The geography of airfares: modeling market and spatial forces in the U.S. Airline Industry

The deregulation of the airline industry created a myriad of changes in the U.S. air transport system that has both defended and sparked debate on the wisdom of such policy change for over three decades. One of the promises of deregulation from its proponents in the 1970s was increased competition that would lead to a reduction in fares for consumers. Historic data and literature has indeed shown this to be to the case as average airfares have trended downward especially over the last twenty years. Nonetheless, the industry has become much more complex since deregulation in terms of pricing to the point that very sophisticated yield management computer models are used to achieve an optimum balance between load factors and price. Consequently, this has in turn translated into a haphazard experience for most air travelers in the United States; for instance, the cost of a ticket is sometimes lower traveling from coast to coast than within a particular region of the U.S. and paid fares for the exact same trip can deviate dramatically, often based on variation in the date of purchase. Additionally, this has also resulted in a spatial pattern where certain regions throughout the country have enjoyed lower airfares more so than others. This research seeks to identify this regional disparity using a geographically weighted regression and spatial autoregressive models in a sample of 6,200 routes between 80 primary U.S. airports. The results from the global model showed that variables which measure competition (airlines), operating cost (flights, distance) and elasticity (layover time) proved to be statistically significant and had a positive relationship with airfare The GWR results indicated that while some factors like distance, and hub size, were statistically significant almost nationwide, other factors such as frequency, presence of low cost carriers, and numbers of airlines were only statistically significant at certain airports. Finally, the spatial regressions models indicate that the spatial autocorrelation found in U.S. airfares resemble the first order properties of spatial autocorrelation (i.e. spatial heterogeneity) and not the second order properties (i.e. spatial dependence).