Inter-industry Wage Premia in Portugal: Evidence from EU-SILC data

Transcription

1 MPRA Munch Personal RePEc Archve Inter-ndustry Wage Prema n Portugal: Evdence from EU-SILC data Bagett, Marco and Sccchtano, Sergo Unversty of Rome La Sapenza, Mnstry of Economc Development, Italy 08. December 2009 Onlne at MPRA Paper No , posted 08. December 2009 / 16:54

2 Inter-ndustry Wage Prema n Portugal: Evdence from EU-SILC data - Marco Bagett a and Sergo Sccchtano a,b * - a Mnstry of Economc Development, Department for the Development and the Economc Coheson, va Scla 162, 00187, Rome, Italy. b Unversty "La Sapenza", Faculty of Economcs, Department of Publc Economcs, Va del Castro Laurenzano 9, 00161, Rome, Italy. Ths verson December 8 th 2009 Abstract In ths paper we nvestgate whether nequalty n the nter-ndustry wage prema may be explaned by unobserved dfferences n workers educatonal sklls. We use the 2007 EU-SILC data set for Portugal, a naton whch can be consdered a case-study, due to ts hgh nterndustry wage dsperson. Applyng both OLS and quantle regresson technques, our results suggest that ths unobserved heterogenety s not a relevant matter n the wage prema determnaton. We thus corroborate the prevous emprcal contrbuton to Economc Letters performed by Martns (2004). Keywords: Returns to educaton; nter-ndustry wage nequalty; Quantle regresson; Portugal JEL Classfcaton: C14, I21, J31. * Correspondng author. E mal: sergo.sccchtano@unroma1.t, Tel: , fax

3 1. Introducton 1 There s a wde consensus that the nter-ndustry wage nequalty across countres remans a relevant topc. As a matter of fact, snce the semnal paper by Krueger and Summers (1988) some emprcal analyss have explored possble explanatons. Partcularly, unobserved qualty dfferences across workers have also been called to explan the behavor of ncome nequalty among ndustres, n dfferent countres (e.g., Gbbons and Katz, 1992). In ths context, Portugal s a chefly sutable country, because ts nter-ndustry wage dsperson s found to be hgh when compared wth other European countres (Hartog et al. 2001). In partcular, n a contrbuton to Economcs Letters, Martns (2004, M henceforth), by applyng a Quantle Regresson (QR) approach to a 1995 Portuguese data set, ponted out that unobservable dfferences across workers are not a crtcal element n determnng ndustry wage prema n that country. He tested the relevance of the unobserved worker qualty hypothess by comparng the dfferences n the returns across hgh and low-wage ndustres and evaluatng the correlaton between OLS returns and both QR coeffcents and nterquantle dfferences. The ntuton s that for the unobserved ablty explanaton to be relevant, hgh wage ndustres should also show a large dfference n returns between the top of the wage dstrbuton, where hgh ablty workers are expected to be found, and ts bottom, where are lkely to be low ablty employees. In ths paper we apply a smlar approach, usng the last 2007 wave of the European Unon Statstcs on Income and Lvng Condtons nqury (EU-SILC), the new European homogenzed panel survey. More specfcally, our prmary purpose s to explore the potental for EU-SILC data to shed some lght on the nter-ndustry nequalty n Portugal s wage prema. 1 The vews expressed n ths artcle are those of the authors and, n partcular, do not necessarly reflect those of the Mnstry of Economc Development. The usual dsclamer apples. 2

4 To address these ssues we apply the QR sem-parametrc approach whch s more nterestng, as well as more sutable, for t allows us to get a more precse pcture of the dynamcs of the dependent varable at dfferent ponts of the dstrbuton, rather than at the condtonal mean. We deeply examne QR and OLS, n order to provde a cross-ndustres comparable vew. The paper supports the dea that n Portugal there s not only a relevant cross-ndustres dsperson n returns to educaton, but also a hgh heterogenety n wage prema at dfferent ponts of the wage dstrbuton, whch OLS modellng of condtonal average of a dependent varable completely fals to account for. In partcular, 12 years later the data used n M s analyss and through a dfferent data-set, we confrm that unobservable qualty dfference across workers cannot be called to spell out Portuguese nequalty n the nter-ndustry wage prema. Ths emprcal paper s organzed as follows. The next secton descrbes the data. Secton 3 llustrates our econometrc specfcaton. Secton 4 reports the results as well as a robustness check. Secton 5 dsplay the robustness check, whle n the last secton we present our man conclusons. 2. Data Data are collected from the 2007 EU-SILC survey verson 1 of March contanng nformaton for 24 European countres amongst whch Portugal. It s the new homogenzed panel survey that has replaced European Communty Household Panel (ECHP). Our analyss focuses on Portuguese full-tme male workers aged between 25 and 65: women have been dsregarded on account of potental selectvty bases. Younger males have been also dropped because they are stll n the almost exclusvely educatonal perod of ther lfe,.e. they are very much lkely enrolled n a secondary or tertary course than performng a work actvty whatsoever. In the

