Faculdade de Economia da Universidade de Coimbra Grupo de Esudos Moneários e Financeiros (GEMF) Av. Dias da Silva, 165 3004-512 COIMBRA, PORTUGAL gemf@fe.uc.p hp://gemf.fe.uc.p MICAELA ANTUNES & ELIAS SOUKIAZIS How well he balance-of- paymens consrain approach explains he Poruguese growh performance. Empirical evidence for he 1965-2008 period ESTUDOS DO GEMF N.º 13 2009 PUBLICAÇÃO CO-FINANCIADA PELA FUNDAÇÃO PARA A CIÊNCIA E TECNOLOGIA Impresso na Secção de Texos da FEUC COIMBRA 2009
How well he balance-of- paymens consrain approach explains he Poruguese Absrac growh performance. Empirical evidence for he 1965-2008 period. Micaela Anunes Faculy of Economics, Universiy of Coimbra, Porugal and Elias Soukiazis Faculy of Economics, Universiy of Coimbra and GEMF The presen sudy aims o verify wheher he balance-of-paymens consrained growh approach is suiable for explaining he Poruguese growh performance during he las decades. For ha, we adop Thirlwall s Law ha predics acual growh by he raio of he expors growh relaive o he income-elasiciy of he demand for impors. The income-elasiciy of impors, essenial for he enire analysis, is obained from he esimaion of he impors funcion by 2SLS, assuming ha domesic growh is endogenous. To smooh cyclical variaions, 15-year overlapping periods are considered in he compuaion of Thirlwall s Law, assuming ha income-elasiciy wih respec o impors is eiher consan or variable over ime. I is found ha he Law is a good insrumen for predicing acual growh in Porugal and his resul is reinforced by performing he McCombie es. Our resuls reveal ha Porugal grew slighly higher relaively o he OECD counries in he enire period and his is consisen wih he income-elasiciy of he demand for expors exceeding ha of impors, as Thirlwall s Law implies. Porugal also grew slighly faser han he rae consisen wih he balance-of-paymens equilibrium, accumulaing exernal deficis over ime. Dividing he sample in he pre and posadhesion period o he EU, i is shown ha Porugal grew a a lower rae in he laer, and his is consisen wih lower expor growh and higher income-elasiciy wih respec o impors. To overcome his problem, policies are needed o improve he supply characerisics of expors relaed o non-price compeiiveness and reduce he impors sensiiviy wih respec o domesic income changes. JEL code: C13, E12, F43, O24 Keywords: balance-of-paymens equilibrium growh rae, income-elasiciies wih respec o rade, overlapping periods, 2SLS regressions Auhor for correspondence: Elias Soukiazis, Faculy of Economics, Universiy of Coimbra, Av. Dias da Silva, 165, 3004-512 Coimbra, Porugal, Tel: +351239790534, Fax: +351239790514, e-mail: elias@fe.uc.p 1
1. Inroducion In he core of he debae beween he supply-consrained and he demand-led growh sands he very relevan conribuion of Thirlwall, under he Pos Keynesian framework, hrough wha became known as Thirlwall s Law. The growh of an economy is ulimaely deermined by effecive demand, especially exernal demand, insead of being explained by he accumulaion of facor inpus as he neoclassical heory assumes. The balance-of-paymens equilibrium growh rae is deermined by he growh of expors over he income-elasiciy of he demand for impors and his is relaed o he Harrod foreign rade muliplier when i is expressed in a dynamic form. The general proposiion of Thirlwall s Law is ha whenever an economy grows a a rae higher han ha consisen wih he balance-of-paymens equilibrium, i will run ino exernal deficis which are no susainable in he long-run, unless capial inflows can finance he ever growing imbalances. In case a counry falls in such a rap, domesic producion mus be adjused downwards resuling in higher unemploymen. Therefore, income is adjused o bring he economy back o equilibrium and no relaive prices as he neoclassical heory assumes. Compaible devaluaions are no he soluion, since in he long-run hey aggravae domesic inflaion, lowering compeiiveness and worsening even farher exernal imbalances. Srucural soluions are needed o urn expors more aracive in exernal markes and impors less sensiive o changes in domesic income. The aim of his paper is o analyse wheher Thirlwall s Law predics accuraely he acual growh in Porugal over he 1965-2008 period, a ask ha has no been ye underaken, a leas a he auhors knowledge. More specifically, he sudy is carried ou no only for he global period bu also for smaller overlapping sages, o which he McCombie es is applied o assess he validiy of he Law. The whole sample is also divided in he pre and pos-adhesion periods o deec differen endencies in growh performance. In secion 2 we reconsider he model developed by Thirlwall o predic a counry s acual growh. In secion 3 we explain he variables and he daa behaviour over he ime period considered. Uni roo ess are also performed o jusify he specificaion of he expors and impors demand funcions. The expors and impors demand funcions are esimaed in secion 4, o provide he income-elasiciies of demand for impors and expors necessary o se up Thirlwall s Law. In secion 5 we compue he growh raes consisen wih he balance-of-paymens equilibrium and 2
compare hem o he acual growh raes, for 15-year overlapping periods eiher using a consan or a varying income-elasiciy of demand for impors over ime. The final secion concludes on he accuracy of Thirlwall s Law as an insrumen for predicing acual growh in Porugal. 2. Thirlwall s Law reconsidered In his secion we focus on Thirlwall s Law (Thirlwall, 1979) assuming boh ha he curren accoun is iniially in equilibrium 1 and relaive prices are sicky, a leas in he long-run. The reason for his specificaion may res in he exisence of compeiive markes or oligopolisic marke srucures (Thirlwall, 1986). Thirlwall s Law serves o predic he growh rae of he economy consisen wih he balance-of-paymens equilibrium (on curren accoun). Alhough i is an old-fashioned approach, he model brings very imporan insighs which become very relevan in he era of globalizaion and regional inegraion. The Law can be compued by assuming ha he demand for impors is a funcion of domesic income and relaive prices of impors and he demand for expors is a funcion of foreign income and relaive prices of expors. The model can be described by expressing he impors and expors demand funcions, as well as, he equilibrium condiion on curren accoun, in growh raes: gm gx ( pd pf e ) = π gy + ψ Impors demand funcion (1) ( pd pf e ) = ε gz + η Expors demand funcion (2) pd + gx = pf + e + gm Curren accoun equilibrium (3) In hese equaions gm, gx, gy and gz are he raes of growh of real impors, expors, domesic and foreign income, respecively. 2 As for he remaining variables, pd and pf are he raes of growh of domesic and impor prices and e is he rae of change of he exchange rae. π and ε are he income-elasiciies of demand for impors and expors, boh expeced o be posiive and ψ and η are he price-elasiciies of demand for impors and expors (ψ>0 and η<0). 1 Thirlwall and Hussain (1982) also consider a model saring from disequilibrium in curren accoun (implying he exisence of capial inflows). We will no adop ha version in he presen sudy. 2 For a descripion of he variables and daa sources, see he Appendix. 3
Subsiuing equaions (1) and (2) ino (3) and solving for gy, we obain he economy s rae of growh consisen wih he balance-of-paymens equilibrium (gy BP, ) given by: gy BP, (1 + η ψ )( pd pf e ) + ε ( gz ) = (4) π Assuming ha relaive prices remain unchanged in he long-run, ha is, pd pf e = 0 he expression simplifies o: ε ( gz ) gy BP, = (5a) π or gx gy BP, = (5b) 3 π Generally speaking, he rae of growh of a counry is approximaely given by he raio of he expors growh relaive o he income-elasiciy of demand for impors (5b). Tha is o say, he balance-of-paymens consrained growh rae is given by he growh of expors and he Harrod foreign rade muliplier ( 1 ). 4 Therefore, i is income ha adjuss π o preserve equilibrium and no relaive prices. If a counry wishes o conrol exernal deficis (coming from gy>gy BP ) i mus increase he consrain on he balance-ofpaymens, eiher hrough an increase in expors growh (gx) or a decrease in he incomeelasiciy of he demand for impors (π), or combining boh cases. Alernaively equaion (5a) can be rewrien as: gy BP, ε = (5c) gz π This expression ells us ha relaive income growh beween he domesic counry and he res of he world is given by he raio of he income-elasiciy of he demand for expors over he income-elasiciy of he demand for impors of he domesic economy. In oher words, a counry can grow faser han he res of he world (gy BP >gz) wihou 3 Equaion (5b) is obained from he resricion pd pf e = 0 on equaion (2). 4 For deails on his explanaion, see Thirlwall (1982). 4
