Online Publication Date: 10 March, 2012 Publisher: Asian Economic and Social Society

Transcription

1 Onlne Publcaon Dae: 10 March, 2012 Publsher: Asan Economc and Socal Socey Economc Growh and Income Inequaly: Emprcal Evdence from Norh Afrcan Counres Zouher ABIDA (Unversy of Sfax, Faculy of Economcs and Managemen of Sfax - Tunsa) Imen Mohamed SGHAIER (Unversy of Sfax, Faculy of Economcs and Managemen of Sfax Tunsa) Caon: Zouher ABIDA, Imen Mohamed SGHAIER (2012): Economc Growh and Income Inequaly: Emprcal Evdence from Norh Afrcan Counres Asan Economc and Fnancal Revew Vol.2,No.1, pp

2 Asan Economc and Fnancal Revew, 2(1),pp Economc Growh and Income Inequaly: Emprcal Evdence from Norh Afrcan Counres Absrac Auhor (s) Zouher ABIDA Unversy of Sfax, Faculy of Economcs and Managemen of Sfax Tunsa. Emal: Imen Mohamed SGHAIER Unversy of Sfax, Faculy of Economcs and Managemen of Sfax Tunsa. Emal: Ths paper examnes he emprcal relaonshp beween economc growh and ncome nequaly for 3 counres of Norh Afrca (Tunsa, Morocco and Egyp) over he perod The resuls of hs paper ndcae ha he long-run growh elascy of ncome nequaly s negave and sgnfcan mplyng ha keepng oher facors consan; more ncome nequaly reduces economc growh. Moreover, hs paper fnds evdence ha more physcal and human capal nvesmen and hgher openness o rade have sascally sgnfcan mpac on enhancng economc growh and reducng povery. Keywords: Economc Growh, Income Inequaly, Panel conegraon JEL Classfcaon: O4, I3, O15 Inroducon Economc growh s consdered o be a powerful force for reducng povery. Hgh and susaned economc growh ncreases he labor demand and wages whch n reurn wll reduce povery. Smlarly, beer earnngs as a resul of reducon n povery lead o ncrease producvy and growh. Bu he exen of povery reducon as a resul of economc growh depends on how he dsrbuon of ncome changes wh economc growh and on nal Inequales n ncome. If ncome nequaly ncreases, hen economc growh does no lead o a sgnfcan povery reducon. A large number of emprcal sudes have aemped o explore he relaonshp beween ncome nequaly and economc growh. 1 Bu here are only few sudes ha dscuss he role of cred marke mperfecons n growh nequaly relaonshp. Mos of earler sudes ha hghlgh he role of cred marke mperfecons n growh nequaly relaonshp used Ordnary Leas Squares o esmae he cross-counry growh regresson, whch has a problem of omed varable bas. Secondly, due o lmed avalably of comparable nequaly sascs, sample selecon remaned a problem n mos of earler sudes. The resulng esmaes of mos of hese sudes found a negave coeffcen on nequaly suggesed ha counres wh a more equal ncome dsrbuon (ha s a lower Gn ndex) end o have hgher levels of ncome. 2 No counry has acheved rapd economc growh by closng hemselves off o nernaonal rade. Trade openness s defned as he degree o whch foregners and domesc czen can ransac whou governmen mposed coss ha are leved on a ransacon beween hem. For example, arff, non arff 1 Ravallon (1997), Dennger and Squre (1996, 1998), Barro (2000), ec. 2 Persson and Tabelln (1994), Aghon and Bolon (1997), Galor (2000) ec. 142

