jup presentation, winter '98, faa tech...

M I T I n t e r n a t i o n a l C e n t e r f o r A i r T r a n s p o r t a t i o nM I T I n t e r n a t i o n a l C e n t e r f o r A i r T r a n s p o r t a t i o n

MIT MIT ICAT ICAT

PRICING AND REVENUE MANAGEMENT RESEARCHAirline Competition and Pricing Power

Presentations to Industry Advisory Board Meeting

November 4, 2005

MIT MIT ICAT ICAT

2

PRESENTATIONS

“Pricing and Competition in Top US Markets” (Celia Geslin)

Fare, Traffic and Revenue Changes 2000 to 2004

“Impacts of Airline Fare Simplification” (Maital Dar)MIT PODS Research ConsortiumSimulations of Revenue and Traffic Impacts

“Adapting Revenue Management Systems” (Peter Belobaba)

Development of New Forecasting and Optimization Algorithms


MIT MIT ICAT ICAT

AIRLINE PRICING AND COMPETITION IN TOP US MARKETS

Célia Geslin

MIT MIT ICAT ICAT

4

Objectives and Approach

Preliminary analysis of airline pricing power in US markets:

How have air fares changed in domestic markets in the past 5 years? Differences by length of haul?Differences between LCC and non-LCC markets?

Empirical analysis of largest domestic markets

Top 100 US 2004 Markets from O&D Plus DataAggregate analysis and overall trends between 2000 and 2004Analysis by carrier and type of carrier (legacy, LCC)

MIT MIT ICAT ICAT

5

Average Fares in Top 100 US Markets

Fares continue to decrease. On average, fares were 19.3% lower in 2004 compared to 2000.

Average Fares - Top 100 Markets

$100

$110

$120

$130

$140

$150

2000 2001 2002 2003 2004

-8.4%-15.6% -14.7%

-19.3%

MIT MIT ICAT ICAT

6

Total Passengers in Top 100 US Markets

Passenger volumes have rebounded to 2000 levels after dropping by over 11%.

-8.5% -12.6% -10.4%-0.6%

Total Passengers per day - Top 100 Markets

120,000

125,000

130,000

135,000

140,000

145,000

150,000

2000 2001 2002 2003 2004

- 8.5% -11.7% -9.5%

+ 0.4%

MIT MIT ICAT ICAT

7

Total Revenues in Top 100 US Markets

Huge revenue drop of 25.4% by 2002. Slow recovery since then, but still 19% below 2000.

Total Revenues per day - Top 100 Markets

$10

$12

$14

$16

$18

$20

$22

2000 2001 2002 2003 2004

Mill

ions

-16.2% -25.4% -22.8% -19%

MIT MIT ICAT ICAT

8

Carrier Market Share Losses and Gains

Market share losses for network carriers, gains for LCCs – led by JetBlueSouthwest is MS leader in Top 100 Markets, in both 2000 and 2004

% Market Share Change 2000-2004 - Top 100 Markets

-4%-3%-2%-1%0%1%

2%3%4%5%6%7%

US Airw

ays

United

Airli

nesCon

tinen

tal A

irlines

Delta A

ir Lines

Southw

est A

irlines

America

Wes

t Airli

nesSpir

it Air L

ines

Jet B

lueM

arke

t Sha

re

BIGGEST LOSSES BIGGEST GAINS

MIT MIT ICAT ICAT

9

Market Share by Carrier Group

Overall, LCC group MS increased from 26% to 37%, while Legacy group MS dropped from 60% to 53%

% Market Share - Top 100 Markets

0.00%

10.00%

20.00%

30.00%

40.00%

50.00%

60.00%

70.00%

Legacy LCC Others

Mar

ket S

hare

20002004

MIT MIT ICAT ICAT

10

Fares by Distance Category

Average Fare comparison 2000-2004

$-

$50

$100

$150

$200

$250

$300

Short Haul Medium Haul Long Haul

20002004

+ 0.92%

- 21.34%

- 36.08%

Average fares have dropped by 36% in long haul markets, while short haul fares actually increased slightly compared with 2000.

