DOCUMENTOS DE ECONOMÍA Y FINANZAS INTERNACIONALES Working Papers on Inernaional Economics and Finance DEFI 11-07 Ocober 2011 Susainabiliy of exernal imbalances in he OECD counries Oscar Bajo-Rubio Carmen Díaz-Roldán Vicene Eseve Asociación Española de Economía y Finanzas Inernacionales www.aeefi.com ISSN: 1696-6376
Susainabiliy of exernal imbalances in he OECD counries * Oscar Bajo-Rubio (Universidad de Casilla-La Mancha) Carmen Díaz-Roldán (Universidad de Casilla-La Mancha) Vicene Eseve (Universidad de Valencia and Universidad de La Laguna) Absrac In his paper, we provide a es of he susainabiliy of exernal imbalances in he OECD counries, over he years 1970-2007. Specifically, we deal wih he case of hose counries ha have experienced curren accoun deficis in more han half of he years hroughou he period of analysis, and address he recen criique of Bohn (2007) on uni roo and coinegraion ess of he ineremporal budge consrain. JEL classificaion: F32, F41. Keywords: Exernal imbalances, Susainabiliy, Curren accoun. * The auhors acknowledge financial suppor from he Spanish Insiue for Fiscal Sudies, he Spanish Minisry of Science and Innovaion (Projecs ECO2008-05072-C02-01 O. Bajo-Rubio and C. Díaz-Roldán and ECO2008-05908-C02-02 V. Eseve ), and he Deparmen of Educaion and Science of he regional governmen of Casilla-La Mancha (Projec PEII09-0072- 7392). V. Eseve also acknowledges suppor from he Generalia Valenciana (Projec GVPROMETEO2009-098).
1. Inroducion Global exernal imbalances seem nowadays o be quie differen as compared o hose prevailing in he pas. Firs, hey mosly affec o rich counries, boh he US and wihin he euro area. In addiion, hey are primarily driven by privae saving and invesmen decisions, raher han by governmen deficis. Moreover, hese imbalances are financed in a more orhodox way (i.e., hrough eiher direc or porfolio invesmen), raher han hrough bank lending. Accordingly, hese imbalances are a maer of concern for such counries, requiring an adequae policy answer (Blanchard, 2007). The problem of global exernal disequilibria relaes in urn o he curren inernaional financial crisis, as discussed a lengh in Obsfeld and Rogoff (2010). Regarding he case of he European Union (EU), hose counries wih he highes curren accoun disequilibria are hose ha have experienced a greaer fall in heir levels of domesic demand (Lane, 2010). Moreover, i has been argued ha for hese counries, usually he eurozone members wih lower income levels, borrowing in inernaional markes would have become easier before he beginning of he crisis. In fac, he greaer financial inegraion, ogeher wih he adopion of he euro, would have mean a reducion in he cos of capial and he disappearance of exchange rae risk. As a resul, his would have ranslaed ino boh a decrease in saving and an increase in invesmen, and hence ino a deerioraion of he curren accoun balance (Blanchard and Giavazzi, 2002). Even more, he prospecs of convergence as regards he richer counries would have favour growh expecaions in hose counries, which conribued addiionally o greaer deficis; see Lane (2010). However, unlike he case of he US, where he size of he exernal defici has led o a wide academic debae [see, e.g., Mann (2002), Blanchard, Giavazzi and Sa (2005), Edwards (2005) or Obsfeld and Rogoff (2007)], his has no been he norm in he European case, wih a predominance of descripive sudies of a limied analyical conen. The usual way o analyzing curren accoun imbalances makes use of he ineremporal approach o he curren accoun (Sachs, 1981; Obsfeld and Rogoff, 1995; Razin, 1995). According o his approach, given ha, from he perspecive of he naional accouns, he curren accoun equals he difference beween savings and invesmen, and, because savings and invesmen decisions are based on ineremporal facors (such as life-cycle feaures, he expeced reurns of invesmen projecs, and he like) he curren accoun is necessarily an ineremporal phenomenon. In an imporan conribuion, Milesi-Ferrei and Razin (1996) discuss he usual noion of