3 Economc Growh and Income Inequaly.. barrers, local conen requremens, nspecon delays rase he cos of buyng from abroad. Despe of havng conssen emphass on how rade promoes growh, he heory also suggesed n he presence of dsorons lke cred marke mperfecons, polcal nsably, less mproved nfra srucure ec., free rade mgh no be bes for growh. For nsance, a hgh real reurn o capal n unsklled labor abundan counres explos her comparave advanage. Even f rade openness leads o more rapd growh, does no necessarly mply ha s an effecve nsrumen for reducng povery. If a growh sraegy based on rade openness leads o a sgnfcan worsenng of ncome nequaly of households, does no lead o sgnfcan reducon n povery. How rade affecs ncome dsrbuon of a counry s purely an emprcal queson. Ths paper also consders he role of rade openness, physcal and human capal nvesmen and governmen spendng n enhancng economc growh and reducng nequales. Ths sudy uses panel conegraon mehods and mproved daa on ncome nequaly o assess he possble seady-sae relaonshp beween ncome nequaly and economc growh for 3 counres of Norh Afrca (Tunsa, Morocco and Egyp) over he perod Ths arcle s organzed as follows. Secon 2 presens he general heores descrbng he causal relaonshp from ncome nequaly o economc growh. Secon 3 presens he daa and repors he resuls of panel un roo and conegraon ess. Esmaon deals and resuls are gven n Secon 4 and Secon 5 concludes. Economc Growh and Income Inequaly: Theory and Evdence Income dsrbuon s a conroversal opc n macroeconomcs. In he early 1955, Kuznes (1955) suded he relaonshp beween he economc growh and ncome dsrbuon and nroduced he famous nvered-u shape relaonshp beween nequaly and ncome, whch saes ha he dsrbuon of ncome frs becomes hen, many heorecal works and emprcal works have been rased o rees Kuznes hypohess. Kaldor (1957) and Pasne (1962) assumed ha ncome nequaly affeced economc growh hrough he savngnvesmen mechansm. Because ndvduals wh dfferen ncome levels wll choose dfferen savngs rae, he ncome nequaly leads he agen o ncrease he savngs and nvesmen, whch n urn ncreases he growh rae. Dennger and Squre (1996) usng he daa for 108 counres over he perod found no sysemac relaonshp beween growh and changes n aggregae nequaly. Accordng o her analyss, perods of aggregae growh were assocaed wh ncreased nequaly n foryhree cases and wh a decrease n nequaly n fory-fve cases. Smlarly, perods of economc declne were assocaed wh ncreased nequaly n fve cases and wh a more equable dsrbuon of ncome n wo cases. The smple relaonshp beween curren as well as lagged ncome growh and he change n he Gn coeffcen s nsgnfcan for he whole sample as well as for sub samples defned n erms of counry characerscs lke rch or poor, equal or unequal, fas-growng or slowgrowng economes, suggesng no srong relaonshp beween growh and changes n aggregae nequaly. Dennger and Squre (1998) use new crosscounry daa on ncome and asse (land) dsrbuon o show ha () here s a srong negave relaonshp beween nal nequaly n he asse dsrbuon and long-erm growh, () nequaly reduces ncome growh for he poor, bu no for he rch, and () avalable longudnal daa provde lle suppor for he Kuznes hypohess. Polces ha ncrease aggregae nvesmen and faclae acquson of asses by he poor mgh hus be doubly benefcal for growh and povery reducon Forbes (2000) found posve effecs of ncome nequaly on growh. The auhor argued ha counry-specfc effecs and omed varables are he cause of a sgnfcan negave bas n he esmaons of he effecs of nequaly on growh. She also concluded ha fxed-effec esmaons yeld he conssen resul of a 143

4 Asan Economc and Fnancal Revew, 2(1),pp posve shor and medum erm correlaon beween nequaly and growh. Smh (2001), examned emprcally wo hypoheses subssence consumpon and cred marke mperfecons of specfc channels of nequaly o affec prvae savng raes, he found ha here s economerc evdence ha especally a low per capa ncome levels, ncome nequaly may be assocaed wh hgher aggregae savng. Garbs (2005) examnes he emprcal relaonshp beween nequaly and growh, and analyzes he mpacs of growh, nequaly, and governmen spendng on povery reducon. A panel daase for 82 counres for he perod has been assembled wh he daa averaged over perods of hree o seven years, dependng on he avalably of nequaly and povery daa. The emprcal resuls challenge he belef ha ncome nequaly has a negave effec on growh and confrm he valdy of he Kuznes curve. Cred marke mperfecons n low and medum-ncome counres are denfed as he lkely reason for he posve lnk beween nequaly and growh over he shor o medum erm. In he long erm, nequaly may have an adverse mpac on growh. In a neoclasscal growh framework wh a ypcal polcal-economy mechansm, Yn and al. (2006) reexamnes he relaonshp beween he ncome nequaly and economc growh by nroducng governmen spendng no he producon funcon and he uly funcon. I demonsraes ha Kuznes famous nvered-u shape relaonshp beween nequaly and economc growh wll hold - he growh rae wll be frs ncreasng wh he ncome nequaly before he growh rae decreases wh nequaly. Penalosa and Turnovsky (2006) develop an endogenous growh model wh elasc labor supply, n whch agens dffer n her nal endowmens of physcal capal. In hs framework, he growh rae and he dsrbuon of ncome are jonly deermned. The key equlbrang varable s he equlbrum labor supply. I deermnes he rae of reurn o capal, whch n urn affecs boh he rae of capal accumulaon and he dsrbuon of ncome across agens. Then hey examne he mpac of varous srucural shocks on growh and dsrbuon. They found ha faser growh s assocaed wh a more unequal, conemporaneous dsrbuon of ncome, conssen wh recen emprcal fndngs. The resuls could be summarzed n hree pons. Frs, nal nequaly n he dsrbuon of land appears o be assocaed wh lower subsequen growh. Second, here s no suppor for a redsrbuve medan-voer based explanaon of nal nequaly's effec on growh. Thrd, mperfecons n fnancal markes for cred appear o be more relevan for nvesmen n human capal raher han physcal capal. However, daa on land nequaly was very lmed and could no be used n he panel daa model o check f cross seconal resuls hold afer conrollng for omed varable bas. Role of Cred Marke Imperfecons The ncome approach emphaszes he effec of ncome nequaly on savngs and on physcal capal accumulaon. Cred marke mperfecons approach consders he effec of ncome nequaly on he accumulaon of human capal (Galor and Zera, 1993). In a model by Galor and Moav (2004), he engne of economc growh changes from physcal capal o physcal and human capal n he process of economc developmen. The process of economc developmen s dvded no wo regmes, whch have her own seady-sae growh pahs. Economes n he frs regme are underdeveloped, aggregae physcal capal s small, and he rae of reurn o human capal s lower han he rae of reurn o physcal capal (Galor and Moav, 2004). There are wo ypes of ndvduals n he economy: hose who own he physcal capal (he rch) and hose who do no (he poor). The poor consume her enre ncome (wages) and are no engaged n savng and on capal accumulaon. Thus, here s emporary seady-sae equlbrum where he poor are n povery rap and he rch ge rcher. Inequaly ncreases aggregae savngs by ncreasng he ncome of he rch and greaer 144

