Journal of Air Transport Management 31 (2013) 1e5

Contents lists available at SciVerse ScienceDirect

Journal of Air Transport Management journal homepage: www.elsevier.com/locate/jairtraman

Note

Physical infrastructure and flight consolidation efficiency drivers in Brazilian airports: A two-stage network-DEA approach Peter F. Wanke Federal University of Rio de Janeiro, Rua Paschoal Lemme, 355, Rio de Janeiro, Brazil

a b s t r a c t Keywords: Brazilian airports Flight consolidation Airport efficiency

Efficiency in Brazilian airports is measured using a two-stage process. In the first stage, physical infrastructure efficiency, assets, such as terminal area, aircraft parking spaces, and runways are related to the number of landings and take-offs per year. The second focuses on flight consolidation efficiency in terms of the number of passengers carried and cargo handled per year. A network-DEA centralized efficiency model is used to optimize the stages simultaneously. Results indicate that contextual variables or efficiency drivers, such as hub operations and airport location, impact physical infrastructure and flight consolidation efficiency levels differently. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction Although there is evidence that airports owned/managed by governments are significantly less efficient than airports with a private majority ownership, only recently has the Brazilian government initiated moves towards privatization and deregulation. The lack of pressure on airports to be more competitive and productive, together with insufficient investments in infrastructure, led to operational bottlenecks and capacity shortfalls that are a concern for the forthcoming World Cup in 2014 and Olympic Games in 2016 (Wanke, 2012). Here we look at the way the number of movements at an airport links the longer and shorter-term perspectives of its physical infrastructure and flight consolidation. It also examines factors affecting the capacity airport shortfall in Brazil. 2. Data and model Secondary data from a sample of 63 Brazilian public airports operated by the state company, Infraero, are taken from the National Agency for Civil Aviation (ANAC) sources.1 A two-stage (Fig. 1) approach to modelling is adopted. In the first stage, physical infrastructure efficiency assets, such as terminal area, aircraft parking spaces, and runways are related to the number of landings and take-offs per year. The second focuses on flight consolidation efficiency in terms of the number of passengers carried and cargo handled per year.

1

E-mail address: [email protected]. http://www.anac.gov.br.

0969-6997/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.jairtraman.2012.09.001

In the first productive stage, the terminal area, the number of the aircraft parking spaces, and the runways are inputs to the number of landing and take-offs per year; our single intermediate measure. This is to minimize the physical infrastructure required to achieve a given number of annual movements per. In the second stage, the landings and take-offs per year are used as a determinant of the number of passengers per year, as well as a cargo throughput. Although the number of runways is the input/output variable with the smallest coefficient of variation, there are two airports with runways exclusively dedicated for use by general aviation being less than 1000 m; Jacarepaguá (RJ) and Carlos Prates (BH). Regarding the relationship between the number of decisionmaking units (DMUs) and inputs/outputs, the ratio used by Cooper et al. (2001) is relevant; the number of DMUs should be at least three times the number of inputs and outputs. Correlation analyses indicate significant positive relationships between the input and the output variables, justifying their inclusion (Wang et al., 2011). The data related to 2009 are presented in Table 1. Contextual variables presented in Table 1 relate to the airport type: hub (1) or non-hub (0); international (1) or non-international (0); metropolitan or (1) non-metropolitan (0); regular flights (1) or non-regular flights (0). An underlying assumption considered here is that these contextual variables or efficiency drivers are exogenous, that is, they affect efficiency levels without being affected by them. Here, hub, international, location, and regular flights represent, therefore, decision variables based on Infraero’s discretion, rather than endogenous variables generated within the ambit of an efficiency model or a production process. To use the two-stage network Data Envelopment Analysis (DEA) model (Liang et al., 2008), it is assumed that DMUj (j ¼ 1, 2, . n) has

2

P.F. Wanke / Journal of Air Transport Management 31 (2013) 1e5

Fig. 1. Airport two-stage DEA model.

D intermediate measures zdj (d ¼ 1, 2, ., D), besides the initial inputs xij (i ¼ 1, 2, ., m) and the final outputs yrj (r ¼ 1, 2, ., s) (Fig. 2). Also considering vi, wd, and ur as unknown non-negative weights, the two-stage network DEA model is:

qGlobal ¼ Max o s:t: s P

ur yrj 

r¼1 D P d¼1 m P i¼1

s P r¼1

D P d¼1 m P

wd zdj 

ur yro

wd zdj  0; j ¼ 1; 2; .n;

i¼1

(1)

vi xij  0; j ¼ 1; 2; .n;

vi xio ¼ 1;

wd  0; d ¼ 1; 2; .D; vi  0; i ¼ 1; 2; .m; ur  0; r ¼ 1; 2; .s;

where qo is overall (global) efficiency level of the two-stage process for DMUo. Assuming that model 1 yields a unique solution, the efficiencies for the first and second stages are given by: Global

