You are on page 1of 8

ARTICLE IN PRESS

Journal of Air Transport Management 13 (2007) 184–191


www.elsevier.com/locate/jairtraman

Performance evaluation of Italian airports: A data envelopment analysis


Carlos Pestana Barrosa, Peter U.C. Diekeb,
a
Instituto Superior de Economia e Gestão, Technical University of Lisbon, Rua Miguel Lupi, 20, 1249-078 Lisbon, Portugal
b
School of Recreation, Health, and Tourism, College of Education and Human Development, George Mason University,10900 University Boulevard,
Manassas, VA 20110, USA

Abstract

This paper addresses empirically financial and operational performance of Italian airports using data envelopment analysis (DEA)
methodology. With panel data for 2001–2003, the study tests variable relationships—the relative roles of dimension, managerial status
and workload unit—to measure the proximity of the airports to the frontiers of best practices. The implications of the research for
managerial purposes are then drawn to validate prior findings of the same relationships in other geographical contexts.
r 2007 Elsevier Ltd. All rights reserved.

Keywords: Italian airports; Data envelopment analysis; Economic efficiency

1. Introduction data for 3 years (2001–2003), to examine the relative roles


of dimension, managerial status and workload unit (WLU)
The importance of efficiency of airports has been of in determining the proximity of airports to the frontier of
interest in contemporary economics, given airports’ in- best practices.
creasingly strategic importance in the movement of people DEA methodology can enable management to bench-
and cargo in the globalized world (Adler and Berechman, mark the best-practice decision-making unit (DMU), here
2001). Moreover, the increased airline competition brought the Italian airport authority, by calculating scores denoting
about by deregulation and liberalization has heightened their efficiency within a linear programming procedure.
this recognition and placed airports in a much more Furthermore, DEA provides estimates of the potential
competitive environment. As a result, airports are now improvement that can be made by inefficient DMU.
under pressure to improve their efficiency relative to
competitors. De Borger et al. (2002) opine that bench- 2. Methodology
marking analysis is one of the ways to drive airports
towards the frontier of best practices. Whilst there is extensive literature on benchmarking
This paper analyzes, by means of four data envelopment applied to a diverse range of activities, there are few
analysis (DEA) models, the financial and operational looking at European airports. Table 1 highlights the main
performance of Italian airports. These model types are: cases.
the DEA-CCR (or Charnes, Cooper and Rhodes) model There is a tradition of analyzing airports by separating
(Charnes et al., 1978); the DEA-BCC (or Banker, Charnes activities into terminals and movements. Several papers
and Cooper) model (Banker et al., 1984); the Cross- compare the DEA model with the frontier model, while
Efficiency DEA model (Sexton et al., 1986; Doyle and others mix principal component analysis with a DEA
Green, 1994); and the Super-Efficiency DEA model model but there has been use of a stochastic frontier
(Andersen and Petersen, 1993). The study utilizes panel allowing contextual variables.
Following Farrell (1957), Charnes et al. (1978) first
Corresponding author. introduced the term DEA to describe a mathematical
E-mail address: pdieke@gmu.edu (P.U.C. Dieke). programming approach to the construction of production

0969-6997/$ - see front matter r 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jairtraman.2007.03.001
ARTICLE IN PRESS
C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191 185

Table 1
Research into Airport Efficiency

Papers Method Units Inputs Outputs

Gillen and Lall DEA-BCC model and a 21 US airports (i) Terminal services model (i)Terminal services model
(1997) Tobit model (1) Number of runways (1) Number of passengers
(2) Number of gates (2) Pounds of cargo
(3) Terminal area
(4) Number of baggage collection belts
(5) Number of public parking spots

(ii) Movement model (ii) Movements model


(1) Airport area (1) Air carrier movements
(2) Number of runways (2) Commuter
(3) Runway area movements
(4) Number of employees

Parker (1999) DEA-BCC and CCR 32 UK regulated airports, (1) Number of employees (1) Turnover
models 1979/1980 to 1995/1996. (2) Capital input estimated as an (2) Passengers handled
In a second model, uses 22 annual rental based on a real rate of (3) Cargo and mail
airports from 1988/89 to return of 8% each year applied to business
1996/97 net capital stock
(3) Other inputs defined as the residual
of total operating costs

