Efficiency in research, collaboration, and innovation: Parametric and nonparametric approaches in Italian universities

Rella, Angela; Dipie o, Anna Ri a
A icle
E iciency in esea ch, collabo a ion, and inno a ion:
Pa ame ic and nonpa ame ic app oaches in I alian
uni e si ies
Jou nal o Inno a ion & Knowledge (JIK)
P o ided in Coope a ion wi h:
Else ie
Sugges ed Ci a ion: Rella, Angela; Dipie o, Anna Ri a (2025) : E iciency in esea ch, collabo a ion,
and inno a ion: Pa ame ic and nonpa ame ic app oaches in I alian uni e si ies, Jou nal o
Inno a ion & Knowledge (JIK), ISSN 2444-569X, Else ie , Ams e dam, Vol. 10, Iss. 3, pp. 1-11,
h ps://doi.o g/10.1016/j.jik.2025.100724
This Ve sion is a ailable a :
h ps://hdl.handle.ne /10419/327619
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-nc-nd/4.0/
E iciency in esea ch, collabo a ion, and inno a ion: Pa ame ic and
nonpa ame ic app oaches in I alian uni e si ies
Angela Rella
a
, Anna Ri a Dipie o
b,*
a
LUM Uni e si y “Giuseppe Degenna o”, Depa men o Managemen , Finance and Technology, 70010, Casamassima, Ba i, I aly
b
Uni e si y o Calab ia, Depa men o Economics, S a is ics and Finance "Gio anni Anania", 87036, A ca aca a di Rende, Cosenza, I aly
ARTICLE INFO
JEL classi ica ions:
I20
Keywo ds:
Uni e si y
Da a en elopmen analysis
S ochas ic on ie analysis
I alian con ex
ABSTRACT
O e he yea s, echnological ad ancemen s and digi al ans o ma ion ha e b ough signi ican changes o he
I alian educa ion sec o . Uni e si y pe o mance measu emen , pa icula ly in e ms o e iciency, has been
widely deba ed, gi en uni e si ies’ c i ical ole in d i ing inno a ion and disco e y. This s udy aims o compa e
nonpa ame ic and pa ame ic app oaches o es ima e he e iciency o I alian uni e si ies, wi h a ocus on
esea ch, collabo a ion, and inno a ion. Speci ically, da a en elopmen analysis and s ochas ic on ie analysis
we e applied using h ee di e en inpu -ou pu combina ions o e alua e he e iciency o 62 I alian public
uni e si ies in 2021. The indings a e aluable om bo h heo e ical and p ac ical pe spec i es. By iden i ying
bes p ac ices and o e ing a ge ed insigh s in o a eas o pe o mance imp o emen , his s udy con ibu es o
he ongoing de elopmen o I alian uni e si ies and hei global compe i i eness in esea ch, coope a ion, and
inno a ion.
In oduc ion
In ecen yea s, I alian uni e si ies ha e unde gone a majo ans-
o ma ion, aiming o s eng hen connec ions be ween academia and
ex e nal ac o s wi hin he inno a ion ecosys em (Compagnucci & Spi-
ga elli, 2024). Th ough he dissemina ion o knowledge, uni e si ies
inc easingly con ibu e o socie al and economic de elopmen while
also playing a c i ical ole in wo k o ce aining (Ma al & Çe in, 2024).
In his con ex , he concep o e iciency is essen ial (Dipie o & De
Wi e, 2024; Kallio e al., 2020) because i conce ns how inpu s a e
con e ed in o ou pu s. This capabili y is especially impo an in he
uni e si y se ing, whe e ins i u ions o en ope a e unde esou ce
cons ain s (Agasis i & Johnes, 2015).
Renowned o i s ich his o y and di e se academic adi ions, he
I alian highe educa ion sec o wa an s he a en ion o p ac i ione s
who can suppo policymake s in guiding ongoing uni e si y ans-
o ma ions. Recen challenges demand p oac i e measu es om I alian
uni e si ies o emain aligned wi h ad ancemen s in knowledge, p o-
mo e collabo a ion wi h indus y and ins i u ions, and suppo inno a-
ion (F ondizi e al., 2019). Howe e , o he bes o he au ho s’
knowledge, empi ical esea ch on he e iciency o I alian uni-
e si ies—speci ically in he con ex o esea ch, collabo a ion, and
inno a ion— emains limi ed.
Acco ding o he sys ema ic li e a u e e iew by Rella and Vi olla
(2024), he e has been limi ed e o o join ly apply he wo mos widely
used echniques o measu ing uni e si y e iciency: da a en elopmen
analysis (DEA) and s ochas ic on ie analysis (SFA). To add ess his
gap, we aim o unde ake a compa a i e e alua ion o he e iciency o
I alian uni e si ies, ocusing on key dimensions such as esea ch,
coope a ion, and inno a ion (Ab amo e al., 2020). Speci ically, we
apply he mos commonly used nonpa ame ic and pa ame ic
app oaches—DEA and SFA (Agasis i & Dal Bianco, 2006; Ta a es e al.,
2021)— o highligh hei espec i e heo e ical and me hodological
ad an ages and limi a ions. Each me hod p o ides dis inc bene i s and
add esses speci ic sho comings, making hem complemen a y ools o
assessing he e iciency o complex sys ems such as uni e si ies. These
ools a e essen ial o e alua ing uni e si y pe o mance, especially
om a manage ial pe spec i e commi ed o os e ing knowledge
de elopmen . Th ough a compa a i e lens, his s udy employs DEA and
SFA o o e a comp ehensi e assessmen o I alian uni e si y e iciency
in ela ion o esea ch, coope a ion, and inno a ion ac i i ies. DEA and
SFA a e applied o a da a se comp ising 62 I alian public uni e si ies
using h ee di e en inpu -ou pu combina ions. This allows us o assess
e iciency ac oss mul iple ac i i y domains. The da a se includes
* Co esponding au ho a : Uni e si y o Calab ia, 87036, A ca aca a di Rende, Cosenza, I aly.
E-mail add esses: [email p o ec ed] (A. Rella), [email p o ec ed] (A.R. Dipie o).
