152-2013: Using SAS® to Measure Airport Connectivity: An Analysis of Airport Centrality in the US Network with SAS/IML® Studio

The U.S. Federal Aviation Administration (FAA) estimates that $52.2 billion will be available over the years 2011– 2015 to fund airport infrastructure developments. Because one of the main objectives is to reduce congestion and delays, there is a need to acknowledge the importance of connectivity (measured with a centrality indicator) when establishing funding priorities. Currently, the FAA does not do this. In this paper, we exploit the capabilities of SAS/IML® Studio to implement a range of centrality measures, construct a graphical representation of the U.S. air transport network from airline ticketing data, test the algorithms to identify hub airports, and study the evolution of these indicators during the last decades in order to analyze the impact of airline decisions on airport connectivity. INTRODUCTION Airport classification and benchmarking is typically used for both policy and management purposes. In a context of centralized capacity development, one of the crucial aspects is the measurement of airport connectivity, especially when capacity expansions aim to reduce congestions and delays within the domestic network (e.g. the Airport Improvement Program run by the US Federal Aviation Administration). From a social perspective it seems reasonable that funding priority should be given to airports playing a central role in the network, not just because they process a significant proportion of nationwide traffic but also because passengers and airlines are connecting through them to other destinations. Hence, there is a potential for optimizing the social benefits from any public investment by introducing connectivity considerations in regulatory airport classifications. However, despite the significance of this issue With this background, we aim to test the sensibility of several indicators to airline de-hubbing in order to assess their suitability to characterize airport connectivity. To that end, this paper uses all the available data on passenger demand from the US Department of Transportation to perform a time-series analysis of airport hubbing patterns in the US domestic network between 1993 and 2012.The well-known indicator of flow-centrality is adapted from its original social network setting to an air transport context and used to develop a novel measure of each airport’s contribution to the network in terms of actual connectivity. The final indicator is directly proportional to the number of transit passengers going through each airport, and inversely proportional to the total number of passengers in those same markets. A survey of high-profile de-hubbing cases that occurred in the US during the last decades is obtained from the previous literature and the individual cases are analyzed. Besides our flow-based centrality, results are presented for other indicators that have been used in the same context such as degree centrality and betweenness centrality. The sensibility of the different indicators is established by comparing the temporal evolution of the connectivity measures immediately before and after the documented de-hubbing process. A suitable indicator should present a significant decrease in the airports’ degree of connectivity. From a methodological perspective, results are expected to establish a clear difference between the concepts of airport “hubbing” and “centrality”. From a policy perspective, results can be useful to improve airport classification and benchmarking within a centralized capacity management context. Finally, from a managerial perspective, results provide new insights on airport recovery patterns, not only after airline de-hubbing, but also after natural disasters or major industrial actions. , the existing literature does not provide an established approach to measure airport connectivity, and the choice of an appropriate indicator is still an unresolved question. This paper is therefore an extension of the work presented by Rodriguez-Deniz (2012) at the last SAS Global Forum in Orlando, Florida. On that occasion, a market-based variation of the betweenness centrality index, implemented using SAS/IML modules, was presented and tested for the US airports using a single-year sample, as an alternative to the airport classification criteria used by the FAA in their National Plan of Integrated Airport Systems (FAA, 2011). Clearly, the scope of the present work is wider in both methodological and applied terms. Consequently, and given the amount of data involved in the calculations (flight coupon data for nearly two decades), we chose SAS, particularly SAS/IML Studio, as the leading tool for the accomplishment of our goal, once again. SAS/IML Studio 1 The FAA estimates that $52.2 billion will be available over the period 2011-2015 under the Airport Improvement Program (AIP). Operations Research SAS Global Forum 2013