Baumol’s cost disease in acute versus long-term care: Do the differences loom large?

Celebi, Kaan; Ha wig, Jochen; Sandq is , Anna Pauliina
A icle — Published Ve sion
Baumol’s cos disease in acu e e sus long- e m ca e: Do
he di e ences loom la ge?
In e na ional Jou nal o Heal h Economics and Managemen
P o ided in Coope a ion wi h:
Sp inge Na u e
Sugges ed Ci a ion: Celebi, Kaan; Ha wig, Jochen; Sandq is , Anna Pauliina (2025) : Baumol’s cos
disease in acu e e sus long- e m ca e: Do he di e ences loom la ge?, In e na ional Jou nal o
Heal h Economics and Managemen , ISSN 2199-9031, Sp inge US, New Yo k, NY, Vol. 25, Iss. 2, pp.
159-191,
h ps://doi.o g/10.1007/s10754-025-09392-9
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/330429
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/4.0/
Vol.:(0123456789)
In e na ional Jou nal o Heal h Economics and Managemen (2025) 25:159–191
h ps://doi.o g/10.1007/s10754-025-09392-9
RESEARCH ARTICLE
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do
hedi e ences loom la ge?
KaanCelebi1· JochenHa wig1,2,3· AnnaPauliinaSandq is 4
Recei ed: 14 Ap il 2024 / Accep ed: 9 Janua y 2025 / Published online: 24 Feb ua y 2025
© The Au ho (s) 2025
Abs ac
Baumol’s (Am Econ Re 57: 415–426, 1967) model o ‘unbalanced g ow h’ yields a sup-
ply-side explana ion o he ‘cos explosion’ in heal h ca e. Applying a es ing s a egy sug-
ges ed by Ha wig (J Heal h Econ 27: 603–623, 2008), a sp awling li e a u e a i ms ha
he ‘Baumol e ec ’ has bo h a s a is ically and economically signi ican impac on heal h
ca e expendi u e g ow h. Skep ics main ain, howe e , ha he p oli e a ion o hi- ech medi-
cine in acu e ca e is clea ly a odds wi h he assump ion unde lying Baumol’s model ha
p oduc i i y-enhancing machine y and equipmen is only ins alled in he ‘p og essi e’ (i.e.
manu ac u ing) sec o o he economy. They a gue ha Baumol’s cos disease may a ec
long- e m ca e, bu no acu e ca e. Ou aim in his pape is o es whe he Baumol’s cos
disease a ec s long- e m ca e and acu e ca e di e en ly. Ou es ing s a egy consis s in
combining Ex eme Bounds Analysis (EBA) wi h an ou lie - obus MM es ima o . Using
panel da a o 23 OECD coun ies, ou esul s p o ide obus and s a is ically signi ican
e idence ha expendi u es on bo h acu e ca e and long- e m ca e a e d i en by Baumol’s
cos disease, e en hough he e ec on long- e m ca e expendi u es is mo e p onounced.
Keywo ds Heal h ca e expendi u e· Baumol’s cos disease· Ex eme Bounds Analysis·
MM es ima o · OECD panel
JEL Classi ica ion C12· C23· I10
* Jochen Ha wig
[email p o ec ed]
1 Facul y o Economics andBusiness Adminis a ion, Chemni z Uni e si y o Technology,
Chemni z, Ge many
2 KOF Swiss Economic Ins i u e, ETH Zu ich, Zu ich, Swi ze land
3 Fo um o Mac oeconomics andMac oeconomic Policies, Hans Böckle S i ung, Düsseldo ,
Ge many
4 Deloi e GmbH Wi scha sp ü ungsgesellscha , Munich, Ge many
160
K.Celebi e al.
In oduc ion
Baumol’s (1967) model o ‘unbalanced g ow h’ yields a supply-side explana ion o he
‘cos explosion’ in heal h ca e. Baumol di ides he economy in o wo pa s: a ‘p og es-
si e’ and a ‘non-p og essi e’ sec o . He assumes ha p oduc i i y g ow h is highe in he
p og essi e (seconda y) han in he non-p og essi e—o ‘s agnan ’—( e ia y) sec o o
he economy, bu wages g ow mo e o less he same in bo h sec o s. The e o e, uni cos s
and also p ices ise much as e in he e ia y sec o han in he seconda y. Demand o
ce ain se ices, like heal h ca e and educa ion o ins ance, is ha dly p ice-elas ic, hence
consume s a e willing o pay he highe p ices. The e o e, e en i he wo sec o s keep hei
p opo ion in e ms o eal p oduc ion, an e e -highe sha e o o al expendi u es will be
channeled in o he s agnan sec o . This phenomenon is known as ‘Baumol’s cos disease’.1
Ha wig (2008) has sugges ed a es o whe he Baumol’s cos disease d i es heal h ca e
expendi u e (HCE) in OECD coun ies ha does no equi e p ice o p oduc i i y da a o
he heal h sec o , which a e no o iously un eliable (see Be nd e al., 2000, p. 171). This
es consis s in eg essing HCE g ow h a es (log di e ences) on he di e ence be ween
nominal wage g ow h and labo p oduc i i y g ow h in he o e all economy (plus con ols).
Rossen and Fa oque (2016, p. 192) nea ly summa ize he in ui ion behind his app oach as
ollows: “His [Ha wig’s] key insigh is ha since wage g ow h in heal h ca e depends on
he highe p oduc i i y g ow h in he es o he economy, g ow h in he uni labo cos and
p ice o heal h ca e se ices, and he e o e g ow h in heal h ca e spending, mus bea a
p opo ional ela ionship o he excess wage g ow h o e labo p oduc i i y g ow h in he
o e all economy”.2
E iden ly, Ha wig’s ‘Baumol a iable’, i.e. he di e ence be ween nominal wage
g ow h and labo p oduc i i y g ow h, equals he g ow h a e o agg ega e (nominal) uni
labo cos (NULC). To check whe he he ‘Baumol a iable’ is no jus picking up pu ely
mone a y changes, Ha wig (2008) de la ed bo h pe -capi a HCE— he dependen a ia-
ble—and nominal wages pe employee on he igh -hand side o he eg ession equa ion by
he GDP de la o . Hence, his ul ima e es o whe he Baumol’s cos disease d i es heal h
ca e expendi u e consis s in eg essing eal HCE g ow h on he g ow h a e o agg ega e
eal uni labo cos (RULC) plus con ols.3 Ha wig (2008) and schola s ollowing his lead
o es ing Baumol’s cos disease in heal h ca e (see, e.g., Ba es & San e e, 2013, Medei os
& Schwie z, 2013, Ha wig & S u m, 2014, Rossen & Fa oque, 2016, Colombie , 2017,
Tian e al., 2018, Bellido e al., 2019, Lo enzoni e al., 2019, Jee oo, 2020, Wang & Chen,
2021) ha e ho oughly con i med ha he ‘Baumol e ec ’4 has bo h a s a is ically and eco-
nomically signi ican impac on heal h ca e expendi u e g ow h.5
1 The e m ‘Baumol’s cos disease’ was coined by Alice Vande meulen (1968), see Baumol (2012, p. xii).
2 See also Lo enzoni e al. (2019, p. 40) o a single-page de i a ion o Ha wig’s cos disease a iable.
3 The g ow h a e o agg ega e RULC equals he g ow h a e o he wage sha e in GDP.
4 Helland and Taba ok (2019) p e e he exp ession ‘Baumol e ec ’ since i a oids he nega i e conno a-
ions o he e m ‘Baumol’s (cos ) disease’.
5 Rossen and Fa oque (2016, p. 203), who only pe o m he nominal e sion o he es , conclude ha
“Baumol’s cos disease on heal h-ca e spending inc eases in Canada may no be economically e y impo -
an ”. Ba es and San e e (2013) and Wang and Chen (2021) also ind ela i ely mino e ec s. The main
eason o his inding, howe e , seems o be ha hese au ho s apply a ‘co ec ion’ o Ha wig’s ‘Baumol
a iable’ sugges ed by Colombie (2012). The main e ec o his ‘co ec ion’ is ha i scales he es ima ed
coe icien down by a ac o a ound 10 (see Table2 in Rossen and Fa oque 2016).