creaing balance-of-paymens problems only if is income-elasiciy wih respec o expors is higher han ha of is impors (ε>π). This inerpreaion is ineresing and conneced o he concep of convergence or caching-up, where compeiiveness is he key facor for such endency o occur. In erms of policy, he counry has o improve he supply characerisics of he goods and services produced and urn he economy more compeiive in inernaional markes. These supply characerisics are relaed o he qualiy, design, produc differeniaion, innovaion, pos sale services, ec., which deermine he non-price compeiiveness. The hypohesis of consan relaive prices has been criicised in he lieraure for long (McGregor and Swales, 1985; 1991; Alonso and Garcimarín, 1998-99). By adoping his specificaion in he presen sudy we do no assume ha relaive prices do no maer in inernaional rade performance; only ha i is of minor significance. In mos empirical sudies in his field relaive prices have been shown o be saisically insignifican and even when hey are significan he price-elasiciies wih respec o impors and expors are very low when compared o he income-elasiciies, showing ha impors and expors are less sensiive o price changes han o income changes. In our sudy, when relaive prices are regressed on a ime rend here is evidence of absence of a significan rend during he period of analysis showing ha he hypohesis of consan relaive prices in he long-run is reasonable. Thirlwall s Law as has been defined in equaion (5b) will be esed empirically for he Poruguese economy over he period 1965-2008 in he following secions. 3. Variables and daa analysis The empirical analysis is focused on Porugal, for he period 1965-2008 (44 annual observaions). Our aim is o examine wheher he balance-of-paymens consrained growh approach is adequae o explain he performance of he Poruguese economy over his period. Despie he conroversy involving he ype of he variables o be used levels or raes of growh - we op for he laer o avoid he exisence of spurious relaions since in principle variables in growh raes are saionary. Thus, he opion for dynamic impors and expors demand funcions is suiable and in line wih oher sudies, as in Bairam (1993). 5
Impors and Expors growh performance The firs sep is o analyse informally he emporal evoluion of he variables o deec wheher some regular endencies exis for he whole period, as well as for he periods of he pre and pos-adhesion o he EU. Combining he informaion from Table 1 (he wo firs rows) and Char 1, we observe ha he peak for he growh of impors and expors (gm and gx, respecively) is 1987 and 1979, whereas he lowes record is regisered for boh in 1975. The annual average growh rae of expors (6.05%) is slighly lower han ha of impors (6.53%) in he whole period. 1979 is he year wih he highes growh of ne expors, whereas in 1974 (year of he change of he poliical regime) he mos negaive growh is achieved. From 1993 onwards he gap beween he growh of impors and expors is more sable which coincides wih he pos Maasrich period and he effor made owards a nominal convergence and a fixed exchange rae regime. Addiionally, dividing he global sample in he pre and pos-adhesion periods, i is observed ha impors grow faser in he pos-adhesion period (7.6% agains 5.37%) and expors grow slower in he same period (5.51% agains 6.65%). 5 Anoher remarkable resul is ha expors growh (6.65%) is higher han impors growh (5.37%) in he preadhesion period bu his endency is reversed in he pos-adhesion period, wih 5.51% for expors and 7.6% for impors. Therefore, Porugal is loosing compeiiveness in he pos-adhesion period compeing in a free marke and moving owards a fixed exchange rae sysem. [Inser Table 1 around here] [Inser Char 1 around here] Relaive income growh performance From he analysis of Table 1 (rows 3 and 4) and mosly from Char 2, i is shown ha boh he growh of domesic income (gy) and he growh of foreign income (gz) 6 follow a downward rend hroughou he whole period. Sill, for he global period he annual average growh rae of domesic income (3.58%) surpasses ha of exernal income 5 Moreover, he esimaion of he impors funcion by 2SLS (as in secion 4) wih an addiive dummy variable for he 1986-2008 period indicaes ha he adhesion o he EU increased he growh rae of impors. The resuls are available upon reques. 6 Foreign income is proxied by he growh rae of he OECD counries. This is a reasonable proxy, since more han 80% of Poruguese impors and expors are associaed wih hese counries. For deails on he compuaion of gz, see he Appendix. 6
(3.19%) showing evidence of moderae convergence. However, here are signs of divergence in many occasions: 1966, 1969, 1975, 1978, 1981, 1983-1985, 1992-1994, 2000 and finally from 2002 onwards, where Porugal is growing a a slower rae han ha of he OECD counries. I is imporan o noe ha Porugal grows a a higher rae in he pre-adhesion period (4.39%) han in he pos-adhesion period (2.84%) and ha he difference beween he growh of he Poruguese economy and ha of he OECD counries is higher in he former (0.69 percenage poins - p.p.) han in he laer (0.12 p.p.). Therefore, no only Porugal grows more rapidly in he pre-adhesion period bu also faser relaively o he OECD counries. In fac, he raio of relaive income growh for he global period, (gy/gz), is abou 1.12, bu i is higher in he pre-adhesion period (1.19) han in he pos-adhesion period (1.04). In general, he daa shows ha Porugal grew on average a a slighly higher rae han ha of he res of he world and for ha o be feasible, according o equaion (5c), he income-elasiciy of he demand for expors (ε) mus be higher han ha of impors (π). [Inser Char 2 around here] Relaive prices of impors and expors Turning o he analysis of relaive prices in Char 3 (and Table 1, rows 7 and 8), he annual growh rae of relaive prices of impors (grpm), defined as he difference beween he growh of domesic and impors prices, reaches a minimum in 1974 (gain in price compeiiveness) and he highes value in 1986 (loss of price compeiiveness). As for he relaive prices of expors (grpx), defined as he growh of expors less impors prices, he behaviour is similar. For boh proxies, he 1965-1985 negaive average implies a favourable posiion in erms of price compeiiveness, since domesic (for grpm) and expors prices (for grpx) are no growing as much as impors prices, and his is he pre-adhesion period. Tha paern is reversed during 1986-2008 (he pos-adhesion period), wih impor prices growing a a slower rae han domesic and expors prices. This endency is of course explained by he removal of impor ariffs and exchange rae sabiliy no allowing compeiive devaluaions. Combining hese resuls wih he expors and impors behaviour of Char 1, we conclude ha Porugal looses compeiiveness afer joining he EU in 1986, and his is associaed wih a lower growh performance in he same period relaively o he pre-adhesion period. 7
An ineresing aspec o noice is a long-run movemen of relaive prices of impors and expors owards zero. This can be aken as evidence ha relaive prices remain consan in he long-run ( pd pf e = 0 ) hus jusifying he use of equaions (5a) or (5b) for predicing he Poruguese acual growh. [Inser Char 3 around here] Consumpion and invesmen growh behaviour Privae consumpion (gc) and invesmen (gi) growh raes will be used as insrumens for domesic growh in he 2SLS esimaion of he impors funcion. This is he reason why hese wo variables appear in Table 1 (rows 5 and 6) and hey exhibi differen growh behaviours. Consumpion growh performance is more or less sable, alhough in he pos-adhesion period a small fall occurs (3.12% agains 3.66%). On he conrary, invesmen has been growing faser afer Porugal has joined he EU (4.29% agains 3.21%). Therefore, he slower growh rae of he Poruguese economy in he posadhesion period is no due o he lack of invesmen bu raher o he poorer performance of expors and loss of compeiiveness of he economy as we observed before. Curren accoun performance The las row of Table 1 repors he curren accoun average (as a percenage of GDP) for he whole period and he wo sub-periods before and afer Porugal has joined he EU. The curren accoun average is always negaive, bu he sriking evidence is ha he average exernal defici is wice higher in he pos-adhesion han in he pre-adhesion period (-5.51% agains -2.73%). This is consisen wih he poorer performance of expors and he relaively higher increase in impors in he pos-adhesion period as we have seen before. The accumulaion of higher exernal deficis could explain he slower growh performance of he Poruguese economy in he laer period and his is consisen wih he balance-of-paymens consrain hypohesis which will be esed in he following secions. The correlaion coefficien beween gy and ca is posiive as expeced (0.5367) and significan a he 5% level (see Table 3) and hrough he analysis of Char 4 i is possible o observe ha generally, boh variables move in he same direcion. [Inser Char 4 around here] 8
