DOES INMIGRATION CONTRIBUTE TO CONVERGENCE AMONG REGIONAL LABOUR MARKETS IN SPAIN?

Transcription

1 DOES INMIGRATION CONTRIBUTE TO CONVERGENCE AMONG REGIONAL LABOUR MARKETS IN SPAIN? Cuadrad-Rura, Juan R. (*) (**), Iglesias-Fernández, Carls (*) (**) and Llrente-Heras, Raquel (**) 3 ABSTRACT Freign immigratin is still a recent phenmenn in Spain, but its magnitude and grwth has been extrardinarily imprtant during recent years. Accrding t the available data (Freigners Official Register), freign residents fficially registered have increased by 700 percent between 1991 and 2005 that is, frm arund t mre than 2.7 millin peple. It is generally accepted that immigrants may have, accrding t their magnitude, imprtant effects n labur markets f the destinatin cuntries. Firstly, because immigrants frm a rather different ppulatin grup, due t their persnal and labur characteristics when cmpared with the dmestic ppulatin. S, immigrants may cnstitute a ptential factr in changing the dynamics f the dmestic labur markets. On the ther hand, in the Spanish case, the immigrant ppulatin is characterized by displaying an unequal territrial distributin, with remarkable reginal differences in terms f immigratin rates and aggregate figures. On the basis f these ideas, the aim f the paper is t analyse t what extent the massive immigratin flws t Spain in recent years have mdified the parameters f reginal labur markets, cntributing r nt t change the reginal differences f the main characteristics f the dmestic markets. KEY WORDS: Immigratin, Regins, Cnvergence, Labur markets (*) University f Alcala. Madrid. Dept. f Applied Ecnmics (**) Institute f Ecnmic and Scial Analysis SERVILAB. Univesity f Alcala. 3 jr.cuadrad@uah.es; carls.iglesias@uah.es; raquel.llrente@uah.es 1

2 1. Intrductin. Cntrary t what has happened in ther cuntries, the receptin f immigrants in Spain has been particularly fast. This has led t the immigrant ppulatin reaching nticeably relevant levels within the ttal ppulatin in a shrt time. While freigners residing in Spain in 1991, with residence permits, were n mre than 360 thusand, accrding t data frm the Ministry f Labur and Scial Affairs (MTAS), at 31 st December 2005 this figure had reached 2.7 millin peple. In the last 14 years the number f freigners has increased mre than 700% 4, an evlutin that has led t an immigrant ppulatin in 2005 f smething mre than 6% f the ttal ppulatin. In additin, this migratry prcess has been particularly intense in recent years, accelerating mst nticeably since the year Frm this year n the increase in residence permits has reached 205%. Of curse, these data nly take int accunt the fficial r legal immigrant ppulatin. Outside these figures there is anther immigrant ppulatin withut papers which, in Spain, is estimated at arund mre than a millin peple. Therefre, when cnsidering this grup the afrementined figures culd be even greater. In additin t its size, immigratin is relevant fr its pssible effects, psitive and/r negative, n the labur market f the receiving cuntry. These affect mre specifically the permanent r transitry cntributin f labur patterns, which are different frm thse f the natinal wrkfrce (Chiswick, 1978), the labur pprtunities f native wrkers (Brjas, 1994, 1999, 2003; Friedberg and Hunt,1995), the intrductin f a "disciplining effect" n the reserve wage f the native-brn wrkers, reducing their pssible pwer as insiders (Dlad, 2001) and their preferential allcatin t lw incme psts (Thurw, 1976, Pire, 1979), amng thers. The immigrant ppulatin presents strng differences cmpared t the native wrkfrce with regard t its labur characteristics (Izquierd, 2003; Cuadrad, Iglesias and Llrente, 2006, t illustrate the Spanish case). S, ntably different parameters can be bserved in relatin t the labur market (rates f activity, ccupatin and unemplyment), their wrking cnditins (activity sectr, wrk ccupatin, type f cntract, type f wrking day, labur situatin, etc) and even their wrk dynamics (greater presence in flws f entry t and exit frm unemplyment, fr example 5 ). Finally, the immigrant ppulatin, far frm being hmgeneusly distributed thrughut the cuntry, presents gegraphic patterns f very unequal lcatin. The first theretical mdels attributed the migratin decisins t the different salaries existing between the rigin and destinatin f migratin (Raimn, 1962). S the unequal territrial distributin f the immigrants in the destinatin cuntry shuld reflect the wage differences between different regins, explaining that the freign ppulatin established its residence in the regins with higher wages. The literature als ntes, frm perspectives f analysis based n search mdels, that the existence f pprtunities f emplyment in the place f destinatin (Harris and Tdar, 1970) shuld als be a relevant explanatry variable. Accrding t this, immigrants shuld lcate especially in thse regins with higher rates f emplyment r with mre active labur market. Later, the mdels based n the thery f Human Capital explain the decisins f migratin n the basis f cmparisn f csts (transprt, psychlgical, 4 A mre detailed descriptin f this prcess can be fund in Cuadrad, Iglesias and Llrente (2006). 5 In Cuadrad, Iglesias and Llrente (2006) these cnclusins can be fund in detail. 2