161
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
E en hough empi ical esea ch o e he pas decade and a hal has buil up a s ong
case in a o o Baumol’s cos disease being one o he main d i e s o HCE g ow h, he e
emains one piece o skep icism ha mo i a es ou p esen pape . This skep icism was i s
b ough o ou a en ion by he la e Gebha d Ki chgässne , hen p esiden o he Swiss ed-
e al Commission o Business Cycle A ai s (K K). The 2006 annual epo o ha commis-
sion i led ‘Re o ming he heal h sys em’ (Kommission ü Konjunk u agen, 2006, pp.
36–37) quo es he wo king pape e sion o Ha wig (2008) a he disapp o ingly. Techno-
logical p og ess, so he a gumen goes, is i e in acu e ca e, becoming mani es in hi- ech
medicine. The e o e, he assump ion unde lying Baumol’s model o ‘unbalanced g ow h’
ha p oduc i i y-enhancing machine y and equipmen is only ins alled in he seconda y
sec o o he economy is clea ly lawed. The epo does concede an impac o Baumol’s
cos disease on HCE g ow h, bu es ic s i o he long- e m ca e (LTC) sec o (see also
Ki chgässne , 2009).
Simila ly, de la Maisonneu e and Oli ei a Ma ins (2014), in hei heal h expendi u e
p ojec ions un il 2060 on behal o he OECD, model (public) HCE and LTC expendi u e
sepa a ely and allow Baumol’s cos disease ( o which hey use he le el o he g ow h
a e o labo p oduc i i y in he o al economy as a p oxy) only o a ec he la e . In hei
upda e o he OECD’s spending p ojec ions, howe e , Lo enzoni e al. (2019, p. 25), no e
ha he allowance o “ he impac o he Baumol e ec on heal h ca e as a whole (ins ead o
only o long- e m ca e)” was one o he main di e ences agains p e ious s udies.
Ou aim in his pape is o es whe he Baumol’s cos disease a ec s long- e m ca e
and acu e ca e di e en ly. Acu e ca e, acco ding o he OECD (2008, p. 17) “is one in
which he p incipal in en is one o mo e o he ollowing: (i) o manage labou (obs e ics),
(ii) o cu e illness o o p o ide de ini i e ea men o inju y, (iii) o pe o m su ge y, (i )
o elie e symp oms o illness o inju y (excluding pallia i e ca e), ( ) o educe se e i y
o an illness o inju y, ( i) o p o ec agains exace ba ion and/o complica ion o an ill-
ness and/o inju y which could h ea en li e o no mal unc ion, ( ii) o pe o m diagnos ic
o he apeu ic p ocedu es”. In he Wikipedia en y on ‘acu e ca e’ i eads: “In medical
e ms, ca e o acu e heal h condi ions is he opposi e om ch onic ca e, o longe - e m
ca e”.6 We, he e o e, de ine acu e ca e expendi u e (ACE) e y b oadly as o al cu en
heal h ca e expendi u e (HCE) minus long- e m ca e expendi u e (LTCE). Acu e ca e is
p o ided as inpa ien o ou pa ien ca e by hospi als and medical and den al p ac ices o
o he heal hca e p o ide s.
Long- e m ca e, on he o he hand, comp ises (i) medical o nu sing ca e, such as elie -
ing pain and o he symp oms, (ii) pe sonal ca e se ices, i.e. help wi h ac i i ies o daily
li ing, pe o med ei he by ela i es o nu sing s a and (iii) assis ance se ices, which
enable pe sons o li e independen ly a home, e.g. shopping o pe o ming housewo k
(OECD e al., 2017, p. 91). The modes o p o ision o long- e m ca e a e (i) inpa ien long-
e m ca e in hospi als o nu sing homes equi ing an o e nigh s ay wi h medical supe -
ision, (ii) day cases o long- e m ca e deli e ed by he same p o ide s, bu wi hou an
o e nigh s ay and (iii) ou pa ien o home-based long- e m ca e, which ypically in ol es
p o ide s o nu sing se ices egula ly isi ing elde ly people who a e becoming mo e
dependen (OECD e al., 2017, pp. 94–95).
In o de o es whe he Baumol’s cos disease a ec s long- e m ca e and acu e ca e
di e en ly we use he same s a egy as Ha wig and S u m (2014), i.e. Ex eme Bounds
Analysis (EBA) combined wi h an ou lie - obus MM es ima o . As EBA includes many (i
6 See h ps:// en. wikip edia. o g/ wiki/ Acu e_ ca e.
162
K.Celebi e al.
no all) o he HCE d i e s ha ha e been sugges ed in he li e a u e, and omi ed a iables
a e an impo an sou ce o endogenei y, conside ing many explana o y a iables mi iga es
he la e . Howe e , we do no claim o p ope ly iden i y causal e ec s. When we use he
e m ‘e ec ’ in ou empi ical analysis and o en when we e e o he li e a u e, i ela es o
condi ional co ela ions, no a causal ela ionship. In o he wo ds, we a e sugges ing a he
han es ing o causal ela ionships.
The emainde o his pape is s uc u ed as ollows. The nex sec ion discusses ou
da ase . Sec . "Me hodology" explains he me hodologies o Ex eme Bounds Analysis
and ou lie - obus MM-es ima ion. Sec s. "Resul s" and "Robus ness es s" p esen he
esul s—including hose o obus ness checks—and Sec ."Conclusion" concludes. Da a
issues a ound LTCE a e discussed in Appendix1.
Da a
The da a sou ce o mos o he a iables is he OECD Heal h Da abase, which also con-
ains economic, socio-demog aphic, and e en echnological da a (as long as hey a e
heal h- ela ed).7 Conside ing he dependen a iables, da a on o al heal h expendi u es a e
a ailable o qui e a long ime pe iod and many coun ies while da a on long- e m ca e
expendi u es (LTCE) end o be ela i ely sca ce. We exclude I eland, I aly, New Zealand,
and he Uni ed Kingdom (because o he small numbe o obse a ions) and coun ies wi h
a e y low sha e (smalle han 5 pe cen in 2017) o LTCE in HCE (Aus alia, G eece,
Hunga y, La ia, Po ugal, and Slo akia) as hei da a migh be o low quali y and/o noisy.
Ou da ase hus co e s he ollowing 23 OECD coun ies: Aus ia, Belgium, Canada, he
Czech Republic, Denma k, Es onia, Finland, F ance, Ge many, Iceland, Is ael, Japan,
Sou h Ko ea, Li huania, Luxembou g, he Ne he lands, No way, Poland, Slo enia, Spain,
Sweden, Swi ze land, and he Uni ed S a es.
Wi h espec o he explana o y a iables, nume ous possible de e minan s ha e been
in oduced. We d aw on Ha wig and S u m (2014), who ha e conduc ed an ex ensi e li e -
a u e e iew o unco e all mac oeconomic and ins i u ional de e minan s o HCE g ow h
ha ha e been sugges ed in he li e a u e, and in oduce hem in an EBA amewo k o
be explained in he nex sec ion.8 Mos o hese a iables a e also d awn om he OECD
Heal h Da abase. Agains Ha wig and S u m (2014), we upda ed he sample as a as pos-
sible and included suga in ake, he impo ance o which as HCE d i e was demons a ed
by Cas o (2017).9 Ou sample pe iod uns om 1971 o 2019.
Va iable desc ip ions and summa y s a is ics a e gi en in Tables1 and 2.
7 We mos ly used he 2020 e sion o he OECD Heal h da abase. All da a a e a ailable on eques o he
co esponding au ho .
8 Ha wig and S u m’s da a we e also used by Hauck and Zhang (2016).
9 We ound no sou ces o upda e he dummy a iables used by Ha wig and S u m (2014) o model he
ins i u ional speci ics o he na ional heal h sys ems. Since none o hese ins i u ional dummy a iables
u ned ou as a obus and s a is ical signi ican explana o y a iable o HCE g ow h in Ha wig and
S u m’s EBAs, and in o de no o o ego he mos ecen yea s by including he non-upda ed dummies, we
decided o d op hem om ou analysis.

