School composition effects in Spain

DOCUMENT DE TREBALL
XREAP2010-13
School composi ion e ec s in Spain
An onio Di Paolo (GEAP & IEB)
1
SCHOOL COMPOSITION EFFECTS IN SPAIN
An onio Di Paolo
Uni e si a Au ònoma de Ba celona & IEB
Abs ac :
D awing on he PISA 2006 da a se , his s udy examines he impac o school’s socio-
economic composi ion on he science es sco es o Spanish s uden s en olled in
compulso y seconda y educa ion. We de ine school’s composi ion in e ms o he a e age
pa en al human capi al o s uden s a ha same school. These con ex ual pee e ec s a e
es ima ed using a semi-pa ame ic me hodology, which enables spillo e s o a ec all he
pa ame e s in he educa ional p oduc ion unc ion. We also deal wi h he po en ial
p oblem o s uden sel -selec ion in o speci ic schools by using an a i icial so ing
mechanism, which we belie e o be independen o a s uden ’s unobse ed abili ies. The
esul s indica e ha he associa ion be ween a school’s socio-economic composi ion and
es sco e esul s is clea ly posi i e and signi ican ly highe when compu ed using a semi-
pa ame ic app oach. Howe e , we ind ha he endogenous so ing o s uden s in o
schools also plays a undamen al ole, gi en ha spillo e s a e signi ican ly educed
when his selec ion p ocess is elimina ed om ou measu e o school’s composi ion
e ec s. Speci ically, he es ima ions sugges ha con ex ual pee e ec s a e mode a ely
posi i e only in hose schools whe e he socio-economic composi ion is compa a i ely
high. In addi ion, we ind some e idence o asymme y as o how he ex e nal e ec s and
he so ing p ocess ac ually ope a e, appa en ly a ec ing males and emales and high and
low pe o mance s uden s di e en ly.
Keywo ds: Educa ional A ainmen s, Pee E ec s, PISA, Spain
JEL Classi ica ions: I20, I21, I29
Co esponding au ho : An onio Di Paolo; e-mail: an onio.d[email p o ec ed], el.: +34935813415; ax: +34935812292.
Depa amen d'Economia Aplicada, Uni e si a Au ònoma de Ba celona (UAB); Campus de Bella e a, Edi ici B 08193
Bella e a (Ce danyola), Spain. Ins i u d’Economia de Ba celona, Uni e si a de Ba celona.
1. In oduc ion
This pape in es iga es school composi ion e ec s on Spain’s lowe -seconda y schools
(Educación Secunda ia Obliga o ia, ESO), using PISA da a om 2006, and as such is he
i s s udy o examine his ques ion explici ly in he Spanish case. Quan i ying he impac o
he socio-economic composi ion o Spain’s schools is especially impo an in his coun y
because, whe e he e is an excess o demand — as is common in la ge ci ies — he admission
p ocess in public and public- unded (conce ada) p i a e schools is e y closely ela ed o
zoning laws and school dis ic policies. The e o e, gi en ha school admission c i e ia
assign g ea es weigh o he p oximi y o he s uden ’s home o he school, hese educa ional
policies a e inex icably linked o he e ec s o he schools’ socio-economic mix. In ac , hey
2
may esul in he di ec ans e o he exis ing socio-economic esiden ial seg ega ion in o
he schools (Hoxby 2000, Go a d e al. 2003). Mo eo e , school composi ion e ec s migh
ha e gained addi ional ele ance in Spain as a esul o he signi ican inc ease in he numbe
o immig an s uden s om less a luen social backg ounds in ecen yea s and he
subsequen in e ac ion wi h exis ing zoning laws — i.e. less ad an aged immig an amilies
end o eside wi hin e hnic encla es and, as a consequence, hei child en ine i ably end o
concen a e in schools cha ac e ised by a low socio-economic composi ion.
Empi ically, he e a e many channels ia which he ea u es o an indi idual’s
schoolma es o classma es — namely, he pee e ec s — migh in luence indi idual
a ainmen . In he gene al amewo k p oposed by Manski (1993, 2000), he o e all e ec o
he pee g oup on indi idual ou comes p ima ily in ol es elemen s o social in e ac ion ha
include bo h endogenous and con ex ual (o exogenous) e ec s. The o me a e he di ec
e ec s ha pee beha iou o ou comes can ha e on indi idual ou comes; ha is, s uden s
may well lea n mo e because hei school/classma es lea n mo e. The la e a e he impac
ha ce ain exogenous cha ac e is ics o he pee g oup can ha e on a s uden ’s achie emen
— i.e. indi idual pe o mance depends on he socio-economic composi ion o his/he g oup.
In addi ion, he ex en o pee e ec s migh be con ounded by he p esence o sha ed
en i onmen al/school elemen s o indi idual cha ac e is ics (e.g. cogni i e and non-cogni i e
skills) ha go unobse ed by he econome ician; he so-called con ex ual e ec s.
Ob aining sepa a e es ima es o endogenous and con ex ual e ec s is augh wi h
empi ical complica ions
1
and, mo eo e , is highly da a-demanding. Thus, his s udy conce ns
i sel solely wi h con ex ual e ec s, which has been a ai ly common app oach in he
empi ical li e a u e o da e. Mo e speci ically, his pape uses a b oad measu e o he socio-
economic composi ion o schools, based on he a e age pa en al educa ional backg ound
(de ined as he highes educa ional le el comple ed by ei he one o he wo pa en s) o each
school. I s main con ibu ion o he exis ing li e a u e consis s in he implemen a ion o a
semi-pa ame ic me hodology ha allows school-con ex ual e ec s o in luence all
pa ame e s in he educa ional p oduc ion unc ion (as such, adap ing he o iginal p oposal
made by Raymond & Roig 2010). In ui i ely, aking as a e e ence he mos disad an aged
schools in e ms o hei socio-economic composi ion (i.e. schools in he lowes quin ile o
a e age pa en al backg ound), he pape shows ha mo ing o be e endowed schools migh
gene a e a le el shi as well as a ious po en ial g adien shi s. Indeed, he measu e o
1
Speci ically, wha is commonly e e ed o as e lec ion p oblems, which in ol e he simul aneous de e mina ion o
achie emen o all s uden s wi hin a pee g oup (i.e. a simul anei y bias p oblem).
3
con ex ual pee e ec s p oposed he e should cap u e he global impac o school composi ion
on he educa ional p oduc ion unc ion. In addi ion, he lexible s a egy adop ed ensu es ha
such school composi ion e ec s a e non-linea , since hey a e sepa a ely compu ed o each
successi e quin ile o he schools’ a e age pa en al educa ion.
Finally, he pape deals wi h he mos common p oblem encoun e ed in such s udies,
namely he sel -selec ion o s uden s in o di e en schools (i.e. a speci ic ype o co ela ed
e ec ). Speci ically, he p esence o a so ing mechanism ha alloca es hose s uden s ha a e
be e endowed o unobse ed cha ac e is ics in o schools wi h highe a e age pa en al
schooling migh bias ou measu e o pee e ec s. The e o e, an al e na i e so ing
mechanism is p o ided ha can be assumed o be un ela ed o an indi idual s uden ’s
unobse ed cha ac e is ics. Such eo de ing is based on he p edic ed linea sco e, ob ained
om an o de ed p obi model ha es ima es he p obabili y o membe ship in each quin ile o
he schools’ a e age pa en al educa ion. This a i icial so ing is hen used o educe selec ion
bias in he de ini ion o e e ence and non- e e ence g oups. Thus, he s udy is able o p o ide
a measu e o a school’s composi ion e ec s ha is signi ican ly less a ec ed by co ela ed
e ec s.
Wi h hese pu poses in mind, he es o he pape is o ganized as ollows: sec ion 2
con ains a b ie e iew o selec ed pape s examining pee e ec s, ocusing on he a ious
es ima ion s a egies adop ed o elimina e co ela ed e ec s. Sec ion 3 desc ibes he empi ical
me hodology ha is used in his s udy and Sec ion 4 is dedica ed o a desc ip ion o he da a.
Sec ion 5 con ains he empi ical esul s, as well as a obus ness check and an analysis o he
po en ial asymme ies o school composi ion e ec s. Sec ion 6 concludes.
