Be linski, Samuel G.; Gage e-Mi anda, Jessica
Wo king Pape
En o cemen spillo e s unde di e en ne wo ks: The case
o quo as o pe sons wi h disabili ies in B azil
IDB Wo king Pape Se ies, No. IDB-WP-1613
P o ided in Coope a ion wi h:
In e -Ame ican De elopmen Bank (IDB), Washing on, DC
Sugges ed Ci a ion: Be linski, Samuel G.; Gage e-Mi anda, Jessica (2024) : En o cemen spillo e s
unde di e en ne wo ks: The case o quo as o pe sons wi h disabili ies in B azil, IDB Wo king
Pape Se ies, No. IDB-WP-1613, In e -Ame ican De elopmen Bank (IDB), Washing on, DC,
h ps://doi.o g/10.18235/0013112
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/302219
S anda d-Nu zungsbedingungen:
Die Dokumen e au EconS o dü en zu eigenen wissenscha lichen
Zwecken und zum P i a geb auch gespeiche und kopie we den.
Sie dü en die Dokumen e nich ü ö en liche ode komme zielle
Zwecke e iel äl igen, ö en lich auss ellen, ö en lich zugänglich
machen, e eiben ode ande wei ig nu zen.
So e n die Ve asse die Dokumen e un e Open-Con en -Lizenzen
(insbesonde e CC-Lizenzen) zu Ve ügung ges ell haben soll en,
gel en abweichend on diesen Nu zungsbedingungen die in de do
genann en Lizenz gewäh en Nu zungs ech e.
Te ms o use:
Documen s in EconS o may be sa ed and copied o you pe sonal
and schola ly pu poses.
You a e no o copy documen s o public o comme cial pu poses, o
exhibi he documen s publicly, o make hem publicly a ailable on he
in e ne , o o dis ibu e o o he wise use he documen s in public.
I he documen s ha e been made a ailable unde an Open Con en
Licence (especially C ea i e Commons Licences), you may exe cise
u he usage igh s as speci ied in he indica ed licence.
h ps://c ea i ecommons.o g/licenses/by/3.0/igo/
En o cemen Spillo e s unde Di e en
Ne wo ks:
The Case o Quo as o Pe sons wi h Disabili ies in B azil
Samuel Be linski
Jessica Gage e-Mi anda
WORKING PAPER No IDB-WP-1613
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
Augus 2024
* In e -
A
me ican De elopmen Bank and IZA
** Uni e si y o Milano-Bicocca
En o cemen Spillo e s unde Di e en
Ne wo ks:
The Case o Quo as o Pe sons wi h Disabili ies in B azil
Samuel Be linski*
Jessica Gage e-Mi anda**
In e -
A
me ican De elopmen Bank
Depa men o Resea ch and Chie Economis
Augus 2024
Ca aloging-in-Publica ion da a p o ided by he
In e -Ame ican De elopmen Bank
Felipe He e a Lib a y
Be linski, Samuel, 1970-
En o cemen spillo e s unde di e en ne wo ks: he case o quo as o
pe sons wi h disabili ies in B azil / Samuel Be linski, Jessica Gage e-Mi anda.
p. cm. — (IDB Wo king Pape Se ies ; 1613)
Includes bibliog aphical e e ences.
1. People wi h disabili ies-B azil. 2. People wi h disabili ies-Legal s a us, laws,
e c.-B azil. 3. People wi h disabili ies-Employmen -B azil. I. Gage e-Mi anda,
Jessica. II. In e -Ame ican De elopmen Bank. Depa men o Resea ch and
Chie Economis . III. Ti le. IV. Se ies.
IDB-WP-1613
h p://www.iadb.o g
Copy igh © 2024 In e -Ame ican De elopmen Bank ("IDB"). This wo k is subjec o a C ea i e
Commons license CC BY 3.0 IGO (h ps://c ea i ecommons.o g/licenses/by/3.0/igo/legalcode). The
e ms and condi ions indica ed in he URL link mus be me and he espec i e ecogni ion mus be
g an ed o he IDB.
Fu he o sec ion 8 o he abo e license, any media ion ela ing o dispu es a ising unde such license
shall be conduc ed in acco dance wi h he WIPO Media ion Rules. Any dispu e ela ed o he use o
he wo ks o he IDB ha canno be se led amicably shall be submi ed o a bi a ion pu suan o he
Uni ed Na ions Commission on In e na ional T ade Law (UNCITRAL) ules. The use o he IDB's name
o any pu pose o he han o a ibu ion, and he use o IDB's logo shall be subjec o a sepa a e
w i en license ag eemen be ween he IDB and he use and is no au ho ized as pa o his license.
No e ha he URL link includes e ms and condi ions ha a e an in eg al pa o his license.
The opinions exp essed in his wo k a e hose o he au ho s and do no necessa ily e lec he iews o
he In e -Ame ican De elopmen Bank, i s Boa d o Di ec o s, o he coun ies hey ep esen .
Abs ac
This s udy examines labo law en o cemen spillo e s in B azil’s highly in o mal
economy, ocusing on disabili y quo a en o cemen o o mal i ms. New inspec ion
p ocedu es inc eased compliance h ough heigh ened inspec ions and ines, boos ing
disabili y hi ing. We in es iga e spillo e e ec s ac oss a ious i m ne wo ks:
neighbo hood, owne ship, and human esou ces specialis s. Resul s show ha spillo e s
can ha e up o wice he impac on disabili y employmen compa ed o di ec ines.
These indings highligh he po en ial o a ge ed en o cemen s a egies o ampli y
policy e ec i eness beyond di ec ly a ec ed i ms e en in de eloping economies
cha ac e ized by low compliance wi h employmen laws.
JEL codes: I38, J68, K31
Keywo ds: En o cemen spillo e s, Ne wo ks, Pe sons wi h disabili y, B azil
∗We acknowledge he aluable commen s o Ma hew S. Johnson, Gab iel Ulyssea, and pa icipan s a
he 2023 Wo kshop on Ne wo ks and De elopmen , he NEUDC 2023 Con e ence, and semina s a CAF –
De elopmen Bank o La in Ame ica and he Ca ibbean, he In e -Ame ican De elopmen Bank, Uni e si y
o Milano-Bicocca, Uni e si y o P e o ia, and he Uni e si y o Sao Paulo. The opinions exp essed in
his publica ion a e hose o he au ho s and do no necessa ily e lec he iews o he In e -Ame ican
De elopmen Bank, i s Boa d o Di ec o s, o he coun ies hey ep esen .
1 In oduc ion
Compliance is a c ucial issue policymake s ace in i ually all legisla i e sphe es, such as
ax (Slem od,2019), a ic (Lu e al.,2016), en i onmen al (Shimshack,2014) and labo
(Almeida and Ronconi,2016) egula ions. De e ence achie ed h ough he inspec ion and
punishmen o non-complie s is cos ly and eaches e y ew indi iduals and i ms in mos
en i onmen s. To imp o e gene al compliance, in o ma ion abou he inspec ion and pun-
ishmen o a ew mus a el be ween indi iduals and i ms (see, Slem od,2019;Johnson,
2020). This pape in es iga es how in o ma ion on employmen ines is ansmi ed h ough
di e en i m ne wo ks. In pa icula , we s udy he case o quo as o pe sons wi h disabil-
i ies in B azil (hence o h Quo a Law o QL) and explo e i) he impac o new inspec ion
and punishmen p ocedu es ha inc eased he egula o y en o cemen o he QL; and ii)
how i ms lea n om di e en ne wo ks abou he QL en o cemen and hei isk o being
punished.
The B azilian Quo a Law manda es ha companies employing mo e han 100 wo ke s
alloca e a leas wo pe cen o hei employmen posi ions o pe sons wi h disabili ies. This
legal equi emen p o ides an ideal se ing o ou in es iga ion o se e al easons. Fi s ,
e en hough he equi emen was in oduced in 1991, labo egula o y o ices only s a ed o
e ec i ely en o ce i and punish non-complie s abou 20 yea s la e . Hence, o a conside able
ime, i ms’ p io was ha such a de ju e law was no a de ac o law, so hey would no
ace he isk o punishmen in case o non-compliance. En o cemen became igh e a e
he in oduc ion o an adminis a i e ac in 2012, enabling us o s udy i ms’ adjus men
p ocesses once a pa icula law s a s o be en o ced. Second, he B azilian con ex allows
us o combine di e en ypes o da a o in es iga e how law en o cemen spills o e unde
di e en i ms’ ne wo ks. Thi d, B azil is a good example o he gap be ween de ju e and
de ac o laws in de eloping coun ies, which e lec s endemic p oblems o go e nmen s in
en o cing compliance wi h laws (Acemoglu e al.,2015).
