Revenue Management for Airline gAlliances

H. JainMIT, Cambridge, MANovember 4 2010November 4, 2010

Global Alliance Market Shares

Available Seat Kilometres 2010 Revenue Pass. Kilometres 2010

Star Alliance26.2%Others

41.9%

Star Alliance26.4%Others

41.1%

SkyTeam16.1%

oneworld15 8%

SkyTeam16.6%

oneworld15.8%

S Sk T ld

ASK(Bn)

15.9%

S Sk T ld

RPK(Bn)

Star SkyTeam oneworld

1569.1 963.9 944.6

Star SkyTeam oneworld

1205.1 755.1 725.1

Source: Airline Consolidation, Dr. Olaf Backofen, Deutsche Lufthansa AG, MIT, June 12, 2010

2

Revenue Management for Alliances

Alliances formed with a goal of increasing revenues for the member airlines

Alli t d th i t k b Alliance partners expand their network coverage by use of codeshare on each other’s flights

Sub-optimal benefits or potentially negative effects can arise from:

Lack of joint network optimization solutionLack of joint network optimization solutionPartners using arbitrary codeshare valuation in their Revenue Management (RM) systemsDifferent RM capabilities of each partner technical Different RM capabilities of each partner, technical distribution system constraints

3

Codeshare Example

Operated Flights: UA101 LH202Operated Flights: UA101 LH202

LAX BOS FRA

Codeshare: LH*2101 UA*1202

Seats must be made available by RM systems of both operating carriers to accept the codeshare b ki LAX FRAbooking: LAX-FRA

4

PODS ALLIANCE NETWORK: Cities & Hubs

BZNGEG

YVR

SEA

PDXMSP

BZNPDX

PHL

LGA

CLE

BOS

GRR

ANC

NRT

AMS

LHR

ORD

SFO

SLC

LAS

DEN STL

CLEPIT

MCI

CVGCMH

BWIDCAHNL

AMS

CDG

FRA

DFW

SAN

LAXSNA

LAS

PHX

MEMATL

TPAMSY

MIA

IAHLIM

BUE

MEX

5

Code Share Paths via ORD Hub

BZNGEG

YVR

SEA

PDXMSP

BZNPDX

PHL

LGA

BOS

GRR

CLE

ANC

NRT

AMS

LHR

SFO

SLC

DEN STL

BWIDCA

ORD PIT

CVGCMH

CLE

LAS

MCI

HNL

CDG

FRA

DFW

SAN

LAXSNA

ATLMEM

LAS

PHX

AUS

TPAMSY

MIA

IAHLIM

BUE

MEX

6

Code Share Paths via DFW Hub

BZNGEG

YVR

SEA

PDXMSP

BZNPDX

PHL

LGA

CLE

BOS

GRR

ANC

NRTAMS

LHR

ORD

SFO

SLC

LAS

DEN STL

CLEPIT

MCI

CVGCMH

BWIDCAHNL

CDG

FRA

DFW

SAN

LAXSNA

LAS

PHX

MEMATL

TPAMSY

MIA

IAHLIM

BUE

MEX

7

Examples of Double Connect CS Paths

BZNGEG

YVR

SEA

PDXMSP

BZNPDX

PHL

LGA

CLE

BOS

GRR

ANC

NRTAMS

LHR

ORD

SFO

SLC

LAS

DEN STL

CLEPIT

MCI

CVGCMH

BWIDCAHNL

CDG

FRA

DFW

SAN

LAXSNA

LAS

PHX

MEMATL

TPAMSY

MIA

IAHLIM

BUE

MEX

8

1. Different Levels of Information

Itinerary information (AVS and Cascading)BASELINE: Under standard AVS practices, operating airline does not know complete itinerary airline does not know complete itinerary “Cascading” gives both partners complete itinerary information for making availability decisions

Each alliance partner performs optimization for own network separately:p y

Separate network optimization assuming local fare valuation of code-share connecting passengers

9

Benefits of Network RM and Full InformationRevenues Compared to Baseline: Leg RM

1.20%

1.40%

0.80%

1.00%

CASCADING

0 20%

0.40%

0.60% AVS

0.00%

0.20%

PARTNER 1 PARTNER 2 ALLIANCE COMPETITOR

The control is sub-optimal for the alliance because of the arbitrary local fare valuation on codeshare pathsCascading leads to slightly higher revenues than AVS Cascading leads to slightly higher revenues than AVS (red stacks)