2. Selec ed Con ibu ions
P e ious s udies o pee e ec s on scholas ic achie emen p esen qui e mixed indings
and, o da e, he e is no uni ied e idence as o he exis ence o o he ac ual o m ha hese
e ec s migh ake. This line o esea ch has sough o cap u e hese po en ial spillo e s a
se e al poin s in he educa ional p ocess ( om p ima y o e ia y s ages), and by conside ing
di e en pee ea u es (ac ual o lagged pee s’ es sco es, e hnic and socio-economic
composi ion o he pee g oup, e c.). This poin s o he ac ha he esul ing spillo e s will
be ei he posi i e o nega i e (o e en ze o), while dependen a all imes on he na u e o he
pee a iables, and as such he inal ne e ec s become an empi ical ques ion. Fu he mo e,
4
go e ned p ima ily by da a a ailabili y, he de ini ion o hese pee g oups has been ma kedly
he e ogeneous, anging om school, school-by-g ade and class oom o o he social pee s
such as oomma es o iends. As a esul , he indings end o be highly case-speci ic and no
always s ic ly compa able. In gene al, his lack o explici compa abili y is a ibu able o i)
he speci ic cha ac e is ics o he sample used, and ii) he (subsequen ) econome ic echnique
adop ed in iden i ying pee e ec s o he han he co ela ed e ec s.
In e es ingly, some s udies a e based on special samples in which s uden s a e assigned
andomly in o pee g oups, he eby possibly elimina ing he bias a ibu able o co ela ed
e ec s. Mo e speci ically, such quasi-expe imen al s udies exploi he andomized ials
gene a ed by he chance ma ching o s uden s wi h i s -yea oomma es in college
accommoda ion (see Sace do e 2001, Zimme man 2003, Fos e 2006 and B unello e al. 2010
among o he s), o class assignmen on he basis o su name du ing i s yea uni e si y
cou ses (De Paola & Scoppa 2010).
Se e al o he pape s, which ocus p incipally on p ima y and seconda y schools, adop
ixed e ec amewo ks in o de o con ol o any po en ial bias in pee e ec es ima es. Fo
example, McEwan (2003) con ols o bo h school and amily ixed e ec s when es ima ing
pee con ex ual e ec s a he class oom le el, inding a posi i e and sligh ly conca e e ec o
he class oom mean o he mo he s’ educa ion. Hanushek e al. (2003) exploi s a unique panel
da ase co e ing h ee successi e coho s o s uden s; hence, hey a e able o con ol o
indi idual, school and school-by-g ade ixed e ec s in a alue-added speci ica ion o he
educa ional p oduc ion unc ion. They epo a posi i e e ec o mean pee achie emen on
imp o emen s eco ded in es sco es, which emains almos cons an o e he es sco e
dis ibu ion. They also ound no impo an e ec o he a e age amily income o he pee s
and es sco e he e ogenei y in he pee g oup. La y e al. (2008) exploi coho - o-coho and
wi hin-school changes in he p opo ion o low achie e s (i.e. hei p oxy o pee abili y) o
iden i y endogenous pee e ec s and he mechanisms ia which hey impac on an
indi idual’s own achie emen . They epo a clea nega i e impac o he p opo ion o low
achie e s in he class oom, which ends o be mo e p onounced o s uden s o low socio-
economic backg ounds
2
. Amme mulle & Pischke (2009) conside he con ex ual e ec s in
p ima y schools o se e al Eu opean coun ies (using PIRLS da a). They conside ha
con ex ual pee e ec s a he class oom le el a e cap u ed by he a e age numbe o books a
home. These pee e ec s a e iden i ied by exploi ing a ia ions ac oss he class ooms wi hin
2
Mo eo e , hei esul s also sugges ha he nega i e impac o he p opo ion o low achie e s mainly ope a es ia he
dis up i e in luence i has on eache s’ pedagogical p ac ices, in e ac ion wi h o he s uden s and class oom diso de .

5
he same g ade o he same coho o s uden s (once es ablished ha hese classes had been
o med in wha was a la gely andom manne ). Thei esul s indica e ha , in gene al,
con ex ual pee e ec s do exis ; howe e , hey also poin ou ha simple OLS es ima ions
migh be equally a ec ed by selec ion bias as well as by measu emen e o in he pee
a iable, which end o ope a e in opposi e di ec ions.
The p esen s udy is mos closely ela ed o hose unde aken by Fe ig (2003),
Schneeweis & Win e -Ebme (2007) and Rang id (2007), which also d aw on PISA da a.
Speci ically, Fe ig (2003) in es iga es he e ec o eading achie emen he e ogenei y in US
schools, which is iden i ied h ough Ins umen al Va iables (IV)
3
— namely, dummies o
p i a e and selec i e schools and a se o a iables ela ed o he p e alence o pa en al ca ing
beha iou in each school. His esul s indica e ha a ending a he e ogeneous school in e ms
o s uden achie emen unde mines indi idual pe o mance; howe e , he nega i e e ec he
epo s appea s o be excessi e when es ima ed using IV (which aises he ques ion abou he
alidi y o he ins umen s used).
The pape by Schneeweis & Win e -Ebme (2007) explo es he e ec o socio-economic
composi ion a he school-by-g ade le el in Aus ia. The au ho s p esen e idence ob ained,
on he one hand, om OLS es ima ions based on an ex ensi e se o indi idual and school
con ols and, on he o he hand, om he applica ion o school ixed e ec s. They a gue ha ,
when accoun ing o school ype — gi en he ma ked ack sys em in Aus ian lowe and
uppe seconda y schools — school ixed e ec s educe he selec ion bias in he es ima ion o
pee e ec s. Thei esul s highligh a signi ican asymme y in he pee e ec s on eading
4
,
which seems o ha e a mo e bene icial e ec in he case o s uden s o a low socio-economic
backg ound. Mo eo e , hey also adop a quan ile eg ession s a egy, which e eals ha
s uden s in he lowe pa o he abili y dis ibu ion a e mo e posi i ely a ec ed by he socio-
economic composi ion o hei pee g oup.
Finally, Rang id (2007) analyses he e ec o he socio-economic composi ion o a
school in e ms o he h ee PISA subjec s ( eading, ma hs and science) d awing on Danish
da a, which a e complemen ed wi h adminis a i e egis e s o o e come he po en ial
p oblems caused by he limi ed sample o s uden s wi hin each school
5
. Gi en he
3
O he pape s in which he iden i ica ion o pee e ec s elies on IV s a egies include hose by Feins ein & Symons (1999)
and Robe son & Symons (2003), whe e he ins umen s consis o loca ion a iables and eache assessmen o a s uden ’s
p e ious abili y combined wi h egion o bi h dummies, espec i ely.
4
By con as , hey also sugges ha he appa en pee e ec s in ma hema ics, as es ima ed by OLS, a e due only o selec ion
e ec s, gi en ha hei ixed-e ec es ima es a e no s a is ically signi ican . Addi ionally, in his case, pee g oup
he e ogenei y seems o play a e y limi ed ole in explaining es sco e a ainmen .
5
As a gued by Micklew igh e al. (2010), he limi ed s uden sampling made by PISA can esul in a measu emen e o in
he es ima ion o pee e ec s. This would bias he e ec o school composi ion owa ds ze o. Un o una ely, such
6
comp ehensi e na u e o he Danish seconda y school educa ion sys em, he au ho canno
ely on he school- ixed e ec es ima ion as was he case in Schneeweis & Win e -Ebme
(2007); indeed, she canno assume ha indi iduals (and hei amilies) who a e placed in a
gi en school o a ce ain ack sha e simila unobse ed cha ac e is ics
6
. He iden i ica ion
s a egy is ins ead based on con olling o a la ge se o indi idual, amily and school
a iables, wi hou explici ly conside ing he ole o selec ion on unobse able ea u es. The
esul s in his s udy sugges a clea posi i e e ec o a ending a school wi h a highe socio-
economic composi ion in he middle o he es sco e dis ibu ion, whe eas no signi ican
e ec is ound o he socio-economic he e ogenei y a he school le el. Mo eo e , he
quan ile es ima ion e eals ha school composi ion e ec s end o be highe o low-abili y
s uden s on he eading es sco e, bu he au ho inds a U-shaped e ec o science, which
means ha low and high abili y s uden s bene i equally om a be e socio-economic school
composi ion.
3. Empi ical F amewo k
The es ima ion s a egy p oposed in his pape ep esen s a s ep o wa d in e ms o he
measu emen o pee e ec s. Indeed, he main inno a ion wi h espec o p e ious s udies
consis s, as b ie ly commen ed in he in oduc ion, in he idea ha he spillo e s p oduced by
an imp o emen in a school’s socio-economic composi ion may a ec no only he in e cep ,
bu all he pa ame e s o he educa ional p oduc ion unc ion. This o iginal p oposal has been
aken (and adap ed) om he pape by Raymond & Roig (2010), in which hey es ima e he
ex e nali y p oduced by he a e age human capi al o wo ke s in he same i m. In keeping
wi h his ex e nali y, his pape akes as i s s a ing poin he s anda d educa ional p oduc ion
unc ion,
, ,
i s s i s i s
T X Z
α β δ ε
′ ′
= + + + (1)
adminis a i e da a a e no a ailable o public use in he Spanish case; he e o e, i should be bo ne in mind ha he
es ima es epo ed in his s udy ep esen a lowe bounda y o he ue impac o school-a e age pa en al educa ion.