Murillo- DEA-Malmquist 33 Spanish civil airports, (1) Number of workers Number of passengers
Melchor (1999) 1992 to 1994 (2) Accumulated capital stock proxied
by amortization
(3) Intermediate expenses

Gillen and Lall DEA-Malmquist 22 major US airports, (i) Terminal services model (i) Terminal services model
(2001) 1989 to 1993
(1) Number of runways (1) Number of passengers
(2) Number of gates (2) Number of pounds
(3) Terminal area
(4) Number of employees
(5) Number of baggage collection belts
(6) Number of public parking places

(ii) Movement model (ii) Movement model


(1) Airport area (1) Air carrier movements
(2) Number of runways (2) Commuter
(3) Runway area movements
(4) Number of employees

Pels, Nijkamp DEA-BCC model 34 European airports, (i) Terminal model


and Rietveld 1995 to 1997 (1) Terminal size in square metres (1) Number of passengers
(2001)* (2) Number of aircraft parking
positions at the terminal (ii) Movement model
(3) Number of remote aircraft parking (1) Aircraft transport
positions movement
(4) Number of check-in desks
(5) Number of baggage claims

frontiers and the efficiency measurement of the constructed et al., 1986) and the Super-Efficiency DEA model
frontiers assuming constant returns-to-scale (CRS). (Andersen and Petersen, 1993) have been used.
In terms of subsequent variants, the DEA-CCR and
DEA-BCC models are strong in identifying the inefficient 2.1. Data
units, but are weak in discriminating between the efficient
units (Seiford and Zhu, 1999). They often rate too many Forty-nine airports are registered in the Italian Statis-
units as efficient. To overcome this deficiency, the Cross- tical Register of Air Transportation and 31 airports for
Efficiency DEA model (Doyle and Green, 1994; Sexton which balance sheet data allowing the combination of
ARTICLE IN PRESS
186 C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191

financial and operational data are available are examined evidence as can the professional opinion of the airport
here using DEA (Table 1). managers.
Regarding WLU and total passenger numbers, the The output is measured by number of planes, number of
standard deviation is higher than the mean, signifying that passengers, cargo, aeronautical receipts, handling receipts,
the sample is not very homogenous. This is largely because and commercial receipts. We measure inputs as labor costs,
there are large airports in the sample alongside small ones, a capital invested and operational costs excluding the labor
feature stemming from the local population densities. costs. The monetary variables were deflated by the GDP
Relative to managerial status, only the main airports are deflator and denoted at constant 2001 prices. As DEA is
fully privatized, while the remainder operate under mixed designed to handle physical measures, but the balance sheets
regimes with the managing director a public appointee who provided only financial variables, the deflation of the
manages the airport by outsourcing its activities to private variables is of importance to minimize the role of prices.
enterprises. DEA that uses financial variables is common in the literature,
To estimate the cost frontier, balanced panel data on the but despite this, it remains a caveat in productivity analysis.
31 Italian airports for the 2001–2003 are used (Table 2). The combination of indicators meets DEA convention
The production of the airport authorities using a general- that the minimum number of DMU observations should be
ized Cobb–Douglas production function is measured. greater than three times the number of inputs plus outputs
Frontier models require the identification of inputs and [120X3(6+3)] (Raab and Lichty, 2002). Moreover, we also
outputs. Several criteria can be used in the selection. The respected the convention that the minimum number of
first empirical criterion is the availability of inputs and units is equal to or larger than the product of the number
outputs. Second, a literature survey can provide supporting of outputs and inputs (Boussofiane and Dyson, 1991).

Table 2
Characteristics of the Italian airports used in the analysis (2003)

No. Airport WLU (euro) Rotation of capital Management status Total passenger
(production/capital) (private total ¼ 1, arrivals (number)
private partial ¼ 0)

