Uni e si y o he Basque Coun y UPV/EHU
Mas e in Economics: Empi ical Applica ions and
Policies
Te i o ial Dis ibu ion o
Immig an s in Eu ope
Au ho : Aie Gu u xaga U zaa
Supe iso s: Ma ía José Gu ié ez & Susan O be
2019/20
Index
Abs ac ......................................................................................................................................... 4
1.- In oduc ion ............................................................................................................................. 5
2.- Da a ........................................................................................................................................ 12
3.- Seg ega ion me ics ............................................................................................................... 12
3.1.- Indices ............................................................................................................................. 12
3.2.- Resul s ............................................................................................................................ 17
4.- De e minan s o seg ega ion me ics ................................................................................... 23
5.- Conclusions ............................................................................................................................ 40
6.- Bibliog aphy ........................................................................................................................... 42
Figu es Index
Figu e 1: Example o he Kuzne s cu e be ween seg ega ion and GDP ...................................... 6
Figu e 2: Pe cen age o immig an popula ion a egional le el (NUTS2), 1999-2019 ................. 9
Figu e 3: Social ne wo k o a seconda y educa ion school. Sou ce: “Ne wo ks, C owds and
Ma ke s. Reasoning abou a highly connec ed wo ld” ............................................................... 11
Figu e 4:E olu ion o dissimila i y, en opy and A kinson 0.5 indices a Eu opean le el, 1999-
2019 ............................................................................................................................................. 18
Figu e 5:Dissimila i y index o each coun y, 1999-2019 .......................................................... 19
Figu e 6: Dissimila i y index o Romania, Spain and I eland, 1999-2019 .................................. 20
Figu e 7: A kinson 0.5 index o Romania, Spain and I eland, 1999-2019 .................................. 21
Figu e 8: En opy index o Romania, Spain and I eland, 1999-2019 ......................................... 22
Figu e 9:E olu ion o di e en mac oeconomic a iables, 1999-2019 ...................................... 25
Figu e 10: E olu ion o densi y, 1999-2019 ................................................................................ 25
Tables Index
Table 1: Top 5 coun ies wi h he highes pe cen age o immig an popula ion in EU-15, 1999-
2019 ............................................................................................................................................... 7
Table 2: Top 5 egions wi h he highes pe cen age o immig an popula ion in EU-15, 1999-
2019 ............................................................................................................................................... 8
Table 3:De ini ion and o mula o dissimila i y, en opy and A kinson indices .......................... 16
Table 4: Mac oeconomic a iables and hei measu e ............................................................... 24
Table 5: Es ima ion o he empi ical model o he dissimila i y index ...................................... 27
Table 6. Econome ic model o dissimila i y index ..................................................................... 29
Table 7: Es ima ion o he empi ical model o he A kinson 0.5 index .................................... 32
Table 8: Econome ical model o he A kinson 0.5 index ........................................................... 34
Table 9: Es ima ion o he empi ical model o he en opy index ........................................... 36
Table 10: Econome ic model o en opy index .......................................................................... 38
Table 11: Dissimila i y, en opy and A kinson 0.5 indices' alue and numbe index a Eu opean
le el, 1999-2019 .......................................................................................................................... 44
Abs ac
Une en dis ibu ion o immig an popula ion has di e se consequences in
coun ies and socie ies. We call immig an s hose ha ha e a na ionali y di e en
o he epo ing coun y. Some Eu opean egions ha e an impo an pe cen age
o immig an popula ion. This a icles aims o quan i y his une enness in he
dis ibu ion om 1999 o 2019. Fo his we use seg ega ion indices o he
e enness dimension. The h ee indices we use show ha immig an popula ion
is une enly dis ibu ed a Eu opean le el and a coun y le el. Howe e , his
he e ogenei y has dec eased. We also look o mac oeconomic de e minan s o
hese indices. We ind a e y in e es ing ela ion be ween GDP and he indices,
as he e is Kuzne s e ec . Being awa e ha he e a e inequali ies and how
di e en socioeconomic a iables can a ec is help ul in o de o adop success ul
policies o enjoy he posi i e e ec s immig a ion can p o ide.
Keywo ds: NUTS2 egions; Immig an s; Seg ega ion indices; Inequali y
measu es; Kuzne s e ec
5
1.- In oduc ion
The aim o his a icle is o s udy he e i o ial dis ibu ion o immig an s ac oss
Eu ope in ecen yea s. This is a e y impo an socioeconomic subjec in
nowadays Eu ope, pa ly because, acco ding o Sa dad a & Rocha (2016), o e
he pas qua e cen u y, Eu ope has expe ienced h ee impo an e en s which
eased mig a ion wi hin he con inen : he dissolu ion o he Comecon, he T ea y
on Eu opean Union and he Eu opean Union accession o new coun ies.
One o he easons o why popula ion exoduses a e eally impo an phenomena
is ha hey ha e a huge impac in he des ina ion coun y. People lea e hei
coun y o o igin o many easons, such as imp o ing li ing condi ion, o , simply,
o escape social dis ess (Jones, 2015), o om na u al disas e s (López-Ca ,
Ma e -Kenyon, 2015) o wa . Once hey make he decision o lea e, hey sea ch
o wha migh be he bes des ina ion. Mac oeconomic a iables o he
des ina ion coun y, such as, employmen , educa ion le el o how ich he coun y
is may a ec he immig an ’s selec ion. On he o he hand, immig an s also a ec
des ina ion coun ies in se e al dimensions, o example, hey inc ease he
numbe o a ailable labo , p opelling he economic g ow h.
Howe e , we can see ha o mig an popula ion, o a leas o hose wi h some
non-na i e cha ac e is ics is much mo e di icul o ob ain ce ain jobs, as explain
Sche e and Slaugh e (2001) and Mayda (2006)
.
