Directional Price Discrimination. in the U.S. Airline Industry

Transcription

1 Evidence of in the U.S. Airline Industry University of California, Irvine June 21st, 2017

2 Summary First paper to explore possible determinants that may factor into an airline s decision to charge passengers different roundtrip fares depending on trip origin Why do these differences exist? Cannot be due to differences in cost, as the cost of flying a roundtrip passenger does not significantly differ depending on direction flown Incentive for airlines to directionally price discriminate if passenger price elasticities of demand differ between route endpoints Findings: Significant evidence of directional price discrimination based on income, with passengers at high-income origins charged higher fares.

3 Roundtrip Fare Differences Exist

4 Example 1: United s Average Roundtrip Fares on San Francisco (SFO) - Chicago (ORD) $600 United Passenger Weighted Average Fare $500 $400 $300 $200 $ $ $ $ $ $ SFO Origin ORD Origin $ $ $100 $ Quarter

5 Example 2: Virgin America s Average Roundtrip Fares on San Francisco (SFO) - Chicago (ORD) Virgin America Passenger Weighted Average Fare $500 $400 $300 $200 $100 $ $ $ $ $ $ SFO Origin ORD Origin $ $ $ Quarter

6 Previous Literature on Price Discrimination in the Airline Industry Focused on the various mechanisms used by airlines to segment passengers with different price elasticities of demand 1 Advance-purchase restrictions effective in segmenting customers by their value of time (Gale and Holmes (1993), Dana (1998)) 2 Saturday night stay, minimum stay, and non-refundability of tickets designed to discourage price-inelastic consumers from buying cheaper tickets (Stavins (2001)) 3 Fares purchased on weekends 5% lower (Puller and Taylor (2012))

7 Borenstein and Rose (1994) Examine price dispersion on a route basis Expected absolute deviation in fares an airline charges to two different passengers on the same route found to be 36% of the airline s average ticket price Extent of price dispersion increases as the number of competitors in a market grows Present paper differs by examining price dispersion on a directional basis JFK-LAX roundtrips treated differently than LAX-JFK roundtrips

8 Income Discrimination? Airline wishing to maximize revenue may income discriminate by charging passengers beginning roundtrip travel in the higher income locality a higher average fare (3rd degree price discrimination) Passengers residing in the higher income area must have a lower price elasticity of demand for air travel relative to passengers residing in the lower income area Rich passengers less sensitive to the cost of travel

9 Income Differential and Demand Income difference cause a demand differential? Unlikely to affect fares given: Substantial level of competition in the U.S. airline industry Airlines facing constant costs in delivering seats to a market Possible for demand to affect fares in markets where airlines operate out of: Congested airports Slot-controlled airports

10 Income Differential and Yield Management How could differences in income between route endpoints be reflected in an airline s yield management strategy? Simple Yield Management example: 156 available seats Airline allocates seats to either a discount or full fare bucket Discount fare is $100 and full fare is $200 Problem is to choose the maximum number of seats sold at the discounted fare ( booking limit ) to maximize revenue

11 Yield Management Example

12 Yield Management Continued What if an airline chooses a booking limit that depends on direction? Income discrimination arises if the airline lowers the booking limit for passengers beginning roundtrip travel in the metro area with the higher average income Consider the SFO-ORD route. Suppose booking limit is 100 in each direction Average roundtrip fares are $ What if booking limit is lowered to 80 for passengers beginning at SFO but kept at 100 for passengers at ORD? Average roundtrip fares are $ for passengers originating at SFO and $ at ORD

13 Population Differential and the Price Elasticity How can a difference in population between route endpoints be reflected in average fares? May impact price elasticity if passengers originating travel at more populous endpoints benefit from greater flight frequency Larger metro areas demand more air travel

14 Differences in the Mix of Traveling Passengers What if the proportion of business and leisure travelers on a route differs by point of origin? Roundtrip fares will be cheaper in the direction where the mix of passengers is more likely to contain a higher proportion of price-elastic leisure travelers Examples: Routes into Las Vegas, Orlando ect.

15 Fare Data US Department of Transportation s Airline Origin and Destination Survey (database DB1B) 10 percent random sample of all airline tickets that originate in the United States on U.S. based carriers Released quarterly Data used to determine air traffic patterns, air carrier market shares and passenger flows Limitations Don t know date of purchase, date of travel, or time of day Ticket Restrictions?