1 Alghero—Fertilia 812.01 0.37 0 440 606


2 Ancona—Falconara 519.81 0.65 0 235 409
3 Bari-Palese Macchie 1919.36 0.31 1 715 940
4 Bergamo-Orio aal Serio 2431.31 0.49 1 14 06 640
5 Bologna-Borgo Panigale 3554.03 0.49 0 17 54 485
6 Bolzano 36.60 0.11 0 20 843
7 Cagliari—Elmas 2201.08 0.13 0 11 45 426
8 Catania—Fontanarossa 4115.25 0.95 0 23 80 289
9 Crotone 38.35 0.23 0 44 153
10 Cuneo—Levaldigi 7.84 0.12 0 14 520
11 Firenze—Peretola 1385.74 0.62 1 685 840
12 Forli 161.35 0.5 0 175 719
13 Genova—Sestri 1030.54 0.69 1 517 044
14 Lamezia Terme 795.91 0.75 0 556 032
15 Milano—Malpensa 28356.95 0.36 1 87 65 195
16 Napoli—Capodichino 4158.05 0.51 1 2266937
17 Olbia—Costa Smeralda 1389.95 0.73 0 745 607
18 Palermo—Punta Raisi 3604.30 0.9 0 18 25 976
19 Parma 53.10 0.16 0 28 847
20 Perugia—Sant’Egidio 54.51 0.28 0 28 138
21 Pescara 310.82 0.84 0 148 598
22 Pisa—San Giusto 1701.30 0.83 0 985 267
23 Reggio Calabria 467.84 0.32 0 223 115
24 Rimini—Miramare 255.74 0.73 0 107 007
25 Roma—Fiumicino 27687.96 0.17 1 1 28 36 515
26 Torino—Caselle 2831.87 0.48 1 14 05 183
27 Trapani—Birgi 43.22 0.86 0 119 518
28 Treviso—Sant’Angelo 674.06 0.83 0 340 050
29 Trieste—Ronchi dei Legionari 654.20 1.12 0 302 795
30 Venezia—Tessera 4312.35 0.26 1 26 33 850
31 Verona—Villafranca 2231.05 0.59 0 11 95 367
Mean 3154.72 0.528 0.290 14 20 997
Median 1030.54 0.500 — 556 032
Standard Deviation 6779.30 0.281 — 26 67 912
ARTICLE IN PRESS
C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191 187

Table 3
Characteristics of the variables

Variables Definition Minimum Maximum Mean Square


deviation

Outputs
Number of planes Number of planes that landed at and take off from the airport 629 293790 35996.45 60640.01
Number of passengers Number of passengers who arrive at and depart from the 7222 25809828 2690814.68 5256722.37
airport
General cargo Number of tons of cargo that arrive and depart from the 0 26103618 976067.38 3377147.64
airport
Handling receipts Sales to planes in milliard Euro 0 272486 14599.09 39723.23
Aeronautical sales Sales to planes in milliard euro 0 206550 17582.90 42556.60
Commercial sales Sales to passengers in milliard euro 0 235406 16319.26 46357.69

Inputs
Labor costs Labor costs in milliard euro 574.886 170141.7 14530.01 34367.59
Capital invested Book value of assets in milliard euro 354.43 2795018 137532.64 455246.70
Operational costs excluding labour Operational in costs of airports in milliard euro 430.76 498970 41553.64 98288.04
costs