We employ a di e ence in discon inui y design on ma ched employee-employe da a and
documen ha he 2012 Adminis a i e Ac led o an inc ease in inspec ions and ines issued
due o non-compliance wi h he Quo a Law and, consequen ly, o an inc ease in he hi ing
o pe sons wi h a disabili y a e ha yea by abou 7%. Fu he mo e, we show non-causal
e idence o wha migh be in e p e ed as en o cemen spillo e s. The impac o he 2012
Adminis a i e Ac on issuing QL ines dec eases as en o cemen capaci y dec eases. How-
e e , he impac on hi ing pe sons wi h disabili ies is independen o he local en o cemen
capaci y.
Finally, we p esen he key con ibu ion o he pape whe e we es o he p esence o
1
en o cemen spillo e s by employing a s acked di e ences-in-di e ences design o unde s and
how he occu ence o a QL ine in a i m’s ne wo k impac s he likelihood ha such a i m
will inc ease i s hi ing o wo ke s wi h disabili ies. We look a h ee di e en ne wo ks:
neighbo ne wo k (i.e., i ms loca ed in he same zip code o a speci ic i m), owne ne wo k
(i.e., i ms ha belong o he same owne o a speci ic i m o an associa e o such an owne )
and HR wo ke s ne wo k (i.e., i ms whe e he HR wo ke s o a speci ic i m we e wo king
be o e joining such a i m). We compa e i ms wi hin ne wo ks whe e a Quo a Law ine was
issued o i ms wi hin ne wo ks whe e a Quo a Law ine will be issued in he u u e bu ha e
no ecei ed one ye .
We ind s ong e idence o en o cemen spillo e s when a QL ine happens in he neighbo ,
owne , o HR wo ke s’ ne wo ks. I ano he i m in he ne wo k o a i m i ecei es a QL ine,
his inc eases he numbe o wo ke s wi h a disabili y p esen in i m iin he ollowing yea s
by 7.4% in he neighbo ne wo k, 7% in he owne ne wo k, and 4.6% in he HR wo ke s’
ne wo k. A back-o - he-en elope calcula ion indica es ha he o al numbe o wo ke s wi h
disabili ies hi ed due o spillo e e ec s in he neighbo and HR wo ke s’ ne wo ks is abou
wice as la ge as he di ec impac o ecei ing a QL ine. This igu e dec eases conside ably
o he owne ne wo ks (only 30% as la ge as he di ec impac ) due o he small size o such
ne wo ks.
Spillo e s a e s onge o i ms ha we e no complying wi h he Quo a Law when he
ine was issued. We show ha he likelihood o being inspec ed does no inc ease a e he
occu ence o a QL ine in any o he i m’s ne wo ks, which sugges s ha he sp ead o
in o ma ion and no he local inc ease in en o cemen is causing he eme gence o spillo e s.
This pape con ibu es o wo di e en s eams o he li e a u e. Fi s , i con ibu es
o he li e a u e on egula ions o imp o e he employmen oppo uni ies o people wi h a
disabili y. Ti le I o he Ame icans wi h Disabili ies Ac (ADA) o 1990 ensu es ha p i a e
employe s, s a e and local go e nmen s, employmen agencies, and labo unions canno dis-
c imina e agains quali ied indi iduals wi h disabili ies du ing job applica ion p ocedu es,
hi ing, i ing, ad ancemen , compensa ion, job aining, and o he aspec s o employmen .
E idence on i s impac is mixed, wi h some pape s showing a nega i e e ec on labo ma ke
pa icipa ion o pe sons wi h disabili ies (Acemoglu and Ang is ,2001;DeLei e,2000a,b),
while o he s udies dispu e such indings (Ho chkiss,2004;Jolls and P esco ,2004).1Ou -
side o he Uni ed S a es, quo a sys ems like he one analyzed in his s udy ha e been adop ed
by o e wo- hi ds o OECD coun ies (OECD,2003). In Aus ia (Lali e e al.,2013), Hun-
ga y (K ekó and Telegdy,2022), and Japan (Mo i and Sakamo o,2018) , esea ch has ound
1Fo a e iew on he impac s o he ADA and o he policies a ge ing he inclusion o pe sons wi h
disabili ies in he Uni ed S a es, see Li e mo e and Goodman (2009).
2
ha i ms comply wi h such egula ions. Pape s ha s udy he quo a sys em in de eloping
coun ies, whe e egula o y compliance is a guably wo se, usually le e age he ole o law
en o cemen in imp o ing compliance wi h he quo a o calcula e i s wel a e e ec s, like
Sze man (2022) and de Souza (2023), who also s udy he B azilian quo a sys em. In his
pape , we p o ide e idence ha en o cing he Quo a Law gene a es spillo e e ec s, which
should be aken in o conside a ion in any wel a e calcula ion.
Second, his pape con ibu es o he li e a u e in es iga ing he impac s o law en o ce-
men and, mo e speci ically, he eme gence o en o cemen spillo e s among i ms2. Inspec-
ions, audi ing, and ines ha e been p o en e ec i e in inc easing egula o y compliance
(see G ay and Shimshack,2011;Le ine e al.,2012, o examples on en i onmen al and oc-
cupa ional heal h and sa e y egula o s). Howe e , de eloping coun ies s uggle wi h low
en o cemen capaci y, challenging compliance e o s in many sec o s and locali ies (Almeida
and Ronconi,2016;Ponczek and Ulyssea,2022). Hence, such coun ies could signi ican ly
bene i om en o cemen spillo e s since he eme gence o such indi ec impac would ha e a
mul iplie e ec on each a omis ic en o cemen e o . The e idence on en o cemen spillo e s
so a comes om de eloped coun ies. In he Uni ed S a es, o ins ance, Shimshack and
Wa d (2005) and E ans e al. (2018) p o ide e idence ha en o cing en i onmen al egula-
ions inc eases u u e compliance o o he i ms loca ed in he same s a e whe e en o cemen
occu s bu may c ea e nega i e ex e nali ies in a eas ha a e no inspec ed. Also in he
Uni ed S a es, Johnson (2020) shows ha publicizing i ms’ heal h iola ions led o o he
i ms o comply mo e wi h such a egula ion. Howe e , he eme gence o such spillo e s in de-
eloping coun ies is a om ob ious due o he weake en o cemen capaci y and he la ge
gap be ween de ju e and de ac o laws. Besides being he i s pape o p esen e idence
o en o cemen spillo e s in de eloping coun ies, his pape also con ibu es o his li e a-
u e by showing how in o ma ion abou law en o cemen lows h ough di e en ne wo ks
connec ed o he i m, which can help in o m policymake s on how o imp o e a ge ing o
le e age he eme gence o spillo e s. Fo example, he esul s o a se o inspec ions could
be dissemina ed h ough in o ma ion le e s o companies in a i m’s ne wo k.
2The li e a u e has shown ha he e a e en o cemen spillo e s among indi iduals’ ne wo ks ( amily, co-
wo ke s, and neighbo s) in de eloped coun ies o di idend and capi al axa ion, commu e ax allowances,
and TV license paymen s (e.g., Als adsæ e e al.,2019;D ago e al.,2020;Pae zold and Winne ,2016;
Rincke and T axle ,2011).
3
2 Backg ound
The B azilian Quo a Law is an impo an example o he b oade emphasis ha bo h de-
eloping and de eloped coun ies a e placing on di e si y and inclusion policies.3This is an
impo an policy goal o e hical, e iciency, and edis ibu i e easons. Be linski e al. (2021)
es ima e ha in La in Ame ica and he Ca ibbean (LAC) coun ies, 88 million people we e
li ing wi h a disabili y in 2020 (a ound 15 pe cen o he popula ion). By 2050, his igu e
could ise by 60 million. People li ing wi h disabili ies ha e lowe educa ional a endance
and school comple ion a es and la ge gaps in labo ma ke ou comes wi h espec o hose
li ing wi hou disabili ies. Fo example, Be linski e al. (2021) epo ha he employmen
disabili y gap o people aged 25–34 in he eigh na ional censuses o LAC coun ies hey
analyze is, on a e age, 18.5 pe cen age poin s.
Go e nmen s o en implemen employmen quo as o p omo e he hi ing o pe sons wi h
disabili ies. (See, Mon e al.,2004;Fö s e ,2007). These quo as a e commonly used among
OECD and pa ne coun ies (Fö s e ,2007;OECD,2003). The speci ic egula ions ega d-
ing which i ms a e subjec o quo as and he pe cen age o acancies hey should ese e o
wo ke s wi h disabili ies a y ac oss coun ies. Majo co po a ions a e he p ima y ocus o
such measu es, wi h he ypical p opo ion o jobs se aside o indi iduals wi h disabili ies
ho e ing a ound 4%.
In B azil, he Quo a Law4is he mos ele an legisla ion ega ding he employmen
oppo uni ies o pe sons wi h disabili y. I es ablishes ha , espec i ely, i ms wi h mo e
han 100, 200, 500, and 1,000 employees mus ill a leas 2%, 3%, 4%, and 5% o hei
pay oll wi h people wi h a ce i ied disabili y.
