Asymmetric regulation and airport dominance in international aviation: evidence from the London-New York market

Southern Economic Journal, Oct, 2007 by Volodymyr Bilotkach

My task is to identify the airport dominance and the regulation effects. For this, I will need to difference out the airline effect and the nonstop flight effect. To be able to do so, I restrict the conditional mean fare function for CO in the following way:

E ([P.sub.LONNYC] | Airline = CO, NS) = [[delta].sub.CO] [[delta].sub.NS] [[delta].sup.CO.sub.Reg] [[delta].sub.Hub]. (1)

Comparison Carrier--Nonstop London-New York Flights

For some other carrier on the same city-pair market (the comparison carrier), the conditional mean fare function will be

E ([P.sub.LONNYC] | Airline, NS) = [[delta].sub.Airline] [[delta].sub.NS] [[delta].sup.Airline.sub.Reg], (2)

where [[delta].sub.Airline] is the airline effect, again assumed to be the same for all flights of the comparison carrier, regardless of the number of stops; and [[delta].sub.NS] is the same across carriers and airport-pair markets. On the other hand, [[delta].sup.Airline.sub.Reg] varies across carriers (yet, it can be easily normalized to zero for the comparison airline).

Getting Rid of the Nonstop Flight Effect

Given Equations 1 and 2, with AA as the comparison carrier and taking the difference,

[[DELTA].sub.NS] = E([P.sub.LONNYC] | Airline = CO, NS) - E([P.sub.LONNYC] | Airline = AA, NS) (3)

obtains

[[DELTA].sub.NS] = [[delta].sub.CO] [[delta].sup.CO.sub.Reg] [[delta].sub.Hub] - [[delta].sub.AA] - [[delta].sup.AA.sub.Reg], (4)

that is, I am successfully rid of the nonstop flight effect.

Getting Rid of the Airline Effects

To difference out the airline effects, I can, given my assumption that it is the same for all flights of a given carrier, use fares for the London-New York segment of what I will call "through itineraries" (e.g., fare for the New York to London segment of the Nashville-New York-London trip). For CO, these fares will not include the hub effect, because it exists only for trips to or from (not through) the hub airport. I will also assume that regulation effects for through trips are the same for all carriers. The plausibility of this assumption depends on the source of the postulated asymmetric regulation effect. If it arises because of the differences in consumer preferences between LGW airports, my assumption is clearly not defensible. But, in this case, the resulting estimate of the difference in regulation effects will be biased upward. That is, if a negative value for the difference in regulation effects obtains in the end, I can be confident that, even if this estimate is biased as described, correction for this bias will make it even more negative. Thus, a possible violation of my identifying assumption will not move me to conclude that the negative regulation effect does exist where it does not, but it can lead me to claim that less favorable regulation does not negatively affect fares, when in fact it does. Given my assumptions, the airline effect is the only one I will have in the expected fare function for the London-New York segment of one-stop trips. Then, taking the difference