Con en s lis s a ailable a ScienceDi ec
Jou nal o Inno a ion & Knowledge
jou nal homepage: www.else ie .com/loca e/jik
h ps://doi.o g/10.1016/j.jik.2025.100724
Recei ed 23 Decembe 2024; Accep ed 5 May 2025
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
A ailable online 13 May 2025
2444-569X/© 2025 The Au ho s. Published by Else ie España, S.L.U. on behal o Jou nal o Inno a ion & Knowledge. This is an open access a icle unde he CC
BY-NC-ND license ( h p://c ea i ecommons.o g/licenses/by-nc-nd/4.0/ ).
in o ma ion on esea ch publica ions, academic collabo a ions, pa en
applica ions, and inancial esou ces, p o iding a mul idimensional and
de ailed iew o uni e si y e iciency. By adop ing his in eg a ed
app oach, we aim o answe he ollowing empi ical esea ch ques ion
(RQ):
RQ: To wha ex en a e I alian uni e si ies e icien when assessed
using DEA and SFA app oaches?
By iden i ying bes p ac ices and o e ing speci ic insigh s in o a eas
whe e uni e si ies can imp o e pe o mance, his s udy aims o suppo
he ongoing de elopmen o I alian uni e si ies and s eng hen hei
global compe i i eness in esea ch, coope a ion, and inno a ion. Uni-
e si ies oday ope a e in an inc easingly challenging en i onmen ,
o en acing signi ican budge cons ain s ha necessi a e a ocus on
compe i i eness. By op imizing esou ce managemen and enhancing
e iciency, uni e si ies can inc ease bo h hei a ac i eness and
compe i i e posi ioning (Dipie o & De Wi e, 2024). E ec i e esou ce
managemen also imp o es he abili y o uni e si ies o mee he needs
o in e nal and ex e nal s akeholde s (Ashma ina e al., 2015).
In pu suing hese objec i es, his s udy con ibu es o he academic
ield by ecommending ac ionable s a egies ha can enhance he
impac o uni e si ies on inno a ion and knowledge p oduc ion in I aly,
he eby suppo ing b oade socio-economic de elopmen . The p oposed
pe spec i e is also adap able o o he Eu opean con ex s. Speci ically,
he indings o e policy- ele an insigh s o esou ce alloca ion and
ins i u ional e o ms wi hin he highe educa ion sec o .
This a icle is s uc u ed as ollows: Sec ion 2 ou lines he li e a u e
e iew, Sec ion 3 de ails he me hodology, Sec ion 4 desc ibes he ma-
e ials, Sec ion 5 p esen s he esul s, and Sec ion 6 p o ides he dis-
cussion and conclusions.
Li e a u e e iew
The concep o e iciency
De ining e iciency is a complex ask due o he wide ange o in-
e p e a ions p oposed by schola s and p ac i ione s o e ime. In his
s udy, we adop he de ini ion p o ided by Cha nes e al. (1978), which
conside s e iciency as he abili y o con e inpu s in o ou pu s. The e
a e a ious ways o calcula e e iciency sco es, bu he mos widely used
app oach in ol es es ima ing a on ie ha iden i ies op imal alues,
ollowed by compa ing he e iciency sco es de i ed om inpu -ou pu
combina ions (Fø sund e al., 1980; Lo ell, 1993; Wo hing on & Dol-
le y, 2000).
Following his o e iew o he e iciency concep , i is essen ial o
examine how i is measu ed empi ically. As no ed ea lie , me hodo-
logical app oaches a e ypically classi ied in o wo main ca ego ies:
pa ame ic and nonpa ame ic. While bo h aim o quan i y e iciency,
hey di e subs an ially in e ms o s uc u e and unde lying
assump ions.
Nei he me hod is inhe en ly supe io because each o e s dis inc
s eng hs and limi a ions (Sale no, 2003). In his con ex , he p esen
s udy ocuses on wo widely adop ed app oaches: DEA, a nonpa ame ic,
non-s a is ical, and de e minis ic me hod, and SFA, a pa ame ic, s a-
is ical, and s ochas ic me hod. These echniques a e ex ensi ely dis-
cussed in he li e a u e and we e ecen ly e iewed in a sys ema ic
analysis by Rella and Vi olla (2024), o which eade s a e e e ed o
addi ional insigh s. Thei e iew highligh s a signi ican gap in p io
s udies, pa icula ly he unde u iliza ion o combined DEA and SFA
app oaches and he limi ed a en ion gi en o he esea ch, collabo a-
ion, and inno a ion goals o uni e si ies. The ollowing sec ions p esen
he heo e ical ounda ions and empi ical conside a ions necessa y o
unde s and and apply DEA and SFA in he con ex o highe educa ion.
A pano amic iew o da a en elopmen analysis
DEA has become a cen al analy ical ool in ope a ional esea ch
(OR), signi ican ly con ibu ing o he e olu ion o pe o mance mea-
su emen me hodologies (Me goni e al., 2024). DEA is a
decision-making echnique g ounded in linea p og amming (Cha nes
e al., 1978). Speci ically, i e alua es p oduc ion on ie s and measu es
he ela i e e iciency o a se o decision-making uni s (DMUs) ha
u ilize simila inpu s o p oduce simila ou pu s (Nguyen & Pham,
2020). G aphically, DMUs loca ed on he on ie a e conside ed e i-
cien , while hose below he on ie a e deemed ine icien . The ela i e
e iciency o each DMU is calcula ed based on i s dis ance om his
on ie , as measu ed by he DEA model.
Building on Fa ell’s (1957) ounda ional de ini ion o e iciency,
DEA e ol ed in o wo p ima y models. Cha nes e al. (1978) de eloped a
model based on cons an e u ns o scale (CRS) and an inpu -o ien ed
pe spec i e. Subsequen ly, Banke e al. (1984) in oduced he a i-
able e u ns o scale (VRS) model, which is mo e widely suppo ed in
ecen li e a u e due o i s abili y o accoun o non-p opo ional e-
la ionships be ween inpu s and ou pu s.
In empi ical applica ions, Johnes (2004, 2006) and Fø sund e al.
(2018) explo ed a ious dimensions o DEA in he highe educa ion
con ex . Thei con ibu ions discuss he me hodological ounda ions,
echnical backg ound, s eng hs and weaknesses, and he mul iple pe -
spec i es h ough which DEA can be applied. In line wi h his, A ki an
(2001) assessed he e iciency o 36 Aus alian uni e si ies, ocusing on
se ice deli e y, ee-paying en ollmen s, and o e all pe o mance.
Simila ly, Glass e al. (2006) calcula ed he e iciency sco es o 98
non-specialis UK uni e si ies in 1996, wi h emphasis on eaching and
esea ch quali y. Con inuing in his di ec ion, Naza ko and ˇ
Sapa auskas
(2014) examined he e iciency o 19 Polish uni e si ies wi h espec o
educa ional and esea ch quali y, whe eas Con e as and Lozano (2022)
assessed he o e all e iciency o 47 Spanish uni e si ies. Gui onne and
Peypoch (2018) e alua ed he pe o mance o 7735 Ame ican colleges
and uni e si ies in he pe iod 2012 o 2013, ocusing on eaching and
esea ch ou comes. A simila app oach was adop ed by Ta a es e al.
