And ews, An ony; Em aloma is, G igo ios
A icle
E iciency measu emen in heal hca e: The ounda ions,
a iables, and models. A na a i e li e a u e e iew
Economics: The Open-Access, Open-Assessmen Jou nal
P o ided in Coope a ion wi h:
De G uy e B ill
Sugges ed Ci a ion: And ews, An ony; Em aloma is, G igo ios (2024) : E iciency measu emen in
heal hca e: The ounda ions, a iables, and models. A na a i e li e a u e e iew, Economics: The
Open-Access, Open-Assessmen Jou nal, ISSN 1864-6042, De G uy e , Be lin, Vol. 18, Iss. 1, pp. 1-24,
h ps://doi.o g/10.1515/econ-2022-0062
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/306078
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/
Re iew A icle
An ony And ews* and G igo ios Em aloma is
Efficiency Measu emen in Heal hca e: The
Founda ions, Va iables, and Models –A Na a i e
Li e a u e Re iew
h ps://doi.o g/10.1515/econ-2022-0062
ecei ed Oc obe 28, 2022; accep ed Decembe 11, 2023
Abs ac : Efficiency and p oduc i i y analysis ha e been
c i ical in heal hca e and economics li e a u e. Despi e he
emendous inno a ion in me hodology and da a a ail-
abili y, a comp ehensi e li e a u e e iew on his opic
has no been conduc ed ecen ly. This a icle p o ides a
h ee-pa li e a u e e iew o heal hca e efficiency and
p oduc i i y s udies. I begins by e iewing he wo p i-
ma y empi ical me hods used in heal hca e efficiency s u-
dies, emphasising he ea men o inefficiency pe sis ence.
Second, p e ious con ibu ions o heal hca e p oduc i i y
esea ch a e discussed wi h a ocus on me hodology and
findings. In he hi d sec ion, a ious measu es o ou pu s,
inpu s, and p ices in heal h li e a u e a e explo ed o
de e mine he ex en o consensus in he li e a u e. On
he me hodological on , he li e a u e e iew shows ha
while he Da a En elopmen Analysis and he S ochas ic
F on ie Analysis ha e been used ex ensi ely in heal hca e
p oduc i i y and efficiency s udies, hei applica ion in he
con ex o longi udinal da a is limi ed. Fu he , no s udy
cu en ly unde akes o measu e he TFP changes and i s
componen s ha use bo h p imal and dual app oaches.
The e is also a conside able a ia ion in he use o inpu s,
ou pu s, and p ice a iables, sugges ing ha he use o a i-
ables in heal hca e p oduc i i y and efficiency li e a u e
es s on he balance be ween da a a ailabili y and he
esea ch scope.
Keywo ds: s ochas ic on ie , hospi al efficiency, dea
1 In oduc ion
The las 30 yea s ha e wi nessed conside able momen um
in he numbe o s udies published on he opic o
heal hca e efficiency. The heo y o p oduc ion and
cos unc ions, ollowing he seminal wo k o Fa ell
(1957), influences he cu en me hods o efficiency e a-
lua ion. Many o he heal hca e li e a u e’s empi ical
me hods e ol e a ound es ima ing ei he echnical o
alloca i e efficiency o bo h (Wo hing on, 2004).
Resea che s ha e ex ensi ely employed on ie -based
efficiency echniques o measu e heal hca e uni s’p oduc-
i i y and efficiency. F on ie echniques a e di ided in o
pa ame ic and nonpa ame ic me hods. Bo h me hods
in ol e es ima ing a on ie agains which he pe o -
mance o heal hca e p o ide s is compa ed. A heal hca e
p o ide on he on ie is belie ed o be able o p o ide a
gi en le el o se ice using he leas amoun o inpu s/
minimum cos o he maximum le el o se ices o a gi en
le el o inpu s/cos (Hollingswo h & Peacock, 2008, p. 2).
The deg ee o de ia ion om he efficien on ie p o ides
an es ima e o he le el o inefficiency.
Da a En elopmen Analysis (DEA) is a nonpa ame ic
me hodology based on linea p og amming ools de el-
oped by Cha nes e al. (1978) and is one o he commonly
used on ie -based me hodologies in heal h efficiency s u-
dies. The DEA on ie includes a se ies o linea segmen s
connec ing one efficien decision-making uni (DMU
1
) o
ano he . The on ie ’s cons uc ion is based on “bes -
obse ed p ac ice,”whe e inefficien DMUs a e “en el-
oped”by he efficiency on ie . A no able ea u e o he
adi ional DEA, and o en conside ed a d awback, is ha
all de ia ions om he on ie a e a ibu ed o ineffi-
ciency. Howe e , his issue has been add essed by he boo -
s ap me hods o Sima and Wilson (1998), which p o ide a
* Co esponding au ho : An ony And ews, College o Business
Adminis a ion, Ajman Uni e si y, Al Ju 1, Ajman, UAE,
e-mail: [email p o ec ed]
G igo ios Em aloma is: Depa men o Economics, Uni e si y o C e e,
Re hymno, G eece
1DMU is a e m used o no - o -p ofi en i ies by Cha nes e al.
(1978), in he seminal pape ha es ima ed hei efficiency.
Economics 2024; 18: 20220062
Open Access. © 2024 he au ho (s), published by De G uy e . This wo k is licensed unde he C ea i e Commons A ibu ion 4.0 In e na ional License.
mechanism o dis inguish be ween inefficiency and s a is-
ical noise, he eby efining he accu acy o he efficiency
es ima es de i ed om DEA.
One o he ea lies applica ions o efficiency measu e-
men echniques was unde aken by Nunamake (1983),
who used he DEA o es ima e he echnical efficiency o
16 hospi als in he s a e o Wisconsin, USA. Soon a e ,
Bo den (1988) and She man (1984) also employed he
DEA me hodology o compu e he echnical efficiency sco es
o hospi als in he USA.
A majo limi a ion o using he DEA comes om he
ac ha i makes an un e ifiable and s ong assump ion o
no measu emen e o o andom a ia ion in ou pu
(Newhouse, 1994). In pa icula , in e p e a ion o DEA-
based esul s may be p oblema ic, as on ie s may be
affec ed by s ochas ic a iance, measu emen e o , o
unobse ed he e ogenei y o da a (Hollingswo h & Pea-
cock, 2008, p. 37).
In he heal hca e sec o , he e a e occasions whe e he
heal hca e uni ’s capaci y o deli e se ices is affec ed by
ac o s ou side he heal hca e p o ide ’s con ol. Fo
example, he sudden onse o a pandemic in a egion, med-
ical equipmen may suddenly b eak down,
2
o he e may
be e o s in he measu emen o he le el o esou ces
used. Since he DEA ails o accoun o andom shocks, i
can in oduce bias o he efficiency sco es (Jacobs e al.,
2006, p. 153). Ne e heless, he DEA and i s a ian s a e s ill
he mos widely used ool in heal hca e s udies, possibly
due o i s ease o use and e sa ili y (Jacobs e al., 2006,
p. 13).
In he las 20 yea s, a ious s udies ha e been pu
o wa d ha can be used in conjunc ion wi h he DEA o
deal wi h efficiency sco es’sensi i i y. One o he mos
popula among hese echniques is he applica ion o he
boo s ap me hodology in oduced by Sima and Wilson
(1998, 2007). The boo s ap me hodology, o some ex en ,
has add essed he issue o he sensi i i y o efficiency
sco es o he sampling a ia ion and has p o ided he s a-
is ical p ope ies o he nonpa ame ic es ima o s. Some
o he ecen s udies in heal hca e li e a u e using boo -
s ap me hodology wi h he DEA include wo k by Alonso
e al. (2015), And ews (2020a), And ews (2020b), Chowdhu y
and Zelenyuk (2016), and Jiang and And ews (2020).
S ochas ic F on ie Analysis (SFA) is, on he o he
hand, a pa ame ic app oach de eloped independen ly
by Aigne e al. (1977) and Meeusen and Van den B oeck
(1977). SFA diffe s omDEAini sassump ion ha
disc epancies be ween ac ual and op imal o ganiza ional
pe o mance a e due o inefficiencies and andom shocks.
In o de o inco po a e he concep o s ochas ic
shocks and inefficiency in SFA, he e o e m is defined
as he sum o wo componen s –a one-sided, non-nega i e
e m ha ep esen s inefficiency and he o he componen ,
which ep esen s andom o s ochas ic fluc ua ions. In
addi ion o he dis ibu ional assump ion, he p oduc ion
unc ion specifica ion is also equi ed in SFA. On he o he
hand, DEA equi es no specifica ion o he p oduc ion
unc ions o dis ibu ional assump ions whe e he effi-
ciency on ie is cons uc ed pu ely based on obse ed
da a (Jacobs e al., 2006, p. 90; Nedelea & Fannin, 2013).
E en hough he e a e challenges associa ed wi h
SFA’s assump ions and specifica ions, i s abili y o sepa a e
andom fluc ua ions beyond a hospi al’s con ol has made
i e y popula . Fu he mo e, SFA allows esea che s o
es ima e he ela ionships be ween ou pu s, inpu s, and
cos s. Fu he , SFA allows esea che s o sepa a e heal h-
ca e p o ide -specificeffec s (he e ogenei y) and ime-spe-
cificeffec s when longi udinal da a a e a ailable. Hence,
applying SFA o longi udinal da a also allows o a mo e
obus es ima e o pa ame e s.
While he majo i y o s udies in heal hca e efficiency
li e a u e use classical in e ences o es ima e he model
pa ame e s, Koop e al. (1997) employed Bayesian in e -
ence o es ima e he model pa ame e s and cos efficiency
by using longi udinal da a on 382 non- eaching U.S. hospi-
als. Mo e ecen ly, Chen e al. (2016) used Bayesian SFA o
es ima e hospi al cos efficiency in 31 p o inces in China.
Al hough SFA’s implemen a ion is mo e demanding in
e ms o modelling and in e p e i e skills (Jacobs e al.,
2006, p. 13), i has been gaining mo e p ominence in heal h-
ca e p oduc i i y and efficiency s udies (Wo hing on,
2004). Some heal hca e s udies ha employ SFA include
wo k by Al-Amin e al. (2016), Chen e al. (2016), Colombi
e al. (2017), and Jiang and And ews (2020). Some mo e
s udies include he ollowing:
Al hough SFA’s implemen a ion is mo e demanding in
e ms o modelling and in e p e i e skills (Jacobs e al.,
2006), i has been gaining p ominence in heal hca e p o-
duc i i y and efficiency s udies (Wo hing on, 2004). Some
heal hca e s udies employing SFA include he wo ks o Al-
Amin e al. (2016), Chen e al. (2016), Colombi e al. (2017),
and Jiang and And ews (2020). Recen addi ions o his
body o li e a u e a e he analysis o heal h acili y
efficiency o non-communicable diseases by Bala e al.
