Using Online Data to Estimate Flight-Level Price Elasticities

Click to enlarge

Sponsors:

National Science Foundation CAREER award SES-0846758

Researchers:

Matt Higgins

Students:

Stacey Mumbower

Network-planning models are used to forecast the profitability of airline schedules. Airlines use these models to evaluate the costs and revenues associated with network design decisions and strategic planning initiatives (e.g., equipment purchase decisions, minimum connection time studies, code-share and joint-venture scenarios). The majority of network-planning models use discrete choice models to represent how customers select itineraries by making trade-offs among carriers, time of day, level of service, and price. Network planning models use comprehensive schedule and passenger booking data, but do not have accurate information on the actual fares and itinerary choice sets faced by consumers at the time of booking. As a result, current network-planning models may not incorporate realistic measures of individuals’ price sensitivities. In an era dominated by online bookings and low cost carriers, it has become critical for airlines to investigate whether new data sources can be used to provide more accurate price elasticity estimates.

To investigate this issue, former doctoral student Stacey Mumbower collected airline pricing, itinerary, and seat map data from the internet and are using that data to model how customers make choices among several itineraries. The results of the models provide information about how departure time of day, pricing, and seat availabilities (how many and which seats are available at a given point in time) influence customer decisions. The data collected is much more detailed than any other data currently available to researchers (and airlines). 

Another interesting piece of this research explores an issue termed “endogeneity” of price. Airline prices influence demand, and demand in turn influences prices. This simultaneous influence of price and demand leads to endogenous prices, an issue well known and researched in economics literature. In economics, there are several methods that can be used to correct models that suffer from endogenous prices. After correction, the models give much more realistic results. Without correction, the model coefficients will be biased, which could potentially influence predictions and policy decisions. In the worst case scenario, this bias may be so large that models predict raising prices will lead to more demand! Although the endogeneity issue has been well documented and researched in the economics field, this issue has been almost completely ignored in airline research.

As part of this research, Mumbower estimated the first flight-level price elasticities from online data that corrected for price endogeneity. In contrast to market-level and route-level elasticities reported in the literature, flight-level elasticities can forecast responses in demand due to day-to-day price fluctuations. Knowing how elasticities vary by flight and booking characteristics and in response to competitors’ pricing actions allows airlines to design better promotions. It also allows policy makers the ability to evaluate the impacts of proposed tax increases or time-of-day congestion pricing policies.

Publications resulting from this research

  1. Mumbower, S., Garrow, L.A., Higgins, M.J. (2014). Estimating flight-level price elasticities using online airline data: A first step towards integrating pricing, demand, and revenue optimization. Transportation Research Part A 66: 196-212.
  2. Mumbower, S. and Garrow, L.A. (2014). Data set: Online pricing data from multiple U.S. carriers. Manufacturing and Service Operations Management 16(2): 198-203.
  3. Mumbower, S. and Garrow, L.A. (2010). Using online data to explore competitive airline pricing policies: A case study approach. Transportation Research Record 2184: 1-12.