Net migration and convergence in Portugal. An alternative analysis

Transcription

1 MPRA Munich Personal RePEc Archive Ne migraion and convergence in Porugal. An alernaive analysis Víor João Pereira Domingues Marinho Escola Superior Agrária, Insiuo Poliécnico de Viseu 211 Online a hps://mpra.ub.uni-muenchen.de/3281/ MPRA Paper No. 3281, posed 15. Augus 211 3:35 UTC

2 NET MIGRATION AND CONVERGENCE IN PORTUGAL. AN ALTERNATIVE ANALYSIS Vior João Pereira Domingues Marinho Unidade de I&D do Insiuo Poliécnico de Viseu Av. Cor. José Maria Vale de Andrade Campus Poliécnico Viseu (PORTUGAL ABSTRACT In his work we preend o presen a projec of research abou he idenificaion of he deerminans ha affec he mobiliy of labor. The empirical par of he work will be performed for he NUTS II of Porugal, from 1996 o 22. As main conclusion i can be said which is confirmed he exisence of some labor mobiliy in Porugal and ha regional mobiliy is mainly influenced posiively by he oupu growh and negaively by he unemploymen raes and by he weigh of he agriculural secor. This sudy analyses, also, hrough cross-secion esimaion mehods, he influence of spaial effecs and human capial in he condiional produciviy convergence in he economic secors of NUTs III of mainland Porugal beween 1995 and 22. Keyword: ne migraion; convergence; panel and cross-secion esimaions; Poruguese regions. 1. INTRODUCTION There are many auhors who have dedicaed hemselves o issues of labor mobiliy, wih very differen heoreical assumpions, rying o invesigae how hese issues or do no explain he regional differences. For example, he auhors associaed wih he Neoclassical heory, as (1Solow (1956, consider ha he endency is, for he labor mobiliy, o alleviae, in he medium and long erm, he regional dispariies. This, because hese auhors consider he mobiliy of facors as a funcion of wages and he supply of resources as exogenous. Thus, wha deermines he mobiliy facor is heir compensaion. In anoher conex, i appears ha he curren rend of several economic heories is o consider ha he labor mobiliy accenuaes regional dispariies. Even wriers in he line of neoclassical heory, as Barro and (5Sala-i-Marin (1991, associaed wih endogenous growh heory, now admi ha he mobiliy of labor reacs o processes of convergence and reduce regional dispariies, bu only if some condiions are me. Tha is, lef o disappear he idea of absolue convergence for he same "seady sae" of neoclassical influence, o a perspecive of condiional convergence for differens "seady saes" (2-4(Marinho, 211a, 211b and 211c. 2. THE THEORETICAL MODEL FOR NET MIGRATION WITH PANEL DATA We consider here, he models relaed o he migraory balance of (5Salvaore (1977, (6Kaseli e al. (1989 and (7Soukiazis (1995 and he models of he New Economic Geography of (8Epifani e al. (25. The choice of hese models has o do wih he fac ha seem o be more closely aligned wih he objecives se for his work iniially jus in he absrac. Tha is, models Salvaore (1977, Kaseli e al. (1989 and Soukiazis (1995 are models simpler and can idenify he deerminans of labor mobiliy and he Epifani e al. (25 is a more complee model ha allows us o analyze he dynamics associaed wih he spaial evoluion wih implicaions for labor migraion and unemploymen. The model esimaed in his sudy is wha is presened below in equaion 1. Are represened in he model presened below some new facors, menioned in he economic heory, such as he effecs of congesion, hrough he availabiliy of housing. ( SM / PA c c1 ( ri re c2( DI DE c3( AI c4( si se c5( f I f (1 SM/PA = ne migraion from one counry or region wih he ouside, as a percenage of oal acive populaion of he counry or region; r I-r E = difference beween he growh raes of real oupu, wih r I o be he annual growh rae of real oupu of he originaing counry or region and r E being he average growh raes of real GDP in all counries or regions desinaion; D I-D E = difference beween he inernal unemploymen rae and he exernal average; AI = number of employees in agriculure of he counry or region of origin; s I-s E = difference beween he inernal growh rae of wage and exernal average; E