(2023), an examina ion o hospi al cos efficiency in US
acu e ca e by Linde (2023), a dynamic analysis o cos effi-
ciency in New Zealand heal hca e p o ide s by And ews
and Em aloma is (2023), and an e alua ion o he
2The ailu e o equipmen in his ins ance is no assumed o be he
esul o unmain ained o ou -o -da e capi al s ock.
2An ony And ews and G igo ios Em aloma is
empo al–spa ial e olu ion o Heal hca e Se ices Effi-
ciency in 31 Chinese p o inces by Ye and Tao (2023).
Since he in oduc ion o SFA in efficiency li e a u e,
se e al app oaches ha e been pu o wa d ha employ i
in he con ex o longi udinal da a in a ious o he sec o s.
Ea ly esea ch in o longi udinal SFA ocussed on es i-
ma ing ime-in a ian (pe sis en ) o long- un efficiency
(Ba ese & Coelli, 1988; Kumbhaka , 1987; Pi & Lee, 1981;
Schmid & Sickles, 1984), ime- a ying ( ansien ) o sho -
un efficiency (Ba ese & Coelli, 1992, 1995; Co nwell e al.,
1990; Kumbhaka , 1990). O he s udies, such as hose by
Kumbhaka and Heshma i (1995) and Kumbhaka and
Hjalma sson (1995), es ima ed pe sis en efficiency and
ansien efficiency. On he o he hand, s udies by G eene
(2005a,b), Kumbhaka and Wang (2005), and Wang and Ho
(2010) es ima ed ansien efficiency while accoun ing o
he e ogenei y a he cos o igno ing pe sis en inefficiency.
Cu en ly, wo main app oaches exis in he efficiency
li e a u e ha inco po a es he idea o pe sis ence in ine -
ficiency. The diffe ence be ween hese wo app oaches is
due o he specific ea men o he adjus men cos hypo h-
esis in efficiency analysis. S udies such as hose by Colombi
e al. (2014), Filippini and G eene (2016), Filippini and Hun
(2015), Kumbhaka and Heshma i (1995), Kumbhaka and
Hjalma sson (1995), Kumbhaka e al. (2014), and Tsionas
and Kumbhaka (2014) employ SFA wi hou inco po a ing
he adjus men cos heo y. Ins ead, hey di ide o al ine -
ficiency in o sho - un ( ansien ) and long- un inefficiency
(pe sis en inefficiency). In heal hca e li e a u e, s udies
inco po a ing he pe sis en na u e o inefficiency a e
sca ce. So a , only one s udy by Colombi e al. (2017) has
used SFA o e alua e he ansien and pe sis en efficiency
o 133 I alian hospi als.
An impo an poin o no e abou hese s udies is ha
he au ho s diffe en ia e be ween ansien (sho - un) and
pe sis en (long- un) inefficiency by speci ying a ime-
a ying and one-sided ime-in a ian skewed e o e m,
espec i ely. In such specifica ions, bo h sho - un and
long- un inefficiency e ms necessi a e one-sided dis ibu-
ional assump ions ha only ake posi i e alues, such as
hal -no mal, exponen ial, and gamma dis ibu ions. While
hese e ms o en employ he same ype o dis ibu ional
specifica ion, hey a e conside ed independen o each
o he . Also, he sho - un inefficiency es ima es in hese
models a e assumed o be independen be ween diffe en
ime pe iods.
The mo i a ion behind he exis ence o long- un ine -
ficiency in hese models undamen ally es s on he idea
ha he e a e long- un ac o s ha gi e ime-in a ian
cha ac e is ics o pe sis en inefficiency. Examples o
such ac o s include obsole e p oduc ion equipmen and
echnology, subs anda d buildings, subs anda d anspo
sys ems, he con inuous lack o wo k o ce de elopmen
leading o unde exploi ed echnologies, and o he manage-
men igidi ies associa ed wi h adminis a i e p ac ices.
In o he wo ds, long- un inefficiency s ems om s uc-
u al issues ha cons ain efficien me hods due o ope a-
ional igidi ies o e a longe ime ho izon. These ope a ional
igidi ies a e heo ized o be ela ed o physical capaci y,
in as uc u al p oblems, ecu ing manage ial incompe-
ence, and mode n echnology a ailabili y. Though he mo i-
a ion behind pe sis ence in inefficiency makes economic
sense, none o hese s udies inco po a es dependencies in
inefficiency h ough ime.
Ano he s and o efficiency li e a u e p o ides a mo e
comp ehensi e and economically in ui i e way o com-
bining he idea o sho - and long- un inefficiencies in
SFA h ough a dynamic p ocess. This app oach explici ly
highligh s he exis ence o he adjus men cos s o quasi-
fixed inpu s o be he p ima y eason o pe sis ence in
inefficiency o e ime. As a esul , he o ganiza ion would
p e e o emain pa ly inefficien in he sho un due o
high adjus men cos s and ins ead seek o achie e i s a -
ge ed long- un efficiency le el (i.e. s eady-s a e efficiency
le el) in he long un. Fu he mo e, hese models a e
dynamic, allowing he sho - un inefficiency be ween pe -
iods o be dependen .
Ahn and Sickles (2000) pionee ed he dynamic s o-
chas ic on ie app oach by speci ying an au o eg essi e
p ocess o accommoda e pe sis ence in inefficiency due o
adjus men cos s. The dynamic models hey specified use
gene alized non-linea me hods o momen s (GMM) o es i-
ma e he pa ame e s. Howe e , a s udy by Bun and Wind-
meije (2010) highligh s ha in dynamic longi udinal da a
models es ima ed ia GMM me hods, weak ins umen s –
a iables ha inadequa ely co ela e wi h endogenous
p edic o s –can become pa icula ly p oblema ic nea
he uni oo bounda y, whe e a iables exhibi s ochas ic
ends wi hou e e ing o a long- e m mean. This issue
can lead o a skewed a iance a io o e o s, a measu e
compa ing he a iabili y due o inefficiency e sus o he
andom effec s, s aying om he ideal alue o uni y. This
di e gence affec s he model’s accu acy, especially in di -
e en ia ing be ween sho - un and long- un inefficiencies,
emphasizing he need o ca e ul ins umen selec ion in
econome ic analyses o ensu e eliable esul s.
Desli e al. (2003) pu o h a e sion o he dynamic
s ochas ic on ie model es ima ed using maximum like-
lihood me hods (ML), assuming ha he heal hca e p o-
ide -specific in e cep is au o eg essi e, wi h a se o
co a ia es ha influence a heal hca e p o ide ’sp oduc ion
on ie o e ime. Howe e , as Khala and Saunde s (2016)
Efficiency Measu emen in Heal hca e 3
highligh ed, his specifica ion is p one o inciden al pa a-
me e bias due o he co ela ion be ween unobse ed
he e ogenei y and efficiency-specificco a ia esin he
la en equa ion.
Using he Bayesian app oach, Tsionas (2006) p esen ed
a dynamic model whe e an au o eg essi e p ocess was
applied o a ans o med efficiency ha can ake any alue
on he eal line, and hus, a s anda d au o eg essi e p o-
cess can be imposed on i . Simila ly, using he Bayesian
app oach, Em aloma is (2012) used he in e se o he
logis ic unc ion o echnical efficiency as a ans o ma ion
in he au o eg essi e p ocess. Building on Tsionas (2006),
many o he Bayesian dynamic model e sions ha e been
p esen ed in s udies such as hose by Em aloma is e al.
(2011), Galán e al. (2015), Lamba aa e al. (2015), and
Ske as e al. (2018).
These dynamic models a e mo i a ed by he adjus -
men cos hypo hesis, whe e he sho - un efficiency is
de i ed based on he o ganiza ion’s pe o mance ela i e
o he p oduc ion possibili y on ie . In con as , he long-
un efficiency co esponds o he long- un equilib ium
alue o efficiency specified by he au o eg essi e p ocess.
Hence, he dynamic model is mo e flexible, as i accommo-
da es he ela ionship be ween ansien efficiency be ween
diffe en pe iods. Howe e , no such ela ionship is ound in
he models whe e ansien efficiency is assumed o be a
one-sided ime- a ying e o componen . Fu he , in he
non-dynamic model, sho - un efficiency is ob ained om
a sys em ha is always assumed o be in equilib ium. The
assump ion o cons an equilib ium is un ealis ic in he p e-
sence o adjus men cos s and o he igidi ies a ising om
he sec o ’s egula o y amewo k.
To he bes o cu en knowledge, he e a e no s udies
ha ha e inco po a ed he idea o dynamic models in
assessing heal hca e p o ide s’efficiency o p oduc i i y
pe o mances. Jacobs e al. (2006, pp. 174–176) a gue ha in
he sho un, heal hca e p o ide s migh only be able o
pe o m ela i e o he a ious cons ain s imposed by
in as uc u e and a ailable inpu s (e.g. quali y o clinical
equipmen and echnology). The e o e, sho - un efficiency
le els should only be assessed based on he configu a ion o
he inpu s ha a heal hca e p o ide has a ailable. On he
o he hand, heal hca e p o ide s may econfigu e hei
esou ces o b ing abou efficiency imp o emen s in he
long un. This implies ha he heal hca e p oduc ion p o-
cess should be modelled h ough a dynamic link be ween
i s p esen and pas pe o mances Jacobs e al. (2006,
pp. 177–178).
Speci ying a dynamic link basically makes i possible
o he cu en ou pu o a heal hca e p o ide o depend
on he ease wi h which he inpu s can be econfigu ed o
h ough which echnology may be adop ed in he p esence
o adjus men cos s. Gi en he p e alence o public finance
as a undamen al sou ce o heal hca e se ices in he
majo i y o de eloped coun ies and he exis ence o a
highly egula ed ope a ing en i onmen (Jacobs e al.,
2006, p. 3), i is su p ising how li le a en ion is paid o
his dynamic link in heal hca e s udies.