5 Economc Growh and Income Inequaly.. aggregae savngs fuel physcal capal accumulaon. 3 In he second regme, physcal capal accumulaon by he rch has ncreased he rae of reurn o human capal so hgh ha nduces human capal accumulaon (Galor and Moav, 2004). In hs regme, boh human and physcal capals are engnes for economc developmen. Snce ndvduals nvesmen n human capal s subjeced o dmnshng margnal reurns, he reurn o human capal nvesmens s maxmzed when nvesmen n human capal s wdely spread among he populaon. Because access o cred s consraned, human capal nvesmen s maxmzed when ncome n he economy s dsrbued evenly. However, n a ceran phase of economc developmen ncome of every ndvdual becomes so hgh ha credconsrans become less bndng. In hs locally sable seady-sae equlbrum, he effec of nequaly on growh becomes less sgnfcan. Openness o Trade, Economc Growh and Income Inequaly The dea ha rade lberalzaon has an mpac on he counry's growh s no new and goes back a leas o Adam Smh. New classcal model based on consan and decreasng reurns o scale as n Solow (1956) and Swan (1956) predced ha a counry would have sac gans from lowerng s rade barrers. Mos of he recen sudes ncludng Dollar and Davd (1992), Edwards (1993), Sachs and Warner (1995) and Dollar and Kraay (2001) have found a posve assocaon beween rade lberalzaon and growh. There are number of channels hrough whch rade promoes growh raes by allocang he resources more effcenly. Trade promoes growh by encouragng economes o specalze and produce n areas where hey have relave cos advanage over oher economes. Overme, hs helps economes o employ more of her human, physcal and capal resources n secors where hey ge reurns n open nernaonal 3 In modern less developed economes, s possble ha also human capal drves growh, f he capal and skllbased echnology s mpored. In hs case, he effec of nequaly on growh would be mxed or negave (Galor and Moav, 2004). markes, boosng producvy and reurns o workers. Trade also expands he markes ha local producers can access, allowng hem o produce a mos effcen scale o keep down he coss. Trade dsperses new echnologes and deas, ncreasng he producvy of local workers and managers. Technology ransfers hrough rade are also more valuable for developng counres, whch employ less advance echnologes and have lle capacy o develop new echnologes hemselves. Removng rade barrers e.g. arff on mpors gves consumers access o cheaper producs, ncreasng her Purchasng power and lvng sandard. I also provdes producers an access o cheap npus, reducng coss and boosng her compeveness. Dollar and Kraay (2001) usng daa on rade lberalzaon as a share of GDP n consan prces for 101 counres ncludng 73 developng counres found ha rade lberalzaon leads o faser growh n average ncomes, and ha hs growh n average ncomes n urn ncreases he ncomes of he poor proporonaely. The poor counres ha have reduced rade barrers and parcpaed more n nernaonal rade over he pas weny years have seen her growh raes accelerae. In he 1990s hey grew far more rapdly han he rch counres, and hence reduced he gap beween hemselves and he developed world. A he same me he developng counres ha are no parcpang n globalzaon are fallng furher and furher behnd. Whn he globalzng developng counres here has been no general rend n nequaly. Thus, rapd growh has ranslaed no dramac declnes n absolue povery n counres such as Chna, Inda, Thaland, and Venam. OLS esmaon resuls showed ha per capa GDP growh n he pos-1980 globalzers acceleraed from 1.4 percen a year n he 1960s and 2.9 percen a year n he 1970s o 3.5 percen n he 1980s and 5.0 percen n he 1990s. Framework of Analyss and Esmaon Technque Framework of Analyss There are dfferen channels hrough whch 145