Ano he example is he e ec
hey ha e on demog aphy. Usually immig an s a e young and end up con ibu ing
o inc easing bi h a es in des ina ion coun ies. An example o nega i e e ec
could be, o example, ha he economic gap be ween he o igin and des iny
coun ies can inc ease, especially i hose lea ing he coun y a e he mos
quali ied ones. Also a pa o he popula ion does no like o eigne s coming o
hei coun y and ha ing jobs, hey claim ha hese indi iduals s eal he jobs
(Sche e & Slaugh e , 2001; Mayda, 2006) and ha hey a e c iminals. The u h
is ha hey ha e less job oppo uni ies, as said be o e. As a consequence o his
s a emen s made by many populis poli ical pa ies, in some coun ies wi h a high
sha e o o eigne s he e has been a ise in he acis and xenophobic beha iou s
and ul a- igh is poli ical pa ies’ powe . In he second pa o he a icle we ocus
6
on how some hese mac oeconomic a iables a ec di e en seg ega ion
indices.
The mos impo an hing we wan o s udy when sea ching o de e minan s o
hese seg ega ion me ics is he possible e ec o he G oss Domes ic P oduc .
One o he objec i es will be o see i he e is a Kuzne s cu e be ween he GDP
and di e en seg ega ion indices used. I his esul is ob ained we will see ha
he seg ega ion o immig an s inc eases as coun ies become iche , bu only
un il a poin , once he coun y has a pe capi a GDP highe han ha , he ela ion
be ween indices and GDP will be nega i e. We can see his e olu ion on Figu e
1.
Figu e 1: Example o he Kuzne s cu e be ween seg ega ion and GDP
I is because o seg ega ions impo ance, in all aspec s, and because “mig a ion
is a ubiqui ous phenomenon” (Mazzoli e al, 2020) ha he e a e se e al wo ks in
ela ion o seg ega ion and di e en ways o s udying i . The impo ance o he
subjec has an inc easing endency, and e en mo e wi h he possible c isis ha
is coming. Acco ding o Eu os a , a ound 9.29% o he popula ion o he Eu opean
Union (15 coun ies, EU-15) had a di e en na ionali y o he espec i e coun y
in 2019; his numbe doubles ha o 1999 (4.5%). This shows ha he weigh o
non-na ional popula ion has inc eased in Eu ope du ing hese wo decades.
Howe e , i mus be aken in o accoun ha hese igu es may no ep esen he
7
eal numbe s o immig an s o se e al easons. Among o he s, illegal immig an s
a e di icul o be eco ded, as hey do no always appea in any egis e .
Table 1: Top 5 coun ies wi h he highes pe cen age o immig an popula ion in EU-15, 1999-2019
Table 1 shows he i e coun ies o EU-15 wi h he highes pe cen age o
immig an popula ion o yea s 1999 and 2019. We can clea ly see ha ; like
Czaika and Di Lillo (2018) poin ou immig an s a e making up a con inuously
g owing p opo ion o he Eu opean popula ion. Fo example, Belgium is he
coun y wi h he highes pe cen age o immig an popula ion in 1999, bu s ill has
a lowe pe cen age han he i h coun y in 2019, shows his. Du ing hese wen y
yea s he ela i e impo ance o F ance and Sweden in his aspec has
dec eased, since hey do no appea in he op i e o 2019, being subs i u ed by
Spain and I eland. As p e iously said, in some coun ies ul a- igh mo emen s
ha e inc eased oge he wi h he inc ease o he pe cen age o immig an
popula ion. This is he case o Aus ia, whe e we can see ha he pe cen age has
nea ly doubled du ing he ime pe iod s udied, Aus ia is one o he coun ies
whe e he a - igh mo emen s mo e ha e inc eased oo, ha ing he F eedom
Pa y Aus ia (FPÖ) a high pe cen age o o es. This phenomenon is no
exclusi e o Aus ia, i has been obse ed in coun ies like F ance and G eece.
Howe e , his da a is a coun y le el, so di e ences in in e nal le el a e pe ec ly
possible, as we see in he nex able.
Top 5coun ies in EU-15
1999 2019
Belgium 7.39%
12.89%
Aus ia
Ge many 6.74%
11.36%
I eland
Aus ia 6.48%
10.57%
Ge many
F ance 4.63%
9.05%
Spain
Sweden 3.63%
8.89%
Belgium
8
Top 5 egions in
EU-15
1999
2019
Coun y
Regions
%
%
Regions
Coun y
Belgium
Région de B uxelles-
Capi ale / B ussels
Hoo ds edelijk Gewes
26.98%
33.74%
Comunidad de Mad id
Spain
Uni ed
Kingdom
Inne London
21.32%
33.31%
Région de B uxelles-
Capi ale / B ussels
Hoo ds edelijk Gewes
Belgium
F ance
Co se
16.86%
31.29%
Région lémanique
Swi ze land
Aus ia
Wien
15.58%
29.06%
Wien
Aus ia
Belgium
P o . Hainau
14.88%
27.57%
Ticino
Swi ze land
Table 2: Top 5 egions wi h he highes pe cen age o immig an popula ion in EU-15, 1999-2019
In Table 2 we see he op 5 egions om EU-15. We obse e ha he e a e
di e ences be ween coun ies and egions; o example, he second egion wi h
he highes pe cen age o immig an s is Inne London, bu Uni ed Kingdom does
no appea in Table 1. In 2019 he same happens, Swiss egions o Ticino and
Région Iémanique. We also obse e ha Comunidad de Mad id is he egion wi h
he highes pe cen age o immi an s in 2019, being Spain he second coun y in
his yea . Table 2 shows he ac ha coun ies change when aking a look a
egions. Apa om ha , Table 2, like happened wi h Table 1, shows ha he e
has been an inc ease in he weigh o immig an popula ion. Like happened in
Table 1, he i s egion in 1999 has a lowe pe cen age o immig an s han he
i h egion in 2019.
