Fundamentals of QSI. Khalid Usman Jan 28, 2014 AVIATION, AEROSPACE & DEFENCE 2012 OLIVER WYMAN

Transcription

1 Fundamentals of QSI Khalid Usman Jan 28, 2014 AVIATION, AEROSPACE & DEFENCE 2012 OLIVER WYMAN

2 Agenda QSI overview Uses of QSI QSI methodology Framework of QSI forecasting Summary and Discussion Oliver Wyman 1

3 Section 1 QSI Overview

4 QSI Overview QSI stands for Quality of Service Index History dates back to 1970 s where it was developed by the Civil Aeronautics Board (CAB) Based on the product attributes, quantifies the relative attractiveness of consumer choices QSI is a quantitative score that is assigned to each travel choice available to the traveler Higher QSI score would mean that the itinerary is preferable Lower QSI score would mean less attractiveness QSI scores then help determine market share that each itinerary will capture Market share : Itinerary QSI / Market QSI Where Market QSI is the total QSI in the market Oliver Wyman 3

5 QSI Overview QSI Models are calibrated based on historical data for passenger travel Revealed preference (ticketing/booking) vs. stated preference (survey) Ticketing/booking data is used for calibration QSI methodology originated in deregulated era where fares were not a determinant of choice Currently the landscape has changed, with fares being a very important determinant of choice QSI methodology still proves to be robust Very important to properly calibrate and maintain the models Alternate methodologies to QSI are available Logit modeling An advanced technique that has been used in transportation modeling and planning Oliver Wyman 4

6 Section 2 Uses of QSI

7 Uses of QSI QSI is used to forecast passenger demand on flights, and associated revenue and profit/loss QSI is generally an integral part of Network planning model (or called Network-simulation model/schedule profitability forecasting models) Mainly used to answers questions such as: Market share that a new flight will capture Load factors and passengers carried on a flight Local and connecting passenger mix, and what O&D s are participating in the flow traffic Diversion of passengers from competitors with new flight introduction Revenue associated with the flight (local revenue and connecting revenue) Network contribution and P&L (achieved through cost modeling) Oliver Wyman 6

8 Uses of QSI Different entities in the aviation industry use QSI for answering important business questions Airlines Airports Manufacturers Government/ Regulatory authorities Forecast route load factors and passengers Passenger mix: local/connecting Network flow Revenue forecast P & L Fleet decisions Mergers and acquisition analysis New route opportunities for airports Presentation of business cases with same methodology that airlines use Fleet changes and more economic flying/matching of supply and demand with fleet types Competition studies, merger analysis Oliver Wyman 7

9 Section 3 QSI Methodology

10 Factors affecting QSI Typically QSI models will include the following factors: Level of service - Non-stop - Direct - Single connection - Double connection Codeshare and Interline factor Aircraft size preference, penalty for turbo-props Departure time preference Elapsed time preference Longer connection time penalty Airline preference factors Additionally, other factors can be included Oliver Wyman 9

11 QSI and market share calculations Market share models are based on itinerary scoring mechanism Illustrative Example: Albuquerque New York Hypothetical market with 4 itineraries/day between ABQ and JFK Non-stop itinerary Score: 7.0 One-stop itinerary Score: 0.27 ORD JFK Itinerary Index score Expected share Non-stop % 1-stop via ORD % ABQ DFW RDU Two-stop itinerary Score: stop via DFW, RDU % Total market % Invalid itinerary Circuity > 1.35 MCO *Any alternate airport effect such as LGA or EWR are not included for simplicity Oliver Wyman 10

12 QSI and demand allocation Total marketsize is split using predicted market share Illustrative Example: Albuquerque New York Hypothetical market with 3 valid itineraries/day between ABQ and JFK Marketsize/ Size of Pie ABQ Non-stop JFK QSI Sore: 7.0 Single Connection share: 3.7% QSI Model/ Share of Pie Double Connection share: 0.3% Forecasted Passengers Non-stop: PDEW Single-connection Double-connection DFW ORD RDU QSI Sore: 0.27 QSI Sore: Non-stop share: 96% Single connect: 3.7 Double connect: 0.3 Market QSI: 7.29 *Any alternate airport effect such as LGA or EWR are not included for simplicity Oliver Wyman 11