6 Asan Economc and Fnancal Revew, 2(1),pp ncome nequaly affecs growh raes. Kaldor (1957) suggess ha margnal propensy o save of he rch s hgher han ha of he poor, mplyng ha ha a hgher degree of nequaly wll yeld hgher aggregae savngs, hgher capal accumulaon and growh. In conras, Persson and Tabelln (1994) and Alsena and Rodrck (1994) emphasze he four man channels hrough whch ncome nequaly lowers growh raes. Frs, he mpac of nequaly on encouragng ren - seekng acves ha reduce he secury of propery rghs, second, unequal socees face more dffcules n collecve acon possbly refleced n polcal nsably, a propensy for populs redsrbuve polces, or greaer volaly n polces - all of whch can lower growh, hrd, he medan voer n a more unequal socey s relavely poorer and favors a hgher (and hus more neffcen) ax burden, fourh, o he exen ha nequaly n ncome or asses coexss wh mperfec cred markes, poorer people may be unable o nves n her human and physcal capal, wh adverse consequences for long-run growh. Galor and Zera (1993) argued ha ncome dsrbuon plays an mporan role n he deermnaon of aggregae economc acvy and economc growh. In conras o he represenave agen approach ha has domnaed he feld of macroeconomcs for several decades, hs sudy analyzes he role of heerogeney n he deermnaon of macroeconomc behavor. The research demonsraed ha n he presence of capal markes mperfecons and local nonconvexes n he producon of human capal, ncome dsrbuon affecs aggregae oupu n he long run as well as n he shor-run. The research developed he hypohess ha equaly n suffcenly wealhy economes smulaes nvesmen n human capal and n (ndvdual specfc projecs) and enhances economc growh, whereas nequaly promoes growh n a suffcenly poor economes, a predcon ha was confrmed by nal emprcal sudes. Galor's (2000) argues ha he classcal approach holds a low-ncome levels bu no a laer sages of developmen. In he early sage of developmen, nequaly would promoe growh because physcal capal s scarce a hs sage and s accumulaon requres savng. Inequaly n ncome would hen resul n hgher savngs and rapd growh. In laer sages of economc developmen, however, as he reurn o human capal ncreases owng o capal-skll complemenarly, human capal becomes he man engne of growh. Cred consrans, however, become less bndng as wages ncrease, and he adverse effec of ncome nequaly on human capal accumulaon subsdes, and hus he effec of nequaly on he growh process becomes nsgnfcan. The unfed approach complemens he research of Galor and Wel (1999, 2000) who developed unfed models ha encompasses he ranson beween hree dsnc regmes ha have characerzed he process of economc developmen: he Malhusan Regme, he Pos- Malhusan Regme, and he Modern Growh Regme, focusng on he hsorcal evoluon of he relaonshp beween populaon growh, echnologcal change, and economc growh. Galor and Moav (1999) argue ha nequaly has a posve effec on capal accumulaon bu negave effec on human capal accumulaon n he presence of cred consrans. In he early sages of developmen physcal capal s scarce, he rae of reurn o human capal s lower han he reurn on physcal capal and he process of furher developmen s drven manly by capal accumulaon. In he early sages of developmen, he posve effec of nequaly on aggregae savng more han offses he negave effec on nvesmen n human capal and, snce he margnal propensy o save s an ncreasng funcon of he ndvdual's wealh, nequaly ncreases aggregae savngs and capal accumulaon, enhancng he process of developmen. In he laer sages of developmen, however, he posve effec of nequaly on savng s offse by he negave effec on nvesmen n human capal. Based on heorecal leraure on economc nequales and some oher poenal facors ha deermne economc growh, we develop he followng model, whch s also n lnes wh Garbs (2005). 146

7 Economc Growh and Income Inequaly.. GR INV 3 GINI TRADE 5 Where: 1 SCH 4 (1) y 2 1 GR = Growh rae of GDP per capa, GINI = Gn ndex n he curren perod, y = Inal GDP per capa, 1 INV, = Gross fxed capal formaon as a % of GDP, SCH, = Secondary school enrolmen rae, TRADE = Sum of expors and mpors of goods and servces as a % of GDP, represens unobserved counry-specfc facors and s a perod specfc effec. The me-specfc effec,, allows o conrol for nernaonal condons ha change over me and affec he growh performance of counres n he sample, whle accouns for unobserved counry-specfc facors ha boh drve growh and are poenally correlaed wh he explanaory varables. represens he dsurbance erm. Tme Seres Analyss of Panel Daa The heorecal models presened above predc seady-sae equlbrum relaons, or saonary dsrbuons, ha may exs beween ncome nequaly and he evoluon of oupu. The esmaon of hese heorecal saonary dsrbuons requres ha we know he me seres feaures of he varables n he model. Many models also assume ha ncome dsrbuon and economc developmen are deermned endogenously, whch has o be aken no accoun n he esmaon. 4 Daa We use Gn coeffcen o measure ncome nequaly, whch s one of he mos popular represenaons of ncome nequaly. I s based on Lorenz Curve, whch plos he share of populaon agans he share of ncome receved and has a mnmum value of 0 (case of perfec equaly) and maxmum value of 1 (perfec nequaly). Mssng values n Income nequaly daa are he major problem n cross counry analyss. Many of developng counres have only one or wo observaons. Therefore, we expanded he exsng daabase by ncludng he comparable daa on povery and nequaly from recen household surveys ncluded n World Bank, IMF Saff repors and Povery Reducon Sraegy Papers. The human capal ndcaor measured by secondary school enrolmen rae (SCH) and he oher ndependen varables - he GDP per capa, he physcal capal (INV), he openness rae (TRADE) and he growh rae of GDP per capa (GR) - are aken from he World Bank Developmen Indcaors (WDI, 2009). Un Roo Tesng To es for he presence of un roos on panel daa, we use he Im, Pesaran and Shn (2003) - IPS hereafer. IPS usng he lkelhood framework, sugges a new more flexble and compuaonally smple un roo esng procedure for panels (whch s referred as bar sasc), ha allows for smulaneous saonary and non-saonary seres. Moreover, hs es allows for resdual seral correlaon and heerogeney of he dynamcs and error varances across groups. The IPS es s based on he esmaon of he followng equaon: 4 Bénabou (2005) has acually suggesed ha endogeney of ncome nequaly n growh regressons s he prmary reason for he observed conroversy n emprcal growh sudes. 147