(2021) in hei analysis o 45 B azilian ede al uni e si ies.
Despi e he conside able schola ly a en ion, an impo an empi ical
gap emains. To he bes o he au ho s’ knowledge, no exis ing s udy
has independen ly e alua ed he e iciency o uni e si ies in e ms o
esea ch, collabo a ion, and inno a ion objec i es.
A pano amic iew o s ochas ic on ie analysis
De eloped independen ly by Aigne e al. (1977), Ba ese and Co a
(1977), and Meeusen and an den B oeck (1977), SFA is a pa ame ic
app oach ha inco po a es bo h ine iciency and andom e o in he
es ima ion p ocess. This me hodology es ima es and decomposes he
composi e e o e m in a single s ep. Thus, i accoun s o he p esence
o s ochas ic e o when cons uc ing he e iciency on ie a he han
a ibu ing all de ia ions om he on ie solely o ine iciency (Nguyen
& Pham, 2020). The academic li e a u e iden i ies wo key es ima o s
o SFA: he JLMS es ima o (Jond ow e al., 1982), which calcula es
echnical ine iciency, and he BC es ima o (Ba ese & Coelli, 1988),
which ocuses on es ima ing echnical e iciency (Sena, 2003). In
con as o DEA, which is de e minis ic, SFA adop s a s ochas ic
app oach ha conside s da a noise when calcula ing he dis ance om
he e iciency on ie , o e ing a mo e lexible e alua ion o pe o -
mance (Le i e al., 2022).
Empi ical applica ions o SFA in he con ex o highe educa ion
emain ela i ely limi ed. Agasis i and G alka (2019) applied SFA o
measu e he o e all e iciency o 55 I alian and 70 Ge man public uni-
e si ies o e he pe iod 2001 o 2011. Simila ly, Le i e al. (2022)
examined a sample o 56 B azilian ede al uni e si ies om 2010 o
2016, whe eas Nade i (2022) in es iga ed he e iciency o a comp e-
hensi e uni e si y in I an.
Despi e hese con ibu ions, exis ing s udies p ima ily ocus on
o e all ins i u ional e iciency. To da e, no s udy has applied SFA o
e alua e uni e si y e iciency wi h a speci ic ocus on esea ch,
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
2
collabo a ion, and inno a ion ac i i ies.
A compa ison o da a en elopmen analysis and s ochas ic on ie
analysis
The numbe o s udies compa ing nonpa ame ic and pa ame ic
app oaches, speci ically DEA and SFA, emains limi ed and somewha
inconsis en . In he highe educa ion con ex , McMillan and Chan
(2006) showed ha e iciency measu es ob ained h ough DEA and SFA
yield di e en uni e si y ankings. Simila pa e ns ha e also been
obse ed in o he sec o s, including banking (Nguyen & Pham, 2020;
Weill, 2004), heal hca e (Hollingswo h, 2008; Jacobs, 2001), and en-
e gy (Baue e al. (1998); Hossin e al., 2023) a gued ha no uni e sally
supe io me hod exis s o es ima ing e iciency. Howe e , wi h he
suppo o obus s a is ical es s, dis ibu ional analysis, e iciency
ankings, and co ela ion assessmen s, compa isons be ween nonpa a-
me ic and pa ame ic me hods can gene a e aluable insigh s and
highligh no able di e ences in ou comes.
In empi ical applica ions, Mi anda e al. (2012) epo ed minimal
di e ences be ween DEA and SFA, no ing ha bo h app oaches iden i-
ied he same op and bo om 20 highe educa ion ins i u ions (HEIs) in
e ms o e iciency. In con as , G alka e al. (2019) ound di e gen
ou comes. Thei DEA-based esul s p oduced consis en e iciency
le els, while SFA es ima es showed mino a ia ions o e ime and
weake co ela ions. Simila ly, Le i e al. (2022) emphasized he
di e ing conclusions o he wo me hods: DEA sugges ed ha special-
ized HEIs a e mo e e icien , whe eas SFA indica ed ha gene alis HEIs
demons a e g ea e e iciency.
Based on he li e a u e e iewed, he e is a compelling need o
u he empi ical e idence on he e iciency measu emen o unde -
explo ed dimensions o uni e si y ac i i y, such as esea ch, coope a-
ion, and inno a ion. This need is pa icula ly e iden in he I alian
con ex . The ex en o which I alian uni e si ies ope a e e icien-
ly—especially in e ms o esea ch ou pu , collabo a i e ne wo ks, and
inno a ion—has no been ho oughly in es iga ed in he academic
li e a u e. Mo eo e , addi ional inqui y is equi ed o de e mine
whe he a single me hodological amewo k can e ec i ely es ima e and
e alua e he e iciency o HEIs. In esponse o hese gaps, his s udy
in es iga es he pe o mance o I alian uni e si ies ac oss esea ch,
coope a ion, and inno a ion by applying bo h DEA and SFA and sys-
ema ically compa ing he esul s. This in ol es examining he heo-
e ical ounda ions, p ac ical applica ions, ad an ages, and limi a ions
o each app oach. Guided by he me hodology p esen ed in he
ollowing sec ion, he s udy ocuses on I alian public uni e si ies—an
unde -examined se ing ha dese es close a en ion due o ongoing
s uc u al ans o ma ions, dependence on public unding, and pe sis-
en egional dispa i ies.
Me hodology
This s udy applies bo h nonpa ame ic and pa ame ic me hodolo-
gies: DEA and SFA, espec i ely. The wo app oaches a e hen compa ed
o e alua e he e iciency le els o I alian public uni e si ies wi h espec
o esea ch, coope a ion, and inno a ion.
Da a en elopmen analysis
This s udy employs DEA, a nonpa ame ic echnique ini ially in o-
duced by Fa ell (1957), la e ex ended by Cha nes, Coope , and Rhodes
(1978), and u he e ined by F¨
a e e al. (1994). DEA measu es e i-
ciency by assessing how inpu s a e ans o med in o ou pu s. In his
con ex , he uni s o analysis— e e ed o as DMUs—a e I alian public
uni e si ies. E iciency is ep esen ed by a on ie , and he dis ance o
each DMU om his on ie e lec s i s le el o ine iciency.
The analysis in es iga es he inpu –ou pu ans o ma ion p ocess
h ough h ee sepa a e DEA models, each co esponding o a dis inc
dimension o uni e si y pe o mance: esea ch, coope a ion, and inno-
a ion. All h ee models use he same inpu —cu en expendi u es—-
while he ou pu s a y o e lec he di e en pe o mance dimensions
being assessed.