3 cultural) and benefits (increase f incme in the cuntry f destinatin cnditined by the prbability f btaining a jb) assciated t displacement (Sjaastad, 19 62), s that the immigrants will mve t thse regins where the benefits (r the difference between csts and benefits) are greatest. In these prcesses f cmparisn, individuals can als value the existence f ethnic netwrks r previus cncentratins f immigrants f the same natinality r culture, as this fact will reduce the csts and als perhaps increase the benefits f migratin (Byd, 1989, Massey,1990). The implicatin is that immigrant ppulatins tend t cncentrate gegraphically. Accrding t the data f the Natinal Institute f Statistics (INE) fr the year 2005, in Spain the different regins shw marked differences in their rates f immigratin. The greatest presence f immigrants is in the Balearics Ils., where practically 16% f the resident ppulatin is freign. This is fllwed by Madrid, Murcia and the Cmmunity f Valencia with between 12 and 13 per cent. On the cntrary, in Galicia, Extremadura and Asturias the resident freign ppulatin des nt reach 3%. This paper aims t analyse the extent t which the arrival f immigrants in Spain and their uneven reginal distributin can be cntributing t a mdificatin f the existing labur differences amng Spanish regins, thus cnstituting a relevant factr in the explanatin f reginal cnvergence patterns. Fr this purpse, in sectin 2 a descriptin f the principal labur differences existing amng Spanish regins is ffered and an pinin is given n whether a cnvergence patterns exists amng them, n the basis f a beta-cnvergence calculatin. Later (sectin 3), the imprtance f immigratin in Spanish regins will be analysed and the different wrk patterns that immigrants present accrding t their place f rigin and in the regin in which they are. Using this analytical framewrk, in sectin 4 we determine what the cntributin f the immigrants is t existing labur differences in Spanish regins. Firstly, a breakdwn f the evlutin f the active emplyed and unemplyed ppulatin is made by varius effects, determining in this way what the cntributin f the immigrant ppulatin is and its cmpsitin in this evlutin. Then, a series f virtual scenaris n the evlutin f reginal labur markets is built frm this breakdwn, determine the evlutin f the basic reginal wrking ppulatins nce the effect induced by immigratin has been discunted. Finally, we re-estimate reginal cnvergence in labur terms, having discunted the induced effect f the immigrants. Using this strategy, the imprtance f the effects induced by immigratin can be checked. The paper finishes with sectin 5 shwing the main cnclusins reached and evaluating if the arrival f immigrants has in effect cntributed t the increase r decrease f the differences existing in reginal labur markets. Mst f the data used cme frm the Survey f Active Ppulatin ( EPA, Encuesta de Pblación Activa ) relative t the secnd quarter f the perid between 1987 and 2005 and refer t the active emplyed r unemplyed ppulatin brken dwn by regins and the immigrants cuntry f rigin. Additinally, data frm the Anuaris de Extranjería (Freigners Official annual reprts) f the MTAS are used and the Padrón Municipal de Habitantes (Municipal Ppulatin Census) elabrated by the INE. This paper, data n immigrants refer t the ppulatin that appears in the statistics. Fr this reasn the results btained shuld be taken with cautin as a large number f immigrants exist wh d nt nrmally appear in the fficial statistics. 3

4 2. Labur hetergeneity in Spanish regins and its evlutin in the perid 1987 t In Spain, the hetergeneity f reginal labur markets is ntably high. This fact is illustrated in figures 1-3 where the different basic labur rates by regins are cmpared t natinal averages. In terms f rate f activity, with the average rate at 57%, Madrid, Catalnia and Navarre shw rates f activity f 61 t 62%. On the cntrary, Asturias limits the value f this rate t values f 48% and Extremadura is n higher than 51%. Frm the emplyment rate perspective, the differences are smewhat mre enlarged, abve all in relatin t the least favured regins. Thus, while Madrid, Catalnia and Navarre have rates f ccupatin f abut 57% (the natinal average is 52%), Extremadura and Asturias are the regins with the lwest emplyment rates, 43 t 44%, nearly 10 percentage pints belw natinal average. Finally, the differences in terms f unemplyment rates are even mre ntable. Six regins have unemplyment levels higher than the average and 10 regins have unemplyment rates belw. The differences are particularly high fr Extremadura (15%), Andalusia (13%) and the Canaries (10.5%). On the cntrary, the lwest levels f unemplyment can be seen in Aragón, Madrid and Navarra with rates belw 6%. Figure1. Rates f activity by regins and cmparisn with the natinal average. 4 th. Quarter, Surce: EPA. (INE) Madrid Cataluña Navarra Baleares Rija Canarias C. Valenciana Murcia País Vasc Aragón Cantabria Andalucía Castilla - La Mancha Galicia Castilla y León Extremadura Asturias Figure 2. Rates f emplyment by regins and cmparisn with the natinal average. 4th. Quarter Surce: EPA (INE) Madrid Cataluña Navarra Rija Baleares C. Valenciana Murcia País Vasc Canarias Aragón Cantabria Castilla - La Mancha Castilla y León Galicia Andalucía Asturias Extremadura 4

5 Figure 3. Rates f unemplyment by regins and cmparisn with the natinal average, 4th. Quarter Surce: EPA (INE) Extremadura Andalucía Canarias Asturias (Principad de) Castilla - La Mancha Galicia Castilla y León Cantabria Cmunidad Valenciana Baleares (Islas) Murcia (Región de) Cataluña Rija (La) País Vasc Navarra (Cmunidad Fral de) Madrid (Cmunidad de) Aragón T quantify these differences and t cmpare and mnitr them ver the time, a reginal variatin cefficient has been calculated frm the afrementined rates. The results are shwn in Figures 4 t 6 6. Frm them it can be deduced that reginal labur markets present ntable differences especially in terms f unemplyment rates. The differences in the rates f activity and ccupatin, althugh imprtant, are smaller. On the ther hand, the three rates shw a fairly similar evlutin pattern. There was an increase in the labur differences between regins until the year 2000, in the rates f ccupatin and unemplyment, which in the case f the rate f activity cntinued until the third quarter f the year Frm then n, the cefficients f variatin shw a clear inflectin in their evlutin beginning with at least a mderate reductin in the reginal variability f activity, ccupatin and unemplyment. Figure 4. Evlutin f the cefficient f variatin f the rate f activity fr regins Surce: EPA (INE). 0,1 0,09 0,08 0,07 0,06 0,05 0,04 6 The cefficient f variatin is defined as the qutient between the standard deviatin f the series and its average value. The change in the definitin f unemplyment in 2001 and the new methdlgy intrduced in 2005 impse n the EPA a rupture f the principal labur rates (activity rate and rate f unemplyment) which, hwever, shuld nt be an influence in the reginal distributin f these rates. 5