Howe e , compliance wi h he Quo a Law has his o ically emained limi ed since i s
in oduc ion in 1991. Fo ins ance, in 2009, less han 30% o i ms wi h mo e han 100
wo ke s we e employing he minimum numbe o wo ke s wi h disabili ies es ablished by he
law. Compliance has been an issue in o he coun ies as well (see, OECD,2003) despi e
some e idence o he e ec i eness o quo as a inc easing he employmen o people wi h
3Fo example, he Uni ed Na ions’ Sus ainable De elopmen Goals o he yea 2030 aspi e o: SDG
4: “Ensu e inclusi e and equi able quali y educa ion and p omo e li elong lea ning oppo uni ies o all”;
SDG 5: “Achie e gende equali y and empowe all women and gi ls”; SDG 8: “P omo e sus ained, inclusi e
and sus ainable economic g ow h, ull and p oduc i e employmen , and decen wo k o all”; SDG 10: “Re-
duce inequali y wi hin and among coun ies.”; SDG 11: Make ci ies and human se lemen s inclusi e, sa e,
esilien , and sus ainable; and SDG 16, “Peace, jus ice and s ong ins i u ions,” which p omo es building
e ec i e, accoun able, and inclusi e ins i u ions a all le els o ensu e peace ul and inclusi e socie ies o
all. Addi ionally, his SFD is consis en wi h he social model o disabili y embedded in he 2008 Uni ed
Na ions Con en ion on he Righ s o Pe sons wi h Disabili ies, and i is aligned wi h ILO Con en ion 169
and a icles 3 and 4 o he UN Decla a ion on he Righ s o Indigenous Peoples, which ecognize hei igh
o make au onomous decisions ega ding hei de elopmen p io i ies.
4A . 93. o Law 8.213/1991.
4
duced by G embi e al. (2016), who combined he adi ional RD design wi h ideas om
he di e ence-in-di e ences design. The au ho s added a second dimension o he eg es-
sion discon inui y (RD) design, whe e a s uc u al change happened be ween wo di e en
pe iods while a discon inui y holds du ing bo h pe iods. In ou case, such a change is he
in oduc ion o he new inspec ion p ocedu es in 2012, while he h esholds de ined by he
Quo a Law emained he same du ing he whole pe iod. The main idea o he di e ences-
in-discon inui ies design is ha i akes he di e ence be ween he p e-2012 and pos -2012
discon inui ies in he Quo a Law h eshold o sepa a e he e ec o he Quo a Law om he
e ec o en o cing such a policy h ough he new inspec ion p ocedu es.
We iden i y he e ec o in oducing he new inspec ion p ocedu es h ough he ollowing
equa ion (es ima ed wi hin he bandwid h p oposed by Calonico e al. (2014a,b)):
Yim =β0+β1Pim +β2Sim (γ0+γ1Pim ) + T[α0+α1Pim +
Sim (δ0+δ1Pim )] + θm+θ +θm +εim .(1)
Whe e, Yim is he ou come o i m iloca ed a municipali y ma yea ,Sim = 1(Fi m Size ≥
100), and Pim =Fi m Size −100. Mo eo e , T= 1( ≥2012). Ou coe icien o in e es in
equa ion 1is δ0. I iden i ies he impac o su passing he cu -o h eshold a e 2012. The
es ima ion includes con ols o municipal and yea ixed e ec s (θmand θ , espec i ely)
and in e ac ions be ween hese wo (θm ). The inclusion o municipali y-by-yea ixed e ec s
in ou es ima ions add esses wo iden i ica ion h ea s. Fi s , i ensu es ha he supply
o pe sons wi h disabili y is held cons an ac oss municipali ies and ime. This alle ia es
conce ns ha he new inspec ion p ocedu es migh ha e changed no only i ms’ demand
o wo ke s wi h disabili y bu also he willingness o pe sons wi h disabili y o look o
jobs. Second, i con ols o ime-speci ic s ingency o labo inspec ions – o ins ance, new
openings o labo inspec ion o ices.
Th ee assump ions need o hold o iden i ica ion. Fi s , as in he adi ional RD design,
all po en ial ou comes should be con inuous a ound he discon inui y each yea . Second,
simila o he di e ence-in-di e ences design, he obse a ions jus below and abo e he
discon inui y mus ollow (local) pa allel ends in he coun e ac ual scena io o no new
inspec ion p ocedu es. Thi d, he e ec s o he new inspec ion p ocedu es should be inde-
penden o he h esholds o he Quo a Law.
A u he p oblem in ou analysis is measu emen e o in i ms’ size since ou a iables a e
agg ega ed by yea . To add ess his issue, we implemen a donu ing s a egy (Ba eca e al.,
2011), whe e we exclude i ms wi h sizes wi hin one o wo wo ke s om he 100 h esholds.
Th ough such exclusion, we a oid inclusion and exclusion e o s whe e we w ongly conside a
11
i m la ge o smalle han he QL h eshold. The donu s a egy also helps in add essing he
issue o po en ially endogenous manipula ion o he unning a iable a ound he h eshold,
e en hough Figu e A1 in he Appendix Ashows ha manipula ion is no a conce n in ou
analysis.
Table 2p esen s he impac o he 2012 Adminis a i e Ac on law en o cemen ( he
likelihood o ecei ing an inspec ion in columns (1) and (2) and o being ined due o non-
compliance wi h he Quo a Law in columns (3) and (4)15), and on he numbe o wo ke s
wi h disabili ies p esen in he i m (columns (5) and (6)). Columns (1), (3), and (5) p esen
es ima ions conside ing a donu ing o one, while columns (2), (4), and (6) p esen obus ness
checks implemen ing a donu ing o wo in he i m-size a iable.
The in oduc ion o he 2012 Adminis a i e Ac inc eases he likelihood ha i ms would
be inspec ed by 2.4 pe cen age poin s o a 5.3% inc ease. The inc eased likelihood o ecei ing
a Quo a Law ine, shown in column (3), is much mo e s iking: a e he 2012 Adminis a i e
Ac , such a likelihood inc eased by 1.2 pe cen age poin s o 44.4%. Finally, he esul in
column (5) shows ha i ms eac ed o he inc ease in QL en o cemen a e 2012 by hi ing
mo e pe sons wi h disabili ies. Mo e speci ically, we can see ha i ms la ge han 100
wo ke s inc eased hei numbe o wo ke s wi h a disabili y by 6.9%.16 The esul s a e qui e
simila ega dless o he donu ing implemen ed.
Figu e 3p esen s he dynamic esul s o such es ima ions, whe e we subs i u e he bi-
na y a iable Tin equa ion 1 o yea ly dummies. The igu e shows no di e ence be ween
i ms la ge and smalle han 100 wo ke s be o e 2012. Howe e , we obse e an inc ease in
inspec ions, Quo a Law ines, and he numbe o wo ke s wi h disabili ies o i ms la ge
han 100 wo ke s a e ha yea .
Recen wo k by Chen and Ro h (2024) shows ha one should be ca e ul when in e p e ing
esul s wi h log o in e se hype bolic sine ans o ma ions, especially i he ea men a ec s
he ex ensi e ma gins – in ou case, he likelihood o a i m passing om ha ing no wo ke s
wi h a disabili y o ha ing one o mo e. We p esen in Table A2 and Figu e A2 wo obus ness
checks o deal wi h his issue. Fi s , we p esen ex ensi e ma gin es ima ions, calcula ing
he likelihood ha i ms ha e a leas one employee wi h a disabili y. Second, we es ima e a
linea eg ession model wi h he numbe o wo ke s wi h disabili y as he dependen a iable.
Resul s a e he same o he ex ensi e ma gin es ima ion and poin in he same di ec ion as
15The likelihood o ecei ing a QL ine is ze o o i ms smalle han 100 wo ke s since hey do no need
o comply wi h he law. Hence, he esul s in columns (3) and (4) show he di e ence ac oss ime in he
likelihood ha i ms la ge han 100 will ecei e such a ine.
16As explained in sec ion 3, we use he hype bolic sine ans o ma ion o he numbe o wo ke s wi h a
disabili y. We include in Table 2 he elas ici y o he numbe o wo ke s wi h disabili ies, using he calcula ion
de i ed by Bellema e and Wichman (2020).
12
he main es ima ion in he eg ession model wi h he coun a iable, e en hough we do no
ha e enough powe o ejec he hypo hesis o null e ec s in he las case.
Table 2: The 2012 Adminis a i e Ac , Law En o cemen , and Wo ke s wi h Disabili ies
(1) (2) (3) (4) (5) (6)
Inspec ion QL ine Wo ke s w/ a disabili y
(hyp. sine ans.)