8 Asan Economc and Fnancal Revew, 2(1),pp y p y 1 m, d m, j y j j 1, 1,..., T, 1,..., N (2) where T s he number of observaons over me, N denoes he number of ndvdual members n he panel and d, conans m deermnsc varables. The null hypohess s defned as H : 0 for all 1,..., N 0 and he alernave hypohess s H : 0 1,..., N and 0 for 1 a for N1 1,..., N, wh 0 N1 N ha allows for some (bu no all) of ndvdual seres o have un roos. IPS (2003) compue separae un roo es for he N cross-secon uns and defne her sasc as a smple average of he bar ndvdual ADF sascs, T, for he null as: N T 1 bar ( 1/ N). IPS (2003) argues ha are..d. and have fne mean and T varance. Therefore, he sandardzed bar N, T sasc converges o a sandard normal dsrbuon as N under he null hypohess. In order o propose a sandardzaon of he bar N, T sasc, he values of he mean and he varance have been compued va Mone Carlo mehods for dfferen values of T and p s and abulaed by IPS (2003). The resuls of each one of our fve varables are repored n able 1, where all he ess have a un roo under he null hypohess. As ndcaed n able1, he ess of panel un roo of accordng o IPS (2003) confrm ha all varables are no saonary n levels bu saonary n frs dfferences. We now es for he exsence of a long-run relaonshp beween he ncome nequaly and economc growh. Conegraon Tess The possble conegraon beween nequaly and GDP s esed wh panel conegraon es developed by Pedron (1999, 2004). Pedron proposes a resdual-based es for he null of conegraon for dynamc panels wh mulple regressors n whch he shor-run dynamcs and he long-run slope coeffcens are permed o be heerogeneous across ndvduals. The es allows for ndvdual heerogeneous fxed effecs and rend erms and no exogeney requremens are mposed on he regressors on he conegrang regressons. Specally, he ess ask for he resduals esmaon from sac conegrang long-run relaon for a me seres panel of observables y : y x x... k, k, e 1, 1, x 2, 2,, 1,..., T ; 1,..., N (3) where as usual T s he number of observaons over me and N s he number of uns n he panel. I s possble o nerpre he model (3) as N dfferen equaons, each of whch has K regressors. The varables y and x are assumed o be I(1), for each member of he panel, and under he null of no conegraon he resdual e wll also be I(1). and are scalars denong fxed effecs and un-specfc lnear rend parameers, respecvely and are he conegraon slopes; noe ha all hs coeffcens are permed o vary across ndvduals, so ha consderable heerogeney s allowed by hs specfcaon. Pedron consders he use of seven resdualbased panel conegraon sascs, four based on poolng he daa along he whn-dmenson (denoed panel conegraon sascs ) and hree based on poolng along he beweendmenson (denoed group mean conegraon sascs ). Anoher dsncon beween he wo ses of es s based on he alernave hypohess specfcaon. In fac, even f boh ses of es 148