Infrastructure q1;Physical ¼ o

and Consolidation q2;Flight ¼ o

D X d¼1

s X

w*d zdo ;

u*r yro

D .X

r¼1

d¼1

(2)

w*d zdo :

(3)

Since a unique solution is assumed, it is possible to define: 1;Physical Infrastructure 2;Flight Consolidation qGlobal ¼ qo *qo ; o

(4)

that is, the overall efficiency level is equal to the product of the individual efficiency levels for each stage. 3. Results The physical infrastructure and flight consolidation efficiency levels calculated using the two-stage network DEA model for each DMU are seen in Fig. 3. Visual inspection suggests that the networkDEA model has more discriminate power than does the standard DEA model, which usually finds more DMUs as the most efficient ones (Zhu, 2011). In 2009, only four of the 63 airports achieved 100% efficiency in the first stage of physical infrastructure efficiency, all of them old, small or former major city airports surrounded by populated vicinities: Campo de Marte (SP), Jacarepaguá (RJ), Carlos Prates (BH), and Bacacheri (CWB). These airports belong to a group of small-sized DMUs that handles a significant amount of the landing/ take-offs at Brazil’s airports, although the passenger and cargo shares are negligible because their operations are focused on, and in some cases, restricted to, smaller aircrafts. In a broader sense, this also indicates that most of the newer and larger Brazilian airports were not using their physical infrastructure efficiently to generate enough landings and take-offs a year. On the other hand, only one of the 63 airports achieved 100% efficiency in the second stage of flight consolidation efficiency: Viracopos (Campinas), the largest Brazilian airport geared to cargo transportation. The median value for the physical infrastructure efficiency, however, is lower than that for flight consolidation e 0.19 versus 0.38 e suggesting the airports tend to be comparatively more efficient in turning aircraft movements into passenger and cargo flows than in turning physical infrastructure into aircraft movements.

Table 1 Summary statistics for the sample. Descriptive

Initial inputs

Final outputs

Terminal area (m2)

Aircraft parking spaces

Runways

Passengers (per year)

Cargo throughput (kg/yr)

Mean Standard deviation Coefficient of variation Minimum Maximum

17,471.6 28,711.2 1.6 157.1 147,834.0

18.2 13.3 0.7 2.0 61.0

1.2 0.4 0.3 1.0 2.0

2,032,320.6 3,902,928.6 1.9 2046.0 21,727,649.0

17,651,456.2 53,003,637.3 3.0 e 351,787,564.0

Descriptive

Intermediate output/input

Mean Standard deviation Coefficient of variation Minimum Maximum

Contextual variables

Landings and take-offs per year

Regular flights

Location

International

Hub

36,227.0 45,427.2 1.3 1301.0 209,636.0

0.8 0.4 0.4 e 1.0

0.6 0.5 0.9 e 1.0

0.5 0.5 1.0 e 1.0

0.1 0.2 3.9 e 1.0

P.F. Wanke / Journal of Air Transport Management 31 (2013) 1e5

3

Fig. 2. Efficiency decomposition.

An exploratory analysis, considering the median values for the physical infrastructure and flight consolidation efficiency levels (Fig. 3), delineates four quadrants or groups. For each one of these, descriptive statistics are computed on the efficiency estimates, on the inputs/outputs, and on the remainder contextual variables (Table 2). DMUs located in Group No. 1 have reasonable infrastructure; they are large airports with high physical infrastructure and flight consolidation efficiency levels, most of them located in large metropolitan areas, offering international flights, and operating as hubs. They account for more than 80% of passengers, 90% of cargo, and 60% of movements at Brazilian airports. This group encompasses the relevant airports of São Paulo (Guarulhos, Campinas,

and Congonhas) and Rio de Janeiro states (Galeão and Santos Dumont), besides the airports of Brasília, Belém, Manaus, and other major Northeast capitals. On the other hand, DMUs located in Group No. 2 consist of poorly infrastructured, small airports with low physical infrastructure and flight consolidation efficiency levels. Most of them located at extreme southern, northern, and mid-western regions, with limited opportunities to operate connecting flights to other regions. They account for only 2% of passengers, 2% of cargo, and 5% of the movements of Brazilian airports. Group No. 3, consists of old, poorly infrastructured airports located near or within large metropolitan areas. They handle 22% of the movements of Brazilian airports, although the passenger and

Fig. 3. Distributions of efficiency estimates.