susainabiliy in relaion o he counry s ineremporal solvency, ha is, when he presen discouned value of fuure rade surpluses equals curren exernal indebedness. Pu in oher words, curren accoun susainabiliy would be defined as he abiliy of an economy of saisfying is ineremporal consrain in he long run, in absence of a drasic change eiher in he behaviour of he privae secor or in economic policy (Taylor, 2002). In general, a curren accoun defici in excess of 5% of GDP is regarded as unsusainable, so ha above his hreshold he adjusmen process of he curren accoun usually begins (Freund, 2005). Following his general approach several papers have appeared, which es for curren accoun susainabiliy by using coinegraion analysis. Specifically, a curren accoun defici would be susainable if he series for expors and impors were coinegraed; see, e.g., Hused (1992), Wickens and Ucum (1993), Wu, Founas and 1
Chen (1996), Apergis, Karakilidis and Tabakis (2000), or Wu, Chen and Lee (2001), o name a few. However, radiional analyses of susainabiliy (for eiher he public defici or he exernal defici) have been recenly criicized by Bohn (2007). According o his auhor, hese ess, based on uni roo and coinegraion ess of he (governmen s or naion s) ineremporal budge consrain, are incapable of rejecing he exisence of susainabiliy. In place of he radiional mehod, Bohn proposes an alernaive approach, based on he exisence of an arbirarily high order of inegraion of he variables involved, and on error-correcion-ype policy reacion funcions. Moreover, his previous lieraure neglecs he role of capial gains or losses on ne foreign asse posiions. According o Gourinchas and Rey (2007), for a counry facing an exernal disequilibrium, adjusmen hrough fuure rade surpluses (he so called rade channel, i.e., ha sressed by he ineremporal approach o he curren accoun) will be complemened by changes in he reurns on domesic asses held by foreigners relaive o he reurn on foreign asses held by domesic residens. The laer effec (he so called valuaion channel ) may occur in urn via a depreciaion of he domesic currency. Therefore, in his paper we will use Bohn s approach o assess he susainabiliy of exernal imbalances in hose OECD counries experiencing curren accoun deficis in more han half of he years along he period 1970-2007, allowing for he valuaion effecs emphasized by Gourinchas and Rey. The paper is organized as follows. In secion 2, we describe he underlying heoreical framework. Nex, in secion 3 we inroduce he empirical mehodology, and discuss he daa and he empirical resuls. The main conclusions are summarized in secion 4. 2. Theoreical framework The susainabiliy of exernal deficis is a maer of concern for governmens, and is relaed o he issue of long-run solvency. A curren accoun defici is regarded as susainable when, if mainained in he indefinie fuure, i does no violae he naion s solvency consrain; and a naion is said o be solven if he presen-value budge consrain, i.e., is ineremporal budge consrain (IBC) holds. In oher words, a defici can be susainable if he counry can borrow. However, if he ineres rae on he exernal deb exceeds he growh rae of he economy, deb dynamics would lead o an ever-increasing raio of deb o GDP. The dynamics of deb accumulaion could be sopped only if he raio of he exernal defici o GDP would urn o be a surplus. The cusomary approach for analyzing exernal imbalances is based on he ineremporal approach o he curren accoun. Under his approach, he curren accoun (i.e., changes in a counry s ne indebedness) is considered as an ineremporal issue, since decisions on indebedness imply changes in fuure consumpion possibiliies and hese are based on expecaions of he enire fuure pah of a number of variables. The ineremporal model of he curren accoun originaes in he work of, among ohers, Sachs (1981, 1982), Obsfeld (1982), and Svensson and Razin (1983); some overviews are provided in Obsfeld and Rogoff (1995) and Razin (1995). 2
