Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
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
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
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
Short Haul Medium Haul Long Haul
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
Short Haul Medium Haul Long Haul
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?
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
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
Remove AllRestr, Keep
AP
Remove AllRestr and 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
Remove AllRestr, Keep
AP
Remove AllRestr and 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
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
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