Vous êtes sur la page 1sur 3

Journal of Air Transport Management 16 (2010) 334e336

Contents lists available at ScienceDirect

Journal of Air Transport Management


journal homepage: www.elsevier.com/locate/jairtraman

Note

Measuring the inuence of congestion on efciency in worldwide airports


Rui Cunha Marques*, Pedro Simes
Center of Urban and Regional Systems, Technical University of Lisbon, Avenida Rovisco Pais, Lisbon, Portugal

a b s t r a c t
Keywords:
Airports
Congestion
Efciency

This short communication evaluates the inuence of congestion on the technical efciency of airports
using three different approaches. To accomplish this aim a sample of 141 worldwide airports is used. The
results show considerable signs of congestion inefciency in some airports, highlighting the importance
of studying this phenomenon.
2010 Elsevier Ltd. All rights reserved.

1. The problem of congestion


Congestion is an issue in the airport industry; for example, due
to the number of movements, particularly in the peak hours.
Congestion is characterized by the decrease of outputs produced as
a consequence of a large increase of inputs used or the decrease of
one or more outputs by the increase of other. In this communication we compare three different approaches available to compute
the congestion effect on efciency. The rst was introduced by Fre
et al. (1985). It was the rst study to dene congestion algebraically
using a radial measure. Later, other two alternatives to measure
congestion efciency were proposed respectively by Cooper et al.
(2001) and Tone and Sahoo (2004) using non-radial measures.
Cooper et al. argue that input congestion occurs whenever more
than one input is employed while all other inputs are held constant,
and this leads to a fall in the outputs. This approach is based on the
hypothesis of diminishing marginal returns. Congestion requires
a negative marginal product to take place. Tone and Sahoo uses
a unied approach to measure congestion efciency rooted in the
slacks-based measure (SBM).
2. Different approaches to measure congestion
2.1. Fre et al. approach
The Fre et al. method to measure congestion comprises two
stages whereby each one distinguishes the strong disposal from
weak disposal properties given by the inequality and equality in
the traditional Charnes, Cooper and Rhodes (CCR) model (Charnes
et al., 1978), respectively. Therefore, it is possible to compute two
measures of technical efciency (q* e b*). With more restrictive

* Corresponding author. Tel.: 351 218418305; fax: 351 218409884.


E-mail address: rcmar@civil.ist.utl.pt (R.C. Marques).
0969-6997/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.jairtraman.2010.03.002

constraints in the weak disposal model, we will obtain 0  q  b .


This property allows for measuring the congestion by means of 0 
*
* *
Cq ; b q*  1. The computation of the two models in a twob
stage procedure provides the Fre et al. estimation of congestion. In
this case, an airport shows signs of congestion if and only if
* *
* *
Cq ; b < 1. In contrast, congestion is not present if Cq ; b 1.
2.2. Cooper et al. approach
The Cooper et al. approach also proceeds in a two-stage method
to compute the congestion. The comparison with the Fre et al.
approach reveals some differences between them, inter alia, the
addition of the convexity condition and consequently it includes
the Banker, Charnes and Cooper (BCC) model (Banker et al., 1984)
and also the incorporation of slacks into the objective function,
being multiplied by the non-Archimedean value (3 > 0). The latter is
a relevant difference when compared with the Fre et al. approach.
Under a general perspective, in the Cooper et al. approach the nonzero slacks have more weight. Furthermore, the second stage used
as Brockett et al. (1998) increases even more the differences for the
treatment of slacks adopted by Fre et al.
2.3. Tone and Sahoos approach
Tone and Sahoo have proposed a new unied approach to
measure congestion efciency. This method allows us to evaluate
congestion of inputs via an output oriented approach and it is
similar to Cooper et al. output oriented method in so far as a BCC
output-oriented model is used in the rst stage. However, this is
distinguished by the SBM procedure used in the second stage (Tone,
2001). The main interpretation of this ratio is that the correlation
between an average reduction in inputs corresponds to a certain
increase of outputs produced.
Tone and Sahoo developed two denitions distinguishing strong
from weak congestion. In practical terms, strong congestion

R.C. Marques, P. Simes / Journal of Air Transport Management 16 (2010) 334e336

335

Fig. 1. Technical inefciency of worldwide airports according to CRS and VRS models.

Fig. 2. Levels of congestion inefciency of airports according to Fre et al. approach.

Fig. 3. Levels of congestion inefciency of airports according to Cooper et al. approach.

corresponds to the congestion of all inputs while in the weak


congestion not all inputs are congested. Strong congestion implies
weak congestion but not vice-versa. This approach has several
appealing features, such as the fact that negative marginal
productivity is linked to the existence of congestion (unlike Fre
et al. method), the analysis can easily be carried out using friendly
software and, nally, the result is comprehensive and easily
understood.
3. Are the worldwide airports congested?
3.1. Data, model specication and data envelopment analysis
results
We considered a model with four inputs and three outputs. The
outputs adopted were the number of movements, the number of
passengers and the cargo transported whereas the inputs were the
number of runways, the number of gates, the terminal area and the

number of employees. The orientation of the model is not


a consensual aspect of the efciency models in airports. An input
minimization orientation, which favors the public service view, is
usually adopted but nowadays airport industry is much more than
just this and although the public services component is
unchangeable, we privileged here an output orientation.
This work uses airports data relative to the year 2006. We used the
database of Air Transport Research Society (ATRS), although in some
cases we collected data from airport websites and from their annual
account reports to complement it. The sample includes 141 airports.
With the exception of Tone and Sahoos approach, all the computations are carried out with Matlab codes developed by the authors.
The performance of worldwide airports was measured by the
application of data envelopment analysis (DEA) with an output
orientation (see, about DEA technique, Fried et al., 2008). Therefore,
two models were computed, namely DEA-CCR (considering
constant returns to scale e CRS) and DEA-BCC (considering variable
returns to scale e VRS).