We sar wih some accouning ideniies. In period, he curren accoun, i.e., he change in ne foreign asses vis-à-vis he res of he world, equals ne expors of goods and services plus ne facor paymens from abroad 1 : NFA CA NX rnfa 1 (1) where CA, NFA and NX sand for he curren accoun, ne foreign asses and ne expors, respecively, all of hem in real erms, and r is a (consan) real ineres rae. Noice ha when NFA > 0 he counry is a ne credior, and when NFA < 0 he counry is a ne debor. Alernaively: NFA CA Q rnfa 1 C I S I (2) where Q is gross domesic produc (GDP, so Q + rnfa 1 is gross naional produc), and C, S and I denoe oal (i.e., privae plus public) consumpion, saving and invesmen, respecively. As can be seen, equaion (2) links he curren accoun balance wih decisions on saving and invesmen. Since (1) holds every period, solving for NFA and ieraing forward over an infinie horizon yields he naion s IBC, wrien in erms of GDP shares: j1 j1 1 g 1 g Enx j1 lim Enfa j1 0 1 j j r 1 r nfa (3) where nfa and nx denoe, respecively, ne foreign asses and ne expors, boh as raios o GDP; E is he expecaions operaor; and g sands for he rae of growh of real GDP, assumed (as he real ineres rae) o be consan for simpliciy. The condiion for curren accoun susainabiliy is: 1 g lim Enfa j1 0 j 1 r i.e., he ransversaliy condiion; or, equivalenly: j1 1 g nfa Enx j1 (5) j0 1 r By muliplying boh sides of (5) by 1, so ha he counry is a ne debor, we can see ha solvency requires ha he counry mus run expeced fuure rade surpluses equal, in presen-value erms, o he curren value of is ousanding ne liabiliies vis-à-vis he res of he world. The sandard approach o es for susainabiliy of he curren accoun consiss of esimaing a coinegraion relaionship beween ne expors and he (lagged) level of ne foreign asses, boh as raios o GDP: nx α βnfa1 v (6) where ν denoes an error erm. In his equaion, a negaive and significan esimae of β would be a sufficien condiion for solvency, indicaing ha he naion saisfies is presen-value budge consrain. Tesing wheher β < 0 from he esimaion of (6) or, alernaively, wheher β = 1 from he esimaion of a coinegraion relaionship such as: exp α βimp ε (7) j1 (4) 1 Noice ha we are omiing here unilaeral ransfers, usually a small iem in he balance of paymens. Alernaively, ne expors could be assumed ne of ransfers. 3
where exp and imp denoe, respecively, he GDP raios of he expors of goods and services, and he impors of goods and services plus ne ineres paymens and ne ransfer paymens, and ε is an error erm, are cusomary approaches o es for he susainabiliy of exernal imbalances. However, his kind of assessmens of exernal susainabiliy based on uni roo and coinegraion ess have been recenly criicized by Bohn (2007), on he grounds ha such ess are incapable of rejecing susainabiliy. Specifically, Bohn derives he following hree proposiions, relaed o he order of inegraion of ne foreign asses, ne expors, expors, and impors, in order o verify under which condiions he ransversaliy condiion and he IBC hold (see Bohn (2007) for deails): (i) If nfa is inegraed of order m for any finie m 0, hen nfa saisfies he ransversaliy condiion, and nfa and nx saisfy he IBC. (ii) If exp and imp are inegraed of order m X and m M, respecively, where nfa = exp imp ; hen nfa is inegraed of order m wih m max(m X, m M ) + 1, so he ransversaliy condiion and he IBC hold. (iii) If nfa and nx follow an error-correcion specificaion of he form nx + ρnfa 1 = z, and z is inegraed of order m for some ρ < 0 such ha 0, 1 r ρ where r is a consan ineres rae, hen nfa saisfies he ransversaliy condiion and he IBC holds. Noice, on he oher hand, ha hese are jus sufficien condiions, so ha a failure of he ess would no mean a rejecion of susainabiliy. 