9
Figu e 2: Pe cen age o immig an popula ion a egional le el (NUTS2), 1999-2019
Figu e 2 shows he e olu ion and dis ibu ion o he immig an s a egional le el
(NUTS2). Fo his we ha e maps o 1999 and 2019. The maps show he
pe cen age o popula ion wi h a na ionali y di e en o he one o he epo ing
coun y. We can obse e ha he immig an s a e he e ogeneously di ided ac oss
coun ies, and he same happens inside he coun ies, as Table 1 and Table 2
showed. In he case o hese maps we ha e in e als o pe cen age o
immig an s. Those a eas wi h he ligh g een ep esen coun ies whe e he
immig an popula ion is i e pe cen o less o o al popula ion. A eas ge s da ke
when he e is mo e pe cen age o immig an s. The second ligh es g een shows
coun ies whe e immig an popula ion is be ween i e and en pe cen o o al
popula ion. The second da kes g een a e coun ies whe e he pe cen age o
immig an popula ion is be ween i een and wen y pe cen . Las ly we ha e he
da k g een, which ep esen s coun ies whe e immig an popula ion is abo e
wen y pe cen .
No ice ha da a in 2019 co e s mo e coun ies han da a om 1999, we ha e
coun ies such as I aly o Swi ze land, o which we ha e no da a in 1999. We
also obse e a highe immig an concen a ion in 1999. Fo example, we see ha
in Spain he e is no egion ha s ands ou mo e han he es ; howe e , his can
be because o he in e als selec ed o colou he map. We can also see ha in
2019 coun ies, and especially egions, a e much mo e di e en ia ed han in
1999.
1999 2019
16
Table 3:De ini ion and o mula o dissimila i y, en opy and A kinson indices
DEFINITION / CHARACTERISTICS
FORMULA
Dissimila i y Index o Del a (D)
The mos widely used index. Measu es
he pe cen age o he g oup’s popula ion
ha would ha e o change esidence o
each egion o ha e he same pe cen age
o he g oups as he coun y o e all.
∑[|(−)|]
[2(1−)]
En opy Index o In o ma ion Index (H)
I measu es he weigh ed a e age
de ia ion o each egion om he coun y
di e si y. I is bigge when each g oup is
equally ep esen ed in he a ea.
(
−
)
= 1
!+(1−) 1
1−!
=
1
!
+
(
1
−
)
1
1
−
!
A kinson Index (A)
Allows o weigh egions a di e en poin s
on he dis ibu ion. Pe mi ing ha whe e
mino i ies a e unde o o e - ep esen ed
con ibu e in a mo e impo an way o he
o e all index.
1−
1−!&(1−)'((
&
'(
17
These indexes sa is y some o he ou c i e ia ha we e es ablished by James
and Taeube (1985). These p ope ies a e cha ac e is ics o ha e a good index;
his is, indices ha ul il he ou p inciples a e good indices. This a e he ans e
p inciple, which says ha a measu e should be sensi i e o edis ibu ion o he
immig an popula ion among coun ies wi h immig an p opo ions abo e o below
he egion’s. The second c i e ion is he composi ional in a iance, so he ela i e
size o he immig an popula ion should no a ec he index. Thi d we ha e he
size in a iance, which s a es ha he measu e should no be a ec ed i he
numbe o immig an and non-immig an popula ion is mul iplied by a cons an .
And ou h, he o ganiza ional equi alence, which means ha he index should
be una ec ed by agg ega ing uni s wi h he same mino i y composi ion. The h ee
indices ha we s udy do no ul il all he p inciples. In he case o he dissimila i y
index, i ails o sa is y he ans e p inciple. On he o he hand, he same
happens wi h he composi ional in a iance in he case o he en opy index. The
A kinson index is he only index, ou o hose we use, ha sa is ies wi hou any
p oblem he ou c i e ia.
Massey & Den on wo ked wi h a o al o 20 indices, bu he educed he lis o
i e, jus one index o each one o he dimensions. So, o summa ize, in he
e enness case, ha is he one in which we a e in e es ed, he bes op ion is he
dissimila i y index by Duncan and Duncan as i con ains mos in o ma ion ha he
o he indices can p o ide. Apa om ha , he e a e con inui y and ease easons,
as mos o p e iously w i en li e a u e uses his index. As said we will be using
wo mo e indices, he en opy index, which is an index ha wo ks well oo, and i
can be di ided o see he weigh o each one o he g oups analysed. This
decomposi ion migh be use ul o see o which ex en he inequali ies a e
gene a ed wi hin g oups and o which ex en a e caused because o di e ences
be ween g oups. Howe e , we do no use ha p ope y in his pape ; and he
A kinson index, which is also a g ea choice, since i ul ils he ou p inciples
p e iously explained. Tha is why we decide o ocus on hese h ee indices.
3.2.- Resul s
We a e no in e es ed only in he esul s ob ained by using he indices, we also
wan o obse e he e olu ion hey had du ing he ime pe iod s udied, o see how
hey all e ol e in he same way, e en i hey make he measu es in di e en ways.
18
We calcula e he uni o mi y in he dis ibu ion o immig an s in wo le els. Fi s o
all, we measu e i a Eu opean le el, using e e y NUTS2 egion om ou coun y
sample. Second, we measu e i a coun y le el, using e e y NUTS2 egion o
each coun y. This analysis is made o each one o he yea s o ou sample,
1999-2019.
Figu e 4:E olu ion o dissimila i y, en opy and A kinson 0.5 indices a Eu opean le el, 1999-2019
Figu e 4 shows he e olu ion o he dissimila i y index, he en opy index and he
A kinson index wi h 0.5 pa ame e o Eu ope. In his case, in which we s udy he
indices a Eu opean le el, we analyse Eu ope as i i was a coun y as a whole,
wi h all he NUTS2 egions aken inside. In o de o see esul s in a clea e way,
he alues ob ained ha e been no malized, using 1999 as he base yea . I can
be obse ed ha he h ee indices ha e a simila e olu ion, and, in gene al, a
nega i e end. Wha means ha he immig an popula ion is mo e e enly
dis ibu ed in 2019 han in 1999. These esul s co obo a e he in ui ion gene a ed
om he i s glance. This is, he in ui ion ha seg ega ion dec eases wi h ime.