9 Economc Growh and Income Inequaly.. verfy he null hypohess of no conegraon: H : 1 where s 0 he auoregressve coeffcen of esmaed resduals under he alernave hypohess ˆ e v ), alernave hypohess ( e ˆ 1 specfcaon s dfferen: - he panel conegraon sascs mpose a common coeffcen under he alernave w hypohess whch resuls: H a : 1, - he group mean conegraon sascs allow for heerogeneous coeffcens under he alernave hypohess and resuls: H : 1. b a I s sraghforward o observe ha he frs caegory of four sascs ncludes a ype of non - paramerc varance rao sasc, a panel verson of a non-paramerc Phllps and Perron (1988) -sasc, a non-paramerc form of he average of he Phllps and Perron - sasc and an ADF ype -sasc. The second caegory of panel conegraon sascs s based on a group mean approach and ncludes a Phllps and Perron ype - sasc, a Phllps and Perron ype -sasc and an ADF ype -sasc. The comparave advanage of each of hese sascs wll depend on he underlyng daa-generang process. Afer he calculaon of he panel conegraon es sascs he approprae mean and varance adjusmen erms are appled, so ha he es sascs are asympocally sandard normally dsrbued: N, T where N N(0,1) v N, T s one of he seven sascs of Pedron and v are he funcons of momens of he underlyng Brownan moon funconal. The approprae mean and varance adjusmen erms for dfferen number of regressors and dfferen panel conegraon es sascs are gven n Table 2 n Pedron (1999). 5 Pedron (2004) explored fne sample performances of he seven sascs. He showed ha n erms of power all he proposed sascs do farly well for T > 100. Moreover Pedron s (1996) smulaons showed ha for small me span (T < 20), he beween dmenson (group - sasc) s he mos powerful.gven our relavely shor me span (T = 29), we wll pay a parcular aenon o he group paramerc- sasc ( ADF sa ) when esng for conegraon. The resul of panel conegraon ess are dsplayed n able 2. Snce smulaons made by Pedron (2004) show ha, n small samples, he group-mean paramerc-es s more powerful han he oher ess, we can conclude ha he null hypohess of no conegraon s rejeced n our sudy, and now urn o he esmaon of he long run relaonshp beween he ncome nequaly and economc growh. Esmaon of he Conegrang Coeffcen of Inequaly Our esmaon echnque addresses ssues of endogeney and unobserved counry characerscs. Therefore, o accoun for endogeney and counry-specfc unobserved characerscs, we use he Sysem GMM dynamc panel esmaon mehod. The opon o use Sysem GMM s based on he argumen ha he exsence of weak nsrumens mples asympocally ha he varance of he coeffcen ncreases and n small samples he coeffcens can be based. To reduce he poenal bas and naccuracy assocaed wh he use of Dfference GMM (Arellano and Bond, 1991), Arellano and Bover (1995) and Blundell and Bond (1998) develop a sysem of regressons n dfferences and levels. The 5 Ths able conans he mean and varance values for he cases when here s no heerogeneous nercep, or when here s a heerogeneous nercep or/and a me rend n he heerogeneous regresson equaon. k s he number of regressors whou akng he heerogeneous deermnsc erms no accoun. 149

10 Asan Economc and Fnancal Revew, 2(1),pp nsrumens for he regresson n dfferences are he lagged levels of he explanaory varables and he nsrumens for he regresson n levels are he lagged dfferences of explanaory varables. These are consdered as approprae nsrumens under he assumpon ha alhough here may be correlaon beween he levels of explanaory varables and he counry specfc effec, here s no correlaon beween hose varables n dfferences and he counry specfc effec. The conssency of he Sysem GMM esmaor s assessed by wo specfcaon ess. The Sargan es of over denfyng resrcons ess he overall valdy of he nsrumens. Falure o rejec he null hypohess gves suppor o he model. The second es examnes he null hypohess ha he error erm s no serally correlaed. Agan, falure o rejec he null hypohess gves suppor o he model. In Table 3 we repor our regresson esmaes usng he Sysem GMM esmaon echnque. Before we descrbe our resuls, we should menon ha he specfcaon ess - boh he Sargan es of over-denfyng resrcons and he es for hgher order correlaon - valdae our regressons for nference. Tha s, our nsrumens are no correlaed wh he error erm and he laer does no dsplay hgher order seral correlaon. The panel regresson resuls regardng growh nequaly relaonshp gven n Table 3 confrms he negave relaonshp beween growh and nequaly n Norh Afrca economes. The conegrang coeffcen of nequaly s negave and sascally sgnfcan a he 5% level when panel GMM esmaor s used. Wh respec o he dfferen emprcal sudes (Alesna and Rodrck (1994), Garbs (2005) and Yn and al. (2006)), we fnd ha ncome nequaly has a subsanal negave mpac on economc growh. On he oher hand, an ncrease of he Gn coeffcen level (say, 10%) would mply a growh declne of approxmaely 0.65 percenage pons. We nroduce he level of nal GDP per capa (he naural logarhm) as ndependen varable accordng o he condonal convergence hypohess. The nal GDP per capa coeffcen s negave, meanng ha he condonal convergence hypohess s evdenced: holdng consan oher growh deermnans, counres wh lower GDP per capa end o grow faser. The nal poson of he economy s hus a sgnfcan deermnan of growh, as recognsed by he neoclasscal heory. The nal ncome has a negave effec on economc growh coheren o he heorecal sudy and sascally sgnfcan a a 1% level. The esmaed coeffcen of secondary school enrolmen s posve and sascally sgnfcan a a 10% level. Ths resul s conssen wh common fndngs of heorecal leraure ha suggess a posve relaon beween human capal and economc growh and of emprcal leraure (Romer (1990), Mankw and al. (1992) and Benhabb and Spegel (1994)). The nvesmen varable has also he rgh sgn snce here found a posve relaonshp beween capal accumulaon and growh. Trade openness also posvely affecs growh. Thus, he more counres are ouward-orened he more hs conrbues favorably o economc growh. These resuls are n lne wh hose found by Agure and Calderon (2005) and Dufréno and al. (2009), and, more generally wh he neoclasscal approach accordng o whch he posve mpac of rade on growh s explaned by comparave advanages, be hey n resource endowmen or dfferences n echnology. Concluson Ths sudy aemps o examne he emprcal relaonshp beween growh and ncome nequaly for 3 counres of Norh Afrca over he perod The resuls of hs paper clearly ndcae ha he long-run growh elascy of ncome nequaly s negave and sgnfcan when panel GMM esmaor s used. The resuls also show negave and hghly sgnfcan relaonshp beween growh and nal ncome per capa. Physcal capal nvesmen has posve effec on economc growh. The resuls also sugges ha 150