6 Figure 5. Evlutin f the cefficient f variatin f the rate f unemplyment fr regins Surce: EPA (INE). 0,12 0,11 0,1 0,09 0,08 0,07 0,06 Figure 6. Evlutin f the cefficient f variatin f the unemplyment rate fr regins Surce: EPA (INE). 0,5 0,45 0,4 0,35 0,3 0,25 0,2 0,15 T give greater supprt t the afrementined cnclusins, we have estimated the equatins f cnvergence r beta-cnvergence n the evlutin f the rate f activity, emplyment and unemplyment. ln Rate 7 t, t 1 = α 0 + α1 ln Ratet 1 (1) This permits us t determine whether regins with highest labur rates at the beginning f the perid have managed t clse the gap with the thers reducing the existing differential (s, in the case f the existence f cnvergence cefficient α 1 shuld be negative) 8. 7 Frm this equatin we can deduce a measure f apprximatin thrugh the estimatin f the rate f cnvergence ( β ) twards a statinary situatin as fllws: α ( e βt 1 = 1 ) and f the time by reducing the existing differential by half: ln 2 t * = β See: Barr, R. J. y Sala i Martín (1991) and (Sala i Martín, 1994). 8 This affirmatin is valid fr the case f the rate f activity and emplyment but nt fr the rate f unemplyment. In this case, the cnvergence r apprximatin between the regins is defined as the 6

7 Table1. Regressins f Cnvergence 9. Surce: Authr s figures based n EPA (INE). Rate f activity Initial level -0,119 (-0,453) 0,017 (-0,002) -0,175 (-0,556) Cnstant 0,013 (0.511) (0.170) 0,815 (0,661) R 2 0,013 0,001 0,020 Emplyment rate Initial level -0,329 (-1,396) -0,178 (-2.890)** -0,101 (-0,402) Cnstant 1,426 (1,678) (3,366)** 0,508 (0,546) R 2 0,115 0,357 0,010 Speed f cnvergence 0,008 Time necessary fr halving the differential. 88 (quarters) 22 (years) Unemplyment rate Initial level 0,224 (0,645) -0,412 (-6,387)** -0,017 (-0,026) Cnstant -1,110 (1,090) 0,729 (4,520)** -0,017 (-1,081) R 2 0,027 0,731 0,072 Speed f cnvergente. 0,016 Time necessary fr halving the differential. 42 (quarters) 11 (years) Number f bservatins: 17. The values f the t- student appear in brackets. ** Significant with a prbability f 95 %. Table 1 shws the results reached. Mst f them indicate that the existence f cnvergence cannt be cncluded as mst f the cefficients are nt significant. Hwever, fr the perid the hypthesis f cnvergence is significant bth fr the rate f emplyment and the rate f unemplyment, in bth cases reaching rbust and significant results. As we can see in sectin 3, this perid crrespnds t an intense grwth f migratry flws twards the Spanish labur market. Our bjective is t determine t what degree the tw issues are linked 10. Fr thse perids fr which significant results have been btained, the rate f cnvergence amng the Spanish regins has been calculated and an estimatin was made f the number f perids (quarters r years) which wuld be necessary fr reducing the existing reginal difference by half. The results btained shw that the speed f apprximatin is nt very high in the case f the emplyment rate, but is high reductin f the rates f unemplyment in thse regins with greater unemplyment at the beginning f the perid. 9 Althugh the methdlgical changes in the EPA affect the time series f the activity, emplyment and unemplyment, in the calculatin f these regressin we maintain the hypthesis that the changes affect the different Spanish regins in a similar way, and therefre they are nt relevant fr the analysis f reginal dispersin r reginal cnvergence. 10 Fr apprximately the same perid, , the Bank f Spain (Bletín Ecnómic, Octber 2005) reaches similar results in the case f unemplyment rates. In the case f emplyment rates, their analysis als btains similar but nt significant results. Regarding the Eurpean case, reginal cnvergence f unemplyment is still lw, with persist and high reginal disparities (Llrente, R. 2004). 7

8 in the rate f unemplyment. In the first case, apprximately 22 years wuld be necessary fr halving the disparities, but in the secnd case, nly 11 years. 3. The freign ppulatin resident in Spain and its regins. As it was previusly stated, the grwing immigratin that is taking place in Spain is nt distributed unifrmly amng the Spanish regins. At the same time, different labur patterns are detected within the immigrant ppulatin depending n their rigin. These tw facts lead us t think that the labur effects intrduced by immigratin culd be substantially different at reginal level. In this sectin we will analyse the reginal distributin f migratin as well as its different labur characteristics. Fr this, we will firstly intrduce the issues frm an aggregate perspective (3.1) and fllw n by intrducing the reginal detail (3.2) Immigratin and its labur characteristics: an aggregate analysis. Figure 7 shws the evlutin f the number f freigners with residence permits residing in Spain. In little mre than 10 years, the number f immigrants has multiplied by five. The grwth f the freign ppulatin has taken place abve all since 1994, but reaches its greatest intensity after the year Figure 7- Evlutin f the number f resident freigners in Spain with residence permit. Surce: Freigners Annual Reprt. MTAS Hwever, we shuld differentiate between tw different types f immigratin, n the basis f its ecnmic and nn-ecnmic character. Our analysis is centred n the frmer, as this is mre relevant fr the analysis f immigratin. T define the ecnmic immigrant ppulatin, the cuntry f rigin is used as a criterin thereby establishing that ecnmic immigrants are thse riginating in a less develped cuntry 11. Accrding t the data crrespnding t the secnd quarter f 2005, the rate f migratin f an ecnmic character reached 6.29 per cent f the ttal ppulatin, a surprising 11 The rigins cnsidered fr the immigrant ppulatin f ecnmic character are the rest f Eurpe (Eurpean cuntries with the exceptin f EU-15), Africa, Central America and the Caribbean, Suth America and thers (Asia, Oceania and Stateless peple). This classificatin is based n data frm the EPA. Therefre, it is nt pssible t determine in greater detail thse cuntries with less develpment. 8