By using an output orientation, we can determine but five airports are efficient in the period. The rationale
whether an airport is capable of producing the same level for interpreting BCC as management skills is based on the
of output with less input (Table 3). contrast between the CCR and BCC models. The CCR
We observe that the Italian airport authorities are model identifies the overall inefficiency, whereas BCC
relatively heterogeneous, with the standard deviation being differentiates between technical efficiency and scale effi-
higher than the mean for some variables. ciency (Gollani and Roll, 1989). Based on this differentia-
tion, the ratio between CCR and BCC enables the
3. DEA results estimation of scale efficiency (Table 3) and, assuming
efficiency is due to managerial skills and scale effects, the
The DEA index can be calculated in several ways. Here BCC scores are interpreted as managerial skills. Thus,
we estimate an output-oriented, technically efficient (TE) according to the BCC scores, only three airports are
DEA index, assuming that airports aim to maximize profits inefficient.1
resulting from their activity. In this context, inputs are Fifth, according to the scale efficiency, almost all Italian
exogenous and the outputs are endogenous because of the airport authorities are efficient, while a small number
competitive environment in which the units compete is not.
(Kumbhakar, 1987). The variable returns-to-scale (VRS) Sixth, the efficiency scores are average values for the
hypothesis is chosen, disentangling technical efficiency into period, but when we analyze the airport authorities for
two components: pure technical efficiency and scale each year, the result is the same: all Italian airport
efficiency. The VRS scores measure pure technical effi- authorities display pure technical efficiency, but some of
ciency only. However, the CRS index is composed of a them do not display scale efficiency. Italian airports are
non-additive combination of pure technical and scale thus well managed as far as pure technical efficiency is
efficiencies. A ratio of overall efficiency scores to pure concerned, but dimension makes a difference and there-
technical efficiency scores provides a measurement of scale fore, some airports have decreasing returns to scale, while
efficiency. The relative efficiency of Italian airport autho- others have increasing returns to scale.
rities is presented in Table 4, with the airport authorities Table 5 presents the results of the Cross-Efficiency DEA
being ranked according to the BCC model, using GAMS model and the Super-Efficiency DEA model to cross-
software (Brooks et al., 1992). validate the DEA-CCR and DEA-BCC models and to
A number of points emerge. First, there are too many restrict the number of DMUs on the best practice frontier.
airport authorities on the efficient frontier; i.e. obtaining The scores from the Cross-Efficiency and the Super-
scores on the CCR and BCC of unity. Secondly, best- Efficiency DEA models, unlike those seen in Table 4, rank
practice calculations indicate that almost all Italian airport the Italian airports unequivocally, and that they maintain
authorities operated at a high level of pure technical
1
efficiency in the period. Thirdly, all technically efficient There may be an alternative reason for the level of technical efficiency.
CRS airports are also technically efficient in VRS, The assumption here is essentially that associated with Farrell’s work on
efficient frontiers, namely that a lack of skilled resources might lead to
signifying that the dominant source of efficiency is scale. production within it. An alternative view is that of Leibenstein whose
Fourthly, on the basis of the BCC results regarding pure work on X-inefficiency suggests that it is not always a matter of skills but
technical efficiency attributable to management skills, all can also involve motivation.
ARTICLE IN PRESS
188 C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191

Table 4
CCR-DEA Model and BCC-DEA Model

No. Airport Technical efficiency, Technical efficiency, Scale efficiency Position of the
constant returns-to- variable returns-to- company on the
scale CCR model scale BCC model frontier

1 Alghero—Fertilia 0.936 0.940 0.997 Drs


2 Ancona—Falconara 1.000 1.000 1.000 —
3 Bari-Palese Macchie 0.759 0.763 0.994 Drs
4 Bergamo-Orio aal 1.000 1.000 1.000 —
Serio
5 Bologna-Borgo 1.000 1.000 1.000 —
Panigale
6 Bolzano 1.000 1.000 1.000 —
7 Cagliari—Elmas 1.000 1.000 1.000 —
8 Catania— 1.000 1.000 1.000 —
Fontanarossa
9 Crotone 0.533 1.000 0.533 Irs
10 Cuneo—Levaldigi 1.000 1.000 1.000 —
11 Firenze—Peretola 1.000 1.000 1.000 —
12 Forli 1.000 1.000 1.000 —
13 Genova—Sestri 1.000 1.000 1.000 —
14 Lamezia Terme 1.000 1.000 1.000 —
15 Milano—Malpensa 1.000 1.000 1.000 —
16 Napoli—Capodichino 1.000 1.000 1.000 —
17 Olbia—Costa 1.000 1.000 1.000 —
Smeralda
18 Palermo—Punta Raisi 1.000 1.000 1.000 —
19 Parma 0.918 1.000 0.918 Drs
20 Perugia—Sant’Egidio 0.703 0.741 0.948 Irs
21 Pescara 1.000 1.000 1.000 —
22 Pisa—San Giusto 1.000 1.000 1.000 —
23 Reggio Calabria 1.000 1.000 1.000 —
24 Rimini—Miramare 1.000 1.000 1.000 —
25 Roma—Fiumicino 1.000 1.000 1.000 —
26 Torino—Caselle 1.000 1.000 1.000 —
27 Trapani—Birgi 1.000 1.000 1.000 —
28 Treviso—Sant’Angelo 1.000 1.000 1.000 —
29 Trieste—Ronchi dei 1.000 1.000 1.000 —
Legionari
30 Venezia—Tessera 1.000 1.000 1.000 —
31 Verona—Villafranca 1.000 1.000 1.000 —
Mean 0.962 0.982 0.980
Median 1.000 1.000 1.000
Standard Deviation 0.105 0.062 0.084

Technical efficiency scores for Italian airport authorities, average values for the period 2001–2003.

