Market segmentation and dynamic price discrimination in the U.S. airline industry

Airfares are affected by a variety of factors, but it is less clear which factors are the key determinants and how they interact. Based on a unique transaction level data set, this paper introduces a widely used, machine learning based pricing tool to investigate the airline market segmentation and dynamic price discrimination problems. The empirical results suggest that purchasing time, city distance, market structure, market size, and seat availability are the five most important pricing factors in order. Airlines first partition their markets into an early market and a late market, and split the market further by city distance and other factors. While intertemporal price discrimination explains the majority of fare variations, there are strong indications that airlines use their market power in the late market and charge higher fares on late-arriving consumers (but not on early consumers), in response to extra seats sold.