Is Mercosur an Optimum Currency Area? An Assessment Using Generalized Purchasing Power Parity

Transcription

1 Unversdade Federal de Santa Catarna From the SelectedWorks of Sergo Da Slva July, 2008 Is Mercosur an Optmum Currency Area? An Assessment Usng Generalzed Purchasng Power Party J. Ancheta Neves Leandro Stocco Sergo Da Slva Avalable at:

2 Is Mercosur an optmum currency area? An assessment usng generalzed purchasng power party J. Ancheta Neves Petrobras Leandro Stocco Department of Economcs, Unversty of Sao Paulo Sergo Da Slva Department of Economcs, Federal Unversty of Santa Catarna Abstract We consder the contegraton approach of generalzed purchasng power party to show that a necessary condton for Mercosur to be an optmum currency area s met. Yet there are stll large cross-country dfferences as to cast doubt on the success of ether monetary unon or offcal dollarzaton. The PPP puzzle s also found to occur n Mercosur. SDS acknowledges fnancal support from the Brazlan agences CNPq and CAPES-Procad. Ctaton: Neves, J. Ancheta, Leandro Stocco, and Sergo Da Slva, (2008) "Is Mercosur an optmum currency area? An assessment usng generalzed purchasng power party." Economcs Bulletn, Vol. 6, No. 29 pp Submtted: June 6, Accepted: July 23, URL:

3 1. Introducton In an optmum currency area (Mundell 1961) effcency s maxmzed f the area shares a sngle currency. One ratonale behnd the creaton of the euro, for nstance, s that the ndvdual countres of Europe do not each form an optmum currency area but Europe as a whole does. Even f the fundamental economc varables determnng real exchange rates are nonstatonary (Da Slva 2002) and accordngly the rates are nonstatonary, the fundamentals can stll be suffcently ntegrated as n a currency area. Here the real rates wll share common trends (Enders and Hurn 1994). The exstence of at least one contegraton vector n a set of natonal economes nonstatonary seres suggests both an optmum currency area and generalzed purchasng power party (G- PPP). One purpose of ths paper s to evaluate whether Mercosur (the South Amercan trade group) s an optmum currency area n that sense. The dea of G-PPP was poneered by Enders and Hurn (1994), who apply t to Pacfc Rm countres. G-PPP was rejected and the Pacfc Rm natons were found not to consttute an optmum currency area. Enders and Hurn (1997) also tested G-PPP to the G7 countres. They found one contegraton vector at the 5 percent sgnfcance level, whch means that those countres real exchange rates seem to be lnked by a sngle long run equlbrum relatonshp, and a shock to any one rate s lkely to affect the long run values of the others. Lang (1999) found that G-PPP holds for Chna, Hong Kong, Japan, and the Unted States, and then that those countres consttute an optmum currency area. Bernsten (2000) tested G-PPP for the euro area and found that the null of noncontegraton cannot be rejected. Lee (2003) found that Australa, New Zealand, and Japan comprse an optmum currency area, but ths s not true of Australa, New Zealand, and the USA. And the East Asa countres were found not to consttute an optmum currency area (Choudhry 2005, Ahn et al. 2006, Kawasak and Ogawa 2006). Prevous work on Mercosur roughly suggests that t s stll a mrage. Hallwood et al. (2006) examned the case for ether Latn Amercan monetary unon (Argentna, Brazl, Chle, Uruguay, and Venezuela) or monetary unon wth the USA through offcal dollarzaton. Usng VAR technques they found that macroeconomc shocks are so hghly asymmetrc n Latn Amerca and between the Latn Amercan countres and the USA as to make monetary unon or offcal dollarzaton questonable. We wll replcate ths fndng below. It contrasts wth Fratann (2004) and Alexander and Von Furstenberg (2000), who argued that the Mercosur members are suted to creatng a common currency (though not to adoptng the US dollar). Of course, one cannot neglect the hypothess that an optmum currency area may be endogenous (Frankel and Rose 1998) n the sense that there can be a postve assocaton between trade ntensty and busness cycle correlaton. But emprcal evdence suggests that ths s unlkely for Mercosur (Hallwood et al. 2006, Ahumada and Martrena-Mantel 2001, Lcandro- Ferrando 2000). Intraregonal trade n Mercosur s stll modest, thanks manly to the low openness of the Argentne and Brazlan economes (Machnea 2004). The Brazlan economy can be consdered more relatvely dversfed than Argentna s (Barenbom 2004) and, as a result, less prone to large asymmetres of shocks (Kenen 1969, Calderon et al. 2007). But ncreasng ntra-mercosur trade s unlkely to make t more sutable for monetary unon because the macroeconomc shocks between the countres and between them and the USA do not become more symmetrc as tme goes by (Hallwood et al. 2006). Ths contrasts wth the experence of the European countres (Bayoum and Echengreen 1994). The ssue of the usefulness of monetary unon for the Mercosur countres has also been addressed n a number of other papers (e.g. Busse et al. 2006, Camarero et al.