163
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
Me hodology
Baseline esul s
To p oduce baseline esul s we eg ess he g ow h a e o HCE/ACE/LTCE in eal e ms on
he g ow h a e o eal GDP pe capi a and on he Baumol a iable (i.e. he g ow h a e o
he wage sha e). We include eal GDP because o he longs anding insigh o igina ing om
Newhouse (1977) ha GDP (o income) d i es HCE. Resea ch in o he de e minan s o
HCE g ow h since Newhouse’s pionee ing s udy has o a long ime ailed o disclose o he
obus explana o y a iables beyond na ional income g ow h (see Robe s, 1999). The e-
o e, we include only GDP g ow h and he Baumol a iable in ou baseline model as well
as in he M- ec o o he Ex eme Bounds Analysis o be discussed below.
The models a e es ima ed wi h OLS using coun y-clus e ed obus s anda d e o s as
well as wi h an ou lie - obus MM es ima o . The ollowing speci ica ions a e es ima ed:
he i s model includes only a cons an , he second one addi ionally ixed coun y e ec s
(FE coun y) and he hi d model ixed yea e ec s (FE ime). The ou h speci ica ion
includes bo h ixed coun y and yea e ec s (FE bo h). Fu he mo e, a i h OLS speci ica-
ion inco po a ing coun y-speci ic ends (CST) is included.10
Ex eme bounds analysis
To examine he sensi i i y o he indi idual a iables on pe -capi a HCE g ow h, we apply
( a ian s o ) EBA, as sugges ed by Leame (1985) and Le ine and Renel (1992).11 This
app oach, which has been widely used in he economic g ow h li e a u e, has become a
popula ool o economis s who wan o es he obus ness o he esul s o hei empi ical
wo k. In addi ion, he EBA p o ides an oppo uni y o es whe he a pa icula de e mi-
nan is obus ly ela ed o he dependen a iable.
The cen al di icul y in his esea ch—which also applies o he esea ch opic o he
p esen pape —is ha se e al di e en models may all seem easonable gi en he da a bu
yield di e en conclusions abou he pa ame e s o in e es . As a gued by Temple (2000),
i is a e in empi ical esea ch ha we can say wi h ce ain y ha one model domina es
all o he possibili ies in all dimensions. In hese ci cums ances, i makes sense o p o ide
in o ma ion abou how sensi i e he indings a e o al e na i e modelling choices. EBA
p o ides a ela i ely simple means o doing exac ly his. I in ol es sys ema ically es ing
all possible combina ions o a iables in a eg ession model. Speci ically, he EBA in ol es
unning a la ge numbe o eg essions, each wi h a di e en combina ion o a iables o
see how sensi i e he es ima ed coe icien s a e o changes in he speci ica ion. Fo each
eg ession, he coe icien o in e es and he associa ed -s a is ic a e eco ded. Finally, he
dis ibu ion o hese coe icien s and -s a is ics ac oss all he eg essions is examined o
de e mine whe he he coe icien o in e es is obus o changes in he speci ica ion. Equa-
ions o he ollowing gene al o m a e es ima ed:
whe e Y is he dependen a iable; M is a ec o o ‘s anda d’ explana o y a iables ha
will be included in each eg ession model; F is he a iable o in e es ; Z is a ec o o
(1)
Y
=
𝛼M
+
𝛽F
+
𝛾Z
+
u,
10 Fo echnical easons, he unc ion lm ob canno calcula e MM es ima es wi h coun y-speci ic ends.
11 Pa s o his sec ion ely upon p e ious wo k (Ha wig and S u m 2014).
164
K.Celebi e al.
Table 1 Va iable desc ip ions
Va iable code Va iable label Desc ip ion Sou ce T ans o ma ion
Baumol Baumol Compensa ion o employees as pe cen age o g oss alue added OECD Di e ence o log
gdp GDP p.c GDP pe capi a in US-dolla s PPP OECD Di e ence o log
pop65 Popula ion ≥ 65yea s Sha e o popula ion 65yea s and o e OECD/Eu os a Fi s di e ence
pop80 Popula ion ≥ 80yea s Sha e o popula ion 80yea s and o e OECD/Eu os a Fi s di e ence
p1564 Female p. Female pa icipa ion in he labo o ce (% o ac i e pop.) OECD Fi s di e ence
u Unemploymen a e Unemployed as a sha e o he labo o ce (%) OECD Fi s di e ence
a Heal h adminis a ion spending Pe capi a eal expendi u e on heal h adminis a ion, go e nance and
heal h sys em and inancing adminis a ion, pe capi a, cons an p ices,
OECD base yea
OECD Di e ence o log
acciden Road a ali ies Land a ic acciden s, dea hs pe 100,000 popula ion OECD Di e ence o log
alcc Alcohol consump ion Alcohol in ake, li e s pe capi a 15 + OECD Di e ence o log
dp Popula ion densi y Popula ion densi y pe squa e kilome e OECD Fi s di e ence
LE65.F Female li e expec ancy (65) Li e expec ancy a age 65 o emales OECD Di e ence o log
LE65.M Male li e expec ancy (65) Li e expec ancy a age 65 o males OECD Di e ence o log
mo Mo ali y a e (0–69) Mo ali y a e, po en ial yea s o li e los pe 100 000 popula ion, 0–69 OECD Fi s di e ence
obc Tobacco consump ion Tobacco consump ion, g ams pe capi a pe yea 15 + OECD Di e ence o log
co e o Insu ance co e age Insu ance co e age % o o al popula ion co e ed OECD Fi s di e ence
dl.ge d Heal h R&D G oss expendi u e on R&D, compound annual g ow h a e OECD –
suga Suga supply Suga supply in kilog ams pe capi a pe yea OECD Di e ence o log
bedsi Cu a i e beds (pe 1,000) Cu a i e ca e beds pe 1000 inhabi an s OECD Di e ence o log
bedsh Cu a i e beds (pe hospi al) Cu a i e ca e beds pe gene al hospi al OECD Di e ence o log
gsh Public expendi u e Public expendi u e as pe cen age o GDP OECD One yea lagged
i s di e ence
puhes Public- o-heal h expendi u e a io Public expendi u e as a sha e o o al heal h expendi u e OECD Fi s di e ence
exmc Inpa ien expendi u e Sha e o inpa ien expendi u e in o al heal h expendi u e OECD Fi s di e ence
hpi Heal h p iceindex P ice index o o al expendi u e on heal h Eu os a Di e ence o log
doc ca Physicians The s ock o p ac icing physicians pe 1000 popula ion OECD Di e ence o log
165
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
Table 1 (con inued)
Va iable code Va iable label Desc ip ion Sou ce T ans o ma ion
nu ca Nu ses Numbe o ac i ely employed nu ses pe 1000 popula ion OECD Di e ence o log
pe sh Hospi al employmen To al hospi al employmen OECD Di e ence o log
physh Physicians pe 100 beds Physicians pe 100 hospi al beds OECD Di e ence o log
a Specialis - o-GP a io The a io o specialis o gene al p ac i ione s OECD Fi s di e ence
end Renal dialysis Renal dialysis a e pe million inhabi an s OECD Di e ence o log
166
K.Celebi e al.
up o h ee possible addi ional explana o y a iables, which he li e a u e sugges s may
be ela ed o he dependen a iable; and u is an e o e m. The ex eme bounds es o
a iable F s a es ha i he lowe ex eme bound o β— he lowes alue o β minus wo
s anda d de ia ions—is nega i e, and he uppe ex eme bound o β— he highes alue o
β plus wo s anda d de ia ions—is posi i e, he a iable F is no obus ly ela ed o Y.