3. Daa and empirical resuls In his secion, we provide a es of Bohn s hree proposiions for he case of he susainabiliy of curren accoun imbalances in he OECD counries. We use daa on ne expors and ne foreign asses, as well as on expors and impors of goods and services (he laer augmened wih ne ineres paymens and ne ransfer paymens), all of hem as percenages of GDP, for hose OECD counries experiencing curren accoun deficis in more han half of he years along our sample period. These counries are Ausralia, Ausria, Canada, Greece, Ireland, Ialy, New Zealand, Porugal, Spain, he UK, and he US. The daa are annual, and have been aken from he Inernaional Moneary Fund s Inernaional Financial Saisics. In urn, he ne foreign asse posiions have been aken from he updaed and exended version of he Exernal Wealh of Naions Mark II daabase developed by Lane and Milesi-Ferrei (2007), which includes he valuaion effecs menioned above. The sample period runs from 1970 o 2007, i.e., he las year for which he daa on ne foreign asse posiions are available. In a relaed paper, Durdu, Mendoza and Terrones (2010) analyze he susainabiliy of exernal imbalances, using a panel of 50 counries (21 indusrial and 29 emerging) along he period 1970-2004. In his paper, however, we will focus only on hose counries experiencing exernal deficis, since susainabiliy should apply o deficis raher han surpluses, for he case of indusrial counries. In addiion, we will perform he analysis on a counry-by-counry basis, since panel esimaion can hide he differen behaviour of specific counries regarding susainabiliy of heir exernal imbalances. 4
We begin by esing for he order of inegraion of he variables nfa, exp, and imp, using he ess of Ng and Perron (2001). These auhors proposed using he ess saisics and, which are modified versions of he Z α and Z ess of Phillips and Perron (1988), compued afer derending he series under analysis using he mehod of Generalized Leas Squares (GLS). Such modificaions improve he ess wih regard o boh size disorions and power. According o he resuls in Table 1, he presence of wo uni roos is clearly rejeced a he convenional significance levels; and he null hypohesis of no saionariy canno be rejeced in all cases bu he variable imp for Ausralia, where i would be rejeced a he 10% level. Therefore, wih he only excepion of imp for Ausralia, he hree series would be concluded o be non-saionary, and he firs wo proposiions of Bohn (2007) would hold. [Table 1 here] Nex, we esimae, using he mehod of Non-Linear Leas Squares, he errorcorrecion specificaion analogue o (6): nx = ω + δ(l) nfa 1 + ρ(nx 1 α βnfa 2 ) + γ(l) nx + η (8) where η is an error erm, and he resuls are shown in Table 2. As can be seen, he errorcorrecion coefficien ρ always shows he expeced negaive sign, and is significan a he convenional levels in all cases wih he only excepion of Ireland, where significance only appears a he 15% level. Regarding he long-run coefficien β, we find: - a negaive and saisically significan coefficien for Ausria, Canada, Ialy, and New Zealand - a posiive and saisically significan coefficien for Ausralia, Porugal, Spain, and he US - a non significan coefficien for Greece, Ireland, and he UK [Table 2 here] Hence, he hird proposiion of Bohn (2007) would hold, and he curren accoun defici would be susainable, only for he cases of Ausria, Canada, Ialy, and New Zealand. In paricular, he adjusmen of he ne expors-gdp raio o a given change in he ne foreign asses-gdp raio would have had an average half-life of abou one, hree, one, and half a year, respecively 2. On he conrary, in he cases of Ausralia, Greece, Ireland, Porugal, Spain, he UK, and he US, Bohn s (2007) hird proposiion would no hold, so no evidence is found on he fulfilmen of he naion s IBC for hese counries. Noice, however, ha Bohn s approach gives only sufficiency condiions for susainabiliy o hold; in oher words, if he ess yield posiive resuls his means evidence indicaing ha he IBC holds, bu failure of he ess does no rejec i. 4. Conclusions In his paper, we have esed for he susainabiliy of exernal imbalances in he OECD counries over he years 1970-2007, addressing he recen criique of Bohn (2007) on previous uni roo and coinegraion ess of he IBC, and allowing for he valuaion 2 Compued as 0.5 log1 ρˆ column of Table 2. log, where ρˆ is he esimae of ρ in equaion (8), from he second 5