This could happen because, wi h he passage o ime, immig an s y o mo e o
coun ies whe e wo k possibili ies exis bu whe e he i s wa es o mig a ion did
no a i e. We obse e an inc ease in he indices’ alue in he yea s 2004 and
2005; his may happen because we ob ain da a o some coun ies a i s ime in
80
85
90
95
100
105
110
115
120
Eu ope
Dissimila i y En opy A kinson 0.5
19
his pe iod. Examples o hese coun ies a e I aly o Romania, which ha e no
da a un il hese yea s.
I we ake a look o he Eu opean indices o dissimila i y, we can see ha he
dissimila i y index is he one ha shows highes alues un il 2003. A e his yea
i is he one below o e e y o he yea . I is also he index wi h he mos uni o m
end. Wi h his in o ma ion we can say ha , acco ding o he dissimila i y index,
he seg ega ion has been educed o e ime. The o he wo indices will gi e a
simila esul , bu hey show a smalle educ ion o he seg ega ion han he
dissimila i y index, since du ing hese wo decades he alue o he dissimila i y
index has educed in 18.1% agains he educ ion o 16.35% and 16.63% o he
en opy and A kinson indices espec i ely.
Figu e 5:Dissimila i y index o each coun y, 1999-2019
Fi s o all, we calcula e he dissimila i y index, which is he mos widely one used.
Figu e 5 shows he index o he 1999-2019 pe iod o he Eu opean coun ies o
which we ha e a ailable da a. We can see ha he alues a e below 0.4 in mos
cases. The highes alue is ob ained by Romania, being a ound 0.55. A he same
ime, i e idences di e ences in seg ega ion ac oss coun ies; which mean ha
inside each coun y “immig an s” a e no homogeneously dis ibu ed. We also
obse e ha he e olu ion o he index changes be ween coun ies. Howe e ,
du ing hese wo decades he alue o he index has dec eased in mos o hese
0
0,1
0,2
0,3
0,4
0,5
0,6
199920002001200220032004200520062007200820092010201120122013201420152016201720182019
Dissimila i y index o each coun y
Belgium Bulga ia Czechia Ge many
I eland G eece Spain F ance
I aly Ne he lands Aus ia Po ugal
Romania Slo enia Slo akia Finland
Sweden Uni d Kingdom No way Swi ze land
20
coun ies, which means ha he e ogenei y in he e i o ial dis ibu ion has
dec eased acco ding o his index. In his Figu e 5 we obse e ha some
coun ies a e missing, his happens because coun ies such as Luxembou g o
Mal a ha e only one egions, and, as a consequence, he index alue hey ob ain
is always ze o.
Figu e 6: Dissimila i y index o Romania, Spain and I eland, 1999-2019
Figu e 6 shows he e olu ion o he dissimila i y index in he h ee chosen
coun ies. In his case we ha e chosen Romania (blue line), Spain ( ed line) and
I eland (g een line). We can see ha , o example, in Romania immig an s a e
mo e une enly dis ibu ed han in Spain o I eland, ha is all i shows, i has
no hing o do wi h he numbe o immig an s. In he case o Romania, we see an
i egula inc ease. I is he coun y ha eaches he highes alues and i ends
being he coun y wi h he highes immig an concen a ion. In he case o Spain,
we see alues ha show ha i is nei he he coun y whe e he immig an s mo e
seg ega ed a e no he coun y wi h he mos homogeneous geog aphical
dis ibu ion. We also see ha i has a ela i ely la cu e, which means ha in
e ms o concen a ion i has no changed oo much. Las we ha e he case o
I eland, whe e we obse e an inc ease in he une enness dis ibu ion o
0
0,1
0,2
0,3
0,4
0,5
0,6
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Dissimila i y index Romania, Spain and I eland
Belgium Bulga ia Czechia Ge many I eland
G eece Spain F ance I aly Ne he lands
Aus ia Po ugal Romania Slo enia Slo akia
Finland Sweden Uni d Kingdom No way Swi ze land
21
immig an s. Du ing mos pa o he pe iod is he coun y analysed wi h he lowes
index alue. The e can be many easons o make his happen, o example, i
could be ha mos immig an s in Romania a e he e o wo k easons and ha
speci ic ypes o jobs a e concen a ed, causing an i egula dis ibu ion o hese
immig an s.
Figu e 7: A kinson 0.5 index o Romania, Spain and I eland, 1999-2019
In Figu e 7 we ha e he same analysis o he A kinson index. This index allows
us o choose he alue o he pa ame e b. Figu e 7 shows he calcula ed index
o b=0.5. This is, as said by Iceland e al. (2002), he index will be mo e sensi i e
o changes in he middle pa o he dis ibu ion, when unde ep esen ed and
o e ep esen ed a eas con ibu e equally. In his case we ob ain lowe alues
han in he dissimila i y case. Romania has he peak alue oo, and also he
highes alue o 2019, while Sweden has he lowes alue o his yea . We can
see ha he e olu ion o di e en coun ies is i ually he same ha wha we ha e
jus seen wi h he dissimila i y index. I we ake a look o he same h ee coun ies
p e iously p ojec ed, we see in a clea way ha esul s a e, al hough
quan i a i ely di e en , quali a i ely simila .
0
0,05
0,1
0,15
0,2
0,25
0,3
0,35
0,4
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
A kinson index 0.5 o Romania, Spain and I eland
Belgium Bulga ia Czechia Ge many I eland
G eece Spain F ance I aly Ne he lands
Aus ia Po ugal Romania Slo enia Slo akia
inland Sweden Uni ed Kingdom No way Swi ze land
22
Figu e 8: En opy index o Romania, Spain and I eland, 1999-2019
Las ly, we ha e he en opy index, also known as he in o ma ion index. As we
can see in Figu e 8, he alues ob ained a e di e en o he ones ob ained abo e,
since hey a e calcula ed in a di e en way. Howe e , we see ha in all cases he
alues a e no eally high o any coun y. Once again we obse e simila
quan i a i e esul s han wi h he o he indices used. In he h ee cases we
obse e an i egula inc ease o Romania, a p e y simila end in Spain and low
alues wi h a li le inc ease in I eland.