13 Determinants of QSI score Non-stop flights get highest QSI relative to connections In long-haul markets, connections get higher relative preference compared to short-haul markets Double connections get very low preference compared to non-stops and single connections Codeshare flights are becoming more and more important Bi-lateral JVs typically get higher preference than regular codeshares Differentiation between local codeshare and connection codeshare, local codeshares get less preference Interline flights get lower preference than codeshares Flights having higher elapsed time get lower preference, similarly flights having longer ground time get lower preference Oliver Wyman 12

14 Determinants of QSI score Aircraft size matters (regional jets and turbo-prop preference) Relation to distance Some aircraft types get higher preference, more in long-haul Time of day preference Important for business markets S-curve effects Can be built in QSI models, tendency is to treat any adjustments outside of model Airline preference factors and share gaps Some airlines perform better than their fair share Low cost carriers with lower fares than market averages get higher share Oliver Wyman 13

15 Non-stop flights get highest preference In US domestic markets with non-stop service, 90% of traffic is captured on non-stop flights (based on Q2, 2013 Db1b data) Top 10 US Mainland markets 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% ATLLGA BOSDCA FLLLGA JFKLAX JFKSFO LASLAX LASSFO LAXORD LAXSFO LGAORD Non-stop Single Connect Double Connect Oliver Wyman 14

16 Market distance is an important determinant of relative preference In short-haul markets non-stop flights get much higher preference as compared to connections Top 10 US Mainland markets 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% LASLAX LAXSFO BOSDCA LASSFO LGAORD ATLLGA FLLLGA LAXORD JFKLAX JFKSFO Non-stop Single Connect Double Connect Non-stop Miles Oliver Wyman 15

17 How QSI parameters are calculated Historical data is used for calibration of parameters Historical data is used to calculate QSI parameters Typically data used is DOT Db1b data, MIDT and PaxIS data Co-efficients are based on large amounts of passenger booking or ticketing data Different models are calibrated: Level of service Elapsed time Codeshare/interline preference Ground time penalty Aircraft size Typically, regression modeling and statistical curve fitting techniques are used to calibrate parameters Oliver Wyman 16

18 QSI can be calculated for limited set of markets or the whole world In an airline network, each flight gets flow traffic from large number of O&D markets A world-wide itinerary QSI file has several million itineraries Jun 2013: 7 million+ itineraries File size over 800 Mb Example of a QSI data file is given below Oliver Wyman 17

19 Section 4 Framework of QSI forecasting

20 QSI Forecasting Process The QSI model takes schedules, airport details, and alliance information into consideration as it builds itineraries and QSI scores QSI market share models are only one component of the overall forecasting process As airport planners and air service development practitioners, it important to understand the main components of the forecasting process QSI calculations for particular O&D markets can be done manually, however due to complexity of forecasting flights that involve network flow it becomes impractical First step is to take the input data and build all possible itineraries (choices) based on the business rules/constraints Spill & recapture is another component of the modeling that should be used as capacity constraint Oliver Wyman 19

21 QSI Forecasting Process Forecasting process goes through a series of steps to produce final output Input Schedule: SSIM, OAG/Innovata, Codeshare Airport Specific Details: Minimum connection time (MCT), airport reference list Demand and fare databases: DOT data, MIDT, PaxIS, Vendor calibrated Connection generator Build all possible connection itineraries: Online, codeshare, interline Connections are subject to: MCT, circuitry rules, Maximum connection time limits Cabotage, traffic rights are handled through traffic suppression codes QSI Scoring Assignment Service levels (non-stop, single connection, double connection, through flight) Aircraft size (Mainline, RJ, turbo-prop) and other preferences Codeshare and interline preference Elapsed time and ground time preference Share gaps and airline preference Market share calculation and demand allocation Market share for each itinerary is calculated based on the QSI score generated Itinerary demand is calculated based on the itinerary markets share Segment allocation & Spill and recapture Itinerary demand is allocated to flight segments Available demand is constrained by the capacity of the aircraft Boeing spill model is used Final Output Forecast is obtained at the flight/market level Details of network flow available Misconnection analysis, revenue diversion analysis Alliance valuation and codeshare analysis Oliver Wyman 20