A main limi a ion o he me hod o EBA is ha i canno decen ly cope wi h s ong
mul icollinea i y. Two highly co ela ed a iables o en u n indi idually insigni ican
when en e ed join ly and should he e o e ideally no bo h en e he EBA. Gi en he la ge
Table 2 Summa y s a is ics
*indica es di e ence o log, which is ep esen ed wi h he p e ix “dl_” h oughou he pape . ** indica es
i s di e ence, which is ep esen ed wi h he p e ix “d_” h oughou he pape
Va iable label Obse a ions Mean SD Min Max
To al heal h ca e expendi u e* 581 0.5 4.2 − 27.8 23.0
Long- e m ca e expendi u e* 581 5.2 25.1 − 24.8 487.2
Acu e ca e expendi u e* 581 0.2 3.9 − 28.5 12.7
Baumol* 580 0.0 2.4 − 13.5 13.2
GDP p.c.* 581 1.8 2.7 − 15.4 11.7
Popula ion ≥ 65yea s** 581 0.2 0.2 − 0.4 1.0
Popula ion ≥ 80yea s** 579 0.1 0.1 − 0.4 0.7
Female p. .** 574 0.5 0.8 − 2.6 7.7
Unemploymen a e** 480 − 0.1 1.2 − 4.4 8.1
Heal h adminis a ion spending* 572 3.1 17.0 − 99.4 166.5
Road a ali ies* 580 − 4.6 15.2 − 133.3 137.6
Alcohol consump ion* 576 − 0.2 4.3 − 27.1 26.2
Popula ion densi y** 480 1.0 1.6 − 4.5 8.1
Female li e expec ancy (65)* 575 0.7 1.4 − 5.7 8.1
Male li e expec ancy (65)* 575 0.9 1.4 − 6.7 8.1
Mo ali y a e (0–69)* 531 − 1.1 21.8 − 127.9 94.1
Tobacco consump ion* 398 − 2.5 10.5 − 80.8 86.0
Insu ance co e age** 520 0.1 1.7 − 5.6 36.4
Heal h R&D* 486 4.2 6.3 − 17.7 56.7
Suga supply* 555 0.9 9.3 − 37.9 115.8
Cu a i e beds (pe 1,000)* 419 − 1.3 7.1 -53.2 103.2
Cu a i e beds (pe hospi al)* 419 − 0.7 8.5 − 93.2 53.9
Public expendi u e** 575 0.1 0.4 − 1.0 5.9
Public- o-heal h expendi u e a io** 581 0.1 1.9 − 7.8 35.2
Inpa ien expendi u e** 580 − 0.1 1.9 − 35.2 7.8
Heal h p ice index* 365 2.3 3.0 − 7.9 27.0
Physicians* 435 1.5 2.5 − 14.7 29.4
Nu ses* 369 1.7 2.9 − 9.0 20.7
Hospi al employmen * 304 1.0 14.1 − 94.4 131.7
Physicians pe 100 beds** 329 − 0.3 12.4 − 171.7 18.2
Specialis - o-GP a io** 406 0.0 0.5 − 2.2 4.8
Renal dialysis* 410 2.5 9.7 − 116.8 36.4
173
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
ACE and LTCE in bo h OLS and MM es ima ions is signi ican a he 1% le el.20 This
sugges s ha he Baumol coe icien is signi ican ly highe o LTCE as dependen a iable
han o ACE.21
The impo ance o income (GDP) seems o be lowe in explaining long- e m ca e
expendi u e compa ed wi h acu e ca e o o e all heal h expendi u e. GDP g ow h emains
obus , bu he coe icien is lowe han in Tables8 and 9, and he a iable is s a is ically
signi ican in less han 50% o he eg essions in he MM-EBA (and less han 10% in he
OLS-EBA). O he obus a iables ‘ou pe o m’ GDP g ow h in e ms o highe p o-
po ions o signi ican coe icien s, o ins ance, he (change in he) unemploymen a e
(d.u ),22 he change in he sha e o he popula ion 80yea s and o e (d.pop80) and he
g ow h a e o he s ock o p ac icing physicians pe 1000 popula ion (dl.doc ca).23 One
eason why we hink he MM-EBA esul s a e mo e plausible han he OLS-EBA esul s
o LTCE is ha he coe icien on d.pop80 is nega i e on a e age in he OLS es ima ions.
Table 6 Tes s—Baumol coe icien LTCE e sus ACE
This able p esen s he es s a is ics and co esponding p alues o a Wald es which assesses whe he he
es ima ed coe icien o he Baumol a iable is equal in he ACE and LTCE es ima es
OLS MM
Cons an FE coun y FE ime FE bo h CST Cons an FE coun y FE ime FE bo h
Tes -s a-
is ic
1.923 0.032 0.517 0.012 0.009 − 0.567 0.236 0.564 0.833
P alue 0.166 0.859 0.472 0.914 0.924 0.285 0.593 0.713 0.797
Table 7 Tes s—GDP coe icien LTCE e sus ACE
This able p esen s he es s a is ics and co esponding p alues o a Wald es which assesses whe he he
es ima ed coe icien o GDP is equal in he ACE and LTCE es ima es
OLS MM
Cons an FE coun y FE ime FE bo h CST Cons an FE coun y FE ime FE bo h
Tes -s a-
is ic
2.951 1.577 2.497 0.952 0.005 − 1.681 − 1.145 − 1.440 − 1.269
P alue 0.086 0.209 0.114 0.329 0.945 0.047 0.126 0.075 0.103
20 In he case o he Baumol coe icien s es ima ed wi h OLS, we ob ain:
(4)
=𝜇ACE
−
𝜇LTCE
√
𝜎2
ACE
n
ACE
+
𝜎2
LTCE
n
LTCE
=
0.642
−
0.787
√
0.05002
20853 +0.25592
20853
=−
80.39
,
whe eas o he Baumol coe icien s es ima ed wi h MM we ge :
(5)
=
0.693−0.923
√
0.04072
1179
+0.11262
1972
=−
81.92
.
21 This esul pu s in o pe spec i e ou ea lie inding based on a seemingly un ela ed eg ession ha he
impac o he Baumol a iable on ACE is no di e en om i s impac on LTCE (see Table6).
22 The impac o unemploymen on LTCE g ow h is much highe han on ACE g ow h.
23 Tha he coe icien on dl.doc ca is nega i e on a e age may seem implausible gi en ha ‘ oo many doc-
o s’ a e hough o inc ease heal h cos s. This seems no o be he case in long- e m ca e, hough.

174
K.Celebi e al.