9 figure taking int accunt the starting pint: hardly 1 percent in the secnd quarter f In additin, the immigrants shw different wrk patterns depending n their rigin. Table 2 shws these differences in relatin t the principal labur rates. Analysing the data f 2005, it is cncluded that immigrants cming frm the rest f Eurpe and Suth America are thse wh btain best labur results: higher rate f activity and emplyment and lwer rates f unemplyment. The immigrants cming frm Asia, Nrth America, Central America and the Caribbean and Suth America shw uneven labur results. Finally, immigrants riginating in Africa shw the wrst results: lwest rate f activity and emplyment and quite high rates f unemplyment. This pattern persists thrughut the perid in the case f Central America and f the Caribbean, Suth America and Africa. Table 2. Rates f activity, unemplyment and emplyment f ecnmic immigrants by rigin. Surce: authr s figures frm EPA (INE) data. Rest f Eurpe Activity rate Unemplyment rate Emplyment rate 2 tri tri Crec. 2 tri tri Crec. 2 tri tri Crec. 72,82 83,28 10,46 14,56 10,12-4,44 62,22 74,85 12,64 África 65,13 69,94 4,81 24,87 20,71-4,17 48,93 55,46 6,53 Nrth America 71,25 50,39-20,86 14,26 1,66-12,61 61,09 49,56-11,53 Central America and Caribbean 73,71 69,86-3,85 14,59 15,89 1,30 62,95 58,76-4,19 Suth America 76,57 83,34 6,76 13,28 10,09-3,18 66,41 74,93 8,52 Asia 66,23 70,44 4,21 6,09 4,68-1,41 62,20 67,14 4,95 Average 72,85 70,38-2,47 15,08 10,60-4,48 62,05 62,61 0,56 T summarise, bth the intensity f immigratin and its different cmpsitin by rigin are relevant and imprtant aspects t be taken int accunt in the analysis f the dmestic labur market. 3.2 Immigratin and its labur characteristics: a reginal analysis. Data f Table 3, rdered accrding t the present rate f migratin, shw that the rates f migratin in Spanish regins vary in a range f n less than 10 percentage pints frm a minimum in Extremadura (1.4 6%) t a maximum in Murcia (11.5 2% ) 12. On the ther hand, we can bserve a high cncentratin f immigrants in Madrid, Catalnia, the Balearics and Murcia, regins that have undergne spectacular increases in their rates f immigratin. The cnclusin is therefre, the existence f a reginally uneven distributin f the immigrant ppulatin resident in Spain. 12 Reginal rates figures are different frm the afrementined in the first sectin because the analysis is restricted here t the immigratin f ecnmic character. 9

10 Table 3. Ecnmic immigratin rates per regin. Surce: EPA (INE). RATE ECONOMIC IMMIGRATION 2 qtr qtr Grth. Extremadura Asturias Castile & Len Cantabria Galicia The Basque Cuntry Andalusia Aragn Castile-La Mancha Navarre Average Cmmunity f Valencia Ceuta & Melilla La Rija Canaries Madrid Catalnia Balearics Murcia Furthermre, the immigrant ppulatin presents reginal differences frm the pint f view f their prfessinal activities. This culd be due, as underlined in the previus sectin, t the fact that immigrants are prfessinally different depending n their rigin, r that the different regins create distinct behaviur due t peculiarities in their respective labur markets. T answer these questins table 4 includes the activity, emplyment and unemplyment rates f ecnmic immigrants accrding t regin in which they live. Frm this data sme interesting bservatins can be made. Aragn, the Balearic Islands, the Canary Islands, Castile-Len, Madrid, Murcia, Navarre, the Basque Cuntry and the Cmmunity f Valencia are regins where ecnmic immigrants enjy the mst favurable rates f emplyment and the lwest unemplyment rates. a) Althugh in terms f activity rates its results are belw average, in Catalnia immigrants enjy higher emplyment rates and lwer unemplyment rates. b) On the ther hand, emplyment rates are wrse in Extremadura, Galicia and Ceuta and Melilla, where they are always belw average. Andalusia, Asturias, Cantabria and the Rija als present pr emplyment pprtunities fr immigrants thugh nt as general as thse in the regins mentined previusly. They cincide with thse regins that are badly situated in s far as the relative characteristics f their labur markets are cncerned. c) Only in Castile-Len, Extremadura and Galicia has there been an abve average imprvement in the immigrants emplyment prspects in the three indicatrs used. Hwever, they are regins with a cmparatively lw number f immigrants. 10