Uni roo ess In ime series analysis i is pruden o ensure ha he series used are saionary, o avoid he exisence of spurious relaions. Thus, we use Augmened Dickey-Fuller ess (ADF) o check for uni roos based on hree alernaive specificaions: wih no consan and no rend; wih consan and no rend; wih consan and rend. The choice of he lag lengh ha ensures he absence of serial auocorrelaion follows Adkins and Hill (2008). We sar wih four lags, esing he saisical significance of he las lagged coefficien and eliminaing i in case of saisical insignificance. The process ends when he las lag is significan a leas a a 10% level. Addiionally, a Breusch-Godfrey LM es is performed afer he ADF regressions o check he exisence of serial auocorrelaion in he residuals. The idea is o choose he mos parsimonious ADF model wih no error auocorrelaion. Alernaively, he Phillips-Perron (PP) es for uni roo is also performed o check if he conclusions from he ADF ess are robus, using Newey-Wes sandard errors o accoun for serial correlaion. ADF ess are criicized for failing whenever a srucural break occurs in he period under analysis, for no considering he change in he mean ha i implies (McCombie, 1997). The uni roo ess are displayed in Table 2. In he same able we include also hose variables ha will be used as exra insrumens in he esimaion of he impors demand funcion: he growh of privae consumpion (gc) and he growh of real invesmen (gi). [Inser Table 2 around here] All variables are inegraed of order 0, I(0), implying ha we always rejec he null hypohesis of he exisence of a uni roo. Therefore, all variables we consider in he esimaion approach are saionary when expressed in growh raes ensuring ha no spurious relaions are involved when he impors and expors funcions are esimaed. 4. Esimaion of he impors and expors demand funcions As a saring poin, in Table 3 we presen a simple correlaion marix conaining only he significan correlaions (a he 5% significance level) beween he variables o consider in he esimaion approach. This preliminary analysis may help us finding he mos relevan explanaory variables boh in he expors and impors demand funcions. 9
Moreover, i also enables us o check he variables more closely linked o he (endogenous) domesic growh, o jusify he choice of insrumens. [Inser Table 3 around here] The firs aspec o noice is ha he growh of impors (gm) is linearly and posiively correlaed wih he growh of domesic oupu (gy), and expors growh (gx) is posiively correlaed wih he exernal oupu growh (gz) as expeced. The expeced posiive correlaion beween impors growh (gm) and relaive prices of impors growh (grpm) is only saisically confirmed when lagged values are used for he laer. As for expors growh (gx), i is posiively correlaed eiher wih he relaive prices of expors growh (grpx) or is lagged value (grpx -1 ). However his correlaion is modes, 0.32, and has a wrong sign. Regarding he growh of domesic oupu (gy), i is posiively relaed o privae consumpion (gc), invesmen (gi), expors (gx) and exernal oupu growh (gz), as expeced. I is imporan o highligh here ha he correlaion beween domesic growh and he curren accoun is posiive (0.54) implying ha higher curren accoun deficis are associaed wih lower growh raes, or wha i urns o be he same hing, higher curren accoun surpluses are associaed wih higher growh raes of domesic oupu, and his is consisen wih he balance-of-paymens consrained growh hypohesis. Accordingly, he impors demand funcion is specified as follows: gm = a + π ( gy ) + ψ ( grpm 1 ) + ω (6) I is expeced ha he growh of impors is posiively relaed o he growh of domesic income and he lagged value of he growh of relaive prices of impors (defined as he difference beween he growh of domesic and impors prices). We use lagged insead of curren prices essenially because he laer displays no saisical significance. In heoreical erms, i may be jusified by he fac ha relaive price changes have no an immediae impac on impors growh, given ha inernaional ransacions are based in conracs wih fixed erms in he shor-run. I is hus in conformiy wih he J-curve effec (Aesoglu, 1993). Analogically, he expors demand funcion is defined as: 10
gx = β + ε η ) + ν ( gz ) + ( grpx 1 (7) I is expeced ha a higher growh of foreign income (OECD counries) simulaes he growh of expors and ha he lagged relaive price of expors (defined as he difference beween he growh of expors and impors prices) have a negaive impac on expors growh. The firs sep is o esimae separaely each equaion by OLS and hese resuls are available on Table 4. 7 [Inser Table 4 around here] The oucomes for he expors funcion reveal a posiive and saisically significan income-elasiciy of he demand for expors bu he price-elasiciy carries a wrong posiive sign and i is saisically significan only a he 10% level. 8 An ineresing aspec o highligh is ha he income-elasiciy of he demand for expors (2.57) is higher han ha of impors (1.56) and his jusifies our earlier finding from he previous secion ha Porugal grew on average a a higher rae han ha of he OECD counries over he whole period, reflecing some kind of convergence or caching-up endency. Our focus is on he impors demand funcion and we observe ha boh he incomeelasiciy and price-elasiciy of he demand for impors display heir expeced signs and are saisically significan a he 1% level. However, hey may be biased and inconsisen due o he endogeneiy of he growh of domesic income. Two reasons can explain his endogeneiy, wih adverse endencies: a higher domesic growh may induce more impors and if impors rely on raw maerials, machinery and invesmen equipmen, hen growh will be induced farher. The second reason is from he balanceof-paymens perspecive: a higher increase in impors relaively o expor may deeriorae he rade balance posiion affecing negaively he growh of domesic income. Thus, an insrumenal variables approach is required o urn he resuls more consisen. From he correlaion analysis of Table 3, we sugges as insrumens for (gy) he growh of privae consumpion (gc), he growh of invesmen (gi) and he growh of 7 The regressions were run in Saa 10. 8 The (unexpeced) posiive impac of relaive prices on expors was also found by Bairam (1988), for Porugal, during 1970-1985. However, he magniude of he impac is very low when compared o ha of income. 11
expors (gx). The suiabiliy of hese insrumens will be esed in he 2SLS esimaion approach. Before urning o he insrumenal variables esimaion, we joinly run he expors and impors demand funcions by he SUR (Seemingly Unrelaed Regression) esimaion echnique (see Table 4). In case he error erms across equaions are conemporaneously correlaed, here are gains in efficiency from using his mehod in comparison o OLS (AlDakhil, 1998; Baum, 2006). The drawback is ha in SUR all regressors are exogenous (conradicing our assumpion of gy being endogenous). The resuls from he SUR esimaion do no differ subsanially from hose of OLS and according o he Breusch-Pagan (BP) es of cross error independence we rejec he null hypohesis of error independence beween equaions a he 5% significance level bu only marginally (a he 1% significance level he null is no rejeced). Thus, no significan efficiency gains arise from using full informaion esimaion echniques applied o sysem equaions. Since our aim is o obain esimaes for he income-elasiciy of demand for impors (πˆ ) in order o deermine he balance-of-paymens equilibrium growh rae relying on he assumpion ha gy is endogenous, we esimae he impors demand funcion using he 2SLS mehod, 9 as in Bairam (1988), Aesoglu (1993; 1995) and Léon-Ledesma (1999). The esimaes of income and price-elasiciies display he expeced signs and are saisically significan. The income-elasiciy of demand for impors (2.15) is higher han in he OLS (1.56) and SUR (1.53) mehods. Comparing he raio of he elasiciies (ε OLS /π 2SLS =1.20) wih he relaive income raio (gy/gz =1.12) he approximaion is closer han wih he OLS (ε/π=1.65) and SUR (ε/π=1.81) mehods, giving evidence in favour of Thirlwall s Law as has been expressed in equaion (5c). The Pagan-Hall heeroscedasiciy es indicaes he exisence of homoscedasiciy and hus here is no need for robus sandard errors. Addiionally, he Cumby-Huizinga es 10 shows he absence of 1 s order error auocorrelaion. The diagnosic ess from he 2SLS regression are saisfacory. The rank condiion for idenificaion is checked hrough he 9 For more informaion on insrumenal variables esimaion, see Baum e al. (2003). 10 The Cumby-Huizinga es is a generalizaion of he Breusch-Godfrey procedure o analyse he independence of he regression errors. I becomes especially useful in conexs of endogenous regressors, exisence of overlapping daa and condiional heeroscedasiciy of he regression error erm (Baum e al., 2007). 12