5 EU-SILC wave 4,665 Portuguese men are ntervewed: yet, our dependent varable - the hourly (logarthmc) gross wage s avalable for 1,735 full-tme Portuguese male-workers aged between 25 and 65 years. These consttute our reference sample. Our regressors are schoolng years, whch has been bult up followng the usual framework 2, the number of years spent n pad work and ts squared: the second s regarded as beng a proxy for ndvdual experence whle the thrd takes account of possble non lneartes. Sectors surveyed are those consdered n EU-SILC,.e. the 12 macro-sectors obtaned by jonng several NACE 1.1 classfcaton codes (see table 1 for further detals). [Table 1 here] 3. Econometrc specfcaton: OLS versus quantle regresson form: We assume our OLS specfcaton to have the followng smple 2 ln w = α + β S + γ X + δ X + ε (1) As s well known, OLS mplctly assume that the mpact of the regressors along the condtonal dstrbuton of the response varable are rrelevant. But as s already equally known covarates may nfluence t 2 That usual framework refers to makng use of the hghest ISCED level of educaton attaned by a male worker, and for each level assgnng the legal mnmum number of years typcally requred to acheve t: more precsely, those who reached only an ISCED 1 grade have been gven 5 years of schoolng; 8 years of school have been assgned to those wth an ISCED 2 grade; 13 years to those wth an ISCED 3; 14 to people who attaned an ISCED 4 grade and 18 years to those who reached an ISCED 5 4

6 on ts whole shape. Ths case can be studed by performng a quantle regresson (QR), whch has the followng functonal form (Koenker & Basset, 1978): Quant + :ln w x β ln w 2 ( α + β S + γ X + δ X ) + 2 ( ) ( ) 1 ln w α + βs + γ X + δ X :ln w xβ (2) The equaton (2) s normally wrtten as: mn k β R ρ 2 ( ln w α β S γ X δ X ) (3) where ( z ) z ρ = f 0 z or ρ ( z ) ( 1)z = f z<0. Ths problem s solved usng lnear programmng methods. Standard errors for the vector of coeffcents are obtanable by usng a bootstrap procedure. 4. Results In table 2 we show OLS returns to educaton as well as condtonal returns at 5 representatve quantles n the 12 Portuguese ndustres. Dfferences between percentles of the wage dstrbuton computed for 4 dfferent extremes taken by twos (90-10 and 75-25) are also reported. A hgh dsperson across ndustres clearly emerges as a stylzed fact. The hghest OLS sectoral coeffcent (health and socal work) s almost 3 tmes hgher than the lowest (Other communty, socal and personal servce actvtes, Prvate households wth employed persons, Extra-terrtoral organzatons and bodes). [Table 2 here] 5

7 Comparng OLS and QR coeffcents, t can be noted that the ndustry wth the hghest (lowest) value n terms of OLS s not the ndustry wth the hghest (lowest) spread between the top and the bottom of the dstrbuton, both for the and dfferences. Smlarly to M - who compares only the quantle dfference n returns between the 15 hgh-payng and 15 low-payng ndustres (ranked n terms of OLS) - we fnd that the nterquantle dfferences between the extremes of the dstrbuton are not bgger for the ndustres wth hgh OLS returns compared to the ndustres wth low OLS returns. Wth regard to the spread, results evdence that the gap between the 6 ndustres wth hgher OLS returns and ther 6 counterparts wth lower estmated OLS coeffcents s tghter and only equal to The same dfference was about n M. Usng the same OLS rankng of ndustres, such gap s found to be even negatve for the spread. Accordng to M, for the unobserved heterogenety to be a crtcal element for wage prema determnaton, we would expect a really bgger dfferental between the top of the dstrbuton, where the hghest ablty workers are expected to be, and ts bottom. 5. Robustness check Further, we test whether gaps between quantle coeffcents estmated n our QR are statstcally sgnfcant. The test has been carred out wth respect to the 2 spreads consdered n the paper (90-10=0 and 75-25=0) and to all quantles. More specfcally, p-values are obtaned through a bootstrapped varance-covarance matrx that ncludes between quantle blocks. Smlarly to M, the results ndcate that n most cases the 2 lnear hypothess (95-5=0 and 75-25=0) are found to be not sgnfcant. As expected, sgnfcance decreases 6