4 2006, Berg et al. 2002, Hochreter et al. 2002, Corbo 2001, Salvatore 2001), but usage of the G-PPP approach s novel. Our contrbuton n ths paper s thus to show that although a necessary condton for Mercosur to be an optmum currency area s met, there are stll large cross-country dfferences as to cast doubt on the success of ether monetary unon or offcal dollarzaton. Here we abstan from any dscusson of the polcy mplcatons of our fndngs, and confne ourselves to make our case n a rather techncal way. However, our results can stll be contextualzed n the debate by consderng the above references. The rest of ths paper s organzed as follows. Secton 2 presents data. Secton 3 analyzes the data. And Secton 4 concludes. 2. Data We consdered quarterly data from 1973Q3 to 2006Q3 for the Mercosur s full members (Argentna (ARG), Brazl (BRA), Paraguay (PAR), Uruguay (URU)), the applcant Venezuela (VEN) as well as the USA, consdered as the benchmark country. We also used data for Bolva (BOL) and Chle (CHI) n the statonarty tests. The seres of consumer prce ndex (CPI) and average dollar prce of the currences were taken from the IMF s Internatonal Fnancal Statstcs. Yet our results could be replcated for wholesale prce ndces rather than CPIs (avalable upon request). The seres of real exchange rates were bult accordng to the formula * E, CPI *, ln t r t t = CPI = e t, t, + pt pt,, where r t, s the natural log of the real exchange rate n country at tme perod t, e t, s the natural log of the nomnal exchange rate n country, p t, s country s natural log of the CPI, and * p t s the natural log of the US CPI. Fgure 1 dsplays the real exchange rates. A frst look suggests that the Mercosur rates are nonstatonary. Ths wll be confrmed by the unt root tests below. 3. Analyss Table 1 presents results of the unt roots tests of Dckey-Pantula (DP) (1987), augmented Dckey-Fuller (ADF), Phllps-Perron (PP), and Ellott-Rothenberg-Stock (ERS) (1996). The DP tests suggested that the seres do not present two or more unt roots. The ERS tests are more approprate for slow adjustment processes, and thus were performed for the cases where the autoregressve parameter n the ADF tests fell above 0.9. The ADF and PP tests suggested rejecton of the null of unt root for Bolva and Chle at the sgnfcance level of one percent. However, the tests dd not provde evdence of rejecton (at one percent) of the null for the Mercosur countres. Thus Bolva and Chle could be left out from the subsequent contegraton analyss. Because the seres present structural breaks, we also performed Lee-Strazcch (2003, 2004) mnmum Lagrange multpler unt root tests, whch allow for one and two endogenous breaks under both the null and alternatve hypotheses. The seres volatlty made t hard for us to tell whether the breaks occurred n ether ntercept or trend. So we consdered two cases, namely (1) a crash model where the breaks occur n the ntercept, and (2) a trend model wth breaks n both ntercept and trend. Apart from Bolva (one percent sgnfcant), the tests dd not reject the null (Table 2), thus confrmng those n Table 1. We also performed Zvot-Andrews (1992) and Perron (1997) unt root tests wth one endogenous break and the result of nonstatonarty appeared agan (output avalable upon request). The break dates dentfed n the Lee- Strazcch tests (Table 2) captured the breaks related to both the ol shock of 1979 and