Anderson canonical correlaion LM saisic and shows ha he excluded insrumens are correlaed wih he endogenous regressor and he equaion is hus idenified. Furhermore, he Cragg-Donald Wald F-saisic indicaes ha he insrumens are no weak. The endogeneiy es for gy reveals ha his variable canno be reaed as exogenous in he impors demand funcion. We also confirm he hypoheses of exogeneiy of he insrumens and of non-redundancy (for gx, gi and gc). 11 Finally, he Sargan saisic leads us o accep he validiy of he insrumens se. 5. Balance-of-paymens equilibrium growh rae Overall, pre and pos-adhesion periods Afer he esimaion of he impors demand funcion, i is possible o compue he growh rae consisen wih he balance-of-paymens equilibrium o compare i wih he acual growh rae of he economy over he period 1965-2008. The expression (5b) is he preferred one o compue he balance-of-paymens consrained growh rae (Bairam, 1997), due o he insabiliy of he income-elasiciy of demand for expors over ime. The resuls can be observed in Table 5. [Inser Table 5 around here] The annual average growh rae of domesic income for he global period is abou 3.58% (aken from Table 1), which is higher han he average growh rae consisen wih he balance-of-paymens equilibrium (2.82%), meaning ha during he period 1965-2008 Porugal was growing beyond is capaciy, accumulaing balance-ofpaymens deficis. In fac, he average curren accoun defici (as percenage of GDP a marke prices), ca, is -4.18% for he whole period. Considering once more he pre and pos-adhesion periods o EU here are some ineresing remarks o make. In he pos-adhesion period, Porugal no only grows a a lower rae boh in erms of income and expors (as we menioned earlier) bu addiionally, he income-elasiciy wih respec o impors is higher in his period (2.66) relaively o he pre-adhesion period (2.22). As a consequence, he growh rae consisen wih he balance-of-paymens equilibrium is lower in his period (2.07) 11 The exogeneiy and redundancy ess are no included in he aricle for reasons of space, bu are available upon auhors reques. 13
relaively o he pre-adhesion period (2.99). Therefore, he increase in he incomeelasiciy of demand for impors afer Porugal joining he EU was no counerbalanced by he growh of expors o allow a higher growh of domesic income. On he oher hand, he increase in impors and he poorer performance in expors growh explain he higher exernal imbalances in he pos-adhesion period measured by he curren accoun defici (-5.51) which more han doubled relaively o he pre-adhesion period (-2.73). Table 5 also shows how close is he balance-of-paymens equilibrium growh rae (gy BP ) o he acual growh rae (gy) in Porugal. Comparing he difference beween he wo (gy BP -gy) we observe ha i is always negaive, revealing ha Porugal grew a a higher rae han ha consisen wih he balance-of-paymens equilibrium and his is in line wih he curren accoun deficis accumulaed over ime. The approximaion beween he wo growh raes is closer for he whole period (0.76 p.p. unexplained) and he pos-adhesion period (0.77 p.p. unexplained) bu wider in he pre-adhesion period (1.40 p.p. unexplained). Overall, Thirlwall s Law is a useful insrumen for predicing he growh performance in Porugal. Overlapping periods wih consan income-elasiciy of impors Criicisms on Thirlwall s Law refer ha he compuaion of a single growh rae for he global period is a shor-sided vision (Aesoglu, 1993). Therefore, we oped for analysing he same Law considering 30 overlapping periods wih a 15-year span. Firsly, we es he validiy of he Law by considering ha income-elasiciy of he demand for impors is he same for all periods, in line wih Léon-Ledesma (1999). The resuls are repored in Table 6, where we also display averages of he curren accoun (ca), he annual growh raes of domesic income (gy) and expors (gx), he annual growh rae of income consisen wih he balance-of-paymens equilibrium given by gy BP = gx πˆ, as well as he corresponding differences relaively o he acual growh raes (gy BP -gy). Following McCombie (1989) we also repor π * = gx, ha is, he gy income-elasiciy of demand for impors assuming equilibrium in he balance-ofpaymens (on curren accoun). If he average π * for he se of overlapping periods is no significanly differen fromπˆ, neiher is gy from gy BP, confirming herefore he validiy of Thirlwall s Law. 14
[Inser Table 6 around here] The difference beween gy BP and gy is negaive in mos of he periods, meaning ha he growh rae of he economy surpasses ha compaible wih he balance-of-paymens equilibrium, which in he long-run can no be susainable. In eleven ou of he hiry overlapping periods, from 1976 o 1997 and laer from 1992 o 2008, he difference is posiive, which would heoreically imply ha Porugal was growing less han i was capable of from he poin of view of he balance-of-paymens equilibrium and was herefore reducing he exernal imbalances. From he second column of Table 6 he reducion in curren accoun deficis is confirmed for he former bu no for he laer overlapping periods. This resul could eiher sugges ha πˆ is no sable over ime or he consideraion of capial inflows (in he form of srucural funds) is needed in he analysis, for gy BP o predic more accuraely he acual growh rae in Porugal. Following McCombie (1989), he average π * for he 30 overlapping periods is 1.89. When we esimae he impors funcion for he global period and es wheher π = π * using he -es, we conclude ha hey are saisically equal implying ha he condiion gy BP = gy is valid. Therefore, Thirlwall s Law is apparenly relevan o predic he Poruguese acual growh rae for he period 1965-2008. Overlapping periods wih income-elasiciy of impors changing over ime Conrary o Léon-Ledesma (1999), we compue he income-elasiciy of demand for impors by esimaing he impors demand funcion for each of he 15-year overlapping periods. The specificaion of he equaion is he same as for he global period, i.e., he growh of impors is relaed o he growh of domesic income and he growh of lagged relaive prices of impors and he esimaion mehod is again 2SLS, assuming ha domesic income is endogenous. Apar from he wo firs overlapping periods, he resuls indicae ha he equaions are idenified. Sill, we presen he resuls even for hose wo problemaic esimaions, o be coheren wih he global period 1965-2008, defined in he beginning of he analysis. In Table 7 we display he income-elasiciy of demand for impors obained from he 2SLS regression for each overlapping period and from Char 5 we are able o deec is general endency. 15
[Inser Char 5 around here] As we may infer from he char, here is a general favourable downward endency up o 1986. Afer he adhesion o he EU, here is a sharp increase in he income-elasiciy of impors ha is mainained approximaely ill 1990. In fac, he EU membership urned he Poruguese economy more vulnerable o impors due o he free circulaion of goods and services and abolishmen of any kind of ariffs on impors. In he following periods, he elasiciy drops from he 3% barrier and keeps more or less sable around ha limi unil 2004. In he las periods he income-elasiciy of he demand of impors is declining moderaely, bu is value remains higher han 2 implying an increase in impors growh wice higher han he increase in domesic income. Also in Table 7, we replicae he income-elasiciy of he demand for impors (π*) compaible wih he equilibrium in he balance-of-paymens (on curren accoun) for an easier comparison of he resuls. The McCombie es is performed for each of he overlapping periods and he absolue value of he -es is displayed. [Inser Table 7 around here] Comparing he esimaed income-elasiciy of demand for impors π derived from he 2SLS regressions o π*, he hypohesis ha gy BP is a good predicor of gy (ha is π = π * ) is never rejeced a he 5% significance level (he Law is rejeced hree imes only, a he 10% significance level: 1974-1988, 1985-1999 and 1987-2001). Thus, by implemening he McCombie es our evidence shows ha Thirlwall s Law is accurae for predicing acual growh in Porugal, for he period considered. Also relevan is he fac ha in his able he acual growh rae is always higher han ha compaible wih he balance-of-paymens equilibrium (excep for he las overlapping period), indicaing he exisence of exernal deficis. This evidence is now much more in conformiy wih he negaive averages of he curren accoun (as a percenage of GDP) found for each overlapping sage, as repored in Table 6. The approach based on he overlapping periods esimaion of he income-elasiciy of he demand of impors is apparenly more appropriae o analyse Thirlwall s Law, insead of considering a single π esimaed for he global period and hen using i o compue he balance-of-paymens equilibrium growh rae eiher for he oal period or 16