MIT MIT ICAT ICAT

11

Passengers by Distance Category

Passenger traffic in short haul markets dropped 18%, while increasing 10-13% in medium and long haul markets

Total Passengers comparison 2000-2004

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

80,000


20002004 + 13.4%

- 17.8%

+ 9.9%

MIT MIT ICAT ICAT

12

Revenues by Distance Category

Total Revenues decreased most in long haul markets despite traffic growth – down 27% overall

Total Revenue comparison 2000-2004

$-

$2

$4

$6

$8

$10

$12


Mill

ions

20002004

- 27.5%

- 17%

-13.5%

MIT MIT ICAT ICAT

13

Markets Grouped by LCC Presence

In 2000, 27 of Top 100 US Markets without LCC presence

By 2004, only 10 Top 100 US Markets without LCC presence (6 when Hawaii markets excluded)

84 of the Top 100 US Markets with more than 10% LCC MS

LCC Market Share Distribution ( 2000-2004)

27

7

65

10 6

84

0102030405060708090

LCC=0 LCC<10% LCC>10%

Freq

uenc

y

2000

2004

MIT MIT ICAT ICAT

14

Average Fares and LCC Presence

Average Fare decreased more for markets with a small 2004 LCC market share than the markets with well-established LCC presence.

Largest (31%) decrease in fares observed for markets with new entry by LCC between 2000 and 2004.

Average Fares Comparison (2000-2004)

$-

$50

$100

$150

$200

$250

LCC MS < 10% 2004 LCC MS > 10% 2004 LCC New Entry2000-2004

20002004

-20.4%

-17.8%

-31.3%

MIT MIT ICAT ICAT

15

Passenger Traffic and LCC Presence

Markets with LCC presence showed traffic growth of 4.51%

But in O&D markets with small or no LCC market share, traffic is still 16% below the 2000 level.

Total Passengers Comparison (2000-2004)

0

20,000

40,000

60,000

80,000

100,000

120,000

140,000

LCC MS > 10% LCC MS < 10%

20002004

+4.51%

-16.3%

MIT MIT ICAT ICAT

16

Conclusions: Top 100 Markets

Overall trends in largest US markets 2000-2004Traffic has rebounded to peak 2000 levelsBut average fares have dropped 19%, with a corresponding total revenue decrease

Major differences identified:By carrier type – Legacy carriers have lost 5% market share and over 9% revenue shareLong-haul market fares have dropped the most, with greatest traffic growth. On the other hand, short-haul traffic is down, and average fares stable. Substantially lower total revenues in all distance categories.Markets with LCC new entry saw the greatest drop in average fares between 2000 and 2004

MIT MIT ICAT ICAT

17

Future Research

Expand the sample to 500 or 1000 Top US Markets

Identify relevant factors in the evolution of pricing and competition in airline markets:

Length of haulLow-fare carrier competitionHub vs. non-hub markets

Broader questions include:How has willingness to pay (price elasticity) changed? Are people less willing to pay for air travel?How has airline pricing power been reduced? How can we quantify this effect?


MIT MIT ICAT ICAT

IMPACTS OF AIRLINE FARE SIMPLIFICATION

Maital Dar

MIT MIT ICAT ICAT

19

PODS RM Research Consortium

Airline revenue management research at MIT funded in large part by PODS Research Consortium

Focus on forecasting and optimization models for seat inventory control (seat allocation)Findings used to help guide each airline’s RM system development

Most member airlines have renewed; new member added in 2005

Continental Airlines Lufthansa German AirlinesScandinavian Airlines System Northwest AirlinesDelta Air Lines KLM/Air FranceAir New Zealand LAN Airlines (new)

MIT MIT ICAT ICAT

20

Tumbling Airline Revenues

Fares have been decreasingThe lower fares are due in part to LFA competition, but not exclusivelyRM system shortcomings are also involved

Passenger choice process has changed, but RM systems have not

Airline customers have learned how to get cheaper fares, but existing revenue management systems in use largely don’t take this new reality into account

Traditional RM systems all based on:Identifiable and independent demand for different fare products with restrictions associated with lower fares

MIT MIT ICAT ICAT

21

BOS-SEA Traditional Fare StructureAmerican Airlines, October 2001

Roundtrip Fare ($)

Cls Advance Purchase

Minimum Stay

Change Fee?