Table 8 EBA esul s o HCE (wi h yea and coun y FE)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl.Baumol 0.69 0.53 0.84 0.07 100.0 100.0 0.72 0.60 0.85 0.08 100.0 100.0
dl.gdp 0.61 0.34 1.18 0.10 100.0 100.0 0.59 0.38 1.26 0.10 100.0 100.0
d.u 0.40 0.04 1.16 0.21 37.5 100.0 0.48 0.28 0.99 0.19 76.6 100.0
dl.hpi − 0.15 − 0.32 0.02 0.06 63.1 99.1 0.00 − 0.15 0.12 0.07 0.6 57.0
dl. obc 0.01 0.00 0.02 0.01 0.0 98.1 0.01 0.00 0.02 0.01 0.7 98.7
d.gsh_l1 0.60 − 0.26 2.78 0.46 12.7 96.0 0.61 0.14 2.18 0.32 25.5 100.0
d.physh 0.02 − 0.04 0.12 0.02 33.8 93.0 0.03 − 0.06 0.10 0.02 30.2 93.4
dl.suga 0.02 − 0.02 0.04 0.02 0.0 91.1 0.00 − 0.01 0.02 0.01 0.0 74.4
dl.bedsh 0.01 − 0.16 0.06 0.02 1.6 86.4 0.00 − 0.04 0.03 0.02 0.0 79.2
dl. a 0.01 − 0.02 0.08 0.01 22.5 85.7 0.01 − 0.01 0.06 0.01 13.2 95.6
d.puhes 0.28 -2.53 16.75 0.24 3.7 83.2 0.07 − 2.31 7.69 0.08 3.0 56.8
d. a − 0.14 − 0.81 0.52 0.37 0.0 82.8 − 0.27 − 0.59 0.18 0.35 3.4 98.6
dl.pe sh 0.01 − 0.04 0.10 0.02 13.6 82.2 0.02 − 0.04 0.07 0.02 17.9 85.5
dl.acciden 0.00 − 0.03 0.01 0.01 2.1 81.9 − 0.01 − 0.02 0.00 0.01 0.8 100.0
d. exmc 0.20 − 2.54 16.71 0.24 3.0 81.8 0.06 − 2.33 7.69 0.08 3.0 53.8
d.pop80 1.03 − 3.15 6.66 2.16 0.5 80.9 0.09 − 3.64 2.31 1.52 0.0 53.2
dl.LE65.M − 0.10 − 0.37 0.40 0.13 5.3 80.6 − 0.03 − 0.19 0.19 0.12 0.0 60.2
dl.ge d − 0.01 − 0.07 0.07 0.03 0.3 79.1 − 0.01 − 0.04 0.06 0.03 0.0 83.2
dl.nu ca 0.02 − 0.08 0.14 0.07 0.0 78.6 0.02 − 0.03 0.15 0.04 1.6 83.3
d. p1564 0.14 − 0.29 0.68 0.24 0.0 78.4 − 0.08 − 0.30 0.11 0.21 0.0 66.4
d.mo 0.01 − 0.02 0.04 0.01 2.6 78.1 0.00 0.00 0.02 0.01 0.0 96.4
dl.doc ca − 0.05 − 0.29 0.22 0.09 11.0 74.6 − 0.08 − 0.23 0.20 0.09 1.1 97.8
dl.bedsi 0.01 − 0.11 0.11 0.03 0.7 73.5 0.03 − 0.03 0.07 0.03 8.6 97.1
d.dp 0.07 − 0.36 0.61 0.26 0.0 69.7 0.01 − 0.24 0.48 0.17 0.5 52.2
dl.LE65.F − 0.08 − 0.55 0.47 0.16 6.5 67.0 − 0.02 − 0.37 0.27 0.13 0.8 62.0
175
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
This able shows he esul s o OLS-EBA and MM-EBA es ima ions using HCE g ow h as he dependen a iable. Fo he explana o y a iables, he a e age es ima ed coe i-
cien (mean β), minimum and maximum o he coe icien s (min β and max β), hei a e age s anda d de ia ion (Ø se), he p opo ion o signi ican coe icien s a he 5% le el
(%Sign.) and he cumula i e dis ibu ion unc ion CDF(0) o he es ima ed coe icien s a e epo ed. CDF(0) alues > 90% and %Sign alues > 90% a e highligh ed in bold
Table 8 (con inued)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl.alcc 0.02 − 0.04 0.28 0.04 3.5 65.2 0.01 − 0.01 0.08 0.03 0.0 71.3
dl. end − 0.01 − 0.05 0.04 0.02 5.4 62.3 0.00 − 0.03 0.05 0.02 0.0 60.7
d.co e o − 0.01 − 0.49 2.72 0.23 0.0 62.0 0.14 − 0.19 2.25 0.17 7.3 66.0
d.pop65 0.18 − 2.87 2.59 1.08 0.1 57.6 0.04 − 0.74 0.93 0.75 0.0 51.4
176
K.Celebi e al.
Table 9 EBA esul s o ACE (wi h yea and coun y FE)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl.Baumol 0.64 0.50 0.84 0.07 100.0 100.0 0.69 0.55 0.85 0.08 100.0 100.0
dl.gdp 0.62 0.34 1.17 0.09 100.0 100.0 0.57 0.34 1.20 0.10 100.0 100.0
d.u 0.39 0.05 1.27 0.20 48.5 100.0 0.49 0.30 0.98 0.22 67.4 100.0
dl. obc 0.01 0.00 0.03 0.01 0.0 100.0 0.01 0.00 0.02 0.01 1.4 100.0
dl.hpi − 0.15 − 0.31 0.07 0.06 69.5 97.1 − 0.10 − 0.29 0.14 0.12 9.8 93.0
dl.acciden − 0.01 − 0.04 0.00 0.01 11.3 97.0 − 0.01 − 0.04 − 0.01 0.01 28.2 100.0
d.gsh_l1 0.64 − 0.30 2.87 0.43 18.8 96.7 0.42 0.08 1.73 0.27 8.7 100.0
dl.pe sh 0.02 − 0.04 0.08 0.01 10.5 90.5 0.01 − 0.02 0.07 0.02 12.2 82.7
dl.suga 0.01 − 0.02 0.05 0.02 0.0 85.7 0.00 − 0.03 0.01 0.01 0.0 53.7
dl.ge d − 0.02 − 0.10 0.06 0.03 3.8 82.8 − 0.01 − 0.05 0.04 0.03 1.6 81.3
dl.alcc 0.02 − 0.05 0.21 0.03 6.8 82.0 0.01 − 0.05 0.08 0.03 0.0 70.5
d.physh 0.01 − 0.07 0.10 0.02 4.0 81.5 0.02 − 0.10 0.10 0.02 15.0 90.0
dl.nu ca 0.02 − 0.08 0.10 0.07 0.0 77.3 0.02 − 0.03 0.13 0.05 0.0 68.0
dl. a 0.01 − 0.02 0.12 0.01 14.1 74.1 0.02 − 0.01 0.15 0.01 52.3 99.2
d. a − 0.09 − 0.79 0.83 0.35 0.1 72.3 − 0.33 − 0.77 0.23 0.41 2.8 98.6
dl.LE65.M − 0.06 − 0.36 0.34 0.12 0.3 69.0 − 0.03 − 0.19 0.15 0.13 0.0 74.0
dl.bedsi 0.01 − 0.07 0.15 0.03 0.1 69.0 0.01 − 0.04 0.05 0.02 0.0 91.5
dl.bedsh 0.00 − 0.14 0.06 0.02 1.0 68.7 0.00 − 0.05 0.02 0.01 0.5 70.5
d.dp 0.04 − 0.30 0.59 0.23 0.0 62.4 − 0.04 − 0.23 0.33 0.20 0.0 71.3
d. p1564 − 0.05 − 0.43 0.42 0.22 0.0 62.3 − 0.10 − 0.30 0.19 0.23 0.0 84.7
dl.LE65.F 0.00 − 0.45 0.45 0.15 0.4 56.9 0.01 − 0.29 0.35 0.13 0.0 59.5
d.co e o 0.01 − 0.46 3.16 0.21 0.0 56.6 0.09 − 0.27 1.40 0.20 0.0 55.2
d. exmc 0.15 − 4.06 18.49 0.23 2.3 55.1 0.33 − 1.87 7.74 0.15 5.6 62.6
d.mo 0.00 − 0.02 0.03 0.01 0.0 54.4 0.00 − 0.01 0.02 0.01 0.0 88.3
d.pop65 − 0.11 − 2.23 2.37 1.00 0.0 53.7 0.08 − 0.61 1.40 0.81 0.0 60.2
177
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
This able shows he esul s o OLS-EBA and MM-EBA es ima ions using ACE g ow h as he dependen a iable. Fo he explana o y a iables, he a e age es ima ed coe i-
cien (mean β), minimum and maximum o he coe icien s (min β and max β), hei a e age s anda d de ia ion (Ø se), he p opo ion o signi ican coe icien s a he 5% le el
(%Sign.) and he cumula i e dis ibu ion unc ion CDF(0) o he es ima ed coe icien s a e epo ed. CDF(0) alues > 90% and %Sign alues > 90% a e highligh ed in bold
Table 9 (con inued)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl.doc ca − 0.02 − 0.29 0.30 0.08 7.4 53.7 − 0.05 − 0.24 0.22 0.08 2.3 86.4
dl. end 0.00 − 0.04 0.06 0.02 0.1 53.3 0.00 − 0.02 0.03 0.02 0.0 67.4
d.puhes 0.17 − 4.08 18.52 0.23 2.4 52.3 0.33 − 1.86 7.73 0.14 5.4 59.5
d.pop80 − 0.05 − 4.71 3.79 2.00 0.0 50.7 − 0.69 − 4.47 1.29 1.41 0.0 82.5
178
K.Celebi e al.
