Airline and Alliance Networks

Abstract

Increased market deregulation and the accompanied rise of LCCs over the last decades has put profit for many of the old FSCs under pressure forcing them to merge and form alliances. This sparked the research into finding the most efficient structure for a single airline network in terms of profit and passenger mobility. Along with this, the social and economical dependancy on air transport grew and hence the need to assess the robustness of the network rose. Complex network theory offers a way to assess the efficiency of the networks using amongst other the degree (distribution), the betweenness, the average path length and the clustering coefficient. The main focus of current literature is on the analysis of global and regional airport networks, with limited coverage of separate airline networks and codeshare and alliance formation. Furthermore current research uses very standardized methods of assessing robustness and more realistic assumptions are needed. The first aim of this study is to get an insight into the differences in structure of FSCs and LCCs by analyze the topology and robustness of 17 European separate airline networks, using complex network theory. The second aim is to investigate the influence of codeshare and alliance formation on the topology and robustness of ATNs. Finally the third aim is to improve the methods used to analyze the robustness of ATNs. First the topology of the seperate airline networks of both FSCs and LCCs is analyzed in order to distinguish between the business solutions used by the airlines (PPs and HSs). Additionaly the influence of using codeshares and an alliances on the topology of the airline networks is investigated. This is performed by using complex network indicators to compare the seperate and combined (both codeshare and alliance) network layouts. The analysis confirms literature regarding FSCs, which turn out to use SFN associated with HS. LCCs however, are found not found to have RN associated with PP as suggested in literature, but a SFN with multiple interconnected hubs. The most important difference found between FSCs and LCCs is that LCCs tend to focus on diversity of destinations over frequency, whilst FSCs tend to focus on frequency over diversity. Combining networks into codeshare networks or the Skyteam alliance, increases the diversity of the network, the size and number of hubs and brings the behaviour closer to LCCs, however still with a focus on frequency over diversity. After this the synthetic static robustness of the seperate airline, codeshare and the Skyteam alliance networks is investigated in order to distinguish between the robustness behaviour of both FSCs and LCCs. The link between the complex network indicators and the synthetic static robustness of the ATN is also explored. This is performed by simulating error and attack on the separate airline networks and the codeshare and Skyteam alliance networks. Error is based on the random removal of airports from the network, while attack is based on the consecutive removal of nodes based on the heights of the degree, seat strength and (weighted) betweenness of the airports. The analysis confirms literature regarding FSCs, which shows low robustness against attack and high robustness against error. LCCs again show similar behaviour as the FSCs, contradicting literature, but confirming the results from Chapter 3. The shape of the curve of the cumulative degree distribution can be directly linked to the robustness independant of the size of the network. The higher the amount of hubs (with relative high degree), the higher robustness against attack. The robustness against error is much higher and similar for all networks. Combining networks into codeshare netrworks or the Skyteam alliance, will thus increase the robustness against attack. Finally the robustness analysis of ATNs is improved by introducing new methods of simulating more realistic error and attack scenarios. The link between the realistic robustness analysis and the synthetic robustness analysis is also investigated Three different phenomena are simulated: weather, strikes and volcano eruptions. Weather and volcano eruptions are simulated using the introduced geographic attack. Geographic attack is based on starting at an initiation airport and removing the other airports using geographic radial spreading. Strikes are simulated using the geographic degree, which groups the airport into FIR. The analysis puts the synthetic robustness in perspective. It shows that not only the number of hubs is important in order to improve the robustness of an ATN, but also the geographic spreading of the hubs.