the same ranking, thereby overcoming the difficulty that The following hypotheses are defined:
the CCR-DEA and BCC-DEA models have in discriminat-
Hypothesis 1. Large airport authorities are more efficient
ing between the efficient units.
than small airport authorities.
3.1. Efficiency by different types of Italian airports This is a common hypothesis in airport analysis where
dimension and efficiency are possibly related. Airports are
Having established the efficiency rankings of the Italian classified by the book values of their assets and then
airport authorities, we now test some hypotheses related to dividing the sample into two based on this. It is expected
the rankings obtained. The Mann–Whitney U-test, which that larger airports are more efficient because of economies
tests for differences between the efficiency scores, is of scale.
adopted. Golany (1996) recommends the Mann–Whitney
Hypothesis 2. Airports under fully private management are
U-test for the non-parametric analysis of DEA results that
more efficient than those under partially private manage-
is used here because the efficiency scores do not follow a
ment.
standard normal distribution. The super-efficient scores are
chosen because they adequately discriminates the units The separation of Italian airport authorities into
analyzed. partially private and fully private is based on their property
ARTICLE IN PRESS
C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191 189

Table 5
Cross-efficiency DEA model and Super-Efficiency DEA model, Technical Efficiency scores for Italian airport authorities, average values for the period
1991–2003

No. Airport Technical efficiency, cross- Technical efficiency, super-efficiency scores


efficiency scores

1 Alghero—Fertilia 0.961 0.982


2 Ancona—Falconara 1.036 1.172
3 Bari-Palese Macchie 0.759 0.763
4 Bergamo-Orio aal Serio 1.129 1.135
5 Bologna-Borgo Panigale 1.135 1.139
6 Bolzano 1.082 1.117
7 Cagliari—Elmas 1.053 1.192
8 Catania—Fontanarossa 1.032 1.125
9 Crotone 0.638 1.052
10 Cuneo—Levaldigi 1.028 1.152
11 Firenze—Peretola 1.067 1.173
12 Forli 1.118 1.282
13 Genova—Sestri 1.052 1.245
14 Lamezia Terme 1.053 1.182
15 Milano—Malpensa 1.378 1.452
16 Napoli—Capodichino 1.219 1.352
17 Olbia—Costa Smeralda 1.191 1.238
18 Palermo—Punta Raisi 1.102 1.218
19 Parma 0.927 1.021
20 Perugia—Sant’Egidio 0.718 0.821
21 Pescara 1.052 1.091
22 Pisa—San Giusto 1.231 1.327
23 Reggio Calabria 1.093 1.112
24 Rimini—Miramare 1.116 1.135
25 Roma—Fiumicino 1.532 1.163
26 Torino—Caselle 1.025 1.051
27 Trapani—Birgi 1.032 1.063
28 Treviso—Sant’Angelo 1.045 1.052
29 Trieste—Ronchi dei Legionari 1.063 1.085
30 Venezia—Tessera 1.072 1.086
31 Verona—Villafranca 1.067 1.073
Mean 1.064 1.130
Median 1.063 1.135
Standard Deviation 0.166 0.136

status. There are 10 Italian airports under private manage- The minus sign of the Z-score indicates that large
ment. We compare the efficiency scores relative to these 10 airports tend to have higher efficiency scores than smaller
airports (30 observations) with the efficiency scores of the ones, validating the first hypothesis and is in line with
10 most efficient public airports. We expect the privately previous research on airport efficiency (Murillo-Melchor,
managed airports to be more efficient, since they have 1999). Moreover, privately managed airports are found to
higher efficiency scores. This expectation gains support have higher efficiency scores than publicly managed
from the presence of principal-agent relationships observed facilities validating the second hypothesis that private
in public entities (Jensen and Meckling, 1976). airports are more efficient than their public counterparts
(Parker, 1999).
Hypothesis 3. Airports with higher WLU during the period
Finally, the test shows that airports with higher WLU
are more efficient than airports with lower WLU.
tend to be more efficient than airport authorities with lower
The WLU is considered a natural benchmark element WLU supporting the notion that WLU is a good proxy for
when comparing airport (Graham, 2005) and the stated the efficiency scores (Graham, 2005).
hypothesis is an intuitive one based on the definition of We thus find that the majority of Italian airports are
WLU, as a standard measure of output defined as one managed with pure technical efficiency but that scale
passenger or 100 kg of cargo. The sample is split into two effects differentiate airport authorities, with some display-
on the basis of the WLU rankings. It is expected that the ing scale efficiency and others not. Moreover, large
higher the WLU airports are the more efficient due to airports, with a higher book value of assets, tend to have
economies of scale effects. Table 6 presents the Mann– higher efficiency scores than those with lower values, an
Whitney Z-test results. effect that can be explained by the economies of scale
ARTICLE IN PRESS
190 C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191