All analyses ollow an ou pu -o ien ed app oach, whe e ou pu s a e
maximized while inpu s a e held cons an . This o ien a ion was chosen
due o he na u e o he highe educa ion sec o , whe e inpu s a e
ypically exogenously de e mined and beyond he di ec con ol o in-
s i u ions (Agasis i & Johnes, 2015). The analysis is conduc ed unde he
assump ion o VRS, which is p e e ed o e he CRS assump ion when
assessing uni e si y e iciency. VRS be e accommoda es
non-p opo ional ela ionships be ween inpu s and ou pu s (Dipie o &
De Wi e, 2024), a c i ical conside a ion in a sec o a ec ed by ech-
nological ad ances, impe ec compe i ion, egula o y changes, budge
cons ain s, and en y- ela ed condi ions. VRS is especially ele an in
cap u ing he complexi y and he e ogenei y o educa ional ins i u ions
(Dipie o and De Wi e, 2024).
The esul s o he DEA models a e exp essed as e iciency sco es
anging om 0 o 1. A sco e o 1 indica es ha a uni e si y is e icien . A
sco e below 1 signi ies ela i e ine iciency. A sco e o 0 deno es com-
ple e ine iciency.
The DEA analysis was conduc ed using R so wa e ia he DeaR
package. The me hodological ounda ion is based on he linea p o-
g amming model de eloped by Cha nes e al. (1978).
Max 0=∑m
i=1 ixi0
∑s
=1u y 0
subjec o :∑m
i=1 ixij
∑s
=1u y j
≥1;
whe e j=1,…,n, i,u ≥0,and whe e and u ep esen he weigh
o inpu and ou pu a iables, espec i ely.
S ochas ic on ie analysis
The second me hodological app oach employed in his s udy is SFA,
a pa ame ic echnique in oduced by Aigne e al. (1977). I s p ima y
ad an age lies in he abili y o measu e e iciency while simul aneously
accoun ing o s a is ical noise.
Pa ame ic models assume ha he e iciency on ie can be iden-
i ied using a unc ional o m ha closely i s he obse ed da a se . This
is exp essed as:
y= (x;β) +
ε
whe e y ep esen s he ou pu , x is he ec o o inpu s, β is he ec o o
pa ame e s, and
ε
is he composi e e o e m.
A key ea u e o SFA is i s decomposi ion o he e o e m (
ε
) in o
wo componen s: a symme ic e m ep esen ing andom noise ( ) and a
one-sided e m cap u ing ine iciency (u). These wo componen s a e
assumed o be s a is ically independen . The decomposi ion is exp essed
as:
ε
i= i−ui
whe e
ᵢ
deno es he andom noise, and u
ᵢ
ep esen s he ine iciency
componen .
In his s udy, he s ochas ic on ie is modeled using a Cobb–Douglas
unc ional o m combined wi h a hal -no mal dis ibu ion o he in-
e iciency e m. The model is speci ied as a cos on ie and ollows his
o mula ion:
y=
α
+xiβ+ i+uii=1,…,N;
whe e
u∼iid N+(0,
σ
2
u)
F om his pe spec i e, selec ing an ine iciency dis ibu ion is a
c i ical assump ion ha mus be made a he ou se o he analysis. I is
no a choice based on model cha ac e is ics, and no s a is ical es s a e
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
3
a ailable o alida e he dis ibu ional assump ions.
To calcula e echnical ine iciency, his s udy adop s he BC app oach
de eloped by Ba ese and Coelli (1988), which es ima es echnical e -
iciency as he expec ed alue o exp( − u)| e,E(exp( − u)|e).
Th ee SFA models a e es ima ed. All models use he same
inpu —cu en expendi u es—while he ou pu s a y depending on he
dimension being assessed. The i s model e alua es esea ch e iciency
using he numbe o publica ions as ou pu . The second ocuses on
coope a ion, wi h he numbe o a ilia ions as ou pu . The hi d assesses
inno a ion, using he numbe o pa en s as he ou pu a iable.
The e iciency sco es gene a ed om hese models ange om 0 o 1.
A sco e o 1 indica es ha a uni e si y is e icien . A sco e below 1
signi ies ela i e ine iciency. A sco e o 0 e lec s comple e ine iciency.
This pa o he analysis was conduc ed using he on ie package in R.
Da a en elopmen analysis e sus s ochas ic on ie analysis
P e ious esea ch highligh s a subs an ial body o li e a u e on DEA,
whe eas conside ably ewe s udies ha e applied SFA o employed a
combined app oach (Rella and Vi olla, 2024). In ligh o his imbalance,
he p esen s udy aims o compa e he wo me hodologies o explo e
hei espec i e ad an ages and limi a ions. The key cha ac e is ics,
s eng hs, and d awbacks o DEA and SFA a e summa ized in Table 1.
Ma e ials
Sample
The I alian uni e si y sys em comp ises 97 uni e si ies, including 67
s a e uni e si ies, 19 legally ecognized non-s a e uni e si ies, and 11
online uni e si ies. This s udy ocuses exclusi ely on s a e uni e si ies,
selec ed based on mul iple c i e ia o ensu e da a se obus ness and
compa abili y. S a e uni e si ies we e chosen due o hei adhe ence o
anspa ency and accoun abili y s anda ds, which suppo accu a e e -
iciency assessmen s. A o al o 62 s a e uni e si ies we e included in he
analysis. Ins i u ions issuing equi alen quali ica ions o uni e si y de-
g ees and hose ha did no disclose inancial in o ma ion o he
e e ence yea we e excluded. This selec ion ensu es ha he
uni e si ies included o e consis en and comp ehensi e inpu -ou pu
da a, as sou ced and desc ibed in he ollowing sec ion. The use o
publicly a ailable seconda y da a enhances he eliabili y and compa-
abili y o he analysis while minimizing bias and s eng hening ans-
pa ency and alidi y. The selec ion p ocess also ensu es he sample is
ep esen a i e o he I alian public highe educa ion sys em. The anal-
ysis was conduc ed o he yea 2021 o compa e wo me hodological
app oaches—nonpa ame ic and pa ame ic— o e alua ing he pe -
o mance o I alian uni e si ies in e ms o esea ch, coope a ion, and
inno a ion.