Yea >2012 X Dis . QL h eshold>0 0.020∗∗∗ 0.019∗∗∗ 0.013∗∗∗ 0.013∗∗∗ 0.062∗∗∗ 0.068∗∗∗
(0.006) (0.007) (0.003) (0.002) (0.016) (0.018)
N 446497 437528 331122 322136 222233 213288
Mean Dep. Va . 0.440 0.441 0.024 0.024 0.841 0.855
Elas ici y 0.064 0.070
h (le ) 33.305 33.305 19.665 19.665 13.910 13.910
h ( igh ) 118.828 118.828 127.249 127.249 66.489 66.489
R2 0.176 0.177 0.106 0.107 0.208 0.210
Donu ing 1 2 1 2 1 2
No e: This able p esen s es ima ions om Equa ion 1. Elas ici ies o wo ke s wi h disabili ies a e calcula ed based on
Bellema e and Wichman (2020). All es ima ions include ci y-by-yea ixed e ec s. Signi icance le els a e indica ed by ∗< .1,
** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in pa en heses.
We also in es iga e whe he in oducing he new inspec ion mechanisms led o changes
in ines un ela ed o he QL (non-QL ines) o o ou comes po en ially ela ed o i ms’
p o i abili y. Table A3 and Figu e A3 in he Appendix show he esul o ha analysis.
We es o he e ec s o he new inspec ion mechanisms on non-QL ines as a placebo
exe cise since he new adminis a i e ac should no in e e e wi h he en o cemen o o he
labo egula ions. Indeed, we do no obse e any impac on he likelihood o ecei ing ines
no ela ed o he Quo a Law a ound he h esholds, which suppo s ou hypo hesis ha
he obse ed impac o inspec ions and QL ines is due o he in oduc ion o he 2012
Adminis a i e Ac and no o an o e all inc ease in en o cemen o labo egula ions a e
2012. Fu he mo e, we analyze he impac o highe en o cemen a ound he h eshold on
i m closu e (we p oxied o closu e by looking a whe he a i m p esen in he da a a ime
is no ound a + 1), i m o al wage bill, and u no e a e.17 We ind no impac on
hese ou comes sugges ing ha he en o cemen did no lead o majo p o i abili y issues.
These esul s a e in line wi h e idence on he en o cemen o labo egula ion in he Uni ed
S a es (Le ine e al.,2012) and in he B azilian con ex (Sze man,2022), which also does
no ind ha hi ing people wi h disabili ies had nega i e impac s on i ms o wo ke s wi hou
disabili ies.18
17We use sepa a ion a e, de ined as he numbe o wo ke s ha le he i m a ime di ided by all spells
p esen in he i m a as ou measu e o u no e a e.(P ies and Roge son,2022)
18de Souza (2023), in con as , inds nega i e e ec s om he Quo a Law en o cemen o he employmen
and wages o wo ke s no ca ying a disabili y in B azilian i ms. A possible eason o such a di e ence
is ha we exploi an ins i u ional change, while de Souza (2023) exploi s he iming o i m inspec ions.
Recei ing an inspec ion, howe e , migh lead i ms o ush in o hi ing pe sons wi h disabili y, which migh
dec ease hei p oduc i i y, a leas momen a ily.
13
Figu e 3: The 2012 Adminis a i e Ac , Law En o cemen , and Wo ke s wi h Disabili ies
-.05
0
.05
.1
Di . in Likelihood: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(a) Inspec ion
-.01
0
.01
.02
.03
.04
Di . in Likelihood: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(b) QL Fines
-.1
0
.1
.2
.3
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(c) Wo ke s wi h Disabili y
No e: These g aphs p esen es ima ions om a model simila o Equa ion 1, whe e we subs i u e he pos -2012 dummy T
wi h yea dummies, lea ing 2011 as he benchma k. We conside a donu ing o wo in he i m-size a iable. All es ima ions
include ci y-by-yea ixed e ec s. 90% con idence in e al shown in he g aphs.
14
A inal ques ion we ask is whe he en o cing he Quo a Law led i ms o compe e o
wo ke s wi h disabili y. Hence, we in es iga e whe he he wages o pe sons wi h disabili ies
inc eased a e he 2012 Adminis a i e Ac . We p esen his esul in he las column o
Table A3 and he las sub- igu e o Figu e A3 in he Appendix. As we can see, he e is no
e idence o such a compe i ion o wo ke s wi h disabili y.
O e all, hese esul s show ha he only ema kable di e ence a e 2012 in he p e alence
o law en o cemen be ween i ms la ge and smalle han 100 was he inc ease in inspec ions
and p ima ily he inc ease in issuing o Quo a Law ines. The esul s also show ha i ms
eac ed o such an inc ease in QL en o cemen by hi ing mo e wo ke s wi h disabili ies, and
his does no seem o impac o he i ms’ economic ou comes o hei compe i ion o he
employmen o pe sons wi h disabili y.
4.1 Compliance in he Absence o Di ec En o cemen
En o cemen spillo e s a e a powe ul ool in achie ing gene al compliance due o he sp ead
o in o ma ion abou law en o cemen o agen s no di ec ly impac ed by i . We nex p esen
e idence ha i ms inc eased hei compliance wi h he QL a e he 2012 Adminis a i e
Ac , e en i loca ed in places wi h lowe en o cemen capaci y o in cases whe e hey ne e
ecei ed a QL ine hemsel es. In he nex sec ion, we di ec ly in es iga e he eme gence o
en o cemen spillo e s.
P e ious esea ch on labo egula ions highligh s he signi ican impac o en o cemen
capaci y on compliance (e.g., Almeida and Ca nei o,2012;Ponczek and Ulyssea,2022).
Howe e , ou analysis e eals ha while he issuance o QL ines dec eases wi h educed
en o cemen capaci y, his does no hold o he hi ing o wo ke s wi h disabili ies. In pa -
icula , we es ima e a model ha includes in e ac ions be ween he a iable Absence o LO
and all elemen s o equa ion 1. We de ine Absence o LO as a bina y a iable indica ing ha
he e is no labo o ice in he municipali y whe e i m iis loca ed (i.e., he dis ance be ween
municipali y mwhe e i m iis loca ed and he nea es labo o ice is g ea e han ze o). The
esul s a e p esen ed in Table 3. The impac o he 2012 Adminis a i e Ac on inspec ions
does no change depending on he i m’s dis ance o he nea es labo o ice (see columns (1)
and (2)). Howe e , he issuance o ines dec eases conside ably in places wi h no labo o ice
(see columns (3) and (4)).
E en wi h such a dec ease in he issuing o ines in he absence o labo o ices, he impac
o he 2012 Adminis a i e Ac on he hi ing o pe sons wi h a disabili y is cons an ac oss
locali ies, ega dless o whe he labo o ices a e p esen in hei municipali y o no (see
columns (5) and (6)).
15
Table 3: He e ogeneous Resul s by En o cemen Capaci y Le el
(1) (2) (3) (4) (5) (6)
Inspec ion QL ine Wo ke s w/ a disabili y
(hyp. sine ans.)
Yea >2012 X Dis . QL h eshold>0 0.014∗0.012 0.016∗∗∗ 0.015∗∗∗ 0.062∗∗∗ 0.063∗∗∗
(0.008) (0.009) (0.003) (0.003) (0.017) (0.020)
Yea >2012 X Dis . QL h eshold>0 X Absence o LO 0.014 0.017 -0.007∗∗ -0.006∗0.001 0.012
(0.013) (0.013) (0.004) (0.004) (0.029) (0.032)
N 432081 423407 320222 311529 214932 206282
Mean Dep. Va . 0.437 0.438 0.024 0.024 0.465 0.472
h (le ) 33.305 33.305 19.665 19.665 13.910 13.910
h ( igh ) 118.828 118.828 127.249 127.249 66.489 66.489
R2 0.175 0.176 0.107 0.108 0.209 0.211
Donu ing 1 2 1 2 1 2
No e: This able p esen s es ima ions om equa ion 1. All es ima ions include ci y-by-yea ixed e ec s. Signi icance le els
a e indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in pa en heses.
Figu e 4: The 2012 Adminis a i e Ac and Wo ke s wi h Disabili ies
Sub-sample: Fi ms ha Ne e Recei ed a QL Fine
-.1
0
.1
.2
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
No e: This igu e p esen s es ima ion om a model simila o equa ion 1, whe e we subs i u e he pos -2012 dummy Twi h
yea dummies, lea ing 2011 as he benchma k. We keep in ou es ima ion he sub-sample o i ms ha ne e ecei ed a QL
ine. The es ima ion includes ci y-by-yea ixed e ec s. 90% con idence in e al shown in he g aphs.
Mo eo e , e en in places whe e en o cemen capaci y is highe , he numbe o QL ines
issued is ela i ely small: a e 2012, only 8.4% o i ms la ge han 100 wo ke s loca ed
in municipali ies wi h he p esence o labo egula o y o ices (i.e., close o he egula ion
en o ce s) we e ined due o non-compliance wi h he QL, e en hough compliance was only
nea 20% o hese i ms du ing ha pe iod. Howe e , e en i ms ha ne e ecei ed a
QL ine eac ed o he 2012 Adminis a i e Ac by inc easing hei hi ing o people wi h
disabili ies. We show his in Figu e 4, whe e we ep oduce he es ima ions shown in Figu e
3 es ic ing ou sample o i ms ha ne e ecei ed a QL ine. As we can see, he impac o
he 2012 Adminis a i e Ac in his sub-sample is ema kably simila o he one conside ing
he whole sample o i ms.
