Social cost of airline delays: Assessment by the use of revenue management data

We develop a schedule delay model using advanced econometrics techniques, while computing the social cost of passenger delays, in order to propose an alternative to the method of defining such cost that is actually found in the literature. This study has been made possible by obtaining private revenue management and operational data directly from a main European legacy airline. Thanks to MNL models, we compute the arrival time preferences, for all groups of passengers, namely arriving or departing, and short or long stay passengers, and for every day of departure that we have collected in our dataset. Then, we compute the effect of the ex-post modification of the arrival time, created by a delay, on the passenger's utility. We finally compute two effects on the social welfare cost of delays regarding ex-post schedule displacement: the price effect, which is the sensitivity of passengers regarding schedule displacement, and the volume effect measured by the level of airline delays in terms of time. Then, we compare those two effects across the day, in order to consider the social cost of airline delays as a two-effect phenomenon, instead of being only defined by the aggregated airline delay, in terms of time. Our results show that delays are costlier for passengers making the departing portion of their journey, with a short stay upon arrival, and during certain periods of the day. The originality of our paper principally lies in two aspects: first, studies about the precise computation of the social cost of airline delays regarding ex-post schedule displacement do not exist in the literature. Our paper fills this gap by providing an original method aiming at defining this cost. Second, our study is, to the best of our knowledge, the first to consider the effect of airline delays on the passenger's utility, at the time of day level, by jointly analyzing revenue management and operational data.