6
No ice ha since he LOGSE e o m o 1990, he Spanish seconda y educa ion sys em has been compulso y and
comp ehensi e un il he age o six een, which (as in he case o Denma k) makes he school- ixed e ec amewo k
un easible o con olling endogenous pee g oup selec ion. See sec ion 3 o de ails as o how such a p oblem is add essed
in his pape .
7
whe e es sco e T
i,s
o s uden i in school s depends on a se o indi idual and amily
cha ac e is ics (X
i
) as well as on a se o school cha ac e is ics (Z
s
), plus a composi e e o
e m (ε
i,s
). Usually, exogenous pee e ec s a e simply es ima ed by conside ing ha he
in e cep e m (α
s
) is no ixed, bu ins ead dependen on an a e age cha ac e is ic o he pee
g oup — i.e. in his case, he a e age pa en al educa ion o s uden s a ha school s (
s
PE
).
This means ha he in e cep e m in (1) can be ew i en as,
(
)
s
s
PE
α α µ
= + ⋅
(1a)
which indica es ha a uni inc ease in he a e age pa en al educa ion in he school modi ies
he mean es sco e by
µ
poin s, h ough a shi in he in e cep e m. We could also adop a
non-linea speci ica ion, whe e he impac o he school’s composi ion o pa en al human
capi al is allowed o a y o each successi e quin ile o school-a e age pa en al educa ion
(
(
)
, 1,..,5
s
j
Q PE j =
). In his case, he in e cep e m in (1) can be exp essed as,
(
)
5
12
s
s j j
j
Q PE
α α α
=
= + ⋅
∑
(1b)
whe e he con ex ual pee e ec s a e now α
j
(
j=1,..,5
) and a e allowed o be di e en o each
quin ile o a e age pa en al schooling. E en in his case, he impac o he pee g oup’s
cha ac e is ics is only p oduced by a le el e ec , which ope a e h ough a modi ica ion o he
educa ional p oduc ion unc ion in e cep ; in ac , once he exp ession (1b) is subs i u ed in o
equa ion (1) we ob ain,
(
)
5
, 1 ,
2
s
i s j j i s i s
j
T Q PE X Z
α α β δ ε
=
′ ′
= + ⋅ + + +
∑
. (2)
This co esponds o he s anda d equa ion used in he pee e ec s li e a u e, excep o he
non-linea speci ica ion o he con ex ual pee e ec s.
Equa ion (2) clea ly speci ies ha he s anda d app oach cons ains school composi ion
spillo e s so as o a ec only he in e cep e m and no o he pa ame e in he educa ional
p oduc ion unc ion (e en allowing o a non-linea e ec ). Howe e , he e is no heo e ical
eason o belie e ha he con ex ual pee e ec s consis only o a simple le el e ec . Fo
example, an imp o emen in he socio-economic composi ion o he pee g oup migh modi y
he g adien o he e ec o a s uden ’s amily backg ound and home en i onmen on his/he
es sco e. Addi ionally, belonging o a “good” pee g oup in e ms o a e age pa en al human
capi al migh elax he ela ionship be ween o he school cha ac e is ics and an indi idual’s
achie emen s.
8
In o de o cap u e any po en ial shape e ec o school composi ion, we conside a
e e ence g oup, which consis s o all he s uden s who belong o he leas -ad an aged
schools in e ms o a e age pa en al educa ional backg ound. In he p esen applica ion, he
leas -ad an aged schools a e de ined as hose schools ha appea in he i s quin ile o he
a e age pa en al educa ion
7
(i.e.
(
)
(
)
ss
j
PEQPEQ
1
=
). The e o e, he educa ional p oduc ion
unc ion is sepa a ely es ima ed o he e e ence ca ego y, as in equa ion (3):
(
)
(
)
(
)
(
)
1 1 1 1 1
, 1 1 , , , 1
ˆ ˆ
ˆ ˆ ˆ ˆ
| i
Q Q Q Q Q
s s s
i s i s i s i s i s j
T Q PE X Z R Q PE Q PE
α β δ ε ψ ε
′ ′ ′
= + + + = + = . (3)
F om he ob ained pa ame e es ima es (
ψ
ˆ
1
Q
), we hen p oceed o o ecas he es sco e o
all he indi iduals who do no belong o he e e ence g oup, ha is,
(
)
(
)
(
)
1 1 1
, 1 1
ˆ ˆ
ˆˆ ˆ
| ; , 1
Q Q Q
s s
i s i s j
T Q PE X Z i Q PE j
ψ α β δ
′ ′
= + + ∀ ∈ >
. (4)
Finally, o each successi e quin ile o he school-a e age pa en al educa ion, he measu e o
school composi ion spillo e s p esen ed he e consis s o he a e age di e ence be ween he
ac ual and he o ecas ed es sco e wi hin each quin ile:
( )
(
)
( )
1
, ,
1
ˆˆ
| ;
, 1
j
NQ
s
i s i s j s
i
j j
j
T T Q PE
IEX i Q PE j
N
ψ
=
−
= ∀ ∈ >
∑
(5)
In o he wo ds, his measu e o con ex ual pee e ec s consis s o coun e ac ual
e idence, which is based on he ce e is pa ibus wi hin-quin ile mean di e en ial be ween he
obse ed and he p edic ed es sco e, whe e he la e is ob ained by using he pa ame e s
es ima ed o s uden s in he leas -ad an aged schools. Mo e in ui i ely, his me hodology
ep esen s a semi-pa ame ic app oach o cap u e he ce e is pa ibus change in he es sco e,
p oduced by mo ing a ep esen a i e s uden om he i s quin ile o successi e quin iles o
he school-a e age pa en al educa ion. No e ha his measu e o he e ec s o school
composi ion cap u es in a semi-pa ame ic way he change in each pa ame e making up he
whole educa ional p oduc ion unc ion (bo h le el and shape e ec s), p oduced by
inc emen ing he a e age pa en al schooling om he i s o he highe quin iles. In his way,
we a e able o p o ide mo e compelling and comple e e idence abou he e ec o school
7
As no ed by Raymond & Roig (2010), he de ini ion o he e e ence g oup is always subjec o some deg ee o
a bi a iness; in hei case, hey de ine he e e ence g oup as hose p oduc i e es ablishmen s in which he a e age wo ke s’
human capi al is equal o o less han eigh yea s o schooling. This de ini ion ollows he logic ha eigh yea s o educa ion
co esponds o he compulso y leng h o educa ion unde he ins i u ional amewo k ha was hen alid o indi iduals in
hei sample; mo eo e , i should ep esen hose i ms ha chie ly employ unskilled wo ke s. In ou case, we conside i
be e o de ine he e e ence g oup in an endogenous way — i.e. di iding he sample in o quin iles and aking he i s one
as he e e ence g oup. This de ini ion allows us i) o conside schools as being mo e he e ogeneous uni s han i ms, and ii)
o main ain a su icien numbe o obse a ions in he e e ence and non- e e ence g oups.
15
pa en al educa ion and also calcula ed he mean alue o all he quin iles (excep ha o he
i s , which is he e e ence ca ego y). The esul s om he semi-pa ame ic me hodology
con i med ha he e ec o he pa en al educa ion o he pee g oup was subs an ial and
clea ly non-linea . As in he p e ious case, mo ing om he leas -ad an aged g oup o he
second quin ile o he school’s socio-economic composi ion had almos no e ec on
indi idual es sco es (almos 5 poin s, bu no s a is ically di e en om ze o), whe eas he
s ep o he hi d quin ile p oduced a posi i e inc ease o abou 12 poin s. Howe e , he
mo emen o highe quin iles gene a ed subs an ial (and posi i e) slope e ec s, which we e
hidden by he implici cons ains o eq. (2). Indeed, school composi ion e ec s could be
quan i ied in o 26 addi ional es sco e poin s o s uden s in he ou h quin ile o he a e age
pa en al schooling and up o 71 poin s o s uden s in he highes quin ile. Addi ionally, he
mean alue o all he non- e e ence g oups was also s a is ically signi ican , app oaching 28
es sco e poin s.
