Intermodal competition and temporal interdependencies in passenger flows: evidence form the Emerald Coast

Travellers commonly perceive available transport modes (e.g. planes and ships) as substitutes, forcing transportation providers into competition for the same routes. Existing research in the literature on intermodal competition focuses on the determinants of traveller preference and the effects of intermodal competition on prices and quality of service. Another strand of transport literature focuses on time series analysis of passenger flows. Most of these studies analyse univariate or multivariate air passenger arrival time series using monthly data and, to a lesser extent, higher-frequency data such as daily time series. Very little attention has been paid, however, to the connection between the two lines of research, i.e. intermodal competition and time series analysis of passenger flows. In order to start filling this gap, we develop a Threshold-VAR (TVAR) regression model that simultaneously takes into account the intertemporal interdependencies between the airport and seaport arrival time series generated by the intermodal competition and the existence of (possibly) multiple time regimes due to the temporal heterogeneity in traveller preferences. Our analysis focuses on daily passenger arrivals at Olbia (Italy) airport and seaport from 2005 to 2008. Olbia is a major regional logistics hub, and the most populous city of North-eastern Sardinia, the second largest island in the Mediterranean Sea. Situated on the Emerald Coast, Olbia is a well-established elite tourist destination, and one of the most exclusive seaside resorts of Europe. This island provides an ideal setting to test the implications of intermodal competition for passenger arrival time series, since travellers can only arrive by air or sea, meaning these two modes of transport are (at least partially) substitutable and in competition with each other. Additionally, the distribution of traveller preference for mode of transport is likely characterized by a certain degree of temporal heterogeneity across the year, since tourists typically travel during the summer period. Our results show that intertemporal interdependencies and multiple time regimes do indeed exist. In particular, we find negative correlations in the two time series both within (autocorrelation) and across (cross-correlation) transport modes when the level of demand is low (i.e. in the low-season regime) and positive correlations (at one lag) when the level of demand is high (i.e. in the high-season regime). In addition, during the high season, the strongest correlation is observed within the mode (instead of across modes). These empirical results are interpreted within a conceptual framework that includes key factors such as traveller preferences for both mode of transportation and departure date, and capacity constraints in the transport mode.