5 the fnancal crses of the eghtes. Yet we further consdered n the contegraton analyss the changes occurred n the countres monetary regmes (as n Sngh et al. 2005) (Table 3). And an ol shock dummy was also nserted to track the short run dynamcs. Then we carred out the contegraton analyss allowng for two breaks n the long run relatonshp (as n Johansen et al. 2000). We also consdered the tradtonal Johansen (1988, 1991) full nformaton maxmum lkelhood approach wthout breaks. Here, the lkelhood rato test statstcs were adjusted by the Cheung-La (1993) correcton term. We performed the contegraton tests through two types of model, namely (1) a restrcted determnstc lnear trend (RDLT) model, and (2) a determnstc lnear trend (DLT) model. The RDLT model takes only one ntercept (or none) n the determnstc lnear trend, and the DLT model consders both ntercept and trend. Table 4 shows the results of the bvarate contegraton analyss. We rejected the null of noncontegraton for Argentna-Uruguay (at one percent), Brazl-Paraguay and Paraguay-Venezuela (at 5 percent), and for Argentna-Venezuela and Paraguay- Uruguay (at 10 percent). Overall the contegraton vector parameters β s were sgnfcant. In partcular, for Argentna-Uruguay and Paraguay-Uruguay the sum of the β s were not sgnfcantly dfferent from zero. Ths at frst suggests leavng the USA out from the potental currency area (see Enders and Hurn 1994). However, the contegraton vector parameter reflects not only the trade relaton but also broader fundamental macro varables trackng the lnkages between the countres, such as technology transfers, mmgraton, and fnancal resource movements (Enders and Hurn 1994). The more smlar the aggregate demand parameters, the smaller the contegraton vector parameter. Tables 5 9 show that all the contegraton vector parameters fell above 0.8 n absolute value (see also Fgures 2 6). Ths means that the countres are very dssmlar, regardless of whether a long run relatonshp exsts. The α s n the tables are the adjustment speed parameters from the short run to the long run. The sgnfcant absolute values were lesser than 0.21, whch means large devaton persstence. It s necessary from 1.2 to 4 years for the short run devatons to damp out. Ths fndng s consstent wth the PPP puzzle,.e. hgh short run real exchange rate volatlty accompaned by slow adjustment process toward the long run path. Whenever the α s were nonsgnfcant (meanng weak exogenety ), we also performed extra Granger-causalty tests to check for strong exogenety, whch means that the other country n a relaton does not affect the country wth the nonsgnfcant adjustment coeffcent. The null of strong exogenety could not be rejected at one percent n the bvarate relatons. By frst consderng core Mercosur (wthout Venezuela), we could reject the null of noncontegraton at one percent (Table 10 and Fgure 7). The β s were all sgnfcant at one percent, and ther sum was not nl (10 percent sgnfcant). Ths means that the USA cannot be excluded from the potental currency area. The core Mercosur countres depended on the US fundamentals. Moreover, the aggregate demand parameters revealed cross-country dssmlartes. In partcular, the coeffcents of adjustment speed for Brazl were nonsgnfcant, and the null of strong exogenety could not be rejected. For Argentna, Paraguay, and Uruguay the α -persstence ranged from 1.3 to 3.7 years. The null of noncontegraton could also be rejected for full Mercosur (ncludng Venezuela). Table 11 shows the estmates of the contegraton vector and the adjustment coeffcents (see also Fgure 8). The β s were all sgnfcant and ther sum was not nl (5 percent sgnfcant). Ths replcated the former fndng that the USA

6 cannot be excluded from the potental currency area. Also, the β s all fell above 0.5 n absolute value, thus renforcng the fndng of cross-country dssmlartes. The estmates of the adjustment speed parameter also presented the PPP puzzle. And Argentna and Venezuela were found to be strongly exogenous to the other Mercosur partners. In short, G-PPP held for Mercosur and then a necessary condton for an optmum currency area was met. Yet we cannot push ths result too far and conclude that Mercosur s ready for ether a sngle currency or offcal dollarzaton because we also found large dssmlartes between the countres. 4. Concluson Mercosur s real exchange rates were found nonstatonary, and ths allowed us to assess whether G-PPP holds through a contegraton analyss. Our fndngs suggested that the null of noncontegraton could be rejected for Mercosur. But one cannot jump to the concluson that t consttutes an optmum currency area because the contegraton vector estmates ponted to large cross-country dssmlartes. We also found that Mercosur presents the PPP puzzle.

