion coefficiens and adjusmens We have observed from Table 2 ha, here is evidence of a long-erm relaionship beween he REERs and heir idenified fundamenals in he four CEMAC counries invesigaed. Table 3 conains he resuls from esimaing he VAR/VECMs in Eq. (3) for each of he four counries. The able is divided ino wo panels, wih Panel A reporing esimaes for he coinegraing vecors (he s ) ogeher wih heir sandard errors and Panel B reporing he adjusing (feedback) coefficiens esimaes (he s ) and heir -saisics. The resuling coinegraion equaions have some signs ha are consisen wih he predicions from economic heory, while some do no. These asymmeric counry-specific dynamics canno be obained from a panel-based analysis and hus, provide jusificaion for our problem saemen and he complemenary characer of he presen sudy wih respec o Abdih & Tsangarides (2010). 8 The major findings in he curren simulaion sudy are previewed as follows. Firs, hese crieria managed o pick up he correc lag lengh a leas half of he ime in small sample. Second, his performance increases subsanially as sample size grows. Third, wih relaively large sample (120 or more observaions), HQC is found o oudo he res in correcly idenifying he rue lag lengh. In conras, AIC and FPE should be a beer choice for smaller sample. Fourh, AIC and FPE are found o produce he leas probabiliy of under esimaion among all crieria under sudy. Finally, he problem of over esimaion, however, is negligible in all cases. The findings in his simulaion sudy, besides providing formal groundwork supporive of he popular choice of AIC in previous empirical researches, may as well serve as useful guiding principles for fuure economic researches in he deerminaion of auoregressive lag lengh (Liew, 2004, p. 2). 10

12 From Panel A, he following could be esablished. (1) Bu for Cameroon, he TOT are posiively correlaed wih he REER indicaors such ha, an improvemen in he TOT would resul in an appreciaion of he long-run EREER hrough a possible wealh-effec. The Cameroonian case implies ha, an increase in he TOT does no lead o an increase in nonradables hrough he wealh-effec. (2) Wih he excepion of he CAR, governmen consumpion has a posiive (appreciaing) impac on REER, suggesing ha mos governmen spending is direced owards non-radables. (3) Bu for Cameroon, increases in openness are associaed wih depreciaion of he REER hrough increases in impors. (4) The relaive high long-erm impac of echnological progress (proxied by he relaive real GDP per capia) confirms he Balassa-Samuelson effec for Gabon and Equaorial Guinea. Hence, produciviy in Cameroon and he CAR does no lead o an increase in he expor of radables and corresponding increase in wages and demand for non-radables ha ulimaely appreciae he REER. (5) Invesmen is negaively correlaed wih he REER only in Cameroon, confirming he hypohesis ha invesmen has a high impor conen and increases spending owards radable goods. This is no he case wih he oher counries. Panel B of Table 1 shows he feedback coefficiens for he coinegraing vecors or he shor-run relaionships of he REER and is fundamenals. Some adjusmens are significanly differen from zero, implying ha hese fundamenals are no weakly exogenous wih respec o he parameers of he coinegraion relaionship. In he face of any deviaion from he long-erm equilibrium, hese variables joinly respond and adjus he sysem back o equilibrium. The fundamenals of Cameroon are significanly srong in adjusing he sysem o he equilibrium. Openness and produciviy are no significan fundamenals in adjusing he REER of all counries under consideraion. Conversely, he TOT is a significan insrumen for adjusing he REER in all counries, followed by governmen spending and invesmen in wo of he four counries. Furhermore, he feedback coefficien for he REER is only significan for Cameroon 11