Table 6
Mann–Whitney test of differences in efficiency

Reference Mann–Whitney U-test Mann–Whitney Z-test Asymptotic significance


(two-tailed)

Large airport authorities vs. 199.00 1.38 0.029*


small airport authorities
Privately managed airport 135.00 1.75 0.025*
authorities vs. publicly managed
airport authorities
Airport authorities with higher 162.00 1.63 0.051*
WLU vs. airport authorities
with lower WLU

*Indicates significance at a 5% level.

(Murillo-Melchor, 1999). Additionally, totally private air- Brooks, A.D., Kendrick, D., Meerhaus, A., 1992. GAMS: A User’s
ports tend to have higher efficiency scores than partially Guide. Scientific Press, Los Angeles.
Charnes, A., Cooper, W.W., Rhodes, E., 1978. Measuring the efficiency of
private airport authorities possibly because of differences
decision-making units. European Journal of Operations Research 2,
in the principal-agent relationship (Jensen and Meckling, 429–444.
1976). This relates to the difficulty of controlling those De Borger, B., Kerstens, K., Costa, A., 2002. Public transit performances:
empowered as managers to act on behalf of the govern- What does one learn from frontier studies? Transport Reviews 22 (1),
ment. Since partially private Italian airports have public- 1–38.
appointed Managing Directors, they may be unable to full Doyle, J., Green, R., 1994. Efficiency and cross-efficiency in DEA:
derivations, meanings and uses. Journal of the Operational Research
achieve efficiency. Finally, airports with higher WLU tend Society 45, 567–578.
to be more efficient than those with lower WLU, an effect Farrell, M.J., 1957. The Measurement of productive efficiency. Journal of
that is explained by the economies of scale (Graham, 2005). the Royal Statistical Society, Series A 120, 253–290.
The general conclusion is that scale is of paramount Gillen, D., Lall, A., 1997. Non-parametric measures of efficiency
importance and therefore, the DEA-CCR models should of US airports. International Journal of Transport Economics 28,
283–306.
not be used alone in the evaluation of the airports’ Gillen, D., Lall, A., 2001. Developing measures of airport productivity
performance. and performance: an application of data envelopment analysis.
Transportation Research E 33, 261–273.
Gollani, B., Roll, Y., 1989. An application procedure for DEA. Omega
4. Conclusions International Journal of Management Sciences 17, 237–250.
Graham, A., 2005. Airport benchmarking: a review of the current
This article used a simple framework for the comparative situation. Benchmarking: An International Journal 12, 99–111.
evaluation of the Italian airport authorities and the Jensen, M.C., Meckling, W., 1976. Theory of the firm: managerial
behavior, agency costs and capital structure. Journal of Financial
rationalization of their operational activities. The analysis
Economics 3, 305–360.
is based on a DEA model that allows for the incorporation Kumbhakar, S.C., 1987. Production frontiers and panel data: an
of multiple inputs and outputs in determining relative application to US Class 1 railroads. Journal of Business and
efficiencies. Benchmarks are provided for improving the Economics Statistics 5, 249–255.
operations of poorly performing airports. The general Murillo-Melchor, C., 1999. An analysis of technical efficiency and
productive change in Spanish airports using the Malmquist Index.
conclusion is that the Italian airports display relatively high
International Journal of Transport Economics 26, 271–292.
managerial skills, being VRS-efficient for the most part. Parker, D., 1999. The performance of the BAA before and after
However, there are some inefficient airport authorities and privatization. Journal of Transport Economics and Policy 33, 133–146.
the analysis highlights some of the reasons for this. Raab, R., Lichty, R., 2002. Identifying sub-areas that comprise a greater
metropolitan area: the criterion of county relative efficiency. Journal of
Regional Science 42, 579–594.
References Seiford, L.M., Zhu, J., 1999. Infeasibility of super-efficiency data
envelopment analysis models. INFOR 37, 174–187.
Adler, N., Berechman, J., 2001. Measuring airport quality from the Sexton, T.R., Silkman, R.H., Hogan, A., 1986. Data envelopment
airlines viewpoint: An application of data envelopment analysis. analysis: critique and extension. In: Silkman, R.H. (Ed.), Measuring
Transport Policy 8, 171–181. Efficiency: an Assessment of Data Envelopment Analysis. Jossey-Bass,
Andersen, P., Petersen, N.C., 1993. A procedure for ranking efficiency San Francisco.
units in data envelopment analysis. Management Science 39,
1261–1264.
Banker, R.D., Charnes, A., Cooper, W.W., 1984. Some models for Further reading
estimating technical and scale inefficiencies in data envelopment
analysis. Management Science 30, 1078–1092. Barros, C.P., Sampaio, A., 2004. Technical and allocative efficiency
Boussofiane, A., Dyson, R.G., 1991. Applied data envelopment analysis. in airports. International Journal of Transport Economics 31,
European Journal of Operational Research 52, 1–15. 355–377.
ARTICLE IN PRESS
C. Pestana Barros, P.U.C. Dieke / Journal of Air Transport Management 13 (2007) 184–191 191