7 ARG BOL BRA CHIL PAR URU VEN Fgure 1. Real exchange rates aganst the US dollar, 1973Q3 2006Q3 Fgure 2. Contegraton relaton n the DLT model: Argentna and Uruguay Fgure 3. Contegraton relaton n the RDLT: Argentna and Venezuela

8 Fgure 4. Contegraton relaton n the RDLT model: Brazl and Paraguay Fgure 5. Contegraton relaton n the DLT model: Paraguay and Uruguay Fgure 6. Contegraton relaton n the DLT model: Paraguay and Venezuela Fgure 7. Contegraton relaton n the DLT model: core Mercosur Fgure 8. Contegraton relaton n the DLT model: full Mercosur

9 Table 1. Unt root tests Country Seres wth DP ADF PP ERS ARG BOL BRA CHI PAR URU VEN Intercept * *** Determnstc Lnear Trend * Intercept * ** Determnstc Lnear Trend * 3.295*** 4.934* Intercept 8.753* Determnstc Lnear Trend 8.753* Intercept 4.809* 5.081* 4.589* Determnstc Lnear Trend 4.809* 4.836* 4.235* Intercept 8.490* Determnstc Lnear Trend 8.490* Intercept * Determnstc Lnear Trend * Intercept 9.803* ** Determnstc Lnear Trend 9.803* * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10% Table 2. Lee-Strazcch unt root tests Country Seres wth model of Break Date 1 Break Date 2 t-statstc ARG Crash 1982Q2* 1990Q1** NS Trend Break 1990Q1* 2002Q1** NS BOL Crash 1985Q1* NS Trend Break 1983Q1* 1986Q4* 6.970* BRA Crash 1979Q4** 2002Q2* NS Trend Break 1984Q3 NS 1998Q3* NS CHI Crash 1982Q2* 1999Q4*** NS Trend Break 1985Q3* 1999Q2*** NS PAR Crash 1986Q4* NS Trend Break 1983Q3* 1986Q3** NS URU Crash 1982Q3* 2002Q2* NS Trend Break 1985Q2 NS 2000Q4* NS VEN Crash 1986Q4* 1999Q3*** NS Trend Break 1989Q3* 4.516*** * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant Table 3. Monetary regme changes Country Break Date 1 Break Date 2 ARG 1991Q2 2001Q4 BRA 1994Q3 1998Q4 PAR 1989Q2 2001Q4 URU 1990Q4 2001Q4 VEN 1989Q2 2002Q1 Table 4. Bvarate contegraton analyss ARG BRA PAR URU BRA λ trace 0 PAR λ trace 0 1** (no break) URU λ trace 1* (breaks) 0 1*** (breaks) VEN λ trace 1*** (no breaks) 0 1** (breaks) 0 * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%

10 Table 5. Contegraton vector for Argentna and Uruguay (breaks at 1990Q4 and 2001Q4) Country Determnstc Lnear Trend Model C0 = 3.024* C1 = NS C2 = 4.901** α T0 = NS T1 = NS T2 = 0.037** β ARG 0.210* 1.000* URU 0.158* 1.224* Total NS * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant Table 6. Contegraton vector for Argentna and Venezuela Country Restrcted Determnstc Lnear Trend Model α β ARG 0.094* 1.000* VEN NS (1) 1.367** Total 2.367* * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant (1) The hypothess that ARG does not Granger-cause VEN could not be rejected at 1% Table 7. Contegraton vector for Brazl and Paraguay Country Restrcted Determnstc Lnear Trend Model α β BRA NS (1) 1.000* PAR 0.062* 1.404* Total 0.404*** * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant (1) The hypothess that PAR does not Granger-cause BRA could not be rejected at 1% Table 8. Contegraton vector for Paraguay and Uruguay (breaks at 1989Q2 and 2001Q4) Country Determnstc Lnear Trend Model C0 = 5.128* C1 = 0.907** T0 = NS T1 = 0.015* C2 = 3.917*** T2 = 0.032** α β PAR 0.179* 1.000* URU NS (1) 0.868* Total NS * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant (1) The hypothess that PAR does not Granger -cause URU could not be rejected at 1%