To summa ize, and as a conclusion o his i s pa , when analysing a Eu opean
le el, we can see ha , in gene al, he e olu ion du ing he pe iod is e y simila
o he di e en indices used. Taking a look o he alues o he calcula ed
seg ega ion indices p esen ed in Table 11 in appendix, we can see ha
seg ega ion has dec eased a Eu opean le el be ween 1999 and 2019. Howe e ,
wi h each index he in ensi y o his end is di e en . This s eng h is bigge in
he case o he dissimila i y index, whe e he alue o he index educes in mo e
o less 18% while in he case o he o he wo indices we can see ha he alue
is a ound 16% smalle in 2019 han in 1999.
0
0,02
0,04
0,06
0,08
0,1
0,12
0,14
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
En opy index o Romania, Spain and F ance
Belgium Bulga ia Czechia Ge many I eland
G eece Spain F ance I aly Ne he lands
Aus ia Po ugal Romania Slo enia Slo akia
Finland Sweden Uni ed Kingdom No way Swi ze land
23
In he case o he coun y le el analysis, we see ha e en i he alues ob ained
wi h each index a e di e en , hey show i ually he same. Da a shows
he e ogenei y a coun y le el. Mos o he coun ies show a dec easing end
h oughou he yea s, ul illing he expec a ion in his case oo. In some coun ies
immig an s a e much mo e concen a ed, o example in Romania and I aly, while
in o he coun ies he dis ibu ion is nea ly uni o m, he cases o I eland and
Sweden. So wi h all he h ee indices we each he conclusion ha immig an s
a e geog aphically une enly dis ibu ed. Fo example, immig an popula ion is
mo e homogeneously dis ibu ed in he e i o y in I eland han in Romania.
Howe e , he mos impo an and ema kable poin is ha he esul s a e
quali a i ely simila e en i hey a e quan i a i ely di e en .
4.- De e minan s o seg ega ion me ics
In his second pa o he pape we wan o see wha mac oeconomic a iables
ha e an e ec on he seg ega ion indices and how a ec hem. Fo his we use
all da a oge he aking in o accoun i s panel s uc u e. In o de o analyse he
e ec ha some mac oeconomic a iables ha e o e he seg ega ion indices. A
ixed e ec s model is speci ied,
)
,+
=,
+-.
,+
+/
,+
whe e he dependen a iable, Y
i, ,
is he seg ega ion index. We ha e an index o
each coun y and yea . Subsc ip i is used o indica e he coun y, and subindex
indica es he yea , we ha e 19 coun ies and 21 yea s in his case. In his pa
we wo k wi h 19 coun ies because we do no use coun ies wi h only one egion,
like Luxembou g o Es onia, because he alue o hei indices is always ze o. X
i,
is he eg esso . These eg esso s a e some mac oeconomic a iables ha can
be a p io i de e minan s o he seg ega ion. Las , u
i,
is he e o e m. Wi h hese
ixed coun y and ime e ec s, Ben i ogli and Pagano (1999) s a e ha egional
ne mig a ion is sensi i e o changes in income dispa i ies, bu un esponsi e o
changes in he ela i e unemploymen a es.
24
The mac oeconomic a iables we will be using a e: Unemploymen a e, Densi y,
Elde ly popula ion, R&D expendi u e, G oss Domes ic P oduc , Fo eigne s,
Fe ili y a e, New ci izenship, Educa ion and Go e nmen expendi u e. We can
see how each one is measu ed in Table 4.
Va iables Measu es
Unemploymen a e Pe cen age o labo o ce ha is
jobless
Densi y Popula ion pe squa e kilome e
Elde ly popula ion Pe cen age o popula ion wi h age
o e 65
R&D Expendi u e Pe cen age o g oss domes ic p oduc
used in R&D
GDP In housands pe capi a
Fo eigne s Pe cen age o non-na ional popula ion
Fe ili y a e A e age numbe o child en ha would
be bo n o a woman o e he li e ime.
New ci izenship The popula ion ha each yea has
ecei ed he na ionali y o he
epo ing coun y as pe cen age o
o al popula ion
Educa ion Adul popula ion wi h e ia y
educa ion as pe cen age o o al
popula ion
Go e nmen expendi u e Pe cen age o he G oss Domes ic
P oduc
Table 4: Mac oeconomic a iables and hei measu e
25
Figu e 9:E olu ion o di e en mac oeconomic a iables, 1999-2019
Figu e 10: E olu ion o densi y, 1999-2019
In Figu e 9 we can see he e olu ion o mos o hese mac oeconomic a iables.