A possible eason o his migh be he complexi y
associa ed wi h he Es ima ion o dynamic longi udinal
models and he small da a sizes ha a e p e alen in
heal hca e efficiency s udies (Jacobs e al., 2006,pp.37–38).
Ne e heless, Colombi e al. (2017) and Hollingswo h and
S ee (2006) a gue ha iden i ying he na u e and he o m
o inefficiency in heal hca e sys ems is c i ical o o mu-
la ing app op ia e policy measu es. Fo example, i a high
deg ee o pe sis ence in inefficiency exis s, especially among
public heal hca e p o ide s, hen unless he e is a econfi-
gu a ion o he cu en o ganiza ional s uc u e o /and a
significan change o go e nmen policy owa ds he o e all
sys em, effo s o imp o e efficiency will no yield expec ed
ou comes.
A selec i e lis o heal hca e efficiency s udies om he
ea ly 1990s o 2020 ha ha e used on ie -based app oaches
is p esen ed in Table A1. While no a comple e lis , i p o ides
a ai ep esen a ion o he me hodology used in he las
h ee decades. A comp ehensi e lis o on ie -based heal h-
ca e efficiency s udies can be ound a Hollingswo h and
Peacock (2008,pp.102–117) and Wo hing on (2004).
I is also wo h no ing ha , based on he s udies lis ed
in Table A1, 36 o he 40 s udies ha applied DEA, wi h o
wi hou boo s apping, 17 o hem exclusi ely used long-
i udinal da a wi h no con ol o unobse ed o uni -spe-
cific he e ogenei y. This is likely o esul in a subs an ial
bias in he measu e o efficiency. Addi ionally, o he 13
s udies ha used SFA on longi udinal da a, only Ba os
e al. (2013), Chen e al. (2016), Colombi e al. (2017), and
Koop e al. (1997) con olled o unobse ed he e ogenei y.
2 P e ious Con ibu ion o TFP
S udies in Heal hca e
When longi udinal da a a e a ailable, i is insigh ul o
in es iga e he changes in p oduc i i y o e ime and o
decompose i in o i s componen s o in es iga e he ela i e
con ibu ions. In he heal hca e sec o , such a s udy will
help analyse he effec o a ge ed policies on he p o ision
o heal h se ices and ul ima ely de e mine he impac o
a ious ini ia i es on he popula ion’s heal h ou comes.
4An ony And ews and G igo ios Em aloma is
Fo example, suppose he in en ion is o examine he p o-
duc i i y o a g oup o heal hca e p o ide s. In ha case,
one could de e mine whe he p oduc i i y change o a
specific heal hca e p o ide is d i en by imp o emen in
he p o ide ’s ela i e efficiency, scale imp o emen , o
echnological p og ess.
In p ac ice, TFP can ei he be es ima ed by using index
numbe me hods o econome ic echniques. Examples o
index numbe me hods include he Malmquis p oduc-
i i y index, he Hicks–Moo s een p oduc i i y index, he
Tö nq is p oduc i i y index, and he Fishe p oduc i i y
index (Jacobs e al., 2006, p. 129). In he econome ic
app oach, eg ession analysis and he SFA a e o en used
o es ima e a p oduc ion o cos unc ion wi h dis ibu-
ional assump ions o es ima e he TFP change and i s
componen s.
A selec i e lis o heal hca e li e a u e s udies ocusing
on TFP change and i s componen s is p o ided in Table A2.
Among o he s, his lis summa izes he me hodology, a i-
ables, and esul s o se e al heal hca e s udies ela ed o
he assessmen o TFP and componen s. While he s udies
in Table A2 decompose he TFP changes in efficiency, scale,
and echnological componen s, hey end o concen a e on
he ela i e con ibu ion o efficiency and echnological
change o changes in he TFP.
The lis o s udies in Table A2 shows he Malmquis
p oduc i i y index (Malmquis , 1953) o be he mos common
app oach o heal hca e p oduc i i y. Malmquis ’s p oduc-
i i y index was in oduced in o he li e a u e h ough he
seminal s udy by Ca es e al. (1982), which adop ed Malm-
quis ’s app oach o cons uc ing quan i y indices as dis ance
unc ion a ios.
Fä e e al. (1992) unde ook one o he ea lies applica-
ions o he Malmquis p oduc i i y index in heal hca e.
The s udy analysed he p oduc i i y changes o a g oup
o pha macies in Sweden and concluded ha mos o he
imp o emen s in he TFP we e due o echnological p o-
g ess. The heal hca e s udies ha ollowed he s udies by
Linna (1998); Maniadakis e al. (1999); Ng (2011); Tambou
(1997)) and ha used he Malmquis p oduc i i y index also
ound posi i e echnological p og ess.
Howe e , ollowing he s udy by Fä e e al. (1992),
ano he applica ion o he Malmquis p oduc i i y index
was unde aken by Bu gess and Wilson (1995). Thei s udy
ound ha echnological decline domina ed he effec o
echnical efficiency on TFP o a g oup o 137 hospi als in
he USA. O he s udies, such as hose by Giuff ida (1999),
González and Gascón (2004), and Jiménez e al. (2003), ha e
also ei he epo ed echnological eg ess o no significan
impac o echnological change on he TFP.
As o he effec o efficiency changes on TFP, s udies
by Dismuke and Sena (1999), Gannon (2008), Linna (1998),
and Maniadakis and Thanassoulis (2000) ound ha bo h
efficiency and echnological p og ess con ibu ed o he
inc ease in he TFP. On he o he hand, Giuff ida (1999),
who assessed he TFP g ow h o 90 English amily heal h
se ice au ho i ies o e he pe iod 1991–1995, ound ha
only echnical and scale efficiency con ibu ed o TFP
g ow h while he e was no no iceable echnological p o-
g ess. Fu he , he s udy highligh ed ha imp o emen s in
he TFP we e minimal and exp essed a limi ed scope o
p oduc i i y g ow h in he heal hca e sec o .
Two Malmquis index s udies ha e inco po a ed he
quali y o heal hca e ou pu s in he assessmen o he
TFP. The ea lies s udy was by Fä e e al. (1995), which
showed ha he inco po a ion o quali y significan ly affec s
he measu e o TFP change. Mo e ecen ly, Ka mann and
Roesel (2017) ound ha quali y imp o emen s con ibu ed
mo e g ow h owa ds TFP han jus ou pu olumes o a
sample o Ge man hospi al da a.
While Malmquis indices a e equen ly u ilized in
heal hca e s udies and demand da a on inpu s and ou pu s
o be consis en ly measu ed o e ime (Jacobs e al., 2006,
p. 137), hei applica ion in heal hca e is challenging due o
equen policy shi s and a iable da a collec ion p ac-
ices. I is impo an o no e ha , al hough he compu a ion
o Malmquis indices does no equi e he assump ion o
cons an e u ns o scale (CRS), hey a e o en calcula ed
unde his assump ion. This common p ac ice, albei no a
necessi y, is p ima ily o acili a e he e alua ion o p o-
duc i i y changes inclusi e o scale effec s. This app oach,
while no always ideal, allows o a simplified analysis o
scale efficiency alongside efficiency and echnical changes
(Coelli e al., 2005, p. 293).
Howe e , e y ew heal hca e s udies use econome ic
app oaches o assess imp o emen s in p oduc i i y and i s
componen s. Mo ikawa (2010) used a fixed-effec s panel-
da a model o es ima e he effec s o inc easing hospi al
size on TFP, based on da a om 239 Japanese medical
acili ies. The s udy ound ha TFP inc eases mo e han
10% when he size o a hospi al doubles. Using ime-se ies
eg ession analysis, Blank and Eggink (2014) used Du ch
hospi al da a o he pe iod 1972–2010 o analyse p oduc-
i i y imp o emen s. Thei s udies looked in o he effec s
o egula ions on changes in he TFP and ound ha hos-
pi al compe i ion e o m ailed o imp o e p oduc i i y
among hospi als. Bo h o hese s udies excluded ineffi-
ciency in he cos /p oduc ion unc ions. Conce ning he
econome ic app oach, only one s udy by Dismuke and
Sena (1999) used he SFA o decompose he TFP change
Efficiency Measu emen in Heal hca e 5
o a g oup o hospi als in Po ugal. Thei esul ound
echnical p og ess o mos o he hospi als.
One o he ad an ages o using SFA is ha i allows o
he con ol o unobse ed ime-in a ian he e ogenei y
when compu ing cos s and inpu elas ici ies. Fu he , i
offe s an oppo uni y o decompose p oduc i i y changes
in o pa s ha ha e a s aigh o wa d economic in e p e-
a ion. Despi e nume ous ad an ages, cu en ly, SFA is no
widely used in he heal hca e sec o o assess changes in
he TFP and i s componen s.
Ano he no ewo hy poin is ha all he s udies in
Table A2, excep o Linna (1998), ha e used he p imal
app oach o es ima e he TFP changes and hei compo-
nen s. No s udy has unde aken a TFP analysis unde
bo h he p imal and dual app oaches. In TFP analysis,
he p imal and dual app oaches offe dis inc me hodol-
ogies o es ima ing p oduc i i y g ow h. The p imal
app oach, o ou pu -based me hod, ocuses on quan i ying
he ou pu g ow h ela i e o he g ow h o inpu s, di ec ly
measu ing he changes in he quan i ies o inpu s and ou -
pu s. I essen ially examines he p oduc ion unc ion o
de i e p oduc i i y changes.
The dual app oach, in con as , is o en cha ac e ized
by he use o o al cos as he dependen a iable, wi h he
p ices o inpu s and ou pu s se ing as independen a i-
ables. This inpu -based o p ice-based me hod de i es TFP
g ow h by analysing how inpu and ou pu p ices impac
he cos s uc u e. I looks a he cos unc ion, ocusing on
he ela ionship be ween cos s and p ices o deduce p o-
duc i i y changes.
The duali y heo y, as ou lined by Jo genson and
G iliches (1967), sugges s ha hese wo app oaches should
heo e ically yield consis en esul s since hey a e bo h
g ounded in he same economic beha iou bu iewed
om diffe en pe spec i es. Howe e , p ac ical diffe ences
can a ise in hei ou comes due o ac o s like measu e-
men e o s, model misspecifica ion, o da a inconsis en-
cies (Kee, 2004). Fo ins ance, he p imal me hod migh be
skewed by inaccu a e measu emen s o physical quan i-
ies, whe eas he dual me hod could be influenced by
p ice fluc ua ions o ma ke changes ha affec cos s
and he p icing o inpu s and ou pu s. Hence, al hough
he p imal and dual me hods a e heo e ically aligned,
hei empi ical implemen a ion can yield di e gen con-
clusions, unde sco ing he need o me iculous da a sc u-
iny and p ecise model specifica ion in TFP esea ch.