[TABLE 3 ABOUT HERE]
Howe e , hese esul s may well be biased by he ac ha s uden s wi h a be e
endowmen o unobse ed abili ies a e mo e likely o en ol in he be e schools (in e ms,
ha is, o hei socio-economic composi ion). In o de o educe his po en ial selec ion bias,
we i s es ima ed eq. (6) using an o de ed p obi model, he dependen a iable o which was
he i e quin iles o he ac ual school-a e age pa en al educa ion. The es ima es (see Table
2A in he Appendix) indica e ha immig an pupils who a i ed in Spain a an ea lie da e a e
signi ican ly mo e likely o be en olled in schools whe e hei schoolma es’ pa en al
educa ion is highe ; mo eo e , condi ional on egion and municipali y size, he chances o
being in be e schools is also highe o hose who eside close o o he schools. In gene al,
he a iables included p o ide a sa is ac o y explana ion o he p obabili y o being in each o
he quin iles o he school-a e age pa en al educa ion. Subsequen ly, we used he p edic ed
linea sco e o ob ain a p oxy o he school-a e age pa en al human capi al ha was
independen o he s uden s’ unobse ed cha ac e is ics.
When s uden s we e so ed in o e e ence and non- e e ence g oups acco ding o he
p edic ed linea sco e, he e idence conce ning school composi ion spillo e s was ma kedly
di e en . As epo ed in he lowe panel o Table 3, he mean e ec o all he non- e e ence
g oups was s a is ically non exis en , which is he esul o a clea con exi y o school
composi ion e ec s wi h espec o he di e en quin iles o
s
PE
( he same in o ma ion is
depic ed in Figu e 1). In ac , s uden s om he second quin ile o he p oxied a e age

16
pa en al educa ion we e penalized by abou 10 poin s wi h espec o pupils a he leas -
ad an aged schools (i.e. he e e ence g oup), and he spillo e e ec s o s uden s in he
hi d quin ile we e no s a is ically di e en om ze o. In addi ion, when he selec i e so ing
o pupils in o schools was accoun ed o , he e ec o school composi ion was s ongly
educed o s uden s en olled in he be e endowed schools (abou 15 es sco e poin s o
bo h he ou h and he i h quin iles). This e idence sugges s ha , especially o s uden s in
he highes quin ile o he a e age pa en al schooling, he e is a conside able so ing p ocess
in hei a ou wi h espec o less-ad an aged s uden s. Summing up, a signi ican con ex ual
pee e ec was s ill de ec ed, bu i seemed o gene a e a posi i e and modes spillo e only
in hose schools whe e he a e age le el o pa en al educa ion was highe . Howe e , he
p ocess o s uden so ing would seem o be e en mo e impo an han he ex e nali y
p oduced by he socio-economic o igins o an indi idual’s schoolma es.
[FIGURE 1 ABOUT HERE]
5.3 Robus ness Checks
The e idence p esen ed abo e sough o de e mine whe he a mo e a ou able school
composi ion p oduced be e indi idual esul s in he science es sco e (ce e is pa ibus). The
baseline esul s seem o sugges ha he exogenous cha ac e is ics o an indi idual’s
schoolma es ( he con ex ual pee e ec , he e de ined in e ms o pa en al educa ion) exe a
posi i e ex e nali y on he indi idual acquisi ion p ocess o compe ence in his ield. This
spillo e was e en highe when we conside ed no only he le el e ec , bu also he whole
slope e ec s in he educa ional p oduc ion unc ion. Howe e , hese esul s a e likely o ha e
been con ounded by he p esence o an endogenous so ing p ocess ha alloca ed he s uden s
wi h a be e endowmen o unobse ed abili ies o he be e schools (in e ms o hei socio-
economic composi ion). On a emp ing o educe his po en ial bias, he esul s a e ma kedly
di e en : he e was a small and posi i e e ec o he school’s socio-economic composi ion
only in hose schools whe e he a e age le el o pa en al educa ion was conside ably highe .
Wha e e he case, hese esul s migh s ill be biased i he a iables used as exclusion
es ic ions had been sys ema ically ela ed o a s uden ’s unobse ed abili y. Recall ha he
alidi y o hese esul s is based on he assump ion ha , ha ing con olled o he a he ’s and
mo he ’s educa ion, he amily’s socio-economic s a us, mig a ion s a us, language spoken a
17
home and o he amily cha ac e is ics, he p esence o one o mo e a ailable schools and he
a i al age o immig an child en a e un ela ed o he unobse ed abili ies. Un o una ely,
he e is no o mal way o p o e he alidi y o his assump ion, gi en ha i in ol es
elemen s ha a e, by de ini ion, unobse able. E en so, we ha e p o ided an in ui i e
alsi ica ion es , which helps us o co obo a e ou excludabili y assump ion. This es is
based on he idea ha i he excluded a iables had had some e ec on he es sco e equa ion
(e en ia co ela ion wi h he unobse able), including hem in he equa ion o he e e ence
g oup would ha e modi ied he esul s ob ained wi h ou measu e o pee e ec s. In ac , he
logic behind he exclusion es ic ions is ha hese a iables only a ec he es sco e ( o he
e e ence g oup) h ough hei e ec on he p obabili y o being in each quin ile o he
school-a e age pa en al educa ion.
Fi s , we pe o med se e al s a is ical es s o analyse he signi icance o he a iables
excluded om he educa ional p oduc ion unc ion; he esul s (no shown he e) sugges ha
bo h a iables (indi idually and join ly) do no di e om ze o a any con en ional le el o
signi icance. Mo eo e , we g adually included he dummies o school a ailabili y and age
on a i al in he es sco e equa ions (3) and (7) and, hen, we ecompu ed he measu es o
school composi ion e ec s (5) and (8), wi hou and wi h he endogenous so ing co ec ion
espec i ely. The esul s, epo ed in Table 4, showed i) he baseline measu e o spillo e s o
be indis inguishable om he o iginal one compu ed wi hou he excluded a iables; in
addi ion, ii) he esul s we e only sligh ly di e en (bu iden ical in s a is ical e ms) when
s uden s we e e-so ed in o e e ence and non- e e ence g oups acco ding o eq. (6) and he
wo a iables we e included in he es sco e equa ion o he e e ence g oup. In p inciple, i
he excluded a iables had had an e ec on a s uden ’s es sco e and/o had con ibu ed o
explain a school’s composi ion e ec s, we would ha e obse ed a ma ked al e a ion in he
measu e p oposed in his pape . The e idence ha can be d awn om he ac ha when
school a ailabili y and age on a i al a e included in he es sco e equa ion and no signi ican
changes a e obse ed makes he excludabili y assump ion made in his pape mo e eliable.
Wha e e he case, i should be bo ne in mind ha , we e ano he so ing mechanism o be
ope a ing — especially wi h espec o schools’ unobse able cha ac e is ics — he esul s
could s ill con ain some bias and mus be conside ed wi h cau ion.
[TABLE 4 ABOUT HERE]
18
5.4 Asymme ies in School Composi ion E ec s
This subsec ion is conce ned wi h he po en ial asymme ies in he e ec s o school
composi ion on di e en subg oups o s uden s, which would ep esen key e idence o he
design o public educa ion policies. Speci ically, i seeks o in es iga e he ex en o which 1)
males and emales, 2) s uden s wi h a high and low backg ound and 3) high and low
pe o mance s uden s (in e ms o hei es sco es) eac di e en ly o a gi en imp o emen
in he socio-economic composi ion a he school le el. The e o e, we p oceeded as ollows:
equa ions (3) and (7) and he espec i e indexes o school composi ion e ec s (5 and 8) we e
compu ed sepa a ely o each subg oup in he o iginal sample
13
. Table 5 con ains he
es ima ed alues o he con ex ual pee e ec measu e o each subg oup, as well as he -
S a is ics o he s a is ical signi icance o he es ima ed di e ence.
[TABLE 5 ABOUT HERE]
Fi s , he esul s o males and emales indica e ha wi hou aking in o accoun he
endogenous so ing p ocess, bo h subg oups we e posi i ely a ec ed by he con ex ual
e ec s. Howe e , i seems ha male s uden s p o i mo e han emales om an inc ease in he
school’s socio-economic composi ion; in addi ion, he di e ence (and i s s a is ical
signi icance) inc eases wi h he quin iles o he school-a e age pa en al human capi al. E en
so, his e idence is comple ely con ounded by he p esence o sel -selec ion. In ac , once i is
accoun ed o , males appea o be penalized when mo ing om he i s o he second quin ile
o he school-a e age pa en al educa ion and a e gene ally insensi i e o u he
imp o emen s. By con as , once he selec ion p ocess has been accoun ed o , emales seem
o ake signi ican ad an age om an imp o emen in he socio-economic mix o he school,
which inc eased mono onically wi h he quin iles o he a e age pa en al backg ound. This
esul migh indica e ha male pupils a e, in gene al, mo e dependen on he e ec o
endogenous g ouping o s uden s in o schools, whe eas emales a e mo e di ec ly sensi i e o
he con ex ual e ec c ea ed by an inc ease in he a e age pa en al educa ion in he school.