Taking a look o Figu e 9, we can see h ee g oups. In he i s g oup we ha e
hose a iables ha ha e a li le change du ing his pe iod. In his i s g oup we
ha e “New ci izenship”, “R&D Expendi u e”, “Fe ili y a e” and “Unemploymen
a e”. Then we ha e a iables ha ha e inc eased h ough ime, mos ly in a
uni o m way. Examples o his a e “Fo eign popula ion”, “Elde ly popula ion” and
0
5
10
15
20
25
30
35
40
45
1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019
Mac oeconomic a iable
Pop.O e 65 Unemploymen R&DExpendi u e Fo eigne s
Fe ili y Ci izenship Educa ion GDP/1000
130
132
134
136
138
140
142
144
Densi y
32
Table 7: Es ima ion o he empi ical model o he A kinson 0.5 index
Model 1 Model 2
Coe icien
Signi icance
R
2
, e o c i e ia and p alue
Coe icien
Signi icance
R
2
, e o c i e ia and p alue
Elde ly popula ion 0.0005
R
2
= 0.787739
Schwa z c i e ia= -1,282.430
Akaike c i e ia= -1,397.298
Hannan-Quinn c i e ia= -
1,351.528
p- alue=1.25e-055
0.0004
R
2
= 0.789555
Schwa z c i e ia= -1,313.996
Akaike c i e ia= -1,425.710
Hannan-Quinn c i e ia= -1,381.234
p- alue=3.02e-057
Unemploymen a e -0.0009
-0.0013 *
GDP 0.0012
**
0.0011 **
GDP_sq -9.41e-06
**
-9.20e-06 **
R&D Expendi u e 0.0078
0.0085
Densi y -0.0018
***
-0.0018 ***
Fo eigne s -0.0043
**
-0.0042 **
Fe ili y a e 0.0088 0.0087
New ci izenship -0.0027
Go . Expendi u e 0.0002 0.0003
Educa ion 0.0017 0.0018
33
Model 3 Model 4
Coe icien
Signi icance
R
2
, e o c i e ia and p alue
Coe icien
Signi icance
R
2
, e o c i e ia and p alue
Elde ly popula ion
R
2
= 0.789536
Schwa z c i e ia= -1,319.817
Akaike c i e ia= -1,427.678
Hannan-Quinn c i e ia= -1,384.737
p- alue= 2.92e-055
R
2
= 0.789425
Schwa z c i e ia= -1,325.486
Akaike c i e ia= -1,429.495
Hannan-Quinn c i e ia= -1,388.087
p- alue=6.13e-056
Unemploymen a e -0.0013
*
-0.0014
**
GDP 0.0012
**
0.0013
***
GDP_sq -9.41e-06
**
-1.01e-05
***
R&D Expendi u e 0.0087
0.0094
*
Densi y -0.0019
***
-0.0019
***
Fo eigne s -0.0044
***
-0.0045
***
Fe ili y a e 0.0078
New ci izenship
Go . Expendi u e 0.0003
0.0003
Educa ion 0.0020 * 0.0020
*
34
Fixed e ec s, using 348 obse a ions
Dependen a iable: A kinson 0.5 Index
Beck-Ka z s anda d de ia ions
Coe icien -s a is ic
Cons an 0.3099 9.544 ***
Unemploymen
a e
-0.0013 -2.302 **
GDP 0.0013 3.510 ***
GDP_sq -1.07e-05 -3.686 ***
R&D Expendi u e 0.0099 1.897 *
Densi y -0.0019 -6.663 ***
Fo eigne s -0.0043 -3.444 ***
Educa ion 0.0019 1.869 *
R-squa e MCVF (LSDV)
0.789250
Log- e osimili ud 741.6032
Akaike c i e ia -1431.206
Schwa z c i e ia -1331.049
Hannan-Quinn c i e ia -1391.332
ho 0.450756
Du bin-Wa son 0.997583
Robus con as o di e en in e cep s in g oups -
Null hypo hesis: G oups ha e a common in e cep
Con as s a is ic: Welch F (18, 119.5) = 77.0527
wi h p alue = P (F (18, 119.5) > 77.0527) = 6.51e-057
Table 8: Econome ical model o he A kinson 0.5 index
As be o e, many speci ica ions ha e been es ima ed and compa ed in e ms o
e o c i e ia. The es ima ion esul s o he selec ed speci ica ion is shown in
Table 8. Some signi ican a iables a e amilia om he dissimila i y index
analysis: unemploymen a e, densi y and o eigne s. These a iables main ain
he same sign hey had be o e; his is; he h ee a iables a ec in a nega i e
way. So, i unemploymen inc eases in one pe cen , he immig an popula ion will
be mo e e enly dis ibu ed, as he alue o he index is lowe . In he same way, i
densi y inc eases, o he popula ion o o eigne s inc eases in one pe cen , he
immig an popula ion will be mo e homogeneously dis ibu ed in he e i o y. We
also see ha he adjus men is good, since 78.925% o he a iabili y o he
A kinson index wi h pa ame e 0.5 is explained by he a iabili y o hese
independen a iables. Like happened in he analysis o he dissimila i y index,
he log-likelihood ob ained by his model is he highes ou o he speci ica ions
used, also he alues o he c i e ia a e he lowes .
35
In his case we ha e new signi ican a iables, he “R&D Expendi u e” and
“Educa ion”. In bo h cases he e ec is posi i e. So, i he expendi u e made in
R&D, as pe cen age o GDP, inc eases in one pe cen , he es ima ed inc ease in
he expec ed seg ega ion index is 0.0099, so he dis ibu ion o immig an
popula ion will be mo e une en. This may be because Resea ch and
De elopmen may concen a e quali ied wo k. In he case o educa ion, i adul
popula ion wi h e ia y educa ion, as pe cen age o o al popula ion, inc eases in
one uni , he es ima ed inc ease in he expec ed seg ega ion index is 0.0019,
causing a mo e he e ogeneous dis ibu ion o immig an popula ion. This migh
be caused due o quali ied jobs equi ing his ype o educa ion being
geog aphically concen a ed.
I we ake a look o he GDP and i s’ squa e we can see ha he Kuzne s e ec
we saw in he dissimila i y index case is s ill p esen . So, hose coun ies wi h
GDP abo e minimum h eshold ha e a nega i e ela ion be ween GDP and he
index. This is, when GDP inc eases he alue o he index dec eases in hese
coun ies. Following he same p ocess p e iously done, we ge ha he minimum
h eshold GDP is, in his case, 22.067. This means ha he ela ion be ween
hese wo a iables will be nega i e when he GDP pe capi a is bigge han
22,067. This h eshold is smalle han in he p e ious case, so hose coun ies
plus some new will be included. These addi ional coun ies a e he Czech
Republic, Po ugal and Slo enia.
Las , we will analyse he en opy index. In he same way han in p e ious cases,
many speci ica ions ha e been es ima ed. Table 9 shows his p ocess un il
ha ing ob ained he las model. In his case we ejec he equali y o he in e cep s
in he es ima ed models oo.