Brockett, P.L., Golany, B., 1996. Using rank statistics for determining Sarkis, J., 2000. Operational efficiency of major US airports. Journal of
programming efficiency differences in data envelopment analysis. Operation Management 18, 251–335.
Management Science 42, 466–472. Sarkis, J., Talluri, S., 2004. Performance-based clustering for
Charnes, A., Cooper, W.W., Gollany, B., Seiford, L., Stutz, J., 1985. benchmarking of US airports. Transportation Research A 38,
Foundations of data envelopment analysis for Pareto-Koopmans 329–346.
efficient empirical productions functions. Journal of Econometrics 30, Short, J., Kopp, A., 2005. Transport infrastructure: investment
91–107. and planning. Policy and research aspects. Transport Policy 12,
Charnes, A., Cooper, W.W., Seiford, L., Stutz, J.A., 1982. Multiplicative 360–367.
model of efficiency analysis. Socio-Economic Planning Sciences 16, Seiford, L.M., Thrall, R., 1990. Recent developments in DEA: the
223–224. mathematical programming approach to frontier analysis. Journal of
Charnes, A., Cooper, W.W., Huang, Z.M., 1990. Polyhedral Cone-Ratio Econometrics 46, 7–38.
DEA with an illustrative application to large commercial banks. Thompson, R.G., Langemeier, L.N., Lee, C., Thrall, R.M., 1990.
Journal of Econometrics 46, 73–91. The role of multiplier bounds in efficiency analysis with
Charnes, A., Cooper, W.W., Lewin, A.Y., Seiford, L.M., 1995. Data application to Kansas farming. Journal of Econometrics 46,
Envelopment Analysis: Theory, Methodology and Applications. 93–108.
Kluwer, Boston. Thompson, R.G., Singleton, F.D., Thrall, R.M., Smith, B.A., 1986.
Cooper, W.W., Seiford, L.M., Tone, K., 2000. Data Envelopment Comparative site evaluation for locating a high-energy Physics lab in
Analysis. Kluwer Academic Press, Boston. Texas. Interfaces 16, 35–49.
Hooper, P.G., Hensher, D.A., 1997. Measuring total factor productivity Yoshida, Y., Fujimoto, H., 2004. Japanese-airport benchmarking with
of airports: an index number approach. Transportation Research E 33, DEA and endogenous-weight TFP methods: testing the criticism of
249–259. over-investment in Japanese regional airports. Transportation Re-
Humphreys, I., Francis, G., 2002. Performance measurement: a review of search E 40, 533–546.
airports. International Journal of Transport Management 1, 79–85. Varian, H.R., 1987. Intermediate Microeconomics: A Modern Approach.
Pels, E., Nijkamp, P., Rietveld, P., 2001. Relative efficiency of European W.W. Norton & Co, New York.
airports. Transport Policy 8, 183–192. Yoshida, Y., 2004. Endogenous-weight TFP measurement: methodology
Pels, E., Nijkamp, P., Rietveld, P., 2003. Inefficiency and scale economics of and its application to Japanese-airport benchmarking. Transportation
European airport operations. Transportation Research E 39, 341–361. Research E 40, 151–182.

You might also like