4

P.F. Wanke / Journal of Air Transport Management 31 (2013) 1e5

Table 2 Differences between groups. Variables

Initial inputs

Final outputs Intermediate Efficiency levels

Number of cases Airport type within each group

Group relevance

Groups of airports

Terminal area (m2) Aircraft parking spaces Runways Passengers (per year) Cargo throughput (kg/yr) Landings and take-offs (per year) Global Physical infrastructure Flight consolidation Regular flights Location International Hub Passenger share Cargo share Movement share

Airports in each group

Both high physical infrastructure and flight consolidation efficiency levels

Both low physical infrastructure and flight consolidation efficiency levels

High physical infrastructure efficiency level but low flight consolidation efficiency level

Low physical infrastructure efficiency level but high flight consolidation efficiency level

42,102.63 21.72 1.50 5,712,135.89 56,480,536.89 79,708.67

6698.13 14.58 1.16 113,463.95 885,889.53 6207.89

4630.68 19.62 1.00 649,073.85 2,023,129.77 37,860.31

11,953.78 17.00 1.23 1,124,921.46 4,019,960.69 18,262.62

0.21 0.33 0.66 18 100% 94% 72% 22% 80% 91% 63% Florianópolis Navegantes Curitiba e International Campinas Guarulhos Congonhas Galeão Santos_Dumont Vitória Cuiabá Goiânia Brasília Manaus Belém e International Teresina Fortaleza Recife Salvador

0.01 0.09 0.13 19 74% 16% 42% 0% 2% 2% 5% Bagé Pelotas Uruguaiana Campos dos Goytacazes Montes Claros Corumbá Ponta Porã Rio Branco Boa Vista Altamira Marabá Parauapebas Santarém Palmas Imperatriz Parnaíba Campina Grande Petrolina Paulo Afonso

0.05 0.62 0.12 13 62% 62% 15% 0% 7% 2% 22% Porto Alegre Londrina Bacacheri (CWB) São José dos Campos Campo de Marte (SP) Macaé Jacarepaguá (RJ) Uberaba Pampulha Carlos Prates (BH) Uberlândia Cruzeiro do Sul Belém

0.08 0.13 0.63 13 100% 62% 69% 0% 11% 5% 10% Joinville Foz do Iguaçu Confins Campo Grande Porto Velho Macapá São Luís Juazeiro do Norte Parnamirim João Pessoa Maceió Aracaju Ilhéus

Table 3 Differences among airport types. Efficiency levels

Median values e KruskaleWallis test Airport type Regular flights

Global Physical infrastructure Flight consolidation

Yes

No

0.08 0.19 0.42

0.01 0.22 0.02

Sig.

0.00 0.89 0.00

Location Yes

No

0.12 0.26 0.51

0.03 0.12 0.19

the cargo shares are negligible because their operations are focused on smaller aircraft, and thus their flight consolidation efficiency levels are relatively low. Finally, Group No. 4 represents mid-sized infrastructured airports, located at touristic, sea-shore regions; most in the northeast region, where cargo movement of high value added products is negligible adversely affecting physical infrastructure efficiency levels. They account for significant share of passengers and cargo, which partly explains their high flight consolidation efficiency levels. Non-parametric KruskaleWallis tests are used for testing significant differences in global, physical infrastructure, and flight consolidation efficiency levels between efficiency drivers (Table 3). Although the positive impacts of hub and regular flight operations on physical infrastructure efficiency levels were not found to be

Sig.

International Yes

No

0.00 0.00 0.00

0.09 0.16 0.51

0.03 0.25 0.21

Sig.

Hub

Sig.

Yes

No

0.07 0.04 0.00

0.18 0.23 0.70

0.05 0.18 0.31

0.01 0.30 0.03

significant; regular flight operation, metropolitan area location, hub operation, and international status present a significant, positive impact on flight consolidation efficiency levels. International status negatively impacts physical infrastructure efficiency levels, probably due to the lack of discipline required by internationalization, due to flight delays and other additional requirements that consume capacity. Acknowledgements The author would like to thank the editors and the reviewers for their helpful comments on this paper. This research was supported by FAPERJ (Fundação Carlos Chagas Filho de Amparo à Pesquisa do Estado do Rio de Janeiro) e Project ID: E-26/103.286/2011.

P.F. Wanke / Journal of Air Transport Management 31 (2013) 1e5

References Cooper, W.W., Li, S., Seiford, L.M., Thrall, R.M., Zhu, J., 2001. Sensitivity and stability analysis in DEA: some recent developments. Journal of Productivity Analysis 15, 217e246. Liang, L., Cook, W.D., Zhu, J., 2008. DEA models for two-stage processes: game approach and efficiency decomposition. Naval Research Logistics 55, 643e653.

5

Wang, W.-K., Lu, W.-M., Tsai, C.-J., 2011. The relationship between airline performance and corporate governance amongst US Listed companies. Journal of Air Transport Management 17, 148e152. Wanke, P.F., 2012. Capacity shortfall and efficiency determinants in Brazilian airports: evidence from bootstrapped DEA estimates. Socio-Economic Planning Sciences 46, 216e229. Zhu, J., 2011. Airlines performance via two-stage network DEA approach. Journal of CENTRUM Cathedra 4, 260e269.