Second, o analyse he di e ences by pa en al backg ound he sample was di ided in a
non-pa ame ic way: ollowing Schneeweis & Win e -Ebme (2007), s uden s o a “high
backg ound” a e de ined as hose wi h pa en s whose educa ional a ainmen is highe han
13
In his way he pa ame e s o he es sco e equa ion o he e e ence g oup can di e o each subg oup. No ice ha he
p edic ed linea sco e ha we use o e-so s uden s in o e e ence and non- e e ence g oups is he same o he o e all
sample; his decision makes sense because we a e in e es ed in analysing how di e en subg oups o s uden s en olled in a
gi en school espond o a ce e is pa ibus inc ease in he school-a e age pa en al educa ion — i.e. conside ing ha hey a e
o iginally in he same school, which belongs o a gi en quin ile o he school-a e age pa en al educa ion.
19
he school a e age, while s uden s o a “low backg ound” a e de ined as hose wi h pa en s
whose educa ional a ainmen is he same as o less han he school a e age. In his ins ance,
esul s we e less han conclusi e. In gene al, when he measu e o he e ec s o school
composi ion includes he endogenous selec ion we ailed o de ec any signi ican di e ences
be ween he subg oups. In addi ion, when selec ion was uled ou , we only obse ed a
posi i e and signi ican di e ence in a ou o “high backg ound” s uden s in he be e
endowed schools ( i h quin ile).
Finally, he same logic was used o di ide he sample in o s uden s wi h ei he high o low
es sco es; speci ically, we conside he o me o be s uden s whose es sco e was highe
han he a e age sco e o hei g ade in ha school, whe eas he la e a e s uden s wi h a
poo e es sco e pe o mance han he school-by-g ade a e age
14
. The esul s o he
asymme ies by s uden es sco e indica e ha , i we igno e endogenous s uden so ing, a
signi ican di e ence is only obse ed in a ou o poo e s uden s in he second quin ile o
he school-a e age pa en al educa ion. Howe e , no signi ican di e ence was obse ed
be ween he wo g oups o highe alues o school socio-economic composi ion.
Ne e heless, once endogenous selec ion is accoun ed o , high es sco e s uden s appea ed
o be ba ely a ec ed by an imp o emen in he school’s socio-economic composi ion. By
con as , s uden s wi h low es sco es showed a posi i e and signi ican eac ion, especially
in he highes quin iles o he school-a e age pa en al backg ound.
6. Conclusions
D awing on PISA 2006 da a (p ima ily he science es sco e), his pape has in es iga ed
he e ec s o school composi ion on Spanish seconda y schools. A no el me hodology has
been designed o measu e he spillo e s p oduced by one speci ic exogenous cha ac e is ic o
a s uden ’s schoolma es, namely he highes le el o educa ion comple ed by he pa en s as an
index o he school’s socio-economic composi ion. The p oposed me hodology elaxes he
implici cons ain — common o any pee e ec s udy — whe eby he con ex ual elemen
can only a ec he a e age ou come h ough an in e cep shi (i.e. a le el e ec ).
14
In his case we conside he school-by-g ade a e age ins ead o he simple wi hin-school a e age in o de o ake in o
accoun he ma ked di e ences in he es sco e ha a e explained by he g ade a ended. Howe e , hese esul s should be
ea ed wi h cau ion since his kind o sample sepa a ion migh gene a e addi ional selec ion bias in he es ima es. Howe e ,
he non-pa ame ic s a egy used o di ide he sample should educe he po en ial bias, e en main aining he in o ma i e
con en o his e idence.
20
When accoun ing o all he changes in he educa ional p oduc ion unc ion pa ame e
gene a ed by an imp o emen in he school’s socio-economic composi ion (le el and slope
e ec s), i was ound ha school composi ion e ec s a e subs an ial and signi ican ly highe
han hose ob ained wi h he cons ained speci ica ion. Mo e speci ically, he esul s indica e
ha he e ec o mo ing om he leas -ad an aged schools ( hose in he i s quin ile o he
school-a e age pa en al educa ion) o be e endowed schools imp o es he science es sco e
in a non-linea way, wi h he posi i e e ec being pa icula ly p onounced o pupils om op
schools — i.e. hose en olled in schools whe e mos o he pa en s had comple ed uppe -
seconda y o e ia y educa ion ( i h quin ile o he school-a e age pa en al educa ion).
Howe e , his p elimina y e idence should no be unde s ood as being he pu e
con ex ual e ec o he school’s socio-economic composi ion, gi en ha i migh be
con ounded by he p esence o co ela ed e ec s. This pape has explici ly a emp ed o deal
wi h he endogenous selec ion p ocess whe eby s uden s endowed wi h highe unobse ed
abili ies a e alloca ed o be e schools (in e ms o hei socio-economic composi ion). This
was achie ed by e-so ing s uden s acco ding o a p edic ed linea sco e ob ained om an
o de ed p obi model, which es ima es he p obabili y o membe ship o each quin ile o
school-a e age pa en al educa ion. I is a gued ha , by p oxying pa en al human capi al,
s uden s can be e-so ed in a way ha is unco ela ed wi h unobse ed indi idual
cha ac e is ics.
When school-composi ion spillo e s we e ecompu ed on he basis o his a i icial e-
so ing, he e idence was signi ican ly di e en . The ex e nali ies p oduced by he pa en al
human capi al o schoolma es we e d as ically educed and hey we e mode a ely posi i e
only when he school socio-economic composi ion was compa a i ely high (in he ou h and
i h quin iles). Mo eo e , addi ional e idence conce ning he asymme ies o he e ec s o
school composi ion e ealed majo di e ences be ween male and emale and be ween high
and low pe o mance s uden s. I seems ha he esul s o male s uden s a e mo e closely
a ec ed by endogenous so ing han hey a e by he exogenous cha ac e is ics o hei
schoolma es; by con as , he esul s o hei emale coun e pa s a e mo e sensi i e o he
posi i e con ex ual e ec gi en ha school composi ion e ec s we e g ea e when “cleaned”
by sel -selec ion. Fu he mo e, he subg oup o low pe o mance s uden s appea s o be
posi i ely a ec ed by an imp o emen in school-a e age pa en al backg ound, e en a e
accoun ing o he p esence o endogenous so ing.
Ha ing said his, i is impo an o bea in mind he po en ial pi alls o his s udy, which
a e linked p ima ily o he limi a ions o he da abase d awn upon. Fi s , i is mo e han likely

21
ha he bounda y epo ed o he e ec s o school composi ion is lowe han i is in eali y,
gi en ha he somewha es ic ed sample o PISA s uden s will cause an a enua ion bias in
ou es ima e (compa e Micklew igh e al. 2010). Second, in he case whe e selec ion is made
on he basis o a school’s unobse ed cha ac e is ics (i.e., hose ha a e no cap u ed by he
ex ensi e lis o school con ols included he e), he measu e o school composi ion e ec s
migh s ill con ain some bias. Thi d, i he a iables used as exclusion es ic ions a e in some
way co ela ed wi h unobse ed s uden abili ies, he me hodology adhe ed o he e o
educing he bias gene a ed by endogenous so ing would no be e ec i e.
Wha e e he case, and e en aking hese po en ial limi a ions in o conside a ion, he
e idence p esen ed abo e makes impo an con ibu ions o he on-going public deba e
conce ning school laws and he ( e)alloca ion o ce ain ypes o s uden in o o he schools.
Fi s , he ele ance o endogenous s uden so ing aises he ques ion as o jus how equi able
and e icien he zoning laws egula ing access o Spain’s seconda y schools a e. This
becomes a ma e o u gency when i is seen ha , wi h he sel -selec ion o s uden s uled ou ,
he posi i e impac o enhancing a school’s socio-economic composi ion is only possible
when he a e age pa en al educa ional backg ound is compa a i ely high. This esul would
seem o sugges ha he zoning laws a e ac ually impeding s uden s o a low socio-economic
backg ound om bene i ing om a mo e a ou able socio-economic school en i onmen ,
gi en ha hey appea o lead o he concen a ion o such s uden s in disad an aged school
en i onmen s. This is because amilies o lowe socio-economic s anding end o loca e
sys ema ically in ce ain esiden ial a eas and, as such, hei child en ha e es ic ed access o
he “be e ” schools in e ms o hei socio-economic composi ion. Al hough hese esul s do
no in hemsel es jus i y he supp ession o he a o emen ioned zoning laws, hey migh be
seen as a jus i ica ion o seeking o compensa e less ad an aged s uden s.