3 f I-f E = difference beween he inernal growh rae of housing and exernal average. 3. THE THEORETICAL MODEL FOR CONDITIONAL CONVERGENCE WITH SPATIAL EFFECTS AND CROSS-SECTION DATA Bearing in mind he heoreical consideraions, wha is presened nex is he model used o analyse condiional produciviy convergence wih spaial effecs, a a secor and regional level (NUTs III in mainland Porugal: ( 1/ Tlog( Pi Wij pi b Pi X log i, wih e (2 In his equaion (2 P is secor produciviy, p is he rae of growh of secor produciviy in various regions, W is he marix of disances, X is he vecor of variables which represen human capial (levels of schooling primary, secondary and higher b is he convergence coefficien, is he auoregressive spaial coefficien (of he spaial lag componen and is he error erm (of he spaial error componen, wih, W. The indices i, j and, represen he regions under sudy, he neighbouring regions and he period of ime respecively. 4. EMPIRICAL EVIDENCES Then we presen empirical evidence for he differen NUTS II, from 1996 o 22, and for he NUTS III in 1995 and EMPIRICAL EVIDENCES ON THE LEVEL OF NUTS II Analyzing he resuls presened below in Table 1 for he esimaion of equaion (1, we verify which he esimaion mehod which we mus o ake in coun is ha of random effecs, given he value of he Hausman es (no significan saisics. On he oher hand, only he coefficiens associaed wih he relaive growh raes of real oupu, unemploymen raes and he relaive share of agriculural employmen are ha have saisical significance. The firs coefficien referred has posiive effec (only significan for 1% and he las wo negaive effecs (as i was expeced, given he heory. I should be noed, however, ha he coefficien associaed wih he share of employmen has he highes marginal effec ( For hese reasons, we conclude ha he regional mobiliy of labor in mainland Porugal is posiively affeced by growh raes of real oupu, in oher words, greaer is he difference beween he rae of growh of real oupu of a region and he average growh raes of oher regions mos is he migraion of workers ino he region. On he oher hand, i appears ha mobiliy is negaively relaed o unemploymen raes and he relaive share of agriculural employmen. Tha is, higher he unemploymen rae of a region and greaer he weigh of he agriculural secor, lower is he labor migraion o his region. The growh raes for wages and growh raes on he housing sock does no have saisical significance and because his hey have no influence on naional labor mobiliy. Wha is no a surprising, given he Poruguese regional conex. Table 1: Resuls of panel esimaions, wih he equaion of ne migraion for he NUTS II in he period ( SM / PA c c1 ( ri re c2( DI DE c3( AI c4( si se c5( f I c c 1 c 2 c 3 c 4 c 5 G.L. R 2 SEE T.H ** -.746* LSDV (# (1.62 (-1.89 ( (-.86 ( *.31** -.2* * (.188 GLS (2.627 (1.82 ( ( (-.333 (1.395 Noe: LSDV, mehod of esimaion wih fixed effecs; GLS esimaion mehod wih random effecs; * Coefficien saisically significan a 5%; ** Coefficien saisically significan a 1%; GL, Degrees of freedom; SEE, sandard deviaion esimaion; TH, Hausman Tes; (#, all "dummies" saisical significance and values are very close. Figures in brackes are he -saisics EMPIRICAL EVIDENCES ON THE LEVEL OF NUTS III The resuls concerning OLS esimaes of condiional convergence wih ess of spaial specificaion are presen in Table 2, which follows. f E 2