for each of he overlapping sages. Wih he approach from Table 7, he McCombie es enables us o analyse he performance of Thirlwall s Law period by period. The general conclusion is ha he Law predics accuraely acual growh of he Poruguese economy giving suppor o he balance-of-paymens consrain hypohesis. 6. Conclusions The presen sudy analyses wheher he demand-orienaed approach based on he balance-of-paymens consrain hypohesis is suiable for explaining he Poruguese growh in he 1965-2008 period. The model developed by Thirlwall o compue he balance-of-paymens equilibrium growh rae is adoped, assuming consan relaive prices in he long-run (a plausible hypohesis) and iniial equilibrium on curren accoun. The impors and expors demand funcions are esimaed o obain he incomeelasiciies wih respec o impors and expors, which are crucial parameers for compuing Thirlwall s Law. A preliminary daa analysis shows ha Porugal grew on average a a higher rae han he OECD counries in he whole period, 1965-2008, and his is consisen wih he empirical finding ha he income-elasiciy of he demand for expors is higher han ha of impors, as Thirlwall s Law implies. This corroboraes he hypohesis ha a counry can grow faser han he res of he world only when is income-elasiciy of he demand for expors exceeds ha of impors, unless capial inflows can compensae exernal imbalances. I is also observed ha Porugal grew faser in he pre han in he pos-adhesion period o he EU and his is consisen wih higher curren accoun deficis accumulaed in he laer as a resul of boh higher impors growh and lower expors growh. The crucial parameer of he income-elasiciy of he demand for impors is obained by esimaing he impors demand funcion by 2SLS, wih domesic income growh being endogenous. Knowing ha parameer, he balance-of-paymens equilibrium growh raes are compued for a series of 15-year overlapping periods and are compared o he acual growh raes. The approximaion of he wo raes is quie close, validaing Thirlwall s Law as a good insrumen for predicing acual growh of he Poruguese economy. The McCombie es reinforces his conclusion. Generally i is found ha Porugal grew slighly higher han he rae consisen wih he balance-of-paymens 17
equilibrium, and his is consisen wih he accumulaion of curren accoun deficis over he period considered. Alernaively he income-elasiciy of he demand of impors is esimaed individually for each overlapping period and i is observed a sharp increase of is value afer Porugal joining he EU. Assuming ha he income-elasiciy of impors is changing over ime he confirmaion of Thirlwall s Law becomes more saisfacory. When he McCombie es is performed, i shows ha he acual growh in Porugal can be accuraely prediced by he balance-of-paymens equilibrium growh approach in almos all he overlapping periods. The overall analysis shows ha exernal demand consrains are crucial for explaining he growh performance of he Poruguese economy, especially in he pos-adhesion period. For he counry o achieve susainable growh raes expors mus increase and impors sensiiviy o domesic income changes mus be reduced, urning he economy more compeiive boh in domesic and foreign markes and his is compaible wih he increase in he balance-of-paymens equilibrium growh rae. A he micro level, policies are needed o improve he non-price characerisics of he goods and services associaed o qualiy, design, innovaion, produc differeniaion, markeing and efficien disribuion. 18
Appendix o gx Annual growh rae of real expors. Expors of goods and services a 1995 (2000) prices (naional currency; annual percenage change). o gm Annual growh rae of real impors. Impors of goods and services a 1995 (2000) prices (naional currency; annual percenage change). o gy Annual growh rae of real GDP. GDP a 1995 (2000) marke prices (naional currency; annual percenage change). o gc Annual growh rae of real privae consumpion. Privae final consumpion expendiure a 1995 (2000) prices (naional currency; annual percenage change). o gi Annual growh rae of real invesmen. Gross fixed capial formaion a 1995 (2000) prices (naional currency; annual percenage change). o gpx Annual growh rae of expor prices. Price deflaor expors of goods and services (naional currency; annual percenage change). o gpm Annual growh rae of impor prices. Price deflaor impors of goods and services (naional currency; annual percenage change). o gpy Annual growh rae of domesic prices. Price deflaor GDP a marke prices (naional currency; annual percenage change). o grpm Annual growh rae of he relaive price of impors (gpy-gpm). o grpx Annual growh rae of he relaive price of expors (gpx-gpm). o ca Balance on curren ransacions wih he res of he world (%of GDP a marke prices). o gz Annual growh rae of real foreign income (OECD counries). Compued by he auhors from 1965 o 1994. 1965 1970: GDP a he price levels and exchange raes of 1990 (billions of US dollars) OECD (1997) 1971 1994: GDP a he price levels and exchange raes of 2000 (billions of US dollars) OECD (2006) 1995-2008: Real GDP (% change from previous year) OECD(2009) Noes: Daa on gx, gm, gy, gc, gi, gpx, gpm, gpy and ca was aken from European Commission (2002, 2009). Consan figures are a 1995 prices (for 1965-1980) and 2000 prices (for 1981-2008), depending on he Saisical Annex from which hey were obained (2002 and 2009, respecively). 19
Table 1. Descripive saisics of variables Variable Period Obs Mean Sd. Deviaion Min Max (1) gm % Impors (2) gx % Expors (3) gy % Domesic income (4) gz % Foreign income (5) gc % Consumpion (6) gi % Invesmen (7) grpm % Relaive price of impors (8) grpx % Relaive price of expors (9) ca Curren accoun 1965-2008 44 6.53 7.78-24.2 23.1 1965-1985 21 5.37 9.03-24.2 14.6 1986-2008 23 7.60 6.46-3.3 23.1 1965-2008 44 6.05 7.82-16.4 33.0 1965-1985 21 6.65 10.66-16.4 33.0 1986-2008 23 5.51 3.95-3.3 12.2 1965-2008 44 3.58 3.24-4.3 11.2 1965-1985 21 4.39 3.85-4.3 11.2 1986-2008 23 2.84 2.41-2.0 7.5 1965-2008 44 3.19 1.52 0.1 6.3 1965-1985 21 3.70 1.83 0.1 6.3 1986-2008 23 2.72 0.99 0.8 4.6 1965-2008 44 3.38 3.20-2.9 13.0 1965-1985 21 3.66 4.18-2.9 13.0 1986-2008 23 3.12 2.00-0.1 6.9 1965-2008 44 3.78 7.87-17.4 18.0 1965-1985 21 3.21 9.28-17.4 17.9 1986-2008 23 4.29 6.49-7.4 18.0 1965-2008 44 1.14 7.76-24.9 27.3 1965-1985 21-1.55 7.95-24.9 8.7 1986-2008 23 3.60 6.85-5.5 27.3 1965-2008 44 0.10 3.99-12.9 11.3 1965-1985 21-1.04 4.74-12.9 6.1 1986-2008 23 1.14 2.88-3.2 11.3 1965-2008 44-4.18 5.03-13.5 5.5 1965-1985 21-2.73 5.59-13.5 5.5 1986-2008 23-5.51 4.14-11.9 2.1 Noes: Variables (1) o (8) are annual growh raes. Variable (9) is curren accoun as a percenage of GDP a marke prices. Char 1. Annual growh rae of expors (gx) and impors (gm), 1965-2008 40.0 30.0 20.0 % 10.0 0.0-10.0-20.0-30.0 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 gx gm 20
Char 2. Annual growh rae of domesic (gy) and foreign income (gz), 1965-2008 % 12 10 8 6 4 2 0-2 -4-6 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 gy gz Char 3. Annual growh rae of relaive price of impors (grpm) and expors (grpx), 1965-2008 30 20 10 % 0-10 -20-30 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 grpm grpx Char 4. Evoluion of acual growh rae (gy) and he curren accoun (ca), 1965-2008 15 10 5 % 0-5 -10-15 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 gy ca 21