16 Data Processing Analysis sample spans all four quarters of 2015 Only nonstop roundtrip travel in coach class between airport-pairs in the US is included Roundtrip fares less than $25 excluded (similar to restriction in Brueckner, Lee, Singer (2013)) Tickets involving a connection excluded to directly compare how legacy and low-cost carriers differ in their directional pricing of nonstop roundtrip flights Recoding of operating carrier variable for regional carriers that are owned or have a partnership with a major airline

17 Weighted Average Fares Each itinerary in the DB1B data contains a passenger count corresponding to the number of sampled passengers observed paying that particular fare for travel on a given carrier during the quarter Different observed fares on a given carrier for travel in the same market generate separate observations Using passenger counts as weights, observations are aggregated to the origin-destination-carrier-quarter level to produce a weighted average fare Resulting weighted average fares are compared with the corresponding weighted average fares for roundtrip travel in the opposite direction on the same carrier

18 Model Observation is the passenger-weighted average fare for nonstop roundtrip travel from origin airport i to destination airport j on carrier k in quarter l. To analyze the directional effect on roundtrip fares, this weighted average fare is subtracted from the weighted average fare for direct roundtrip travel from origin airport j to destination airport i on the same carrier k in the same quarter l. FARE ijkl FARE jikl = β 0 + β 1 DINC i,j + β 2 DPOP i,j + β 3 DTEMP i,j +ɛ ijkl (1) ln(fare ijkl ) ln(fare jikl ) = β 0 + β 1 DLINC i,j + β 2 DLPOP i,j +β 3 DLTEMP i,j + ɛ ijkl (2)

19 Levels Regression DINC DPOP DTEMP Per $1,000 difference in per capita income between airports i and j 2014 estimates at the MSA level from the BEA Difference in total population (in millions) between airports i and j 2014 estimates at the MSA level from ACS Difference in the average quarterly high temperatures between airports i and j Airport weather data from NOAA

20 Analysis Sample Observation included only if at least 100 sampled passengers are observed traveling in each direction during the quarter To prevent double counting by including both ij and ji observations, an observation is included only if the origin airport code comes after the destination code in the alphabet. 6,944 observations encompassing 1,406 city-pair markets

21 Summary Statistics

22 Base Regression Results

23 Interpretation for SFO-ORD Fares are $0.32 (.055%) higher on average for each $1,000 difference in income between the endpoint cities SF per capita income $72,364, Chicago $50,690 $21,674 difference predicted to be $6.91 (1.64%) higher on average for SFO originating passengers How much additional revenue did United generate from SFO passengers? 8,694 sampled roundtrip passengers traveling on United with an origin at SFO implies 86,490 total RT passengers in 2015 additional $597,992 in revenue Similar calculation across all routes with positive fare differences $160 million

24 Potential Threats to the ID of the Income Discrimination Effect

25 Additional Controls & Interpretation for SFO-ORD Fares are $0.19 (.043%) higher on average for each $1,000 difference in income between the endpoint cities How much additional revenue did United generate from SFO passengers? 8,694 sampled roundtrip passengers traveling on United with an origin at SFO in 2015 additional $361,795 in revenue

26 Legacy vs. Low-Cost Carriers

27 Interpretation For each $1,000 difference in average per capita income between endpoint metro areas: Low-cost carriers fares are $0.29 ($0.45) higher Legacy carriers fares are $0.08 ($0.08) higher Potential explanation: Elasticities display more variation among the low-cost passenger base than among the legacy passenger base. Low-cost carriers exploit this variation to income discriminate more among their price-elastic customer base than legacy carriers.

28 Effects by Time of Year

29 Impact of Route Competition

30 Interpretation for SFO-ORD Coefficients of $0.052 on DINC*COMPETITORS & $0.097 on DINC in Column 2: 1 Three competitors on Route (American, United, Virgin) 2 Roundtrip fares $5.51 higher for SFO originating pax Addition of adjacent competition variables has little impact on the overall magnitude of the income discrimination effect

31 Conclusion Paper provides first exploration of an airline s decision to charge passengers different roundtrip fares on a nonstop route depending on trip origin Evidence found of directional price discrimination based on income, with passengers at high-income origins charged higher fares Issues & Possible Extensions What about time of day, day of week, peak vs off-peak? Connecting routes? International Routes?

32 Thank You for Listening! QUESTIONS OR COMMENTS?