4 Table 2: OLS esimaion resuls for he equaion of absolue convergence wih spaial specificaion ess ( 1/ Tlog( Pi log b Pi i Con. Coef. b JB BP KB M I LM l LMR l LM e LMR e _ R 2 N.O. Agriculure -.399*.46* ( ( ** ** Indusry.49* -.47* (5.431 ( * 13.65*.12** Services.181** -.14 (1.928 ( * 4.94* * * Toal of.138* -.1 secors (2.212 ( Noe: JB, Jarque-Bera es; BP, Breusch-Pagan es; KB, Koenker-Basse es: M I, Moran s I; LM l, LM es for spaial lag componen; LMR l, robus LM es for spaial lag componen; LM e, LM es for spaial error componen; LMR e, robus LM es for spaial error componen;r 2, coefficien of adjused deerminaion; N.O., number of observaions; *, saisically significan o 5%; **, saisically significan o 1%. Produciviy convergence is only seen in indusry, alhough he values of he convergence coefficien presen indicaions of heeroskedasiciy, according o he BP and KB ess. Agriculure presens clear signs of divergence, since he convergence coefficien is posiive and saisically significan. Convergence in he produciviy secor will be condiioned by spillover effecs spaial error in agriculure, evenually, and spillover effecs spaial lag in services, according o he LM ess. Table 3 presens he resuls of he esimaes of spillover effecs spaial error for agriculure and spillover effecs spaial lag for services. Table 3: ML esimaion resuls for he equaion of condiional convergence o spaial effecs ( 1/ Tlog( Pi log Wij pi b Pi Consan Coefficien Spaial Breusch- _ coefficien Pagan R 2 N.Observaions Agriculure -.46*.53* ( (6.558 ( Services (1.365 (-1.65 ( * Noe: *, saisically significan o 5%; **, saisically significan o 1%; ***, spaial coefficien of he spaial error model for agriculure and spaial lag model for services. The convergence coefficien for agriculure is similar o wha is presened in Table 2, alhough i has improved slighly saisically. In services he convergence coefficien is slighly differen in erms of values obained and saisical significance. On he oher hand, he coefficiens of spaial variables have no saisical significance. As a resul, convergence in agriculure and services is no condiioned by spaial effecs. Table 4 presens a series of esimaes for condiional secor produciviy convergence, wih he level of schooling as a proxy for human capial (NUTs III. Three levels of schooling were considered (primary, secondary and higher educaion represened by differen variables. These variables were obained hrough he percenage of he populaion wih each level of schooling in relaion o he oal number of people, aking ino accoun he daa from he Census 21. Differen esimaes for each secor were carried ou for level of schooling so as o avoid problems of mulicollineariy. Table 4: Empirical evidence of he imporance of he level of schooling in he convergence of produciviy in he various economic secors ( 1/ Tlog( Pi blog Pi X i Con. Coef.1 Coef.2 JB BP KB M I LM l LMR l LM e LMR e R 2 N.O. i Agriculure Prim. -.2 ( * (-4.41 High * ( * (3.32.4* ( * ( * ( ( * ( * 5.7** ** ** * 7.98* ** * * ** Indusry Prim..578* ( * (4.89 High...521* ( * ( * ( * ( * ( ( * ( * * * 1.875*.19** * *