11 Economc Growh and Income Inequaly.. coeffcens of openness o rade and human capal nvesmen are posve and robusly sgnfcan ndcang ha boh facors have srong mpac on economc growh. A pro-poor economc growh leadng o a rapd and susanable povery reducon depends upon he neracon of a wde range of polcy measures. Frs, a pro-poor growh sraegy does no have o only focus on economc growh, bu could also be combned wh an acve polcy of ncome redsrbuon. Then, he hgher he level of boh physcal and human capal nvesmen, he hgher s he level of oupu per capa. A beer-educaed labor force can mprove producvy and echnologcal level n he Table 1. Panel Un Roo Tess of IPS Varables n levels economy, whch have a long-run posve effec on economc growh. Fnally, governmens mus creae an envronmen ha s conducve o growh. Macroeconomc polcy should am a sably, and openness owards he res of he world. For all hese effors o be effecve, he governmen mus develop good nsuons, and provde good governance. Varables n frs dfferences Consan Consan wh rend consan Consan wh rend Per capa GDP *** -3.15*** Gn ndex *** -3.75*** Trade *** -4.52*** Invesmen *** -5.25*** Secondary School Enrol *** -4.12*** Noes: * (resp.**,***): rejecon of he null hypohess a 10% (resp. 5%, 1%) sgnfcance level. Lags seleced accordng o he SIC wh a maxmum lag lengh of 3. Table 2. Pedron s Panel Conegraon Tess Tes Sasc p-values Panel conegraon ess v sa rho sa PP sa ADF sa -2.36*** Group mean conegraon ess 151

12 Asan Economc and Fnancal Revew, 2(1),pp rho sa -4.25*** PP sa ADF sa -2.12** Noes: *(resp.**,***): rejecon of he null hypohess a he 10% (resp. 5%, 1%) sgnfcance level. Lags seleced accordng o he SIC wh a maxmum lag lengh of 3. Table 3. GMM Esmaes of he Conegrang Coeffcen of Inequaly Dependan varable : Growh rae of GDP per capa Inal GDP per capa (-3.2)*** Inequaly (Gn ndex) (-2.11)** Trade (0.44) Invesmen (1.88)* Secondary School Enrolmen (1.79)* Consan (0.51) Observaons 105 Specfcaon Tess (p-values) - Sargan Tes nd order correclaon 0.55 Noes: -sa n parenheses. *, ** and *** ndcaes sgnfcance a 10%, 5% and 1% respecvely. References Aghon, P., and P. Bolon (1997) Dsrbuon and Economc Growh n Model of Imperfec Capal Markes, European Economc Revew, Vol.36, pp Alesna, A., and D. Rodrck (1994) Dsrbuve Polcs and Economc Growh, Quarerly Journal of Economcs, Vol.108, pp Agurre, A. and C. Calderon (2005) Real Exchange Rae Msalgnmens and Economc Performance, Cenral Bank of Chle Workng Paper, n 315. Arellano, M., and S. Bond (1991) Some Tess of Specfcaon for Panel Daa: Mone 152