11 Table 4. Activity rate, unemplyment and emplyment rates f ecnmic immigrants accrding t the regin in which they reside. Surce: Authrs. Frm EPA (INE). Activity rate Unemplyment rate Emplyment rate 2 qtr. 2 qtr. 2 qtr. 2 qtr. 2 qtr. 2 qtr. Grth. Grth. Grth Andalusia Aragn Asturias Balearics Canaries Cantabria C-La Mancha C-Len Catalnia Extremadura Galicia La Rija Madrid Murcia Navarre Basque. C C. Valencia Ceuta & Melilla Average Typical deviatin T sum up, it can be cncluded that the intense wave f immigratin taking place in Spain shws ntable reginal differences, related as much t the territrial distributin f their residence, their cuntry f rigin, their emplyers and the registered grwth in emplyment. In general, regins with better labur markets cntain larger numbers f immigrants. Besides which, the immigrant ppulatin resident in these cmmunities are als assciated with better labur results. The ppsite results are usually prduced in regins that are in a wrse situatin in the natinal labur market. Departing frm these results, the fcus f the wrk will be t determine the influence this behaviur has n the reginal cnvergence f the Spanish labur market. 4. The cntributin immigratin makes t reginal differences and cnvergence: the cnstructin f virtual ecnmies. T determine the influence an unequal gegraphic distributin f immigrants has n the prcess f reginal cnvergence a duble level strategy will be fllwed. Firstly, we are ging t carry ut different shift-share analyses with the bjective f determining which cmpnents are respnsible fr the evlutin f the reginal labur statistics we have been using (activities, ccupatin and unemplyment). Secndly, we will build a series f virtual ecnmies (Marimn and Zilibtti, 1996 and 1998; Murill, Nuñez and Usabiaga, 2003), enclsed t different stages f grwth, depending n sme f the results previusly btained by means f the shift-share analyses, and try t islate the 11

12 effect the immigrant ppulatin has n the questins raised (reginal differences and cnvergence) 4.1Explanatry factrs in the reginal grwth f the active ppulatin, emplyed and unemplyed: a shift-share analysis. The shift-share analysis allws us t differentiate between thse factrs respnsible fr the registered variatin between determinate variables, s that the cntributin t grwth made by each f the variables cmpnents can be islated. In ur case, and with the aim f analysing the cntributin made by the unequal reginal distributin f the immigrant ppulatin, tw grups have been frmed, ne fr the grwth f the ttal ppulatin irrespective f natinality and the ther limited t the ecnmic immigrant ppulatin. In the first analysis carried ut, the active ppulatin, emplyed and unemplyed by regins, has been divided int three effects: the natinal effect, the mix r cmpsitin effect and the reginal effect. Accrding t the fllwing equatin, grwth in emplyment can be divided in the fllwing way (Mayr, López y Pérez, 2004): E ij E ij = E ij = E ij r + E ij (r i r) + E ij (r ij r i ) (1) being, r s r i= 1 j= 1 = s ( E r i= 1 j= 1 ij E ) E ij ij r i r ( E ij j= 1 = r j= 1 E ) E ij ij r ij E = ij E E ij ij (2) Each term f the equatin indicates the fllwing effects: a) Natinal Effect: NE ij = E ij r, represents the change in the active ppulatin, emplyed and unemplyed, that wuld have been prduced in each f the regins if these had grwn at a similar rate t that registered fr the cuntry as a whle. b) Mix Effect: ME ij = E ij (r i r), represents the change in the active ppulatin, emplyed and unemplyed, f the regins due t a change r mdificatin in the structure f the natinality f the resident ppulatin. T achieve this, we have cnsidered that the ppulatin f a regin is cmpsed f native individuals and immigrants. This term assumes that part f the changes registered in the labur variatins f each regin is due t the relative specialisatin f ne natinality r anther. c) Reginal Effect: RE ij = E ij (r ij r i ), represents the change in the active ppulatin, emplyed and unemplyed, due t the dynamism f the regin itself. It btains, in relative terms, what the regin itself cntributes t the variatins in the ppulatins under cnsideratin. 12

13 Table 5. Breakdwn f the percentage f grwth f the active ppulatin, emplyed and unemplyed, fr all Spanish regins. Surce: Authrs. Frm EPA (INE), Percentage f grwth f the active ppulatin Natinal Mix Reginal Effect effect effect Percentage f grwth f the emplyed ppulatin Natinal Mix Reginal Effect effect effect Percentage f grwth f the unemplyed ppulatin Natinal Mix Reginal Effect effect effect Andalusia Aragn Asturias Balearics Canaries Cantabria C-La Mancha C-Len Catalnia Extremadura Galicia La Rija Madrid Murcia Navarre Basque. C C. Valencia Ceuta & Melilla Nte: The majr effects f each regin have been highlighted. Table 5 summarises the results btained and shws the percentage due t each f the effects under cnsideratin. Frm table a number f imprtant questins are raised: A) First, natinal effect is larger in each f the labur grups cnsidered (active, emplyed and unemplyed). Independent f the arrival f immigrants, the main grwth in the labur market is due t the general grwth in the cuntry as a whle. B) The cntributin made by immigrants (mix effect) is nt the largest in any f the regins. S althugh the immigrant flws are imprtant and grwing but the levels reached are nt exterminating t the evlutin f reginal emplyment and unemplyment. C) Nevertheless, the arrival f immigrants has been especially psitive in the grwth f the active ppulatin in Murcia, the Balearics, Madrid and the Canaries. On the ther hand it can be seen that in Extremadura, Asturias and the Basque cuntry the mix effect has been negative. D) The arrival f immigrants and the change they have prduced in the structure f the reginal ppulatin has als cntributed t an increase in emplyment in the Balearics, Murcia, the Canaries, Catalnia and Madrid. Equally the lack f immigratin t the Basque Cuntry, Extremadura and Asturias has impeded greater grwth in these regins. E) With respect t the unemplyed ppulatin, the regins with larger mix effects (Castile-La Mancha, the Canaries and the Balearics Ils.) are thse in which the arrival f immigrants has cntributed mre t an increase in unemplyment. The psitive effect here representing a negative aspect. On the ther hand, C. 13