336

R.C. Marques, P. Simes / Journal of Air Transport Management 16 (2010) 334e336

Fig. 4. Levels of congestion inefciency of airports according to Tone and Sahoos approach.

Fig. 1 provides the inefciency estimates using the DEA-CCR and


DEA-BCC models. The computation of scale inefciency from the
gure is immediate. The results show that 28 airports out of 141 are
efcient under the DEA-CCR model and 53 under the DEA-BCC
model. As expected, the DEA-BCC model shows higher efciency
than the DEA-CCR model. The DEA-BCC model presents an average
value of inefciency of 0.19 while DEA-CCR model gives a value of
0.3.

Table 1
Comparison of congestion results between the three approaches.

3.2. Fre et al. approach

3.5. Concluding remarks

Fre et al. approach, assuming VRS (with the CRS the results are
comparable), measures the congestion through the ratio between
strong efciency and weak efciency. The results show that 73 out
of the 141 airports are congested. On average, the worldwide
airports are 5.7% inefciently congested. However, congestion
inefciency increases to 11.0% in airports when we consider just the
congested ones. Fig. 2 presents the congestion inefciency results
of the 141 airports analyzed. Richmond, Dubai, Munich, London
Stansted, Albany, Paris Orly, Christchurch, Winnipeg, Orlando and
Birmingham are the airports which display the greatest congestion
inefciency.

From the comparison between different approaches we identied important differences in the results, since the assumptions of
each approach are diverse. However, there are several congested
airports in the three methodologies and with similar levels of
inefciency. A priori there are no doubts about the existence of
congestion in these airports. In this circumstance we found 34
airports. Table 1 systematizes the results obtained with the three
approaches adopted and available in the literature. Although there
is lack of consensus about how to measure congestion inefciency
and what is the best technique to use, this paper highlights the
importance of this phenomenon. In our sample, we found that
more than 50% are inefciently congested and that they represent
at least about 6% of inefciency for the whole sample and 11% only
for the inefciently congested ones.

3.3. Cooper et al. approach


Regarding Cooper et al. approach, the results show that 77
airports are inefciently congested. The average value obtained for
all the sample was 7.7% and 13.8% for inefciently congested
airports. Fig. 3 shows the congestion inefciency results of each
airport. We found that the airports of Munich, Dubai, Detroit, Birmingham, Richmond, Vienna, Christchurch, Jakarta, Paris-Orly and
Portland have the highest congestion inefciencies. In general the
results are quite similar to the Fre et al. approach.
3.4. Tone and Sahoos approach
At last we applied the Tone and Sahoos approach and we found
that 80 airports are inefciently congested. The average of
congestion inefciency in the whole sample corresponds to 6.3% of
inefciency. For the inefciently congested airports this percentage
rises to 11.1%. From the highest congested airports the Tone and
Sahoos approach identied Vienna, Chiang Kai-Check, Macau,
Zurich, Ottawa, Albany, Portland, London Heathrow, London
Stansted and Paris Orly. Fig. 4 shows the congestion inefciency
results of each airport with the Tone and Sahoos approach.

Fre et al. Cooper et al. Tone and Sahoo


Congestion average (whole sample) 0.057
Maximum
0.58
Congested airports (no.)
73
Average of congested airports
11.0

0.077
0.65
77
13.8

0.063
0.67
80
11.1

References
Banker, R., Charnes, A., Cooper, W., 1984. Some models for estimating technical and
scale inefciencies in data envelopment analysis. Management Science 30,
1078e1092.
Brockett, P., Cooper, W., Shin, H., Wang, Y., 1998. Inefciency and congestion in
Chinese production before and after the 1978 economic reforms. SocioEconomic Planning Sciences 32, 1e20.
Charnes, A., Cooper, W., Rhodes, E., 1978. Measuring the efciency of decision
making units. European Journal of Operational Research 2, 429e444.
Cooper, W., Gu, B., Li, S., 2001. Comparisons and evaluations of alternative
approaches to the treatment of congestion in DEA. European Journal of Operational Research 132, 62e74.
Fre, R., Grosskopf, S., Lovell, C., 1985. The Measurement of Efciency of Production.
Kluwer-Nijhoff, Boston.
Fried, H., Lovell, K., Schmidt, S., 2008. The Measurement of Productive Efciency and
Productivity Change. Oxford University Press, New York.
Tone, K., 2001. A slacks-based measure of efciency in data envelopment analysis.
European Journal of Operational Research 130, 498e509.
Tone, K., Sahoo, B., 2004. Degree of scale economies and congestion: a unied DEA
approach. European Journal of Operational Research 158, 755e772.

Vous aimerez peut-être aussi