11 Table 9. Contegraton vector for Paraguay and Venezuela (breaks at 1989Q2 and 2001Q4) Country Determnstc Lnear Trend Model C0 = 5.995* C1 = 3.846* C2 = NS α T0 = NS T1 = 0.037* T2 = NS β PAR NS (1) 1.000* VEN 0.110* 2.138* Total 1.138* * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant (1) The hypothess that PAR does not Granger -cause VEN could not be rejected at 1% Table 10. Contegraton vector for core Mercosur (breaks at 1990Q4 and 1998Q4) Country Determnstc Lnear Trend Model C0 = NS T0 = 0.013* C1 = 0.291** C2 = NS α β ARG 0.193* 1.000* BRA NS 0.578* PAR 0.068** 0.466* URU 0.164* 1.722* Total 0.322*** * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant The hypothess that ARG, PAR, and URU do not Granger-cause BRA could not be rejected at 1% Table 11. Contegraton vector for full Mercosur (breaks at 1994Q3 and 2002Q1) Country Determnstc Lnear Trend Model C0 = NS T0 = 0.012** C1 = 7.284* T1 = 0.073* C2 = 9.490* T2 = 0.076* α β ARG NS 1.000* BRA 0.070* 0.549*** PAR 0.059* 1.853* URU 0.108* 1.371* VEN NS 2.570* Total 0.895** * sgnfcant at 1%, ** sgnfcant at 5%, *** sgnfcant at 10%, NS nonsgnfcant The hypotheses that BRA, PAR, URU, and VEN do not Granger-cause ARG, and that ARG, BRA, PAR, and URU do not Granger-cause VEN could not be rejected at 1%

12 References Ahn, C., H. B. Km, and D. Chang (2006) Is East Asa ft for an optmum currency area? An assessment of the economc feasblty of a hgher degree of monetary cooperaton n East Asa The Developng Economes 44, Ahumada, H., and A. M. Martrena-Mantel (2001) Towards a Potental Monetary Unon n Latn Amerca: Testng the Endogenety Crtera for Mercosur XVIII Latn Amercan Meetng of the Econometrc Socety. Alexander, V., and G. M. Von Furstenberg (2000) Monetary unons: a superor alternatve to full dollarzaton n the long run North Amercan Journal of Economcs and Fnance 11, Barebom, I. (2004) Unão Monetára para Brasl e Argentna: uma Análse Empírca n La Cordnacón Macroeconómca y la Cooperacón Monetara, sus Costos, Benefcos y Aplcabldad en Acuerdos Regonales de Integracón by M. De La Cuba et al., Eds., INTAL/Inter-Amercan Development Bank: Buenos Ares, Bayoum, T., and B. Echengreen (1994) One Money or Many? Analyzng the Prospects for Monetary Unfcaton n Varous Parts of the World Prnceton studes n nternatonal fnance 76. Berg, A., E. Borenszten, and P. Mauro (2002) An evaluaton of monetary regme optons for Latn Amerca The North Amercan Journal of Economcs and Fnance 13, Bernsten, D. J. (2000) Generalzed purchasng power party and the case of the European Unon as a successful currency area Atlantc Economc Journal 28, Busse, M., C. Hefeker, and G. Koopmann (2006) Between two poles: a dual currency board for Mercosur The North Amercan Journal of Economcs and Fnance 17, Calderon, C., A. Chong, and E. Sten (2007) Trade ntensty and busness cycle synchronzaton: Are developng countres any dfferent? Journal of Internatonal Economcs 71, Camarero, M., R. G. Flores, and C. R. Tamart (2006) Monetary unon and productvty dfferences n Mercosur countres Journal of Polcy Modelng 28, Cheung, Y. W., and K. S. La (1993) Fnte-sample szes of Johansen s lkelhood rato test for contegraton Oxford Bulletn of Economcs and Statstcs 55, Choudhry, T. (2005) Asan currency crss and the generalzed PPP: evdence from the Far East Asan Economc Journal 19, Corbo, V. (2001) Is t tme for a common currency for the Amercas? Journal of Polcy Modelng 23,