22 Different Levels of route forecasts While flight/route centric forecasts are important network contribution should not be ignored which provides better system view Basic Passenger/Load Factor Provides the basic guidance whether operating flight in a new market, or adding frequency makes sense or not Revenue Differentiates markets based on the average fares paid Cargo and ancillary revenue are important contributors. Markets operated by wide-body aircraft could have significant cargo revenue Comprehensive P &L Forecasts While passenger and resulting revenue are important metric, final evaluation criteria is flight profitability Includes costs also: P& L = Revenue - Cost Route profitability Short-term decision horizon Route profitability with aircraft ownership cost Fully allocated costs (including ownership) Long-term decision horizon Oliver Wyman 21

23 QSI forecasting software Software packages utilize input data to produce market forecasts While QSI forecasting software provides a lot of advantages, however it should be used with proper caution Calibration of the model is very important Comparing historical actuals with historical forecast Applying share gaps and any other required adjustments Input data is most critical, bad data feed will provide bad decisions New markets require stimulation study to stimulate demand generally done outside the model, although automated features can be built within model Proper maintenance of the model is required for getting good results and output. Includes parameter calibration, marketsizes, schedule and MCT data updates Oliver Wyman 22

24 QSI forecasting software QSI Software allows modeling of network flows. Example UA ORD-SFO flight Flight Composition Illustrative Double Connections 13% SFO ORD Single Connections 44 % Local O&D 43% Network flow over flights means that each flight will have passengers from several to several hundred markets ORD-SFO flight serves a lot of connecting markets both domestic and international Since QSIs and shares need to be calculated for several hundred markets for a flight forecast, it is not practical via spreadsheets Powerful QSI software area ideally suited for forecasting complex airline networks Oliver Wyman 23

25 QSI forecasting schematic Important to understand the various components of forecasting Base Marketsize (historical) Forecasted Marketsize Boeing/Airbus Market Outlook, IMF GDP Forecasts Segment based Load factor information (Internal) True O&D passenger and fare information (Internal) Model Calibration Process (historical forecast vs. actual) QSI Forecasting Model Worldwide Connection Generator Market Share Model (QSI) Marketsize Stimulation Model Industry Schedule Data, Minimum connection time (MCT data) Estimation of Fares for new routes Spill and Recapture Model Schedule alternatives for Evaluation Cost Information (Internal) P&L Forecast Seat Information (Internal) Legend: Final Output: By Route,Market forecast Flight forecast Diversion, share analysis Internal data Oliver Wyman 24

26 Section 5 Summary and discussion

27 Summary QSI modeling helps build the business case for new services However, it should be accompanied by complete business case Important to utilize credible and similar data sources as airlines are using Market size is one of the most critical inputs to QSI models For US domestic markets, US DOT is the standard International market sizes are MIDT, PaxIS based or vendor calibrated Calibration of model is very important, modeling bias would distort the results Running the model against historical month and checking against actual data New markets or additional service may require marketsize stimulation Based on historical data Oliver Wyman 26

28 QSI Departures Interesting facts Top 15 QSI producers in the world 90,000 90,000 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 - DL UA AA US WN LH CZ AF AC MU FR TK BA CA U2 80,000 70,000 60,000 50,000 40,000 30,000 20,000 10,000 - Overall, QSI gets correlated with number of departures Airlines with higher network connectivity get higher overall QSI values since they generate more connecting service LCCs tend to get capture higher share than just predicted by schedule preference due to fare effect QSI (Weekly) Departures (Weekly) Oliver Wyman 27

29 About Oliver Wyman With offices in 50+ cities across 25 countries, Oliver Wyman is a leading global management consulting firm that combines deep industry knowledge with specialized expertise in strategy, operations, risk management, organizational transformation, and leadership development. The firm s 3,000 professionals help clients optimize their businesses, improve their operations and risk profile, and accelerate their organizational performance to seize the most attractive opportunities. Oliver Wyman is part of Marsh & McLennan Companies [NYSE: MMC]. For more information, visit Oliver Wyman s global Aviation, Aerospace & Defense practice helps passenger and cargo carriers, OEM and parts manufacturers, aerospace/ defense companies, airports, and MRO and other service providers develop value growth strategies, improve operations, and maximize organizational effectiveness. Our deep industry expertise and our specialized capabilities make us a leader in serving the needs of the industry. Also, Oliver Wyman offers a powerful suite of industry data and analytical tools to drive key business insights through For more information, please contact: Khalid Usman Khalid.usman@oliverwyman.com