36
Table 9: Es ima ion o he empi ical model o he en opy index
Model 1 Model 2 Model 3
Coe icien
Signi icance
R
2
, e o c i e ia
and p alue
Coe icien Signi icance
R
2
, e o c i e ia
and p alue
Coe icien Signi icance R
2
, e o c i e ia
and p alue
Elde ly popula ion 0.0006
R
2
= 0.824461
Schwa z c i e ia=
-1,852.645
Akaike c i e ia= -
1,967.425
Hannan-Quinn
c i e ia= -
1,921.685
p- alue= 1.76e-
049
0.0006
R
2
= 0.824437
Schwa z
c i e ia= -
1,858.425
Akaike c i e ia= -
1,969.379
Hannan-Quinn
c i e ia= -
1,925.164
p- alue= 1.41e-
049
0.0005
R
2
= 0.822473
Schwa z c i e ia= -
1,901.886
Akaike c i e ia= -
2,009.586
Hannan-Quinn
c i e ia= -
1,966.699
p- alue= 1.39e-
051
Unemploymen
a e
-0.0003 -0.0003 -0.0003
GDP 0.0009
***
0.0009
***
0.0009
***
GDP_sq -6.48e-06
***
-6.64e-06
***
-6.39e-06
***
R&D Expendi u e 0.0107
***
0.0109
***
0.0106
***
Densi y -0.0007
***
-0.0007
***
-0.0007
***
Fo eigne s -0.0007 -0.0007 -0.0006
Fe ili y a e 0.0017
New ci izenship 0.0033 0.0032
Go . Expendi u e 0.0002 0.0002 0.0002
Educa ion -0.0006 -0.0006 -0.0006
37
Model 4 Model 5 Model 6
Coe icien
Signi icance
R
2
, e o c i e ia
and p alue
Coe icien
Signi icance
R
2
, e o c i e ia
and p alue
Coe icien
Signi icance
R
2
, e o c i e ia
and p alue
Elde ly
popula ion
0.0004
R
2
= 0.822196
Schwa z c i e ia= -
1,907.191
Akaike c i e ia= -
2,011.045
Hannan-Quinn
c i e ia= -1,969.690
p- alue= 6.58e-061
R
2
= 0.822066
Schwa z c i e ia= -
1,912.786
Akaike c i e ia= -
2,012.793
Hannan-Quinn
c i e ia= -
1,972.970
p- alue= 3.41e-
059
R
2
= 0.821696
Schwa z c i e ia=
-1,917.913
Akaike c i e ia= -
2,014.074
Hannan-Quinn
c i e ia= -
1,975.782
p- alue= 3.48e-
059
Unemploymen
a e
-0.0002 -0.0002 -0.0003
GDP 0.0009
***
0.0009
***
0.0008
***
GDP_sq -6.76e-06
***
-6.89e-06
***
-6.47e-06
***
R&D Expendi u e 0.0109
***
0.0111
***
0.0102
***
Densi y -0.0007
***
-0.0007
***
-0.0007
***
Fo eigne s -0.0005 -0.0007 -0.0009
*
Fe ili y a e
New ci izenship
Go . Expendi u e
Educa ion -0.0006 -0.0004
38
Fixed e ec s, using 350 obse a ions
Dependen a iable: En opy Index
Beck-Ka z s anda d de ia ions
Coe icien -s a is ic
Cons an 0.1212 10.54 ***
GDP 0.0008 4.281 ***
GDP_sq -5.84e-06 -4.164 ***
R&D Expendi u e 0.0101 3.015 ***
Densi y -0.0008 -8.814 ***
Fo eigne s -0.0009 -1.867 *
R-squa e MCVF (LSDV)
0.819703
Log-likelihood 1042.936
Akaike c i e ia -2037.872
Schwa z c i e ia -1945.282
Hannan-Quinn c i e ia -2001.018
ho 0.398579
Du bin-Wa son 1.100795
Robus con as o di e en in e cep s in g oups -
Null hypo hesis: G oups ha e a common in e cep
Con as s a is ic: Welch F (18, 119.8) = 80.0174
wi h p- alue = P (F(18, 119.8) > 80.0174) = 6.49e-058
Table 10: Econome ic model o en opy index
Once ha ing gone h ough di e en models, we decide ha he one shown in
Table 10 is he bes one, in e ms o log-likelihood and e o c i e ia. We obse e
ha 81.9703% o he en opy indexes’ a iabili y is explained by he a iabili y o
he eg esso s o his de ini i e model. This is he bes adjus men ou o he h ee
indices’ models, and his happens being his he model wi h leas a iables ou o
e e yone ha has been done. We obse e an inc ease in log-likelihood a he
same ime ha e o s dec ease, gi ing a highe eliabili y o his las model, in he
case whe e he en opy index is he dependen a iable. We see ha “R&D
Expendi u e”, “Densi y” and “Fo eigne s” a e signi ican a iables. We a e no
going o explain again wha his means, as hey main ain he sign seen be o e o
he o he indices.
In his las analysis he e s ill is p esence o Kuzne s e ec . So, like in he p e ious
cases we calcula e he minimum h eshold o GDP, o see he ela ion be ween
hese wo a iables in di e en coun ies. So, ollowing he p ocedu e p e iously
done we see ha he h eshold has a alue o 15.7201, being he smalles
h eshold ou o he h ee. his means ha coun ies wi h a GDP highe han
15,720.1 will ha e a nega i e ela ion be ween GDP and he seg ega ion index.
39
I we ake a look o which coun ies, om he s udied ones, a e hese, we ha e
all he p e ious ones plus G eece and Slo akia. In hese coun ies and inc ease
o he GDP in housand pe capi a will cause a dec ease in he alue o he index,
so he immig an popula ion will be mo e une enly dis ibu ed. This may happen
because in coun ies ich enough hey need mo e indi iduals e e ywhe e, so
immig an s will be dis ibu ed in all e i o y.