Va iables
The a iables used in his s udy a e di ided in o inpu s and ou pu s
and a e applied consis en ly ac oss bo h DEA and SFA. The selec ion o
a iables is g ounded in he ele an li e a u e. To e alua e he h ee
co e dimensions— esea ch, collabo a ion, and inno a ion—di e en
inpu -ou pu combina ions a e compu ed. No ably, he inpu emains
he same ac oss all h ee con igu a ions, while he ou pu s a y
depending on he ac i i y being assessed.
The selec ed inpu is cu en expendi u es, de ined as he o al o
se e al ope a ing cos s exp essed in eu os (
€
). This choice is suppo ed
by he li e a u e, pa icula ly o i s impo ance in e alua ing uni e si y
e iciency (Rella e al., 2024). Acco ding o Thanassoulis e al. (2011),
cu en expendi u es a e especially ele an o public manage s and
policymake s and a e p e e able o mo e g anula ca ego ies o spending
due o hei comp ehensi e na u e and pa simony in a iable use, an
impo an conside a ion when applying DEA and SFA me hodologies.
Cu en expendi u es encompass cos s ela ed o all h ee e alua ed ac-
i i ies and a e hus compa ible wi h he ou pu s desc ibed in he
ollowing sec ion. This a iable was ex ac ed om he income s a e-
men s o each uni e si y, as published on he websi es o he 62 in-
s i u ions included in he analysis. Speci ically, wi hin each balance
shee , cu en expendi u es a e epo ed unde Sec ion B, I em IX, and
a e exp essed in eu os. The b eakdown o cos componen s comp ising
cu en expendi u es is de ailed in Table 2.
The ou pu a iables used in his analysis se e as p oxies o he
esea ch, coope a ion, and inno a ion missions o uni e si ies. Speci -
ically, he numbe o publica ions is used o ep esen esea ch ac i i y,
he numbe o a ilia ions e lec s coope a ion, and he numbe o pa -
en s se es as an indica o o inno a ion. These da a e e o he yea
2021 and we e e ie ed om he Scopus da abase. Each uni e si y was
indi idually que ied o ob ain he numbe o publica ions, a ilia ions,
and pa en s speci ic o he yea unde analysis.
Desc ip i e s a is ics
The inpu and ou pu a iables a e desc ibed using he minimum
(Min), i s qua ile (Q1), mean (Mean), median (Median), hi d qua ile
(Q3), maximum (Max), and s anda d de ia ion (SD). Table 3 p esen s
he desc ip i e s a is ics o he I alian uni e si ies included in he 2021
Table 1
Ad an ages and limi a ions o DEA and SFA.
DEA SFA
Ad an ages - No speci ica ion on he
unc ional o m o he
adop ed echnology
- Es ima ing e iciency in
mul i-ou pu as well as single-
ou pu se ings
- Two-s age eg ession o
ac o s explaining
ine iciency wi h he
possibili y o co ec ing
collinea i y p oblems
- Dis inc ion o he andom e o
om he e o ela ed o he
e iciency a ia ion o he
s udied en i y
- Reduced ine iciencies may
ha e s a is ical p ope ies
- No e y sensi i e o ou lie s
Limi a ions - Sensi i e o ou lie s: La ge
e o s in measu emen and/
o a iables can a ec
measu es o ine iciency
- Reduced ine iciencies ha e
no s a is ical p ope ies
- Rela i e measu e o
e iciency by compa ison o
all o he uni s o he e e ence
aken as “bes p ac ice”: isk
o o e - o unde -es ima ion
ela ed o he cha ac e is ics
o he e e ence
- Sensi i e o sample size: isk o
misspeci ica ion due o small
sample size
- Need o ep esen he
echnology by a pa icula
pa ame ic o m
- Mode a ely sui able o mul i-
ou pu s
- Simul aneous eg ession ( i s
s age) o he a iables
explaining ine iciency: isk o
collinea i y be ween ou pu s
and inpu s and he
en i onmen a iables
in eg a ed in o he p oduc ion
unc ion
Table 2
Componen s o cu en expendi u es.
IX) CURRENT EXPENDITURES
1Cos s o suppo o s uden s
2Cos s o he igh o educa ion
3Cos s o edi o ial ac i i ies
4T ans e s o he pa ne o coo dina ed p ojec s
5Pu chase o consumables o labo a o ies
6Change in in en o ies o consumables o labo a o ies
7Pu chase o books, pe iodicals, and bibliog aphic ma e ial
8Pu chase o se ices and echnical collabo a ion managemen
9Pu chase o o he ma e ials
10 Change in in en o ies o ma e ials
11 Cos s o using hi d-pa y asse s
12 O he cos s
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
4

analysis. The inpu a iable, cu en expendi u es, displays subs an ial
a ia ion ac oss uni e si ies. The maximum alue is conside ably highe
han he minimum, indica ing he p esence o small ins i u ions wi h
limi ed inancial esou ces. Speci ically, he minimum alue is 3.5
million eu os and e e s o Is i u o Uni e si a io di S udi Supe io i di Pa ia,
while he maximum alue is 234 million eu os and cap u es Alma Ma e
S udio um - Uni e si `
a’ di Bologna. The dis ibu ion o cu en expendi u es
is highly asymme ic, as e lec ed by he di e ence be ween he mean
and qua ile alues. A la ge p opo ion o alues lie on he igh side o
he dis ibu ion, sugges ing conside able dispa i ies in inancial esou ce
alloca ion. Focusing on ou pu a iables (i.e., publica ions, a ilia ions,
and pa en s), e eals he g ea disc epancy be ween minimum and
maximum alues, as well as among he qua ile alues. F om his
pe spec i e, he minimum numbe o publica ions is 41 and e e s o
Uni e si `
a’ degli S udi di Napoli – L’ O ien ale, whe eas he maximum
numbe o publica ions is 11,363 and e e s o Uni e si `
a’ degli S udi di
Roma - La Sapienza. The mean alue is 2758.6. These esul s demons a e
a e y low numbe o publica ions p oduced by I alian uni e si ies
compa ed o he global a e age, and i e lec s he low in es men s in
esea ch and de elopmen in I aly. Rega ding a ilia ions, he maximum
alue is 844 and e e s o Alma Ma e S udio um - Uni e si `
a’ di Bologna,
whe eas he minimum alue is 0, co esponding o 28 o he 62 sampled
uni e si ies o almos hal o he sample. Finally, o pa en s, he
maximum is 953 and e e s o Uni e si `
a degli S udi di Pa ma, whe eas he
minimum alue is 0 and co esponds o only 10 o he sampled I alian
uni e si ies.