5 Services Prim..371* ( ** (1.81 High..284* ( ** ( ( ** ( ( ( ( * 5.55** * * ** 5.31** * * * 7.736* ** ** Toal of secors Prim..37* -.24* -.7* (3.45 (-2.9 ( * -.18*.72** (2.816 ( ( High...213* -.19*.16** (3.1 ( ( ** Noe: Prim., esimae wih primary educaion;, esimae wih secondary educaion; High., esimae wih higher educaion; Con., consan; Coef.1, coefficien of convergence; Coef. 2 coefficien of level of schooling; JB, Jarque-Bera es; BP, Breusch-Pagan es; KB, Koenker-Basse es: M I, Moran s I; LM l, LM es for spaial lag componen ; LMR l, robus LM es for spaial lag componen; LM e, LM es for spaial error componen; LMR e, robus LM es for spaial error componen; R 2, r squared adjused; N.O., number of observaions *, saisically significan o 5%; **, saisically significan o 1%. In agriculure, for he hree levels of schooling, he indicaions of divergence are mainained, since he coefficiens for convergence presen a posiive sign wih saisical significance, alhough he values are slighly lower, which is a sign ha he level of schooling produciviy convergence in his secor, albei slighly. On he oher hand, as could be expeced, primary educaion has a negaive effec on he growh of produciviy in agriculure for he period 1995 o 22, while higher educaion has a posiive effec. Therefore, he progress in he level of schooling in his secor improves produciviy performances. As far as he LM es of specificaion are concerned, wih he excepion of he resuls obained from he esimaions of higher educaion, all figures confirm he previous resuls for his secor, or, in oher words, he beer specificaion of he model is wih he spaial error componen. Indusry confirms in hese esimaions he signs of produciviy convergence across he NUTs III of mainland Porugal from 1995 o 22, a fac which is only favoured by higher educaion (since he effec of higher educaion is posiive and increases convergence. The non-exisence of indicaions of spaial auocorrelaion was also confirmed, given he values of he LM ess. Conrary o wha was seen in he resuls for absolue convergence, in hese esimaions of condiional produciviy convergence in services, has he level of schooling as a condiioning variable. Some indicaions of convergence can be seen in he equaions of primary educaion and higher educaion, which is sign ha evenually convergence will be condiioned o human capial in his secor (since none of he coefficiens associaed o he variables of he level of schooling has saisical significance. On he oher hand, aking ino accoun he LM ess, i is confirmed ha he beer specificaion of he model is wih he spaial lag componen. In he oal of secors, somehing similar o wha was verified in services can see, or, in oher words, he convergence coefficien has no saisical significance in he esimaions for absolue convergence, bu is presen in he esimaions for condiional produciviy convergence wih human capial. The difference is ha here he coefficiens of condiioning variables demonsrae saisical significance, an indicaor ha convergence in he oal of secor sis condiioned by level of schooling. Finally, i should be noed ha he greaes marginal effec is hrough higher educaion schooling, which indicaes ha he higher he level of schooling, he greaer he growh in produciviy. 5. CONCLUSIONS I is noed also ha in he period 1996 o 22, he Algarve was he region wih higher percenages for ne migraion. Differen rend showed he Alenejo region which has even negaive migraion balance a he beginning of he period, which is undersandable, since i is he region wih he highes raes of unemploymen and highes percenage of employmen in agriculure. A NUTs III level and wih cross-secion daa, i can be seen ha secor by secor he endency for produciviy convergence is greaes in indusry. Wih reference o spaial auocorrelaion i is also confirmed ha his possibly exiss in agriculure and services, when aking ino accoun he LM ess. Following he procedures of Florax e al. (23 he equaion is esimaed wih he spaial error componen for agriculure and he spaial lag componen for services, and i can be seen ha he consideraion of hese spaial effecs does no significanly aler he resuls obained previously wih he OLS esimaion. The level of schooling as proxy for human capial condiioning produciviy convergence, improves he value and he saisical significance of convergence coefficiens. On he oher hand, above all he variable which represens higher educaion shows indicaions which direcly favour he growh of produciviy, since he coefficien associaed o i presens in all economic secors he greaes marginal posiive effec. Beside, he differen regional level used, he conclusions are more or less consisen and we can say which he endency in his period is for he regional divergence. 4

6 6. REFERENCES 1. R. Solow. A Conribuion o he Theory of Economic Growh. Quarerly Journal of Economics ( V.J.P.D. Marinho. Analysis of ne migraion beween he Poruguese regions. MPRA Paper 32311, Universiy Library of Munich, Germany (211a. 3. V.J.P.D. Marinho. Spaial effecs and convergence heory in he Poruguese siuaion. MPRA Paper 32185, Universiy Library of Munich, Germany (211b. 4. V.J.P.D. Marinho. Wha said he neoclassical and endogenous growh heories abou Porugal?. MPRA Paper 32631, Universiy Library of Munich, Germany (211c. 5. D. Salvaore. An economeric analysis of inernal migraion in Ialy. Journal of Regional Science, 17, ( L. Kaseli and N. Glysos. Theoreical and empirical deerminans of inernaional labour mobiliy: a Greek- German perspecives. In Gordon, I. and Thirlwall, A., European facor mobiliy, Macmillan E. Soukiazis. The endogeneiy of facor inpus and he imporance of Balance of Paymens on Growh. Am empirical sudy for he OECD counries wih special reference o Greece and Porugal, PhD Disseraion P. Epifani and G.A. Gancia. Trade, migraion and regional unemploymen. Regional Science and Urban Economics, 35, (25. 5