10

2. Codeshare Valuation

Valuation of CS bookings in RM systems affects:Own network because of potential displacement of own local and connecting trafficown local and connecting trafficPartner’s network due to interaction with their RM system and availability calculations for CS bookings

Two codeshare (CS) valuation schemes are compared:compared:

Local Fare Valuation: CS paths are valued at the local fares by each partner regardless of the total fareY Prorate Valuation: Total fare is divided exactly into Y-Prorate Valuation: Total fare is divided exactly into two parts, in the ratio of the Y-Prorates (highest fares)

11

Valuation Schemes

LAX BOS FRA

Booking (O‐D) Marketing Airline OD Fare

LAX‐BOS UA $ 200

BOS FRA LH $ 500BOS‐FRA LH $ 500

LAX‐FRA Codeshare(UA/ LH )

$ 600

Valuation of  Valuation of Airline LAX‐FRA

UA $ 200

LH $ 500

Airline LAX‐FRA

UA $ 150

LH $ 450LH $ 500

Total $ 700

LH $ 450

Total $ 600

12

Local and Y-Prorate ValuationRevenues Compared to Baseline: Leg RM

1.00%

1.50%

0 00%

0.50%

‐0.50%

0.00%

PARTNER 1 PARTNER 2 ALLIANCE

LOCAL FARE INPUTS

‐1.50%

‐1.00%Y‐PRORATE INPUTS

Y-Prorate leads to slightly higher gains for the allianceThough the difference in gains in small, the revenuecomponents are quite different in the two schemescomponents are quite different in the two schemes

13

Revenue ComponentsY-Prorate vs. Local Valuation

4000

6000

8000

)

Y‐Prorate Valuation compared to Local

‐2000

0

2000

LOC CNX CS LOC CNX CSte Change ($)

8000

‐6000

‐4000

2000

Absolut

‐10000

‐8000PARTNER 1 PARTNER 2

*Codeshare (CS) revenues are pre-resolution

Y-Prorate values the codeshare bookings at a lowervalue and hence take fewer codeshare bookings

Codeshare (CS) revenues are pre-resolution

14

3. Bid Price Sharing

B ki R t

Bid price = marginal network revenue value of available seat on each leg

Bid Price Inventory Control

Bid PricesPartner 1:

Booking Request

Decision

Computation ControlAt the end of each time

frame

Separate Optimization

Partner 2:Booking Request

Bid Price Sharing

Bid Price Computation Bid Prices

Inventory Control

Decision

15

Bid Price Sharing ResultsRevenues Compared to Baseline: Leg RM

1 00%

1.50% Incremental gain of 0.3% using Bid price sharing is equivalent to $ 200M for an alliance like United-Lufthansa

0.50%

1.00%

NO BPS

0.00%

LOCAL Y‐PRORATE LOCAL Y‐PRORATE LOCAL Y‐PRORATE

WITH BPS

‐0.50%PARTNER 1 ALLIANCEPARTNER 2

‐1.50%

‐1.00%Bid Price sharing yields higher gains for both Local and Y-Prorate Valuation

16

Next Step: Dynamic Codeshare Valuation

Until now, only own airline bid prices are used for the network optimization by each partnerIncorporating estimates of the value of a partner’s seat Incorporating estimates of the value of a partner s seat into own optimization gets closer to the joint network revenue solution

Booking Request

Bid Price Comp.

Inventory Control

Bid Prices

Partner 1:Decision

At the end of each

Bid Price Comp.

Bid Prices

Separate Optimization

Booking Request

time frame

Use the Partner’s Bid Prices in valuation:

Next Time Frame

Total Fare - Partner’s Bid Price

Bid Price Comp. Bid Prices

Partner 2:

Inventory Control

oo g equest

DecisionBid Price Sharing

Bid Price Comp. Bid Prices

17

Conclusions

Airline revenue gains can be affected by alliances:Valuation scheme of code share passengers affects seat availability decisions on both partner networks With separate and uncoordinated RM, one partner can benefit more than the other

Information sharing improves revenues: Cascading yields higher revenues than AVSBid i h i i ld b t ti ll hi h Bid price sharing yields substantially higher revenues, of the order of $ 100M (each) for big alliance carriers

Dynamic codeshare valuation using bid prices can Dynamic codeshare valuation using bid prices can lead to even greater revenues

18