Table 10 EBA esul s o LTCE (wi h yea and coun y FE)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl. obc − 0.04 − 0.09 0.00 0.07 0.0 100.0 0.00 − 0.02 0.02 0.02 0.0 87.0
dl.Baumol 0.79 − 0.29 1.59 0.51 49.5 99.7 0.92 0.52 1.22 0.13 99.9 100.0
dl.LE65.M − 1.27 − 3.93 0.86 0.96 9.2 98.0 − 0.05 − 0.31 0.25 0.18 0.0 57.7
dl.gdp 0.88 − 0.86 3.45 0.69 7.7 97.7 0.41 0.10 1.27 0.21 43.5 100.0
d.mo 0.12 − 0.09 0.42 0.08 24.3 95.7 0.01 − 0.02 0.03 0.01 0.0 84.5
dl.ge d 0.38 − 0.12 0.96 0.23 32.7 95.1 − 0.03 − 0.09 0.20 0.05 2.4 82.0
dl.suga 0.11 − 0.19 0.23 0.14 0.0 95.1 0.02 − 0.01 0.04 0.02 0.0 92.3
dl. a 0.12 − 0.16 0.51 0.07 39.6 94.1 − 0.02 − 0.14 0.03 0.02 9.7 71.3
d. exmc − 0.56 − 14.87 27.99 1.69 15.5 91.0 0.20 − 3.87 15.34 0.25 15.9 64.0
d.physh 0.08 − 0.20 0.85 0.10 0.3 90.9 0.02 − 0.42 0.23 0.04 27.8 50.4
d.puhes 1.07 − 13.72 30.23 1.69 14.8 90.3 0.21 − 3.86 15.36 0.24 18.6 54.3
dl.alcc 0.31 − 0.51 1.03 0.27 0.4 88.3 − 0.02 − 0.19 0.20 0.08 0.5 75.1
d.u 0.98 − 1.56 3.40 1.67 0.6 87.1 0.81 0.40 1.17 0.32 87.1 100.0
d.pop65 9.82 − 6.69 31.89 8.01 19.1 86.9 − 1.24 − 2.83 0.77 1.29 1.9 94.7
dl.pe sh − 0.08 − 0.36 0.24 0.09 2.1 85.1 0.01 − 0.04 0.13 0.03 7.1 69.5
d. p1564 1.41 − 2.62 6.36 1.75 25.1 81.9 0.06 − 0.32 0.65 0.30 0.0 56.9
dl.LE65.F − 0.43 − 2.24 5.63 1.19 3.9 79.6 − 0.21 − 0.74 0.01 0.21 7.8 99.1
d.dp 0.69 − 2.29 3.14 1.98 0.0 79.6 0.54 − 0.04 1.30 0.30 38.6 99.6
dl.nu ca − 0.29 − 1.28 0.63 0.58 0.0 76.4 0.02 − 0.05 0.25 0.06 6.1 58.2
dl.acciden 0.02 − 0.09 0.11 0.07 0.0 75.5 0.02 -0.01 0.03 0.01 2.7 99.1
dl.doc ca − 1.38 − 7.10 0.86 0.67 56.8 69.6 − 0.18 − 0.40 0.11 0.10 64.1 99.5
dl.bedsi − 0.14 − 2.19 0.56 0.28 2.4 67.7 − 0.04 − 0.26 0.59 0.06 5.4 98.2
dl. end − 0.03 − 0.21 0.10 0.10 4.4 66.9 − 0.02 − 0.15 0.04 0.03 4.2 82.1
d.pop80 − 15.49 − 85.83 28.79 16.03 5.7 65.0 6.57 1.10 11.00 3.06 68.6 100.0
d.gsh_l1 − 0.98 − 10.95 3.09 3.43 0.0 63.5 0.64 − 0.22 1.73 0.42 20.4 98.1

179
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
This able shows he esul s o OLS-EBA and MM-EBA es ima ions using LTCE g ow h as he dependen a iable. Fo he explana o y a iables, he a e age es ima ed coe -
icien (mean β), minimum and maximum o he coe icien s (min β and max β), hei a e age s anda d de ia ion (Ø se), he p opo ion o signi ican coe icien s a he 5%
le el (%Sign.) and he cumula i e dis ibu ion unc ion CDF(0) o he es ima ed coe icien s a e epo ed. CDF(0) alues > 90% and %Sign alues > 90% a e highligh ed in
bold
Table 10 (con inued)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
dl.bedsh − 0.16 − 1.14 0.44 0.18 7.1 63.3 0.03 − 0.06 0.13 0.03 7.9 98.7
dl.hpi 0.39 − 1.16 1.57 0.52 17.0 53.1 − 0.14 − 0.72 0.17 0.19 4.7 91.8
d.co e o − 0.02 − 6.87 3.04 1.22 2.7 51.9 − 0.04 − 0.50 0.71 0.27 2.1 54.2
d. a − 0.28 − 4.23 5.83 3.02 0.0 50.7 − 0.60 − 1.97 0.48 0.81 8.1 89.2
180
K.Celebi e al.
Robus ness es s
Figu e1 shows he esul s o he jackkni e es o HCE. The lowes es ima ed coe icien s
o he GDP and Baumol a iables a e 0.50 and 0.64, espec i ely. These we e ob ained
when Sou h Ko ea and No way ( espec i ely) we e excluded. The highes es ima ed coe -
icien o GDP is ound by excluding Canada (0.56). Fo he Baumol a iable, he highes
coe icien (0.71) esul s when excluding Spain. The mean alues o he 23 coe icien s
calcula ed by he jackkni e es a e
𝜇GDP
=
0.53
and
𝜇Baumol
=
0.69
. Wi h a s anda d de ia-
ion o 0.0144 and 0.0138 espec i ely, he es ima ed coe icien s on GDP and he Baumol
a iable emained ela i ely obus h oughou his expe imen . I is also s iking ha in
all i e a ions he es ima ed coe icien s o bo h independen a iables a e signi ican a he
1% le el. Resul s o he jackkni e es a e e y simila when ACE ins ead o HCE is he
dependen a iable (see Fig.2).
The jackkni e esul s o LTCE as dependen a iable a e displayed in Fig. 3.
The mean alues and s anda d de ia ions o he coe icien s a e:
𝜇GDP
=
0.24
,
𝜎GDP
=
0.0475, 𝜇Baumol
=
0.74
,
𝜎Baumol
=
0.0203
.24 Excluding Iceland esul s in he lowes
GDP coe icien (0.10), while excluding Es onia p oduces he highes one (0.31). The low-
es Baumol coe icien (0.71) is ob ained by excluding Iceland, while he highes one (0.78)
is ob ained by excluding Sweden. All es ima ed Baumol coe icien s emain s a is ically
signi ican a he 1% le el; GDP g ow h, howe e , does no con ibu e o explaining LCTE
g ow h a con en ional le els o signi icance. This con i ms ou inding om Table5 o
he ou lie - obus MM-es ima ion wi h coun y and ime ixed e ec s.
The esul s o he EBA es ima ions pe o med wi h lagged independen a iables
a e shown in Tables 11, 12 and 13 in Appendix 2. When HCE o ACE a e conside ed
as dependen a iables, high CDF(0) alues indica e ha he Baumol a iable emains a
obus explana o y a iable, al hough i s lagged e sion is s a is ically signi ican in ewe
eg essions han i s un-lagged coun e pa , and he coe icien es ima es a e conside ably
lowe . Lagged GDP g ow h is also a qui e obus explana o y a iable o bo h HCE and
ACE g ow h, wi h ela i ely high CDF(0) alues. This is no he case o LTCE g ow h,
howe e . In he LTCE EBA es ima ed wi h OLS, he lagged Baumol a iable d ops below
he 90% CDF(0) h eshold. This is due o ou lie s, howe e , as he MM es ima ion demon-
s a es. E en when lagged by one yea , he Baumol a iable emains a obus explana o y
a iable o LTCE g ow h.
