Choice of Aircraft Fleets in the US NAS: Findings from a Multinomial Logit Analysis

How does the passenger demand influence the choice of aircraft? Can we derive the choice of aircraft and fleet mix for origin and destination (O&D) pairs from knowledge of passengers' demand for scheduled air services? How do these demands affect the overall demand for air traffic services (i.e., en route, and terminal radar approach control (TRACON) facilities), in the short- and medium-term? The National Airspace System (NAS) in the United States (US) had an inventory of 5156 big jets at the end of December 2002, of which 4085 were narrow bodies, and 1071 can be classified as wide bodies. There were 1180 regional jets. In addition, there were 660 turboprops in the system at that time. Empirical research reveals that there is a critical link between the flow of scheduled passenger services and the choice of aircraft by the airlines in any O&D market pair. This relationship can be empirically retrieved without the detailed knowledge of airlines' behavior and used for analyzing the traffic patterns in the NAS. This is a natural segué from the econometric modeling of passenger demand [Bhadra (2003)]. Although the demand for scheduled passenger services provides important information, it cannot be directly used to generate demand for air transportation management (ATM) services. Hence, the empirical linkages between demand for scheduled air services and the demand for aircraft fleets by O&D pairs will have to be established. This paper is an attempt to establish this empirical linkage. The fleet mix in O&D market (T100 market) and segment pairs (T100 segment) of Form 41 are the primary data used for this work. Using the T100 market and segment data from the latter part of the last decade (1995-2002), we build multinomial qualitative choice models, e.g., logit choice method. In this paper, we use two sample periods, 2002: month 3; and 2002: month 6 to demonstrate empirical relationships between aircraft choice and passengers, distance, and types of airports. This framework establishes empirical linkages between aircraft choice, six categories based on all observed equipment types in the system, and passenger flows in addition to distance, and types of airport hubs. Estimated models demonstrate that both passengers and distance play important roles in selecting types of aircraft. Using the estimated coefficients from the qualitative econometric choice model and varying assumptions (i.e., number of passengers in particular), we can easily generate forecasts of aircraft choices for O&D pairs and the fleet mix. This can, then, be used to derive demand for ATM services and distribution of the TRACON facilities and en route workload.