Comment

458 N 21 days Sat. Night Yes Tue/Wed/Sat 707 M 21 days Sat. Night Yes Tue/Wed 760 M 21 days Sat. Night Yes Thu-Mon 927 H 14 days Sat. Night Yes Tue/Wed 1001 H 14 days Sat. Night Yes Thu-Mon 2083 B 3 days none No 2 X OW Fare 2262 Y none none No 2 X OW Fare

2783 F none none No First Class

MIT MIT ICAT ICAT

22

Simulation of Leg-Based RM Benefits Differentiated Fare Structure

R e ve n u e G a in W h e n B o th A irlin e s Im p le m e n t E M S R b

3 .8 5 %

8 .6 3 %

1 4 .7 4 %

2 .1 8 %

5 .7 2 %

1 0 .6 2 %

0 .0 0 %

2 .0 0 %

4 .0 0 %

6 .0 0 %

8 .0 0 %

1 0 .0 0 %

1 2 .0 0 %

1 4 .0 0 %

1 6 .0 0 %

E M S R b A L F = 7 8 % E M S R b A L F = 8 4 % E M S R b A L F = 8 9 %

A L 1 A L 2

MIT MIT ICAT ICAT

23

Fare Simplification:Less Restricted and Lower Fares

Recent trend toward “simplified” fares – compressed fare structures with fewer restrictions

Initiated by low-fare airlines in many parts of the worldEarly in 2005, implemented in all US domestic markets by Delta, matched selectively by legacy competitors

Simplified fare structures characterized by:Little or no minimum stay restrictions, but advance purchase andnon-refundable/change feesLower fare ratios from highest to lowest published fares, typically no higher than 5:1 in affected US domestic markets

MIT MIT ICAT ICAT

24

Example: BOS-ATL Simplified FaresDelta Air Lines, September 2005

One Way Fare ($)

Bkg Cls

Advance Purchase

Minimum Stay

Change Fee?

Comment

$124 T 21 days 0 $50 Non-refundable $139 U 14 days 0 $50 Non-refundable $199 L 7 days 0 $50 Non-refundable $224 K 3 days 0 $50 Non-refundable $259 Q 0 0 $50 Non-refundable $444 B 3 days 0 $50 Non-refundable $494 Y 0 0 No Full Fare $294 A 0 0 No First Class $594 F 0 0 No First Class

MIT MIT ICAT ICAT

25

LEG RM SIMULATIONS:Impacts of Fare Restriction Removal

2 carriers, single market, both use EMSRb leg RM controls6 fare classes, 3.5:1 fare ratio:

Class 1 2 3 4 5 6

Fare 425.00 310.00 200.00 175.00 150.00 125.00

BASE CASE: Restricted and Differentiated Fares

Fare Class AP MIN Sat Night

Chg Fee Non-Refund

1 0 0 0 0

2 3 0 1 0

3 7 1 0 0

4 10 1 1 0

5 14 1 1 1

6 21 1 1 1

MIT MIT ICAT ICAT

26

Revenue Impact of Each “Simplification”

30,000

35,000

40,000

45,000

50,000

55,000

60,000

65,000

FullyRestricted

Remove AP Remove SatNight Min Stay

Remove AllRestr, Keep

AP

Remove AllRestr and AP

-0.5

%

-29

.6%

-45

%

-16

.8%

MIT MIT ICAT ICAT

27

Loads by Fare Class

0

10

20

30

40

50

60

70

80

90

100

Fully Restricted Remove AP Remove SatNight Min Stay

Remove AllRestr, Keep AP

Remove All Restrand AP

FC 6

FC 5FC 4

FC 3FC 2

FC 1

81.6

87.8

79.8 82.7

88.1

MIT MIT ICAT ICAT

28

Revenues by Fare Class

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

FullyRestricted

Remove AP Remove SatNight Min Stay


AP


FC 6

FC 5FC 4

FC 3FC 2

FC 1

MIT MIT ICAT ICAT

29

Effectiveness of Traditional Leg RM

30,000

35,000

40,000

45,000

50,000

55,000

60,000

65,000

FullyRestricted

Remove AP Remove SatNight Min

Stay


AP


EMSRb

FCFS

8.2% 9.9%

11.8%

6.7%

0.1%

Percentage improvement over No RM Controls

MIT MIT ICAT ICAT

30

Summary – Impacts of Fare Simplification

Simplified fares have contributed to large revenue losses for US airlines

PODS simulated revenue losses in line with 15% impacts quoted by airlines

Fare class mix is also affected“Simplified” fare structures have changed the types of products passengers buy