16
5 En o cemen Spillo e s
Ha ing es ablished ha in oducing new en o cemen mechanisms e ec i ely changed i ms’
beha io ega ding hi ing pe sons wi h a disabili y, we now ocus on he spillo e e ec s
s emming om such an inc ease in he s ingency o law en o cemen . I i ms lea n om
hei ne wo ks abou he inc ease in he likelihood o being punished due o non-compliance
wi h he Quo a Law, hey migh change hei beha io , e en i no di ec ly exposed o law
en o cemen .
We explo e he iming o Quo a Law ines and in es iga e how a i m i eac s when
ano he i m in i s ne wo k ecei es such a ine. We look a h ee di e en ne wo ks om
which i m icould lea n abou he Quo a Law en o cemen : i) neighbo ne wo k, de ined as
he i ms loca ed in he same zip code as i m i;ii) owne ne wo k , de ined as he i ms ha
belong o he same owne o i m io i ms ha belong o a business associa e o such an
owne ;19 and iii) HR wo ke s ne wo k, de ined as he i ms whe e human esou ces wo ke s
wo king o i m ia ime we e wo king up o h ee yea s be o e .
We implemen an e en -s udy me hodology whe e we analyze ends in he p esence o
wo ke s wi h a disabili y be o e and a e he occu ence o a QL ine in a i m’s ne wo k.
We use as con ol g oup i ms ha belong o ne wo ks ha will ecei e a QL ine in he
u u e (Deshpande and Li,2019;Fadlon and Nielsen,2020). The main iden i ica ion as-
sump ion behind he choice o such a con ol g oup is ha , while ecei ing a QL ine migh
be endogenous o a ne wo k, he iming o such a ine can be conside ed exogenous. Table
A4 in he Appendix shows ha he occu ence o a ine in i ms’ ne wo ks al eady seems
qui e exogenous, e en i we conside ne wo ks ha ha e ne e ecei ed a ine. O e all, i ms’
p e ious cha ac e is ics, such as hei size o hei numbe o wo ke s wi h a disabili y, a e
no able o p edic he occu ence o a QL ine in hei ne wo k. Howe e , he exogenei y is
e en mo e e iden when we es ic he compa ison g oup o i ms whose ne wo ks ecei ed
a QL ine in he u u e: he li le p edic i e powe ha we obse e in he e en columns o
Table A4 (i.e., he es ima ion ha included ne e - ea ed ne wo ks) usually anishes in he
subsequen es ima ions (odd-columns) when we exclude om he es ima ion i ms belonging
o ne wo ks ha ha e ne e ecei ed a QL ine.
We cons uc ou es ima ion sample o i ms in ou s eps. Fi s , we ake he ne wo ks
whe e he QL ine happened a any ime a e 2012.20 Fo ins ance, o look a he impac
o a QL ine in he neighbo ne wo k, his means es ic ing he sample o zip codes whe e
any QL ine happened be ween 2012 and 2018. The same idea applies o he owne ne wo k
19We de ine business associa es as indi iduals who sha e he owne ship o a i m.
20We ocus on he pe iod a e implemen ing he new inspec ion p ocedu es since he numbe o ines
inc eased conside ably a e i .
17
and he HR wo ke s ne wo k. O e his pe iod, some ne wo ks a e ea ed ea lie han
o he s. Second, a e e y yea , we label ime 0 he i s ime a ne wo k ecei es he QL ine.
A ha poin , he ne wo k is conside ed ea ed. Thi d, a e e y yea , any ne wo k ha
ecei es a QL ine o he i s ime a leas wo yea s in he u u e is conside ed a con ol
ne wo k. Fou h, we s ack o e e y yea a se o ea ed and con ol ne wo ks epea ing his
p ocedu e. E e y ea ed and con ol ne wo k has h ee yea s o da a be o e and wo a e
he e en . To ensu e we in es iga e spillo e s om law en o cemen and no i s di ec e ec ,
we d op om ou sample he es ablishmen s ha ecei ed a QL ine a ime 0.
We es ima e he ollowing model using o dina y leas squa es eg essions:
Yimc =δ0T ea edimq +
τ=2
X
τ6=−1
τ=−3
Dτ
+
τ=2
X
τ6=−1
τ=−3
δτ(T ea edimq ×Dτ
)
+θm+θ +θc+θmc +imq
(2)
whe e Yimc is he ou come o in e es ( o ins ance, numbe o wo ke s wi h disabili ies) o
i m i, in municipali y m, in he yea o ea men (o coho ) c, a e en - ime , he Dτ
a e
indica o s equal o one o each e en - ime window (i.e., τ= -3, ..., 0, ...., 2), and imc is a
andom speci ica ion e o . As in equa ion 1, we include con ols o ime and municipali y
ixed e ec s and he in e ac ion be ween hese wo e ms (θ and θm, and θm espec i ely)
o con ol o ixed cha ac e is ics o he local labo ma ke which could be co ela ed wi h
he likelihood o obse ing a ine, and he p esence o labo egula o y o ices. Besides, we
also include coho ixed e ec s θc o con ol o o he shocks happening simul aneously o
he QL ine.
The coe icien s o in e es a e he es ima es o δτ. A e e y e en - ime window, hey
ep esen he causal impac on he employmen o wo ke s wi h disabili ies o i ms belonging
o a ne wo k whe e ano he i m ecei ed a QL ine. We also p esen a summa y es ima e
wi h a pos -dummy ins ead o each e en - ime dummy.
We es ima e equa ion 2 i s o he sample o all i ms la ge han 100 wo ke s and second
sepa a ely o he sub-sample o i ms ha we e a leas pa ially complying wi h he QL
be o e he QL ine hi hei ne wo k and he sub-sample o i ms ha we e no complying
wi h he QL a ha ime.21 We hypo hesize ha i ms al eady complying wi h he law
should no be a ec ed by he in o ma ion abou en o cing such a law in hei ne wo ks,
while non-complian i ms should be he mos impac ed by he new in o ma ion.
21Since ully complying wi h he QL is a a e e en , we conside ha a i m pa ially complies wi h he
law i i has a leas 50% o he numbe o wo ke s wi h a disabili y ha i should ha e acco ding o he
quo a.
18
Fi s , i ms eac o a QL ine in hei ne wo k by hi ing mo e wo ke s wi h disabili ies,
ega dless o he ne wo k whe e such a ine happens. We see such a pa e n in Figu e 5 ha
p esen s he esul s o es ima ions conside ing all i ms la ge han 100 wo ke s o he e en
o a QL ine in he i m’s neighbo ne wo k (Figu e 5a), owne ne wo k (Figu e 5b), and HR
wo ke s ne wo k (Figu e 5c). In all cases, we obse e an inc ease in he numbe o wo ke s
wi h a disabili y in he i m a e he e en o a QL ine in hei ne wo k. Tha inc ease is
pe sis en , and i g ows wi h ime in he neighbo and he owne ne wo ks while i ades ou
in he HR wo ke s’ ne wo k.
Second, only i ms ha isk ecei ing a QL ine – ha is, hose no complying wi h he
law – eac o he occu ence o a ine in hei ne wo ks. Figu e 6p esen s such a esul .
O e all, we see ha he posi i e e ec o a QL ine in he i m’s ne wo k on i s numbe
o wo ke s wi h a disabili y is concen a ed in i ms ha we e no complying wi h he QL
be o e ha e en ( igu es 6a,6c, and 6e). The esul is imp ecisely es ima ed o he HR
wo ke s’ ne wo k (Figu e 6e), such ha hey a e no signi ican ly di e en om ze o, e en
hough we obse e an inc ease in he coe icien s’ size. In u n, i ms ha we e al eady a
leas pa ially complying wi h he QL do no change hei beha io a all when exposed o
he e en o a QL ine (Figu es 6b,6d, and 6 ).22
Thi d, one could hypo hesize ha , a e ining a i m in a pa icula ne wo k, labo
inspec o s become mo e likely o inspec o he i ms in ha same ne wo k, and his is wha
d i es he inc ease in he numbe o wo ke s wi h disabili ies in i ms ins ead o hei di ec
communica ion wi h o he i ms in hei ne wo k. Howe e , his does no seem o be he
case: we can see in Figu e 7, which p esen s esul s o es ima ions whe e we in es iga e
whe he he occu ence o a Quo a Law ine impac s he likelihood o i ms being inspec ed.
We es ima e models simila o he one in equa ion 2, whe e now ou dependen a iable is a
bina y a iable indica ing whe he he i m was inspec ed. As shown in he igu e, ha ing a
QL ine in hei ne wo k does no inc ease he likelihood ha i ms will ecei e an inspec ion
a e such an e en .