Conclusion
Baumol’s cos disease has been a gued o be a majo d i e o heal h ca e expendi u e
g ow h. Skep ics doub i s impac on expendi u es on acu e ca e, howe e , main aining ha
only labo in ensi e asks in long- e m ca e a e likely o be a ec ed by he cos disease.
Ou aim in his pape was o es his p oposi ion empi ically.
In ou analysis, we combine Ex eme Bounds Analysis (EBA) wi h ou lie - obus MM
es ima ion and use da a o 23 OECD coun ies o e he pe iod 1971–2019. We ind ha ,
al hough he impac o Baumol’s cos disease is s onges on long- e m ca e expendi u e,
24 Re isi ing he ques ion whe he he Baumol e ec a ec s ACE and LTCE di e en ly, hese indings ein-
o ce he conclusion ha he e ec on LTCE is s onge .
181
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
Fig. 1 Jackkni e esul s o HCE No e: The dependen a iable is HCE g ow h. The igu e shows he es i-
ma ed coe icien s and hei co esponding p alues o GDP g ow h and Baumol in each i e a ion, omi -
ing one o he 23 coun ies in each s ep. The coe icien s we e es ima ed using ou lie - obus MM-es i-
ma ion wi h coun y and ime ixed e ec s, using clus e ed s anda d e o s o calcula e he p alues. The
mean
𝜇
and s anda d de ia ion
𝜎
o he 23 coe icien s calcula ed by he jackkni e es a e
𝜇GDP
=
0.53
,
𝜇Baumol
=
0.69
,
𝜎GDP
=
0.0144
and
𝜎Baumol
=
0.0138
Fig. 2 Jackkni e esul s o ACE No e: The dependen a iable is ACE g ow h. The igu e shows he es i-
ma ed coe icien s and hei co esponding p alues o GDP g ow h and Baumol in each i e a ion, omi -
ing one o he 23 coun ies in each s ep. The coe icien s we e es ima ed using ou lie - obus MM-es i-
ma ion wi h coun y and ime ixed e ec s, using clus e ed s anda d e o s o calcula e he p alues. The
mean
𝜇
and s anda d de ia ion
𝜎
o he 23 coe icien s calcula ed by he jackkni e es a e
𝜇GDP
=
0.54
,
𝜇Baumol
=
0.68
,
𝜎GDP
=
0.0129
and
𝜎Baumol
=
0.0159
182
K.Celebi e al.
expendi u e on acu e ca e is ne e heless also a ec ed. These indings a e obus o exclud-
ing single coun ies om he sample as well as o lagging all explana o y a iables by one
yea .
We conclude ha Baumol’s cos disease d i es he whole ange o heal h ca e expen-
di u es, no jus hose on labo in ensi e ca e wo k. Ou esul s hence gi e succo o he
decision aken by he OECD in 2019 o e ise he me hodology used o he o ganiza ion’s
heal h spending p ojec ions o allow o an “impac o he Baumol e ec on heal h ca e as a
whole (ins ead o only o long- e m ca e)” (Lo enzoni e al., 2019, p. 25).
Appendix1: Da a issues a oundLTCE
As OECD e al. (2017, p. 88) no e, “( )o da e, es ima es o spending on se ices o long-
e m heal h ca e ha e been mos ly limi ed o highe -income coun ies, due o he ac ha
in mos low- and middle-income coun ies (LMIC) long- e m heal h ca e is p o ided as
Fig. 3 Jackkni e esul s o LTCE No e: The dependen a iable is LTCE g ow h. The igu e shows he es i-
ma ed coe icien s and hei co esponding p alues o GDP g ow h and Baumol in each i e a ion, omi -
ing one o he 23 coun ies in each s ep. The coe icien s we e es ima ed using ou lie - obus MM-es i-
ma ion wi h coun y and ime ixed e ec s, using clus e ed s anda d e o s o calcula e he p alues. The
mean
𝜇
and s anda d de ia ion
𝜎
o he 23 coe icien s calcula ed by he jackkni e es a e
𝜇GDP
=
0.24
,
𝜇Baumol
=
0.74
,
𝜎GDP
=
0.0475
and
𝜎Baumol
=
0.0203
189
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
OLS-EBA and MM-EBA es ima ions, wi h LTCE g ow h as he dependen a iable and explana o y a iables lagged by one yea . Fo he explana o y a iables, he a e age
es ima ed coe icien (mean β), minimum and maximum o he coe icien s (min β and max β), hei a e age s anda d de ia ion (Ø se), he p opo ion o signi ican coe i-
cien s (%Sign.) and he cumula i e dis ibu ion unc ion CDF(0) o he es ima ed coe icien s a e epo ed.CDF(0) alues>90% and %Sign alues>90% a e highligh ed in
bold
Table 13 (con inued)
OLS MM
Va iables Mean βMin βMax βØ se %Sign CDF(0) Mean βMin βMax βØ se %Sign CDF(0)
L.dl.ge d − 0.03 − 0.39 0.51 0.23 0.3 58.9 − 0.01 − 0.08 0.20 0.06 0.4 78.8
L.dl. end − 0.02 − 0.30 0.11 0.09 0.1 56.1 0.03 − 0.10 0.09 0.04 40.3 88.1
L.dl.hpi 0.04 − 1.18 1.82 0.52 6.8 55.2 − 0.11 − 0.27 0.09 0.15 0.0 91.8
L.dl.suga − 0.06 − 0.42 0.74 0.14 23.8 53.2 0.02 − 0.05 0.06 0.03 0.6 93.2

190
K.Celebi e al.
Acknowledgemen We would like o hank pa icipan s o he 16 h annual con e ence o he Ge man
Socie y o Heal h Economics (dggö) in Halle/Saale and an anonymous e iewe  o his jou nal o alu-
able commen s on an ea lie d a o his pape . Sandq is was a ilia ed a KOF Swiss Economic Ins i u e,
ETH Zu ich, Swi ze land and a i o Ins i u e—Leibniz Ins i u e o Economic Resea ch a he Uni e si y o
Munich, Ge many when esea ch o his pape was ca ied ou .
Funding Open Access unding enabled and o ganized by P ojek DEAL. This esea ch did no ecei e any
speci ic g an om unding agencies in he public, comme cial, o no - o -p o i sec o s.
Decla a ions
Con lic o in e es s None.
Open Access This a icle is licensed unde a C ea i e Commons A ibu ion 4.0 In e na ional License,
which pe mi s use, sha ing, adap a ion, dis ibu ion and ep oduc ion in any medium o o ma , as long
as you gi e app op ia e c edi o he o iginal au ho (s) and he sou ce, p o ide a link o he C ea i e Com-
mons licence, and indica e i changes we e made. The images o o he hi d pa y ma e ial in his a icle
a e included in he a icle’s C ea i e Commons licence, unless indica ed o he wise in a c edi line o he
ma e ial. I ma e ial is no included in he a icle’s C ea i e Commons licence and you in ended use is no
pe mi ed by s a u o y egula ion o exceeds he pe mi ed use, you will need o ob ain pe mission di ec ly
om he copy igh holde . To iew a copy o his licence, isi h p://c ea i ecommons.o g/licenses/by/4.0/.
Re e ences
Ang is , J. D., & Pischke, J. (2010). The c edibili y e olu ion in empi ical economics: how be e
esea ch design is aking he con ou o econome ics. Jou nal o Economic Pe spec i es, 24(2),
3–30.
Ba ne , V., & Lewis, T. (1994). Ou lie s in S a is ical Da a. Wiley.
Ba es, L. J., & San e e, R. E. (2013). Does he U.S. heal h ca e sec o su e om Baumol’s cos dis-
ease? E idence om he 50 s a es. Jou nal o Heal h Economics, 32(2), 386–391.
Baumol, W. J. (1967). Mac oeconomics o unbalanced g ow h: The ana omy o u ban c isis. Ame ican
Economic Re iew, 57(3), 415–426.