As we see h ough his pa o he wo k, di e en mac oeconomic a iables a ec
o seg ega ion indices. Howe e , a a iable being signi ican o explain ce ain
index does no mean ha i will be signi ican in e e y o he index; we obse e
his wi h he a iable “Educa ion”. Each a iable a ec s in di e en ways o
indices. The e a e some a iables we hough o ha happen o ne e be
signi ican . This is he case o “Fe ili y a e”, “New ci izenship” and “Go e nmen
expendi u e”. The o he a iables a e signi ican a leas once, being ema kable
he cases o “Densi y” and “Fo eigne s” ha a e always signi ican . Some o he
a iables ha a e signi ican ha e a posi i e e ec in he index alue; his is he
case o “Elde ly popula ion”, “R&D Expendi u e” and “Educa ion”. An inc ease o
hese a iables means an inc ease in he immig an seg ega ion, meaning ha
immig an popula ion will be geog aphically mo e une enly dis ibu ed. We also
ha e a iables like “Unemploymen a e”, “Densi y” and “Fo eigne s” ha
dec ease he immig an seg ega ion. So, i hese a iables inc ease he
immig an popula ion will be e i o ially dis ibu ed in a mo e homogeneous way.
This does no mean ha we ha e o discou age he educa ion o he expendi u e
in R&D; i also does no mean ha we ha e o adop policies ha inc ease
unemploymen .
I is e y impo an o ema k he nonlinea ela ion o he GDP, as i shows ha
he e ec o he GDP in he seg ega ion index is wha we we e looking o , a
Kuzne s cu e. We ha e seen ha his Kuzne s e ec is p esen in he h ee
models. We ha e also seen ha his allows us o calcula e whe e he minimum
h eshold is, in o de o see in which coun ies he nega i e ela ion be ween GDP
and he dependen a iable is p esen . Fo example, we see ha in Romania,
whe e he GDP is below he h eshold, i he GDP in housand pe capi a,
inc eases in one uni , he alue o he index will inc ease, being ansla ed in a
40
mo e une en dis ibu ion o he immig an popula ion. Howe e , in o he coun ies
such as No way, whe e he GDP is abo e he h eshold o he GDP, an inc ease
o he GDP, in housand pe capi a, in one uni means an es ima ed dec ease o
he seg ega ion index; ha is, he immig an popula ion will be mo e
homogeneously dis ibu ed in he e i o y.
5.- Conclusions
The gold o his pape is wo old. Fi s , measu ing he une enness o he e i o ial
dis ibu ion o immig an popula ion in Eu ope wi h egions’ agg ega e da a. I is
also in e es ing o see how hese indices e ol ed h oughou he pe iod s udied,
since his quan i ica ion enables he end o he une enness o be analysed.
Second, s udying he mac oeconomic a iables ha a ec he geog aphical
une enness o immig an s’ loca ion. In pa icula , we ind ha he a iable ha
mos a ec o immig an e i o ial dis ibu ion is GDP. We analyse he ela ion
be ween he GDP and he indices, ocusing in ying o see i he e was a Kuzne s
e ec .
Rega ding wi h he une enness quan i ica ion, we obse e ha , e en i he alues
o he seg ega ion indices used a e di e en , he e olu ion is simila in he h ee
cases. So, we see ha he esul s a e quali a i ely simila e en i hey a e
quan i a i ely di e en . We also see in a clea way ha seg ega ion dec eases
o e ime, ul illing he expec a ion. This may be because as ime goes by
immig an s a i e o egions whe e he e a e employmen oppo uni ies bu whe e
i s wa e immig an s did no a i e. We can say ha a Eu opean le el immig an
popula ion is mo e e enly dis ibu ed in 2019 han in 1999. When he analysis is
ca ied ou a coun y le el he conclusion is simila ; une enness in he
geog aphical dis ibu ion o immig an popula ion has educed.
Rega ding he sea ch o de e minan s o he seg ega ion indices used. We
choose some a iables ha we conside ep esen a i es. In ela ion wi h he
mac oeconomic de e minan s o he geog aphical dis ibu ion o immig an s, we
ob ain some in e es ing esul s. Fi s , we ind ha a iables, like densi y, a ec
nega i ely o he indices. So ha egions being dense in e ms o popula ions
will lead o a lowe e i o ial dispe sion o immig an popula ion. This ela ion may
41
e lex ha when he densi y is highe he immig an popula ion has o dis ibu e
in o de o ix in a be e way. In he same way, o he s like “Elde ly popula ion”
inc ease he alue o he indices. This means ha when he pe cen age o elde ly
popula ion inc eases, immig an popula ion is mo e une enly dis ibu ed. This
migh be caused because o he wo kplaces o hose ha jus e i e being mo e
concen a ed. Talking abou he ela ion be ween hese indices and he GDP, he
ela ion is in e es ing, as he exis ence o he Kuzne s e ec in his case is p o ed.
This is in e es ing o policymake s, since depending o a coun ies si ua ion he
p io i ies will change.
The conclusion o he second pa is especially ele an o policymake s, in o de
o adop di e en policies o enjoy he bene i s o immig a ion. This does no mean
ha i is a good idea o make d as ic changes in hose a iables ha inc ease he
alue o he indices; ha is, ha inc ease he une en dis ibu ion. Bu i helps o
hink abou o he policies in o de o minimize he nega i e e ec s some decisions
can ha e in his aspec .
Fo u he esea ch i could be in e es ing o use he decomposabili y p ope y
some indices ha e. This p ope y allows us o analyse he inequali y in he
dis ibu ion o immig an popula ion in Eu ope, dis inguishing be ween he
inequali y gi en be ween di e en coun ies and he inequali y gi en wi hin
coun ies; his is, be ween egions inside a coun y. I would also be in e es ing
o see o wha ex en he p esence o a - igh pa ies wi h some powe a ec s o
hese indices, as hey a e winning powe in many coun ies. Fo example, apa
om Aus ia ha was men ioned be o e, in Sweden he e is
S e igedemok a e na; he Dansk Folkepa i in Denma k, he Schweize ische
Volkspa ei in Swi ze land, he Al e na i e o Ge many in Ge many, o The
Uni ed Kingdom Independence Pa y in Uni ed Kingdom. Wi h so many cases i
could be in e es ing o see how his a ec s o he homogenei y in he e i o ial
dis ibu ion o immig an popula ion.