14 Valencia, Catalnia and Madrid are the regins with the highest negative mix effect and s the regins where immigrants have had a greater impact n a reductin in unemplyment. Secndly, the same analysis has been carried ut, but taking int accunt nly the grwth f the ecnmic immigrant ppulatin. In this way the mix effect cmes clser t the implicatin that each regin has gt a different immigrant ppulatin structure depending n its rigins which has therefre evlved differently in each regin. In this case, each f the effects f the shift-share respnds t the fllwing interpretatin: - The natinal effect represents the change t the active ecnmic immigrant ppulatin, bth emplyed and unemplyed, f each regin, if within the regin they have grwn at the same pace as that f the ecnmic immigrant ppulatin at a natinal level. - The mix effect represents the change t the active immigrant ppulatin, bth emplyed and unemplyed, in each regin because said regin pssesses an immigrant ppulatin structure with rigins that are different t thse f the cuntry as a whle. This effect reflects the imprtance that an unequal distributin f ecnmic immigrants accrding t their rigins has. - And the reginal effect shws the change t the active ecnmic immigrant ppulatin, emplyed and unemplyed, due t the fact that the ttal immigrant ppulatin has increased mre in sme regins than thers. It is imprtant t emphasise the aim f crrectly interpreting the results, which unlike the previus shift-share, where the breakdwn was cncerned with the activities, emplyment and unemplyment f bth the native and immigrant ppulatin f the regin, this secnd definitin f the analysis is limited t explaining the evlutin f the basic labur indicatrs fr the ecnmic immigrant ppulatin. In table 6 the results f the secnd breakdwn are presented as percentage pints. At a general level the new data shws that reginal differences are nw mre marked. 14

15 Table 6. Breakdwn f the percentage f active, emplyed and unemplyed, ecnmic immigrant ppulatin f each regin. Surce: Authrs. Frm EPA (INE), Percentage f grwth f the active ppulatin Natinal Mix Reginal Effect effect effect Percentage f grwth f the emplyed ppulatin Natinal Mix Reginal Effect effect effect Percentage f grwth f the unemplyed ppulatin Natinal Mix Reginal Effect effect effect Andalusia Aragn Asturias Balearics Canaries Cantabria C-La Mancha C-Len Catalnia Extremadura Galicia La Rija Madrid Murcia Navarre Basque. C C. Valencia Ceuta & Melilla Nte: The majr effects f each regin have been highlighted. Nevertheless, in ur pinin the mst relevant results are the fllwing: a) Fr the active ppulatin, the natinal effect remains predminant, thugh the exceptins are high. Mix effect shw that the change t the immigrant structure accrding t their rigin is nt a determinant factr. The reginal effect is nly in the majrity in Cantabria, Extremadura, La Rija and Navarre. b) The results fr the emplyed ppulatin are identical: The natinal effect predminates except in the regins mentined previusly. The nly exceptin being La Rija. c) We fund the greatest differences amngst the unemplyed. The grwth in unemplyed immigrants is, in the cases f Aragn, the Balearics, Cantabria, Extremadura, La Rija, Murcia and Basque Cuntry due mainly t reginal issues. In the ther Spanish regins unemplyment amngst the immigrant ppulatin can be explained by the grwth f immigrant unemplyment natinally r as a cmbinatin f all the issues. T sum up, the mix effect des nt appear t be excessively relevant in explaining the grwth f the main labur variables cnsidered. The natinal behaviur is predminantly respnsible fr the grwth f the active and emplyed sectins f the ppulatin. The reginal effect is imprtant in a ntable number f regins, but nly t explain the grwth f the unemplyed ppulatin. The main effects f immigratin n reginal labur markets are due t the number f immigrants and their unequal distributin rather than the rigin f the immigrants 15

16 within each regin.the unequal distributin f the immigrants depending n their rigin is respnsible fr certain reginal differences but is nt a fundamental reasn fr differences in reginal labur markets. 4.2 Virtual settings: Immigratin and reginal cnvergence. Frm the previus results we have cnstructed a series f virtual settings t intentin f check the effect the immigrant ppulatin has n the labur parameters within the Spanish regins. Given previus calculatins in which we have registered the grwth f the active, emplyed and unemplyed ppulatins accrding t a series f effects, we can estimate the virtual grwth f said ppulatins withut taking int accunt any f these effects. In ur case, we are interested in eliminating the cntributin made by the mix effect, which reflects the cntributin made by the arrival f ecnmic immigrants t the grwth f the different wrking ppulatins under cnsideratin, just as the reginal differences that exist as far as their cmpsitin is cncerned. As a cnsequence we calculate a scenari fr the grwth f the active, emplyed and unemplyed ppulatins based nly n the natinal and reginal effect withut taking int accunt the mix effect. Given that we have defined tw shift-share analyses, ne fr the ecnmic immigrant ppulatin and the ther fr ppulatin as a whle, we can estimate tw alternative scenaris. In the first, we assume that the number f natives and ecnmic immigrants within the ppulatins is the same fr all regins. (The mix effect is zer in the first shift-share analysis). In this way we can see what the labur results wuld be if the immigrants were distributed equally amngst all the Spanish regins. In the secnd scenari, using the data frm the shift-share analysis carried ut fr the immigrant ppulatin alne, we cnsider a scenari in which the ecnmic immigrants f different rigins are distributed equally amngst the Spanish regins. Once again, we cnsider that the grwth f the active, emplyed and unemplyed, immigrant ppulatin is based n the natinal and reginal effect and that the mix effect is zer. a) Virtual activity, unemplyment and emplyment, grwth rates f the regins fr the ppulatin as a whle. Using the previus scenaris, we have btained virtual active, emplyed and unemplyed ppulatins and with them we have calculated the main labur rates, which we have als called virtual rates. Table 7 shws the virtual activity, emplyment and unemplyment rates fr the Spanish regins and the differences between them and the real rates; that is t say, thse that wuld exist if the ppulatin f native and ecnmic immigrants were f equal vlume. 16