effecs emphasized by Gourinchas and Rey (2007). Unlike Durdu, Mendoza and Terrones (2010), we analyze he case of only hose counries undergoing curren accoun deficis in more han half of he years hroughou his period, since susainabiliy should apply o deficis raher han surpluses; and on a counry-by-counry basis, since panel esimaion can hide he differen behaviour of specific counries. Our resuls show ha he hree variables ne foreign asses, expors of goods and services, and impors of goods and services (augmened wih ne ineres paymens and ne ransfer paymens) would be inegraed of order one in he all counries analyzed, excep for imp for Ausralia (even hough a a 10% significance level). Accordingly, he IBC would hold in principle for all of hem, wih he possible excepion of Ausralia. However, when esimaing an error-correcion relaionship beween ne expors and ne foreign asses, he long-run coefficien had he expeced (negaive) sign, and was saisically significan, for Ausria, Canada, Ialy, and New Zealand, so ha for hese counries he curren accoun defici would be susainable. On he conrary, in he cases of Ausralia, Greece, Ireland, Porugal, Spain, he UK, and he US, no clear-cu resuls emerge, i.e., he IBC would fail in principle o hold bu, since Bohn s approach gives only sufficiency condiions, a failure of he ess does no mean a rejecion of susainabiliy. 6
References Apergis, N., Karakilidis, K. P. and Tabakis, N. M. (2000): Curren accoun defici susainabiliy: The case of Greece, Applied Economics Leers 7, 599-603. Blanchard, O. (2007): Curren accoun deficis in rich counries, IMF Saff Papers 54, 191-219. Blanchard, O. and Giavazzi, F. (2002): Curren accoun deficis in he euro area: The end of he Feldsein-Horioka puzzle?, Brookings Papers on Economic Aciviy 2, 147-209. Blanchard, O., Giavazzi, F. and Sa, F. (2005): Inernaional invesors, he U.S. curren accoun, and he dollar, Brookings Papers on Economic Aciviy 1, 1-49. Bohn, H. (2007): Are saionariy and coinegraion resricions really necessary for he ineremporal budge consrain?, Journal of Moneary Economics 54, 1837-1847. Durdu, C. B., Mendoza, E. G. and Terrones, M. E. (2010): On he solvency of naions: Are global imbalances consisen wih ineremporal budge consrains?, Working Paper WP/10/50, Inernaional Moneary Fund. Edwards, S. (2005): Is he U.S. curren accoun defici susainable? If no, how cosly is adjusmen likely o be?, Brookings Papers on Economic Aciviy 1, 211-288. Freund, C. (2005): Curren accoun adjusmen in indusrial counries, Journal of Inernaional Money and Finance 24, 1278-1298. Gourinchas, P.-O. and Rey, H. (2007): Inernaional financial adjusmen, Journal of Poliical Economy 115, 665-703. Hused, S. (1992): The emerging U.S. curren accoun defici in he 1980s: A coinegraion analysis, Review of Economics and Saisics 74, 159-166. Lane, P. R. (2010): A European perspecive on exernal imbalances, Repor 2010:5, Svenska Insiue for Europapoliiska Sudier. Lane, P. R. and Milesi-Ferrei, G. M. (2007): The exernal wealh of naions mark II: Revised and exended esimaes of foreign asses and liabiliies, 1970-2004, Journal of Inernaional Economics 73, 223-250. Mann, C. L. (2002): Perspecives on he U.S. curren accoun defici and susainabiliy, Journal of Economic Perspecives 16, 131-152. Milesi-Ferrei, G. M. and Razin, A. (1996): Curren-accoun susainabiliy, Princeon Sudies in Inernaional Finance 81, Princeon Universiy. Ng, S. and Perron, P. (2001): Lag lengh selecion and he consrucion of uni roo ess wih good size and power, Economerica 69, 1529-1554. Obsfeld, M. (1982): Aggregae spending and he erms of rade: Is here a Laursen- Mezler effec?, Quarerly Journal of Economics 97, 251-270. 7