Thus, applying bo h me hods can p o ide a comp ehen-
si e iew and help iden i y he easons o a ia ions in
TFP es ima ions.
3 Va iables used in Heal hca e
Efficiency and P oduc i i y
S udies
3.1 Ou pu Va iables
The measu emen o ou pu in he heal hca e sec o is no
s aigh o wa d, as he demand o heal hca e se ices
a ises om he need o imp o e heal h s a us. A heal hca e
ins i u ion combines esou ces such as labou and capi al
o p o ide heal hca e se ices, which indi iduals hen con-
sume, leading o imp o ed heal h (Hollingswo h & Peacock,
2008, p. 21). The e o e, p oduc ion and efficiency analysis
should ideally be based on imp o ing he popula ion’s
heal h s a us (Jacobs e al., 2006, p. 22).
Using heal h esul s when s udying effec i eness and
pe o mance analyses, p oponen s claim ha ou come
me ics a e he p ima y pu pose o deli e ing heal h se -
ices. While he a gumen is con incing, i lacks consis ency
in p ac ical applica ions. The quali y o li e measu es is
o en o mula ed using diffe en key indica o s and me h-
odologies (Hollingswo h & Peacock, 2008,p.24).Fu he -
mo e, heal hca e ou comes may ake yea s o be ealized,
and he collec ion o heal hca e ou come da a may impose
imp ac ically high cos s on he heal h sys em (Jacobs e al.,
2006, p. 27). Addi ionally, he expec ed imp o emen in an
indi idual’s heal h s a us depends on o he ac o s which
may be ou side he heal hca e p o ide s’con ol.
Due o he p ac ical difficul ies in ol ed in measu ing
heal hca e ou comes and he associa ed cos s o collec ion,
a ious measu es o heal hca e ac i i ies a e used as p oxies
o heal hca e ou comes (Jacobs e al., 2006,p.27).These
p oxies measu e heal hca e ou pu s in inpa ien ca e epi-
sodes, ou pa ien isi s, and he leng h o inpa ien s ay (Hol-
lingswo h & Peacock, 2008, p. 24). Fo example, measu es o
heal hca e ac i i ies can include a coun o he pa ien s
admi ed, su gical p ocedu es pe o med, ou pa ien num-
be s, o immuniza ions gi en.
The s udies in Tables A1 and A2 show ha almos all
heal hca e p oduc i i y and efficiency s udies use heal h-
ca e ac i i ies o measu e heal hca e ou pu s. Howe e ,
he excep ions a e s udies ha ocus on measu ing heal h-
ca e p oduc i i y a he egional o c oss-coun y le el. Fo
example, Cozad and Wichmann (2013) used s a e-le el da a
om U.S. hospi als, including su i al a es, heal h s a us,
and popula ion sha e wi hou disabili ies, o measu e ech-
nical efficiency. Simila ly, Kin u (2013) used da a on unde -
6An ony And ews and G igo ios Em aloma is
fi e-yea -old mo ali y a es o 52 dis ic s in Sou h A ica.
In a c oss-coun y analysis, Ce in and Bahce (2016) used
DEA o measu e li e expec ancy and in an mo ali y a es
om 26 OECD coun ies o assess hei ela i e echnical
efficiency using DEA.
F om Tables A1 and A2, i is also e iden ha he
majo i y o s udies used inpa ien admissions o discha ges
as one o he measu es, along wi h some e sions o ou -
pa ien isi s. A hand ul o s udies also used ancilla y se -
ices, such as he numbe o X- ays aken (Pilya sky and
S aa (2008)), labo a o y es s pe o med (A hanassopoulos
& Gouna is, 2001; Pilya sky & S aa , 2008), and ambula o y
isi s (Anca ani e al., 2009; Bu gess & Wilson, 1995;
Chowdhu y & Zelenyuk, 2016) as a measu e o heal hca e
ou pu s.
The numbe o inpa ien s can be conside ed he mos
c i ical measu e o hospi al ou pu in esou ce consump-
ion. The measu ing o inpa ien se ices can u he be
di ided in o he numbe o admissions, he numbe o
inpa ien bed days, and he numbe o sepa a ions.
3
While he majo i y o s udies use sepa a ion as a measu e
o ou pu , he e a e a ew s udies such as hose by Mu e
e al. (2008), Pilya sky and S aa (2008), and Pilya sky
e al. (2006) ha use inpa ien admissions as a measu e
o ou pu .
In an a emp o inco po a e bo h case complexi y and
se e i y in o he measu emen o heal hca e ou pu s, s u-
dies o en use he numbe o inpa ien days o accoun o
case complexi y and esou ce use. One o he ea lies s u-
dies o use his a iable is by G osskop and Valdmanis
(1987). They used acu e and in ensi e ca e inpa ien bed
days along wi h o he a iables o assess he efficiency o
80 hospi als in Cali o nia, USA. Mo e ecen ly, Giménez
e al. (2019) and Jiang e al. (2017) included measu ing
inpa ien days as one o he ou pu a iables in e alua ing
efficiency le els.
Howe e , he use o inpa ien days s ill does no ully
cap u e he case complexi y and can only be conside ed a
c ude measu e (Hollingswo h & Peacock, 2008, p. 24). Fo
example, a one-day inpa ien s ays in he ge ia ic wa d
canno be coun ed as equal o a one-day s ay by a newbo n
in a paedia ic wa d. The ea men s and cos s diffe
g ea ly depending on pa ien s’heal h condi ions and cha -
ac e is ics. Ne e heless, inpa ien days and ea men s
p o ide some eliabili y in e ms o ou pu measu emen
bu do no ully eflec he he e ogenei y o ou pu s (Hol-
lingswo h & Peacock, 2008, p. 25).
In assessing heal hca e efficiency, s udies like Thanassoulis
e al. (2016, 2020) ake an al e na i e app oach by exam-
ining he inpa ien episode a he indi idual pa ien le el.
He e, each case is inhe en ly homogenized by i s diagnosis,
nega ing he need o “case mix”adjus men s. This me hod
ea s each inpa ien episode as a dis inc en i y ca ego -
ized by a specific diagnos ic g oup, hus simpli ying com-
pa isons. None heless, i is c ucial o ensu e he a e age
leng h o s ay is accu a ely conside ed, as his can signifi-
can ly impac he assessmen o heal hca e p oduc i i y.
Howe e , i is impo an o ecognize ha no all s udies
can u ilize such a fixed ca ego iza ion. As Hollingswo h &
Peacock (2008) explain, indi idual inpa ien s ays o en
diffe g ea ly due o he complexi y and se e i y o he
pa ien ’s condi ion. This a ia ion has necessi a ed he
de elopmen o “Case-Mix”adjus ed ou pu s, which s an-
da dize he ou pu s by accoun ing o he a ying se e i y
o pa ien condi ions. Commonly, hese adjus ed measu es
a e based on Diagnosis Rela ed G oups (DRGs), p o iding a
mo e equi able compa ison ac oss di e se pa ien g oups.
While DRGs offe a solu ion o ou pu measu emen incon-
sis ency, hey also in oduce an addi ional laye o com-
plexi y o he analysis, hence he me i in conside ing bo h
me hodologies o a comp ehensi e e iew.
The DRG sys em was pionee ed by Fe e e al. (1980)
who s a ed ha “ he p ima y objec i e in he cons uc ion
o he DRGs was a defini ion o case ypes, each o which
could be expec ed o ecei e ou pu s o se ices om a
hospi al”(p. 5). In o he wo ds, DRG is essen ially a s a is-
ical sys em o classi ying any inpa ien s ay by conside ing
he diagnosis in ol ed and he hospi al esou ces neces-
sa y o ea he condi ion. Each DRG is assigned a specific
p ice based on a leas fi e cha ac e is ics: he pe son’s age,
hei p ima y and seconda y diagnosis, he p ima y and
seconda y su gical p ocedu es and, in some cases, he
gende o he pa ien (Fe e e al., 1980).
A s udy by Rosko and Chilinge ian (1999) ound ha
he inclusion o case-mix ou pu a iables educes he
mean efficiency sco e by mo e han 50%. Ano he s udy
by Bjö kg en e al. (2004) showed ha he efficiency sco es
a y conside ably, depending on he case-mix adjus men s
used o inpa ien se ices. The popula i y o case-mix
ou pu measu es has g own since i was fi s used by Wag-
s aff(1989) o assess he efficiency o 49 Spanish hospi als.
Simila ly, using DRG-based case-mix measu es, B own
(2003) sepa a ed discha ges in o h ee classes o accoun
o ela i e esou ce use and he complexi y o ea men s.
Soon a e , Linna e al. (2006) used DRG-based ou pu mea-
su es o assess hospi als’efficiency in No way and Finland.
A mo e ecen s udy on hospi als in On a io used case-mix
adjus ed weigh ed inpa ien bed days and ambula o y
3Sepa a ion occu s when a pa ien lea es a heal hca e acili y due o
dea h, discha ge o lea es wi hou au ho iza ion.
Efficiency Measu emen in Heal hca e 7
isi s as ou pu s Chowdhu y and Zelenyuk (2016). In New
Zealand, s udies by And ews (2020a,b), and Jiang and
And ews (2020) ha e used DRG-based case-weigh ed inpa-
ien s and p ice-weigh ed ou pa ien isi s as a measu e o
ou pu s.
3.2 Inpu s and P ice Va iables
In heal hca e li e a u e, he measu emen o inpu s ends
o be ela i ely less challenging han ou pu s, as physical
inpu s can o en be measu ed mo e p ecisely han ou pu s
(Jacobs e al., 2006, p. 29). In i s simples o m, he p oduc-
ion o heal hca e se ices in ol es combining esou ces
such as labou , capi al, and o he in e media y inpu s o
p oduce heal hca e se ices, which he indi iduals hen
consume o imp o e hei heal h s a us (Hollingswo h &
Peacock, 2008, p. 21).