We summa ize all hese esul s in Table 4, whe e we subs i u e he e en - ime dummies
wi h a bina y a iable indica ing he pos -e en pe iods. O e all, hese esul s show ha
22We employ he me hodology p oposed by Rambachan and Ro h (2023) o assess he sensi i i y o pa allel
ends iola ions in Figu e A4 (Appendix A). Speci ically, we allow o di e ences in linea ends be ween
ea ed and no -ye - ea ed and quan i y how la ge any depa u es om such linea i y should be so ha we
would ha e null esul s. We conside a ange o alues M, whe e M= 0 means no di e ence in linea ends
and M > 0allows o de ia ions in linea i y. I we conside all i ms, we do no nulli y he esul s o he
neighbo ne wo k e en o alues o Mas la ge as M= 1. The b eakdown alue o Min he owne ne wo k
is 0.3, and esul s a e al eady imp ecise in he wo ke ne wo k, e en o M= 0. I we conside non-complie
i ms, we do no nulli y he esul s o he neighbo and owne ne wo k e en o alues o Mand la ge as
M= 1, and he b eakdown alue o Mis 0.8 in he wo ke ne wo k.
19
Figu e 5: Law En o cemen a Fi m Ne wo ks and he Numbe o Wo ke s wi h a Disabili y
-.1
-.05
0
.05
.1
.15
Coe icien
-3
-2
-1
0
1
2
E en Time
(a) Neighbo Ne wo k
-.05
0
.05
.1
.15
.2
Coe icien
-3
-2
-1
0
1
2
E en Time
(b) Owne Ne wo k
-.2
-.1
0
.1
.2
Coe icien
-3
-2
-1
0
1
2
E en Time
(c) HR Wo ke s Ne wo k
No e: These g aphs p esen es ima ions om equa ion 2. The dependen a iable is he hype bolic sine ans o ma ion o he
numbe o wo ke s wi h a disabili y in he i m. The sample comp ises i ms la ge han 100 wo ke s ha did no ecei e a QL
ine in = 0. "E en Time" is he ime a e he occu ence o he QL ine in he i m’s ne wo k. All es ima ions include coho
and ci y-by-yea ixed e ec s. 90% con idence in e al shown in he g aphs.
20
Re e ences
Acemoglu, D. and J. D. Ang is (2001). Consequences o employmen p o ec ion? he case
o he ame icans wi h disabili ies ac . Jou nal o Poli ical Economy 109(5), 915–957.
Acemoglu, D., C. Ga cía-Jimeno, and J. A. Robinson (2015). S a e capaci y and economic
de elopmen : A ne wo k app oach. Ame ican economic e iew 105(8), 2364–2409.
Almeida, R. and P. Ca nei o (2012). En o cemen o labo egula ion and in o mali y. Ame -
ican Economic Jou nal: Applied Economics 4(3), 64–89.
Almeida, R. and L. Ronconi (2016). Labo inspec ions in he de eloping wo ld: S ylized ac s
om he en e p ise su ey. Indus ial Rela ions: A Jou nal o Economy and Socie y 55(3),
468–489.
Als adsæ e , A., W. Kopczuk, and K. Telle (2019). Social ne wo ks and ax a oidance: E i-
dence om a well-de ined no wegian ax shel e . In e na ional Tax and Public Finance 26,
1291–1328.
Ba eca, A. I., M. Guldi, J. M. Lindo, and G. R. Waddell (2011). Sa ing babies? e isi ing he
e ec o e y low bi h weigh classi ica ion. The Qua e ly Jou nal o Economics 126(4),
2117–2123.
Bellema e, M. F. and C. J. Wichman (2020). Elas ici ies and he in e se hype bolic sine
ans o ma ion. Ox o d Bulle in o Economics and S a is ics 82(1), 50–61.
Be linski, S., S. Du yea, and S. M. Pe ez-Vincen (2021). P e alence and co ela es o
disabili y in la in ame ica and he ca ibbean: E idence om 8 na ional censuses. PloS
one 16(10), e0258825.
Callaway, B. and P. H. San ’Anna (2021). Di e ence-in-di e ences wi h mul iple ime pe i-
ods. Jou nal o econome ics 225(2), 200–230.
Calonico, S., M. D. Ca aneo, and R. Ti iunik (2014a). Robus da a-d i en in e ence in he
eg ession-discon inui y design. The S a a Jou nal 14(4), 909–946.
Calonico, S., M. D. Ca aneo, and R. Ti iunik (2014b). Robus nonpa ame ic con idence
in e als o eg ession-discon inui y designs. Econome ica 82(6), 2295–2326.
Ca aneo, M. D., N. Id obo, and R. Ti iunik (2019). A p ac ical in oduc ion o eg ession
discon inui y designs: Founda ions. Camb idge Uni e si y P ess.
Ca aneo, M. D., M. Jansson, and X. Ma (2018). Manipula ion es ing based on densi y
discon inui y. The S a a Jou nal 18(1), 234–261.
Chen, J. and J. Ro h (2024). Logs wi h ze os? some p oblems and solu ions. The Qua e ly
Jou nal o Economics 139(2), 891–936.
de Souza, G. (2023). Employmen and wel a e e ec s o he quo a o disabled wo ke s in
b azil.
27
DeLei e, T. (2000a). The unin ended consequences o he ame icans wi h disabili ies ac .
Regula ion 23, 21.
DeLei e, T. (2000b). The wage and employmen e ec s o he ame icans wi h disabili ies
ac . Jou nal o human esou ces, 693–715.
Deshpande, M. and Y. Li (2019). Who is sc eened ou ? applica ion cos s and he a ge ing
o disabili y p og ams. Ame ican Economic Jou nal: Economic Policy 11(4), 213–48.
D ago, F., F. Mengel, and C. T axle (2020). Compliance beha io in ne wo ks: E idence
om a ield expe imen . Ame ican Economic Jou nal: Applied Economics 12(2), 96–133.
E ans, M. F., S. M. Gilpa ic, and J. P. Shimshack (2018). En o cemen spillo e s: Lessons
om s a egic in e ac ions in egula ion and p oduc ma ke s. The Jou nal o Law and
Economics 61(4), 739–769.
Fadlon, I. and T. H. Nielsen (2020). Family labo supply esponses o se e e heal h shocks:
E idence om danish adminis a i e eco ds. Ame ican Economic Jou nal: Applied Eco-
nomics.
Fö s e , M. (2007). F om disabili y o abili y. policy challenges and ends in oecd coun ies.
G ay, W. B. and J. P. Shimshack (2011). The e ec i eness o en i onmen al moni o ing and
en o cemen : A e iew o he empi ical e idence. Re iew o En i onmen al Economics
and Policy 5(1), 3–24.
G embi, V., T. Nannicini, and U. T oiano (2016). Do iscal ules ma e ? Ame ican Economic
Jou nal: Applied Economics, 1–30.
Ho chkiss, J. L. (2004). A close look a he employmen impac o he ame icans wi h
disabili ies ac . Jou nal o Human Resou ces 39(4), 887–911.
Johnson, M. S. (2020). Regula ion by shaming: De e ence e ec s o publicizing iola ions
o wo kplace sa e y and heal h laws. Ame ican economic e iew 110(6), 1866–1904.
Jolls, C. and J. J. P esco (2004). Disagg ega ing employmen p o ec ion: The case o
disabili y disc imina ion. Technical epo , Na ional Bu eau o Economic Resea ch.
K ekó, J. and Á. Telegdy (2022). The e ec s o a disabili y employmen quo a when com-
pliance is cheape han de iance.
Lali e, R., J.-P. Wuell ich, and J. Zweimülle (2013). Do inancial incen i es a ec i ms’
demand o disabled wo ke s? Jou nal o he Eu opean Economic Associa ion 11(1),
25–58.
Le ine, D. I., M. W. To el, and M. S. Johnson (2012). Randomized go e nmen sa e y
inspec ions educe wo ke inju ies wi h no de ec able job loss. Science 336(6083), 907–
911.
28
Li e mo e, G. A. and N. Goodman (2009). A e iew o ecen e alua ion e o s associa ed
wi h p og ams and policies designed o p omo e he employmen o adul s wi h disabili ies.
epo . Rehabili a ion Resea ch and T aining Cen e on Employmen Policy o Pe sons
wi h Disabili ies.
Lu, F., J. Zhang, and J. M. Pe lo (2016). Gene al and speci ic in o ma ion in de e ing
a ic iola ions: e idence om a andomized expe imen . Jou nal o Economic Beha io
& O ganiza ion 123, 97–107.
Mon , D. e al. (2004). Disabili y employmen policy. Social P o ec ion, Labo Ma ke s,
Pensions, Social Assis ance, Wo ld Bank.
Mo i, Y. and N. Sakamo o (2018). Economic consequences o employmen quo a sys em o
disabled people: E idence om a eg ession discon inui y design in japan. Jou nal o he
Japanese and In e na ional Economies 48, 1–14.
OECD (2003). T ans o ming disabili y in o abili y: policies o p omo e wo k and income
secu i y o disabled people. OECD Pa is.