Table 2. Uni roo ess, 1965-2008 Z() 1% criical value Variable ADF PP ADF PP gm -3.959*** -4.750*** -3.648-3.628 gx -5.388*** -4.468*** -3.634-3.628 gy -4.517*** -4.602*** -4.242-4.214 gz -5.116*** -4.634*** -4.224-4.214 gc -3.876*** -3.887*** -3.628-3.628 gi -5.416*** -3.421*** -3.634-2.631 grpm -4.781*** -4.694*** -2.631-2.631 grpx -5.016*** -4.213*** -2.633-2.631 Noes: ADF - Augmened Dickey-Fuller es for uni roo Regression wihou consan and rend for: grpm and grpx. Regression wih consan and wih no rend for: gm, gx, gc and gi. Regression wih consan and rend for: gy and gz. 0 lags for gc and grpm, 1 lag for gx, gz, gi and grpx, 3 lags for gm and gy. PP - Phillips-Perron es for uni roo Regression wihou consan and rend for: gi, grpm and grpx. Regression wih consan and wih no rend for: gm, gx and gc. Regression wih consan and rend for: gy and gz. 3 Newey-Wes lags. *** Coefficien significan a he 1% significance level. Criical values are provided by Saa. Table 3. Correlaion marix, 1965-2008 gm gx gy gz gc gi grpm grpx ca gm 1.0000 gx 0.4823 1.0000 gy 0.7657 0.4333 1.0000 gz 0.4451 0.5671 0.6568 1.0000 gc 0.5574 0.6153 1.0000 gi 0.6998 0.3488 0.6282 0.3613 0.3686 1.0000 grpm 1.0000 grpx 0.4833 0.3204 0.6138 1.0000 ca 0.3995 0.3776 0.5367 0.5832 0.4328 0.3091 0.2991 1.0000 Noes: n 2 The significance of he correlaion coefficien is given by he following -es: = r *, where n is 2 1 r he number of observaions and r he correlaion coefficien. The null hypohesis is ha he correlaion beween a pair of variables is null in he populaion. The correlaions displayed are saisically significan a he 5% significance level. 22
Table 4. Esimaion resuls from he expors and impors demand funcions OLS SUR 2SLS Variable Expors Impors Expors Impors Impors gz 2.5736*** 2.7690*** (3.81) (4.37) grpx (-1) 0.4730* 0.4263* (1.86) (1.79) gy 1.5590*** 1.5280*** 2.1483*** (7.04) (7.35) (7.22) grpm (-1) 0.3639*** 0.3670*** 0.2881*** (3.99) (4.30) (2.84) Consan -2.2876 0.4612-2.8944 0.5653-1.4985 (-0.98) (0.46) (-1.32) (0.59) (-1.22) Obs 43 43 43 43 R-squared 0.36 0.70 0.79 F (2,40) 11.42 45.98 44.12 (0.0001) (0.0000) (0.0000) BP es of error independence across equaions χ 2 1 = 3. 890 (0.0486) Pagan-Hall heeroscedasiciy es χ 2 4 = 6. 5 (0.1648) Cumby-Huizinga auocorrelaion 2 1 0.0611 es (0.8048) Anderson canon. corr. LM saisic χ 2 3 = 28. 076 (0.0000) Cragg-Donald Wald F- saisic # ( 3,38) = 23. 83 F (0.0000) Endogeneiy es χ 2 1 = 14. 305 (0.0002) Sargan saisic χ 2 2 = 0. 908 (0.6351) Noes: Numbers in parenhesis are -raio (for esimaed coefficiens) and p-values (for ess). *** Coefficien significan a he 1% level. ** Coefficien significan a he 5% level. * Coefficien significan a he 10% level. #A saisic higher han 10 indicaes no weak insrumens and consequenly, no biases in he coefficiens. Table 5. Evidence from Thirlwall s Law gy gx π gy BP =gx/π (gy BP -gy) ca 1965-2008 3.58 6.05 2.15 2.82-0.76-4.18 1965-1985 4.39 6.65 2.22 2.99-1.40-2.73 1986-2008 2.84 5.51 2.66 2.07-0.77-5.51 Noes: gy, gx and ca were aken from Table 1. π 1965-2008 was aken from he 2SLS esimaion in Table 4. π 1965-1985 and π 1986-2008 come from he 2SLS regressions for he corresponding sub-periods (he resuls are available upon reques). In he 1965-1985 period, here is a problem wih he endogeneiy of gy and he insrumens are weak. For 1986-2008, gc is redundan, bu is removal worsens he global resuls. Bearing in mind hese limiaions, he main purpose is o have a global view of he performance of he mos imporan variables before and afer he adhesion o he EU. 23
Table 6. Acual and balance-of-paymens equilibrium growh raes, 15-year overlapping periods. Period ca gy gx gy BP (gy BP - gy) π* 1965-1979 -0.96 5.55 7.01 3.26-2.28 1.26 1966-1980 -1.33 5.35 6.26 2.91-2.43 1.17 1967-1981 -2.19 5.19 5.11 2.38-2.81 0.98 1968-1982 -3.34 4.79 4.87 2.27-2.52 1.02 1969-1983 -3.99 4.17 5.81 2.71-1.46 1.40 1970-1984 -4.46 3.81 6.01 2.80-1.02 1.58 1971-1985 -4.56 3.49 6.09 2.84-0.66 1.74 1972-1986 -4.59 3.33 5.75 2.68-0.65 1.73 1973-1987 -4.93 3.22 5.15 2.40-0.82 1.60 1974-1988 -5.31 2.97 5.09 2.37-0.61 1.71 1975-1989 -4.90 3.33 6.79 3.16-0.17 2.04 1976-1990 -4.60 3.88 8.51 3.96 0.08 2.19 1977-1991 -4.20 3.71 8.65 4.02 0.31 2.33 1978-1992 -3.73 3.42 8.59 4.00 0.58 2.51 1979-1993 -3.49 3.10 7.76 3.61 0.51 2.50 1980-1994 -3.63 2.79 6.12 2.85 0.06 2.19 1981-1995 -3.43 2.77 6.56 3.05 0.28 2.37 1982-1996 -2.89 2.91 7.23 3.37 0.46 2.49 1983-1997 -2.41 3.05 7.33 3.41 0.36 2.40 1984-1998 -2.35 3.39 6.99 3.25-0.13 2.06 1985-1999 -2.71 3.77 6.41 2.99-0.78 1.70 1986-2000 -3.45 3.84 6.53 3.04-0.80 1.70 1987-2001 -4.29 3.70 6.19 2.88-0.82 1.67 1988-2002 -4.87 3.33 5.55 2.58-0.74 1.67 1989-2003 -5.13 2.77 5.26 2.45-0.32 1.90 1990-2004 -5.64 2.45 4.71 2.19-0.25 1.93 1991-2005 -6.23 2.24 4.21 1.96-0.28 1.88 1992-2006 -6.79 2.04 4.71 2.19 0.15 2.31 1993-2007 -7.28 2.09 5.00 2.33 0.23 2.39 1994-2008 -7.93 2.23 5.19 2.41 0.19 2.33 Average 1.89 Noes: ca Curren accoun as % of GDP a marke prices gy Annual growh rae of real GDP gx Annual growh rae of real expors gy BP - balance-of-paymens equilibrium growh rae, given by gx gy BP = πˆ gx π * = - income-elasiciy of demand for impors assuming a balanced curren accoun gy 24
Char 5. Evoluion of he esimaed income-elasiciy of demand for impors in he overlapping periods 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.00 1965-1979 1967-1981 1969-1983 1971-1985 1973-1987 1975-1989 1977-1991 1979-1993 1981-1995 1983-1997 1985-1999 1987-2001 1989-2003 1991-2005 1993-2007 25
Table 7. Balance-of-paymens equilibrium growh raes wih income-elasiciy of impors varying over he 15-year overlapping periods. gy (gy - gy) Period π (1) π* Abs. value of he -es (2) BP BP 1965-1979 2.70 1.26 1.31 2.60-2.95 1966-1980 2.67 1.17 1.35 2.34-3.01 1967-1981 2.53 0.98 1.55 2.02-3.17 1968-1982 2.49 1.02 1.50 1.96-2.84 1969-1983 2.43 1.40 1.64 2.39-1.78 1970-1984 2.10 1.58 1.02 2.85-0.96 1971-1985 2.13 1.74 0.74 2.86-0.63 1972-1986 1.93 1.73 0.33 2.98-0.34 1973-1987 2.01 1.60 0.67 2.57-0.65 1974-1988 2.96 1.71 1.80* 1.72-1.25 1975-1989 3.04 2.04 1.37 2.24-1.09 1976-1990 3.22 2.19 1.10 2.64-1.24 1977-1991 2.89 2.33 0.89 2.99-0.72 1978-1992 2.73 2.51 0.35 3.15-0.27 1979-1993 2.74 2.50 0.48 2.84-0.26 1980-1994 2.77 2.19 1.08 2.21-0.59 1981-1995 2.76 2.37 0.70 2.38-0.40 1982-1996 2.72 2.49 0.42 2.65-0.25 1983-1997 2.74 2.40 0.60 2.68-0.37 1984-1998 2.68 2.06 1.11 2.61-0.78 1985-1999 2.99 1.70 1.81* 2.14-1.62 1986-2000 2.96 1.70 1.71 2.21-1.63 1987-2001 2.85 1.67 1.78* 2.17-1.53 1988-2002 3.18 1.67 1.67 1.75-1.58 1989-2003 3.05 1.90 1.25 1.73-1.05 1990-2004 3.03 1.93 1.38 1.56-0.89 1991-2005 2.68 1.88 1.20 1.57-0.67 1992-2006 2.75 2.31 0.68 1.72-0.32 1993-2007 2.44 2.39 0.10 2.05-0.05 1994-2008 2.30 2.33 0.06 2.26 0.03 Noes: (1) The esimaed coefficien from he 2SLS regression is always saisically significan. The Anderson canonical correlaion LM saisic indicaes ha he equaion is underidenified, for he wo firs sub-periods. (2) The null hypohesis is ha π = π *, for each overlapping period. π = π * always for a 5% significance level. * denoes ha π π *, for a 10% significance level. 26
References Adkins, L.; Hill, R. (2008) Using Saa for principles of economerics, John Wiley and Sons (3 rd ediion). AlDakhil, K. (1998) A mehod for esimaing simulaneous equaions models wih ime-series and cross-secion daa, Journal of King Saud Universiy - Adminisraive Sciences, 10, 1, 13-28. Alonso, José A.; Garcimarín, C. (1998-99) A new approach o Balance-of-paymens consrain: some empirical evidence, Journal of Pos Keynesian Economics, Winer, 21, 2, 259-282. Aesoglu, H. (1993) Balance-of-paymens-consrained growh evidence from he Unied Saes, Journal of Pos Keynesian Economics, Summer, 15, 4, 507-514. Aesoglu, H. (1995) An explanaion of he slowdown in US economic growh, Applied Economic Leers, 2, 91-94. Bairam, E. (1988) Balance-of-paymens, he Harrod foreign rade muliplier and economic growh: he European and Norh American experience, 1970-1985, Applied Economics, 20, 1635-1642. Bairam, E. (1993) Saic versus dynamic specificaions and he Harrod foreign rade muliplier, Applied Economics, 25, 739-742. Bairam, E. (1997) Levels of economic developmen and appropriae specificaion of he Harrod foreign-rade muliplier, Journal of Pos Keynesian Economics, Spring, 19, 3, 337-344. Baum, C. (2006) An inroducion o modern economerics using Saa, College Saion, Saa Press. Baum, C.; Schaffer, M.; Sillman, S. (2003) Insrumenal variables and GMM: esimaion and esing, Boson College-Deparmen of Economics Working Paper 545, February. Baum, C.; Schaffer, M.; Sillman, S. (2007) Enhanced rouines for insrumenal variables/gmm esimaion and esing, Boson College-Deparmen of Economics Working Paper 667, Sepember. European Commission (2002) Saisical Annex of European Economy, Direcorae General Economic and Financial Affairs, Spring. European Commission (2009) Saisical Annex of European Economy, Direcorae General Economic and Financial Affairs, Spring. Léon-Ledesma, M. (1999) An applicaion of Thirlwall s law o he Spanish economy, Journal of Pos Keynesian Economics, Spring, 21-3, 431-439. McCombie, J.S.L. (1989) Thirlwall s Law and balance-of-paymens consrained growh a commen on he debae, Applied Economics, 21, 611-629. McCombie, J.S.L. (1997) On he empirics of balance-of-paymens-consrained growh, Journal of Pos Keynesian Economics, Spring, 19, 3, 345-375. McGregor, P.; Swales, J. (1985) Professor Thirlwall and he balance-of-paymens consrained growh, Applied Economics, 17, 17-32. 27