8 when the nterquantle spread also decreases. Fnally, n addton to M, we fnd that even the jont equalty of coeffcents at all quantles s not rejected n many cases. Moreover and smlarly to M, whle the OLS returns are well correlated wth almost all the QR returns (table 3), both n terms of Pearson and Spearman s correlatons, they are not correlated wth the dfference between the top and the bottom of the dstrbuton. The Pearson s correlaton coeffcent between OLS and the quantle dfference s even negatve. [Table 3 here] 6. Conclusons In ths paper we have appled OLS and QR technques to the last 2007 EU-SILC wave n order to clear up a possble puzzlng presence of nter-ndustry nequalty wage prema for full-tme male workers n Portugal. In dong so, we have updated results acheved by Martns (2004) for the year We have found hgh dsperson of OLS returns amongst the 12 consdered sectors of the Portuguese economy, but a weak correlaton between OLS and the magntude of 2 quantle benchmarkng dfferences (95-5 and 75-25). Furthermore, n accordance to Martns (2004) we have demonstrated that f sectors are ranked by OLS coeffcents and dvded n 2 groups (hgh and low returns), the nterquantle dfference s small (and for the dfference s even negatve). Tests on the 2 lnear hypothess 95-5=0 and 75-25=0 are found to be not sgnfcant whle OLS are well correlated wth almost all the QR coeffcents but not wth the dfference between the top and the bottom of the dstrbuton. 7

9 These results suggest that for Portuguese adult male workers, unobserved heterogenety possbly generated by ndvdual sklls s an rrelevant matter n the wage determnaton, whle other factors such as the relatve strength of ndustres n a small country may play a more mportant role. 8

10 References Gbbons, R. and L. Katz, 1992, Does Unmeasured Ablty Explan Inter- Industry Wage Dfferentals?, Revew of Economc Studes 59, Hartog, J., Perera, P.T., and J.C. Vera, 2000, Inter-ndustry wage dsperson n Portugal: hgh but fallng. Emprca 27, Koenker, R. and G. Bassett, 1978, Regresson quantles. Econometrca 46, Krueger, A. and L. Summers, 1988, Effcency Wages and the Inter- Industry Wage Structure, Econometrca 56, Martns P., 2004, Industry Wage Dsperson: Evdence from the Wage Dstrbuton, Economcs Letters, 83(2),

11 Tables Tab. 1. Summary statstcs Industres - NACE 1.1 Class. EU-SILC class. % Fshng, Agrculture, huntng and forestry a+b 7.17 Mnng and quarryng, Manufacturng, Electrcty, gas and water supply c+d+e Constructon f Wholesale and retal trade; repar of motor vehcles, motorcycles and personal and household good g Hotels and restaurants h 4.49 Transport, storage and communcaton 5.67 Fnancal ntermedaton j 2.31 Real estate, rentng and busness actvtes k 4.81 Publc admnstraton and defence, compulsory socal securty l Educaton m 3.13 Health and socal work n 1.86 Other communty, socal and personal servce actvtes, Prvate households wth employed persons, Extra-terrtoral organzatons and bodes o+p+q 2.31 Total 100 Tab. 2. Condtonal returns to schoolng - OLS and QR - and sgnfcance tests a+b c+d+e f g H j k l m n o+p+q OLS = = = = = All eq Note. Data are from cross sectonal UDB SILC 2007 verson 1 of March 2009; p-values n talcs. 10

12 Tab. 3. Correlaton between OLS and QR coeffcents Pearson Spearman = = = = =