Pae zold, J. and H. Winne (2016). Taking he high oad? compliance wi h commu e ax
allowances and he ole o e asion spillo e s. Jou nal o Public Economics 143, 1–14.
Ponczek, V. and G. Ulyssea (2022). En o cemen o labou egula ion and he labou ma ke
e ec s o ade: E idence om b azil. The Economic Jou nal 132(641), 361–390.
P ies, M. J. and R. Roge son (2022). Declining wo ke u no e : The ole o sho -du a ion
employmen spells. Ame ican Economic Jou nal: Mac oeconomics 14(1), 260–300.
Rambachan, A. and J. Ro h (2023). A mo e c edible app oach o pa allel ends. Re iew o
Economic S udies 90(5), 2555–2591.
Rincke, J. and C. T axle (2011). En o cemen spillo e s. Re iew o Economics and S a is-
ics 93(4), 1224–1234.
Shimshack, J. P. (2014). The economics o en i onmen al moni o ing and en o cemen .
Annu. Re . Resou . Econ. 6(1), 339–360.
Shimshack, J. P. and M. B. Wa d (2005). Regula o epu a ion, en o cemen , and en i-
onmen al compliance. Jou nal o En i onmen al Economics and Managemen 50(3),
519–540.
Slem od, J. (2019). Tax compliance and en o cemen . Jou nal o Economic Li e a u e 57(4),
904–954.
Sze man, C. (2022). The labo ma ke e ec s o disabili y hi ing quo as. A ailable a SSRN
4267622.
29
A Appendix
Table A1: RDD Quo a Law Th eshold: Di e en Bandwid h Selec ion P ocedu es
Dep. a .: wo ke s w/ a disabili y (hyp. sine ans.)
(1) (2) (3) (4) (5)
Panel A: Be o e 2012
RD Es ima e 0.016 0.041*** 0.016 0.016 0.016
(0.014) (0.010) (0.012) (0.014) (0.013)
N 991510 991510 991510 991510 991510
Mean dep. a . wi hin bandwid h 0.198 0.314 0.195 0.198 0.206
h (le ) 16.830 17.820 20.725 16.830 17.820
h ( igh ) 16.830 69.080 20.725 16.830 20.725
Bandwid h selec ion p ocedu e mse d mse wo msesum msecomb1 msecomb2
Panel B: A e 2012
RD Es ima e 0.057** 0.080*** 0.059*** 0.057** 0.055***
(0.025) (0.017) (0.018) (0.025) (0.020)
N 1593962 1593962 1593962 1593962 1593962
Mean dep. a . wi hin bandwid h 0.343 0.548 0.333 0.343 0.371
h (le ) 9.844 10.476 14.293 9.844 10.476
h ( igh ) 9.844 52.741 14.293 9.844 14.293
Bandwid h selec ion p ocedu e mse d mse wo msesum msecomb1 msecomb2
No e: This able shows esul s om local polynomial eg essions whe e we es ima e i ms’ hi ing beha io ega ding
wo ke s wi h disabili ies once hey pass he 100 wo ke s h eshold es ablished by he Quo a Law (see Ca aneo e al.,
2019;Calonico e al.,2014a,b, o de ails on ou RDD es ima ion). The dependen a iable is he hype bolic sine
ans o ma ion o he numbe o wo ke s ca ying a disabili y. Panel A shows es ima ions o he yea s be o e 2012,
i.e., be o e he in oduc ion o he new inspec ion p ocedu es. Panel B shows es ima ions o he yea s a e 2012, i.e.,
a e he in oduc ion o he new inspec ion p ocedu es. Each column o he able shows he esul s o es ima ions
using di e en bandwid hs op imally compu ed ollowing he algo i hm de eloped by Calonico e al. (2014a,b). Due o
measu emen e o s in he es ima ion o he i m’s size, we exclude i ms wi hin a donu ing o size wo om he 100
h eshold. Signi icance le els a e indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el
shown in pa en heses.
Figu e A1: Manipula ion es
0
.01
.02
.03
-100 -50 0 50 100
P_100
Manipula ion Tes ing Plo
(a) Be o e 2012: T s a = -0.4719 ; P>|T|=0.6370
0
.01
.02
.03
-100 -50 0 50 100
P_100
Manipula ion Tes ing Plo
(b) A e 2012: T s a =-2.7434 ; P>|T|=0.0061
See Ca aneo e al. (2018) o de ails abou he implemen a ion o manipula ion es s.
30
Table A2: The 2012 Adminis a i e Ac and P esence o Wo ke s wi h Disabili y: Robus ness
check
One o mo e wo ke w/ disabili y Numbe o wo ke s w/ disabili y
(1) (2) (3) (4)
Yea >2012X Dis ance o QL h eshold>0 0.049∗∗∗ 0.062∗∗∗ 0.052 0.042
(0.009) (0.011) (0.070) (0.097)
N 282594 265016 282594 265016
Mean Dep. Va . 0.328 0.944 0.919 0.944
h (le ) 16.175 16.175 16.175 16.175
h ( igh ) 90.242 90.242 90.242 90.242
R2 0.195 0.198 0.108 0.108
Donu ing 1 2 1 2
No e: This able p esen s es ima ions om equa ion 1. All es ima ions include ci y-by-yea ixed e ec s. Signi icance le els a e
indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in pa en heses.
Figu e A2: The 2012 Adminis a i e Ac and P esence o Wo ke s wi h Disabili y:
Robus ness Checks
-.05
0
.05
.1
.15
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(a) One o mo e wo ke w/ disabili y
-.2
0
.2
.4
.6
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(b) Numbe o wo ke s w/ disabili y
No e:These g aphs p esen es ima ions om a model simila o equa ion 1, whe e we subs i u e he pos -2012 dummy Twi h
yea dummies, lea ing 2011 as he benchma k. We conside a donu ing o wo in he i m-size a iable. All es ima ions include
ci y-by-yea ixed e ec s. 90% con idence in e al shown in he g aphs.
Table A3: The 2012 Adminis a i e Ac and O he Fi m Ou comes
Non-QL To al wage Tu no e Fi m Wage wo ke s w/
ines bill a e closu e disabili y (ln)
(1) (2) (3) (4) (5)
Yea >2012 X Dis ance o QL h eshold>0 -0.007∗-0.015 -0.002 -0.003 -0.056
(0.003) (0.013) (0.003) (0.005) (0.038)
N 641525 298207 426112 357065 113520
Mean Dep. Va . 0.125 12.051 0.314 0.035 7.952
h (le ) 45.211 22.223 24.076 13.552 30.213
h ( igh ) 185.839 82.449 246.241 277.317 126.129
R2 0.090 0.440 0.102 0.061 0.267
Donu ing 2 2 2 2 2
No e: This able p esen s es ima ions om equa ion 1. All es ima ions include ci y-by-yea ixed e ec s. Signi icance le els a e
indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in pa en heses.
31
Figu e A3: The 2012 Adminis a i e Ac and o he Fi m Ou comes
-.04
-.02
0
.02
.04
Di . in Likelihood: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(a) Non-QL Fines
-.1
-.05
0
.05
.1
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(b) To al Wage Bill
-.02
-.01
0
.01
.02
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(c) Tu no e Ra e
-.04
-.02
0
.02
.04
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(d) Fi m Closu e
-.2
-.1
0
.1
.2
.3
% G ow h: La ge s. Small Fi ms
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
(e) Wage Wo ke s wi h disabili y (ln)
No e: These g aphs p esen es ima ions om a model simila o equa ion 1, whe e we subs i u e he pos -2012 dummy Twi h
yea dummies, lea ing 2011 as he benchma k. The dependen a iables a e he hype bolic sine ans o ma ion o he a iables
indica ed in each sub- igu e. All es ima ions include ci y-by-yea ixed e ec s. 90% con idence in e al shown in he g aphs.