As heal hca e is a labou -in ensi e sec o , medical and
non-medical s aff’s con ibu ion is c ucial in p o iding se -
ices o he popula ion. The use o measu es o labou
inpu in efficiency and p oduc i i y s udies a ies signifi-
can ly. O he 40 s udies p esen ed in Table A1 on echnical
efficiency, 18 used he s affnumbe s, and 15 used FTEs o
accoun o labou consump ion. The use o coun s may no
be app op ia e as he numbe o labou uni s does no
accoun o ac ual wo k o ce use; mos impo an ly, hey
do no eflec he ac ual ime spen doing asks. Mo eo e ,
coun s obscu e he mix o s affwho a e employed on a
pa - ime o casual basis o who wo k o e ime (Peacock
e al., 2001). In such cases, compa ed o headcoun s, he FTE
measu e is mo e app op ia e in accoun ing o he mix o
a ious ypes o s aff ime.
Ano he conside a ion in he heal hca e efficiency li -
e a u e ela es o he le el o labou disagg ega ion, based
on he skill le el ha is deemed o be app op ia e. Jacobs
e al. (2006, p. 30) a gue ha unless he e is a pa icula
in e es in analysing he inpu ela ionships o add essing
specific policy- ela ed ques ions, i may be easonable o
agg ega e labou inpu s by weigh ing hem acco ding o
hei ela i e wages. Howe e , such da a on labou p ices
migh no always be a ailable. S udies such as hose by
Mi opoulos e al. (2015) and Somme sgu e -Reichmann
(2000) appea o ha e used an unweighed agg ega ed mea-
su e o labou inpu s. And ews (2020b) used disagg ega ed
labou da a on medical, nu ses, allied, suppo and man-
agemen s aff, And ews (2020a) and Jiang and And ews
(2020) weigh ed and agg ega ed he FTEs o nu ses, allied,
suppo and managemen s aff, based on hei ela i e
p ice.
Suppose he esea ch in e es is o ex ac he inpu
elas ici ies o measu e how each labou g oup’s inpu
in e ac s wi h each o he o wi h ou pu s. In ha case, i
may be app op ia e o use disagg ega ed da a. This may be
pa icula ly impo an when he SFA is used o es ima e
he efficiency and ela ionships be ween a ious inpu s
and ou pu s.
Acco ding o he s udies in Tables A1 and A2, mos
hospi al-based efficiency and p oduc i i y s udies ha e a
leas disagg ega ed labou inpu s in o doc o s, nu ses, and
all o he s aff. On he o he hand, s udies such as hose by
Ahmed e al. (2019), Alonso e al. (2015), and Anca ani e al.
(2016) ha e agg ega ed only he labou inpu s in o doc o s
and nu ses while comple ely igno ing he con ibu ion o
adminis a ion and o he labou g oups. Once again, his
migh be due o he una ailabili y o inconsis ency o da a
o a ious non-medical skill g oups.
Jacobs e al. (2006, p. 30) sugges ha labou inpu s can
also be measu ed in e ms o expendi u e, as physical
inpu s ail o cap u e a ia ions in he wage a es be ween
diffe en labou g oups and o ganiza ions. S udies by Giokas
(2001) and S einmann and Zwei el (2003) used labou expen-
di u e da a as a p oxy o inpu consump ion o es ima e
echnical efficiency. Howe e , due o comme cial and poli-
ical sensi i i y, access o financial da a may be limi ed in
many cases.
Capi al is he second c ucial inpu ac o needed o
unde ake any kind o p oduc i i y and efficiency s udy.
Howe e , measu ing capi al in he heal hca e sec o is
mo e complica ed han labou inpu . This is due o he diffi-
cul y in dis inguishing and measu ing he flow o capi al
se ices om he capi al s ock a any gi en ime. As a esul ,
esea che s o en ely on e y udimen a y measu es, such
as hospi al beds, dep ecia ion o hospi al floo space. Ideally,
he bes indica o o capi al inpu is he flow o capi al se -
ices om capi al s ock (Jacobs e al., 2006,pp.31–32). How-
e e , such se ices a e ha d o measu e in p ac ice, and he
associa ed da a is challenging o ob ain.
In he heal hca e li e a u e, he numbe o beds is he
mos widely used p oxy o he measu e o capi al s ock
(Wo hing on, 2004). Simila ly, based on he s udies lis ed
in Tables A1 and A2, 26 ou o 40 s udies on echnical effi-
ciency and nine ou o 19 s udies on TFP decomposi ion use
beds o measu e capi al s ock. Some ecen s udies ha use
hospi al beds include hose by Ahmed e al. (2019), Colombi
e al. (2017), and Sul an and C ispim (2018). Though widely
used, he use o hospi al beds is a om ideal and can lead
o an o e es ima ion o capi al use, which may esul in
biased es ima es o efficiency (Jacobs e al., 2006, p. 32).
Ano he a ely used measu e o capi al is he “capi al
cha ge,”which was fi s used by Pa kin and Hollingswo h
8An ony And ews and G igo ios Em aloma is
Appendix
Table A1 and Table A2
Table A1: P e ious con ibu ion o heal hca e echnical, alloca i e, and cos -efficiency s udies
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
Vi aliano and
To en (1994)
USA Cos
efficiency
604 Nu sing and
o he heal h- ela ed
acili ies. Da a
ela es o he yea s
1987 and 1990
SFA Ou pu s: pa ien days, admissions and ans e s.
P ices: wages o nu sing aids, nu ses and p ope y
expenses pe squa e ee o a nu sing home. An
indica o a iable o he ype o owne and a a iable
o con olling quali y was also used
Koop e al. (1997) USA Cos
efficiency
382 Non- eaching
hospi als om
1987–1991
Bayesian SFA Ou pu s: numbe o discha ges, inpa ien days, beds,
ou pa ien isi s and case mix index. P ices: wage-
p ice index
F ied e al. (1999) USA Technical
efficiency
990 Nu sing homes.
Da a ela es o he
yea 1993
DEA Ou pu s: inpa ien days. Labou inpu s: egis e ed
nu ses (FTEs), licensed p ac ical nu ses (FTEs), o he
pe sonnel (FTEs). O he inpu : non-pay oll expenses
Maniadakis and
Thanassoulis
(2000)
Sco land Technical,
alloca i e
and scale
efficiency
75 Acu e hospi als
o e he pe iod
1992 o 1996
DEA Ou pu s: acciden and eme gency a endances, case-
mix adjus ed ou pa ien a endances, day cases and
inpa ien discha ges. Labou inpu s: FTEs o doc o s,
nu ses and o he pe sonnel. Capi al inpu s: hospi al
beds and he cubic me es o he hospi al buildings
Giokas (2001) G eece Technical
efficiency
91 Hospi als (72
gene al and 19
eaching hospi als)
o he yea 1992
DEA & SFA Ou pu s: inpa ien days, ou pa ien isi s and ancilla y
se ices. Labou inpu : o al s affea nings. O he
inpu : expendi u e on ope a ing se ices and supplies
Rosko (2001) USA Cos
efficiency
1631 U ban
hospi als o he
pe iod 1990–1996
SFA Ou pu s: ou pa ien isi s and case-mix adjus ed
inpa ien discha ges. Labou p ices: a e age annual
sala y pe FTE employee. Capi al p ice: dep ecia ion
and in e es expenses pe bed
A hanassopoulos
and
Gouna is (2001)
G eece Technical
and
alloca i e
efficiency
98 Public hospi als
in he yea 1992
DEA Ou pu s: medical pa ien s, su gical pa ien s, medical
examina ions and labo a o y es s. Labou inpu s: a
coun o medical, adminis a i e and nu sing
pe sonnel. O he inpu s: ope a ing and
pha maceu ical cos s, medical supply and o he
supply cos s. P ices: only he labou p ice: a e age
annual cos s pe hospi al employee
S einmann and
Zwei el (2003)
Swi ze l-
and
Technical
efficiency
89 Swiss hospi als
co e ing he yea s
1993–1996
DEA Ou pu s: inpa ien s days. Labou inpu s:
expendi u e on academic, nu sing and adminis a i e
s aff.O he inpu : non-labou expendi u e
B own (2003) USA Technical
efficiency
613 Hospi als
ela ing o yea s
1992–1996
SFA Ou pu s: Case-mix discha ges. Labou inpu s: The
FTEs o employees. Capi al inpu : o al beds and o al
expenses minus labou expenses a e p oxies o
capi al equipmen . Indica o a iables o yea -specific
effec s, p ofi and public hospi als we e used
Chang
e al. (2004)
Taiwan Technical
efficiency
1996: 43 Regional
hospi als and 440
dis ic hospi als. In
1997, he 44
egional hospi als
and 429 dis ic
hospi als
DEA Ou pu s: pa ien days, ou pa ien isi s and su ge ies.
Labou inpu s: he numbe o physicians, nu ses and
ancilla y se ice pe sonnel. Capi al inpu : numbe
o beds
Blank and
Valdmanis (2005)
The
Ne he la-
nds
Cos
efficiency
71 Homes o he
disabled. Da a o
he yea 1998
DEA Ou pu s: numbe o pa ien days. Inpu s: numbe o
gene al pe sonnel, nu sing and medical pe sonnel,
auxilia y pe sonnel and weigh ed ma e ial supplies
cos s. Inpu p ices: he egional p ice index was used
(Con inued)
Efficiency Measu emen in Heal hca e 15
Table A1: Con inued
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
Pilya sky
e al. (2006)
Uk aine Technical
efficiency
61 Communi y
hospi als
DEA-
boo s ap
Ou pu s: numbe o medical admissions and su gical
admissions. Labou inpu s: numbe o physicians and
nu ses. Capi al inpu s: numbe o hospi al beds
Ale as
e al. (2007)
G eece Technical
and scale
efficiency
51 Gene al hospi als
o he yea s 2000
and 2003.