13 and Gabon, confirming hese counries as he backbones of he CEMAC economy. The significan negaive adjusing erms of he REER change for hese wo counries also sugges he sabiliy of he error correcion mechanism. As a maer of principle, he speed of adjusmen of he parameers should be beween zero and minus one (0, -1). If he Error Correcion Terms (ECTs) are no wihin his inerval, hen eiher he model is misspecified (and needs adjusmen) or he daa is inadequae (perhaps owing o issues wih degrees of freedom) 9. Table 3: Coinegraion and shor-erm adjusmen coefficiens Cameroon Cenral A. R Gabon Equaorial G. Panel A: Esimaes of coinegraion relaionships TOT [0.641] 1.886[0.232] 2.678[0.508] 0.862[0.000] Gov Spending 0.262[0.040] [0.009] 0.364[0.044] 0.097[0.000] Openness 0.026[0.009] [0.004] [0.014] [0.000] Produciviy(ln) [0.213] [0.108] 2.425[0.281] 0.392[0.000] Invesmen [0.033] [0.027] 0.024[0.000] consan 1.523[3.213] [1.276] [3.654] [0.000] Panel B: Esimaes of shor erm adjusmen coefficiens D[REER(ln)] (-4.083)*** 0.009(0.049) (-2.355)** (-0.852) D[TOT] 0.143(2.169)** (-2.660)** (-1.868)* (-2.874)** D[Gov Spending] (-1.951)* (4.771)*** (-0.949) 2.081(0.606) D[Openness] 3.940(1.525) 13.06(1.223) (-0.029) (1.603) D[Produciviy(ln)] 0.038(0.866) 0.072(0.504) 0.000(0.035) (-0.304) D[Invesmen] (-2.817)** (1.120) (1.952)* Noes. *, **, ***: denoe significance levels a 10%, 5% and 1% respecively. Model specificaion is by AIC wih 2 maximum lags. The deerminisic rend assumpion is a resriced consan. [ ]: sandard errors. ( ): - saisics. D[ ]: Firs difference. 4. Conclusion and policy recommendaions Wih he help of a dynamic model of a small open economy and he Johansen coinegraion mehodology, he EREERs have been analyzed for counries in he CEMAC zone. The objecive has been o analyze REERs imbalances and assess wheher he movemens in he aggregae real exchange raes are consisen wih he underlying macroeconomic fundamenals. We have shown ha, from counry-specific perspecives he long-erm behavior 9 The error correcion erm ells us he speed wih which our model reurns o equilibrium following an exogenous shock. I should be negaively signed, indicaing a move back owards equilibrium, a posiive sign indicaes movemen away from equilibrium. The coefficien should lie beween 0 and 1, 0 suggesing no adjusmen one ime period laer, 1 indicaes full adjusmen. The error correcion erm can be eiher he difference beween hedependen and explanaory variable (lagged once) or he error erm (lagged once), hey are in effec he same hing (Babazadeh & Farrokhnejad, 2012, p.73). 12

14 of he REERs can be explained by flucuaions in governmen expendiure, erms of rade, openness, produciviy and invesmen. We have found evidence of significan misalignmens in he REER only in Cameroon and Gabon. The negaive feedback erms of he exchange raes sugges ha adjusmens will resore he long-run equilibrium. Had he adjusmens been posiively significan, his would have implied a furher deviaion from he long-run equilibrium o a new equilibrium. Hence, we can infer ha based on he available weigh of evidence, here is ye no need for an immediae adjusmen in he level of he peg. In heory, fixed exchange rae regimes can be susainable as long as acual deviaions from long-erm equilibrium raes are small and mean revering. Conversely, if deviaions are one-sided and build up o longer-erm significan misalignmens, i is generally argued ha (in addiion o demand-side managemen policies) real exchange rae acion may be required o resore balance. Hence, based on he esimaed pahs here is ye no a very clear paern of overvaluaion as was he case prior o he 1994 devaluaion. As a policy implicaion, here is no need for an adjusmen in he level of he peg based he presen counry-specific quaniaive analysis of he pah of he REERs in he CEMAC zone. However, exhausive analysis of he environmen ha affecs he overall susainabiliy of he CFA franc arrangemen necessiaes an examinaion of possible pressures on balance of paymen flows, reserve levels, losses of compeiiveness, susained deviaions from counryspecific EREERs and unfavorable marke percepions. We have also observed ha, bu for a few excepions, he impac of he fundamenals on he REER are similar across counries. For fundamenals ha did no mee expeced signs (wih respec o he REER), i is up o he auhoriies in place o adop relevan measures o adjus he endencies. This is paricularly he case of Cameroon wih respec o he erms of rade, governmen expendiure and openness. 13