32
Table A4: Fi ms’ Cha ac e is ics in −1and QL Fine in Ne wo k in
2012 2013 2014 2015 2016 2017
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)
Panel A: Neighbo ne wo k
L.hyp_disabled -0.000 0.000 0.001 0.001 -0.002∗-0.004∗0.001 0.000 -0.001 -0.003 0.000 -0.001
(0.001) (0.003) (0.002) (0.004) (0.001) (0.003) (0.001) (0.002) (0.001) (0.002) (0.001) (0.002)
L.hyp_ i m_size_QL 0.001 -0.002 -0.000 -0.002 0.002 0.003 -0.000 -0.003 0.001 0.002 -0.001 -0.002
(0.001) (0.003) (0.002) (0.004) (0.002) (0.004) (0.002) (0.004) (0.001) (0.003) (0.001) (0.003)
Ne e ea ed included Yes No Yes No Yes No Yes No Yes No Yes No
N 22744 9926 23222 10033 22962 9927 21249 9286 19892 8717 19822 8145
R2 0.429 0.484 0.357 0.417 0.329 0.403 0.367 0.439 0.407 0.473 0.213 0.329
Panel B: Owne ne wo k
L.hyp_disabled 0.002∗∗ 0.003 0.000 -0.002 0.002 -0.001 0.001∗-0.003 -0.001 -0.008∗∗ 0.001 -0.004
(0.001) (0.004) (0.001) (0.004) (0.001) (0.005) (0.001) (0.004) (0.001) (0.004) (0.001) (0.003)
L.hyp_ i m_size_QL 0.001 -0.012∗∗ 0.004∗∗∗ 0.000 0.001 -0.009 0.001 -0.004 0.002 0.002 0.001 -0.007
(0.001) (0.005) (0.001) (0.006) (0.002) (0.008) (0.002) (0.009) (0.001) (0.006) (0.001) (0.007)
Ne e ea ed included Yes No Yes No Yes No Yes No Yes No Yes No
N 22744 3719 23222 3654 22962 3564 21249 3361 19892 3166 19822 3089
R2 0.055 0.124 0.064 0.125 0.061 0.137 0.071 0.149 0.067 0.138 0.061 0.138
Panel C: HR wo ke s ne wo k
L.hyp_disabled 0.001 -0.002 0.001 -0.002 -0.000 -0.003 0.001 0.001 0.000 -0.002 0.000 -0.002
(0.001) (0.004) (0.001) (0.003) (0.001) (0.003) (0.001) (0.003) (0.001) (0.002) (0.001) (0.002)
L.hyp_ i m_size_QL 0.006∗∗∗ -0.004 0.005∗∗∗ -0.004 0.002 -0.017∗∗∗ 0.002 -0.014∗∗∗ 0.000 -0.006∗0.002∗-0.002
(0.001) (0.005) (0.001) (0.004) (0.001) (0.004) (0.002) (0.005) (0.001) (0.003) (0.001) (0.004)
Ne e ea ed included Yes No Yes No Yes No Yes No Yes No Yes No
N 22744 4332 23222 4559 22962 4652 21249 4426 19892 4277 19822 4194
R2 0.051 0.118 0.054 0.134 0.041 0.117 0.046 0.121 0.062 0.121 0.045 0.110
No e: This able shows es ima ions o he likelihood ha a ne wo k ecei es a Quo a Law ine each yea , depending on he cha ac e is ics o i ms belonging
o such a ne wo k in he p e ious yea . Odd-numbe ed columns include i ms om ne e - ea ed ne wo ks, ha is, ha ne e ecei ed a Quo a Law ine, while
e en-numbe ed columns exclude such i ms, es ic ing he sample o ne wo ks used in ou analysis. All es ima ions include ci y-by-yea ixed e ec s. Signi icance
le els a e indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in pa en heses.
33
Table A5: Law En o cemen a Fi m’s Ne wo ks: Robus ness Checks
One o mo e wo ke w/ disabili y Numbe o wo ke s w/ disabili y
(1) (2) (3) (4) (5) (6)
All i ms Non-complie s Pa ially complie s All i ms Non-complie s Pa ially complie s
Panel A: Neighbo ne wo k
Pos -e en X T ea ed 0.023*** 0.038*** -0.001 0.432 0.719*** 0.046
(0.007) (0.010) (0.010) (0.508) (0.275) (1.399)
N 118160 70940 42603 118160 70940 42603
N ( i ms) 12248.000 7861.000 5610.000 12248.000 7861.000 5610.000
A g. i m size 368.404 385.448 351.608 368.404 385.448 351.608
Mean Dep. Va . 5.579 2.607 10.739 5.579 2.607 10.739
R2 0.172 0.193 0.193 0.044 0.101 0.082
Panel B: Owne ne wo k
Pos -e en X T ea ed 0.014 0.032* -0.019 1.279** 1.063** 1.145
(0.013) (0.019) (0.011) (0.643) (0.434) (1.621)
N 37049 20846 13948 37049 20846 13948
N ( i ms) 3794.000 2322.000 1729.000 3794.000 2322.000 1729.000
A g. i m size 485.848 499.083 492.306 485.848 499.083 492.306
Mean Dep. Va . 8.649 3.983 16.025 8.649 3.983 16.025
R2 0.254 0.287 0.216 0.140 0.195 0.173
Panel C: HR wo ke s ne wo k
Pos -e en X T ea ed 0.003 0.011 -0.003 0.552 1.106** 0.078
(0.010) (0.015) (0.013) (0.442) (0.523) (0.760)
N 37237 22122 13385 37237 22122 13385
N ( i ms) 3671.000 2415.000 1615.000 3671.000 2415.000 1615.000
A g. i m size 513.923 575.726 419.657 513.923 575.726 419.657
Mean Dep. Va . 7.698 4.692 12.547 7.698 4.692 12.547
R2 0.218 0.252 0.203 0.218 0.301 0.273
No e: This able p esen s es ima ions om a model simila o equa ion 2, whe e we subs i u e he ime dummies wi h a pos -e en
dummy. The sample is composed o i ms la ge han 100 wo ke s, which did no ecei e an inspec ion o any ype o ine in
= 0. "T ea ed" is an indica o ha some ype o law en o cemen ook place in he i m owne ’s ne wo k a = 0. Elas ici ies o
wo ke s wi h disabili ies a e calcula ed based on Bellema e and Wichman (2020). All es ima ions include coho and ci y-by-yea
ixed e ec s. Signi icance le els a e indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he ci y le el shown in
pa en heses.
34
Table A6: QL Fine in Fi m’s Ne wo k: Robus ness Check Using Callaway and San ’Anna
(2021)
Neighbo ne wo k Onwe ne wo k HR wo ke s ne wo k
(1) (2) (3)
P e_a g -0.022 -0.038 0.011
(0.030) (0.030) (0.022)
Pos _a g 0.104∗∗∗ 0.154∗∗∗ 0.102∗∗∗
(0.027) (0.038) (0.031)
Tm3 -0.067 -0.090 0.011
(0.067) (0.081) (0.066)
Tm2 -0.017 -0.032 0.008
(0.032) (0.047) (0.039)
Tm1 0.019 0.008 0.013
(0.023) (0.032) (0.028)
Tp0 0.050∗∗∗ 0.073∗∗∗ 0.047∗∗
(0.019) (0.027) (0.023)
Tp1 0.125∗∗∗ 0.158∗∗∗ 0.108∗∗∗
(0.028) (0.044) (0.034)
Tp2 0.138∗∗∗ 0.233∗∗∗ 0.150∗∗∗
(0.046) (0.057) (0.050)
N 24477 9887 10439
No e: This able p esen s es ima ions om equa ion 2, bu using he me hod p oposed
by Callaway and San ’Anna (2021) ins ead o he s acked di e ences-in-di e ences used
in ou main es ima ions. All es ima ions include ci y-by-yea ixed e ec s. Signi icance
le els a e indica ed by ∗< .1, ** < .05, *** < .01. S anda d e o s clus e ed a he
ci y le el shown in pa en heses.
Table A7: Law En o cemen a Fi m’s Ne wo ks: Placebo wi h Fi ms Smalle han 100
Neighbo ne wo k Owne ne wo k HR wo ke s ne wo k
(1) (2) (3)
Pos -e en X T ea ed 0.000 -0.006 -0.014
(0.002) (0.010) (0.012)
N 599867 69379 29689
N ( i ms) 70133 7952 3614
A g. i m size 41.717 49.181 54.135
Mean Dep. Va . 0.097 0.167 0.227
Elas ici y 0.000 -0.006 -0.013
R2 0.045 0.164 0.194
No e: This able p esen s es ima ions om equa ion 1 ocusing on a sample o i ms smalle
han 100 wo ke s, ins ead o i ms la ge han 100, as in ou main es ima ions. All es ima ions
include ci y-by-yea ixed e ec s. Signi icance le els a e indica ed by ∗< .1, ** < .05, *** < .01.
S anda d e o s clus e ed a he ci y le el shown in pa en heses.
35
Figu e A4: P e- end Robus ness
.02
.04
.06
.08
.1
.12
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
(a) Neighbo NW: All Fi ms
.05
.1
.15
.2
.25
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
(b) Neighbo NW: Non-complie s
-.05
0
.05
.1
.15
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
(c) Owne NW: All Fi ms
0
.05
.1
.15
.2
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
(d) Owne NW: Non-complie s
-.05
0
.05
.1
.15
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
(e) HR wo ke s NW: All Fi ms
0
.1
.2
.3
90% Robus CI
O iginal .1 .2 .3 .4 .5 .6 .7 .8 .9 1
Mba
( ) HR wo ke s NW: Non-complie s
No e: These igu es epo 90% con idence in e als o di e en de ia ions om linea ends be ween ea ed and no -ye -
ea ed i ms, employing he me hodology p oposed by Rambachan and Ro h (2023) o assess he sensi i i y o pa allel ends
iola ions. Mba = 0 means no di e ence in linea ends and Mba > 0allows o de ia ions in linea i y.
36