Obsfeld, M. and Rogoff, K. (1995): The ineremporal approach o he curren accoun, in Grossman, G. and Rogoff, K. (eds.): Handbook of Inernaional Economics (Vol. 3), Norh-Holland, Amserdam, 1731-1799. Obsfeld, M. and Rogoff, K. (2007): The unsusainable U.S. curren accoun posiion revisied, in Clarida, R. H. (ed.): G7 curren accoun imbalances: Susainabiliy and adjusmen, The Universiy of Chicago Press, Chicago, 339-366. Obsfeld, M. and Rogoff, K. (2010): Global imbalances and he financial crisis: Producs of common causes, in Glick, R. and Spiegel, M. M. (eds.): Asia and he global financial crisis, Federal Reserve Bank of San Francisco, 131-172. Perron, P. and Ng, S. (1996): Useful modificaions o some uni roo ess wih dependen errors and heir local asympoic properies, Review of Economic Sudies 63, 435-463. Phillips, P. C. B. and Perron, P. (1988): Tesing for a uni roo in ime series regression, Biomerika 75, 335-346. Razin, A. (1995): The dynamic-opimizing approach o he curren accoun: Theory and evidence, in Kenen, P. B. (ed.): Undersanding inerdependence: The macroeconomics of he open economy, Princeon Universiy Press, Princeon, 169-198. Sachs, J. (1981): The curren accoun and macroeconomic adjusmen in he 1970 s, Brookings Papers on Economic Aciviy 1, 201-268. Sachs, J. (1982): The curren accoun in he macroeconomic adjusmen process, Scandinavian Journal of Economics 84, 147-159. Svensson, L. E. O. and Razin, A. (1983): The erms of rade and he curren accoun: The Harberger-Laursen-Mezler effec, Journal of Poliical Economy 91, 97-125. Taylor, A. M. (2002): A cenury of curren accoun dynamics, Journal of Inernaional Money and Finance 21, 725-748. Wickens, M. R. and Ucum, M. (1993): The susainabiliy of curren accoun deficis: A es of he US ineremporal budge consrain, Journal of Economic Dynamics and Conrol 17, 423-441. Wu, J.-L., Chen, S.-L. and Lee, H.-Y. (2001): Are curren accoun deficis susainable?: Evidence from panel coinegraion, Economics Leers 72, 219-224. Wu, J.-L., Founas, S. and Chen, S.-L. (1996): Tesing for he susainabiliy of he curren accoun defici in wo indusrial counries, Economics Leers 52, 193-198. 8
Table 1 Ng-Perron ess for uni roos I(2) vs. I(1) Ausralia Ausria Canada Greece Ireland Ialy nfa 17.90 *** 2.96 *** 17.93 *** 2.93 *** 17.28 *** 2.90 *** 15.18 *** 2.54 *** 17.98 *** 2.95 *** 17.92 *** 2.98 *** exp 17.95 *** 2.99 *** 16.40 *** 2.80 *** 12.64 *** 2.50 *** 17.08 *** 2.92 *** 15.53 *** 2.78 *** 17.94 *** 2.97 *** imp 17.84 *** 2.94 *** 17.58 *** 2.95 *** 16.60 *** 2.86 *** 16.90 *** 2.85 *** 17.70 *** 2.97 *** 17.69 *** 2.96 *** New Zealand Porugal Spain Unied Kingdom Unied Saes nfa 17.07 *** 2.91 *** 11.31 *** 2.26 ** 14.60 *** 2.45 ** 17.96 *** 2.92 *** 16.98 *** 2.88 *** exp 17.45 *** 2.95 *** 15.66 *** 2.78 *** 13.88 *** 2.63 *** 17.99 *** 2.94 *** 14.05 *** 2.59 *** imp 17.80 *** 2.98 *** 16.86 *** 2.90 *** 16.77 *** 2.86 *** 17.97 *** 2.93 *** 17.38 *** 2.93 *** I(1) vs. I(0) Ausralia Ausria Canada Greece Ireland Ialy nfa 10.57 2.28 13.32 2.55 1.37 0.58 1.59 0.60 4.40 1.48 10.19 2.15 exp 12.91 2.54 0.98 0.36 3.13 1.10 6.02 1.72 4.57 1.35 5.78 1.65 imp 16.85 * 2.87 * 4.95 1.37 4.19 1.23 6.06 1.61 7.29 1.87 7.83 1.87 New Zealand Porugal Spain Unied Kingdom Unied Saes nfa 5.28 1.56 1.13 0.49 1.78 0.54 5.12 1.56 4.34 1.47 exp 9.31 2.07 7.00 1.85 4.57 1.50 8.19 2.01 6.13 1.74 imp 11.91 2.43 11.11 2.33 5.92 1.57 11.65 2.40 11.88 2.39 Noes: (i) ** and *** denoe significance a he 5% and 1% levels, respecively. The criical values are aken from Ng and Perron (2001), Table I. (ii) The auoregressive runcaion lag has been seleced using he modified Akaike informaion crierion, as proposed by Perron and Ng (1996). 9
Table 2 Esimaion of long-run nonlinear relaionships beween ne expors and ne foreign asses Long-run coefficien Ausralia 0.02 *** (4.35) Ausria 0.40 *** ( 6.26) Canada 0.27 ** ( 2.37) Greece 0.03 (0.58) Ireland 0.12 ( 0.48) Ialy 0.15 *** ( 2.65) New Zealand 0.05 *** ( 2.85) Porugal 0.19 *** (2.44) Spain 0.11 *** (3.90) Unied Kingdom 0.03 (0.69) Unied Saes 0.13 ** (2.21) Error-correcion coefficien 0.74 *** ( 4.20) 0.50 *** ( 3.54) 0.21 ** ( 2.26) 0.20 * ( 1.70) 0.07 ( 1.58) 0.44 *** ( 3.68) 0.76 *** ( 4.46) 0.14 ** ( 1.97) 0.47 *** ( 3.30) 0.35 ** ( 2.45) 0.20 ** ( 2.11) Noes: (i) (ii) -saisics in parenheses. *, ** and *** denoe significance a he 10%, 5% and 1% levels, respecively. 10