McGregor, P.; Swales, J. (1991) Thirlwall s Law and balance-of-paymens consrained growh: furher commen on he debae, Applied Economics, 23, 9-23. Organizaion for Economic Co-operaion and Developmen - OECD (1997) Naional Accouns Main Aggregaes (vol.i) 1960-1995. Organizaion for Economic Co-operaion and Developmen - OECD (2006) Saisical Compendium: Naional Accouns and Hisorical Saisics 1970-2004. Organizaion for Economic Co-operaion and Developmen - OECD (2009) Economic Oulook no. 85 Annex Tables, June. Thirlwall, A. (1979) The balance-of-paymens consrain as an explanaion of inernaional growh rae differences, Banca Nazionale del Lavoro, 128, 45-53. Thirlwall, A. (1982) The Harrod rade muliplier and he imporance of expor-led growh, Pakisan Journal of Applied Economics, 1, 1, 1-21. Thirlwall, A. (1986) Balance-of-paymens consrained growh: a reply o McGregor and Swales, Applied Economics, 18, 1259-1263. Thirlwall, A.; Hussain, N. (1982) The balance-of-paymens consrain, capial flows and growh rae differences beween developing counries, Oxford Economic Papers, 34, 498-510. 28
ESTUDOS DO G.E.M.F. (Available on-line a hp://gemf.fe.uc.p) 2009-13 How well he balance-of- paymens consrain approach explains he Poruguese growh performance. Empirical evidence for he 1965-2008 period - Micaela Anunes & Elias Soukiazis 2009-12 Aypical Work: Who Ges I, and Where Does I Lead? Some U.S. Evidence Using he NLSY79 - John T. Addison, Chad Coi & Chrisopher J. Surfield 2009-11 The PIGS, does he Group Exis? An empirical macroeconomic analysis based on he Okun Law - João Sousa Andrade 2009-10 A Políica Moneária do BCE. Uma esraégia original para a esabilidade nominal - João Sousa Andrade 2009-09 Wage Dispersion in a Parially Unionized Labor Force - John T. Addison, Ralph W. Bailey & W. Sanley Sieber 2009-08 Employmen and exchange raes: he role of openness and echnology - Fernando Alexandre, Pedro Bação, João Cerejeira & Miguel Porela 2009-07 Channels of ransmission of inequaliy o growh: A survey of he heory and evidence from a Poruguese perspecive - Adelaide Duare & Mara Simões 2009-06 No Deep Pockes: Some sylized resuls on firms' financial consrains - Filipe Silva & Carlos Carreira 2009-05 Aggregae and secor-specific exchange rae indexes for he Poruguese economy - Fernando Alexandre, Pedro Bação, João Cerejeira & Miguel Porela 2009-04 Ren Seeking a Plan Level: An Applicaion of he Card-De La Rica Tenure Model o Workers in German Works Councils - John T. Addison, Paulino Teixeira & Thomas Zwick 2009-03 Unobserved Worker Abiliy, Firm Heerogeneiy, and he Reurns o Schooling and Training - Ana Sofia Lopes & Paulino Teixeira 2009-02 Worker Direcors: A German Produc ha Didn Expor? - John T. Addison & Claus Schnabel 2009-01 Fiscal and Moneary Policies in a Keynesian Sock-flow Consisen Model - Edwin Le Heron 2008-08 Uniform Price Marke and Behaviour Paern: Wha does he Iberian Elecriciy Marke Poin Ou - Víor Marques, Isabel Soares & Adelino Forunao 2008-07 The parial adjusmen facors of FTSE 100 sock index and sock index fuures: The informaional impac of elecronic rading sysems - Helder M. C. V. Sebasião 2008-06 Waer Losses and Hydrographical Regions Influence on he Cos Srucure of he Poruguese Waer Indusry - Ria Marins, Fernando Coelho& Adelino Forunao 2008-05 The Shadow of Deah: Analysing he Pre-Exi Produciviy of Poruguese Manufacuring Firms - Carlos Carreira & Paulino Teixeira 2008-04 A Noe on he Deerminans and Consequences of Ousourcing Using German Daa - John T. Addison, Luz Bellmann, André Pahnke & Paulino Teixeira 2008-03 Exchange Rae and Ineres Rae Volailiy in a Targe Zone: The Poruguese Case - Anónio Porugal Duare, João Sousa Andrade & Adelaide Duare 2008-02 Taylor-ype rules versus opimal policy in a Markov-swiching economy - Fernando Alexandre, Pedro Bação & Vasco Gabriel
Esudos do GEMF 2008-01 Enry and exi as a source of aggregae produciviy growh in wo alernaive echnological regimes - Carlos Carreira & Paulino Teixeira 2007-09 Opimal moneary policy wih a regime-swiching exchange rae in a forward-looking model - Fernando Alexandre, Pedro Bação & John Driffill 2007-08 Esruura económica, inensidade energéica e emissões de CO2: Uma abordagem Inpu-Oupu - Luís Cruz & Eduardo Baraa 2007-07 The Sabiliy and Growh Pac, Fiscal Policy Insiuions, and Sabilizaion in Europe - Carlos Fonseca Marinheiro 2007-06 The Consumpion-Wealh Raio Under Asymmeric Adjusmen - Vasco J. Gabriel, Fernando Alexandre & Pedro Bação 2007-05 European Inegraion and Exernal Susainabiliy of he European Union An applicaion of he hesis of Feldsein-Horioka - João Sousa Andrade 2007-04 Uma Aplicação da Lei de Okun em Porugal - João Sousa Andrade 2007-03 Educaion and growh: an indusry-level analysis of he Poruguese manufacuring secor - Mara Simões & Adelaide Duare 2007-02 Levels of educaion, growh and policy complemenariies - Mara Simões & Adelaide Duare 2007-01 Inernal and Exernal Resrucuring over he Cycle: A Firm-Based Analysis of Gross Flows and Produciviy Growh in Porugal - Carlos Carreira & Paulino Teixeira 2006-09 Cos Srucure of he Poruguese Waer Indusry: a Cubic Cos Funcion Applicaion - Ria Marins, Adelino Forunao & Fernando Coelho 2006-08 The Impac of Works Councils on Wages - John T. Addison, Paulino Teixeira & Thomas Zwick 2006-07 Ricardian Equivalence, Twin Deficis, and he Feldsein-Horioka puzzle in Egyp - Carlos Fonseca Marinheiro 2006-06 L inégraion des marchés financiers - José Soares da Fonseca 2006-05 The Inegraion of European Sock Markes and Marke Timing - José Soares da Fonseca 2006-04 Mobilidade do Capial e Susenabilidade Exerna uma aplicação da ese de F-H a Porugal (1910-2004) - João Sousa Andrade 2006-03 Works Councils, Labor Produciviy and Plan Heerogeneiy: Firs Evidence from Quanile Regressions - Joachim Wagner, Thorsen Schank, Claus Schnabel & John T. Addison 2006-02 Does he Qualiy of Indusrial Relaions Maer for he Macroeconomy? A Cross-Counry Analysis Using Srikes Daa - John T. Addison & Paulino Teixeira 2006-01 Mone Carlo Esimaion of Projec Volailiy for Real Opions Analysis - Pedro Manuel Coresão Godinho 2005-17 On he Sabiliy of he Wealh Effec - Fernando Alexandre, Pedro Bação & Vasco J. Gabriel 2005-16 Building Blocks in he Economics of Mandaes - John T. Addison, C. R. Barre & W. S. Sieber
Esudos do GEMF 2005-15 Horizonal Differeniaion and he survival of Train and Coach modes in medium range passenger ranspor, a welfare analysis comprising economies of scope and scale - Adelino Forunao & Daniel Mura 2005-14 Aypical Work and Compensaion - John T. Addison & Chrisopher J. Surfield 2005-13 The Demand for Labor: An Analysis Using Mached Employer-Employee Daa from he German LIAB. Will he High Unskilled Worker Own-Wage Elasiciy Please Sand Up? - John T. Addison, Luz Bellmann, Thorsen Schank & Paulino Teixeira 2005-12 Works Councils in he Producion Process - John T. Addison, Thorsen Schank, Claus Schnabel & Joachim Wagnerd 2005-11 Second Order Filer Disribuion Approximaions for Financial Time Series wih Exreme Ouliers - J. Q. Smih & Anónio A. F. Sanos 2005-10 Firm Growh and Persisence of Chance: Evidence from Poruguese Microdaa - Blandina Oliveira & Adelino Forunao 2005-09 Residenial waer demand under block raes a Poruguese case sudy - Ria Marins & Adelino Forunao 2005-08 Poliico-Economic Causes of Labor Regulaion in he Unied Saes: Alliances and Raising Rivals Coss (and Someimes Lowering One s Own) - John T. Addison 2005-07 Firm Growh and Liquidiy Consrains: A Dynamic Analysis - Blandina Oliveira & Adelino Forunao 2005-06 The Effec of Works Councils on Employmen Change - John T. Addison & Paulino Teixeira 2005-05 Le Rôle de la Consommaion Publique dans la Croissance: le cas de l'union Européenne - João Sousa Andrade, Maria Adelaide Silva Duare & Claude Berhomieu 2005-04 The Dynamics of he Growh of Firms: Evidence from he Services Secor - Blandina Oliveira & Adelino Forunao 2005-03 The Deerminans of Firm Performance: Unions, Works Councils, and Employee Involvemen/High Performance Work Pracices - John T. Addison 2005-02 Has he Sabiliy and Growh Pac sabilised? Evidence from a panel of 12 European counries and some implicaions for he reform of he Pac - Carlos Fonseca Marinheiro 2005-01 Susainabiliy of Poruguese Fiscal Policy in Hisorical Perspecive - Carlos Fonseca Marinheiro 2004-03 Human capial, mechanisms of echnological diffusion and he role of echnological shocks in he speed of diffusion. Evidence from a panel of Medierranean counries - Maria Adelaide Duare & Mara Simões 2004-02 Wha Have We Learned Abou The Employmen Effecs of Severance Pay? Furher Ieraions of Lazear e al. - John T. Addison & Paulino Teixeira 2004-01 How he Gold Sandard Funcioned in Porugal: an analysis of some macroeconomic aspecs - Anónio Porugal Duare & João Sousa Andrade