Secondly, e idence poin ing o an asymme ic e ec in a ou o low pe o mance
s uden s migh be deemed an a gumen in a ou o he in oduc ion o mo e lexible school
admission policies. Indeed, a mo e equi able s uden mix, achie ed by ealloca ing ma ginal,
low-pe o mance s uden s in o schools wi h a highe han a e age socio-economic
composi ion, migh well educe inequali ies in educa ional achie emen and e en inc ease
o e all esul s. Wha e e he case, a mo e de ailed examina ion o he ela ionship be ween
exis ing school laws, school seg ega ion and he e ec s o school composi ion is essen ial i
we a e o cla i y any o his e idence. These ep esen in e es ing ques ions o u u e
esea ch, pa icula ly i mo e exhaus i e da a can be d awn upon.
22
Re e ences
Amme muelle , A. & Pischke, J.S. (2009). "Pee E ec s in Eu opean P ima y Schools: E idence om he
P og ess in In e na ional Reading Li e acy S udy". Jou nal o Labo Economics, Uni e si y o Chicago P ess,
ol. 27(3), pages 315-348, 07.
B unello, G., De Paola, M. & Scoppa, S. (2010). "Pee E ec s In Highe Educa ion: Does The Field O S udy
Ma e ?". Economic Inqui y, Wes e n Economic Associa ion In e na ional, ol. 48(3), pages 621-634, 07.
Hoxby, C. (2000). "Pee E ec s in he Class oom: Lea ning om Gende and Race Va ia ion". NBER Wo king
Pape s 7867, Na ional Bu eau o Economic Resea ch, Inc.
De Paola, M. & Scoppa, V. (2010). "Pee g oup e ec s on he academic pe o mance o I alian s uden s".
Applied Economics, Taylo and F ancis Jou nals, ol. 42(17), pages 2203-2215.
Feins ein, L. & Symons, J. (1999). “A ainmen in seconda y school”. Ox o d Economic Pape s, Ox o d
Uni e si y P ess, Vol. 51 (2), pages 300-321.
Michael, F. (2003). "Educa ional P oduc ion, Endogenous Pee G oup Fo ma ion and Class Composi ion -
E idence om he PISA 2000 S udy". IZA discussion pape No. 714, Ins i u e o he S udy o Labo , Bonn.
Fos e , G. (2006). "I 's no you pee s, and i 's no you iends: Some p og ess owa d unde s anding he
educa ional pee e ec mechanism". Jou nal o Public Economics, Else ie , ol. 90(8-9), pages 1455-1475,
Sep embe .
Go a d, S., Taylo , C., & Fi z, J. (2003). Schools, Ma ke s and Choice Policies. Rou ledgeFa me .
Hanushek, E.A., Ma kman, J.M., Kain, J.F. & Ri kin, S. (2003). "Does pee abili y a ec s uden
achie emen ?". Jou nal o Applied Econome ics, John Wiley & Sons, L d., ol. 18(5), pages 527-544.
Lauzon D. (2004), "Va iance es ima ion wi h plausible alue achie emen da a: Two STATA p og ams o use
wi h YITS/PISA da a". In o ma ion and Technical Bulle in, S a is ics Canada, O awa.
La y, V. & Pase man, M.D. & Schlosse , A. (2008). "Inside he Black o Box o Abili y Pee E ec s: E idence
om Va ia ion in he P opo ion o Low Achie e s in he Class oom". NBER Wo king Pape s 14415, Na ional
Bu eau o Economic Resea ch, Inc.
Macdonald, K. (2008). "PV: S a a module o pe o m es ima ion wi h plausible alues". S a is ical So wa e
Componen s S456951, Bos on College Depa men o Economics, e ised 05 Sep 2008.
Manski, C.F. (1993). "Iden i ica ion o Endogenous Social E ec s: The Re lec ion P oblem". Re iew o
Economic S udies, Blackwell Publishing, ol. 60(3), pages 531-42, July.
Manski, C.F. (2000). "Economic Analysis o Social In e ac ions". Jou nal o Economic Pe spec i es, Ame ican
Economic Associa ion, ol. 14(3), pages 115-136, Summe .
McEwan, P.J. (2003). "Pee e ec s on s uden achie emen : e idence om Chile". Economics o Educa ion
Re iew, Else ie , ol. 22(2), pages 131-141, Ap il.
Micklew igh , J., Schnep , S.V. & Sil a, P.N. (2010). "Pee e ec s and measu emen e o : he impac o
sampling a ia ion in school su ey da a". DoQSS Wo king Pape s 10-13, Depa men o Quan i a i e Social
Science - Ins i u e o Educa ion, Uni e si y o London.
O ganisa ion o Economic Co-ope a ion and De elopmen (OECD) 2009. PISA 2006 echnical epo . Pa is:
OECD.
23
Rang id, B.S. (2007). "School composi ion e ec s in Denma k: quan ile eg ession e idence om PISA 2000".
Empi ical Economics, Sp inge , ol. 33(2), pages 359-388, Sep embe .
Raymond, J.L. & Roig, L. (2010) “In a i m human capi al ex e nali ies and selec ion bias”, Applied
Economics, Taylo and F ancis Jou nals ( o hcoming).
Robe son, D. & Symons, J. (2003). "Do Pee G oups Ma e ? Pee G oup e sus Schooling E ec s on
Academic A ainmen " Economica, London School o Economics and Poli ical Science, ol. 70(277), pages 31-
53, Feb ua y.
Sace do e, B. (2001). "Pee E ec s Wi h Random Assignmen : Resul s Fo Da mou h Roomma es". The
Qua e ly Jou nal o Economics, MIT P ess, ol. 116(2), pages 681-704, May.
Schneeweis, N. & Win e -Ebme , R. (2007). "Pee e ec s in Aus ian schools". Empi ical Economics, Sp inge ,
ol. 32(2), pages 387-409, May.
Zimme man, D.J. (2003). "Pee E ec s in Academic Ou comes: E idence om a Na u al Expe imen ". The
Re iew o Economics and S a is ics, MIT P ess, ol. 85(1), pages 9-23, No embe .
24
TABLES AND FIGURES
Table 1: Desc ip ion o a iables and summa y s a is ics
VARI
ABLE DESCRIPTION
mean
s.d.
INDIVIDUAL CONTROLS
age S uden ’s age in yea s 15.83 0.286
emale Female s uden (0-1) 0.496 0.5
4 h g ade S uden a ending 4 h g ade (0-1) 0.605 0.489
3 d g ade S uden a ending 3 d g ade (0-1) 0.328 0.469
2nd o 1s g ade S uden a ending 1s g ade (0-1) 0.067 0.249
immig an o i s gen. S uden ’s immig a ion s a us =
i s gene a ion (0-1) 0.059 0.236
na ional language a home Na i e-bo n immig an s uden speaking
he na ional language a home (0-1) 0.974 0.159
na i e- o eigne language a home Na i e-bo n s uden speaking
a o eign language a home (0-1) 0.006 .0075
immig an - o eigne language a home Immig an s uden speaking
a o eign language a home (0-1) 0.02 .141
FAMILY CONTROLS
amily socio-econ. s a us Highes socio-economic s a us index
(ISEI) o he pa en s 44.95 17
a he 's schooling Pa en al yea s o schooling
(based on comple ed ISCED97 le els) 9.945 4.342
mo he 's schooling Ma e nal yea s o schooling
(based on comple ed ISCED97 le els) 9.892 4.269
mo he wo king Mo he in egula wo k (0-1) 0.712 0.453
# books a home Numbe o books a home
(mid poin o o iginal ca ego ies) 155 154.7
educa ional esou ces PISA index o home educa ional esou ces 0.238 0.955
SCHOOL CONTROLS
% immig an s a school P opo ion o i s gene a ion immig an s 0.057 0.092
% gi ls a school P opo ion o emale s uden s 0.497 0.073
% pa - ime eache s P opo ion o pa - ime eache s 0.156 0.419
abili y s eaming be ween S uden s g ouped by abili y
be ween classes (0-1) 0.501 0.5
abili y s eaming wi hin S uden s a e g ouped by abili y
wi hin each class (0-1) 0.541 0.498
pc o ins uc ion a io Ra io o PCs used o ins uc ion o school size 0.099 0.089
public school Public school (0-1) 0.65 0.477
p i a e public- unded Public unded p i a e school (conce ada) (0-1) 0.247 0.431
p i a e school P i a e school (0-1) 0.103 0.304
lack quali ied eache s Sho age o quali ied eache s (0-1) 0.102 0.302
school size To al en ollmen in school 695.7 347.1
s uden / eache a io S uden eache a io 12.4 4.354
ex . guidance employee Ex e nal employee p o iding ca ee
guidance o s uden s (0-1) 0.8 0.4
budge au onomy School wi h budge au onomy (0-1) 0.566 0.496
eache hi ing au onomy School wi h eache hi ing au onomy (0-1) 0.315 0.464
ex book au onomy School wi h ex book selec ion au onomy (0-1) 0.927 0.26
cou se con en au onomy School wi h cou se con en au onomy (0-1) 0.555 0.497
cou se o e au onomy School wi h cou se o e au onomy (0-1) 0.522 0.499
No e: desc ip i e s a is ics compu ed wi h he inal s uden weigh .