DEA Ou pu s: case-mix adjus ed inpa ien cases,
ou pa ien isi s and su gical ope a ions. Labou
inpu s: FTEs o medical and o he s aff.Capi al inpu :
s affed hospi al beds
Linna
e al. (2006)
No way
& Finland
Cos
efficiency
47 Finnish and 51
No wegian public
hospi als in 1999
we e s udied
DEA Ou pu s: weigh ed discha ges, bed days, dayca e and
ou pa ien isi s. P ices: wage expendi u e pe FTE
employee and an inpu p ice index o ope a ing cos s
He (2008) Ge many Technical &
cos
efficiency
1556–1635 Gene al
hospi als each yea
o 2000 and 2003
SFA Ou pu : weigh ed hospi al cases. Labou inpu s: FTE
coun s o doc o s, nu ses, and o he s aff.Capi al
inpu : numbe o beds. Labou inpu p ices: cos o
each labou g oup di ided by espec i e FTEs. Capi al
p ice: cos s o all medical equi emen s
(pha maceu ical d ugs, medical ins umen s,
ansplan s, e c.) di ided by he numbe o ins alled
beds. Va ious exogenous a iables a e included o
con ol o obse able he e ogenei y and o measu e
he effec s on inefficiency
Pilya sky and
S aa (2008)
Uk aine Technical
efficiency
193 Communi y
hospi als and
polyclinics o he
yea s 1997–2001
O de -m
es ima o
( ela ed o
FDH/DEA)
Hospi al Ou pu s: admissions and su gical
p ocedu es. Polyclinics Ou pu s: admissions,
su gical p ocedu es, labo a o y es s and X- ays.
Hospi al inpu s: a coun o nu ses, physicians and
beds. Polyclinics Inpu s: a coun o nu ses and
physicians
Mu e
e al. (2008)
USA Cos
efficiency
1,290 U ban
hospi als in 20
s a es ope a ing
in 2001.
SFA Ou pu s: inpa ien admissions, ou pa ien isi s and
pa ien days in nonacu e ca e uni s. Labou p ice:
a e age sala y and benefi s pe FTE employee. Capi al
p ice: dep ecia ion and in e es expenses pe bed.
Quali y a iable: eaching and he excess in-hospi al
mo ali y a e index
F iesne
e al. (2008)
USA Technical,
alloca i e &
scale
efficiency
80 Hospi als and
1076 obse a ions,
balanced
longi udinal da a
o he pe iod
1998–2001
DEA Ou pu s: case-mix ou pa ien isi s, inpa ien days.
Inpu s: hospi al beds, squa e ee o hospi al, and
paid labou hou s. Labou p ice: a e age eal wage
paid by he hospi al. In e media e inpu p ice:
supply expenses di ided by he numbe o licensed
beds and he p oduce p ice index. Capi al p ice: he
sum o in e es and dep ecia ion expenses di ided by
he squa e oo age o he hospi al and he p oduce
p ice index
Shimshak
e al. (2009)
USA Technical
efficiency
38 Res homes o
he yea 2003
DEA Ou pu s: numbe o esiden s, sepa a ed by who
needs ba hing, d essing, ans e ing, oile ing and
ea ing. Inpu s: FTEs o nu ses, nu sing aids, ancilla y
and adminis a i e s aff
Anca ani
e al. (2009)
I aly Technical
efficiency
48 Hospi al wa ds
o he yea 2004.
DEA Ou pu s: ambula o y isi s, discha ges and day
su ge ies. Inpu s: numbe o physicians, non-medical
pe sonnel, numbe o beds, shi s o su ge y ooms
and main enance cos s o medical equipmen .
He e al. (2011) Ge many Cos ,
echnical &
p ofi
efficiency
541 Hospi als
be ween pe iod
2002 and 2006.
Unbalanced
longi udinal
SFA Ou pu s: weigh ed hospi al cases. Inpu s: FTEs o
doc o s, nu ses, and o he s aff.Labou inpu p ices:
sala y o doc o s, nu ses, and o he s affdi ided FTEs.
O he inpu p ices: adminis a ion cos s pe bed and
ma e ial cos pe bed. Capi al inpu : ins alled beds
(Con inued)
16 An ony And ews and G igo ios Em aloma is
Table A1: Con inued
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
Ng (2011) China Scale and
echnical
efficiency
Da a o 2004–08
on 463 hospi als
(balanced
longi udinal)
DEA Ou pu s: ou pa ien s and inpa ien cases. Inpu s:
numbe o doc o s, nu ses, pha macis s and o he
s aff.Capi al inpu : numbe o beds
Medin
e al. (2011)
No way,
Finland,
Denma k
and
Sweden.
Cos
efficiency
70 uni e si y
hospi als in he
No dic coun ies
o e 3 yea s
(2002–2004).
Unbalanced
longi udinal da a
DEA-
boo s ap
Ou pu s: case-mix medical dayca e and inpa ien
discha ges, su gical dayca e and inpa ien discha ges,
and clinical eaching ac i i ies. Inpu s: ope a ing
cos s, cos s o physicians and nu ses. Inpu p ices:
wage index o espec i e coun ies. The au ho s also
used quali y indica o s
Hu e al. (2012) China Technical
efficiency
30 P o ince-le el
hospi al da a o he
yea 2002–2008.
DEA Ou pu s: numbe o ou pa ien and eme gency oom
isi s and he o al numbe o inpa ien days.
Undesi able ou pu - pa ien mo ali y. Labou
inpu s: numbe o doc o s, medical echnicians
(nu ses and physicians), and o he pe sonnel (mainly
adminis a i e s aff). Capi al inpu s: hospi al beds
and alue o fixed asse s
Nedelea and
Fannin (2013)
USA Cos
efficiency
Unbalanced
longi udinal da a
o a se o C i ical
Access Hospi als in
he pe iod
1999–2006
DEA-
boo s ap
Ou pu s: ou pa ien isi s, admissions, pos -
admission days, eme gency oom isi s, ou pa ien
su ge ies, and o al bi hs. Labou inpu : FTEs o
pe sonnel. Capi al inpu : s affed and licensed beds.
Labou p ices: he p ice o labou (pay oll expenses +
employee benefi s) di ided by o al FTEs. Capi al
p ice: dep ecia ion expenses plus in e es expenses
di ided by he numbe o beds in each acili y. A
quali y p oxy a iable was used in he second s age o
unca ed eg ession
Fe ie and
T i i (2013)
USA Technical
efficiency
1,074 Gene al acu e-
ca e hospi als
ope a ing in 2005
DEA Ou pu s: case-mix measu e o inpa ien days,
eme gency oom isi s, ou pa ien isi s, ou pa ien
su ge ies and inpa ien su ge ies. Inpu s: FTEs o
egis e ed nu ses, licensed p ac ical nu ses, medical
esiden s and o he labou . Va ious measu es o
quali y we e also used
Ba os
e al. (2013)
Po ugal Cos
efficiency
51 Hospi als
ela ing o he yea
1997–2008
(balanced
longi udinal da a)
La en
class SFA
Ou pu s: numbe o discha ged pa ien s, ex e nal
consul a ions and eme gency isi s. Inpu p ices:
a io o wages o he numbe o employees and he
egional p ice index. Capi al inpu p oxied by he
numbe o beds
Cozad and
Wichmann (2013)
USA Technical
efficiency
48 S a e-le el
balanced
longi udinal da a
om 2000 o 2007.
DEA-
boo s ap
Ou pu s: su i al a es, heal h s a us, and popula ion
sha e wi hou disabili ies. Labou inpu : numbe o
gene al p ac i ione s and egis e ed nu ses. Capi al
inpu : numbe o hospi al beds
Kin u (2013) Sou h
A ica
Technical
efficiency
52 Dis ic s in Sou h
A ica o he
yea 2001
SFA Ou pu s: unde -fi e mo ali y and co e age o bi h
ca e. Inpu s: pe -capi a public expendi u es on
heal h, heal h insu ance co e age, he p opo ion o
he popula ion wi h access o sa e d inking wa e ,
sani a ion and was e disposal, he densi y o hospi al
beds and he numbe o heal h wo ke s
Yang and
Zeng (2014)
China Technical
efficiency
46 A e public
hospi als o he
pe iod 2006–2010
(balanced
longi udinal)
DEA Ou pu s: numbe o ou pa ien isi s and inpa ien s.
Labou inpu s: numbe o doc o s, nu ses,
adminis a i e s affand o he s aff.Capi al inpu :
numbe o beds
(Con inued)
Efficiency Measu emen in Heal hca e 17
Table A1: Con inued
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
Alonso
e al. (2015)
Spain Technical
efficiency
25 Public hospi als,
in he yea 2009
DEA-
boo s ap
Ou pu : desi able ou pu s: Case-mix adjus ed numbe
o discha ges and he numbe o ou pa ien isi s.
Undesi able ou pu s: In-hospi al mo ali y a e and
he a io be ween pa ien eadmissions and
discha ges. Labou inpu s: FTEs o physicians and
nu sing s aff.Capi al inpu : numbe o beds
Ma eus
e al. (2015)
England,
Po ugal,
Spain
and
Slo enia
Technical
efficiency
Po ugal
(2002–2009) o 102
hospi als. England
(2005–2008) o 163
hospi als. Spain
(2003–2009) o 287
hospi als. Slo enia
(2005–2009) o 19
hospi als
SFA Ou pu s: weigh ed hospi al discha ges. Labou
inpu s: headcoun s o physicians, nu ses and o he
employees. Capi al inpu : numbe o beds
Gok and
Al ındağ(2015)
Tu key Technical
efficiency
251 Hospi als o
he pe iod
2011–2008
(balanced
longi udinal da a)
DEA Ou pu s: bed u iliza ion a e, bed u no e a e, o al
su gical ope a ions, numbe o bi hs, o al ou pa ien
isi s, a e age acili y inpa ien days, and numbe o
discha ges. Labou inpu s: numbe o specialized
physicians and non-specialized physicians. Capi al
inpu s: numbe o hospi al beds
Mi opoulos
e al. (2015)
G eece Technical
efficiency
117 Gene al public
hospi als o he
yea 2009.
DEA Ou pu s: numbe s o inpa ien admissions and
agg ega ed scheduled and eme gency ou pa ien
isi s. Labou inpu s: numbe o doc o s as an
agg ega ion o all special ies o doc o s in he
hospi al, numbe o o he pe sonnel as an
agg ega ion o nu ses, adminis a i e and suppo
s affin he hospi al. Capi al inpu : numbe o
hospi al beds
Co de o
e al. (2015)
Spain Technical
efficiency
132 P ima y ca e
p o ide s in he
yea 2010
DEA Ou pu s: hospi aliza ion a es. Inpu s: numbe o
GPs, nu ses and numbe o p esc ip ions
Anca ani
e al. (2016)
UAE Technical
efficiency
48 Wa ds o h ee
main hospi als in
Dubai o he
yea 2013
DEA Ou pu s: inpa ien su ge y discha ges, inpa ien non-
su ge y discha ges and ou pa ien s. Labou inpu s:
numbe o doc o s and nu ses. Capi al inpu : numbe
o beds
Widme (2015) Swi ze l-
and
Cos
efficiency
333 Hospi als o
pe iod 2004–2009
Bayesian SFA Ou pu s: numbe o case-mix adjus ed inpa ien
cases and e enue om ou pa ien ’s ea men .