17 Table 7. Virtual activity, emplyment and unemplyment rates with a zer mix effect. The ppulatin as a whle. Surce: Authrs. Frm EPA (INE). Rate f activity Virtual Rates Rate de unemplyment Rate f emplyment Rate f activity Differences ( Virtual-Real) Rate de unemplyment Rate f emplyment Andalusia Aragn Asturias Balearics Canaries Cantabria C-La Mancha C-Len Catalnia Extremadura Galicia La Rija Madrid Murcia Navarre Basque. C C. Valencia Ceuta & Melilla Nte: The psitive differences have been highlighted. Accrding t the data btained, if the ppulatins had been f the same cmpsitin; that is t say, if the arrival f the ecnmic immigrants had been hmgenus, the activity and emplyment rates wuld have been larger and the unemplyment rates wuld have been mre similar in mst f the Spanish regins. Furthermre, the regins with lwer immigratin rates wuld have reached higher rates f activity and emplyment and lwer rates f unemplyment. Regarding the activity rate, the regins f Extremadura, Asturias and the Basque Cuntry stand ut as regins that culd have btained much higher activity rates had they received mre immigrants. On the ther hand, Murcia, the Balearics, Madrid and Catalnia present negative differences. S their activity rates wuld have been much lwer had it nt been fr the in-flw f immigrants t these regins. Regarding the unemplyment rate, Murcia and Andalusia are the regins that present the greatest differences. Had they received the same number f immigrants as the ther regins, their unemplyment rates wuld have been lwer. Lastly, it must be emphasized that, Galicia, the Basque Cuntry and Cantabria wuld have btained higher emplyment rates if the ecnmic immigrants in their ppulatin structure had carried mre weight. T sum up, lking at the results, we cme t the cnclusin that if the regins with mre/less immigratin had the same rates f immigratin as thse registered natinally, they wuld see their activity and emplyment rates rise/fall accrdingly as wuld their unemplyment rates. Therefre, immigratin culd cmpensate fr reginal differences, 17

18 but nly if it is distributed hmgenusly amngst the regins. As this is nt happening at the mment, it can be cncluded that the unequal distributin f immigrants is, if nt increasing inequality amngst the different regins, at least nt helping t reduce it. b) Virtual activity, emplyment and unemplyment grwth rates f the regins fr the ecnmic immigrant ppulatin. In this sectin we repeat the previus scenari s as t take int accunt the results frm the secnd shift-share analysis. In this way, we limit the analysis t the reginal grwth f the ecnmic immigrant ppulatin. Therefre, we analyse the principal labur rates fr the ecnmic immigrant ppulatin within each Spanish regin. As in the previus case, we have cnstructed virtually what the labur rates wuld be if the ppulatin structure f the ecnmic immigrants residing within each regin were nt differentiated by rigin. That is t say, assuming that the mix effect is zer, we frce the ecnmic immigrants t distribute hmgenusly thrughut the Spanish regins accrding t their rigin. The difference between the real rates and the virtual nes demnstrate the fact that the regins present greater r lesser immigrant ppulatins with higher r lwer activity, emplyment and unemplyment grwth rates. This highlights the imprtance f the differences in the cmpsitin f the immigrant ppulatins in the Spanish regins. Table 8 cntains these calculatins. Table 8. Virtual activity, emplyment and unemplyment rate with a zer mix effect. Ecnmic immigrant ppulatin. Surce: Authrs, frm EPA data (INE), 2 nd. quarter Activity rate Virtual rates Unemplyment rate Emplyment rate Activity rate Difference ( Virtual-Real) Unemplyment rate Emplyment rate Andalusia Aragn Asturias Balearics Canaries Cantabria C-La Mancha C-Len Catalnia Extremadura Galicia La Rija Madrid Murcia Navarre Basque. C C. Valencia Ceuta & Melilla

19 Firstly, in relatin t the activity rate, it can be seen that if the regins did nt differentiate their immigrant ppulatin by rigin the majrity f them wuld ntice a reductin in the value f their participatin in the labur frce. This is true fr all the regins except Asturias and Murcia where the activity rates wuld increase. Furthermre the unequal natinal mix f the immigrants has had imprtant cnsequences n the grwth f unemplyment rates amngst the immigrant ppulatin in the majrity f the Spanish regins. The immigrant unemplyment rates wuld be higher in thse regins with very few immigrants, cmpensating in this way fr the reginal differences. Results fr the emplyment rates are similar t thse btained fr the activity rates, thugh with a higher degree f hetergeneity. In the majrity f regins the results btained shw that if the immigrants were distributed mre hmgenusly accrding t rigin, there wuld be a greater similarity in the different reginal emplyment rates fr this sectin f the ppulatin, thus reducing the reginal differences amngst immigrants. Therefre the unequal distributin f immigrants accrding t rigin affects the labur results they btain. Hwever, as was demnstrated earlier, the mst relevant factr in explaining the reginal differences in the labur market is the unequal glbal distributin f the ecnmic immigrants. c) Cnvergence equatins and the speed f apprach derived frm the virtual settings. Finally, t determine the imprtance f the the arrival f immigrants and their unequal gegraphic distributin n the cnvergence f the reginal labur markets, we have reestimated the cnvergence equatins using the virtual data btained in the first shiftshare analysis. Given that the analyses carried ut previusly have shwn us that the effect f immigratin n the reginal labur markets is based principally n its unequal cncentratin rather than n its unequal distributin accrding t rigin. We have calculated the beta-cnvergence between the 2000 and 2005, taken fr the last year thse activity, emplyment and unemplyment rates that wuld be btained fr the ppulatin as a whle if the mix effect were null; that is t say, if the immigrant ppulatin had grwn at the same rate in each and every Spanish regin. The results btained shw us whether the unequal arrival f immigrants has benefited the cnvergence f the Spanish regins in labur terms, r nt. Table 9 shws the results btained. 19

