Air connection dropouts and isolation risks across European regions

ABSTRACT

This work focuses on the duration of air connections in Europe and investigates the main factors affecting the probability that specific air routes survive over time. Socio-economic and geographical characteristics as well as modal competition, public subsidies and other airport/carrier-related factors are investigated. Moreover, we derive from our analysis the probabilities of connection dropouts to characterize European regions on the basis of their expected risks of increasing isolation. Our results might be of particular interest for policymakers and airport managers who intend to prevent their regions/airports from being affected by the negative consequences of increasing isolation. This is especially true in a context of expected increasing consolidation by major European network carriers.

KEYWORDS:

• air connections
• isolation risks
• routes' survival
• European regions

JEL:

• C41
• L93
• O18
• R58

DISCLOSURE STATEMENT

No potential conflict of interest was reported by the authors.

Notes

1. Airline consolidation has mainly centred in the United States, where the trend started in 2005 and continued after the global financial crisis with four further major deals up to 2013 (e.g., United Airlines/Continental Airlines in 2010; and Southwest Airlines/AirTran Airways in 2011). Despite some relevant deals (e.g., Air France/KLM, British Airways/Iberia), the European airline industry has lagged behind, but experts agree on the fact that there will be more consolidation in the near future (The Telegraph, 2017).

2. This happened even to some ‘old’ US industrial cities such as Cincinnati, Pittsburgh, Memphis, St. Louis and Minneapolis, which were increasingly cut off from each other and from the global economy as a result of the airline consolidation process (Washington Monthly, 2012).

3. For example, when the decision to exit is driven by competitive pressure, the route is kept in operation by the remaining competitors.

4. Survival and duration in airline alliances is studied by Gudmundsson and Rhoades (2001), while the survival odds of new entrants in the liberalized intra-European airline industry is investigated by Fan (2009).

5. Indeed, the hazard rate $h(t)$ can be rewritten as $h(t)=\frac{f(t)}{S(t)}$, where $f(t)$ is the probability density function of the failure function, that is, $f(t)={\displaystyle \frac{\mathrm{\partial }F(t)}{\mathrm{\partial }t}=-{\displaystyle \frac{\mathrm{\partial }S(t)}{\mathrm{\partial }t}}}$. Therefore, by estimating $h(t)$, it is possible to estimate the related survival function. See Lewin et al. (2018) and Litzel (2017) for more technical details.

6. Iceland, Switzerland and Turkey are not included in the analysis due to missing data in the Eurostat database.

7. The choice of including socio-economic characteristics at the NUTS-2 level is due to the fact that the attractivity of an origin/destination may be influenced significantly by the characteristics of its surrounding areas. Therefore, including these data at the NUTS-3 level may be too limiting. On the other hand, the classification of remoteness is mainly geographical, and therefore it is reasonable to include information at the NUTS-3 level.

8. We thank an anonymous referee for this suggestion. The choice to use a dummy variable is made also to avoid the problems that may arise because of the tourism contribution to GDP. Results do not change significantly when the percentage is slightly increased or reduced.

9. We reduce the measure of remoteness to three classes: identifying urban regions scored 1 (predominantly urban regions), regions close to a city scored 2 (intermediate or rural regions close to a city), and remote regions scored 3 (intermediate or rural remote areas).

10. European PSOs are similar to the essential air service (EAS) US government programme.

11. The list has been integrated through desk research based on existing literature and reports.

12. The HSR variable is time-varying since we included any line starting to operate during the period considered.

13. As correctly suggested by an anonymous referee, the presence of LCCs can be considered, on average, a proxy of taxes, given that these carriers tend to choose low-tax airports in Europe (even if with some exceptions). Results do not change significantly when the percentage is set to 40% or 60%. We thank the same referee for the suggestion to consider airport taxes in the analysis.

14. Network carriers and LCCs offering the highest number of seats on European routes are considered.

15. For example, Antunes et al. (2020) find evidence of the fact that European NUTS-2 regions that have highly connected neighbours tend to exhibit lower air connectivity levels.