Labou p ices: labou expendi u e di ided by FTEs.
O he inpu p ices: p ice o o he inpu s such as
ene gy, ma e ial, and pu chased se ices, compu ed
by di iding o al cos s by he numbe o admissions
Chowdhu y and
Zelenyuk (2016)
Canada Technical
efficiency
113 acu e-ca e
hospi als in On a io
o he yea s 2003
and 2006
DEA-
boo s ap
Ou pu s: ambula o y isi s and case-mix weigh ed
inpa ien days. Labou inpu s: FTEs o nu ses and
adminis a i e wo ke s. O he inpu s: Medical/
o he ical supplies cos s and equipmen cos s. Capi al
inpu : numbe o s affed beds
Ce in and
Bahce (2016)
OECD
coun ies
Technical
efficiency
26 OECD coun ies
in he yea 2016
DEA Ou pu s: li e expec ancy and in an mo ali y a es.
Inpu s: numbe o doc o s, beds and heal h
expendi u e pe capi a
Al-Amin
e al. (2016)
USA Cos
efficiency
1108 Hospi als ha
epo ed HCAPHS
da a in bo h Augus
2008 and July 2009
SFA Ou pu s: a io o eme gency depa men isi s o
o al ou pa ien isi s, he a io o ou pa ien su ge ies
o o al ou pa ien isi s, he p opo ion o o al
hospi al beds classified as acu e ca e, and he a io o
bi hs o o al admissions. Labou inpu p ices: he
(Con inued)
18 An ony And ews and G igo ios Em aloma is
Table A1: Con inued
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
p ice o labou was app oxima ed by he a ea a e age
annual sala y pe ull- ime-equi alen employee.
Capi al p ice: dep ecia ion and in e es expenses
pe bed
Chen e al. (2016) China Cos
efficiency
31 P o incial-le el
hospi al da a om
2002–2011
Bayesian SFA Ou pu s: numbe o su ge ies and o al e enue.
Inpu p ices: sala y expendi u e by he s affo he
hospi als. Capi al p ice: o al dep ecia ion by o al
asse s
Jiang e al. (2017) China Technical
efficiency
1105 Hospi als
ac oss 31 p o inces
o pe iod
2008–2012
DEA Ou pu s: ou pa ien & eme gency isi s and inpa ien
days. Labou inpu s: numbe o physicians, nu ses,
medical echnicians. Capi al inpu : numbe o
open beds
DePuccio and
Ozcan (2017)
USA Technical
efficiency
2212 Gene al
medical-su gical
hospi als in he
yea 2012
DEA Ou pu s: medica e case mix-adjus ed inpa ien
admissions, ou pa ien isi s, and ED isi s. Labou
inpu s: hospi al se ice-mix, non-physician FTEs.
O he inpu : non-labou ope a ing expenses. Capi al
inpu : numbe o s affed and se -up beds
Colombi
e al. (2017)
I aly Technical
efficiency
133 Acu e hospi als
du ing he pe iod
2008–2013
SFA Ou pu s: hospi al annual acu e discha ges co ec ed
by ea men cos . Labou inpu s: annual wo king
hou s o physicians, nu ses and o he wo ke s.
Capi al inpu : o al beds o acu e discha ges
S e ko
e al. (2018)
Slo ak
Republic
Technical
efficiency
8 Regions du ing
he pe iod
2008–2015
DEA Ou pu s: use o beds and a e age nu sing ime.
Labou inpu s: numbe o medical s aff.O he inpu :
quan i y o medical equipmen , magne ic esonance
and compu ed omog aphy. Capi al inpu : numbe
o beds
Sul an and
C ispim (2018)
Pales ine Technical
efficiency
11 Public hospi als
om 2010 o 2015
DEA Ou pu s: o al numbe o annual ca e days, annual
ou pa ien isi s and cases se ed wi hou admission.
Inpu s: FTEs o nu ses, echnicians, and o he
employees in pa amedical depa men s and he
adminis a i e s aff.Capi al inpu : numbe o
hospi al beds
Fe ei a and
Ma ques (2019)
Po ugal Technical
efficiency
7 Hospi als and 20
hospi al cen es,
ope a ing be ween
2013 and 2016
DEA Ou pu s: numbe o inpa ien discha ges, eme gency
cases, fi s medical appoin men s, ollow-up medical
appoin men s, ou pa ien su ge ies, con en ional
su ge ies, u gen su ge ies and numbe o bi hs.
Labou inpu s: FTEs o doc o s, nu ses, hospi al days.
Also, he use o a ious expendi u es as inpu s
Giménez
e al. (2019)
Mexico Technical
efficiency
606 public and 182
p i a e hospi als
DEA Ou pu s: su gical medical p ocedu es, medical
consul a ions, days o s ay and hospi al discha ges.
Labou inpu s: numbe o doc o s in di ec con ac
wi h he pa ien and nu ses. Capi al inpu s:
ope a ing ooms and licensed beds
Ahmed
e al. (2019)
Banglad-
esh
Technical
efficiency
62 Dis ic hospi als
o he yea 2015
DEA Ou pu s: numbe o women ecei ing ANC se ices,
egula deli e ies, caesa ean-sec ion se ices, PNC
se ices, ou pa ien isi s and inpa ien admissions.
Labou inpu s: numbe o doc o s and nu ses.
Capi al inpu : numbe o beds
Jiang and
And ews (2020)
New
Zealand
Technical
efficiency
and cos
efficiency
20 Dis ic heal h
boa ds o pe iod
2011–2017.
SFA & DEA Ou pu s: case-weigh ed inpa ien discha ges and
p ice-weigh ed ou pa ien isi s. Labou inpu s: FTEs
o medical and weigh ed nu ses and o he s aff.
Capi al inpu : dep ecia ion and capi al cha ges.
In e media e inpu s: expendi u e on clinical
supplies. Labou p ice: o al expendi u e di ided by
FTEs. Capi al p ice: capi al cha ges di ided by
(Con inued)
Efficiency Measu emen in Heal hca e 19
Table A1: Con inued
S udy Coun y Efficiency
ype
Facili y ype and
pe iod
Me hodology Va iables
inpa ien discha ges. In e media e inpu p ice: o al
expendi u e di ided by inpa ien discha ges
And ews (2020a) New
Zealand
Technical
efficiency
20 Dis ic heal h
boa ds o he
pe iod 2011–2017
DEA-
boo s ap
Ou pu s: case-weigh ed inpa ien discha ges and
p ice-weigh ed ou pa ien isi s. Labou inpu s: FTEs
o medical, nu ses, allied, suppo and managemen
s aff.Capi al inpu : capi al asse s alue.
In e media e inpu s: clinical supply expendi u e
And ews (2020b) New
Zealand
Technical
efficiency
20 Dis ic heal h
boa ds o pe iod
2011–2018
DEA-
boo s ap
Ou pu s: case-weigh ed inpa ien discha ges and
p ice-weigh ed ou pa ien isi s. Labou inpu s: FTEs
o medical and weigh ed nu ses & o he s aff.Capi al
inpu : capi al asse s alue. In e media e inpu :
clinical supply expendi u e
20 An ony And ews and G igo ios Em aloma is
Table A2: P e ious con ibu ion o heal hca e TFP s udies
S udy Coun y Facili y ype and pe iod Me hodology Va iables Findings
Fä e e al. (1992) Sweden 42 Swedish g oup pha macies
o pe iod 1980–1989
Malmquis index Ou pu s: numbe o d ug deli e ies,
p esc ip ion d ugs, medical appliances and
o e he coun e goods. Labou inpu s:
numbe o hou s o pha macis s, echnical
s aff, o he building and equipmen se ices
s aff.Capi al inpu : dep ecia ion amoun .
A e age TFP inc eased in se en pe iods and
dec eased in wo pe iods. On a e age,
p og ess in TFP du ing he la e pa o he
1980s was due o he posi i e shi s in he
on ie
Bu gess and
Wilson (1995)
USA 137 Nonpsychia ic hospi als
o he pe iod 1985–1988.
DEA-Malmquis index Ou pu s: numbe o acu e ca e inpa ien days,
case-mix weigh ed acu e ca e inpa ien
discha ges, long- e m ca e inpa ien days,
ou pa ien isi s; ambula o y su gical
p ocedu es and inpa ien su gical p ocedu es.
Labou inpu s: FTEs o nu ses, o he clinical
labou , nonclinical labou , and long- e m ca e
labou s aff.Capi al inpu : numbe o acu e-
ca e beds and long- e m hospi al beds
On a e age, he e was echnical eg ess
which domina ed changes in inefficiency in
de e mining changes in TFP
(RNs) measu ed in ull ime equi alen s (RNF/
E); licensed p ac ical nu ses (LPNs) measu ed
in ull ime equi alen s (LPFTE); o he clinical
labo (excluding RNs and LPNs) measu ed in
ull ime equi alen s (XCF/∼); nonclinical labo
measu ed in ull ime equi alen s (NCFTE); and
long- e m ca e labo measu ed in ufl- ime
equi alen s. Capi al inpu : Dep ecia ion
amoun .
Fä e e al. (1995) Sweden 257 Pha macies in ci ies and
subu ban a eas o he pe iod
1990–1991
Malmquis index Ou pu s: numbe o p esc ip ions, d ug
deli e ies, p esc ip ion d ugs, medical
appliances and o e - he-coun e goods.
Labou inpu s: numbe o hou s o
pha macis s, echnical s aff, building and
equipmen se ices s affCapi al inpu :
dep ecia ion amoun .
The esul s sugges ha he inco po a ion o
quali y makes a diffe ence in measu ed
p oduc i i y change
Tambou (1997) Sweden 20 Oph halmology
depa men s in a ious
hospi als om 1988 o 1993
DEA-boo s ap Malmquis
index
Ou pu s: numbe o pe o med ope a ions o
ca a ac , glaucoma, squin diseases and
numbe o physician isi s. Labou inpu s:
FTEs o specialis s and o he physicians.