20 Table 9. Cnvergence regressins taking int accunt the virtual labur rates, assuming that the mix effect is zer Activity rate Unemplyment rate Emplyment rate Initial level (-6.580)** (-7.195)** (-6.462)** Cnstant (6.692)** (7.206)** (6.565)** R squared Speed f cnvergence Time needed t 18.5 (quarters) 15.4 (quarters) 16.6 (quarters) reduce the 5 (years) 4 (years) 4 (years) difference by half Number f bservatins: 17. The values f the t-student are in brackets. ** Significant under a prbability f 95 %. Results shw that cnvergence (r the apprach between the principal labur rates - activity, emplyment and unemplyment-) f the Spanish regins wuld have been much higher if the arrival and distributin f the immigrants had been mre hmgenus. Remving the mix effect, the speed f cnvergence increases cnsiderably and the time needed t reduce by half the present reginal differential is estimated t be arund 4 r 5 years depending n the rate under cnsideratin; much lwer than previus figures, 22 years fr the emplyment rate and 11 years fr the unemplyment rate. Besides this, cnvergence in terms f unemplyment rates wuld be very high. Given the extensive reginal inequalities that exist in Spain regarding the distributin f unemplyment, this result is extremely imprtant. These results are in agreement with the data bserved previusly. The arrival f immigrants is greater in thse regins with mre favurable r dynamic labur markets; areas with greater emplyment pprtunities, fr example the regins f Madrid and Catalnia, which present elevated immigratin rates and where the pssibilities f getting emplyment are easiest. Given that the immigrants are at the same time a mre active labur grup (they usually present activity and emplyment rates higher than natinals 14 ), there arrival cntributes twards making the reginal labur market they enter mre dynamic. Therefre, due as much t the cncentratin f the immigrants in thse regins with a mre dynamic labur market, as t the existence f a higher level f labur activity amng the immigrants, it leads us t cnclude that the arrival f immigrants cntributes t an increase in the labur differences that exist between the Spanish regins, and helps t explain the limited labur cnvergence prcesses between the regins. 5. Cnclusins. The purpse f this paper was t ascertain whether the migratry flws t Spain and their uneven gegraphical distributin have had effects n the reginal labur markets 13 Althugh methdlgical changes f the EPA affect the tempral series f the activity, emplyment and unemplyment rates, estimatin f these regressins has been dne maintaining the earlier assumptins in s far as said methdlgical changes affect all the regins in the same way and s shuld nt be relevant in the analysis f the reginal distributin r f the reginal cnvergence. 14 In this respect see Iglesias, C. y Llrente, R (2006). 20

21 by reducing r increasing the differences amng them. Althugh a slight prcess f cnvergence starting frm the years 2000 and 2001 was bserved, the existence f high hetergeneity amng the reginal labur markets was disclsed, especially sharp in the case f unemplyment. The shift-share analyses shwed that the arrival f immigrants has had imprtant cnsequences in the evlutin f the main labur rates, cntributing t the increase in the activity and emplyment rates and t the reductin f the unemplyment rate. The shift-share analysis f the immigrant ppulatin f an ecnmic nature shwed that its uneven gegraphical distributin by rigin did nt have imprtant labur results, except in the case f unemplyment. The reginal differences between the unemplyment rates f immigrants wuld be smaller if they were mre unifrmly distributed with respect t their gegraphical rigin. The cnstructin f virtual scenaris disclsed the great imprtance f immigrants in the present rates f activity, emplyment and unemplyment. Regins with lwer rates f immigratin wuld get better labur results if they received the same migratry flws as all the ther regins. Therefre, immigratin culd ffset reginal differences if their patterns f territrial lcatin were different. A fundamental cnclusin is that immigrants tend t cncentrate in the regins with better labur cnditins, and there are fewer f them in the regins with wrse placed jb markets in the cntext f the cuntry as a whle. Mrever, althugh in aggregate terms the immigrants increase the activity and emplyment rates and reduce the unemplyment rate, these results are intensified in the regins with the best jb markets and mderated in the wrse placed regins. Thus, the present gegraphical distributin f immigratin and its differing labur perfrmance have ultimately cntributed t increasing the reginal differences. The Spanish regins with highest immigratin rates are thse that have the mst dynamic jb markets, and since the arrival f immigrants adds dynamism t the market which they enter, the verall effect has been t increase the reginal labur differences. The ppsite pattern can be seen fr the less favured regins. The freging bservatins were cnfirmed by the cnvergence analysis perfrmed using the virtual rates. The apprximatin f Spanish regins in labur terms wuld have been much greater had the immigrant presence in the ppulatin structure in each regin been similar. Therefre, the cnvergence achieved in recent years amng Spanish regins was nt driven by the arrival f immigrants. Hence, in rder t establish what the determining factrs f cnvergence wuld be, it wuld be necessary t analyse anther series f issues such as the evlutin f labur prductivity, the evlutin f wage differences r the inter-reginal mbility f wrkers. 21