The fundamental assumptions of RM systems:Are no longer appropriate under changing conditionsMay even be hurting airline revenues


MIT MIT ICAT ICAT

ADAPTING RM SYSTEMS AND MODELS

Peter Belobaba

MIT MIT ICAT ICAT

32

Existing Airline RM Systems Need to be Modified for Changing Fare Structures

RM systems were developed for restricted faresAssumed independent fare class demands, because restrictions kept full-fare passengers from buying lower fares

Without modification, these RM systems will not maximize revenues in less restricted fare structures

Unless demand forecasts are adjusted to reflect potential sell-up, high-fare demand will be consistently under-forecastOptimizer then under-protects, allowing more “spiral down”

RM system limitations are affecting airline revenuesExisting systems, left unadjusted, generate high load factors but do not increase yields

MIT MIT ICAT ICAT

33

Models for Undifferentiated Fares

Need to forecast demand by willingness to pay (WTP) higher fares with same restrictions (i.e., sell-up)

“Q-forecasting” approach requires estimates of passenger WTP by time to departure for each flight

Approach is to forecast maximum demand potential at lowest (Q) fare, and convert into “partitioned” forecasts for each fare class

Then, modified WTP forecasts can be fed as demand inputs to RM optimizers:

Standard EMSRb for Leg-based RMDynamic Programming methodsNetwork optimization methods for O+D Controls

MIT MIT ICAT ICAT

34

Example of Expected WTP Behavior

Typical values exhibit an S-shape reflecting the changing business/leisure mix across time frames

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Time Frame

FRA

T5 FRAT5 A

FRAT5 C

FRAT5 E

MIT MIT ICAT ICAT

35

Hybrid Forecasting For Simplified Fare Structures

• Separate forecasts for price and product oriented demandA passenger is counted as price-oriented if the next lower class from the one booked is closedA passenger is counted as product-oriented if the next lower class from the one booked was open.

• Combine standard RM forecasts and WTP forecastsFor product-oriented demand, bookings are treated as a historical data for the given class, and standard time series forecasting applied.For price-oriented demand, forecasts by WTP based on expected sell-up behaviorCombined forecasts fed into optimizers

MIT MIT ICAT ICAT

36

Impacts of Hybrid Forecasting

Airline 1 Hybrid Forecasting and EMSRb Airline 2 Standard Pick-up Forecasting and EMSRb

Airline 1 revenues increase by 1.36%, with greater protection for higher classes and fewer seats sold in classes 5 and 6, leading to lower Load Factor

01020304050

60708090

100

FC 1 FC 2 FC 3 FC 4 FC 5 FC 6

Boo

king

s Product-oriented

AP Price-oriented

RM Price-oriented

MIT MIT ICAT ICAT

37

New Forecasting and Optimization for Simplified Fare Structures

Combining Hybrid Forecasting and Dynamic Programming (DP) for optimization of seat inventory further improves revenues.

Revenues of Airline 1

490005000051000520005300054000

Trad RM DP DP-HFRM of Airline 1

+2.6%+4.1%

MIT MIT ICAT ICAT

38

Impact on Fare Class Mix: DP w/HF

Fare Mix of Airline 1

0

20

40

60

80

100

120

1 2 3 4 5 6

Fare Mix of Airline 1

0

20

40

60

80

100

120

1 2 3 4 5 6

LF = 80.4 LF = 77.4

Traditional RM DP w/ Hybrid Forecasts

DP with hybrid forecasting increases revenues by capturing more high yield passengers in middle and upper classes.

MIT MIT ICAT ICAT

39

Conclusions: RM Systems in “Simplified” Fare Structures

Relaxed fare restrictions increase the importance of effective RM controls to airline revenues

But, traditional RM methods do not maximize revenuesModifications required to better forecast consumer choice

New approaches to “hybrid” forecasting of price- vs. product-oriented demand show good potential

Incremental revenue gains over traditional RM methods

Need to estimate passenger WTP, affected by competitor’s RM method and seat availability

Focus of current research is how to actually ESTIMATE these values, required to generate the modified forecasts

jup presentation, winter '98, faa tech...

Documents