13 Economc Growh and Income Inequaly.. Carlo Evdence and an Applcaon o Employmen Evdence, Revew of Economc Sudes, Vol.58, No.2, pp Arellano, M. and O. Bover (1995) Anoher Look a he Insrumenal Varables Esmaon of Error Componens Models, Journal of Economercs, Vol.68, No.1, pp Benhabb, J. and M. Spegel (1994) The Role of Human Capal n Economc Developmen Evdence for Aggregae Cross- Counry Daa, Journal of moneary Economcs Vol.34, pp Balag B. H. (2001) Economerc Analyss of Panel Daa, Chcheser: UK: Wley and Sons. Barro, R. (2000) Inequaly and Growh n a Panel of Counres, Journal of Economc Growh Vol.5, pp Bénabou, R. (2005) Inequaly, echnology, and he socal conrac. In: Aghon P, Durlauf S (eds) Handbook of economc growh. Elsever, Amserdam, pp Blundell, R., and S. Bond (1998) Inal Condons and Momen Resrcons n Dynamc Developmen, Journal of Economercs Vol.87, No.1, pp Dennger, K., and L. Squre (1996) A New Daa Se Measurng Income Inequaly. World Bank Economc Revew, Vol.10, pp Dennger, K., and L. Squre (1998) New Ways of Lookng a Old Issues: Inequaly and Growh, Journal of Developmen Economcs, Vol.57, pp Dollar, D., and A. Davd (1992) Ouward- Orened Developng Economes Really Do Grow More Rapdly: Evdence from 95 LDCs , Economc Developmen and Culural Change Vol.40, No.3, pp Dufréno, G., V. Mgnon, and C. Tsangardes (2009) The Trade Growh Nexus n he Developng Counres: A Quanle Regresson Approach, CEPII Workng Paper, n 4. Dollar, D., and A. Kraay (2001) Trade, Growh, and Povery, Fnance and Developmen Vol.38, pp Edwards, S. (1993) Openness, Trade Lberalzaon, and Growh n Developng Counres, Journal of Economc Leraure Vol.31, pp Forbes, K. (2000) A Reassessmen of he Relaonshp beween Inequaly and Growh, Amercan Economc Revew Vol.90, pp Galor, O. (2000) Income Dsrbuon and he Process of Developmen, European Economc Revew Vol.44, pp Galor, O., and A. Wel (1999) From Malhusan Sagnaon o Modern Growh, Amercan Economc Revew Vol.89, pp Galor, O., and A. Wel (2000) Populaon, Technology and Growh: From Malhusan Sagnaon o Growh and Beyond, Amercan Economc Revew Vol.90, pp Galor, O., and J. Zera (1993) Income Dsrbuon and Macroeconomcs, Revew of Economc Sudes Vol.60, pp Galor, O., and O. Moav (1999) From Physcal o Human Capal Accumulaon: Inequaly n he Process of Developmen, CEPR Dscusson Paper, n Galor, O., and O. Moav (2004) From Physcal o human capal accumulaon: nequaly and he process of developmen, Revew of Economc Sudes Vol.71, No.4, pp Garbs, I., (2005) Inequaly, Povery and Growh: Cross Counry Evdence, IMF Workng Paper, n 28. Im, K. S., M. H. Pesaran, and Y. Shn (2003) Tesng for Un Roo n Heerogeneous Panel, Journal of Economercs Vol.115, No.1, pp Kaldor, N. (1957) A Model of Economc Growh, Economc Journal Vol.57, pp Kuznes, S. (1955) Economc Growh and Income Inequaly Amercan Economc Revew Vol.45, No.1, pp Mankw, N., D. Romer, and D. Wel (1992) A conrbuon o he emprcs of economc growh Quarerly Journal of Economcs Vol.152, pp Pasne L. (1962) Rae of prof and ncome dsrbuon n relaon o he rae of economc growh, Revew of Economc Sudes Vol.29, pp Penalosa, C.C., and S.J. Turnovsky (2006) Growh and ncome nequaly: a canoncal model, Economc Theory Vol.28, pp Persson, T., and G. Tabelln (1994) Is Inequaly Harmful for Growh?, Amercan Economc Revew Vol.84, pp Pedron P. (1996) Fully modfed OLS for heerogeneous conegraed panels and he case of purchasng power pary, Indana 153

14 Asan Economc and Fnancal Revew, 2(1),pp Unversy, Workng Paper n Economcs, n 20. Pedron P. (1999) Crcal for Conegraon Tess n Heerogeneous Panels wh Mulple Regressors, Oxford Bullen of Economcs and Sascs Vol.61, No.1, pp Pedron P. (2004) Panel Conegraon Asympoc and Fne Sample Properes of Pooled Tme Seres Tess wh an Applcaon o he PPP Hypohess, Economerc Theory Vol.20, No.3, pp Phllps, P.C.B., and P. Perron (1988) Tesng for a Un Roo n Tme Seres Regresson, Bomerka Vol.75, pp Romer, P. M (1990) Endegenous Technologcal Change, The Journal of Polcal Economy Vol. 98, No.5, pp Ravallon, M. (1997) Good and Bad Growh: The Human Developmen Repors, World Developmen Vol.25, pp Sachs, J.D., and A.M. Warner (1995) Economc Reform and he Process of Global Inegraon, Brookngs Papers on Economc Acvy, pp Solow, R. (1956) A Conrbuon o he Theory of Economc Growh, Quarerly Journal of Economcs Vol.70, pp Swan, T. (1956) Economc Growh and Capal Accumulaon, Economc Record Vol.32, pp Smh, D. (2001) Inernaonal Evdence on How Income Inequaly and Cred Marke Imperfecons Affec Prvae Savng Raes, Journal of Developmen Economcs Vol.64, pp World Bank (2009) World Developmen Indcaors Daabase. Yn, H., L. Gong, and H. Zou (2006) Income nequaly and economc growh - he Kuznes curve revsed, Emprcal Economcs, Vol.1, No.2, pp