Summary of the article written by Joanna Stavins,
Price Discrimination in the Airline Market: The Effect of Market Concentration

The article is based on the assumption of the price discrimination increases with the competition in the airline market.
Introduction:
The relationship between price discrimination and market concentration has already been studied by Dana (1998) and Gale&Holmes (1993); the first states that PD1 could be present even if MC2 is low and there is no particular strength of a carrier. Gale&Homes point out that monopoly airline may want to avoid all those customers with high valuation of time.
Purchase discount and Saturday-night stayover are the more used techniques to price discriminate the airline market.
The model implemented calculates the differences in marginal effects of ticket restriction over the main routes characterized by different levels of MC.
A negative effect of MC on price dispersion has been discovered by Borenstein and Rose in their 1994 study.
An effective way to discriminate the nature of their customer is to attach some ticket restrictions to the cheaper one, causing the Higher Type of customer to dislike those kinds of fares.
Dana highlighted that the discrimination techniques described above mainly affect the consumers with heterogeneous and uncertain demand
Data:
A certain date has been chosen to be surveyed, September 28 1995, All data comes from Official Airline Guide. That day was a Thursday, so all the holiday or weekend travellers are just a small minority.
It is useful to notice that the MC is low due to the small number of routes (12) included in the 5804 tickets available.
To test for robustness, we use two different ticket restrictions in the estimation, and both deliver the same results.
The restriction that could be attached are: 1. Cancellation Penalty 2. Number of Days in Advance that Purchase is Required 3. Saturday Night Stayover Required or Not 4. Other (Unspecified)
To avoid multicollinearity, one restriction at a time was included in the estimation, according to Dana and Gale&Holmes, the restriction number 2 and 3 were used as proxies for price discrimination.
The number of passenger is not included in the data, so, in order to carry out each carrier’s market share and the Herfindahl-Hirschman index (HHI) the number of passenger has been substituted with the numbers of route.
Model:
Two different equations have been estimated, leading to a reduced-form regression model.
The first equation doesn’t show any interaction between restriction and concentration while the second one allows for a separate effect of the restriction-concentration interaction on airfares. lnPijk=β0+β1Rijk+β2HHIi+β3Sij+β4DISTi+β5DISTSQi+β6AVGPOPi+β7AVGINCi+β8TEMPi+β9HUBij+β10SLOTSi+β11ONEWAYijk+β12DIRECTij+β13FIRSTijk+β14DAYSijk+ϵijk In both equations * i denotes route * j denotes carrier * k is a particular ticket for the carrier on the route * DIST and DISTQ are the distance between two endpoints and the distance squared * AVGPOP is the avg population and AVGINC is average per capita income * TEMP is the difference in mean January temperatures between the origin and the destination * HUB is boolean variable, 1 if carrier has a hub and 0 if not * SLOTS is boolean variable, 1 if the number of takeoff and landing slots at either airport is regulated * ONEWAYis boolean variable equal to 1 for one-way ticket * DIRECT is a boolean variable equal to 1 for direct flights * FIRST is a boolean variable equal to 1 for first class flights * DAYS indicates the number of days prior to departure that the fare was last offered
The second equation is the following lnPijk=α0+Rijk(γ0+γ1HHIi)+α1HHIi+α2Sij+α3DISTi+α4DISTSQi+α5AVGPOPi+α6AVGINCi+α7TEMPi+α8HUBij+α9SLOTSi+α10ONEWAYijk+α11DIRECTij+α12FIRSTijk+α13DAYSijk+νijk From this equation we get price discrimination estimated as:
∂P∂R= γ0+γ1HHIiP