Baumol, W. J. (2012). The cos disease. Why compu e s ge cheape and heal h ca e doesn’ . Yale Uni-
e si y P ess, New Ha en (CT) and London.
Bellido, H., Olmos, L., & Román-Aso, J. A. (2019). Do poli ical ac o s in luence public heal h expen-
di u es? E idence p e- and pos -g ea ecession. Eu opean Jou nal o Heal h Economics, 20(3),
455–474.
Be nd , E. R., Cu le , D. M., F ank, R. G., G iliches, Z., Newhouse, J. P., & T iple , J. E. (2000). Medi-
cal ca e p ices and ou pu . In A. J. Culye , & J. P. Newhouse (Eds.), Handbook o heal h economics,
ol. 1A. Ams e dam, Else ie .
B eusch, T. S. (1990). Modelling economic se ies. Ox o d Uni e si y P ess.
Cas o, V. (2017). Pu e, whi e and deadly … expensi e: A bi e swee ness in heal h ca e expendi u e.
Heal h Economics, 26(12), 1644–1666.
Colombie , C. (2012). D i e s o heal h ca e expendi u es: does Baumol’s cos disease loom La ge?
FiFo Discussion Pape , No. 12–5.
Colombie , C. (2017). D i e s o heal h-ca e expendi u e: Wha ole does Baumol’s cos disease play?
Social Science Qua e ly, 95(5), 1603–1621.
C oux, C., Dhaene, G., & Hoo elbeke, D. (2008). Robus s anda d e o s o obus es ima o s, mimeo.
www. econ. kuleu en. be. ch is ophe. c oux/ public.
de la Maisonneu e, C., & Oli ei a Ma ins, J. (2014). The u u e o heal h and long- e m ca e spending.
OECD Jou nal: Economic S udies, 61–96.
Ha wig, J. (2008). Wha d i es heal h ca e expendi u e? – Baumol’s model o ‘unbalanced g ow h’
e isi ed. Jou nal o Heal h Economics, 27(3), 603–623.
Ha wig, J., & S u m, J.-E. (2014). Robus de e minan s o heal h ca e expendi u e g ow h. Applied Eco-
nomics, 46(36), 4455–4474.
Hauck, K., & Zhang, X. (2016). He e ogenei y in he e ec o common shocks on heal hca e expendi u e
g ow h. Heal h Economics, 25(9), 1090–1103.
191
Baumol’s cos disease inacu e e suslong‑ e m ca e: Do he…
Helland, E., & Taba ok, A. (2019). Why a e he p ices so damn high? Heal h, educa ion, and he Bau-
mol e ec . Me ca us Cen e , Geo ge Mason Uni e si y, A ling on, VA.
Hla ac, M. (2016). Ex emeBounds: Ex eme bounds analysis in R. Jou nal o S a is ical So wa e,
72(9), 1–22.
Jee oo, J. (2020). Heal hca e expendi u e and Baumol cos disease in Sub-Saha a A ica. Economics
Bulle in, 40(4), 2704–2716.
Ki chgässne , G. (2009). Das schweize ische Gesundhei swesen: Kos enen wicklung und g undsä zliche
P obleme. Die Volkswi scha , 11–2009, 4–8.
Kolle , M., & S ahel, W. A. (2011). Sha pening Wald- ype in e ence in obus eg ession o small sam-
ples. Compu a ional S a is ics & Da a Analysis, 55(8), 2504–2515.
Kommission ü Konjunk u agen (2006). Re o m des Gesundhei swesens, Jah esbe ich , Be n, 8. Sep-
embe 2006. 385. Mi eilung, Beilage zum Magazin Die Volkswi scha (www. seco. admin. ch).
Leame , E. E. (1985). Sensi i i y analyses would help. Ame ican Economic Re iew, 75(3), 308–313.
Le ine, R., & Renel , D. (1992). A sensi i i y analysis o c oss-coun y g ow h eg essions. Ame ican
Economic Re iew, 82(4), 942–963.
Lo enzoni, L., Ma ino, A., Mo gan, D., & James, C. (2019). Heal h spending p ojec ions o 2030: new
esul s based on a e ised OECD me hodology. OECD Heal h Wo king Pape s No. 110, OECD
Publishing, Pa is.
McAlee , M., Pagan, A. R., & Volke , P. A. (1985). Wha will ake he con ou o econome ics? Ame i-
can Economic Re iew, 75(3), 293–307.
Medei os, J., & Schwie z, C. (2013). Es ima ing he d i e s and p ojec ing long- e m public heal h
expendi u e in he Eu opean Union: Baumol’s ‘cos disease’ e isi ed. Eu opean Commission DG
EcFin Economic Pape s No. 507.
Newhouse, J. P. (1977). Medical-ca e expendi u e: A c oss-na ional su ey. The Jou nal o Human
Resou ces, 12(1), 115–125.
OECD. (2008). OECD Glossa y o s a is ical e ms. OECD Publishing.
OECD. (2023). Heal h a a glance: OECD indica o s. OECD Publishing.
OECD, Eu os a and Wo ld Heal h O ganiza ion. (2017). A sys em o heal h accoun s 2011 (Re ised).
OECD Publishing.
Robe s, J. (1999). Sensi i i y o elas ici y es ima es o OECD heal h ca e spending: Analysis o a
dynamic he e ogeneous da a ield. Heal h Economics, 8(5), 459–472.
Rossen, B., & Fa oque, A. (2016). Diagnosing he causes o ising heal h-ca e expendi u e in Canada:
Does Baumol’s cos disease loom la ge? Ame ican Jou nal o Heal h Economics, 2(2), 184–212.
Rousseeuw, P., & Yohai, V. (1984). Robus eg ession by means o S-Es ima o s. In: J. F anke, W. Hä -
dle, & D. Ma in (Eds.), Robus and nonlinea ime se ies analysis, Lec u e No es in S a is ics, ol
26, Sp inge , New Yo k, NY.
Sala-i-Ma in, X. (1997). I jus an wo million eg essions. Ame ican Economic Re iew, 87(2), 178–183.
Sala-i-Ma in, X., Doppelho e , G., & Mille , R. I. (2004). De e minan s o long- e m g ow h – a
Bayesian a e aging o classical es ima es (BACE) app oach. Ame ican Economic Re iew, 94(4),
813–835.
Salibian-Ba e a, M., & Yohai, V. (2006). A as algo i hm o S- eg ession es ima es. Jou nal o Com-
pu a ional and G aphical S a is ics, 15(2), 414–427.
Swa z, S., & Welsch, R.E. (1986). Applica ions o bounded-in luence and diagnos ic me hods in ene gy
modeling. In: D. A. Belsley, & E. Kuh (Eds.), Model Reliabili y, MIT P ess, Camb idge, MA.
Temple, J. (2000). G ow h eg essions and wha ex books don’ ell you. Bulle in o Economic Resea ch,
52(3), 181–205.
Tian, F., Gao, J., & Yang, K. (2018). A quan ile eg ession app oach o panel da a analysis o heal h-ca e
expendi u e in O ganisa ion o Economic Co-ope a ion and De elopmen coun ies. Heal h Eco-
nomics, 27(12), 1921–1944.
Vande meulen, A. (1968). A emission om Baumol’s disease: Ways o publish mo e a icles. Sou he n
Economic Jou nal, 35(2), 189–191.
Ve a di, V., & C oux, C. (2008) Robus eg ession in S a a, KU Leu en, Wo king Pape KBI 0823.
Wang, L., & Chen, Y. (2021). De e minan s o China’s heal h expendi u e g ow h: based on Baumol’s
cos disease heo y. In e na ional Jou nal o Equi y in Heal h, 20(1), A icle 213.
Yohai, V. (1987). High b eakdown-poin and high e iciency es ima es o eg ession. The Annals o S a-
is ics, 15(2), 642–665.
Publishe ’s No e Sp inge Na u e emains neu al wi h ega d o ju isdic ional claims in published maps and
ins i u ional a ilia ions.