Resul s
E iciency acco ding o da a en elopmen analysis
In his s udy, h ee DEA models we e applied o measu e he e i-
ciency o I alian public uni e si ies in he a eas o esea ch, coope a ion,
and inno a ion. Table 4 p esen s desc ip i e s a is ics o he e iciency
sco es, including he minimum (Min), i s qua ile (Q1), median, mean,
hi d qua ile (Q3), maximum (Max), SD, skewness, and ku osis. The
i s DEA model e alua es esea ch e iciency, using one inpu —cu en
expendi u es—and one ou pu — he numbe o publica ions. The
a e age e iciency sco e o his model is ela i ely high (0.7618),
al hough he e is oom o imp o emen . The minimum sco e, 0.0592,
was eco ded by Uni e si `
a degli S udi “G. d’Annunzio,” whe eas he
maximum sco e o 1.000 was achie ed by 11 o he 62 sampled
uni e si ies. The second DEA model assesses coope a ion e iciency, again
using cu en expendi u es as he inpu and he numbe o a ilia ions as
he ou pu . The mean e iciency sco e is 0.8006, which is also ela i ely
high. The minimum alue, 0.0592, co esponds o Uni e si `
a degli S udi
di Napoli “L’O ien ale,” whe eas he maximum sco e o 1.000 was
achie ed by 14 uni e si ies in he sample. The hi d DEA model in-
es iga es inno a ion e iciency, using cu en expendi u es as he inpu
and he numbe o pa en s as he ou pu . The mean e iciency sco e o
his model is 0.3086, indica ing ela i ely low pe o mance in inno a-
ion. Ele en uni e si ies sco ed he minimum alue o 0.000, whe eas
se en uni e si ies achie ed he maximum e iciency sco e o 1.000.
The h ee DEA models e eal no able di e ences in e iciency ou -
comes. The mos e iden disc epancy lies in he mean e iciency sco es.
The i s wo models— ocused on esea ch and coope a ion—demon-
s a e ela i ely high mean sco es, wi h coope a ion appea ing o be
sligh ly mo e e icien han esea ch. In con as , inno a ion e iciency
displays a signi ican ly lowe mean sco e. This esul can be in e p e ed
om bo h nega i e and posi i e pe spec i es. F om a nega i e s and-
poin , he numbe o pa en s p oduced by I alian uni e si ies is small.
Digi al pla o ms, inno a i e ools, and sma echnologies emain un-
de de eloped and unde u ilized. This e lec s a lack o echnical
expe ise and he pe sis en challenge o ansla ing scien i ic indings
and in ellec ual p ope y in o p ac ical inno a ion wi hin he I alian
uni e si y sys em. On a mo e posi i e no e, he low inno a ion sco e
may se e as a s imulus o imp o emen . As I aly’s educa ion sec o
con inues o g ow, p og ess in inno a ion will equi e dedica ion, long-
e m planning, and consis en suppo . Encou agingly, he ecen allo-
ca ion o public unding sugges s a commi men o os e ing mo e
inno a i e p ac ices.
The addi ional s a is ical measu es epo ed in Table 4 u he sup-
po hese disc epancies ac oss he h ee dimensions o e iciency
assessed in I alian public uni e si ies.
E iciency acco ding o s ochas ic on ie analysis
To measu e e iciency wi hin I alian public uni e si ies in e ms o
esea ch, coope a ion, and inno a ion, h ee SFA models we e also
p oduced. Table 4 p o ides desc ip i e s a is ics o he e iciency sco es
o he SFA, including he minimum (Min), i s qua ile (Q1), median,
mean, hi d qua ile (Q3), maximum (Max), SD, skewness, and ku osis.
The i s SFA model aims o analyze he esea ch e iciency le el and
conside s one inpu (cu en expendi u es) and one ou pu (numbe o
Table 3
Desc ip i e s a is ics o inpu and ou pu s.
Va iables Min Q1 Mean Median Q3 Max SD
Inpu s       
Cu en expendi u e 3529,781 18,958,069 63,922,608 51,882,888 81,945,250 233,997,913 59,151,095
Ou pu s       
No. Publica ions 41.0 837.2 2758.6 2026.0 3636.8 11,363.0 2583.3
No. A ilia ions 0.0 0.10 75.88 1.0 105.25 844.0 147.9
No. Pa en s 0.0 2.0 199.4 22.0 310.5 953.0 261.9
Table 4
Desc ip i e s a is ics o he e iciency sco es ob ained om DEA and SFA.
DEA SFA
Resea ch Coope a ion Inno a ion Resea ch Coope a ion Inno a ion
Min 0.0592 0.0592 0.0000 0.0813 0.5008 0.1623
1s Qu. 0.6882 0.7539 0.0112 0.6091 0.6383 0.4856
Median 0.7619 0.8357 0.3086 0.7345 0.7123 0.6261
Mean 0.7618 0.8006 0.3086 0.6950 0.7022 0.5929
3 d Qu. 0.8644 0.9907 0.4731 0.7851 0.7561 0.7057
Max 1.0000 1.0000 1.0000 0.9444 0.8373 0.8316
SD 0.1998 0.2067 0.3465 0.1678 0.0758 0.1543
Skewness -1.3139 -1.5578 0.9208 -1.2904 -0.6061 -0.6524
Ku osis 5.2287 5.5950 2.5941 5.2623 2.7666 2.9094
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
5
publica ions). The mean e iciency sco e is mode a ely high (0.6950),
al hough he e is signi ican oom o imp o emen . The minimum alue
is 0.0813, eco ded by Uni e si `
a degli S udi di Napoli – “L’O ien ale,”
whe eas he maximum alue is 0.9444, which e e s o Uni e si `
a degli
S udi del Molise.
The second SFA model in es iga es he coope a ion e iciency le el
and conside s one inpu (cu en expendi u es) and one ou pu (numbe
o a ilia ions). The mean sco e o his model is 0.7022, which is also
mode a ely high. The minimum alue is 0.5008 and e e s o Poli ecnico
di Milano, whe eas he maximum alue is 0.8373 and e e s o Is i u o
Uni e si a io di S udi Supe io i di Pa ia.
The hi d SFA model aims o explo e he inno a ion e iciency le el
and conside s one inpu (cu en expendi u es) and one ou pu (numbe
o pa en s). The mean sco e o his model is 0.5929, which ep esen s a
medium alue. The minimum alue is 0.1623 and e e s o IMT Al i
S udi di Lucca, whe eas he maximum alue is 0.8316 and e e s o
Uni e si `
a degli S udi della Campania - ’Luigi Van i elli’ and Uni e si `
a degli
S udi di Mace a a.