Capi al inpu : numbe o beds
The posi i e changes in TFP a e mainly due
o posi i e changes in p oduc ion
echnology a he han an o e all posi i e
change in ela i e ( echnical) efficiency o
scale efficiency
Linna (1998) Finland 43 Acu e hospi als in pe iod
1988–1994.
Malmquis index Ou pu s: DRG weigh ed inpa ien episodes,
numbe o ou pa ien s, eme gency isi s,
esiden s, esea ch ou pu s and nu sing
s uden s. Labou p ice: pe sonnel p ice index
Resul s showed a 3–5% annual a e age
inc ease in TFP, hal o which was due o an
imp o emen in cos efficiency and he o he
hal due o echnological change
(Con inued)
Efficiency Measu emen in Heal hca e 21
Table A2: Con inued
S udy Coun y Facili y ype and pe iod Me hodology Va iables Findings
Dismuke and
Sena (1999)
Po ugal 58 Hospi als du ing he yea s
1992–1994
SFA and DEA-Malmquis
index
Ou pu s: DRG weigh ed desi able and
undesi able discha ges. Inpu s: au ho s
concen a e on diagnos ic echnology
u iliza ion on h ee echnological inpu s: he
compu e ized axial omog aphy scanne , he
elec oca diog am and he echoca diog am in
he p oduc ion o discha ges
Imp o emen o echnical efficiency has no
been accompanied by an equi alen
imp o emen in he quali y o ou pu in
dis ic hospi als. The pa ame ic on ie s
show echnical p og ess in mos ou pu s,
excep echoca diog ams which expe ienced
echnical eg ess
Giuff ida (1999) Uni ed Kingdom 90 English Family Heal h
Se ice Au ho i ies o e he
pe iod 1991–1995
DEA-Malmquis index Ou pu s: he o al numbe o people
egis e ed wi h a gene al p ac i ione b oken
down by a ious demog aphics. Also, a
measu e o in e media e ou pu s, such as p e-
de e mined a ge s o child en, was also
included. Labou inpu s: numbe o gene al
p ac i ione s and p ac ice nu ses
The imp o emen in TFP was e y small. The
ise was due o pu e p og ess in echnical
efficiency and posi i e changes in scale
efficiency, al hough he echnology shows
no no iceable change. Analysis indica es
e y a limi ed scope o p oduc i i y g ow h
in his sec o
Maniadakis
e al. (1999)
Uni ed Kingdom 72 Acu e Sco ish hospi als o
he pe iod 1992–1996
DEA-Malmquis index Ou pu s: numbe o acciden and eme gency
a endances, case-mix adjus ed ou pa ien
a endances, day cases and inpa ien
discha ges. Labou inpu s: numbe o
doc o s, nu ses, o he pe sonnel. Capi al
inpu : numbe o beds
The imp o emen in TFP was domina ed by
echnical change a he han hospi al-
ela i e efficiency changes
Maniadakis and
Thanassoulis (2000)
Uni ed Kingdom 75 Sco ish hospi als o he
pe iod 1992–1996
DEA-Malmquis index Ou pu s: numbe o acciden and eme gency
a endances, case-mix adjus ed ou pa ien
a endances, day cases and inpa ien
discha ges. Labou inpu s: numbe o
doc o s, nu ses, o he pe sonnel. Capi al
inpu : numbe o beds
The imp o emen in TFP is due o o e all
p og ess in efficiency, which, in u n, is
p ima ily a ibu ed o an inc ease in
alloca i e efficiency. Technical p og ess
esul ed in a small educ ion in he numbe
o inpu s used, bu also a highe cos o
p oduc ion due o he wo sening o he
ma ch be ween inpu mixes and ela i e
inpu p ices
Somme sgu e -
Reichmann (2000)
Aus ia 22 Aus ian hospi als o pe iod
1994 and 1998
DEA-Malmquis index Ou pu s: he o al numbe o pa ien s ea ed
in he ou pa ien s and he numbe o c edi
poin s epo ed by each hospi al, mul iplied by
a s ee ing ac o . Labou inpu s: FTEs o
labou . In e media e inpu : expenses o
ex e nal medical se ices. Capi al inpu :
numbe o beds
TFP dec eased om 1994 o 1995, while i
inc eased om 1995 o 1996. The esul s
showed a posi i e shi in echnology
be ween 1996 and 1998, wi hou any
echnical efficiency imp o emen
(Con inued)
22 An ony And ews and G igo ios Em aloma is
Table A2: Con inued
S udy Coun y Facili y ype and pe iod Me hodology Va iables Findings
Jiménez e al. (2003) Uni ed Kingdom 39 English coun y council
hospi als o pe iod 1992–1995
DEA-Malmquis index Ou pu s: numbe o people who ecei e
esiden ial ca e a day cen es; he numbe o
hou s o domicilia y ca e deli e ed; he
numbe o meals deli e ed o people a home;
and he magni ude o he use cha ges aised
om hose in ca e. Inpu s: g oss cos o all
se ices o olde people
The TFP shows a s eady inc ease, om 0.7%
in yea 2 o 2.3% in yea 5. The e was
minimal imp o emen in any o he
componen s in 1993/1994. Subsequen ly, a
dip o 3.5% in echnological p og ess in
1994/1995 was offse by a 13.3% ise in he
ollowing yea . Con e sely, bo h pu e and
scale efficiencies ell back in he final yea
González and
Gascón (2004)
Spain 80 pha maceu ical labs o he
pe iod 1994–2000
DEA-Malmquis index Ou pu s: ne sales. Labou inpu : labou
cos s. Capi al inpu : fixed asse s dep ecia ion
(capi al). In e media e inpu : o he cos s
The esul s indica e ha imp o emen s in
echnical efficiency and changing
echnology explain mos o he obse ed
TFP g ow h. Howe e , he con ibu ion o
echnological imp o emen s o p oduc i i y
g ow h is minimal
Gannon (2008) I eland Se o hospi als om 1995
o 1998.
DEA-Malmquis index Ou pu s: numbe o case-mix adjus ed
inpa ien s, ou pa ien s and day cases. Labou
inpu s: FTEs o people employed in each
hospi al. Capi al inpu : numbe o beds in
each hospi al
Resul s show ha , on a e age, bo h
echnical and efficiency changes con ibu e
o a highe TFP in la ge hospi als bu lead
o lowe p oduc i i y le els in smalle
hospi als. Howe e , he con ibu ion o
hese p oduc i i y componen s a ies o e
ime, and echnical imp o emen s play a
mo e c i ical ole in inc easing he
p oduc i i y o la ge hospi als
Pilya sky and
S aa (2008)
Uk aine 193 Communi y hospi als o
he yea s 1997–2001
DEA-Malmquis index Ou pu s: numbe o admissions and su gical
p ocedu es. Labou inpu s: FTEs o people
employed in each hospi al. Capi al inpu :
numbe o beds in each hospi al
The o e all a e age TFP did no change
h oughou he obse a ion pe iod.
Howe e , subs an ial de ia ions om uni y
can be obse ed depending on he pe iod
and he egion
Mo ikawa (2010) Japan 239 Seconda y medical a eas
o he pe iod 1998–2007
Fixed-effec s eg ession Ou pu s: numbe o inpa ien days and
ou pa ien isi s. Labou inpu s: FTEs o he
physicians and he a io o physicians o he
o al numbe o o he s aff.Capi al inpu :
numbe o beds mul iplied by he
u iliza ion a e
TFP inc eases by mo e han 10% when he
size o he hospi al doubles
Ng (2011) China 463 Hospi als om Guangdong
p o ince o he pe iod
2004–2008.
DEA-Malmquis index Ou pu s: numbe o inpa ien and ou pa ien
cases. Labou inpu s: numbe o doc o s,
nu ses, pha macis s and o he s aff.Capi al
inpu : numbe o hospi al beds
TFP g ew be ween 2004 and 2008, mainly
d i en by echnological p og ess. Howe e ,
echnical efficiency de e io a ed in he
pe iod unde s udy
(Con inued)
Efficiency Measu emen in Heal hca e 23
Table A2: Con inued
S udy Coun y Facili y ype and pe iod Me hodology Va iables Findings
Blank and
Eggink (2014)
The Ne he lands Agg ega ed hospi al da a o e
he pe iod 1972–2010 which
yielded 39 obse a ions
Time se ies eg ession Ou pu s: numbe o su ge ies and o al
e enue. Inpu p ices: p ice o pe sonnel pe
FTE, p ice o ma e ial supplies is p oxied by he
consume p ice index. Capi al p ice: o al
capi al cos s di ided by dep ecia ion and
in es men
The esul s indica e ha he a e age
p oduc i i y o he hospi al sec o in
diffe en pe iods a ies and ha hese
diffe ences a e ela ed o he s uc u e o
egula ion in hose pe iods. Fu he , he
au ho s a gue ha compe i ion e o m
ailed o imp o e hospi al sec o
p oduc i i y
Ki elsen e al. (2015) Denma k, Finland,
No way, and
Sweden
Public acu e soma ic hospi als
o he pe iod 2005–2007
DEA-Malmquis index Ou pu s: numbe o ou pa ien isi s, DRG
weigh ed inpa ien s and day pa ien s. Inpu s:
eal ope a ing cos s
The esul s show small diffe ences in scale
and echnical efficiency be ween coun ies
bu significan diffe ences in p oduc ion
possibili ies ( on ie posi ion). The coun y-
specific Finnish on ie is he key sou ce o
he Finnish p oduc i i y ad an age
Ka mann and
Roesel (2017)
Ge many Hospi als om 16 ede al
s a es o he pe iod 1993–2013
F on ie -based Malmquis
app oach and Non- on ie
To nq is app oach
Ou pu s: he numbe o discha ges, a quali y
index, and he quali y‐adjus ed numbe o
discha ges (ou come). Labou inpu s: FTEs o
physicians, nu ses, and o he s aff.
In e media e inpu s: defla ed cos s o
ene gy, ma e ials, and se ice expenses.
Capi al inpu : p oxied by he amoun o
defla ed capi al s ocks
The au ho s find ha quali y imp o emen s
a he han inc eases in quan i y olumes
gene a e TFP g ow h in hospi al ca e. Also,
educing he leng h o s ay is a p ope way
o enhance hospi al TFP
24 An ony And ews and G igo ios Em aloma is