13 Da Slva, S. (2002) Classroom gude to the equlbrum exchange rate model Economc Issues 7, Dckey, D. A., and S. G. Pantula (1987) Determnng the order of dfferencng n autoregressve processes Journal of Busness and Economc Statstcs 5, Ellott, G., T. J. Rothenberg, and J. H. Stock (1996) Effcent tests for an autoregressve unt root Econometrca 64, Enders, W., and S. Hurn (1994) Theory and tests of generalzed purchasng power party: common trends and real exchange rates n the Pacfc Rm Revew of Internatonal Economcs 2, Enders, W., and S. Hurn (1997) Common trends and generalzed purchasng power party Mathematcs and Computers n Smulaton 43, Frankel, J. A., and A. K. Rose (1998) The endogenety of the optmum currency area crtera Economc Journal 108, Fratann, M. (2004) The Case for Monetary Unon n Mercosur n Money Matters: Essays n Honor of Alan Walters by P. Mnford, Ed., Edward Elgar: Cheltenham, Hallwood, P., I. W. Marsh, and J. Schebe (2006) An assessment of the case for monetary unon or offcal dollarzaton n fve Latn Amercan countres Emergng Markets Revew 7, Hochreter, E., K. Schmdt-Hebbel, and G. Wnckler (2002) Monetary unon: European lessons, Latn Amercan prospects The North Amercan Journal of Economcs and Fnance 13, Johansen, S. (1988) Statstcal analyss of contegraton vectors Journal of Economc Dynamcs and Control 12, Johansen, S. (1991) Estmaton and hypothess testng of contegraton vectors n Gaussan vector autoregressve models Econometrca 59, Johansen, S., R. Moscon, and B. Nelsen (2000) Contegraton analyss n the presence of structural breaks n the determnstc trend The Econometrcs Journal 3, Kawasak, K., and E. Ogawa (2006) What should the weghts of the three major currences be n a common currency basket n East Asa? Asan Economc Journal 20, Kenen, P. B. (1969) Theory of Optmum Currency Areas: An Eclectc Vew n Monetary Problems of the Internatonal Economy by R. A. Mundell and A. K. Swoboda, Eds., Unversty of Chcago Press: Chcago,

14 Lee, M. (2003) Common trend and common currency: Australa and New Zealand Internatonal Journal of Busness and Economcs 2, Lee, J., and M. C. Strazcch (2003) Mnmum Lagrange multpler unt root test wth two structural breaks Revew of Economcs and Statstcs 85, Lee, J., and M. C. Strazcch (2004) Mnmum LM Unt Root Test wth One Structural Break Appalachan State Unversty Department of Economcs workng paper Lang, H. (1999) Do Hong Kong SAR and Chna Consttute an Optmal Currency Area? An Emprcal Test of the Generalzed Purchasng Power Party Hypothess Internatonal Monetary Fund workng paper 99/79. Lcandro-Ferrando, G. (2000) Un Área Monetára para el Mercosur? n Coordnacón de Polítcas Macroeconómcas en el MERCOSUR by J. Carrera and F. Sturzenegger, Eds., Fondo de Cultura Económca: Buenos Ares, Machnea, J. L. (2004) Mercosur: n Search of a New Agenda; Exchange Rate Instablty n Mercosur: Causes, Problems and Possble Solutons, INTAL/Inter- Amercan Development Bank: Buenos Ares. Mundell, R. (1961) The theory of optmum currency areas Amercan Economc Revew 51, Perron, P. (1997) Further evdence on breakng trend functons n macroeconomc varables Journal of Econometrcs 80, Salvatore, D. (2001) Whch countres n the Amercas should dollarze? Journal of Polcy Modelng 23, Sngh, A., A. Belasch, C. Collyns, P. De Mas, R. Kreger, G. Meredth, and R. Rennhack (2005) Stablzaton and Reform n Latn Amerca: A Macroeconomc Perspectve on the Experence Snce the Early 1990s Internatonal Monetary Fund occasonal paper 238. Zvot, E., and D. W. K. Andrews (1992) Further evdence on the Great Crash, the olprce shock, and the unt-root hypothess Journal of Busness and Economc Statstcs 10,