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
2006
CREAP2006-01
Ma as, A. (GEAP); Raymond, J.Ll. (GEAP)
"Economic de elopmen and changes in ca owne ship pa e ns"
(Juny 2006)
CREAP2006-02
T illas, F. (IEB); Mon olio, D. (IEB); Duch, N. (IEB)
"P oduc i e e iciency and egula o y e o m: The case o Vehicle Inspec ion Se ices"
(Se emb e 2006)
CREAP2006-03
Bel, G. (PPRE-IREA); Fageda, X. (PPRE-IREA)
"Fac o s explaining local p i a iza ion: A me a- eg ession analysis"
(Oc ub e 2006)
CREAP2006-04
Fe nàndez-Villadangos, L. (PPRE-IREA)
"A e wo-pa a i s e icien when consume s plan ahead?: An empi ical s udy"
(Oc ub e 2006)
CREAP2006-05
A ís, M. (AQR-IREA); Ramos, R. (AQR-IREA); Su iñach, J. (AQR-IREA)
"Job losses, ou sou cing and eloca ion: Empi ical e idence using mic oda a"
(Oc ub e 2006)
CREAP2006-06
Alcañiz, M. (RISC-IREA); Cos a, A.; Guillén, M. (RISC-IREA); Luna, C.; Ro i a, C.
"Calcula ion o he a iance in su eys o he economic clima e”
(No emb e 2006)
CREAP2006-07
Albala e, D. (PPRE-IREA)
"Lowe ing blood alcohol con en le els o sa e li es: The Eu opean Expe ience”
(Desemb e 2006)
CREAP2006-08
Ga ido, A. (IEB); A qué, P. (IEB)
“The choice o banking i m: A e he in e es a e a signi ican c i e ia?”
(Desemb e 2006)

SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
CREAP2006-09
Sega a, A. (GRIT); Te uel-Ca izosa, M. (GRIT)
"P oduc i i y g ow h and compe i ion in spanish manu ac u ing i ms:
Wha has happened in ecen yea s?”
(Desemb e 2006)
CREAP2006-10
Andono a, V.; Díaz-Se ano, Luis. (CREB)
"Poli ical ins i u ions and he de elopmen o elecommunica ions”
(Desemb e 2006)
CREAP2006-11
Raymond, J.L.(GEAP); Roig, J.L.. (GEAP)
"Capi al humano: un análisis compa a i o Ca alunya-España”
(Desemb e 2006)
CREAP2006-12
Rod íguez, M.(CREB); S oyano a, A. (CREB)
"Changes in he demand o p i a e medical insu ance ollowing a shi in ax incen i es”
(Desemb e 2006)
CREAP2006-13
Royuela, V. (AQR-IREA); Lambi i, D.; Biagi, B.
"Economía u bana y calidad de ida. Una e isión del es ado del conocimien o en España”
(Desemb e 2006)
CREAP2006-14
Cama e o, M.; Ca ion-i-Sil es e, J.LL. (AQR-IREA).;Tama i , C.
"New e idence o he eal in e es a e pa i y o OECD coun ies using panel uni oo es s wi h b eaks”
(Desemb e 2006)
CREAP2006-15
Ka anassou, M.; Sala, H. (GEAP).;Snowe , D. J.
"The mac oeconomics o he labo ma ke : Th ee undamen al iews”
(Desemb e 2006)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
2007
XREAP2007-01
Cas any, L (AQR-IREA); López-Bazo, E. (AQR-IREA).;Mo eno , R. (AQR-IREA)
"Decomposing di e ences in o al ac o p oduc i i y ac oss i m size”
(Ma ç 2007)
XREAP2007-02
Raymond, J. Ll. (GEAP); Roig, J. Ll. (GEAP)
“Una p opues a de e aluación de las ex e nalidades de capi al humano en la emp esa"
(Ab il 2007)
XREAP2007-03
Du án, J. M. (IEB); Es elle , A. (IEB)
“An empi ical analysis o weal h axa ion: Equi y s. Tax compliance”
(Juny 2007)
XREAP2007-04
Ma as, A. (GEAP); Raymond, J.Ll. (GEAP)
“C oss-sec ion da a, disequilib ium si ua ions and es ima ed coe icien s: e idence om ca owne ship
demand”
(Juny 2007)
XREAP2007-05
Jo e-Mon seny, J. (IEB); Solé-Ollé, A. (IEB)
“Tax di e en ials and agglome a ion economies in in a egional i m loca ion”
(Juny 2007)
XREAP2007-06
Ál a ez-Albelo, C. (CREB); He nández-Ma ín, R.
“Explaining high economic g ow h in small ou ism coun ies wi h a dynamic gene al equilib ium model”
(Juliol 2007)
XREAP2007-07
Duch, N. (IEB); Mon olio, D. (IEB); Media illa, M.
“E alua ing he impac o public subsidies on a i m’s pe o mance: a quasi-expe imen al app oach”
(Juliol 2007)
XREAP2007-08
Sega a-Blasco, A. (GRIT)
“Inno a ion sou ces and p oduc i i y: a quan ile eg ession analysis”
(Oc ub e 2007)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
XREAP2007-09
Albala e, D. (PPRE-IREA)
“Shi ing dea h o hei Al e na i es: The case o Toll Mo o ways”
(Oc ub e 2007)
XREAP2007-10
Sega a-Blasco, A. (GRIT); Ga cia-Que edo, J. (IEB); Te uel-Ca izosa, M. (GRIT)
“Ba ie s o inno a ion and public policy in ca alonia”
(No emb e 2007)
XREAP2007-11
Bel, G. (PPRE-IREA); Foo e, J.
“Compa ison o ecen oll oad concession ansac ions in he Uni ed S a es and F ance”
(No emb e 2007)
XREAP2007-12
Sega a-Blasco, A. (GRIT);
“Inno a ion, R&D spillo e s and p oduc i i y: he ole o knowledge-in ensi e se ices”
(No emb e 2007)
XREAP2007-13
Be múdez Mo a a, Ll. (RFA-IREA); Guillén Es any, M. (RFA-IREA), Solé Au ó, A. (RFA-IREA)
“Impac o de la inmig ación sob e la espe anza de ida en salud y en discapacidad de la población
española”
(No emb e 2007)
XREAP2007-14
Calaeys, P. (AQR-IREA); Ramos, R. (AQR-IREA), Su iñach, J. (AQR-IREA)
“Fiscal sus ainabili y ac oss go e nmen ie s”
(Desemb e 2007)
XREAP2007-15
Sánchez Hugalbe, A. (IEB)
“In luencia de la inmig ación en la elección escola ”
(Desemb e 2007)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
2008
XREAP2008-01
Du án Wei kamp, C. (GRIT); Ma ín Bo a ull, M. (GRIT) ; Pablo Ma í, F.
“Economic e ec s o oad accessibili y in he Py enees: Use pe spec i e”
(Gene 2008)
XREAP2008-02
Díaz-Se ano, L.; S oyano a, A. P. (CREB)
“The Causal Rela ionship be ween Indi idual’s Choice Beha io and Sel -Repo ed Sa is ac ion: he Case
o Residen ial Mobili y in he EU”
(Ma ç 2008)
XREAP2008-03
Ma as, A. (GEAP); Raymond, J. L. (GEAP); Roig, J. L. (GEAP)
“Ca owne ship and access o jobs in Spain”
(Ab il 2008)
XREAP2008-04
Bel, G. (PPRE-IREA) ; Fageda, X. (PPRE-IREA)
“P i a iza ion and compe i ion in he deli e y o local se ices: An empi ical examina ion o he dual
ma ke hypo hesis”
(Ab il 2008)
XREAP2008-05
Ma as, A. (GEAP); Raymond, J. L. (GEAP); Roig, J. L. (GEAP)
“Job accessibili y and employmen p obabili y”
(Maig 2008)
XREAP2008-06
Bashe , S. A.; Ca ión, J. Ll. (AQR-IREA)
Decons uc ing Shocks and Pe sis ence in OECD Real Exchange Ra es
(Juny 2008)
XREAP2008-07
San omá, E. (IEB); Ramos, R. (AQR-IREA); Simón, H.