Esudos do GEMF 2003-07 Tesing Gibra s Law: Empirical Evidence from a Panel of Poruguese Manufacuring Firms - Blandina Oliveira & Adelino Forunao 2003-06 Régimes Monéaires e Théorie Quaniaive du Produi Nominal au Porugal (1854-1998) - João Sousa Andrade 2003-05 Causas do Araso na Esabilização da Inflação: Abordagem Teórica e Empírica - Víor Casro 2003-04 The Effecs of Households and Firms Borrowing Consrains on Economic Growh - Maria da Conceição Cosa Pereira 2003-03 Second Order Filer Disribuion Approximaions for Financial Time Series wih Exreme Ouliers - J. Q. Smih & Anónio A. F. Sanos 2003-02 Oupu Smoohing in EMU and OECD: Can We Forego Governmen Conribuion? A risk sharing approach - Carlos Fonseca Marinheiro 2003-01 Um modelo VAR para uma Avaliação Macroeconómica de Efeios da Inegração Europeia da Economia Poruguesa - João Sousa Andrade 2002-08 Discriminaion des faceurs poeniels de croissance e ype de convergence de l économie porugaise dans l UE à ravers la spécificaion de la foncion de producion macroéconomique. Une éude appliquée de données de panel e de séries emporelles - Mara Simões & Maria Adelaide Duare 2002-07 Privaisaion in Porugal: employee owners or jus happy employees? -Luís Moura Ramos & Ria Marins 2002-06 The Poruguese Money Marke: An analysis of he daily session - Fáima Teresa Sol Mura 2002-05 As eorias de ciclo políicos e o caso poruguês - Rodrigo Marins 2002-04 Fundos de acções inernacionais: uma avaliação de desempenho - Nuno M. Silva 2002-03 The consisency of opimal policy rules in sochasic raional expecaions models - David Backus & John Driffill 2002-02 The erm srucure of he spreads beween Poruguese and German ineres raes during sage II of EMU - José Soares da Fonseca 2002-01 O processo desinflacionisa poruguês: análise de alguns cusos e benefícios - Anónio Porugal Duare 2001-14 Equiy prices and moneary policy: an overview wih an exploraory model - Fernando Alexandre & Pedro Bação 2001-13 A convergência das axas de juro poruguesas para os níveis europeus durane a segunda meade da década de novena - José Soares da Fonseca 2001-12 Le rôle de l invesissemen dans l éducaion sur la croissance selon différenes spécificaions du capial humain. - Adelaide Duare & Mara Simões
Esudos do GEMF 2001-11 Ricardian Equivalence: An Empirical Applicaion o he Poruguese Economy - Carlos Fonseca Marinheiro 2001-10 A Especificação da Função de Produção Macro-Económica em Esudos de Crescimeno Económico. - Maria Adelaide Duare e Mara Simões 2001-09 Eficácia da Análise Técnica no Mercado Accionisa Poruguês - Nuno Silva 2001-08 The Risk Premiums in he Poruguese Treasury Bills Ineres Raes: Esimaion by a coinegraion mehod - José Soares da Fonseca 2001-07 Principais facores de crescimeno da economia poruguesa no espaço europeu - Maria Adelaide Duare e Mara Simões 2001-06 Inflaion Targeing and Exchange Rae Co-ordinaion - Fernando Alexandre, John Driffill e Fabio Spagnolo 2001-05 Labour Marke Transiion in Porugal, Spain, and Poland: A Comparaive Perspecive - Paulino Teixeira 2001-04 Paridade do Poder de Compra e das Taxas de Juro: Um esudo aplicado a rês países da UEM - Anónio Porugal Duare 2001-03 Technology, Employmen and Wages - John T. Addison & Paulino Teixeira 2001-02 Human capial invesmen hrough educaion and economic growh. A panel daa analysis based on a group of Lain American counries - Maria Adelaide Duare & Mara Simões 2001-01 Risk Premiums in he Poruguese Treasury Bills Ineres Raes from 1990 o 1998. An ARCH-M Approach - José Soares da Fonseca 2000-08 Idenificação de Vecores de Coinegração: Análise de Alguns Exemplos - Pedro Miguel Avelino Bação 2000-07 Imunização e M-quadrado: Que relação? - Jorge Cunha 2000-06 Eficiência Informacional nos Fuuros Lisbor 3M - Nuno M. Silva 2000-05 Esimaion of Defaul Probabiliies Using Incomplee Conracs Daa - J. Sanos Silva & J. Mureira 2000-04 Un Essaie d'applicaion de la Théorie Quaniaive de la Monnaie à l économie porugaise, 1854-1998 - João Sousa Andrade 2000-03 Le Taux de Chômage Naurel comme un Indicaeur de Poliique Economique? Une applicaion à l économie porugaise - Adelaide Duare & João Sousa Andrade 2000-02 La Convergence Réelle Selon la Théorie de la Croissance: Quelles Explicaions pour l'union Européenne? - Mara Crisina Nunes Simões 2000-01 Políica de Esabilização e Independência dos Bancos Cenrais - João Sousa Andrade
Esudos do GEMF 1999-09 Noa sobre a Esimação de Vecores de Coinegração com os Programas CATS in RATS, PCFIML e EVIEWS - Pedro Miguel Avelino Bação 1999-08 A Aberura do Mercado de Telecomunicações Celulares ao Terceiro Operador: Uma Decisão Racional? - Carlos Carreira 1999-07 Is Porugal Really so Arerioscleroic? Resuls from a Cross-Counry Analysis of Labour Adjusmen - John T. Addison & Paulino Teixeira 1999-06 The Effec of Dismissals Proecion on Employmen: More on a Vexed Theme - John T. Addison, Paulino Teixeira e Jean-Luc Grosso 1999-05 A Coberura Esáica e Dinâmica aravés do Conrao de Fuuros PSI-20. Esimação das Rácios e Eficácia Ex Pos e Ex Ane - Helder Miguel C. V. Sebasião 1999-04 Mobilização de Poupança, Financiameno e Inernacionalização de Careiras - João Sousa Andrade 1999-03 Naural Resources and Environmen - Adelaide Duare 1999-02 L'Analyse Posiive de la Poliique Monéaire - Chisian Aubin 1999-01 Economias de Escala e de Gama nos Hospiais Públicos Porugueses: Uma Aplicação da Função de Cuso Variável Translog - Carlos Carreira 1998-11 Equilíbrio Moneário no Longo e Curo Prazos - Uma Aplicação à Economia Poruguesa - João Sousa Andrade 1998-10 Algumas Observações Sobre o Méodo da Economia - João Sousa Andrade 1998-09 Mudança Tecnológica na Indúsria Transformadora: Que Tipo de Viés Afinal? - Paulino Teixeira 1998-08 Porfolio Insurance and Bond Managemen in a Vasicek's Term Srucure of Ineres Raes - José Albero Soares da Fonseca 1998-07 Financial Innovaion and Money Demand in Porugal: A Preliminary Sudy - Pedro Miguel Avelino Bação 1998-06 The Sabiliy Pac and Poruguese Fiscal Policy: he Applicaion of a VAR Model - Carlos Fonseca Marinheiro 1998-05 A Moeda Única e o Processo de Difusão da Base Moneária - José Albero Soares da Fonseca 1998-04 La Srucure par Termes e la Volailié des Taux d'inérê LISBOR - José Albero Soares da Fonseca 1998-03 Regras de Comporameno e Reformas Moneárias no Novo SMI - João Sousa Andrade 1998-02 Um Esudo da Flexibilidade dos Salários: o Caso Espanhol e Poruguês - Adelaide Duare e João Sousa Andrade
Esudos do GEMF 1998-01 Moeda Única e Inernacionalização: Apresenação do Tema - João Sousa Andrade 1997-09 Inovação e Aplicações Financeiras em Porugal - Pedro Miguel Avelino Bação 1997-08 Esudo do Efeio Liquidez Aplicado à Economia Poruguesa - João Sousa Andrade 1997-07 An Inroducion o Condiional Expecaions and Saionariy - Rui Manuel de Almeida 1997-06 Definição de Moeda e Efeio Berlusconi - João Sousa Andrade 1997-05 A Esimação do Risco na Escolha dos Porafólios: Uma Visão Seleciva - Anónio Albero Ferreira dos Sanos 1997-04 A Previsão Não Paramérica de Taxas de Renabilidade - Pedro Manuel Coresão Godinho 1997-03 Propriedades Assimpóicas de Densidades - Rui Manuel de Almeida 1997-02 Co-Inegraion and VAR Analysis of he Term Srucure of Ineres Raes: an empirical sudy of he Poruguese money and bond markes -João Sousa Andrade & José Soares da Fonseca 1997-01 Reparição e Capialização. Duas Modalidades Complemenares de Financiameno das Reformas - Maria Clara Mureira 1996-08 A Crise e o Ressurgimeno do Sisema Moneário Europeu - Luis Manuel de Aguiar Dias 1996-07 Housing Shorage and Housing Invesmen in Porugal a Preliminary View - Víor Neves 1996-06 Housing, Morgage Finance and he Briish Economy - Kenneh Gibb & Nile Isephan 1996-05 The Social Policy of The European Communiy, Reporing Informaion o Employees, a U.K. perspecive: Hisorical Analysis and Prognosis - Ken Shackleon 1996-04 O Teorema da Equivalência Ricardiana: aplicação à economia poruguesa - Carlos Fonseca Marinheiro 1996-03 O Teorema da Equivalência Ricardiana: discussão eórica - Carlos Fonseca Marinheiro 1996-02 As axas de juro no MMI e a Resrição das Reservas Obrigaórias dos Bancos - Fáima Assunção Sol e José Albero Soares da Fonseca 1996-01 Uma Análise de Curo Prazo do Consumo, do Produo e dos Salários - João Sousa Andrade