The h ee SFA models also p o ide esul s ha unde line some dis-
c epancies be ween he di e en dimensions o e iciency wi hin I alian
uni e si ies. Howe e , hese di e ences a e no as no iceable as in he
DEA models, e en hough hey ollow he same end. The a e age
alues o hese h ee models demons a e a mode a ely low sco e ela ed
o inno a ion e iciency. Mo eo e , he e iciency alues om he SFA
models a e sligh ly mo e balanced han hose om he DEA models. A
possible explana ion is ha nonpa ame ic sco es en i ely a ibu e de-
ia ion om he on ie o ine iciency, whe eas he pa ame ic SFA
allows o a andom e o e m (Cummins and Weiss, 2000; Sua ez-
Fe nandez e al., 2021). In his case, he o he s a is ical measu es in
Table 4 highligh he same pa e n o disc epancy ac oss he h ee as-
pec s o I alian uni e si ies’ e iciency.
Compa ing e iciency acco ding o da a en elopmen analysis and
s ochas ic on ie analysis
To compa e he DEA and he SFA models, se e al aspec s can be
u he explo ed. Speci ically, desc ip i e s a is ics and co ela ion an-
alyses a e ca ied ou o iden i y clea signals o di e ence be ween he
esul s p o ided by he wo me hodological app oaches.
S a ing wi h he desc ip i e s a is ics, i is impo an o highligh he
sub le bu meaning ul disc epancies among he models. The ac ha
nonpa ame ic sco es en i ely a ibu e de ia ion om he on ie o
ine iciency, whe eas he pa ame ic SFA allows o a andom e o
e m, sugges s ha be o e d awing inal conclusions, i is essen ial o
conside whe he andomness (i.e., DEA o SFA) has been aken in o
accoun when es ima ing e iciency. Mo eo e , skewness and ku osis
a e also examined. The skewness alues a e nega i e o all models
excep he hi d DEA model. These alues indica e ha he densi y
unc ions o he e iciency sco es ha e longe ails on he le side and
ha he mass o he dis ibu ion is concen a ed on he igh . In con as ,
he ku osis alues a e posi i e and high o all DEA and SFA models,
sugges ing ha he da a a e p edominan ly clus e ed a ound he a e age
alue.
Addi ional obse a ions om he desc ip i e s a is ics in ol e he
qua ile alues. These esul s a e clea ly isible in he boxplo (see
Fig. 1), which o e s s ong suppo o a isual compa ison o he ange
and dis ibu ion o he e iciency sco es.
Fou o he six models p esen a median be ween 0.6 and 0.8
(Resea ch_DEA, Resea ch_SFA, Coope a ion_SFA, and Inno a ion_SFA),
whe eas he emaining wo models (Coope a ion_DEA and Inno a-
ion_DEA) show median alues o 0.9 and 0.3, espec i ely. Fu he -
mo e, in he i s wo DEA models, bo h he median and he i s qua ile
a e highe han in he o he models, and in he Coope a ion_DEA model,
he hi d qua ile is posi ioned highe han he o he s. Mo eo e , some
ou lie s a e isible, pa icula ly a ec ing wo o he DEA models
(Resea ch_DEA and Coope a ion_DEA) and one o he SFA models
(Resea ch_SFA).
Fig. 2 p esen s he sca e plo s, which demons a e he ela ionship
be ween he di e en nonpa ame ic and pa ame ic models. In his
case, he au ho s analyzed he phenomenon by pai ing he e iciency
le els o he same dimension o I alian public uni e si ies ob ained using
di e en me hodological app oaches.
The dis ibu ions appea di e en , which is also due o he dis inc
se s o assump ions unde lying he wo me hodological app oaches.
F om his pe spec i e, he i s sca e plo , which in es iga es he
esea ch aspec (Resea ch_DEA and Resea ch_SFA), does no show a
subs an ial disc epancy be ween he wo me hods. Simila ly, he second
sca e plo , which explo es he coope a ion aspec (Coope a ion_DEA
and Coope a ion_SFA), p esen s mo e o less he same end in he e -
iciency sco es. In con as , he las sca e plo , which conside s he
inno a ion aspec (Inno a ion_DEA and Inno a ion_SFA), demons a es
a conside able di e ence be ween he wo me hodological app oaches.
Despi e hese disc epancies, he choice o me hodology should no
de e mine he esul s. The indings om bo h a e conside ed alid and
Fig. 1. Boxplo s o compa ison o he e iciency sco es ob ained om DEA and SFA models.
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
6
Fig. 2. Sca e plo s DEA and SFA models by coupling he same e iciency dimension.
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
7
ele an , including o policymake s and public manage s o apply po-
li ical and manage ial policies in he es ima ion o he e iciency o
I alian public uni e si ies.
Fu he mo e, we ein o ce wha has al eady been obse ed by
p o iding ankings o he e icien I alian public uni e si ies, acco ding
o he di e en combina ions o me hodologies and dimensions o e i-
ciency. Table 5 p o ides a op 5 anking o e icien I alian public uni-
e si ies o gi e a snapsho o he indings.
In a sample o 62 uni e si ies, 21 ins i u ions appea wi hin he op
i e posi ions o he e iciency ankings. This esul sugges s ha ,
depending on he di e en e iciency dimensions and me hodological
app oaches, uni e si ies occupy a ying posi ions in he e iciency
measu emen ankings. Howe e , a no ewo hy pa e n conce ns he
Is i u o Uni e si a io degli S udi Supe io i di Pa ia, which anks in he
op posi ion ou imes—speci ically in Resea ch_DEA, Coope a-
ion_DEA, Inno a ion_DEA, and Coope a ion_SFA—and once in he i h
posi ion. Mo eo e , Uni e si `
a degli S udi del Molise appea s h ee
imes, including wo i s -place ankings and one second-place anking.
IMT Al i S udi di Lucca, Uni e si `
a degli S udi di Geno a, and Uni e si `
a
degli S udi di Milano – Bicocca each appea wice among he op 5 mos
e icien uni e si ies. In o he cases, a high a iance is obse ed in he
e iciency ankings o he uni e si ies.
Following he e iciency ankings, a co ela ion analysis is now
p esen ed o compa e he Pea son and Spea man indexes. Addi ionally,
Fig. 3 and Table 6 g aphically and nume ically p esen he mos
commonly used me hod o compa e DEA and SFA model esul s
(Agasis i & Dal Bianco, 2006), namely co ela ion analysis.