Po abilidad del capi al humano y asimilación de los inmig an es. E idencia pa a España
(Juliol 2008)
XREAP2008-08
Bashe , S. A.; Ca ión, J. Ll. (AQR-IREA)
P ice le el con e gence, pu chasing powe pa i y and mul iple s uc u al b eaks: An applica ion o US
ci ies
(Juliol 2008)
XREAP2008-09
Be múdez, Ll. (RFA-IREA)
A p io i a emaking using bi a ia e poisson eg ession models
(Juliol 2008)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
XREAP2008-10
Solé-Ollé, A. (IEB), Ho as Rico, M. (IEB)
Does u ban sp awl inc ease he cos s o p o iding local public se ices? E idence om Spanish
municipali ies
(No emb e 2008)
XREAP2008-11
Te uel-Ca izosa, M. (GRIT), Sega a-Blasco, A. (GRIT)
Immig a ion and Fi m G ow h: E idence om Spanish ci ies
(No emb e 2008)
XREAP2008-12
Duch-B own, N. (IEB), Ga cía-Que edo, J. (IEB), Mon olio, D. (IEB)
Assessing he assigna ion o public subsidies: Do he expe s choose he mos e icien R&D p ojec s?
(No emb e 2008)
XREAP2008-13
Bilo kach, V., Fageda, X. (PPRE-IREA), Flo es-Fillol, R.
Scheduled se ice e sus pe sonal anspo a ion: he ole o dis ance
(Desemb e 2008)
XREAP2008-14
Albala e, D. (PPRE-IREA), Gel, G. (PPRE-IREA)
Tou ism and u ban anspo : Holding demand p essu e unde supply cons ain s
(Desemb e 2008)

SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
2009
XREAP2009-01
Calonge, S. (CREB); Tejada, O.
“A heo e ical and p ac ical s udy on linea e o ms o dual axes”
(Feb e 2009)
XREAP2009-02
Albala e, D. (PPRE-IREA); Fe nández-Villadangos, L. (PPRE-IREA)
“Explo ing De e minan s o U ban Mo o cycle Acciden Se e i y: The Case o Ba celona”
(Ma ç 2009)
XREAP2009-03
Bo ell, J. R. (PPRE-IREA); Fe nández-Villadangos, L. (PPRE-IREA)
“Assessing excess p o i s om di e en en y egula ions”
(Ab il 2009)
XREAP2009-04
San omá, E. (IEB); Ramos, R. (AQR-IREA), Simon, H.
“Los sala ios de los inmig an es en el me cado de abajo español. ¿Impo a el o igen del capi al
humano?”
(Ab il 2009)
XREAP2009-05
Jiménez, J. L.; Pe digue o, J. (PPRE-IREA)
“(No)compe i ion in he Spanish e ailing gasoline ma ke : a a iance il e app oach”
(Maig 2009)
XREAP2009-06
Ál a ez-Albelo,C. D. (CREB), Man esa, A. (CREB), Pigem-Vigo, M. (CREB)
“In e na ional ade as he sole engine o g ow h o an economy”
(Juny 2009)
XREAP2009-07
Callejón, M. (PPRE-IREA), O ún V, M.
“The Black Box o Business Dynamics”
(Se emb e 2009)
XREAP2009-08
Lucena, A. (CREB)
“The an eceden s and inno a ion consequences o o ganiza ional sea ch: empi ical e idence o Spain”
(Oc ub e 2009)
XREAP2009-09
Domènech Campmajó, L. (PPRE-IREA)
“Compe i ion be ween TV Pla o ms”
(Oc ub e 2009)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
XREAP2009-10
Solé-Au ó, A. (RFA-IREA),Guillén, M. (RFA-IREA), C immins, E. M.
“Heal h ca e u iliza ion among immig an s and na i e-bo n popula ions in 11 Eu opean coun ies. Resul s
om he Su ey o Heal h, Ageing and Re i emen in Eu ope”
(Oc ub e 2009)
XREAP2009-11
Sega a, A. (GRIT), Te uel, M. (GRIT)
“Small i ms, g ow h and inancial cons ain s”
(Oc ub e 2009)
XREAP2009-12
Ma as, A. (GEAP), Raymond, J.Ll. (GEAP), Ruiz, A. (GEAP)
“T a ic o ecas s unde unce ain y and capaci y cons ain s”
(No emb e 2009)
XREAP2009-13
Sole-Ollé, A. (IEB)
“In e - egional edis ibu ion h ough in as uc u e in es men : ac ical o p og amma ic?”
(No emb e 2009)
XREAP2009-14
Del Ba io-Cas o, T., Ga cía-Que edo, J. (IEB)
“The de e minan s o uni e si y pa en ing: Do incen i es ma e ?”
(No emb e 2009)
XREAP2009-15
Ramos, R. (AQR-IREA), Su iñach, J. (AQR-IREA), A ís, M. (AQR-IREA)
“Human capi al spillo e s, p oduc i i y and egional con e gence in Spain”
(No emb e 2009)
XREAP2009-16
Ál a ez-Albelo, C. D. (CREB), He nández-Ma ín, R.
“The commons and an i-commons p oblems in he ou ism economy”
(Desemb e 2009)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
2010
XREAP2010-01
Ga cía-López, M. A. (GEAP)
“The Accessibili y Ci y. When T anspo In as uc u e Ma e s in U ban Spa ial S uc u e”
(Feb e 2010)
XREAP2010-02
Ga cía-Que edo, J. (IEB), Mas-Ve dú, F. (IEB), Polo-O e o, J. (IEB)
“Which i ms wan PhDs? The e ec o he uni e si y-indus y ela ionship on he PhD labou ma ke ”
(Ma ç 2010)
XREAP2010-03
Pi , D., Guillén, M. (RFA-IREA)
“An in oduc ion o pa ame ic and non-pa ame ic models o bi a ia e posi i e insu ance claim se e i y
dis ibu ions”
(Ma ç 2010)
XREAP2010-04
Be múdez, Ll. (RFA-IREA), Ka lis, D.
“Modelling dependence in a a emaking p ocedu e wi h mul i a ia e Poisson eg ession models”
(Ab il 2010)
XREAP2010-05
Di Paolo, A. (IEB)
“Pa en al educa ion and amily cha ac e is ics: educa ional oppo uni ies ac oss coho s in I aly and
Spain”
(Maig 2010)
XREAP2010-06
Simón, H. (IEB), Ramos, R. (AQR-IREA), San omá, E. (IEB)
“Mo ilidad ocupacional de los inmig an es en una economía de bajas cuali icaciones. El caso de España”
(Juny 2010)
XREAP2010-07
Di Paolo, A. (GEAP & IEB), Raymond, J. Ll. (GEAP & IEB)
“Language knowledge and ea nings in Ca alonia”
(Juliol 2010)
XREAP2010-08
Bolancé, C. (RFA-IREA), Alemany, R. (RFA-IREA), Guillén, M. (RFA-IREA)
“P edic ion o he economic cos o indi idual long- e m ca e in he Spanish popula ion”
(Se emb e 2010)
XREAP2010-09
Di Paolo, A. (GEAP & IEB)
“Knowledge o ca alan, public/p i a e sec o choice and ea nings: E idence om a double sample
selec ion model”
(Se emb e 2010)
SÈRIE DE DOCUMENTS DE TREBALL DE LA XREAP
XREAP2010-10
Coad, A., Sega a, A. (GRIT), Te uel, M. (GRIT)
“Like milk o wine: Does i m pe o mance imp o e wi h age?”
(Se emb e 2010)
XREAP2010-11
Di Paolo, A. (GEAP & IEB), Raymond, J. Ll. (GEAP & IEB), Cale o, J. (IEB)
“Explo ing educa ional mobili y in Eu ope”
(Oc ub e 2010)
XREAP2010-12
Bo ell, A. (GiM-IREA), Fe nández-Villadangos, L. (GiM-IREA)
“Clus e ing o sca e ing: he unde lying eason o egula ing dis ance among e ail ou le s”
(Desemb e 2010)
XREAP2010-13
Di Paolo, A. (GEAP & IEB)
“School composi ion e ec s in Spain”
(Desemb e 2010)