15 Appendix Appendix 1: Variable definiions Variables Signs Variable definiions Source Real Effecive Exchange Rae REER Naural Log. of REER WDI (World Bank) Terms of Trade TOT Exporable Commodiies/Imporable Commodiies Governmen Spending Gov Governmen final consumpion expendiure (% of GDP) Openness Trade Impors plus Expors of Commodiies (% of GDP) WDI (World Bank) WDI (World Bank) WDI (World Bank) Produciviy Prod. Naural Log. of real GDP per capia. WDI (World Bank) Invesmen Inv. Gross Fixed Capial Formaion (% of GDP) WDI (World Bank) Log: Logarihm. GDP: Gross Domesic Produc. WDI: World Domesic Indicaors. References Abdih, Y., & Tsangarides, C., (2010). FEER for he CFA franc, Applied Economics, 42, pp Acemoglu, D., Johnson, S., Robinson, J. A., & Thaicharoen, Y., (2003). Insiuional causes, macroeconomic sympoms: volailiy, crises and growh, Journal of Moneary Economics, 50, pp Ahlers, T., & Hinkle, L., (1999). Esimaing he equilibrium real exchange empirically in Exchange Rae Misalignmen: Conceps and Measuremen for Developing Counries, (Eds) Hinkle, L., & Moniel, P., World Bank, Washingon, pp Asongu, S. A., (2012a). Are Proposed African Moneary Unions Opimal Currency Areas? Real, Moneary and Fiscal Policy Convergence Analysis, African Journal of Economics and Managemen Sudies: Forhcoming. Asongu, S. A., (2012b). Real and Moneary Policy Convergence: EMU Crisis o he CFA zone, MPRA Paper No Babazadeh, M., & Farrokhnejad, F., (2012). Effecs of Shor-run and Long-run Changes in Foreign Exchange Raes on Banks Profi. Inernaional Journal of Business and Managemen, 7 (17), pp Baffes, J., Elbadawi, I., O Connell, S., (1999). Single equaion esimaion of he equilibrium real exchange rae in Exchange Rae Misalignmen: Conceps and Measuremen for Developing Counries, (Eds) Hinkle, L., and Moniel, P., World Bank, Washingon, pp

16 Chen, R., Milesi-Ferrei, G. M., & Tressel, T., (2012). Exernal Imbalances in he Euro Area, IMF Working Paper 12/236. Clémen, J. A. P., Mueiller, J., Cossé, S., & Dem, J., (1996). Afermah of he CFA franc devaluaion, IMF Occasional Paper No Devarajan, S., (1997). Real exchange rae misalignmen in he CFA zone, Journal of African Economies, 6(1), pp Edwards, S., (1989). Real Exchange Raes, Devaluaion and Adjusmen: Exchange Rae Policy in Developing Counries, MIT Press, Cambridge, MA. Edwards, S., & van Wijnbergen, S., (1986). The welfare effecs of rade and capial marke liberalizaion, Inernaional Economic Review, 27(1), pp Johansen, S., (1988). Saisical analysis of coinegraing vecors, Journal of Economic Dynamics and Conrol, 12(2-3), pp Johansen, S., (1991). Esimaion and hypohesis esing of coinegraing vecors in Gaussian vecor auoregressive models, Economerica, 59(6), pp Johansen, S., (1995). Likelihood-based Inference in Coinegraed Vecor Auoregressive Models, Oxford Universiy Press, Unied Kingdom. Liew, V. K., (2004). Which lag selecion crieria should we employ?. Economics Bullein, 3(33), pp Mussa, M., (1974). Tariffs and he disribuion of income: he imporance of facor specificiy, subsiuabiliy and inensiy in he shor and long run Journal of Poliical Economy, 82(6), pp Obsfeld, M., & Rogoff, K., (1996). Foundaions of Inernaional Macroeconomics, MIT Press, Cambridge. 15