In PANEL A o Table 6, he Pea son co ela ion shows a posi i e
co ela ion be ween Inno a ion_DEA and Resea ch_DEA (0.336) a he
1% signi icance le el and a s ong posi i e co ela ion be ween
Resea ch_SFA and Coope a ion_DEA (0.827) a he 1% signi icance
le el. In con as , he e is a nega i e co ela ion be ween Inno a-
ion_SFA and Inno a ion_DEA (-0.325) a he 5% signi icance le el, and
be ween Inno a ion_SFA and Coope a ion_SFA (-0.352) a he 1% sig-
ni icance le el. In PANEL B o Table 6, he Spea man co ela ion p o-
ides simila esul s as he Pea son co ela ion, hough a di e en
signi icance le els. In his case, he e is a posi i e co ela ion be ween
Inno a ion_DEA and Resea ch_DEA (0.315) a he 5% signi icance le el
and a posi i e co ela ion be ween Resea ch_SFA and Coope a ion_DEA
(0.652) a he 1% signi icance le el. A nega i e co ela ion is obse ed
be ween Inno a ion_SFA and Inno a ion_DEA (-0.342) a he 1% sig-
ni icance le el and be ween Inno a ion_SFA and Coope a ion_SFA
(-0.313) a he 5% signi icance le el. The indings ob ained om hese
wo co ela ion coe icien s a e e y simila , highligh ing he impo -
ance o conside ing mo e han one me hodological app oach when
es ima ing e iciency.
Discussion and conclusions
In he e ol ing landscape o uni e si ies, hei g ea e scope o ac-
i i y places uni e si ies a he o e on o socie al ad ancemen
(Godonoga & Spo n, 2023). This s udy applied nonpa ame ic and
pa ame ic e iciency app oaches—DEA and SFA, espec i ely— o es-
ima e he e iciency le els o I alian public uni e si ies in e ms o
esea ch, coope a ion, and inno a ion. The no el y o his s udy lies in
in es iga ing he di e en dimensions o e iciency wi hin I alian public
uni e si ies, ocusing on he ans o ma ion p ocess cu en ly impac ing
he educa ion sec o .
Speci ically, he s udy analyzed 62 I alian public uni e si ies o he
yea 2021, wi h pa icula a en ion o applying di e en me hodologies
o compu e and es ima e e iciency le els using ele an ou pu s such as
publica ions, a ilia ions, and pa en s. The me hodological app oaches
used we e DEA and SFA. Mo e speci ically, bo h we e applied o es i-
ma e he h ee e iciency dimensions o I alian public uni e si ies:
esea ch, coope a ion, and inno a ion. DEA models we e employed
using a VRS model and an ou pu -o ien ed con igu a ion, ollowing he
exis ing scien i ic li e a u e. SFA models we e compu ed using he Cobb-
Douglas unc ional o m and hal -no mal dis ibu ion.
Due o hei me hodological ea u es, DEA and SFA a e widely used
by schola s o es ima e e iciency wi hin he educa ion sec o . Howe e ,
no empi ical s udy o da e has explo ed he e iciency o esea ch,
coope a ion, and inno a ion using bo h app oaches in he con ex o
I alian public uni e si ies. In ligh o his, he p esen s udy o e s wo
main con ibu ions: he applica ion, implemen a ion, and compa ison o
wo di e en e iciency app oaches—nonpa ame ic and pa a-
me ic—and he explo a ion o he e iciency le els o I alian public
uni e si ies in e ms o esea ch, coope a ion and inno a ion.
The indings e eal he easibili y and accu acy o bo h me hods in
es ima ing he e iciency le els o I alian public uni e si ies. Ce ainly,
each me hod includes speci ic cha ac e is ics, wi h s eng hs and
weaknesses, as ecognized in he scien i ic li e a u e. Ne e heless, he
di e ences in assump ions and me hodological ea u es p o ide mean-
ing ul insigh s o he es ima ion and compa ison o e iciency esul s.
The compa ison be ween DEA and SFA e ealed no able disc epancies,
con i ming ha no single me hodology is uni e sally supe io . Ra he ,
he di e en app oaches a e capable o cap u ing di e se aspec s in he
es ima ion and e alua ion o e iciency, depending on hei unde lying
backg ound, ounda ions, ad an ages, and limi a ions. These esul s
suppo he pe spec i es o Baue e al. (1998), McMillan and Chan
(2006), and Mi anda e al. (2012).
F om he e iciency sco es o he six compu ed models, i was
possible o ex ac aluable in o ma ion in e ms o di e ing esul s and
pe spec i es. O e all, he e is subs an ial oom o imp o emen ,
pa icula ly wi hin he inno a ion dimension. Subsequen ly, he uni-
e si y ankings we e de eloped based on he e iciency sco es o
examine whe he he same uni e si ies emain e icien ac oss he
di e en me hodologies and e iciency aspec s. Finally, Pea son and
Table 5
Top 5 anking o e icien I alian public uni e si ies.
RANK DMUs
Resea ch_DEA
1Is i u o Uni e si a io degli S udi Supe io i di Pa ia
2Uni e si `
a degli S udi del Piemon e O ien ale
3Uni e si `
a degli S udi Roma – La Sapienza
4Scuola Supe io e San ’Anna
5Uni e si `
a degli S udi di Siena
Coope a ion_DEA
1Is i u o Uni e si a io degli S udi Supe io i di Pa ia
2Uni e si `
a degli S udi del Molise
3Uni e si `
a degli S udi di Sale no
4Uni e si `
a degli S udi di Geno a
5Uni e si `
a della Calab ia
Inno a ion_DEA
1Is i u o Uni e si a io degli S udi Supe io i di Pa ia
2Uni e si `
a degli S udi del Molise
3Uni e si `
a degli S udi di Geno a
4Im Al i S udi di Lucca
5Uni e si `
a degli S udi di Milano - Bicocca
Resea ch_SFA
1Uni e si `
a degli S udi del Molise
2Uni e si `
a degli S udi del Sannio
3Uni e si `
a degli S udi della Basilica a
4Uni e si `
a degli S udi dell’Aquila
5Is i u o Uni e si a io degli S udi Supe io i di Pa ia
Coope a ion_SFA
1Is i u o Uni e si a io degli S udi Supe io i di Pa ia
2Uni e si `
a degli S udi di Milano - Bicocca
3Im Al i S udi di Lucca
4Uni e si `
a degli S udi di Sassa i
5Uni e si `
a degli S udi di Modena e Reggio Emilia
Inno a ion_SFA
1Uni e si `
a degli S udi di Mace a a
2Uni e si `
a degli S udi della Campania – ‘Luigi Van i elli’
3Alma Ma e S udio um – Uni e si `
a di Bologna
4Uni e si `
a di Pisa
5Poli ecnico di Milano
A. Rella and A.R. Dipie o
Jou nal o Inno a ion & Knowledge 10 (2025) 100724
8